Diabetes Methods in Molecular Biology Stocker 2020

Methods in
Molecular Biology 2076
Claire J. Stocker Editor
Type 2
Diabetes
Methods and Protocols
Second Edition
METHODS IN MOLECULAR BIOLOGY
Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, UK
For further volumes:

http://www.springer.com/series/7651
For over 35 years, biological scientists have come to rely on the research protocols and
methodologies in the critically acclaimed Methods in Molecular Biology series. The series was
the first to introduce the step-by-step protocols approach that has become the standard in all
biomedical protocol publishing. Each protocol is provided in readily-reproducible step-by-
step fashion, opening with an introductory overview, a list of the materials and reagents
needed to complete the experiment, and followed by a detailed procedure that is supported
with a helpful notes section offering tips and tricks of the trade as well as troubleshooting
advice. These hallmark features were introduced by series editor Dr. John Walker and
constitute the key ingredient in each and every volume of the Methods in Molecular Biology
series. Tested and trusted, comprehensive and reliable, all protocols from the series are
indexed in PubMed.
Type 2 Diabetes
Methods and Protocols
Second Edition
Edited by
Claire J. Stocker
Medical School, University of Buckingham, Buckingham, Buckinghamshire, UK
Editor
Claire J. Stocker
Medical School
University of Buckingham
Buckingham, Buckinghamshire, UK
ISSN 1064-3745 ISSN 1940-6029 (electronic)

ISBN 978-1-4939-9880-7 ISBN 978-1-4939-9882-1 (eBook)
https://doi.org/10.1007/978-1-4939-9882-1
© Springer Science+Business Media, LLC, part of Springer Nature 2020

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations
and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.
The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This Humana imprint is published by the registered company Springer Science+Business Media, LLC, part of Springer
Nature.
The registered company address is: 233 Spring Street, New York, NY 10013, U.S.A.
Preface
Diabetes is now reaching pandemic proportions, and the associated complications are often
life-threatening and disabling. Diabetes and its comorbidities are considered among the
biggest killers of UK society and by many estimate a worse threat than cancer. Over the last
decade, both clinical and basic researchers have increased our understanding of the physio-
logical and molecular mechanisms involved in the progression of diabetes and importantly
the relative roles played by both environmental and genetic factors.
The primary purpose of this edition is to provide up-to-date and relevant explanations of
commonly used protocols and analyses used in diabetes research. Specifically, the chapters
start with a series of overarching reviews and will provide its readers with simple explanations
of practical techniques or models in a short succinct format, thus allowing research scientist
and early career clinicians with quick and easy practical information to address a particular
question. There are key practical notes at the end of each chapter, as well as numerous
helpful figures and tables.
Buckingham, UK Claire J. Stocker
v
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
1 Sixty Years of Drug Discovery for Type 2 Diabetes:
Where Are We Now? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
John C. Clapham
2 Practical Considerations for In Vivo Mouse Studies . . . . . . . . . . . . . . . . . . . . . . . . . 31
Edward T. Wargent
3 Nutritional Models of Type 2 Diabetes Mellitus . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Beverly Sara Mühlh€ a usler, Carla Toop, and Sheridan Gentili
4 Cheminformatics in the Identification of Drug Classes
for the Treatment of Type 2 Diabetes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Paul W. Finn
5 Whole-Exome Sequencing (WES) for Illumina Short Read Sequencers
Using Solution-Based Capture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Milind C. Mahajan and Andrew S. McLellan
6 Gene Expression Mining in Type 2 Diabetes Research. . . . . . . . . . . . . . . . . . . . . . . 109
Donald R. Dunbar
7 Pathways Enrichment Analysis of Gene Expression Data
in Type 2 Diabetes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Maysson Ibrahim
8 Diagnostic Genetic Testing for Monogenic Diabetes and Congenital
Hyperinsulinemia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Jayne A. L. Houghton
9 Isolation and Purification of Rodent Pancreatic Islets of Langerhans . . . . . . . . . . 179
Jacqueline F. O’Dowd and Claire J. Stocker
10 Characterization of Islet Leukocyte Populations in Human
and Murine Islets by Flow Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Matthew J. Butcher, Michelle B. Trevino, Yumi Imai,
and Elena V. Galkina
11 Analysis of Histone Modifications in Rodent Pancreatic
Islets by Native Chromatin Immunoprecipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Ionel Sandovici, Lisa M. Nicholas, and Laura P. O’Neill
12 Quantification of Pancreatic Islets: Using Image Analysis Tools. . . . . . . . . . . . . . . 215
Parvathy E. Harikumar
13 Measurement of Calcium Signaling in Beta-Cell Lines
Using Epifluorescence and Confocal Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Joanne L. Selway
14 Nitric Oxide and Redox State Measurements in Pancreatic Beta Cells . . . . . . . . . 241
Rodrigo Carlessi, Vinicius Cruzat, Younan Chen,
and Philip Newsholme
vii
viii Contents
15 Primary Adipocytes as a Model for Insulin Sensitivity. . . . . . . . . . . . . . . . . . . . . . . . 255

Mohamed S. Zaibi
16 Assessing Islet Transplantation Outcome in Mice . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Aileen J. F. King and Chloe L. Rackham
17 Measurement of Insulin Secretion Using Pancreas Perfusion
in the Rodent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Edward T. Wargent
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Contributors
MATTHEW J. BUTCHER Department of Microbiology and Molecular Cell Biology, Eastern

Virginia Medical School, Norfolk, VA, USA
RODRIGO CARLESSI School of Pharmacy and Biomedical Sciences, Curtin Health Innovation
Research Institute, Perth, WA, Australia
YOUNAN CHEN Key Laboratory of Transplant Engineering and Immunology, NHFPC,
Regenerative Medicine Research Center, West China Hospital, Sichuan University,
Chengdu, PR China
JOHN C. CLAPHAM Medical School, University of Buckingham, Buckingham, UK
VINICIUS CRUZAT Faculty of Health, Torrens University Australia, Melbourne, Victoria,
Australia
DONALD R. DUNBAR Edinburgh Genomics, University of Edinburgh, Edinburgh, UK
PAUL W. FINN School of Computing, University of Buckingham, Buckingham, UK
ELENA V. GALKINA Department of Microbiology and Molecular Cell Biology, Eastern
SHERIDAN GENTILI Sansom Institute for Health Research, School of Pharmacy and Medical
Sciences, University of South Australia, Adelaide, SA, Australia
PARVATHY E. HARIKUMAR National Institute for Biological Standards and Control
(NIBSC) and Buckingham Institute of Translational Medicine, University of
Buckingham, Buckingham, UK
JAYNE A. L. HOUGHTON Royal Devon and Exeter Foundation Trust, Exeter, UK
MAYSSON IBRAHIM Clinical Trial Service Unit and Epidemiological Studies Unit, Nuffield
Department of Population Health, University of Oxford, Oxford, UK
YUMI IMAI Department of Internal Medicine, Strelitz Diabetes Center, Eastern Virginia
Medical School, Norfolk, VA, USA
AILEEN J. F. KING Diabetes Research Group, School of Life Course Sciences, King’s College
London, London, UK
MILIND C. MAHAJAN Sema4, a Mount Sinai venture, Stamford, CT, USA
ANDREW S. MCLELLAN Sema4, a Mount Sinai venture, Stamford, CT, USA
€
BEVERLY SARA MÜHLHAUSLER Food and Nutrition Research Group, Department of Food
and Wine Sciences, School of Agriculture, Food and Wine, The University of Adelaide,
Adelaide, SA, Australia; FOODplus Research Centre, School of Agriculture, Food and
Wine, The University of Adelaide, Adelaide, SA, Australia; CSIRO, Health and
Biosecurity, Adelaide, SA, Australia
PHILIP NEWSHOLME School of Pharmacy and Biomedical Sciences, Curtin Health
Innovation Research Institute, Perth, WA, Australia
LISA M. NICHOLAS University of Cambridge Metabolic Research Laboratories and MRC
Metabolic Diseases Unit, Institute of Metabolic Science, Addenbrooke’s Hospital,
Cambridge, UK
JACQUELINE F. O’DOWD Medical School, University of Buckingham, Buckingham,
Buckinghamshire, UK
LAURA P. O’NEILL Institute of Biomedical Research, College of Medical and Dental
Sciences, University of Birmingham, Birmingham, UK
ix
x Contributors
CHLOE L. RACKHAM Diabetes Research Group, School of Life Course Sciences, King’s College
London, London, UK
IONEL SANDOVICI Metabolic Research Laboratories, Medical Research Council Metabolic
Diseases Unit, Department of Obstetrics and Gynaecology, University of Cambridge,
Cambridge, UK; Centre for Trophoblast Research, Department of Physiology, Development
and Neuroscience, University of Cambridge, Cambridge, UK
JOANNE L. SELWAY Medical School, University of Buckingham, Buckingham, UK
CLAIRE J. STOCKER Medical School, University of Buckingham, Buckingham,
Buckinghamshire, UK
CARLA TOOP Sansom Institute for Health Research, School of Pharmacy and Medical
Sciences, University of South Australia, Adelaide, SA, Australia
MICHELLE B. TREVINO Department of Internal Medicine, Strelitz Diabetes Center, Eastern
EDWARD T. WARGENT Buckingham Institute of Translational Medicine, University of
MOHAMED S. ZAIBI Buckingham Institute of Translational Medicine, University of
Chapter 1
Sixty Years of Drug Discovery for Type 2 Diabetes: Where Are

We Now?
John C. Clapham
Abstract
Today, excluding insulin, there are eight classes of anti-diabetic medicines that have been added to the
pharmacy since the introduction of metformin in the mid-1950s; the sulfonylureas, biguanides, thiazoli-
dinediones, α-glucosidase inhibitors, meglitinides, incretins, and sodium glucose transport 2 inhibitors.
Does the fact that metformin is still first-line treatment suggest that our drug discovery efforts over the past
60 years have not been good enough? Or does it suggest that diabetes is such a complex disorder that no
single treatment, other than gastric bypass surgery, can affect true normalization of not only blood sugar
but also the underlying pathologies? Our understanding of the disease has most definitely improved which
may bring hope for the future in terms of science, but for it to be beneficial, this science has to be translated
into better drug treatments for the disease. In this review, I have examined the eight classes of anti-diabetes
drugs from a drug discovery perspective.
Key words Type-2-diabetes, Drug discovery, Metformin, Sulfonylureas, Biguanides, Thiazolidine-

diones, α-Glucosidase inhibitors, DPPIV inhibitors, GLP-1, SGLT2 inhibitors, Pharmaceutical
industry
1 Introduction
Modern drug discovery programs often start with a hypothesis,

based on some prior knowledge, that modulation of target X in
pathway Y will result in a positive therapeutic effect in a particular
disease area. The key to this approach is discovering the right target
in the right pathway in the right tissue which is amenable to specific
interaction with a ligand, a target characteristic that has been
referred to as “druggability” [1]. This is “target-based” drug dis-
covery and validates drug targets using molecular genetics techni-
ques [2], though this approach does have its critics [3]. Judging by
the huge failure rate, on top of patent expiration [4], this is chal-
lenging in practice, more so in complex diseases where the risks are
much higher. Normal glucose homeostasis and the etiology of type
2 diabetes are highly complex processes, so, predictably, drug
Claire J. Stocker (ed.), Type 2 Diabetes: Methods and Protocols, Methods in Molecular Biology, vol. 2076,
https://doi.org/10.1007/978-1-4939-9882-1_1, © Springer Science+Business Media, LLC, part of Springer Nature 2020
1
2 John C. Clapham
discovery in this area has been slow with relatively few new entrants
to the market in the 60 years since the introduction of today’s first-
line drug treatment, metformin.
Maintenance of a normal blood glucose level involves a
dynamic interplay between glucose absorption, glucose produc-
tion, and glucose utilization. These in turn are controlled by inter-
actions between circulating hormones (primarily insulin, though
there are many others) and the cellular processes involved in insulin
(and other hormone) signaling, glucose uptake, and glucose dis-
posal. The principal tissues involved in this interplay are the liver,
brain, skeletal muscle, and adipose tissue which in turn differ in
their patterns of substrate utilization, production, and recycling
which are critical to maintaining normal blood glucose levels in
the range of 3.8–6.1 mmol/L (68.4–109.8 mg/dL). Following a
meal, the degree to which blood glucose increases is a function of
the amount of glucose absorbed, the pancreatic insulin secretory
response, suppression of hepatic glucose output and increased glu-
cose uptake by insulin-sensitive (skeletal muscle, liver, adipose tis-
sue—65–70%) and insulin-insensitive organs (brain, kidney—30%)
[5–7].
Type 2 diabetes is the pathologic consequence of two concur-
rent and interacting conditions of insulin resistance and relative
insulin deficiency. On the one hand, insulin’s ability to suppress
hepatic glucose output and stimulate glucose uptake and utilization
is impaired (resulting in chronically raised insulin levels), and on the
other hand, the capacity of the pancreatic β-cell to maintain this
hyperinsulinemic state also begins to fail [8, 9]. Type 2 diabetic
patients are rarely hypoinsulinemic when compared to nondiabetic
individuals [8]. Another twist in the natural history of type 2 diabe-
tes story is that there is also a degenerative process that affects the
normal healthy existence of the pancreatic β-cell. In a landmark
study, it has been shown that after a point, a relatively small decrease
in β-cell mass is all that is required to have profound effects on
fasting blood glucose levels [10].
The Diabetes Control and Complications Trial (DCCT) and
the UK Prospective Diabetes Study (UKPDS) have shown us that
maintaining blood glucose levels to near-normal levels significantly
improves type 2 diabetes and its ensuing complications
[11–13]. The American Diabetes Association has recommended
that treatment should reduce HbA1c levels to less than 7%
[14, 15]. This is a relaxation of earlier guidelines which recom-
mended reducing HbA1c to 6.5% or below [16] as such aggressive
treatment was found to be associated at best with no additional
cardiovascular benefit but at worst with increased mortality
[17, 18]. However, in practice, HbA1c of 7% or above should
initiate treatment regimens. Regulatory guidelines for new drug
treatments usually stipulate an endpoint for a clinically meaningful
and statistically significant reduction in HbA1c, usually at least 1%
Drug Discovery in Type 2 Diabetes 3
over placebo with noninferiority against standard therapy, usually

metformin.
There is a challenge that any anti-diabetic drug will face. The
natural history of type 2 diabetes occurs as a continuum of the
interplay described above and that occurs over many years. By the
time of diagnosis, the disease is already in its mid stages before any
drug treatment can be initiated. The costs of treating diabetes are
huge and the drugs bill is a favorite political punch bag. It is of
course true; in the UK, the diabetes drugs bill is expensive; in 2010,
it was around £2 billion. However, the cost of hospitalization as a
consequence of diabetes was £8 billion (Diabetes UK). It is worth
noting that drug treatment with oral anti-diabetic agents may
actually reduce the hospitalization costs [19].
“Type 2 diabetes therapies” is a topic that is extensively and
regularly reviewed. The aim of this chapter, therefore, is to provide
an overview of current drug therapy of type 2 diabetes resulting
from research dating back many decades but with a flavor of the
type of things that might influence drug discovery efforts in this
area. It will be confined to small molecule drug inventions (though
peptide drugs of the exenatide type are included). “Industrial”
drug discovery for type 2 diabetes appears to have started around
the mid-twentieth century and since then eight broad classes,
excluding insulin, of agents are on the market today. These are
presented below in the order of their introduction.
2 Sulfonylureas (First Generation)
Insulin secretion from the pancreatic β-cell occurs in response to an

increase in blood glucose levels after a meal. Briefly, intracellular
ATP levels in the pancreatic β-cell increase in response to a rise in
blood glucose levels and this causes closure of an inward-rectifying
ion channel, the ATP-sensitive potassium channel (KATP). This
depolarizes the β-cell plasma membrane, via calcium entry through
L-type (verapamil sensitive) calcium channels, and stimulates insu-
lin exocytosis [20].
Our understanding of the electrophysiology of the KATP chan-
nel, now combined with molecular and genomic insights, allows a
comprehensive description of how sulfonylureas work. The KATP
channel is actually a multi-subunit protein complex consisting of
four inward rectifying potassium channels (Kir6.2) that form a pore
and four regulatory subunits [21, 22]. There are two subtypes of
regulatory subunits, SUR1 and SUR2, which differ in their binding
affinities for sulfonylureas [23, 24]. The SUR1 regulatory subunit
is found in the pancreatic β-cell, while SUR2 is found in cardiac and
smooth muscle cells [25].
The anti-diabetic sulfonylureas stimulate insulin secretion from
the pancreatic β-cells by mimicking the effect of ATP in the β-cell to
4 John C. Clapham
block opening of KATP channels [26, 27]. They are termed insulin
secretagogues.
A first-generation sulfonylurea is tolbutamide. Tolbutamide
originated from a war-effort search for antibiotics and the unpleas-
ant side effect of blackouts resulting from hypoglycemia; while
unacceptable for an antibiotic, it turned out to be a new therapeutic
approach for the treatment of diabetes. Tolbutamide, and an ana-
log, carbutamide, were launched in the mid-1950s [28]. Some
studies have suggested that tolbutamide, because of a short dura-
tion of action, could have a place today in the treatment of elderly
type 2 diabetics [29]. The elderly type 2 diabetic patient is particu-
larly vulnerable to the consequences of hypoglycemic episodes that
can make them prone to falls [30] and cardiovascular issues [29]. In
practice, however, tolbutamide is rarely prescribed today and it
carries an FDA warning regarding cardiovascular mortality.
3 Biguanides
The only representative of this class of anti-diabetic drug remaining

in use is metformin. Although chemically synthesized, metformin is
derived from biguanide compounds obtained from the medicinal
plant Galega officinalis [31]. It was launched in 1957 as an oral
hypoglycemic agent. Other examples in this class were phenformin
and buformin.
Metformin is poorly absorbed from the gut [32, 33], a process
dependent on the plasma membrane monoamine transporter [34],
and large doses (500–2500 mg per day) are required for efficacy.
The metformin molecule is not metabolized into any other product
by the body and is excreted in the urine as the parent drug with the
half-life of elimination of around 5 h [35]. The liver appears to be
the main site of action for metformin and uptake into the liver is
also facilitated by transporters, primarily OCT1 [36] but its phar-
macodynamic mechanism of action in the liver has not been fully
elucidated. One wonders how far metformin would have advanced
through a modern drug discovery process based on current experi-
ence, yet metformin is probably the most widely prescribed anti-
diabetic agent in the world. The UKPDS showed that metformin
outperformed standard therapies, including sulfonylurea and insu-
lin, on all diabetic endpoints [12]. It is now a first-line therapy in
almost all markets and can be used as monotherapy or in combina-
tion [37, 38]. Indeed, as we shall see later, it has been used in
combination with all of the other classes of anti-diabetic drugs
because of its front-line role. Since its use is so widespread in a
patient population that may already suffer from other problems
requiring medication, attention needs to be paid to the potential
risk for drug–drug interactions. Its reliance on the action of trans-
porters for absorption and delivery to target organ means that
other agents using these transporters will influence metformin’s

activity.
Unlike sulfonylureas, metformin reduces raised blood glucose
levels only in the presence of hyperglycemia and without stimulat-
ing insulin levels [39–41]. Metformin exerts its anti-diabetic effect
primarily by enhancing the effect of insulin in suppressing gluco-
neogenesis and thereby reducing hepatic glucose output [42]. Sec-
ondarily, metformin increases muscle tissue insulin sensitivity
[43]. Information is being gleaned on the pathways involved in
the mechanism of action of metformin but its precise molecular
target is still unknown; we do not yet know what it binds to. One of
the first breakthroughs was reported in 2002 [44] which showed
that metformin, along with thiazolidinedione and rosiglitazone,
activated AMP-activated protein kinase (AMPK) by two different
mechanisms.
AMPK is a multi-subunit enzyme playing a central role in
cellular stress responses through marked pleiotropic effects on
metabolism [45]. It initiates cellular cascades to increase fat oxida-
tion, decrease anabolic pathways in fat metabolism, and enhance
glucose uptake in order to preserve cellular energy stores
[46]. Thus, AMPK orchestrates the flux of fatty acids away from
triglyceride synthesis and into β-oxidation via modulation of many
downstream proteins. However, metformin does not bind to the
AMPK complex directly in order to activate it; AMPK is sensitive to
the cellular milieu, in this case the AMP:ATP intracellular ratio. It is
suggested that metformin induces changes in the AMP:ATP ratio
by disrupting normal mitochondrial function by interfering with
complex 1 of the respiratory transport chain [47]. This alteration of
the energetic state of the cell elicits an effect on AMPK but inde-
pendently of a direct interaction with the AMPK molecule. Fur-
thermore, AMPK is itself upregulated by other kinases lying
upstream, including the LKB1 tumor suppressor protein kinase
[48], which, incidentally, may also be a pathway affected by met-
formin. A later study, however, has shown that metformin could
still inhibit gluconeogenesis in hepatocytes lacking either AMPK or
LKB1, perhaps via a reduction in glucose-6-phosphatase expres-
sion. Thus, the precise molecular target(s) of metformin have yet to
be elucidated—somewhat anachronistic in todays “Big Data”
world with emphasis on identifying precise molecular targets even
before a drug discovery project starts.
Another advantage of metformin is that it is associated with
weight loss in obese subjects with or without type 2 diabetes
[49]. This could be via activation of the oxidative pathways out-
lined above which can be thermogenic and increase energy expen-
diture [50] with an additional effect to reduce food intake [51] or a
combination of both.
A rare side effect of metformin is lactic acidosis. Lactic acidosis
has been reported to occur in patients with or without renal
6 John C. Clapham
insufficiency [52]. Metformin is extensively cleared via the kidneys

[33] so care must be exercised in the elderly or people with a history
of kidney disease. The effect of inhibiting complex 1 of the mito-
chondrial respiratory chain, while possibly explaining inhibition of
gluconeogenesis from lactate, may also contribute to the rare cases
of lactic acidosis [47]. Early competitors to metformin, phenformin
and buformin, were more potent and more efficacious but were
eventually withdrawn in the 1970s because of higher incidence of
lactic acidosis [31]. Metformin can also cause severe gastrointesti-
nal problems in some patients, possibly via an effect on bile salt
absorption [53], though this can be mitigated to a large extent
using a prolonged release formulation [54]. There have been occa-
sional reports of metformin-induced pancreatitis in both with [55]
or without [56] pre-existing renal disease.
An aspect of drug discovery that has been gaining prominence
over the last decade or so is the search for new indications for old
(in this case very old) drugs in a drug discovery paradigm called
“repurposing” [57]. The potentially positive effects of metformin
in cancer, polycystic ovary syndrome, and nonalcohol fatty liver
disease are in themselves well-reviewed areas [58–60].
4 Sulfonylureas (Second Generation)
This group is represented by the drugs glibenclamide (glyburide),

gliclizide, and glimepiride. Glibenclamide was approved by the
FDA in 1984. These agents are still used but predominantly as
second-line add-on after metformin. As a drug class, they have
been extensively reviewed and meta-analysis of clinical trial data
reveals that they lower HbA1c by around 1.5% [61]. Hypoglycemia
is a common issue with these agents but that is hardly surprising
since their mechanism of action is independent of any glucose-
sensing remaining in the β-cell or the prevailing blood glucose
levels. The liability for hypoglycemia seems to be greater for gly-
buride than the others. Weight gain is another common unwanted
effect of this class of oral anti-diabetic agent.
The second-generation sulfonylureas are metabolized in the
liver via the cytochrome P450 system and eliminated via the
kidneys. The biotransformation of glyburide in the liver results in
the production of active metabolites [62] which will be particularly
troublesome in patents with renal insufficiency. No active metabo-
lites of gliclizide and glimeripiride have been identified to date.
When drugs have been in clinical use for so long and in so many
patients, any lurking skeletons in the cupboard usually make them-
selves known. Drug discovery efforts are inconveniently, and fre-
quently, hampered by two major causes of attrition: failure in
efficacy, failure in safety or both [63].
4.1 Failure Although the meta-analysis reassures us that they are effective in
in Efficacy lowering HbA1c, both as monotherapy or on top of other agents,
the issue with sulfonylureas with regard to efficacy is durability. In
other words, while reasonable efficacy is seen at the start of treat-
ment, it declines with continued use [64, 65] in a phenomenon
called “secondary failure.” Secondary failure in response to sulfo-
nylurea appears to be specific as long-term treatment with sulfony-
lureas can selectively reduce the insulin secretory response to an
acute dose of another sulfonylurea but not to glucagon [66]. The
exact mechanism is not known though sulfonylureas appear to
induce β-cell apoptosis in cultured human islets [67].
4.2 Failure in Safety The other issue for the sulfonylureas is the increased risk of cardio-
vascular side effects, an issue that has been debated since the 1970s.
The Sulfonylurea glyburide was shown to have a greater risk of
cardiovascular mortality than metformin [68] in a study that raised
the question about the manner in which blood glucose can be safely
reduced. Since then, several studies, using metformin as a compar-
ator, have reported that a number second-generation sulfonylureas
share this cardiovascular risk [64, 65]. There also appears to be
differences between different sulfonylureas on the degree of risk
[67]. Increased cardiovascular risk with these agents is probably
beyond doubt now and has serious implications for developing
economies that rely on cheap drugs [69] as preferred treatments.
The increased cardiovascular risk is probably due to the inter-
action with cardiac KATP channels [70]. In patients with already
increased risk of cardiovascular disease, sulfonylureas acting on
cardiac KATP could mask ST-segment elevation causing opportu-
nities for life-saving interventions to be missed [71].
5 Thiazolidinediones
The thiazolidinediones, troglitazone, rosiglitazone, and pioglita-

zone, were launched in the late 1990s. They exert their anti-
diabetic effects principally by alleviating insulin resistance. Unlike
metformin, whose main site of action is the liver and secondarily
skeletal muscle, the thiazolidinediones act primarily on adipose
tissue and secondarily skeletal muscle. The molecular mechanism
of the thiazolidinediones is activation of the nuclear hormone
receptor, Peroxisome Proliferator Activated Receptor-γ (PPAR-γ).
There are a number of classes of PPAR [72] and all appear to be
involved in the regulation of metabolism; for example, PPAR-α
activates pathways for fatty acid oxidation [73]. The thiazolidine-
diones mentioned above were not a result of rational drug design
based on knowledge of their molecular target but rather through
the observation that certain drugs, fibrates, originally developed for
dyslipidemia and acting via PPAR-α, lowered blood glucose levels
8 John C. Clapham
but through a different, then unknown, mechanism. They were

well into their clinical development when PPAR-γ was identified as
their primary molecular target; the thiazolidinediones were identi-
fied by what we now call phenotypic screening, the precursor
methodology to target-based approaches. However, almost from
the start, they were beset with problems.
Troglitazone, which was approved in 1997, was shown to be
efficacious as monotherapy or in combination with sulfonylureas or
metformin [74]. But there were early signs of trouble; even in
clinical trials, there were indications of the hepatotoxicity [75]
that eventually led to its withdrawal from the UK market in 1997
and the US market in 2000. The background to this story and the
reasons for the almost immediate withdrawal by GlaxoSmithKline
in the UK and its persistence on the US and Japanese markets are
sobering reading, and underlies how different companies have very
different attitudes to patient safety risk [76]. The other thiazolidi-
nediones, rosiglitazone and pioglitazone, did not share this partic-
ular liability.
Rosiglitazone and pioglitazone are efficacious; they lower
HbA1c by between 1.0% and 1.5% over placebo control in mono-
therapy trials [65]. They did however cause fluid retention and an
increase in body weight [77]. The thiazolidinediones also lower
elevated free fatty acids, particularly in combination with metfor-
min [78, 79]. This is good and has the potential to reduce lipo-
toxicity which is thought to be a mechanism contributing to β-cell
dysfunction in type 2 diabetes [80, 81]. Long-term treatment of
fa/fa rats with rosiglitazone, using either prevention or interven-
tion protocols, has hugely beneficial effects on islet morphology
[82]. Together these sets of results suggest that the thiazolidine-
diones could have had the potential for improving β-cell health.
Sadly, we will never know for sure, as further calamity overtook this
class of agent.
Two very controversial meta-analyses suggested that rosiglita-
zone increased risk of myocardial infarction [83, 84]. The resulting
public outcry and media storm [85] condemned the drug and led
to its withdrawal by the company in 2007 even though an FDA
advisory committee did not think that there was sufficient evidence
to actually call for its withdrawal [86]. The methodologies in the
meta-analyses have been subsequently questioned and other studies
did not find the increased risk of myocardial infarction, and perhaps
even a positive finding for rosiglitazone [87, 88]. This controversy
seriously knocked the confidence of patients and physicians but it is
sad to note that the manufacturer no longer invests in in-house
diabetes drug discovery research. In late 2013, after reviewing the
data of Rosiglitazone Evaluated for Cardiovascular Outcomes and
Regulation of Glycemia in Diabetes (RECORD) trial [89], the
FDA removed marketing restrictions on rosiglitazone but the pat-
ent on rosiglitazone had expired in 2012.
But the pain did not end there. Pioglitazone sparked some
further concern surrounding increased risk of bladder cancer
[90]. Thiazolidinediones are also associated with fractures, particu-
larly in postmenopausal women [91] and macular oedema [92],
which appears to be reversible on discontinuation of therapy
[93]. Thiazolidinediones are available in their generic form but
really only as third-line therapies.
6 α-Glucosidase Inhibitors
Carbohydrates constitute a large proportion of our energy intake,

especially for those of us living in the Western world. Worryingly,
over the last 50 years or so, the proportion of refined sugars
entering our diet has also increased dramatically [94, 95] as we
consume more and more processed foods. The majority of the
carbohydrate consumed in food is polysaccharide in the form of
starch or oligosaccharides such as sucrose.
Monosaccharides, such as glucose, are readily absorbed
through the intestinal epithelium via the sodium-dependent glu-
cose co-transporter-1 (SGLT1) [96]. Oligosaccharides and poly-
saccharides are not readily absorbed and have to be broken down to
monosaccharides first. Starch is broken down into smaller oligosac-
charides by α-amylase in the small intestine. The resulting oligo-
mers (or dietary oligomers) are further broken down by a
membrane-bound enzyme family into (primarily) glucose, galac-
tose, and fructose [97] which can then be transported through the
gut epithelia and on into the blood stream. These latter enzymes
reside in the brush border of the small intestine and are the
α-glucosidases (formally known as the maltases) and they hydrolyze
O-linked glycosidic bonds [98].
As mentioned in the introduction with regard to a potential
drug discovery program, a clear hypothesis presents itself here;
sugar consumption is increasing and so is type 2 diabetes. In the
absence of restraint from ourselves or moderation of added sugars
to processed foods by manufacturers, inhibition of α-glucosidases
by drugs will reduce absorption of dietary sugar and the hypothesis
assumes that this will improve glycemic control, and this is essen-
tially what has happened. The α-glucosidase inhibitors, acarbose,
voglibose, and miglitol, are pseudo-carbohydrates and were
launched between 1994 and 1996. Acarbose and voglibose are
poorly absorbed and are consequently excreted in the feces [99]
while miglitol is fully absorbed and excreted as parent compound
by the kidneys [100].
Acarbose has been shown to be an effective monotherapy
where diet modification is insufficient to control fasting plasma
glucose. A marked reduction in postprandial hyperglycemia was
reported in one study where a 0.65% reduction in HbA1c after
10 John C. Clapham
24 weeks treatment was observed [101]. The efficacy of acarbose is

reasonable compared to other oral anti-diabetic agents such as
metformin and gliclizide [102, 103]. α-Glucosidase inhibitors
may also be useful in reducing insulin requirement of type 1 diabetic
patients [104]. Side effects of the α-glucosidase inhibitors are,
unsurprisingly, mainly gastrointestinal: diarrhea, flatulence, and
abdominal distension (meteorism).
The α-glucosidase inhibitors also appear to have a positive
effect on disease markers of cardiovascular dysfunction [105], an
effect that may be related to the suppression of postprandial hyper-
glycemia seen with these agents. Postprandial hyperglycemia is
associated with oxidative stress [106] and oxidative stress is linked
to vascular endothelial dysfunction [107, 108]. Thus, the
α-glucosidase inhibitors have the potential of being a useful
add-on treatment to existing anti-diabetic agents to improve post-
prandial hyperglycemia and perhaps could confer benefits with
regard to cardiovascular function.
7 Meglitinides
The insulin response to glucose occurs in two phases: an immediate

early phase increase within minutes and, a later, more prolonged
increase over 2–3 h. The biphasic nature of the insulin response is
more discernible in response to intravenous administration of glu-
cose but the episodic nature of the insulin response to prandial
glucose is less clearly defined [109]. The liver is an important target
organ for the early-phase insulin response, where insulin acts to
suppress hepatic glucose output [110, 111].
In type 2 diabetes, there is a loss of the early insulin response to
a meal and this may be a marker for impairment at the level of the
β-cell [112, 113]. Thus, restoring the first-phase insulin response is
an important therapeutic goal [114]. An early proof of concept to
this idea was shown in newly diagnosed type 2 diabetics; insulin was
infused to mimic the first-phase insulin response and resulted in a
marked reduction in elevated postprandial glucose excursions
[115]. Thus, the hypothesis that drugs that restore first-phase
insulin response will produce an important beneficial therapeutic
effect was a clear “go” signal for project work.
The meglitinides are short-acting insulin secretagogues that are
readily absorbed from the gastrointestinal tract [116]. This makes
them a much more suitable option than infusing insulin. Examples
of these agents are repaglinide and nateglinide launched in 1997
and 2000, respectively. Like the sulfonylureas, the meglitinides
reduce the probability of KATP channel opening in β-cells to depo-
larize the cell by allowing calcium influx through L-type calcium
channels [117]. Repaglinide and nateglinide also bind competi-
tively to the SUR1 sulfonylurea receptor with Ki of 100 nM and
240 nM, respectively, compared to 2.3 nM for glibenclamide

[118]. Furthermore, nateglinide differs from repaglinide in that it
dissociates from the SUR1 receptor very rapidly [118].
Clinical experience, either as monotherapy or in combination,
show that this class of anti-diabetic agent is effective, particularly
with respect to mealtime glucose excursions. For example, in a
double-blind, placebo-controlled study, repaglinide (0.5–1.0 mg)
taken at meal times improved glycemic control and reduced HbA1c
by 1.14% after 4 weeks of dosing without a significant effect on
body weight [119]. This study also highlighted the flexibility of
these agents; repaglinide has a rapid onset of action and short
duration and has utility where meal patterns may change from day
to day. In a 16-week head-to-head trial, repaglinide and nateglinide
were compared as monotherapies in type 2 diabetic patients
[120]. While both agents have similar effects on postprandial glu-
cose excursions, repaglinide was more effective at reducing HbA1c
and fasting plasma glucose though it did induce more weight gain
than nateglinide. While repaglinide and nateglinide both block the
KATP channel, their mode of interaction with the regulatory sub-
unit, SUR1, differs [121] which may partly explain their therapeu-
tic differences. Similarly, repaglinide was shown to be slightly more
effective that the sulfonylurea, glipizide [122].
The meglitinides are unlikely to be a first-line drug therapy; as
the majority of patients will be on metformin, however, where
metformin is contraindicated, these agents may be used [123] but
there are questions with regard to durability of effect and long-term
usefulness as monotherapy [124]. Meglitinides may still be a useful
add-on therapy in subjects where their diabetes is hard to control
despite diet, exercise, and metformin treatment. In such a case,
addition of repaglinide resulted in better control compared to
metformin (or repaglinide) monotherapy [125]. Weight gain was
reported as well as a greater incidence of mild-to-moderate hypo-
glycemic incidents.
The rapid onset and short duration of action of the meglitinides
seems to confer advantages over the second-generation sulfonylur-
eas with respect to postprandial glucose control, and their pharma-
cokinetic properties allow flexibility with respect to meal patterns.
As we saw with the α-glucosidase inhibitors, this property will have
other beneficial effects, particularly on cardiovascular risk, and this
does seem to be the case for repaglinide compared to the
sulfonylureas [126].
8 Incretins
8.1 GLP-1 Mimetics The discovery of the polypeptide hormones, glucose-dependent

insulinotropic polypeptide (GIP), and glucagon-like peptide-1
(GLP-1), followed from the well-known observation that oral,
12 John C. Clapham
but not intravenous, administration of glucose enhanced insulin

secretion from pancreatic β-cells [127]. These are the incretin
hormones and both GIP and GLP-1 are major contributors to
meal-induced insulin secretion [128]. GIP is secreted from duode-
nal K-cells while GLP-1 is secreted from L-cells located further
down the gastrointestinal tract at the distal ileum and colon
[129]. Glucose sensing by SGLT1 plays a crucial role in the stimu-
lus–secretion coupling in L-cells [130]. GLP-1 reduces the secre-
tion of glucagon from pancreatic β-cells to reduce glycogenolysis
while GIP increases glucagon secretion. In the blood stream, both
GIP and GLP-1 are rapidly inactivated by N-terminal cleavage of
two amino acids by dipeptidyl peptidase IV (DPPIV). Within the
2–4 min that the incretins are active, they interact with GIP recep-
tors and GLP-1 G-protein-coupled receptors on the pancreatic
β-cell to effect an increase in cAMP and ultimately insulin secretion
[131–133].
The biology of GIP and GLP-1 has been well documented and
the reader is referred to a comprehensive review where their biology
has been compared and contrasted [134]. The first objective for a
therapeutic agent activating the incretin pathway will be to have an
active half-life for considerably longer than the 2 min enjoyed by
GLP-1. This goal spurned a great deal of industrial activity to
screen and identify a small molecular agent to activate the GLP-1
receptor. Agonists are harder than antagonists to identify and Class-
B GPCRs are notoriously resistant to drug discovery efforts. This is
because of a large and complex N-terminal domain that appears to
shield the extracellular binding face of the receptor; to activate the
receptor, peptide ligands have to bind to both the N-terminal
domain and to the active site [135]. Since the GLP-1 receptor is a
Class-B GPCR and an agonist is the therapeutic modality, the odds
were stacked against small molecule discovery efforts from the start.
Small molecule approaches have been traditionally preferred
because they, in the main, are oral therapies. There have been
many attempts to identify small molecule activators of the GLP-1
receptor using the shotgun approach of high throughput screen-
ing. It is likely that millions of compounds from a variety of libraries
from several companies were screened to largely no avail. However,
there are some examples of small molecule agonists of the GLP-1
receptor [136]. Molecules of interest, arising from research in
China, are substituted cyclobutane compounds which appear to
avoid the barriers to small molecule activation of the GLP-1 recep-
tor. The lead compound, designated Boc5, appears to exert GLP-1
agonistic effects in animal models of diabetes [137, 138]. Sadly,
these are non-druggable molecules [136], meaning that there is
little or no scope for optimization, but they at least encourage us
that a small molecule approach is not as improbable as we originally
believed.
However, in the present world, the main therapeutic GLP-1

agonists are all peptidic, and therefore require injection. Analogs of
GLP-1, such as exenatide which was approved in 2005 and liraglu-
tide which was approved in 2010, were the results of the effort to
find longer acting GLP-1s. Unlike the very short half-life of native
GLP-1, exenatide has a half-life of around 2.5 h [139] which still
necessitates multiple daily doses. Liraglutide, a fatty acid derivative
of GLP-1, was approved in 2010 and displays a much longer half-
life as a result of this modification [140]. The HbA1c lowering of
GLP-1 mimetics is related to half-life, and a longer acting version of
exenatide was introduced which allowed for once weekly
dosing [141].
The anti-diabetic effects of exenatide and liraglutide have been
extensively reviewed [142] and have been shown to be effective
treatments in trials either as monotherapy or in combination; in
real-world medicine, they are more likely to be given in combina-
tion. It is possible that liraglutide outperforms exenatide with
respect to HbA1c [143] but that the effect is marginal and the
longer acting version of exenatide may mitigate this difference.
These agents have some interesting side properties that are benefi-
cial—in the future, these additional properties (over and above
HbA1c lowering) will determine the commercial success of any
agent for the treatment for type 2 diabetes. Exenatide has been
shown to lower body weight, most likely mediated by a central
effect on appetite [144, 145]. This may be very useful in the
treatment of hyperphagic obesity of hypothalamic origin
[146, 147]. Given the experience with the thiazolidinediones and
sulfonylureas, this is a desirable add-on effect in drug discovery
programs.
β-Cell decompensation [148] and declining β-cell mass [10]
are key morbidities in the type 2 diabetes continuum. Therapeutic
agents that could reverse this decline in β-cell mass would be game
changers as this would result in the in vivo generation of new
β-cells. As we have seen, the thiazolidinedione drug class may
have had this ability via an attenuation of lipotoxicity, but the
world will never really know. However, the general idea of reversing
the decline in β-cell mass is generating plenty of interest as a
biological effect area in its own right with the realization that the
pancreas is actually flexible enough to generate new β-cells either by
replication or by neogenesis [149].
It seems that drugs acting via the incretin system appear to have
very interesting effects on β-cell biology with the finding that
GLP-1 mimetics appear to improve measures of β-cell function
[150] or responsiveness [151] in diabetic patients. However, this
sort of finding does not imply the effects are due to neogenesis
which is currently impossible to measure in living patients. Experi-
mental data from animals [152] and from isolated human islets
[153] suggest that this is an area indeed worthy of continued
14 John C. Clapham
study especially in the clinical setting. In another twist to the

biology of GLP-1 mimetics, exenatide has been shown to reduce
β-cell apoptosis in neonatal rat or human isolated islet preparations
[154, 155], one of the destructive influences on β-cells in type
2 diabetes [156]. Sadly, since β-cell depletion by apoptosis is a
process that occurs over many years, it is hard to envisage a viable
clinical study protocol designed to confirm that an anti-apoptotic
effect contributes to improvements in β-cell function or mass. As
we have seen before [10], after a certain point, only a small reduc-
tion in β-cell mass results in overt hyperglycemia. Conversely, per-
haps only a small increase in β-cell mass could delay the switch-over
to overt, insulin requiring diabetes. However, expansion of β-cell
mass in mice may be different and perhaps more amenable than in
human islets as the difference in islet structure, and perhaps behav-
ior, between the two species is marked [157–159].
Despite the promise of some very interesting pharmacology
with regard to glycemic control, weight loss, and perhaps β-cell
mass, a potentially very serious safety concern has been flagged and
this has been hotly debated; acute pancreatitis and pancreatic can-
cer. If probable expansion of β-cell number is due to a proliferative
effect, could this occur in other cell types?
Pancreatitis was reported as a potential risk by the FDA in late
2006 and has been subsequently confirmed in many studies
[160]. This article summarizes these reports and offers a plausible
mechanism for this side effect. Briefly, GLP-1 receptors are present
in pancreatic duct cells [161] and following chronic stimulation of
these receptors during treatment with GLP-1 mimetics, the duct
cells may proliferate and block pancreatic ducts leading to pancrea-
titis and ultimately pancreatic cancer [160]. The degree of risk
posed by these agents has been challenged on the basis that the
data neither prove nor disprove the hypothesis that the acute pan-
creatitis progresses to malignant disease [162]. Nevertheless, dia-
betes is itself a risk factor for pancreatitis [163] and extreme care
must therefore be taken if agents have the potential to superimpose
risk where there is already vulnerability.
8.2 DPPIV Inhibitors An alternative strategy to the very challenging approach of produc-
ing Class-B GPCR agonists is to extend the half-life of the endoge-
nous hormones; this has been achieved by inhibiting the DPPIV
enzyme. This is generally an easier option for the drug hunter and
pharmaceutical industry.
DPPIV is a 766-amino acid glycoprotein and is well conserved
across species, which is very useful from a drug discovery perspec-
tive to reduce the number of screens required. It is a serine prote-
ase, so called due to the presence of a serine residue in a catalytic
triad comprising serine-624, aspartic acid-702, and histidine-734.
As a serine protease, DPPIV removes a dipeptide from the
N-terminus of peptide hormones of around 30 amino acids in
length and recognizes its substrates via amino acid motifs. For
DPPIV, peptides which have proline or alanine as the penultimate
amino acid are favored substrates (i.e., N2H-x-Pro-). Although well
conserved across species, DPPIV lacks similarity to the classical
serine proteases such as chymotrypsin [164]. However, there are
other peptidases that favor the N2H-x-Pro- motif, DPPVIII, and
DPPIX, where inhibition of these enzymes, as an off-target effect,
could result in severe adverse effects [165].
DPPIV is a type 2 cell surface protein in terms of its spatial
arrangement; most of the heavily glycosylated molecule is exposed
to the extracellular space and is anchored to the cell membrane by
virtue of a transmembrane helix anchor [166]. A soluble form of
DPPIV, which lacks the anchor, was discovered in human plasma
[167]. In the context of its therapeutic role, its most important
substrates are GIP and GLP-1. However, neuropeptide Y (NPY),
peptide YY, Substance P, and chemokines such as RANTES are also
substrates; indeed, we have found that DPPIV inhibition enhances
the antilipolytic activity of NPY in human adipose tissue [168]. The
crystal structure of DPPIV was solved in 2003 [169] and subse-
quently many structures have been published co-crystallizing
DPPIV with small molecule ligands and peptides [170]. This is a
hugely valuable resource in drug discovery as it increases confidence
and facilitates rational drug design.
Therapeutic inhibitors of DPPIV fall broadly into three types:
reversible substrate analogs (no example on market), covalently
bound substrate analogs such as vildagliptin and saxagliptin and
reversible nonpeptidic heterocycles such as sitagliptin [171]. The
development path for launch of sitagliptin is worthy of mention.
Sitaglitptin (MK-0431) was nominated as the lead candidate for
development in January 2002 and by only 2006 was approved by
the FDA and by the EMEA the following year; this was a spectacu-
lar achievement for Merck and was an example to the industry of
what clear planning and crisp decision-making looks like. Sitagliptin
was the first DPPIV inhibitor to market.
This drug class has been extensively reviewed (for example, see
[165, 172–175]). All marketed DPP-IV inhibitors show augmen-
tation of GIP and GLP-1 levels and produce broadly similar reduc-
tions in HbA1c of around 1–1.5%, close to that produced by
metformin. However, metformin is superior with respect to fasting
plasma glucose levels. DPP-IV inhibitors have a low risk for hypo-
glycemia in monotherapy and are weight-neutral. All DPPIV inhi-
bitors can be given with other oral hypoglycemic agents, and as the
diabetes advances, they can be given with insulin in the late stages of
the disease where perhaps combination with metformin fails to
produce the desired level of glucose control [176].
As expected for agents working in a common pathway, similar
effects of DPP-IV inhibitors to GLP-1 mimetics on β-cell biology
have been reported. Clinical studies in type 2 diabetics also show an
16 John C. Clapham
apparently beneficial effect on the β-cell when using insulin: proin-

sulin ratio as a surrogate of β-cell function [177–179]. Animal
studies, where β-cell mass can be measured directly, show that
DPP-IV inhibition preserves (perhaps increases) β-cell mass
[180]. Also not surprising is that acute pancreatitis, and the same
associated issues, has also been reported with the DPP-IV inhibi-
tors in some [181, 182] but not all [183] studies. While there are
no reports of improved cardiovascular outcomes, DPP-IV inhibi-
tors as a class, with perhaps the exception of sitagliptin, appear to be
relatively free from major adverse cardiovascular events [184].
Where there are agents with different modalities acting in the
same pathway, comparison is inevitable. In one study comparing
outcomes of GLP-1 receptor agonists and DPP-IV outcomes, the
conclusion was that the GLP-1 receptor agonists were superior in
terms of HBA1c and body weight reduction [185]. They also
conclude, quite diplomatically, that since DPP-IV inhibitors can
be taken orally rather than by injection and where overweight is not
a medical issue, DPP-IV inhibitors may be preferred. Nevertheless,
it does indicate that there is useful flexibility in treatment options.
9 SGLT2 Inhibitors
The SGLT2 inhibitors are the latest oral anti-hyperglycemic agents

to be approved and launched. Their rationale is based on the
hypothesis that increasing glucosuria will facilitate glycemic control
by removing glucose from the blood. In healthy individuals, plasma
glucose levels are maintained between 3.8 and 6.1 mmol/L
(68.4–109.8 mg/dL). All plasma glucose is filtered freely through
the glomerulus and into the proximal convoluted tubule, yet less
than 1% of this glucose enters the loop of Henle, glucose is virtually
all reabsorbed into the circulation. Glucose is a highly polar mole-
cule; it will not diffuse across biological membranes on its own and
a sodium–potassium ATPase is required to power a sodium con-
centration gradient that provides the drive to transport glucose
across the membrane via a sodium–glucose co-transporter. There
is a family of renal sodium–glucose transporters and their molecular
biology and biochemistry are very well reviewed [186, 187].
The highly industrialized process of glucose reabsorption from
the tubular lumen occurs in the earlier part of the proximal convo-
luted tubule where over 80% of all the glucose is reabsorbed
[186]. It is mediated by a high-capacity, low-affinity sodium–glu-
cose co-transporter (SGLT2) present in the luminal brush border
membrane. Further along the proximal tubule, there is a
low-capacity, high-affinity sodium–glucose co-transporter
(SGLT1) which accounts for the remaining glucose removal.
Although SGLT2 and SGLT1 differ in their capacities and affi-
nities, they both operate via a common mechanism [188]. SGLT2
was cloned in 1992 [189] and is expressed at very high levels in the
early proximal tubule [190], very suggestive of a bulk
transport role.
Over the course of 24 h, the kidneys of a normal individual will
have reabsorbed over 160 g of glucose [191] which equates to
“energy conservation” of the order of 2560 kJ (600 kcal) per day.
This degree of reabsorption is well within the maximum renal
absorptive capacity for a normal individual and gives the system
resilience. However, when plasma glucose levels exceed
10.0 mmol/L (180.0 mg/dL), as seen in diabetes, glucose starts
appearing in the urine, a condition known as glucosuria.
Before significant investment is committed to initiating and
progressing drug discovery programs, various levels of comfort
are required. The highest level is of course when a first-in-class
drug is successfully launched; but then it really should be too late
for anyone else to follow because of the decade-long lead-time.
Commonly, in drug discovery projects, some form of human target
validation is sought early in the projects lifetime, perhaps even
before a formal investment decision. For SGLT2, there was luck
in two respects. First, a form of human target validation presented
itself in the form of loss of function mutations (human gene knock-
outs if you will) that exist and result in glucosuria [192, 193]. Sec-
ond, there was a competitive inhibitor that has been known to
science since the mid-1880s: phlorizin, a naturally occurring dihy-
drochalcone glucoside found in the bark of pear, apple, and cherry
trees. In a study conducted in the 1930s, phlorizin was reported to
promote glucosuria and reduce plasma glucose levels in humans
[194]. In a reversal of the normal linear process, 50 years elapsed in
order to find a study in animals where phlorizin was shown to
ameliorate hyperglycemia and, interestingly, improve insulin sensi-
tivity in partially pancreatectomized diabetic rats [195]. Phlorizin
was subsequently found to be a competitive inhibitor of SGLT2
with a Ki of around 220 nM [187]. For a project start-up, this is
“gold dust”!
However, phlorizin has a number of issues: [1] it is poorly
absorbed, [2] it is nonselective with regard to other SGLTs, [3] it
is poorly bioavailable, and it needs to get to its target, and [4] has
unpleasant gastric side effects probably due to point 2. These are all
features that can be fixed by a decent medicinal chemistry approach
and indeed the currently approved SGLT2 inhibitors used phlor-
izin as a molecular starting point. Early attempts were not so
successful since they, like phlorizin itself, were O-linked glycosides
and very prone to rapid metabolism [196]. This was solved by
changing the O-linked bond to a C-glucosidic bond to yield cana-
gliflozin [197], dapagliflozin [198], and empagliflozin [199] which
gained FDA approval in the order cited here between late 2013 and
autumn of 2014. The structural parameters in order to mass-
produce SGLT2 inhibitors are well defined [200] as evidenced by
18 John C. Clapham
these approvals occurring in rapid succession with yet others wait-

ing in the wings [201, 202]. This may well be a drug class where the
“me-too” criticism may well resurface. Differentiation will then be
a challenge to get a decent return on investment and additional
benefit to the patient.
However, depite being on the market for a relatively short time,
there have been several reviews of the SGLT2 inhibitors in diabetes
[199, 201–204]. As a class, they appear to have good pharmacoki-
netic properties [205, 206] and access their molecular target from
the luminal side of the proximal tubule [207]. They are well toler-
ated [205, 208] and reduce HbA1c by the usual 1% or so over
placebo [197, 199, 203]. They also appear to be good, perhaps
excellent, add-on therapy to metformin, glimeperide, sitagliptin,
and insulin [203, 204].
Given that all of the energy contained in glucose is not recycled
following SGLT2 inhibition, weight loss is perhaps not surprising
[209, 210] and could even be a useful add-on to problematic
agents like sulfonylureas where weight gain is a concern [211].
One potentially serious issue that needs careful monitoring is
the increased risk of urinary tract infections and an increase inci-
dence has been reported for all members of this class [205, 208]. In
an analysis of the literature sponsored by one of the manufacturers,
they could not find a definitive dose relationship between SGLT2
inhibitors and urinary tract infection [212]. Of course, the infec-
tion risk may well be real, just not related to the dose of drug; it
boils down to a question of the integrity of protocol design.
Despite this, there has not yet been sufficient time to lapse given
the years that diabetic patients will be on their treatments, and
given the track record of other classes of drugs when they get into
clinical practice, careful monitoring of safety and durability is
essential.
There are early signs of additional cardiovascular benefit in this
class—a major boon for an oral anti-diabetes therapy—the use of
these agents as add-on therapy to metformin and the results of large
cardiovascular outcomes trials, DECLARE TIMI-58 for dapagli-
flozon [213] and EMPA-REG for empagliflozin [214], are eagerly
awaited.
An increasingly important challenge for pharmaceutical com-
panies is to demonstrate that their new drug is cost-effective,
especially where taxpayers money is involved. Although not men-
tioned for the previous drug classes, these agents were launched at a
time where costs of medicines have never been higher on the
agenda of the funders and are worth commenting on. Despite a
very optimistic cost–benefit analysis performed by the manufac-
turer, it does seem that these agents do have some modest cost
benefits when modeled using more stringent methods, particularly
when compared to sulfonylureas, pioglitazone, and DPP-IV inhi-
bitors [215]. In subjects with uncontrolled diabetes, cost savings
may be indirect. A retrospective study in such patients from India

showed that in “real-world clinical practice,” there was a compen-
satory reduction in the amount of insulin required by these
patients [216].
10 Where Are We Now?
In this chapter, I have outlined the eight classes of anti-diabetic

drugs in the chronological order of their appearance as medicines
since the 1950s from a drug discovery perspective; so quite a
narrow perspective. What is evident is that these agents are the
result of research and development activities performed in pharma-
ceutical companies using knowledge originating from academic
research. In future, this paradigm will have to remain the norm.
However, pipeline attrition in this disease area is high; numerous
publications have appeared on new drug targets optimistically pre-
dicting a novel approach only to ultimately end in failure, with
stearoyl-CoA desaturase-1 (SCD-1) [217], 11β-hydroxysteroid
dehydrogenase-1 (11β-HSD-1) [218], and diacylglycerol acyl
transferase-1 (DGAT-1) [219] being relatively recent examples.
Nevertheless, potential future targets for type 2 diabetes are
reviewed frequently [220–223] showing that science can deliver
new ideas. In particular, a class of GPCRs, the fatty acid receptors,
seem to be very interesting in this therapeutic area [221].
However, for there to be a real future for anti-diabetes drug
development, there needs to be a vibrant pharmaceutical industry
with the ability to invest in, or facilitate, the whole journey. Today
the costs involved in bringing a drug to market are enormous and
appear to have risen hugely over the past decade or so. A report
published in 2011 suggests that the cost is between 0.3 and 0.9 bil-
lion dollars per drug [224]. However, this analysis does not seem to
account for the global costs involved which includes the potentially
debilitating costs of failure which have caused serious productivity
issues in the industry [225]. An article in Forbes [226] used a rather
simplistic but useful calculation where total research and develop-
ment expenditure over a 10-year period was divided by the number
of product launches over the same period. This revealed a truly
shocking reality—drug discovery in the pharmaceutical industry, as
we know it today, is unsustainable. The numbers suggest that the
giants with greatest loss of productivity are spending in excess of
$10 billion per drug launch. The industry has been forced to
undergo change, often traumatic, as many ex-employees will testify,
and there is also the risk of retrenching effort away from the high-
risk, high-cost endeavors such as chronic diseases to areas where risk
can more comfortably be predicted and managed.
Apart from cost, the next challenge is return on investment:
what can new drugs be reasonably sold for? New modalities such as
20 John C. Clapham
antibody therapeutics seem to hold many possibilities not dreamed

of by the small molecule mindset that I was involved in when I
started my career in the 1970s, but they are very expensive which
could render them unaffordable no matter how good they are,
particularly in the developing economies where future need will
be greatest. Then there is the developing sophistication of users and
prescribers who will demand more than just HbA1c lowering;
weight loss and improved cardiovascular outcomes have been seen
with agents; though this was probably a result of serendipity, this is
much harder to design in from the beginning. Disease reversal may
also be a demand if β-cell biology delivers. Also, new drugs should
have to be very safe, probably the hardest thing of all, as nasty
surprises are often revealed post-launch when drugs are prescribed
to much larger numbers of patients, and the thiazolidinedione story
and the more recent potential pancreatitis issue with the incretin
agents and urinary tract infection risk with the SGLT2 inhibitors
are heightening awareness. Highly effective, safe, and affordable
drugs are still urgently needed and type 2 diabetes is, I believe, still
an unmet medical need despite six decades of drug discovery
research. The prospects are not impossible but the probability is
hard to predict.
References
1. Sugiyama Y (2005) Druggability: selecting 8. Reaven GM (2014) Pathophysiology of insu-
optimized drug candidates. Drug Discov lin resistance in human disease. Physiol Rev
Today 10:1577–1579 75:473–486
2. Allen MJ, Carey AH (2004) Target identifica- 9. Reaven GM (1988) Role of insulin resistance
tion and validation through genetics. Drug in human disease. Diabetes 37:1595–1607
Discov Today TARGETS 3:183–190 10. Ritzel RA, Butler AE, Rizza RA, Veldhuis JD,
3. Sams-Dodd F (2005) Target-based drug dis- Butler PC (2006) Relationship between b-cell
covery: is something wrong? Drug Discov mass and fasting blood glucose concentration
Today 10:139–147 in humans. Diabetes Care 29:717–718
4. Paul SM, Mytelka DS, Dunwiddie CT, Per- 11. The Diabetes Control and Complications
singer CC, Munos BH, Lindborg SR, Schact Trial Research Group (1993) The effect of
AL (2010) How to improve R&D productiv- intensive treatment of diabetes on the devel-
ity: the pharmaceutical industry’s grand chal- opment and progression of long-term compli-
lenge. Nat Rev Drug Discov 9:203–214 cations in insulin-dependent diabetes
5. Bouzakri K, Koistinen HA, Zierath JR (2005) mellitus. N Engl J Med 329:977–986
Molecular mechanisms of skeletal muscle 12. UK Prospective Diabetes Study (UKPDS)
insulin resistance in type 2 diabetes. Curr Dia- Group (1998) Effect of intensive blood-
betes Rev 1:167–174 glucose control with metformin on complica-
6. Gerich JE, Mitrakou A, Kelley D, tions in overweight patients with type 2 diabe-
Mandarino L, Nurjhan N, Reilly J, tes (UKPDS 34). Lancet 352:854–865
Jenssen T, Veneman T, Consoli A (1990) 13. UK Prospective Diabetes Study (UKPDS)
Contribution of impaired muscle glucose Group (1998) Intensive blood-glucose con-
clearance to reduced postabsorptive systemic trol with sulphonylureas or insulin compared
glucose clearance in NIDDM. Diabetes with conventional treatment and risk of com-
39:211–216 plications in patients with type 2 diabetes
7. Consoli A (1992) Role of liver in pathophysi- (UKPDS 33). Lancet 352:837–853
ology of NIDDM. Diabetes Care
15:430–441
14. American Diabetes Association (2011) Exec- 25. Gribble FM, Tucker SJ, Seino S, Ashcroft FM
utive summary: standards of medical care in (1998) Tissue specificity of sulfonylureas:
diabetes - 2011. Diabetes Care 34:S4–S10 studies on cloned cardiac and beta-cell K
15. Nathan DM, Buse JB, Davidson MB, (ATP) channels. Diabetes 47:1412–1418
Ferrannini E, Holman RR, Sherwin R, Zin- 26. Zünkler BJ, Lenzen S, M€anner K, Panten U,
man B (2009) Medical management of hyper- Trube G (2014) Concentration-dependent
glycemia in type 2 diabetes: a consensus effects of tolbutamide, meglitinide, glipizide,
algorithm for the initiation and adjustment glibenclamide and diazoxide on
of therapy: a consensus statement of the ATP-regulated K+ currents in pancreatic
American Diabetes Association and the b-cells. Naunyn Schmiedebergs Arch Pharma-
European Association for the Study of Diabe- col 337:225–230
tes. Diabetes Care 32:193–203 27. Sturgess NC, Kozlowski RZ, Carrington CA,
16. Nathan DM, Buse JB, Davidson MB, Heine Hales CN, Ashford ML (1988) Effects of sul-
RJ, Holman RR, Sherwin R, Zinman B phonylureas and diazoxide on insulin secre-
(2006) Management of hyperglycemia in tion and nucleotide-sensitive channels in an
type 2 diabetes: a consensus algorithm for insulin-secreting cell line. Br J Pharmacol
the initiation and adjustment of therapy: a 95:83–94
consensus statement from the American Dia- 28. Splitter S, Brown FR, Friskey RW, Grindell L,
betes Association and the European Associa- Kinsell LW (1956) Treatment of diabetic
tion for the study of diabetes. Diabetes Care patients-observations on the use of carbuta-
29:1963–1972 mide and tolbutamide. Calif Med
17. The ADVANCE Collaborative Group (2008) 85:285–288
Intensive blood glucose control and vascular 29. Graal MB, Wolffenbuttel BHR (1999) The
outcomes in patients with type 2 diabetes. N use of sulphonylureas in the elderly. Drugs
Engl J Med 358:2560–2572 Aging 15:471–481
18. The Action to Control Cardiovascular Risk in 30. Malabu UH, Vangaveti VN, Kennedy LR
Diabetes Study Group (2008) Effects of (2014) Disease burden evaluation of fall-
intensive glucose lowering in type 2 diabetes. related events in the elderly due to hypoglyce-
N Engl J Med 358:2545–2559 mia and other diabetic complications: a clini-
19. Herman WH, Dirani RG, O’Neill MC, cal review. Clin Epidemiol 6:287–294
Kravitz B, Heise A, Bakst A, Freed.M.I. 31. Bailey CJ, Day C (2004) Metformin: its
(2005) Reduction in use of healthcare services botanical background. Pract Diabetes Int
with combination sulfonylurea and rosiglita- 21:115–117
zone: findings from the Rosiglitazone Early vs 32. Bailey CJ (1992) Biguanides and NIDDM.
SULfonylurea Titration (RESULT) study. Diabetes Care 15:755–772
Am J Manag Care 11:273–278
33. Tucker GT, Casey C, Phillips PJ, Connor H,
20. Proks P, Reimann F, Green N, Gribble F, Ward JD, Woods HF (1981) Metformin
Ashcroft F (2002) Sulfonylurea stimulation kinetics in healthy subjects and in patients
of insulin secretion. Diabetes 51:S368–S376 with diabetes mellitus. Br J Clin Pharmacol
21. Nichols CG (2006) KATP channels as molec- 12:235–246
ular sensors of cellular metabolism. Nature 34. Zhou M, Xia L, Wang J (2007) Metformin
440:470–476 transport by a newly cloned proton-
22. Ashcroft FM (2007) ATP-sensitive K channels stimulated organic cation transporter (Plasma
and disease: from molecule to malady. Am J Membrane Monoamine Transporter)
Physiol Endocrinol Metab 293:E880–E889 expressed in human intestine. Drug Metab
23. Inagaki N, Gonoi T, Clement JP, Wang CZ, Dispos 35:1956–1962
Aguilar-Bryan L, Bryan J, Seino S (1996) A 35. Gong L, Goswami S, Giacomini KM, Altman
family of sulfonylurea receptors determines RB, Klein TE (2012) Metformin pathways:
the pharmacological properties of pharmacokinetics and pharmacodynamics.
ATP-sensitive K+ channels. Neuron Pharmacogenet Genomics 22:820–827
16:1011–1017 36. Graham GG, Punt J, Arora M, Day RO, Doo-
24. Aittoniemi J, Fotinou C, Craig TJ, de Wet H, gue MP, Duong J, Furlong TJ, Greenfield JR,
Proks P, Ashcroft FM (2009) SUR1: a unique Greenup LC, Kirkpatrick CM, Ray JE,
ATP-binding cassette protein that functions Timmins P, Williams KM (2011) Clinical
as an ion channel regulator. Philos Trans R pharmacokinetics of metformin. Clin Phar-
Soc B 364:257–267 macokinet 50:81–98
22 John C. Clapham
37. Strack T (2008) Metformin: a review. Drugs intake in obese subjects. Eur J Clin Invest
Today 44:303–314 28:441–446
38. Campbell RK, White JR, Saulie BA (1996) 52. Bruijstens LA, van Luin M, Buscher-
Metformin: a new oral biguanide. Clin Ther Jungerhans PM, Bosch FH (2008) Reality of
18:360–371 severe metformin-induced lactic acidosis in
39. DeFronzo RA, Goodman AM (1995) Efficacy the absence of chronic renal impairment.
of metformin in patients with non-insulin- Neth J Med 66:185–190
dependent diabetes mellitus. N Engl J Med 53. Carter D, Howlett HCS, Wiernsperger NF,
333:541–549 Bailey CJ (2003) Differential effects of met-
40. Bailey CJ, Turner RC (1996) Metformin. N formin on bile salt absorption from the jeju-
Engl J Med 334:574–579 num and ileum. Diabetes Obes Metab
41. Stumvoll M, Nurjhan N, Perriello G, 5:120–125
Dailey G, Gerich JE (1995) Metabolic effects 54. Davidson J, Howlett H (2004) New
of metformin in non-insulin-dependent dia- prolonged-release metformin improves gas-
betes mellitus. N Engl J Med 333:550–554 trointestinal tolerability. Br J Diabetes Vasc
42. Hundal RS, Krssak M, Dufour S, Laurent D, Dis 4:273–277
Lebon V, Chandramouli V, Inzucchi SE, 55. Mallick S (2004) Metformin induced acute
Schumann WC, Petersen KF, Landau BR, pancreatitis precipitated by renal failure. Post-
Shulman GI (2000) Mechanism by which grad Med J 80:239–240
metformin reduces glucose production in 56. Alsubaie S, Almalki MH (2013) Metformin
type 2 diabetes. Diabetes 49:2063–2069 induced acute pancreatitis. Derm Endocrinol
43. Klip A, Leiter LA (1990) Cellular mechanism 5:317–318
of action of metformin. Diabetes Care 57. Oprea TI, Bauman JE, Bologa CG,
13:696–704 Buranda T, Chigaev A, Edwards BS, Jarvik
44. Fryer LGD, Parbu-Patel A, Carling D (2002) JW, Gresham HD, Haynes MK, Hjelle B,
The anti-diabetic drugs rosiglitazone and Hromas R, Hudson L, Mackenzie DA, Muller
metformin stimulate AMP-activated protein CY, Reed JC, Simons PC, Smagley Y,
kinase through distinct signaling pathways. J Strouse J, Surviladze Z, Thompson T,
Biol Chem 277:25226–25232 Ursu O, Waller A, Wandinger-Ness A, Winter
45. Hardie DG (2008) AMPK: a key regulator of SS, Wu Y, Young SM, Larson RS, Willman C,
energy balance in the single cell and the whole Sklar LA (2011) Drug repurposing from an
organism. Int J Obes Relat Metab Disord 32: academic perspective. Drug Discov Today
S7–S12 Ther Strat 8:61
46. Ruderman N, Prentki M (2004) AMP-kinase 58. Kasznicki J, Sliwinska A, Drzewoski J (2014)
and malonyl-CoA: targets for therapy of the Metformin in cancer prevention and therapy.
metabolic syndrome. Nat Rev Drug Discov Ann Transl Med 2:57
3:340–351 59. Diamanti-Kandarakis E, Christakou CD,
47. Owen MR, Doran E, Halestrap AP (2000) Kandaraki E, Economou FN (2010) Metfor-
Evidence that metformin exerts its anti- min: an old medication of new fashion: evol-
diabetic effects through inhibition of complex ving new molecular mechanisms and clinical
1 of the mitochondrial respiratory chain. Bio- implications in polycystic ovary syndrome.
chem J 348:607–614 Eur J Endocrinol 162:193–212
48. Sakamoto K, Göransson O, Hardie DG, Alessi 60. Scarpello JHB, Howlett HCS (2008) Metfor-
DR (2004) Activity of LKB1 and AMPK- min therapy and clinical uses. Diabetes Vasc
related kinases in skeletal muscle: effects of Dis Res 5:157–167
contraction, phenformin, and AICAR. Am J 61. Hirst JA, Farmer AJ, Dyar A, Lung TWC,
Physiol Endocrinol Metab 287:E310–E317 Stevens RJ (2013) Estimating the effect of
49. Golay A (2007) Metformin and body weight. sulfonylurea on HbA1c in diabetes: a system-
Int J Obes Relat Metab Disord 32:61–72 atic review and meta-analysis. Diabetologia
56:973–984
50. Clapham JC, Arch JR (2008) Influencing
energy expenditure and substrate utilisation. 62. Rydberg T, Jönsson A, Karlsson MO, Melan-
In: Pharmacotherapy of obesity, Milestones in der A (1997) Concentration-effect relations
drug therapy. Birkh€auser, Basel, pp 101–115 of glibenclamide and its active metabolites in
man: modelling of Pharmacokinetics and
51. Paolisso G, Amato L, Eccellente R, Pharmacodynamics. Br J Clin Pharmacol
Gambardella A, Tagliamonte MR, 43:373–381
Varricchio G, Carella C, Giugliano D, D’ono-
frio F (1998) Effect of metformin on food
63. Kola I, Landis J (2004) Can the pharmaceuti- and therapeutic possibilities. Prog Drug Res
cal industry reduce attrition rates? Nat Rev 51:33–94
Drug Discov 3:711–716 75. Henry RR (1997) Thiazolidinediones. Endo-
64. Rosengren A, Jing X, Eliasson L, Renstrøm E crinol Metab Clin 26:553–573
(2008) Why treatment fails in type 2 diabetes. 76. Gale EAM (2006) Troglitazone: the lesson
PLoS Med 5:e215 that nobody learned? Diabetologia 49:1–6
65. Kahn SE, Haffner SM, Heise MA, Herman 77. Barnett AH (2009) Redefining the role of
WH, Holman RR, Jones NP, Kravitz BG, thiazolidinediones in the management of
Lachin JM, O’Neill MC, Zinman B, Viberti type 2 diabetes. Vasc Health Risk Manag
G (2006) Glycemic durability of rosiglita- 5:141–151
zone, metformin, or glyburide monotherapy. 78. Phillips LS, Grunberger G, Miller E,
N Engl J Med 355:2427–2443 Patwardhan R, Rappaport EB, Salzman A
66. Karam JH, Sanz N, Salamon E, Nolte MS (2001) Once- and twice-daily dosing with
(1986) Selective unresponsiveness of pancre- rosiglitazone improves glycemic control in
atic b-cells to acute sulfonylurea stimulation patients with type 2 diabetes. Diabetes Care
during sulfonylurea therapy in NIDDM. Dia- 24:308–315
betes 35:1314–1320 79. Fonseca V, Rosenstock J, Patwardhan R, Salz-
67. Pantalone KM, Kattan MW, Yu C, Wells BJ, man A (2000) Effect of metformin and rosi-
Arrigain S, Jain A, Atreja A, Zimmerman RS glitazone combination therapy in patients
(2012) Increase in overall mortality risk in with type 2 diabetes mellitus: a randomized
patients with type 2 diabetes receiving glipi- controlled trial. JAMA 283:1695–1702
zide, glyburide or glimepiride monotherapy 80. Poitout V (2004) b-cell lipotoxicity: burning
versus metformin: a retrospective analysis. fat into heat? Endocrinology 145:3563–3565
Diabetes Obes Metab 14:803–809
81. Nolan CJ, Prentki M (2008) The islet b-cell:
68. Rustenbeck I, Wienbergen A, Bleck C, Jörns fuel responsive and vulnerable. Trends Endo-
A (2004) Desensitization of insulin secretion crinol Metab 19:285–291
by depolarizing insulin secretagogues. Diabe-
tes 53:S140–S150 82. Buckingham RE, Al-Barazanji KA, Toseland
CDN, Slaughter M, Connor SC, West A,
69. Sadikot SM, Mogensen CE (2008) Risk of Bond B, Turner NC, Clapham JC (1998)
coronary artery disease associated with initial Peroxisome proliferator-activated receptor-g
sulphonylurea treatment of patients with type agonist, rosiglitazone, protects against
2 diabetes: a matched case-control study. Dia- nephropathy and pancreatic islet abnormal-
betes Res Clin Pract 82:391–395 ities in Zucker fatty rats. Diabetes
70. Lefer DJ, Nichols CG, Coetzee WA (2009) 47:1326–1334
Sulfonylurea receptor 1 subunits of 83. Nissen SE, Wolski K (2007) Effect of rosigli-
ATP-sensitive potassium channels and myo- tazone on the risk of myocardial infarction
cardial Ischemia/reperfusion Injury. Trends and death from cardiovascular causes. N
Cardiovasc Med 19:61–67 Engl J Med 356:2457–2471
71. Bell DSH (2006) Do sulfonylurea drugs 84. Singh S, Loke YK, Furberg CD (2007) Thia-
increase the risk of cardiac events? Can Med zolidinediones and heart failure: a teleo-
Assoc J 174:185–186 analysis. Diabetes Care 30:2148–2153
72. Grygiel-Górniak B (2014) Peroxisome 85. Bloomgarden ZT (2007) The avandia debate.
proliferator-activated receptors and their Diabetes Care 30:2401–2408
ligands: nutritional and clinical implications -
a review. Nutr J 13:17 86. Misbin RI (2007) Lessons from the avandia
controversy: a new paradigm for the develop-
73. Glosli H, Gudbrandsen OA, Mullen AJ, ment of drugs to treat type 2 diabetes. Diabe-
Halvorsen B, Røst TH, Wergedahl H, tes Care 30:3141–3144
Prydz H, Aukrust P, Berge RK (2005)
Down-regulated expression of PPARa target 87. Bhatt DL, Chew DP, Grines C, Mukherjee D,
genes, reduced fatty acid oxidation and Leesar M, Gilchrist IC, Corbelli JC, Blanken-
altered fatty acid composition in the liver of ship JC, Eres A, Steinhubl S, Tan WA, Resar
mice transgenic for hTNFa. Biochim Biophys JR, Al Mahameed A, Abdel-Latif A, Tang
Acta 1734:235–246 HW, Brennan D, McErlean E, Hazen SL,
Topol EJ (2007) Peroxisome proliferator-
74. Turner NC, Clapham JC (1998) Insulin resis- activated receptor g agonists for the preven-
tance, impaired glucose tolerance and non- tion of adverse events following percutaneous
insulin-dependent diabetes, pathologic coronary revascularizationG-results of the
mechanisms and treatment: current status PPAR study. Am Heart J 154:137–143
24 John C. Clapham
88. McAfee AT, Koro C, Landon J, Ziyadeh N, 101. Hanefeld M, Fischer S, Schulze J,
Walker AM (2007) Coronary heart disease Spengler M, Wargenau M, Schollberg K,
outcomes in patients receiving antidiabetic Fücker K (1991) Therapeutic potentials of
agents. Pharmacoepidem Drug Safe acarbose as first-line drug in NIDDM insuffi-
16:711–725 ciently treated with diet alone. Diabetes Care
89. Home PD, Pocock SJ, Beck-Nielsen H, 14:732–737
Gomis R, Hanefeld M, Dargie H, 102. Salman S, Salman F, Satman I, Yilmaz Y,
Komajda M, Gubb J, Biswas N, Jones NP Özer E, Sengül A, Özer H, Demirel HO,
(2005) Rosiglitazone evaluated for cardiac Karsidag K, Dinççag N, Yilmaz MT (2001)
outcomes and regulation of glycaemia in dia- Comparison of acarbose and gliclazide as first-
betes (RECORD): study design and protocol. line agents in patients with type 2 diabetes.
Diabetologia 48:1726–1735 Curr Med Res Opin 16:296–306
90. Hillaire-Buys D, Faillie JL, Montastruc JL 103. Wang G, Liu J, Yang N, Gao X, Fan H, Xu Y,
(2011) Pioglitazone and bladder cancer. Lan- Yang W (2014) MARCH2:comparative
cet 378:1543–1544 assessment of therapeutic effects of acarbose
91. Grey A (2009) Thiazolidinedione-induced and metformin in newly diagnosed type 2 dia-
skeletal fragility - mechanisms and implica- betes patients. PLoS One 9:e105698
tions. Diabetes Obes Metab 11:275–284 104. Dabhi AS, Bhatt NR, Shah MJ (2013) Vogli-
92. Oshitari T, Asaumi N, Watanabe M, bose: an alpha glucosidase inhibitor. J Clin
Kumagai K, Mitamura Y (2008) Severe mac- Diagn Res 7:3023–3027
ular edema induced by pioglitazone in a 105. Standl E, Theodorakis MJ, Erbach M,
patient with diabetic retinopathy: a case Schnell O, Tuomilehto J (2014) On the
study. Vasc Health Risk Manag 4:1137–1140 potential of acarbose to reduce cardiovascular
93. Liazos E, Broadbent DM, Beare N, Kumar N disease. Cardiovasc Diabetol 13:81
(2008) Spontaneous resolution of diabetic 106. Ceriello A, Taboga C, Giacomello R, Stel L,
macular oedema after discontinuation of thia- Motz E, Pirisi M (1996) Post-meal coagula-
zolidenediones. Diabet Med 25:860–862 tion activation in diabetes mellitus: the effect
94. Cordain L, Eaton SB, Sebastian A, Mann N, of acarbose. Diabetologia 39:469–473
Lindeberg S, Watkins BA, O’Keefe JH, 107. Kawano H, Motoyama T, Hirashima O,
Brand-Miller J (2005) Origins and evolution Hirai N, Miyao Y, Sakamoto T, Kugiyama K,
of the Western diet: health implications for Ogawa H, Yasue H (1999) Hyperglycemia
the 21st century. Am J Clin Nutr 81:341–354 rapidly suppresses flow-mediated endothe-
95. Prentice AM, Jebb SA (1995) Obesity in Brit- lium-dependent vasodilation of brachial
ain: gluttony or sloth? Br Med J 311:437–439 artery. J Am Coll Cardiol 34:146–154
96. Röder PV, Geillinger KE, Zietek TS, 108. Heitzer T, Schlinzig T, Krohn K, Meinertz T,
Thorens B, Koepsell H, Hannelore D (2014) Münzel T (2001) Endothelial dysfunction,
The role of SGLT1 and GLUT2 in intestinal oxidative stress, and risk of cardiovascular
glucose transport and sensing. PLoS One 9: events in patients with coronary artery dis-
e89977 ease. Circulation 104:2673–2678
97. Caspary WF (1992) Physiology and patho- 109. Caumo A, Luzi L (2004) First-phase insulin
physiology of intestinal absorption. Am J secretion: does it exist in real life? Considera-
Clin Nutr 55:299S–308S tions on shape and function. Am J Physiol
98. Chiba S (1997) Molecular mechanism in Endocrinol Metab 287:E371–E385
alpha-glucosidase and glucoamylase. Biosci 110. Lewis GF, Zinman B, Groenewoud Y,
Biotechnol Biochem 61:1233–1239 Vranic M, Giacca A (1996) Hepatic glucose
99. Göke B, Fuder H, Wieckhorst G, Theiss U, production is regulated both by direct hepatic
Stridde E, Littke T, Kleist P, Arnold R, Lücker and extrahepatic effects of insulin in Humans.
PW (1995) Voglibose (AO-128) is an efficient Diabetes 45:454–462
a-glucosidase inhibitor and mobilizes the 111. Calles-Escandon J, Robbins DC (1987) Loss
endogenous GLP-1 reserve. Digestion of early phase of insulin release in humans
56:493–501 impairs glucose tolerance and blunts thermic
100. Standl E, Schernthaner G, Rybka J, effect of glucose. Diabetes 36:1167–1172
Hanefeld M, Raptis SA, Naditch L (2001) 112. Del Prato S, Miccoli R, Penno G (2005) The
Improved glycaemic control with miglitol in importance of effective early phase insulin
inadequately-controlled type 2 diabetics. Dia- secretion. Br J Diabetes Vasc Dis 5:198–202
betes Res Clin Pract 51:205–213 113. Gerich JE (2002) Is reduced first-phase insu-
lin release the earliest detectable abnormality
in individuals destined to develop type 2 dia- type 2 diabetes: a patient-centered approach.

betes? Diabetes 51:S117–S121 Diabetes Care 35:1364
114. Bruce DG, Chisholm DJ, Storlien LH, Krae- 124. Ekström N, Svensson AM, Miftaraj M,
gen EW (1988) Physiological importance of Andersson SK, Cederholm J, Zethelius B,
deficiency in early prandial insulin secretion in Eliasson B, Gudbjörnsdottir S (2015) Dura-
non-insulin-dependent diabetes. Diabetes bility of oral hypoglycemic agents in drug
37:736–744 naı̈ve patients with type 2 diabetes: report
115. Luzio SD, Owens DR, Vora J, Dolben J, from the Swedish National Diabetes Register
Smith H (1991) Intravenous insulin simulates (NDR). BMJ Open Diabetes Res Care 3:
early insulin peak and reduces post-prandial e000059
hyperglycaemia/hyperinsulinaemia in type 125. Moses R, Slobodniuk R, Boyages S,
2 (non-insulin-dependent) diabetes mellitus. Colagiuri S, Kidson W, Carter J, Donnelly T,
Diabetes Res 16:63–67 Moffitt P, Hopkins H (1999) Effect of repa-
116. Owens DR, Luzio SD, Ismail I, Bayer T glinide addition to metformin monotherapy
(2000) Increased prandial insulin secretion on glycemic control in patients with type
after administration of a single preprandial 2 diabetes. Diabetes Care 22:119–124
oral dose of repaglinide in patients with type 126. Schramm TK, Gislason GH, Vaag A, Rasmus-
2 diabetes. Diabetes Care 23:518–523 sen JN, Folke F, Hansen ML, Fosbøl EL,
117. Gromada J, Dissing S, Kofod H, Frøkjær-Jen- Køber L, Norgaard ML, Madsen M, Hansen
sen J (1995) Effects of the hypoglycaemic PR, Torp-Pedersen C (2011) Mortality and
drugs repaglinide and glibenclamide on cardiovascular risk associated with different
ATP-sensitive potassium-channels and cyto- insulin secretagogues compared with metfor-
solic calcium levels in TC3 cells and rat pan- min in type 2 diabetes, with or without a
creatic beta cells. Diabetologia previous myocardial infarction: a nationwide
38:1025–1032 study. Eur Heart J 32:1900–1908
118. Hu S, Wang S, Fanelli B, Bell PA, Dunning 127. Creutzfeldt W (2005) The [pre-] history of
BE, Geisse S, Schmitz R, Boettcher BR the incretin concept. Regul Pept 128:87–91
(2000) Pancreatic b-cell K ATP channel activ- 128. Vilsbøll T, Krarup T, Madsbad S, Holst JJ
ity and membrane-binding studies with nate- (2003) Both GLP-1 and GIP are insulinotro-
glinide: a comparison with sulfonylureas and pic at basal and postprandial glucose levels
repaglinide. J Pharmacol Exp Ther and contribute nearly equally to the incretin
293:444–452 effect of a meal in healthy subjects. Regul Pept
119. Moses RG, Gomis R, Frandsen KB, Schlien- 114:115–121
ger JL, Dedov I (2001) Flexible meal-related 129. Eissele R, Göke R, Willemer S, Harthus HP,
dosing with repaglinide facilitates glycemic Vermeer H, Arnold R, Göke B (1992)
control in therapy-naive Type 2 diabetes. Dia- Glucagon-like peptide-1 cells in the gastroin-
betes Care 24:11–15 testinal tract and pancreas of rat, pig and man.
120. Rosenstock J, Hassman DR, Madder RD, Eur J Clin Invest 22:283–291
Brazinsky SA, Farrell J, Khutoryansky N, 130. Gorboulev V, Schürmann A, Vallon V,
Hale PM (2004) Repaglinide versus nategli- Kipp H, Jaschke A, Klessen D, Friedrich A,
nide monotherapy: a randomized, multicen- Scherneck S, Rieg T, Cunard R, Veyhl-
ter study. Diabetes Care 27:1265–1270 Wichmann M, Srinivasan A, Balen D,
121. Hansen AM, Christensen IT, Hansen JB, Carr Breljak D, Rexhepaj R, Parker HE, Gribble
RD, Ashcroft FM, Wahl P (2002) Differential FM, Reimann F, Lang F, Wiese S, Sabolic I,
interactions of nateglinide and repaglinide on Sendtner M, Koepsell H (2012) Na+-d-glu-
the human b-cell sulphonylurea receptor cose cotransporter SGLT1 is pivotal for intes-
1. Diabetes 51:2789–2795 tinal glucose absorption and glucose-
122. Madsbad S, Kilhovd B, Lager I, Mustajoki P, dependent incretin secretion. Diabetes
Dejgaard A, for the Scandinavian Repaglinide 61:187–196
Group (2001) Comparison between repagli- 131. Thorens B (1992) Expression cloning of the
nide and glipizide in type 2 diabetes mellitus: pancreatic beta cell receptor for the gluco-
a 1-year multicentre study. Diabet Med incretin hormone glucagon-like peptide
18:395–401 1. Proc Natl Acad Sci U S A 89:8641–8645
123. Inzucchi SE, Bergenstal RM, Buse JB, 132. Dillon JS, Tanizawa Y, Wheeler MB, Leng
Diamant M, Ferrannini E, Nauck M, Peters XH, Ligon BB, Rabin DU, Yoo-Warren H,
AL, Tsapas A, Wender R, Matthews DR Permutt MA, Boyd AE (1993) Cloning and
(2012) Management of hyperglycemia in functional expression of the human glucagon-
26 John C. Clapham
like peptide-1 (GLP-1) receptor. Endocrinol- 144. Schlögl H, Kabisch S, Horstmann A,

ogy 133:1907–1910 Lohmann G, Müller K, Lepsien J, Busse-
133. Gremlich S, Porret A, Hani EH, Cherif D, Voigt F, Kratzsch J, Pleger B, Villringer A,
Vionnet N, Froguel P, Thorens B (1995) Stumvoll M (2013) Exenatide-induced
Cloning, functional expression, and chromo- reduction in energy intake is associated with
somal localization of the human pancreatic increase in hypothalamic connectivity. Diabe-
islet glucose-dependent insulinotropic poly- tes Care 36:1933–1940
peptide receptor. Diabetes 44:1202–1208 145. Edwards CM, Stanley SA, Davis R, Brynes
134. Seino Y, Fukushima M, Yabe D (2010) GIP AE, Frost GS, Seal LJ, Ghatei MA, Bloom
and GLP-1, the two incretin hormones: simi- SR (2001) Exendin-4 reduces fasting and
larities and differences. J Diabetes Invest postprandial glucose and decreases energy
1:8–23 intake in healthy volunteers. Am J Physiol
135. Salon JA, Lodowski DT, Palczewski K (2011) Endocrinol Metab 281:E155–E161
The significance of G protein-coupled recep- 146. Ando T, Haraguchi A, Matsunaga T,
tor crystallography for drug discovery. Phar- Natsuda S, Yamasaki H, Usa T, Kawakami A
macol Rev 63:901–937 (2014) Liraglutide as a potentially useful
136. Wang MW, Liu Q, Zhou C (2010) agent for regulating appetite in diabetic
Non-peptidic glucose-like peptide-1 receptor patients with hypothalamic hyperphagia and
agonists: aftermath of a serendipitous discov- obesity. Intern Med 53:1791–1795
ery. Acta Pharmacol Sin 31:1026–1030 147. Zoicas F, Droste M, Mayr B, Buchfelder M,
137. Su H, He M, Li H, Liu Q, Wang J, Wang Y, Schöfl C (2013) GLP-1 analogues as a new
Gao W, Zhou L, Liao J, Young AA, Wang treatment option for hypothalamic obesity in
MW (2008) Boc5, a non-peptidic glucagon- adults: report of nine cases. Eur J Endocrinol
like peptide-1 receptor agonist, invokes sus- 168:699–706
tained glycemic control and weight loss in 148. Matveyenko AV, Butler PC (2008) Relation-
diabetic mice. PLoS One 3:e2892 ship between b-cell mass and diabetes onset.
138. Chen D, Liao J, Li N, Zhou C, Liu Q, Diabetes Obes Metab 10:23–31
Wang G, Zhang R, Zhang S, Lin L, Chen K, 149. Demeterco C, Hao E, Lee SH, Itkin-Ansari P,
Nan F, Young AA, Wang MW (2007) A non- Levine F (2009) Adult human b-cell neogen-
peptidic agonist of glucagon-like peptide esis? Diabetes Obes Metab 11:46–53
1 receptors with efficacy in diabetic db/db 150. Bunck MC, Cornér A, Eliasson B, Heine RJ,
mice. Proc Natl Acad Sci U S A 104:943–948 Shaginian RM, Taskinen MR, Smith U,
139. Kolterman OG, Kim DD, Shen L, Ruggles Yki-J€arvinen H, Diamant M (2011) Effects
JA, Nielsen LL, Fineman MS, Baron AD of exenatide on measures of b-cell function
(2005) Pharmacokinetics, pharmacodynam- after 3 years in metformin-treated patients
ics, and safety of exenatide in patients with with type 2 diabetes. Diabetes Care
type 2 diabetes mellitus. Am J Health Syst 34:2041–2047
Pharm 62:173–181 151. Kjems LL, Holst JJ, Vølund A, Madsbad S
140. Agersø H, Jensen LB, Elbrønd B, Rolan P, (2003) The influence of GLP-1 on glucose-
Zdravkovic M (2002) The pharmacokinetics, stimulated insulin secretion: effects on β-cell
pharmacodynamics, safety and tolerability of sensitivity in type 2 and nondiabetic subjects.
NN2211, a new long-acting GLP-1 deriva- Diabetes 52:380–386
tive, in healthy men. Diabetologia 152. Mondragon A, Davidsson D, Kyriakoudi S,
45:195–202 Bertling A, Gomes-Faria R, Cohen P,
141. Cai Y, Wei L, Ma L, Huang X, Tao A, Liu Z, Rothery S, Chabosseau P, Rutter GA, da
Yuan W (2013) Long-acting preparations of Silva Xavier G (2014) Divergent effects of
exenatide. Drug Des Devel Ther 7:963–970 liraglutide, exendin-4, and sitagliptin on
142. Hansen KB, Vilsbøll T, Knop FK (2010) beta-cell mass and indicators of pancreatitis
Incretin mimetics: a novel therapeutic option in a mouse model of hyperglycaemia. PLoS
for patients with type 2 diabetes - a review. One 9:e104873
Diabetes Metab Syndr Obes 17:155–163 153. Rutti S, Sauter NS, Bouzakri K, Prazak R,
143. Pinkney J, Fox T, Ranganath L (2010) Select- Halban PA, Donath MY (2012) In vitro pro-
ing GLP-1 agonists in the management of liferation of adult human beta-cells. PLoS
type 2 diabetes: differential pharmacology One 7:e35801
and therapeutic benefits of liraglutide and 154. Toso C, McCall M, Emamaullee J, Merani S,
exenatide. Ther Clin Risk Manag 6:401–411 Davis J, Edgar R, Pawlick R, Kin T, Knudsen
LB, Shapiro AMJ (2010) Liraglutide, a long-
acting human glucagon-like peptide 1 ana- Ippolito DE, Kim D, Lyons KA, Ok HO,
logue, improves human islet survival in cul- Patel RA, Petrov AN, Pryor KA, Qian X,
ture. Transpl Int 23:259–265 Reigle L, Woods A, Wu JK, Zaller D,
155. Bregenholt S, Møldrup A, Blume N, Karlsen Zhang X, Zhu L, Weber AE, Thornberry
AE, Friedrichsen BN, Tornhave D, Knudsen NA (2005) Dipeptidyl peptidase IV inhibi-
LB, Petersen JS (2005) The long-acting glu- tion for the treatment of type 2 diabetes:
cagon-like peptide-1 analogue, liraglutide, potential importance of selectivity over dipep-
inhibits b-cell apoptosis in vitro. Biochem tidyl peptidases 8 and 9. Diabetes
Biophys Res Commun 330:577–584 54:2988–2994
156. Butler AE, Janson J, Bonner-Weir S, Ritzel R, 166. Lambeir A-M, Durinx C, Scharpé S, De Mee-
Rizza RA, Butler PC (2003) b-Cell deficit and ster I (2013) Dipeptidyl-peptidase IV from
Increased β-cell apoptosis in humans with bench to bedside: an update on structural
type 2 diabetes. Diabetes 52:102–110 properties, functions, and clinical aspects of
157. Bosco D, Armanet M, Morel P, Niclauss N, the enzyme DPPIV. Crit Rev Clin Lab Sci
Sgroi A, Muller YD, Giovannoni L, 40:209–294
Parnaud G, Berney T (2010) Unique arrange- 167. Durinx C, Lambeir AM, Bosmans E, Fal-
ment of α- and β-cells in human Islets of magne JB, Berghmans R, Haemers A,
Langerhans. Diabetes 59:1202–1210 Scharpé S, De Meester I (2000) Molecular
158. Cabrera O, Berman DM, Kenyon NS, characterization of dipeptidyl peptidase activ-
Ricordi C, Berggren PO, Caicedo A (2006) ity in serum. Eur J Biochem 267:5608–5613
The unique cytoarchitecture of human pan- 168. Kos K, Baker AR, Jernas M, Harte AL, Clap-
creatic islets has implications for islet cell func- ham JC, O’Hare JP, Carlsson L, Kumar S,
tion. Proc Natl Acad Sci U S A McTernan PG (2009) DPP-IV inhibition
103:2334–2339 enhances the antilipolytic action of NPY in
159. Steiner DJ, Kim A, Miller K, Hara M (2010) human adipose tissue. Diabetes Obes Metab
Pancreatic islet plasticity: interspecies compar- 11:285–292
ison of islet architecture and composition. 169. Engel M, Hoffmann T, Wagner L,
Islets 2:135–145 Wermann M, Heiser U, Kiefersauer R,
160. Butler PC, Elashoff M, Elashoff R, Gale EAM Huber R, Bode W, Demuth HU, Brandstetter
(2013) A critical analysis of the clinical use of H (2003) The crystal structure of dipeptidyl
incretin-based therapies: are the GLP-1 thera- peptidase IV (CD26) reveals its functional
pies safe? Diabetes Care 36:2118–2125 regulation and enzymatic mechanism. Proc
Natl Acad Sci U S A 100:5063–5068
161. Perfetti R, Zhou J, Doyle ME, Egan JM
(2000) Glucagon-like peptide-1 induces cell 170. Aertgeerts K, Ye S, Tennant MG, Kraus ML,
proliferation and pancreatic-duodenum Rogers J, Sang BC, Skene RJ, Webb DR,
homeobox-1 expression and increases endo- Prasad GS (2004) Crystal structure of
crine cell mass in the pancreas of old, glucose human dipeptidyl peptidase IV in complex
intolerant rats. Endocrinology with a decapeptide reveals details on substrate
141:4600–4605 specificity and tetrahedral intermediate for-
mation. Protein Sci 13:412–421
162. Nauck MA, Friedrich N (2013) Do GLP-1-
based therapies increase cancer risk? Diabetes 171. Sebokova E, Christ AD, Boehringer M, Miz-
Care 36:S245–S252 rahi J (2015) Dipeptidyl peptidase IV inhibi-
tors: the next generation of new promising
163. Noel RA, Braun DK, Patterson RE, Bloomg- therapies for the management of type 2 diabe-
ren GL (2009) Increased risk of acute pancre- tes. Curr Top Med Chem 7:547–555
atitis and biliary disease observed in patients
with type 2 diabetes: a retrospective cohort 172. Shubrook J, Colucci R, Guo A, Schwartz F
study. Diabetes Care 32:834–838 (2011) Saxagliptin: a selective DPP-4 inhibi-
tor for the treatment of type 2 diabetes melli-
164. David F, Bernard AM, Pierres M, Marguet D tus. Clin Med Insights Endocrinol Diab
(1993) Identification of serine 624, aspartic 4:1–12
acid 702, and histidine 734 as the catalytic
triad residues of mouse dipeptidyl-peptidase 173. Toth PP (2015) Overview of saxagliptin effi-
IV (CD26). A member of a novel family of cacy and safety in patients with type 2 diabetes
nonclassical serine hydrolases. J Biol Chem and cardiovascular disease or risk factors for
268:17247–17252 cardiovascular disease. Vasc Health Risk
Manag 11:9–23
165. Lankas GR, Leiting B, Roy RS, Eiermann GJ,
Beconi MG, Biftu T, Chan CC, 174. Banerjee M, Younis N, Soran H (2009) Vil-
Edmondson S, Feeney WP, Huaibing H, dagliptin in clinical practice: a review of
28 John C. Clapham
literature. Expert Opin Pharmacother 186. Wright EM, Loo DDF, Hirayama BA (2011)
10:2745–2757 Biology of human sodium glucose transpor-
175. Dhillon S (2010) Sitagliptin: a review of its ters. Physiol Rev 91:733–794
use in the management of type 2 diabetes 187. Wright EM (2001) Renal Na+-glucose
mellitus. Drugs 70:489–512 cotransporters. Am J Physiol Renal Physiol
176. Ahrén B (2014) Insulin plus incretin: a 280:F10–F18
glucose-lowering strategy for type 188. Mackenzie B, Loo DDF, Panayotova-
2-diabetes. World J Diabetes 5:40–51 Heiermann M, Wright EM (1996) Biophysi-
177. Pratley RE, Schweizer A, Rosenstock J, Foley cal characteristics of the pig kidney Na+/Glu-
JE, Banerji MA, Pi-Sunyer FX, Mills D, Deja- cose cotransporter SGLT2 reveal a common
ger S (2008) Robust improvements in fasting mechanism for SGLT1 and SGLT2. J Biol
and prandial measures of b-cell function with Chem 271:32678–32683
vildagliptin in drug-naı̈ve patients: analysis of 189. Wells RG, Pajor AM, Kanai Y, Turk E, Wright
pooled vildagliptin monotherapy database. EM, Hediger MA (1992) Cloning of a human
Diabetes Obes Metab 10:931–938 kidney cDNA with similarity to the sodium-
178. Del Prato S, Barnett AH, Huisman H, glucose cotransporter. Am J Physiol Renal
Neubacher D, Woerle HJ, Dugi KA (2011) Physiol 263:F459–F465
Effect of linagliptin monotherapy on glycae- 190. Kanai Y, Lee WS, You G, Brown D, Hediger
mic control and markers of b-cell function in MA (1994) The human kidney low affinity Na
patients with inadequately controlled type +/glucose cotransporter SGLT2. Delineation
2 diabetes: a randomized controlled trial. Dia- of the major renal reabsorptive mechanism for
betes Obes Metab 13:258–267 D-glucose. J Clin Invest 93:397–404
179. Ahrén B, Pacini G, Tura A, Foley JE, Schwei- 191. DeFronzo RA, Hompesch M,
zer A (2007) Improved meal-related insulin Kasichayanula S, Liu X, Hong Y, Pfister M,
processing contributes to the enhancement of Morrow LA, Leslie BR, Boulton DW,
b-cell function by the DPP-4 inhibitor vlda- Ching A, LaCreta FP, Griffen SC (2013)
gliptin in patients with type 2 diabetes. Horm Characterization of renal glucose reabsorp-
Metab Res 39:826–829 tion in response to dapagliflozin in healthy
180. Yeom JA, Kim ES, Park HS, Ham DS, Sun K, subjects and subjects with type 2 diabetes.
Kim JW, Cho JH, Yoon KH (2011) Both Diabetes Care 36:3169–3176
sitagliptin analogue & pioglitazone preserve 192. Calado J, Soto K, Clemente C, Correia P,
the beta-cell proportion in the islets with dif- Rueff J (2004) Novel compound heterozy-
ferent mechanism in non-obese and obese gous mutations in SLC5A2 are responsible
diabetic mice. BMB Rep 44:713–718 for autosomal recessive renal glucosuria.
181. Girgis CM, Champion BL (2011) Hum Genet 114:314–316
Vildagliptin-induced acute pancreatitis. 193. van den Heuvel L, Assink K, Willemsen M,
Endocr Pract 17:e48–e50 Monnens L (2002) Autosomal recessive renal
182. Lee CF, Sun MS, Tai YK (2014) Saxagliptin- glucosuria attributable to a mutation in the
induced recurrent acute pancreatitis. Intern sodium glucose cotransporter (SGLT2).
Med 53:1351–1354 Hum Genet 111:544–547
183. Chang CH, Lin JW, Chen ST, Lai MS, Cuang 194. Gallo LA, Wright EM, Vallon V (2015) Prob-
LM, Chang YC (2016) Dipeptidyl peptidase- ing SGLT2 as a therapeutic target for diabe-
4 inhibitor use is not associated with acute tes: basic physiology and consequences.
pancreatitis in high-risk type 2 diabetic Diabetes Vasc Dis Res 12:78–89
patients: a nationwide cohort study. Medicine 195. Rossetti L, Smith D, Shulman GI,
95:e2603–e2609 Papachristou D, DeFronzo RA (1987) Cor-
184. Karagiannis T, Bekiari E, Boura P, Tsapas A rection of hyperglycemia with phlorizin nor-
(2015) Cardiovascular risk with DPP-4 inhi- malizes tissue sensitivity to insulin in diabetic
bitors: latest evidence and clinical implica- rats. J Clin Invest 79:1510–1515
tions. Ther Adv Drug Saf 7:36–38 196. Isaji M (2011) SGLT2 inhibitors: molecular
185. Brunton S (2014) GLP-1 receptor design and potential differences in effect. Kid-
agonists vs. DPP-4 inhibitors for type 2 diabe- ney Int 79:S14–S19
tes: is one approach more successful or prefer- 197. Nomura S, Sakamaki S, Hongu M,
able than the other? Int J Clin Pract Kawanishi E, Koga Y, Sakamoto T,
68:557–567 Yamamoto Y, Ueta K, Kimata H,
Nakayama K, Tsuda-Tsukimoto M (2010)
Discovery of canagliflozin, a novel
C-glucoside with thiophene ring, as sodium- 208. Anderson SL (2014) Dapagliflozin efficacy
dependent glucose cotransporter 2 inhibitor and safety: a perspective review. Ther Adv
for the treatment of type 2 diabetes Mellitus. J Drug Saf 5:242–254
Med Chem 53:6355–6360 209. Bolinder J, Ljunggren Ö, Kullberg J,
198. Meng W, Ellsworth BA, Nirschl AA, McCann Johansson L, Wilding J, Langkilde AM,
PJ, Patel M, Girotra RN, Wu G, Sher PM, Sugg J, Parikh S (2011) Effects of dapagliflo-
Morrison EP, Biller SA, Zahler R, Deshpande zin on body weight, total fat mass, and
PP, Pullockaran A, Hagan DL, Morgan N, regional adipose tissue distribution in patients
Taylor JR, Obermeier MT, Humphreys WG, with type 2 diabetes mellitus with inadequate
Khanna A, Discenza L, Robertson JG, glycemic control on metformin. J Clin Endo-
Wang A, Han S, Wetterau JR, Janovitz EB, crinol Metabol 97:1020–1031
Flint OP, Whaley JM, Washburn WN (2008) 210. Nauck MA, Del Prato S, Meier JJ, Durán-
Discovery of dapagliflozin: a potent, selective Garcı́a S, Rohwedder K, Elze M, Parikh SJ
renal sodium-dependent glucose cotranspor- (2011) Dapagliflozin versus glipizide as
ter 2 (SGLT2) inhibitor for the treatment of add-on therapy in patients with type 2 diabetes
type 2 diabetes. J Med Chem 51:1145–1149 who have inadequate glycemic control with
199. Neumiller JJ (2014) Empagliflozin: a new metformin: a randomized, 52-week, double-
sodium-glucose co-transporter 2 (SGLT2) blind, active-controlled noninferiority trial.
inhibitor for the treatment of type 2 diabetes. Diabetes Care 34:2015–2022
Drugs Context 3:212262 211. Strojek K, Yoon KH, Hruba V, Sugg J,
200. Hummel CS, Lu C, Liu J, Ghezzi C, Hir- Langkilde A, Parikh S (2014) Dapagliflozin
ayama BA, Loo DDF, Kepe V, Barrio JR, added to glimepiride in patients with type
Wright EM (2011) Structural selectivity of 2 diabetes mellitus sustains glycemic control
human SGLT inhibitors. Am J Physiol Cell and weight loss Over 48-weeks: a rando-
Physiol 302:C373–C382 mized, double-blind, parallel-group,
201. Valentine V (2015) The role of the kidney and placebo-controlled trial. Diabetes Ther
sodium-glucose cotransporter-2 inhibition in 5:267–283
diabetes management. Clin Diab 30:151–155 212. Geerlings S, Fonseca V, Castro-Diaz D, List J,
202. Kalra S (2014) Sodium glucose co-transpor- Parikh S (2014) Genital and urinary tract
ter-2 (SGLT2) inhibitors: a review of their infections in diabetes: impact of
basic and clinical pharmacology. Diabetes pharmacologically-induced glucosuria. Dia-
Ther 5:355–366 betes Res Clin Pract 103:373–381
203. Dietrich E, Powell J, Taylor JR (2013) Cana- 213. Sonesson C, Johansson PA, Johnsson E,
gliflozin: a novel treatment option for type Gause-Nilsson I (2016) Cardiovascular
2 diabetes. Drug Des Devel Ther effects of dapagliflozin in patients with type
7:1399–1408 2 diabetes and different risk categories: a
204. Aylsworth A, Dean Z, VanNorman C, Okere meta-analysis. Cardiovasc Diabetol 15:37
AN (2014) Dapagliflozin for the treatment of 214. MacIsaac RJ, Jerums G, Ekinci EI (2016)
type 2 diabetes mellitus. Ann Pharmacother Cardio-renal protection with empagliflozin.
48:1202–1208 Ann Transl Med 4:409–412
205. Heise T, Seman L, Macha S, Jones P, 215. Liakos A, Karagiannis T, Bekiari E, Boura P,
Marquart A, Pinnetti S, Woerle HJ, Dugi K Tsapas A (2014) Update on long-term effi-
(2013) Safety, tolerability, pharmacokinetics, cacy and safety of dapagliflozin in patients
and pharmacodynamics of multiple rising with type 2 diabetes mellitus. Ther Adv Endo-
doses of empagliflozin in patients with type crinol Metab 6:61–67
2 diabetes mellitus. Diabetes Ther 4:331–345 216. Sosale B, Sosale A, Bhattacharyya A (2016)
206. Kasichayanula S, Liu X, LaCreta F, Griffen Clinical effectiveness and impact on insulin
SC, Boulton DW (2014) Clinical pharmacotherapy cost after addition of dapagliflozin to
kinetics and pharmacodynamics of dapagliflo- patients with uncontrolled type 2 diabetes.
zin, a selective inhibitor of sodium-glucose Diabetes Ther 7:765–776
co-transporter type 2. Clin Pharmacokinet 217. Issandou M, Bouillot A, Brusq J, Forest M,
53:17–27 Grillot D, Guillard R, Martin S, Michiels C,
207. Ghezzi C, Hirayama BA, Gorraitz E, Loo Sulpice T, Daugan A (2009) Pharmacological
DDF, Liang Y, Wright EM (2014) SGLT2 inhibition of Stearoyl-CoA Desaturase
inhibitors act from the extracellular surface 1 improves insulin sensitivity in insulin-
of the cell membrane. Physiol Rep 2:e12058 resistant rat models. Eur J Pharmacol
618:28–36
30 John C. Clapham
218. Lepifre F, Christmann-Franck S, Roche D, drugs and targets for type 2 diabetes: review-
Leriche C, Carniato D, Charon C, Bozec S, ing the evidence. Am Health Drug Benefits
Doare L, Schmidlin F, Lecomte M, Valeur E 7:452–463
(2009) Discovery and structure-guided drug 223. Cornell S (2015) Continual evolution of type
design of inhibitors of 11b-hydroxysteroid- 2 diabetes: an update on pathophysiology and
dehydrogenase type I based on a spiro- emerging treatment options. Ther Clin Risk
carboxamide scaffold. Bioorg Med Chem Manag 11:621–632
Lett 19:3682 224. Morgan S, Grootendorst P, Lexchin J,
219. Birch AM, Buckett LK, Turnbull AV (2010) Cunningham C, Greyson D (2011) The cost
DGAT1 inhibitors as anti-obesity and anti- of drug development: a systematic review.
diabetic agents. Curr Opin Drug Discov Health Policy 100:4–17
Devel 13:489–496 225. Paul SM, Mytelka DS, Dunwiddie CT, Per-
220. Kaiser D, Oetjen E (2014) Something old, singer CC, Munos BH, Lindborg SR, Schacht
something new and something very old: AL (2010) How to improve R&D productiv-
drugs for treating type 2 diabetes. Br J Phar- ity: the pharmaceutical industry’s grand chal-
macol 171:2940–2950 lenge. Nat Rev Drug Discov 9:203–214
221. Ichimura A, Hasegawa S, Kasubuchi M, 226. Herper M (2012) The truly staggering cost of
Kimura I (2014) Free fatty acid receptors as inventing new drugs. Forbes. https://www.
therapeutic targets for the treatment of diabe- forbes.com/sites/matthewherper/2012/
tes. Front Pharmacol 5:236 02/10/the-truly-staggering-cost-of-
222. Miller BR, Nguyen H, Hu CJH, Lin C, inventing-new-drugs/
Nguyen QT (2014) New and emerging
Chapter 2
Practical Considerations for In Vivo Mouse Studies

Edward T. Wargent
Abstract
Obesity and type 2 diabetes are serious conditions that have reached pandemic proportions. The underlying
physiology is complex with multiorgan interactions involved and, consequently, multiorgan approaches are
necessary for researchers to elucidate and find treatments. As such, in vivo models are an invaluable resource
for these studies and mice are, for many reasons, by far the most common species used. The use of animals
comes with responsibilities to ensure their welfare and well-being: primarily for the sake of the mice
themselves, but also to ensure quality of data. Physiological stress responses, such as adrenalin (epinephrine)
and corticosterone release among others, have major consequences on metabolism. Additionally, behavioral
stress responses are also a source of data variance. This chapter looks at the main in vivo procedures
incorporated in mouse obesity/diabetes protocols and considers the practical factors involved that can be
considered to minimize animal stress and improve study data quality.
Key words Mouse, In vivo, Stress, Data quality, 3Rs
1 Introduction
Type 2 diabetes and obesity are worldwide major public health

problems. As well as its adverse psychosocial impact, obesity
increases susceptibility to type 2 diabetes (markedly), cardiovascular
disease (both dependent and independent of diabetes), and certain
cancers. By some estimates, it is close to becoming the biggest killer
in our society. Diabetes increases myocardial infarction, stroke,
kidney failure, blindness, and peripheral vascular disease and is a
major cause of male impotence.
The onus of obesity treatment traditionally lies with the
patients themselves expected to take responsibility for management
of their own condition [1]. This approach persists despite the
increasing prevalence of both diseases. Patients are rightly advised
to eat less and exercise more, but this simplistic therapy is evidently
insufficient. Currently, there are several anti-obesity medications
available, but orlistat is the only worldwide-approved drug
31
32 Edward T. Wargent
remaining for the treatment of obesity (the only one licensed in the
UK, for example). However, the effectiveness of orlistat is slim with
an average weight loss of only 2.9 kg (or about 2.3 kg for the over-
the-counter dose). There are more therapeutic options for type
2 diabetes, but the older drugs merely slow the advance of the
disease by a few years (at most), and the newer ones are either
suspected of having a poor benefit-to-risk ratio, or their ability to
halt the advance of the disease has yet to be demonstrated. There is
a need for better treatments for diabetes and obesity in the form of
drugs, nutraceuticals (since much of the diabetes epidemic is occur-
ring in countries and populations that cannot afford Western
drugs), diets or lifestyle modifications. To achieve this, we need to
understand better the causes of the diseases, identify potential
therapeutic targets, and investigate whether and how potential
interventions work. Control of metabolism, growth, and body
composition involve the interaction of many body organs including
brain, liver, adipose tissue, muscle, and endocrine organs. Neither
cell or tissue cultures nor nonvertebrate organisms can replicate
such mammalian whole body systems. Computational methods are
still in a developmental stage, requiring further experimental data
before they can reliably replace whole animals.
The use of animals in scientific research comes with responsi-
bilities to ensure the welfare and ethical treatment of those animals.
It is every in vivo researcher’s duty to explore ways to replace
animals with other means to answer the questions postulated; to
reduce the numbers of animals used; and the refine the procedures
employed to reduce harm and distress experienced by experimental
animals. When studying conditions such as obesity and diabetes,
the reduction of stress is also necessary as it impacts greatly upon
metabolism [2]. Consequently, reducing stress is the single most
important factor to be considered when designing reliable and
reproducible experiments and reducing the variation of the data
generated.
For manifold reasons, including finance, ethics, genetics, time
scales, etc., rodents and mice are by far the most widely employed
species in obesity and diabetes studies and this chapter describes the
typical in vivo procedures used in a manner intended to minimize
stress by exposing the rodents to no more than transient pain,
distress or harm. When carried out in the ways described, each
procedure can be classified as mild. However, animals should be
monitored before and after each procedure to ensure no compound
suffering from consecutive procedures. The recovery times detailed
below are minimal and the health of each animal should be moni-
tored carefully during and after each procedure.
Practical Considerations for In Vivo Mouse Studies 33
2 General Output Measures
Relevant output measures fall into three groups: obesity, insulin

sensitivity/glucose tolerance, and energy balance. Specific details
are stated where animal quality of life and experimental variance will
be improved. Further details can be obtained from the National
Centre for the Replacement, Refinement and Reduction of Animals
in Research (https://www.nc3rs.org.uk).
2.1 Obesity Body weight and composition may be determined in vivo by quan-
titative magnetic resonance (QMR) or dual-energy X-ray absorpti-
ometry (DEXA) scanning. QMR measurements can be made
without the need for anesthesia. In this procedure, mice are unre-
strained in a plastic cylinder with freedom to turn about but limited
vertical movement and placed in the scanner. Measurement is com-
pleted within 1–2 min. DEXA scanning requires animals to be
under light general anesthesia to prevent movement.
2.2 Insulin Blood levels of markers of insulin resistance and type 2 diabetes in
Resistance either the fed state or fasted state include glucose, insulin, lipids,
and Glucose Tolerance incretins, and pro- and anti-inflammatory molecules such as cyto-
kines and chemokines. Blood samples will typically be taken from a
2.2.1 Blood Levels superficial blood vessel or by removing the tip of the tail (~1 mm).
of Markers of Insulin Repeat samples can then be collected by removing the scab and
Resistance gently massaging the tail. Not more than 10% blood volume should
be taken in a single day and not more than 15% in a 28-day period.
On average, mice have around 58.5 mL of blood per kg of body-
weight and this approximation should be used for the above calcu-
lation, e.g., a 20.0 g mouse will be estimated to have 1.17 mL
blood and therefore blood sampling should be no more than
117 μL in a day or 175 μL in any 28-day period.
2.2.2 Glucose Tolerance GTTs assess the disposal of a glucose load administered via oral
Test (GTT) gavage or intraperitoneal/intravenous injection. The response to a
GTT is determined by insulin secretion and insulin resistance and is
impaired in obesity-related diabetes. Mice are fasted for 5–6 h to
ensure steady-state blood glucose levels. An aqueous glucose load is
administered in a volume of 10 mL/kg and blood glucose is
measured over a span of 2–3 h. The glucose load is typically
2–3 g/kg and may be adjusted based on lean body mass body
composition data if available due to muscle being the main tissue
responsible for glucose clearance in a GTT (the liver second). A
glucose load of 3 g/kg is sufficient to cause a doubling of blood
glucose concentration after 30 min in normal mice. Insulin-
resistant mice require less glucose to effectuate a doubling in
blood glucose concentration: 2.5 g/kg is recommended in diet-
induced obese mice and 2 g/kg in more extreme models of insulin
resistance such as ob/ob or db/db mice. A doubling in blood
glucose concentration is desirable as it produces a statistically sig-

nificant effect without overloading the system, thus providing a
window to show treatments that either increase or decrease insulin
sensitivity. The GTT duration varies for the same reason, with
blood glucose concentration taking 2 h to return to pre-dose levels
in normal mice. GTTs in insulin-resistant mice may not have
returned to pre-dose levels even after 3 h, but the purpose of the
GTT will have been fulfilled after this time and so the GTT should
be no longer than this. Typically, a blood sample (10 μL or less) is
taken prior to the glucose load (for baseline measurements) and
then at 30-min intervals for the duration of the experiment (for 3-h
GTTs, a 150 min time point does not improve the quality of the
data and should be omitted for refinement purposes). Thirty-
minute intervals are recommended, as the stress-induced increase
in blood glucose following handling/bleeding will usually be insig-
nificant by then. An additional blood 30 min before glucose load
can be performed to demonstrate this (and this may be a conve-
nient time for a treatment dose, should the GTT incorporate one in
addition to the glucose load). Oral GTTs represent the most phys-
iological route of entry of glucose. Mice like the taste of glucose
water and it is sometimes possible to train unstressed mice to drink
a small volume from the tip of a syringe. Experimenters should
conduct preliminary tests to see if their mice will voluntarily drink
the glucose load. Alternatively, it may be administered by gavage or
administered with a feeding needle. The appearance and clearance
of blood glucose during an oral GTT is affected by the rate of
gastric emptying and the incretin effect and if it is necessary to
circumvent these processes, the glucose may be given by the intra-
peritoneal or intravenous routes. Blood samples for the measure-
ment of plasma insulin are taken 30 min before and 30 min after the
administration of glucose. These are much larger sample sizes,
taking longer to collect and increasing the stress the mouse experi-
ences, and should be limited to these time points only for the sake
of the animal and the quality of the GTT.
2.2.3 Sensitivity This is the gold standard technique for measuring insulin sensitiv-
to Insulin, by Euglycemic/ ity. Mice are fasted for 4–6 h to prevent the appearance of meal-
Hyperinsulinemic Clamp derived glucose, but not deplete liver glycogen stores. Mice then
receive a constant infusion of insulin to achieve a steady-state
hyperinsulinemia above fasting insulin level. A variable glucose
infusion is administered to maintain euglycemia. Infusions are
administered via a previously implanted indwelling catheter. The
glucose infusion rate is determined by measuring blood glucose at
regular intervals, typically every 10 min, and adjusting the glucose
infusion accordingly. This variable rate of glucose infusion indicates
whole-body insulin action, as mice with enhanced insulin action
require a greater infusion of glucose. Isotopic tracer infusions can
be applied to assess sites where insulin action is affected.
2.2.4 Sensitivity Following a short (4 h) fast, glucose concentration is monitored

to Insulin, by Insulin every 15–30 min for 60–90 min following a bolus of insulin admi-
Tolerance Test (ITT) nistered via intraperitoneal or intravenous injection. A bolus insulin
dose to fasting mice comes with a risk of hypoglycemia and death,
and the shorter fasting period in ITTs compared to GTTs, albeit
only an hour, provides a significantly safer procedure in terms of
mouse mortality. The safety factor far outweighs the risk of having a
small amount of glucose still entering the blood from the gut. The
degree to which glucose falls following the insulin bolus is indica-
tive of whole-body insulin action. The insulin dose will be deter-
mined according to lean body mass where appropriate and may be
higher if there is insulin resistance. A dose is chosen which is not
expected to produce hypoglycemia: typically, 0.5 mU/kg in normal
mice and 0.75 mU/kg in insulin-resistant mice (severely insulin-
resistant phenotypes may require an even higher insulin dose, e.g.,
ob/ob and db/db mice may require up to 1.5 mU/kg, but alterna-
tives to such studies should be carefully considered due to the
variability and uncertain success rate). A lower dose can be used
in phenotyping studies as it gives a good response in wild-type
animals without reducing blood glucose to dangerous levels. If
the genotype causes insulin resistance, then no reduction is seen
in this group. In mice with diet-induced insulin resistance, the
insulin dose used needs to be higher to elicit a borderline nonsig-
nificant reduction in glucose. Drugs that improve insulin sensitivity
therefore have the best chance to have their effect detected at this
insulin dose. For the same reason, the insulin dose will alter in
genetically insulin-resistant mice, though the exact dose will vary
for each transgenic/mutant. Any animal that shows signs of torpor
after insulin administration should immediately be given glucose
and glucagon (amount of glucagon equal to counteract the insulin
dose) by the intraperitoneal route, monitored continuously, and
culled if it fails to respond to stimulation or does not recover within
20 min (treatment with both glucose and glucagon should recover
95% of mice experiencing torpor). If it is suspected that a strain of
mouse may be overly sensitive to insulin, then small pilot studies
should be performed.
2.2.5 Real-Time Minute- Under in vivo conditions, the study of physiological and pharma-
by-Minute Changes cological functions of an organ is difficult due to whole-body
in Insulin Secretion interactions with the organ. In vitro perfusion of isolated pancreas
is required for physiologic and response studies, including the
investigation of endocrine function and secretory responsiveness
under a variety of diabetes-associated conditions and treatments.
Animals are terminally anesthetized, and the pancreas isolated from
the connecting spleen, stomach, and duodenum and transferred to
a pre-warmed chamber, where it is perfused in isolation from all
other organs and the perfusate collected for analysis of insulin
secretion.
2.2.6 Blood Pressure Hypertension is associated with hyperinsulinemia, and the same
programming conditions that cause increased risk of obesity and
diabetes also increase the susceptibility to hypertension. Therefore,
blood pressure is an important measure and can be done satisfacto-
rily by the noninvasive tail cuff method. Systolic blood pressure
equal to or greater to 140 mmHg is considered to be hypertensive.
A positive outcome of treatment would be to reduce systolic blood
pressure below 130 mmHg.
2.3 Energy Balance (a) Measurement of food intake and measurement of overall daily
energy expenditure using indirect calorimetry measuring the
amount of oxygen consumed and carbon dioxide produced in
expired air.
(b) Measurement of basal metabolic rate by measuring the utiliza-
tion of glucose and release of carbon dioxide in animals in a
thermoneutral environment over 8 h. The only difference
from normal measurements of energy expenditure is that the
ambient temperature raised to 28–30 C and measurements
are limited to 8 h during the light period.
(c) Measurement of brown adipose tissue activity after adminis-
tration of a thermogenic agent (such as a beta 3-adrenoceptor
agonist) by measuring oxygen consumption and carbon diox-
ide output and blood pressure indirectly by the tail cuff
method. Brown adipose tissue thermogenic capacity will also
be assessed ex vivo by determining the expression of uncou-
pling protein-1.
3 Considerations for In Vivo Procedures in Obesity and Diabetes Protocols
3.1 Strain Many strains of mice are available for the study of obesity and type
2 diabetes and the question of a “right model” per se a complex,
and possibly unanswerable, one [3]. Genetically modified mice will
usually be generated on a 129 background and thereafter trans-
ferred to the strain of choice, usually C57Bl/6 mice because of
their susceptibility to environmentally induced obesity and insulin
resistance. However, the strain employed will depend on the ques-
tion being posed by the study. For example, studies into browning
of adipose tissue may well be suited to being on the 129 strain
because they exhibit a propensity for browning of white adipose
tissue and so have a window to both upregulate and downregulate
this physiological effect. Experimental consequences will occur
with all strain, e.g., in the case of 129 mice, the propensity for
browning might make them suitable to study promoters of brown-
ing, but will make them resistant to the effects of a high-fat diet and
so require a longer period before they develop insulin resistance and
require a 60% fat diet rather than a 40% fat diet. The complex
physiology of obesity and diabetes makes a study of strain suitability

within the confines of a single book chapter impossible. Fortu-
nately, there are mouse strain resources that researchers can consult
in addition to literature tapping, e.g., International Mouse Strain
Resource (https://www.jax.org/research-and-faculty/resources/
international-mouse-strain-resource).
3.2 Single Housing Animals should be group housed wherever possible. For some
studies, it may be necessary to house animals singly so that individ-
ual food intake can be measured, or to remove any effect of hierar-
chy or aggression on feeding behavior.
3.3 Food Modified diets should be adequate with respect to all known nutri-
and Feeding ents including vitamins and minerals. As such, commercial rodent
feeds are preferable to handcrafted feeds, and these are also consis-
3.3.1 Feeding
tent in constitution. Diets should be stored under appropriate
a Modified Diet
conditions to prevent deterioration and be given a defined shelf-
life. Feeding high-fat diets may reduce wear of the teeth. Over-
growth of teeth will reduce the animal’s ability to feed and drink
resulting in loss of weight. All experimental animals should have
their teeth checked for malocclusion, with the weighing times a
good time to do this. Where possible, animals should be provided
with environmental enrichment, such as wooden sticks, to help
increase teeth wear. Advice on treatment or culling should be
sought from a veterinary surgeon if teeth become overgrown.
Likewise, diets with a high sugar intake may increase the risk of
dental caries, which may cause pain and a reduction in food intake.
Mice on these diets should have their teeth inspected weekly (when
weighed would be ideal). If evidence of dental caries is observed,
the advice on tooth management should be sought from a veteri-
nary surgeon.
3.3.2 Pair-Feeding In the event of an experimental group having reduced food intake,
it may be necessary to pair-feed a group of control animals so that
they receive the same amount of food. A reduction in food intake
results in an improvement in insulin sensitivity and pair-feeding
allows differentiation between the insulin sensitizing effects of
reduced food intake and the primary mechanistic actions being
investigated in the study. This reduction in food intake should be
no less than 40% of normal on any one occasion (up to a maximum
of 72 h) and no more than 25% over the period of an experiment to
minimize the stress imposed on an animal. Body weight must be
measured to assess the impact of pair-feeding on the control ani-
mals and should not drop below 80% of the weight of normal mice
on standard diet.
3.3.3 Fasting The withholding of food from rodents is stressful but some studies
require deprivation of food for a defined period. The minimal
period of food withdrawal commensurate with the objectives of
the experiment should therefore always be used and should never
be greater than 18 h, which may be required if it is necessary to
deplete the primary nutrient stores of glycogen and mobilize lipid
stores from adipose tissue. Measurements of lipid dysregulation are
more likely to be seen if the animal is fasted for 18 h in mice
weighing 20 g or more and 5 h in mice less than 20 g. Animals
should have a minimum of 3 days with continuous availability of
food between any periods of fasting. There is no evidence that mice
that are hyperphagic show an enhanced level of fasting stress.
For some studies, animals may need to be deprived of food for a
defined period. Glucose tolerance and insulin sensitivity tests
require there to be no uncontrolled glucose appearance into the
blood and require a fasting period (typically 5 h) that is long
enough to ensure that the gut is empty, but short enough not to
stimulate hepatic gluconeogenesis. This is typically 5 h in mice.
3.4 Temperature The topic of appropriate housing temperature for mice is currently
under debate. With appropriate nesting material, mouse housing
temperatures are generally maintained 1–2 C above the room
temperature. Thermoneutrality (the temperature with the mini-
mum of physiological thermoregulation) is achieved with a room
temperature of 28–30 C. Below this zone, mice start to generate
their own heat by increased skeletal muscle and brown adipose
tissue activity, resulting in increased fat catabolism. Therefore,
studies designed to study the adverse effects of fat accumulation
(particularly those investigating chemical or genetic resistance to
these effects) would benefit from higher temperatures, e.g., high-
fat diet-induced obesity and insulin resistance is promoted at tem-
peratures close to thermoneutrality (with a tendency toward
reduced variation in data generated). However, mouse aggression
also increases with temperature and mice housed at thermoneutral-
ity tend to fight more and for this reason, mice should only be
housed at thermoneutrality when necessary. C57Bl/6 mice, a strain
commonly employed in metabolic studies because of its physiolog-
ical suitability, are naturally aggressive and an alternative strain
should be considered if housing at thermoneutrality for extended
periods be intended. Behavioral studies have shown that mice
prefer temperatures higher than the usual housing conditions of
around 20–24 C; however, temperatures in the thermoneutral
range are only preferred during periods of rest. Taking the meta-
bolic and behavioral aspects into consideration, the ethically recom-
mended temperature range for in vivo mice protocols is 24–26 C.
Procedures investigating energy expenditure at thermoneutrality
should be performed during the light phase for a maximum period
of 8 h.
3.5 Blood Sampling Blood sampling would preferably be taken either from a superficial
blood vessel or from the tip of the tail after removal of no more than
0.5 mm of the tip of the tail on the first occasion and on subsequent
occasions after removal of the scab. Blood sampling from a superfi-
cial vessel will be as noninvasive as possible. Pain from tail tipping
for blood samples can be controlled by local anesthetic. The volume
of blood taken should not exceed 10% of total blood volume on a
single occasion or 15% in a 28-day period.
Blood sampling may be facilitated by placing the animals in a
warm environment for the period of the sampling, but this will not
have any adverse effects on the animal. Animals should be checked
for bleeding at 1 and 12 h after the procedure and more frequently
if required. Any other than minor bleeding after blood sampling
should be controlled by the temporary application of pressure and a
clotting agent. A veterinary surgeon should be consulted for advice
on wound healing if any animal showing signs of bleeding more
than 24 h after the last sampling. Any animal showing signs of
excessive blood loss such as torpor lethargy or anemia should be
culled.
Repeat sampling from the tip of the tail may result in bruising if
too much pressure is applied when massaging. If hematomas and
localized inflammation are noted at the tip of the tail, then animals
should be carefully monitored. Localized inflammation and redness
should recede within 24 h. If prolonged redness is observed or is
accompanied by swelling or darkening of the tail, the animal should
be removed from any further procedures. Any animal showing signs
of necrosis, infection or delayed healing more than 24 h after the
last sampling should be culled.
Long-term studies should have a maximum of five tolerance
tests each requiring repeat blood sampling during the procedure.
Animals should be assessed prior to each test to ensure the animal
has healed fully from the previous procedure, and if there is any sign
of hyperalgesia or abnormal healing, the animal should not be used.
3.6 Administration The mini-pump could be either intraperitoneal or subcutaneous

of Test Compounds and should not be removed or re-used. Animals are expected to
recover quickly and to show no adverse effects from the surgery
3.6.1 Implantation of an
other than a small and transient reduction in food intake. It is
Osmotic Mini-pump
expected that the mini-pump will initiate the formation of a fibrous
for Delivery of Compound
coat around itself, but this will cause minimum pain, distress, or
by Constant Infusion
lasting harm. Pumps should be neither replaced nor removed.
Following surgery, body weight should be measured daily to assess
for any post-operative loss in bodyweight. A loss of 20% body
weight should be the humane end-point at which an animal will
be culled. The skin of obese and diabetic animals does not have the
same elasticity as that of normal animals and wound healing can be
impaired. A special watch should be kept on such animals and the
animals culled by an appropriate method if tears appear. Animals
with implanted pumps should not be kept more than 7 days after
the expected exhaustion of the pump, nor should they undergo
measurement of body composition by QMR.
3.6.2 Administration This is the first choice of route for chronic dosing as no discomfort
of Test Agents in the Food or distress is expected from feeding compounds in the diet and
or Water should be employed if compounds can be homogenously
distributed with no degradation. If this is not possible, or if taste
aversion occurs, another method will need to be considered. Waste
measurement for accurate assessment of dosage is also a consider-
ation. Leaky water bottles, “shredding” of solid diets and dispersal
of powdered diets out of the food tray are complications that may
arise. There are no fool proof methods for overcoming these issues,
e.g., trays with a wire top tend to accumulate urinary and fecal
waste and wire-bottomed cages with a waste tray underneath can
cause weight loss in the mice. Another disadvantage to this route is
that this is not a method that would be used in humans, and so
some studies, by necessity, require a bolus dosing route.
3.6.3 Oral Gavage Dosing by oral gavage in mice should be at a maximum of 20 mL/
kg, and preferably 10 mL/kg. Some minor discomfort and tran-
sient distress is expected on each occasion, and on rare occasions,
the esophagus becomes damaged or a small volume may inadver-
tently enter the lungs. Animals should be observed immediately
after dosing. Any animal showing signs of mis-dosing or damage
such as by coughing/choking or collapsing after administration of
substances should be culled. A mis-dose directly into the lungs is
extremely unlikely and results in immediate collapse shortly fol-
lowed by death. Should collapse occur, the mouse should be imme-
diately culled. More likely than a direct dose into the lungs is the
event of a small amount of fluid ending up in the lungs. This
happens when the dosing tube is inserted into the esophagus rather
than the stomach, combined with too rapid a dosing. Detection is
by careful listening immediately after dosing—a rasping sound will
indicate this has happened. If this happens, the animal should be
immediately culled, as death will occur slowly within 2 days and
cause pain and distress.
3.6.4 Subcutaneous Substances should be administered by the subcutaneous route at a

Injection maximum volume of 20 mL/kg, and preferably 10 mL/kg. Some
minor discomfort and transient distress is expected on each occa-
sion. On repeat dosing, the injection site should be alternated over
five subcutaneous areas in the flank and scapula (except for when
direct exposure of brown adipose tissue is undesired, which case the
scapula will not be used). Animals should be observed immediately
after dosing and monitored throughout the treatment period for
signs of discomfort, pain, or distress. Signs to look for are lumps/
bumps or skin ulcerations and localized inflammation/redness:
should this exceed a diameter of 5 mm, these mice should receive
no further doses.
3.6.5 Intraperitoneal Substances should be administered by the intravenous route at a

Injection maximum volume of 20 mL/kg, and preferably 10 mL/kg. Some
minor discomfort and transient distress is expected on each occa-
sion. Occasionally, a superficial blood vessel may be punctured. On
repeat dosing, the injection site should be alternated over four areas
in the peritoneum. Animals should be observed immediately after
dosing and if treatment is for a prolonged period of time, the mice
should be monitored for signs of pain and distress that may indicate
peritonitis (hunching, subdued behavior, hind limb extension).
Any animal showing signs of peritonitis should be culled. In
obese mice, intraperitoneal dosing should be performed only
when no other route is suitable as a proportion of animals will
receive the dose in the fat pads instead. Data from tolerance tests
can be analyzed for lack of appearance of glucose (although exclu-
sion of data solely on this basis is not good practice).
3.6.6 Intravenous Substances may be administered by the intravenous route at a

Injection maximum volume of 10 mL/kg, but preferably at 5 mL/kg.
Some minor discomfort and transient distress is expected on each
occasion. The main adverse effects of intravenous administration
are needle stick injuries, bruising (which may be a result of the
animal aggravating the site by licking and rubbing particularly
when placed back in the cage), and loss of visibility of the vein.
These adverse effects can be minimized using a suitable restraint,
and good dosing technique (by administering slowly), including
use of appropriately sized needles and warming the procedure room
to ~25 C to facilitate vasodilatation. Any bleeding after dosing
should be controlled by the application of local pressure. On repeat
dosing, the site of injection should be altered.
3.7 Energy Energy expenditure may be measured by whole body calorimetry at

Expenditure room temperature or at thermoneutrality (28–30 C). The equip-
ment draws air out of the enclosed animal cage which has a small air
inlet hole to allow replacement and an alarm should be
incorporated to detect if air ceases to be drawn from the cage.
This can be linked to a mobile phone device.
3.7.1 Measurement Energy expenditure should be measured in the normal housing

of Energy Expenditure cages so as not to change behavior. No stress is expected from
at Room Temperature measuring energy expenditure at room temperature.
3.7.2 Measurement Energy expenditure measured at thermoneutrality (28–30 C)

of Energy Expenditure should be done for no more than 8 h during the light period and
at Thermoneutrality the temperature of the room should be regularly monitored. Ani-
mals kept at thermoneutrality are expected to show a relaxed pos-
tural extension.
They should be observed at least every hour to ensure that this

behavior is not altered to one of heat stress. This can be recognized
by locomotor activity (to escape to a cooler environment) and exces-
sive grooming (to spread saliva). Animals showing signs of heat stress
should be moved to a normal environment, and if these signs do not
begin to decrease within 20 min, they should be culled by an
approved method. When animals have been treated in a way that
might increase their energy expenditure or interfere with heat loss
mechanisms, they must be checked every 30 min for the first 2 h.
References
1. Bessesen DH, Van Gaal LF (2018) Progress and 3. Leiter EH (2009) Selecting the “right” mouse
challenges in anti-obesity pharmacotherapy. model for metabolic syndrome and type 2 diabe-
Lancet Diabetes Endocrinol 6:237–248 tes research. Methods Mol Biol 560:1–17
2. Rabassa C, Dickson SL (2016) Impact of stress
on metabolism and energy balance. Curr Opin
Behav Sci 9:71–77
Chapter 3
Nutritional Models of Type 2 Diabetes Mellitus

Beverly Sara Mühlh€ausler, Carla Toop, and Sheridan Gentili
Abstract
In order to better understand the events that precede and precipitate the onset of type 2 diabetes (T2DM),
several nutritional animal models have been developed. These models are generated by manipulating the
diet of either the animal itself, or its mother during her pregnancy, and in comparison to traditional genetic
and knock out models, have the advantage that they more accurately reflect the etiology of human T2DM.
This chapter will discuss some of the most widely used nutritional models of T2DM: Diet-induced obesity
(DIO) in adult rodents, and studies of offspring of mothers fed a low-protein, high-fat and/or high-sugar
diet during pregnancy and/or lactation. Several common mechanisms have been identified through which
these nutritional manipulations can lead to metabolic disease, including pancreatic beta-cell dysfunction,
impaired insulin signaling in skeletal muscle, and the excess accumulation of visceral adipose tissue and
consequent deposition of nonesterified fatty acids in peripheral tissues. In addition, there is an emerging
concept that obesity/poor quality diets result in increased production and release of pro-inflammatory
cytokines from adipose tissue leading to a state of chronic low-grade inflammation, and that this is likely to
represent an important link between obesity/diet and metabolic dysfunction. The following chapter will
discuss the most common nutritional models of T2DM in experimental animals, their application, and
relationship to human etiology, and will highlight the important insights these models have provided into
the pathogenesis of T2DM.
Key words Type 2 diabetes, Obesity, Insulin resistance, Animal models, Nutrition, High-fat diet,
Programming
1 Introduction
While the human is undoubtedly the model of choice when study-

ing the pathophysiology of human disease, the study of the under-
lying mechanisms of disease in living humans has a number of
logistical and ethical limitations. Therefore, to better understand
the events which precede and precipitate the onset of type 2 diabetes
(T2DM), there is a need to develop in vivo animal models of this
disease. Commonly used genetic models of T2DM, including ob/
ob and db/db mice and Zucker fa/fa rats, have been useful in
understanding some of the mechanisms which may contribute to
altered glucose and insulin metabolism; however, none of these are
43
44 €usler et al.
Beverly Sara Mühlha
an ideal disease model, since these gene mutations are extremely

rare in human populations [1]. Similarly, experimental animal mod-
els of T2DM induced by chemical destruction or removal of a
portion of the pancreas [2] are not representative of the etiology
of T2DM in humans, which is typically preceded by obesity [3, 4].
This has led to the development of a number of nutritional
animal models of T2DM, which more closely mimic the etiology of
the disease in human populations. These nutritional models involve
the manipulation of the diet of either the animal itself, or its mother
during pregnancy and/or lactation. A number of dietary
approaches have been utilized in an attempt to best mimic the
features of insulin resistance and T2DM in humans; however, the
majority involve increases in dietary fat, in particular saturated fat,
and/or increases in sugar (sucrose/fructose). This chapter will
present an overview of these nutritional models of T2DM, with a
particular focus on different dietary approaches used in these stud-
ies. It will discuss the application and relationship of these models
to human T2DM etiology and the mechanisms contributing to the
development of insulin resistance/T2DM that these nutritional
models have provided.
2 Diet-Induced Obesity
A significant proportion of T2DM in human populations is linked

to an excessive accumulation of body fat, particularly in the abdom-
inal region [4–6]. This fact makes the diet-induced obesity model
(DIO) particularly relevant for studying the underlying mechan-
isms through which an excessive accumulation of body fat and/or
an excessive dietary fat and/or sugar intake can lead to the devel-
opment of insulin resistance and T2DM. The two most widely used
models for the study of obesity-induced T2DM are high-fat/high-
sugar feeding in rodents and the sand rat (Psammomys obesus),
which develops obesity spontaneously when fed on a standard
rodent laboratory diet.
2.1 Diet-Induced The DIO rodent model involves a regimen in which healthy, non-
Obesity: Rats and Mice obese mice or rats are provided with ad libitum access to a highly
palatable high-fat, high-energy diet. C57BL/6J mice were origi-
nally selected for this model because in early studies, these mice
developed clear-cut diabetes more rapidly than other strains fed the
same high-fat diet, suggesting a genetic predisposition to T2DM in
this strain [7]. In this model, male C57BL/6J (B6) mice are main-
tained on these diets for 8–12 weeks, and as a result, become obese,
mildly to moderately hyperglycemic, and develop impaired glucose
tolerance [7]. Studies where the animals are maintained on the diets
for longer periods have also demonstrated that after 16 weeks on a
high-fat diet, C57BL/6 mice exhibit adipocyte hyperplasia and
hypertrophy, and are insulin resistant [8]. For similar reasons, that
Models of T2DM 45
is, due to their propensity toward diet-induced insulin resistance in

comparison to other common breeds, Wistar and Sprague-Dawley
rats have been the rat strains of choice for studying the metabolic
effects of diet-induced obesity [1].
The use of high-energy diets to induce obesity in laboratory
animals for the purpose of studying obesity-related disorders has
been widely adopted, and the scope of these diets is now quite
extensive, ranging from an oil-based diet to free access to a variety
of human junk foods, including pies, cakes and chocolate, termed a
“cafeteria diet” [9–13]. The composition of these diets varies con-
siderably between studies. However, the most common nutritional
models involve increasing the dietary content of fat and/or sugar in
combination with an increase in energy density. The response of the
animals to these diets depends both on the specific dietary compo-
sition and the breed/strain of the animal, and both must therefore
be considered when selecting an appropriate nutritional approach
for a given study. We will briefly consider the different dietary
approaches below.
2.1.1 High-Fat Diets The majority of high-fat diets used in rodents provide between 40%
and 60% energy as fat, as compared to ~10% energy from fat in
standard laboratory diets [14]. Historically, the major source of fat
in these diets has been lard, and therefore, most studies to date have
predominately increased the dietary saturated fat content. Feeding
rodents diets containing high levels of saturated fat has consistently
been shown to promote fat deposition, increase the production of
pro-inflammatory cytokines, and result in the development of insu-
lin resistance, hypertriglyceridemia, hyperphagia, hypertension,
and nonalcoholic fatty liver disease [14–16].
Over recent years, it has also become increasingly clear, how-
ever, that different types of dietary fats can have profoundly differ-
ent physiological effects. Consequently, not all high-fat diets
provide a suitable nutritional approach for inducing insulin resis-
tance/T2DM. This is particularly true of diets which contain high
levels of the omega-3 long-chain polyunsaturated fatty acids (n 3
LCPUFA), eicosapentaenoic acid (EPA) and docosahexaenoic acid
(DHA), since these n 3 LCPUFA, unlike saturated and omega-6
polyunsaturated fats, have been reported to have both antilipo-
genic/adipogenic and insulin-sensitizing effects in vitro and
in vivo [17, 18].
The insulin sensitizing and anti-obesity actions of n 3
LCPUFA have been attributed to both direct effects on insulin-
sensitive tissues, and the effects of the bioactive derivatives of this
class of fatty acids. Studies in vitro and in adult rodents have shown
that both EPA and DHA act to suppress expression of genes
involved in preadipocyte proliferation/differentiation, thereby
inhibiting the expansion of adipose depots by hyperplasia [17]. In
addition, these fatty acids also suppress expression/activity of the
46 €usler et al.
lipogenic transcription factor, SREBP-1c in mature adipocytes, and

thus inhibit the accumulation of triglycerides in mature adipose
depots [18, 19]. This suppression of both hyperplastic and hyper-
trophic expansion of fat depots is thought to underlie the beneficial
effect of n 3 LCPUFA on fat deposition. However, despite the
strong biochemical data, and the encouraging results of rodent
studies, results from human trials investigating the potential of
n 3 LCPUFA in assisting weight loss have been disappointing.
One possible explanation for this is the high level of omega-6
PUFA which are present in most typical human diets in Western
countries, since this class of fat has been shown to have proadipo-
genic and prolipogenic actions in vitro and in vivo [20, 21].
The insulin-sensitizing effects of the n 3 LCPUFA are again
driven both by direct and indirect effects. Both EPA and DHA
activate the transcription factor PPARγ in adipose tissue, which
increases the synthesis and secretion of the insulin-sensitizing adi-
pokine, adiponectin [22]. Adiponectin then acts via its receptors on
peripheral insulin-sensitive tissues, in particular, the skeletal muscle
and liver, to improve peripheral insulin sensitivity [22]. The n 3
LCPUFA also give rise to bioactive derivatives which either
oppose/dampen the pro-inflammatory effects of the n 6 PUFA
derivatives or have inherent anti-inflammatory properties [23].
These therefore act to suppress inflammatory processes and pro-
duction of pro-inflammatory cytokines in adipose and other tissues
which is thought to be a major contributor to the development of
insulin resistance and T2DM in humans [24].
These data are important to consider when designing nutri-
tional models of insulin resistance/T2DM, since they imply that
the balance of n 6 and n 3 PUFA in the diet are likely to be
important determinants of the effects of the diet on fat deposition
and insulin action. Therefore, the content of both these fatty acid
classes needs to be taken into account in the design of nutritional
approaches for inducing insulin resistance/T2DM in experimental
animal models.
2.1.2 High-Sucrose In recent years, there has been an increasing focus on the contribu-
and High-Fructose Diets tion of added sugars in foods/beverages to the current epidemic of
obesity and T2DM in human populations worldwide. This interest
has been heightened by data from epidemiological studies, which
have demonstrated that consumption of one or more sugar-
sweetened beverage per day is associated with an elevated risk of
developing high blood pressure, metabolic syndrome [25], and a
22% increase in the risk of developing T2DM [26]. These data have
resulted in the American Heart Association advising that daily
added sugar intake be limited to 100 cal and 150 cal for women
and men respectively [27], which is equivalent to approximately
5–10% of daily energy intake. More recently, the World Health
Models of T2DM 47
Organization recommended that guidelines for added sugar intake

be reduced to 25 g per day, less than a single can of sugar-sweetened
beverage [28]. Depsite these recommendations, however, the aver-
age daily consumption of added sugars in Western countries
remains high, with average daily intakes of 15% total energy intake
in the United States, United Kingdom, and Australia [27, 29, 30].
Sucrose (cane/table sugar) is the most commonly used sweet-
ener worldwide; however, the use of high-fructose corn syrup-55
(HFCS-55), is becoming more widespread, particulary in the
United States. Following ingestion, sucrose (a disaccharide of glu-
cose and fructose) is hydrolyzed into its monosaccharides prior to
absorption, whereas the free glucose and fructose in HFCS-55
(55% fructose, 42% glucose) can be absorbed directly into the
circulation [31]. The absorption of free fructose is limited by the
physiological capacity of the glucose transporters (GLUTs), and
this can result in malabsorption if consumed in large amounts
[32]. The metabolic effects of sucrose and HFCS-55 are thought
to more closely mimic those of fructose consumption alone; hence,
the adverse metabolic outcomes, including dyslipidemia, ectopic fat
deposition, impaired glucose tolerance and insulin resistance, asso-
ciated with excess sugar consumption have been attributed to the
fructose component [33].
A number of rodent studies have begun to utilize high-sucrose
and high-fructose diets as a nutritional strategy for inducing obesity
and insulin resistance. However, the results of these studies have
not been consistent, apparently because of variation in the meta-
bolic response to the diets both between species/strains, concen-
tration and the way in which the sucrose/fructose are administered.
Studies looking at fructose and sucrose range from a 10% w/v
beverage to 60–70% fructose or sucrose content of chow, and
many use a supra-physiological dose of sucrose or fructose, com-
monly 60% content of chow [33, 34]. In Wistar rats, providing
sucrose either in solid form (e.g., as a component added to the
feed) or in the drinking water (as a 10–30% w/v sucrose solution)
has been associated with both increased visceral fat accumulation
and insulin resistance in both liver and skeletal muscle in male
animals [35–37]. In mice, however, adding sucrose to drinking
water has had more varied effects. Providing additional sucrose to
C57BL/6 mice fails to induce obesity; however, recent data sug-
gests that this does induce more subtle changes in the metabolic
phenotype of the animals, including adipocyte hypertrophy, glu-
cose intolerance, hyperinsulinemia, hyperlipidemia, fatty liver, and
increased levels of inflammatory cytokines [38]. Further studies in
the area of high-sucrose feeding are required in order to establish
whether this is an appropriate model of T2DM in humans and to
fully elucidate the effects of sucrose/fructose consumption on male
and female animals.
48 €usler et al.
While fructose does not acutely increase insulin levels, chronic

exposure to high-fructose diets in both rats and mice has been
associated with the development of obesity, insulin resistance,
impaired glucose tolerance, hyperinsulinemia, and hypertriglycer-
idemia [34]. In one study, feeding rats a 66% fructose diet for only
2 weeks was associated with a reduced abundance of the insulin
receptor in both skeletal muscle and liver, and these changes pre-
ceded any changes in insulin sensitivity, plasma glucose or insulin
concentrations in these animals [39]. High-fructose diets in
rodents have also been associated with increased accumulation of
nonesterified free fatty acids in tissues, resulting in lipotoxicity and
ultimately leading to the development of insulin resistance
[33]. Fructose feeding also stimulates expression of pro-lipogenic
genes, SREBP-1 mRNA and FAS mRNA, in liver, and these molec-
ular changes are implicated in the development of fatty liver and
hepatic insulin resistance [39, 40]. While there are still relatively few
studies which have utilized high-fructose feeding as a model of
T2DM, the data to date are encouraging and support the utility
of this nutritional strategy as an appropriate model of T2DM in
humans [34].
2.1.3 Studies of Obesity- It is now widely accepted that the susceptibility to developing
Prone and Obesity- obesity and T2DM varies between individuals, and there is consid-
Resistant Animals erable interest in understanding the genetic and physiological basis
of this difference. This question has been addressed using an adap-
tation of the DIO model in which adult Sprague-Dawley rats are
fed a purified diet with a moderately high fat content. The rats are
subsequently selected as either obesity-prone or obesity-resistant
according to their response (i.e., degree of weight gain and increase
in percentage body fat) on the high-energy diet [41]. The obesity-
prone rats typically eat approximately 16% more calories over the
first 30 days compared to the obesity-resistant strain and exhibit a
phenotype comparable to human metabolic syndrome, including
glucose intolerance, hyperinsulinemia, and T2DM [41, 42]. Selec-
tive breeding has enabled these researchers to generate obesity-
resistant and obesity-prone strains, and to study these two popula-
tions in order to determine factors which may underlie an increased
propensity to obesity [43, 44]. This is a useful model, in that it
provides a means of distinguishing between the effects due to high
dietary fat content as distinct from those related to the effects of
excess body fat.
2.1.4 Summary: Diet- Irrespective of the precise nature of the diet, diet-induced obesity
Induced Obesity as a Model results in many of the same changes as seen in human obesity,
of Human T2DM including the development of central and peripheral insulin and
leptin resistance [9], increased lipid accumulation in peripheral
tissues, a pro-inflammatory state and in the altered expression of
Models of T2DM 49
adipokines which are known to contribute to the regulation of

peripheral insulin sensitivity, in particular adiponectin and
resistin [45].
DIO rodent models have been widely utilized in order to
investigate the defects within specific tissues that may contribute
to the development of insulin resistance in obese individuals. By
comparing the profile of gene expression and protein content in
specific tissues of DIO mice with those from lean controls using
micro-array or 2D gel electrophoresis, it is possible to quickly
identify genes which may be involved in the development of
T2DM. Using this approach, DIO was shown to be associated
with a reduced pancreatic abundance of enzymes involved in the
clearance of reactive oxygen species, providing a potential mecha-
nism for the deterioration of pancreatic function seen in the later
stages of T2DM [46]. In addition, DIO is associated with adipo-
cyte hypertrophy [47] and the induction of hepatic steatosis and
hepatic insulin resistance [48], which again resembles the human
obese state. More recently, high-fat diets, high-sugar diets, and
excess fat accumulation have been linked to increased production
of pro-inflammatory cytokines by adipose tissue and infiltration of
these into peripheral tissues, resulting in a state of chronic
low-grade inflammation [49]. This heightened inflammatory state
is now considered to be a key factor which links excess adipose
tissue deposition with the development of insulin resistance
(Fig. 1).
Overall, the major advantage of the DIO model lies in the fact
that these rats share many of the same characteristics as humans
with obesity-related diabetes (“diabesity”). These therefore provide
a model for measuring the cellular events through which excess
accumulation of body fat and/or excessive dietary fat and/or sugar
intake result in the degradation of insulin action in key insulin-
sensitive tissues. Another advantage of this model is that DIO mice
and rats can now be ordered directly from commercial suppliers.
While this is obviously more expensive than an in-house option, it
does have the potential to save the researcher the 8–12 weeks that it
would normally take to generate the DIO rodents. In addition, a
variety of high-fat diets are now available from commercial animal
feed manufacturers. Most of these suppliers are able to adapt these
diets according to the particular requirements of the study, which
provides researchers with added flexibility and the potential to focus
studies on one or more specific dietary component.
While the DIO model has many positive attributes, it also has
some potential limitations. Perhaps, the major limitation is the lack
of standardization of the feeding regimen of different studies,
which has meant that the phenotype of the animals varies between
experiments. The high-fat diets which have been used consist of
anywhere from 20% to 60% energy content from fat, and the fat
50 €usler et al.
Diet-induced Obesity
PANCREAS ADIPOSE TISSUE

LIVER
(1) Reduced pancreatic (3) Visceral Adipocyte (2) Hepatic steatosis
enzyme activity hypertrophy Hepatic inflammation
Pancreatic (4) Increased plasma Hepatic insulin

dysfunction NEFA and pro-inflammatory resistance
cytokines
(7) Reduced Insulin (6) Increased glucose

secretion SKELETAL MUSCLE output
(5) Increased inflammation

Increased lipid content
Skeletal muscle insulin

resistance
6) Reduced glucose
uptake
Insulin Resistance
Hyperglycemia
Type 2 Diabetes
Fig. 1 Summary of proposed mechanisms involved in the development of type

2 diabetes in models of diet-induced obesity in adult rodents. Increased dietary
fat intake is associated with (1) reduced activity of pancreatic enzymes, leading
to impaired pancreatic function and reduced insulin secretion, and (2) increased
accumulation of lipids in the liver (hepatic steatosis) which results in hepatic
insulin resistance and increased hepatic glucose output. In addition, increased
accumulation of adipose tissue in the visceral compartment results in
(3) increased production and secretion of pro-inflammatory cytokines resulting
in chronic low-grade inflammation and (4) increased plasma concentrations of
nonesterified free fatty acids (NEFAs) (4) which are deposited in liver and skeletal
muscle. Both the chronic low-grade inflammation and ectopic fat deposition
further contribute to insulin resistance in peripheral tissues. Together, this
results in an impaired glucose uptake by skeletal muscle (5), increased hepatic
glucose output (6) and impaired insulin secretion (7) and precipitates the
development of whole-body insulin resistance and type 2 diabetes
source (i.e., animal vs. plant), fatty acid composition (i.e., the
content of saturated, monounsaturated, and polyunsaturated
fats), the balance of n 6 and n 3 PUFA and the level of trans
fats also varies between studies [1, 14]. Similarly, high-sucrose
Models of T2DM 51
and/or high-fructose diets vary in their sugar content and the form
in which the sugar is delivered, as well as the background fat and
protein content of the diet. While there is no clear indication of
which type of nutritional treatment represents the best model of the
metabolic disturbances seen in human obesity, two publications
focused on various types of diets employed in these experiments
concluded that a “cafeteria diet” approach, in which rats/mice were
provided with a selection of high-fat/high-sugar “junk foods,”
produced the metabolic phenotype that was most similar to that
seen in human diet-induced obesity [1, 50]. However, it is also
important to note that the composition of “cafeteria diets” is highly
variable between laboratories and their effects variable between
breeds. Therefore, it is important to critically evaluate the extent
to which the diet used in nutritional models of human T2DM
reflects the typical “poor quality western style diet,” including the
proportion of energy derived from fat/sugar and fat composition,
and the precise phenotype of the strain under investigation when
interpreting results and extrapolating these to humans.
2.2 Diet-Induced The sand rat (Psammomys obesus) has been another popular model
Obesity: Psammomys for studying the degradation of metabolic function associated with
Obesus obesity. This is an attractive model because the sand rat rapidly
develops obesity when fed a standard laboratory chow in standard
animal housing conditions, which is considerably less expensive
than custom-made high-fat diets. Many of the pathophysiological
changes found in the obese sand rat are similar to those seen in
human type 2 diabetic patients [51], and this model therefore
appears to provide an appropriate nutritional model of human
T2DM. When provided with ad libitum access to laboratory
chow, the adult sand rat progress through from normoglycemia
and normoinsulinemia, to hyperinsulinemia with marked insulin
resistance, followed by pronounced hyperinsulinemia with hyper-
glycemia, and after a further 6–10 weeks of chow feeding, gradual
beta-cell degradation and disappearance of beta-cell insulin which
eventually results in severe insulin deficiency and overt diabetes
[52]. The obese sand rat exhibits profound insulin resistance in
both skeletal muscle and liver, and provides an attractive model for
studying the mechanisms which underlie the development of dia-
betes in human obesity [53]. Studies utilizing this model have
reported that GLUT4 protein content is reduced in skeletal muscle,
resulting in a reduced peripheral glucose uptake and hyperglycemia
[52]. In addition, hepatic PEPCK activity is increased, suggesting
that the ability of insulin to inhibit hepatic glucose production is
impaired [54]. Extensive use has been made of the obese sand rats
for the purpose of testing potential T2DM drugs, including tyro-
sine phosphatase inhibitors and glucagon like peptide-1 (GLP-1)
analogues [52].
52 €usler et al.
The major limitations of this model are that the sand rat exhi-
bits insulin resistance even when fed on a low energy diet, suggest-
ing that the physiology of this desert-adapted animal may not be
entirely comparable with humans. The link between weight gain
and the development of insulin resistance in this model has yet to be
clearly defined; however, the phenotype is normalized by dietary
restriction. Nevertheless, this model has been utilized extensively
for the study of obesity-induced diabetes, and continues to provide
important insights into the cellular defects, which contribute to the
development of central and peripheral insulin resistance.
3 Prenatal Nutritional Models of T2DM
The fetal or developmental origins of adult disease hypothesis states

that insults during critical windows of development result in adap-
tive changes within fetal tissues and organ systems, which have
lifetime consequences for the health of an individual. The hypothe-
sis was first derived from studies of the Hertfordshire birth cohort
by Professor David Barker in the early 1990s which showed that
there was an inverse relationship between birth weight and the
incidence of cardiovascular disease in adult life [55]. This was
followed by a series of studies in human populations which all
demonstrated that being of low birth weight was associated with
an increased incidence of adult metabolic and cardiovascular disease
[56, 57].
The Dutch Winter Hunger Famine was a 5-month period in
World War II during which the food supply to Amsterdam, Holland
was severely restricted, resulting in a substantial decrease in the
daily energy intake of the population [58]. Subsequent studies of
the children of women who were pregnant during the famine
showed that those exposed to famine in the last 5 months of
gestation had a significantly greater incidence of glucose intoler-
ance, obesity and T2DM in adult life compared to those children
whose mothers were not exposed to the famine [58, 59]. Similarly,
early studies in infants of diabetic mothers clearly showed that
exposure to high glucose levels in the pre- and perinatal period
was associated with an increased incidence of hyperglycemia and
T2DM in the offspring in postnatal life [60]. These studies
provided the first evidence that exposure to an inappropriately
low, or inappropriately high nutrient supply during early develop-
ment was associated with a permanent alteration of metabolic
function in the offspring. An increasing number of epidemiological
and experimental animal studies have since continued to highlight
the importance of both the prenatal and early postnatal nutritional
environment for the determination of later metabolic health
[57, 61, 62].
Models of T2DM 53
Together, these studies have clearly demonstrated that a low

birth weight followed by a period of accelerated postnatal growth,
or a high birth weight as a consequence of prenatal overnutrition,
are each associated with an increased propensity toward the devel-
opment of insulin resistance, glucose intolerance, and T2DM in
adult life [63–65]. As a result, both spontaneous and experimen-
tally induced fetal growth restriction (in utero growth restriction
(IUGR)) and prenatal overnutrition in animal models have been
widely employed in order to understand the physiological basis of
reduced insulin sensitivity. It has also been demonstrated that
global undernutrition in the pregnant rodent is associated with
later onset of T2DM in the offspring [66]. This is, however, con-
sidered to be largely a consequence of impaired insulin secretion
rather than defects in insulin signaling [67], and is therefore not
regarded as an appropriate model for the pathophysiological pro-
cess which contributes to T2DM in the majority of human patients.
3.1 The Maternal The maternal low-protein model is one of the most extensively
Low-Protein Model studied models of fetal growth restriction. In this model, pregnant
dams are fed a diet containing approximately 8% of energy as
protein, compared to approximately 20% in controls, and this is
associated with low birth weight in the offspring, followed by
accelerated postnatal growth [68, 69]. The period of rapid postna-
tal growth is due to enhanced insulin sensitivity in the period
immediately after birth; however, this is not maintained beyond
the early postnatal period, and the offspring develop insulin resis-
tance by the age of 15 months, frank T2DM by 17 months
[65, 70], and the deterioration of metabolic function is accelerated
by a postnatal high-fat diet [71]. The phenotype of the low-protein
offspring has many similarities to that of human type 2 diabetics,
including insulin resistance, altered regulation of hepatic glucose
output and pancreatic dysfunction [68]. As a result, this model has
been used extensively to explore and characterize the cellular
defects within the insulin signaling pathway which may contribute
to reduced insulin sensitivity. These studies have demonstrated that
prenatal exposure to a low-protein diet results in defects in several
peripheral insulin-sensitive tissues, which are central to the mainte-
nance of glucose homeostasis. In the pancreas, the islets of Langer-
hans are smaller and exhibit a reduced insulin secretion in response
to amino acid stimulation [72]. While there is no obvious defect in
glucose-stimulated insulin release in offspring fed a control chow
diet postweaning, a reduction in the capacity of glucose to stimulate
insulin release emerges if these offspring are fed on a high-fat diet
postweaning. The liver of the low-protein offspring is unresponsive
to the action of glucagon, and insulin stimulates, rather than sup-
presses, hepatic glucose output [73]. Furthermore, adipose cells
exhibit increased basal and insulin-stimulated glucose uptake, lead-
ing to an increased accumulation of fat in the visceral compartment
54 €usler et al.
Maternal Low protein diet
ADIPOCYTE SKELETAL MUSCLE LIVER PANCREAS
4) Beta cell
1) Increased 2) Impaired insulin 3) Hepatic insulin dysfunction
Glucose uptake signalling resistance
5) Increased 6) Reduced 7) Increased 8) Reduced Insulin

plasma NEFA Glucose uptake Glucose output secretion
Hyperglycemia
Insulin resistance
Type 2 Diabetes
Fig. 2 Summary of mechanisms proposed to contribute to the programming of type 2 diabetes of offspring in
the maternal low-protein model. Prenatal exposure to a low-protein diet results in (1) an increased capacity for
glucose uptake by visceral adipocytes (2) impaired insulin signaling in skeletal muscle, (3) hepatic insulin
resistance and (4) impaired pancreatic function. The increased glucose uptake by visceral adipocytes results
in an increased accumulation of visceral adipose tissue, resulting in elevated plasma concentrations of
nonesterified free fatty acids (NEFAs) (5) which are deposited in peripheral tissues (liver and skeletal muscle),
further reducing insulin sensitivity. This reduced insulin sensitivity is associated with impaired glucose uptake
by skeletal muscle (6), increased glucose output by the liver (7) and reduced insulin secretion by the pancreas
(8), resulting in peripheral hyperglycemia, insulin resistance and type 2 diabetes
and ectopic fat storage in liver and skeletal muscle, which further
contributes to the peripheral insulin resistance [74] (Fig. 2).
Given that muscle represents the major site of postprandial
glucose disposal, it is not surprising that changes in the functional
characteristics of muscle fibers during the perinatal period are
important in the programming of insulin resistance and diabetes
(Fig. 2). Isolated muscle strips from these low-protein animals
exhibit enhanced basal and insulin-stimulated glucose uptake, and
this increase in insulin sensitivity is associated with a twofold
Models of T2DM 55
increase in the abundance of insulin receptors in muscle membranes

[75]. By 15 months of age, however, there is a decrease in the
insulin sensitivity of glucose uptake in skeletal muscle from the
group exposed to the low-protein diet in utero [76]. This impaired
insulin action is not associated with changes in the expression of
either the insulin receptor or GLUT4, but is associated with a
decrease in the abundance of signaling molecules downstream of
the insulin receptor, including the zeta-isoform of protein kinase C,
an isoform that is positively involved in GLUT4-mediated glucose
transport [76] and p85 PI3K [69] (Fig. 2). In addition to defects in
insulin signaling, it has also been reported that prenatal undernu-
trition is associated with impaired mitochondrial biogenesis and
impaired mitochondrial oxidative capacity in skeletal muscle,
which has previously been associated with reductions in insulin
sensitivity in this tissue [77, 78].
While these studies have provided important insights into the
cellular defects which contribute to deteriorations in insulin sensi-
tivity in peripheral tissues, it is still unclear whether these defects are
the same as those seen in human T2DM, which typically develops
secondary to increased body fat accumulation. There is recent
evidence that the mechanisms involved in the prenatal program-
ming of insulin resistance may be quite distinct from those in
insulin resistance which develops in response to diet-induced obe-
sity [67]. Nevertheless, the obvious importance of the early envi-
ronment in determining metabolic health in later life may well
explain why it is that some individuals are more susceptible to
insulin resistance than others at the same degree of body fatness
and same nutritional/environmental conditions.
3.2 Prenatal There is increasing evidence that prenatal exposure to a high plane
Overnutrition, of nutrition is also associated with an increased risk of obesity and
Postnatal Obesity, T2DM in postnatal life [61]. The rat and sheep represent the two
and T2DM main animal models that have been utilized thus far for investigat-
ing the effects of prenatal overnutrition on the offspring, and there
is increasing evidence pointing to the negative impact of intrauter-
ine exposure to maternal obesity and/or high-fat/high-sugar diets
on insulin signaling in the offspring [61, 79]. Studies to date have
suggested that the deterioration of metabolic function in these
individuals is associated with altered development of pancreatic
beta cells, mitochondrial dysfunction and programming of the
central appetite-regulating network [11, 80–82]. Several studies
have also suggested that the deterioration of metabolic function
may be secondary to the increased accumulation of adipose tissue in
these offspring (Fig. 3) [82, 83]; however, it is not clear whether
this is the case in all models of prenatal overnutrition.
56 €usler et al.
Prenatal
Overnutrition
(1) ADIPOSE TISSUE

PANCREAS ALL METABOLIC
PPARγ TISSUES
(7) Epigenetic
LEPTIN
Programming
Birth
Obesity
(2) (3) β-cell dysfunction
Inflammation
(4) Increased plasma NEFA Reduced insulin

and pro-inflammatory secretion
cytokines
LIVER SKELETAL
MUSCLE
(5) Increased (6) Reduced
glucose output glucose uptake
Insulin resistance
Hyperglycemia
Type 2 Diabetes
Fig. 3 Potential mechanisms contributing to the programming of type 2 diabetes following prenatal overnutri-
tion. It has been demonstrated that prenatal overnutrition is associated with an increased lipogenic capacity of
adipose cells (1), which results in increased fat accumulation after birth (2) and predisposes these offspring to
obesity in later life. In addition, prenatal high-fat diets result in impaired pancreatic development, and beta-
cell dysfunction in postnatal life (3) and increased inflammation in key metabolic tissues, both of which
contribute to increased insulin resistance. The increased accumulation of body fat further contributes to
metabolic dysfunction by elevating plasma nonesterified free fatty acid (NEFA) concentrations (4) which are
Models of T2DM 57
In the sheep, feeding ewes approximately 55% above their

maintenance energy requirements in later pregnancy results in an
increase in fetal glucose and insulin concentrations in the last third
of gestation [84], and is associated with an increased accumulation
of subcutaneous adipose tissue at the end of the first month of
postnatal life [82]. The offspring of overfed ewes exhibit elevated
glucose concentrations during the first month after birth [82],
although whether this is associated with reduced insulin sensitivity
during this period has yet to be determined.
In the rat, maternal high-fat feeding results in disturbed glu-
cose homeostasis in the offspring at weaning and in adult life,
including an impaired glucose tolerance and reduced whole-body
insulin sensitivity [81, 85]. As with adult rodents, however, the type
of fat in the maternal diet also appears to be an important determi-
nant of the effects of offspring physiology/metabolism. Patel and
colleagues found that feeding rats on a chow containing high levels
of saturated fat during pregnancy and lactation resulted in offspring
that were significantly heavier than controls, and had elevated
plasma concentrations of insulin, glucose, free fatty acids and tri-
glycerides and exhibited glucose intolerance [80]. When they inves-
tigated the pancreatic islets in these animals, they found that those
from male offspring exhibited an increased insulin secretory
response at low glucose concentrations compared to controls, sug-
gesting that high-fat feeding resulted in altered pancreatic develop-
ment [86]. This was also the case in a separate study, in which
feeding pregnant rats a high-saturated fat diet during pregnancy
and lactation was associated with hyperglycemia and evidence of
compromised beta-cell function in the weanling offspring [87].
Similarly, the offspring of Sprague-Dawley rats fed a lard-rich diet
during pregnancy and lactation exhibited whole-body insulin resis-
tance and a reduced glucose-stimulated insulin secretion in isolated
islets at 9 months [11], again supporting the thesis that maternal
high-fat feeding is associated with impaired pancreatic function in
the offspring. Maternal consumption of diets high in saturated fat
has also been linked to increased hepatic triglyceride accumulation
in the offspring both in nonhuman primates and rodents, with a
consequent reduction in hepatic insulin sensitivity [88].
3.3 Omega-6 The pro-adipogenic and pro-lipogenic actions of omega-6 PUFA,

and Omega-3 PUFA as discussed above, coupled with the substantial increase in the level
and Metabolic of these fatty acids in the typical diets of many Western countries
Programming over the past 3–4 decades, have led to the suggestion that these
Fig. 3 (continued) deposited in skeletal muscle and liver, resulting in insulin resistance in these tissues
(5) and (6). Epigenetic changes which occur before birth (7) may also play a role in metabolic programming
after prenatal overnutrition. Together, these changes result in peripheral insulin resistance and, ultimately,
type 2 diabetes
58 €usler et al.
fatty acids in the maternal diet could program an increased propen-

sity to obesity and T2DM in the offspring [89, 90]. In support of
this, a recent study by Massiera and colleagues demonstrated that
intakes of omega-6 PUFA at levels comparable to those in modern
western diets, in the absence of an increase in saturated fat intake,
was sufficient to program a transgenerational increase in adipose
tissue deposition in C57BL6/J mice [91]. Further studies are
required, however, to establish whether maternal diets high in
omega-6 fats could represent a suitable model to study the inter-
generational programming of insulin resistance/T2DM in humans.
Despite the established role of n 3 LCPUFA in suppressing
fat storage and promoting insulin resistance in vitro and in adults,
the data from studies in which maternal n 3 LCPUFA intake is
increased have failed to provide robust evidence that this is asso-
ciated with an improved metabolic phenotype in the offspring
[92, 93]. Indeed, in one study, offspring of dams who were
provided with a DHA-supplemented diet during pregnancy and
lactation actually had a higher relative fat mass in adolescence
[94]. The potential for maternal n 3 LCPUFA supplementation
to produce improvements in insulin sensitivity of the offspring has
yet to be studied directly in either humans or animal models, and
this remains an important area for future research.
3.4 Maternal High- The adverse metabolic effects associated with excess consumption
Sucrose/High- of additional sweeteners in adults have led to a growing interest in
Fructose Diets the potential role of these substances in the maternal diet in the
and Metabolic programming of insulin resistance and T2DM. In epidemiological
Programming studies, excess consumption of sugar-sweetened beverages prior to
pregnancy has been associated with an increased risk of developing
gestational diabetes [95], which is known to be a risk factor for
obesity and T2DM in the offspring [79]. The consumption of these
sugar-sweetened beverage has also been linked to an increased birth
weight [96]. Thus far, however, there have been relatively few
experimental studies which have attempted to address the effects
of sucrose/fructose in the maternal diet at levels typically encoun-
tered in Western diets on the long-term metabolic health of the
offspring.
There is evidence from rodent studies, however, that maternal
consumption of high levels of fructose or fructose containing
sugars during pregnancy is associated with metabolic dysfunction
in the mother, including increased adiposity, weight gain and
hepatic insulin resistance, and adverse effects on placental and
fetal development [97–103]. As discussed above, the metabolic
effects of fructose in combination with glucose are different from
fructose alone; however, few studies have investigated the separate
effects of sucrose or fructose during pregnancy on maternal and
Models of T2DM 59
offspring outcomes [104]. Further studies in this area are clearly

warranted, given that the evidence from the studies to date suggests
that elevated intakes of both sucrose and fructose are associated
with negative metabolic outcomes in the offspring. Therefore,
maternal high-sucrose and/or high-fructose diets may offer an
alternative experimental animal model of T2DM to maternal
high-fat diets.
4 Possible Mechanisms for Metabolic Programming
4.1 The Adipocyte In large animal models, there is evidence that exposure to prenatal
overnutrition or undernutrition can act to permanently alter the
function of adipocytes, and result in an increased lipogenic capacity
in adipose depots in postnatal life. In sheep, which have a similar
profile of adipose cell development to humans, exposure to
increased glucose concentrations in late fetal life increases the
expression of genes within adipose cells that are responsible for
promoting lipid storage and forming new adipocytes [105]. This
is associated with increased adipose tissue mass by the end of the
first month of life, due primarily to an increase in adipocyte cell size
[82]. In pigs, adipocytes exposed to high glucose levels before birth
also exhibit a dramatic increase in their capacity for lipogenesis
[106] and this precedes the development of obesity in the piglets
[107]. These findings have since been supported by work in
rodents, in which maternal junk food feeding during pregnancy
and lactation in rats and an obesogenic diet during this same period
in mice were each associated with increased adiposity and increased
expression of lipogenic genes and insulin-independent glucose
transporters in the perirenal adipose depot [83, 108]. It would
therefore appear that fat cells exposed to an excess substrate supply
during critical windows in their development have an increased
capacity for storing lipid in postnatal life. This enhanced lipogenic
capacity would render these individuals more likely to store excess
energy in the form of fat, and increase their susceptibility to weight
gain, obesity and, consequently, to the excess deposition of fatty
acids in liver and skeletal muscle, resulting in peripheral insulin
resistance, hyperglycemia, and ultimately, T2DM [109] (Fig. 3).
In individuals exposed to low nutrition levels before birth,
adipocyte development is initially sacrificed in favor of “essential”
organs [57, 110]. If an in utero “restricted” individual is born into
a postnatal environment where nutrient supply is no longer con-
strained, a period of “catch-up” fat deposition ensues, mainly in the
visceral adipose depot [111]. These individuals are thus at increased
risk of visceral obesity [112], and consequently, to the development
of insulin resistance and T2DM [113–115]. It is the increased
60 €usler et al.
accumulation of visceral adipose tissue that has also been demon-

strated in animal models of IUGR, including the sheep, guinea pig,
and rodent [115–117]. Therefore, exposure of the developing
adipocyte to suboptimal nutrition, particularly of the visceral adi-
pose tissue which is the first adipose depot to develop in sheep and
humans [118, 119], also appears to play a critical role in defining an
individual’s propensity for accumulating visceral body fat later in
life, that is, a central pattern of fat accumulation, which is the fat
pattern linked to increased risk of metabolic dysfunction and
T2DM [6].
4.2 Mitochondrial Mitochondria play a central role in the regulation of cellular energy
Biogenesis metabolism, and impaired mitochondrial function and reduction in
mitochondrial oxidative capacity in skeletal muscle have been asso-
ciated with the onset of insulin resistance and T2DM both in
experimental animal models and in human subjects [120]. In exper-
imental animal studies, both diet-induced obesity and prenatal
undernutrition have been associated with impaired mitochondrial
biogenesis and reduced abundance of oxidative enzymes in skeletal
muscle. Importantly, there is evidence that the reduction of mito-
chondrial biogenesis in adult rats exposed to prenatal undernutri-
tion precedes the development of insulin resistance and glucose
intolerance in this model [77, 78], therefore, implicating impaired
mitochondrial function in the causal pathway linking prenatal
undernutrition to later metabolic disease. The role of mitochon-
drial dysfunction in metabolic programming has yet to be explored
in large animal models, and the potential role of mitochondrial
dysfunction in the development of T2DM remains an important
area for future investigation.
4.3 Increased The established role of chronic low-grade inflammation in the

Inflammation etiology of metabolic disease and T2DM in adults [49] has led to
the suggestion that the impact of nutritional exposures experienced
during the perinatal period on insulin sensitivity in the offspring
could be mediated via altered inflammatory status in the offspring
[121]. This premise has been supported by a number of experimen-
tal animal studies. These studies have demonstrated that both
maternal undernutrition and maternal high-fat feeding results in a
state of chronic low-grade inflammation in the mother, and that
this is transferred both to the placenta and to a number of organs of
the offspring, including the liver, adipose tissue, cardiovascular
system and brain [122–124]. In one study in mice, offspring of
high-fat fed dams exhibited increased expression of a number of key
inflammatory factors, including TNFα, CD68, and MCP-1, in their
subcutaneous adipose tissue, in conjunction with decreased
GLUT4 mRNA expression, which supports the suggestion that
maternal obesity can affect fetal insulin sensitivity via the alteration
of inflammatory pathways [124]. Similarly, in a sheep model of
maternal obesity, suppression of insulin receptor expression and
Models of T2DM 61
insulin signaling occurred in conjunction with an upregulation of

inflammation in skeletal muscle in the offspring [123]. While fewer
studies have focused on the potential role of chronic inflammation
in the programming of T2DM by maternal nutritional deficits,
there is recent evidence implicating exposure to maternal undernu-
trition in altered immune development in the offspring [122].
4.4 Epigenetics The concept that changes in phenotype could be elicited by modi-
fication of the DNA in the absence of changes in DNA sequence,
that is, epigenetic changes, provides a new basis for understanding
of phenotypic programming. The two most well-characterized epi-
genetic modifications include DNA methylation and histone acety-
lation, which both act to suppress gene expression
[125, 126]. Recent evidence has demonstrated that exposure to
an inappropriate nutrient supply during early development may
result in epigenetic modifications in the fetus which have the
potential to permanently alter the function of important metabolic
systems. Waterland and colleagues demonstrated that when mice
with a genetic tendency toward obesity were fed a standard diet
during pregnancy, the degree of obesity in the offspring increased
progressively with each generation. This effect was, however,
completely abolished if mice were fed a diet high in methyl groups
(which increases DNA methylation) during pregnancy, and the
offspring remained lean. These findings suggested that reduced
DNA methylation in genes which play a role in the regulation of
energy balance and fat storage, as a result of early nutritional
programming, may contribute to the later development of obesity
and metabolic disease [127]. The early work of Waterland has since
been supported by a large number of subsequent experimental
animal studies, and the importance of epigenetics in mediating
metabolic programming is now well established. There is also
increasing evidence that exposure to a suboptimal nutrient supply
before birth can also alter methylation of a series of genes involved
in growth and energy homeostasis, and that this may explain the
link between both fetal growth restriction and maternal nutritional
imbalance and later metabolic dysfunction. In rodent studies,
Burdge and Lilycrop have reported altered methylation of a num-
ber of key hepatic genes, including those involved in carbohydrate
metabolism, in offspring exposed to maternal undernutrition in
utero [128]. Importantly, these studies have provided the evidence
of transgenerational programming effects, such that epigenetic
alterations persist into the F2 generation [129]. There is also
emerging evidence of epigenetic programming in human studies,
including a study reporting altered Igf2 methylation in adults
whose mothers were exposed to the Dutch Winter Hunger Famine
during pregnancy [130]. Overall, the role of epigenetics in the
developmental origins of metabolic disease is certainly an exciting
area of research, and has been the subject of a number of excellent
reviews [131–134].
62 €usler et al.
5 Summary
The dramatic increase in the incidence of T2DM over the past

decade has highlighted the need to better understand the patho-
physiology of this disease. The use of nutritional animal models
which mimic the disease progression in humans is essential in order
to investigate the cellular mechanisms which contribute to the
deterioration of insulin metabolism in key insulin-sensitive tissues.
This criterion is not fulfilled by genetic models, or models in which
the pancreas is surgically ablated, since these causes account for only
a relatively minor proportion of the humans who will develop
T2DM. A number of nutritional models of T2DM have now
been developed, and have made a significant contribution to our
current understanding of the cellular defects which emerge in
insulin-sensitive tissues in response to an excess accumulation of
body fat or excessive intake of dietary fat, and which result in the
development of peripheral insulin resistance. It has also become
clear that exposure to an inappropriate nutritional environment in
utero is associated with an increased susceptibility to T2DM in
adult life, and that the detailed study of the cellular and molecular
defects in these offspring has the potential to provide novel insights
into why some individuals are more prone to the development of
insulin resistance and T2DM than others.
The study of diet-induced obesity and models of prenatal
undernutrition and overnutrition has revealed several common
mechanisms which contribute to the onset of insulin resistance
and T2DM in these different models (Fig. 4). This has provided
important insights into the etiology of T2DM in humans, and
identified possible targets for intervention. As with all animal mod-
els, it is important to recognize that animal data may not always be
directly extrapolated to the human situation, where other factors,
in particular, socioeconomic status and demographic variations,
also contribute to the risk of metabolic disease [135]. It is therefore
important to exercise caution when extrapolating results to human
clinical practice. Nevertheless, it is clear that nutritional studies in
the animal models discussed in this chapter have been critical in
shaping our current understanding of the pathophysiology of
T2DM and offer significant opportunities for identifying and test-
ing potential clinical interventions.
Acknowledgments
BSM is supported by a Career Development Fellowship from the

National Health and Medical Research Council of Australia
(NHMRC).
Models of T2DM 63
Diet Induced Obesity

Maternal low protein diet
Prenatal Overnutrition
PANCREAS
7) Epigenetic
changes
8) Mitochondrial
dysfunction
3) Adipocyte
hypertrophy
1) Pancreatic 4) Increased plasma

dysfunction NEFA and pro-
inflammatory
cytokines
LIVER
2) Reduced Insulin
secretion SKELETAL MUSCLE
6) Increased glucose
output
5) Reduced glucose
uptake
Hyperglycemia
Type 2 Diabetes
Fig. 4 Summary of common mechanisms identified in all nutritional models of type 2 diabetes. Both diet-
induced obesity and prenatal nutritional interventions are associated with pancreatic dysfunction (1) and
impaired insulin secretion (2), which contributes to peripheral hyperglycemia. In addition, adipocyte hypertro-
phy, in particular within the visceral compartment (3), is associated with increased circulating concentrations
of nonesterified free fatty acids (NEFA) (4) and increased production and release of pro-inflammatory cytokines
(5). The infiltration of NEFA and pro-inflammatory cytokines liver and skeletal muscle results in development of
insulin resistance in these tissues, and consequently, reduced glucose uptake by skeletal muscle (5) and
increased hepatic glucose output (6). In addition, epigenetic changes and mitochondrial dysfunction in liver
and skeletal muscle may also contribute to development of metabolic disorders
64 €usler et al.
References
1. Buettner R, Scholmerich J, Bollheimer LC 12. Ong ZY, Wanasuria AF, Lin MZP, Hiscock J,
(2007) High-fat diets: modeling the meta- Muhlhausler BS (2013) Chronic intake of a
bolic disorders of human obesity in rodents. cafeteria diet and subsequent abstinence.
Obesity 15:798–808 Sex-specific effects on gene expression in the
2. Portha B, Blondel O, Serradas P, McEvoy R, mesolimbic reward system. Appetite
Giroix MH, Kergoat M, Bailbe D (1989) The 65:189–199
rat models of non-insulin dependent diabetes 13. Reichelt AC, Morris MJ, Westbrook RF
induced by neonatal streptozotocin. Diabete (2014) Cafeteria diet impairs expression of
Metab 15:61–75 sensory-specific satiety and stimulus-outcome
3. Bray GA (2004) Medical consequences of learning. Front Psychol 5:852
obesity. J Clin Endocrinol Metab 14. Hariri N, Thibault L (2010) High-fat diet-
89:2583–2589 induced obesity in animal models. Nutr Res
4. Goralski KB, Sinal CJ (2007) Type 2 diabetes Rev 23:270–299
and cardiovascular disease: getting to the fat 15. Flanagan AM, Brown JL, Santiago CA, Aad
of the matter. Can J Physiol Pharmacol PY, Spicer LJ, Spicer MT (2008) High-fat
85:113–132 diets promote insulin resistance through cyto-
5. Smith SR, Lovejoy JC, Greenway F, Ryan D, kine gene expression in growing female rats. J
deJonge L, de la Bretonne J, Volafova J, Bray Nutr Biochem 19:505–513
GA (2001) Contributions of total body fat, 16. Barbosa-da-Silva S, da Silva NC, Aguila MB,
abdominal subcutaneous adipose tissue com- Mandarim-de-Lacerda CA (2013) Liver dam-
partments, and visceral adipose tissue to the age is not reversed during the lean period in
metabolic complications of obesity. Metabo- diet-induced weight cycling in mice. Hepatol
lism 50:425–435 Res 18:12138
6. Ravussin E, Smith SR (2002) Increased fat 17. Flachs P, Rossmeisl M, Bryhn M, Kopecky J
uptake, impaired fat oxidation, and failure of (2009) Cellular and molecular effects of n-3
fat cell proliferation result in ectopic fat stor- polyunsaturated fatty acids on adipose tissue
age, insulin resistance, and type 2 diabetes. biology and metabolism. Clin Sci (Lond)
Ann N Y Acad Sci 967:363–378 116:1–16
7. Surwit R, Kuhn C, Cochrane C, McCubbin J, 18. Muhlhausler BS, Cook-Johnson R, James M,
Feinglos M (1988) Diet-induced type II dia- Miljkovic D, Duthoit E, Gibson R (2010)
betes in C57BL/6J mice. Diabetes Opposing effects of omega-3 and omega-6
37:1163–1167 long chain polyunsaturated fatty acids on the
8. Black BL, Croom J, Eisen EJ, Petro AE, expression of lipogenic genes in omental and
Edwards CL, Surwit RS (1998) Differential retroperitoneal adipose depots in the rat. J
effects of fat and sucrose on body composi- Nutr Metab
tion in A/J and C57BL/6 mice. Metabolism 19. Ruzickova J, Rossmeisl M, Prazak T, Flachs P,
47:1354–1359 Sponarova J, Veck M, Tvrzicka E, Bryhn M,
9. Roberts CK, Berger JJ, Barnard RJ (2002) Kopecky J (2004) Omega-3 PUFA of marine
Long-term effects of diet on leptin, energy origin limit diet-induced obesity in mice by
intake, and activity in a model of diet-induced reducing cellularity of adipose tissue. Lipids
obesity. J Appl Physiol 93:887–893 39:1177–1185
10. Bayol SA, Farrington SJ, Stickland NC (2007) 20. Massiera F, Guesnet P, Ailhaud G (2006) The
A maternal ‘junk food’ diet in pregnancy and crucial role of dietary n-6 polyunsaturated
lactation promotes an exacerbated taste for fatty acids in excessive adipose tissue develop-
‘junk food’ and a greater propensity for obe- ment: relationship to childhood obesity. Nes-
sity in rat offspring. Br J Nutr 98:843–851 tle Nutr Workshop Ser Pediatr Program
11. Taylor PD, McConnell J, Khan IY, 57:235–242
Holemans K, Lawrence KM, Asare-Anane H, 21. Massiera F, Saint-Marc P, Seydoux J,
Persaud SJ, Jones PM, Petrie L, Hanson MA, Murata T, Kobayashi T, Narumiya S,
Poston L (2005) Impaired glucose homeosta- Guesnet P, Amri E-Z, Negrel R, Ailhaud G
sis and mitochondrial abnormalities in off- (2003) Arachidonic acid and prostacyclin sig-
spring of rats fed a fat-rich diet in pregnancy. naling promote adipose tissue development: a
Am J Physiol Regul Integr Comp Physiol 288: human health concern? J Lipid Res
R134–R139 44:271–279
Models of T2DM 65
22. Semple RK, Chatterjee VK, O’Rahilly S 34. Tran LT, Yuen VG, McNeill JH (2009) The
(2006) PPAR gamma and human metabolic fructose-fed rat: a review on the mechanisms of
disease. J Clin Invest 116:581–589 fructose-induced insulin resistance and hyper-
23. De Caterina R, Basta G (2001) n-3 Fatty acids tension. Mol Cell Biochem 332:145–159
and the inflammatory response — biological 35. Pagliassotti MJ, Prach PA, Koppenhafer TA,
background. Eur Heart J Suppl 3:D42–D49 Pan DA (1996) Changes in insulin action,
24. Chandalia M, Abate N (2007) Metabolic triglycerides, and lipid composition during
complications of obesity: inflated or inflamed? sucrose feeding in rats. Am J Physiol 271:
J Diabetes Complications 21:128–136 R1319–R1326
25. Malik VS, Popkin BM, Bray GA, Despres JP, 36. Sheludiakova A, Rooney K, Boakes RA
Hu FB (2010) Sugar-sweetened beverages, (2012) Metabolic and behavioural effects of
obesity, type 2 diabetes mellitus, and cardio- sucrose and fructose/glucose drinks in the
vascular disease risk. Circulation rat. Eur J Nutr 51:445–454
121:1356–1364 37. Chicco A, D’Alessandro ME, Karabatas L,
26. Romaguera D, Norat T, Wark PA, Vergnaud Pastorale C, Basabe JC, Lombardo YB
AC, Schulze MB, van Woudenbergh GJ, (2003) Muscle lipid metabolism and insulin
Drogan D, Amiano P, Molina-Montes E, San- secretion are altered in insulin-resistant rats
chez MJ, Balkau B, Barricarte A, Beulens JW, fed a high sucrose diet. J Nutr 133:127–133
Clavel-Chapelon F, Crispim SP, Fagherazzi G, 38. Oliveira LS, Santos DA, Barbosa-da-Silva S,
Franks PW, Grote VA, Huybrechts I, Key TJ, Mandarim-de-Lacerda CA, Aguila MB
Khaw KT, Nilsson P, Overvad K, Palli D, (2014) The inflammatory profile and liver
Panico S, Quiros JR, Rolandsson O, damage of a sucrose-rich diet in mice. J Nutr
Sacerdote C, Sieri S, Slimani N, Spijkerman Biochem 25:193–200
AM, Tjonneland A, Tormo MJ, Tumino R, 39. Catena C, Giacchetti G, Novello M,
van den Berg SW, Wermeling PR, Zamara- Colussi G, Cavarape A, Sechi L (2003) Cellu-
Ros R, Feskens EJ, Langenberg C, Sharp SJ, lar mechanisms of insulin resistance in rats
Foroughi NG, Riboli E, Wareham NJ (2013) with fructose-induced hypertension. Am J
Consumption of sweet beverages and type Hypertens 16:973–978
2 diabetes incidence in European adults: 40. Kazumi T, Odaka H, Hozumi T, Ishida Y,
results from EPIC-InterAct. Diabetologia Amano N, Yoshino G (1997) Effects of die-
56:1520–1530 tary fructose or glucose on triglyceride pro-
27. Johnson RK, Appel LJ, Brands M, Howard duction and lipogenic enzyme activities in the
BV, Lefevre M, Lustig RH, Sacks F, Steffen liver of Wistar fatty rats, an animal model of
LM, Wylie-Rosett J (2009) Dietary sugars NIDDM. Endocr J 44:239–245
intake and cardiovascular health: a scientific 41. Levin BE, Hogan S, Sullivan AC (1989) Initi-
statement from the American Heart Associa- ation and perpetuation of obesity and obesity
tion. Circulation 120:1011–1020 resistance in rats. Am J Physiol Regul Integr
28. Te Morenga L, Mallard S, Mann J (2012) Comp Physiol 256:R766–R771
Dietary sugars and body weight: systematic 42. Clegg DJ, Benoit SC, Reed JA, Woods SC,
review and meta-analyses of randomised con- Dunn-Meynell A, Levin BE (2005) Reduced
trolled trials and cohort studies. BMJ 346: anorexic effects of insulin in obesity-prone
e7492 rats fed a moderate-fat diet. Am J Physiol
29. Willett WC, Ludwig DS (2013) Science sour- Regul Integr Comp Physiol 288:R981–R986
ing on sugar. BMJ 346:e8077 43. Levin BE, Dunn-Meynell AA, Balkan B, Kee-
30. McLennan W, Podger A (1998) National sey RE (1997) Selective breeding for diet-
nutrition survey nutrient intakes and physical induced obesity and resistance in Sprague-
measurements 1995. Canberra, ACT, Dawley rats. Am J Physiol Regul Integr
Australian Bureau of Statistics Comp Physiol 273:R725–R730
31. Tappy L, Le KA (2010) Metabolic effects of 44. Tkacs NC, Levin BE (2004) Obesity-prone
fructose and the worldwide increase in obe- rats have preexisting defects in their counter
sity. Physiol Rev 90:23–46 regulatory response to insulin-induced hypo-
32. Douard V, Ferraris RP (2013) The role of glycemia. Am J Physiol Regul Integr Comp
fructose transporters in diseases linked to Physiol 287:R1110–R1115
excessive fructose intake. J Physiol 45. Woods SC, Seeley RJ, Rushing PA,
591:401–414 D’Alessio D, Tso P (2003) A controlled
33. Lustig RH (2013) Fructose: it’s “alcohol high-fat diet induces an obese syndrome in
without the buzz”. Adv Nutr 4:226–235 rats. J Nutr 133:1081–1087
66 €usler et al.
46. Qiu L, List EO, Kopchick JJ (2005) Differen- Bleker OP (1998) Glucose tolerance in adults
tially expressed proteins in the pancreas of after prenatal exposure to famine. Lancet
diet-induced diabetic mice. Mol Cell Proteo- 351:173–177
mics 4:1311–1318 60. Pettit DJ, Knowler WC (1998) Long-term
47. Corbett SW, Stern JS, Keesey RE (1986) effects of the intrauterine environment, birth
Energy expenditure in rats with diet-induced weight, and breast-feeding in Pima Indians.
obesity. Am J Clin Nutr 44:173–180 Diabetes Care 21:B138–B141
48. Yaqoob P, Sherrington EJ, Jeffery NM, 61. Rkhzay-Jaf J, O’Dowd JF, Stocker CJ (2012)
Sanderson P, Harvey DJ, Newsholme EA, Maternal obesity and the fetal origins of the
Calder PC (1995) Comparison of the effects metabolic syndrome. Curr Cardiovasc Risk
of a range of dietary lipids upon serum and Rep 6:487–495
tissue lipid composition in the rat. Int J Bio- 62. Muhlhausler BS, Ong ZY (2011) The fetal
chem Cell Biol 27:297–310 origins of obesity: early origins of altered
49. Dandona P, Aljada A, Bandyopadhyay A food intake. Endocr Metab Immune Disord
(2004) Inflammation: the link between insu- Drug Targets 11:189–197
lin resistance, obesity and diabetes. Trends 63. Bavdekar A, Yajnik C, Fall C, Bapat S,
Immunol 25:4–7 Pandit A, Deshpande V, Bhave S,
50. Higa TS, Spinola AV, Fonseca-Alaniz MH, Kellingray S, Joglekar C (1999) Insulin resis-
Evangelista FSA (2014) Comparison between tance syndrome in 8-year-old Indian children:
cafeteria and high-fat diets in the induction of small at birth, big at 8 years, or both? Diabetes
metabolic dysfunction in mice. Int J Physiol 48:2422–2429
Pathophysiol Pharmacol 6:47–54 64. McMillen IC, Adam CL, Muhlhausler BS
51. Wassink AM, Olijhoek JK, Visseren FL (2005) Early origins of obesity: programming
(2007) The metabolic syndrome: metabolic the appetite regulatory system. J Physiol
changes with vascular consequences. Eur J 565:9–17
Clin Invest 37:8–17 65. Hales CN, Barker DJP (2001) The thrifty
52. Shafrir E, Ziv E, Kalman R (2006) Nutrition- phenotype hypothesis: type 2 diabetes. Br
ally induced diabetes in desert rodents as Med Bull 60:5–20
models of type 2 diabetes: acomys cahirinus 66. Holemans K, Verhaeghe J, Dequeker J, Van
(spiny mice) and Psammomys obesus (desert Assche FA (1996) Insulin sensitivity in adult
gerbil). ILAR J 47:212–224 female rats subjected to malnutrition during
53. Kaiser N, Nesher R, Donath MY, Fraenkel M, the perinatal period. J Soc Gynecol Investig
Behar V, Magnan C, Ktorza A, Cerasi E, Lei- 3:71–77
bowitz G (2005) Psammomys obesus, a 67. Thompson NM, Norman AM, Donkin SS,
model for environment-gene interactions in Shankar RR, Vickers MH, Miles JL, Breier
type 2 diabetes. Diabetes 54:S137–S144 BH (2007) Prenatal and postnatal pathways
54. Maislos M, Medvedovskv V, Sztarkier I, to obesity: different underlying mechanisms,
Yaari A, Sikuler E (2006) Psammomys obesus different metabolic outcomes. Endocrinol
(sand rat), a new animal model of Metab Clin North Am 148:2345–2354
non-alcoholic fatty liver disease. Diabetes 68. Ozanne SE (2001) Metabolic programming
Res Clin Pract 72:1–5 in animals: type 2 diabetes. Br Med Bull
55. Barker DJP, Bull AR, Osmond C, Simmonds 60:143–152
SJ (1990) Fetal and placental size and risk of 69. Ozanne SE, Jensen CB, Tingey KJ,
hypertension in adult life. Br Med J Storgaard H, Madsbad S, Vaag AA (2005)
301:259–262 Low birthweight is associated with specific
56. Hales CN, Barker DJP (2001) The thrifty changes in muscle insulin-signalling protein
phenotype hypothesis. Br Med Bull 60:5–20 expression. Diabetologia 48:547–552
57. McMillen IC, Robinson JS (2005) Develop- 70. Hales CN, Desai M, Ozanne SE, Crowther
mental origins of the metabolic syndrome: NJ (1996) Fishing in the stream of diabetes:
prediction, plasticity, and programming. from measuring insulin to the control of fetal
Physiol Rev 85:571–633 organogenesis. Biochem Soc Trans
58. Ravelli AC, van der Meulen JH, Osmond C, 24:341–350
Barker DJ, Bleker OP (1999) Obesity at the 71. Petry CJ, Ozanne SE, Wang CL, Hales CN
age of 50 y in men and women exposed to (1997) Early protein restriction and obesity
famine prenatally. Am J Clin Nutr independently induce hypertension in
70:811–816 1-year-old rats. Clin Sci (Lond) 93:147–152
59. Ravelli ACJ, van der Meulen JHP, Michels 72. Snoeck A, Remacle C, Reusens B, Hoet JJ
RPJ, Osmond C, Barker DJP, Hales CN, (1990) Effect of a low protein diet during
Models of T2DM 67
pregnancy on the fetal rat endocrine pancreas. 83. Samuelsson A-M, Matthews PA, Argenton M,
Biol Neonate 57:107–118 Christie MR, McConnell JM, Jansen EHJM,
73. Ozanne SE, Smith GD, Tikerpae J, Hales CN Piersma AH, Ozanne SE, Twinn DF,
(1996) Altered regulation of hepatic glucose Remacle C, Rowlerson A, Poston L, Taylor
output in the male offspring of protein- PD (2008) Diet-induced obesity in female
malnourished rat dams. Am J Physiol Endo- mice leads to offspring hyperphagia, adiposity,
crinol Metab 270:E559–E564 hypertension, and insulin resistance: a novel
74. Ozanne SE, Nave BT, Wang CL, Shepherd murine model of developmental program-
PR, Prins J, Smith GD (1997) Poor fetal ming. Hypertension 51:383–392
nutrition causes long-term changes in expres- 84. Muhlhausler BS, Roberts CT, McFarlane JR,
sion of insulin signaling components in adi- Kauter KG, McMillen IC (2002) Fetal leptin
pocytes. Am J Physiol Endocrinol Metab 273: is a signal of fat mass independent of maternal
E46–E51 nutrition in ewes fed at or above maintenance
75. Ozanne SE, Wang CL, Coleman N, Smith energy requirements. Biol Reprod
GD (1996) Altered muscle insulin sensitivity 67:493–499
in the male offspring of protein-malnourished 85. Bruce KD, Cagampang FR, Argenton M,
rats. Am J Physiol Endocrinol Metab 271: Zhang J, Ethirajan PL, Burdge GC, Bateman
E1128–E1134 AC, Clough GF, Poston L, Hanson MA,
76. Ozanne SE, Olsen GS, Hansen LL, Tingey McConnell JM, Byrne CD (2009) Maternal
KJ, Nave BT, Wang CL, Hartil K, Petry CJ, high-fat feeding primes steatohepatitis in
Buckley AJ, Mosthaf-Seedorf L (2003) Early adult mice offspring, involving mitochondrial
growth restriction leads to down regulation of dysfunction and altered lipogenesis gene
protein kinase C zeta and insulin resistance in expression. Hepatology 50:1796–1808
skeletal muscle. J Endocrinol 177:235–241 86. Srinivasan M, Aalinkeel R, Song F, Mitrani P,
77. Wadley GD, Siebel AL, Cooney GJ, McConell Pandya JD, Strutt B, Hill DJ, Patel MS
GK, Wlodek ME, Owens JA (2008) Utero- (2006) Maternal hyperinsulinemia predis-
placental insufficiency and reducing litter size poses rat fetuses for hyperinsulinemia, and
alters skeletal muscle mitochondrial biogene- adult-onset obesity and maternal mild food
sis in a sex-specific manner in the adult rat. Am restriction reverses this phenotype. Am J
J Physiol Endocrinol Metab 294:E861–E869 Physiol Endocrinol Metab 290:E129–E134
78. Siebel AL, Mibus A, De Blasio MJ, Westcott 87. Cerf ME, Williams K, Chapman CS, Louw J
KT, Morris MJ, Prior L, Owens JA, Wlodek (2007) Compromised beta-cell development
ME (2008) Improved lactational nutrition and beta-cell dysfunction in weanling off-
and postnatal growth ameliorates impairment spring from dams maintained on a high-fat
of glucose tolerance by uteroplacental insuffi- diet during gestation. Pancreas 34:347–353
ciency in male rat offspring. Endocrinology 88. McCurdy CE, Bishop JM, Williams SM,
149:3067–3076 Grayson BE, Smith MS, Friedman JE, Grove
79. Poston L (2011) Intergenerational transmis- KL (2009) Maternal high-fat diet triggers
sion of insulin resistance and type 2 diabetes. lipotoxicity in the fetal livers of nonhuman
Prog Biophys Mol Biol 106:315–322 primates. J Clin Invest 119:323–335
80. Srinivasan M, Katewa SD, Palaniyappan A, 89. Ailhaud G, Massiera F, Weill P, Legrand P,
Pandya JD, Patel MS (2006) Maternal high- Alessandri JM, Guesnet P (2006) Temporal
fat diet consumption results in fetal malpro- changes in dietary fats: role of n-6 polyunsat-
gramming predisposing to the onset of meta- urated fatty acids in excessive adipose tissue
bolic syndrome-like phenotype in adulthood. development and relationship to obesity. Prog
Am J Physiol Endocrinol Metab 291: Lipid Res 45:203–236
E792–E799 90. Muhlhausler BS, Ailhaud GP (2013) Omega-
81. Cerf ME, Muller CJ, Du Toit DF, Louw J, 6 polyunsaturated fatty acids and the early
Wolfe-Coote SA (2006) Hyperglycaemia and origins of obesity. Curr Opin Endocrinol Dia-
reduced glucokinase expression in weanling betes Obes 20:56–61
offspring from dams maintained on a high- 91. Massiera F, Barbry P, Guesnet P, Joly A,
fat diet. Br J Nutr 95:391–396 Luquet S, Moreilhon-Brest C, Mohsen-
82. Muhlhausler BS, Adam CL, Findlay PA, Duf- Kanson T, Amri E-Z, Ailhaud G (2010) A
field JA, McMillen IC (2006) Increased Western-like fat diet is sufficient to induce a
maternal nutrition alters development of the gradual enhancement in fat mass over genera-
appetite-regulating network in the brain. tions. J Lipid Res 51:2352–2361
FASEB J 20:1257–1259 92. Muhlhausler BS, Gibson RA, Makrides M
(2011) The effect of maternal omega-3
68 €usler et al.
long-chain polyunsaturated fatty acid (n-3 affects maternal and fetal leptin signaling. J
LCPUFA) supplementation during preg- Nutr Biochem 24:1709–1716
nancy and/or lactation on body fat mass in 103. Samuelsson AM, Matthews PA, Jansen E,
the offspring: a systematic review of animal Taylor PD, Poston L (2013) Sucrose feeding
studies. Prostaglandins Leukot Essent Fatty in mouse pregnancy leads to hypertension,
Acids 85:83–88 and sex-linked obesity and insulin resistance
93. Muhlhausler BS, Gibson RA, Makrides M in female offspring. Front Physiol 4:14
(2010) Effect of long-chain polyunsaturated 104. Regnault TR, Gentili S, Sarr O, Toop CR,
fatty acid supplementation during pregnancy Sloboda DM (2013) Fructose, pregnancy
or lactation on infant and child body compo- and later life impacts. Clin Exp Pharmacol
sition: a systematic review. Am J Clin Nutr Physiol 40:824–837
92:857–863 105. Muhlhausler BS, Duffield JA, McMillen IC
94. Muhlhausler BS, Miljkovic D, Fong L, Xian (2007) Increased maternal nutrition stimu-
CJ, Duthoit E, Gibson RA (2011) Maternal lates peroxisome proliferator activated
omega-3 supplementation increases fat mass receptor-{gamma} (PPAR{gamma}), adipo-
in male and female rat offspring. Front Genet nectin and leptin mRNA expression in adi-
2:48 pose tissue before birth. Endocrinology
95. Chen L, Hu FB, Yeung E, Willett W, Zhang C 148:878–885
(2009) Prospective study of pre-gravid sugar- 106. Kasser TR, Martin RJ, Allen CE (1981) Effect
sweetened beverage consumption and the risk of gestational alloxan diabetes and fasting on
of gestational diabetes mellitus. Diabetes Care fetal lipogenesis and lipid deposition in pigs.
32:2236–2241 Biol Neonate 40:105–112
96. Phelan S, Hart C, Phipps M, Abrams B, 107. Ezekwe MO, Martin RJ (1980) The effects of
Schaffner A, Adams A, Wing R (2011) Mater- maternal alloxan diabetes on body composi-
nal behaviors during pregnancy impact off- tion, liver enzymes and metabolism and
spring obesity risk. Exp Diabetes Res serum metabolites and hormones of fetal
985139:26 pigs. Horm Metab Res 12:136–139
97. Toop CR, Muhlhausler BS, O’Dea K, Gentili 108. Bayol SA, Simbi BH, Bertrand JA, Stickland
S (2014) Consumption of sucrose, but not NC (2008) Offspring from mothers fed a
high fructose corn syrup, leads to increased ‘junk food’ diet in pregnancy and lactation
adiposity and dyslipidaemia in the pregnant exhibit exacerbated adiposity that is more
and lactating rat. J Dev Orig Health Dis pronounced in females. J Physiol
19:1–9 586:3219–3230
98. Goran MI, Dumke K, Bouret SG, Kayser B, 109. Muhlhausler B, Smith SR (2009) Early-life ori-
Walker RW, Blumberg B (2013) The obeso- gins of metabolic dysfunction: role of the adi-
genic effect of high fructose exposure during pocyte. Trends Endocrinol Metab 20:51–57
early development. Nat Rev Endocrinol 110. Padoan A, Rigano S, Ferrazzi E, Beaty BL,
9:494–500 Battaglia FC, Galan HL (2004) Differences in
99. Sloboda DM, Li M, Patel R, Clayton ZE, fat and lean mass proportions in normal and
Yap C, Vickers MH (2014) Early life exposure growth-restricted fetuses. Am J Obstet Gyne-
to fructose and offspring phenotype: implica- col 191:1459–1464
tions for long term metabolic homeostasis. J 111. Crescenzo R, Samec S, Antic V, Rohner-
Obesity 203474:23 Jeanrenaud F, Seydoux J, Montani J-P, Dul-
100. Vickers MH, Clayton ZE, Yap C, Sloboda loo AG (2003) A role for suppressed thermo-
DM (2011) Maternal fructose intake during genesis favoring catch-up fat in the
pregnancy and lactation alters placental pathophysiology of catch-up growth. Acta
growth and leads to sex-specific changes in Paediatr 52:1090–1097
fetal and neonatal endocrine function. Endo- 112. Ibanez L, Ong K, Dunger DB, de Zegher F
crinology 152:1378–1387 (2006) Early development of aiposity and
101. Mukai Y, Kumazawa M, Sato S (2013) Fruc- insulin resistance after catch-up weight gain
tose intake during pregnancy up-regulates the in small-for-gestational-age children. J Clin
expression of maternal and fetal hepatic sterol Endocrinol Metab 91:2153–2158
regulatory element-binding protein-1c in 113. Jaquet D, Gaboriau A, Czernichow P, Levy-
rats. Endocrine 44:79–86 Marchal C (2000) Insulin resistance early in
102. Rodriguez L, Panadero MI, Roglans N, adulthood in subjects born with intrauterine
Otero P, Alvarez-Millan JJ, Laguna JC, growth retardation. J Clin Endocrinol Metab
Bocos C (2013) Fructose during pregnancy 85:1401–1406
Models of T2DM 69
114. De Blasio MJ, Gatford KL, McMillen IC, 126. Gallou-Kabani C, Junien C (2005) Nutri-
Robinson JS, Owens JA (2006) Placental tional epigenomics of metabolic syndrome:
restriction of fetal growth increases insulin new perspective against the epidemic. Diabe-
action, growth and adiposity in the young tes 54:1899–1906
lamb. Endocrinology 148:1350–1358 127. Waterland RA, Travisano M, Tahiliani KG,
115. De Blasio MJ, Gatford KL, Robinson JS, Rached MT, Mirza S (2008) Methyl donor
Owens JA (2007) Placental restriction of supplementation prevents transgenerational
fetal growth reduces size at birth and alters amplification of obesity. Int J Obes (Lond)
postnatal growth, feeding activity, and adipos- 32:1373–1379
ity in the young lamb. Am J Physiol Regul 128. Burdge GC, Hanson MA, Slater-Jefferies JL,
Integr Comp Physiol 292:R875–R886 Lillycrop KA (2007) Epigenetic regulation of
116. Ozanne SE (2001) Metabolic programming transcription: a mechanism for inducing var-
in animals. Br Med Bull 60:143–152 iations in phenotype (fetal programming) by
117. Kind KL, Clifton PM, Grant PA, Owens PC, differences in nutrition during early life? Br J
Sohlstrom A, Roberts CT, Robinson JS, Nutr 97:1036–1046
Owens JA (2003) Effect of maternal feed 129. Burdge GC, Slater-Jefferies J, Torrens C,
restriction during pregnancy on glucose tol- Phillips ES, Hanson MA, Lillycrop KA
erance in the adult guinea pig. Am J Physiol (2007) Dietary protein restriction of preg-
Regul Integr Comp Physiol 284:R140–R152 nant rats in the F0 generation induces altered
118. Alexander G (1978) Quantitative develop- methylation of hepatic gene promoters in the
ment of adipose tissue in foetal sheep. Aust J adult male offspring in the F1 and F2 genera-
Biol Sci 31:489–503 tions. Br J Nutr 97:435–439
119. Merklin RJ (1973) Growth and distribution of 130. Tobi EW, Slagboom PE, van Dongen J,
human fetal brown fat. Anat Res 178:637–646 Kremer D, Stein AD, Putter H, Heijmans
120. Højlund K, Mogensen M, Sahlin K, Beck- BT, Lumey LH (2012) Prenatal famine and
Nielsen H (2008) Mitochondrial dysfunction genetic variation are independently and addi-
in type 2 diabetes and obesity. Endocrinol tively associated with DNA methylation at
Metab Clin North Am 37:713–731 regulatory loci within IGF2/H19. PLoS
One 7:30
121. Howie GJ, Sloboda DM, Reynolds CM, Vick-
ers MH (2013) Timing of maternal exposure 131. Hanley B, Dijane J, Fewtrell M, Grynberg A,
to a high fat diet and development of obesity Hummel S, Junien C, Koletzko B, Lewis S,
and hyperinsulinemia in male rat offspring: Renz H, Symonds M, Gros M,
same metabolic phenotype, different develop- Harthoorn L, Mace K, Samuels F, van Der
mental pathways? J Nutr Metab 517384:13 Beek EM (2010) Metabolic imprinting, pro-
gramming and epigenetics – a review of pres-
122. Palmer AC (2011) Nutritionally mediated ent priorities and future opportunities. Br J
programming of the developing immune sys- Nutr 104:S1–S25
tem. Adv Nutr 2:377–395
132. Heerwagen MJR, Miller MR, Barbour LA,
123. Zhu MJ, Du M, Nathanielsz PW, Ford SP Friedman JE (2010) Maternal obesity and
(2010) Maternal obesity up-regulates inflam- fetal metabolic programming: a fertile epige-
matory signaling pathways and enhances netic soil. Am J Physiol 299:R711–R722
cytokine expression in the mid-gestation
sheep placenta. Placenta 31:387–391 133. Li CC, Maloney CA, Cropley JE, Suter CM
(2010) Epigenetic programming by maternal
124. Murabayashi N, Sugiyama T, Zhang L, nutrition: shaping future generations. Epige-
Kamimoto Y, Umekawa T, Ma N, Sagawa N nomics 2:539–549
(2013) Maternal high-fat diets cause insulin
resistance through inflammatory changes in 134. Li CC, Young PE, Maloney CA, Eaton SA,
fetal adipose tissue. Eur J Obstet Gynecol Cowley MJ, Buckland ME, Preiss T, Hen-
Reprod Biol 169:39–44 stridge DC, Cooney GJ, Febbraio MA, Mar-
tin DI, Cropley JE, Suter CM (2013)
125. Junien C, Gallou-Kabani C, Vigé A, Gross Maternal obesity and diabetes induces latent
MS (2005) Nutritional epigenomics: conse- metabolic defects and widespread epigenetic
quences of unbalanced diets on epigenetics changes in isogenic mice. Epigenetics 8:602
processes of programming during lifespan
and between generations. Ann Endocrinol 135. Haslam DW, James WP (2005) Obesity. Lan-
(Paris) 66:S19–S28 cet 366:1197–1209
Chapter 4
Cheminformatics in the Identification of Drug Classes

for the Treatment of Type 2 Diabetes
Paul W. Finn
Abstract
Computer-Aided Drug Design has developed into a powerful suite of methods that complement experi-
mental approaches to the identification of new pharmacologically active compounds. In particular, virtual
screening has become a standard tool for lead identification. Diverse examples of the application of virtual
screening applied to T2DM target proteins have been reported. While several of these indicate successful
identification of new lead compounds from synthetic chemical and natural product databases, many of them
have been performed on a small scale and with limited validation. Careful study design and collaboration
with cheminformaticians and computational chemists will enable these approaches to fulfil their potential
for T2DM.
Key words Cheminformatics, Virtual screening, Molecular modeling, Chemical database
1 Introduction
The goal of Computer-Aided Drug Design (CADD) is to use

computational approaches to understand the relationship between
the chemical structures of drug molecules and their biological
activities, and to use this understanding to assist in identifying
new and improved compounds. CADD began in the 1960s with
simple statistical regression models. As our theoretical understand-
ing of molecular interactions and the processing power of compu-
ters has increased, CADD has grown in scope and sophistication to
now encompass a large toolbox of machine learning and simulation
methods.
One of the most widely used CADD methods is “virtual
screening.” Virtual screening simulates experimental testing; in
particular, the High-Throughput Screening performed in Pharma
and Biotech companies. To perform a virtual screen, two compo-
nents are required. The first is a cheminformatics database, which
stores the compounds’ chemical structures, names (identifiers), and
typically a range of computed and/or experimentally determined
71
72 Paul W. Finn
Fig. 1 A typical virtual screening pipeline. An input database, which may contain tens of millions of unique
chemical structures, is searched against either a ligand or structure-based model of activity. Typically, the
initial rank-ordered list of “hits” is further processed by additional cheminformatic criteria and by manual
selection to identify the compounds to be tested experimentally. The final numbers chosen range from a
handful to several hundreds or thousands
chemical properties. The second is a computer model of the

requirements for a particular biological activity. A diverse range of
models have been used, but they can be categorized at a high level
into two main classes. Ligand-based virtual screening compares the
database compounds to a query compound of known activity
through a quantitative measure of molecular similarity. The similar-
ity metric could, for example, be overall molecular shape or a
geometric pattern of interaction features (known as a pharmaco-
phore) the presence of which in a compound confers the desired
biological activity. Structure-based virtual screening assesses the
shape and property complementarity of each database compound
to a binding site (usually the active site, but possibly an allosteric
site) of a target protein (Fig. 1).
One of the main driving forces behind the development of
virtual screening methods has been the exponential growth in the
availability of experimental data on which these methods can be
applied. Protein X-ray crystallography methodology and invest-
ment in structural genomics initiatives are delivering the atomic
resolution three-dimensional structures of the protein targets of
drugs at an ever-increasing rate (Fig. 2), such that in 2016, over
10,000 new protein X-ray structures were deposited in the Protein
Data Bank.
New Diabetes Drugs from Cheminformatics 73
Fig. 2 Since the mid-1990s there has been continual growth in the number of new protein structures that can
be used as input to computational methods
Equally important has been the growth in public domain

sources of chemical structure databases of commercially available
and literature compounds. The popular ZINC database [1] is in
the public domain and contains approximately 4.5 million chemical
structures appropriate for virtual screening experiments and avail-
able for commercial suppliers. Other examples of useful cheminfor-
matics databases include ChemSpider [2], maintained by the UK
Royal Society of Chemistry, which includes information on over
30 million structures collated from over 450 data sources, includ-
ing analytical data and literature and patent references and
ChEMBL [3], a database of over 1.5 million compounds combin-
ing chemical structure and biological activity data mostly extracted
from the open scientific literature. A non-exhaustive summary of
software tools and databases for virtual screening is provided in
Table 1.
Type 2 diabetes (T2DM) is a complex disease presenting a
great variety of potential drug targets. Many of these can be inves-
tigated by CADD methodologies. The remaining sections of this
review chapter summarize some of the recent T2DM CADD liter-
ature to provide an overview of the scope of these methods to
uncover new chemotypes (drug classes) as lead compounds for
drug discovery and development. While there are a number of
success stories in the drug discovery literature, CADD methods,
and their application are far from mature technologies and there are
many pitfalls to their successful application. Some of these
74 Paul W. Finn
Table 1
Tools and databases for virtual screening
Structure-based virtual screening

AutoDock/autodock vina Autdock.scripps.edu
Dock Dock.compbio.ucsf.edu/Overview_of_DOCK/index.htm
GOLD www.ccdc.cam.ac.uk/solutions/csd-discovery/components/gold/
Glide www.schrodinger.com/Glide
FlexX www.biosolveit.de/flexx/index.htm
Blaster Blaster.docking.org
Ligand-based virtual screening
Pharmer smoothdock.ccbb.pitt.edu/pharmer/
LiSiCa insilab.org/lisica/
LigandScout www.inteligand.com/ligandscout/
Compound databases
Zinc zinc15.docking.org/
ChEMBL www.ebi.ac.uk/chembl/
ChemSpider www.chemspider.com
DrugBank www.drugbank.ca
limitations are highlighted in the context of T2DM studies in the

hope that this will be helpful to diabetes researchers stimulated to
apply CADD in their own research.
2 Virtual Screening
There are many examples in the literature of the use of virtual

screening and related approaches to identify novel compounds
with the potential to treat type 2 diabetes. It is not possible to
provide a detailed review of this extensive literature, but the follow-
ing recent examples provide an impression of the current state-of-
the-art.
Virtual screening and “core-hopping” studies have been
reported to find dual agonists of PPAR-α and PPAR-γ. One study
was based on GW409554 [4] and the other employed the ZINC
screening database [5]. The protocol combines multiple computa-
tional approaches and included prediction of ADME properties,
resulting in the design of ten putative dual agonists in each case.
Using similar methods, the same group has recently reported the
design of a multitargeted PPARα/γ/δ pan agonist [6]. However,
no experimental confirmation of the design ideas is reported in any
of these publications, and thus, the hypotheses must currently be

regarded as speculative given the current limited performance of
computational design approaches.
In another application of the ZINC database, a combination of
QSAR and docking methods was used to perform virtual screening
of more than 25 million compounds in the ZINC library. The
QSAR model was developed using 1,517 compounds, identifying
42,378 potential PPARγ agonists. Of these, 10,000 were selected
for docking with PPARγ based on their diversity. Several steps were
used to refine the docking results, and finally 30 potentially highly
active ligands were identified. Four compounds were subsequently
tested for their in vitro activity, with one compound found to have a
Ki value of <5 μM [7].
Tanwar et al. [8] performed structure-based virtual screening
against dipeptidyl peptidase IV using the Molecular Diversity Pres-
ervation International (MDPI) database and the docking programs
Glide and GOLD. Six hits were tested for DPP-IV inhibition, of
which the most potent had an IC50 of 0.73 μM and glucose
lowering effects in fed hyperglycemic female Wistar rats. For the
same target, pharmacophore-based virtual screening against the
NCI database identified a small number of hits, although activity
was very weak with the most potent compound, paroxetine,
showing only 20% inhibition at 10 μM.
Another widely studied diabetes target is aldose reductase
(AR). AR is an aldo–keto reductase that has been widely investi-
gated as an enzyme crucially involved in the pathogenesis of chronic
complications associated with diabetes mellitus as well as acting as a
key mediator of oxidative and inflammatory signaling pathways
involved in the development of other human pathologies, such as
cardiovascular disorders, sepsis, and cancer [9]. Wang et al. [10]
have described the identification of novel aldose reductase inhibi-
tors through virtual screening. Notable aspects of this study are the
use of multiple binding site conformations and the extensive use of
molecular dynamics, a computationally expensive process which
require the use of Graphics-Processor-Unit (GPU) hardware to
make the computation tractable. The library screened was very
small for a virtual screening experiment, at only 7,249 compounds,
but a respectable 14% hit rate was obtained.
The farnesoid X receptor (FXR), a ligand-dependent transcrip-
tion factor of the nuclear hormone receptor superfamily, is involved
in the regulation of glucose and lipid metabolism and is thus a
potential target for T2DM and other metabolic diseases. Workers
at the University of Innsbruck and their collaborators have under-
taken a thorough pharmacophore-based approach to discover FXR
agonists. Pharmacophore models were developed based on multi-
ple structures available in the PDB. Known FXR agonists were
obtained from ChEMBL-DB (FXR-Actives, 221 compounds) as
were a property-matched set of inactive compounds (FXR-Decoys,
76 Paul W. Finn
5,598 compounds). Property matching involves ensuring that the

inactive compounds display a similar distribution of physicochemi-
cal and pharmacophoric feature ranges (logP, molecular weight,
hydrogen bonds, etc.). If this is not done, then it is easy to develop
computer models that distinguish actives from inactives on the basis
of these simple properties, but such models have no predictive
value. Three other databases were also used in this study, DrugBank
[11] and an in-house “Virtual DB” were used for validation and the
NCI Developmental Therapeutics Program database for predic-
tion. Validation studies on these datasets enabled the selection of
the best-performing pharmacophore models and the selection of
virtual screening hits from the NCI database for biological testing.
Only eight compounds were tested; two showed dose-dependent
FXR agonism, with the remaining six being inactive [12]. The same
approach was also applied to the identification of novel FXR ago-
nists from natural sources (see below).
3 Natural Products
A significant percentage of small molecule drugs are either natural

products or natural product derivatives. Plants and plant extracts
have been used since ancient times for the treatment of T2DM,
with a great number of species being used [13].
Historically, many of the sources of natural product used for
experimental screening consisted of uncharacterized or only partly
characterized samples. This limited the potential for computational
virtual screening approaches, which typically require a well-defined
chemical structure to associate with the screening sample. How-
ever, a number of natural product computer databases have now
been developed, for example, the Analyticon database of over 5,000
natural products is available for download (www.ac-discovery.com)
and several other companies maintain natural product databases.
The website Super Natural II (bioinf-applied.charite.de/supernatu
ral_new/index.php) allows access to a database of 325,508 natural
compounds through collecting offerings from a number of com-
mercial suppliers and academic collections.
There are a number of literature reports of efforts to explore
the PPAR-γ-activating potential of a wide range of natural products
originating from traditionally used medicinal plants or dietary
sources. Cheminformatics approaches have been applied to natural
product databases and again PPAR-γ has been a frequently used
target structure.
Fakhrudin et al. [14] used a structure-based pharmacophore
model to screen natural compound databases. The model was based
on the PDB entry identifier 2G0G and consists of three hydropho-
bic features, one aromatic ring, one hydrogen bond acceptor, and
exclusion volume spheres lining the ligand-binding domain of
PPAR-γ. Sixty-one hits were identified, including several neo-

lignans, including magnolol, as partial PPAR-γ agonists with
EC50 values in the low micromolar or submicromolar range and
the ability to induce adipocyte differentiation in 3T3-L1 adipo-
cytes. The prediction that magnolol binds to PPAR-γ was subse-
quently confirmed by X-ray crystallography, although the binding
mode was different from that predicted by modeling [15]. Magno-
lol was also demonstrated to bind to the RXRα receptor, indicating
potential polypharmacology (see below). Subsequently [16], fur-
ther computational modeling experiments identified another natu-
ral product neolignan, honokiol, a major constituent of the
traditional Chinese herbal drug Magnolia bark. Honokiol was
reported to directly bind to the purified human PPARγ ligand-
binding domain in vitro with a Ki of 22.86 μM, but unlike previ-
ously identified neolignans, it did not induce PPAR-γ-dependent
lipid accumulation in 3T3-L1 adipocytes. However, other reports
question the ability of honokiol to interact directly with PPAR
receptors [17]. Also, using a structure-based pharmacophore
model, in this case based on the common interactions of four
different agonists with the PPARγ structure, Tanrikulu et al. [18]
screened the Analyticon database, leading to the discovery of two
α-santonin derivatives as PPARγ activators.
In another example [19], a ligand-derived PPARγ pharmaco-
phore was created, starting from an analysis of 13 selective partial
agonists. The derived pharmacophore, simpler than the structure-
based one as it consists only of a hydrogen bond acceptor and three
hydrophobic features, was used to search the Chinese Natural
Product Database, comprising over 57,000 constituents of Tradi-
tional Chinese Medicine. One compound selected from the hitlist,
methyl oleanonate, is found in Chios mastic gum. The acid of
methyl oleanonate, oleanonic acid, was identified as a modestly
active PPARγ agonist through bioassay-guided chromatographic
fractionation of Chios mastic gum fractions.
Rupp et al. [20] used a machine learning method, descriptor-
based Gaussian process regression, to search for PPARγ agonists
based on a data set of 144 published PPAR-γ ligands. Fifteen
compounds were selected for experimental testing in PPAR-α and
PPAR-γ agonist assays, based on a combination of their perfor-
mance in the prediction models and manual selection. Eight of
these exhibited agonist activity toward at least one of the receptors.
The most active compound, a truxillic acid derivative, was a selec-
tive PPAR-γ agonist with an EC50 of 10 μM.
Lewis et al. [21] used docking to evaluate various natural
products as putative PPARγ agonists. They were able to demon-
strate a docking pose for α-eleostearic acid, a conjugated triene, and
a compound already known as an activator of PPARγ [22].
A further structure-based virtual screening study [23] of the
PPARγ- ligand-binding domain against a natural product library
78 Paul W. Finn
identified 29 potential agonists. Of these, six flavonoids stimulated

PPARγ transcriptional activity in a transcriptional factor assay with
the most potent, psi-baptigenin-possessing an EC50 of 2.9 μM.
Evidence that the compound was working through the intended
mechanism of action was provided by the finding that psi-bapti-
genin’s-induced PPARγ activity was abolished in the presence of a
selective PPARγ antagonist, GW9662.
Much of the computational analysis of natural product libraries
has focused on PPAR-γ as the molecular target, but other proteins
have also been investigated. Building on their previously developed
FXR pharmacophore work (see above) Grienke et al. have explored
an in-house Chinese Herbal Medicine database. This identified
mainly lanostane-type triterpenes of the fungus Ganoderma luci-
dum Karst as putative FXR ligands. Extracts of Ganoderma fruit
bodies and isolated compounds exhibited FXR inducing effects at
100 μg/mL.
DPP-IV inhibitors from natural sources have also been studied
[24]. Starting with a database of 89,000 natural compounds taken
from ZINC, a structure-derived pharmacophore and docking study
identified 446 potential hit compounds. To focus on those com-
pounds that contained novel elements, the hitlist was merged with
2,342 known DPP-IV inhibitors taken from the BindingDB data-
base [25] and clustered based on 2D structural fingerprints. Of the
resulting 50 clusters, 12 exclusively contained natural products. A
total of nine compounds were selected from seven of these 12 clus-
ters and tested in an in vitro assay. The compounds showed activity,
but were tested at very high concentrations (between 100 μM and
1 mM). A computational analysis suggested analogues with pre-
dicted higher binding affinity, but no compounds were synthesized
or tested. The weak affinity of the identified compounds and the
lack of any experimental confirmation of the proposed binding
mode make the conclusions drawn highly speculative at this time.
A follow-up study suggested further compounds from natural
extracts as DPP-IV inhibitors, but without any supporting
evidence [26].
An alternative strategy to the search for natural product anti-
diabetic compounds has been described by Pathania et al.
[27]. Here, the search was focused on compounds contained in a
specific plant Rauvolfia serpentina, a medicinal plant endemic to
the Himalayas known to be effective in relieving diabetes and its
complications. This focus naturally led to a rather small dataset of
compounds, only 142 unique chemical structures. The search spe-
cifically focused on the identification of putative inhibitors of aldose
reductase. Docking studies against a high-resolution aldose reduc-
tase crystal structure were followed by molecular dynamics refine-
ment, leading to two indole alkaloids, indobine and indobinine
being suggested as novel aldose reductase inhibitors. The compu-
tational characterization of the compounds is relatively thorough,
but the conclusions are again unsupported by experimental valida-

tion. It can be difficult, of course, to obtain sufficient sample of
well-characterized natural products. The authors, however, suggest
structurally related compounds from the ZINC database that they
also predict to be active through the same screening workflow. Such
compounds are often available commercially at modest cost for a
screening sample. Testing these commercial compounds could have
been an effective way to add support to their conclusions.
4 Polypharmacology
The vast majority of drug discovery activities undertaken over the

last 50 years have conformed to the classical one-compound–one-
target paradigm derived from the “magic bullet” hypothesis of
Ehrlich [28]. Nonetheless, modulation of unintended targets that
can play a fundamental role in explaining pharmacological profiles,
for example, of drugs acting on the central nervous system, which
can bind to many different receptors [29] and for some multi-
kinase inhibitors in oncology [30]. In many cases, such “polyphar-
macology” arose serendipitously, but it suggests that there is poten-
tial in purposely designing compounds with multiple targets, and
researchers have begun to explicitly consider systems pharmacology
and the functioning of drug targets as part of a network, rather than
individually [31–33].
In this context, Gu et al. have created a natural product data-
base and have explored its network properties as a source of com-
pounds with polypharmacology [34, 35]. The analysis was limited
by a lack of experimental activity data on the compounds, so this
was supplemented with computational virtual screening data
against 332 protein targets. The structure of the derived network
indicated that the database should be rich in compounds with
polypharmacological properties. Related work has investigated the
mechanism of action of components of a Traditional Chinese Med-
icine (TCM) used for the treatment of T2DM. It was found that
multiple compounds in the preparation were predicted to target
proteins implicated in T2DM, and that this would have an effect at
the network level. Several additional compounds with no reported
pharmacological activity in T2DM were also highlighted. More
recently, Tian et al. [36], combined docking, pharmacophore
mapping and machine learning methods to establish a compound–-
target interaction network for TCMs. Approximately, 20% of the
predicted inhibitors could interact with multiple targets and some
of these predictions were experimentally validated. The authors
speculate that the antidiabetic efficacy of TCMs derives from a
combination of substituents that are active against T2DM-related
targets supplemented by bioactivities that are more indirectly
related, in particular compounds with free radical scavenging and
antioxidant activities.
80 Paul W. Finn
5 Conclusions
Computational methods continue to develop in speed and sophis-

tication and have great potential to identify novel synthetic and
natural product compounds as leads for the development of new
T2DM therapies. However, the computational methods present
many technical difficulties at every stage of their application and
there are many pitfalls which can trap nonexperts and lead to false
conclusions.
Most molecular modeling protocols involve the sequential
application of a number of steps, many of these requiring choices
to be made by the scientists involved. In many studies, it is difficult
to determine the degree to which the various modeling steps
involved contributed to the identification of the active compound,
and more importantly, the lack of any control experiments makes
judging the effectiveness of the method overall problematic. For
example, could a simpler method have performed as well, or is the
success largely determined by the manual selection of compounds
for testing, which is typically the final step in many protocols?
Several 2D and 3D virtual screening approaches summarized
above have reported structurally diverse natural products modula-
tors of T2DM targets, which at face value indicates that natural
products could be a rich source for novel compounds for the
treatment of T2DM. However, in many of these reports, the num-
ber of compounds subjected to experimental testing is very small,
often barely into double figures. If, say, two of these compounds are
active in the assay, it is very hard to judge the significance of the
finding. This is particularly the case if the compounds have been
tested at high concentrations. It is not rare to see concentrations as
high as 100 μM, or even higher, being used. The pharmacological
relevance of effects at these concentrations must be doubtful.
Of the compounds reported as, for example, PPARγ modula-
tors, certain chemical families (chemotypes) occur quite frequently.
Of course, at one level, this is expected, as from the similarity
principle that guides medicinal chemistry, similar compounds will
tend to share similar biological effects. However, often the binding
modes of these compounds are different to the thiazolidinedione
agonists and their effects in animal models are different (sometimes
beneficially so, through reduction in side effects). Example com-
pounds include genistein, biochanin A, sargaquinoic acid, resvera-
trol, amorphastilbol, magnolol, honokiol, and amorfrutin
1 (Fig. 3). In most cases, predicted binding modes were not fol-
lowed up with experimental verification through X-ray
crystallography.
There is also growing awareness that many compounds have
the potential to interact nonspecifically with many biological sys-
tems and are thus frequently identified in high-throughput and
Fig. 3 Chemical structures of selected natural product PPAR modulators reported in the literature
computational screening exercises. Such compounds have been

dubbed PAINS, pan-assay interference compounds [37] and it
has been estimated that 5–12% of a typical academic screening
library consists of PAINS. Well-known examples include polyhy-
droxylated natural compounds such as resveratrol, curcumin, and
genistein. Compounds of this type require careful characterization
if they are to be suggested as interesting drug leads, as their poten-
tial to interact nonspecifically with multiple biological systems is
well documented [38, 39]. This may be particularly the case with
receptors such as PPARγ, which appears to be rather promiscuous
in its ability to bind a wide variety of structurally diverse com-
pounds. A good example of the characterization that can be per-
formed can be seen in the identification of the amorfrutins
(through experimental screening) as partial agonists of PPARγ
and as potent antidiabetic natural products [40].
Among the important points for someone wishing to apply
virtual screening to carefully consider are the following:
1. Choice of software tool. Comparative studies of the methodol-
ogies listed in Table 1, and others, shows that they tend to have
82 Paul W. Finn
similar levels of performance overall. However, there is much

variability from target to target and it can be difficult to know in
advance which method will work best for any particular target.
Ideally, one may wish to choose a method that has been shown
to perform well against similar targets or binding sites.
2. Beware of artifacts that could arise from the choice of com-
pound database or methodology. Compare performance with a
simpler method and evaluate the performance statistically
[41]. Guidelines are available to help avoid the most serious
pitfalls [42].
3. The “hit rates” for virtual screening reported in the literature
of often derived from retrospective studies. The performance in
prospective studies is usually much lower. It is unlikely that
active compounds will be found unless a large number (proba-
bly hundreds) of compounds are tested experimentally. Also,
test at a reasonable concentration (at most low micromolar).
Beware PAINS and potential nonspecific activities—demon-
strate a dose–response.
There has never been a greater need for the development of
new treatments for T2DM, and there would be great benefit glob-
ally from identification of compounds of natural origin as aspects of
therapeutic regimens. Our improving knowledge of the pathophys-
iology of the disease and the development of ever-expanding data-
bases and screening methods offers much promise. To deliver on
this promise, closer collaboration between experimentalists with
T2DM domain expertise and expert cheminformaticians is needed
if we are to identify compounds with genuine potential to address
the unmet medical need.
References
1. Sterling T, Irwin JJ (2015) ZINC 15 – ligand by means of cheminformatics. Drug Des Devel
discovery for everyone. J Chem Inf Model Ther 7:279–288
55:2324–2337 6. Wang X-J, Zhang J, Wang S-Q et al (2014)
2. Pence HE, Williams A (2010) ChemSpider: an Identification of novel multitargeted PPARα/
online chemical information resource. J Chem γ/δ pan agonists by core hopping of rosiglita-
Educ 87:1123–1124 zone. Drug Des Devel Ther 8:2255–2262
3. Gaulton A, Bellis LJ, Bento AP et al (2011) 7. Gee VM, Wong FS, Ramachandran L et al
ChEMBL: a large-scale bioactivity database for (2014) Identification of novel peroxisome
drug discovery. Nucleic Acids Res 40: proliferator-activated receptor-gamma
D1100–D1107 (PPARγ) agonists using molecular modeling
4. Ma Y, Wang S-Q, Xu W-R et al (2012) Design method. J Comput Aided Mol Des
novel dual agonists for treating type-2 diabetes 28:1143–1151
by targeting peroxisome proliferator-activated 8. Tanwar O, Tanwar L, Shaquiquzzaman M et al
receptors with core hopping approach. PLoS (2014) Structure based virtual screening of
One 7:e38546. https://doi.org/10.1371/ MDPI database: discovery of structurally
journal.pone.0038546 diverse and novel DPP-IV inhibitors. Bioorg
5. Liu L, Ma Y, Wang R-L et al (2013) Find novel Med Chem Lett 24:3447–3451
dual-agonist drugs for treating type 2 diabetes
9. Maccari R, Ottanà R (2015) Targeting aldose mice by activating peroxisome proliferator-

reductase for the treatment of diabetes compli- activated receptor-gamma. PLoS One 6:
cations and inflammatory diseases: new insights e24031
and future directions. J Med Chem 22. Moon H-S, Guo D-D, Lee H-G et al (2009)
58:2047–2067 Alpha-eleostearic acid suppresses proliferation
10. Wang L, Gu Q, Zheng X et al (2013) Discov- of MCF-7 breast cancer cells via activation of
ery of new selective human aldose reductase PPARγ and inhibition of ERK1/2. Cancer Sci
inhibitors through virtual screening multiple 101:396–402
binding pocket conformations. J Chem Inf 23. Salam NK, Huang TH, Kota BP et al (2008)
Model 53:2409–2422 Novel PPAR-gamma agonists identified from a
11. Wishart DS, Knox C, Guo AC et al (2006) natural product library: a virtual screening,
DrugBank: a comprehensive resource for in induced-fit docking and biological assay study.
silico drug discovery and exploration. Nucleic Chem Biol Drug Des 71:57–70
Acids Res 34:D668–D672 24. Guash L, Ojeda MJ, Gonzalez-Abuin N et al
12. Schuster D, Markt P, Grienke U et al (2011) (2012) Identification of novel human dipepti-
Pharmacophore-based discovery of FXR ago- dyl peptidase-IV inhibitors of natural origin
nists. Part 1. Model development and experi- (Part I): virtual screening and activity assays.
mental validation. Bioorg Med Chem PLoS One 7:e44791
19:7168–7180 25. Chen X, Liu M, Gilson MK (2001) Bin-
13. Bnouham N, Ziyyat A, Mekhfi H et al (2006) dingDB: a web-accessible molecular recogni-
Medicinal plants with potential antidiabetic tion database. Comb Chem High
activity – a review of ten years of herbal medi- Throughput Screen 4:719–725
cine research (1990-2000). Int J Diabetes & 26. Guasch L, Sala E, Ojeda MJ et al (2012) Iden-
Metab 14:1–25 tification of novel human dipeptidyl-IV inhibi-
14. Fakhrudin N, Ladurner A, Atanasov AG et al tors of natural origin (Part II): in silico
(2010) Computer-aided discovery, validation, prediction in antidiabetic extracts. PLoS One
and mechanistic characterization of novel neo- 7:e44792
lignan activators of peroxisome proliferator- 27. Pathania S, Randhawa V, Bagler G (2013) Pro-
activated receptor gamma. Mol Pharmacol specting for novel plant-derived molecules of
77:559–566 Rauvolfia serpentina as inhibitors of aldose
15. Zhang H, Xu X et al (2011) Molecular deter- reductase, a potent drug target for diabetes
minants of magnolol targeting both RXRα and and its complications. PLoS One 8:e61327
PPARγ. PLoS One 6(11):e28253 28. Ehrlich P (1907) On immunity with special
16. Atanasov AG, Wang JN, Gu SP et al (2013) reference to the relationship between distribu-
Honokiol: a non-adipogenic PPARγ agonist tion and action of antigens. J R Inst Pub Health
from nature. Biochim Biophys Acta 15:321–340
1830:4813–4819 29. Roth BL, Sheffler DJ, Kroeze WK (2004)
17. Choi SS, Cha BY, Iida K et al (2011) Honokiol Magic shotguns versus magic bullets: selec-
enhances adipocyte differentiation by poten- tively non-selective drugs for mood disorders
tiating insulin signaling in 3T3-L1 preadipo- and schizophrenia. Nat Rev Drug Discov
cytes. J Nat Med 65:424–430 3:353–359
18. Tanrikulu Y, Rau O, Schwarz O et al (2009) 30. Knight ZA, Lin H, Shokat KM (2010) Target-
Structure-based pharmacophore screening for ing the cancer kinome through polypharmacol-
natural-product-derived PPARgamma ago- ogy. Nat Rev Cancer 10:130–137
nists. Chembiochem 10:75–78 31. Hopkins AL (2008) Network pharmacology:
19. Petersen RK, Christensen KB, Assimopoulou the next paradigm in drug discovery. Nat
AN et al (2011) Pharmacophore-driven identi- Chem Biol 4:682–690
fication of PPARgamma agonists from natural 32. Berger SI, Iyengar R (2009) Network analyses
sources. J Comput Aided Mol Des 25:107–116 in systems pharmacology. Bioinformatics
20. Rupp M, Schroeter T, Steri R et al (2010) 25:2466–2472
From machine learning to natural product 33. Vogt I, Mestres J (2010) Drug-target net-
derivatives that selectively activate transcription works. Mol Informatics 29:10–14
factor PPARgamma. ChemMedChem 34. Gu J, Gui Y, Chen L et al (2013) Use of natural
5:191–194 products as chemical library for drug discovery
21. Lewis SN, Brannan L, Guri AJ et al (2011) and network pharmacology. PLoS One 8:
Dietary alpha-eleostearic acid ameliorates e62839
experimental inflammatory bowel disease in
84 Paul W. Finn
35. Gu J, Chen L, Yuan G et al (2013) A drug- 38. Priyadarsini KI (2013) Chemical and structural
target network-based approach to evaluate the features influencing the biological activity of
efficacy of medicinal plants for type II diabetes curcumin. Curr Pharm Des 19:2093–2100
mellitus. Evid Based Complement Alternat 39. Ingólfsson HI et al (2014) Phytochemicals per-
Med 2013:203614 turb membranes and promiscuously alter pro-
36. Tian S, Li Y, Li D et al (2013) Modelling tein function. ACS Chem Biol 9:1788–1798
compound-target interaction network of tradi- 40. Weidner C et al (2012) Amorfrutins are potent
tional Chinese medicines for type II diabetes antidiabetic dietary natural products. Proc Natl
mellitus: insight for polypharmacology and Acad Sci U S A 109:7257–7262
drug design. J Chem Inf Model 53:1787–1803 41. Zhou W, Hevener KE et al (2009) A statistical
37. Baell JB, Holloway GA (2010) New substruc- framework to evaluate virtual screening. BMC
ture filters for removal of pan assay interference Bioinformatics 10:225
compounds (PAINS) from screening libraries 42. Scior T, Bender A et al (2012) Recognizing
and for their exclusion in bioassays. J Med pitfalls in virtual screening: a critical review. J
Chem 53:2719–2740 Chem Inf Model 52:867–881
Chapter 5
Whole-Exome Sequencing (WES) for Illumina Short Read

Sequencers Using Solution-Based Capture
Milind C. Mahajan and Andrew S. McLellan
Abstract
Next-generation sequencing (NGS) is transforming clinical research and diagnostics, vastly enhancing our
ability to identify novel disease-causing genetic mutations and perform comprehensive diagnostic testing in
the clinic. Whole-exome sequencing (WES) is a commonly used method which captures the majority of
coding regions of the genome for sequencing, as these regions contain the majority of disease-causing
mutations. The clinical applications of WES are not limited to diagnosis; the technique can be employed to
help determine an optimal therapeutic strategy for a patient considering their mutation profile. WES may
also be used to predict a patient’s risk of developing a disease, e.g., type 2 diabetes (T2D), and can therefore
be used to tailor advice for the patient about lifestyle choices that could mitigate those risks. Thus, genome
sequencing strategies, such as WES, underpin the emerging field of personalized medicine. Initiatives also
exist for sharing WES data in public repositories, e.g., the Exome Aggregation Consortium (ExAC)
database. In time, by mining these valuable data resources, we will acquire a better understanding of the
roles of both single rare mutations and specific combinations of common mutations (mutation signatures)
in the pathology of complex diseases such as diabetes.
Herein, we describe a protocol for performing WES on genomic DNA extracted from blood or saliva.
Starting with gDNA extraction, we document preparation of a library for sequencing on Illumina instru-
ments and the enrichment of the protein-coding regions from the library using the Roche NimbleGen
SeqCap EZ Exome v3 kit; a solution-based capture method. We include details of how to efficiently purify
the products of each step using the AMPure XP System and describe how to use qPCR to test the efficiency
of capture, and thus determine finished library quality.
Key words NGS, WES, Exome-seq, Sequencing, AMPure XP, NEBNext, Illumina, qPCR
1 Introduction
Next-generation sequencing (NGS) technology emerged only a

decade and a half ago, but is fast becoming an essential tool in
research and the clinic. It aids in the discovery of genetic variation
that causes disease and can assist with diagnosis and disease preven-
tion. For type 2 diabetes (T2D) is a complex disease, which arises
out of the interaction of many genes and the environment
[1, 2]. Deep sequencing of exonic regions of the genome
85
86 Milind C. Mahajan and Andrew S. McLellan
employing NGS serves as a high precision tool for the identification

of new disease-causing variants, being capable of efficiently detect-
ing complex and rare variants that have not been previously uncov-
ered by non-NGS methods [1].
Since the publication of the human genome sequence in 2001
[3], the technology used to sequence DNA has evolved at an
incredible pace exceeding Moore’s law (an observation originally
used to define the exponential rate of increase in the number of
transistors on integrated circuits which double approximately every
2 years) [4, 5]. The first-generation (Sanger) sequencing-based
techniques used to sequence the human genome were slow, labori-
ous, and expensive to complete. The desire for cost-effective
cheaper and faster sequencing has brought us into the era of next-
generation sequencing (NGS) which has progressed to the point
that a human genome can be sequenced for as low as $1000 in less
than a week, compared to the estimated $3 billion accomplished in
about 13 years for the first human genome sequenced in 2003
[5]. Over the past decade, the enhancements and reduced cost of
sequencing technology and host of methods being developed
around it have unleashed significant potential for understanding
mechanisms of complex diseases, e.g., type 2 diabetes. While
genome-wide association studies (GWAS) have helped identify
over 153 variants from 120 loci which reliably associate with T2D
[1, 6], most of the heritable basis for T2D still remains undiscov-
ered and it was thought to be explained by rare variants with minor
allele frequencies (MAFs) <5% (omitted from GWAS studies)
[7]. NGS facilitates the contribution of rare variants to be assessed,
and a recent NGS-based study of more than 120,000 Europeans,
with and without T2D, has revealed that rare variants do not
provide a significant contribution to T2D development [8]. It
was concluded that most variation associated with the disease was
still mostly linked to common variants observed in the same regions
of the genome implicated by GWAS.
NGS technologies for assessment of genome variation, e.g.,
custom targeted sequencing, whole-exome sequencing (WES),
and whole genome sequencing (WGS), are becoming cheaper and
more commonly utilized in both research and clinical settings as the
cost per base decreases and the utility in discovering pathogenic
(disease-causing) variants and using them for genetic testing gains
traction [9]. NGS techniques are capable of uncovering any type of
genomic variation, all of which have the potential to cause disease
by altering the structure and function of a protein product, or by
modifying the regulation of expression of a protein, which can lead
to loss or gain of function. The variant types include single nucleo-
tide variants (SNVs), e.g., an A ! T mutation, small insertions and
deletions (indels), e.g., an AGT ! TAAA change, copy number
variants (CNVs), e.g., deletion or duplication of a genomic region
or even whole chromosomes, and other structural variants (SVs)
Whole Exome Sequencing (WES) for Illumina HiSeq 87
including genomic rearrangements. Custom targeted NGS is com-

monly used clinically for diagnostic sequencing, where a set of
known high impact pathogenic variants (typically rare with a popu-
lation frequency of <1%) can be screened in a patient at low cost.
For example, a targeted NGS-based diagnostic test has been devel-
oped to detect mutations in all genes known to cause monogenic
and neonatal diabetes [10]. This technology typically uses hybridi-
zation capture or amplicon-based methods to enrich only the
regions of the genome of interest. Hybridization capture-based
methods use libraries of synthesized DNA or RNA oligonucleotide
sequences, called “probes” or “baits,” that span the genomic
regions to be sequenced [11]. They can be either attached to
magnetic beads (solution-based hybridization) or a solid surface
(array-based hybridization) to capture complementary DNA from
ultrasonically or enzymatically fragmented genomic DNA
[11, 12]. “Solution”-based hybridization capture is a more recent
development, requiring less input DNA, being more efficient and
enabling capture of smaller genomic fragments [11, 12]. Newer
“amplicon”-based methods simplify preparation and require even
less starting material, but a recent study showed that they can miss
variants detected by traditional hybridization capture methods
[12]. Costs are reduced by combining many samples within a single
sequencing reaction (multiplexing). Of course, use of targeted
sequencing comes with the caveat of potentially missing causative
mutations, because only a limited portion of the genome is
sequenced, although, in a clinical setting, restricting the genomic
regions being analyzed leads to less risk of incidental findings
(observing potentially disease-causing mutations that were not
part of the contracted test) which may have ethical implications.
WES is a specific implementation of targeted sequencing, tar-
geting 95% of the protein-encoding parts of the genome [13]. This
helps with the discovery of causative mutations, but is more expen-
sive and involves a greater burden of analysis, increased computa-
tion, and greater storage needs. In a clinical setting, WES comes
with challenges with respect to filtering the variants so that all the
potentially pathogenic and uncertain variants are retained while the
majority of benign variants are discarded; a signal to noise
ratio problem. An example of WES being used in the clinic is
documented by Bonneford et al. [14] who describe its application
in the discovery of a novel non-synonymous mutation in the
ABCC8 gene using DNA from a patient presenting with
non-autoimmune neonatal diabetes mellitus (NDM). This finding
indicated that oral sulfonylurea may be effective as a treatment
instead of insulin therapy. In a recent clinical case, a family study,
WES was used to study the heritability of T2D in a family from
which three brothers and their mother were affected, implicating
four genes in their disease pathology [15].
To maximize the chances of discovering disease-causing genetic

aberrations, WGS is the technique of choice [9, 13, 16]. Targeted
sequencing and WES may indicate the existence of SVs and large
genomic rearrangements, e.g., CNV detection is possible; however,
WGS is also capable of accurately describing the precise nature of
the variation. Moreover, capture methods require greater coverage
(the number of uniquely captured DNA fragments that overlap a
genomic region) in the order of two- to threefold at each locus to
achieve a sensitivity comparable to WGS [11, 13]. This results from
capture bias, as baits have different target binding affinities and
coverage loss may occur due to off-target binding [13]. In a recent
comparison using a cohort of more than 500 patients, WGS was
pitted against a commonly used targeted NGS test for the diagnosis
of inherited retinal disease (IRD) [16]. WGS increased diagnostic
yield by 29% by detecting disease-causing variants that were missed
by targeted NGSs. An obvious limitation of WGS, given the enor-
mous amount of sequencing required, is that the cost of both
sequencing and analysis is significantly higher for WGS. Costs of
data storage also increase as more raw data is collected, and depend-
ing on local regulations, in clinical settings, all raw data may need to
be archived for in excess of a decade. Over time, however, decreas-
ing sequencing costs, improved analytical algorithms, faster com-
putation, and cheaper storage will lead to the increasing use of
WGS. At this juncture, as the exome represents approximately
1–2% of the genome but contains an estimated 85% of pathogenic
variants [17, 18], WES is, therefore, a good compromise for most
purposes.
The NGS techniques described are currently employed in clini-
cal use for diagnosing simple (Mendelian) diseases, e.g., monogenic
diabetes [14, 19]. For complex diseases, such as T2D, the complex
interplay of myriad genetic and environmental factors means that
NGS is not yet readily amenable to diagnostic use [20]; neverthe-
less, employment of NGS is already advancing our ability to under-
stand the genetic basis of the disease [1, 8, 15]. With continually
increasing amounts of NGS data becoming available for bioinfor-
matic interrogation, improved classification of disease subtypes and
development of better artificial intelligence (AI) approaches to
elucidate predictive signals from the data, it may one day be possible
to use a patient’s WES or WGS data routinely in the clinic to help
target lifestyle advice to genetically at risk but asymptomatic indi-
viduals and prescribe the most appropriate treatments for patients
as part of a personalized medicine strategy [1, 17, 18]. Two such
data repositories are maintained by The Exome Aggregation Con-
sortium (ExAC) [21] and the T2D-GENES (Type 2 Diabetes
Genetic Exploration by Next-generation sequencing in multi-
Ethnic Samples) consortium [8]. The latter aims to identify genetic
variants for T2D through multiethnic NGS, and are performing
deep WES in 10,000 people from five ethnicities. The more WES
datasets annotated and placed into the public domain, the more
accurate predictions of pathogenicity become.
WES is commonly performed using short-read sequencing
using Illumina platforms (NextSeq, HiSeq series, and NovaSeq
series), which use the sequencing by synthesis (SBS) method
[22, 23]. Sequencing may be up to 150 bp, but can be performed
“paired-end” meaning that both ends of a DNA fragment are
sequenced, aiding alignment. Illumina are the market leader in
short read NGS technologies [22, 23], but there are other plat-
forms available which have their strengths and weaknesses. For
example, short read sequencing has limited use in de-novo assem-
bly or for resolving large structural variants and complex or repeti-
tive regions. Long-read sequencing platforms, e.g., the Pacific
BioSciences RSII and Sequel, excel in this space, being able to
sequence regions 8–20 kb in length [22, 23].
We herein provide a working protocol for the preparation of a
gDNA library followed by exome enrichment using the Roche
NimbleGen SeqCap v3 kit to obtain a final WES library designed
for sequencing on the Illumina family of sequencers. A diagram of
the entire workflow is shown in Fig. 1. This protocol is by no means
a promotion of the reagents and kits used, but is intended as a
general practical guide that can be used for any WES library prepa-
ration kit of choice. Increasingly, kits for library preparation are
becoming available that provide faster turnaround time with
reduced amplification cycles and which combine several reactions
into one step. We have chosen the NEBNext kit, which yields
excellent quality WES libraries compatible with the Illumina
HiSeq and NovaSeq sequencers (we currently use the NovaSeq).
This protocol is designed to be used for the preparation of WES
libraries from 10 to 12 DNA samples simultaneously using Eppen-
dorf tubes. However, the protocol can easily be adapted for high-
throughput WES library preparation in a 96-well format, using
liquid handlers and automation.
2 Materials
2.1 Purification 1. AMPure XP System (SPRI beads) (Beckman Coulter, Brea,

and Size Selection CA, USA).
of DNA 2. SPRI binding buffer (20% PEG, 2.5 M NaCl): Prepare 40%
PEG8000 by weighing out 16 g of PEG into a sterile 50 mL
conical tube. Bring the volume up to 40 μL with nuclease-free
water. Shake on a plate shaker machine for 4 h, or overnight.
Mix equal volumes of 5 M NaCl and 40% PEG for 20% PEG,
2.5 M NaCl solution. Vortex well and allow any air bubbles to
settle before using. Store at 4 C and bring to room tempera-
ture before use.
Fig. 1 A diagram to summarize the workflow of the whole-exome capture procedure
3. SPRI elution buffer: 20 mM Tris-HCl, pH 8 or can use Qiagen

Buffer EB (Qiagen, Germantown, MD, USA): 10 mM Tris-
HCl, pH 8.5.
4. Magnet stand (Life Technologies Ref # 12027).
5. Covaris sonicator (LE220) (Covaris, Woburn, MA, USA).
6. Bioanalyzer 2100 (Agilent, Santa Clara, CA, USA).
2.2 Preparation 1. NEBNext DNA Library Prep Master Mix Set for Illumina
of the gDNA (NEB, Ipswich, MA).
Sequencing Library 2. KAPA HiFi HotStart ReadyMix (KAPA Biosystems, Wilming-
ton, MA, USA): KAPA HiFi HotStart DNA Polymerase in a
proprietary reaction buffer, dNTPs (0.3 mM of each dNTP at
1), MgCl2 (2.5 mM at 1) and stabilizers.
3. Qubit dsDNA BR Assay Kit (ThermoFisher Scientific, Wal-
tham, MA, USA).
2.3 Exonic Sequence 1. Roche NimbleGen SeqCap EZ Exome v3 (Roche,

Library Capture Indianapolis, IN, USA).
and Quality 2. COT Human DNA, Fluorometric Grade 1 mg/mL, 1 mL
Assessment (Roche, Indianapolis, IN, USA).
3. Dynabeads M-270 Streptavidin (ThermoFisher Scientific, Wal-
tham, MA, USA).
4. NanoDrop ND-1000 (ThermoFisher Scientific, Waltham,
MA, USA).
5. Qubit Fluorometer (Thermofisher Scientific, Waltham, MA,
USA).
6. SYBR Green PCR Master Mix (Roche Molecular Systems, Ct #
04707516001).
7. ABI 7900HT Real-Time PCR System (Applied Biosystems,
Foster City, CA, USA).
3 Methods
3.1 Purification Purification and size selection of DNA can be carried out by con-
and Size Selection ventional agarose gel chromatography. However, it is a
of DNA During WES low-throughput and time-consuming process that requires a large
Library Preparation amount of input DNA and the gel extraction process results in the
Steps loss of a large proportion of the gel-isolated fraction. Thus, Solid
Phase Reversible Immobilisation (SPRI) beads are used to achieve
efficient purification and size selection of nanogram to microgram
levels of DNA [24]. SPRI beads are paramagnetic, which prevents
them from clumping and falling out of solution. Each bead is made
of polystyrene surrounded by a layer of magnetite and is coated
with carboxyl molecules that are variably protonated/deprotonated
depending on pH. These carboxylic groups reversibly bind DNA in
the presence of polyethylene glycol (PEG) and salt (20% PEG,
2.5 M NaCl) binding buffer and can be eluted by 10–20 mM
Tris-HCl pH 8–8.5. This system is ideally suited to purifying
from low concentrations of DNA, is easy to use, and is amenable
to automation, facilitating high-throughput DNA purification and
size selection (see Note 1).
3.1.1 Random Classically, fragmentation of gDNA has been achieved via physical
Fragmentation of Genomic shearing of the DNA, which yields unbiased random DNA frag-
DNA by Sonication ments representing the whole genome in the final library. Recently,
however, several vendors including Kapa Biosystems and NEB have
developed enzymatic shearing protocols, using endonucleases to
create double-stranded breaks in the DNA. The advantage of enzy-
matic fragmentation is that it does not require any specialized
equipment; however, efficacy, consistency, and reliability of enzy-
matic shearing needs to be established. Hence, we describe Covaris-
based ultrasonic shearing of DNA for the preparation of WES
libraries.
The Covaris sonicator transmits focused acoustic energy waves
to randomly fragment DNA (by physical shearing). Prior to use, the
sonicator must be set up (see Note 2).
1. Transfer 0.5–3 μg of gDNA in 120 μL of nuclease-free water
into Covaris MicroTubes.
2. Vortex MicroTubes containing DNA samples briefly, then cen-
trifuge to remove air bubbles (see Note 3). If air bubbles
remain, a second centrifugation step may be necessary.
3. Place the MicroTubes in the MicroTube rack (see Note 4), then
place the rack in the sonication chamber. We use the following
parameters to obtain a DNA size range of 150–400 base pairs,
with mean peak size in the range of 200–250 base pairs:
time ¼ 600 s, duty ¼ 5%, intensity ¼ 2, cycle per burst ¼ 50
(see Note 5).
4. After fragmentation is complete, remove the shearing plate
from the sonicator, wipe the water from the outside of the
tubes and pulse spin in a centrifuge.
5. For each sample, transfer the sheared DNA to a 1.5 mL
Eppendorf tube.
6. Assess the quality of the sheared genomic DNA using an Agi-
lent 2100 Bioanalyzer: Run sheared DNA on the DNA HS
chip using the Bioanalyzer to determine the size of sheared
DNA (Fig. 2).
7. You may pause the protocol at this point (see Note 6).
3.1.2 Purification 1. AMPure XP beads should be removed from storage at 4 C, at

of Sheared DNA least 30 min prior to use, to allow them to equilibrate to room
temperature.
2. Mix the reagents well so that the bead suspension appears
homogeneous.
3. Aliquot 216 μL of completely resuspended room temperature
AMPure XP beads to an Eppendorf tube (one tube per
sample).
Fig. 2 Determining optimal shearing of DNA using the Bioanalyzer. (a) Bioanalyzer trace of the input DNA,
showing size and integrity. (b) Optimally sheared gDNA. (c) Poorly sheared gDNA
4. For each sample, transfer the total volume of fragmented DNA

(0.5–3 μg in ~119 μL) from the Covaris tube into an Eppen-
dorf tube containing AMPure XP beads (the volume of beads
added is 1.8 the original DNA volume).
5. Mix well and incubate at room temperature for 5 min with
intermittent mixing using a pipette to keep the beads in sus-
pension throughout.
6. Spin the tubes for 1 min at 3000 g in a microcentrifuge, then
place them in a magnetic stand for 3 min or until the solution is
clear.
7. Without disturbing the beads, remove 330 μL of the superna-
tant and discard (see Note 7).
8. Keeping the tubes in the magnetic stand, dispense 400 μL of
freshly prepared 70% ethanol into each and incubate for 30 s.
9. Remove and discard the ethanol and any remaining
supernatant.
10. Dry the beads at 37 C for 5 min or until the residual ethanol
has completely evaporated.
11. Add 42.5 μL of SPRI elution buffer and resuspend the beads by
gently pipetting.
12. You may pause the protocol at this point (see Note 8).
3.2 Preparation The NEB Next™ DNA Sample Prep kit is used for constructing the
of the Whole Genome gDNA library from the fragmented DNA. This kit contains master
DNA (gDNA) mixes of enzymes and reagents for end repair of fragmented DNA,
Sequencing Library dA-tailing, ligation of synthetic oligonucleotide adaptors and
amplification of the adaptor-ligated DNA, using adaptor-specific
barcoded PCR primers, to obtain the final gDNA library. After each
of these four steps, DNA is purified using AMPure beads (see Note
9), selecting a binding buffer with the appropriate concentration of
PEG and NaCl (as detailed above).
3.2.1 End Repair Bead-bound purified sheared DNA from step 11 above (Subhead-
of Sheared DNA ing 3.1.2) is used to create blunt-ended DNA by end repair.
1. For each sample, set up the end repair reaction mix as follows:
5 μL of NEB Next End Repair Reaction Buffer (10), 2.5 μL
of NEBNext End Repair Enzyme Mix, 42.5 μL of AMPure
bead-bound sheared DNA (50 μL total).
2. Incubate in a thermocycler for 30 min at 20 C to complete the
end repair reaction.
3.2.2 Cleaning Up Prior to using the end-repaired DNA, it must first be purified as
End-Repaired DNA follows:
1. Add 90 μL of SPRI binding buffer (20% PEG-2.5 M NaCl)
solution to the 50 μL of end repair reaction mix to bind the
DNA back to the beads.
2. Pipette up and down 20 times to resuspend the beads and
incubate at room temperature for 5 min.
3. Spin the tube for 1 min at 3000 g in a microcentrifuge. Place
the tube on the magnetic stand for 3 min or until the solution is
clear.
4. Without disturbing the beads, remove 135 μL of supernatant
and discard. A small volume of supernatant will be left in
each well.
5. Continue to keep the tube on the magnetic stand and add
200 μL of freshly prepared 70% ethanol, incubate for 30 s
then discard the ethanol.
6. Repeat step 5 for a total of two ethanol washes.
7. Dry the samples at room temperature for 10 min or until the
residual ethanol completely evaporates.
8. Add 42 μL of SPRI elution buffer and pipette up and down
15 times to resuspend the beads. The bead-bound DNA is now
ready for dA-tailing.
9. You may pause the protocol at this point to continue next day
or proceed to the dA-tailing step described below in Subhead-
ing 3.2.3 (see Note 8).
3.2.3 Addition 1. Prepare the dA-tailing reaction mix using reagents from the
of Deoxyadenine to the 30 NEBNext DNA Library Prep Master Mix Set for Illumina as
End of Blunt-Ended DNA follows: 5 μL of NEBNext dA-Tailing Reaction Buffer (10),
(dA-Tailing) 3 μL of Klenow Fragment (30 ! 50 exo), 42 μL of DNA-bead
mix after end repair step (total of 50 μL).
2. Incubate the reaction mix at 37 C for 30 min to obtain DNA
fragments with dA-overhangs on the 30 end.
3.2.4 Cleaning Up 1. Add 90 μL of 20% PEG, 2.5 M NaCl to ~50 μL of dA-tailed

the dA-Tailed DNA DNA reaction mix, then mix the contents thoroughly using a
pipette and incubate at room temperature for 5 min.
2. Spin tube at 3000 g in a microcentrifuge. Place the tube on
the magnetic stand for 3 min or until the solution is clear.
each well.
6. Dry the samples at 37 C for 5 min or until any residual ethanol
7. Add 25 μL of Qiagen elution buffer and pipette up and down
15 times to resuspend the beads. You should proceed directly
to the next step (see Note 10).
3.2.5 Ligation of Illumina 1. Prepare the Adaptor Ligation reaction mix as follows: 10 μL of
Adaptors to dA-Tailed DNA NEBNext Quick Ligation Reaction Buffer (5), 10 μL of
DNA Multiplex Adaptor (10 μM), 5 μL of Quick T4 DNA
Ligase, 25 μL of Quick T4 DNA Ligase (total volume of
50 μL). Adaptor sequences are shown in Table 1.
2. Incubate for 60 min at 20 C to allow ligation to occur.
3.2.6 Cleaning Up 1. Add 60 μL of SPRI binding buffer to 50 μL of adaptor ligation

and Size Selection reaction mix to get to 1.2 ratio of SPRI binding buffer/beads
of the Adaptor-Ligated DNA to ligation mix (see Note 11).
2. Mix well by briefly vortexing each tube then incubate at room
temperature for 5 min. Place the tubes in the magnetic stand
for 3 min or until the solution becomes clear.
3. Without disturbing the beads, remove the supernatant and
discard. A small volume of supernatant will be left in each well.
6. Dry the sample at 37 C for 5 min or until any residual ethanol
7. Add 20 μL of nuclease-free PCR grade water and pipette up
and down 15 times to resuspend the beads and incubate for
5 min at room temperature, pipetting intermittently to keep
the beads in suspension.
Table 1
Illumina Adaptor sequences ligated to dA-tailed gDNA during library preparation (rows 1 and 2)
Primer name Sequence (50 –30 ) Direction

Multiplex adap. 1 P-GATCGGAAGAGCACACGTCT
Multiplex adap. 2 ACACTCTTTCCCTACACGACGCTCTTCCGATC∗T
PE PCR primer-F AATGATACGGCGACCACCGAGATCTACACTCTTTCCC F
TACACGACGCTCTTCCGATCT
PE PCR primer-R CAAGCAGAAGACGGCATACGAGAT[CGTGAT] R
GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT
PE PCR primer AATGATACGGCGACCACCGAGATCTACACTCTTTCCC F
F-BLOCK TACACGACGCTCTTCCGATCT
Idx01-Long- CAAGCAGAAGACGGCATACGAGAT[CGTGAT∗] R
BLOCK GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT
IS 5 AATGATACGGCGACCACCGA F
IS 6 CAAGCAGAAGACGGCATACGA R
NEBNext Universal PCR Primer (F) and an example NEBNext Index Primer (R) (rows 3 and 4). Blocking oligos used
during exome-capture step (rows 5 and 6). PCR primers used for the post-capture amplification of WES library (rows
7 and 8). The index sequence of the blocking oligos is the same as that used for the NEBNext Index Primer in the gDNA
library preparation for a given sample. Index sequences are shown within square brackets. ∗indicates a phosphorothioate
bond
8. Pulse spin in a microcentrifuge, then place the tube on the

magnetic stand for 3 min or until the solution is clear.
9. Transfer the eluate (~20 μL) to a new PCR tube and discard the
beads.
3.2.7 Amplification The quantity of adaptor-ligated gDNA library is currently too small
of the gDNA Library to proceed with exome capture by solution hybridization. It must
first be amplified by a few PCR cycles as necessary to obtain a yield
of 500 ng.
1. Prepare the PCR mix in a pre-PCR hood (to reduce the risk of
contamination with aerosolized PCR products) as follows:
25 μL KAPA HiFi HotStart ReadyMix, 2.5 μL of forward
PCR primer (10 μM), 2.5 μL of indexed reverse primer
(10 μM), 20 μL of adaptor-ligated DNA (50 μL total volume).
PCR primer sequences are shown in Table 1.
2. Set up the PCR reaction on a thermocycler using the following
program: 98 C for 5 min (denaturation), followed by 8 cycles
of 98 C for 1 min, 60 C for 30 s and 72 C for 45 s (amplifi-
cation), then finish with 72 C for 10 min (extension). Hold at
4 C until ready for the next stage.
3. Clean up 50 μL of PCR amplified DNA using 60 μL of AMPure
XP beads as described in Subheading 3.2.6 (see Note 12).
Fig. 3 A Bioanalyzer trace showing the profile of an optimally prepared genomic DNA library
4. Elute DNA in 30 μL of SPRI elution buffer.

At this point the gDNA library is ready for exome enrichment,
but, before proceeding, it is necessary to estimate the quantity and
size of the gDNA library on the Bioanalyzer and verify that the PCR
primers have been removed. Note the following features on the
electropherogram provided by the Bioanalyzer:
l Concentration of each library.
l Average library size.
l Any other peaks besides the library.
For most samples, the majority of the gDNA obtained after
initial shearing should have been sized between 200 and 250 base
pairs (Fig. 2b). Thus, the mean peak size of the finished library will
be around 325–350 base pairs, which is suitable for sequencing
(Fig. 3). If, for some reason, the library exhibits a broad molecular
weight range or shows a peak greater than 400 base pairs (Fig. 4),
then use the following protocol to remove high molecular weight
DNA from the gDNA Seq library.
3.2.8 Removal of High 1. Remove a stock tube of AMPure XP beads from storage at 4 C
Molecular Weight DNA and allow to equilibrate to room temperature.
from the gDNA Library 2. Centrifuge the sample plate at 300 g for 30 s to ensure all
components are collected at the bottom of each well.
3. To 30 μL of PCR amplified WES library, add 19.5 μL of
AMPure beads. This is a 0.65 ratio of beads to sample
Fig. 4 A Bioanalyzer trace obtained before removal of high molecular weight DNA from a genomic DNA library
volume, optimal for binding the high molecular weight mate-

rial in the library.
4. Pipette up and down 20 times to resuspend the beads.
5. Incubate at room temperature for 5 min with intermittent
mixing by gentle pipetting.
6. Spin the tube for 1 min at 3000 g in a microcentrifuge then
place the tube on the magnetic stand for 3 min or until the
solution is clear.
7. Transfer the supernatant containing unbound ~350 base pairs
of WES library DNA to a new Eppendorf tube. Add 16.5 μL of
AMPure beds to the DNA in the supernatant to obtain a 1.2
ratio of beads to sample volume.
8. Repeat steps 5 and 6.
the tube.
12. Dry the samples in an incubator at 37 C for 5 min or until the
residual ethanol has evaporated.
13. Add 20 μL of nuclease-free PCR grade water and pipette up
and down 15 times to resuspend the beads.

15. Transfer the eluate (~20 μL) to a new 1.5 mL Eppendorf tube.
Run some of the eluted DNA on a Bioanalyzer for size deter-
mination and estimation of concentration. This is the final
gDNA library ready for exome enrichment (Fig. 3).
16. Store the gDNA library overnight at 20 C or until needed
for the exome enrichment process described as in Subheading
3.3 below.
3.3 Enrichment The gDNA library prepared as described above comprises sequence
of Exonic Content content from the entire genome out of which only ~2% belongs to
of the gDNA Library the coding portion of the genome. To isolate/enrich this relatively
(Exome Enrichment) tiny fraction of the total gDNA library, a solution hybridization-
based capture method can be employed. There is an assortment of
vendors producing kits for generating capture libraries for WES and
most use tiled or overlapping biotinylated DNA or RNA capture
probes (baits) and streptavidin-coated magnetic beads to isolate the
fragments containing exonic sequence. Of these, the most popular
kits are produced by Illumina, Roche NimbleGen, Agi-
lent, Integrated DNA Technologies (IDT) and Twist Bioscience.
Illumina offers three kits based on the Nextera Rapid Capture
enrichment method. The Nextera Rapid Capture Exome kit targets
all coding sequence (~37 megabases), the Nextera Rapid Capture
Expanded Exome kit also captures non-coding regions, UTRs and
miRNAs (~62 megabases) and the TrueSight One kit captures a
reduced set of exons (~12 megabases) that have been implicated in
at least one disease. Roche NimbleGen provides a similar selection
with its SeqCap EZ Exome Library v3.0 (~64 megabases), SeqCap
EZ Exome + UTR Library (~96 megabases), and SeqCap EZ
HGSC VCRome (~45 megabases) solution-based capture kits.
Agilent’s SureSelectXT Human All Exon V6 (~58 megabases),
SureSelectXT Human All Exon V6 + UTR (size not specified),
and SureSelectXT Focused Exon Capture (~12 megabases) are
equivalent. IDT offers the xGen Exome Research Panel (~39Mb),
and several disease specific and custom targeted panels, boasting a
comprehensively high on-target rate and uniform coverage, even
across GC-rich regions. Twist Bioscience is the latest entrant in the
growing list of vendors offering whole exome and targeted capture
solutions. It offers the Human Core Exome Kit (~33 Mb of highly
conserved protein-coding regions) and custom panels. The com-
pany has developed high-fidelity double stranded oligos which
are uniquely designed to capture both DNA strands, providing
enhanced specificity and uniformity across targeted regions without
unexpected dropouts. Some of these manufacturers also allow for
the addition of custom probes to the exome panel. While there are
widely reported advantages and disadvantages to specific kits
[17, 25] in general they appear to perform similarly and it is
important to realize that no single kit has complete coverage of all

exons. It is difficult to use comparisons reported in the literature as
the results tend to vary considerably, particularly as kits are mod-
ified and improved over time; however, the choice of kit should
include consideration of relative coverage of exome databases, tar-
get coverage sensitivity, GC bias (which is traditionally more pro-
nounced in Illumina Nextera kits as they exhibit an increase in read
depth for sequences with at least 60% GC content [25]), and
sensitivity in detecting single nucleotide variants (SNVs) and indels.
Other factors to take into account include platform compatibility,
cost, fragmentation preference (enzymatic or sonication), required
input DNA quantity and amenability toward automation should
this feature be desirable.
Here we describe exome enrichment procedure using the
Roche SeqCap EZ Exome v3 reagent kit, a solution-based capture
method using a 64 megabase sequence capture design based on the
human genome build GRCh37/hg19 [26].
3.3.1 Hybridization Before commencing the exome capture procedure it is necessary to

of gDNA Library complete the following steps:
to the Exome Capture
1. Set a heat block to 95 C and allow to equilibrate.
System
2. Remove an appropriate number of 4.5 μL SeqCap Exome
Oligo pool aliquots (1 per capture) from 65 C storage and
allow them to thaw on ice. Thaw freezer stored COT DNA and
Blocking Oligos (Table 1) to room temperature.
3. Set up a pre-hybridization as follows: 1 μL of COT DNA
(1 mg/mL), 1 μg forward blocking oligo (1 mM), 1 μL of
indexed reverse blocking oligo (1 mM) and 500–750 ng of
gDNA library in 10–15 μL (as in Subheading 3.2.7).
4. Lyophilize the amplified sample library/COT DNA/PE-HE
Oligos in a DNA vacuum concentrator at 65 C (see Note 13).
5. Add 7.5 μL of 2 SC Hybridization Buffer and 3 μL of SC
Hybridization Component.
6. Cap and vortex the tube for 10 s then microcentrifuge for 10 s.
7. Place the samples in a 95 C heat block for 10 min to denature
the DNA, then pulse-spin in a microcentrifuge.
8. Transfer the ~10.5 μL of amplified sample library/COT
DNA/PE-HE Oligos/Hybridization cocktail to the 4.5 μL
aliquot of SeqCap EZ Exome v3 oligonucleotide pool that
has been thawing on ice.
9. Vortex the contents of the tube briefly and centrifuge at maxi-
mum speed for 10 s.
10. Incubate in a thermocycler (with a heated lid) at 47 C for
64–72 h (see Note 14).
Table 2
Preparation of wash buffers for washing and recovery of hybridized samples using Streptavidin
Dynabeads
Volume of concentrated Volume of water

Stock concentrated buffer buffer for 1 (μL) for 1 (μL) Total volume (μL)
10 Stringent wash buffer (vial 4) 40 360 400
10 Wash buffer I (vial 1) 30 270 300
10 Wash buffer II (vial 2) 20 180 200
10 Wash buffer III (vial 3) 20 180 200
2.5 Bead wash buffer (vial 7) 200 300 500
3.3.2 Wash and Recovery 1. Dilute stock Stringent Wash Buffer, SC Wash Buffers (I, II, and
of Hybridized Samples III), and Bead Wash Buffer to create 1 working solutions as
shown in Table 2. The values in the table are for a single exome
Step 1. Prepare Sequence capture. For more exome captures, simply multiply each vol-
Capture Wash Buffers ume by the number of samples (see Notes 15–17).
2. Preheat 1 Stringent Wash Buffer and one-third of the volume
of 1 SC Wash Buffer I to 47 C in a water bath.
Step 2. Prepare 1. Allow the Streptavidin Dynabeads M270 to equilibrate at room

the Streptavidin Dynabeads temperature for 20 min prior to use.
2. Mix the beads thoroughly by vortexing for 1 min.
3. Aliquot 100 μL of beads per capture into a single 1.5 mL tube
(i.e., for one capture use 100 μL of beads and for four captures
use 400 μL of beads—see Note 18).
4. Centrifuge the 1.5 mL tube at maximum speed for 10 s in a
microcentrifuge, then place the tube in a magnetic stand. When
the beads separate and liquid becomes clear, carefully remove
and discard the liquid without disturbing the beads. Any
remaining traces of liquid will be removed with subsequent
wash steps. Proceed immediately to the next step (see Note 19).
5. Take the 1.5 mL tube off the magnetic stand and add twice the
initial bead volume of 1 Bead Wash Buffer (i.e., for one
capture add 200 μL of buffer and for four captures add
800 μL of buffer).
6. Vortex for 10 s, then centrifuge at full speed for 10 s.
7. Return the tube to the magnetic stand to bind the beads.
Again, once clear, remove and discard the liquid.
8. Repeat steps 5–7 for a total of two washes.
9. After removing the buffer following the second wash, resus-

pend the beads by vortexing in 1 the original volume using
the Streptavidin Dynabead Binding and Wash Buffer that was
prepared in Step 1 (i.e., for one capture use 100 μL of buffer
and for four captures use 400 μL of buffer).
10. Aliquot 100 μL of resuspended beads into a new individual
PCR tube or strip tube.
11. Use the 96-well magnetic plate to bind the beads by placing the
tube on the magnet for 5 min. Remove and discard the liquid
when clear.
12. The Streptavidin Dynabeads are now ready to bind the cap-
tured DNA. Remove the tubes containing the beads from the
magnetic plate.
Step 3. Bind DNA 1. Transfer the ~15 μL of amplified sample library/COT DNA/
to the Streptavidin PE-HE Oligos/Hybridization cocktail (which should have
Dynabeads been hybridized for 64–72 h) to the Streptavidin Dynabeads
prepared in step 2.
2. Mix thoroughly by pipetting up and down 10 times.
3. Bind the captured sample to the beads by placing the tubes
containing the beads and DNA in a thermocycler at 47 C for
45 min. Mix the samples by vortexing for 3 s every 15 min to
ensure that the beads remain in suspension (see Note 20).
Step 4. Wash 1. If more than eight captures are being washed, transfer the
the Streptavidin Dynabeads entire contents (~115 μL) of each tube to a 96-deep well
Plus Bound DNA plate. If less than eight captures are being washed, samples
may remain in strip tubes.
2. Place the strip tube or 96-deep well plate on the magnet.
Remove the supernatant immediately once clear.
3. Add 100 μL of SC Wash Buffer I preheated to 47 C (see Note
21).
4. Cap tubes or seal plate and mix by vortexing for 10 s.
5. Place the tubes or plate on the magnet to bind the beads.
Remove and discard the liquid immediately once clear.
6. Remove the tubes or plate from the magnet and add 200 μL of
Stringent Wash Buffer preheated to 47 C. Pipette up and
down 5 times to mix (see Note 21).
7. Incubate at 47 C for 5 min. If captures are in a 96-deep well
plate, use an incubation chamber. If samples are in a strip tube,
use a PCR block with heated lid.
8. Repeat steps 5–7 for a total of two washes with Stringent Wash
Buffer heated to 47 C.
9. Place the tubes or plate on the magnet to bind the beads. After
5 min remove and discard the supernatant.
10. Add 200 μL of room temperature SC Wash Buffer I and mix by
vortexing for 2 min. If liquid has collected in the tube’s cap,
pulse-spin in a microcentrifuge to collect the liquid into the
tube’s bottom before continuing to the next step.
Remove and discard the liquid once clear.
12. Add 200 μL of room temperature SC Wash Buffer II and mix
by vortexing for 1 min. Pulse-spin in a microcentrifuge if
necessary.
14. Add 200 μL of room temperature SC Wash Buffer III and mix
by vortexing for 30 s. Pulse-spin in a microcentrifuge if
necessary.
16. Remove the tubes or plate from the magnet and add 20 μL of
PCR grade water to each tube of bead-bound exome-captured
library (see Note 22).
3.4 Final The quantity of exome enriched library is usually less than 1 ng,
Amplification which is too low to be used for sequencing on the Illumina
of Captured Library sequencer directly and must be amplified using a limited number
of PCR cycles.
1. Generate a PCR Master Mix by adding the following compo-
nents: 25 μL of KAPA HiFi HotStart ReadyMix, 2.5 μL of IS
5 Primer (10 μM), 2.5 μL of IS 6 Primer (10 μM), 20 μL of
exome captured library with beads (total volume of 50 μL). See
Table 1 for primer sequences.
2. Run the PCR reaction on a thermocycler using the following
program: 98 C for 10 min (denaturation), followed by
12 cycles of 98 C for 1 min, 60 C for 30 s and 72 C for
45 s, then finish with 72 C for 10 min. Hold at 4 C until
ready for the next stage.
3. Clean up the post-capture PCR product using AMPure XP
beads as shown in Subheading 3.2.6.
4. Elute WES library DNA in a final volume of 30 μL of nuclease-
free water.
5. Assess library size on a Bioanalyzer and quantify using the
Qubit dsDNA BR Assay Kit. On the Bioanalyzer, the trace
should look as shown in Fig. 5.
Fig. 5 A bioanalyzer profile of a successfully captured whole-exome library
3.5 Assessing To assess if the exonic regions of the gDNA library were success-
the Quality of Exonic fully enriched, qPCR can be performed using exon-specific primers
Enrichment in a WES for a couple of genes. Significant enrichment of targeted exonic
Library by qPCR regions in a WES library over the original gDNA library indicates a
successful outcome of the exome enrichment process described
above. Roche NimbleGen recommends verifying two exonic
regions, and also two intronic regions as controls. When running
qPCR for verification of exon enrichment, it is necessary to include
both the input gDNA Seq library (prepared in Subheading 3.2) and
the amplified capture library (prepared in Subheading 3.4). Every
qPCR assay must include standard curves for all four amplicons
(two exons and two introns) using a dilution series of wild-type
human genomic DNA, and a no template control (NTC) must be
included. The dissociation curves for all four amplicons must be
reviewed to ensure that no unexpected PCR products are gener-
ated. Compare the ratios of the results (Ct values) from qPCR of all
the exonic and intronic regions in the capture libraries vs. the input
libraries to determine whether enrichment of exonic regions has
been successful.
1. Determine the concentration of each of the samples using
either the Qubit or Bioanalyzer. Each 15 μL of qPCR reaction
will contain 0.5 ng of library DNA. Make working dilutions of
all libraries to a concentration of 0.1 ng/μL.
2. For quantitation of amplicon enrichment for a given qPCR
reaction, each qPCR assay must include a standard curve gen-
erated using a dilution series of normal human genomic DNA
with a concentration determined by fluorescence-based quan-

titation using a Qubit Fluorometer. Within each dilution must
be included 5 ng/μL of lambda phage DNA (non-human
carrier DNA) to ensure efficient PCR across the dilution series.
The dilution series should comprise the following gDNA con-
centrations: 5, 0.5, and 0.05 ng/μL. The negative control will
contain only the 5 ng/μL of lambda (carrier) DNA.
3. Prepare the qPCR mix as follows: 7.5 μL of SYBR Green PCR
Master Mix, 0.3 μL Forward/Reverse Primer Mix (5 μM), 5 μL
of template DNA (0.1 ng/μL) (total volume of 10 μL).
4. Run qPCR using any standard qPCR instrument using follow-
ing cycle conditions (in our lab we use the ABI 7900HT):
95 C for 10 min (denaturation step), followed by 40 cycles
of 95 C for 15 s and 60 C for 1 min (amplification step), then
1 cycle of 95 C for 15 s, 60 C for 15 s, and 95 C for 15 s
(final step to generate a dissociation curve).
The qPCR results obtained with two example samples: “WES
Seq library-1” and “WES Seq library-2” are summarized in Tables 3
Table 3
qPCR results for quality control analysis of an example WES library “WES-Seq library-1”
NSC-247.a (exon) PRKG.b (intron) THAP3.a (exon) THAP2.a (intron)

gDNA input Ct 29.06 28.88 29.58 30.35
Post-capture WES library Ct 21.32 30.98 20.34 30.80
delta Ct 7.74 2.10 9.23 0.45
delta Ct/3.32 2.33 0.63 2.78 0.14
Fold increase 213.88 0.23 604.41 0.73
The cycle parameters show the enrichment/depletion profile for the capture library using four different primer pair sets
(two exonic and two intronic)
Table 4
qPCR results for quality control analysis of an example WES library “WES-Seq library-2”
NSC-247.a (exon) PRKG.b (intron) THAP3.a (exon) THAP2.a (intron)

gDNA input Ct 28.89 29.51 29.58 30.12
Post-capture WES library Ct 20.86 29.66 20.92 31.51
delta Ct 8.03 0.15 8.65 1.39
delta Ct/3.32 2.42 0.04 2.61 0.42
Fold increase 262.59 0.90 403.59 0.38
The cycle parameters show the enrichment/depletion profile for the capture library using four different primer pair sets
(two exonic and two intronic)
and 4. The tables indicate the cycle enrichment/depletion for the

capture library for four different primer pair sets. For the regions to
be captured (exons) there is usually a 6–9 cycle difference or ~65 to
500-fold enrichment (a difference of 3.32 cycles is equivalent to a
tenfold enrichment). For regions that were not targeted in the
capture (introns), one should see either a decrease in representation
or little to no enrichment.
The results in Tables 3 and 4 show an increase of 200- to
600-fold in the exonic regions in the captured WES libraries,
indicating successful exome enrichment had been achieved. These
libraries are ready for sequencing (see Note 23).
4 Notes
1. Charge-dependent binding of DNA to SPRI beads is directly

proportional to the concentration of the PEG and NaCl in the
binding buffer (as the charge on a DNA molecule is directly
proportional to its length). Consequently, by altering the com-
position of the PEG and NaCl in the binding buffer, DNA
fragments within specific size ranges can be isolated using the
SPRI beads.
2. The Covaris sonicator must be turned on, chilled to 6–8 C and
degassed for at least 1.5 h before samples are sheared.
3. Centrifugation helps to remove air bubbles, which can lead to
uneven sharing of the DNA.
4. Position the MicroTubes in the bottom half of the MicroTube
rack and position the top half of the rack so that that the
samples are held tightly.
5. The timing and intensity of sonication needs to be standardized
to obtain sheared DNA of an appropriate size.
6. If you do not wish to continue to the next step immediately, the
samples may be stored at 4 C for 24 h or at 20 C for a longer
period.
7. A small volume of supernatant should remain to avoid aspira-
tion of the beads.
8. If you do not wish to continue to the next step immediately, the
samples may be stored at 4 C. Do not freeze the sample/bead
mixture.
9. To save on the cost of AMPure beads, and avoid loss of DNA
during each purification step and the size selection step, we add
AMPure-bound DNA to all the biochemical reactions
described. At the end of each step, a binding buffer containing
an appropriate amount of PEG and NaCl is added to bind the
biochemically modified DNA back to the beads, which are then
purified by placing in the magnetic stand and aspirating off the

spent reagents and buffer. The bead-bound purified DNA is
then used in the subsequent biochemical reaction step. We
refer to this modified protocol as the “on-bead” protocol.
10. Do not stop after dA-tailing the DNA. Long-term storage and
freeze-thawing causes the dA-tail to degrade. For optimal
results, proceed immediately to the adaptor ligation procedure.
11. A 1.2 ratio of SPRI binding buffer/beads to sample mix
ensures that only adaptor-ligated DNA will bind to the beads
and any non-ligated adaptors will remain in solution.
12. The 1.2 ratio of SPRI binding buffer/beads to sample mix
facilitates removal of any residual primer dimers.
13. Before lyophilizing the sample, cover the opening of the tube
with a piece of paraffin tape and make an X-shaped slit through
the tape using a clean blade/scalpel. This minimizes the risk of
contamination during concentration of the sample.
14. When hybridizing the capture probes to the genomic library, it
is preferable to set up the incubation on Friday and hybridize
over the weekend to maximize productivity.
15. Be sure to generate enough stock of sequence capture wash
buffers for 1–2 extra reactions, or 10% extra, to ensure there is
enough for all samples.
16. Diluted 1 sequence capture wash buffers may be stored at
room temperature for up to 2 weeks.
17. 1 sequence capture wash buffers must always be equilibrated
to 47 C before use.
18. Enough Streptavidin Dynabeads for six captures can be
prepared in a single tube.
19. Do not allow the Streptavidin Dynabeads to dry out. Small
amounts of residual Streptavidin Dynabead Binding and
Wash Buffer will not interfere with the binding of DNA to
the Streptavidin Dynabeads.
20. It is helpful to have a vortex mixer located close to the thermo-
cycler for this step.
21. Work quickly so that the temperature does not drop much
below 47 C.
22. There is no need to elute DNA off the beads. The beads plus
captured DNA will be used as the template in the following
steps.
23. If the calculated fold enrichment values are less than 50-fold,
the exome capture (i.e., enrichment) has probably failed. If this
is the case, the whole-exome library preparation must be
repeated.
References
1. Prasad RB, Groop L (2015) Genetics of type mellitus using next-generation sequencing of
2 diabetes—pitfalls and possibilities. Genes 6 the whole exome. PLoS One 5(10):e13630
(1):87–123 15. Shaer N, Khan JJ, Dallol A, Abuzenadah A
2. Murea M, Ma L, Freedman BI (2012) Genetic (2014) Diabetes personalized: the use of
and environmental factors associated with type whole-exome sequencing in order to identify
2 diabetes and diabetic vascular complications. familial type 2 diabetes mellitus susceptibility
Rev Diabet Stud 9(1):6–22 factors. BMC Genomics 15(Suppl 2):P51
3. Lander ES, Linton LM, Birren B et al (2001) 16. Ellingford JM, Barton S, Bhaskar S et al (2016)
Initial sequencing and analysis of the human Whole genome sequencing increases molecular
genome. Nature 409(6822):860–921 diagnostic yield compared with current diag-
4. Muir P, Li S, Lou S et al (2016) The real cost of nostic testing for inherited retinal disease.
sequencing: scaling computation to keep pace Ophthalmology 123(5):1143–1150
with data generation. Genome Biol 17:53 17. Meienberg J, Zerjavic K, Keller I et al (2015)
5. Wetterstrand KA (2016) DNA sequencing New insights into the performance of human
costs: data from the NHGRI genome sequenc- whole-exome capture platforms. Nucleic Acids
ing program (GSP). www.genome.gov/ Res 43(11):e76
sequencingcostsdata. Accessed 27 Feb 2017 18. Rabbani B, Tekin M, Mahdieh N (2014) The
6. Dorajoo R, Liu J, Boehm BO (2015) Genetics promise of whole-exome sequencing in medi-
of type 2 diabetes and clinical utility. Genes cal genetics. J Hum Genet 59(1):5–15
6:372–384 19. Kwak SH, Jung CH, Ahn CH et al (2016)
7. Dickson SP, Wang K, Krantz I, Hakonarson H Clinical whole exome sequencing in early
et al (2010) Rare variants create synthetic onset diabetes patients. Diabetes Res Clin
genome-wide associations. PLoS Biol 8: Pract 122:71–77
e1000294. https://doi.org/10.1371/journal. 20. Lyssenko V, Laakso M (2013) Genetic screen-
pbio.1000294 ing for the risk of type 2 diabetes: worthless or
8. Fuchsberger C, Flannick J, Teslovich TM et al valuable? Diabetes Care 36(Suppl 2):
(2016) The genetic architecture of type 2 dia- S120–S126
betes. Nature 536(7614):41–47 21. Exome Aggregation Consortium, Lek M,
9. Katsanis SH, Katsanis N (2013) Molecular Karczewski KJ, Minikel EV et al (2016) Analy-
genetic testing and the future of clinical geno- sis of protein-coding genetic variation in
mics. Nat Rev Genet 14(6):415–426 60,706 humans. Nature 536(7616):285–291
10. Ellard S, Lango Allen H, De Franco E et al 22. Goodwin S, McPherson JD, McCombie WR
(2013) Improved genetic testing for mono- (2016) Coming of age: ten years of next-
genic diabetes using targeted next-generation generation sequencing technologies. Nat Rev
sequencing. Diabetologia 56:1958 Genet 17(6):333–351
11. Warr A, Robert C, Hume D et al (2015) 23. Levy SE, Myers RM (2016) Advancements in
Exome sequencing: current and future per- next-generation sequencing. Annu Rev Geno-
spectives. G3 5(8):1543–1550 mics Hum Genet 17:95–115
12. Samorodnitsky E, Jewell BM, Hagopian R et al 24. DeAngelis MM, Wang DG, Hawkins TL
(2015) Evaluation of hybridization capture (1995) Solid-phase reversible immobilization
versus amplicon‐based methods for whole‐ for the isolation of PCR products. Nucleic
exome sequencing. Hum Mutat 36 Acids Res 23(22):4742–4743
(9):903–914 25. Chilamakuri CSR, Lorenz S, Madoui M-A,
13. Lelieveld SH, Spielmann M, Mundlos S et al Vodák D, Sun J, Hovig E, Myklebost O,
(2015) Comparison of exome and genome Meza-Zepeda LA (2014) Performance com-
sequencing technologies for the complete cap- parison of four exome capture systems for
ture of protein‐coding regions. Hum Mutat 36 deep sequencing. BMC Genomics 15(1):449
(8):815–822 26. Roche Sequencing (2017) SeqCap EZ System.
14. Bonnefond A, Durand E, Sand O et al (2010) http://sequencing.roche.com/products/
Molecular diagnosis of neonatal diabetes nimblegen-seqcap-target-enrichment/seqcap-
ez-system.html. Accessed 27 Feb 2017
Chapter 6
Gene Expression Mining in Type 2 Diabetes Research

Donald R. Dunbar
Abstract
Gene expression analysis by microarray and more recently by next-generation sequencing has become a core
part of biomedical research and its value can be seen in thousands of research papers. A successful gene
expression experiment needs to be augmented by specialized data mining techniques if the data are to be
fully exploited. Here, tools that concentrate on three areas—gene enrichment analysis, literature mining,
and transcription factor binding site analysis—are described for the novice user of microarray and next
generation sequencing technologies. The focus of this chapter is on free, publicly available, web-based
tools.
Key words Bioinformatics, Microarray, Next generation sequencing, Gene expression, Data mining,
Gene enrichment, Literature mining, Transcription factor binding site
1 Introduction
The methods and utility of microarray gene expression studies have

been described in the previous chapter by Drs. White and Kaestner
in the first edition of this series. In recent years, next generation
sequencing methodologies have begun to compete with microar-
rays. Although the initial analysis methodologies are very different
for microarrays and sequencing (often called RNA-seq) once a gene
list is generated, similar challenges remain for both. For biologists,
the main limiting step in the process from experiment design to
publication of results is usually the interpretation of the data.
Modern microarrays and next generation sequencing experiments
measure gene expression on a genome-wide level, and depending
on the experimental system being studied, routinely identify from
tens to thousands of genes that are differentially expressed. This
creates a problem for the biologist. The biological annotation for a
handful of genes is relatively easy to collate and any strong themes
within the data will be simple to spot. However, when presented
with a list of a hundred or more genes, interpretation of the
biological changes underlying the data becomes an enormous
109
110 Donald R. Dunbar
challenge. High-throughput methods that help the biologist make

sense of their gene expression data have been developed over the
last few years, and the purpose of this chapter is to guide the
inexperienced user of microarrays through a selection of them.
The selection includes tools to identify overrepresented biological
functions and processes associated with the gene list, to mine the
biomedical literature for associations between genes and biological
terms of interest, and to identify transcription factor binding sites in
the gene sequences. The focus of this chapter will be on free,
publicly available, web-based tools.
The method section of this chapter is split into five short
sections (one method for preparing gene lists and four web-based
tools). Each describes one method or tool that helps the biologist
to gain some insight into the biology behind their data. The aim is
to give enough information to allow the novice user of microarrays
and next-generation sequencing, with little or no bioinformatics
support, to use each tool. Each of the tools has additional function-
ality (and usually adequate online help pages), and indeed, there are
many other similar tools available in addition to those
described here.
1.1 Gene Enrichment Gene enrichment analysis uses statistical procedures to discover
Analysis overrepresented features within a dataset. Given a gene list and
some biological information, these tools can identify categories
(e.g., biological processes and molecular functions from the Gene
Ontology consortium or molecular pathways from KEGG) that
appear more often in the gene set that they would by chance in a
random gene set of the same size. If a category is overrepresented,
then this might have biological significance and point to that part of
biology being perturbed in the experiment [1]. Gene enrichment
analysis with the online tool DAVID (Database for Annotation,
Visualization, and Integrated Discovery) will be described [2]. A
tool with a broader and complementary scope than DAVID,
Enrichr, has similar functionality to look for enriched biological
functions [3, 4] and is used in a very similar way to DAVID. Enrichr
gives access to libraries of data including transcription, pathways,
ontologies, diseases, and drug and is very intuitive to explore.
1.2 Literature Mining Literature Mining probes the huge body of information available to
biomedical scientists, for example in the Medline database [5]. If
two or more genes or proteins occur in the same paper, it is possible
that they have some sort of association. Two main problems face
the biologist when mining the literature, however. First, because
there are often hundreds of genes of interest, looking for pairwise
associations using manual web-based searching (e.g., through the
PubMed web interface) is unfeasible. Computationally, however,
these searches are trivial. The PubMatrix text mining tool will be
described [6]. Second, genes often have multiple synonyms that are
used in the literature. For example, the human gene transcription
Microarrays in T2D 111
factor 7-like 2 (T-cell specific, HMG-box) (official gene symbol,

TCF7L2) has several synonyms: HMG box transcription factor
4, hTCF-4, T-cell-specific transcription factor 4, TCF4, TCF-4,
and transcription factor 7-like 2. Several of these are used in scien-
tific publications and this makes exact text matching difficult unless
a complicated query string is built for each gene. The Information
Hyperlinked over Proteins (iHOP) tool makes good use of syno-
nym information, and offers several useful features, and its use will
be described here [7]. These text-mining tools give the biologist a
good entry to mining the literature. Further tools that include
more advanced information extraction and retrieval, entity recog-
nition, and natural language processing will become available to the
biologist over time.
1.3 Transcription Transcription factor binding site (TFBS) analysis can identify bind-
Factor Binding Site ing sites that are statistically overrepresented in the sequences of a
(TFBS) Analysis gene list. Finding such enriched sites can give an idea of the factors
that are driving gene expression locally at the cellular level in the
experimental model. TFBSs often bind to several similar sequences
and computational tools that access TFBS databases can use
sequence models to predict binding sites in gene sequences
[8]. In addition, if a binding site is conserved between species, it
is more likely to be functionally important. Whole Genome RVista
allows for identification of conserved (between pairs of species)
statistically overrepresented TFBSs in specified regions (e.g., 5 kb
upstream) of groups of genes [9].
2 Materials
This description of protocols for downstream analysis of high-

throughput gene expression data is strongly aimed at the inexperi-
enced user of microarray technology with little or no bioinformatics
support. As such, the reader only needs access to the processed,
statistically analyzed data, and a computer with web access, spread-
sheet (e.g., Microsoft Excel or OpenOffice) software, and text file
editing software.
3 Methods
The following methods require the user to have a list of selected

genes. This will be generated by a statistical tool, often used by a
microarray or next generation sequencing facility, and genes will be
included or excluded based on a threshold score for some statistics
of the gene expression (e.g., p-value or fold change). These proto-
cols require that the user can copy a list of gene identifiers (e.g.,
Affymetrix Probeset IDs or Entrez Gene IDs for Subheadings 3.2
and 3.5; gene symbols and names for Subheading 3.3) for further
use. Other identifiers are available and documentation in each of
the tools will describe the appropriate one. Data should come to the
user in the form of a spreadsheet or a tab delimited text file.
3.1 Gene List 1. Open the file containing gene expression data with spreadsheet
Preparation software.
2. Copy the gene identifiers.
3. Paste into a new spreadsheet.
4. Sort the identifiers.
5. Remove any duplicate entries (see Note 1).
6. Save as a plain text file with appropriate name, e.g.,
“exp1_up_2fold_AffyIDs.txt.”
3.2 Gene Enrichment 1. Open the gene list text file (see Note 2).
Analysis with DAVID 2. Copy the identifiers, noting their type (e.g., Affymetrix ID,
Entrez Gene IDs).
3. Go to DAVID website: https://david.ncifcrf.gov/.
4. Click “Start Analysis” in the menu bar.
5. Paste the gene list into “Step 1: Enter Gene List” box.
6. Select the type of identifier (e.g., AFFY_ID) in “Step 2: Select
Identifier”.
7. Select “Gene List” from “Step 3: List Type.”
8. Submit List (this will take you to the analysis wizard).
9. Click “Rename,” rename the list in the dialog box, click “OK.”
10. The “Current Background” will be automatically selected (see
Note 3).
11. If your data are Affymetrix IDs, click “Background” in the
navigation bar.
12. Click the “+” beside “Affymetrix.” Choose the appropriate
GeneChip.
13. Click “Functional Annotation Tool” in the analysis wizard.
14. Click “Functional Annotation Chart.”
15. Click the “+” for “Options.” Check “Fold Enrichment” then
“Re-run Using Options.”
16. Explore the results (see Note 4).
3.3 Literature Mining 1. Go to the PubMatrix site (http://pubmatrix.grc.nia.nih.gov/)

with PubMatrix and register for an account, and then log on to the “authenti-
cated site.”
2. Prepare a list of gene names and symbols up to a maximum of
100 (see Note 5).
3. Prepare a list of relevant words to analyze alongside the genes

(e.g., diabetes, obesity, insulin, diet, glucose, metformin, regu-
late, and expression).
4. Give the “Search Terms” and “Modifier Terms” appropriate
names (e.g., “upregulated genes experiment 2,” “T2D terms
1–10”).
5. Copy the gene list into the “Search Terms” list (or upload
a file).
6. Copy the word list into the “Modifier Terms” list (or upload
a file).
7. Submit to PubMatrix (see Note 6).
8. The following day, go to the PubMatrix site and click “Your
past results.” Check the status of your search, and if completed,
click the link.
9. Explore your results. In the matrix displayed, each combination
of Search Term (gene) and Modifier Term (extra word) has a
number indicating how many abstracts were found. Click one.
10. This takes you to a PubMed search for the titles and abstracts
that contain both the search and modifier term (see Note 7).
3.4 Literature Mining 1. Go to the iHOP website: http://www.ihop-net.org.

with iHOP 2. Enter a gene name, symbol or ID in the first box.
3. Select the appropriate fields to search (leaving “all fields” is
usually fine, if it is a gene ID, select “NCBI Gene”). Select a
species if required.
4. Click the “SEARCH” link.
5. Explore your results.
6. If successful, the search gives one or more rows. Click any of
the links in the row that matches your gene (all the links in a
row are to the same page).
7. This page is the current iHOP page for that gene (highlighted
red). Sentences from PubMed abstracts are displayed and other
genes and other annotation highlighted and hyperlinked (see
Note 8).
8. Click a link for another gene mentioned in the sentence along
with your gene of interest and you will be taken to the iHOP
page for that gene.
9. Use some of the other functions available in iHOP (see Note 9).
3.5 Transcription 1. Go to the Whole Genome RVista site: http://genome-test.lbl.

Factor Binding Site gov/cgi-bin/WGRVistaInputCommon.pl.
Analysis with Whole 2. Choose a genome alignment to use (e.g., Human Feb. 2009
Genome RVista (hg19) vs. Mouse Jul. 2007 (mm9)) and “GO.”
3. Select the length of upstream sequence you would like to

analyze (5 kb default) and a p-value threshold (0.005 by
default).
4. Open the file with the Entrez Gene IDs. Copy the IDs into the
box select “I am submitting locus link ids” (see Note 10).
Submit.
5. Explore your results. The results table shows overrepresented
(statistically significant) conserved transcription factor binding
sites, ordered by log10 p-value (a p-value of 0.005 gives a
log10 p-value of 2.3). Click one of the buttons on the right,
marked with a TFBS symbol. This then shows the genes in your
list that have this sequence at least once in the region searched
(see Note 11).
6. Back on the main results page, scroll down to the second table.
Click on one of the “show” links in the “Summary of all
conserved TFBS upstream of this gene” column. This will
then show a table of conserved TFBSs upstream of the gene
with those overrepresented in the input gene list marked in
green.
4 Notes
1. Often, lists of genes that are differentially expressed will have

two or more entries for the same gene. This is the case for some
outputs of microarray analyses where probes/probe sets have
been kept separate. This can cause problems in further analysis,
especially where statistics require no redundancy. If a gene
(e.g., its Entrez Gene ID) is present two or more times in a
list, then it will be counted twice in the analyses used, for
example, by DAVID (Subheading 3.2) and rVISTA (Subhead-
ing 3.5). When working with tools that accept chip identifiers
(such as Affymetrix probe set IDs), redundancy is not usually a
problem as the chip identifiers will normally be unique.
Although many tools deal with this redundancy at the gene
level, it is safer to remove it at source. Paste the IDs into a
worksheet, sort, and then remove duplicate items, either man-
ually or using a spreadsheet function (Excel has a remove
duplicates tool).
2. Many tools allow identifiers or sequences to be input by pasting
text or directly form a file. If the file is in the appropriate
format, then these methods are equivalent. Choose the method
that suits you.
3. Enrichment analysis requires a “benchmark” to test against. A
background gene set is used for this and can be the comple-
ment of genes on a microarray, or in the whole genome. (The
“Current Background” in DAVID defaults to complete knowl-

edgebase gene set for the input species). If, for example, ten
genes in your gene list of 200 (5%) belong to (or map to) one
Gene Ontology term (GO term) this might sound exciting.
However, if 5% of the genes on the microarray also belong to
that GO term, then your finding is no different from chance.
But if only 1% of the genes on the microarray map to the GO
term, then it is likely that your result is a real enrichment. You
can then proceed to investigate the biological significance of
the result. Other background gene sets, for example, the lit of
genes expressed in a tissue of interest, can be used and may give
better enrichment results; however, this may be more advanced
than novice users can manage.
4. The DAVID functional annotation chart is a table displaying
those terms that are statistically enriched in your gene list. The
“term” column gives a hyperlinked name of the enriched term.
The hyperlink goes to the database outlined in the “category”
column: for example, GOTERM_ goes to QuickGO at EBI,
SP_ goes to Uniprot, and KEGG_ and BIOCARTA_ go to
those pathway databases stored within DAVID. The links are
not all ideal as they can be to text searches rather than direct
links to database entries using the ID of the term: these text
searches, using the term name can often match more than the
term of interest. Further information about the term can be
obtained from the “RT” column, which gives a hyperlink to
related terms for each term, ranked by relatedness based on
overlap of gene lists. This can be useful to get a wider view of
the biology around the enriched term. There is usually plenty
of redundancy in this list, mainly due to the hierarchical nature
of the GO database (e.g., genes in GOTERM_BP_5 are a
subset of genes in GOTERM_BP_4). The table is ordered by
the p-value for the statistical test (Fisher Exact) used to deter-
mine enrichment. In addition, there is a column “Benjamini”
that lists the p-value corrected for multiple testing. The
“count” indicates the number of genes from our list that is
mapped to the term, and “%” is this number as a percentage of
the full list of genes mapped to that term. The “fold enrich-
ment” column gives a ratio for the enrichment. Clicking on the
blue bar in the “Genes” column will show the genes in the list
that map to the term and useful links to the database entry for
each gene and a set of related genes.
5. PubMatrix does not map a gene’s ID onto its name or symbol,
so it is essential to prepare a gene name and symbol list. These
of course can be used separately, but they can also be used as a
query together separated by “OR.” Copy and paste name and
symbol for your gene list into a worksheet. Insert a column
between them and add the word “OR” to each cell in the
column by filling down, so that we have gene name in column

1, “OR” in column 2, and gene symbol in column 3. Now, in
the column 4, add the following function: ¼CONCATENATE
(A1,“ ”,B2,“ ”,C1), and fill or copy it down the column. This
will give terms like “aquaporin 4 OR Aqp4.” These can then be
stored in a text file with an appropriate name and used in a
PubMatrix search.
6. Due to restrictions placed on using PubMed servers at NCBI,
PubMatrix will only run searches from 5 p.m. to 8 a.m. Eastern
Standard Time. Jobs will, however, go into the queue if sub-
mitted outside this window and are almost always completed by
the next day.
7. PubMatrix searches are fairly crude and as with other text
searches of PubMed, and are sensitive to false positives and
negatives. False negatives usually happen because none of a
gene’s synonyms were used in the search. False positives are
frequently the fault of gene symbols having an alternative
meaning in the literature. Symbols that are also short words
or abbreviations are often the culprits. Adding extra synonyms
where appropriate and removing problematic symbols can
help. Where this occurs, simply resubmit the new list to
PubMatrix.
8. The main advantage of using iHOP is its use of synonym data
to search the literature and link genes/proteins. This means
that searching with one synonym will also find publications that
cite other synonyms or symbols for the gene or protein. iHOP
makes extensive use of links and markup. Each iHOP page has
links to the pages for other genes: this is the essence of iHOP.
“Marking up” abstracts means that it is immediately obvious
where in the abstract the current and other genes are in addi-
tion to informative verbs such as associated, interacts, biding,
complexed, affect, inhibits, suppresses, and activates. MESH
(Medical Subject Headings) terms and chemical compounds
are also marked up and have additional links. Links on the right
of the screen will take you to the abstracts within PubMed.
9. An extremely useful feature of iHOP is the “show overview”
link. This link takes you to a list of genes that are associated
with the gene of interest, ordered by the number of sentences
in the literature that co-cite the two genes and whether there is
any interaction evidence in several databases including IntAct.
This immediately guides you to genes that are highly co-cited
with the gene of interest. Click a gene link to see that both the
original and linked genes are marked up in the sentences.
Enhanced PubMed and Google searches (using all synonyms,
symbols, and additional terms) are useful for further searching,
and genes can be added to a useful “Gene Model” that shows

interactions visually.
10. Locus link was the predecessor of the Entrez Gene database.
Locus link IDs map exactly to Entrez Gene IDs.
11. In the Whole Genome rVISTA output, the “number of hits in
the submitted regions” column tells us the total number of hits
rather than the number of genes that contain the conserved
TFBS sequence. This is compared with the “total number of
hits on genome” to generate the statistics. Clicking the “show
genes where. . .” buttons will then show the genes that have
the TFBS.
References
1. Curtis RK, Oresic M, Vidal-Puig A (2005) Path- 6. Becker KG, Hosack DA, Dennis G, Lempicki
ways to the analysis of microarray data. Trends RA, Bright TJ, Cheadle C et al (2003) PubMa-
Biotechnol 23(8):429–435 trix: a tool for multiplex literature mining. BMC
2. Huang DW, Sherman BT, Lempicki RA (2009) Bioinformatics 4:61
Systematic and integrative analysis of large gene 7. Hoffmann R, Valencia A (2004) A gene network
lists using DAVID bioinformatics resources. Nat for navigating the literature. Nat Genet 36
Protoc 4(1):44–57 (7):664
3. Chen EY, Tan CM, Kou Y, Duan Q, Wang Z, 8. Elnitski L, Jin VX, Farnham PJ, Jones SJM
Meirelles GV et al (2013) Enrichr: interactive (2006) Locating mammalian transcription factor
and collaborative HTML5 gene list enrichment binding sites: a survey of computational and
analysis tool. BMC Bioinformatics 14:128 experimental techniques. Genome Res 16
4. Kuleshov MV, Jones MR, Rouillard AD, Fernan- (12):1455–1464
dez NF, Duan Q, Wang Z et al (2016) Enrichr: a 9. Zambon AC, Zhang L, Minovitsky S, Kanter JR,
comprehensive gene set enrichment analysis web Prabhakar S, Salomonis N et al (2005) Gene
server 2016 update. Nucleic Acids Res 44(W1): expression patterns define key transcriptional
W90–W97 events in cell-cycle regulation by cAMP and pro-
5. Jensen LJ, Saric J, Bork P (2006) Literature tein kinase A. Proc Natl Acad Sci U S A 102
mining for the biologist: from information (24):8561–8566
retrieval to biological discovery. Nat Rev Genet
7(2):119–129
Chapter 7
Pathways Enrichment Analysis of Gene Expression Data

in Type 2 Diabetes
Maysson Ibrahim
Abstract
Profiling genome-wide transcriptional changes with advanced high-throughput transcriptional profiling
techniques has led to a revolution in biomedical science. It has been challenging to handle the massive data
generated by these techniques and draw meaningful conclusions from it. Therefore, computational biolo-
gists have developed a number of innovative methods of varying complexity and effectiveness to analyze
such complex data. Over the past decade, rich information in pathway repositories has attracted and
motivated researchers to incorporate such existing biological knowledge into computational analysis tools
to develop what is known as pathway enrichment analysis tools. This chapter describes a new sophisticated
pathway enrichment tool that exploits topology of pathway as well as expression of significantly changed
genes to identify biologically significant pathways for high-dimensional gene expression datasets. Also, we
demonstrate the use of this tool to analyze gene expression data from a type 2 diabetes dataset to identify a
list of significantly enriched metabolic pathways.
Key words Pathway analysis, Metabolic pathway, Gene expression, Type 2 diabetes, Microarray
1 Introduction
Genome-wide transcriptional profiling techniques provide

researchers with tools to explore life at the molecular level by
quantifying temporal and spatial changes in gene activity. However,
meaningful analysis of complex data generated by these techniques
remains a challenge despite recent substantial efforts to develop
sophisticated analysis methods. Pathway enrichment is one of these
analysis methods that have facilitated our understanding of big
genetic data by reducing the dimensionality of analyzed data from
tens of thousands of individual genes into hundreds of predefined
biological pathways curated and stored in public databases
[1]. Based on differences in structure and mechanism, biological
pathways are grouped and stored in different categories such as
biochemical enzyme–substrate pathways, linear or branching sig-
naling cascades pathways or protein–DNA binding. Therefore,
119
120 Maysson Ibrahim
Fig. 1 Pathway enrichment analysis approaches (Adapted from [2])
defining a one-size-fits-all set of mathematical principles to analyze

different groups of pathways is a very challenging task. Pathway
enrichment approaches can be generally divided into three genera-
tions [2]: Over-Representation Analysis (ORA), Functional Class
Scoring (FCS) approaches, and Pathway Topology (PT)-based
approaches (Fig. 1). Arguably, the third-generation approaches
outperform both ORA and FCS approaches in the rational exploi-
tation of rich biological information such as topology of pathway
represented by genes location and interaction (e.g., activation,
inhibition, etc.). Several topology-based pathway enrichment
tools have been proposed in the literature over the past few years
[2]. Some tools rely on topology only in scoring pathways [3, 4],
whereas others use topology in addition to other gene expression
measurements [5].
This chapter presents a new topology-based pathway enrich-
ment tool for analyzing gene expression data [6]. The underpin-
ning approach utilizes both pathway topology (i.e., the
relationships between genes of the network) and magnitude of
gene expression changes in formulating a new score for impacted
pathways called Pathway Regulation Score (PRS) [7]. The PRS
approach was developed to analyze signaling as well as metabolic
pathways taking into consideration differences in structural com-
plexity between the two groups. The list of pathways used by this
tool was imported from KEGG database (http://www.kegg.jp/
kegg/rest/) [8] and then maps were redrawn in MATLAB for
two reasons. Firstly, removing any redundancy of genes in pathways
map (i.e., some genes are represented multiple times in one path-
way map) to mitigate any bias can affect the reliability of the final
pathway score, and secondly to facilitate the use of graph theory for
Pathways Analysis 121
solving the loops problem in pathways in addition to other pro-

blems that emerged from working on annotations and topology
information. The PRS tool gives users the option of running a
statistical test to evaluate the significance of their results. For com-
parison purposes with traditional pathway enrichment scoring
approaches, the tool uses the known z-score approach [9] to run
overrepresentation analysis in parallel with the PRS approach.
2 Materials
1. Microarray gene expression data. There are a number of public

gene expression dataset repositories such as Gene Expression
Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/
geo/), or ArrayExpress (http://www.ebi.ac.uk/arrayexpress/).
2. Access to PRS software, which is a standalone tool that needs to
be downloaded on your PC (see Subheading 3.1 for details).
3. Microsoft Excel spreadsheet to create an input file for the tool.
4. Internet access to browse enriched pathways map.
3 Methods
3.1 Software 1. If you do not have MATLAB on your PC, you should initially
Installation install the MCR (MATLAB Compiler Runtime) as explained in
the tool webpage (http://www.buckingham.ac.uk/research/
clore-laboratory-diabetes-obesity-and-metabolic-research/
staff/maysson-al-haj-ibrahim/prs-tool/).
2. Click on the “Topology-based Pathway Analysis Tool” hyper-
link in the tool webpage to download the .zip file.
3. Right-click on the .zip format file to extract all relevant files,
and then run the PRS_interface file.
3.2 Data 1. Prior to the analysis of gene expression data, some information
Preprocessing should be checked such as the array platform, number of
samples per group, species, and whether data is raw or
normalized.
2. The tool does not provide a filter to normalize data. Therefore,
users should normalize the data before using the tool for
analysis (see Note 1).
3. Data must be saved in a form of simple Excel spreadsheet, in
which the first column should be a probe ID, and the following
columns are normalized gene expression values from control
and test groups (see Note 2).
122 Maysson Ibrahim
Fig. 2 PRS tool interface
3.3 Pathways 1. Right-click on the main file to open the GUI (see Fig. 2).
Enrichment with PRS 2. Select the type of species (e.g., Human, Mouse, Rat) from the
drop-down menu.
3. Click the “browse” button to select the input data file (.xls
format) pre-prepared in step 3 of Subheading 3.2.
4. Enter the number of samples in group 1 and 2 (i.e., group1
might refer to control or test group based on which one comes
first in the input file).
5. Enter the fold-change threshold for selecting highly changed
genes (see Note 3).
6. Enter the p-value threshold for selecting significantly changed
genes (see Note 4).
7. Select the normalization method (e.g., RMA/GCRMA, MAS)
used for normalizing your data.
8. Select the platform (e.g., Affy HG U133A, Affy HG U133B,
Affy HG U133 Plus 2.0, Affy HG U95A, Agilent) that is
compatible with your data.
9. Finally, select which group of pathways (signaling or metabolic
pathways, or both) you would like to map your data onto to
run the analysis.
10. Press the “Analyze” button to start the analysis and see the
results in the main interface (see Note 5).
3.4 Statistical 1. Click the “browse” button to select a results’ file (.mat format)
Evaluation that was generated from a previous analysis (see Note 6).
2. Input the number of permutations for the statistical test (see

Note 7).
3. Input the p-value threshold for selecting significant pathways
(see Note 8).
4. Click the “calculate” button to run the evaluation.
3.5 Visualizing 1. In the main interface, activate the “Select a pathway to visua-
Enriched Pathway Map lize” checkbox option.
2. From the results’ table in the main interface, click on the
pathway that you want to visualize.
3. Click the “Apply” button and wait for the pathway map web-
page to open (see Note 9).
4 Notes
1. It is important to mention that effective normalization techni-

ques are usually platform-dependent. For example, RMA [10],
dChip [11], and MAS5 [12] are between-arrays normalization
techniques used with one-channel arrays such as Affymetrix
[13]. Whereas Lowess is within-array normalization method
and commonly used with Agilent platforms [10]. These meth-
ods are implemented in a number of existing software such as
Affymetrix Expression Console Software, which can be down-
loaded from (http://www.affymetrix.com/estore) or other
software development platforms such as MATLAB (www.
mathworks.com) or Bioconductor (http://www.bio
conductor.org/).
2. To format the input file, make sure that your data start from the
first row with no headings. Also remember that the order and
size of the groups stored in the file should match those you will
input in the main interface (step 4 in Subheading 3.3).
3. Fold-change is a known method for identifying genes that
change significantly between two groups. In this method,
gene expression medians are calculated for each gene in both
control and test groups. The fold-change will be estimated by
the ratio of the medians of the two groups. All genes that have
fold-change greater than a user-defined threshold are defined
as Differentially Expressed Genes (DEGs). Although the
threshold is traditionally set to be 1.5 or more, this is relatively
arbitrary and may be adapted depending on the number of
replicates. Indeed, the selected value should yield a sufficient
number of impacted genes to allow pathway mapping.
4. p-Value threshold is the other criterion used in combination
with fold-change to identify significantly changed genes
between groups. Only DEGs that have p-values less than a
124 Maysson Ibrahim
threshold (user-defined) are considered as significant DEGs.

Although the p-value threshold is an arbitrary value, 0.01 and
0.05 are most commonly used in this area.
5. By clicking the “Analyze” button, the PRS tool will:
(a) Read data (probe IDs, groups’ samples) from the
input file.
(b) Convert probe IDs to KEGG IDs. It is not always easy to
convert from any platform to KEGG ID. For that reason,
Entrez ID is used as an intermediate step in the conver-
sion. Firstly, probe IDs are converted into Entrez IDs
based on existing mapping tables for different platforms.
Secondly, Entrez IDs can be easily converted into KEGG
IDs by adding a known prefix to Entrez IDs (e.g., “has”:
for human, “mmu”: for mouse, “rno”: for rat).
(c) Identify a list of significant DEGs based on fold-change
and p-value criteria. Unlike significant DEGs, nonsignifi-
cant genes are assigned a value 1 instead of their fold-
change value to mitigate their contribution in the final
pathways score.
(d) Map genes and their values onto the list of curated
pathways.
(e) Build a graph for each pathway. Then, a Depth-First
Search (DFS) algorithm is used to solve the loops problem
and facilitate traversing the nodes (genes) of each graph
(pathway). It also assigns a weight to each node based on
its location and the number of significant downstream
nodes.
(f) Calculate a pathway regulation score (PRS) for each
weighted graph (pathway) using the weights of significant
nodes in that graph in addition to other parameters.
(g) Calculate the z-score for each weighted graph (pathway)
for comparison purposes with the PRS approach.
(h) Rank the pathways in descending order based on their
PRS values then export the ranked list to the table in the
main interface.
(i) Repeat the latter step for z-score and export the ranked list
to a different table for comparison purposes.
(j) Save the results in two different format files (.mat and .
xls).
6. After the analysis and generating a list of scored pathways,
statistical significance of the results can be evaluated to test if
any of these scores (particularly high scoring pathways) were
likely to occur by chance. The null hypothesis assumes that
there is no difference between an observed score (PRS) and
any random score for the same pathway.
7. A nonparametric permutation approach is used to generate

random PRS scores for each pathway. It starts by permuting
all genes’ values, mapping them back onto pathways, and then
recalculating the PRS score for each pathway. This process is
repeated n times to generate the null distribution, where n is
the number of permutations provided by the user through the
interface. This number should be high enough (e.g., n ¼ 500,
n ¼ 1000) to get a reliable test.
8. To decide if a pathway score is statistically significant, all ran-
dom scores created in step 7 for a pathway are compared to its
observed score. Only the ones that match or exceed the
observed score are counted, then the resultant number is
divided by the total number of random scores (number of
permutations) to give us what is known as the p-value. Next,
the p-values of all pathways are adjusted by the false discovery
rate (FDR) to assign a refined p-value to each pathway. Finally,
all pathways that have a p-value smaller than the user-defined
threshold are considered statistically significant.
9. The PRS software calls a pathway mapping web service (REST-
based API service) hosted on the KEGG website and passes a
number of parameters including a list of all expressed genes
with their fold-change values, as well as setting colors to differ-
entiate significant genes from nonsignificant genes (see Fig. 3).
Herein, we used the PRS tool to run pathways enrichment
analysis on type 2 diabetes and pancreatic islets dataset down-
loaded from the Gene Expression Omnibus database (https://
www.ncbi.nlm.nih.gov/geo/; accession #GDS4337). The
Fig. 3 “Arachidonic acid metabolism pathway” enriched in T2D dataset (significant genes are colored in red)
126 Maysson Ibrahim
Table 1
Input data information for pathways analysis and significance evaluation in T2D and pancreatic islets
dataset
PRS analysis
Species Human
# of samples in group1 9
# of samples in group2 54
Fold-change threshold 1.3
p-Value threshold 0.05
Normalization method RMA
Platform Affymetrix human gene 1.0 ST
Pathway type Both (metabolic and signaling)
Statistical significance evaluation
Number of permutations 1000
p-Value threshold 0.05
experiment was designed to compare the gene expression levels

in RNA isolated from human pancreatic islets taken from 9 type
2 diabetes (T2D) cadaver donors with RNA samples of pancre-
atic islets derived from 54 nondiabetic cadaver donors. Affyme-
trix Human Gene 1.0 ST Array was used for measuring gene
expression levels and the resulting values were normalized by
Robust Multi-array Analysis (RMA) before being uploaded to
the GEO database. To use the PRS tool, we created an input
file containing Affymetrix probe IDs and the samples from the
two groups. Also, we input in the GUI other parameters (see
Table 1 for more details about the parameters used in this
example) such as number of samples in each group taking
into account the groups order in the input file, (group 1 is
diabetic and group 2 is nondiabetic in this case), fold-change
(FC) and p-value thresholds (FC 1.3 and p-value <0.05 in
this case, as FC 1.5 yields small number of significant genes
which is not sufficient number to allow pathways mapping).
Also, we chose to enrich for both signaling and metabolic
pathways. Accordingly, the tool found the significant genes,
mapped them onto the pathways, and created two scores (PRS
and z-score) for each pathway. Then, pathways were ranked
according to PRS and z-score and shown on the GUI in two
separate tables (Fig. 2). Table 2 shows the top ten significant
pathways ranked by PRS score where only statistically signifi-
cant pathways (FDR < 0.05) were selected, and Table 3 shows
the top ten significant pathways ranked by z-score.
Table 2
Top ten pathways ranked by PRS score (T2D and pancreatic islets dataset)
Rank Pathway name PRS p-Value FDR

1 Arachidonic acid metabolism 3.450412 0 0
2 Cytokine–cytokine receptor interaction 1.443531 0 0
3 TGF-beta signaling pathway 1.345376 0 0
4 Complement and coagulation cascades 1.180362 0 0
5 PPAR signaling pathway 1.030316 0.002 0.0065
6 Pathways in cancer 0.910555 0.004 0.0104
7 Type II diabetes mellitus 0.793327 0.002 0.0065
8 Tryptophan metabolism 0.754089 0.001 0.004875
9 MAPK signaling pathway 0.736616 0.001 0.004875
10 Fatty acid metabolism 0.701842 0.004 0.0104
Table 3
Top ten pathways ranked by z-score (T2D and pancreatic islets dataset)
Rank Pathway name z-Score

1 Arachidonic acid metabolism 6.103672
2 TGF-beta signaling pathway 5.571651
3 Complement and coagulation cascades 5.468563
4 PPAR signaling pathway 5.302763
5 Cytokine–cytokine receptor interaction 5.102405
6 Fatty acid metabolism 5.050608
7 Intestinal immune network for IgA production 4.748036
8 Cell adhesion molecules (CAMs) 4.601507
9 Allograft rejection 4.480696
10 Staphylococcus aureus infection 4.416682
Interestingly, some relevant pathways come in the top of the list

such as “Arachidonic acid metabolism” [14], “TGF-beta sig-
naling pathway” [15],” Fatty acid metabolism” [16, 17],
“PPAR signaling pathway” [18, 19]. In comparison with z-
score, only PRS picked up “MAPK signaling” and “Type II
diabetes mellitus” pathways which are highly related to T2DM.
128 Maysson Ibrahim
References
1. Bauer-Mehren A, Furlong LI, Sanz F (2009) microarray platforms comparable? Genomics
Pathway databases and tools for their exploita- 83(6):1164–1168
tion: benefits, current limitations and chal- 11. Li C, Wong W (2003) DNA-chip analyzer
lenges. Mol Syst Biol 5:290 (dChip). http://www.springerlink.com/
2. Khatri P, Sirota M, Butte AJ (2012) Ten years index/G5V0232463245R24.pdf
of pathway analysis: current approaches and 12. Pepper SD, Saunders EK, Edwards LE, Wilson
outstanding challenges. PLoS Comput Biol 8 CL, Miller CJ (2007) The utility of MAS5
(2):e1002375 expression summary and detection call algo-
3. Gao S, Wang X (2007) TAPPA: topological rithms. BMC Bioinformatics 8(1):273
analysis of pathway phenotype association. Bio- 13. Do JH, Choi D (2006) Normalization of
informatics 23(22):3100–3102 microarray data: single-labeled and dual-
4. Vert JP, Kanehisa M (2003) Extracting active labeled arrays. Mol Cells 22(3):254
pathways from gene expression data. Bioinfor- 14. Persaud SJ, Muller D, Belin VD, Kitsou-
matics 19(Suppl 2):ii238–ii244 Mylona I, Asare-Anane H, Papadimitriou A
5. Tarca AL, Draghici S, Khatri P, Hassan SS, et al (2007) The role of Arachidonic acid and
Mittal P, Kim J et al (2009) A novel signaling its metabolites in insulin secretion from human
pathway impact analysis. Bioinformatics 25 islets of Langerhans. Diabetes 56(1):197–203
(1):75–82 15. Prentki M, Nolan CJ (2006) Islet cell failure in
6. Ibrahim MA, Jassim S, Cawthorne MA, Lang- type 2 diabetes. J Clin Invest 116
lands K (2014) A MATLAB tool for pathway (7):1802–1812
enrichment using a topology-based pathway reg- 16. Yaney GC, Corkey BE (2003) Fatty acid
ulation score. BMC Bioinformatics 15(1):358 metabolism and insulin secretion in pancreatic
7. Ibrahim MA, Jassim S, Cawthorne MA, Lang- beta cells. Diabetologia 46(10):1297–1312
lands K (2012) A topology-based score for 17. McGarry JD (2002) Banting lecture 2001 Dys-
pathway enrichment. J Comput Biol 19 regulation of fatty acid metabolism in the etiol-
(5):563–573 ogy of type 2 diabetes. Diabetes 51(1):7–18
8. Kanehisa M, Goto S, Kawashima S, Nakaya A 18. Kim H-S, Hwang Y-C, Koo S-H, Park KS, Lee
(2001) The KEGG databases at GenomeNet. M-S, Kim K-W et al (2013) PPAR-γ activation
Nucleic Acids Res 2(1):42–46 increases insulin secretion through the
9. Cheadle C, Vawter MP, Freed WJ, Becker KG up-regulation of the free fatty acid receptor
(2003) Analysis of microarray data using Z GPR40 in pancreatic β-cells. PLoS One 8(1):
score transformation. J Mol Diagn 5(2):73–81 e50128
10. J€arvinen A-K, Hautaniemi S, Edgren H, 19. Sugden MC, Holness MJ (2004) Potential role
Auvinen P, Saarela J, Kallioniemi O-P et al of peroxisome proliferator-activated receptor-α
(2004) Are data from different gene expression in the modulation of glucose-stimulated insu-
lin secretion. Diabetes 53(1):S71–S81
Chapter 8
Diagnostic Genetic Testing for Monogenic Diabetes

and Congenital Hyperinsulinemia
Abstract
Monogenic diabetes and hyperinsulinism are genetically heterogeneous disorders. The determination of
the genetic etiology defines the diagnostic subtype, predicts prognosis, and importantly can guide clinical
management. This chapter focuses on the processes and methodologies utilized in the diagnostic testing for
monogenic diabetes and congenital hyperinsulinism (i.e., Sanger sequencing and targeted next-generation
sequencing).
Key words Sanger sequencing, Targeted next-generation sequencing, Neonatal diabetes, MODY,
Hyperinsulinism, KCNJ11 and ABCC8
1 Introduction
Monogenic diabetes and hyperinsulinemic hypoglycemia result

from a mutation or mutations in a single gene and can be domi-
nantly inherited, recessively inherited, or can arise de novo. Mono-
genic diabetes is rare and accounts for ~3% of all cases of diabetes in
young people. Neonatal diabetes affects ~1:100,000 live births and
can be transient, permanent, or part of a syndrome. A genetic
diagnosis often leads to improved treatment and can predict prog-
nosis. Genetic testing can define the subtype in ~80% patients
diagnosed with neonatal diabetes and about 40% of these patients
will have an activating mutation in one of the KATP channel genes
KCNJ11 and ABCC8. A rapid genetic test is important, as patients
with KATP channel mutations are able to transfer from insulin
injections to sulfonylurea tablets which results in an improvement
in glycemic control [1]. Approximately 52% of patients with a
confirmed diagnosis of maturity onset diabetes of the young
(MODY) have a mutation in HNF1A, 10% have a mutation in
HNF4A, and 32% have GCK mutations [2]. A genetic diagnosis
of MODY is essential to ensure that patients are on the most
appropriate treatment, reducing the risk of diabetic complications
129
130 Jayne A. L. Houghton
in later life. MODY subtypes HNF1A and HNF4A can be treated

with low-dose sulfonylureas and those with MODY subtype GCK
do not require any pharmacological treatment.
Congenital hyperinsulinism affects about 1:50,000 newborns,
although the prevalence is more common in certain populations
~1:2500 [3]. It is characterized by the inappropriate secretion of
insulin despite low glucose levels which if untreated can lead to
seizures, coma, and death. It has two distinct histological forms,
diffuse and focal. In patients with diffuse disease, the whole of the
pancreas is affected, whereas a patient with focal disease has only a
portion of the pancreas affected. Diffuse disease can be recessively
or dominantly inherited, but the inheritance of focal congenital
hyperinsulinism is more complex. Focal disease results from a
paternally inherited recessive inactivating KCNJ11 or ABCC8
mutation and loss of the maternal allele within the pancreas due
to paternal uniparental isodisomy during embryonic development.
Genetic testing can help to distinguish between the diffuse and
focal forms in ~40% of patients with congenital hyperinsulinism
and plays an important role in helping to guide appropriate clinical
management as it can help identify those patients with congenital
hyperinsulinism who can be cured by lesionectomy. A genetic
diagnosis also informs families of sibling recurrence risk and the
risk for future offspring and can enable predictive genetic testing
for asymptomatic relatives.
Current comprehensive genetic testing for patients with mono-
genic diabetes and congenital hyperinsulinism is performed using a
combination of Sanger sequencing and targeted next-generation
sequencing. Sanger sequencing of the common causes of mono-
genic diabetes and hyperinsulinism is performed as the first-line test
as this can help guide urgent clinical management. For patients in
whom a mutation is not identified in one of the common genetic
causes, testing of all the known causes of monogenic diabetes and
hyperinsulinism is performed using targeted next-generation
sequencing. The aim of this chapter is to provide information for
users regarding the methods utilized for the detection of mutations
in the diagnostic testing for monogenic diabetes and congenital
hyperinsulinism.
2 Materials
2.1 DNA Sample DNA is extracted from peripheral blood in EDTA within 5 days of
Preparation venesection using standard laboratory protocols or commercial
DNA extraction and purification kits. High-quality DNA (nonde-
graded, A260/A280 is 1.8–2.0) is essential for reliable PCR amplifi-
cation and sequencing.
Diabetes and Hyperinsulinism Genetic Testing 131
2.2 Sanger 1. MegaMix Royal (Microzone) contain Taq polymerase in

Sequencing: PCR 2 enhancing buffer (6 mM MgCl2) with 400 μM dNTPs,
Reagents blue MiZn loading dye, and stabilizer.
2. Microforce (MF) (Microzone) enhancing buffer for the ampli-
fication of GC-rich templates.
3. AmpliTaq Gold Polymerase (Applied Biosystems) with 10
PCR buffer I containing 15 mM MgCl2.
4. 5 M Betaine.
5. 7-Deaza-dGTP (2 mM) (Roche).
6. dNTPS (Mix 1) (Cat no. AB-0196).
7. dNTPs (Mix 2) to make up 100 μl: 5 μl dTTP (100 mM), 5 μl
dTTP (100 mM), 5 μl dTTP (100 mM), 5 μl dTTP (100 mM),
and 81.5 μl of sterile water.
8. Deaza dGTP (Mix 3) to make up 100 μl: 10 μl Deaza dGTP
(2 mM), 90 μl of sterile water.
9. Dimethyl sulfoxide (DMSO).
10. DNA (10 ng/μl).
11. PCR primers (see Table 1 for sequences) (see Note 1).
2.3 Sanger 1. 5 μl PCR product.

Sequencing: PCR 2. Ethanol (85% v/v) (Sigma-Aldrich E7023—500 ml).
Clean-Up Reagents
3. Agencourt AMPure XP-PCR purification system.
4. 384-Well cycling plate (40 μl well capacity).
2.4 Sanger Sequencing reagents will vary because they will be specific to the
Sequencing: laboratory’s available sequencing equipment. We have found that
Sequencing Reagents the following produce high-quality data and fit with our automated
laboratory system.
1. 5 BigDye dilution buffer (Life Technologies Cat# 4336697).
2. BigDye terminator v3.1 (Life Technologies Cat# 4337455).
3. dH2O (Baxter (UKF7114).
4. M13 sequencing primers (Stock 100 pmol/μl).
5. POP7 Polymer (Life Technologies Cat#4342759).
6. 10 3730 running buffer (Life Technologies Cat# 4335613).
7. EDTA (0.5 mM).
2.5 Sanger 1. Sequencing reaction product.

Sequencing: 2. Agencourt CleanSeq Dye terminator removal.
Sequencing Reaction
3. Ethanol (85% v/v) (Sigma-Aldrich E7023—500 ml).
Clean-Up Reagents
4. Elution buffer (dH2O).
Table 1
132
List of genes, transcripts, primers and PCR conditions
Gene: ABCC8 Accession No.: U63421 and L78208
Primer DNA
PCR concentration concentration Size
Exon method (pmol/μl) (ng/μl) Forward and reverse primer sequences (M13 tailed) chromosome number (bp)
1 MMR 2 (80% MF) 10 ABCC8ex1F_and_ABCC8ex1R agctgcaagggacagagg gagtgaagggatgagctgg 11 417
2 MMR 2 10 ABCC8ex2F_and_ABCC8ex2R GAGGCAACAGAGCAAGACC 422
ACCCTGGAGCAGATTCACTT 11
3 MMR 2 10 ABCC8ex3F_and_ABCC8ex3R gccctgcagcctataaagtg ctccatgaaggcagggatt 11 390
4 MMR 2 10 ABCC8ex4F_and_ABCC8ex4R AAATGTACACACCCAGGCAC 411
GGGTAAAACAAGCTGATCCC 11
5 MMR 2 10 ABCC8ex5F_and_ABCC8ex5R gtgttggggaatccttttcc ccctttgaggtccctctctg 11 478
6 MMR 2 10 ABCC8ex6F_and_ABCC8ex6R GTTTCCCCAGACAACAGGAG 522
TGGTAGTGACGGTGAGAGGA 11
7 MMR 2 10 ABCC8ex7F_and_ABCC8ex7R CAGGGTGTAAGCAACCTTCC 453
TGAGGATGAATAACACTCATGGAC 11
8 MMR 2 10 ABCC8ex8F_and_ABCC8ex8R AAGTTGGAACGGTGATACAG 435
TGTGAAAGGTACAGGCAAGC 11
9 MMR 2 10 ABCC8ex9F_and_ABCC8ex9R GATAATTTGGAAACCTGGGC 387
TGAAGTGGCCTACTCAAAGTC 11
10 MMR 2 10 ABCC8ex10F_and_ABCC8ex10R TCTGGGAAATGGAGTCAATG 432
GAGTCGGATAATCTCAAGGC 11
11 MMR 2 10 ABCC8ex11F_and_ABCC8ex11R TGCCCCTAGCCTACTGGAG 248
CTGGGCAGCCTGTCACTG 11
12 MMR 2 10 ABCC8ex12F_and_ABCC8ex12R ATGAAGGTGTCTCCAACTAAAAGAT 362
ATCACTCGAGCAAGCCTTG 11
13 MMR 2 10 ABCC8ex13F_and_ABCC8ex13R TTCAGTGTGGGCTTTGTGG 390
GGTGGTTTGGAGGTGAGGA 11
14 MMR 2 10 ABCC8ex14F_and_ABCC8ex14R GCTGTGTCGGACTTCTGCCTTT 266
GCTCCCTCTGGGAGTTGGTG 11
15 MMR 2 10 ABCC8ex15F_and_ABCC8ex15R TTTTGGCTTTCATGGAGGAG 251
TGCAGCTTTGTCTTTTTATCTCTATG 11
16 MMR 2 10 ABCC8ex16F_and_ABCC8ex16R GAGGATGTTGATTTCCAGAAGG 346
TGAGGAGGATGGTTAAAAGGAG 11
17 MMR 2 10 ABCC8ex17F_and_ABCC8ex17R ACAGAGGCCATTTGGAAAC 298
TCTGAAAATATGTAGGCTGCAC 11
18 MMR 2 10 ABCC8ex18F_and_ABCC8ex18R TCTCTATGCAGCATTTGTGG 366
AATGGATGCACAGAAACAGC 11
19 MMR 2 10 ABCC8ex19F_and_ABCC8ex19R AGACCCAGACCTCTCAAACC 439
GGTGCACCATATGGAGAGG 11
20 MMR 2 10 ABCC8ex20F_and_ABCC8ex20R GAGGCCTATTAAAGCCATTGC 350
CATGTTTGACCTTACTGCAGGC 11
21 MMR 2 10 ABCC8ex21F_and_ABCC8ex21R AGGTGAGAAGCAGGCAAAGA 277
GGTGGAGGTGGGCAGTTAG 11
22 MMR 2 10 ABCC8ex22F_and_ABCC8ex22R TCCAAAGCCACACAGCTAAC 404
CCAGTGCTGGTCTCTTATGC 11
23 MMR 2 10 ABCC8ex23F_and_ABCC8ex23R AGGAGTATGTTGGTTGGGGTAG 399
GGGCACTAAGGACAGGAAGA 11
24 MMR 2 10 ABCC8ex24F_and_ABCC8ex24R TGAATGTGTGTCTGTCTGCC 370
CAGAGGGAAGCCATTTAATC 11
25 MMR 2 10 ABCC8ex25F_and_ABCC8ex25R CCCGTTGTCCCCTCAGTAAG 393
CTCAGCCCTTCCCCCATC 11
Diabetes and Hyperinsulinism Genetic Testing
26 MMR 2 10 ABCC8ex26F_and_ABCC8ex26R CTGCAGCCAGGAACTGCTC 375

CCATTTTATAGATGGGAAGACTAAGG 11
133
(continued)
Table 1
134
(continued)
Gene: ABCC8 Accession No.: U63421 and L78208
Primer DNA
27 MMR 2 10 ABCC8ex27F_and_ABCC8ex27R TGAATGACTCCAGAGACACTTA 246
AGACAGGAGAAGCCCCCAG 11
28 MMR 2 10 ABCC8ex28F_and_ABCC8ex28R AGTCTGGGCAACAGTGAGAC 466

TAGGGCGGTGGAATAAGATG 11
29 MMR 2 10 ABCC8ex29F_and_ABCC8ex29R CACGGGGTAAGAAGCTGAG 349
GCTTGAGAGAGAACGTGTCC 11
30 MMR 2 10 ABCC8ex30F_and_ABCC8ex30R GACATTCCAGAGAGGGATAGC 411
ACACTAGGAGGACCACCAGG 11
31 MMR 2 10 ABCC8ex31F_and_ABCC8ex31R CCCTTGTGTGTGTCTGGTG 454
AACCTCCACCTGTCTGGG 11
32 MMR 2 10 ABCC8ex32F_and_ABCC8ex32R GATGGCAGCAAAAGGAATC 411
AGTTCTTTGGGATCAGCG 11
33 MMR 2 10 ABCC8ex33F_and_ABCC8ex33R AGTCCAAGGAGGAGTGTGTC 274
AGCATTGGGTTGGGCCCG 11
34 MMR 2 10 ABCC8ex34F_and_ABCC8ex34R GAAACAAGCCCAAACCTGTG 286
GGTGGCTGTGGGTACACG 11
35 MMR 2 10 ABCC8ex35F_and_ABCC8ex35R GTGTACCCACAGCCACCAG 300
CAACCCCCTCCTCTTTGTG 11
36 MMR 2 10 ABCC8ex36F_and_ABCC8ex36R ACCACCTCGGTGCTTCTC 363
TAGGACTAAATGGTCCTGCC 11
37 MMR 2 10 ABCC8ex37F_and_ABCC8ex37R CCATGCACACATTTTCCAAC 289
ATCCCACTAAACCCTTTCCAA 11
38 MMR 2 10 ABCC8ex38F_and_ABCC8ex38R GGACTAGGATCGGGGTCAG 307
CTGCTTCAGGGTTCTTTCTTG 11
39 MMR 2 10 ABCC8ex39F_and_ABCC8ex39R ACCCCAGGAAAGTGCAGTC 495
tttgctcacacagcttctgc 11
IVS8 MMR 2 10 ABCC8_Cryptic_Donor_FandR AGACACCGGCTCACAAGGT 290
TCTTAGAGGCTGGGAAGTGG 11
Gene: EIF2AK3 Accession No.: AF110146.1
PCR Primer concentration DNA concentration Size

1F MMR (5% 2 (5% DMSO) 10 EIF2AK3ex1F_and_EIF2AK3ex1R CCTAGCACGTCCTTGCCTTC 509
DMSO) CCCCTACACCGCATCCTC 2
1R MMR 2 (5% DMSO) 10 EIF2AK3ex1F_and_EIF2AK3ex1R CCTAGCACGTCCTTGCCTTC 509
CCCCTACACCGCATCCTC 2
2 MMR 2 10 EIF2AK3ex2F_and_EIF2AK3ex2R TGAGCATGTGGGATAAGTG 369
TGCCCTAAAGGGACACAAAC 2
3 MMR 2 10 EIF2AK3ex3F_and_EIF2AK3ex3R TCAGGATCAAGACTCCAGCTC 484
TGACAACCTCAGGGGAAAAT 2
4 MMR 2 10 EIF2AK3ex4F_and_EIF2AK3ex4R GTTGGTAATCTAACTGATGC 358
CCAACAGCAACATTA 2
5 MMR 2 10 EIF2AK3ex5F_and_EIF2AK3ex5R GCCCTCTTGTGGCATAAATC 485
GGGAGAGGAAGAACCGTA 2
6 MMR 2 10 EIF2AK3ex6F_and_EIF2AK3ex6R TACTTGGGGCTCTCAGCTTG 410
CACTCCTGAAGTAGGAAGG 2
7 MMR 2 10 EIF2AK3ex7F_and_EIF2AK3ex7R CCCTCCCTGTTTTTGTTGAA 425
GGGCAAAGACAGTCAGGATT 2
8 MMR 2 10 EIF2AK3ex8F_and_EIF2AK3ex8R CTGGGCCATTTGTTTAACTT 420

TGAAATTGTCTCCCAAGATG 2
(continued)
135
Table 1
136
(continued)
Gene: EIF2AK3 Accession No.: AF110146.1

9 MMR 2 10 EIF2AK3ex9F_and_EIF2AK3ex9R 388
AAGAAGAGAGACAAAACTTAAAAGGAA
GGAAGATCACTGAGAACTTTGG 2
10 MMR 2 10 EIF2AK3ex10F_and_EIF2AK3ex10R AAGACTGGAGGGATAGCAGT 407

AGATCTTAGGTCATTTCTTCTTTG 2
11 MMR 2 10 EIF2AK3ex11F_and_EIF2AK3ex11R TGAACTGATTTTCACATTACCAC 376
AATTGGCAGCACTTAGAACC 2
12 MMR 2 10 EIF2AK3ex12F_and_EIF2AK3ex12R GCCTTCAGTGTTGTCTTACT 420
CATTGTAATCACACAAGCAAA 2
13A MMR 2 10 EIF2AK3ex13AF_and_EIF2AK3ex13AR ACAGAGGGTGCAGTTCAGGT 572
GCTACTGGTGGGCTTGAAAG 2
13B MMR 2 10 EIF2AK3ex13BF_and_EIF2AK3ex13BR GAGGGGCACTCCTTTGAAC 589
GATGCTTTTACTCTCCCCAACTC 2
14 MMR 2 10 EIF2AK3ex14F_and_EIF2AK3ex14R TTGTCACTATTTTCCTGTTAGCC 384
TGGGTTTTAGATTACTGGGTATTT 2
15 MMR 2 10 EIF2AK3ex15F_and_EIF2AK3ex15R CCTGGGCTTTCCTTCTGTAA 500
TGAGCTTTAAATGGAAGCAAA 2
16 MMR 2 10 EIF2AK3ex16F_and_EIF2AK3ex16R 371
GATGTACAACCTCTTAGTCATTTTGTT
GGGAGCAGGTCTCTTTCCTC 2
17 MMR 2 10 EIF2AK3ex17F_and_EIF2AK3ex17R TTTTGCCAGCACTGATTTTA 403
TTTCAAGTCTGCAATTTTGG 2
Gene: FOXP3 Accession No.: NM_014009.2

1 MMR 2 10 FOXP3ex1F_and_FOXP3ex1R TCAAGAAAAGGAGAAACACAGAGAG 359
CACGGTAGCTGGGTACATCC X
2 MMR 2 10 FOXP3ex2F_and_FOXP3ex2R TTTGACCAGAGGAGTGTCCA 414
ACAGTAAAGGTCGGCACCTG X
3 MMR 2 10 FOXP3ex3F_and_FOXP3EX3R tgagcctcagtttccatacg 395
GGGCATCCACCGTTGAG X
4 MMR 2 10 FOXP3ex4F_and_FOXP3EX4R GACAGGCCACATTTCATGC 427
CCAAGCCTCTGAGACCTGAC X
5 and MMR 2 10 FOXP3ex5-6F_and_FOXP3ex5-6R CGCTCAAAATGAGAGGCCTTG 391
6 CACCCTAGACCTCTCCCCAC X
7 and MMR 2 10 FOXP3ex7-8F_and_FOXP3ex7-8R CCATTCAGAGCATTGAGCCAG 439
8 CACCGTGCAGACCTCCTC X
9 MMR 2 10 FOXP3ex9F_and_FOXP3ex9R CTTGAATCTGGGAGGTGGG 546
CCGAAAGGAAGCTTTTGTGA X
10 MMR 2 10 FOXP3ex10F_and_FOXP3ex10R ACGGGTGTTGACGGTGAG 297
GGAGGAACCCACTCTGAGG X
11 and MMR 2 10 FOXP3ex11-12F_and_FOXP3ex11-12R CCCTGATTACCTGCCCCTAC 605
12 CTGCCTCCCACCAGTTTG X
Gene: GATA4 Accession No.: NM_002052.3

1 MMR 2 10 GATA4ex1F_and_GATA4ex1R CGGAGATGTGGAGTGATTGG 394
CGGAAGGGGAAACTGAGG 8
2A AG60 5 10 GATA4ex2AF_and_GATA4ex2AR AAATTGGGATTTTCCGGAGT 477
GTGTGGGCACGTAGACTGG 8
137
(continued)
Table 1
(continued)
138

2B AG60 5 10 GATA4ex2BF_and_GATA4ex2BR CTCAGCTCGACACGGAGG 350
GTATGGAGGGCTGTCGGC 8
2C AG60 5 10 GATA4ex2CF_and_GATA4ex2CR CGCAGGGACCATGTATCAG 497
GGTAGGGGCTGGAGTAGGAG 8
2D AG60 5 10 GATA4ex2DF_and_GATA4ex2DR CTGCGGCCTACAGCAGTG 299
CCTCGACAGGGCTCAAGAC 8
3 AG60 5 10 GATA4ex3F_and_GATA4ex3R AGCCCGAGGTGGTCTTCT 435
TATTATGGGGCTCTCACCCA 8
4 MMR 2 10 GATA4ex4F_and_GATA4ex4R CGCAGGTGACAGGAGAGTTAG 373
AAGGAAGAAGACAAGGGAGGA 8
5 MMR 2 10 GATA4ex5F_and_GATA4ex5R TTGCTTAGGTGTTGCCTTCTC 353
TTTTTGCTGGGCTCTTCATC 8
6 MMR 2 10 GATA4ex6F_and_GATA4ex6R AGCCATCCCTGTGAGAACTG 387
GCTGGCCTCTGGGACTCT 8
7 MMR 2 10 GATA4ex7F_and_GATA4ex7R AGAAGTGCTCCTTGGTCCCT 449
TTCCCAGTTGTTGTTCTGGA 8

2A AG60 5 10 GATA6ex2AF_and_GATA6ex2AR AAGGAGTGGAGGCGAGGTAG 482
GGTCGAGGTCAGTGAACAGC 18
2B AG60 5 10 GATA6ex2BF_and_GATA6ex2BR CTCAGCTCGACACGGAGG 518
GTATGGAGGGCTGTCGGC 18
2C AG60 5 10 GATA6ex2CF_and_GATA6ex2CR CTCTCTCCAGCCAGGGTCC 484
GACAGCGAGCTGTACTGGG 18
2D AG60 5 10 GATA6ex2DF_and_GATA6ex2DR GCGCTTCCCCTACTCTCC 501
CTGCAAATCCTTCCTGGGAC 18
3 AG60 5 10 GATA6ex3F_and_GATA6ex3R AAAAGCTCAGCCGGGAAG 435
CTAGGGTGGGACCGCAG 18
4 MMR 2 10 GATA6ex4F_and_GATA6ex4R TCTTGGCCCAGAAAAGTCAG 491
AAAAGCACCTTCAATTCAATAA 18
5 and 6 MMR 2 10 GATA6ex5and6F_and_GATA6ex5and6R CATGCGCCAACAAGTCTG 801
CAAAACTTCTTGCTTTTACTTGG 18
7 MMR 2 10 GATA6ex7F_and_GATA6ex7R GAGCAGCTCTGGCCCTG 412
CAAAATAAAGGCACGAGAATCAC 18
Gene: GCK Accession No.: NM_000162.2
PCR Primer concentration DNA concentration Forward and reverse primer sequences (M13 tailed) chromosome Size
Exon method (pmol/μl) (ng/μl) number (bp)
1 MMR 2 10 GCKex1F_and_GCKex1R TCCACTTCAGAAGCCTACTG 195
TCAGATTCTGAGGCTCAAAC 7
2 MMR 2 10 GCKex2F_and_GCKex2R GGGGTCAGAAGACAGAAGGA 511
AGAGGAGCCAAGGGTGAGA 7
3 MMR 2 10 GCKex3F_and_GCKex3R ATATCCGGGCTCAGTCACC 332
CACCTCCCGTCAGGACTAGC 7
4 MMR 2 10 GCKex4F_and_GCKex4R TAGCTTGGCTTGAGGCCGTG 308
TGAAGGCAGAGTTCCTCTGG 7
5 MMR 2 10 GCKex5F_and_GCKex5R tcactcacggcagaagca tggaagccaaggagaaagg 7 634
6 MMR 2 10 GCKex6F_and_GCKex6R TGCGAGACGCTATCAAACG 358
GGCTCTGCTCTGACATCACC 7
7 MMR 2 10 GCKex7F_and_GCKex7R ccattgttccagacaaagca caagcccattatctgcaatg 7 436
139
(continued)
Table 1
(continued) 140
Gene: GCK Accession No.: NM_000162.2
8 MMR 2 10 GCKex8F_and_GCKex8R CATTTCTAAAGCTCTGGCTCATT 634
GTCCTGTCCCAGCCTCCT 7
9 MMR 2 10 GCKex9F_and_GCKex9R ctttggctgggtgaggtg agggggacgagaagaggac 7 574
10 MMR 2 10 GCKex10F_and_GCKex10R CCTCTTCTCGTCCCCCTTG 360
ATGGAGCCTGGGTGCTGT 7
PROM MMR 2 (80% MF) 10 GCKpromF_and_GCKpromR gctcggcatttcctgct agtcaggctgccaagaggt 7 512
Gene: GLIS3 Accession No.: NM_001042413.1

Prom A MMR 2 10 GLIS3 5’UTR(A)F_and_GLIS3 5’UTR(A)R aggaagcagctggctacagg 398
GCTCATTCTCCCCTGCTACA 9
Prom B MMR 2 10 GLIS3 50 UTR(B)F_and_GLIS3 50 UTR(B)R 244
ACGAGGATGAACAGGACAGG GCCTCCTTCGGAAATGAAA 9
Prom C MMR 2 10 GLIS3 50 UTR©F_and_GLIS3 50 UTR©R TGTAGCAGGGGAGAATGAGC 333
gcaaagttcctatggcatcc 9
2A MMR 2 10 GLIS3ex2AF_GLIS3Ex2AR ttgccgagagcactgtattg 642
CTTCCCATTGGTGAGCATTT 9
2B MMR 2 10 GLIS3ex2BF_and_GLIS3ex2BR ATCTCAAGATGCCCTCAGGA 284
CTGGCAGAAAATGGGATGG 9
3 MMR 2 10 GLIS3ex3F_and_GLIS3ex3R attggaattggagaaagcc 437
CGAACGGAAAACACTTGTGGG 9
4A MMR 2 10 GLIS3ex4AF_and_GLIS3ex4AR gggattgtcaatgcagtg aagccctcgacccgttg 9 510
4B MMR 2 10 GLIS3ex4BF_and_GLIS3ex4BR gggattgtcaatgcagtg aagccctcgacccgttg 9 510
4C MMR 2 10 GLIS3ex4CF_and_GLIS3ex4CR gggattgtcaatgcagtg aagccctcgacccgttg 9 510
5 MMR 2 10 GLIS3ex5F_and_GLIS3ex5R ccaaatctgatacagaaa cttttaccaggctccact 9 527
6 MMR 2 10 GLIS3ex6F_and_GLIS3ex6R Caattggtaacagttcat 338
TCTAAAGCAGGTTATACC 9
7 MMR 2 10 GLIS3ex7F_and_GLIS3ex7R Atagagcatgatgccttc 305
TCACGTGAGCATTCCTTT 9
8 MMR 2 10 GLIS3ex8F_and_GLIS3ex8R Tatgcttgctgtcaggct 334
TTCAGCAACTGTCAAGGC 9
9 MMR 2 10 GLIS3ex9F_and_GLIS3ex9R Gaatttggctttggagag 345
GCATCTGAAATCCACGAC 9
10 MMR 2 10 GLIS3ex10F_and_GLIS3ex10R Agctgggttggaatgatc 351
GTGCTTGGTCACGTCCAG 9
11 MMR 2 10 GLIS3ex11F_and_GLIS3ex11R tgtagtgcatttccagtg 304
GCTGACATCCTTCCTCAA 9
Gene: GLUD1 Accession No.: NM_005271.1

6 MMR 2 10 GLUD1ex6F_and_GLUD1ex6R TTAAATGAGAATGTGCTTTGACTT 496
TGAATTTGGTGATAGTTTGGTTG 10
7 MMR 2 10 GLUD1ex7F_and_GLUD1ex7R AACCAGTTAGTACATTGTTTCTTTTGG 393
AGAATGGACCCATGTTGCTG 10
10 MMR 2 10 GLUD1ex10F_and_GLUD1ex10R gtgggatgggaaggagtgt ttgtgcatttttggtctaagtttc 455
10
11 MMR 2 10 GLUD1ex11F_and_GLUD1ex11R GCAGAGTTTGCAGTGAGCTG 374
TGCCGCAGATGAAATCCA 10
12 MMR 2 10 GLUD1ex12F_and_GLUD1ex12R gttctgtgtggtgtccctgtt ggctgagatagcatggttgag 546

10
141
(continued)
Table 1
(continued)
142
Gene: HADH (SCHAD) Accession No.: NM_005327.2
DNA
PCR Primer concentration concentration Size
1 MMR 2 10 HADHex1F_and_HADHex1R CGTGTATACCCGCTCAACG 502
GTGAAAACTCCCTGGTGTCG 4
2 MMR 2 10 HADHex2F_and_HADHex2R TCGGTATTTTGAATGGAATGG 538
AACCTCAAATCCCACCCAAC 4
3 MMR 2 10 HADHex3F_and_HADHex3R GAACAGATAGGGTAGGCCAAT 436
GCATGGAACAAAACTGCAC 4
4 MMR 2 10 HADHex4F_and_HADHex4R CCCAATGGGTCAGGACAAC 425
TATGAGGCTCAGGGACAACC 4
5 MMR 2 10 HADHex5F_and_HADHex5R TTGTTTCTGGCTTGAGATTCC 439
AACTAGAACAGGAGGCACGG 4
6 MMR 2 10 HADHex6F_and_HADHex6R TGTACAGCTTGATAAATGGGG 363
TCACAAAGCTAGAAAACAGACACTG 4
7 MMR 2 10 HADHex7F_and_HADHex7R CCAAGCCAGAAAGTCTCAGTC 413
GTGAGGCAGGATTTTGGAAG 4
8 MMR 2 10 HADHex8F_and_HADHex8R CACCAGCACAGCCTTCCT 490
TGCTGTAACTGGTGTAATAATCACTTC 4
IVS5 MMR 2 10 HADH_Cryptic_Donor_FandR ttttaggtacagcttggtaaacaca tcatttcagcaactgggatt 4 297
Gene: HNF1A Accession No.: NM_000545.3

1 MMR 2 10 HNF1aEx1F_and_HNF1aEx1R gggtgcaaggagtttggttt ggcccctctaggctctcc 12 524
2 MMR 2 10 HNF1aEx2F_and_HNF1aEx2R TGGGCTCCATAACTGCTTTC 588
TCCCACTGACTTCCTTTCC 12
3 MMR 2 10 HNF1aEx3F_and_HNF1aEx3R GCATGTGTGCTGTGTGTTTG 501
AAGCCAATATCAGGAGTTCTCG 12
4 MMR 2 10 HNF1aEx4F_and_HNF1aEx4R ACTGTCAATTGCCCAAGGTC 540
GAATGGAATGGAACCAAACTG 12
5 MMR 2 10 HNF1Aex5F_and_HNF1Aex5R tggagtttgaagtgctgagg gccaaggaaagatgaggttg 12 320
6 MMR 2 10 HNF1aEx6F_and_HNF1aEx6R TGTAAGGAAAACCCAACCTCA 430
AGTGGCTCTTCCAGCTCCT 12
7 MMR 2 10 HNF1aEx7F_and_HNF1aEx7R AAGTCACCGCCTGCCTCT 516
GCCAACCTCTATCATCATCTCC 12
8 MMR 2 (80% MF) 10 HNF1aEx8F_and_HNF1aEx8R ACAAGAGGAGCTGGAGTTGG 513
AGACCTGGGGGAGCAGAG 12
9 MMR 2 (80% MF) 10 HNF1aEx9F_and_HNF1aEx9R ACCAAGCAGGTAAGGTCCAG 331
GTGACGGACAGCAACAGAAG 12
10 MMR 2 10 HNF1aEx10F_and_HNF1aEx10R GTACCCCTAGGGACAGGCAGG 328
ACCCCCCAAGCAGGCAGTACA 12
Prom MMR 2 10 HNF1aPrmF_and_HNF1aPrmR ACAAGGTTCTTTCGGGGGTG 550
TCTTTGCTCAGCCCTGACTC 12
Gene: HNF1B Accession No.: NM_000458.1

1 MMR 2 10 HNF1bEx1F_and_HNF1bEx1R CTGGATTTGGGGTTTGCTT 592
CGGGGGACTTCTCTGGTG 17
2 MMR 2 10 HNF1Bex2F_and_HNF1Bex2R TTTTGGCCTCATGTCTACCC 534
GCCACCTTCCTCATATCTGC 17
3 MMR 2 10 HNF1bEx3F_and_HNF1bEx3R CATCTCCAGCTCCACATGC 472
GAGGGTTCCTGGGTCTGTG 17
4 MMR 2 10 HNF1Bex4F_and_HNF1Bex4R actcccaaccaagactgctg 415
GATCCGTGGCAAGAACCA 17
143
(continued)
Table 1
(continued)
144
Gene: HNF1B Accession No.: NM_000458.1

5 MMR 2 10 HNF1bEx5F_and_HNF1Bex5R cctcgtggcgcttacattc ctggacagccctcattttcc 17 535
6 MMR 2 10 HNF1bEx6F_and_HNF1bEx6R TAGTCATGCCAAGGAATCG 361
GAGTTTGAAGGAGACCTACAG 17
7 MMR 2 10 HNF1bEx7F_and_HNF1bEx7R GGTGACTGGGACATTGAGC 414

ACTTCCGAGAAAGTTCAGACC 17
8 MMR 2 10 HNF1bEx8F_and_HNF1bEx8R CTCACAGCCAAGCATCCAC 631
TGTCATCCAGTCTCCAGCAA 17
9 MMR 2 10 HNF1bEx9F_and_HNF1bEx9R TTCAGGATTCAGGTCAGCAAG 430
AGGTCACTGGGCTTTTCCA 17
Gene: HNF4A Accession No.: NM_175914.4

P2 + 1D MMR 2 10 HNF4aEx1F_and_HNF4aEx1R CCAGGTTGGACTCTCACCTCTC 535
GTGTCCCATGGCCTCCCAAAG 20
2 MMR 2 10 HNF4aEx2F_and_HNF4aEx2R AGGTGATGGAGTGGGAACAG 380
AGACCTTGGGGCCATGAG 20
3 MMR 2 10 HNF4aEx3F_and_HNF4aEx3R AGATGAGAGCACTGAGGTTGG 309
GCCTGCCACTGAGTCATAAAG 20
4 MMR 2 10 HNF4aEx4F_and_HNF4aEx4R GCTCCCACTCCTCATCAGTC 250
CAAACTGGGCCATGTGAAAC 20
5 MMR 2 10 HNF4aEx5F_and_HNF4aEx5R CCTCCGTTTTTACCCTGAGC 435
CCACGGCTATATCCCAGGT 20
6 MMR 2 10 HNF4aEx6F_and_HNF4aEx6R CACAGTTCAGGCAGGTAGAGG 333
GCCACCATGTGAATCTCCTT 20
7 MMR 2 10 HNF4aEx7F_and_HNF4aEx7R TTGCCAACTTAAAAGCCAAAAC 280
TGGAGAGAGAGTCAGGGATGG 20
8 MMR 2 (80% MF) 10 HNF4aEx8F_and_HNF4aEx8R TTGTTGAGGTCCCTGAATCC 467
CAAGGAGCTGAGGGGTGAG 20
9 MMR 2 10 HNF4aEx9F_and_HNF4aEx9R TGGTTGATTGGCCACGCCTG 377
ATCCTGGTTCTACCTTCTAG 20
10 MMR 2 10 HNF4aEx10F_and_HNF4aEx10R TGGGACTCACAGAAGGTTGA 434
CACCAGGTGCTCTCTTAGGG 20
Gene: INS Accession No.: NM_000207.2

1 MMR 2 10 INSex1F_and_INSex1R CTGTGAGCAGGGACAGGTCT 627
GCACAGGTGTTGGTTCACAA 11
2 MMR 2 10 INSex2F_and_INSex2R ctctgcagcagggaggac gggagctggtcacttttagg 11 522
3 MMR 2 10 INSex3F_and_INSex3R CCCTGACTGTGTCCTCCTGT 423
AGAGAGCGTGGAGAGAGCTG 11
Gene: INSR Accession No.: NM_000208.2

1 MMR 2 (80% MF) 10 INSRex1F_and_INSRex1R GGGCGTGGAAGAGAAGGAC 294
GGCTCGATTTTGGCTTGG 19
2A MMR 2 10 INSRex2AF_and_INSRex2AR GCTCTGCCCCTGATCCTTC 381
TGTAGAGGCCGAGTTCCTTG 19
2B MMR 2 10 INSRex2BF_and_INSRex2BR TCAAAACGAGGCCCGAAG 583
AATGCCACCACCCACTATTC 19
3 MMR 2 10 INSRex3F_and_INSRex3R TGTTTGGTTGGCTTTCACTG 541
AGTTTTAACAAGCGGCATCG 19
145
(continued)
Table 1
(continued)
146
Gene: INSR Accession No.: NM_000208.2

5 MMR 2 10 INSRex5F_and_INSRex5R CAACAGACGCAGCCAGTG 399
TTAGCACTCAGGCCATACACAC 19
6 MMR 2 10 INSRex6F_and_INSRex6R TATGTGCCAAGCAAGTGGAG 384
GGTCCCTCATGCCAAAAAG 19
7 MMR 2 10 INSRex7F_and_INSRex7R TTCCCCCATTCATACTCAGC 367
TGGAGCACAAACGTAGCAAG 19
8 MMR 2 10 INSRex8F_and_INSRex8R TCCTTTCTCTTTGGCTGTTCC 467
AAAGCAAGAGGTCTGATTCACATAC 19
9 MMR 2 10 INSRex9F_and_INSRex9R CATCCTCCCACCAGCTTTC 322
GATAGCTGCTTCCCTAGAGGTG 19
10 MMR 2 10 INSRex10F_and_INSRex10R ATGTGTGTTCAGCCGCAGAG 389
AAGGGCTCCATTCAGACTCC 19
11 MMR 2 10 INSRex11F_and_INSRex11R GATTTCTTCCCCTTCCCTTG 377
AAGCATCTGCTCTCCAGCAC 19
12 MMR 2 10 INSRex12F_and_INSRex12R TAATCCATGTTCCCCATTGC 496
GCACTGCTTAGAGGGTGGAG 19
13 MMR 2 10 INSRex13F_and_INSRex13R TGTGGGATGAGTTTTAGATGTCC 371
GAGGAAGGAGGTACCAAGTGC 19
14 MMR 2 10 INSRex14F_and_INSRex14R AAATTCTGCAATTCTCAAAGTCG 385
ATGTCCCACCATGCTCAGTG 19
15 MMR 2 10 INSRex15F_and_INSRex15R GTGGAAGAGCAGCAGAGAGG 285
CAACTGTTCCCAGCACACC 19
16 MMR 2 10 INSRex16F_and_INSRex16R AGGATTATGGGGATTCTGCTG 249
GAAGGCAAAGGAAGCTGATG 19
17 MMR 2 10 INSRex17F_and_INSRex17R GCATGGGTCCTGGATCAC 467
CTGTGTCCTCTGTCGCTCTG 19
18&19 MMR 2 10 INSRex18&19F_and_INSRex18&19R ATAGACACCAGGGAGGGAGG 546
AAGACTTAACGGCTCATTATAGACAAC 19
20 MMR 2 10 INSRex20F_and_INSRex20R TTACTGGATTTGGCCACCAC 399
CTGCCTTCCTTTCCTTGATG 19
21 MMR 2 10 INSRex21F_and_INSRex21R TTTATGAACTGCAGCATGAATTG 593
GGGCAATACCCTTTCAACG 19
22 MMR 2 10 INSRex22F_and_INSRex22R ACCATCTCTCAGCACCTCTAGC 587
AGGAAAGCGAAAATGGGAAC 19
Gene: IPF1 Accession No.: NM_000209.1

1 MMR 2 10 IPF1ex1F_and_IPF1ex1R AACGCCACACAGTGCCAAATC 651
TTAGTCCGACCCGGGATAATC 13
2A MMR 2 10 IPF1ex2AF_and_IPF1ex2AR CGCTGAAATGGGATGCTG 405
CGCTTCTTGTCCTCCTCCTT 13
2B MMR 2 10 IPF1ex2BF_and_IPF1ex2BR TGAACTTGACCGAGAGACACA 500
CTGAGAGAGCGGGTTTTCC 13
Gene: KCNJ11 Accession No.: NM_000525.3

1B MMR 2 10 KCNJ11ex1BF_and_KCNJ11ex1BR GTGCCCACCGAGAGGACT 524
AGTGGGCACTCCTCAGTCAC 11
1C MMR 2 10 KCNJ11ex1CF_and_KCNJ11ex1CR CACCAGCATCCACTCCTTCT 546
GTTTCCACCACGCCTTCC 11
1D MMR 2 10 KCNJ11ex1DF_and_KCNJ11ex1DR CTACCATGTCATTGATGC 491
CCACATGGTCCGTGTGTA 11
147
(continued)
Table 1
148
(continued)
Gene: LMNA Accession No.: NM_005572.2

1 MMR 2 10 LMNAex1F_and_LMNAex1R CCGAGCAGTCTCTGTCCTTC 564
CCCTCTCACTCCCTTCCTG 1
2 MMR 2 10 LMNAex2F_and_LMNAex2R CTGGTAATTGCAGGCATAGC 344
GTTAGGTGGGGCCATGGGT 1
3 MMR 2 10 LMNAex3F_and_LMNAex3R atgtgtgaaggggtgcacag ccagccctactccatgagc 1 419
4 MMR 2 10 LMNAex4F_and_LMNAex4R TAAAGTGGGGCTGGTAGTGG 380
TAAGGGTAGGGCTGCCAAG 1
5 MMR 2 10 LMNAex5F_and_LMNAex5R GGGATCAGGCAGATGGTG 385
ACCCTTCTCTGTGGTTGTGG 1
6 MMR 2 10 LMNAex6F_and_LMNAex6R CTGGGGAAGCTCTGATTGC 420
CAGAGGACACTGCCAGCAC 1
7 MMR 2 10 LMNAex7F_and_LMNAex7F gtccctccttccccatactt tctcacagccaaagagtcca 1 813
8 MMR 2 10 LMNAex8F_and_LMNAex8R gttgtcaggaagatgaaagataagg 395
cctccccagagtcccaag 1
9 MMR 2 10 LMNAex9F_and_LMNAex9R ATGGAGATGATCCCTTGCTG 420
tctagaaaggggccctgaat 1
10 MMR 2 10 LMNAex10F_and_LMNAex10R CAGGCCACAAGAAAAGTTGC 388
GTTCAAGGTATAGGGAGGAG 1
11 MMR 2 (80% MF) 10 LMNAex11F_and_LMNAex11R GGGCACAGAACCACACCTT 610
CAGACAAGAGGGGCAGGA 1
12 MMR 2 10 LMNAex12F_and_LMNAex12R GGGAGATGCTACCTCCCTTC 240
GGGTTATTTTTCTTTGGCTTCA 1
Gene: NEUROG3 Accession No.: NM_020999.2
2A MMR 2 10 NEUROG3ex2AF_and_NEUROG3ex2AR ctattcttttgcgccggtag 499
cagtgccgagttgaggttg 10
2B MMR 2 10 NEUROG3ex2BF_and_NEUROG3ex2BR gagttggcactgagcaagc 475
ccctctcccttacccttagc 10
Gene: NEUROD1 Accession No.: NM_002500.1

2A MMR 2 10 NeuroD1ex2AF_and_NeuroD1ex2AR CAAGCATTTGTACAGGTTTAG 443
CTCCAGGCGAGCCTTAGTCATC 2
2B MMR 2 10 NeuroD1ex2BF_and_NeuroD1ex2BR CCTCGAAGCCATGAACGCAG 619
GCTGTCCATGGTACCGTAAG 2
2C MMR 2 10 NeuroD1ex2CF_and_NeuroD1ex2CR CCTGCAACTCAATCCTCGGAC 597
CTGTAAGCACAGTGGGTTCG 2
Gene: NKX2-2 Accession No.: NM_002509.3

1 MMR 2 10 NKX22ex1F_and_ NKX22ex1R CAATAATTAAAGCCAATTcccc 491
CTTCCCCTTTCACTCCCAG 20
2A MMR 2 10 NKX22ex2AF_and_ NKX22ex2AR CCAGGGTGCTCCGAGTC 459
GTTTGCCGTCCCTGACC 20
2B MMR 2 10 NKX22ex2BF_and_ NKX22ex2BR CAGCGGTACCTGTCGGC 460
CACCATAAGGACCGAGGC 20
(continued)
149
Table 1
(continued)
150
Gene: PPARG Accession No.: NM_015869.4

1 MMR 2 10 PPARG_1F_and_PPARG_1R AGTTCACGCCCCTCACAA 534
GACAAACTACAAGAGCAATTTTACC 3
2 MMR 2 10 PPARG_2F_and_PPARG_2R CTGGGTAAAGGGTGACTTCC 597
TGGGTGAATAAATGAATGGTGA 3
3 MMR 2 10 PPARG_3F_and_PPARG_3R TTTCTTTTTCTTGCCTTAACTCC 734

CCTTGGAGAAGTCACCTTTTTG 3
4 MMR 2 10 PPARG_4F_and_PPARG_4R GCAGAGAACATTTAGTTCCAGAGA 550
GCCCAACCACAGAGAGAAAG 3
5 MMR 2 10 PPARG_5F_and_PPARG_5R TGATGGTCTGTGCTACTTTTGTG 585
TGAAAACACTTCCTGCGTTTC 3
6A MMR 2 10 PPARG_6AF_and_PPARG_6AR TTGAAGGGAAAGAAGACCAAAA 522
AGGAGGCCAGCATTGTGTAA 3
6B MMR 2 10 PPARG_6BF_and_PPARG_6BR TGGAGGCTGTGCAGGAGA 535
CCGCAAACCTATGACACTCTTT 3
7A MMR 2 10 PPARG_7AF_and_PPARG_7AR acccttcctccccacctatt 456
GGCAGTGGCTCAGGACTCT 3
Gene: PTF1A Accession No.: NM_178161.2

1A AG58 2 10 PTF1Aex1AF_and_PTF1Aex1AR ctcggacgggccttagaaac tcgcggtagcagtactcgtg 10 304
1B AG58 2 10 PTF1Aex1BF_and_PTF1Aex1BR acgacttcttcaccgaccag gggtaggccaggcacgag 10 340
1C AG58 2 10 PTF1Aex1CF_and_PTF1Aex1CR ggctactgctgcgagacg agcgtgtccaccttggag 10 356
1D AG58 2 10 PTF1Aex1DF_and_PTF1Aex1DR atcaacgacgccttcgag gagtgcccgaccccttct 10 332
2 AG58 2 10 PTF1Aex2F_and_PTF1Aex2R gggacttgaagtgtggagga 492
AAACAGTTGATTGCCATTCG 10
REG MMR 2 10 PTF1AexREGF_and_PTF1AexREGR GATGATTCTTCCCTAGAAGTCTCAG 488
CAAGAAAGTCAAGTGGTGCATG 10
Gene: RFX6 Accession No.: NM_173560.3

1 MMR 2 10 RFX6ex1F_and_RFX6ex1R GCCTGGGCTGTACCTGTG 495
TGTTTGTTCGGGATGCATTA 6
2 MMR 2 10 RFX6ex2F_and_RFX6ex2R TGCATCCCGAACAAACATAA 487
CCCAAAAAGTTTGAGGAAGAA 6
3 MMR 2 10 RFX6ex3F_and_RFX6ex3R CATGCTATCTGCCTGACATGA 246
TGAGAAAGTTGAAGGAAAAGGAA 6
4 MMR 2 10 RFX6ex4F_and_RFX6ex4R GTGGCTTGCATTAGGACCAT 172
TCAATCTTCATGCACAAGAGC 6
5 MMR 2 10 RFX6ex5F_and_RFX6ex5R TCATCAGGGTTTGCAGTTCT 249
GGTATCATGCATATTTAAAGAAACACA 6
6 MMR 2 10 RFX6ex6F_and_RFX6ex6R TCCAGCTTATTTCTCTAATCATGG 237
ACATTTGCCAAAGGGAATAA 6
7 MMR 2 10 RFX6ex7F_and_RFX6ex7R TGGTTTGCAGGATGAAGAAG 300
TACCATGAAATTCCCAGCAA 6
8 MMR 2 10 RFX6ex8F_and_RFX6ex8R AAAGGCAGTGTTAAAAGGATGC 201
TGAACTCATTAAGAACAGAACATGAA 6
9 MMR 2 10 RFX6ex9F_and_RFX6ex9R TTTCATGTTCTGTTCTTAATGAGTTC 236
GCTACGTAAAGAAATAAATGGCTCA 6
10 MMR 2 10 RFX6ex10F_and_RFX6ex10R GCTAGGAATAAAAAGTAGTTGACCA 369
CCAGCAAATACAAAGGGACAA 6
11 MMR 2 10 RFX6ex11F_and_RFX6ex11R GGCAGATCTAGGCTGTCTGA 300
TTCCTCAATCTGTCCTTCTCA 6
151
(continued)
Table 1
(continued)
152
Gene: RFX6 Accession No.: NM_173560.3

12 MMR 2 10 RFX6ex12F_and_RFX6ex12R CTCAAGGAACTATTTTTCTGATAGC 383
CAGCTCAAGATGTCAATTTTCTG 6
13 MMR 2 10 RFX6ex13F_and_RFX6ex13R TCTGTCTTTCCCTTTCATGC 297
GCCTGCCTTATTCGTAGAGC 6
14 MMR 2 10 RFX6ex14F_and_RFX6ex14R TGCCATCTGTCCATTTGTTAG 328
GACAGGAACTGTGACTGAGGA 6
15 MMR 2 10 RFX6ex15F_and_RFX6ex15R TCCCCTTTTTACATTTCACACTG 700
GACATGTTTGGAGAAAAACTCTGA 6
16 MMR 2 10 RFX6ex16F_and_RFX6ex16R TGGGTTTATCTACGAGGAATGTG 357
TGAAGCAACAGGCTTCTCTG 6
17A MMR 2 10 RFX6ex17AF_and_RFX6ex17AR TGAAGCAAGCTGGAAAACAA 495
AGAGGTGGAACGCAATGTCT 6
17B MMR 2 10 RFX6ex17BF_and_RFX6ex17BR AGAGGTGGAACGCAATGTCT 495
TGAAGCAAGCTGGAAAACAA 6
18 MMR 2 10 RFX6ex18F_and_RFX6ex18R TTTTCCTTCCTGAAAGTTCTTTGT 487
TGGAAACATGCTTGCTCAGT 6
19 MMR 2 10 RFX6ex19F_and_RFX6ex19R TGAGTGGCATTTGCAGAGAA 246
CAGCCACTGCATATTCATTTT 6
Gene: SLC19A2 Accession No.: NM_006996.2
1 MMR 2 10 SLC19A2ex1F_and_SLC19A2ex1R gcgtccgctgtgattggtt 606
gagaaaagcgctcgagcc 1
2A MMR 2 10 SLC19A2ex2AF_and_SLC19A2ex2AR agatctttgaggtatttgtagg 534
acacaggtaagagagatgaca 1
2B MMR 2 10 SLC19A2ex2BF_and_SLC19A2ex2BR acagccactgaaattgccta 529
agatctaccaagagggagttt 1
3 MMR 2 10 SLC19A2ex3F_and_SLC19A2ex3R ttcgccagaggggataaatg 596
cctgctccacttgagtactt 1
4 MMR 2 10 SLC19A2ex4F_and_SLC19A2ex4R ccctccataatcttgagctatt 476
ttcctcccatttgcctcattt 1
5 MMR 2 10 SLC19A2ex5F_and_SLC19A2ex5R gttggaaaggcaattgacagt 423
actttacatctgttccctattg 1
6 MMR 2 10 SLC19A2ex6F_and_SLC19A2ex6R ctcaggcagtcaggctttatt 470
gctgctgtgaagtcaagaaat 1
Gene: SLC2A2 (GLUT2) Accession No.: NM_000340.1
1 MMR 2 10 SLC2A2_Ex1F_and_SLC2A2_Ex1R ctctccttgctcctcctcct 297
aaacagttcttcaattggcaaag 3
2 MMR 2 10 SLC2A2_Ex2F_and_SLC2A2_EX2R cgatattgaggggaaagaacc 384
gggagtcaaggccaagttatc 3
3 MMR 2 10 SLC2A2_Ex3F_and_SLC2A2_Ex3R tggccaactgtaaacaaatct 518
aaagctattccacaagaagaaaga 3
4 MMR 2 10 SLC2A2_Ex4F_and_SLC2A2_Ex4R ccccaactagcttttacagca 398
tgctattgcctttccctctg 3
5 MMR 2 10 SLC2A2_Ex5F_and_SLC2A2_Ex5R aattcagggaggggctttc 339
gagggacgagatggatgaag 3
6 MMR 2 10 SLC2A2_Ex6F_and_SLC2A2_Ex6R cttgaaccaaagttgtcagaga 418

catgcaccagtcatataaaagc 3
(continued)
153
Table 1
(continued)
154
Gene: SLC2A2 (GLUT2) Accession No.: NM_000340.1
7 MMR 2 10 SLC2A2_Ex7F_and_SLC2A2_Ex7R gaatttccaacacccatgct 432
gacccatgatgccaatacct 3
8 MMR 2 10 SLC2A2_EX8F_and_SLC2A2_Ex8R gcatgccgctatagaaacaa 400
tgcagtggttggcttattca 3
9 MMR 2 10 SLC2A2_EX9F_and_SLC2A2_EX9R tccccaccttgatctcactc 397
ttatgctgttgctctggcttt 3
10 MMR 2 10 SLC2A2_EX10F_and_SLC2A2_EX10R aaaagttgtatgccgcctgt 436
agcactccagcaaagaggaa 3
11 MMR 2 10 SLC2A2_EX11F_and_SLC2A2_EX11R gcattcagcaattggacctg 482
tgacatttctgatgagagcaca 3
Gene: STAT3 Accession No.: NM_139276.2

2 MMR 2 10 STAT3_2F_and_STAT3_2R CCCCAGAGCATCTTTATCCC 396
CTCATTTTCCCCATCACCTG 17
3 MMR 2 10 STAT3_3F_and_STAT3_3R GGGTTATAGCATCAGGTTTGC 432
AAGTATACAGAGCTTTGAGAAAGGG 17
4 MMR 2 10 STAT3_4F_and_STAT3_4R TAGTAACGACCTCCCCTTCG 512
TCTGTTGGATTCTTTTGGTGG 17
5 MMR 2 10 STAT3_5F_and_STAT3_5R TTCCCTTCCTCTTGTGATGG 594
CAAGAGAAGGCTCCCTGTTG 17
6 MMR 2 10 STAT3_6F_and_STAT3_6R AACAGGGAGCCTTCTCTTGG 382
ATGACCAGGCTCCTTTGAGG 17
7 MMR 2 10 STAT3_7F_and_STAT3_7R GGAGGTACGGGTCCTCAAAG 967
CAACTCCAGAGCAGGAACTTCT 17
8 MMR 2 10 STAT3_8F_and_STAT3_8R ATTTCAGCGTCTTGTGGCAG 402
GCTAAATTTGAATATGGAAAAGTCC 17
9 MMR 2 10 STAT3_9F_and_STAT3_9R TTTTCAGCATCCACCCAAC 510
GGAAAGAGAAGATGGGCTCAC 17
10 MMR 2 10 STAT3_10F_and_STAT3_10R GGTAATTTAGCATCCTTGTCCC 362
ATGGCAACAAATTTCAACCC 17
11 MMR 2 10 STAT3_11F_and_STAT3_11R GACAGCTTGGCCTATTTACCTG 295
TGTCCACAAAATGAAGATCTCTG 17
12 MMR 2 10 STAT3_12F_and_STAT3_12R TGCGCTGATCAACTGTAACTG 492
ATTCCCACATCTCTGCTCCC 17
13 MMR 2 10 STAT3_13F_and_STAT3_13R ATTCCCACATCTCTGCTCCC 492
TGCGCTGATCAACTGTAACTG 17
14 MMR 2 10 STAT3_14F_and_STAT3_14R ATGGAAGAATCCAACAACGG 345
GTTCATGTCACTTTGGCCTG 17
15 MMR 2 10 STAT3_15F_and_STAT3_15R TGCTGCTTAGACTGGTCTCG 251
CCCCTGTACGTAGCCTCTCA 17
16 MMR 2 10 STAT3_16F_and_STAT3_16R CACTCCTCGCCTAGAGTTGG 449
GTCCTCGCTTGGTGGTG 17
17 MMR 2 10 STAT3_17F_and_STAT3_17R AACATGCTGACCAACAATCC 397
GCCTTGCTCAGGAAAGAAAC 17
18F MMR 2 10 STAT3_18F_and_STAT3_18R AAATCCTCAGGCCCGTCTAC 323
CCTTCAAAGATGTGAAAGCTG 17
18R MMR 2 10 STAT3_18F_and_STAT3_18R CCTTCAAAGATGTGAAAGCTG 323
AAATCCTCAGGCCCGTCTAC 17
19 MMR 2 10 STAT3_19F_and_STAT3_19R CTGAACTCTTGGTCCAGCG 365

AAAGCCCATGATGTACCTGG 17
155
(continued)
Table 1
(continued)
156
Gene: STAT3 Accession No.: NM_139276.2

20 MMR 2 10 STAT3_20F_and_STAT3_20R GCTGGCAAGGGCTTCTC 524
AAGCAAACCAATCCTTCAGC 17
21 MMR 2 10 STAT3_21F_and_STAT3_21R CACTACAATTCTTTCCCATAAGGAG 483
AACAGGGTGTTCAGGGTCTC 17
22 MMR 2 10 STAT3_22F_and_STAT3_22R TAAATGAGGGCAGACAACCC 397
TCAAACTCTGGTCTCCAACAG 17
23 MMR 2 10 STAT3_23F_and_STAT3_23R AGCCCCTGGGCTATGTTTAG 428
TCTCTTTTGGAAAGCAAAGCTC 17
24 MMR 2 10 STAT3_24F_and_STAT3_24R TCCAGGGAGGAGGGTAAATC 425
AGCAGATCACCCACATTCAC 17
Gene: ZMPSTE24 Accession No.: NM_005857.2
Primer DNA
1 MMR 2 10 ZMPSTE24ex1F_and_ZMPSTE24ex1R TCGGAAAGAACGGATATTGC 381
GACCACAAAGACGAGACTGG 1
2 MMR 2 10 ZMPSTE24ex2F_and_ZMPSTE24ex2R ATCAAGTGCAATATTTGTGTAGG 420
TCTGGGACTTGTAAGTGTGG 1
3 MMR 2 10 ZMPSTE24ex3F_and_ZMPSTE24ex3R GATGGGGTCTCAACTCACCT 373
CCTGCCAAGCTAAAAGAAATACC 1
4 MMR 2 10 ZMPSTE24ex4F_and_ZMPSTE24ex4R GATGATATTTCAGACCTTATGTTATCC 432
ATGATCTTTACCATATCTGTAGTTCC 1
5 MMR 2 10 ZMPSTE24ex5F_and_ZMPSTE24ex5R TGCAAGACATTTACCCATTGT 575
TGGTCAGTCAATACTCCTGTGTC 1
6 MMR 2 10 ZMPSTE24ex6F_and_ZMPSTE24ex6R GGAATACCAGAGCAAGTAAGTTCC 471
TAAAAAGCCAAAAGGTAAACTCC 1
7 MMR 2 10 ZMPSTE24ex7F_and_ZMPSTE24ex7R actgctggtggtgaattatgtt aaggagccatggttttcatct 475
1
8 MMR 2 10 ZMPSTE24ex8F_and_ZMPSTE24ex8R TGTCTCATGTGGAGCAGTGG 415
CAGGTAGCCTGGCTGTGG 1
9 MMR 2 10 ZMPSTE24ex9F_and_ZMPSTE24ex9R ACTTTATGGATGCTACTGATCC 406
TTGGGCTTGAATTTTTCTGC 1
10 MMR 2 10 ZMPSTE24ex10F_and_ZMPSTE24ex10R AGGCTCAATTTATAATTTCCTCACC 504
CATCAAGAGCTGGAACATGC 1
157
2.6 Next-Generation 1. 0.65 ml Nonstick hydrophobic tubes (Anachem Cat

Targeted Sequencing No. 1110-10).
2. TE (10 mM Tris–HCl, 1 mM EDTA, pH 8.0) (Sigma-Aldrich
93,283—100 ml).
3. AMPure (Beckman Coulter A63881).
4. Ethanol (Sigma-Aldrich E7023—500 ml).
5. dH2O (Baxter UKF7114).
6. End-repair buffer (New England Biolabs E6050L).
7. 10 dA tailing buffer (New England Biolabs E6053L).
8. Klenow fragments (New England Biolabs E6053L).
9. Next Flex96 DNA barcodes (adapters; BioScientific).
10. Fast-link DNA ligase (Epicentre, Illumina (LK6201H)).
11. 10 mM ATP (Epicentre, Illumina (LK6201H).
12. 5 Herculase buffer (Agilent).
13. Herculase dNTP mix (100 mM) (Agilent).
14. Herculase II polymerase (Agilent).
15. NEXTflex primers (see Tables 1–3).
16. SureSelect Hybe #1 (Agilent (5972-3551)).
20. SureSelect Index block #1 (Agilent (5972-3641)).
23. SureSelect RNase block (Agilent (5910-1976)).
24. Library: Custom Agilent SureSelect assay [4].
25. Dynabeads MyOne (Invitrogen (2021-02)).
26. SureSelect binding buffer (Agilent (5972-3554)).
27. SureSelect wash buffer 1 (Agilent (5972-3555)).
28. SureSelect wash buffer 2 (Agilent (5972-3556)).
Table 2
NEXTflex primer sequences
NEXTflex primer AATGATACGGCGACCACCGAGATC

forward TACAC
NEXTflex primer reverse CAAGCAGAAGACGGCATACGAGAT
Table 3
Nextflex adapter sequences
Name/sequence Name/sequence Name/sequence Name/sequence

HT1/AACGTG HT38/CCGTGA HT75/ACAGAT HT17/AAGGTA
HT13/AACAAC HT50/GCCACA HT87/CCTAAT HT29/ATTGAG
HT25/AGATCG HT62/TCCGTC HT4/AGTGGT HT41/CGACTG
HT37/CCGAAG HT74/ACACGA HT16/AAGACG HT53/GCTCGG
HT49/GATAGA HT86/CCGACA HT28/ATCCTG HT65/TGGAAC
HT61/TATCAG HT3/ATGCCT HT40/CGAACT HT77/AGCACC
HT73/AATGTT HT15/AACGCT HT52/GCTAAC HT89/CGACAC
HT85/CCATCC HT27/AGTCAC HT64/TGAAGA HT6/ACATTG
HT2/AAACAT HT39/CCTCCT HT76/AGATGT HT18/ACACAG
HT14/AACCGA HT51/GCGAGT HT88/CCTCTA HT30/CAACCA
HT26/AGCAGG HT63/TCTTCA HT5/ACCACT HT42/CGCATA
HT54/GGAGAA HT66/TGGCTT HT78/AGCCAT HT90/CGGATT
2.7 Equipment 1. Thermocycler.

2. Agencourt SPRIPlate® magnetic plate.
3. Agencourt 384 Direct Inject Magnet Plate.
4. ABI 3730XL.
5. Illumina HiSeq2500.
6. Bioruptor®.
7. Agilent 2200 TapeStation®™.
8. Vacuum concentrator (SpeedVac).
2.8 Methods: Sanger The PCR and sequencing methods detail primers and PCR condi-
Sequencing tions for Sanger sequencing of all coding regions and intron–exon
boundaries for single gene analysis to detect missense, nonsense,
frameshift, and splicing mutations. For a list of genes, primer pairs,
amplicon sizes, and optimized PCR conditions for each amplicon
see Table 1.
3 Methods
3.1 PCR A PCR is set up for each of the primer pairs using one of the
Amplification following conditions (Tables 4 and 5).
Table 4
MegaMix Royal (MMR) 10 μl PCR reaction per sample
Per sample (n) (μl)

Megamix 5
Primer mix 2.5
DNA (10 ng/μl) or dH2O for the negative control 2.5
Table 5
AmpliTaq Gold (AG) with dNTP mix 2 and mix 3, 25 μl PCR reaction per sample

dH2O 7.9
10 buffer without MgCl2 2.5
MgCl2 2.5
dNTPs (see mix 2) 1.0
Deaza-dGTP (see mix 3) 1.25
DMSO 1.25
Betaine 5.0
Primer mix 1
AmpliTaq Gold 0.1
DNA (10 ng/μl) or dH2O for the negative control 2.5
3.2 PCR Cycling The following (Tables 5–8) outlines the thermocycler programs for
Conditions using MegaMix Royal and AmpliTaq Gold (see Note 2).
3.3 PCR Product The PCR product is purified prior to sequencing to remove excess
Purification Prior primers, primer dimers, unincorporated dNTPs, salts, and other
to Sanger Sequencing contaminants. There are several commercially available methods for
purifying PCR products. We use the Agencourt AMPure XP PCR
purification system, a solid-phase paramagnetic bead technology
that is highly amenable to automation and high sample numbers.
We use the Biomek NXP (Beckman Coulter) to perform our liquid
handling, though other liquid handling workstations are available.
To summarize, the PCR products are bound to paramagnetic beads
and separated before being washed to remove any contaminants.
The purified PCR products are then eluted from the magnetic
beads ready for the sequencing reaction.
1. Transfer 5 μl of the PCR products to a clean 384-well plate.
2. Briefly centrifuge the plate so that the PCR product is at the
bottom of the well.
Table 6
MegaMix Royal (MMR) PCR cycling conditions
Step Temperature Time

Step 1 95 C 5 min
Step 2 95 C 30 s
60 C 1 min
Step 3 (repeat step 2 for 30 cycles)
Step 4 72 C 10 min
Table 7
AmpliTaq Gold (AG60) PCR cycling conditions

Step 1 95 C 12 min
Step 2 94 C 1 min

60 C 1 min

72 C 2 min
Step 4 72 C 10 min
Table 8
AmpliTaq Gold (AG58) PCR cycling conditions

Step 1 95 C 12 min

Step 2 94 C 1 min
58 C 1 min
72 C 2 min
Step 4 72 C 10 min
3. Gently shake the Agencourt AMPure bottle to resuspend the

solution so that it appears homogenous and consistent in color.
4. Add 9 μl of the Agencourt AMPure XP magnetic beads to the
PCR reaction and mix by pipetting (15 times) so that the color
of the mixture appears homogenous.
5. Place the plate onto the magnetic stand for 3–5 min. This step
separates the beads containing the bound PCR products from
the solution.
6. The beads form a spot on the side of the well. Remove the
cleared solution from the plate and discard. This step needs to
be performed while the plate is situated on the magnetic stand
so that the bead pellet is not disturbed.
7. Dispense 30 μl of 85% ethanol to each well of the plate while
placed on the magnetic stand. Incubate for 30 s at room
temperature. Remove the ethanol and discard (see Note 3).
8. Repeat step 6 (total of two washes).
9. Allow the plate to air dry completely.
10. Add 30 μl of elution buffer (dH2O).
11. Transfer the purified PCR products into a new plate.
3.4 Sanger Prepare a sequencing master mix (n + 5) if there are more than
Sequencing Reaction 10 samples. If there are less than 10 samples, make up a sequencing
mix of n + 1. We perform unidirectional sequencing (forward
direction) and all our primers are M13 tailed so that a universal
sequencing master mix can be used. Note: for samples shown to
contain a frameshift mutation or if mosaicism is suspected, the
reverse sequencing is performed using a revere M13 (M13R)
primer so that bidirectional sequence can be analyzed [5].
1. Add the sequencing reaction mix as in Table 9 (4.52 μl) to 2 μl
of each PCR sample and run on the thermocycler using the
program detailed in Table 10.
3.5 Sequencing The sequencing reaction clean-up is performed using Agencourt

Reaction Clean-Up Cleanseq, a SPRI® (solid-phase reversible immobilization) mag-
netic bead-based sequencing purification system which is amenable
to automation. The methodology is performed directly in the
thermal cycling plate and requires no centrifugation or filtration.
The system efficiently purifies sequencing products to deliver supe-
rior quality sequencing data (see manufacturer’s instructions for
further details).
Table 9
Sequencing reaction

Water 2.0
BigDye 5 sequencing buffer 1.10
M13F primer mix 1.2
BigDye Terminator v3.1 0.22
Table 10
Sequencing PCR cycling conditions

Step 1 96 C 10 s
Step 2 50 C 5s
Step 3 60 C 2 min
Step 4 (repeat steps 1–3 for 24 cycles)
1. Gently shake the Agencourt CleanSEQ solution so that it

appears homogenous and consistent in color.
2. Add 5 μl of Agencourt CleanSEQ to 7 μl of PCR product.
3. Add 17.14 μl of 85% ethanol to each sample and mix by
pipetting seven times until the solution appears homogenous
to ensure that the sequencing products bind to the magnetic
beads (see Note 4).
4. Place the sample plate on the magnetic stand for 2–3 min until
the solution is clear (see Note 5).
5. Aspirate the cleared supernatant from the plate and discard.
This step must be performed while the plate is situated on the
magnetic stand (see Note 6).
6. Dispense 30 μl of 85% ethanol into each well then pipette mix
seven times to wash the beads. Wait for at least 30 s to allow the
beads to resettle before continuing to the next step. This step
should be performed while the plate is situated on the magnetic
stand.
7. Completely remove the ethanol and discard. This step must be
performed while the plate is situated on the magnetic stand.
8. Repeat steps 6 and 7 for a total of two 85% ethanol washes.
9. Let the samples air dry for 10 min at room temperature (the
sample plate can be situated on or off the magnet while drying)
(see Note 7).
10. Add 15–30 μl of elution buffer and incubate the plate for 5 min
at room temperature to elute (see Note 8). Do not denature
samples prior to loading on capillary sequencers.
11. Load the plate on the detector and use the Agencourt
384 Direct Inject Magnet Plate (see Note 9).
12. Seal samples and store at 4 C for up to 24 h prior to loading
(see Note 10).
13. Sequencing is performed on a 3730XL DNA Analyzer
(Applied Biosystems).
3.6 Targeted Next- Targeted next-generation sequencing (also called massive parallel
Generation sequencing, MPS) is a powerful methodology for the analysis of
Sequencing Using multiple genes in a single assay. It enables the surveillance of a
the Agilent SureSelect subset of the genome which contains genes relevant to the pheno-
System type in question. There are many different platforms, chemistries,
and enrichment strategies available. We designed a custom Agilent
SureSelect exon-capture assay with baits for all the diagnostic
monogenic diabetes and hyperinsulinism genes to identify patho-
genic mutations located within 50 bp upstream and 10 bp down-
stream of each exon [4]. Sequencing is performed on an Illumina
HiSeq 2500. This methodology will identify the same types of
mutations detected by Sanger sequencing as well as large deletions
or duplications previously identified by a separate dosage assay.
The majority of the liquid handling in this protocol is amenable
to automation. We use the Bravo liquid handling platform which is
equipped with a ThermoCube thermoelectric and INHECO tem-
perature control system, used to incubate components at defined
temperatures during the procedure, but other systems are available.
Samples for the capture library are individually tagged with unique
barcodes and then combined into 12-plex pools prior to hybridiza-
tion. We process 48 samples at a time.
One library is prepared for all samples, which are individually
tagged with unique barcodes. The process is described below and is
modified from Agilent Technologies SureSelect Automated Library
Prep and Capture System protocol.
3.7 DNA DNA is quantified using the Qubit® dsDNA BR assay kit which is
Quantification designed for use with the Qubit® 2.0 fluorometer. The assay mea-
sures the concentration of double stranded DNA (dsDNA) in the
sample. Used according to the manufacturer’s specific protocols.
3.8 Sample Dilution 1. Label Anachem tubes (1 for each sample).

and Fragmentation 2. Dilute 1000 ng of DNA from each sample in a final volume of
Using the Bioruptor® 80 μl TE buffer.
Instrument
3.9 Fragmentation The Bioruptor® uses high-frequency sound waves at potentially

Using the Bioruptor® damaging noise level, ensure that safety ear protectors are worn at
Instrument all times.
1. Fill the Bioruptor® water bath with fresh deionized water and
allow it to cool to 4 C.
2. Vortex samples briefly and centrifuge briefly to collect the
entire sample at the bottom of the tube.
3. Place 12 samples at a time in the Bioruptor® adapter (see
Note 11).
4. Make sure all the tube hinges face outward and the tubes are
“flush” and the adaptor is tightened properly. Refer to the
Bioruptor® instrument user’s guide for more details.
5. Place the adaptor head in the Bioruptor® and fragment for
54 min.
6. Briefly centrifuge the samples.
7. Transfer each of the samples (80 μl) to a 96-well skirted Eppen-
dorf plate.
8. Repeat steps 2–8 for all of the samples.
9. Centrifuge the plate briefly to collect the sample at the bottom
of the well and seal the plate (see Note 12).
3.10 Purification This system can be used in manual and automated formats. To
of Sheared DNA Using summarize, samples are cleaned up with 120 μl AMPure beads
the Agencourt AMPure (1.5 volumes) and eluted in 40 μl dH2O.
XP System 1. Allow the AMPure XP beads to equilibrate to room tempera-
ture before use and shake well so that the reagent appears
homogenous and consistent in color.
2. Add 80 μl of the sheared DNA to each well and mix by
pipetting.
3. Add 120 μl of the AMPure XP beads to each well of a Nunc
DeepWell plate.
4. Incubate for 5 min at room temperature.
5. Place the plate onto the magnetic stand to separate the beads
from the solution for 3 min.
6. Remove the cleared liquid from the solution while the plate is
situated on the magnetic stand, being careful not to disturb the
separated beads.
7. Add 70% ethanol to each well, enough to cover the beads,
while the plate is situated on the magnetic stand and incubate
for 30 s at room temperature.
8. Repeat for a total of two washes.
9. Remove the plate from the magnetic stand and add 40 μl of
dH2O.
10. Resuspend the beads by pipetting up and down.
11. Incubate for 1 min.
12. Place the plate onto the magnetic stand and transfer the eluate
to a fresh plate.
3.11 Sample Quality A quality control check on D1000 tapes is performed to ensure that
Check Using Agilent each of the samples are fragmented sufficiently. Set up the Agilent
2200 TapeStation®™ 2200 TapeStation®™ according to the manufacturer’s instructions.
1. Take out the appropriate number of tapes and allow to equili-

brate to room temperature.
2. Pipette 3 μl of loading buffer per sample into 96-well plate or
strip tubes.
3. Transfer 1 μl of your sample into the loading buffer and mix
carefully by pipetting against the side of the well.
4. If using a plate, use a foil seal to cover the samples. If using strip
tubes, use strip caps.
5. Vortex the samples at 2000 rpm for 1 min.
6. Briefly spin down the plate/tubes.
7. Transfer your samples to the TapeStation®™ loading block and
start the run according to the Agilent 2200 TapeStation®™
user manual (remove strip caps if using strip tubes).
8. The average sample peak should be 180–250 bp (or the max
sample peak should be no higher than 250 bp).
9. If the samples are not fragmented enough, then perform an
additional sonication of 15–30 min depending on the fragmen-
tation results. Make sure you dilute the samples (which are
AMPure cleaned) with TE buffer to a final volume of 80 μl. If
the DNA quantity of a sample is below 350 ng, repeat the
sonication with more stock DNA and 54 min of sonication
(see Note 12).
3.12 Pre End-Repair 1. Dilute 350 ng of the fragmented DNA in dH2O to 8.75 ng/μl
Dilutions to a final volume of 40 μl in a 96-well skirted Eppendorf plate.
3.13 End Repair Forty microliters diluted fragmented DNA is converted by the
NEBNext End Repair Module to blunt ended DNA having 50
phosphates and 30 hydroxyl, to a final volume of 50 μl.
1. Make an end-repair master mix on ice (see Table 11) and pipette
into a plate. Add 40 μl of DNA from the previous step.
2. Incubate the samples for 30 min at 20 C (see Note 12).
Table 11
End-repair master mix
Volume (n ¼ 1) (μl)
10 End-repair buffer 5.0
End-repair enzyme mix 2.5
dH2O 2.5
Total volume per sample 10
3.14 AMPure XP Samples are cleaned up with 90 μl AMPure XP beads (1.8 volumes)
Clean-Up and eluted in 22 μl dH2O
1. Allow the AMPure XP beads to equilibrate to room tempera-
DeepWell plate.
3. Add the 50 μl sample to each well and mix by pipetting.
4. Incubate for 5 min at room temperature.
5. Place the plate onto the magnetic stand for 3 min to separate
the beads from the solution.
separated beads.
7. Add 70% ethanol (enough to cover the beads) to each well
while the plate is situated on the magnetic stand and incubate
for 30 s at room temperature.
dH2O. Resuspend the beads by pipetting.
11. Place the plate onto the magnetic stand and transfer the elute
to a fresh plate (see Note 12).
3.15 A-Tailing A dAMP is incorporated on the 30 end of the 21-μl blunt

(end-repaired) DNA.
1. Prepare the dA-tailing master mix (Table 12) on ice and mix by
vortexing.
2. Pipette 25 μl into each well of a Nunc DeepWell plate.
3. Add 21 μl of each DNA sample to each well and mix by pipetting.
4. Seal the plate with a film lid and incubate at 37 C for 30 min
(see Note 12).
Table 12
dA tailing master mix
10 Rx buffer 3
Klenow fragment (without exonuclease activity) 1.5
dH2O 4.5
End-repaired DNA 21
Total volume 25
3.16 AMPure Samples are cleaned with 54 μl AMPure beads (1.8 volumes) and
Clean-Up eluted in 12 μl dH2O
DeepWell plate.
3. Add 50 μl sample to each well and mix by pipetting and
incubate for 5 min at room temperature.
separated beads.
6. Add 70% ethanol (enough to cover the beads) to resuspend and
mix by pipetting. Incubate for 1 min at room temperature.
dH2O to elute the samples. Incubate for 2 min.
9. Place the plate onto the magnetic stand for 2 min and transfer
the eluate to a fresh plate (see Note 12).
3.17 Adaptor Ligation with 3.5 μl paired-end adapter to 12 μl A-tailed DNA.

Ligation
1. Make up the ligation master mix (n + 1, see Table 13) on ice and
pipette 4.5 μl on ice into each well.
2. Dilute each of the NEXTflex adapters to 6.25 nM.
3. On ice, add 3.5 μl of the diluted NEXTflex adapters per sample
(note each adapter is used only once for each sequencing run).
4. On ice, add 12 μl of repaired DNA.
5. Seal the plate with a film and incubate 20 C for 15 min.
Table 13
Ligation master mix
Fast-Link DNA ligase 1
10 mM ATP 1
10 buffer 2
Water 0.5
Total volume 4.5
3.18 AMPure Clean samples with 20 μl AMPure beads (1.0 volumes) and eluted
Clean-Up in 42 μl dH2O (see Note 13).
DeepWell plate.
3. Add the 20 μl sample to each well and mix by pipetting and
4. Place the plate onto the magnetic stand for 3 min to separate
the beads from the solution.
separated beads.
6. Add 70% ethanol (enough to cover the beads) and resuspend
and mix by pipetting to each well while the plate is situated on
the magnetic stand and incubate for 1 min at room temperature.
dH2O to elute the samples.
10. Place the plate onto the magnetic stand and allow to separate
for 2 min.
11. Transfer the eluate to a fresh plate (see Note 12).
3.19 Pre Hybe PCR Twenty microliter adapter-ligated DNA is amplified in a final PCR
volume of 50 μl.
1. Prepare the PCR master mix (Table 14) and add 30 μl of the
PCR master mix to each adapter-ligated DNA sample.
2. Run the program in Table 15 on the thermocycler (see Note 12).
Table 14
PCR reaction mix
5 Herculase II buffer 10
NEXTflex primer mix (12.5 μM each primer) 2.0
dNTP mix (Herculase kit only) 0.5
Herculase II polymerase 1.0
PCR-grade water 16.5
Total volume 30
Table 15
PCR cycling conditions

Step 1 98 C 2 min
Step 2 98 C 30 s
Step 3 65 C 30 s

Step 4 72 C 1 min
Step 5 (repeat steps 2–4 for seven cycles)
Step 6 72 C 10 min
Step 7 4 C Hold
3.20 AMPure Clean- Samples are cleaned up with 60 μl AMPure beads (1.2 volumes) and
Up eluted in 30 μl dH2O
DeepWell plate.
3. Add 50 μl sample to each well and mix by pipetting and
separated beads.
6. Add 70% ethanol (enough to cover the beads) and resuspend
and mix by pipetting to each well while the plate is situated on
the magnetic stand and incubate for 1 min at room
temperature.
dH2O to elute the samples.
10. Place the plate onto the magnetic stand and allow to separate
for 2 min.
11. Transfer the eluate to a fresh plate (see Note 12).
3.21 TapeStation®™ A quality control check on D1000 tapes are performed to ensure
Quality and Quantity adapters have been ligated and there is sufficient sample to continue
Check (see the Agilent 220 TapeStation®™ user manual for the manufac-
turer’s instructions).
1. To analyze results, set the region from 115 to 700 bp (this will
exclude any adapter dimer peaks [~135 bp] from the analysis).
The average sample peak should be less than 350–400 bp (due
to the adapter ligation, products will be longer by about
120 bp) (see Note 14).
3.22 Hybe Pools Four 12-plex pools each containing 1200 ng total DNA (i.e.,
100 ng per sample) are prepared for hybe reactions.
1. Label 4 Anachem tubes (e.g., A–D).
2. Add 12 samples per tube (100 ng per sample) to form a pool of
1200 ng.
3. Mix well on a vortex mixer and spin in a microcentrifuge for
1 min.
4. Using a vacuum concentrator dry the samples until no liquid is
left in the tubes (~60 min). A little white pellet may be visible
(see Note 12). We use a SpeedVac, although different instru-
ments are available.
5. If a sample is not completely dehydrated after 60 min, continue
dehydration for an additional 5–10 min (see Note 15).
6. Reconstitute the dehydrated samples in 8.2 μl dH2O to bring
the final concentration up to 147 ng/μl (see Note 16).
3.23 Hybridization For each pool, prepare one hybridization and capture 3.4 μl pooled
DNA (500 ng) hybridized with the Capture library in a final
volume of 28 μl. During the following steps, some of the transfers
are carried out while tubes remain on the PCR thermocycler.
During these steps care should be taken to remove and replace
tube lids without disturbing samples. After transfers, lids should
be replaced as quickly as possible to minimize loss by evaporation.
From this point onward, one sample refers to one pooled
library.
1. Label three rows of strip tubes as follows:
(a) Row 1: Pools A–D prepped library.
(b) Row 2: Pools A–D hybridization buffer.
(c) Row 3: Pools A–D SureSelect capture library.
2. Make up the SureSelect block mix (n + 1) as in Table 16 in strip
tubes for each pool (5.6 μl) and add the DNA.
3. Place the tubes on a PCR thermocycler and incubate at 95 C
for 5 min and then hold at 65 C (see Table 17).
4. In separate strip tubes, make up the hybridization buffer (one
for each pooled sample) (see Table 18), and once the PCR block
has reached 65 C from the previous step, place the tubes
containing the hybridization buffer on the PCR thermocycler
set at 65 C for 5 min.
Table 16
SureSelect block mix
Reagent Volume (n ¼ 1) (μl)

SureSelect indexing block #1 2.5
Total 5.6
Table 17
PCR thermocycler program

Step 1 95 C 5 min
Step 2 65 C Hold
Table 18
Hybridization buffer

SureSelect Hyb#1 25
SureSelect Hyb#2 1
SureSelect Hyb#3 10
SureSelect Hyb#4 13
Total 49
Table 19
RNase block dilution

SureSelect RNase block 0.5
dH2O 4.5
Total 5
5. Dilute the RNase blocker (Tables 19 and 20) and put on ice.
6. The custom library is stored at 80 C, remove from the
freezer, and defrost quickly and gently (i.e., gloved hand).
7. On ice, add 5 μl of the diluted RNase blocker to the labeled
tubes.
Table 20
Capture library (for capture size <3.0 Mb)
SureSelect library 2
RNase block dilution (see Table 19) 5
Total 7 μl
8. Add 2 μl of the capture library to the same tubes and put on the
thermocycler set at 65 C for 2 min.
9. Add 13 μl of hybridization buffer (on the 65 C block) into the
capture library RNase tubes.
10. Then add all of the SureSelect block mix and DNA to the
capture library tubes and pipette to mix.
11. Close the lids and incubate for 16 h at 65 C on a thermocycler.
3.24 Capture Clean-up ~28 μl hybridization library pools with streptavidin-

of the Hybridization coated beads.
Library Pools
1. Vigorously resuspend the Dynabeads MyOne Streptavidin T1
on to Streptavidin- on a vortex mixer.
Coated Beads
2. For each hybridization pool, add 50 μl of Dynabeads MyOne
Streptavidin T1 to a clean 1.5 ml microfuge tube.
Wash the beads:
1. Add 200 μl of SureSelect binding buffer to each of four tubes.
2. Mix the beads on a vortex mixer for 5 s.
3. Place the tubes into the magnetic stand for 2 min.
4. Remove and discard the supernatant.
5. Repeat steps 1 through 4 for a total of three washes.
6. Resuspend the beads in 200 μl of SureSelect binding buffer.
7. Transfer the bead mixture to a Nunc DeepWell plate (200 μl
per well, four wells in total).
8. Using a multichannel pipette, transfer each of the samples from
the hybridization step to the wells of the Nunc plate containing
the beads.
9. Mix thoroughly by pipetting and incubate in a vortex mixer for
30 min at room temperature.
3.25 Washing 1. Place the tubes into the magnetic stand for 2 min and remove
the Capture Library the supernatant.
2. Resuspend the beads in 500 μl of SureSelect wash buffer 2 pre-
warmed to 65 C.
3. Mix on a vortex mixer for 5 s to resuspend the beads.

4. Incubate the samples for 10 min at 65 C in a heat block.
Occasionally mix on a vortex mixer.
5. Briefly centrifuge the tubes so that the sample is at the bottom
of the well.
6. Separate the beads by placing the tubes on a magnetic stand
and remove all the supernatant.
7. Repeat steps 2–6 for a total of three washes.
8. Mix the beads in 30 μl of dH2O on a vortex mixer for 5 s to
resuspend the beads. Captured DNA is retained on the strep-
tavidin beads.
3.26 Fifteen microliter captured DNA amplified in a final PCR volume of

Posthybridization PCR 50 μl.
1. Make a PCR master mix for n + 1 samples according to
Table 21.
2. Pipette 35 μl of posthybridization PCR master mix into a strip
tube for each of the pools.
3. Add 15 μl of the hybridization DNA (including streptavidin
beads) and place on the PCR thermocycler and run the PCR
thermocycler program as in Table 22.
3.27 AMPure Samples are cleaned up with 90 μl AMPure beads and eluted in
Clean-Up 30 μl dH2O
2. Label four tubes.
3. Transfer the entire PCR product of the labeled tubes.
Table 21
Posthybridization PCR master mix
5 Herculase II buffer 10
NEXTflex primer mix (12.5 μM each primer) 2
dNTP mix (Herculase kit only) 0.5
Herculase II polymerase 1.0
PCR-grade water 21.5
Total volume 35
Table 22
PCR thermocycler conditions

Step 1 98 C 2 min
Step 2 98 C 30 s
Step 3 65 C 30 s

Step 4 72 C 1 min
Step 5 (repeat steps 2–4 for 16 cycles)
Step 6 72 C 10 min
Step 7 4 C Hold
4. Add 90 μl AMPure XP beads to each of the tubes containing

the PCR products and mix by pipetting.
5. Incubate at room temperature for 5 min on the vortex.
6. Transfer the tube to the magnetic stand and separate the beads
for 3 min.
7. Remove the supernatant without disturbing the beads.
8. Add 500 μl of 70% ethanol and incubate for 1 min.
9. Remove the ethanol.
10. Add another 500 μl of 70% ethanol and incubate for 1 min.
11. Remove the ethanol.
12. Incubate the samples at 45 C until all of the ethanol has been
removed, little cracks should appear in the bead pellets.
13. Resuspend the beads in 30 μl dH2O by carefully pipetting up
and down.
15. Transfer the tubes to the magnetic stand and allow the beads to
separate for 3 min.
16. Carefully transfer the supernatant to labeled strip tubes.
3.28 TapeStation®™ The concentration of the pooled samples is determined prior to

Quality Control sequencing. Analysis of the samples is performed on HSD1K tape
according to the manufacturer’s instructions.
1. Measure the concentration of the samples with the Qubit
dsDNA BR assay using 1 μl sample.
2. Using the concentration from the Qubit assay, dilute the sam-
ples to 1.5 ng/μl using 2 μl samples.
3. Run the diluted samples in triplicate on a HSD1K tape on the
Agilent 2200 TapeStation®™.
4. To analyze results, set the region from 150 to 700 bp.
3.29 Final Fifty microliter of a 10 nM pool are sequenced on the Illumina

Library Pool 2500HiSeq.
1. Make the final library pool of all the samples in an Anachem
tube to appropriate concentration (50 μl at 10 nM) (see Notes
12 and 17).
3.30 TapeStation®™ A final quality check is run on HSD1K tape to ensure the library
Quality Check pool has a concentration of 10 nM.
1. Dilute the sample 1:3 using 2 μl sample.
2. Run the diluted sample in triplicate on a HSD1K tape on the
TapeStation®™ according to the manufacturer’s instructions.
The samples are now ready to be sequenced. We use the Illu-
mina HiSeq2500 according to standard procedures (see Note 12).
4 Notes
1. All primers are M13 tailed and HPLC purification is

recommended.
2. Variations in these programs for particular amplicons are
detailed in Table 1.
3. Be sure to remove all of the ethanol from the bottom of the
well as it may contain residual contaminants.
4. Volume of ethanol ¼ 1.428 (5 μl + Sample Volume).
5. The magnetic beads will be pulled to the side of the well.
6. Remove as much of the supernatant possible as it contains
excess fluorescent dye and contaminants.
7. Excessive drying can lead to degradation of fluorescent dye.
8. Elution of the sequencing products is rapid. It is not necessary
for the beads to go back into solution for complete recovery.
9. If no Agencourt 384 Direct inject magnet plate is available,
allow the sample plate to separate on the magnet for 3–5 min
until solution is clear. Transfer clear sample into a new plate for
loading on the detector. Leave 2–5 μl of liquid behind to
prevent transfer of beads into the final plate. Residual beads
can interfere with injection, causing late starts or failed
injections.
10. If samples are not loaded within 24 h, store at 20 C.
11. If less than 12 samples, fill the empty spaces with balance
containing 80 μl of dH2O.
12. Note, samples can be stored at 20 C at this point.
13. Note, low AMPure is used to ensure removal of the unligated
adapters.
14. If the samples are too strong and the presence of artifacts
>900 bp long in the TapeStation®™ trace can be seen, repeat
the PCR with fewer cycles. If too weak, repeat with more
cycles.
15. This is not DNA, but precipitated salt from the input sample.
16. The samples should be dried for the minimum time possible,
excessive drying may result in poor resuspension of DNA.
17. The final volume of the pooled library can be adjusted to the
desired concentration. Exact library pool dilutions and proces-
sing can vary based on the cell capacity and analysis pipeline
being used. Refer to appropriate Illumina guide for
instructions.
References
1. Pearson ER, Flechtner I, Njølstad PR, Malecki 3. Flanagan SE, Kapoor RR, Hussain K (2011)
MT, Flanagan SE, Larkin B, Ashcroft FM, Genetic of congenital hyperinsulinemic hypogly-
Klimes I, Codner E, Iotova V, Slingerland AS, cemia. Semin Pediatr Surg 20(1):13–17
Shield J, Robert JJ, Holst JJ, Clark PM, Ellard S, 4. Ellard S, Lango Allen H, De Franco E, Flanagan
Søvik O, Polak M, Hattersley AT, Neonatal Dia- SE, Hysenaj G, Colclough K, Houghton JA,
betes International Collaborative Group (2006) Shepherd M, Hattersley AT, Weedon MN, Cas-
Switching from insulin to oral sulfonylureas in well R (2013) Improved genetic testing for
patients with diabetes due to Kir6.2 mutations. monogenic diabetes using targeted next-
N Engl J Med 355(5):467–477 generation sequencing. Diabetologia 56
2. Shields BM, Hicks S, Shepherd MH, (9):1958–1963
Colclough K, Hattersley AT, Ellard S (2010) 5. Ellard S, Shields B, Tysoe C, Treacy R, Yau S,
Maturity-onset diabetes of the young Mattocks C, Wallace A (2009) Semi-automated
(MODY): how many cases are we missing? Dia- unidirectional sequence analysis for mutation
betologia 53(12):2504–2508 screening in a clinical diagnostic setting. Genet
Test 13(3):381–386
Chapter 9
Isolation and Purification of Rodent Pancreatic Islets

of Langerhans
Jacqueline F. O’Dowd and Claire J. Stocker
Abstract
This chapter describes the detailed protocol for the isolation and purification of islets of Langerhans from
rodent pancreas using collagenase digestion. The first step of the process is to separate and isolate the
insulin-producing islets of Langerhans from the rest of the pancreas. The pancreas is excised from the
animal, trimmed of nonpancreatic tissues before being inflated and chopped into small pieces. The
connective tissue is then broken down with a collagenase enzyme solution to selectively digest the bulk
of the exocrine tissue while leaving the endocrine islets intact and separated from their surrounding
non-islet tissue. Once this process is completed, the islets of Langerhans are separated from the remaining
mixture by centrifugation and purified by the means of hand picking. Once isolated, the subsequent islets
can be used for several varied experimental processes, including transplantation, the study of pathophysio-
logical mechanisms in diabetic conditions, and in the screening of novel therapeutic approaches in phar-
macological research.
Key words Islets of Langerhans, Insulin, Isolation, Endocrine pancreas, β-cell, Collagenase
1 Introduction
The pancreas is a highly vascular retroperitoneal gland located in

the abdomen behind the stomach and on the posterior abdominal
wall surrounded by the liver and intestine. It is composed of both
exocrine and endocrine tissue. Embedded throughout the exocrine
glandular tissue, clusters of secretory endocrine cells, called the
islets of Langerhans, secrete hormones directly into the blood-
stream. Discovered in 1869 by the German pathological anatomist
Paul Langerhans, the islets of Langerhans constitute only 1–3% of
the total pancreatic volume [1] but fulfill a vital role in glucose
homeostasis. The number of islets within a human pancreas can
range from 200,000 to almost two million. Each islet itself can
range in size from a cluster of a few cells less than 40 μm in diameter
to ovoids of 400 μm in diameter [2]. Within the pancreas, islets are
not randomly distributed: small islets (160 nm or less, in diameter)
179
180 Jacqueline F. O’Dowd and Claire J. Stocker
tend to be scattered throughout the exocrine tissue while larger

islets, 250 nm or more in diameter, appear to be located near larger
ducts and blood vessels [3].
The ability to isolate islets from the pancreas enables investiga-
tors to use them in a number of downstream applications [4]. Once
isolated, islets of Langerhans can be maintained as viable units for
extended periods of time in tissue culture or they can be used in
more acute experiments investigating aspects of mechanistic func-
tionality. Isolated islets have long been used for the static and
perfusion incubations (to assess hormone release in response to
compounds, but recent advances such as RNA interference
(RNAi), a powerful and convenient tool for studying gene func-
tion, mean that the ability to isolate islets is a useful tool that allows
investigators to gather functional information without using cell
lines.
Islets are isolated by a modification of the method originally
described by Lacy in 1967 using enzymatic digestion with a rela-
tively crude preparation of collagenase [5]. The less fibrous nature
of the rodent pancreas, as compared to man, makes the release of
islets easier and allows for the preferential use of less purified
collagenase preparations in many laboratories. The first step of the
process is to remove the pancreas from the animal and trim it of
non-pancreatic tissues. The pancreas is then inflated to increase the
surface area and the connective tissue digested with collagenase.
After digestion is complete, the mixture is centrifuged to separate
the islets from non-islet tissue. The pellet is then resuspended in
physiological saline. A small sample of the resuspended pellet is
then placed on a black petri dish containing more buffer. The
black background of the petri dish allows the white islets to be
visible under a dissection microscope with an external light source.
Islets are then purified by hand picking using a very fine glass
pipette. The picking procedure is repeated twice to ensure no
carryover of unwanted exocrine tissue debris and to ensure that
the islets are of the highest quality and purity. It is very important
that the islets being isolated are intact and relatively pure. Contam-
ination of islets with acinar cells when used for functional studies
may lead to a high level of proteases that will later influence islet
integrity and functionality during incubation. The highest level of
purity is required if protein or RNA is to be extracted from the
isolated islets as any contamination of the preparation with acinar
cells will cause spurious data to be obtained.
With the increasing focus on the need to isolate viable islets and
to overcome the berries of islet transplantation, a number of mod-
ifications to this low activity collagenase digestion technique have
been described including low temperature Percoll centrifugation
[6] and sedimentation of islets [7], as well as their embedding in
hydrogel post purification [8].
Isolation and Purification of Rodent Pancreatic Islets of Langerhans 181
In this chapter, a method for the isolation of islets of Langer-

hans from the pancreas by collagenase enzymatic digestion and
then harvesting by handpicking individual islets is described.
Although the process of handpicking islets is both time-consuming
and labor-intensive, this method gives islet preparations with the
highest purity.
2 Materials
2.1 Reagents 1. Collagenase (Type XI, Sigma) (see Note 1).

2. Buffered saline solution: either Gey and Gey or Krebs Ringer
buffer supplemented with 1 mM CaCl2, 4 mM glucose and
gassed with CO2:O2 (95:5).
(a) Gey and Gey Buffer—111 mM NaCl, 27 mM NaHCO3,
5 mM KCl, 1 mM CaCl2, MgCl2, 0.3 mM MgSO4,
1.18 mM Na2HPO4, 0.29 mM KH2PO4, and 4 mM
D-glucose gassed to pH 7.4 by CO2:O2 (95:5).
A ten times stock solution without KCl, CaCl2, MgCl2,

MgSO4, KH2PO4, and D-glucose can be made and kept for
1 month at room temperature.
3. RPMI-1640 culture medium.
2.2 Equipment 1. Sterilized Forceps.

2. Sterilized scissors: One for skin incision and another for
removal of pancreas from adjacent tissue and another for chop-
ping the pancreas into pieces.
3. 10-ml syringe.
4. 25-G needle.
5. 15-ml polypropylene tubes.
6. 25-ml glass measuring beaker.
7. 50-ml conical flask with lid.
8. Water bath with shaker.
9. Plastic Pasteur pipette.
10. 90-mm petri-dish painted ‘in house’ with two layers of black
enamel paint to ensure a darkened background to visualize and
pick islets against.
11. Dissection microscope with external light source.
2.3 Islets Isolation 1. For islets isolated from normal rats (150–180 g) use two ani-
mals per isolation or three normal mice (see Note 2).
2. Just prior to dissection, euthanize the donor rodent by an
appropriate technique.
3. On a sterilized area, make a V-incision starting at the genital

area, move the bowel to the left side of the open rodent. This
will expose the pancreas which is tan in color and can be easily
differentiated from the surrounding fat as can the spleen.
4. Using forceps to grasp the spleen, gently push up and back the
stomach.
5. Cut the pancreas away from the surrounding fat and where
attached at the small intestine.
6. Remove the pancreas from the body and place the pancreas,
still attached to the spleen, into 100 ml of Gey and Gey in a
Petri dish and cut away from the spleen—any retained fat,
lymph nodes, and other non-pancreatic tissues.
7. Inflate the pancreas by inserting a 25-gauge needle attached to
a 10-ml syringe into the pancreas and injecting small volumes
(1–2 ml at a time) of Gey and Gey solution into all the folds of
the tissues so that the pancreas increases in size, thus creating a
larger surface area.
8. Place the pancreata into a glass 50-ml beaker and chop using
scissors until all the pieces were approximately the same size
(1 mm 1 mm).
9. Transfer the contents to a 15-ml tube and centrifuge for 5 min
at 100 g. After centrifugation, remove the supernatant,
which contains fat, and transfer the pellet to a 15-ml
conical tube.
10. Digest the pancreas with collagenase type XI (1–2 mg/pan-
creas) in a 1:1 mixture of Gey and Gey solution in order to
release islets from exocrine tissue.
11. Place the conical tube in a shaking water bath at 37 C and
shake at 800 oscillations per minute until the solution appears
“milky” and most of the pieces of pancreas are digested
(approximately 5–10 min—time varies with strain and age of
the animal). During this time, take a small amount (approxi-
mately 400 μl) in a plastic pipette and examine under a dissect-
ing microscope to confirm whether islets are free from the
exocrine tissue. If there are no islets visible continue the shak-
ing by hand and repeat the process (see Note 3). Normally, a
total shaking time of 6–8 min is enough to release the bulk of
free islets.
12. Place the contents of the conical flask and stop the digestion by
adding 15 ml of cold buffer.
13. Centrifuge at 500 g for 5 min to remove the collagenase in
the solution.
14. Aspirate the supernatant and resuspend the pellet in 15 ml of
the isolation buffer.
Isolation and Purification of Rodent Pancreatic Islets of Langerhans 183
15. Add a sample of the resuspended pellet (1 ml) to a blackened

90-mm Petri dish, add buffer to the top of the Petri dish, and
mix with a plastic Pasteur pipette.
16. Using a dissection microscope and an external light source,
handpick and count individual islets using a drawn out glass
pipette (see Note 4).
17. Only pick clean and intact islets free of exocrine tissues (see
Note 5).
18. Repeat the picking procedure at least twice.
19. Incubate the islets with isolation buffer in a 37 C water bath
until use. Always use the islets within 30 min to 1 h after
isolation.
20. While the yield of islets is variable, approximately 600–1200
islets are obtainable from two Wistar rats and 400 from three
mice using this method (see Note 6).
2.4 Islets Culturing For certain experimental protocols, it may be necessary to culture
the isolated islets. When culturing, it is important to handle the
islets under aseptic conditions and use sterile tubes and flasks.
1. Wash the islets in more than ten volumes of sterile Gey and Gey
solution.
2. Resuspend groups of 200 islets in RPMI-1640 containing
11.1 mM glucose (see Note 7) and transfer to a six-well cell-
culture plate.
3. Incubate the islets in a humidified culture incubator with 5%
CO2; 95% air.
3 Notes
1. A wide variety of collagenase preparations are available for islet

isolation but in our experience Collagenase Type XI from
Sigma-Aldrich is best suited for isolations. This crude mixture
isolated from Clostridium histolyticum contains several
enzymes (collagenase, neutral proteases, clostripain, and case-
inase) that act together to break down tissue. While collagenase
Type XI has one of the highest collagenase activities, not every
batch is equivalent, and it is usually appropriate to test small
quantities from several different batches before choosing to
make a bulk purchase of a particular batch.
2. The age and weight of the animals can influence the numbers of
islets of Langerhans obtained. Rats heavier than 180 g often
yield fewer islets per pancreas. Although younger animals can
give high yields, the islets obtained from these animals should
be considered from a ‘juvenile’ stage of development.
3. If no islets are isolated or isolated islets are not intact, the islets
are over-digested. This can be prevented by decreasing the
collagenase digestion time or collagenase concentration (see
Subheading 2.3 step 10).
4. Normal bore pipettes draw up too much liquid as islets are
picked, so drawn pipettes are preferential. Draw glass pipettes
by heating in a Bunsen flame. Once the glass begins to give,
pull sharply to yield pipettes with a bore of approximately
1 mm.
5. If most of the islets are not discrete and difficult to isolate (see
Subheading 2.3 step 15), the exocrine tissues are under-
digested. In order to overcome this problem, increase the
collagenase digestion time (see Subheading 2.3 step 10)
and/or increase the collagenase concentration (see Subheading
2.3 step 10) and ensure vigorously shaking of the vial to
disrupt the pancreas (see Subheading 2.3 step 11).
6. The yield of islets will vary depending on the age and strain.
Diabetic and older animals will have reduced numbers of viable
islets. Digestion continues as the handpicking selection pro-
ceeds; therefore, speed is important. Islets should be collected
within 15–20 min, and the entire isolation completed within
45 min to 1 h of removal of the organ.
7. The culture medium can be altered depending on the experi-
mental conditions required.
References
1. Hellerstrom C (1984) The life story of the pan- 6. Olack B, Finke E, Scharp DW (1987)
creatic β-cell. Diabetologia 26(6):393–400 Low-temperature culture of human islets by
2. Lifson N, Kramlinger K, Mayrand R, Lender E the distension method and purified with Ficoll
(1980) Blood flow to the rabbit pancreas with or Percoll gradients. Surgery 102(5):869–879
special reference to the islets of Langerhans. 7. Saliba Y, Fares N (2019) Isolation, purification
Gastroenterology 79(3):466–473 and culture of mouse pancreatic islets of Langer-
3. Bonner-Weir S, Orci L (1982) New perspectives hans. Methods Mol Biol 1940:255–265
on the microvasculature of the islet of Langer- 8. Ayenehdeh JM, Niknam B, Hashemi SM,
hans in the rat. Diabetes 31(10):883–889 Rahavi H, Rezaei N, Soleimani M, Tajik N
4. Nolan AL, O’Dowd JF (2009) The measure- (2017) Introducing a new experimental islet
ment of insulin secretion from isolated rodent transplantation model using a biometric hydro-
islets of Langerhans. Methods Mol Biol gel and a simple high yield islet isolation tech-
560:43–51 nique. Iran Biomed J 21(4):218–227
5. Lacy P, Kostianovsky M (1967) Method for the
isolation of intact islets of Langerhans from the
rat pancreas. Diabetes 16(1):35–39
Chapter 10
Characterization of Islet Leukocyte Populations in Human

and Murine Islets by Flow Cytometry
Matthew J. Butcher, Michelle B. Trevino, Yumi Imai, and Elena V. Galkina
Abstract
An increasing body of evidence indicates that a local islet immune response is not only limited to type
1 diabetes, but also is associated with islet dysfunction in type 2 diabetes. Recently, the presence of
pancreatic CD68+ macrophages within islet tissues was demonstrated by RT-PCR and immunohistochemi-
cal methods. However, the precise profile and activation status of intraislet leukocytes, which are present in
both murine and human islets, are poorly defined. Here, we describe a detailed flow cytometry protocol
designed to analyze both human and murine islets for intraislet leukocytes and leukocyte subsets. This
approach permits the simultaneous identification of multiple intraislet leukocyte subsets, as well as their
activation statuses. The use of flow cytometry-based approaches will advance the field of islet biology and
help to identify unique changes in the immune cell composition that accompanies pathological islet
inflammation and dysfunction in type 2 diabetes.
Key words Type 2 diabetes, Leukocytes, Human islets, Murine islets, Flow cytometry
1 Introduction
Type 2 diabetes results from the combination of insulin resistance

and islet dysfunction [1]. It has been well-known that chronic
inflammation within peripheral insulin-responsive tissues contri-
butes to insulin resistance [1]. While the exact mechanisms that
drive islet dysfunction in type 2 diabetes are incompletely under-
stood, recent evidence suggests that the immune system may be
involved [2]. One of the first changes that occurs in response to
metabolic stress from excessive nutrition occurs in the peripheral
circulation, with a contaminant increase of T helper 1 (Th1) and
Th17 cells and a decrease in T regulatory cells (Tregs) [3, 4]. Inter-
estingly, peripheral blood circulating B cells from type 2 diabetic
patients also display a proinflammatory phenotype, and notably
secrete IL-8 and support contact-dependent T-cell activation
[5]. To date, it is unclear whether these systemic alterations also
occur within pancreatic islets. Recent studies have examined the
185
186 Matthew J. Butcher et al.
immune cell content of human type 2 diabetic islets and demon-

strated an elevated number of CD68+ macrophages in histological
samples from type 2 diabetic patients [6, 7]. As type 2 diabetic
conditions may affect not only macrophages, but also lymphocytes
and likely other types of leukocytes, additional studies are necessary
to fully characterize the immune cell composition within islets and
to determine their potential impact on intraislet inflammation in
type 2 diabetes.
While extensive work on animal models of type 2 and type
1 diabetes have provided valuable insights into islet biology,
human islets are known to differ from rodent islets in terms of
their morphology [8, 9] and functionality [10], highlighting the
importance of studying human islets. Nevertheless, there are some
limitations in investigating the human intraislet immune composi-
tion and insulin-producing beta cells. The scarcity and difficulty in
obtaining human islets are significant barriers for performing large-
scale studies using human islets. Additionally, well-established
methods of RT-PCR and immunohistochemistry may provide lim-
ited information about the cell type, stage of activation, and relative
proportions of leukocyte subsets within whole tissues. Flow
cytometry-based analyses of secondary lymphoid organs and several
other nonlymphoid tissues have provided “high-throughput” data
about the overall numbers, subsets, and phenotypes of cells isolated
from dissociated tissues through the use of multiple surface antigen
markers [11]. Several reports have taken advantage of flow
cytometry-based approaches and have characterized endocrine
cells in both human and murine islets [12, 13]. There has been
targeted characterization of myeloid cell populations in murine
islets [14–16]. However, this approach was not used to analyze
leukocyte populations in human islets or study other leukocytes in
murine islets. Thus, we adapted a flow cytometry protocol to better
characterize leukocyte subsets and their activation statuses within
human islets [17]. Here, we describe in detail a method for prepar-
ing both human and murine islets for flow cytometry. In addition,
we provide specific advice on a gating strategy for analyzing all
intraislet leukocytes, and we discuss potential issues and work-
around solutions for preparing and analyzing islets by flow
cytometry.
2 Materials
2.1 Human Islets 1. Supplemented CMRL-1066 media: CMRL-1066, 10% fetal

bovine serum, 1% penicillin–streptomycin.
2. A 100 20-mm polystyrene nontreated tissue culture dish.
Flow Cytometric Analysis of Islet Leukocytes 187
2.2 Murine Islets: 1. Dissection instruments: Hooked nose forceps, scissors, hemo-
Harvest stat clamps, 30 gauge syringe.
and Preparation 2. Surgical microscope.
for Flow Cytometry
3. 1 Dulbecco’s phosphate-buffered saline (PBS) without cal-
cium or magnesium.
4. 12 75 mm, 5 ml polystyrene round bottom test tubes (FACS
tubes).
5. Collagenase P (Roche).
6. Ficoll PM400.
7. 1 Hank’s balanced salt solution (HBSS).
2.3 Preparation 1. Islet digestion buffer: Trypsin/ethylenediaminetetraacetic acid

of Islet and Splenic (EDTA) (0.25% trypsin and 0.2 g/l EDTA) in phosphate-
Cell Suspensions buffered saline.
2. 1 Dulbecco’s phosphate-buffered saline (PBS) without cal-
cium or magnesium.
3. 40 and 70 μm nylon cell sieves prewetted with cold PBS.
4. Temperature controlled, table-top centrifuge 5810R.
5. Red blood cell lysis buffer: 8.3 g/l ammonium chloride
(NH4Cl), 1 g/l potassium bicarbonate (KHCO3), 0.37 g/l
EDTA, pH 7.4.
6. Low-retention pipette tips.
2.4 Cell Counting 1. 0.4% Trypan blue solution.

2. Bright-line hemocytometer.
3. Light microscope.
2.5 Extracellular 1. Human or mouse Fc blocking antibodies (anti-CD16/CD32).

Staining 2. Fc block PBS: 0.25–1.0 μg/ml human or mouse Fc blocking
antibodies in PBS.
3. Appropriate extracellular flow cytometry antibodies.
4. Fixable viability dye (optional).
5. Paraformaldehyde (PFA).
6. HPa2, HPi2 antibodies (for endocrine cell staining).
2.6 Intracellular 1. Commercial fixation and permeabilization reagents.

Staining 2. Appropriate intracellular flow cytometry antibodies.
2.7 Sample 1. Flow cytometer (e.g., a DXP 8-color upgraded BD

Acquisition FACSCalibur).
and Analysis 2. Flow cytometry tubes with cell strainer caps.
3. Flow cytometry data analysis software (such as FlowJo, Kaluza,
etc.).
3 Method
3.1 Human Islets: 1. Upon receiving the islets, culture the islets in a 100 20-mm
Care Post- polystyrene nontreated tissue culture Petri dish in supplemen-
procurement ted CMRL-1066 media overnight (37 C, 5% CO2) to allow
and Preparation the islets to recover from the shipment. 5000–7000 Islet
for Flow Cytometry equivalents (IEQ) of human islets are sufficient for a flow
(See Note 1) cytometry experiment (see Note 2).
2. Prepare a 40-μm nylon mesh cell sieve by wetting it with 4 C
PBS in a 35 10 mm polystyrene nontreated tissue culture
Petri dish, making sure that the entire mesh has been soaked.
Using a pipette, gently collect the islets from culture and
transfer them to the center of a 40-μm cell strainer placed in a
clean 100 20-mm polystyrene nontreated tissue culture Petri
dish. This will remove cellular debris and disintegrated islets,
and only the intact islets will be retained on the cell sieve.
3. Wash the islets retained on the sieve by very gently pipetting
cold PBS over the strainer into the dish. Transfer the strainer
containing intact islets to a new 35 10-mm polystyrene
nontreated tissue culture Petri dish containing 1–2 ml of cold
PBS. Very gently pipette up and down to suspend the islets off
of the sieve and into a 15 ml tube. Repeat by adding 1–2 ml of
cold PBS to the strainer to pick up residual islets. Centrifuge
(800 g, room temperature, 1 min) and aspirate the excess
PBS with a transfer pipette, leaving the islet pellet intact.
4. Resuspend the islets in 0.5 ml PBS by gently pipetting up and
down using low-retention pipette tips to minimize islet loss (see
Note 3).
5. Continue the protocol with Subheading 3.2.
3.2 Murine Islets: 1. After euthanizing the mouse, open the abdominal cavity, locate
Harvest and clamp the common bile duct as it enters the duodenum.
and Preparation Inject 1.4 mg/ml collagenase P solution into the common bile
for Flow Cytometry duct with a 30-gauge needle until the pancreas appears fully
inflated. Transfer to a 15 ml tube containing 1 ml of cold
HBSS. Keep on ice before proceeding to the next step as
collagenase injections are repeated when multiple pancreata
are isolated. Typically, 500–800 islets are used for a single
analysis by flow cytometry. This requires combining pancreata
from three to five mice.
2. Collect the spleen in order to prepare single-color and positive
flow cytometry controls. Place the spleen in a FACS tube with
1–2 ml of PBS and place the tube on ice.
3. Incubate the 15 ml tube containing the pancreas in a 37 C
water bath and mechanically disrupt the pancreatic tissue (see
Notes 4 and 5).
4. Transfer the digested pancreas to a 50-ml tube containing

30 ml HBSS, centrifuge (168 g, 4 C, 1 min) and wash
once more with 30 ml of HBSS.
5. After discarding the supernatant, add 5 ml of 27% Ficoll solu-
tion in HBSS to the tissue pellet and vortex to create a homog-
enous tissue suspension. Carefully layer 3 ml each of 23%, 20%,
and 11% Ficoll solution to generate Ficoll density gradient and
centrifuge (887 g, 4 C, 15 min).
6. Transfer the Ficoll supernatant to an opaque Petri dish and
carefully handpick the islets using a P10 pipette tip to a 1.5-ml
tube containing 1 ml of PBS.
7. Wash islets twice in 1 ml of PBS to remove all the Ficoll by
centrifugation at 887 g at 4 C for 3 min. Suspend the final
islet suspension in 100 μl of PBS.
8. Continue the protocol with Subheading 3.3.
3.3 Preparation 1. Resuspend the islet pellet gently to a final concentration of

of Islet and Splenic trypsin (0.05%)/EDTA (0.04 g/l) PBS. Digest the islets briefly
Cell Suspensions at 37 C for 90 s for human islets and 60 s for mouse islets.
Pipette up and down to mechanically loosen the islet structure
until it appears that 70–80% of islets are dispersed. When the
islets are dispersed, add 1 ml of PBS to dilute the trypsin (see
Note 5).
2. Pass the partially intact islets through a 40-μm nylon mesh cell
sieve into a new 35 10-mm polystyrene nontreated tissue
culture dish, and using the rubber end of a 1-ml syringe
plunger, gently rub any partially intact islets into the nylon
mesh until all islet clumps disappear to create an islet single-
cell suspension. Transfer the dissociated islet cell suspension to
a new 1.5-ml tube and wash the suspension with PBS once,
aspirate the supernatant, and resuspend the cells in 1 ml of PBS.
Place the islet cell suspension on ice.
Note: If mouse spleens or lymph nodes were collected for
single-color and positive flow cytometry controls, resume the
protocol at Subheading 3.3, step 3. Otherwise, proceed to
Subheading 3.4.
3. To create a splenic or lymph node cell suspension, use a 70-μm
cell sieve, a small Petri dish, and a syringe plunger to gently
disrupt the organ and release leukocytes. Collect the cell sus-
pension from the Petri dish and pellet the cells by centrifuga-
tion (400 g, 4 C, 5 min). Discard the supernatant with a
pipette.
4. If a spleen was used to create the cell suspension, add 2 ml of
1 RBC lysis buffer to the suspension and vortex briefly to
resuspend the pellet. Incubate the suspension for 5 min at
room temperature. Proceed to Subheading 3.3, step 5.
If a lymph node was used to create the cell suspension,

gently pellet the cells via centrifugation (400 g, 4 C, 5 min)
and resuspend the cells in 1 ml of PBS. Place the suspension on
ice and resume the protocol at Subheading 3.4.
5. Wash the suspension twice with PBS and resuspend the pellet in
1 ml of PBS. Place the splenic cell suspension on ice and resume
the protocol at Subheading 3.4.
3.4 Cell Counts 1. Pipette or gently vortex the cell suspensions to ensure that they
are homogenous.
2. Collect a 10- to 20-μl aliquot of each sample and determine the
approximate number of viable leukocytes using trypan blue and
a hemocytometer or an automated cell counter that is capable
of distinguishing between leukocytes and stromal cells.
3. Calculate the approximate number of leukocytes present in the
suspension.
4. Centrifuge cells (400 g, 4 C, 5 min) and decant the super-
natant. Resuspend the pellets in an appropriate volume of PBS
and transfer approximately 1 106 cells to labeled FACS tubes.
For a successful flow cytometry experiment, single-color
controls, positive and negative splenic/lymph node controls,
nonstained islet, isotype islet controls, and experimental islet
tubes are necessary (see Notes 6–8 and Fig. 2).
5. Add 1 ml of PBS to each tube and pellet the cells (400 g,
4 C, 5 min, and aspirate the supernatant). Proceed to Sub-
heading 3.5.
3.5 Extracellular 1. Resuspend the pelleted cells (Subheading 3.4, step 5) in 50 μl

Flow Cytometry of Fc block-PBS (0.25–1.0 μg/ml human or mouse Fc block-
Staining ing antibodies in 1 PBS).
2. Incubate for 10–15 min at 4 C to block extracellular Fc
receptors and nonspecific antibody recognition. Prepare the
extracellular antibody cocktails (50 μl/tube) using appropriate
flow cytometry antibodies and Fc block-PBS (see Note 6).
3. Add 50 μl of the extracellular antibody cocktail to the
corresponding FACS tube(s). Gently vortex to mix and incu-
bate the samples for 20–30 min in the dark at 4 C to stain for
extracellular antigens.
4. Add 1 ml PBS to each tube to wash the cells.
5. Gently pellet the cells and decant the supernatant (400 g,
4 C, 5 min).
7. To additionally stain for intracellular antigens, proceed to Sub-
heading 3.5. Otherwise proceed to Subheading 3.5, step 8.
8. Gently resuspend the pelleted cells in 100 μl of 2% paraformal-

dehyde in PBS. The tubes may be stored at 4 C in the dark or
immediately analyzed on the flow cytometer.
3.6 Intracellular 1. Very gently resuspend the pelleted cells (Subheading 3.5, step
Staining 6) in 100–200 μl of a suitable fixation/permeabilization
reagent (i.e., BD Cytofix/Cytoperm), following the manufac-
turer’s instructions for preparation and use (see Note 9).
The rest of the protocol assumes that BD Cytofix/Cyto-
perm reagents were used to fix and permeabilize the cells. If
other intracellular staining reagents were used, the intracellular
staining protocol should be modified following the manufac-
turer’s instructions.
2. Incubate the samples for 20–30 min at 4 C in the dark to
permeabilize the cells. Intracellular antibody cocktails (100 μl/
tube) should be prepared using BD permeabilization/wash
solution. The appropriate amount of intracellular antibody to
use for one test should be empirically determined (see Note 6).
3. Add 1 ml of BD permeabilization/wash solution to each tube
and pellet the cells by centrifugation (400 g, 4 C, 5 min).
Discard the supernatant.
4. Gently resuspend the samples in 100 μl of the correct intracel-
lular antibody cocktail. Incubate the samples for 20–30 min at
4 C in the dark to stain for intracellular antigens.
5. Add 1 ml of 1 permeabilization/wash solution to each tube
to wash the cells.
6. Gently pellet the cells and decant the supernatant (400 g,
4 C, 5 min).
8. Gently resuspend the pelleted cells in 100 μl of 2% paraformal-
dehyde in PBS. The tubes may be stored at 4 C in the dark or
immediately analyzed on the flow cytometer.
3.7 Sample 1. Start and set up the flow cytometer and instrument settings
Acquisition according to manufacturer’s instructions and in-house proto-
on the Flow Cytometer cols. The islet samples should be filtered using a 70-μm cell
strainer cap just before the sample acquisition.
2. The splenic unstained and single-color control samples should
be used to set and adjust the gain values on the forward (FSC)
scatter, side (SSC) scatter, and fluorescent channels. Ideal
single-color controls should have a clearly defined positive
and negative population. Once the acquisition settings have
been finalized, the positive control, negative control, and
experimental samples can be acquired.
3. In order to analyze the maximum number of cells in the sample

tube, when the sample volume reaches the lower limit of the
sample injection port (SIP), the tube can be washed with a
small volume of sheath fluid to acquire the rest of the specimen.
3.8 Analysis 1. Using an appropriate flow cytometry data analysis program

(such as FlowJo), setup and define a compensation matrix
using the single-color controls. Once the compensation matrix
is complete, apply the matrix to the experimental samples.
2. For islet samples, CD45 should be used to distinguish between
endocrine cells and infiltrating leukocytes. In addition, cellular
debris and events that are smaller than ~30K on the FSC axis or
larger than 250K on the SSC axis may be excluded from the
analysis (Fig. 1). Finally, doublets and cellular aggregates
should be excluded from the analysis as well, based on FSC/
FSC-A (Fig. 1) or FSC/FSC-W parameters. As many as possi-
ble cells should be collected. Subsets of single CD45+ FSC+
gated leukocytes can be distinguished using additional markers
(Fig. 2).
3. Critically, as islet-infiltrating leukocytes are relatively low in
abundance, and islet endocrine cells are highly autofluorescent
in comparison to islet leukocytes, nonstained islets, isotype
control islets, and fluorescence minus one control islets are
necessary to determine the correct placement of the gates
(Figs. 1 and 2).
4. If a cell viability dye was used, the experimental islet sample
should be compared with a viability dye positive control islet
sample to correctly distinguish between live and dead cells.
4 Notes
1. We have primarily used human islets obtained through the

Integrated Islet Distribution Program (IIDP, http://iidp.coh.
org/), which provides human pancreatic islets for research to
preapproved investigators. Further information regarding the
standard procedures and policies is available online at http://
iidp.coh.org/. In our experience, human islets within 3 days of
the isolation produce a better quality single-cell suspension in
comparison to the suspensions of islets that were isolated 3 or
more days prior. Typically, human islets were cultured at the
isolation centers for 1–2 days and then shipped overnight in
CMRL-1066 medium supplemented with 10% FBS and 1%
penicillin–streptomycin at 10–20 C. Of note, there have
been changes in the shipping protocol utilized by the IIDP
recently and we have not performed flow cytometry using islets
shipped under the new protocol. Human islets with a purity
Fig. 1 Recommended gating scheme for human and murine islet-infiltrating leukocytes. As multiple cell types
are present within pancreatic islets, including alpha, beta, delta, PP cells, and leukocytes, the initial forward
scatter (FSC) and side scatter (SSC) settings should be set based on a tissue or specimen with clear leukocyte
populations, such as peripheral blood leukocytes, splenocytes, or lymphocytes (not shown). As shown in the
figure, the initial ungated islet cell forward and side scatter profile will have various cell populations that are
dispersed throughout the graph (far left column, a). To specifically examine leukocytes within the islet
suspension, CD45, the common leukocyte marker, can be used (b)—second column first and third rows:
CD45 staining, second and fourth rows: nonstained control). As large granular endocrine cells are more
autofluorescent than leukocytes, islet CD45 staining should be compared with an islet isotype control for each
specimen to place the initial CD45 gate (second column, empty channel). After the CD45 gate has been set,
aggregates of cells should be excluded from the analysis using FSC/FSC-A (c) or FSC/FSC-W (not shown)
plots. Once single CD45+ leukocytes have been gated, a forward versus side scatter should be placed to
exclude debris that are under 40K on the FSC axis and any events that are larger than 250K on the side scatter
axis ((d)—fourth column). Single CD45+ cells ((e)—shown in the fifth column) can be used for subsequent
leukocyte subset, phenotype, and activation analyses. Note: a cell viability dye can also be used to specifically
gate on viable islet-infiltrating leukocytes (not shown). To correctly distinguish between dead cells and viable
cells, an appropriate islet-positive control should be used. HFD High-fat diet, CD Show diet, T2DM
Type 2 diabetes mellitus, Non-DM Nondiabetic, FSC Forward scatter, SSC Side scatter, and FSC-A Forward
scatter-area
and viability above 80% were suitable for flow cytometry

experiments [17].
2. We typically use 5000–7000 islet equivalents (IEQ) of human
islets culture in 5–10 ml of medium for a single flow cytometry
assay. Human islets of variable diameters are converted to IEQ
based on their diameter following a conversion table [18]. One
Fig. 2 Representative CD3+ T-cell and CD20+ B-cell staining within nondiabetic and type 2 diabetic human
and murine islets. To examine intraislet CD3+ T-cell and CD20+ B-cell content, single CD45+ islet leukocytes
were gated as described in Fig. 1 and analyzed for CD3 (top) and CD20 (bottom) staining. Nondiabetic and type
2 diabetic human islets (first and third rows) and their corresponding isotype controls are shown. In addition,
chow diet fed and high-fat western diet fed C57Bl6 islets (second and fourth rows) and their corresponding
isotype controls are also shown. Islet isotype controls are useful and recommended for initially placing the
gates as nonspecific staining or high autofluorescence can be a confounding factor when analyzing non-
lymphoid cell suspensions by flow cytometry
IEQ corresponds to an islet volume equivalent to that of diam-

eter at 150 μm [18].
3. Upon and after trypsinization of the human and mouse islets,
we found it ideal to use low-retention pipette tips such as
TipOne Repel-Polymer Technology (USA Scientific, Ocala,
FL) to reduce islet and cell loss due to islets sticking to the
sides of the tips.
4. The optimum incubation time for the collagenase digestion
period may differ depending on the enzymatic activity of the
collagenase lot preparation and will need to be determined
empirically.
5. It is important to avoid overdigesting the pancreatic tissue at
this stage. Ideally, the pancreatic tissue should be mechanically
disrupted by gently inverting the tube three or four times, or
until the entire solution looks homogenous. Overdigestion
(both enzymatic and/or mechanical) will break the islets.
6. The recommended flow cytometry test volumes found in com-
mercial technical data sheets are typically standardized to 100 μl
test volumes/1 106 cells/tube. The optimal volume of anti-
body to use per test (100 μl test volumes/1 106 cells/tube)
should be empirically determined by the experimenter through
titration curve experiments. The number of cells and the vol-
ume of antibody used in the experiment can be scaled up or
down as desired. As islet leukocytes are relatively low in abun-
dance, our group recommends scaling up the number of islet
cells used in the experimental panel if enough islets are avail-
able. 1–2 106 cells are sufficient to study major leukocyte
populations of interest and additional IEQ are recommended
to study leukocyte subsets (i.e., T helper cell subsets, B-cell
subsets, macrophage subsets, etc.).
7. As islet-infiltrating leukocytes are relatively low in abundance,
and islet endocrine cells are comparatively autofluorescent,
running positive and negative controls (nonstained samples,
isotype, and fluorescence minus one controls) on splenocytes
or lymph nodes are recommended, in addition to negative islet
controls. A successful flow cytometry experiment should have
the following controls:
(a) Unstained splenic or lymph node cell suspensions.
(b) Unstained islet cell suspension.
(c) Single-color controls prepared using splenocytes or lymph
nodes. There should be one separate single-color control
tube for each fluorophore-conjugated antibody that will
be used in the experimental cocktail.
(d) Fluorescent minus one (FMO) controls prepared using
splenocytes or islet cell suspension if enough sample is
available.
(e) Separate isotype controls prepared using splenocytes and

islet cell suspensions.
Isotype controls should contain all of antibodies pres-
ent in the experimental cocktail aside from one marker,
which is replaced by an equivalent amount of a nonspecific
antibody conjugated with the correct fluorophore. For
example, if an experimental panel comprised CD45-
PerCP, CD20-Pacific Blue, and CD3-APC, an appropri-
ate isotype control panel for CD20 would include CD45-
PerCP, the recommended isotype control for the CD20
antibody (e.g., IgG1-Pacific Blue), and CD3-APC.
8. Human and mouse endocrine cells, which are larger and more
granular, are autofluorescent in comparison to islet leukocytes.
We highly recommend running unstained and isotype controls
with portions of the islet cell suspension.
9. Commercial intracellular staining kits typically contain a fixa-
tive as well as a mild detergent (e.g., saponin) in order to
permeabilize the cells and allow the antibodies to access intra-
cellular antigens. However, the cells become fragile when they
are permeabilized. The cells should therefore be carefully han-
dled and resuspended gently throughout the intracellular stain-
ing protocol.
Acknowledgments
We thank the Eastern Virginia Medical School Flow Cytometry

core facility for their excellent technical support. The human islets
were provided by the Integrated Islet Distribution Program
(IIDP). This work was supported by the National Institutes of
Health grants RO1-DK090490 (Y.I.), the IIDP pilot program
(Y.I.), a BD Biosciences research grant (E.G.), and Eastern Virginia
Medical School pilot grant (Y.I. and E.G.).
References
1. Osborn O, Olefsky JM (2012) The cellular and promotes inflammation in type 2 diabetes. J
signaling networks linking the immune system Immunol 186(2):1162–1172
and metabolism in disease. Nat Med 18 4. Zeng C, Shi X, Zhang B, Liu H, Zhang L,
(3):363–374 Ding W, Zhao Y (2012) The imbalance of
2. Imai Y, Dobrian AD, Morris MA, Nadler JL Th17/Th1/Tregs in patients with type 2 dia-
(2013) Islet inflammation: a unifying target for betes: relationship with metabolic factors and
diabetes treatment? Trends Endocrinol Metab complications. J Mol Med (Berl) 90
24(7):351–360 (2):175–186
3. Jagannathan-Bogdan M, McDonnell ME, 5. Jagannathan M, McDonnell M, Liang Y,
Shin H, Rehman Q, Hasturk H, Apovian CM, Hasturk H, Hetzel J, Rubin D, Kantarci A,
Nikolajczyk BS (2011) Elevated proinflamma- Van Dyke TE, Ganley-Leal LM, Nikolajczyk
tory cytokine production by a skewed T cell BS (2010) Toll-like receptors regulate B cell
compartment requires monocytes and
cytokine production in patients with diabetes. (2008) Isolation of major pancreatic cell types
Diabetologia 53(7):1461–1471 and long-term culture-initiating cells using
6. Ehses JA, Perren A, Eppler E, Ribaux P, Pospi- novel human surface markers. Stem Cell Res 1
silik JA, Maor-Cahn R, Gueripel X, (3):183–194
Ellingsgaard H, Schneider MK, Biollaz G, 13. Pechhold K, Zhu X, Harrison VS, Lee J,
Fontana A, Reinecke M, Homo-Delarche F, Chakrabarty S, Koczwara K, Gavrilova O, Har-
Donath MY (2007) Increased number of lan DM (2009) Dynamic changes in pancreatic
islet-associated macrophages in type 2 diabetes. endocrine cell abundance, distribution, and
Diabetes 56(9):2356–2370 function in antigen-induced and spontaneous
7. Richardson AJ, Willcox A, Bone AJ, Foulis AK, autoimmune diabetes. Diabetes 58
Morgan NG (2009) Islet-associated macro- (5):1175–1184
phages in type 2 diabetes. Diabetologia 14. Nackiewicz D, Dan M, He W, Rosa K, Salmi A,
51:1686–1688 Rutti S, Westwell-Roper C, Cunningham A,
8. Brissova M, Fowler MJ, Nicholson WE, Chu A, Speck M, Shuster-Klein C, Guardiola B,
Hirshberg B, Harlan DM, Powers AC (2005) Maedler K, Ehses JA (2014) TLR2/6 and
Assessment of human pancreatic islet architec- TLR4 activated macrophages contribute to
ture and composition by laser scanning confo- islet inflammation and impair beta cell-insulin
cal microscopy. J Histochem Cytochem 53 gene expression via IL-1 and IL-6. Diabetolo-
(9):1087–1097 gia 57:1645–1654
9. Cabera O, Berman DM, Kenyon NS, 15. Westwell-Roper CY, Ehses JA, Verchere CB
Ricordi C, Berggren PO, Caicedo A (2006) (2013) Resident macrophages mediate islet
The unique cytoarchitecture of human pancre- amyloid polypeptide-induced islet IL-1beta
atic islets has implications for islet cell function. production and beta cell dysfunction. Diabetes
Proc Natl Acad Sci U S A 103:2334–2339 63(5):1698–1711
10. MacDonald MJ, Longacre MJ, Stoker SW, 16. Yin N, Xu J, Ginhoux F, Randolph GJ,
Kendrick M, Thonpho A, Brown LJ, Hasan Merad M, Ding Y, Bromberg JS (2012) Func-
NM, Jitrapakdee S, Fukao T, Hanson MS, Fer- tional specialization of islet dendritic cell sub-
nandez LA, Odorico J (2011) Differences sets. J Immunol 188(10):4921–4930
between human and rodent pancreatic islets: 17. Butcher MJ, Hallinger D, Garcia E, Machida Y,
low pyruvate carboxylase, ATP citrate lyase, Chakrabarti S, Nadler J, Galkina EV, Imai Y
and pyruvate carboxylation and high glucose- (2014) Association of proinflammatory cyto-
stimulated acetoacetate in human pancreatic kines and islet resident leucocytes with islet
islets. J Biol Chem 286(21):18383–18396 dysfunction in type 2 diabetes. Diabetologia
11. Beare A, Stockinger H, Zola H, Nicholson I 57(3):491–501
(2008) Monoclonal antibodies to human cell 18. Ricordi C (1991) Quantitative and qualitative
surface antigens. Curr Protoc Immunol 4:4A standards for islet isolation assessment in
12. Dorrell C, Abraham SL, Lanxon-Cookson humans and large mammals. Pancreas 6
KM, Canaday PS, Streeter PR, Grompe M (2):242–244
Chapter 11
Analysis of Histone Modifications in Rodent Pancreatic

Islets by Native Chromatin Immunoprecipitation
Ionel Sandovici, Lisa M. Nicholas, and Laura P. O’Neill
Abstract
The islets of Langerhans are clusters of cells dispersed throughout the pancreas that produce several
hormones essential for controlling a variety of metabolic processes, including glucose homeostasis and
lipid metabolism. Studying the transcriptional control of pancreatic islet cells has important implications for
understanding the mechanisms that control their normal development, as well as the pathogenesis of
metabolic diseases such as diabetes. Histones represent the main protein components of the chromatin
and undergo diverse covalent modifications that are very important for gene regulation. Here we describe
the isolation of pancreatic islets from rodents and subsequently outline the methods used to immunopre-
cipitate and analyze the native chromatin obtained from these cells.
Key words Islets of Langerhans, Histone modifications, Nucleosomes, Native chromatin immuno-
precipitation (NChIP), Antibodies
1 Introduction
The islets of Langerhans represent the endocrine component of the

pancreas and comprise only 1–2% of the volume of the whole
organ. They are composed of several types of hormone-secreting
cells including alpha cells (glucagon), beta cells (insulin and amy-
lin), delta cells (somatostatin), PP cells (pancreatic polypeptide),
and epsilon cells (ghrelin), as well as an intricate network of blood
vessels, which enable the direct release of these hormones into the
bloodstream. In rodents, endocrine cells are first specified around
embryonic day 15.5 (E15.5) and, upon their migration, they form
the first islets of Langerhans between E16.5 and birth [1]. In adult
rodents beta cells constitute the major endocrine cell type
(65–80%) and are generally centrally located in the pancreatic islets,
being surrounded by the other cell types.
There are several methods used for isolation of islets of Lan-
gerhans from adult rodents. Most of these techniques rely on
digestion of the whole pancreas with collagenase injected into the
199
200 Ionel Sandovici et al.
common bile duct, the splenic vein, or after pancreas fragmenta-

tion. This step is then followed by separation of pancreatic islets
using gradient density solutions and manual picking of individual
pancreatic islets [2–4].
Epigenetic mechanisms, such as DNA modifications and post-
translational histone modifications (methylation, acetylation, phos-
phorylation, ubiquitination) play important roles during the
development of the pancreas, as well as in the maintenance of
cellular differentiation states and the normal functioning of adult
pancreatic endocrine cells [5]. Epigenetic alterations induced by
environmental exposures during early development or throughout
the lifetime are currently thought to play important roles in pan-
creatic islets dysfunction and subsequently in the growing incidence
of type 2 diabetes [6, 7].
Chromatin immunoprecipitation (ChIP) is a type of immuno-
precipitation that is often used for characterization of covalent
histone modifications. There are two main types of ChIP: cross-
linked ChIP (XChIP), which uses as starting material reversibly
cross-linked chromatin sheared by sonication and native ChIP
(NChIP), which instead uses chromatin sheared by micrococcal
nuclease digestion. NChIP is the method of choice when histone
modifications are studied [8, 9]. The major advantage of NChIP
over XChIP is antibody specificity. This is because most antibodies
used for recognizing specific histone modifications are raised
against unfixed, synthetic peptides that bear the target covalent
modification. These epitopes can be disrupted or even destroyed
by formaldehyde cross-linking, particularly as the cross-links are
likely to involve lysine amino groups which are often the sites for
many histone modifications. This phenomenon also explains the
consistently lower efficiency of XChIP protocols compared to
NChIP [8, 9].
In this chapter, we will first describe the isolation of pancreatic
islets from adult mice and rats and then outline the NChIP protocol
that can be used for analysis of specific posttranslational histone
modifications. Characterization of histone modifications in pancre-
atic islets collected from fetuses or from neonatal rodents requires
alternative protocols, such as carrier ChIP [10].
2 Materials
Prepare all solutions fresh using nuclease-free water and molecular

biology-grade reagents (unless indicated otherwise). Some solu-
tions can be stored for a short time at 4 C (specified below).
Housing, breeding, and killing of rodents should be performed in
accordance with the local institutional Scientific Procedures act.
NChIP in Pancreatic Islets 201
2.1 Instruments 1. CO2 gas chamber for euthanasia.

and Reagents 2. Surgical board, dissection microscope, and light source.
Required
3. Dissection scissors, blunt forceps, and metallic clamps.
for Pancreatic Islets
Isolation 4. Ethanol.
5. 18 G (gauge), 25 G, 27 G, and 30 G needles.
6. 1 ml, 5 ml, and 10 ml syringes.
7. 15 ml and 50 ml Falcon tubes.
8. Disposable pipettes (5 ml, 10 ml, and 25 ml) and pipette aid.
9. Water bath at 37 C.
10. Bench-top centrifuge with a swing-out bucket rotor for 50 ml
polypropylene tubes and microcentrifuge with cooling system
for 1.5 ml tubes.
11. HBSS buffer.
12. Collagenase P (Sigma Aldrich, 11213865001).
13. Histopaque 1119 (Sigma Aldrich, 11191) and Histopaque
1077 (Sigma Aldrich, 10771).
14. Sterile Petri dishes.
15. Liquid N2 and liquid N2 container.
2.2 Instruments 1. Magnetic stirrers.

and Reagents 2. Glass dounce homogenizers/disruptors.
Required for NChIP
3. 15 ml and 50 ml Falcon tubes.
4. Sigmacote (Sigma Aldrich) (for siliconizing Eppendorf and
Falcon tubes).
5. Rotating wheels at 4 C and room temperature.
6. Dialysis tubing (0.5 mm thick, 10 kDa pore width).
7. Disposable pipettes (5 ml, 10 ml, and 25 ml) and pipette aid.
8. 10 μl, 100 μl, and 1000 μl pipettor and filtered tips.
9. Phosphate-buffered saline (PBS).
10. Sodium butyrate (for analyzing histone acetylation only).
11. Phenylmethanesulfonyl fluoride (PMSF).
12. Tween 40.
13. Protease inhibitors cocktail (Sigma Aldrich, 11697498001).
14. Micrococcal nuclease (MNase; 100 units/μl in 10 mM
Tris–HCl pH 7.5, 50 mM NaCl, 1 mM EDTA, 50% (v/v)
glycerol).
15. Affinity-purified antibodies, ChIP-grade, e.g., rabbit poly-
clonal anti-H3K27ac, rabbit polyclonal anti-H3K27me3, rab-
bit polyclonal anti-H3K4me1, rabbit polyclonal anti-
H3K4me3, rabbit polyclonal anti-H3K9me3 (Diagenode is

preferred).
16. DiaMag protein A-coated magnetic beads (Diagenode is
preferred).
17. Magnetic rack for 1.5 ml Eppendorf tubes.
18. Horizontal electrophoresis system for agarose gels.
19. Gel Doc 2000 gel documentation system (Bio-Rad provides an
excellent set of equipment).
20. Agarose.
21. Ethidium bromide solution (20 mg/ml).
22. 100 bp DNA size ladder.
23. NanoDrop spectrophotometer.
24. Other chemicals: MgCl2, CaCl2, Tris, boric acid, EDTA, SDS,
sucrose—all molecular grade.
2.3 Instruments 1. Proteinase K (Melford Laboratories is preferred).

and Reagents 2. Thermomixer shaker.
Required for DNA
3. Phenol:chloroform:isoamyl alcohol 25:24:1 (v:v:v).
Isolation
and Downstream 4. Chloroform.
Analysis by 5. 2 ml Phase Lock gel tubes.
Quantitative PCR 6. 3 M sodium acetate, pH 5.2.
(qPCR) 7. Absolute ethanol.
8. Glycogen solution 20 mg/ml, molecular grade.
9. 10 mM Tris–Cl, pH 7.5.
10. Qubit® 3.0 Fluorometer.
11. Qubit® dsDNA HS assay (Life Technologies is a recommended
choice).
12. Locus-specific primers (Sigma Aldrich is an excellent recom-
mended supplier).
13. SYBR-Green master mix.
14. QuantStudio6Flex system (Life Technologies) or equivalent
(compatible for SYBR-Green-based quantitative PCR).
3 Methods
3.1 Isolation 1. Prepare fresh collagenase P solution in HBSS (at a concentra-

of Pancreatic Islets tion of 4 mg/ml) and place on ice (see Note 1). Prepare 3 ml
from Adult Mice or and 10 ml solution for each adult mouse and rat, respectively.
Rats (Fig. 1) 2. Adult mice or rats should be euthanized using a CO2 chamber
following institutionally approved instructions. The rodent is
Fig. 1 Isolation of pancreatic islets from mature mice. (a) The metallic clamp is first placed on the common
biliary duct near liver, and then the collagenase P solution is slowly injected into the duct under a microscope.
(b) The red dotted line indicates the fully inflated pancreas at the end of the collagenase injection. (c) Isolated
mouse pancreatic islets
ready for exsanguination after no response to pinching its foot.

Exsanguination is performed by heart puncture using 1 ml
syringes fitted with 25 G needles (for mouse) or 5 ml syringes
fitted with 18 G needles (for rat).
3. Wet the abdominal fur with 70% ethanol to reduce the chance
of hair contamination in the intraperitoneal cavity during
subsequent steps. Open the abdomen with a cross-like incision,
in order to expose all the organs within the peritoneal cavity.
Place the animal with the head toward the surgeon and the tail
away. This allows easier clamping of the common bile duct near
the liver, using metallic clamps. Additional clamping of the
duodenum on each side of the junction with the common
biliary duct is optional but desirable, as it avoids leakage of
the collagenase solution into the gut.
4. Under the microscope, cannulate the common bile duct imme-
diately downstream the junction between the cystic duct and
the left hepatic duct, using a 27 G (rat) or 30 G (mouse)
needle. Slowly inject the entire volume of collagenase P solu-
tion, until the pancreas is completely inflated (see Note 2).
5. Using dissection scissors, remove the pancreas, by carefully

detaching it from the duodenum, jejunum, the greater curva-
ture of the stomach, and finally the spleen. Then, place the
dissected pancreas in a 50-ml Falcon tube containing 1 ml cold
HBSS and immediately place on ice.
6. Repeat the above steps for each rodent (we recommend maxi-
mum four rodents for one collection), and then place all the
Falcon tubes in a water bath prewarmed at 37 C for 20 min
(see Note 3).
7. After incubation, add 10–15 ml ice-cold HBSS to each Falcon
tube to stop the digestion. Place the Falcon tubes in a rack and
shake them hard by hand for approximately 1 min to comple-
ment the enzymatic digestion of the pancreas with mechanical
dissociation of the tissue fragments (the solution should now
have the consistency of a pea soup and should be free of large,
undigested pancreatic fragments).
8. Place the Falcon tubes in a prechilled centrifuge (4 C) and spin
for 1 min at 200 g. Then, carefully remove and discard the
supernatant using a pipette. Wash the pellets with 10 ml
ice-cold HBSS, shake the tubes for 1 min, and centrifuge as
above (repeat for a total of three washes).
9. After the third wash, resuspend the pellet in 10 ml cold HBSS
using a pipette and pass the mixture through a strainer (0.5 mm
metallic mesh) into a new 50-ml Falcon tube and wash one
more time with HBSS as above.
10. Subsequently resuspend the pellet in 5 ml ice-cold Histopaque
1119 (by pipetting until homogeneous), then carefully layer on
top 8 ml ice-cold Histopaque 1077, followed by another 8 ml
of ice-cold HBSS (the three layers should not mix).
11. Carefully place the Falcon tubes in the centrifuge and spin at
1065 g and 4 C for 25 min. At the end of the centrifugation,
the islets should be visible as rounded white specs at the interface
between the Histopaque 1077 and HBSS layers (see Note 4).
12. Carefully aspirate the islets and place them on ice in Petri dishes
containing HBSS. Under an optic microscope, wash the islets
several times with HBSS to remove any traces of Histopaque, as
well as to minimize any contamination with small fragments of
exocrine pancreas. Then, count the islets, hand pick them with
a pipette, and place them in 1.5 ml microcentrifuge tubes (see
Note 5). Pellet the islets by centrifugation for 3 min at 200 g
and 4 C, carefully remove the HBSS, wash three times in
ice-cold PBS, snap-freeze the tubes in a liquid N2 container,
and store them at 80 C until used for NChIP.
Fig. 2 Diagram depicting the principle of NChIP. (a) Isolation of nuclei. (b)
Chromatin digestion. (c) Immunoprecipitation with an antibody that recognizes
a specific chromatin modification (green star). (d) DNA purification. (e) Analysis
by quantitative PCR
3.2 Isolation 1. Isolation of nuclei. Approximately 1000 pancreatic islets snap

of Chromatin from frozen at collection and stored at 80 C (i.e., islets collected
Pancreatic Islets from ~4 adult mice or 2 rats) are placed in a 7-ml all-glass
(Fig. 2) homogenizer (kept on ice) containing 2.5 ml ice-cold 1
TBS (10 mM Tris–HCl pH 7.5, 150 mM NaCl, 3 mM
CaCl2, 2 mM MgCl2, 5 mM Na butyrate pH 8.0) and supple-
mented with protease inhibitors (1 pellet mini inhibitor per
20 ml) (see Note 6). Then, add an equal volume of 1.0% v/v
Tween 40 in 1 TBS and supplement with PMSF to a final
concentration of 0.5 mM (see Note 7).
2. Incubate on ice for a total of 30 min. After the first 10 min,
apply 10 strokes using the loose fitting pestle, then after
another 10 min, apply again 10 strokes with the tight fitting
pestle (see Note 8).
3. The cell lysates are then transferred to 15-ml Falcon tubes,
which are spun at 10,000 g and 4 C for 20 min and the
supernatants are discarded. The nuclei pellets are then resus-
pended in 5 ml 25% [w/v] sucrose/1 TBS and the suspen-
sion is underlayed with 2.5 ml of 50% [w/v] sucrose/1 TBS.
Spin the tubes for 25 min at 14,000 g and 4 C.
4. Discard the supernatants and wash the nuclei pellets in 5 ml
25% [w/v] sucrose/1 TBS. Centrifuge again the tubes for
25 min at 14,000 g and 4 C (see Note 9).
5. Resuspend the nuclei in 0.5 ml digestion buffer (0.32 M
sucrose, 4 mM MgCl2, 1 mM CaCl2, 50 mM Tris–HCl,
pH 7.5, 0.1 mM PMSF, 5 mM Na butyrate, and protease

inhibitors) in 1.7 ml Eppendorf tubes.
6. A small aliquot of 5 μl nuclei is then mixed with 5 μl digestion
buffer supplemented with 0.2% SDS and is then used for
checking the amount of chromatin with a NanoDrop spectro-
photometer. Tubes are then centrifuged at 10,000 g and
4 C for 10 min and the nuclei resuspended in digestion buffer
at a final concentration of 50 μg/ml (see Note 10).
7. Digestion of chromatin: The conditions used for chromatin
digestion with micrococcal nuclease (MNase) should be deter-
mined empirically. We usually obtain good results with 5 min
digestion at 37 C using 5 UI MNase per 50 μg chromatin.
These conditions give us a chromatin ladder rich in mono-, di-,
tri-, tetra-, and penta-nucleosomes. However, if a more detailed
map of histone marks is required, this step can be extended to
7 min, which normally increases the yield of mono- and
di-nucleosomes in the final preparation (see Note 11).
8. At the end of the previous step, the MNase digestion is stopped
by adding 0.5 M EDTA solution, to a final concentration of
5 mM, and placing the tubes on ice for 5 min. The tubes are
then spun at 8000 g and 4 C for 5 min. The supernatant is
placed in a new eppendorf tube (this is fraction S1 (mostly
mono-nucleosomes), which is kept at 4 C until the next day,
when fraction S2 is obtained; the pellet is retained for dialysis.
9. Chromatin dialysis: This step requires special dialysis tubing
(0.5 mm thick, 10 kDa pore width). Before use, the tubing is
washed twice with distilled water (see Note 12).
10. The pellet obtained after fraction S1 is resuspended in 500 μl
dialysis buffer (0.2 mM EDTA, 1 mM Tris–HCl pH 7.5,
0.2 mM PMSF, 5 mM Na butyrate and protease inhibitors).
11. The above mix is placed in the dialysis tubing and the ends are
closed with plastic clamps. The tubing is submerged in 2 l of
dialysis buffer and incubated overnight at 4 C with constant
slow stirring using a magnetic stirring bar.
12. The dialysate is placed in 1.7 ml eppendorf tubes and spun at
4 C and 500 g for 10 min. The supernatant represents
fraction S2. The pellets are resuspended in 200 μl dialysis buffer
(fraction S3).
13. Assessment of chromatin quality: Small aliquots (18 μl) of
chromatin are removed from each fraction (S1, S2, and S3)
and each sample is mixed with 2 μl of 1% [w/v] SDS (final
concentration 0.1% SDS). After the mixtures are gently flicked
and incubated at room temperature for 5 min (which will allow
a good separation of the nucleosome DNA fragments from the
Fig. 3 Analysis of native chromatin fractions isolated from pancreatic islets using
agarose gel electrophoresis. Bands corresponding to chromatin fragments of
one nucleosome (mono) to four nucleosomes (tetra) in length are indicated. The
sizes of the chromatin fragments are verified against the 100 bp DNA ladder
(Hyperladder IV—Bioline) loaded on the right side of the gel
histone proteins), their concentrations are measured with a

NanoDrop spectrophotometer (two readings using 1 μl each).
14. The remaining 18 μl of each mixture receives 4 μl of 6 DNA
loading buffer, then the samples are loaded onto a standard
1.2% agarose gel without ethidium bromide (see Note 13). The
gel is run at 2–3 V/cm for about 3 h (this low voltage facilitates
good separation of the DNA fragments). The gel is then
stained with 0.5 μg/ml ethidium bromide and the sizes of
the DNA fragments are observed under UV light. Typically,
fraction S1 contains mostly mono-nucleosomes, S2 di- to
penta-nucleosomes, and S3 mostly large fragments of undi-
gested DNA (Fig. 3).
3.3 Chromatin 1. Only fractions S1 and S2 are used for immunoprecipitation,

Immunoprecipitation mixed in equal amounts. We obtain good results with as little as
2 μg chromatin per IP (1 μg fraction S1 plus 1 μg fraction S2).
Using this amount, we can perform 3–4 IPs/sample. An equal
aliquot (2 μg) of chromatin should be stored at 4 C (to be
used as input fraction during qPCR quantification). The IPs are
performed in siliconized eppendorf tubes and the total
volumes are adjusted to 1 ml, using incubation buffer (5 mM
EDTA, 20 mM Tris–HCL pH 7.5, 50 mM NaCl, 0.1 mM
PMSF, 20 mM Na butyrate).
2. The amount of antibody required depends on manufacturer’s
recommendations (depending on the antibody, usually 1–2 μg
is required for an IP on the equivalent chromatin isolated from
300,000 pancreatic islet cells). The use of affinity-purified
ChIP-grade antibodies is recommended. At least one mock
tube (containing an isogenic IgG serum that is not directed
against chromatin proteins, for example, rabbit IgG antibody)
should be used in each experiment as negative control for

unspecific binding (see Note 14).
3. Wrap the tubes in parafilm and incubate for 16 h (overnight) in
the cold room (4 C) on a rotator machine, at slow speed
(20 rpm).
4. Next morning, 20 μl DiaMag protein A-coated magnetic beads
are prepared for each immunoprecipitation (they are recom-
mended when rabbit polyclonal antibodies are used, as well as
mouse IgG2a, IgG2b, and IgA, guinea pig IgG, dog IgG, or pig
IgG) (see Note 15). The beads are washed four times with 1 ml
ice-cold incubation buffer (5 mM EDTA, 20 mM Tris–HCL
pH 7.5, 50 mM NaCl, 0.1 mM PMSF, 20 mM Na butyrate) in
1.5 ml siliconized eppendorf tubes, using the magnetic rack to
recover the beads after each wash. After the fourth wash,
resuspend the beads in 20 μl incubation buffer.
5. Using siliconized pipette tips, 20 μl prewashed beads are added
to each IP tube (including the mock tube) and the tubes are
incubated at 4 C on the rotator machine (20 rpm) for 4–6 h.
6. The three washing buffers (A, B, and C) are prepared and
prechilled. The washing buffers have the same composition,
except for increasing concentrations of NaCl: 50 mM Tris–HCl
pH 7.5, 10 mM EDTA, 5 mM Na butyrate, 50 mM NaCl
(wash A), 50 mM Tris–HCl pH 7.5, 10 mM EDTA, 5 mM Na
butyrate, 100 mM NaCl (wash B), 50 mM Tris–HCl pH 7.5,
10 mM EDTA, 5 mM Na butyrate, 150 mM NaCl (wash C).
7. All washes are performed at 4 C. After the incubation with
DiaMag protein A-coated magnetic beads is over, the tubes are
briefly spun, then placed on the magnetic rack for 1 min. The
supernatants are discarded by aspiration, and the beads are
resuspended in 1 ml wash A. Incubate the tubes on the rotating
wheel for 5 min, spin the tubes briefly, place them for 1 min on
the magnetic rack, and discard the supernatants. Repeat in the
same way with washes B and C.
8. After the last supernatant is discarded, resuspend the beads in
500 μl elution buffer (50 mM NaCl, 20 mM Tris–HCl pH 7.5,
20 mM Na butyrate, 0.1 mM PMSF, 5 mM EDTA, and 0.5%
SDS). Incubate for 30 min on a rotating wheel at room tem-
perature (this is the bound fraction).
3.4 DNA Extraction 1. To each bound and mock fraction, add proteinase K (to a final
and Quantification concentration of 20 μg/ml), then incubate the tubes at 55 C
and 800 rpm (on a Thermomixer Comfort shaker or equiva-
lent) for 1–2 h (see Note 16). The input sample (saved in
Subheading 3.3) should also be used at this stage for DNA
extraction. After the proteinase K digestion is over, briefly spin
the tubes and place them in the magnetic rack for 1 min.
Transfer supernatants to phase lock tubes (eppendorf) and

discard the tubes containing the magnetic beads (see Note 16).
2. Next, perform a phenol–chloroform extraction. First, add an
equal volume (500 μl) of phenol:chloroform:isoamyl alcohol
25:24:1 in each tube, vortex for 1 min, and spin (16,000 g)
for 10 min in a microcentrifuge. Transfer the aqueous superna-
tant into a new phase lock tube, add 500 μl chloroform, vor-
tex again for 1 min, and spin for 10 min at 16,000 g (see
Note 17).
3. Finish the DNA extraction with a standard ethanol precipita-
tion (required to reduce the concentration of salts): add 100 μl
of 3 M sodium acetate (pH 5.2), 2 μl glycogen solution
(20 mg/ml), and 700 μl pure ethanol. Vortex thoroughly and
incubate at 20 C overnight. Next day, centrifuge the samples
(16,000 g, 15 min, 4 C) to precipitate the DNA, aspirate the
supernatant, wash the pellets with 1 ml ice-cold 70% ethanol,
spin again (16,000 g, 15 min, 4 C), aspirate the supernatant,
and air dry the pellets. At the end, dissolve the DNA from
input, bound, and mock tubes in 50 μl of 10 mM Tris–Cl
(pH 7.5).
4. Take 5 μl (10%) of each bound, mock, and input DNA and
determine the concentrations using the Qubit® 3.0 Fluorome-
ter. First, mix 10 μl of each standard solution provided in the
Qubit® dsDNA HS assay kit with 190 μl Qubit® working
solution. Vortex the tubes for 2–3 s, incubate at room temper-
ature for 2 min, then read the tubes in the Qubit® Fluorometer.
For each sample (bound, mock, or input), mix 2.5 μl DNA with
197.5 μl working solution and perform duplicate readings. All
fractions are diluted to the same final concentration (see Note
18), then the DNA samples are stored at 20 C until ready to
analyze them by qPCR. Alternatively, the bound and input
DNA can be used for preparation of libraries and for high-
throughput sequencing (NChIP-Seq protocols are described
elsewhere [11, 12]).
3.5 Analysis by 1. Design primers suitable for qPCR (80–150 bp) using freely
Quantitative PCR available software such as Primer3. Before the first use, primers
(qPCR) and Data should be tested for specificity, using input DNA as template.
Interpretation The optimal annealing temperature (no primer dimers or non-
specific PCR products) can be identified using a gradient PCR
machine and by melting curve analysis (ideally this should be
between 58 C and 62 C). Using ten-fold serial dilutions of
input DNA, the efficiency of each primer pair can be calculated
using the formula E ¼ 10(1/slope) 1. The optimal primer
efficiency is 1; however, primer efficiencies between 0.9 and 1.1
are acceptable. A good primer pair should generate Ct values
around 25 when 1 ng input DNA is used as template. If Ct

values are >30, primers should be redesigned.
2. For qPCR analysis of NChIP samples, 1 ng DNA from each
fraction is used as template. qPCR measurements can be done
with primers designed for the regions of interest and the SYBR
Green system. Ratios of bound/input >1 signify local enrich-
ment of the histone mark analyzed, while ratios <1 mean local
depletion.
4 Notes
1. The type of collagenase used is important for maintaining the

integrity of pancreatic islets. Collagenase P, isolated from Clos-
tridium histolyticum, has a very high collagenase activity
(>1.5 U/mg) and has been functionally tested for the isolation
of rodent pancreatic islets.
2. If the integrity of the common biliary duct is compromised
during cannulation, the collagenase P solution can also be
carefully injected into each lob of the pancreas. However, this
may reduce the total number islets isolated per animal.
3. Incubation times may vary between different lots of collagenase
and should be tested prior to use in experimental conditions.
Incomplete digestion of the pancreas results in reduced yield of
pancreatic islets.
4. If fewer islets than expected are observed, resuspend the pellet
produced during the previous step and inspect for islets. This
can happen if more than one pancreas is used per tube during
the gradient centrifugation.
5. Using the protocol described here, we regularly harvest
200–300 islets/adult mouse or 500–800 islets per adult rat.
6. An average rodent islet is estimated to contain approximately
1000–2000 cells [13].
7. It is essential to use 5 mM Na butyrate in all solutions through-
out the chromatin isolation steps when using antibodies against
acetylated histones, in order to prevent histone deacetylation.
PMSF is a serine protease inhibitor with a short half-life in
aqueous solutions (110 min at pH ¼ 7 and 35 min at
pH ¼ 8). PMSF stock is regularly prepared as 0.1 M solution
in isopropanol and stored in aliquots at 20 C.
8. We found that incubation for 30 min in presence of the deter-
gent and the addition of 10 + 10 pestle strokes lead to efficient
lysis of pancreatic islet cells. Excessive extension of this step
increases the probability of nuclei bursting and could lead to
failure of the chromatin extraction. Efficient release of nuclei is
verified using a phase-contrast microscope (intact cells have the

central dark region of the nucleus surrounded by a halo, which
is the less dense cytoplasm). At least 80% free nuclei should be
obtained for a successful NChIP and additional strokes using
the tight-fitting pestle can be applied if required.
9. The gradient sucrose centrifugation leads to very clean nuclei
preparations, which are essential for a successful immunopre-
cipitation. However, when the percentage of free nuclei
obtained during the cell lysis is very high, this step can be
omitted in order to limit the excessive loss of material.
10. Obtaining a suspension of single nuclei (no visible clumps) in
digestion buffer is critical for ensuring even access of the
micrococcal nuclease to chromatin in the following step.
11. With a starting amount of ~1000 islets, we usually obtain up to
10 μg chromatin. It is essential to carefully control the condi-
tions for micrococcal nuclease digestion, especially in the initial
preparations. Excessive digestion of the chromatin leads to
sub-nucleosomal fragments, while underdigestion leads to
excess of larger oligonucleosomes, which severely reduces the
resolution of the assay.
12. Pieces of 10–20 cm dialysis tubing are first boiled for 10 min in
0.5 l of 2% sodium bicarbonate, then rinsed twice in distilled
water, boiled again for 10 min in 1 mM EDTA (pH 8.0), then
cooled and stored until use at 4 C in 1 mM EDTA (pH 8.0).
The tubing can be stored in these conditions for several
months.
13. Do not place ethidium bromide in the agarose gel or the
electrophoresis buffer, because of the presence of SDS.
14. Diagenode offers premium antibodies, which have reached the
highest level of validation from extensive in-house validation,
in combination with numerous collaborations with epigenetics
experts.
15. The binding capacity for DiaMag protein A-coated magnetic
beads is approximately 5 μg antibody per 20 μl beads. If the
amount of antibody used during immunoprecipitation is
higher, the volume of DiaMag protein A-coated magnetic
beads should be increased accordingly. As an alternative to
magnetic beads, a 50% w/v slurry of Protein A Sepharose
(PAS—GE Healthcare) can be used. Add 100 μl of PAS and
incubate for a minimum of 2 h at room temperature on a
rotating platform. Centrifuge 300 g, 5 min at room temper-
ature, and wash the pellets as described in Subheading 3.3
steps 6–7 except with centrifugation after each wash. Elute
the bound material using 100 μl 1% SDS/incubation buffer,
and incubate for 5 min with gentle agitation. Centrifuge
300 g, 5 min at room temperature, and retain the superna-

tant (bound).
16. When required, both the DNA and protein can be isolated
from the same input and bound samples. Add an equal volume
of phenol:chloroform to the input and bound material. Briefly
vortex before centrifugation at 300 g for 5 min. Transfer
supernatant to a phase lock tube and continue from Subhead-
ing 3.4 step 1 to isolate the DNA. Keep the first phenol:
chloroform phase and add 5 μl bovine serum albumin,
1/100th volume 10 M H2SO4 and 12 volumes of acetone.
Following precipitation at 80 C for 16 h, centrifuge at
300 g for 10 min at 4 C, and wash the pellet once in
acidified acetone (one volume of 100 mM H2SO4 mixed with
six volumes of acetone) and three times in dry acetone. Resus-
pend the pellet in 50–100 μl TE (10 mM Tris–HCl, 1 mM
EDTA), before analysis using either SDS-PAGE or
AUT-PAGE.
17. As an alternative to phenol:chloroform and ethanol precipita-
tion for purification of DNA, the Qiaquick purification kit
(Qiagen) can be used, following the manufacturer’s
instructions.
18. The expected yield of DNA depends upon the amount of
antibody used and the amount of starting material, as well as
the distribution of the histone modification in the genome. For
example, the recovery rate in a successful immunoprecipitation
with the H3K4me3 antibody is about 10–20% (i.e.,
200–400 ng when the starting material is 2 μg chromatin).
For less frequent histone modifications such as H3K4me1, the
recovery rate could be around 5%. The recovery in the negative
control (mock) sample should be less than 1%.
References
1. Puri S, Hebrok M (2010) Cellular plasticity epigenetic programming of the endocrine pan-
within the pancreas—lessons learned from creas: consequences for type 2 diabetes. Cell
development. Dev Cell 18:342–356 Mol Life Sci 70:1575–1595
2. Carter JD, Dula SB, Corbin KL et al (2009) A 6. Sandovici I, Smith NH, Nitert MD et al (2011)
practical guide to rodent islet isolation and Maternal diet and aging alter the epigenetic
assessment. Biol Proced Online 11:3–31 control of a promoter-enhancer interaction at
3. Ravier MA, Rutter GA (2010) Isolation and the Hnf4a gene in rat pancreatic islets. Proc
culture of mouse pancreatic islets for ex vivo Natl Acad Sci U S A 108:5449–5454
imaging studies with trappable or recombinant 7. Haumaitre C (2013) Epigenetic regulation of
fluorescent probes. Methods Mol Biol pancreatic islets. Curr Diab Rep 13:624–632
633:171–184 8. O’Neill LP, Turner BM (2003) Immunopre-
4. Li F, Jiang X, Li Y et al (2013) Isolation of cipitation of native chromatin: NChIP. Meth-
mouse islet by collagenase perfusion through ods 31:76–82
the splenic vein. Transplantation 96:e88–e89 9. Wagschal A, Delaval K, Pannetier M et al
5. Sandovici I, Hammerle CM, Ozanne SE et al (2007) Chromatin immunoprecipitation
(2013) Developmental and environmental (ChIP) on unfixed chromatin from cells and
tissues to analyze histone modifications. CSH Protoc. https://doi.org/10.1101/pdb.

Protoc 2007:pdb.prot4767 prot5237
10. O’Neill LP, VerMilyea MD, Turner BM (2006) 12. Brind’Amour J, Liu S, Hudson M et al (2015)
Epigenetic characterization of the early embryo An ultra-low-input native ChIP-seq protocol
with a chromatin immunoprecipitation proto- for genome-wide profiling of rare cell popula-
col applicable to small cell populations. Nat tions. Nat Commun 2015(6):6033
Genet 38:835–841 13. Jo J, Choi MY, Koh DS (2007) Size distribu-
11. Cuddapah S, Barski A, Cui K et al (2009) tion of mouse Langerhans islets. Biophys J
Native chromatin preparation and Illumina/ 93:2655–2666
Solexa library construction. Cold Spring Harb
Chapter 12
Quantification of Pancreatic Islets: Using Image Analysis

Tools
Parvathy E. Harikumar
Abstract
Histological image analysis is becoming an increasingly important tool for research in biological science.
They are important in analyzing biological systems on various scales, from structural details to determina-
tion of number of cells, its area, localization, and concentration. This chapter focuses on analysis of
pancreatic sections stained for insulin and glucagon using a commercially available software.
Key words Histology, Pancreas, Immunohistochemistry, Whole-slide image, Image analysis, Visio-
morphDP, ImageJ
1 Introduction
Histological analysis of tissue sections is very important in clinical

research, histopathology, disease diagnosis, and drug discovery.
Historically, analysis and evaluation of histological sections was
mere qualitative and subjective, and was prone to observer varia-
tions. Also, this type of quantification is laborious and time-
consuming, as it requires examination of multiple tissue sections
from whole-slide images. However, in recent years, the technology
has improved so much that digital slide scanners are used for image
acquisition, and manual or automated analysis is done using
computer-assisted image analysis software for quantitative
histomorphometry.
This chapter aims to focus on the analysis of stained histological
tissue sections of the pancreas using a commercial image analysis
software known as VisiomorphDP. VisiomorphDP software from
Visiopharm (Horgholm, Denmark) was used to create a range of
specific protocols to perform quantitative image analysis by exploit-
ing spectral and morphological features of pancreatic tissue sec-
tions. Image analysis techniques used in the chapter are based on
segmentation, whereby the user identifies a feature of interest, and
215
216 Parvathy E. Harikumar
its constituent pixel values are then used by the software to identify
all similar features in an image. Some of the analyses included in this
chapter are quantification of total pancreas area, total islet area,
alpha and beta cell area, islet number, and size distribution with
reference to specific parameters.
Although the method detailed here is specific for analysis with
VisiomorphDP, the idea is easily transferrable across different
computational image processing systems including imageJ, an
open-source image-processing program. This chapter does not
discuss the algorithms, its implementation, and image processing
techniques in detail as it is beyond the scope. Here, the author
presents a general protocol for detecting and quantifying islets and
its associated measurable features.
2 Materials
2.1 Animals C57/Bl6 mice (Charles River, Manston, UK) were obtained at
5–6 weeks of age. All procedures were conducted in accordance
with the UK Government Animals Scientific Procedures Act 1986,
and approved by the University of Buckingham Ethical Review
Board. The animals were housed in solid-bottomed, sawdust-filled
cages with access to tap water and maintained on a standard chow
diet fed ad libitum (10% kcal fat, 70% kcal carbohydrates, 20% kcal
protein [Beekay Diets, New Brunswick, NJ, USA]). The source of
fat in the chow diet was soybean oil. Mice were maintained at a
controlled temperature of 23 1 C, on a 12 h light–dark cycle. At
12 weeks of age, mice were killed using carbon dioxide inhalation.
2.2 Tissue 1. PBS-T: Phosphate buffered saline (PBS) with 0.1% Triton
Preparation & Staining X-100.
2. IMS: Industrial methylated spirit (Hayman Ltd., Essex, UK).
IMS can be used instead of molecular grade ethanol.
3. 6% hydrogen peroxide: 30% hydrogen peroxide is diluted with
methanol and stored in dark conditions at 4 C until ready to
be used. Use fresh preparations at all times.
4. Acid alcohol: 1% HCl is dissolved in 70% ethanol (v/v).
2.3 Suppliers The following reagents and equipment were purchased from vari-
ous suppliers: Histology cassettes (VWR International Ltd., UK),
automated tissue processor (TP1020, Leica Microsystems, Milton
Keynes, UK), Histo-ClearTM II (National Diagnostics, Hessle,
UK), paraffin wax (Thermo Scientific, Histoplast, UK), microtome
stainless steel blades (Feather S35; pfm medical UK Ltd., Chesire,
UK), microscope slides (Starfrost, Fischer Scientific, UK), VECTA-
SHIELD® HardSet™ Mounting medium (Vector Laboratories,
Peterborough, UK), insulin and glucagon antibody (Dako,
Quantification of Pancreatic Islets: Using Image Analysis Tools 217
Denmark), secondary antibodies and blocking serums (Sigma-

Aldrich, Poole, UK), ImmPACT DAB and SG peroxidase substrate
(SK-4105 and SK-4705;Vector Laboratories, USA), and Aperio
Scanscope CS scanner (Aperio, CA, USA).
3 Methods
3.1 Pancreas 1. Mice were sacrificed using carbon dioxide inhalation, after
Preparation which dissection was carried out to obtain pancreas for
histology.
2. Excess fat and connective tissues bound to the pancreas were
removed using forceps and scissors.
3. Weigh the pancreas and lay each pancreas between two sheets
of filter paper to keep the “head” (pancreas bound to duode-
num) and “tail” (pancreas bound to spleen) orientation and to
maintain the structure of the pancreas.
3.2 Fixation 1. Individual tissues were placed in histology cassettes and fixed in
and Tissue Processing 10% neutral buffered formalin for 17 h in room temperature
(RT) (see Note 1).
2. Specimens were processed using a semi-enclosed benchtop
automated tissue processor. Tissue processing involves dehy-
dration of the samples using graded ethanol (IMS) followed by
clearing in Histo-Clear™ II and finally paraffin wax infiltration
according to the program listed in Table 1.
3.3 Embedding 1. After processing, samples were removed promptly and trans-
and Sectioning ferred to metal cassettes before embedding in paraffin wax at
60 C using a Leica embedding station (model EG1150H) to
form paraffin blocks.
2. The blocks were cooled in ice-cold water for few seconds prior
to sectioning (see Notes 2 and 3).
3. Each block was then placed on a Leica microtome (model
RM2255) set with a clearance angle of 4 to cut 4 μm sections
with stainless steel blades.
4. During sectioning, each pancreas was divided into six different
planes that were separated by an interval of 80 μm, and from
each plane, 10 serial sections of 4 μm thickness were taken (see
Note 4).
5. Sections were then floated in a 37 C water bath and were taken
using positively charged glass slides.
6. Sections were then initially dried for 10 min on hot plate and
then allowed to dry overnight at 35 C in an oven.
7. The resulting slides can be used for various staining purposes as
well as immunohistochemistry.
Table 1
Processing protocol
Station Reagent Time (min)

1 50% IMS 60
2 70% IMS 30
3 80% IMS 30
4 90% IMS 60
5 100% IMS 60
6 100% IMS 60
7 100% IMS 60
8 Histo-Clear™ II 15
11 Paraffin wax (60 C) 15

12 Paraffin wax (60 C) 90
3.4 Rehydration 1. Prior to staining, slides were dewaxed by heating in an oven at

60 C for 15 min followed by rehydration (see Note 5).
2. Rehydration involves incubations as follows: Histo-Clear™ II
[2 2 min], graded IMS [100%, 90%, 70%; each for 1 min],
with a final rinse in running tap water.
3.5 Hematoxylin 1. Routine H&E staining was performed according to [1].

and Eosin (H&E) 2. Briefly, slides were immersed in Harris hematoxylin for 8 min
followed by rinsing in water for few seconds.
3. The slides were then immersed for 30 s in acid alcohol (differ-
entiation) followed by 1 min rinsing in water.
4. This was followed by 30 s in 0.1% (w/v) sodium bicarbonate
and 5 min rinsing in water.
5. The slides were finally incubated in eosin Y 0.5% (w/v) in
acidified 90% ethanol for 3 min before dehydration.
6. Following staining, sections were dehydrated in graded IMS
(90% and 100%; 30 s each) and cleared in Histo-Clear™ II
(30 s) before mounting in VECTASHIELD® HardSet™
Mounting medium and coverslipping.
3.6 Immunohisto- 1. Following dewaxing and rehydration, slides were rinsed in

chemistry PBS-T.
(Chromogenic Method) 2. Slides were then immersed in 6% hydrogen peroxide for 10 min
to block endogenous peroxidase activity (see Note 6) before
blocking in 5% goat serum in PBS for 20 min.
3. Slides were then incubated with polyclonal anti-glucagon anti-
body raised in rabbit (1:200) in PBS and 2% goat serum for
90 min at RT followed by rinsing in PBS-T.
4. Subsequently, slides were incubated with a secondary goat anti-
rabbit peroxidase conjugated antibody (1:200) with PBS and
2% goat serum for 1 h at RT.
5. After washing in PBS-T, the slides were developed using chro-
mogenic detection method by incubating the slides in
ImmPACT DAB peroxidase substrate for 3 min at RT to reveal
glucagon (see Note 7).
6. Following glucagon staining, the slides were rinsed in water for
5 min and subsequently washed in PBS-T for few seconds.
7. The slides were then incubated in 5% rabbit serum in PBS for
20 min.
8. Further to this, the slides were incubated with polyclonal
guinea pig anti-swine antibody (1:25 see Note 8) in PBS and
2% rabbit serum for 30 min at RT followed by rinsing in PBS-T.
9. Subsequently, slides were incubated with rabbit anti-guinea pig
antibody conjugated with peroxidase (1:200) in PBS and 2%
rabbit serum for 30 min at RT.
10. Insulin was detected by incubating the slides in a different
chromogen, ImmPACT SG peroxidase substrate for 4 min at
RT (see Note 7).
11. After immunostaining, slides were counterstained with Nuclear
Fast Red for 10 min followed by rinsing in water for 5 min.
12. The slides were then dehydrated, cleared, and mounted as for
H&E staining.
3.7 Image Capture 1. Whole-slide images of all stained sections were captured with
an Aperio Scanscope CS scanner.
2. Images can be captured at any magnification up to 40.
Whole-slide images were scanned at 20 and 40 original
magnification.
3.8 Image Analysis 1. Development of image analysis protocols was a six-step process
Software and an overview is presented in Fig. 1.
2. Once a protocol was created, it could be run on regions of
interest (ROIs), an entire image, or a batch of whole-slide
Fig. 1 Overview of steps involved in developing image analysis protocols. The specific components of each
step are described for individual protocols
Fig. 2 Training by example. Panel A original image, B training image (blue: background, green: pancreas [both
identified by user] see Table 2) and C final segmented image (blue outline: ROI). Original magnification 10
(A) and 4 (B and C)
images. See Note 9 for an important tip before using images for
analysis.
3. Training by Example: The system was trained to recognize
different image features such as exocrine tissue, islets, blood
vessels, insulin, and glucagon (Figs. 2a and 3b). These were
defined by the operator using a labeling tool to isolate a repre-
sentative area of the feature to be quantified. The new markup
for each object is termed as label and is represented by different
colors.
4. Defining Preprocessing Steps: The purpose of preprocessing was
to allow enhancement of structure of interest by suppressing
noise and nonrelevant image components to improve image
segmentation. This feature combines information such as spec-
tral (RGB or contrast) or morphological information. For a
standard color image, the features are described by red, green,
and blue (RGB) pixel values, where each color band has a
certain intensity at each given pixel location. If the basic input
color band does not provide sufficient robust segmentation
results across multiple images, there are other image processing
filters available to use as required by the user. For this study,
Fig. 3 Islet detection. Islets are shown in green and exocrine are shown in pink. Panel A original image,
B training by example (see Table 3), C preprocessed image with median filter, D segmented image after
Bayesian classification and E final image after postprocessing. Image C shows the outcome of applying a [9, 9]
median filter to an original RGB image. This filter helps to reduce “noise” by smoothing out bright and dark
pixels. Original magnification 40
RGB was insufficient in one case and the use of basic median
filter with a kernel size of 9 9 to RGB color band enhanced
the image by removing noise and preserving the edges. See
Note 10 for the list of all preprocessing filters available in
VisiomorphDP.
5. Defining Image Classification Method: Preprocessed images
undergo segmentation where every pixel sharing a defined set
of characteristics is segmented to allow quantification. There
are many image segmentation methods available, of which
thresholding is the most common. A threshold is defined for
a given feature and a class is assigned to all pixels with a feature
value above or equal to threshold, and remaining pixels are
assigned to another class. A Bayesian classifier can be used
instead of thresholding for image segmentation under two
main reasons. First, colors in specially stained tissues are a
mixture, for which thresholding does not allow for accurate
segmentation. Moreover, Bayesian classification does not
require the user to define an all-inclusive intensity range for
an image class as the representative training set defined will
attribute all pixels in the image to one of the defined classes. See
Note 11 for the list of all classification methods available in
VisiomorphDP.
6. Defining Postprocessing Steps: Following image segmentation,
postprocessing steps can be performed to further refine seg-
mented images based on morphological and contextual infor-
mation. Postprocessing steps allow incorporation of prior
knowledge into the segmentation process, often resulting in
better quantitative results. For example, one can fill holes,
separate objects, and smooth edges of the feature of interest.
Note that these steps create a marked-up image in order that
the user can make a visual assessment of the accuracy of seg-
mentation. By repeating the steps above, one may iteratively
create an effective algorithm. See Note 12 for the list of all
postprocessing operators.
7. Defining Output: The final step is the definition of quantifica-
tion parameters. There are a wide range of output parameters
to choose from to measure objects including number, length,
area, shape, and intensity. See Note 13 for the list of all outputs
available in VisiomorphDP.
3.9 Development 1. The four protocols developed were for (1) tissue detection;
of Image Analysis (2) islets detection; and quantification of (3) islet area and
Protocols (4) alpha and beta cell mass. These protocols were run on
micrographs of pancreas sections and each protocol was run
one after the other sequentially to get the output of interest.
2. Tissue Detection (Protocol 1): As whole-slide images were used
for analysis, non-tissue regions were excluded to decrease the
Table 2
Protocol 1 for tissue detection
Steps Chosen parameters

Magnification setup 0.5
Training (labels) Slide background: blue, tissue: green
Preprocess RGB
Classification Bayesian
Postprocess Close (tissue), fill holes (tissue with tissue), change small (tissue smaller than
6000 μm2 to background), outline as ROI (tissue to ROI), change (tissue to
clear), change (background to clear)
Output None
Table 3
Protocol 2 for islet detection

Magnification setup 20
Training (labels) Insulin (green), glucagon (blue), slide background (gray), blood (yellow), tissue
(pink), fold (light blue), dark tissue (red)
Preprocess RGB median [9,9]
Postprocess Change (background to clear), change (dark tissue to tissue), change large (blood
larger than 100 μm2 to tissue), change nearest (blood to insulin), change (blood
to tissue), change (fold to tissue), close (glucagon with glucagon, pixel 15),
change small (inulin less than 15 μm2), change nearest (glucagon close to insulin
to insulin), change (glucagon to tissue), fill hole (insulin with insulin), close
(insulin with insulin, pixel 70), change small (insulin less than 50 μm2 to tissue)
Output None
time required for analysis. Once protocol 1 was run (Table 2), a
ROI was automatically generated to delineate tissue (Fig. 2).
The resulting ROI was then analyzed using protocol 2 for islet
detection.
3. Islet Detection (Protocol 2): This protocol was developed to
identify and mask islets from the surrounding tissue (Fig. 3).
The steps involved in this protocol are detailed in Table 3.
4. Quantification of Islets (Protocol 3): Run this protocol on the
image resulting from protocol 2. The aim is get quantitative
data from the marked-up image; in this case, number of islets,
area of islets, and area of exocrine region (Fig. 4). The steps
involved in this protocol are detailed in Table 4.
Fig. 4 Islet quantification. Panels A marked-up image (final image from protocol 2), B final segmented image
with islets (green) marked as ROI (red) and C output showing the quantitative data processed from the
marked-up image. Output from this analysis includes number of islets, islet area, and exocrine area. Original
magnification 40
Table 4
Protocol 3 for islet quantification

Training (labels) None
Preprocess RGB
Classification None
Postprocess Change (ROI-1 to clear), outline as ROI (insulin to ROI-1)
Output Number of islets, exocrine area, islet area
Fig. 5 Alpha and beta cell area quantification. Panels A marked-up image (final image after protocol 3),
B preprocessed image, C image segmented after Bayesian classification and D final image after postproces-
sing showing glucagon-producing alpha cells in blue and insulin-producing beta cells in green. Original
magnification 40
5. Quantification of Alpha and Beta Cell Area (Protocol 4): This is

the final protocol to be run in order to get additional data such
as alpha and beta cell area (Fig. 5). The steps involved in this
protocol are detailed in Table 5. Once the data collection is
achieved, alpha and beta cell mass can be calculated by using
the formula below. Likewise, using islet area calculated, an islet
size distribution graph (Fig. 6) can be plotted. Using the
information derived from this plot, one can infer information
about islet neogenesis.
Alpha or beta cell area
Alpha or beta cell mass ¼ Pancreas weight
Exocrine area
Table 5
Protocol 4 for alpha and beta cell quantification

Training (labels) Insulin (green), glucagon (blue), tissue (pink)
Preprocess RGB
Postprocess Change small (insulin less than 10 μm2), change small (glucagon less than
15 μm2), close (glucagon with glucagon, pixel 10), close (insulin with
insulin, pixel 5), change small (tissue less than 200 μm2), change (tissue
to clear), change small (insulin less than 20 μm2 to clear)
Output Insulin (beta cell) area and glucagon (alpha cell area)
a 20
db rosi
db control
db exe
15
lslet Number
10
0
9
99
99
99
99
99
-9
-1
-2
-3
-4
-9
0
0
10
20
30
40
50
lslet Area (um2)

b 22.5
20.5
17.5 db rosi
lslet Number
15.0 db exe
12.5
db control
10.0
7.5
5.0
2.5
0.0
+
0 9
0 9
0 9
0 9
10 - 9 9
00 99
20 - 1 99
25 - 2 99
30 - 2 99
35 - 3 99
40 - 3 99
45 - 4 99
50 - 4 99
55 - 5 9
60 - 5 99
65 - 6 99
70 - 6 99
75 - 7 99
80 - 7 99
85 - 8 99
90 - 8 99
95 - 9 99
-9 9
10 99
0
10 - 9
20 - 19
30 - 29
40 - 39
50 - 49
00 99
00 49
00
00 4
00 9
00 4
00 9
00 4
00 9
00 4
00 4
00 9
00 4
00 9
00 4
00 9
00 4
00 9
9
15 - 1
0
lslet Area (um2)
Fig. 6 Example graphs showing distribution of islets based on its size. Such plots are useful for studies
involving drug treatments to show influence of drugs/compounds on islet neogenesis, islet size, and number
4 Notes
1. In most histopathology laboratories, 10% NBF, a

non-coagulative fixative, is historically the most popular fixative
and continues to be the first choice for many pathologists
[2]. However, the limitations of 10% NBF include its propen-
sity to mask antigenic sites, degrade certain antigens, promote
tissue shrinkage, and reduce nuclear detail [3]. Previous works
[4] have shown that performance of 10% NBF across a range of
stains indicated that it did not provide optimal fixation for all
cutaneous applications. In this study, it was shown that use of
alcoholic fixative (895 ml of 95% ethanol and 105 ml 37%
formaldehyde) instead of 10% NBF provided consistent results
across a range of applications. Thus, it is worth testing which
fixative gives better consistent quality sections, as this is an
important criterion for automated image analysis.
2. Soaking paraffin blocks in ice-cold water for few seconds is very
important for pancreas embedded in paraffin blocks. This helps
to get smooth, wrinkle- and fold-free sections.
3. Also, soaking paraffin blocks for 1–2 min in softening reagents
sometimes helps to get good-quality sections. The formulation
for softening agent is 500 ml of distilled water, 50 ml of
glycerol, 5 tablets of PBS, 2.5 ml of methanol, 2.5 ml of Triton
X-100, and 5 ml of phenol.
4. The problems associated with producing inconsistent sections
can be reduced by changing the microtome blade frequently,
cleaning the pressure plates from any wax deposition, and cool-
ing the blocks on tissue paper soaked with ice-cold water.
5. Preheating in an oven ensures consistent dewaxing, as any
residual wax greatly impairs subsequent analysis.
6. Endogenous peroxidases may react with chromogenic sub-
strates and contribute to background signals that may compro-
mise subsequent analysis. A solution of 3–6% hydrogen
peroxide is commonly used to block such endogenous perox-
idases. Also, one can dilute hydrogen peroxide in either meth-
anol, PBS, or distilled water. Methanol is a superior choice in
this case as it helps to reduce background signal from perox-
idases better than others.
7. Incubation time for DAB and SG chromogen can vary from
2 to 10 min. Ideal time has to be identified through optimiza-
tion by the user according to the experiment being performed.
8. The guinea pig anti-insulin primary antibody can also be used
at a higher dilution such as 1:200. Only use 1:25 dilution if the
expression of insulin is known to be decreased, for example, in
case of pancreas from diabetic mouse models.
9. One of the key elements in image analysis is the preparation of

good-quality sections and subsequently the quality of the
image used. Factors like fixation of specimen/biopsy, tissue
processing, sectioning, and consistency in staining are all very
crucial checkpoints to be assessed in order to get good and
accurate results after image analysis. Therefore, it is very critical
to take care of these factors as much as possible.
10. Preprocessing Steps: Some of the preprocessing filters available in
the software are color transformations (such as RGB, HIS
[intensity, hue, and saturation], chromaticity, contrast, and
color deconvolution), global operations and transformations
(such as abs, negate, not, add, subtract, multiply, divide, min,
max, square, square root, ln, exponential, and scale), local filter
operations (such as mean, median, standard deviation, modus,
median unsharp, mean unsharp, normalize by median, and
normalize by mean) and polynomial filters (such as gradient,
orientation, smoothing, laplace, linear structures, and blob
structures). For example, one can use a polynomial local linear
filter to enhance the linear structures of interest or use a blob
filter to enhance object that has circular shape. Mean and
median filters are very useful to reduce “noise” by smoothing
out bright and dark pixels. Color deconvolution can be used if
the staining contains only hematoxylin and DAB. This hema-
toxylin–DAB deconvolution allows the user to work in “true
stain space” and produces two images of DAB staining and
hematoxylin staining.
11. Classification Methods: Some of the training by example tools
include freehand painting, region growing, ROI, and label
tools. The basic classification tool available is thresholding
and the advanced tools are K-means clustering, fuzzy
K-means clustering, linear Bayesian classification, quadratic
Bayesian classification, and automatic background detection.
12. Postprocessing Steps (Refinement): Some of the postprocessing
categories include change object type (based on object’s area,
circularity, context [neighboring objects]), special operations
(such as enumerate objects), morphological operations (such
as fill holes, skeletonize, separate objects, close, conditional
erode, and dilate), and ROI operations (such as outline objects
as ROIs, create objects from ROIs). For example, close opera-
tion can be used to close holes in the object label with a chosen
label and fill holes are used to fill the holes in the object label
using the fill label. Some of the “change” steps include change
(unconditional change of all label objects of one type (label) to
another type (replace with), change small (change based on
object size), and change nearest (label objects of chosen type
(label) are changed to replace with type if they are sufficiently
close to a third label type (close to)).
13. Quantitative Measurement Options (Output): The outputs are

usually in single values or as histograms. While based on seg-
mented image one can quantify object area, object count,
perimeter, and interface length, based on original image or
feature min/max intensity, mean intensity, standard deviation,
median, modus, and entropy can be quantified.
References
1. Gamble M (2008) The hematoxylins and eosin. other molecules from fixed and processed tis-
In: Theory and practice of histological techni- sues, developmental methods of fixation. J His-
ques. Churchill Livingstone Elsevier, London, totechnol 24:201–210
pp 121–134 4. Al-Habian A∗, Harikumar PE∗, Stocker CJ,
2. Grizzle WE (2009) Special symposium: fixation Langlands K, Selway JL (2014) Histochemical
and tissue processing models. Biotech Histo- and immunohistochemical evaluation of mouse
chem 84:185–193. https://doi.org/10.3109/ skin histology: comparison of fixation with neu-
10520290903039052 tral buffered formalin and alcoholic formalin. J
3. Eltoum I, Fredenburgh J, Grizzle WE (2001) Histotechnol 37:115–124. https://doi.org/10.
Advanced concepts in fixation: effects of fixation 1179/2046023614Y.0000000050. (∗ equal
on immunohistochemistry, reversibility of fixa- contribution)
tion and recovery of proteins, nucleic acid, and
Chapter 13
Measurement of Calcium Signaling in Beta-Cell Lines Using

Epifluorescence and Confocal Microscopy
Joanne L. Selway
Abstract
One of the main consequences of glucose action on pancreatic β-cells is the stimulation of Ca2+ entry as well
as Ca2+ release from intracellular compartments. Therefore, one of the cornerstones of any diabetes research
laboratory should be the ability to measure changes in intracellular calcium concentrations within pancre-
atic β-cells. There are a variety of methods available for the measurement of intracellular calcium, from
multiple regions of interest (ROIs) within single cells to observe oscillating calcium, to population-based
96-well applications to enable high-throughput screening of the effects of novel agonists. These methods
allow calcium signaling to be observed in a single cellular assay to look at oscillations at a cellular level, to
view a population response, and to enable high-throughput assays where the mean reflects a single cell
response.
Key words Calcium, MIN6, Fluorescent microscopy, Beta-cells
1 Introduction
One of the main consequences of glucose action on pancreatic

β-cells is the stimulation of Ca2+ entry and Ca2+ release from
intracellular compartments. This stimulates an array of different
molecular and physiological consequences. One of the major and
most important physiological consequences is the stimulation of
insulin secretion to ensure a feedback and lowering of peripheral
glucose concentrations after a meal. Glucose metabolism and
subsequent membrane depolarization of the pancreatic β-cell
leads to the activation of voltage-gated calcium channels
(VGCCs) and this is believed to mediate Ca2+ release from ER
stores via calcium-induced calcium release (CICR). This process
initiates the fusion of insulin-containing vesicles, already primed at
the plasma membrane, to be fused and insulin released into the
bloodstream [1]. However, as an important second messenger in
intracellular signaling, Ca2+ also initiates other signaling cascades.
The rise in intracellular Ca2+ concentration ([Ca2+]i) can activate
231
232 Joanne L. Selway
additional calcium-sensitive proteins, notably calmodulin, which

can initiate changes in gene transcription through the CREB and
MAPK pathways [2–4].
Rises in the [Ca2+]i in beta-cells can also be used as a pseudo-
marker of any agonist that stimulates insulin secretion or intracel-
lular signaling in beta-cells that permits insulin release, such as
secretagogues. In general, high-throughput screening for new
compounds involved in the treatment of diabetes, for example,
beta-cell proliferation or novel strategies to prevent diabetes-
induced pancreatic decline, relies on the activation of transcription
factors or growth arrest as a primary readout, but no compound is
likely to be validated and taken further in a pharmaceutical devel-
opment pipeline if it is unable to mediate an improvement in Ca2+-
mediated insulin secretion [5, 6].
Strategies to measure [Ca2+]i changes in pancreatic beta-cells
range from simple fluorescence measurement of one-wavelength
with a relatively inexpensive system [4], to fluorescence resonance
energy transfer imaging (FRET)-based imaging that can detect
subtle changes in the calcium stores of the cells that requires more
specialist microscopy equipment [7].
2 Materials
Krebs Ringer bicarbonate buffer (KRB): 115 mM NaCl, 5 mM

KCl, 10 mM NaHCO3, 2.5 mM MgCl2, 2.5 mM CaCl2, 20 mM
HEPES pH 7.4.
Dye loading solution: 1 mg/ml BSA, 1.87 μl/ml 20% pluronic
acid (see Note 1), 2 μM fluorescent dye (Fura-2 AM or Fluo-4-
AM), Krebs Ringer buffer, and D1ER-FRET-GFP vector cameleon
kindly provided by Professor R Tsien, Department of Pharmacol-
ogy and Howard Hughes Medical Institute, University of Califor-
nia [8]. This construct is not commercially available, but readers
may contact the original laboratory.
3 Methods
3.1 Cell Culture 1. MIN6 cells, a cell line derived from murine pancreatic β-cells,
should be used at approximately 80% confluence between pas-
sages 16 and 40.
2. MIN6 cells are grown in DMEM containing 25 mM glucose
supplemented with 15% heat-inactivated FCS, 100 μg/ml
streptomycin, 100 units/ml neomycin 100 units/ml penicillin
sulfate, 40 mM NaHCO3, and 75 μM β-mercaptoethanol,
equilibrated with 5% CO2, 95% air at 37 C.
3. The medium was changed every 2–3 days.
Measurement of Calcium Signaling in Beta-Cell Lines Using Epifluorescence. . . 233
4. When ~80% confluence was reached, cells were washed once in

1 PBS and then placed in 1 ml 0.025% trypsin-EDTA for
2–5 min at 37 C. Cells are resuspended in DMEM as soon as
they start to detach from the plate.
5. MIN6 cells are split 1:3 to 1:4 for maintenance, or as required
for experiments (see Notes 2–4).
3.2 Ca2+ Imaging 1. MIN6 cells are plated on glass coverslips (size dependent on
Techniques microscopy equipment but typically 25 mm) for approximately
72 h to allow adhesion prior to experimentation (see Note 5).
3.2.1 Epifluorescence
Intracellular Ca2+ Imaging 2. Initially, cells should be washed three times in KRB to remove
DMEM components from the cells by gentle removal of media
and then addition and removal of KRB three times to cover the
coverslip (see Note 6).
3. MIN6 cells are then loaded with 2 μM Fura-2-AM by covering
the coverslip in dye loading buffer for 1 h at 37 C. In addition
to enabling the fluorescent dye to enter the cells, it starves the
cells of glucose and reduce spontaneous oscillations of Ca2+.
4. After an hour, wash the cells three times with KRB supplemen-
ted with 2 mM glucose by gentle removal of loading solution
and then addition and removal of KRB three times, sufficient to
cover the coverslip (see Notes 6 and 7).
5. Coverslips can then be mounted on the stage of an inverted
microscope (Nikon Diaphot inverted epifluorescence micro-
scope) and left in 200 μl KRB buffer (or a volume sufficient
to cover the surface of the coverslip when mounted in a cover-
slip bath) at room temperature (see Notes 6 and 7).
6. To measure [Ca2+]i, the cells must be excited at 340 and
380 nm at 1 s intervals and emissions collected at wavelengths
above 520 nm (see Note 8).
7. A period of recording should occur before agonist stimulation
(at least 30 s) to ensure that the cells have a low basal activity
(see Note 9).
8. The agonist of choice (such as glucose) can now be pipetted
onto the coverslip in a concentrated form to ensure the final
concentration is the required dose. Typically, agonists are
applied at 1.4 times the required concentration to the bath
directly above the coverslip (see Notes 6, 7 and 10).
9. Plot the data by looking at the ratio of 340:380 signal as a
pseudo measure of [Ca2+]i. The baseline readings should be
stable, with a value of 1.
3.3 Single-Cell 1. MIN6 cells should be spilt 1:3 onto 25-mm diameter glass
Confocal Intracellular coverslips and left for a minimum of 72 h before experimenta-
Ca2+ Imaging tion (see Note 5).
2. Before imaging, the coverslips should be washed three times

with KRB and left for 30 min at room temperature to ensure
Ca2+ signaling is at baseline levels (see Note 9).
3. After 30 min, the KRB should be replaced, via a pipette, with a
loading solution containing 2 μM Fluo-4 for further 30 min to
load the fluorescence dye (see Note 11).
4. The cells should be then washed three times with KRB supple-
mented with 2 mM glucose by gentle removal of loading
solution and then addition and removal of KRB, sufficient to
cover the coverslip three times (see Notes 6 and 10).
5. A PERKinElmer UltraVIEW confocal microscope, or equiva-
lent model, is used to measure the [Ca2+]i. Cells are excited
using the 488 nm laser line and 485 nm excitation filter, and
the emitted fluorescence captured at wavelengths >520 nm,
with images collected at approximately 2-s intervals.
6. A period of recording should occur before agonist stimulation
(at least 30 s) to ensure that the cells have a low basal activity
(see Note 10).
7. An agonist, of choice (such as glucose), can now be pipetted
onto the coverslip in a concentrated form to ensure the final
concentration is the required dose. Typically, agonists were
applied at 1.4 the required concentration directly above the
coverslip (see Note 11).
8. The data is plotted by looking at the 520 emission value
divided by the baseline emission prior to stimulation. This
ratio provides a pseudo measure of [Ca2+]i.
3.4 Population Based 1. MIN6 cells should be passaged 1:2 into 96-well plates (200 μl
or High-Throughput total volume) and left to culture for 48 h.
Intracellular Ca2+ 2. Wells should be washed three times with KRB supplemented
Imaging with 2 mM glucose by gentle removal of the media and then
addition and removal of KRB, sufficient to cover the well, three
times (see Notes 6 and 10) and then placed at room
temperature.
3. After 30 min, the KRB is replaced with dye loading solution
containing 2 μM Fluo-4.
4. After an additional 30 min, wash the wells three times with
KRB supplemented with 2 mM glucose by gentle removal of
the media and then addition and removal of KRB, sufficient to
cover the well, three times (see Notes 6 and 10).
5. Cells can then be left with 25% of the total volume of KRB as a
starting point (typically this is 200 μl).
6. At least one well of cells that has not been loaded with Fluo-4-
AM should be included in all experiments as an autofluores-
cence control for the data analysis.
7. For stimulation, solutions should be 1.17 higher than
required with 75% of the total volume added from the solution
plate per well. All measurements must be made at room
temperature.
8. The reader required for the population measurement should
measure fluorescence at 520 nm every second during the exper-
iment (see Note 12).
3.5 Fluorescence 1. The Lipofectamine 2000 transfection procedure (Invitrogen,

Resonance Energy Life Technologies, UK) should be utilized as described below.
Transfer Imaging 2. 4 μg D1ER DNA, purified by Qiagen maxi-prep kits, should be
(FRET)-Based added to 225 μl Opti-MEM reduced serum media (Gibco,
Intracellular Ca2+ UK).
Store Imaging 3. Simultaneously, 10 μl of Lipofectamine 2000 is added to 225 μl
3.5.1 Lipofectamine Opti-MEM low serum media. These solutions are incubated
Transfection of MIN6 Cells for 5 min at room temperature and then are combined by
adding the DNA to the Lipofectamine solution (see Notes
2, 3, and 13).
4. The solution is mixed gently by flicking, and then left at room
temperature for 20 min.
5. The media should be changed on the cells to Opti-MEM and
the complexes added to the cells by slowly dripping them from
a 1 ml pipette over a min.
6. The complexes should be left on the cells overnight and then
the media should be changed to complete media and the
cultures left for further 24 h before experimentation (see
Note 14).
3.5.2 FRET-Based 1. MIN6 cells need to be starved of glucose and serum for at least
Imaging 1 h in KRB prior to treatment.
2. All experiments must be carried out at room temperature.
3. Images are captured with the 20 objective of an epifluores-
cence microscope with a CCD camera.
4. The emission ratio imaging of the cameleon is accomplished by
using a 436DF20 excitation filter, 450-nm dichroic mirror, and
two emission filters (475/40 for enhanced CFP and 535/25
for citrine) controlled by a Lambda 10–2 filter changer.
5. Exposure times are typically 100–1000 ms and images can be
collected every 8–20 s for the most optimal results.
6. A period of recording should be taken before agonist stimula-

tion (at least 30 s) to ensure that the cells have a low basal
activity (see Note 9).
7. An agonist of choice (such as glucose) can be pipetted onto the
coverslip in a concentrated form to ensure the final concentra-
tion is the required dose. Typically, agonists were applied at
1.4 the required concentration directly above the coverslip
(see Notes 6 and 10).
8. Data is plotted by looking at the ratio of emission values which
acts as a pseudo measure of [Ca2+]ER.
4 Notes
1. Make the 20% pluronic acid mix 24 h in advance if possible; if

not, warm the mixture to dissolve as it will form an emulsion
initially which can be reversed with heat. If you do not warm
this solution, it may ultimately reduce the concentration of acid
in the loading dye leading to reduced amount of dye being
loaded in the cells.
2. Clumping of MIN6 pancreatic beta-cell lines is vital for the
survival of the cells and individual cells tend to apoptose.
However, clumping of pancreatic beta-cells in the abovemen-
tioned assays can lead to low transfection rates and problems
with imaging.
(a) Lipofectamine-mediated transfection of DNA plasmids
requires equal access to the membranes of the cells to
deposit the coated plasmids into the cells evenly. Clump-
ing of cells prevents the even distribution of plasmid,
reducing the transfection efficiency in these cell lines
further.
(b) Clumped cells in the z-plane leads to problems in imaging
as signal from other z-positions can sometimes be seen,
often known as “bleed through,” and thus can interfere
with individual cell signals increasing noise. One solution
to this is to minimize the laser intensity but this can lead to
a compromise on the observation of true signal.
3. To mitigate both of these problems, a balance in plating of the
cells is required to ensure cells are only in 2–3 groupings. This
can be achieved through the optimization of several different
parameters:
(a) Trypsinization time when plating—regular use of MIN6
cell batches will allow empirical determination of the opti-
mal digestion time required. The optimum time to resus-
pend the cells in DMEM to quench digestion is when the
majority of cells are slowly sliding down the flask or plate

when tilted.
(b) Thorough resuspension in DMEM posttrypsinization
allows clumps of cells to be disaggregated appropriately.
As a rule of thumb, if the above step is optimal, pipetting
up and down through a 10-ml pipette on low speed
10 times will be sufficient. However, the aliquoting of
10 μl of cells and observation under a microscope will
demonstrate whether the clumps have been reduced to
double or triple cell clumps as required for
experimentation.
4. MIN6 cells were used between passages 16 and 40, as higher
than this the cells are not as responsive to glucose, meaning that
the calcium signals generated by glucose signaling become
highly variable.
5. Coating the plates and/or coverslips with poly-D-lysine is an
option for retaining cells on the coverslips and the bottom of
plates but the coating can interfere with imaging as it has a
natural fluorescent property. This interference can be
accounted for in the analysis but can increase the noise due to
inconsistent coating.
6. MIN6 cells (and other pancreatic beta-cell lines such as rat
INS1e cells) can lift from the surface of the plates or coverslips
if not treated gently. Gentle treatment involves:
(a) the use of a bath, rather than perfusion of agonist or
stimulant.
(b) the addition of small volumes slowly onto coverslips.
7. Perfusion rather than bath addition can be used for MIN6 cells
in short-scale (<5 min) experiments but the flow rate needs to
be minimal with the realization that many of the cells being
observed (as well as the individual data points within cells) may
be lost.
8. Any fluorescent microscopy with the correct lasers and filters
can be utilized for these experiments but here a SpectraMAS-
TER II monochromator has been used (PerkinElmer Life
Sciences).
9. A period of recording should occur prior to the addition of any
agonist to evoke a Ca2+ response. The purpose of this recording
is to ensure a stable baseline for recording and as a quality
control measure to ensure that the quality of the recoding is
satisfactory. If the baseline is not linear or if there is a significant
level of activity without agonist, the recording should be
rejected. The most common reason for activity is incorrect
buffer preparation or flow causing a physical movement of the
cells resulting in a Ca2+ response.
Table 1
Plate reader parameters to measure fluorescence at 520 nm
General settings Concentrations/volumes/shaking

Positioning delay 0.2 s Volume
No. of kinetic windows 2 Start volume 0
Kinetic window 1 Factor 1
Measurement start time 0.00 Shaking options
No. of intervals 5 Mode Orbital
No. of flashes per well and interval 10 Shaking width 1 mm
Interval time 1.00 s Additional shaking No shaking
End of kinetic window 1 5s Pump
10. Bath addition of the agonist of choice is the preferred method

of addition. When the coverslip is in a bath on the microscope
stage, it should be covered with 200 μl of KRB. The agonist of
choice should then be added at 1.4X.
11. All Ca2+ imaging experiments must be carried out at room
temperature as incubation at 37 C leads to the increased
leakage of the fluorescent dye during longer (>20 min)
experiments.
12. A plate reader must be set up with the basic parameters given in
Table 1 to measure fluorescence at 520 nm.
13. The transfection efficiency of the MIN6 pancreatic beta-cell
line is typically low ranging from 20 to 40%, so population
FRET imaging (or any imaging requiring a transfection of a
plasmid construct without a fluorescent tag) is not advised.
14. After Lipofectamine transfection, cells should be observed at
the 8–12 h and 24-h mark to check for cell death. A small
proportion of cells may undergo cell death as evidenced by a
lack of attachment to the wells or coverslips and a spherical
shape but this should not exceed 10% of the treated cells.
References
1. Mears D (2004) Regulation of insulin secretion channels mediate Raf-independent activation of

in islets of Langerhans by Ca(2+)channels. J extracellular regulated kinase in response to
Membr Biol 200(2):57–66 glucagon-like peptide-1 in pancreatic beta-cells.
2. Arnette D et al (2003) Regulation of ERK1 and J Biol Chem 277(50):48146–48151
ERK2 by glucose and peptide hormones in pan- 4. Selway J et al (2012) Evidence that Ca2+ within
creatic beta cells. J Biol Chem 278 the microdomain of the L-type voltage gated
(35):32517–32525 Ca2+ channel activates ERK in MIN6 cells in
3. Gomez E, Pritchard C, Herbert TP (2002) response to glucagon-like peptide-1. PLoS One
cAMP-dependent protein kinase and Ca2+ 7(3):e33004
influx through L-type voltage-gated calcium
5. Hill JA et al (2010) A multi-parameter, high- 7. Moore CE et al (2011) PERK activation at low

content, high-throughput screening platform glucose concentration is mediated by SERCA
to identify natural compounds that modulate pump inhibition and confers preemptive cyto-
insulin and Pdx1 expression. PLoS One 5(9): protection to pancreatic beta-cells. Mol Endo-
e12958 crinol 25(2):315–326
6. Wang W et al (2009) Identification of small- 8. Palmer AE et al (2004) Bcl-2-mediated altera-
molecule inducers of pancreatic beta-cell expan- tions in endoplasmic reticulum Ca2+ analyzed
sion. Proc Natl Acad Sci U S A 106 with an improved genetically encoded fluores-
(5):1427–1432 cent sensor. Proc Natl Acad Sci U S A 101
(50):17404–17409
Chapter 14
Nitric Oxide and Redox State Measurements in Pancreatic

Beta Cells
Rodrigo Carlessi, Vinicius Cruzat, Younan Chen, and Philip Newsholme
Abstract
The role of oxidative stress in the pathogenesis of type 2 diabetes (T2D), especially pancreatic β-cell
dysfunction and death, has become apparent in the last two decades. Peroxidase- and catalase-based
antioxidant mechanisms are particularly weak in β-cells and can be easily overwhelmed by excessive
production of reactive oxygen and nitrogen species in the course of pathological processes. Recent research
has attempted to define in detail the mechanistic aspects of oxidative stress-induced β-cell dysfunction.
Here, we describe the procedures for the measurement of various parameters important to assess oxidative
stress in pancreatic β-cells. Detailed protocols for determination of nitric oxide (NO) production, the
glutathione redox status, and general oxidative status in β-cells are presented in this chapter.
Key words Nitric oxide, Glutathione, Oxidative stress, β-cells, Type 2 diabetes mellitus (T2DM)
1 Introduction
Reactive oxygen species (ROS) and reactive nitrogen species (RNS)

are produced during a number of physiological and pathological
processes. The excess production of ROS and RNS, or the inability
to effectively scavenge such species, induces oxidative stress in cells.
Oxidative stress is involved with various pathological conditions,
such as aging, diabetes, inflammation, carcinogenesis, and athero-
sclerosis [1]. In pancreatic β-cells, glucose catabolism to CO2 and
H2O is essential to generate the enormous amount of energy
required to produce and secrete insulin according to demands of
growth and energy storage in target tissues. Glucose-stimulated
insulin secretion (GSIS) requires glucose metabolism and oxidation
in β-cells. Indeed β-cells must satisfy their own energy demands
without hampering their ability to drive and regulate insulin exocy-
tosis. Intense oxidative metabolism, combined with the fact that
β-cells are associated with relatively low expression levels of some
key antioxidant enzymes [2, 3], make β-cells highly susceptible to
ROS/RNS-induced damage. In conditions of excessive nutrient
241
242 Rodrigo Carlessi et al.
overload (normally associated with type 2 diabetes) the resulting

excessive production of ROS and RNS can damage beta cells,
leading to loss of insulin secretion and even cell death [4]. In type
1 diabetes (T1D), proinflammatory cytokine-induced β-cell death
is known to be partially mediated through excessive ROS/RNS
generation [5]. Thus, measurements of reactive species as well as
global oxidative stress in β-cells are important to the field of diabe-
tes research. In this chapter, we describe in detail the procedures
involved in the measurement of nitric oxide (NO), the glutathione
redox state, and general oxidative status in β-cells.
1.1 Measurement NO is a free radical that, under physiological conditions, is rapidly

of Nitric Oxide Free oxidized to nitrite (NO2) and nitrate (NO3). The assay described
Radical (NO) Release in this chapter is a modified version of a diazotization reaction that
was originally described by Peter Griess [6]. It consists of a simple
and well-characterized colorimetric assay for nitrites, with a detec-
tion limit around 0.1 μM, adequate to detect nitrites released by
beta cells in standard tissue culture conditions.
1.2 Glutathione Reduced glutathione (GSH), a tripeptide

Redox State (g-glutamylcysteinylglycine), is the main free thiol in most living
Measurement cells and is the most important antioxidant in animal cells. Oxida-
tion of glutathione leads to the formation of glutathione disulfide
(GSSG), a process that occurs when GSH reacts with free radicals,
protecting cells against oxidative stress. Thus, the ratio
GSH/GSSG within cells is often used as an index of intracellular
redox status. Inflammation and glucolipotoxicity have been shown
to reduce this ratio in β-cells [7, 8]. The protocol detailed here, in
order to determine GSH and GSSG, was described by Kolberg et al.
[9] and Cruzat et al. [10] adapted from Akerboom, Sies [11].
As described in (Fig. 1), in the presence of GSSG reductase and
NADPH, the GSSG resulting from the first reaction (or initially
present in the sample), is reduced to GSH, which is then converted
back to GSSG and 5-thio-2-nitrobenzoic acid (TNB). As the
amount of GSSG reductase is constant and the substrates
NADPH and 5,50 -Dithiobis (2-nitrobenzoic acid) (DTNB) are
added at saturating concentrations, the velocity of TNB formation
is proportional to the initial amount of GSH + GSSH in the sample.
The formation of TNB can be assessed using spectrophotometric
analysis at 412 nm and gives as estimation of the total amount of
glutathione (reduced and oxidized) present in the sample. For the
determination of the initial amount of GSSG in the sample, it is
necessary to eliminate all GSH before the commencement of the
reaction; this is achieved by conjugation with N-Ethylmaleimide
(NEM). The procedure detailed here describes all the steps neces-
sary to determine total GSH and GSSG levels in β-cell lysates. The
amounts of GSH can be calculated by subtracting GSSG values
from total glutathione levels.
Nitric Oxide and Redox State Measurements in Pancreatic Beta Cells 243
2 GSH + DTNB GSSG + 2 TNB
GSSG + H+ + NADPH 2 GSH + NADP+

GSSG Reductase
Fig. 1 Coupled reactions in the DTNB-GSSG reductase recycling assay
1.3 General Oxidative As an alternative to determination of specific ROS or RNE species,

Status general oxidative status can be estimated in living cells. The cell-
permeant 20 , 70 -dichlorodihydrofluorescein diacetate
(H2DCFDA), also known as dichlorofluorescein diacetate, is com-
monly used to detect the generation of free radicals in cells. Upon
cleavage of the acetate groups by intracellular esterases and
subsequent oxidation by ROS and/or RNE, the non-fluorescent
H2DCFDA is converted to the highly fluorescent 20 , 70 - dichloro-
fluorescein (DCF), which can be detected by fluorescence- based
methodologies, such as fluorescence spectroscopy, flow cytometry,
and confocal microscopy.
In this chapter, we describe in detail a method to estimate total
intracellular ROS levels by flow cytometry, using a probe of 5- (and
6-) chloromethyl-20 ,70 -dichlorodihydrofluorescein diacetate, acetyl
ester (CM-H2DCFDA) (Fig. 2). CM-H2DCFDA is a chloro-
methyl derivative of H2DCFDA that passively diffuses into cells,
where its acetate groups are cleaved by intracellular esterases, and its
thiol-reactive chloromethyl group reacts with intracellular glutathi-
one and other thiols. Subsequent oxidation yields a fluorescent
adduct that is trapped inside the cell, thus facilitating detection.
2 Materials
All chemicals can be purchased from Sigma-Aldrich (St. Louis,

MO, USA), unless otherwise stated.
2.1 Measurement 1. 0.1% N-(1-Naphthyl)ethylenediamine dihydrochloride in

of Nitric Oxide Free water (see Note 1).
Radical (NO) Release 2. 1% Sulfanilamide in 5% phosphoric acid (see Note 2).
3. 100 mM sodium nitrite in water (see Note 3).
4. Clear 96-well flat-bottom plate.
5. Plate reader.
6. 1.5 mL disposable polypropylene tubes.
7. Pipettes for ranges 20–200 μL and 200–1000 μL.
8. Disposable 200 and 1000 μL tips.
9. Vortex mixer.
Fig. 2 5-(and 6-) chloromethyl-20 ,70 -dichlorodihydrofluorescein diacetate, acetyl

ester (CM-H2DCFDA, C6827, Life Technology)
10. Centrifuge.
11. Ultrapure or distilled water (see Note 4).
2.2 Glutathione 1. 0.2 M N-Ethylmaleimide (NEM) in 100% ethanol (740.6 mg/

Redox State 30 mL ethanol).
Measurement 2. 0.3 M Pipes dissolved in 2 M potassium hydroxide (KOH)
(9.07 g/100 mL of 2 M KOH).
3. Ethyl acetate.
4. 5% Meta-phosphoric acid (MPA) (w/v) in water.
5. 1 mg/mL Glutathione reduced (GSH) and oxidized (GSSG)
in water.
6. Assay buffer—143 mM Phosphate buffer, pH 7.5 (see Note 7),
containing 6.3 mM EDTA (234.6 mg of ethylenediaminete-
traacetic acid (EDTA) in 100 mL of phosphate buffer).
7. 14.1 mM DTNB (5.59 mg of 5,50 -dithiobis-2-nitrobenzoic
acid (DTNB) in 1 mL of assay buffer).
8. 1.7 mM NADPH (1.42 mg of NADPH in 1 mL of sodium
bicarbonate (NaHCO3) at 5% (w/v) in water).
9. Glutathione Reductase solution—calculate the amount of
enzyme solution necessary for the assay (10 μL/sample).
Resuspend Glutathione Reductase enzyme suspension (Sigma
G3664) to 5 U/mL, and then dilute 1:90 in assay buffer.
10. Clear 96-well plates.
11. Polypropylene microcentrifuge tubes.
12. Microcentrifuge.
13. Vortex.
14. Speed dryer vac.
15. Plate reader.
16. Pipettes for 20–200 μL and 200–1000 μL.
17. Multichannel pipette for 20–200 μL.
2.3 General Oxidative 1. 5-(and-6)-chloromethyl-20 ,70 -dichlorodihydrofluorescein dia-

Activity cetate, acetyl ester (CM-H2DCFDA), mixed isomers, 50 μg/
vial (Life Technologies, NY, USA—C6827).
2. Glucose Oxidase from Aspergillus niger (Sigma G7141) (resus-
pend to 100 U/mL with 0.05 M Sodium Acetate, pH 5.1).
3. Dulbecco’s Modified Eagle Medium (DMEM): Phenol red
free, and fetal bovine serum (FBS) free.
4. Roswell Park Memorial Institute (RPMI) 1640 media: with
10% fetal bovine serum (FBS).
6. Trypsin: 0.25% (w/v) in PBS.
7. Dimethyl sulfoxide (DMSO).
8. 6-well cell culture plate.
9. 1.5 mL and 15 mL disposable polypropylene tubes.
10. Flow cytometer.
3 Methods
3.1 Measurement 1. For each assay series, a nitrite standard curve must be prepared
of Nitric Oxide Free for accurate quantitation of NO2 levels in experimental sam-
Radical (NO) Release ples. Prepare 5 mL of a 100 μM nitrite solution by diluting the
100 mM nitrite stock solution 1 in 1000 in beta cell culture
media (such as RPMI, see Note 5).
2. Label seven 1.5 mL disposable polypropylene tubes with the
following concentrations: 100, 50, 25, 12.5, 6.25, 3.125 and
1.56 μM. Pipette 200 μL of tissue culture media to all tubes,
labeled from 50 μM downwards, leaving the tube labeled
100 μM empty. Add 400 μL of 100 μM nitrite solution to the
tube labeled 100 μM.
3. Perform a serial twofold dilution by pipetting 200 μL from the
100 μM tube into the 50 μM tube. Vortex to mix, then proceed
down the serial dilution as shown in (Fig. 3).
4. Designate columns A, B, and C of the 96-well plate for the
nitrite standard curve. Pipette 50 μL of each nitrite standard
into the wells in rows A–G, starting at A with μM 100 and
down to 1.56 μM at G. Into the row H pipette tissue culture
media only, which will be the blank.
5. Add 50 μL of each experimental sample to wells in triplicate (see
Note 6).
6. Allow the Sulfanilamide Solution and NED Solution to equili-
brate to room temperature.
Fig. 3 Suggested pipetting strategy to prepare the nitrite standard curve
Fig. 4 Representative nitrite standard curve. Trend line and equation are
calculated through linear regression analysis using Microsoft Excel Professional
Plus 2013
7. Pipette 50 μL of the Sulfanilamide Solution to all wells contain-

ing nitrite standards and experimental samples.
8. Incubate for 5 min at room temperature, protected from the
light.
9. Pipette 50 μL of the NED Solution to all wells.
10. Incubate at room temperature for 10 min, protected from the
light. The colors start to develop immediately.
11. Measure absorbance within 30 min using a plate reader with a
filter between 530–550 nm.
12. Subtract the average absorbance value of the blank wells from
all standards and experimental samples wells.
13. Make a standard curve by plotting concentration of the stan-
dards on the X-axis, and average absorbance of the triplicates
measurements on the Y-axis. A representative standard curve is
shown in (Fig. 4).
14. Determine the average absorbance value of each experimental
sample and estimate its concentration by comparison to the
nitrite standard curve.
3.2 Glutathione 1. After treatment, washβ-cells with ice-cold PBS, and then resus-
Redox State pend cell pellet in 5% MPA (150–200 μL of meta-phosphoric
Measurement acid (MPA) for about 107–108 cells population or 50–100 μL
of MPA for 106–107 cells) in order to break the cell mem-
branes. Gently homogenize with a pipette, then centrifuge the
samples 15,000 g for 5 min at 4 C and the supernatants
(from now on referred to as ‘samples’) are collected and stored
for subsequent analysis (see Note 8).
2. Samples designated to measurement of initial GSSG should be
conjugated with NEM to eliminate all GSH before commence-
ment of the reaction (steps 3–6). Samples designated to total
glutathione assay go directly to step 7.
3. Using polypropylene microcentrifuge tubes (1.5 mL), sam-
ples are mixed with NEM and Pipes/KOH (see Note 9).
4. Subsequently, samples are centrifuged at max speed for 3 min
and supernatants discarded.
5. Pellets are resuspended in 500 μL of ethyl acetate, centrifuged
at max speed for 3 min and supernatants discarded. This step is
repeated two times more (see Note 10).
6. Following the last wash with ethyl acetate, supernatants should
be carefully removed with pipette and dried in a SpeedVac
machine at room temperature. Finally, samples are resuspended
in 50 μL of 5% MPA for determination of GSSG.
7. For the determination of GSH and GSSG, the following
components need to be prepared fresh just before the assay:
Assay buffer, 14.1 mM DTNB, 1.7 mM NADPH and
Glutathione Reductase solution (for details see items 6–9 of
Subheading 2.2).
8. A standard curve for GSH and GSSG must be prepared accord-
ing to Table 1. To start, dilute the stock of 1 mg/mL GSH 1:4
in 5% MPA to get 813.5 μM and 1 mg/mL GSSG 1:2 in 5%
MPA to get 816.0 μM.
9. GSH and GSSG assay running in a polystyrene 96-well plate for
a final volume of 100 μL. Reagents must be added to the wells
of the 96-well plate in the following order (see Note 11).
(a) 10 μL sample in 5% MPA (cold).
(b) 70 μL of DTNB (37 C).
(c) Mix gently (5 s).
(d) 10 μL NADPH at room temperature.
(e) Mix gently (5 s) and incubate at 37 C for 2–3 min.
(f) Pipette 10 μL of GSSG reductase (it is highly recom-
mended to use a multichannel pipette for fast distribution
of GSSG reductase across all wells).
Table 1
Series of dilutions for GSH and GSSG standard curves
Dilution GSH (μM) GSH (μL) MPA (μL) GSSG (μM) GSSG (μL) MPA (μL)
75 μL + 813.5 – 0 816.0 – 0
0
25 μL MPA 3/4 1:1 610.1 75 25 612.0 75 25

406.8 50 50 408.0 50 50
1/2
1:1 1/4 203.4 50 50 204.0 50 50
1:1 101.7 50 50 102.0 50 50

1/8
1:1 1/16 50.8 50 50 51.0 50 50

! ZERO ¼ MPA 5% 0 0 – 0 0 0
GSH
250
200
Δ Abs/min
150
y = (0.2269)x + (26.675)
100 R² = 0.99962
50
0
0 200 400 600 800 1000
Concentration (mM)
Fig. 5 Representative GSH standard curve. Trend line and equation are calcu-
lated through linear regression analysis using Microsoft Excel Professional Plus
2013
(g) Mix (10 s) and incubate at 37 C for no more than

2–3 min. Read the absorbance at 415 nm in the plate
reader.
10. Calculations: Make a standard curve by plotting concentration
of the standards on the X-axis, and average absorbance of the
replicate measurements on the Y-axis. A representative stan-
dard curve is shown in (Fig. 5).
11. Using the slope (m) generated by the linear portion of the
standard curve and the formula y ¼ mx + b. Determine average
absorbance value of each experimental sample and estimate its
concentration by comparison to the standard curve. Impor-
tantly, the above calculation will generate the total amount
of GSH and GSSG; however, for final concentrations of
reduced GSH, the following calculation is required.
[GSH] ¼ GSHTOTAL–2 [GSSG].
3.3 General Oxidative 1. Seed the β-cells in 6-well plates, 4 105 cells/well, with 1 mL
Activity RPMI 1640 (10% FBS). Incubate the plate at 37 C overnight
(see Note 12).
2. Incubate the cells in the desired experimental conditions.
3. Reconstitute one vial of CM-H2DCFDA (50 μg) as follows:
dissolve the lyophilized CM-H2DCFDA with appropriate
amount of DMSO to the final concentration of 200 μM
(add 432.7 μL of DMSO to one vial of CM-H2DCFDA).
Keep it protected from light and on ice until ready to use (see
Note 13).
4. Prepare the working solution of CM-H2DCFDA: dilute the
stock CM-H2DCFDA 1:100 (final concentration 2 μM) into
phenol red and FBS free DMEM. The total volume of
CM-H2DCFDA working solution required is 1 mL per well
of 6-well plate. Keep working on ice and protected from light
(see Note 14).
5. Prepare one 15 mL and one 1.5 mL disposable polypropylene
tubes for each sample. Warm up the RPMI 1640 (10% FBS)
and Trypsin at 37 C ahead of time.
6. Collect the supernatants of cell culture into individual 15 mL
tubes (see Note 15).
7. Wash the cells with 0.5 mL PBS, then collet the PBS into the
same 15 mL tube.
8. Harvest the cells: add 0.25 mL of 0.25% Trypsin, and incubate
at 37 C for 5 min. Stop trypsinization by adding 0.5 mL
RPMI (10% FBS). Make sure all the cells are detached by
pipetting several times. Collect the cells to the same previous
15 mL tubes.
9. Centrifuge at 500 g for 5 min to precipitate the cells.
10. Wash the cell pellets by resuspending in 5 mL of PBS per tube,
and centrifuge again under the same conditions.
11. Discard supernatants and resuspend cells with 1 mL of
CM-H2DCFDA working solution. Incubate at 37 C pro-
tected from light for 30 min. Note, the negative control
(unstained sample) should be resuspended with 1 mL of phe-
nol red and FBS-free DMEM.
12. Wash the cells with PBS as in steps 7 and 8 (see Note 16).
13. Resuspend the cells with 1 mL of DMEM (phenol red and FBS
free), and transfer to 1.5 mL tubes. For positive control cells,
add 1 U/mL Glucose Oxidase in 1 mL DMEM, and incubate
for additional 30 min at 37 C.
14. Analyze the samples by flow cytometry as follows: (see
Note 17).
15. Set the instrument parameters:
16. CM-H2DCFDA has approximate fluorescence excitation at
492–495 nm and emission at 517–527 nm. Thus, choose
channel BL1 with filter range of 515–545 nm (Fluorescein
isothiocyanate microscopy (FITC)).
17. Set appropriate gates for the cell population of interest: In the
protocol detailed here, we recommend collecting all the cells in
the wells, including floating dead cells, and cells in the process
of apoptosis. However, the profile of ROS generation is differ-
ent between living, dead, and apoptotic cells. By forward (FSC)
and side (SSC) scatter analysis, it is possible to choose different
target populations according to the experimental aims. For
instance, if only living cells are wanted, a gate must be set to
include only the cell population within typical FSC and SSC
values, and exclude dead or apoptotic cells with smaller FSC
but larger SSC values (Fig. 6a). Otherwise, a gate can be set to
include both populations, and ROS generation analyzed in all
cells (Fig. 6b).
a b
8.4
8.4
6
6
SSC-A (106)
SSC-A (106)
4
4 2
2
E1 E1
31.66% 46.00%
0
0
0 2 4 6 8.4 0 2 4 6 8.4
6
FSC-A (106) FSC-A (10 )
Fig. 6 FSC/SSC dot plot of palmitate treated BRIN-BD11 cells. (a) the gate was drawn to include only living
cells. (b) the gate was drawn to include both living and dead cells
Fig. 7 Histogram of FITC fluorescence of BRIN-BD11 cells. The X-axis indicates fluorescence intensity, and the
Y-axis indicates cell count. (a) cells treated with vehicle (BSA) for 24 h. (b) cells treated with 125 μM palmitate
for 24 h. Palmitate treatment promoted elevated levels of ROS (Geometric Mean of fluorescence intensity
9377 vs. 24,920 respectively)
18. Data analysis: Collect the data of geometric mean of fluores-

cence intensity of the gated population (Fig. 7). Geometric
mean fluorescence intensity values can be plotted in regular bar
plots and an n number of independent samples can be run in
order to estimate experimental variability.
4 Notes
Measurement of Nitric Oxide Free Radical (NO) Release
1. Prepare a stock solution of 0.1% NED in water by adding

50 mg of NED in 50 mL of water. Store at 4 C protected
from light. Can be stored for at least 6 months.
2. Prepare a stock solution of 1% sulfanilamide in 5% phosphoric
acid by adding 500 mg of sulfanilamide in 50 mL of 5%
phosphoric acid. Store at 4 C protected from light. Can be
stored for at least 6 months.
3. Prepare a stock solution of 100 mM sodium nitrite in water by
adding 69 mg of sodium nitrite in 10 mL of water. Store at 4 C
protected from light. Can be stored for at least 6 months.
4. Distilled water, water which has been treated with a mixed

bed ion exchange resin, or ultrapure (17 MW cm or equiva-
lent) water are considered nitrite-free water and suitable for
use. Nitrite-free water should be stored in glass vessels to
prevent any possible leaching of NO2 compounds from the
container.
5. Normally, beta cells are cultured in RPMI 1640 media,
supplemented with 10% fetal bovine serum and 100 units/
mL penicillin, 0.1 mg/mL streptomycin. Use complete beta
cell culture media in order to prepare nitrite standard curve
dilution.
6. Experimental samples are the supernatant of beta cells culture
media. Beta cells are cultured at the conditions to be tested for
NO production and the supernatant of culture media is col-
lected and frozen at 20 C for subsequent analysis.
Glutathione Redox State Measurement
7. Phosphate buffer 143 mM can be made by mixing 3.1 g of

sodium phosphate dibasic heptahydrate (Na2HPO4 7H2O)
and 371.3 mg of sodium phosphate monobasic monohydrate
(NaH2PO4·H2O) in 100 mL of distilled water.
8. Samples can be stored at 20 C for posterior analysis; how-
ever, long-term storage is not advised.
9. Samples must be mixed with NEM and Pipes/KOH in the
following way. To 50 μL of sample, add 17 μL of NEM and
mix with pipette. Then, 10 μL of 2 M Pipes/KOH is added to
the wall of the tube at an angle of about 45 without contact
with the sample. Once closed, the tube is quickly mixed by
vortex.
10. Ethyl acetate washes are necessary in order to remove excess
NEM from the samples. This prevents further reactions with
GSH molecules that are generated during the GSSG reduc-
tase assay recycle.
11. It is recommended to assay all samples and standards in
triplicate. In addition, the timing of GSSG reductase activity
is crucial for the experimental results. Thus, no more
than half of a 96-well plate is recommended for use at any
one time.
General Oxidative Status
12. When seeding the cells, in addition to treatment groups, set

two wells for both, negative and positive controls. Negative
controls are unstained cells (free of probe) to assess autofluor-
escence from experimental system. A positive control can be
any type definite ROS-inducible interference. For example,

125 μM of palmitate for 24 h can strongly stimulate ROS
generation in β-cells, according to our own results. In this
protocol, as a positive control, we use glucose oxidase to react
with glucose in media to form gluconolactone and H2O2, and
the latter is a strong oxidizing agent for CM-H2DCFDA.
13. Concentrated 200 μM CM-H2DCFDA can be aliquoted,
stored, and protected from light at 20 C. Frozen aliquots
must be used within 1 month from reconstitution date.
14. The recommended concentration of the working solution of
CM-H2DCFDA is 1–10 μM. In our routine, we use 2 μM.
15. If only living cells are wanted, without the interference of
dead and apoptotic cells, discard the supernatant.
16. After treatment with CM-H2DCFDA, protection from light
is even more critical since light can induce oxidative reaction.
17. Analyze samples shortly after sample preparation. The fluo-
rescence intensity will decrease with time.
References
1. Halliwell B (1994) Free radicals, antioxidants, beta-cell functional integrity, metabolism and
and human disease: curiosity, cause, or conse- defense from inflammatory challenge. J Endo-
quence? Lancet 344(8924):721–724 crinol 211(1):87–97. https://doi.org/10.
2. Tiedge M, Lortz S, Drinkgern J, Lenzen S 1530/Joe-11-0236
(1997) Relation between antioxidant enzyme 8. Cruzat VF, Keane KN, Scheinpflug AL,
gene expression and antioxidative defense sta- Cordeiro R, Soares MJ, Newsholme P (2015)
tus of insulin-producing cells. Diabetes 46 Alanyl-glutamine improves pancreatic beta-cell
(11):1733–1742. https://doi.org/10.2337/ function following ex vivo inflammatory chal-
diabetes.46.11.1733 lenge. J Endocrinol 224(3):261–271. https://
3. Lenzen S, Drinkgern J, Tiedge M (1996) Low doi.org/10.1530/JOE-14-0677
antioxidant enzyme gene expression in pancre- 9. Kolberg A, Rosa TG, Puhl MT, Scola G, da
atic islets compared with various other mouse Rocha Janner D, Maslinkiewicz A, Lagranha
tissues. Free Radic Biol Med 20(3):463–466 DJ, Heck TG, Curi R, de Bittencourt PI Jr
4. Newsholme P, Cruzat VF, Keane KN, (2006) Low expression of MRP1/GS-X
Carlessi R, de Bittencourt PI Jr (2016) Molec- pump ATPase in lymphocytes of Walker
ular mechanisms of ROS production and oxi- 256 tumour-bearing rats is associated with
dative stress in diabetes. Biochem J 473 cyclopentenone prostaglandin accumulation
(24):4527–4550. https://doi.org/10.1042/ and cancer immunodeficiency. Cell Biochem
BCJ20160503C Funct 24(1):23–39. https://doi.org/10.
5. Rabinovitch A, Suarez-Pinzon WL (1998) 1002/cbf.1290
Cytokines and their roles in pancreatic islet 10. Cruzat VF, Pantaleao LC, Donato J Jr, de Bit-
beta-cell destruction and insulin-dependent tencourt PI Jr, Tirapegui J (2014) Oral supple-
diabetes mellitus. Biochem Pharmacol 55 mentations with free and dipeptide forms of
(8):1139–1149. https://doi.org/10.1016/ L-glutamine in endotoxemic mice: effects on
S0006-2952(97)00492-9 muscle glutamine-glutathione axis and heat
6. Griess P (1879) Bemerkungen zu der Abhan- shock proteins. J Nutr Biochem 25
dlung der HH. Weselsky und Benedikt ?Ueber (3):345–352. https://doi.org/10.1016/j.
einige Azoverbindungen?? Ber Dtsch Chem jnutbio.2013.11.009
Ges 12(1):426–428 11. Akerboom TP, Sies H (1981) Assay of gluta-
7. Krause MS, McClenaghan NH, Flatt PR, de thione, glutathione disulfide, and glutathione
Bittencourt PIH, Murphy C, Newsholme P mixed disulfides in biological samples. Meth-
(2011) L-arginine is essential for pancreatic ods Enzymol 77:373–382
Chapter 15
Primary Adipocytes as a Model for Insulin Sensitivity

Mohamed S. Zaibi
Abstract
Obesity and its comorbidity insulin resistance lead to the development of chronic metabolic diseases, such
as impaired fasted blood glucose and type 2 diabetes. Adipose tissue plays an important role in whole-body
glucose homeostasis, particularly in obese individuals; therefore, many in vivo models of type 2 diabetes are
obese, such as Lepob/ob and Leprdb/db mice or ZDF rats. Primary adipocytes therefore represent an
attractive in vitro model to study insulin-mediated glucose uptake to investigate the mechanisms of insulin
resistance and explore the potential insulin-sensitizing properties of new antidiabetic drugs.
Primary adipocytes are isolated by collagenase digestion of adipose tissue, Glucose transport is evaluated
by the measurement of intracellular uptake of a tracer (D-[U14C] glucose). The uptake of [U-14 C] glucose
reflects directly glucose transport.
In this chapter, we will describe the protocol for the isolation of primary rodent adipocytes and the
measurement of basal and insulin-stimulated glucose uptake.
Key words Primary adipocytes, Collagenase digestion, Insulin sensitivity, Glucose uptake
1 Introduction
In the past 20 years, adipose tissue has long been recognized as a

site for storage of excess energy derived from food intake; this form
of reserve is released during fasting and long-term food deprivation
and used as a source of energy [1]. The apparent simplicity of the
role of white adipose tissue has been replaced by ever-growing
complexity. Indeed, the growth in interest in white adipose tissue
has been mainly generated by the increasing concern with obesity-
induced insulin resistance and type 2 diabetes and its related com-
plications in public health [2]. Thousands of publications have
described the important physiological roles that the adipocytes
play in nutrient homeostasis, including appetite, satiety, fatty acid
oxidation, and glucose uptake.
Although the exact causes of insulin resistance are yet to be
completely understood, the major contributors to insulin resistance
are excess weight and physical inactivity. The decrease in insulin
sensitivity in target tissues, such adipose tissue is compensated by an
255
256 Mohamed S. Zaibi
overproduction of insulin from β-cells in the pancreas, which is a

fundamental defect that precedes the development of the cluster of
abnormalities associated with type 2 diabetes [3–6]. Preventing
insulin resistance and/or improving insulin sensitivity appears to
be a key factor in the prevention and/or the treatment of type
2 diabetes. Because obesity with or without overt hyperglycemia
is associated with insulin resistance [7, 8], and attention has focused
on abnormalities in adipose tissue that could lead to decreased
insulin sensitivity. Given the important role played by adipocytes
in the regulation of systemic energy balance and nutrient homeo-
stasis, it is reasonable to imagine that these cells might be therapeu-
tic targets for metabolic diseases. Primary adipocytes from different
adipose tissue depots represent an interesting and relatively simple
in vitro model to investigate the underlying mechanisms of insulin
resistance, as well as in the research of molecular targets for new
insulin sensitizer drugs [9]. Moreover, elucidating such mechan-
isms will provide new insight into our better understanding of the
mechanisms of insulin resistance in vivo.
2 Materials
2.1 Adipocytes 1. Forceps.

Isolation 2. Scissors: one for skin incision, another for removal of fat pads
and another to mince the adipose tissue.
3. Krebs-Ringer HEPES (KRH) buffer stock solution (see Note 1).
Make up Krebs/Ringer/HEPES (KRH) buffer in 1liter
according to the following.
120 ml NaCl 1 M (120 mM)

100 ml NaHCO3 0.1 M (10 mM)
30 ml HEPES 1 M (30 mM)
12 ml MgSO4, 7H2O 0.1 M (1.2 mM)
4.7 ml of KCl 1 M (4.7 mM)
1.2 ml KH2PO4 1 M (1.2 mM)
4. Collagenase type II, (see Note 2).

5. Bovine serum albumin (fraction V).
6. Adenosine.
7. 50 ml polypropylene tubes.
8. 50 ml syringe and tubing.
9. 15 ml polypropylene tubes.
Primary Adipocytes as a Model for Insulin Sensitivity 257
10. Water bath.

11. Stuart SF1 flask shaker, or equivalent.
12. 200–300 μm nylon mesh.
13. Funnel to hold the nylon mesh.
14. 37 C CO2-incubator.
2.2 Glucose Uptake 1. 0.5 ml polyethylene micro centrifuge tubes.

2. Tracer D-[U14C] Glucose (ref: NEC042V250UC, PerkinEl-
mer) or equivalent generic with specific activity range
250–360 mCi/mmol and radiochemical purity greater than
97%.
3. Insulin from bovine pancreas.
4. Custom made 37 C incubator with rotating system and tubes
rack holder (Fig. 1).
5. Dinonyl phthalate, (see Note 3).
6. Refrigerated centrifuge Sanyo Mistral 3000i, or equivalent.
7. 80 C freezer.
8. Resco guillotine cutter, (Petcetera, UK. Ref: 81243) to cut the
tubes http://www.petcetera.co.uk/grooming-preparations-
and-tools/nail-care/81243-resco-guillotine-nail-cutter-chrome-
small-727.html.
9. 10 ml scintillation vials.
Fig. 1 37 C incubator with rotating system. The 0.5-ml polyethylene microcentrifuge tubes are placed on the
rack and incubated at 37 C for 30 min, under constant rotational movement (30 rpm). Rotating unit (a),
Clamped rack (b), Air flow heating unit (c)
10. Vortex mixer.

11. LKB Wallac 1214 RackBeta Liquid Scintillation Counter, or
equivalent with a selective window for 14C.
3 Methods
3.1 Adipocytes Fat cells are prepared from different adipose tissue depots by using
Preparation the method described previously by Zaibi et al. [10]. Adipose tissue
(perigonadal, visceral, or subcutaneous) is minced and digested
with collagenase type II in Krebs-Ringer HEPES (KRH) buffer
pH 7.4 and 37 C. Isolated cells are filtered through 250–300 μm
nylon mesh, and the floating layer of adipocytes are washed several
times with a fresh buffer, then concentrated to 40% of final volume
of KRH buffer containing 5% bovine serum albumin (BSA) and
0.3 mM glucose and preincubated for 45 min under 95% O2: 5%
CO2.
1. Prepare Prep/Inc. buffers: 500 ml KRH buffer of stock solu-
tion containing 1.25 ml CaCl2, 5 g BSA (1%), 0.5 g Glucose.
2. Adjust the pH to 7.4.
3. Split into:
(a) Preparation buffer 300 ml containing 30 μl of 2 mM
adenosine (see Note 4) to reduce basal lipolysis).
(b) Incubation buffer 200 ml.
4. Warm both solutions to 37 C in water bath (see Note 5).
5. Animals (mice or rats) are euthanized by a method of killing
animals that are considered to be acceptable and humane (see
Note 6).
6. Collect the adipose tissue (approximately 5 g) in a 50-ml
polypropylene tube containing 30–40 ml of pre-warmed prep-
aration buffer.
7. Weigh the adipose tissue and split into two polypropylene tubes
(approximately 2–3 g of adipose tissue/tube), in 6 ml of prep-
aration buffer containing 1.5–2 mg of collagenase/g of adipose
tissue.
8. Dice the adipose tissue into small pieces (2–3 mm diameter
pieces) is this a note.
9. Incubate at 37 C in a water bath, under constant shaking
(250 cycles/min) until full digestion (approximately 60 min).
Gently swirl the tubes every 15 min and check the degree of
digestion (see Note 7).
10. When digestion is complete, adipocytes are separated by filtra-
tion through the 200–300 μm nylon mesh and the adipocytes
transferred to a 50-ml polypropylene tube and re-suspended in

40 ml of fresh KRH buffer.
11. Cells are allowed to stand for 2–3 min, and the infranatant is
removed by aspiration using a 50-ml syringe (see Note 8).
12. Wash the floating layer of adipocytes four times by
re-suspending the cells in 40 ml of fresh preparation KRH
buffer, repeat step 9. After each wash, check the purity of the
cell suspension (see Note 9).
13. Wash adipocytes twice with incubation KRH buffer as
described previously in steps 10 and 11.
14. Transfer the floating layer of adipocytes into 15-ml
polypropylene tube.
15. Adipocytes are concentrated to 40% of final volume (e.g., 4 ml
of adipocytes in 10 ml of final volume) of incubation KRH
buffer, containing 5% BSA and 5.5 mM of glucose, and pre-
incubated for 45 min in 95:5% O2:CO2 atmosphere.
3.2 Measurement Glucose transport is measured by the tracer D-[U14C] Glucose

of Glucose Uptake method. Aliquots of fat cells are incubated at 37 C for 30 min,
with or without insulin in Kreps-Ringer-HEPES buffer containing
5% BSA (fraction V, de Cohn), 0.3 mM glucose and 0.1 μCi/ml of
D-[U14C] glucose. The reaction is terminated by spinning the cell
suspension through dinonyl phthalate. The upper phase which
contains fat cells is collected and subjected to liquid scintillation
counting. The 14C-radioactivity associated with the cells directly
reflects the glucose uptake by the adipocytes.
1. Add 180 μl to 0.5 ml polyethylene microcentrifuge tubes, of
Kreps-Ringer-HEPES buffer containing 5% BSA (fraction V de
Cohn), 0.3 mM glucose, and 0.1 μCi/ml of D-[U14C]
glucose.
2. Aliquot 20 μl of fat cells (see Note 10) and add them carefully
to each tube.
3. Adipocytes are incubated under gentle rotating movement
(30 cycles/min) at 37 C for 30 min with or without tested
compounds (see Note 11), in the absence or presence of vari-
able concentrations of insulin. See example in Fig. 3.
4. After incubation, add 100 μl silicone oil to 400 μl microfuge
tubes. Tap to make sure that no air bubbles are present (see
Note 12).
5. Centrifuge the tubes at 3000 g and 4 C for 15 min.
6. Remove the tubes carefully so as not to disturb the upper phase
containing the adipocytes (see Note 13).
7. Store the tubes overnight in a freezer at 80 C. At this stage
the next steps of the protocol can be delayed (see Note 14).
Fig. 2 Procedure for processing frozen microcentrifuge tubes during the transfer of adipocytes layer into
scintillation vials. Frozen microcentrifuge tube (a) containing (from top to bottom) radioactive adipocytes,
dinonyl phthalate oil and radioactive buffer, container with dry ice (b) to maintain the tubes at low temperature,
tubes’ fraction to be transferred to scintillation vial (c)
8. Cut the upper phase which contains adipose cells (Fig. 2), using
a small ‘Resco guillotine’ nail cutter, or a very sharp scissors and
transfer it to the scintillation vial (see Note 15).
9. Add 0.5 ml of distilled water to each vial.
10. To each vial, add 10 ml of scintillation solution Ecosint™ A, or
a best equivalent (see Note 16).
11. Mix each tube for 10–15 s (using the vortex mixer), and leave
overnight.
12. Measure samples using the scintillation counter for 300 s.
13. Disintegrations per min (dpm) are recorded and glucose
uptake is expressed per mg of total cell protein (see Note 17).
Total activity in 180 μl buffer ðdpmÞ

Specific activity ðdpm=nmolÞ ¼
Total substrate available in 180 μl buffer ðnmolÞ
14
U C glucose counts=tube ðdpmÞ
Glucose uptake=20 μl of cells ðnmolÞ ¼
Specific activity ðdpm=nmolÞ
Data are expressed in nmol/mg of total protein measured in
20 μl of adipocytes aliquots.
4 Notes
1. Prepare stock solutions 1 M NaCl; 0.1 M NaHCO3; 1 M

HEPES; 0.1 M MgSO4, 7H2O; 1 M KCl and 1 M KH2PO4.
Make up your KRH buffer stock solution as described in Sub-
heading 2.1, item 3; the KRH solutions could be kept at room
temperature in the dark.
2. The choice of the type of collagenase is important. Variability
exists between lots in the enzyme activity and performance of
the type of collagenase used. Collagenase Type II (Sigma) is
suitable for adipocytes release. Solutions are typically prepared
at 1–2 mg in KRH buffer (3 ml/g of adipose tissue).
3. Dinonyl Phthalate is used to separate adipocytes layer after
incubation with KRH buffer containing D-[U14C] glucose.
4. The addition of adenosine to the preparation buffer prevents
cell breakage and high lipolytic rates during adipose tissue
digestion.
5. All the media should be warmed to 37 C, during the adipose
tissue collection, digestion, and for isolated adipocyte incuba-
tions and dramatically changing the temperature quickly can
shock the cells.
6. There is a Home Office Code of Practice on Humane Killing of
Animals linked to Schedule 1 to the Animals (Scientific Proce-
dures) Act 1986, which provides the appropriate methods and
explains how to ensure that they are carried out competently.
7. The full digestion is achieved when the collagenase solution
presents a milky aspect. Collagenase digestion should not be
maintained for longer than 60 min, as this reduces adipocytes’
viability.
8. The adipocytes are washed by adding gently a fresh media into
the tubes, and slowly aspirating the infranatant (below the fat
cell layer) using blunt-end needle connected to a syringe.
9. Check the purity of isolated adipocytes suspension, this should
have a clear aspect (does not contain stromal vascular fraction,
blood cells immune cells, or adipocytes precursors).
10. It is not easy to aliquot a small volume of adipocytes suspen-
sion, such as 20 μl, the cells mixture is quiet thick, oily, and
sticky; use preferably P200 pipette, and cut the tip to get a
volume as exact as possible.
11. The tubes are placed in empty pipette tips rack (the most
convenient is P200 rack tips), fixed to the rotating unit by a
clamp, and incubated at 37 C, under a slow rotating move-
ment (30 cycles/min) as described in Fig. 3.
Fig. 3 Effect of an insulin sensitizer compound (P) on glucose uptake in primary adipocytes of female wild-type
mice (Lean mice) and insulin-resistant mice model (Obese mice). Adipocytes were treated with insulin,
compound P (3 μM) alone, or a combination with insulin. n ¼ 8 mice per group for all columns. ∗p < 0.05;
∗∗
p < 0.01 for effect of compound P (P) compared to controls at the same concentration of insulin (C).
{
p < 0.05; {{p < 0.01; {{{p < 0.001 for effect of insulin compared to no insulin at the same concentration of
compound P. (Ngala et al. [9])
12. The separation of adipocytes during the centrifugation will not

be performed properly if air bubbles are present in the mixture.
Make sure that no air bubbles are in the tubes after adding
dinonyl phthalate oil.
13. After centrifugation tubes, should be removed and handled
very carefully so as not to disturb the adipocytes top layer.
14. At this stage, the microtubes could be stored at 80 C, the
next steps of the experiment could be performed later.
15. Before proceeding with the cells’ digestion and measurement
of glucose transport in adipocytes, the microtubes should be
kept frozen, so as not to disturb the top layer of cells. Take the
tubes from 80 C storage and keep them in dry ice before
transferring the top part (containing the cells) to the scintilla-
tion vials, the remaining part should go to the radioactive
waste.
16. Good solvents for liquid scintillation counting are alkylated
aromatic organic solvents. The solvent must absorb the energy
efficiently, safely, and have low vapor pressure and a high flash
point [above 140 C (nonflammable)]. In case where the scin-
tillation solution Ecosint™ (National Diagnostics) is not avail-
able, the Ultima Gold™ cocktails (PerkinElmer Inc.) are an
alternative choice for the vast majority of common counting
applications.
17. Before incubating adipocytes in KRH buffer containing
D-[U14C] glucose, keep a few aliquots of 20 μl of cells at
80 C for the measurement of total protein, using a method
based on the Lowry assay (Bio-Rad, Hemel Hempstead, UK).
References
1. Trayhurn P, Beattie JH (2001) Physiological metabolic and genetic abnormalities. Am J
role of adipose tissue: white adipose tissue as Med 113(Suppl A):3S–11S
an endocrine and secretory organ. Proc Nutr 7. Ferrannini E (1995) Physiological and meta-
Soc 60:329–339 bolic consequences of obesity. Metabolism
2. Bailey CJ (1999) Insulin resistance and antidia- 44:15–17
betic drugs. Biochem Pharmacol 8. Walker M (1995) Obesity, insulin resistance,
58:1511–1520. https://doi.org/10.1016/ and its link to non-insulin-dependent diabetes
S0006-2952(99)00191-4 mellitus. Metabolism 44:18–20
3. Vernon RG, Clegg RA (1985) The metabolism 9. Ngala RA, Zaibi MS, Langlands K, Stocker CJ,
of white adipose tissue in vivo and in vitro. In: Arch JRS, Cawthorne MA (2014) Stimulation
Cryer A, Van RLR (eds) New perspectives in of glucose uptake in murine soleus muscle and
adipose tissue: structure, function and develop- adipocytes by 5-(4-phenoxybutoxy) psoralen
ment. Butterworths, London, pp 65–86 (PAP-1) may be mediated by Kv1.5 rather
4. Boden G (2001) Pathogenesis of type 2 diabe- than Kv1.3. PeerJ 2:e614. https://doi.org/
tes. Insulin resistance. Endocrinol Metab Clin 10.7717/peerj.614.
N Am 30:801–815 10. Zaibi MS, Stocker CJ, O’Dowd J, Davies A,
5. Olefsky JM, Ciaraldi TP, Kolterman OG Bellahcene M, Cawthorne MA, Brown AJ,
(1985) Mechanisms of insulin resistance in Smith DM, Arch JR (2010) Roles of GPR41
non-insulin-dependent (type II) diabetes. Am and GPR43 in leptin secretory responses of
J Med 79:12–22 murine adipocytes to short chain fatty acids.
6. LeRoith D (2002) Beta-cell dysfunction and FEBS Lett 584:2381–2386
insulin resistance in type 2 diabetes: role of
Chapter 16
Assessing Islet Transplantation Outcome in Mice

Aileen J. F. King and Chloe L. Rackham
Abstract
Islet transplantation is a potential treatment for Type 1 diabetes; however, improvements need to be made
before it could become clinically widely available. In preclinical studies, the mouse is often used to model
islet transplantation, with most studies aiming to improve transplantation outcome by manipulating the
islets prior to transplantation or by treating the recipient mouse. Here, we describe the process of islet
transplantation in the mouse, including how one can make the mouse diabetic, isolate donor islets, and
transplant the islets into two different sites. Finally, we discuss how to assess the outcome of the transplan-
tation in order to determine whether the experimental intervention has been beneficial.
Key words Islet transplantation, Mouse, Islet isolation, Intraportal islet transplantation, Renal sub-
capsular islet transplantation
1 Introduction
Transplantation of islets is a potential treatment for Type 1 diabetes

that has clinically been shown to be able to reverse insulin depen-
dence [1]. However, transplantation outcomes in the clinical
setting have been disappointing with many patients reverting to
insulin dependence [1]. Much effort has been focused upon
improving islet transplantations using mouse models. This can
involve a variety of strategies and experimental designs. Islets can
be manipulated prior to transplantation by encapsulation [2, 3],
gene therapy [4, 5] or preculturing with drugs or peptides that may
enhance beta cell survival and/or function [6, 7]. Alternatively, the
relative importance of particular pathways in islet transplantation
outcome can be studied using transgenic [8] or knock-out islets
[9]. Islets have been implanted into a variety of sites in the mouse.
The most common is the subcapsular site of the kidney, which is
chosen for ease of access, ability to easily locate the graft after
transplantation and ability to remove the graft by nephrectomy.
The intraportal site is the clinically used site and is also used in
rodent models although this technique is more invasive and the
265
266 Aileen J. F. King and Chloe L. Rackham
graft harder to locate. Two main experimental designs are typically

used to demonstrate improved efficacy of an islet graft: minimal
mass model or graft survival.
1.1 Minimal Mass In minimal mass models, fewer islets are implanted than would be
Models expected to reverse hyperglycemia in all animals [6, 7, 10]. If the
intervention has had a positive effect, more animals will become
normoglycemic. In mice, implanting between 300 and 500 mouse
islets will typically cure all animals. Therefore, the minimal mass
chosen is often between 150 and 250 islets which will usually cure
around 15–50% of the animals in the control group, depending on
site of transplantation, culture period and starting weight of the
mice. Due to the large variation in the blood glucose concentra-
tions of mice implanted with a minimal mass number, at least
10–14 mice per group are often required to see an effect.
1.2 Graft Survival Graft survival designs are most useful for allogeneic or xenogeneic
transplantations where the intervention intends to prevent or delay
immune rejection [2, 5, 11]. In these models, sufficient numbers of
islets should be implanted to initially reverse hyperglycemia in all
animals. Rising blood glucose concentrations can then be used as
an indicator of rejection. In allogeneic mouse models, rejection
occurs in control animals around 10–20 days after implantation
[2, 5, 11]. Alternatively, the grafts can be harvested at specific
time points after implantation to histologically detect rejection by
staining immune cell infiltration of the graft [11].
1.3 Measurable Prior to transplantation, diabetes can be induced by injection of

Outcomes alloxan [12] or streptozotocin (as described below) [13]. Alterna-
tively, animals developing diabetes spontaneously such as the non-
obese diabetic (NOD) or the Akita mouse can be used as islet
transplant recipients [14]. Reversal of hyperglycemia is a key out-
come of islet transplantation. It is important that blood glucose is
measured at the same time every day, preferably first thing in the
morning after night feeding, as continuous glucose monitoring in
diabetic rodents with an islet graft has shown a substantial diurnal
difference in blood glucose concentrations [15]. Although there
are no clear guidelines on normoglycemia in mice, our laboratory
typically uses a nonfasting blood glucose concentration of less than
11.1 mM on at least three consecutive readings as indicative of
normoglycemia. It should be noted that on the day after surgery,
the blood glucose concentrations may be artificially low due to
leaking of insulin from islets damaged by the transplantation pro-
cess and reduced appetite in the mouse leading to reduced food
intake. It is essential that the weight of the mouse is monitored
together with blood glucose concentrations, as it negates any
effects on blood glucose due to sickness or reduced appetite. In
addition, on an ethical basis, it is undesirable that the mouse loses
Islet Transplantation in Mice 267
considerable weight (>20%). Plasma insulin concentrations can

also be measured, but as more blood is needed for those assays,
this is not generally measured as frequently as blood glucose con-
centrations. In cured mice, glucose tolerance tests can be carried
out to challenge the grafts further.
1.4 Graft Removal If the islets are implanted below the kidney, a nephrectomy of the
graft-bearing kidney can be carried out to demonstrate that the
graft is responsible for normoglycemia. After removing the graft,
the blood glucose concentrations will rise again if it is indeed the
graft that is responsible for maintaining normoglycemia, which
helps to rule out endogenous pancreatic beta cell regeneration
[2, 10]. In addition, the endogenous pancreas can be analyzed for
insulin content or insulin positive cells [16].
1.4.1 Ex Vivo Analysis Depending on the experimental question, the graft can be histo-
logically analyzed. Usually, the graft remains in situ in the implan-
tation organ. Hormone staining can determine the survival and/or
morphology of the islet cells [10], endothelial cell staining can
indicate the revascularization process [10], and staining immune
cells can indicate the involvement of different cells in rejection
[17]. Replication rates of the transplanted islets can be determined
by administering 5-bromo-2-deoxyuridine (BrdU) to the rodents
prior to killing them, which can then be quantified histologically
[18]. Alternatively, Ki-67 staining can detect replicating cells [19].
If more appropriate, the graft can be harvested to measure the
insulin content [10], although it should be noted that in diabetic
animals, the beta cells will be degranulated so it may not form an
accurate basis to determine beta cell mass. There are several steps
involved in measuring islet transplantation outcome in mice which
will be subdivided into the following sections:
1. Inducing diabetes.
2. Isolation of donor islets.
3. Transplantation of islets (under the renal capsule and
intraportally).
4. Posttransplantation monitoring.
5. Graft retrieval.
2 Materials
2.1 Induction 1. Citrate buffer: Dissolve 0.21 g citric acid in 100 ml saline (0.9%
of Diabetes NaCl). Alter the pH to 4.5 and filter through a 0.22 μm filter
into a sterile bottle. Store at 4 C for up to a month or freeze.
2. Streptozotocin solution: Make a 2% streptozotocin solution
immediately before injections by dissolving streptozotocin in
citrate buffer (e.g., 5 ml citrate buffer into 100 mg

streptozotocin).
3. 1 ml syringes and 27 G needles.
2.2 Isolating Islets 1. Suitable cell culture medium for washing the islets (we use
Minimum essential medium (MEM)) supplemented with 10%
Newborn Calf Serum.
2. Collagenase solution: Collagenase type XI dissolved in serum-
free medium (we use MEM) at a concentration of 1 mg/ml.
Approximately 2 ml is needed per mouse.
3. Tools for isolating: one large pair of scissors, one large forceps,
two small forceps, one small pair of scissors, and one bulldog
clamp.
4. Dissecting microscope.
5. 2 ml syringes and 25–30 G needles depending on mouse size.
6. 70% ETOH.
7. Kim wipes.
8. Absorbable paper.
9. Histopaque-1077.
10. Plastic: Sterile 50 ml centrifuge tubes, petri dishes and Eppen-
dorf tubes.
11. Funnel.
12. 10 ml plastic pipettes and autopipettor.
13. 425 micrometer stainless steel mesh.
14. Refrigerated centrifuge with inserts for 50-ml tubes with adjus-
table settings to allow slow acceleration and brake to be
turned off.
16. Culture medium (RPMI 1640 supplemented with 10% fetal
calf serum, 100 U/ml penicillin, and 100 ng/ml
streptomycin).
2.3 Transplantation 1. Anesthetic (we use 1–5% isoflurane, 95% oxygen).

of Islets 2. Analgesic (we use buprenorphine, 30 μg/kg injected
2.3.1 Renal Subcapsular subcutaneously).
Islet Transplantation 3. Heating pad.
4. Saline in a syringe to keep kidney moist.
5. Serum-free RPMI 1640 media.
6. Hamilton syringe (500 μl).
7. Razor.
8. Antiseptic skin scrub.
9. 70% alcohol.
10. Centrifuge with 15 ml tube inserts.
11. Tools for operation (per mouse): one scalpel, four small for-
ceps, two small pairs of scissors, drapes, needle holder, suture,
23-G needle and cauterizer.
12. Siliconized PE50 tubing (treated with SigmaCote and gas
sterilized).
13. Connector tubing (suitable to connect PE50 tubing to the end
of a P200 pipette tip).
14. P200 pipette and tips (+tips cut to fit the end of the Hamilton
syringe).
2.3.2 Intraportal Islet 1. Anesthetic (we use 1–5% isoflurane, 95% oxygen).
Transplantation 2. Analgesic (we use buprenorphine, 30 μg/kg injected
subcutaneously).
3. Heating pad.
4. Hamilton Syringe (500 μl).
5. Razor.
6. Antiseptic skin scrub.
7. 70% alcohol.
8. 25 G butterfly needle.
9. Tools for operation (per mouse): one scalpel, four small for-
ceps, two small pairs of scissors, alm retractor, Dumont forceps,
needle holder, suture, 23 G needle, cauterizer.
10. Drapes.
11. Absorbable hemostatic gelatin sponge (we use Spongostan).
12. Sterile cotton buds.
13. Sterile swabs (we use Topper 8).
14. Sterile saline.
15. Stereo microscope, preferably on an articulated arm.
2.4 Posttransplan- 1. Scales.

tation Monitoring 2. Blood glucose meter and strips.
3. 27 G or 30 G needles.
4. 30% glucose solution.
5. 1 ml syringes and 27 G needles for injection of glucose.
6. Insulin (depending on local ethical guidelines and on advice
from a veterinary surgeon).
7. Heparinized glass pipettes with rubber bulb and Eppendorf
tubes for collecting blood samples for detection of insulin.
8. Sensitive ELISA to measure insulin in small volume samples

(e.g., Mercodia Ultrasensitive Mouse Insulin ELISA).
9. Tools and equipment for operation if graft retrieval by nephrec-
tomy is to be carried out (per mouse): one scalpel, four small
forceps, two small pairs of scissors, drapes, needle holder,
suture. Anesthetic (e.g., 1–5% isoflurane, 95% oxygen), analge-
sic (e.g., buprenorphine, 30 μg/kg injected subcutaneously) as
advised by veterinarian.
10. Appropriate solution for graft depending on further studies:
(a) Histology: 4% formalin.
(b) Graft insulin content: 70% acid alcohol.
3 Methods
3.1 Inducing We generally induce diabetes 5 days prior to transplantation.

Diabetes
1. Inject the mice immediately after making the streptozotocin
solution (see Note 1).
2. Inject between 150 and 220 mg/kg intraperitoneally into each
mouse, depending on mouse strain. For C57Bl/6 mice, use
around 180 mg/kg (see Note 2).
3. Monitor weight and blood glucose of each mouse (see Note 3).
4. By 5–7 days, the mice should be stably diabetic (blood glucose
concentration > 20 mM).
3.2 Isolating Islets 1. This procedure can be carried out under anesthesia with exsan-
guination used to kill the mouse at the end of the procedure or
alternatively the mouse can be killed prior to the procedure by
cervical dislocation.
2. Lay mouse in a supine position and wet the fur with 70%
alcohol.
3. Make a large “V” incision in skin and the peritoneum from the
pubic region to the front legs to expose liver and intestines.
4. Rotate the mouse with its head toward you; place a lengthways
folded Kim wipe on the mouse’s chest across the sternum,
depress chest and flip back liver on top of tissue using half of
the tissue to hold the liver in place to expose posterior-dorsal
surface of the liver and the bile duct (see Note 4).
5. Move the intestines to the right side and gently pull the duo-
denum distally so the common bile duct can be visualized and
is taut.
6. Clamp duodenum at the Vater’s ampulla where the common
bile duct empties into the duodenum (see Note 5).
7. Inject 2–2.5 ml of collagenase into the duct to expand the

pancreas using a bent (90 ) 27 G or 30 G needle and 2 cc
syringe (see Note 6).
8. Detach the duct from the intestine using forceps, and then
carefully remove the inflated pancreas. Handle the pancreas as
little as possible to avoid leakage of the collagenase.
9. Place the pancreas in a 50 ml conical plastic tube on ice. Up to
four pancreata can be added to a single tube prior to incuba-
tion/digestion. It should take less than an hour to finish all
the mice.
10. Place the tubes containing the pancreata in a 37 ˚C water bath
for 10 min (depending upon the activity of the collagenase).
11. After the incubation, add 25 ml of ice-cold wash medium to
stop the reaction.
12. Vigorously shake tube or vortex for 10 s.
13. Centrifuge for 300 G for 1 min and 15 s.
14. Discard the supernatant and resuspend the pellet in 25 ml wash
medium (may need to gently shake to resuspend the pellet).
15. Centrifuge for 300 G for 1 min and 15 s.
16. Repeat wash step (14 and 15) twice.
17. After resuspending the pellet in 25 ml wash buffer, filter the
digest through a 425 μm stainless steel mesh into a new 50 ml
tube using a funnel. This will remove any undigested tissue.
18. Centrifuge at a higher speed of 340 G for 1 min and 30 s.
19. In one movement, turn the tube upside down to discard the
supernatant and with the tube still inverted place on absorbent
paper to soak up excess media (the pellet should be as dry as
possible). If necessary, you can also remove drops of MEM
from tube with a Kim wipe.
20. Resuspend the pellet in 15 ml Histopaque 1077 by vortexing.
21. Carefully, add 10 ml of wash medium by pipetting down the
sides of the tube so as not to mix with the Histopaque. There
should be a clear interphase between the histopaque and
medium.
22. Centrifuge at 1950 RCF for 24 min with slow acceleration and
no brake in a refrigerated centrifuge at 10 ˚C (see Note 7).
23. Remove islets at the interface using a 10 ml plastic pipette and
autopipettor (see Note 8).
24. Place the islets in a new 50 ml tube and fill up to 50 ml with
wash media.
25. Centrifuge at 340 G for 1 min 30 s and remove the supernatant
(see Note 9).
26. Add wash medium up to the 25 ml mark. This in itself will

disrupt the pellet so further shaking is not necessary.
27. Repeat this wash step twice.
28. If the islets are contaminated with exocrine tissue, sedimenta-
tion stages can be carried out for further purification. If the
islets look pure, proceed to step 30.
29. Resuspend the islets in 25 ml total media and let them sit for
4 min on ice. Remove the top 10 ml and add 10 ml new media.
Repeat four times to further purify the islets.
30. Pellet the islets and resuspend in 10 ml media.
31. Place into petri dish (do not use a coated-tissue culture plate).
32. If islets are to be transplanted immediately, pick them into an
Eppendorf tube using a 200 μl pipette and keep on ice or to
place in culture follow steps 33–39.
33. In a hood, swirl the petri dish, in a circulation motion to
coalesce islets in the center. Using the microscope and a
200-μl pipettor, pick the desired number of islets into an
Eppendorf tube containing 1 ml of culture medium.
34. Centrifuge using a low speed mini-centrifuge to pellet islets.
35. Wash the islets (resuspend/pellet/remove supernatant) twice
with cold sterile PBS.
36. After the last centrifuge, remove the PBS and add 1 ml of
culture medium (RPMI 1640 supplemented with 10% fetal
calf serum and antibiotics) into the Eppendorf tube with islets.
37. Add 4 ml culture medium into a sterile 60-mm petri dish.
38. Pipette the islets into the petri dish; wash the tube with 1 ml
borrowed from the plate to recover all islets.
39. Incubate at 37 C 5% CO2. A petri dish containing 5 ml media
should contain a maximum of 200 islets. Change the media
every 48 h.
3.3 Transplantation 1. Anesthetize the mouse with isoflurane or another appropriate

of Islets anesthetic.
3.3.1 Transplantation 2. Shave the fur on the left flank and treat the skin with surgical
of Islets Under the Kidney scrub, followed by 70% alcohol.
Capsule 3. Administer the analgesic and lay the mouse on a heat pad.
4. Lay drapes over the mouse, with the operation area exposed
and check if the mouse is adequately anesthetized (for example,
by lack of reaction to a paw pinch).
5. While the mouse is getting anesthetized, prepare the islets. This
is ideally carried out by a second person as the islets should be
prepared in parallel so they are ready for implantation at the
same time as the mouse kidney is externalized.
(a) Allow islets to settle to the bottom of an Eppendorf tube.

(b) Fill the Hamilton syringe with serum-free media and
attach a 200-μl pipette tip (the top of which may need
to be cut to fit the Hamilton syringe).
(c) Using the plunger, push medium into the tip so that
there is no air in the syringe or tip.
(d) Aspirate the islets into the tip and keep vertical to allow to
islets to settle, pull up a small amount of air to prevent
islet loss from the tip.
(e) Attach the siliconized and sterilized PE50 tubing to the
pipette tip using a 0.5 cm piece of larger connecting
tubing to fit the bottom of the pipette tip.
(f) Inject the islets about 10 cm into the PE tubing using the
plunger.
(g) Fold the PE tubing back on itself just in front of the islet
pellet and use a 1 cm piece of connector tubing to hold it
in place.
(h) Remove the pipette tip with the PE tubing attached and
centrifuge for 2 min at 200 G. The tubing can be kept in
place using the cut outer case of a 1-ml syringe within a
15-ml centrifuge tube.
(i) Screw the plunger of the Hamilton Syringe into place.
(j) After the centrifugation, top up the pipette tip with
medium so that it can be reattached to the Hamilton
syringe without any air bubbles, the islet pellet should
be visible just above the connector tubing (Fig. 1).
Fig. 1 Pelleted islets in the PE50 tubing ready for transplantation. To transplant,
remove the connector tubing carefully and cut a bevel just in front of the islet
pellet
(k) Carefully, remove the connector tubing at the bend of

the PE50 tubing, where the islets should have pelleted.
(l) Cut the PE50 tubing close to the islet pellet as a 45
angle.
(m) Turn the plunger so that the islets are at the opening of
the PE50 tubing.
6. Using a scalpel, make a 1.5–2 cm incision in the skin and using
scissors, cut a 1-cm incision in the peritoneum over the left
kidney.
7. Externalize the kidney by gently pressing under each side of the
kidney to allow it to pop out.
8. Make an incision in the kidney capsule at the bottom quarter by
scoring the capsule surface using a 23-G needle.
9. Keep the capsule moist with sterile saline.
10. Using Dumont forceps, gently lift the kidney capsule and insert
the PE tubing containing the islets under the capsule.
11. Push the tubing up into the top quarter of the kidney and using
the Hamilton syringe carefully inject the islets into the subcap-
sular space using the screw function.
12. Once the islets have been injected into the subcapsular space,
remove the tubing and cauterize the incision.
13. Pull up on the peritoneum around the kidney so that the
kidney falls back into place.
14. Suture the peritoneum and suture the skin.
15. Monitor the recovery of the mouse.
3.3.2 Islet 1. Anesthetize the mouse with isoflurane or another appropriate

Transplantation anesthetic.
Intraportally 2. Place the mouse in the supine position and shave the abdomen
of the mouse and treat the skin with surgical scrub, followed by
70% alcohol.
3. Administer analgesic and lay the mouse on a heat pad with its
paws taped down to retain the position of the mouse.
and check the mouse is adequately anesthetized.
5. In parallel to the operation, the islets should be prepared,
ideally by a second person so that they are ready for implanta-
tion at the correct time point.
(a) Fill a Hamilton Syringe with serum-free medium and
attach to a butterfly needle (25 G).
(b) Push the plunger to fill the tubing of the butterfly needle
with media.
(c) Aspirate the islets into a Butterfly needle.

(d) Make sure that the islets are a few cm up into the tube
before screwing in the plunger.
(e) Keep the tubing vertical to allow the islets to sediment to
the bottom of the tube, but keep the needle perpendicular
to stop the islets entering the needle until ready for the
transplantation.
(f) Immediately prior to the transplantation, screw the
syringe carefully until media appears at the needle
opening.
6. Using a scalpel, make an approximately 3-cm incision in the
skin and then use scissors to make a 3-cm incision in the
peritoneum from just above the xiphysternum. Use an alm
retractor to open up the abdomen.
7. Place a sterile saline-soaked swab over the right arm of the alm
retractor extending over to also cover the skin of the mouse.
Use a thin strip of saline-soaked swab (approximately 2 cm) to
ensure that the upper lobe of the liver is arranged in the
appropriate position so that the portal vein is not blocking
access to the portal vein.
8. Using saline-infused cotton buds, carefully move the intestines
onto the saline-soaked gauze to expose the intraportal vein.
Keep the intestines moist with saline and cover with a saline-
soaked swab, which also runs over the tissue next to the portal
vein and is pulled slightly downwards to help make the vein taut
and helping to place the portal vein in a suitable position for
implantation of the islets as seen in Fig. 2.
Fig. 2 The correct position of the portal vein prior to intraportal transplantation.
Note the swab to the right is slightly pulling the tissue into the correct position
9. Using fine forceps grip the tissue adjacent (operators left hand
side) to the portal vein and cannulate the portal vein.
10. Carefully inject the islets in a volume of approximately 200 μl
by using the screw function of the Hamilton Syringe.
11. Prior to removing the needle from the vein, apply spongostan
gel foam to the point of needle entry using Dumont forceps to
pick up the precut small square of spongostan and ensuring
that the foam is placed flat over the vein to prevent excess loss
of blood (see Note 10).
12. Applying some pressure to the Spongostan using a dry cotton
bud, remove the needle and hold the Spongostan in place for
approximately 2 min.
13. After removing pressure from the Spongostan, observe the vein
to check that there is no bleeding.
14. Leaving the Spongostan in place, carefully remove gauze swabs
and gently replace the intestines into the peritoneal cavity
(using soaked cotton buds).
15. Sew the peritoneum cavity.
16. Sew the skin.
17. Allow the mouse to recover and monitor closely.
3.4 Posttransplan- 1. Blood glucose concentrations and weight together can indicate
tation Monitoring glycemic control and should initially (for the first week) be
measured daily (see Note 11).
3.4.1 Blood Glucose
Monitoring 2. Using a 27-G or 30-G needle, make a small pin prick at the end
of the tail.
3. Apply a small amount of pressure to the tail proximal to the
pinprick and slide fingers toward the pinprick to form a droplet
of blood.
4. Apply the blood drop to a blood glucose meter. Typically, less
than 5 μl is required for most meters.
5. If excessive weight loss is evident, insulin can be administered
to comply with local ethical requirements.
3.4.2 Intraperitoneal 1. Fast mice for at least 6 h (see Note 12).

Glucose Tolerance Test 2. Measure blood glucose as described above.
3. Inject 2 g/kg glucose solution in saline (30% glucose)
intraperitoneally.
4. Measure blood glucose 15, 30, 60, 90, and 120 min after
injection.
3.4.3 Serum Insulin 1. Fasting, nonfasting or glucose-induced plasma insulin concen-

trations can be measured.
2. Using a 27 G or 25 G needle, prick the end of the tail.
3. Apply pressure to form as large a blood drop as possible and use
a heparinized capillary tube to collect blood by capillary action.
Use a rubber bulb to inject the collected blood into a 0.5-ml
Eppendorf tube on ice.
4. Repeat until the required volume is collected.
5. Centrifuge blood samples within 1 h at 1500 G at 4 C for
10 min. Aspirate the serum and freeze at 20 C for later
analysis.
6. Insulin can be measured in small serum volumes (5–10 μl)
using ELISA.
3.5 Graft Retrieval 1. Anesthetize the mouse with isoflurane or another appropriate
anesthetic.
3.5.1 Nephrectomy
to Confirm the Role 2. Shave the fur on the left flank and treat the skin with surgical
of the Graft on Glycemia scrub, followed by 70% ethanol.
3. Administer analgesic.
and check the mouse is adequately anesthetized.
5. Using a scalpel, make a 1.5–2 cm incision in the skin and using
scissors cut a 1-cm incision in the peritoneum over the left
kidney.
6. Use scissors to free the kidney from surrounding tissues (being
very careful not to cut the renal artery or vein).
7. Externalize the kidney.
8. Clamp the renal artery, renal vein and ureter close to the kidney
with a hemostatic clamp.
9. Use a scalpel to cut the kidney free. Place the kidney into
appropriate solution for analysis (e.g., formalin for histology
or acid alcohol for insulin content detection).
10. Ligate the renal artery, renal vein and ureter by tying two to
three times with suture.
11. Remove the hemostatic clamp and monitor whether there is
any bleeding (see Note 13).
12. Suture the peritoneum and suture the skin.
13. Monitor the recovery of the mouse.
14. Measure blood glucose daily until stable hyperglycemia has
been reached (at least three consecutive days of blood glucose
concentrations greater than 20 mM). The mouse should then
be killed and if necessary, the pancreas removed for analysis.
3.5.2 Graft Retrieval 1. If confirmation of return to hyperglycemia is not required or

When Confirmation not relevant (e.g., in mice that have not become normoglyce-
of Hyperglycemia Is Not mic), the graft can be retrieved after the mouse is killed by
Required cervical dislocation.
2. Remove the graft-bearing organ (liver or kidney) as soon as
possible after the mouse is killed and place as soon as possible in
the appropriate solution for analysis. The pancreas may also be
removed for analysis if appropriate.
3. Alternatively, the mouse can be terminally anesthetized and the
kidney removed (see Note 14).
4 Notes
1. Make sure that all mice are weighed and marked, with the dose
for each mouse calculated, prior to mixing the streptozotocin
with the citrate buffer to allow injection as soon as possible
after reconstitution (we generally try to inject all mice within
10 min of making the streptozotocin [20]).
2. The dose may need to be adjusted due to variation in the
streptozotocin lot and/or mouse strain [20, 21]. As glucose
directly competes with streptozotocin, fasting the mice may
increase percentage of mice from becoming diabetic.
3. Depending on your local ethical guidelines, mice may require
to be treated by insulin (we use 2–3 units daily of Caninsulin)
to prevent too much weight loss (more than 15% within 3 days
or more than 20% overall). If weight loss exceeds this, mice
should be killed. Cages will need to be changed more regularly
due to polyuria. In addition, water supply should be checked
daily as the mice will drink considerably more than normal. If
mice do not become diabetic, they can be re-injected again
with streptozotocin (although it is recommended that this is
only done once).
4. It is easier to flip the liver back if you clip off the xiphysternum.
5. Clamping should be carried out with precision; if the clamping
is not adequate, collagenase will leak into the duodenum. If the
clamp is too far down the pancreas, some ducts will be occluded
and thus the pancreas will not be fully inflated.
6. The connective tissue around the duct can be gently removed
by running the forceps up and down around the duct. This
allows the duct to be easily visualized. The forceps can also be
placed under the duct to form a support for injection. When
getting trained in the technique, it is desirable to initially inject
as closely as possible to the liver end of the duct, allowing the
entire length of the duct to be recannulated if first attempts are
initially not successful.
7. A slow acceleration and no brake on the centrifuge setting

prevent the histopaque from mixing with the media during
the centrifugation stage.
8. Islets should be collected from the histopaque gradient as soon
as the centrifugation has been finished to reduce exposure, as
histopaque is toxic to islets.
9. Note that at this stage the pellet is usually not as stable, so
rather than just pouring out the supernatant, it is suggested to
pipette it out leaving about 5 ml left at the bottom of the tube.
10. Spongostan should be split into half layers before starting the
operation to make sure that they are not too thick. A second
piece of Spongostan can be applied if the operator is concerned
that there may be excess blood loss. It is important that the
Spongostan and cotton bud used to apply pressure are dry,
allowing optimal absorption.
11. After the first week, blood glucose concentrations can be
measured less frequently as the mouse is stable and not
showing decline in weight. If the mouse is losing weight
and/or is overtly diabetic then they should continue to be
monitored daily.
12. Although many laboratories fast mice overnight prior to a
glucose tolerance test, it has been suggested that this is too
long and induces metabolic stress [22].
13. Before releasing the clamp, keep hold of the suture that was
used to ligate the vessels and the ureter. If bleeding is evident
on releasing the clamp, reclamp below the suture and repeat
the ligation.
14. In diabetic mice, the graft degranulates, making it more trans-
parent and therefore more difficult to identify. If the mouse is
terminally anesthetized, the graft can be easily identified while
there is still blood circulation. In addition, it is helpful to note
the position of the graft at the time of transplantation.
References
1. Bruni A, Gala-Lopez B, Pepper AR, Abualhas- microcapsules does not improve graft outcome
san NS, Shapiro AJ (2014) Islet cell transplan- at the subcutaneous site. Artif Organs
tation for the treatment of type 1 diabetes: 36:564–570
recent advances and future challenges. Diabe- 4. Estil les E, Tellez N, Escoriza J, Montanya E
tes, Metab Syndr Obes 7:211–223 (2012) Increased beta-cell replication and
2. Zhi ZL, Kerby A, King AJ, Jones PM, Pickup beta-cell mass regeneration in syngeneically
JC (2012) Nano-scale encapsulation enhances transplanted rat islets overexpressing insulin-
allograft survival and function of islets trans- like growth factor II. Cell Transplant
planted in a mouse model of diabetes. Diabe- 21:2119–2129
tologia 55:1081–1090 5. Fernandes JR, Duvivier-Kali VF, Keegan M
3. Kerby A, Bohman S, Westberg H, Jones P, et al (2004) Transplantation of islets trans-
King A (2012) Immunoisolation of islets in duced with CTLA4-Ig and TGFbeta using
high guluronic acid barium-alginate
adenovirus and lentivirus vectors. Transpl 14. King A, Bowe J (2016) Animal models for
Immunol 13:191–200 diabetes: understanding the pathogenesis and
6. King A, Lock J, Xu G, Bonner-Weir S, Weir GC finding new treatments. Biochem Pharmacol
(2005) Islet transplantation outcomes in mice 99:1–10
are better with fresh islets and exendin-4 treat- 15. King AJ, Austin AL, Nandi M, Bowe JE (2016)
ment. Diabetologia 48:2074–2079 Diabetes in rats is cured by islet transplanta-
7. McCall M, Toso C, Emamaullee J et al (2011) tion... but only during daytime. Cell Trans-
The caspase inhibitor IDN-6556 plant 26(1):171–172
(PF3491390) improves marginal mass engraft- 16. Mathews CE, Langley SH, Leiter EH (2002)
ment after islet transplantation in mice. Sur- New mouse model to study islet transplanta-
gery 150:48–55 tion in insulin-dependent diabetes mellitus.
8. King AJ, Guo Y, Cai D et al (2013) Sustained Transplantation 73:1333–1336
NF-kappaB activation and inhibition in beta- 17. Hong J, Yeom HJ, Lee E et al (2013)
cells have minimal effects on function and islet Islet allograft rejection in sensitized mice is
transplant outcomes. PLoS One 8:e77452 refractory to control by combination therapy
9. Toso C, Serre-Beinier V, Emamaullee J et al of immune-modulating agents. Transpl Immu-
(2008) The role of macrophage migration nol 28:86–92
inhibitory factor in mouse islet transplantation. 18. Krautz C, Wolk S, Steffen A et al (2013) Effects
Transplantation 86:1361–1369 of immunosuppression on alpha and beta cell
10. Rackham CL, Chagastelles PC, Nardi NB, renewal in transplanted mouse islets. Diabeto-
Hauge-Evans AC, Jones PM, King AJ (2011) logia 56:1596–1604
Co-transplantation of mesenchymal stem cells 19. Chen X, Zhang X, Chen F, Larson CS, Wang
maintains islet organisation and morphology in LJ, Kaufman DB (2009) Comparative study of
mice. Diabetologia 54:1127–1135 regenerative potential of beta cells from young
11. Pawlick R, Gala-Lopez B, Pepper AR, and aged donor mice using a novel islet trans-
McCall M, Ziff O, Shapiro AM (2014) The plantation model. Transplantation
combination of anti-NKG2D and CTLA-4 Ig 88:496–503
therapy prolongs islet allograft survival in a 20. Leiter EH, Schile A (2013) Genetic and phar-
murine model. Am J Transplant macologic models for type 1 diabetes. Curr
14:2367–2374 Protoc Mouse Biol 3:9–19
12. Bohman S, Andersson A, King A (2006) No 21. Sakata N, Yoshimatsu G, Tsuchiya H, Egawa S,
differences in efficacy between noncultured Unno M (2012) Animal models of diabetes
and cultured islets in reducing hyperglycemia mellitus for islet transplantation. Exp Diabetes
in a nonvascularized islet graft model. Diabetes Res 2012:256707
Technol Ther 8:536–545 22. McGuinness OP, Ayala JE, Laughlin MR, Was-
13. Deeds MC, Anderson JM, Armstrong AS et al serman DH (2009) NIH experiment in centra-
(2011) Single dose streptozotocin-induced lized mouse phenotyping: the Vanderbilt
diabetes: considerations for study design in experience and recommendations for evaluat-
islet transplantation models. Lab Anim ing glucose homeostasis in the mouse. Am J
45:131–140 Phys Endocrinol Metab 297:E849–E855
Chapter 17
Measurement of Insulin Secretion Using Pancreas Perfusion

in the Rodent
Edward T. Wargent
Abstract
Under in vivo conditions, the study of physiological and pharmacological functions of an organ is difficult
due to whole-body interactions with the organ. Thus, an in vitro technique for the perfusion of isolated
pancreata was developed for physiologic and response studies including the investigation of endocrine
function and secretory responsiveness under a variety of diabetes-associated conditions. The pancreas is
isolated from the connecting spleen, stomach, and duodenum and transferred to a pre-warmed chamber
where it is perfused in isolation from all other organs. A detailed description of the isolation of pancreata
from rats and mice and the perfusion apparatus is described, as well as the measurement of glucose-
stimulated insulin secretion using an in-house-developed radioimmunoassay.
Key words Pancreas, Perfusion, Insulin, Radioimmunoassay
1 Introduction
In vivo, the study of physiological and pharmacological functions of

an organ is difficult due to whole-body interactions with the organ.
However, during perfusion, the respective organ is isolated from
these interactions and placed under relatively simple experimental
conditions, approximating the normal state, thereby facilitating
physiological measurements. The pancreas perfusion method is
therefore ideally suited for mechanistic studies examining the
effects of secretagogues and pharmacological agents.
In contrast to studies looking at secretagogue responses in
isolated islet preparations, the isolated whole pancreas perfusion
method allows for the study of both endocrine and exocrine tissue
function simultaneously as this procedure maintains physiological
integrity at both cellular and whole organ levels. There is also a
possibility of a control and experimental period of study in the same
pancreas preparation. Such control in other in vitro settings
requires multiple incubations. Factors affecting insulin secretion
are of great importance in the study of type 2 diabetes, and the
281
isolated pancreas is an invaluable technique of studying insulin

secretion from the whole pancreas. The perfused pancreas can
also be used to evaluate the effects of drugs and secretagogues
that effect pancreatic function. For example, amylin, a protein
secreted by the pancreas and a major constituent of islet amyloid
deposits in patients with type 2 diabetes, while having no effect on
glucose-stimulated insulin secretion, significantly inhibits arginine-
stimulated insulin secretion.
The rodent is anesthetized by an appropriate means and both a
lateral and midline incision made to expose the stomach, intestines,
and mesenteric blood vessels. The pancreas is carefully separated
from the overlaying colon, spleen, and connective tissues. The
jejunum and its blood supply are ligated so that this distal section
can then be removed. All other attachments of the pancreas to the
descending colon are then ligated and cut, as are the splenic vessels.
Once the mesenteric artery and celiac axis have been identified, the
stomach is moved aside and the exposed esophagus and gastric
artery are also ligated. The pancreas is isolated from the stomach
following ligation of the duodenum and the vascular connections
on the posterior wall exposed, which can then be ligated subse-
quently. By careful dissection, the aorta can be exposed, cleared
from the inferior vena cava between the renal artery and superior
mesenteric artery, and ligated. Loose ligatures are placed around
the hepatic portal vein where it leaves the pancreas for the liver.
Once the celiac and mesenteric branches have been identified along
the aorta, the aorta can be freed from the vena cava to the ligature
below the renal artery. Loose ligatures are positioned around the
aorta just below the diaphragm preserving the origin of the celiac
artery, which holds the inlet cannula, while the outlet cannula is
placed in the portal vein. Once the abdominal aortic ligature is
tightened, the pancreatic circulation can be completely isolated by
tightening the ligature around the lower vena cava and perfusion
can commence.
Based on the original apparatus used by [1], perfusate is
pumped from a reservoir by means of a peristaltic pump, into an
enclosed organ perfusion chamber, and enters the pancreatic circu-
lation through the celiac arterial cannula. The pancreas is kept
moist with normal saline and perfused using a modified Krebs-
Ringer bicarbonate-buffered solution containing glucose and
bovine serum albumin. The production of insulin by the prepara-
tion reflects its responsiveness to glucose and other secretagogues
and can be collected by means of perfusate via the portal vein.
This methodology has been applied to a range of investigations
studying pancreatic function with relation to diabetes covering
physiological functionality in response to endogenous compounds
such as nutrients (e.g., glucose-induced insulin, somatostatin, and
orexin-A secretion) or the effects of fatty acids or hormones (e.g.,
amylin, GLP-1, and orexin-A) on the stimulation of insulin release,
Measurement of Insulin Secretion Using Pancreas Perfusion in the Rodent 283
the inhibition of glucagon secretion or the leptin inhibition of basal

insulin release [2–5]. The prediabetic state or diabetic state may be
investigated using rodent models such as the Zucker fatty ( fa/fa)
or Zucker diabetic fatty (ZDF) rats. Furthermore, the isolated
perfused pancreas is also a useful method for investigating potential
therapeutic candidates, most notably those designed to enhance
insulin secretion, such as the GLP-1 receptor antagonists [6].
This chapter describes in detail the ex vivo perfusion technique
for the rodent pancreas and has been updated from the previous
edition to include specific materials and methods required for both
rats and mice (although it should be noted that because of the
complexities of scale the technique in mice has a higher variability
and generally requires a higher number of perfusions per group
than rat studies). When regarding the 3Rs, it is therefore preferable
to preform studies in rats when possible. It presents a detailed
description of the perfusion technique, a study of some of the
functions of the preparation and an observation of the pancreatic
response to glucose as an example of secretagogue stimulation of
insulin secretion. The development of sensitive radioimmunoassays
of biological fluids has enabled the quantitative measurement of
hormone secretion using pancreas perfusion techniques. This chap-
ter will also describe the important salient features and optimiza-
tion stages required to develop an in house radioimmunoassay for
use with this methodology.
2 Materials
2.1 Pancreas 1. Anesthetic and analgesic regime capable of maintaining a level

Excision of anesthesia suitable for major surgery for 40 min, for exam-
ple, isoflurane (see Note 1).
2. Blunt-nosed straight-dissecting scissors 14–15 cm (FST, cat #
14000-14) (see Note 2).
3. Standard pattern forceps 14–15 cm (FST cat # 11000-14).
4. Blunt-nosed curved-dissecting scissors 14–15 cm (FST cat #
14003-14).
5. Straight hemostat 14 cm (FST cat# 13004-14).
6. Saline: 0.9% NaCl(aq).
7. Black-cotton thread cut into approximately 10 cm lengths.
8. Pair of curved-eye dressing forceps 10 cm (FST cat # 11051-
10).
9. Blunt-nosed straight dissecting scissors 10 cm (FST cat #
14078-10).
10. Cotton buds.
11. Pair of serrefines 35 mm (FST cat # 18051-35).
50
A
Turret
15
C
D
10
250
20
120
80
C
B
The turret supports two PVC sheathed copper wires.
It is formed from a tube drilled at 90` to the axis to
accommodate the wires as above. The tube is then
secured with a nylon centre bolt into the top plate.
Fig. 1 Perfusion tank setup. The sheathed copper wires are attached to the two cannulae that are connected to
the celiac artery and hepatic portal vein. The pancreas is housed in the central pool surrounded by perfusion
buffer. (a) inlet and (b) outlet for heated water keeps the central pool at 38 C. (c) inlet and (d) outlet for tubing
transporting the perfusate travels through the heated water re-heating it to perfusion temperature. All
dimensions are expressed in millimeters
12. Sharp straight spring scissors 10 cm (FST cat # 15024-10).

13. Polyethylene tubing with a bore of 0.5 mm, wall 25 mm
(800/100/160, Scientific Laboratory Supplies, Nottingham,
UK, cat # TUB3660).
14. 2.5-ml syringe and needle.
2.2 Pancreas 1. Perfusion tank (see Fig. 1).

Perfusion 2. 20-ml glass scintillation vials and silicon caps.
3. Water bath.
4. Silicone tubing (Scientific Laboratory Supplies, Nottingham,
UK).
5. Three-way male lock stopcocks (Cole-Palmer, Illinois, USA).
6. Water heater/pump.
7. Peristaltic pump.
8. Manometer.
9. Sample collector.
10. Electronic thermometer.
11. Oxygen: carbon dioxide (95:5) cylinder.
12. Perfusion buffer: (5.7 mM KCl, 1.5 mM KH2PO4, 113 mM
NaCl, 1.2 mM MgSO4, 2.3 mM calcium gluconate, 25 mM
NaHCO3, 3% (w/v) dextran T40, 1% human albumin,
pH 7.4). The buffer should be filtered and oxygenated steadily
for 30 min before adjusting the pH to 7.4.
13. Glucose.
14. Arginine.
15. Millex HA 0.45 μm filter, (Millipore, Massachusetts, USA).
2.3 Insulin 1. Polystyrene round-base test tubes capable of holding 2 ml

Radioimmunoassay (e.g., 64 11 mm) (see Note 3).
2. Gamma counter, for example, PerkinElmer (Waltham, MA,
USA).
3. [125I]-insulin (diluted 10,000 cpm per 100-μl assay buffer).
4. Rat insulin standard.
5. Rat insulin antiserum.
6. Secondary antiserum (type determined by the species used to
generate the insulin antisera).
7. 15% polyethylene glycol, MW 8000 (PEG-8000): This sepa-
rates bound from unbound antigen by precipitating the anti-
body–antigen complex (see Note 4).
8. Assay buffer: (50 mM NaH2PO4, 150 mM NaCl, 0.5% BSA
(RIA grade), 15 mM NaN3, pH 7.4). Sera and PEG should be
prepared in this buffer while standards should be prepared in
the perfusion buffer.
3 Methods
The purpose of this surgical procedure is to create an enclosed

system containing an intact pancreas with one inlet (the celiac
artery) and one outlet (the hepatic portal vein) for perfusion, with
all other blood vessels ligated to prevent leakage. For ease of
operation, rat stomach and spleen are excised with the pancreas
but are then excluded from the perfusion system by ligation of the
connecting blood vessels. With smaller rodents, the appropriate
vessels are ligated to isolate the pancreas but remain attached and
the whole body placed in the perfusion chamber. Prior to excision,
the pancreas is perfused with saline and perfusion buffer to flush
out the blood so that clots do not form which would compromise
the system. An anticoagulant such as heparin may be introduced
into the blood system after laparotomy to achieve this but may
require a greater number of perfusions to achieve a similar quality
of data (probably due to the effects of heparin on ATP-associated
signaling). This perfusion method is preferred unless the procedure
is taking longer than 30 min or is performed on smaller rodents
such as mice (see Note 5).
3.1 Pancreas 1. Anesthetize the rat with an aesthetic regimen capable of

Excision providing a degree of surgical anesthesia and analgesia for
40 min (see Note 1).
2. During the surgical period, keep the rat warm by means of a

heated pad or heated operating table.
3. Perform a laparotomy using blunt-nosed dissecting scissors
(14–15 cm is the optimal length for most users). First make
an incision in the skin large enough for the same scissors to be
inserted between the skin and abdominal wall and then using
the scissors tease apart the skin from the abdominal wall so that
the skin can be cut from pubis to sternum along the
mid-central line. Separate the skin and abdominal wall down
the sides (a curved pair of scissors of similar size would be
suited to this task) and cut the skin laterally, avoiding blood
vessels as much as possible. Lift the abdominal wall with forceps
and cut along the mid-central line between the pubis and
xyphoid process being careful not to nick the organs below.
Cut down each side of the abdominal wall after having
occluded the blood vessels with the use of an hemostat (cauter-
izing works equally well).
4. Place pieces of saline-soaked tissue paper immediately to the
left of the subject and place the intestines carefully on top of
these. Pour saline over the organs of the abdomen and repeat
periodically to prevent the organs from drying. At this point, an
anticoagulant such as heparin may be utilized by means of
injection via the inferior vena cava.
5. Doubly ligate the descending colon and cut between the
ligatures.
6. Use the upper ligature to raise the colon and cut the fascia
underneath up to the transverse colon.
7. Tease away the top layer of fascia connecting the intestines and
pancreas using a blunt tool being careful not to damage the
pancreas (fingers are ideal for this task).
8. The bottom layer of fascia contains jejunal blood vessels that
will need to be ligated. First, remove the fascia between the
vessels and ligate the vessels to the right of the transverse colon
and then those of the mesenteric radix.
9. Ligate the duodenum adjacent to the ligature on the mesen-
teric radix and cut the intestines below the ligatures (see Note
6). The lower intestines can now be discarded.
10. Tease apart the fascia linking the pancreas, stomach, and spleen
and cut the vessels connecting the stomach and spleen after
double ligation.
11. Remove the fascia connecting the stomach, liver, and esopha-
gus to expose the esophagus and esophageal artery. Doubly
ligate these and cut.
12. Isolate the pancreas from the stomach by ligating the duode-
num adjacent to the pylorus.
13. Remove the fat from around the descending aorta between the
diaphragm and aortic bifurcation so that the aorta, caudal vena
cava, and any branching vessels are clearly exposed (see Note 7).
14. Using blunt-ended forceps, separate the entire exposed length
of the aorta from the body wall as well as the initial centimeter of
the celiac, superior mesenteric, and renal arteries (see Note 8).
15. Separate a section of aorta from the vena cava between the renal
and iliac arteries using blunt-ended forceps (see Note 9).
16. Lift the liver to the right and clear the hepatic portal vein of any
connective tissue. Place an untied ligature around the vein and
replace the liver exposing the celiac trunk.
17. Place untied ligatures around the aorta superior to the celiac
artery, inferior to the superior mesenteric artery, and between
the renal and iliac arteries. Leave several millimeters between
the celiac artery and the ligature for ease of manipulation of the
inlet cannula and also to avoid risk of cutting the celiac when
severing the aorta.
18. Place untied ligatures around the celiac, mesenteric, and renal
arteries (see Note 10).
19. Place a serrefine (see Note 11) across the aorta immediately
inferior to the renal arteries and another at the aortic bifurca-
tion. Using spring scissors make a cut into the aorta halfway
between the serrefines.
20. Insert a cannula for connection to a saline-containing syringe
into the aorta up to the superior serrefine and secure by tying
the ligature. For rats, the cannula should be polyethylene
tubing (as described in materials and methods) tapered for
easy insertion. Smaller rodents require a needle—rounded so
it can be inserted into the aorta. With experience, a 25-G
needle is possible, especially, in rodents with a lean body weight
over 30 g. A smaller gauge needle, for example, 27 G, may be
necessary for smaller animals or when learning the technique,
but increases the risk of leaks.
21. Remove the superior serrefine and tie the ligatures around the
renal and mesenteric arteries.
22. Tie the uppermost ligature around the aorta and immediately
perfuse the system with 2.5-ml saline over a period of about
20–30 s. Replace the saline syringe with one containing the
perfusion buffer and perfuse with 2.5 ml over 20–30 s. Imme-
diately, tie the remaining aortic ligature followed by the hepatic
portal ligature. Tie the latter ligature as close to the liver as
possible.
23. Cut the hepatic portal vein on the liver side of the ligature and
cut the aorta on the celiac trunk side of both the top two
ligatures. Lift out the section of abdominal organs, cutting
any remaining underlying connective tissue and transfer to the

perfusion tank (see Note 12).
24. Dispose of the rest of the carcass by an appropriate means.
3.2 Ex Vivo 1. Locate the ligature around the celiac artery and hepatic portal
Preparation vein, lift and replace in the center of the tank with the two
for Pancreas Perfusion blood vessels exposed.
2. Maneuver the inlet cannula next to the celiac trunk so that it is
as almost horizontal. Holding the opening of the aorta adja-
cent to the celiac artery, but on the opposite wall, slip the
cannula into the celiac trunk so that the cannula enters the
celiac artery via the aorta. Secure in place by tying the ligature
around the celiac artery and the cannula. The celiac artery
cannula can be fashioned from a blunted 23-G needle for rats
(or no smaller than 25 G). Celiac cannulas for smaller rodents
should already be in place at the surgery stage and now
connected to the inlet tubing.
3. Maneuver the outlet cannula vertically next to the hepatic
portal vein. The outlet cannula should have an outside diame-
ter of 2–3 mm for rats and 5–6 mm for mice. Glass capillary
tubes with the end tapered and melted work well for rats—just
make sure the insertion angle is as horizontal as possible to
avoid the cannula slipping off. A blunted 25-G or 23-G needle
works well for mice. Lift the vein by the ligature and make an
incision into the vein as close to the ligature as possible (see
Note 13). Holding the thread of the ligature in one hand and
the lip of the aperture in the vein with forceps, slide the cannula
into the vein by almost a centimeter. Tape the ligature threads
to the cannula holder so the vein does not slip off and secure a
ligature around both the vein and cannula (see Note 14).
4. Perfuse the pancreas at a rate of 2 ml per gram pancreas weight
per minute until the perfusate begins to flow out and then
perfuse at 2 ml per gram pancreas weight per minute for
20 min to flush out any secretagogues released into the system
during the set-up period (see Note 15).
3.3 Pancreas 1. Set the perfusion equipment up as shown in Fig. 2. The tem-
Perfusion perature of the water bath and water heater/pump must main-
tain the perfusion buffer and perfusion tank temperature at
38 C. Check the perfusion tank temperature prior to surgery
and monitor throughout the experiment (see Note 16).
2. Glucose is required in the perfusion buffer to maintain the
integrity of the pancreas. The perfusion system can detect an
increase in insulin secretion at 4.4-mM glucose. 2.8-mM glu-
cose is used to maintain a basal level of insulin secretion and
pancreas function. 5.6-mM glucose represents the normal level
< 100 mmHg

Air traps
Manometer
Three-way tap
T-junction
Celiac artery cannula
Peristaltic
pump Perfusion tank
Perfusion Water bath
buffer
Fig. 2 Perfusion apparatus. The perfusion buffers are heated in a water bath and connected to air traps. Tubing
connects the air traps to a tap, the number and type depending upon the number of different perfusates to be
analyzed. The system powered by a peristaltic pump; immediately after which the perfusion tank is connected.
A T-junction separates the perfusate into two tubes. One is attached to the cannula inserted into the celiac
artery and the other connected to a manometer
of glucose exposure to the pancreas in a free-fed state. Diabetes

is defined when glucose appears in the urine and occurs when
blood concentrations rise above 11 mM. Concentrations
higher than this are important to include when looking at the
diabetic state.
3. Arginine stimulates insulin secretion by a mechanism indepen-
dent from that induced by glucose. Use 20 mM to demonstrate
integrity of the system when investigating mechanisms where
glucose-induced insulin secretion may be impaired.
4. The sample collector should be set to collect samples over a
1-min period for most experiments. While the perfusion is in
progress, samples already collected can be aliquoted in tripli-
cate into tubes ready for radioimmunoassay.
5. Avoid designing perfusions exceeding 100 min in length,
though this will not include the initial 20-min “flush-out”
period.
6. Check the pH of the perfusion buffer at the end of the perfu-
sion as pH 7.4 is outside the optimal buffering capacity of
bicarbonate buffers (pH 5.1–7.1).
7. Monitor the pressure in the perfusion system continuously
with the manometer. A pressure rise above 100 mmHg is
indicative of a blockage in the pancreas resulting in swelling
followed by leakage. Should this occur, the perfusion should be
terminated and the collected perfusate discarded.
3.4 Insulin In a radioimmunoassay, a fixed concentration of labeled tracer

Radioimmunoassay antigen is incubated with a constant amount of specific antiserum
such that the concentration of antigen binding is limiting. If unla-
beled antigen is added, there is competition between labeled tracer
and unlabeled antigen, so that the amount of tracer bound to the
antibody will decrease as the concentration of unlabeled antigen
increases. This can be measured after separating antibody bound
from free tracer and counting the bound fraction, the free fraction
or both.
Interassay controls are included in each assay system. These
samples should be of identical composition to the material being
assayed, from the same species and medium. These are used to
validate the assay and to measure the variation within and between
assays. The number of these required should be determined statis-
tically during the method evaluation. The interassay drift must be
determined from assay tubes containing identical amounts of
reagents (sample, primary antibody, tracer, and secondary anti-
body). Duplicate sample pools should be placed at the beginning
and end of the assay to determine if drift is a problem.
3.4.1 Setting Up An 1. Determine the primary antibody dilution required. Set up

In-House three tubes each for total counts (T), non-specific binding
Radioimmunoassay (NSB), and at each dilution of primary antibody binding (B0,
a range of 1000- to 1,000,000 fold is usually ample, (see Note
17). Proceed with Subheading 3.4.2 from step 3. Plot the data
as % ((Bo-NSB)/(T-NSB)) versus primary antibody dilution
factor (log10 scale) using a sigmoidal curve fit (Fig. 3). The
primary antibody dilution required corresponds to 30% ((Bo-
NSB)/(T-NSB)) (see Note 18).
2. Determine the secondary antibody titer by setting up three
tubes each for total counts, non-specific binding, and each
dilution of secondary antibody. Proceed with Subheading
3.4.2. Plot the data as % ((Bo-NSB)/(T-NSB)) vs. secondary
antibody dilution factor (log10 scale) using a sigmoidal curve fit
(Fig. 4). The secondary antibody dilution required is the lowest
that gives maximal precipitation (see Note 19).
3. Validate the assay with a serial dilution of perfusate containing
insulin. Either concentrate the perfusate or add enough rat
insulin to make a 10 nM solution and serially dilute for a
range from 1 pM to 10 nM. Perform Subheading 3.4.2 from
step 3 and compare the range to the standard curve. Express
the concentrations relative to the top concentration or stock
solution to ascertain whether the regression curves are verti-
cally parallel. If the dilution curves are parallel then the assay is
valid to analyze insulin secretion (Fig. 5).
100
90
80
% ( Bo- N S B ) / ( T- N S B )
70
60
50
40
30
20
10
0
1000 10000 100000 100000
dilution factor
1st Ab dilution factor = 1 in 86,955
Fig. 3 Primary antibody titer test. The primary antibody will bind to the hormone
of interest as well as the labeled tracer and standard. The bound fraction of the
assay is that part of the labeled and unlabeled antigen specifically bound to the
primary antibody and counted in the gamma counter. The primary antibody titer
test is a preliminary radioimmunoassay to determine what concentration of
primary antibody should be used. The amounts of tracer and secondary
antibody remain constant between groups of NSB’s and 100% tubes, while
only the primary antibody concentration in the Total tubes varies. Typically, three
totals, three NSBs, and three 100%s for each concentration of primary antibody
are tested. The concentration selected depends on the sensitivity desired in the
assay. The least primary antibody used, the lower the standard curve doses can
be reliable. In this example, the final dilution of primary antisera corresponds to
the dilution of the original stock that precipitates 30% of hot antigen added to the
tubes (minus non-specific precipitation). Here this corresponds to a final dilution
of 1:86,955 of the original stock solution. Note that the primary incubation
contains 100 μl of the standard/sample, 100 μl of label and 100 μl of the
primary antibody so in this example the primary antibody would be diluted
1:28,985 dilution (final dilution 1:86,955)
3.4.2 1. Generate radioimmunoassay standard curves from nine con-

Radioimmunoassay centrations in quadruplicate. Ensure three standards fall on the
Protocol steep part of the curve, and three standards at or near both the
top and the bottom of the curve. When analyzing insulin
secretions, from the pancreas perfusion, use the following con-
centrations: 0.02, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, and 10 nM (see
Fig. 6). Add 100 μl of standard to the appropriate tubes.
2. In addition to the samples and standards include total counts
(T), non-specific binding (NSB), maximum binding (B0), and
variance determinants (see Note 20). Set up at least twenty
variance tubes, ten at the start and end of the assay. Set up all
other three groups in quadruplicate.
100
90
80
% (B o- N S B ) / ( T- N S B )
70
60
50
40
30
20
10
0
10000 20000 30000 40000 60000 100000
dilution factor
2nd Ab dilution factor = 1 in 30,000
Fig. 4 Secondary antibody titer test. The secondary antibody binds to the primary
antibody/antigen complex (bound fraction) and after centrifugation, is
precipitated. After the free supernatant is decanted, the assay tube and
precipitate are counted in the gamma counter. A secondary antibody titer test
is a preliminary radioimmunoassay to determine what concentration of
secondary antibody maximally precipitates the bound fraction of the primary
antibody/antigen complex. In this test, the amount of tracer and primary antibody
remain constant while only the secondary antibody concentration varies.
Typically, there are three total count tubes, and for each different amount of
secondary antibody, three NSBs and three 100% tubes. In this example, the final
dilution of secondary antisera corresponding to the lowest concentration that will
still maximally precipitate the label is 1:30,000 would be used
3. Add 100 μl [125I]-insulin solution to all tubes and mix by

vortexing.
4. Add 100 μl of primary antisera to all tubes except T and
NSB (add 100 μl assay buffer to these) and mix thoroughly
(see Note 21).
5. Cover and incubate at 4 C for 24 h.
6. Add 100 μl of secondary antisera and incubate for 2 to 24 h.
7. Add 100 μl of PEG solution (see Note 22).
8. Add 1 ml distilled water at 4 C to each tube, mix and leave for
10 min.
9. Centrifuge all tubes for 10 min at 3000 g.
10. Remove the supernatant and count each tube for a minimum
of 1 min with a gamma counter (see Fig. 7).
100
75
% Bo
50
25
0
1 2 5 10 20 50 100 200 500
dilution factor
rat insulin dilution factor
perfusate dilution factor
Fig. 5 The parallelism test. The parallelism test is performed to validate the assay or sample material. A serial
dilution of a pooled sample is compared to varying standard curve doses within the same assay. Lines of equal
slope demonstrate no significant proportional analytical error within these ranges. This allows the assay of
unknown samples with the confidence that the dose estimate per/ml will not have a significant error. This is
important when unknowns are run in the same assay at different volumes
110
100
90
80
70
60
% Bo
50
40
30
20
10
0
2 5 10 20 50 100 200 500 1000
[insulin] (fmoles)
Fig. 6 Standard curve for the insulin radioimmunoassay. A calibration or standard curve is set up with
increasing amounts of known antigen, and from this curve, the amount of antigen in the unknown samples can
be calculated. The four basic necessities for a radioimmunoassay are an antiserum to the compound to be
measured, the availability of a radioactively labeled form of the compound, a method whereby antibody-bound
tracer can be separated from unbound tracer, and a standard unlabeled material. Each circle represents a
single data point and a sigmoidal regression line generated
2.7 mM 11.1 mM 2.7 mM 20 mM

glucose glucose glucose arginine
1000
insulin (fmoles/min)
100
10
0
0 10 20 30 40 50 60 70 80 90
time (minutes)
Fig. 7 Example of a typical insulin secretion profile from perfused pancreata of

Wistar rats (n ¼ 6). Basal insulin secretion was measured with 2.7 mmol/l
glucose in the pefusate and secretagogue-stimulated insulin secretion was
measured with 11.1 mmol/l glucose or 20 mmol/l arginine in the perfusate
4 Notes
1. Isoflurane is suitable for all rodents and is preferable to inject-

able anesthetics because of the degree of control it allows and
lack of respiratory problems. The fentanyl/fluanisone and mid-
azolam combination described is usually an acceptable inject-
able regime due to low strain variability in duration of
anesthesia. The duration of anesthesia of 40 min described in
the text is conservative—the procedure typically lasts
20–30 min.
2. All surgical tools described may be obtained from Fine Science
Tools and the catalogue numbers are listed (website www.
finescience.com).
3. Borosilicate tubes should be used for radioimmunoassays
where peptides stick to tubes made from polystyrene or poly-
propylene, for example, GLP-1. Borosilicate tubes should also
be used if the pellet formation is inadequate.
4. Some commercial products are a combination of secondary
antibody and precipitating agent, making this step quicker
and easier. If a commercial secondary antibody/precipitator
mix is used, then follow the manufacturer’s instructions.
5. For rodents smaller than rats, it is advisable to use the antico-
agulant method to prevent clotting rather than pre-perfusion
via the aorta. Also, the perfusion should be performed in situ

with the whole animal in the perfusion tank rather than excis-
ing the isolated pancreas.
6. A double ligature is preferable here, as an incomplete ligation at
this site is the most common cause of leakage during the
perfusion stage.
7. The left gastric and iliac arteries are easily disrupted and so the
area must be cleared with care. Cotton buds are useful in
exposing this area, but the entire length of these arteries need
not be exposed as long as the coeliac trunk is cleared of fat and
connective tissue.
8. The right renal artery may prove difficult to access, if so this
ligation may be missed, but it will minimally affect the initial
perfusion step. In some strains of rat, the left and right renal
arteries arise from the same junction of the aorta and may be
ligated together.
9. The connective tissue between the aorta and vena cava may be
difficult to remove but do not be tempted to use sharp imple-
ments as the vena cava is easily disrupted. Should the vena cava
be pierced quickly clamp both the aorta and vena cava where
the aorta would be clamped in step 17, remove traces of blood,
and proceed as rapidly as possible to the initial perfusion step.
In this way, blood clots in the pancreas may possibly be
avoided.
10. Ensure that all loose ends of all untied ligatures are untangled
and easily identified as minimal delay is optimal to prevent clots
occurring in the pancreas.
11. A serrefine is a small spring forceps used to close an artery
during surgery.
12. The section of excised organs will include the pancreas, spleen,
stomach, and part of the intestines, although only the pancreas
will be perfused in this system.
13. A small volume of blood will have drawn back into the portal
vein due to the lack of valves in the hepatic portal system. This
actually aids the insertion of the cannula, as when the internal
solution is translucent, the hole in the vein can be difficult to
locate.
14. The tape is unnecessary if there is a second person available to
tie the ligature around the portal vein and cannula while the
operator is holding it in place.
15. Pancreatic blood flow in rodents is approximately 2 ml per
gram pancreas weight per minute. As pancreata are not isolated
in this technique, they cannot be weighed as part of the proce-
dure. If estimates of pancreas weights are not available, use flow
rates of 2 and 4 ml·min1 for rats and 0.25 and 0.5 ml·min1
for mice. Perfusate collected during this period may be used to

help validate and assess the variability of the secretagogue
assays.
16. Setting up the perfusion equipment in a temperature-
controlled room helps to achieve a consistent perfusion tank
temperature with little or no adjustment of water bath and
water heater/pump temperature.
17. This dilution represents the final dilution, that is, after the label
and cold/blank solution has been added.
18. The lower the concentration of primary antisera used, the
higher the sensitivity of the assay and the more reliable the
lower end of the standard curve. A primary antibody dilution
giving 30% of maximum amount of label precipitated is usually
ideal.
19. When performing the primary antibody titer test for the first
time an excess of secondary antisera may be used. When the
secondary antibody titer test has been carried out the primary
test should be repeated with the newly calculated secondary
antibody dilution. The manufacturer of the secondary anti-
body will recommend a workable dilution factor. This value
should be the middle point dilution in the second antibody
titer test.
20. The total counts (T) are used to confirm that the label is in
excess to the antibody (% B0/T should be about 30%). This can
also be used to estimate gamma counter efficiency (%
cpm/dpm). Non-specific binding (NSB) tubes contain no pri-
mary antibody and represent the proportion of non-bound
label precipitated. This amount should be subtracted from all
other values before any further calculations. It represents 0%
binding. Maximum binding (B0) tubes contain no added cold
insulin and effectively all binding sites are occupied by hot
insulin. These tubes should have the highest counts (apart
from T tubes) and represent 100% binding. All standards and
samples are expressed as a percentage of this value. Intraassay
and interassay variances are determined using the same stock of
insulin for all assays used in the experiment (this has to be at an
insulin concentration not used to generate the standard curve).
Intraassay variance is expressed as a % standard deviation of
variance tubes/mean of variance tubes and should be less than
10%. Interassay variance is expressed as a % standard deviation
of the mean of intraassay means/mean of intraassay means and
should be less than 15%.
21. Incubating with primary antisera before addition of the label
saturates the binding sites with cold antigen. A shortened
incubation time following addition of the hot antigen prevents
equilibrium being achieved and results in a more sensitive
radioimmunoassay. If a lack of sensitivity is seen try incubating

with primary antisera for 12 h prior to the 8–12 h incubation
with hot antigen.
22. A final concentration of 3% PEG has been found to give the
greatest percentage of specific binding and the lowest percent-
age of nonspecific binding in double-antibody
radioimmunoassays [7].
References
1. Sussman KE, Vaughan GD, Timmer RF (1966) Wiedenmann B, Plöckinger U (2008) Orexin-
An in vitro method for studying insulin secretion A inhibits glucagon secretion and gene expres-
in the perfused isolated rat pancreas. Metabolism sion through a Foxo1-dependent pathway.
15:466–476 Endocrinology 149:1618–1626
2. Emilsson V, Liu YL, Cawthorne MA, Morton 5. Egido EM, Hernández R, Marco J, Silvestre RA
NM, Davenport M (1997) Expression of the (2008) Effect of obestatin on insulin, glucagon
functional leptin receptor mRNA in pancreatic and somatostatin secretion in the perfused rat
islets and direct inhibitory action of leptin on pancreas. Regul Pept 152:61–66
insulin secretion. Diabetes 46:313–316 6. Tseng C-C, Zhang X-Y, Wolfe MM (1999)
3. de Heer J, Rasmussen C, Coy DH, Holst JJ Effect of GIP and GLP-1 antagonists on insulin
(2008) Glucagon-like peptide-1, but not release in the rat. Am J Physiol Endocrinol
glucose-dependent insulinotropic peptide, inhi- Metab 276:E1049–E1054
bits glucagon secretion via somatostatin (recep- 7. Eisenman JR, Chew BP (1983) Polyethylene
tor subtype 2) in the perfused rat pancreas. Glycol in conjunction with a secondary antibody
Diabetologia 51:2263–2270 for 24 hour radioimmunoassay of bovine prolac-
4. Göncz E, Strowski MZ, Grötzinger C, Nowak tin and growth hormone. J Dairy Science
KW, Kaczmarek P, Sassek M, Mergler S, 66:1174–1179
El-Zayat BF, Theodoropoulou M, Stalla GK,
INDEX
A Glucose uptake ............................2, 5, 49, 50, 53, 54, 63,

255, 257, 259, 260
α-glucosidase inhibitors .............................................. 9–11 Glutathione (GSH) ........................... 242, 244, 246, 248,
AMPure XP ................................89, 92, 96, 97, 103, 165 249, 252
Animal models.................. 12, 43, 44, 46, 53, 54, 58–60,
62, 80, 186 H
Antibodies ............................................19, 187, 200, 216,
285, 290 High-fat diets ........................................ 44–46, 53, 56, 59
Histology ................................... 130, 186, 215–217, 266,
B 267, 270, 277
Histone modifications.......................................... 199–212
Beta-cells........................................ 13, 20, 49, 54, 56, 57, Human islets............. 7, 13, 14, 186, 188, 189, 192, 194
231, 232, 236–238, 241, 242, 246, 248, 253, 256 Hyperinsulinism ................................................... 130, 164
Biguanides ..................................................................... 4–6
Bioinformatics ................................................88, 110, 111 I
C Illumina HiSeq ............................. 85–107, 159, 164, 176
Image analysis............................................. 215–220, 222,
Calcium.....................................3, 10, 187, 231–238, 284 225, 227–229
Chemical database........................................................... 73 ImageJ............................................................................ 216
Cheminformatics.......................................................71–82
Immunohistochemistry .............................. 186, 217, 219
Collagenase.............. 180–182, 184, 187, 188, 195, 199, In vivo .................................. 13, 31–43, 45, 46, 256, 281
201–203, 210, 256, 258, 261, 268, 271, 278 Incretins ................................................. 11–16, 20, 33, 34
Collagenase digestion ................................ 180, 182, 184,
Insulin ........................ 2, 33, 43, 87, 113, 129, 185, 199,
195, 258, 261 216, 219, 255, 265, 281
Insulin resistance .......................... 2, 33, 44, 53, 185, 255
D
Insulin sensitivity....................... 5, 17, 33–35, 37, 38, 46,
Data mining................................................................... 110 48, 49, 53–55, 57, 58, 60, 255–262
Data quality ..................................................................... 34 Intraportal islet transplantation.................................... 269
DPP-IV inhibitors....................................... 15, 16, 18, 78 Islets isolation....................................................... 181, 182
Drug discovery ...................................... 1–20, 73, 79, 215 Islets of Langerhans ..................... 53, 179–182, 184, 199
Islet transplantation ................... 180, 265–275, 277, 278
E Isolation .......................35, 181, 192, 199, 256–257, 267
Endocrine pancreas ......................................................... 35
K
Exome-sequencing..................................................85–107
KCNJ11 and ABCC8 ................................................... 129
F
L
Flow cytometry ................. 185–192, 195, 196, 243, 250
Fluorescent microscopy ................................................ 237 Leukocytes ...........................................185–192, 195, 196
Literature mining ................................................. 110–113
G
M
Gene enrichment.................................................. 110, 112
Gene expression .....................49, 61, 109–117, 119–127 Maturity onset diabetes of the young
Glinides......................................................................10, 11 (MODY) ............................................................ 129
Glucagon-like peptide-1 (GLP-1)....................11, 15, 16, Metabolic pathway ...................................... 120, 122, 126
49, 282, 294 Metformin .................................. 2–8, 10, 11, 15, 18, 113
https://doi.org/10.1007/978-1-4939-9882-1, © Springer Science+Business Media, LLC, part of Springer Nature 2020
299
TYPE 2 DIABETES: METHODS AND PROTOCOLS
300 Index
Microarray .................................. 109–111, 114, 115, 121 R
MIN6 .................................................................... 232–238
Molecular modeling........................................................ 80 Radioimmunoassay .................... 283, 285, 289–294, 296
Mouse ..........................................33, 113, 122, 187, 202, Renal subcapsular islet transplantation ............... 268, 269
227, 265
S
Murine islets ........................................185–192, 195, 196
Sanger sequencing........................................ 86, 130, 131,
N 159–161, 164
Native chromatin immunoprecipitation Sequencing ............................................................ 85, 109,
(NChIP)................. 200, 201, 204, 205, 209–211 130, 209
NEBNext .............................................89, 90, 94–96, 166 SGLT2 inhibitors ......................................................15–20
Neonatal diabetes................................................... 87, 129 Stress .................................5, 10, 32, 34, 37, 38, 42, 185,
Next generation sequencing (NGS) .......................85–89, 241, 242, 279
109–111, 130, 164 Sulfonylureas ..............................3, 5, 7, 8, 10, 11, 13, 18
Nitric oxide (NO) ..............................241–246, 248, 250, Sulphonylureas .......................................... 4, 6, 7, 11, 129
252, 253
T
Nucleosomes ........................................................ 206, 207
Nutrition........................................................... 43–63, 185 Targeted next generation sequencing................. 130, 164
Thiazolidinediones................................ 5, 7–9, 13, 20, 80
O 3Rs ................................................................................. 283
Obesity.............................................. 13, 31, 44, 113, 255 Transcription factor binding site
Oxidative stress...............................................10, 241, 242 (TFBS) ..................................................... 110, 111,
114, 117
P Type 2 diabetes (T2D) ................................. 1–20, 31–33,
43–63, 71–82, 85, 88, 109–117, 119–127, 185,
Pancreas ..........................13, 35, 44, 130, 179, 188, 199, 193, 200, 242, 256, 282
215, 256, 267, 281
Pathway analysis ............................................................ 121 V
Perfusion...................................... 35, 180, 237, 281–287,
289, 290, 292, 294–297 Virtual screening ......................................................71–77,
Pharmaceutical industry ...........................................12, 19 79–82
Primary adipocytes ............................................... 255–262 VisiomorphDP ............................................ 215, 216, 222
Programming ..................................................... 36, 54–61
W
Q Whole exome sequencing (WES)...........................85–107
Quantitative PCR (qPCR).................................. 104–106, Whole slide image ....................................... 215, 219, 220
202, 206, 209, 210

Diabetes Methods in Molecular Biology Stocker 2020

Uploaded by

Copyright:

Available Formats

Diabetes Methods in Molecular Biology Stocker 2020

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Diabetes Methods in Molecular Biology Stocker 2020

Uploaded by

Copyright:

Available Formats

Methods in

Molecular Biology 2076

Claire J. Stocker Editor

For further volumes:

Methods and Protocols

ISSN 1064-3745 ISSN 1940-6029 (electronic)

© Springer Science+Business Media, LLC, part of Springer Nature 2020

Buckingham, UK Claire J. Stocker

15 Primary Adipocytes as a Model for Insulin Sensitivity. . . . . . . . . . . . . . . . . . . . . . . . 255

MATTHEW J. BUTCHER  Department of Microbiology and Molecular Cell Biology, Eastern

Sixty Years of Drug Discovery for Type 2 Diabetes: Where Are

Key words Type-2-diabetes, Drug discovery, Metformin, Sulfonylureas, Biguanides, Thiazolidine-

Modern drug discovery programs often start with a hypothesis,

over placebo with noninferiority against standard therapy, usually

2 Sulfonylureas (First Generation)

Insulin secretion from the pancreatic β-cell occurs in response to an

The only representative of this class of anti-diabetic drug remaining

other agents using these transporters will influence metformin’s

insufficiency [52]. Metformin is extensively cleared via the kidneys

4 Sulfonylureas (Second Generation)

This group is represented by the drugs glibenclamide (glyburide),

The thiazolidinediones, troglitazone, rosiglitazone, and pioglita-

but through a different, then unknown, mechanism. They were

Carbohydrates constitute a large proportion of our energy intake,

24 weeks treatment was observed [101]. The efficacy of acarbose is

The insulin response to glucose occurs in two phases: an immediate

240 nM, respectively, compared to 2.3 nM for glibenclamide

8.1 GLP-1 Mimetics The discovery of the polypeptide hormones, glucose-dependent

but not intravenous, administration of glucose enhanced insulin

However, in the present world, the main therapeutic GLP-1

study especially in the clinical setting. In another twist to the

apparently beneficial effect on the β-cell when using insulin: proin-

The SGLT2 inhibitors are the latest oral anti-hyperglycemic agents

these approvals occurring in rapid succession with yet others wait-

may be indirect. A retrospective study in such patients from India

10 Where Are We Now?

In this chapter, I have outlined the eight classes of anti-diabetic

antibody therapeutics seem to hold many possibilities not dreamed

in individuals destined to develop type 2 dia- type 2 diabetes: a patient-centered approach.

like peptide-1 (GLP-1) receptor. Endocrinol- 144. Schlögl H, Kabisch S, Horstmann A,

Practical Considerations for In Vivo Mouse Studies

Key words Mouse, In vivo, Stress, Data quality, 3Rs

Type 2 diabetes and obesity are worldwide major public health

2 General Output Measures

Relevant output measures fall into three groups: obesity, insulin

glucose concentration is desirable as it produces a statistically sig-

2.2.4 Sensitivity Following a short (4 h) fast, glucose concentration is monitored

3 Considerations for In Vivo Procedures in Obesity and Diabetes Protocols

physiology of obesity and diabetes makes a study of strain suitability

3.6 Administration The mini-pump could be either intraperitoneal or subcutaneous

3.6.4 Subcutaneous Substances should be administered by the subcutaneous route at a

3.6.5 Intraperitoneal Substances should be administered by the intravenous route at a

3.6.6 Intravenous Substances may be administered by the intravenous route at a

3.7 Energy Energy expenditure may be measured by whole body calorimetry at

3.7.1 Measurement Energy expenditure should be measured in the normal housing

3.7.2 Measurement Energy expenditure measured at thermoneutrality (28–30  C)

They should be observed at least every hour to ensure that this

Nutritional Models of Type 2 Diabetes Mellitus

While the human is undoubtedly the model of choice when study-

an ideal disease model, since these gene mutations are extremely

MATTHEW J. BUTCHER Department of Microbiology and Molecular Cell Biology, Eastern

3.7.2 Measurement Energy expenditure measured at thermoneutrality (28–30 C)