(Methods in Molecular Biology) - Artificial Riboswitches - Methods and Protocols (2014, Humana Press)

Methods in
Molecular Biology 1111
Atsushi Ogawa Editor
Artificial
Riboswitches
Methods and Protocols
METHODS IN M O L E C U L A R B I O LO G Y
Series Editor
John M. Walker
School of Life Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK
For further volumes:

http://www.springer.com/series/7651
Artificial Riboswitches
Methods and Protocols
Edited by
Atsushi Ogawa
Ehime University, Matsuyama, Ehime, Japan
Editor
Atsushi Ogawa
Ehime University
Matsuyama
Ehime, Japan
ISSN 1064-3745 ISSN 1940-6029 (electronic)

ISBN 978-1-62703-754-9 ISBN 978-1-62703-755-6 (eBook)
DOI 10.1007/978-1-62703-755-6
Springer New York Heidelberg Dordrecht London
Library of Congress Control Number: 2013958315
© Springer Science+Business Media New York 2014

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. Exempted from this
legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for
the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.
Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the
Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions
for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution
under the respective Copyright Law.
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.
While the advice and information in this book are believed to be true and accurate at the date of publication, neither
the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be
made. The publisher makes no warranty, express or implied, with respect to the material contained herein.
Printed on acid-free paper
Humana Press is a brand of Springer

Springer is part of Springer Science+Business Media (www.springer.com)
Preface
Ribonucleic acids or RNAs are versatile. mRNAs serve as mediators of hereditary informa-
tion from DNAs to proteins, tRNAs carry amino acids into the ribosome on mRNA, and
rRNAs in the ribosome promote peptide bond formation between the amino acids. In
addition, many diverse noncoding RNAs directly regulate gene expression in cis and/or
trans. There are also many other types of RNAs known in nature, including RNA enzymes
(ribozymes). It is considered that in an ancient RNA world (where proteins and DNAs did
not still exist), a greater variety of functional RNAs would have flourished, though almost
all of them are now defunct.
RNA functions are dependent on their sequences, which are composed of four kinds of
nucleotides: A, G, U, and C. The sequences determine the higher order structures, which
in turn make possible their unique functions. In other words, we could create new func-
tional RNAs by appropriately lining up the nucleotides. Although RNAs are not as diverse
as proteins, which have 20 amino acids as building block, RNAs have a large advantage in
that they are easy to be amplified with the following three steps: reverse transcription, poly-
merase chain reaction (PCR), and transcription. We can therefore obtain functional RNAs
from a randomized RNA pool by repeating selection–amplification cycles (in vitro selec-
tion). In fact, many types of functional RNAs have been selected via in vitro selection. The
representative examples are aptamers that tightly bind to their specific ligand molecules and
ribozymes that catalyze specific biochemical reactions. In vitro selection also enables us to
obtain a ligand-responsive ribozyme called aptazyme, which is generally a conjugate
between two foregoing functional RNAs, an aptamer and a ribozyme.
A riboswitch is also a functional RNA: a ligand-dependent and cis-acting gene regulator
in mRNA. This regulatory RNA is composed of an aptamer domain, which binds to the
specific ligand as described above, and an expression platform, which regulates transcription
termination, translation initiation, self-cleavage, or splicing of its own mRNA. The interac-
tion between the aptamer domain and the ligand induces conformational changes of the
expression platform to up- or down-regulate gene expression. Natural riboswitches acting
in response to endogenous metabolites have been identified mainly in bacteria and rarely in
eukaryotes over the last decade. In parallel with these discoveries, a variety of methods have
also been reported for artificially constructing arbitrary molecule-dependent riboswitches
with the corresponding in vitro-selected aptamers or self-cleaving aptazymes.
This volume is mainly focused on the state-of-the-art methods developed in recent
years for creating artificial riboswitches. Approximately half of the total chapters are devoted
to screening or rational design methods for obtaining artificial riboswitches that function in
either bacterial or eukaryotic translation systems. In these methods, an aptamer or an apta-
zyme is generally required for a gene regulator to acquire ligand responsiveness. Although
an already identified aptamer or aptazyme is available, several chapters cover in vitro
selection methods for obtaining a new aptamer or aptazyme for a ligand molecule that the
reader might be interested in (a small molecule, protein, or photo-responsive molecule).
In this context, one chapter is devoted to a computational method for designing a starting
library of RNA sequences for in vitro selection. Protocols for evaluating the activities of the
v
vi Preface
resultant riboswitches are also presented. Some other chapters include protocols for con-
struction of ligand-dependent, trans-acting gene regulators.
Artificial riboswitches and other ligand-responsive gene regulators that are obtainable
through the cutting-edge methods described here make it possible to switch protein syn-
thesis ON or OFF with arbitrary ligand molecules, which can be freely chosen when select-
ing the corresponding aptamers to be implemented in the gene regulators. Therefore, this
book can be regarded as a collection of recipes for the gene circuit elements in synthetic
biology and metabolic engineering. However, I would recommend this cookbook not only
to bioengineers who aim to reprogram cell behaviors and molecular biologists who leverage
these regulators for genetic studies but also to all researchers who just want to regulate the
expression of a specific gene by an arbitrary molecule in various organisms, to detect a spe-
cific molecule with reporter protein expression in vitro or in vivo, or to design ligand-
dependent RNA switches by using aptamers or aptazymes.
All chapters are written by experts from all around the world who are active in the front
lines of the relevant research areas. I would like to express my gratitude to the contributors,
all of whom were willing to write their lab protocols in an easily comprehensible manner,
despite their busy research lives. I believe that the readers will be able to easily understand
and follow the experimental procedures, thanks to the intelligible explanations and notes,
and to construct their own artificial riboswitches. Last but not least, I am also grateful to
Prof. John Walker for giving me the opportunity to edit this book.
Matsuyama, Ehime, Japan Atsushi Ogawa

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
1 Computational Design of RNA Libraries for In Vitro

Selection of Aptamers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Yaroslav G. Chushak, Jennifer A. Martin, Jorge L. Chávez,
Nancy Kelley-Loughnane, and Morley O. Stone
2 In Vitro Selection of RNA Aptamers for a Small-Molecule Dye . . . . . . . . . . . . 17
Asako Murata and Shin-ichi Sato
3 Development of Photoswitchable RNA Aptamer–Ligand Complexes. . . . . . . . 29
Gosuke Hayashi and Kazuhiko Nakatani
4 Identification of RNA Aptamers Against Recombinant Proteins
with a Hexa-Histidine Tag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Shoji Ohuchi
5 Artificial Riboswitch Selection: A FACS-Based Approach. . . . . . . . . . . . . . . . . 57
Zohaib Ghazi, Casey C. Fowler, and Yingfu Li
6 FRET-Based Optical Assay for Selection of Artificial Riboswitches . . . . . . . . . . 77
Svetlana V. Harbaugh, Molly E. Chapleau, Yaroslav G. Chushak,
Morley O. Stone, and Nancy Kelley-Loughnane
7 Portable Two-Way Riboswitches: Design and Engineering . . . . . . . . . . . . . . . 93
Ye Jin and Jian-Dong Huang
8 Generation of Orthogonally Selective Bacterial Riboswitches
by Targeted Mutagenesis and In Vivo Screening . . . . . . . . . . . . . . . . . . . . . . . 107
Helen A. Vincent, Christopher J. Robinson, Ming-Cheng Wu,
Neil Dixon, and Jason Micklefield
9 Dual Genetic Selection of Synthetic Riboswitches in Escherichia coli . . . . . . . . 131
Yoko Nomura and Yohei Yokobayashi
10 Nucleotide Kinase-Based Selection System for Genetic Switches . . . . . . . . . . . 141
Kohei Ike and Daisuke Umeno
11 Measuring Riboswitch Activity In Vitro and in Artificial Cells
with Purified Transcription–Translation Machinery . . . . . . . . . . . . . . . . . . . . . 153
Laura Martini and Sheref S. Mansy
12 Rational Design of Artificial ON-Riboswitches . . . . . . . . . . . . . . . . . . . . . . . . 165
Atsushi Ogawa
13 Engineering Protein-Responsive mRNA Switch in Mammalian Cells . . . . . . . . 183
Kei Endo and Hirohide Saito
vii
viii Contents
14 Guanine-Tethered Antisense Oligonucleotides

as Synthetic Riboregulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Masaki Hagihara
15 In Vitro Selection of Allosteric Ribozymes that Sense
the Bacterial Second Messenger c-di-GMP . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Kazuhiro Furukawa, Hongzhou Gu, and Ronald R. Breaker
16 Dual-Selection for Evolution of In Vivo Functional Aptazymes
as Riboswitch Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Jonathan A. Goler, James M. Carothers, and Jay D. Keasling
17 In Vivo Screening for Aptazyme-Based Bacterial Riboswitches. . . . . . . . . . . . . 237
Charlotte Rehm and Jörg S. Hartig
18 Engineered Riboswitch as a Gene-Regulatory Platform
for Reducing Antibiotic Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Libing Liu and Shu Wang
19 Construction of Ligand-Responsive MicroRNAs that Operate
Through Inhibition of Drosha Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Chase L. Beisel, Ryan J. Bloom, and Christina D. Smolke
20 A Three-Dimensional Design Strategy for a Protein-Responsive
shRNA Switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Shunnichi Kashida and Hirohide Saito
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Contributors
CHASE L. BEISEL • Department of Chemical and Biomolecular Engineering,

North Carolina State University, Raleigh, NC, USA
RYAN J. BLOOM • Department of Bioengineering, Stanford University, Palo Alto, CA, USA
RONALD R. BREAKER • Department of Molecular, Cellular and Developmental Biology,
Howard Hughes Medical Institute, Yale University, New Haven, CT, USA; Department
of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT, USA
JAMES M. CAROTHERS • University of California, Berkeley, CA, USA; U.S. Department of
Energy Joint BioEnergy Institute, Emeryville, CA, USA
MOLLY E. CHAPLEAU • UES, Inc., Dayton, OH, USA
JORGE L. CHÁVEZ • UES, Inc., Dayton, OH, USA
YAROSLAV G. CHUSHAK • The Henry M. Jackson Foundation, Bethesda, MD, USA
NEIL DIXON • Manchester Institute of Biotechnology and School of Chemistry, The University
of Manchester, Manchester, UK
KEI ENDO • Department of Reprogramming Science, Center for iPS Cell Research
and Application, Kyoto University, Kyoto, Japan
CASEY C. FOWLER • Department of Biochemistry and Biomedical Sciences, McMaster
University, Hamilton, ON, Canada; Department of Chemistry and Chemical Biology,
McMaster University, Hamilton, ON, Canada; Michael G. DeGroote Institute for
Infectious Disease Research, McMaster University, Hamilton, ON, Canada
KAZUHIRO FURUKAWA • Department of Molecular, Cellular and Developmental Biology,
Yale University, New Haven, CT, USA
ZOHAIB GHAZI • Department of Biochemistry and Biomedical Sciences, McMaster
University, Hamilton, ON, Canada; Department of Chemistry and Chemical Biology,
McMaster University, Hamilton, ON, Canada; Michael G. DeGroote Institute for
Infectious Disease Research, McMaster University, Hamilton, ON, Canada
JONATHAN A. GOLER • University of California, Berkeley, CA, USA; U.S. Department of
HONGZHOU GU • Department of Molecular, Cellular and Developmental Biology, Howard
Hughes Medical Institute, Yale University, New Haven, CT, USA
MASAKI HAGIHARA • Graduate School of Science and Technology, Hirosaki University,
Hirosaki, Japan
SVETLANA V. HARBAUGH • The Henry M. Jackson Foundation, Bethesda, MD, USA
JÖRG S. HARTIG • Department of Chemistry, University of Konstanz, Konstanz, Germany
GOSUKE HAYASHI • Department of Chemistry and Biotechnology, Research Center for
Advanced Science and Technology, The University of Tokyo, Tokyo, Japan
JIAN-DONG HUANG • Department of Biochemistry and Shenzhen Institute of Research
and Innovation, The University of Hong Kong, Pokfulam, Hong Kong; The Center
for Synthetic Biology Engineering Research, Shenzhen Institutes of Advanced Technology,
Shenzhen, People’s Republic of China
KOHEI IKE • Department of Applied Chemistry and Biotechnology, Graduate School of
Engineering, Chiba University, Chiba, Japan
ix
x Contributors
YE JIN • Department of Biochemistry, Li Ka Shing Faculty of Medicine, University of Hong

Kong, Hong Kong SAR, People’s Republic of China
SHUNNICHI KASHIDA • Department of Reprogramming Science, Center for iPS Cell
Research and Application, Kyoto University, Kyoto, Japan
JAY D. KEASLING • University of California, Berkeley, CA, USA; U.S. Department of
NANCY KELLEY-LOUGHNANE • Human Effectiveness Directorate, 711 Human Performance
Wing, Air Force Research Laboratory, Wright-Patterson AFB, OH, USA
YINGFU LI • Department of Biochemistry and Biomedical Sciences, McMaster University,
Hamilton, ON, Canada; Department of Chemistry and Chemical Biology, McMaster
University, Hamilton, ON, Canada; Michael G. DeGroote Institute for Infectious Disease
Research, McMaster University, Hamilton, ON, Canada
LIBING LIU • Key Laboratory of Organic Solids, Institute of Chemistry, Chinese Academy of
Sciences, Beijing, People’s Republic of China
SHEREF S. MANSY • CIBIO, University of Trento, Mattarello, Italy
JENNIFER A. MARTIN • The Henry M. Jackson Foundation, Bethesda, MD, USA
LAURA MARTINI • CIBIO, University of Trento, Mattarello, Italy
JASON MICKLEFIELD • Manchester Institute of Biotechnology and School of Chemistry,
The University of Manchester, Manchester, UK
ASAKO MURATA • The Institute of Scientific and Industrial Research, Osaka University,
Ibaraki, Japan
KAZUHIKO NAKATANI • The Institute of Scientific and Industrial Research,
Osaka University, Ibaraki, Japan
YOKO NOMURA • Department of Biomedical Engineering, University of California, Davis,
Davis, CA, USA
ATSUSHI OGAWA • Proteo-Science Center, Ehime University, Matsuyama, Japan
SHOJI OHUCHI • Graduate School of Biostudies, Kyoto University, Kyoto, Japan
CHARLOTTE REHM • Department of Chemistry, University of Konstanz, Konstanz,
Germany
CHRISTOPHER J. ROBINSON • Manchester Institute of Biotechnology and School of Chemistry,
HIROHIDE SAITO • Department of Reprogramming Science, Center for iPS Cell Research
and Application, Kyoto University, Kyoto, Japan
SHIN-ICHI SATO • Institute for Integrated Cell-Material Sciences, Kyoto University, Uji, Japan
CHRISTINA D. SMOLKE • Department of Bioengineering, Stanford University, Palo Alto,
CA, USA
MORLEY O. STONE • Human Effectiveness Directorate, 711 Human Performance Wing,
Air Force Research Laboratory, Wright-Patterson AFB, OH, USA
DAISUKE UMENO • Department of Applied Chemistry and Biotechnology, Graduate School
of Engineering, Chiba University, Chiba, Japan
HELEN A. VINCENT • Manchester Institute of Biotechnology and School of Chemistry,
SHU WANG • Key Laboratory of Organic Solids, Institute of Chemistry, Chinese Academy
of Sciences, Beijing, People’s Republic of China
MING-CHENG WU • Manchester Institute of Biotechnology and School of Chemistry,
YOHEI YOKOBAYASHI • Department of Biomedical Engineering, University of California,
Davis, Davis, CA, USA
Chapter 1
Computational Design of RNA Libraries for In Vitro

Selection of Aptamers
Yaroslav G. Chushak, Jennifer A. Martin, Jorge L. Chávez,
Nancy Kelley-Loughnane, and Morley O. Stone
Abstract
Selection of aptamers that bind a specific ligand usually begins with a random library of RNA sequences,
and many aptamers selected from such random pools have a simple stem–loop structure. We present here
a computational approach for designing a starting library of RNA sequences with increased formation of
complex structural motifs and enhanced affinity to a desired target molecule. Our approach consists of two
steps: (1) generation of RNA sequences based on customized patterning of nucleotides with increased
probability of forming a base pair and (2) a high-throughput virtual screening of the generated library to
select aptamers with binding affinity to a small-molecule target. We developed a set of criteria that allows
one to select a sequence with potential binding affinity from a pool of random sequences and designed a
protocol for RNA 3D structure prediction. The proposed approach significantly reduces the RNA sequence
search space, thus accelerating the experimental screening and selection of high-affinity aptamers.
Key words Screening library design, Patterned sequences, RNA structure prediction, Virtual screening,
RNA–small molecule docking
1 Introduction
Aptamers can be selected in vitro using an iterative process called

systematic evolution of ligands by exponential enrichment (SELEX)
[1, 2]. The SELEX process consists of multiple cycles of selection
and amplification: (1) a random pool of RNA molecules is screened
and sequences demonstrating binding to a target molecule are
separated from non-binders and (2) retained sequences are ampli-
fied by the polymerase chain reaction (PCR) to create a pool of
sequences for the next round of enrichment. The entire selection
process typically requires up to 15 rounds of selection and can take
from a few days to a few months to complete [3]. The initial pool
generally contains 1013–1015 unique sequences designed with two
PCR primer regions flanking a random region of 30–50 nucleotides.
Most of the sequences present in this initial pool are not functional
Atsushi Ogawa (ed.), Artificial Riboswitches: Methods and Protocols, Methods in Molecular Biology, vol. 1111,
DOI 10.1007/978-1-62703-755-6_1, © Springer Science+Business Media New York 2014
1
2 Yaroslav G. Chushak et al.
and have no potential to function as binders, as the majority has a

simple stem–loop structure [4]. The occurrence of more complex,
high-affinity aptamers in the starting random library is very low,
and therefore, many aptamers selected from such random pools
have a simple stem–loop structure. Researchers have investigated
various methods of biasing an initial selection pool to lessen preva-
lence of random or simple structures and to increase the presence
of more complex potential binders. This was experimentally per-
formed by Davis and Szostak [5] who selected aptamers from a
mixture of fully random and partially structured RNA libraries cre-
ated by incorporation of an internal loop which is characteristic of
many known RNA aptamers. They identified six sequence families
of aptamers. Four of these families came from the designed library,
while two families were from the random pool. The aptamers with
the highest affinity were selected from the designed library of
sequences and demonstrated more complex structures, while
aptamers from the random pool had a simple stem–loop structure.
Thus, increasing the complexity of structures formed in the initial
pool (including internal loops, bulges, junctions, and mismatches)
gives a higher probability of selecting high-affinity binders [6].
Recently, there have been several attempts to design enhanced
initial pools of sequences for selection of high-affinity aptamers. In
one approach Ruff et al. [7] designed libraries with varying degrees
of predicted average secondary structure by patterning pyrimidine-
and purine-rich positions. For the three targets used for individual
selection, the patterned library substantially outperformed the ran-
dom library in two cases and performed at least as well as random
in the third case. Kim et al. [8, 9] used 4 × 4 “nucleotide transition
probability” matrices to design and generate pools with an optimal
yield of specific motifs. By using a combination of different matri-
ces it is possible to generate pools with different motif distributions
characteristic to the target [9]. In another approach, Luo et al.
[10] have developed random filtering and genetic filtering meth-
ods to increase the number of five-way junctions or to design a
uniform structure distribution in the starting RNA/DNA pools.
One of the aptamers selected by SELEX from the designed struc-
tural libraries displayed higher affinity for ATP than the previously
selected low-complexity aptamer, while other aptamers showed
weaker affinity.
We present here a computational approach for designing a
starting library of RNA sequences with increased formation of
complex structural motifs and increased affinity to a desired target
molecule. Our approach consists of two steps (Fig. 1): (1) generation
of RNA sequences based on customized patterning of nucleotides
with increased probability of forming a base pair and (2) a high-
throughput virtual screening of the generated library to select
Computational Design 3
Fig. 1 Generation of initial RNA sequence pool. Step 1: Generate large sequence pool of sequences using
patterned libraries with the alternation of purines (R = A, G), pyrimidines (Y = U, C), and random bases (N = A, C,
G, U). At this step generated sequences are not target specific and can be used for SELEX-like selection of
binding aptamers or can be used in step 2. At step 2 sequences are filtered based on their secondary structure,
their 3D structure is generated, and virtual screening is used to identify RNA molecules that bind to a target
molecule. Selected RNA molecules that are now specific to the desired target are placed into a pool of
sequences for experimental screening and selection of high-affinity aptamers
aptamers with binding affinity to a desired small-molecule target.

In step (1), we designed four different patterned libraries (to provide
diversity of structures) of RNA sequences that can be used as the
combined initial pool for aptamer selection using the SELEX pro-
tocol. At this screening level, the selected sequences are not target
specific but rather designed to contain complex structures. In the
second step, we used computational docking to identify which of
the patterned RNA molecules can bind to a specific target ligand.
We have developed a set of criteria that allows one to preselect
sequences with potential binding affinity from a pool of patterned
sequences and developed a protocol for RNA 3D structure predic-
tion and virtual screening. At this point, the selected RNA mole-
cules are specific to the desired target molecule, and they are placed
into a pool of sequences for experimental verification and selection
of high-affinity aptamers, e.g., using microarray techniques. The
developed computational approach allows one to preselect
sequences from the initial pool of RNA molecules and effectively
leads to a reduction in sequence search space by eliminating
sequences with low binding potential before experimental selec-
tion and providing starting libraries with higher percentages of
potential binders present.
2 Materials
Patterned libraries were generated using Perl scripts. The secondary

structure and folding energy of generated sequences were evalu-
ated using the ViennaRNA v.2.0.7 package [11] with a default set
of parameters. ViennaRNA uses energy minimization with experi-
mental thermodynamic parameters [12, 13] for RNA secondary
structure prediction. The package is distributed with Perl utilities
that allows it easy integration with the sequence-generating scripts.
Since the folding algorithm implemented in the ViennaRNA pack-
age cannot predict the formation of structures with pseudoknots,
such structures were excluded from the consideration. The genera-
tion of RNA sequences and analysis of their secondary structure
are the least computationally intensive part of the proposed
approach. It took around 30 min on a single Intel P4 3.06 GHz
CPU to generate a patterned library with 10,000 members.
The tertiary structure of generated sequences in the library was
predicted using the Rosetta v.3.4 software suite [14]. The Rosetta
software suite focuses mainly on the prediction and design of pro-
tein structures and protein–protein interactions, but it also
addresses aspects of RNA design, protein–ligand, and protein–
DNA docking. The prediction of RNA 3D structures consists of
two steps [15]: (a) the fragment assembly of RNA molecule
based on the simplified low-resolution energy function followed
by (b) high-resolution optimization with a full-atom Rosetta
energy function.
While most docking methods have been developed for pro-
teins, recent evaluation of AutoDock and DOCK programs has
demonstrated their ability to dock compounds to RNA molecules
[16]. We used the AutoDock Vina [17] and AutoDockTools
(ADT) packages (http://autodock.scripps.edu/downloads) to
perform screening of a library of RNA molecules.
Docking tools are usually used to screen a library of small
molecules in order to find a ligand that binds to a specific protein
or RNA receptor. In our approach, we screened the library of RNA
molecules to find receptors with the highest binding affinity to a
desired small molecule.
To perform conversion of molecules in different file formats
we used OpenBabel package (http://openbabel.org).
3 Methods
The proposed method for generating starting pools of RNA

sequences consists of two major steps (Fig. 1). In the first step we
used patterned libraries to generate a large pool of RNA sequences.
The using of patterned libraries increases the probability of folding

and formation of a more complex secondary structure.
Four different patterned libraries of 40 nt RNA sequences
were designed as an initial pool for aptamer selection:
PT1: RYRY-N4-(RY)3-N3-(RY)3-N4-(RY)3-N3-RYRY
PT2: RRYY-N3-RRYY-N4-RRYY-N2-RRYY-N4-RRYY-N3-RRYY
PT3: RRYY-N3-RRRYYY-N4-RRRYYY-N4-RRRYYY-N3-RRYY
PT4: RRYY-N3-(RY)3-N4-(RY)3-N4-(RY)3-N3-RRYY
-----------------------------------------
R = (A, G), Y = (U, C), and N = (A, C, G, U).
The first library (PT1) has a pattern similar to one proposed by
Ruff et al. [7]. It consists of alternating purines (R = A, G) and
pyrimidines (Y = U, C) separated by completely random regions of
N4 and N3. The second pattern (PT2) was designed to maximize
the number of three-way junctions as proposed by Luo et al. [10].
Pattern PT1 has 14 completely random bases and 1.8 × 1016 possi-
ble sequences, while pattern PT2 has 16 random bases and
7.2 × 1016 possible sequences. Patterns PT3 and PT4 were designed
with the same number of random bases, 14, but PT3 allows plac-
ing three consecutive purines or pyrimidines in a row, while PT4
forces alternation of purines and pyrimidines. Pools of RNA
sequences generated with the proposed patterns are not target spe-
cific and can be used for experimental aptamer selection using
SELEX or further analyzed in step 2.
In the second step we use virtual screening to identify RNA
molecules that bind to a specific target molecule.
3.1 RNA Secondary Analysis of experimentally selected aptamers shows a significant cor-
Structure Analysis relation between the dissociation constant and the free energy of
secondary structure formation. Based on the literature, free energy
of aptamers is significantly lower than the median same-length ran-
dom sequence value [18]. Figure 2 shows the distribution of folding
energies for 10,000 randomly chosen members of patterned libraries
in comparison with the random N40 library. The mean folding energy
for random library ΔGmean = −6.3 kcal/mol, and the standard devia-
tion SD = 3.2 kcal/mol. The mean energy for pattern PT1 is
−8.98 kcal/mol with standard deviation of 3.55 kcal/mol, for pat-
tern PT2 −7.85 kcal/mol and 3.8 kcal/mol, and for pattern PT3
−8.03 kcal/mol and 3.8 kcal/mol, respectively. The cutoff energy
for sequence selection was chosen as ΔGmean − SD = −9.5 kcal/mol
based on values for the random sequences which is higher than the
mean energy value for the patterned sequences.
Next, we analyzed the interaction of nucleic acids with ligands
for several aptamer–ligand complexes based on their experimental
three-dimensional structures [19]. It was found that RNA bases
involved in molecular recognition do not form Watson–Crick pairs
with other bases. There are two possible reasons for this: first,
Fig. 2 Distribution of folding energies for 10,000 randomly chosen members of

patterned libraries in comparison with the random N40 library. The mean folding
energy for random library ΔGmean = −6.3 kcal/mol and the standard deviation
SD = 3.2 kcal/mol. The mean energy for pattern PT1 is −8.98 kcal/mol with
standard deviation of 3.55 kcal/mol, for pattern PT2 −7.85 kcal/mol and 3.8 kcal/
mol, and for pattern PT3 −8.03 kcal/mol and 3.8 kcal/mol, respectively
unpaired RNA bases are more flexible, so they can easily change
their conformation to form a binding pocket and accommodate a
ligand; secondly, unpaired bases have available donor/acceptor
atoms for potential formation of hydrogen bonds with the ligand.
Therefore, we set a constraint that the secondary structure of our
sequences with 40 bases should have at least 11 unpaired bases.
This number seemed optimal, since a higher number of unpaired
bases will significantly reduce the presence of sequences with high
secondary structure free energy while a lower number will reduce
the potential sites for ligand binding.
Based on these findings, we developed a set of criteria that
limited the presence of sequences with abundant simple structural
motifs and maximized the presence of stable low-energy structures.
For the pool of 40-mer sequences these criteria are as follows [4]:
(1) in the lowest energy conformation bases 1 and 2 form pairs
with bases 40 and 39; (2) the free energy of secondary structure
formation is less than −9.5 kcal/mol; (3) there are at least 11 bases
that do NOT form Watson–Crick pairs. Only the sequences that
passed the selection criteria were forwarded to the next step in the
selection process.
The typical steps for the secondary structure generation and
analysis are outlined below.
3.1.1 Generation To generate random or patterned libraries in Perl script we use

of Sequence Library function rand(). By default it generates a number between 0 and
1. However, we can pass it a specific number to generate random
numbers between 0 and that number. Initially, we create tree
arrays:
my @base4 = ('A', 'U', 'C', 'G');
my @baseR = ('A', 'G');
my @baseY = ('U', 'C');
Next, we specify two numbers for rand() function:
my $range_4 = 4;
my $range_2 = 2;
Now, example of the random part of the sequence:
my $seq1= '';
for($i=0; $i<4; $i++)
{
$ind = int(rand($range_4));
$seq1 .= $base4[$ind];
}
An example of the patterned part of the sequence RRYY:
my $link= '';
$link .= $baseR[$ind];
$link .= $baseR[$ind];
$link .= $baseY[$ind];
$link .= $baseY[$ind];
Finally, we assemble all part of the sequence into a variable
$sequence.
3.1.2 Calculation Next step is to calculate the secondary structure and folding energy
of Folding Energy of the of the generated sequence. We perform that using RNA::fold
Generated Sequences function from the ViennaRNA package:
use RNA;
($s_struct, $f_energy) = RNA::fold($sequence);
The secondary structure of the generated sequence is presented
in bracket-dot notations. In these notations unpaired bases are
presented by dots and base pair between bases i and j is presented
by “(” at position i and “)” at position j, e.g., “(((…((…))..)))”.
3.1.3 Analysis of the Now, perform analysis of the secondary structure. If it satisfies our
Secondary Structure criteria, we save the sequence and its secondary structure into the
of Generated Sequences output files:
open(fasta_FILE, ">Seq1.fasta") || die("Could not
open file:fasta_file !");
open(temp_FILE, ">Seq1.temp") || die("Could not

open file:temp_file !");
my $cut_en = -9.5;
$seq_len = length $sequence;
#Check energy < $cut_en
if ($f_energy < $cut_en )
{
#Check that nt_1 forms base pair with nt_end
@sp_struct = split(//, $s_struct);
if($sp_struct[0] eq "(" && $sp_struct[$seq_len] eq
")" )
{
# Check that we have more than 10
# but less than 30 unpaired bases
@ARG = split(/\./, $s_struct);
$n_points = @ARG;
if (($n_points > 10) && ($n_points < 25))
{
# Save sequence in FASTA format
print fasta_FILE ">RNA_sequence\n$sequence\n";
# Save sequence and structure into temporary file
print temp_FILE "$sequence\n$s_struct\n";
} } }
3.1.4 Generation of Files Saved sequences and their secondary structure are used to generate
for 3D Modeling three-dimensional structures by Rosetta package. RNA 3D struc-
ture modeling in Rosetta requires as input sequence in FASTA for-
mat (saved in step 3.1.3). As optional file it uses “.param” file that
specifies Watson–Crick base pairs, contained in $s_struct.
Parameter file looks like the following:
STEM PAIR 1 40 W W A
Our experience shows that predicted structures are more accurate
with the use of parameter file. There is no easy way to create param-
eter file from the $s_structure string. One possible solution is
to use b2c.c function that is distributed with the ViennaRNA pack-
age and is located in folder “Utilities.” This function converts
bracket-dot notations into mfold *.ct file. We modified write_
ct_file subroutine of b2c.c function to produce output file in
Rosetta parameter format. That modified function is named
b2param.c. As input it uses temporary file created in step 3.1.3,
and it is used inside the Perl script:
system (“b2param < Seq1.temp > Seq1.param”;
3.2 Generation During the last few years, several groups have proposed different
of RNA Three- approaches for de novo prediction of RNA tertiary structure (see
Dimensional ref. 20 for recent review). We used the Rosetta package [14, 15]
Structures developed by the Baker group at the University of Washington to
predict three-dimensional structures of selected sequences. In
Rosetta modeling algorithm, the RNA sequence is divided into
short fragments containing three nucleotides. The assembly of
fragment from a library of experimental RNA crystal structures is
based on Monte Carlo process with a simplified low-resolution
energy function. The generated models are then refined with the
all-atom potential to provide high-resolution atomistic structures.
This two-step protocol has been named Fragment Assembly of
RNA with Full Atom Refinement (FARFAR) [15]. In the previous
section we created two files for each of the sequences in the pool:
a sequence file in FASTA format (Seq1.fasta) and a parameter
file (Seq1.param) that contains information about the Watson–
Crick base pairing. These two files are used as input files for the 3D
structure prediction (see Note 1).
3.2.1 Generation The command to run RNA 3D structure modeling in Rosetta is as

of Three-Dimensional follows:
Structures $ROSETTA_HOME/rosetta_source/bin/rna_denovo.
<exe> @flags_rna - database $ROSETTA_HOME/rosetta_
database
where <exe> is the extension of the rna_denovo command
and file flags_rna contains a list of options for the RNA
modeling:
-fasta Seq1.fasta
-params_file Seq1.param
-nstruct 500
-out:file:silent Seq1.out
-minimize_rna
-vary_geometry
-mute core.io.database
In the above example we generated 500 minimized 3D struc-
tures in the output file Seq1.out. For each of these structures the
program runs 10,000 (default value) of Monte Carlo cycles (see
Note 2). The detailed description of the parameters can be found
on Rosetta website http://www.rosettacommons.org/manuals/
archive/rosetta3.4_user_guide/d2/d82/rna_denovo.html.
3.2.2 Sorting Structures The generated structures in output file Seq1.out are in order
Based on Their Score they were generated and not in order of their score. Therefore,
next step is to sort structures based on their score. The first line in
the Seq1.out is a sequence, e.g.:
SEQUENCE: ggcgucacaccuucgggugaagucgcccaccuucgg
Next line is the description of different score fields:

SCORE: score fa_atr fa_rep fa_intra_rep lk_nonpolar
hack_elec_rna_phos_phos ch_bond rna_torsion
rna_sugar_close hbond_sr_bb_sc hbond_lr_bb_sc
hbond_sc geom_sol atom_pair_constraint angle_constraint
linear_chainbreak rms rms_stem N_WC N_NWC N_BS
f_natWC f_natNWC f_natBP description
We used the first field “score” and the last one “description”
that has a tag for a given model, e.g., S_000001.
The sorting of the output results can be done using another
Perl script. The part of script that performs sorting looks like the
following:
$i = 0;
while ($line = <in_FILE>)
{
chomp($line);
@ARG = split(/ +/,$line);
($a1, $a2) = @ARG;
$name = pop(@ARG);
if (substr($a1, 0, 6) eq "SCORE:" && substr($name,
0, 2) eq "S_")
{
$S_name[$i] = $name;
$S_score[$i] = $a2;
++$i;
}
}
@hash{@S_score} = (0..$#S_score);
@S_sort = sort {$a <=> $b} @S_score;
@S_sort_index = @hash{@S_sort};
3.2.3 Extracting Best Next, we need to extract structures with the lowest score (score is
Structures negative; therefore, best score is the lowest score) from the Seq1.
out file into the *.pdb files. It is widely accepted that ligand binding
can drastically alter the receptor’s conformation [21]. To account
for such conformational flexibility, five to ten of the lowest score
structures for each sequence can be selected and placed into a
library of RNA molecules (see Note 3). This library will be used for
virtual high-throughput screening to select aptamers with binding
affinity to a specified small molecule.
To extract structures we use Rosetta extract_pdb command:
$ROSETTA_HOME/rosetta_source/bin/extract_
pdbs.<exe> @flags_extract -database $ROSETTA_
HOME/rosetta_database
where file flags_extract looks like the following:
-in:file:silent Seq1.out
-in:file:fullatom
-out:prefix Seq1
-in:file:tags S_000469 S_000021 S_000004 S_000145 S_000358
The last line specifies a list of structures with the lowest scores
that need to be extracted. By including five 3D structures we
perform ensemble docking and thus account for conformational
flexibility of the RNA molecules.
3.3 Virtual Screening Computational docking is a common tool used to identify small-
of the RNA Molecule molecule ligands that bind to proteins [22]. While most docking
Library methods have been developed mostly for proteins, recent evalua-
tion of AutoDock and DOCK programs has demonstrated their
ability to dock compounds to RNA molecules [16]. We used the
AutoDock Vina and ADT to prepare RNA molecules and small-
molecule ligand for docking and to perform docking.
In the previous section we generated PDB files with 3D coor-
dinates of RNA molecules. Now, we need to create a PDB file with
the coordinates of a small-molecule ligand. One way of doing that
is to use SMILES representation of the molecule. Usually, SMILES
representation of molecule can be found on Wikipedia on the right
side of the screen. Under the “Chemical data” section there is a tag
SMILES. Online SMILES translator at http://cactus.nci.nih.gov/
translate/ can be used to generate 3D coordinates from SMILES
string and to save them into PDB file. Another way to generate
coordinates is to use a commercial package to build a molecule and
to create a PDB file with the coordinates.
Next step is to prepare RNA and ligand molecules for docking
using AutoDock Vina. But, before we proceed to that step, we need
to center the RNA molecules around (0,0,0) coordinate. This can
be done using Open Babel package. The command is as follows:
babel -ipdb Seq1.pdb -opdb Seq1_Ct.pdb -C
where the output file Seq1_Ct.pdb contains the centered
coordinates of the molecule.
To prepare molecules for docking we can use GUI version of
ADT or scripts provided by ADT for batch mode. To use ADT,
please follow the tutorial at http://autodock.scripps.edu/faqs-
help/tutorial/using-autodock-4-with-autodocktools. To prepare
files in batch mode we used two commands provided by ADT. To
prepare ligand molecule:
prepare_ligand4.py -l ligand.pdb
The input file for this command is a PDB file with 3D of ligand
created above. As output it produces ligand.pdbqt file that
includes Gasteiger charges, AutoDock atom types, and special key-
words that establish the torsional flexibility of ligand molecule
(see Note 4).
The similar command is used to prepare receptor (in our case
RNA) molecules:
prepare_receptor4.py -r Seq1_Ct.pdb
The output file Seq1_Ct.pdbqt contains coordinates of

receptor with added charges, AutoDock atom types, and merging
nonpolar hydrogens. Two files, Seq1_Ct.pdbqt and ligand.
pdbqt, are required for docking software AutoDock Vina. Another
required file is the configuration file config.txt that contains
docking options for Vina. Here is the example of configuration file:
receptor = Seq1_Ct.pdbqt
ligand = ligand.pdbqt
center_x = 0
center_y = 0
center_z = 0
size_x = 25
size_y = 25
size_z = 25
The first two lines specify the receptor and ligand PDBQT
files. The next three lines define the center of the search space. We
use (0,0,0) coordinates as we centered out receptor molecule
before using babel command. The last three lines specify size of
the search space in X, Y, and Z directions for the ligand docking.
The full usage summary can be obtained using the command
“vina --help”. To run the docking of ligand molecule to
RNA molecule specified in file config.dock user need to use the
command “vina --config config.txt --log Seq1.log”.
Two output files are created: Seq1.log file with the summary of
docking results and ligand_out.pdbqt file with the predicted
ligand binding conformations. This file can be analyzed and visual-
ized using ADT.
To run virtual screening of many RNA molecules in Linux/
Mac environment we can use shell scripting. Suppose we are in a
directory containing ligand molecule in file ligand.pdbqt and
RNA molecules in files Seq1.pdbqt, Seq2.pdbqt, etc. We
want to run docking with the same ligand but different receptor
molecules. In the config.txt file we remove a line that specifies
receptor and that file looks now as follows:
ligand = ligand.pdbqt
center_x = 0
center_y = 0
center_z = 0
size_x = 25
size_y = 25
size_z = 25
Now, we are using the script in Bash shell to perform virtual
screening:
#! /bin/bash
for f in Seq*.pdbqt; do
r=`basename $f .pdbqt`
echo Processing receptor $r

vina --config config.txt --receptor $f --out
${r}_out.pdbqt --log ${r}.log
done
For each receptor molecule we have now separate result files
Seq*_out.pdbqt and Seq*.log.
Next step is to analyze the output files and select RNA mole-
cules with the highest binding affinity to the ligand molecule. At
this point, the selected RNA molecules are specific to the desired
target molecule, and they are placed into a pool of sequences for
experimental verification and selection of high-affinity aptamers,
e.g., using microarray technology (see Note 5). The accurate pre-
diction of the three-dimensional structures of RNA molecules and
modeling of RNA–small molecule interactions still faces consider-
able challenges [23, 24] and requires a careful protocol and analy-
sis of the results.
4 Notes
1. Our experience with the prediction of 3D structures using

Rosetta package shows that predicted structures are more
accurate with the use of parameter file with the secondary
structure obtained using ViennaRNA package.
2. Prediction of RNA tertiary structure requires computational
sampling of a huge conformational space. Typically it requires
generation of hundreds to thousands of models that is computa-
tionally very expensive. Therefore, there should be some compro-
mise between the number of structures generated in a reasonable
amount of time and the accuracy of predicted structures.
3. RNA molecules are very flexible, and ligand binding can drasti-
cally change the receptor conformation. To account for such
conformational flexibility, five to ten of the lowest score struc-
tures for each sequence should be selected and placed into a
library of RNA molecules.
4. By default, prepare_receptor4.py and prepare_
ligand4.py assign Gasteiger charges to both RNA mole-
cule and ligand. While for small molecules Gasteiger charges
work well, they are not the best charges for biomolecules,
including RNA. User can consider an option to assign charges
that were specifically designed for biomolecules such as
CHARMM [25] or AMBER [26] to RNA molecules. The
PDB file format does not have a field for atomic charges.
Therefore, RNA molecules should be saved in SYBYL mol2
file format which is readable by AutoDock. In this case the
prepare_receptor4.py should be used with option -C
that means to preserve assigned charges.
5. The biggest challenge in the modeling of RNA–small molecule

interaction lies in the highly charged nature of RNA molecules.
It results in strong solvation and the association of ionic mol-
ecules. It is well documented that divalent ions such as Mg2+
significantly affect the structure of RNA molecules and bind-
ing to small-molecule ligand. Unfortunately, these effects are
not incorporated into current versions of docking programs.
Therefore, we propose to take thousands of designed sequences
for experimental validation and screening.
Disclaimer
The opinions and assertions contained herein are the private views
of the authors and are not to be construed as official or as reflecting
the views of the US Air Force or the US Department of Defense.
This chapter has been approved for public release with unlimited
distribution.
References
1. Ellington AD, Szostak JW (1990) In vitro pools for in vitro selection. Bioinformatics
selection of RNA molecules that bind specific 23:2959–2960
ligands. Nature 346:818–822 10. Luo X, McKeague M, Pitre S et al (2010)
2. Tuerk C, Gold L (1990) Systematic evolution Computational approaches toward the design
of ligands by exponential enrichment: RNA of pools for the in vitro selection of complex
ligands to bacteriophage T4 DNA polymerase. aptamers. RNA 16:2252–2262
Science 249:505–510 11. Lorenz R, Bernhart SH, Hoener Zu
3. Osborne SE, Ellington AD (1997) Nucleic Siederdissen C et al (2011) ViennaRNA pack-
acid selection and the challenge of combinato- age 2.0. Algorithms Mol Biol 6:26
rial chemistry. Chem Rev 97:349–370 12. Mathews DH, Sabina J, Zuker M, Turner DH
4. Chushak Y, Stone MO (2009) In silico selec- (1999) Expanded sequence dependence of
tion of RNA aptamers. Nucleic Acids Res thermodynamic parameters improves predic-
37:e87 tion of RNA secondary structure. J Mol Biol
5. Davis JH, Szostak JW (2002) Isolation of 288:911–940
high-affinity GTP aptamers from partially 13. Mathews DH, Turner DH (2009) NNDB: the
structured RNA libraries. Proc Natl Acad Sci nearest neighbor parameter database for pre-
99:11616–11621 dicting stability of nucleic acid secondary
6. Carothers JM, Oestreich SC, Davis JH, structure. Nucleic Acids Res 38:280–282
Szostak JW (2004) Informational complexity 14. Leaver-Fay A, Tyka M, Lewis SM et al (2011)
and functional activity of RNA structures. J ROSETTA3: an object-oriented software suite
Am Chem Soc 126:5130–5135 for the simulation and design of macromole-
7. Ruff KM, Snyder TM, Liu DR (2010) cules. Methods Enzymol 487:545–574
Enhanced functional potential of nucleic acid 15. Das R, Karanicolas J, Baker D (2010) Atomic
aptamer libraries patterned to increase sec- accuracy in predicting and designing nonca-
ondary structure. J Am Chem Soc 132: nonical RNA structure. Nat Methods
9453–9464 7:291–294
8. Kim N, Gan HH, Schlick T (2007) A compu- 16. Detering C, Varani G (2004) Validation of
tational proposal for designing structured automated docking programs for docking and
RNA pools for in vitro selection of RNAs. database screening against RNA drug targets.
RNA 13:478–492 J Med Chem 47:4188–4201
9. Kim N, Shin JS, Elmetwaly S et al (2007) 17. Trott O, Olson AJ (2010) AutoDock Vina:
RagPools: RNA-as-graph-pools—a web server improving the speed and accuracy of dock-
for assisting the design of structured RNA ing with a new scoring function, efficient
optimization and multithreading. J Comput 22. Taylor RD, Jewsbury PJ, Essex JW (2002) A
Chem 31:455–461 review of protein-small molecule docking
18. Carothers JM, Oestreich SC, Szostak JW methods. J Comput Aided Mol Des 16:
(2006) Aptamers selected for higher-affinity 151–166
binding are not more specific for the target 23. Cruz JA, Blanchet MF, Boniecki M et al
ligand. J Am Chem Soc 128:7929–7937 (2012) RNA-Puzzles: a CASP-like evaluation
19. Patel DJ, Suri AK (2000) Structure, recogni- of RNA three-dimensional structure predic-
tion and discrimination in RNA aptamer com- tion. RNA 18:610–625
plexes with cofactors, amino acids, drugs and 24. Fulle S, Gohlke H (2010) Molecular
aminoglycoside antibiotics. J Biotechnol recognition of RNA: challenges for modelling
74:39–60 interactions and plasticity. J Mol Recognit
20. Laing C, Schlich T (2011) Computational 23:220–231
approaches to RNA structure prediction, anal- 25. Brooks BR, Brooks CL III, Mackerell AD Jr
ysis, and design. Curr Opin Struct Biol et al (2009) CHARMM: the biomolecular
21:306–318 simulation program. J Comput Chem
21. May A, Sieker F, Zacharias M (2008) How to 30:1545–1614
efficiently include receptor flexibility during 26. Case DA, Cheatham TE, Darden T et al
computational docking. Curr Comput Aided (2005) The amber biomolecular simulation
Drug Des 4:143–153 programs. J Comput Chem 26:1668–1688
Chapter 2
In Vitro Selection of RNA Aptamers

for a Small-Molecule Dye
Asako Murata and Shin-ichi Sato
Abstract
Artificial riboswitches can be generated by functional coupling of an RNA aptamer for synthetic small
molecule to an expression platform. RNA aptamers that can bind strongly and selectively to their target are
feasible to be used for obtaining more potent artificial riboswitches. In this chapter, we describe tips and
notes for in vitro selection of RNA aptamers targeting synthetic small molecules.
Key words RNA aptamer, In vitro selection, Small molecule, Riboswitch, Chemical biology
1 Introduction
Bacteria and some eukaryotes use natural RNA aptamers for

metabolite sensing and gene regulation [1–3]. These RNA aptamers
specifically bind to their targets and exhibit conformational changes
that are transduced into the changes in the secondary structure of
a downstream expression platform, resulting in a transcriptional or
a translational modulation of gene expression. This RNA-based
gene regulatory system, termed riboswitch, was first experimentally
validated by RR Breaker in 2002 [4]. To date, more than 20 classes
of natural riboswitches have been identified in many species [3].
The discovery of natural riboswitches has encouraged greater
interest among researchers who seek to create a new type of gene
expression system. In fact, a number of studies have demonstrated
the creation of “artificial riboswitches” for the conditional control
of gene expression in bacteria by engineering natural riboswitches
[5–7]. On the other hand, in vitro-selected RNA aptamers for
synthetic small molecules have also been employed to create artifi-
cial riboswitches [8–12].
In general, the use of RNA aptamers selected against synthetic
small molecules for controlling gene expression might offer several
advantages: orthogonality to endogenous system and choice of a
17
18 Asako Murata and Shin-ichi Sato
Randomized
T7 promoter
sequence (N60)
DNA pool
Transcription by T7
RNA Polymerase
Regeneration by
transcription
Binding to compound-
immobilized resin
RNA pool
DNA pool
Amplification by
PCR
cDNA
Washing
Reverse
transcription
Unbound RNA
Elution
Fig. 1 General scheme of in vitro selection
wide range of target ligands. RNA aptamers can be isolated from

a pool of random RNA sequences through a process called in
vitro selection or Systematic Evolution of Ligands by Exponential
enrichment (SELEX) (Fig. 1) [13]. In vitro selection experiments
consist of several steps: (1) preparation of a ligand-immobilized
matrix, (2) transcription of RNA from a synthetic pool of double-
stranded DNA, (3) isolation of RNA species that can specifically
bind to the ligand on the matrix, and (4) regeneration of dsDNA
pools through reverse transcription and PCR amplification. Steps
2–4 must be repeated several times to yield aptamers having
higher binding affinity to the ligand. Here we focus on steps 1
and 3 that can mostly affect the successful selection of high-affinity
aptamers from a random RNA pool and describe notes and tips
for these steps.
In the first step of an in vitro selection experiment, an affinity
matrix is prepared by immobilizing a target ligand on a resin.
Although there is no consensus for good targets for in vitro selec-
tion, several considerations may help researchers to choose target
ligands. Ligands that contain chemical moieties that are comple-
mentary to those of nucleic acids should be good targets. For
example, in vitro selections against aromatic compounds,
heterocyclic compounds, or nucleotide analogs have often been
Aptamers for Small-Molecules 19
Aptamer for BHQ-1

Fluorophore Aptamer tag
(Quenched) mRNA
Fluorescent !
Quencher dye Detection of
(BHQ-1) target RNA
Fig. 2 Application of in vitro-selected aptamers against a small-molecule dye for RNA tagging
successful. In addition, positively charged functional groups

(e.g., primary amino groups) of a ligand may facilitate an aptamer
selection, as they increase binding affinity to RNA by electrostatic
interaction with the phosphate backbone. Next, the ligand of
choice can be attached to a resin for making a ligand-immobilized
matrix. There are two major methods available for ligand attach-
ment: one is covalent coupling of ligands to a reactive functional
group on matrix and another is immobilization of biotinylated
ligands on avidin-coated matrix. Typically, the former coupling
method gives the matrix with higher concentration of ligand
(~20 mM) than that of the latter one (~100 nM), although it may
depend on the concentration of functional groups that were origi-
nally available and the coupling efficiency. The concentration of
the ligand on the matrix can significantly influence the course of a
selection experiment. Decreasing the concentration of the ligand
on the matrix increases the chance for having aptamers that bind
more strongly to the target ligand, encouraging competition for a
smaller number of binding sites.
Binding and elution conditions also directly affect the strin-
gency of the selections. Binding buffer containing monovalent and
divalent cations (Na+, K+, and Mg2+) may be used to stabilize RNA
structures, which promotes the formation of specific aptamer
structures or interactions with the ligand. Negative selection to
remove matrix-binding species using an underivatized resin is
critically important to obtain specific aptamers, because it may be
difficult to select aptamers if once the matrix binders accumulate
and become dominant. To obtain specific RNA aptamers to the
target ligand, affinity elution with the specific ligand is highly pre-
ferred, though denaturing agents such as 6–8 M urea and 4 M
guanidine can elute bound RNA from the matrix by denaturing
secondary structures of RNA.
We present our recent work on in vitro selection of RNA
aptamers targeting a small-molecule dye (Black hole quencher-1:
BHQ-1). Our study has focused on utilizing the aptamers as a tag
for RNA detection in living cells in combination with the designed
fluorescent probe (Fig. 2) [14]. In vitro selection was performed
to isolate aptamers for BHQ-1 from an RNA pool containing a
60-base random sequence. After 13 rounds of selection against a

BHQ-1-immobilized resin, we obtained four RNA sequences
(A1–A4) that shared a conserved sequence of 17 nucleotides. A1–
A3 had similar Kd values around 5 μM, whereas A4 that contained
a one-base mismatch in the conserved sequence had three times
lower affinity than A1–A3, suggesting that the sequence of the
conserved region is important in the recognition of BHQ-1. A1–
A4 were predicted to form a stem–loop, with the conserved
sequence in the loop region. The prevention of stem–loop forma-
tion caused a marked decrease in binding affinity of these aptamers,
indicating that the stem–loop is the minimal binding motif. In fact,
a short version of A1 containing only the stem–loop retained
binding ability to BHQ-1 with lower affinity than parental A1.
Using A1 sequence as a tag, a specific mRNA was successfully
detected by restoration of fluorescence probe in an E. coli lysate.
2 Materials
2.1 Equipment 1. Thermal cycler.

2. Horizontal and vertical electrophoresis system.
3. UV transilluminator.
4. Tabletop centrifuge.
5. Real-Time PCR System (7500 Fast Real-Time PCR System,
Applied Biosystems).
6. Microplate reader (MTP-800, CORONA).
7. Fluorophotometer (LS-55, PerkinElmer).
2.2 Reagents 1. BHQ-1 carboxylic acid, BHQ-1 amine (Bioresearch Technologies).

and Solutions 2. 4-(4,6-Dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium
chloride: DMT-MM (KOKUSAN CHEMICAL Co., Ltd).
3. Triethylamine (Wako Pure Chemical Industries, Ltd).
4. Platinum® Pfx DNA polymerase (Invitrogen).
5. 10 mM dNTP mix (Invitrogen).
6. T7 RNA polymerase (TAKARA).
7. 25 mM Nucleotides rNTPs Mixture (TOYOBO).
8. RQ1 DNase I (Promega).
9. Prime Script reverse transcriptase (TAKARA).
10. DNA Ligation Kit (TAKARA).
11. Plasmid isolation kit (QIAGEN).
12. MEGAscript T7 Kit (Ambion).
13.
In-Fusion Dry-Down PCR Cloning Kit (Clontech
Laboratory Inc.).
14. RTS 100 E. coli HY Kit, 5 (Prime Inc.).

15. 3 M sodium acetate solution (pH 5.2).
16. 7.5 M ammonium acetate solution (Sigma-Aldrich).
17. Annealing buffer: 10 mM Tris–HCl, pH 7.5, 100 mM KCl,
50 mM NaCl.
18. Binding buffer: 10 mM Tris–HCl, pH 7.5, 100 mM KCl,
50 mM NaCl, 5 mM MgCl2.
19. 250 mM MgCl2.
2.3 Supplies 1. Microtubes, 1.5 and 2.0 mL.

2. Falcon tubes, 15 and 50 mL.
3. 0.2 mL PCR tubes.
4. EAH sepharose™ 4B resins (GE Healthcare).
5. Poly-Prep Chromatography Columns (BIO-RAD).
6. Micro Bio-Spin Chromatography Columns (BIO-RAD).
7. NAP-5 columns (GE Healthcare).
8. 96-well glass bottom black plates (655896, Greiner Bio-One).
3 Methods
All solutions and buffers should be made with DEPC-treated

water. Use RNase/DNase-free disposable plasticware and micropi-
pet tips where possible.
3.1 Preparation 1. Place 0.5 mL of EAH sepharose™ 4B resin in a Poly-Prep

of BHQ-1-Immobilized Chromatography Column and allow to drain by gravity.
Resin 2. Wash the resin thoroughly with at least three bed volumes of
dioxane–H2O (3:1, v/v).
3. Add BHQ-1 carboxylic acid (0.5 mg, 1 μmol), triethylamine
(0.1 mmol), and DMT-MM (29.5 mg, 0.1 μmol) in 0.5 mL of
dioxane–H2O (3:1, v/v) to the drained resin, and shake the
suspension at room temperature overnight (see Note 1).
4. Drain the coupling solution from the resin, and add the freshly
prepared acetic acid (30 mg, 0.5 mmol) and DMT-MM
(147.4 mg, 0.5 mmol) in 0.5 mL of dioxane–H2O (3:1, v/v)
to block unreacted amino groups on the resin.
5. After overnight incubation, drain the resin and wash thor-
oughly with dioxane–H2O (1:1, v/v) followed by methanol–
H2O (1:1, v/v).
6. Suspend the resin with three bed volumes of methanol–H2O
(1:3, v/v) to make a 25 % slurry. Store at −20 °C until use.
Table 1
Reaction mixture for PCR amplification of DNA pools
Template DNA*1 10 μL

10× Pfx amplification buffer 50 μL
10 mM dNTP mixture 15 μL
50 mM MgSO4 10 μL
Forward primer*2 5 μL
Reverse primer*3
5 μL
Platinum Pfx DNA polymerase 10 μL
d.D.W 395 μL
Total 500 μL
3.2 In Vitro Selection In vitro selection was performed to isolate RNA aptamers for
Procedures BHQ-1 according to the standard procedure (Fig. 1).
3.2.1 Preparation 1. Assemble the reaction mix in PCR tubes as described in Table 1
of DNA Pools (see Note 2). *1Template DNA: 5′-GAA TTC CGC GTG TGC
ACA CC-N60–GTC CGT TGG GAT CCT CAT GG-3′ (for
preparation of first DNA pool a reverse-transcribed product
was used for the other selection round). *2Forward primer:
5′-GCT AAT ACG ACT CAC TAT AGG GAA TTC CGC
GTG TGC ACA CC-3′ (T7 promoter sequence is underlined).
*3
Reverse primer: 5′-CCA TGA GGA TCC GAA CGG AG-3′.
2. Place the tubes into a thermal cycler and perform the PCR
amplification using the following program: 2-min initial
denaturation at 94 °C, followed by sequential cycles of dena-
turation at 94 °C for 30 s, annealing at 55 °C for 30 s, exten-
sion at 68 °C for 1 min, and final extension at 68 °C for 7 min
(see Note 3).
3. Precipitate the DNA by adding a half volume of 7.5 M ammo-
nium acetate solution and 1.5 volume of isopropanol. Centrifuge
the sample at 10,000 × g for 30 min to pellet the DNA.
4. Remove the supernatant carefully, and rinse the pellet with
70 % ethanol.
5. Air-dry the pellet for 5 min, and dissolve it in 50 μL of RNase-
free water.
3.2.2 Preparation 1. Assemble the transcription reaction at room temperature in

of RNA Pools the order shown below (Table 2).
2. Incubate the reaction mix at 37 °C overnight.
3. Add 10 μL of DNase I and incubate for 30 min at 37 °C.
Table 2
Reaction mixture for in vitro transcription of reaction
Template DNA X μL

10× T7 RNA polymerase buffer 50 μL
NTP mixture (25 mM each) 15 μL
50 mM DTT 10 μL
T7 RNA polymerase 5 μL
RNase-free water 420 − X μL
Total 500 μL
4. Precipitate the RNA by adding a half volume of 7.5 M

ammonium acetate solution and 1.5 volume of isopropanol.
Centrifuge the sample at 10,000 × g for 30 min at 4 °C to pellet
the RNA.
5. Thoroughly remove the supernatant, and dissolve the pellet in
0.5 mL of RNase-free water.
6. Apply the sample to the pre-equilibrated NAP-5 column, and
elute the sample with 1 mL of RNase-free water (see Note 4).
7. Precipitate the RNA by adding 0.5 mL of 7.5 M ammonium
acetate solution and 1.5 mL of isopropanol. Centrifuge the
sample at 10,000 × g for 30 min at 4 °C to pellet the RNA.
70 % ethanol.
9. Air-dry the pellet for 5 min, and dissolve it in 100 μL of the
annealing buffer.
10. Denature the RNA pool for 3 min at 80 °C, and then let it fold
by slow cooling to room temperature (see Note 5).
11. Add 2 μL of MgCl2 solution to a final concentration of 5 mM
(see Note 6).
3.2.3 Binding of RNA 1. Place 50 μL of the BHQ-1-immobilized resin (25 % slurry)
Pool to BHQ-Immobilized into an empty column (Micro Bio-Spin Chromatography
Resin Columns) (see Note 7), and wash the resin three times with
150 μL of the binding buffer.
2. Apply the RNA pool solution to the resin and incubate for
30 min on ice. Vortex the sample briefly every 10 min.
3. Drain and wash the resin to remove the unbound RNA species
(see Note 6).
4. Elute bound RNA species three times with 130 μL of the binding
buffer saturated with free BHQ-1 amine (see Note 7).
Table 3
Reverse transcription reaction
a
Template RNA 8 μL
10 mM dNTP mix 1 μL
10 μM reverse primer* 1 μL
Total 10 μL
[*Reverse primer: 5′-CCATGAGGATCCGAACGG AC-3′]
↓ 65 °C, 5 min
b
Template RNA/primer mixture 10 μL
5× PrimeScript buffer 4 μL
RNase inhibitor 0.5 μL
PrimeScript reverse transcriptase 1 μL
RNase-free water 4.5 μL
Total 20 μL
↓ 42 °C, 60 min, then 70 °C, 15 min
5. Collect the eluate in a 1.5 mL collection tube, and add 43 μL

of 3 M sodium acetate solution and 1 mL of ethanol to precipi-
tate recovered RNA. Centrifuge the sample at 10,000 × g for
30 min at 4 °C.
70 % ethanol.
7. Air-dry the pellet for 5 min, and dissolve it in 8 μL of RNase-
free water.
3.2.4 Reverse 1. Prepare the following mixture (Table 3a) in a PCR tube. Heat
Transcription the sample for 5 min at 65 °C and then chill immediately on
ice. Assemble the reaction mix as shown in Table 3b, and per-
form cDNA synthesis reaction for 60 min at 42 °C, followed
by heat-inactivation of RTase at 70 °C for 15 min.
2. Use 10 μL of the reverse-transcribed product as a template for
the PCR amplification to generate the DNA pool for the next
round of selection (see Note 8).
3. Repeat the selection steps for several rounds to enrich the RNA
species with high affinity to BHQ-1.
3.3 Cloning and 1. After the final round of selection, digest the resulting DNA
Sequencing of Each pool with EcoRI and BamHI as well as pUC19.
RNA Species in the 2. Ligate the digested insert and vector using DNA Ligation Kit.
Enriched Pool
3. Transform DH5α with the ligation product, and heat shock
and spread the cells on LB plate containing 100 μg/mL of
ampicillin. Incubate the plate at 37 °C overnight.
4. Culture from a single colony, and isolate the plasmid using
standard plasmid isolation procedures.
5. Sequence the insert using primers in the vector backbone.
3.4 Fluorescence To determine if the selected aptamers have the ability to restore the
Measurements fluorescence of the fluorophore–BHQ-1 conjugates, fluorescence
titration of the fluorophore–BHQ-1 conjugates with the aptamers
can be performed. The synthesis of the fluorophore–BHQ-1 con-
jugate was described in our previous report [14].
1. Prepare the dsDNA template containing T7 promoter and each
BHQ-1 aptamer by PCR amplification using the individual clone
as a template. Primers used in the reaction are as follows: the for-
ward primer [5′-GCT AAT ACG ACT CAC TAT AGG GAA
TTC CGC GTG TGC ACA CC-3′ (T7 promoter sequence is
underlined)] and the reverse primer [5′-CCA TGA GGA TCC
GAA CGG AC-3′].
2. Perform a transcription reaction, and purify RNA as described
above in Subheading 3.2.2.
3. Prepare a solution of conjugate AL at the concentration of
2 μM in the binding buffer.
4. Prepare RNA solution at the concentrations of 0, 0.6, 2, 6, 20,
60, and 200 μM in the binding buffer.
5. Mix an equal volume of the RNA solution with the conjugate
solution, and measure the fluorescence of each sample using either
a plate reader or a fluorophotometer. The excitation maximum of
each conjugate and the emission maximum in the RNA-bound
state are as follows: conjugate AL (Ex/Em = 543 nm/610 nm),
conjugate Cy (Ex/Em = 519 nm/562 nm), and conjugate FL
(Ex/Em = 497 nm/515 nm).
The Kd values of each aptamer can be obtained by fitting fluo-
rescence titration data using the following equation:

æ ([conjugate ] + [aptamer ] + K d ) - ö
ç T T
÷
( )
1/ 2
ç ([conjugate ] + [aptamer ] + K )2 - 4 [conjugate ] [aptamer ] ÷
ç T T d T T ÷
Fobs = Aç ÷
è 2 [conjugate ]T ø
Table 4
In transcription/translation reaction mixture
Final conc.
Template DNA X μL 0.5 μg
E. coli lysate 4.8 μL
Reaction mix 4 μL
Amino acids 4.8 μL
Methionine 0.4 μL
Reconstitution buffer 2 μL
100 μM fluorophore–BHQ-1 conjugate 0.5 μL 2.5 μM
RNase-free water 3.5 − X μL
Total 20 μL
where A is the increase in fluorescence at saturating aptamer

concentrations (Fmax − Fmin), Kd is the dissociation constant, and
[conjugate]T and [aptamer]T are the total concentrations of a
conjugate and an aptamer, respectively.
3.5 Real-Time Detailed protocol of the preparation of template DNAs is described

Monitoring of Protein in our previous report [14].
Synthesis
1. Assemble the reaction according to Table 4.
2. Incubate the reaction mixture at 30 °C, and collect the fluores-
cence emission data every 3 min at appropriate wavelength
by using a real-time PCR system with the filter for FAM and
ROX dye.
3. After 1-h incubation, add DNase I to the mixture and incubate
for 30 min at 37 °C to digest the template DNA.
4. Analyze the resulting reaction to verify RNA and protein
synthesis by RT-PCR and western blotting, respectively.
4 Notes
1. We use DMT-MM as a coupling reagent, but any other water-

soluble coupling reagents such as EDC can be used.
2. For the preparation of first DNA pool, use a sufficient amount
of template DNA to minimize the loss of individual sequences
in the random library.
3. Perform a cycle course PCR amplification to determine the
relative quantities of the bound RNA species using one-tenth
aliquot of the reaction mixture. For each sample, 4 μL of aliquots
are removed from the reaction after 4, 8, 12, 16, 20, 24, 28,
and 32 cycles and separated on an 8 % native polyacrylamide
gel. As the population of binding species becomes enriched, a
band of the desired product will appear at an earlier PCR cycle
than the previous selection round.
4. We use NAP-5 column to remove unincorporated NTPs, but
any other gel filtration system can be used.
5. Do not heat the RNA solution in the presence of Mg2+. This is
important to avoid possible chemical degradation of RNA with
divalent cations.
6. We usually measure the UV absorbance of the RNA pool
solution before and after binding to the resin to determine the
percentage of bound RNA in each cycle. The percentage of
RNA bound to the resin is calculated as 100(AbsT − AbsF)/
AbsT, where AbsT is UV absorbance of the RNA solution
before binding to the resin, and AbsF is UV absorbance of the
flow-through fraction after binding to the resin.
7. Do not use a spin column with a microporous membrane such
as PVDF, because BHQ-1 amine may adsorb to the membrane,
probably due to the hydrophobicity of BHQ-1.
8. We usually save a half of the reverse-transcribed product for
archival purposes. If subsequent amplification or selection
reactions fail for any reason, the procedure can be started from
the archival pool.
Acknowledgment
We are indebted to Professor M. Uesugi of Kyoto University for

helpful suggestions.
References
1. Roth A, Breaker RR (2009) The structural and 6. Wieland M, Benz A, Klauser B, Hartig JS
functional diversity of metabolite-binding (2009) Artificial ribozyme switches containing
riboswitches. Annu Rev Biochem 78:305–334 natural riboswitch aptamer domains. Angew
2. Breaker RR (2011) Prospects for riboswitch Chem Int Ed 48:2715–2718
discovery and analysis. Mol Cell 16(43): 7. Dixon N, Duncan JN, Geerlings T, Dunstan
867–879 MS, McCarthy JE, Leys D, Micklefield J
3. Serganov A, Nudler E (2013) A decade of (2010) Reengineering orthogonally selective
riboswitches. Cell 152:17–24 riboswitches. Proc Natl Acad Sci U S A 107:
4. Winkler W, Nahvi A, Breaker RR (2002) 2830–2835
Thiamine derivatives bind messenger RNAs 8. Werstuck G, Green MR (1998) Controlling
directly to regulate bacterial gene expression. gene expression in living cells through small mol-
Nature 419:952–956 ecule-RNA interactions. Science 282:296–298
5. Nomura Y, Yokobayashi Y (2007) Reengineering 9. Suess B, Hanson S, Berens C, Fink B,
a natural riboswitch by dual genetic selection. Schroeder R, Hillen W (2003) Conditional
J Am Chem Soc 129:13814–13815 gene expression by controlling translation with
tetracycline-
binding aptamers. Nucleic Acids 12. Weigand JE, Sanchez M, Gunnesch EB, Zeiher
Res 31:1853–1858 S, Schroeder R, Suess B (2008) Screening for
10. Suess B, Fink B, Berens C, Stentz R, Hillen W engineered neomycin riboswitches that con-
(2004) A theophylline responsive riboswitch trol translation initiation. RNA 14:89–97
based on helix slipping controls gene expres- 13. Ellington AD, Szostak JW (1990) In vitro
sion in vivo. Nucleic Acids Res 32:1610–1614 selection of RNA molecules that bind specific
11. Lynch SA, Desai SK, Sajja HK, Gallivan JP ligands. Nature 346:818–822
(2007) A high-throughput screen for synthetic 14. Murata A, Sato S, Kawazoe Y, Uesugi M (2011)
riboswitches reveals mechanistic insights into Small-molecule fluorescent probes for specific
their function. Chem Biol 14:173–184 RNA targets. Chem Commun 47:4712–4714
Chapter 3
Development of Photoswitchable RNA

Aptamer–Ligand Complexes
Gosuke Hayashi and Kazuhiko Nakatani
Abstract
Photoresponsive artificial riboswitch has the potential to offer a de novo method for spatiotemporal control
of gene expression in living cells. Because, even today, it is difficult to design a small molecule binding to
a specific RNA sequence, generating an artificial riboswitch that possesses highly specific affinity to a ligand
of interest basically depends on in vitro selection procedure where a variety of RNA–ligand complexes can
be obtained in established methods. Here, we describe the protocol for in vitro aptamer selection against
a photoresponsive peptide ligand containing azobenzene moiety that undergoes photoisomerization
through light irradiation. Furthermore, we explain a procedure for surface plasmon resonance assay to
detect photoswitchable association and dissociation of RNA–ligand complex on gold surface.
Key words In vitro selection, RNA aptamer, Azobenzene, Photoresponsive association and dissocia-
tion, Gene regulation
1 Introduction
Riboswitches are functional RNA structures usually positioned at 5′

or 3′ untranslated regions of mRNA, where they interact with spe-
cific metabolite and control gene expression. Upon metabolite
binding, alteration of mRNA secondary structure occurs followed
by translational or transcriptional modulation [1–3]. To date, about
20 classes of riboswitches have been discovered, which recognize
amino acids, nucleobases, sugars, and coenzymes [4]. Such trans-
acting mechanism of riboswitching is attractive for engineers who
aim to develop artificial gene regulation tools. Actually, creating
artificial riboswitches started more than a decade ago [5, 6], and, in
recent years, a broader range of small molecules such as tetracy-
cline, theophylline, and atrazine have been used as their ligands
[7–10]. A specific RNA–ligand complex, which is a key component
of riboswitch, has been obtained through in vitro RNA aptamer
selection procedure from RNA library containing random sequences
[11]. Considering that gene expression levels of these synthetic and
29
30 Gosuke Hayashi and Kazuhiko Nakatani
natural riboswitches depend on concentration of their ligands, it

takes considerable period to turn on/off gene expression.
Furthermore, as the small molecular ligands freely diffuse in cell,
spatiotemporal modulation of gene expression is extremely diffi-
cult. We hypothesized that photoresponsive riboswitches that catch
and release their ligands upon light irradiation would provide a
powerful technology to rapidly turn on/off gene expression. To
this end, we have developed photoresponsive ligand–aptamer com-
plexes, in which aptamer binding to the ligand that undergoes pho-
toisomerization upon photoirradiation can be controlled in a
repetitive fashion [12, 13]. Photoresponsive peptide KRAzR
(Lys-Arg-azobenzene-Arg) containing azobenzene chromophore
was designed and synthesized as a ligand. Here, we report proce-
dures for in vitro RNA aptamer selection against KRAzR and sur-
face plasmon resonance (SPR) analysis, demonstrating that
interactions between KRAzR and selected aptamers were success-
fully controlled by appropriate irradiation to the SPR sensor chip.
2 Materials
All samples and solutions were prepared using ultrapure water

unless otherwise specified.
2.1 In Vitro Selection 1. Affi-Gel 10® Gel (Bio-Rad).

Components 2. 50 mM KRAzR peptide in water. After solid-phase peptide
synthesis, weigh the lyophilized KRAzR powder and add
appropriate amount of water. The synthesis method was previ-
ously reported [12].
3. Econo column (Bio-Rad) for packing selection gels.
4. Buffer Kit (Life technologies). This kit contains RNase-free
solutions including 1 M Tris, pH 7.0, 1 M Tris, pH 8.0, 5 M
NaCl, 1 M MgCl2, 0.5 M EDTA, pH 8.0, 2 M KCl, 5 M
ammonium acetate, and DEPC-treated H2O.
5. TaKaRa Ex Taq® (TaKaRa).
6. AmpliScribe™ T7 High Yield Transcription Kit (Epicentre).
7. Binding buffer (5×): 250 mM Tris, pH 8.0, 1.25 M NaCl,
25 mM MgCl2. Mix 23.75 mL of water, 12.5 mL of 1 M Tris,
pH 8.0, 12.5 mL of 5 M NaCl, and 1.25 mL of 1 M of MgCl2
to prepare 50 mL of binding buffer (5×) (see Note 1).
8. Reverse Transcription System (Promega).
9. Gel washing solution: 1 mM HCl.
10. Coupling buffer: 50 mM NaHCO3, pH 8.
11. Glycine.
12. Synthetic oligonucleotide (131 mer, 5′-GCTAATACG
ACTCACTATAGGGAATTCCGCGTGTGCACACC(N70)
Photoswitchable Ligand-Aptamer Complexes 31
GTCCGTTCGGATCCTCATGG-3′) containing 70 random

nucleotides, T7 promoter (underlined), and recognition sites
for restriction enzymes, EcoRI and BamHI (bold).
13. Primers for DNA library construction: forward primer,
5′-GCTAATACGA CTCACTATAGGGAATTCCGCGTGTG
CACACC-3′, reverse primer, 5′-CCATGA GGATCCGAAC
GGAC-3′.
14. 2-Propanol.
15. 3 M Sodium acetate buffer solution, pH 5.2.
16. 70 % Ethanol.
17. NAP-10 Columns (GE Healthcare).
18. TE buffer solution, pH 8.0.
19. Dr. GenTLE (TaKaRa).
20. BamHI.
21. EcoRI.
22. Mighty Mix (TaKaRa).
23. pUC18 DNA (TaKaRa).
24. ECOS™ Competent E. coli DH5a (Nippon Gene).
25. Difco™ LB Broth, Miller (BD).
26. Agar.
27. Ampicillin sodium.
28. QIAprep Spin Mniprep Kit (50) (QIAGEN).
2.2 Biacore Assay 1. Biacore 2000 (GE Healthcare).

Components 2. Sensor Chip CM5 (GE Healthcare).
3. Running buffer for aptamer binding assay: Same as the binding
buffer used in aptamer selection.
4. 10 mM NaOH.
5. Amine coupling kit (GE Healthcare) contains EDC, NHS, and
1 M ethanolamine hydrochloride solution, pH 8.5 for immo-
bilization of KRAzR peptide to CM5 sensor chip.
6. Running buffer for ligand immobilization: HBS-N Buffer
(GE Healthcare).
7. MAX-302 Xenon Light Source 300 W (ASAHI SPECTRA).
8. αCLEAN DUSTER ECO (TRUSCO).
3 Methods
Carry out all procedures at room temperature and under room

light unless otherwise specified.
3.1 In Vitro Selection 1. Shake the bottle of Affi-Gel 10® Gel to uniformly suspend gel.
Against KRAzR Peptide 2. Take 1 mL of the agarose gel (about 2 mL of the slurry) into
3.1.1 Preparation Econo column (see Note 2).
of KRAzR- and Glycine- 3. Discard the preloaded solution from the bottom of the
Agarose Columns column.
4. Add 3-column volume of gel washing solution (3 mL) to com-
pletely remove the preloaded solution.
5. Equilibrate the gel by adding 2-column volume of coupling
buffer (2 mL).
6. Cap the bottom of the column and add 2 mL of 5 mM KRAzR
dissolved in the coupling buffer.
7. Leave the column on a rotating shaker for 1 h (see Note 3).
8. Drain the reaction solution, and then wash the column with
5-column volume of binding buffer (5 mL).
9. To make agarose gel for negative selection (see Note 4), use
0.5 mL of gel slurry and 1 mL of 10 mM glycine dissolved in
coupling buffer instead of 1 mL gel and 5 mM KRAzR solu-
tion, respectively. The other procedures are same as that
described above.
3.1.2 DNA Library 1. In a 15 mL tube, prepare 8 mL of PCR reaction mixture con-

Construction taining about 70 μg of the 131 mer synthetic oligonucleotide
containing a randomized region, 1 μM of forward and reverse
primers, 200 μM of each dNTP, and 200 units of Ex Taq.
2. Dispense the reaction mixture into 80 tubes (100 μL each),
and perform PCR in four cycles of 98 °C, 10 s; 55 °C, 30 s;
and 72 °C, 30 s (see Note 5).
3. After PCR reaction, mix together the PCR solution in 80 tubes
to two 15 mL tubes (4 mL each) and then add 400 μL of 3 M
sodium acetate buffer solution and 4 mL of 2-propanol to each
tube.
4. Vortex the mixture and leave them on ice for 10 min.
Centrifuge the tubes at 13,000 rpm for 15 min at 4 °C, and
discard the supernatant.
5. Add 3 mL of 70 % ethanol and centrifuge at 13,000 rpm for
5 min.
6. Discard the supernatant, and dry the pellet for 10 min.
7. Add 250 μL of water to both the tubes and mix them together
(see Note 6).
3.1.3 RNA Pool 1. Prepare 2.7 mL of transcription reaction containing 170 μg of

Construction DNA library (PCR product), 10 mM DTT, 7.5 mM each
NTP, and appropriate amount of T7 RNA polymerase and
RNase inhibitor according to the manufacturer’s protocol.
2. Dispense the reaction mixture into 27 tubes (100 μL each) and

incubate at 37 °C for 6 h.
3. Add 5 units (5 μL) of RNase-free DNase I to each tube and
incubate at 37 °C for 30 min.
4. Directly charge the reaction mixture onto an equilibrated
NAP-10 column. For the use of NAP-10 columns including
column preparation and sample elution, see the manufacturer’s
instruction.
5. Add appropriate amount of 3 M sodium acetate buffer solution
and 2-propanol to the eluted sample from NAP-10 column for
further purification by 2-propanol precipitation. Following
procedure is same as that described in the previous section.
6. Dissolve the pellet into 200 μL of TE buffer.
7. Measure UV260 to calculate the concentration of the obtained
RNA pool (see Note 7).
3.1.4 Selection 1. Equilibrate KRAzR-agarose for positive selection and glycine-

agarose for negative selection with 5 and 2.5 mL of binding
buffer, respectively.
2. Fix both the columns to a pole stand and tandemly place
glycine-agarose column over KRAzR column (Fig. 1).
3. Prepare 500 μL of RNA pool containing 550 μg of transcribed
RNA dissolved in binding buffer, and load total volume of the
Fig. 1 Apparatus used in our in vitro selection against KRAzR peptide. Glycine-
agarose packed column and KRAzR-agarose packed column were tandemly
placed to efficiently transfer flow-through of glycine column onto KRAzR column
sample onto glycine column for negative selection. Discard

eluted solution during the sample loading.
4. Rinse the glycine column with 1 mL of binding buffer, and
load the flow-through directly onto the KRAzR column.
5. Add 4.5 mL of binding buffer to wash out non-binding RNAs.
Discard the flow-through.
6. Prepare 3 mL of 6 mM KRAzR solution dissolved in binding
buffer for specific elution, and add the total volume onto the
column. Discard first 500 μL of the eluate, and then start col-
lecting the flow-through.
7. When the elution is done, add 3 mL of binding buffer and col-
lect all eluate (see Notes 8 and 9).
8. Mix all fractions together to a 15 mL tube and add appropriate
amount of 3 M sodium acetate buffer solution (one-tenth vol-
ume of collected solutions), 2-propanol (same volume of col-
lected solutions), and Dr. GenTLE (precipitation carrier) (see
Note 10) for 2-propanol precipitation.
9. Vortex the solution, and leave the tube on ice for 10 min.
Centrifuge the tubes at 13,000 rpm for 15 min at 4 °C, and
discard the supernatant.
10. Add 3 mL of 70 % ethanol and centrifuge at 13,000 rpm for
5 min. Discard the supernatant, and dry the pellet for 10 min.
11. Dissolve the pellet with 20 μL of TE buffer.
3.1.5 Reverse 1. Incubate the recovered RNA solution at 80 °C for 3 min, and
Transcription (RT)-PCR then place it on ice (see Note 11).
2. Add appropriate amount of dNTPs, reverse transcription buf-
fer, MgCl2, RNase inhibitor, primer, and AMV reverse tran-
scriptase according to the manufacturer’s protocol.
3. Incubate the total volume of 50 μL reaction mixture at 42 °C
for 30 min.
4. Using all of RT reaction mixture, prepare 500 μL of PCR reac-
tion with ExTaq polymerase, according to the procedure as
described above. Dispense the reaction mixture into ten tubes
(50 μL each).
5. To determine the optimum PCR cycles, begin with the 50 μL
volume and take 5 μL in each four cycles. PCR amplification
should be done in the condition of 98 °C, 10 s; 55 °C, 30 s;
and 72 °C, 30 s.
6. From PAGE analysis, determine the appropriate PCR cycle
number (Fig. 2a) (see Note 12). In our case, we chose 11 cycles.
7. Run PCR reaction with the remaining 450 μL of the reaction
mixture, and check the PCR product by PAGE (Fig. 2b).
Fig. 2 Native PAGE analysis of RT-PCR products obtained at first round of selection. (a) Optimization of PCR
cycles using 50 μL of PCR mixture. In each four cycles, 5 μL of the reaction mixture was taken for PAGE analysis.
After electrophoresis, the gel was stained with SYBR Green. (b) Confirmation of migration of RT-PCR products
(11 cycles) obtained in large-scale reaction. After electrophoresis, the gel was stained with SYBR Green
8. Purify the obtained solution by 2-propanol precipitation as

described above, and dissolve the obtained pellet in 50 μL of
water.
3.1.6 Next Rounds 1. To proceed to the second round of the selection, prepare
of Selection 200 μL of transcription mixture and react with the same
condition described above using half of the transcribed RNA
for loading to selection column. The other half should be
stored in −20 °C refrigerator. To cycle selection procedures
including negative and positive selection, RT-PCR, and
transcription, basically follow the protocol described above
(see Notes 13 and 14).
2. When you confirm the enrichment of RNA bound to the
KRAzR-gel after several rounds of selection (see Note 15),
stop the selection cycle after RT-PCR.
3.1.7 Isolation and 1. Add BamHI and EcoRI to the obtained PCR product (DNA
Identification of Aptamer library), which is dissolved in appropriate buffer according to
Sequences the manufacturer’s protocol and incubate at 37 °C for 3 h.
2. Treat pUC18 in the same condition in a 200 μL tube.
3. After 37 °C incubation, place both the DNA library and
pUC18 at 70 °C for 10 min for deactivation of restriction
enzymes.
4. Prepare 10 μL of ligation solution including 5 μL of Mighty
Mix, about 25 fmol of cleaved pUC18, and about 25–250 fmol
of cleaved DNA library.
5. Incubate the mixture at 16 °C for 60 min.

6. Prepare 50 μL of chemical competent cell, ECOS™ Competent
E. coli DH5a, and add 5 μL of the ligation mix. For the trans-
formation, use general heat shock procedure.
7. Add 450 μL of LB medium to the competent cell, and incu-
bate it at 37 °C for 1 h.
8. Spread the solution directly or after dilution on the LB-agar
plates containing 100 μg/mL of ampicillin and 40 μg/mL of
X-Gal and incubate at 37 °C overnight.
9. Pick the white colonies (blue-white selection), and check the
insertion of DNA library by colony PCR.
10. Grow the E. coli having proper length of insert DNA on the
vector DNA in 5 mL LB medium containing 100 μg/mL of
ampicillin.
11. Extract and purify the plasmid DNA by QIAprep Spin Mniprep
Kit.
12. Determine the concentration of obtained plasmid DNAs by
measuring the UV260.
13. Read the sequences of obtained RNA aptamers.
3.2 SPR Analysis Basically, follow the instruction manual of Biacore 2000. Use CM5
of Photoresponsive sensor chip throughout all procedures. Immobilization of KRAzR
Interaction Between onto the gold surface was done through amide bond formation.
RNA Aptamers and
KRAzR Peptide
3.2.1 Ligand 1. Flow HBS-N buffer as running buffer followed by mixture of

Immobilization on Gold EDC and NHS to activate carboxylic acid on the surface.
Surface 2. Inject immediately 5 mM KRAzR dissolved in coupling buffer,
followed by HBS-N buffer for washing.
3.2.2 Sample 1. To prepare RNA aptamer samples for Biacore analysis, PCR-
Preparation amplify template DNAs from plasmid DNAs.
2. Transcribe RNA aptamers, and purify them by 8 % denaturing
PAGE and 2-propanol precipitation.
3. Dissolve the obtained pellet with binding buffer used in the
selection procedure, and prepare 1.0 μM of each sample
(see Note 16).
3.2.3 SPR Analyses 1. Set running buffer (binding buffer) and RNA samples in the
Biacore 2000.
2. Flow running buffer for 90 s, and then inject aptamer sample
for 120 s for binding to the KRAzR surface (association).
3. Flow the running buffer again for 120 s for dissociation.
360 nm 430 nm
a b
5 min. 5 min.
Lamp 600 600 600
Response Unit
Response Unit
Response Unit
buffer gold surface 300 300 300
carrying Az-peptide
glass 0 0 0
0 100 200 300 400 0 100 200 300 400 0 100 200 300 400
Time (sec.) Time (sec.) Time (sec.)
Fig. 3 Photoresponsive interaction between RNA aptamers and KRAzR peptide. (a) Illustration of light irradia-
tion to the SPR sensor chip covered with droplet of running buffer. (b) SPR sensorgrams before and after light
irradiation. Two different RNA aptamers (red and blue line) were analyzed. As negative control, running buffer
was injected instead of RNA aptamers (green line)
4. Inject 10 mM NaOH for about 100 s to wash out bound

RNAs on the surface.
5. Flow the running buffer again, and proceed to next
measurement.
3.2.4 Photoisomerization 1. To photo-isomerize trans-KRAzR to cis form, set up a dark-

of KRAzR on Gold Surface room (see Note 17).
2. Cover the gold surface of sensor chip with a droplet of running
buffer (see Note 18), and irradiate 360 nm light for 5 min by
MAX-302 Xenon Light Source 300 W (Fig. 3a).
3. Blow the droplet away by blowing air with αCLEAN DUSTER
ECO, and set the chip back into Biacore 2000 as soon as
possible.
4. Analyze the same aptamer with the same procedure as described
above.
5. In the case of isomerization from cis-KRAzR back to trans-
isomer, irradiate 430 nm light for 5 min. Photoresponsive
binding between RNA aptamers and KRAzR peptide would be
observed (Fig. 3b).
4 Notes
1. Through all experimental procedures including RNA han-

dling, contamination of RNase is destructive. So, always take
care not to contaminate any dirt or dust. The use of gloves and
masks is favorable. Speaking during experiments might cause
contamination of RNase. Pipette tips accidentally contacted
with skins, clothes, or desks should be discarded immediately.
The experiments using RNase such as miniprep and genome
extraction should be done in another room or at a space far
from your RNA experimental space.
2. To take suspension of Affi-Gel 10® Gel by 1,000 mL pipette,

pipette tips should be cut by clean scissors to broaden the
entrance of the tips. Otherwise, you may have a trouble to take
the suspension.
3. It is not necessary to obscure room light, because trans-
KRAzR for the ligand of in vitro selection is stable under room
light condition.
4. To minimize the enrichment of RNA molecules that bind to
the gel itself such as spacer molecules tethered on the gel, neg-
ative selection should be done before KRAzR selection.
5. Before large-scale reaction, test with small scale (e.g., 100 μL)
and determine the optimal cycle number by PAGE analysis.
The excess number of cycles often causes decreased yields of
proper product.
6. Determine the concentration of obtained PCR product using
UV260 measurement. PAGE analysis is also useful to determine
the concentration by comparing the band intensity with that of
an authentic sample.
7. If you want to confirm the purity of the transcription product,
denaturing PAGE analysis is useful.
8. Throughout the procedure, to obtain RNA aptamers against
trans-KRAzR, you do not need to obscure room light because
trans-KRAzR is stable under room light with E/Z ratio of
79:21 [12]. However, if you select aptamers against photo-
unstable compound such as cis-KRAzR, the positive selection
procedure should be done in the dark.
9. The glycine-agarose and KRAzR-agarose are reusable after
selection. Wash each column with 5-gel volume of 7 M urea
for rinsing the remaining molecules, and then equilibrate the
gel with 5-gel volume of binding buffer. When you store the
gels, put the columns in 4 °C refrigerator.
10. Because most of the RNA sequences are not retained in the
column and the recovery rate of bound RNA is very low, use
of precipitating agent may increase the recovery rate. But it is
not indispensable. The precipitating agent is not harmful for
next procedures such as RT-PCR reaction.
11. When you heat your sample containing RNA, magnesium cat-
ion must be excluded. Magnesium cation catalyzes RNA-
cleaving reaction by coordinating oxygen atom of hydroxyl
group and phosphate group on RNA.
12. As seen in Fig. 2a, overcycling of PCR gives decreased proper
band and increased by-product, presumably because DNA
templates with different sequences at the randomized region
would interact with each other and cause undesirable elonga-
tion. Therefore, be careful in choosing the proper cycle number.
You may take samples in shorter period of PCR cycles to

increase the accuracy of your choice.
13. You do not have to choose the same selection conditions of
first round after second round. Increased selection stringency
such as reduced gel volume or increased temperature some-
times provides good RNA aptamers with strong affinity.
14. When you stop your experiment, please stop at RT-PCR, tran-
scription, or 2-propanol precipitation. Do not stop your exper-
iment during negative or positive selection because magnesium
cation present in binding buffer would cause RNA
degradation.
15. When an enrichment of selected RNA aptamers occurs, you
will see the reduced PCR cycle number in PAGE analysis of
RT-PCR products compared to that in the previous rounds.
This is because the amount of RNA bound to KRAzR-agarose
gradually increases as the enrichment proceeds.
16. Sample preparation should be done just before the measure-
ments. Do not leave the sample for a long time. Magnesium
cation included in binding buffer gradually degrades
aptamers.
17. Cis-isomer of KRAzR is unstable under room light and isom-
erizes back to trans-isomer for 5 min under the room light.
Therefore, when your ligand is photolabile, take care to avoid
unwanted light exposure to your ligand.
18. To keep wet condition of the surface, we covered the surface
with droplet of running buffer. But that might not be
necessary.
Acknowledgements
This work was supported by JSPS grant of Research Fellowship

for Young Scientists.
References
1. Mandal M, Breaker RR (2004) Gene regula- underlying riboswitch function. Annu Rev
tion by riboswitches. Nat Rev Mol Cell Biol Biophys 41:343–370
5:451–463 5. Soukup GA, Breaker RR (1999) Engineering
2. Serganov A, Patel DJ (2007) Ribozymes, ribo- precision RNA molecular switches. Proc Natl
switches and beyond: regulation of gene Acad Sci U S A 96:3584–3589
expression without proteins. Nat Rev Genet 6. Werstuck G, Green MR (1998) Controlling
8:776–790 gene expression in living cells through small
3. Winkler WC, Breaker RR (2005) Regulation molecule-RNA interactions. Science
of bacterial gene expression by riboswitches. 282:296–298
Annu Rev Microbiol 59:487–517 7. Bauer G, Suess B (2006) Engineered ribo-
4. Serganov A, Patel DJ (2012) Metabolite rec- switches as novel tools in molecular biology.
ognition principles and molecular mechanisms J Biotechnol 124:4–11
8. Sinha J, Reyes SJ, Gallivan JP (2010) 11. Wilson DS, Szostak JW (1999) In vitro selec-
Reprogramming bacteria to seek and destroy tion of functional nucleic acids. Annu Rev
an herbicide. Nat Chem Biol 6:464–470 Biochem 68:611–647
9. Suess B, Fink B, Berens C, Stentz R, Hillen W 12. Hayashi G, Hagihara M, Dohno C, Nakatani
(2004) A theophylline responsive riboswitch K (2007) Photoregulation of a peptide-RNA
based on helix slipping controls gene expres- interaction on a gold surface. J Am Chem Soc
sion in vivo. Nucleic Acids Res 32:1610–1614 129:8678–8679
10. Suess B, Hanson S, Berens C, Fink B, Schroeder R, 13. Hayashi G, Hagihara M, Nakatani K (2009)
Hillen W (2003) Conditional gene expression by RNA aptamers that reversibly bind photores-
controlling translation with tetracycline-binding ponsive azobenzene-containing peptides.
aptamers. Nucleic Acids Res 31:1853–1858 Chemistry 15:424–432
Chapter 4
Identification of RNA Aptamers Against Recombinant

Proteins with a Hexa-Histidine Tag
Shoji Ohuchi
Abstract
Artificial riboswitches that respond to the concentrations of intracellular proteins are a promising tool with
a variety of applications. They can be designed and engineered using existing RNA aptamers that target
proteins. Aptamers are generated via an iterative selection–amplification process, known as systematic
evolution of ligands by exponential enrichment (SELEX). This chapter describes a SELEX procedure for
the identification of RNA aptamers against hexa-histidine-tagged proteins. For the efficient enrichment of
higher affinity aptamers, the selection stringency should be gradually increased. Undesired RNA species
that bind to affinity resins can be eliminated from the pool by using a negative selection step and alternat-
ing different types of resins.
Key words Aptamer, SELEX, His-tag, In vitro selection, RNA, Artificial riboswitch, Synthetic gene
circuit
1 Introduction
Riboswitches are promising gene components for the construction

of artificial gene circuits and cell-based biosensors. Using existing
aptamers, new artificial riboswitches can be constructed (see later
related chapters in this book). Although most artificial riboswitches
have been generated to respond to metabolites and small cell-
permeable compounds, the generation of artificial riboswitches for
intracellular proteins is also possible. Such riboswitches can be used
to regulate gene expression according to the intracellular environ-
ment; thus, they are valuable for the design of artificial gene circuits
with cascade structures [1] (see later related chapters in this book).
In addition, artificial riboswitches for disease-related proteins may
be applied for novel types of therapy.
Aptamers are generated in vitro from large random sequence
libraries based on their high affinity for a target molecule by a process
known as systematic evolution of ligands by exponential enrichment
41
42 Shoji Ohuchi
Fig. 1 Schematic diagram of SELEX for the identification of RNA aptamers against a hexa-histidine-tagged
protein. Through negative selection, undesired RNA species that bind to the resin (indicated by dark gray) are
eliminated. The unbound RNA is recovered and used for positive selection. After extensive washing, RNA spe-
cies that bind tightly to the target (indicated by black) are isolated and amplified by RT-PCR. Following in vitro
transcription regenerates an enriched RNA pool for the next round of SELEX. These processes are repeated
until the desired aptamers are enriched in the pool
(SELEX) [2–7]. SELEX is an iterative selection–amplification process

(Fig. 1). Since the establishment of this procedure, a wide variety of
aptamers has been generated, and their applications have been exten-
sively studied. In this chapter, I describe a procedure for the identi-
fication of RNA aptamers that target the most popular recombinant
proteins, hexa-histidine (His)-tagged proteins.
2 Materials
Prepare all solutions under ribonuclease (RNase)-free conditions.
2.1 Random Pool 1. Random sequence template DNA (R3-N40): 5′-CAAGG

Constructions AGCGACCAGAGG-N40-TGGCATCCTT CAGCCC-3′ (N40 rep-
resents a random sequence of 40 nucleotides) (see Notes 1 and 2).
2. Forward primer (R3f): 5′-TAATACGACT CACTATAGGG
CTGAAGGATG CCA-3′ (the T7 promoter region is
underlined).
3. Reverse primer (R3r): 5′-CAAGGAGCGA CCAGAGG-3′.
SELEX Against His-Tagged Proteins 43
4. TE buffer: 10 mM Tris–HCl, pH 7.5, 1 mM EDTA.

5. Thermostable DNA polymerase for PCR [e.g., ExTaq (Takara
Bio, Japan)].
6. RiboMAX Large Scale RNA Production System-T7 (Promega).
7. RNase-free water (e.g., MilliQ water).
8. Mixed phenol:chloroform:isoamyl alcohol (PCI) at 25:24:1.
9. Thermocycler (PCR equipment).
10. 0.5× TBE buffer: 44.5 mM Tris–HCl, 44.5 mM borate, and
1 mM EDTA.
11. 8 % polyacrylamide gel in 0.5× TBE buffer.
12. Microcon YM-30 (Millipore).
13. MicroSpin G-25 columns (GE Healthcare).
14. UV spectrometer.
15. Random sequence template DNA for the alternative method
(R3-N40+T7p): 5′-CAAGGAGCGA CCAGAGG-N40-TGG
CATCCTT CAGCCCTATA GTGAGTCGTA TTA-3′ (the
T7 promoter region is underlined; N40 represents a random
sequence of 40 nucleotides).
2.2 SELEX 1. Buffer A: 20 mM Tris–HCl, pH 7.5, 100 mM potassium

acetate, 5 mM magnesium chloride, and 2 mM dithiothreitol
(DTT) (see Note 3).
2. Buffer B: 40 mM Tris–HCl, pH 7.5, 200 mM potassium
acetate, 10 mM magnesium chloride, 4 mM DTT, and 0.1 U/
μL RNase inhibitor (see Note 4).
3. 10 mg/mL tRNA (Sigma) in RNase-free water.
4. 1 % (w/v) BSA (Sigma) in RNase-free water.
5. Recombinant His-tagged protein.
6. HisLink Protein Purification Resin (Promega).
7. TALON Metal Affinity Resin (Clontech Laboratories).
8. Buffer C: Buffer A supplemented with 0.1 U/μL RNase inhib-
itor and 500 mM imidazole (see Note 4).
9. Co-precipitant for ethanol precipitation [e.g., Pellet Paint
Co-Precipitant (Novagen)].
10. Sample tube rotator.
11. Reverse transcriptase [e.g., ReverTra Ace (Toyobo, Japan);
SuperScript III (Invitrogen)].
2.3 Pull-Down Assay 1. Denaturing solution: 90 % formamide, 20 mM EDTA, 0.5 %

bromophenol blue (BPB), 0.1 % xylene cyanol (XC).
2. 12 % denaturing polyacrylamide gel in 0.5× TBE buffer con-
taining 7 M urea.
44 Shoji Ohuchi
3. SYBR Green II (Molecular Probes).

4. Phosphorimager [e.g., FLA 7000 (Fuji Film)].
2.4 Electrophoretic 1. 6× loading solution: 50 % glycerol, 1 % BPB, and 0.2 % XC.

Mobility Shift Assay 2. TBM buffer: 89 mM Tris, 89 mM borate, and 1 mM magne-
sium chloride (see Notes 5 and 6).
3. 10 % polyacrylamide gel in TBM buffer.
2.5 Filter- 1. Radioisotope (e.g., [32P])-labeled RNA (see Note 7).

Binding Assay 2. Blotting unit [e.g., Minifold Dot-blot System (Whatman)].
3. Nitrocellulose blotting membrane (pore size 0.45 μm) (Whatman).
4. A liquid scintillation counter.
3 Methods
3.1 Random DNA Two alternative methods are described blow (Fig. 2). The first
Pool Construction method (Subheading 3.1.1) is more common. The second method
(Subheading 3.1.2) requires longer sequences of synthetic DNA
but no enzymatic reaction.
3.1.1 DNA Pool 1. Extension reaction (Fig. 2a): The total reaction volume is
Preparation by Extension 500 μL: 1× PCR buffer (supplemented), 0.2 mM of each
Reaction dNTP (supplemented), 1 μM R3-N40 (template DNA), 5 μM
Fig. 2 Procedures for the preparation of the random DNA pool: (a) Extension reaction-based procedure.
(b) Annealing-based procedure. Boxes and “N” represent the primer regions and the random sequence,
respectively. Gray arrows indicate the orientation of the extension reaction
R3f (forward primer), and 50 U ExTaq. After denaturing for

1 min at 94 °C, anneal and extend the DNA at 50 °C for 15 s
and 72 °C for 3 min, respectively. Analyze the double-stranded
DNA (dsDNA) product (3 μL of the reaction) by 8 % PAGE
(see Note 8).
2. Purify the DNA by PCI extraction and ethanol precipitation.
Dissolve the DNA in 50 μL of TE buffer.
3.1.2 DNA Pool By using single-stranded DNA template that contains a full pro-
Preparation by Primer moter region, the DNA pool can be prepared without enzymatic
Annealing (Alternative reaction. The annealing of a forward primer to the template makes
Method) a transcriptionally competent, double-stranded promoter region
(Fig. 2b) [8].
The total volume is 40 μL: 0.1× PCR buffer, 10 μM
R3-N30+T7p, and 10 μM R3f. After denaturing for 1 min at
96 °C, anneal the DNA under the following conditions using a
thermocycler: 70 °C for 1 min; 60 °C for 1 min; 50 °C for 1 min;
and 40 °C for 5 min. All of the annealed DNA will be used for in
vitro transcription.
3.2 Preparation of 1. In vitro transcription: Incubate a 100 μL reaction containing

the Random RNA Pool 1× transcription buffer (supplemented), 40 μL of the DNA
pool prepared by extension reaction (Subheading 3.1.1) or
by annealing (Subheading 3.1.2), 5 mM of each rNTP
(supplemented), and 10 μL of T7 RNA polymerase enzyme
mix at 37 °C for 2–4 h.
2. Digest the template DNA by adding 5 U of RQ1 DNase (sup-
plemented) and incubating at 37 °C for an additional 15 min.
3. Extract the transcript with PCI and purify on Microcon YM-30
and MicroSpin G-25 columns.
4. Measure the RNA concentration based on the absorbance at
260 nm using a UV spectrometer.
3.3 SELEX 1. Wash 3 μL of empty resin (HisLink or TALON) with 100 μL

of buffer A once in a 0.2-mL PCR tube. Resuspend the resin
3.3.1 Negative Selection
in 100 μL of buffer A, and store the tube at room temperature
until use.
2. Dilute 2 nmol (1.2 × 1015 molecules) of the purified RNA pool
with water to 48 μL. To facilitate RNA folding, heat the RNA
solution at 90 °C for 3 min and chill quickly to 4 °C using a
thermocycler. Then, add 50 μL of buffer B and incubate at
37 °C for 10 min.
3. Add 1 μL of 10 mg/mL tRNA and 1 μL of 1 % BSA to the
RNA solution (see Note 9).
4. Remove the supernatant from the resin tube, and add the RNA
solution to the resin.
46 Shoji Ohuchi
5. Incubate the resin tube at room temperature with gentle rotation

for 30 min by using a sample tube rotator (see Note 10).
6. Spin down the resin, and collect the supernatant. The recovered
supernatant containing RNA not bound to the empty resin will
be used for positive selection.
3.3.2 Preparation of 1. Wash 3 μL of empty resin (HisLink or TALON) with 100 μL

Target-Immobilized Resin of buffer A once in a 0.2-mL PCR tube. Resuspend the resin
in 100 μL of buffer A and add 25 pmol of the target protein.
2. Incubate the tube at room temperature with gentle rotation
for 30 min using a rotator (see Note 10).
3. Spin down the resin, and remove the supernatant. To remove
un-immobilized proteins, wash the resin with 100 μL of buffer
A three times (see Note 11).
3.3.3 Positive Selection 1. Spin down the target-immobilized resin (Subheading 3.3.2),
and remove the supernatant. Add the RNA solution from the
negative selection (Subheading 3.3.1) to the resin. Incubate
the tube at room temperature with gentle rotation for 30 min
using a rotator (see Note 10).
2. Spin down the resin, and remove the supernatant containing
unbound RNA. To remove weakly bound RNA, wash the resin
with 100 μL of buffer A four times.
3. Spin down the resin, and remove the supernatant. To release
the target protein from the resin, resuspend the resin in 100 μL
of buffer C. Incubate the tube at room temperature with gen-
tle rotation for 10 min using a rotator (see Notes 10 and 12).
4. Collect the supernatant containing the released protein
and RNA bound to the protein. Recover the bound RNA
by PCI extraction followed by ethanol precipitation with a
co-precipitant (see Note 13).
3.3.4 Reverse 1. Reverse transcription: Dissolve the recovered RNA in 31 μL of

Transcription (RT)-PCR RNase-free water. Add 1 μL of 100 μM R3r primer, 10 μL of
5× reaction buffer (supplemented), 5 μL of 10 mM dNTP
mixture, 0.5 μL of 10 U/μL RNase inhibitor, and 2.5 μL of
100 U/μL ReverTra Ace. Incubate the mixture at 50 °C for
1 h and then at 99 °C for 5 min to heat inactivate the enzyme
(see Note 14).
2. PCR: The total reaction volume is 300 μL: 50 μL (all) of the
reverse transcription product, 1× PCR buffer (supplemented),
0.2 mM of each dNTP (supplemented), 3 μM R3f primer,
3 μM R3r primer, and 0.1 U/μL ExTaq. Run eight cycles of
PCR as follows: denaturing at 96 °C for 15 s; annealing at
50 °C for 15 s; and extension at 72 °C for 15 s (see Notes 15
and 16).
Fig. 3 Desired and undesired reactions during PCR. (a) Desired reactions during the extension step.
(b) Undesired side reactions. Because of the random sequence in the DNA pool, unintended mis-annealing may
occur. These side reactions generate longer and shorter by-products. Boxes and “N” represent the primer
regions and the random sequence, respectively. Gray arrows indicate the orientation of the extension reaction
3. Amplification cycle check: Equally divide 50 μL of the PCR

mixture into five new PCR tubes (10 μL each). Run 4, 8, 12,
16, or 20 additional cycles of PCR. Analyze the PCR product
(3 μL from the tube) by 8 % PAGE, and determine the number
of cycles necessary and sufficient for amplification (Figs. 3 and 4)
4. Run additional cycles of PCR for the remaining PCR mixture,
according to the cycle number determined above.
5. Purify the PCR product by PCI extraction and ethanol precipi-
tation. Dissolve the purified DNAs in 20 μL of TE buffer
(see Note 19).
3.3.5 Subsequent Repeat the process described in Subheadings 3.2–3.3.4, typically

Rounds of SELEX for ten rounds. Reaction volumes for in vitro transcription
(Subheading 3.2) and RT-PCR (Subheading 3.3.4) can be reduced
by half or more. To enrich tightly binding RNA, gradually increase
the selection stringency. To achieve this, reducing the target con-
centration and increasing the number of washes during positive
selection are helpful. To avoid enriching RNA that binds tightly to
the resin, alternate the type of resins used. Examples of the typical
conditions for each round of SELEX are listed in Table 1.
3.4 Analysis of The number of SELEX rounds necessary depends on the properties
SELEX Progress of the target protein and the selection conditions employed. Thus,
it is impossible to know how many rounds to perform a priori, and
the progress of SELEX should be experimentally confirmed by
analyzing the affinity enrichment of the RNA pool (see Note 20).
48 Shoji Ohuchi
Fig. 4 Optimization of PCR cycle number. Too many PCR cycles generate unde-
sired longer and shorter by-products, which can be recognized by a smear band
pattern in PAGE analysis. Therefore, the number of cycles that is necessary and
sufficient should be determined. In the case shown, 16–18 cycles are optimal
Table 1
Conditions for positive selection against a His-tagged target protein
Round Target protein (nM) Resin Washing

1 250 HisLink 4
2 250 TALON 5
3 125 HisLink 5
4 125 TALON 6
5 62.5 HisLink 6
6 62.5 TALON 7
7 31.3 HisLink 7
8 31.3 TALON 8
9 15.6 HisLink 8
10 15.6 TALON 10
Fig. 5 Examples of pull-down assay results. (a) Case demonstrating successful

SELEX. If SELEX progresses successfully, the RNA pool (in this case, the Round 10
pool, designated “R10”) binds to the protein-immobilized resin (indicated as “P”)
but not to the empty resin (indicated as “E”). (b) Case demonstrating unsuccessful
SELEX. If undesired RNA species that bind to resin are enriched, the RNA pool
(“R10”) binds to the protein-immobilized resin and to the empty resin. “40N” and
“I” indicate the control RNA and 1/10 volume of an input sample, respectively
3.4.1 Pull-Down Assay 1. Prepare 6 μL of the resin with 200 pmol of the target protein
immobilized as described above (Subheading 3.3.2). Divide it
equally into two 0.2-mL PCR tubes. As a resin control, wash
6 μL of empty resin and divide it equally into two 0.2-mL PCR
tubes (see Notes 21 and 22).
2. Dilute 200 pmol of the RNA pool to be tested with water to
96 μL, fold the RNA, and add tRNA and BSA as described
above (Subheading 3.3.1). The total volume of the RNA sam-
ple is 200 μL. As an RNA control, use 200 pmol of RNA from
the initial random pool.
3. Remove the supernatant from the resin tubes, and add 90 μL
of each RNA sample onto each resin.
4. Incubate the resin tubes at room temperature with gentle rota-
tion for 30 min using a rotator (see Note 10).
5. Spin down the resin and wash with 100 μL of buffer A twice.
After removing the supernatant, resuspend the resin in 10 μL
of the denaturing solution. As an input sample control, transfer
9 μL of the remaining folded RNA to a 0.2-mL PCR tube and
add 9 μL of the denaturing solution.
6. Heat the samples at 90 °C for 3 min and chill quickly to 4 °C
using a thermocycler. Separate the samples on 12 % denaturing
PAGE containing 7 M urea. Soak the gel with SYBR Green II
to stain the RNA bands, and analyze the band image using a
phosphorimager. If SELEX works successfully, and the desired
aptamers are enriched in the pool, the RNA pool binds to the
protein-immobilized resin but not to the empty resin (Fig. 5).
50 Shoji Ohuchi
Fig. 6 Example of EMSA results. When SELEX progresses successfully, the RNA
pool (in this case, the Round 10 pool, designated “R10”) forms complexes with
the target in a protein concentration-dependent manner. Not all RNA molecules
form complexes even when the protein concentration is high due to mis-folding
of the aptamers and the presence of inactive RNA species in the pool. “40N”
indicates the control RNA
3.4.2 Electrophoretic 1. Dilute 5 pmol of the RNA pool to be tested with water to
Mobility Shift Assay 12 μL, fold the RNA, and add tRNA and BSA as described
(Alternative Method) above (Subheading 3.3.1). The total volume of the RNA sam-
ple is 25 μL. As an RNA control, use 5 pmol of RNA from the
initial random pool (see Note 21).
2. Dilute 2.5, 10, or 40 pmol of the target protein with buffer A
to 12.5 μL.
3. Mix 5 μL of each folded RNA and 5 μL of the diluted protein.
As a no-protein control, mix 5 μL of each folded RNA and
5 μL of buffer A. The final concentrations of RNA and protein
are 100 nM and 0 nM, 100 nM, 400 nM, or 1,600 nM,
respectively.
4. Incubate at 37 °C for 30 min.
5. Chill on ice, and add 2 μL of the 6× loading solution.
6. Run the TBM gel at 75 V for 2 h in a cold room (see Note 23).
7. Soak the gel with SYBR Green II to stain the RNA bands, and
analyze the band image using a phosphorimager. If SELEX
works successfully, and the desired aptamers are enriched in the
pool, the RNA pool forms a complex with the protein, and
mobility-shifted bands can be detected (Fig. 6).
3.4.3 Filter-Binding 1. Dilute 0.1 pmol of the radioisotope-labeled RNA pool with
Assay (Alternative Method) water to 120 μL, fold the RNA, and add tRNA and BSA as
described above (Subheading 3.3.1). The total volume of the
RNA sample is 250 μL. As an RNA control, use 0.1 pmol of

RNA from the initial random pool (see Note 24).
2. Dilute 2.5, 25, or 250 pmol of the target protein with buffer A
to 125 μL.
3. Mix 50 μL of each folded RNA and 50 μL of the diluted
protein. As a no-protein control, mix 50 μL of each folded
RNA and 50 μL of buffer A. The final concentrations of RNA
and protein are 1 nM and 0 nM, 10 nM, 100 nM, or 1,000 nM,
respectively.
5. Chill on ice.
6. Soak the nitrocellulose membrane with buffer A for a few minutes.
Place the soaked membrane onto the blotting unit, and assemble
the apparatus.
7. Connect to a vacuum pump, and apply the vacuum.
8. Apply each sample to the membrane. Quickly apply 200 μL of
buffer A twice to wash the membrane (see Note 25).
9. Disassemble the blotting unit and turn off the pump. Place the
membrane onto a paper towel to dry (see Note 26).
10. Cut off each blotted position, and evaluate the amount of
trapped RNA using a liquid scintillation counter. If SELEX
works successfully, and the desired aptamers are enriched in the
pool, the amount of RNA trapped increases in a protein
concentration-dependent manner (see Note 27).
3.5 Cloning Once affinity enrichment of the RNA pool is confirmed, clone the
and Evaluation variants by TA cloning. Several TA cloning vectors are commercially
of Aptamers available [i.e., pGEM-T Easy Vector System (Promega); TA Cloning
Kit with pCR (Invitrogen)]. Sequencing more than 30 clones is
3.5.1 Cloning
recommended. Typically, similar or even identical sequences appear
and Sequencing
multiple times. This sequence convergence indicates successful
enrichment of certain clones via SELEX (see Note 28).
3.5.2 Affinity Estimation To evaluate the affinity of each clone, perform a binding assay as
of Each Clone described for the analysis of SELEX progress (Subheading 3.4). At
this stage, determining the precise value of the affinity constant is
desirable. Thus, an EMSA (Subheading 3.4.2) or a filter-binding
assay (Subheading 3.4.3) with radioisotope-labeled RNA is recom-
mended. Alternatively, other interaction analysis techniques, like
surface plasmon resonance analysis or fluorescence correlation
spectroscopy, are useful [9–12].
3.5.3 Evaluation of Typically, the full-length sequence of the original isolate is not
Essential Elements required for the binding activity. In addition, a shorter sequence
for Binding Activity and simpler structure are beneficial for the design and engineer-
ing of artificial riboswitches employing the obtained aptamers.
52 Shoji Ohuchi
Fig. 7 Generation of a shortened aptamer. Initially, truncations from the 5′- and 3′-ends of the original isolate,
an aptamer against T7 RNA polymerase (T06), were examined. A shortened derivative (T06-38) was generated
without reducing the affinity. Based on the shortened derivative, secondary SELEX was performed, and further
truncation was examined. The resulting 29-nt aptamer (T230-29) is 42 nt shorter than the original aptamer;
its affinity for the target protein is higher. In addition, the secondary SELEX revealed the sequence and struc-
tural features essential for activity (for details, see ref. 13)
To shorten the aptamers, serial truncation of the 5′- and 3′-ends

of the original isolate is the best choice for an initial effort (Fig. 7)
[13]. Alternatively, two-dimensional (2D) structure predictions
facilitate the rational design of shortened aptamers (see Note 29).
Re-selection from a randomly mutagenized pool, in which
~10–25 % of the nucleotides of the original isolate are randomly
replaced with other nucleotides, is also helpful [13, 14]. Based on
the sequence comparison of the isolates from this secondary
SELEX, essential sequence motifs and/or structural features can
be identified. Furthermore, evolved aptamers with higher affinity
may be also isolated (Fig. 7).
4 Notes
1. If the several lines of SELEX are performed (either simultane-

ously or sequentially), use primer sets with different sequences
to prevent cross- and carryover contamination.
2. The length of the random region is typically 30–100 nucleo-
tides. Theoretically, libraries with a longer random sequence are
more likely to identify aptamers [15]. However, longer random
sequences are also more likely to generate by-products
(see Subheading 3.3.4), and they show lower RT-PCR efficiency.
3. Some proteins show a strong tendency to adsorb RNA non-
specifically. For such proteins, this buffer system may not be
suitable for SELEX because the adsorption of RNA severely
reduces the enrichment factor in positive selection. To reduce
the adsorption, the addition of a nonionic, hydrophilic (high
hydrophile–lipophile balance value) surfactant (e.g., 0.05 %
Tween 20) may be effective. Such surfactants do not disrupt
protein structure. In addition to a surfactant, higher ionic

strength (e.g., <500 mM sodium chloride or potassium chloride)
may also be helpful. High ionic strength neutralizes and weak-
ens electrostatic interactions.
4. Buffer B and buffer C should conform to 2× and 1× buffer A,
respectively. When buffer A is modified, these buffers should
be also modified.
5. TBE buffer (89 mM Tris, 89 mM borate, and 2 mM EDTA)
supplemented with 3 mM magnesium chloride can substitute
for TBM buffer.
6. TBE buffer can be used for electrophoretic mobility shift assay
(EMSA) to analyze the varieties of aptamers. However, in
some cases, inclusion of magnesium ion in the electrophoresis
buffer is essential [16]. Because it is unknown whether aptamers
enriched in the pool require magnesium ion in the buffer, the
use of TBM buffer, rather than TBE buffer, is recommended.
7. The radioisotope labeling can be performed by 5′- or 3′-end
labeling and internal labeling. At least 1 × 105 cpm/pmol of the
relative radioactivity is desired.
8. If the reaction efficiency is not sufficiently high (i.e., less than
50 % of the input DNA is converted to dsDNA), repeat the
annealing and extension step several times. The denaturing
step, however, should not be repeated. In contrast to a typical
PCR, the reaction includes only the forward primer; denatur-
ing the dsDNA products will not reproduce dsDNA.
Furthermore, denaturing can enhance the likelihood of side
reactions (see below). When repeated extension does not
improve the yield of dsDNA products, low-quality template
DNA may be at fault. In this case, increase the reaction volume
to obtain sufficient amounts of dsDNA.
9. If the target protein or resin strongly adsorbs RNA nonspecifi-
cally, the tRNA concentration can be increased up to tenfold.
10. If a rotator is not available, invert the tube every 3 min.
11. When using commercially available His-tagged recombinant
proteins, typically, more than 70 % of the input protein is
immobilized on the resin through this procedure. However,
the immobilization efficiency of protein prepared by an indi-
vidual researcher may be very low. Thus, a pilot experiment to
assess protein immobilization is recommended. If immobiliza-
tion is inefficient, modification of the buffers, such as the
addition of a low concentration of denaturant (e.g., 0.5–2 M
urea, guanidine-hydrochloride), may be effective. Alternatively,
completely denatured proteins may be efficiently immobilized
and refolded on the resin [17].
12. When using commercially available His-tagged recombinant
proteins, typically, almost all of the immobilized protein is
54 Shoji Ohuchi
released from the resin under these conditions. However, the

release efficiency of protein prepared by an individual researcher
may be very low. Again a pilot experiment is recommended. If
the protein cannot be released even with higher concentrations
of imidazole, the bound RNA can be directly recovered by PCI
extraction or by heat treatment (i.e., 90 °C for 5 min) of the
resin. These recovery procedures are not selective for RNA
that binds to protein; undesired RNA species that bind to the
resin are also recovered.
13. Typically, the number of RNA molecules at this step is very
low. For the efficient recovery of the desired RNA species, the
use of a co-precipitant is highly recommended.
14. Reverse transcriptases have a tendency to skip strong stem
structures in the template RNA [18]. Such skipping produces
shortened cDNA. To avoid skipping, the use of thermostable
enzymes, like ThermoScript Reverse Transcriptase (Invitrogen),
may be helpful.
15. Typically, eight cycles of PCR is not sufficient amplification
during the early rounds of SELEX. However, according to
the progress of SELEX, the PCR cycle number required may
be lower. If the eight-cycle PCR generates visible amounts of
DNA on PAGE analysis, reduce the cycle number.
16. Typically, efficient PCR amplification of very short DNA frag-
ments, like the DNA pool for SELEX, requires a relatively high
concentration of primers. To confirm the appropriate PCR
conditions for the DNA pool, a pilot experiment using DNA
from an initial random pool is recommended.
17. Because of the random sequence in the template DNA, unin-
tended mis-annealing may occur (Fig. 3). These side reactions
generate longer and shorter by-products. Typically, the early
PCR cycles generate only the desired products. However, the
increased concentration of PCR products, that is DNA con-
taining random sequences, increases the likelihood of side
reactions. Thus, avoid too many cycles of PCR.
18. A high-concentration agarose gel can be used to confirm the
amplification of some PCR products. However, slight size differ-
ences between the desired product and undesired by-products
cannot be distinguished. Thus, the use of polyacrylamide gels,
instead of agarose gels, is recommended.
19. Do not use all the PCR product in the next step (in vitro
transcription). A portion of the DNA pool should be stored.
Useful aptamers in the pools from earlier rounds may disappear
in the pools from later ones.
20. In the author’s experience, the progress of SELEX can be spec-
ulated from PAGE analysis of RT-PCR; a change in the smear
band pattern; and a reduction in the number of PCR cycles
required for sufficient amplification. Although this is not defin-

itive, experienced selectionists may skip the pool analysis step
and proceed to the analysis of each clone in the pool.
21. Smaller amounts of protein and RNA are sufficient for the
analysis when sensitive detection of RNA is possible. For this
purpose, radioisotope labeling (i.e., with [32P]) is least expen-
sive and most sensitive.
22. Alternatively, the detection of RT-PCR products, rather than
the RNA itself, is also a sensitive procedure. However, the
DNA pool usually has a higher RT-PCR efficiency, and the
RT-PCR step may overestimate affinity enrichment.
23. The extent of the mobility shift upon the complex formation
depends on the charge and molecular weight of the target
protein. Thus, the running time of electrophoresis required for
clear discrimination of the complex from free RNA varies from
protein to protein.
24. The filter-binding assay tends to show relatively high data
dispersions; thus, triplicate experiments are recommended.
25. Keep the membrane wet before washing; drying at this step
causes non-ignorable background signal.
26. Do not place the membrane upside down, and do not cover the
membrane with the paper towel. Direct contact of the sample
surface with the paper towel may remove the trapped RNA.
27. The evaluation can be also performed by exposing the mem-
brane to an imaging plate and analyzing the plate using a
phosphorimager.
28. Sequence convergence does not always mean successful enrich-
ment of aptamers. Not only affinity for the target but also
efficiency of RT-PCR and transcription can cause enrichment
bias during SELEX. In some conditions, parasitic RNA species
lacking affinity are efficiently enriched [19]. Thus, the affinity
of each clone should be experimentally validated.
29. Several 2D structure prediction programs are freely available [20].
References
1. Ausländer S, Fussenegger M (2013) From 4. Ellington A, Szostak JW (1990) In vitro

gene switches to mammalian designer cells: selection of RNA molecules that bind specific
present and future prospects. Trends Biotechnol ligands. Nature 346:818–822
31:155–168 5. Tuerk C, Gold L (1990) Systematic evolution
2. Ellington AD, Chen X, Robertson M, Syrett A of ligands by exponential enrichment: RNA
(2009) Evolutionary origins and directed ligands to bacteriophage T4 DNA polymerase.
evolution of RNA. Int J Biochem Cell Biol Science 249:505–510
41:254–265 6. Klussmann S (ed) (2006) The aptamer hand-
3. Joyce GF (2007) Forty years of in vitro evolu- book: functional oligonucleotides and their
tion. Angew Chem Int Ed Engl 46:6420–6436 applications. Wiley-VCH, Weinheim
56 Shoji Ohuchi
7. Stoltenburg R, Reinemann C, Strehlitz B against T7 RNA polymerase. FEBS Open Bio

(2007) SELEX—a (r)evolutionary method to 2:203–207
generate high-affinity nucleic acid ligands. 14. Ekland EH, Bartel DP (1995) The secondary
Biomol Eng 24:381–403 structure and sequence optimization of an
8. Milligan JF, Groebe DR, Witherell GW, RNA ligase ribozyme. Nucleic Acids Res
Uhlenbeck OC (1987) Oligoribonucleotide 23:3231–3238
synthesis using T7 RNA polymerase and 15. Sabeti PC, Unrau PJ, Bartel DP (1997)
synthetic DNA templates. Nucleic Acids Res Accessing rare activities from random RNA
15:8783–8798 sequences: the importance of the length of
9. Di Primo C, Dausse E, Toulmé JJ (2011) molecules in the starting pool. Chem Biol
Surface Plasmon resonance investigation of 4:767–774
RNA aptamer-RNA ligand interactions. 16. Hellman LM, Fried MG (2007) Electrophoretic
Methods Mol Biol 764:279–300 mobility shift assay (EMSA) for detecting
10. Ishino T, Atarashi K, Uchiyama S, Yamami T, protein-nucleic acid interactions. Nat Protoc
Saihara Y, Yoshida T, Hara H, Yokose K, 2:1849–1861
Kobayashi Y, Nakamura Y (2000) Interaction 17. Li M, Su ZG, Janson JC (2004) In vitro
of ribosome recycling factor and elongation protein refolding by chromatographic proce-
factor EF-G with E. coli ribosomes studied by dures. Protein Expr Purif 33:1–10
the surface plasmon resonance technique. 18. Zhang YJ, Pan HY, Gao SJ (2001) Reverse
Genes Cells 5:953–963 transcription slippage over the mRNA second-
11. Werner A, Hahn U (2009) Fluorescence cor- ary structure of the LIP1 gene. Biotechniques
relation spectroscopy (FCS)-based characteri- 31:1286–1294
sation of aptamer ligand interaction. Methods 19. Breaker RR, Joyce GF (1993) Minimonsters:
Mol Biol 535:107–114 evolutionary byproducts of in vitro RNA amplifi-
12. Schürer H, Buchynskyy A, Korn K, Famulok cation. In: Fleischaker GR, Colonna S, Luisi PL
M, Welzei P, Hahn U (2001) Fluorescence (eds) Proceedings of the NATO conference on
correlation spectroscopy as a new method for self-reproduction of supramolecular structures.
the investigation of aptamer/target interac- Kluwer Academic Press, Dordrecht, pp 127–135
tions. Biol Chem 382:479–481 20. Seetin MG, Mathews DH (2012) RNA struc-
13. Ohuchi S, Mori Y, Nakamura Y (2012) ture prediction: an overview of methods.
Evolution of an inhibitory RNA aptamer Methods Mol Biol 905:99–122
Chapter 5
Artificial Riboswitch Selection: A FACS-Based Approach

Zohaib Ghazi, Casey C. Fowler, and Yingfu Li
Abstract
Riboswitches have a number of characteristics that make them ideal regulatory elements for a wide range of
synthetic biology applications. To maximize their utility, methods are required to create custom riboswitches
de novo or to modify existing riboswitches to suit specific experimental needs. This chapter describes such a
method, which exploits fluorescence-activated cell sorting (FACS) to quickly and efficiently sort through
large libraries of riboswitch-like sequences to identify those with the desired activity. Suggestions for the
experimental setup are provided, along with detailed protocols for testing and optimizing FACS conditions
FACS selection steps, and follow-up assays to identify and characterize individual riboswitches.
Key words Artificial riboswitch, Aptamer, Biosensor, Fluorescence-activated cell sorting (FACS),
High-throughput, Enrichment
1 Introduction
Riboswitches are RNA elements that facilitate genetic control

through structural changes that occur in response to directly
binding a specific small-molecule ligand [1]. The modular nature
of riboswitches, coupled with the simple and direct nature of the
regulation that they provide, has sparked the interest of synthetic
biologists who have applied riboswitches toward a range of inter-
esting applications [2–6]. One of the principal advantages of ribo-
switches as regulators in synthetic systems is the potential for novel
elements to be created that can exert a desired regulatory effect in
response to a hand-selected target molecule. The most critical
factor for successfully isolating an effective synthetic riboswitch is
that the aptamer domain must possess desirable binding character-
istics that suit the intended applications. Creating such an aptamer
de novo can be a project in itself, and so riboswitch engineers have
generally made use of previously identified aptamers (either created
artificially using “in vitro selection” protocols or borrowed from
57
58 Zohaib Ghazi et al.
naturally evolved riboswitches) as the foundation for their designs.

Many excellent reviews have been published that describe the
methodology that can be used to create a novel aptamer with
desired properties [7, 8]. Once an appropriate aptamer has been
identified, several creative approaches have been successful in apply-
ing it toward the production of a functional riboswitch [9–13]. In this
chapter, we describe such a method in which fluorescence-activated
cell sorting (FACS) is employed to identify active riboswitches from a
carefully designed, partially random library of riboswitch candidates
[9, 10]. The methodology described here should be generally appli-
cable to bacterial organisms that can take up and maintain plasmid
DNA. For some organisms, steps involving the growth and manipula-
tion of the bacteria will need to be tweaked, and for biologically
hazardous bacteria, additional safety measures will be required that
are not described here.
The distinctive feature of the approach described in this chapter
is the use of FACS to identify sequences that produce a regulatory
response to the desired target molecule. Flow cytometers are
employed for a vast array of purposes in biological research that
exploit their analytical capabilities, sorting capabilities, or both.
For an extensive review of flow cytometry, consult the following:
[14, 15]. With respect to sorting, flow cytometers can be used to
sort individual particles in a given size range (typically ~0.2–
100 μm, depending upon the instrument being used) based on a
number of physical properties. As it is applied here, flow cytometry
is used to sort individual bacterial cells on the basis of fluorescence
intensity. A library of sequences, carefully designed to have the
potential to act as riboswitches, is cloned upstream of a fluorescent
protein on a plasmid that is transformed into the host bacteria.
When its expression is activated, this fluorescent protein represents
the dominant source of fluorescence, and thus the bacteria can be
sorted based on their expression of this gene. To identify bacteria
in which the fluorescent protein is regulated by the target mole-
cule, a procedure is used in which bacteria are sorted under two
different conditions: one where the riboswitch target is present at
high levels and the other where it is absent or present at low levels
(see Fig. 1).
The focus of this chapter is to provide detailed protocols for
conducting FACS-based riboswitch selections. This method is only
as good as the starting library, and the success of this protocol
hinges completely on the presumption that it contains active
riboswitches. While a limited description of the basics of creating a
library is provided in Subheading 3.1, there are endless options for
the design of the library, and a detailed description of these possi-
bilities falls outside the scope of this chapter.
Selecting Artificial Riboswitches by FACS 59
Fig. 1 Isolation of functional riboswitches in E. coli using FACS. Following the transformation of the plasmid
library (small circle), which contains riboswitch candidates placed within the 5′ UTR of an FP gene. E. coli cells
are sorted based on fluorescence intensity to isolate cells that are highly fluorescent in the presence of the
target molecule that regulates the riboswitch, but less so in its absence
2 Materials
2.1 Growth Media 1. Luria Broth (LB).

2. LB + 15 % glycerol.
3. LB + 30 % glycerol.
4. Growth medium to create low target metabolite levels
(see below).
5. Growth medium to create high target metabolite levels
(see below).
The selection of the growth media that will be used for the
culture steps preceding FACS and fluorescence assays is not trivial.
The protocol requires that two culture conditions be created: one
where the riboswitch ligand is absent or present at low levels and
another where this molecule is highly abundant. Some strategies
for creating these conditions are described in Subheading 3.1.1.
In some instances, such as when the target molecule is synthetic,
convenient, chemically undefined media such as LB might be
appropriate. Typically, we suggest using chemically defined media
to more carefully control the levels of the target molecule. Nutrients
such as amino acids and vitamins can be added to simple media for-
mulations to permit more robust growth and gene expression [2].
2.2 Fluorescence- 1. Agar.

Activated Cell Sorting 2. Antibiotics.
3. Sterile polystyrene petri dishes, 100 × 15 mm (Thermo Fisher
Scientific). Prepare LB agar plates supplemented with the
appropriate antibiotic.
4. 10× Dulbecco’s phosphate-buffered saline (DPBS), supple-
mented with calcium and magnesium (Invitrogen). Make up a 1×
solution, filter, and autoclave.
5. 1.5 mL Microfuge tubes.

6. 14 mL Polypropylene round-bottom culturing tubes, 17 × 100 mm
(Becton Dickinson).
7. 5 mL Polystyrene round-bottom tube, 12 × 75 mm (Becton
Dickinson).
8. 5 mL Polystyrene round-bottom tube, 12 × 75 mm with cell
strainer cap (Becton Dickinson).
2.3 96-Well Plate 1. 96-well, flat bottom polystyrene non-treated sterile assay plates
Fluorescence Assay (Corning).
2. 96-well, half-area flat bottom polystyrene NBS plates
(Corning).
3. ABgene 96-well, square-well storage plates (Thermo Fisher
Scientific).
3 Methods
3.1 Considerations Pioneering projects to create synthetic riboswitches have often

Before Beginning the selected a target molecule based primarily on the availability of a
Project high-quality, well-characterized RNA aptamer. While this is a rea-
sonable approach for method development, future endeavors
3.1.1 The Target
should generally be conducted with specific downstream applica-
Molecule
tions in mind. There are, however, certain factors that can make
the development of a synthetic riboswitch for a given ligand par-
ticularly challenging or even impossible using the methodology
described here (and, in general, other methods as well). One con-
sideration is that RNA has limited chemical diversity, and, although
it has shown the potential to tightly and specifically bind a very
wide range of molecules, certain types of ligands will intrinsically
present a greater challenge for aptamer development. For example,
molecules with significant negative charge and highly hydrophobic
molecules are likely to be difficult ligands for RNA.
A second consideration is that this protocol depends on the
ability to predictably manipulate the cellular concentration of
the target molecule. To achieve this for molecules that are not
native to the cell, the primary concern is that the target ligand
must be able to permeate host cells and maintain suitable levels.
In certain instances, strains featuring mutations to broad-specificity
efflux pumps might assist in obtaining elevated concentrations of
molecules that are excreted. For cellular metabolites, a generic
approach that is effective for many molecules is to employ mutant
strains in which the synthesis of the target molecule has been
disrupted. The molecule (or a precursor) is then added to the
growth medium at controlled levels to produce the “high” and
“low” concentrations required for the selection steps. For certain
molecules it might prove difficult or even impossible to manipulate
their cellular levels in a controlled fashion over the desired range,

which would preclude the use of the strategy described in this
chapter.
A final consideration is that all riboswitches are only responsive
over a limited concentration range. The binding characteristics of
the aptamer will be the primary factor in dictating this range, and
it is important for the target molecule concentration range that is
relevant for your purposes to be compatible with the aptamer that
is selected. Importantly, in vitro binding constants will not always
hold true within the context of a cell, and the kinetics of RNA fold-
ing, transcription, and translation have been shown to impact the
responsive range of naturally occurring riboswitches [16–18];
these factors should be considered but can be difficult to predict.
The unpredictability of these issues highlights the advantage of
exploiting the power of selection to identify the best solution to
this complicated problem.
3.1.2 Library Design Once an aptamer has been chosen, the next step before beginning
the riboswitch selection is to design a plasmid-borne library of can-
didate riboswitches. The only region of the plasmid that will vary
between individual members of the library is the “riboswitch
region,” which resides in the 5′ untranslated region (UTR) of a
fluorescent protein. To drive the expression of the transcript that
contains the riboswitch region and the fluorescent protein, it is
advisable to use a well-characterized “constitutive” promoter,
which is as stable as possible under the selection conditions that
will be used. Two important considerations for the library plasmid
are its origin of replication (copy number) and the strength of the
promoter. For each of these two factors, the required signal
strength for the FACS steps must be balanced with the potential
complications and intrusiveness associated with protein overex-
pression. In addition, for riboswitches with high-affinity aptamers
that detect low levels of the target molecule, it is possible that,
within the aptamer’s dynamic range, the target molecule may not
be in excess over the transcript when using high copy plasmids and
strong promoters.
The methodology described in this chapter is intended to be
generally applicable for a range of riboswitch designs and mecha-
nisms. The focus is therefore on setting up and performing the
selection that allows for the identification of active riboswitches
from a starting library and on follow-up experiments to identify
and characterize individual sequences. There are limitless options
for the design of the starting library, which could target ribo-
switches that act at the level of transcription [19, 20], translation
[10–13, 21], or RNA processing [22–24]. Specific library designs
are not discussed here. Generally, the library will consist of a
combination of invariant and random residues in the 5′ UTR
shortly upstream of a fluorescent protein (FP). The aptamer domain,
which is typically held largely constant to maintain the binding
Fig. 2 Two representative library designs, one with a transcription terminator

(a) and the other with an RBS sequestering helix (b). Random or partially random
residues (grey segment) are introduced at key positions that will allow for the
binding event to be transduced into a change in the expression of downstream
genes. More specific examples of library designs can be found in the literature
(see main text) (color figure online)
capabilities of the library, will be arranged in close proximity to a

regulatory element that is capable of hindering the expression of
the downstream fluorescent protein via its ribosomal binding site,
an intrinsic transcriptional terminator, or a self-cleaving ribozyme.
Random residues are introduced at key positions that are likely to
transduce the structural changes that occur in response to aptamer–
ligand binding to the disruption or the activation of the regulatory
element. While many of the sequences in this library might be
inert, a low number of active sequences can be identified using the
selection and screening steps outlined in this chapter. A schematic
demonstrating two generic library designs is shown in Fig. 2. For
specific examples of designs that have been used previously, see refs.
10–13, 19–25.
3.1.3 The Fluorescent Decades of intense research has yielded an abundance of FPs with
Protein diverse properties. Different FPs can vary greatly in their excitation/
emission spectra, extinction coefficient and quantum yield, matu-
ration speed and efficiency, turnover rate, environmental sensitivity,
photostability, oligomerization state, etc. A number of excellent
reviews have been published summarizing these factors and how

they apply to individual FPs that are available; this material should
be carefully reviewed before making a final choice [27–30]. Many
FP reviews focus on microscopy applications; this information is
useful and relevant, but be mindful of the particular requirements of
your experiments. For example, factors such as photostability and
oligomeric state are critical for many imaging experiments but are
not significant factors for FACS.
The optics of the flow cytometer that will be employed for this
project is an essential consideration when selecting an FP; the opti-
mal excitation/emission spectrum of the FP must match up with
the lasers and filter sets of this instrument (see below). The signal-
to-background ratio is another important consideration. The signal
generated by the FP will depend on a number of factors including
the efficiency of its expression and maturation within your strain
and the extinction coefficient and quantum yield of the FP. The
background will be dictated by the level of autofluorescence of
your strain, which will generally be greater in the blue region of the
spectrum and less pronounced in the red range. The amount of
time required for the FP to fold and mature and the length it will
persist in the population (turnover time) are issues that are particu-
larly relevant for this project. Slow-maturing FPs create a signifi-
cant delay between the time that regulatory decisions are made by
potential riboswitches and the appearance of the corresponding
signal. FPs are generally very stable and can persist for hours or
even days within cells. Signal contamination by FPs synthesized at
previous steps of the protocol has the potential to significantly
interfere with the selection process described in this chapter.
Dilution can be effectively used to combat this problem; however,
FPs with a shorter half-life are generally preferable. The caveats of
this are that FPs must persist for at least a few hours to accommo-
date the lag between growth and the time samples are processed by
FACS and that FPs with a shorter half-life will yield a lower signal
due to their decreased abundance.
3.1.4 Instrument High-speed cell sorters are available from several manufacturers,
Considerations such as the FACSAria™ (BD Biosciences), EPICS® ALTRA™
(Beckman Coulter), or the MoFlo® (Dako) cell sorters. Flow
cytometers, particularly those with sorting capabilities, are tremen-
dously valuable for a wide range of applications and are very com-
plex and expensive instruments. Accordingly, many research
institutes have a limited number of flow cytometers that are shared
in common facilities, often operated exclusively by specially trained
personnel. For those that do not have ready access to a capable
instrument that allows the sorting of bacteria, many centralized
facilities permit access to users from other institutes at hourly rates.
While most modern instruments are compatible with the experi-
ments described here, there are several issues to consider before
beginning your experiment. We strongly suggest speaking with an

experienced director or operator from the facility you intend to use
about the specific details of your experiment before getting started.
With regard to the instrument that will be used, the main factors
that must be considered when planning your experiment are the dif-
ficulties associated with sorting particles as small as bacterial cells, the
optics of the instrument, the sorting rates, and any pertinent safety/
containment issues (particularly for those working with BSL-2 or
BSL-3 organisms). Most commercial instruments can be optimized
for small-particle analysis, such as bacteria; however, they approach
the limit of detection, as historically they have been used to detect
much larger particles (i.e., mammalian cells). As much smaller par-
ticles, the probability of bacteria not being in the proper position or
orientation within the stream to be properly assessed by the instru-
ment is much higher. From our experience, using more dilute sam-
ples (and consequently sorting at slower rates) is very important for
dealing with this issue. The choice of the FP that will be used for
your project (see previous section) needs to be coordinated with the
lasers and filter sets that are available with the instrument that will be
used. While most instruments will be compatible with many green
and red FPs, suitable lasers for more exotic FP selections are optional
additions and not always available. The sorting rate achievable for
modern instruments analyzing bacteria depends on a variety of fac-
tors; however, most flow cytometers can efficiently sort at rates in
the 5,000–10,000 events per second range or higher for eukaryotic
cells. The more pressing issue is for the present method is not how
quickly the bacteria can be sorted, but how quickly they can be
sorted without a significant drop off in purity. Be sure to describe
the strain of bacteria that will be used and its safety requirements to
facility personnel prior to the project; many facilities will not be
equipped to deal with pathogenic species.
3.2 Establishing Prior to beginning selection steps, it is important to test the feasibility
Controls and FACS of your particular project, to experience running your samples on the
Parameters flow cytometer that will be employed, to gain insight into the proper-
ties of your samples, and to determine the gating and other settings
that will best suit your selection. This section presents a set of experi-
ments that will help accomplish these goals. Depending upon the
success of these experiments, additional tests along these same lines
may be required in order to prepare for the selection.
At a minimum, these tests require two plasmids to be stably
introduced in the strain that will be used for the selection:
1. The library plasmid containing the FP you have chosen but
lacking any potential regulatory elements in its 5′ UTR
(FP-positive control plasmid).
2. A similar plasmid that contains a different antibiotic resistance
cassette and that lacks a fluorescent protein (FP-negative control
plasmid).
In addition, we strongly suggest also applying a third plasmid.

This plasmid will simulate the FP expression levels of a riboswitch
from your library that is locked in the “expression off” state
(FP-repressed control plasmid). For example, for a riboswitch
library targeting a transcriptional termination mechanism, this
plasmid will contain the intact transcriptional terminator that will
be a part of the library upstream of the FP. Since some leak in
expression will occur with even very efficient riboswitches, this
control plasmid will provide a more realistic representation of the
fluorescence intensity of riboswitches in the “off” state. Like the
FP-negative control plasmid, the FP-repressed control plasmid
should parallel the FP-positive control plasmid (same basic vector
with the same promoter, same origin of replication, etc.) but
should have a different antibiotic resistance cassette.
1. Transform your selected host strain with the FP-positive and
FP-negative control plasmids described above. Plate onto LB
agar plates supplemented with the appropriate antibiotics, and
grow overnight at 37 °C.
2. Inoculate a 2 mL culture of the selected growth medium
(see Subheading 2.1) supplemented with antibiotics with a
single colony picked from each construct transformed in step 1
and grow overnight at 37 °C for ~16–18 h with shaking
(250 rpm).
3. The following day, subculture both samples 1:1,000 into 2 mL
of fresh medium and incubate at 37 °C for ~6–8 h (see Note 1)
with shaking to late log phase (OD600 ~ 0.8).
4. Measure the OD600 of both cultures, and infer the approximate
concentration of bacteria (# bacteria/mL culture) (see Note 2).
5. Transfer 1 mL of each culture from step 4 into a 1.5 mL
microfuge tube and centrifuge for 10 min at 5,000 × g to pellet
the bacteria. Wash several times with an equal volume of DPBS
(see Subheading 2.2). Resuspend in 1 mL of DPBS.
6. Using the approximate cell concentrations determined in step
4, prepare ~5 mL of the following cell suspensions in DPBS at
a concentration of ~106 to 107 cells/mL (see Note 3):
(a) Cells containing the FP-positive control plasmid.
(b) Cells containing the FP-negative control plasmid.
(c) A 50:50 mixture of the FP-positive and FP-negative cells.
(d) A 95:5 mixture of the FP-positive and -negative cells
(FP-positive:FP-negative).
(e) A 95:5 mixture of the FP-positive and -negative cells
(FP-negative:FP-positive).
7. Put all samples on ice, and transport them to the flow cytom-
eter. Keep samples on ice both before and after FACS analysis.
Fig. 3 Schematic illustrations of the expected fluorescence histograms for (a) cells containing the FP-positive
control plasmid, (b) cells containing the FP-negative control plasmid, (c) a 50:50 mixture of the FP-positive and
FP-negative cells, (d) a 95:5 mixture of the FP-positive and FP-negative cells, and (e) a 95:5 mixture of the
FP-negative and FP-positive cells. Each dash line with an arrow indicates an example of the gating threshold
settings that can be used for sorting, with the arrow indicating the cells to be kept from the sort
Keep a small volume of samples c, d, and e. They will be

required for step 13.
8. Beginning with sample a, analyze the bacteria using your flow
cytometry system. As a first step you will need to determine the
forward scatter and side scatter profile of your bacteria
(see Note 4). To assist in determining these parameters, the
fluorescence intensity profile can be used, since intact bacteria
should represent the dominant fluorescent signal for sample a.
9. Examine the fluorescence histograms for each of the five
samples that have been prepared. Figure 3 provides an illustra-
tion that depicts the expected profiles for each of these samples.
See Note 5 if distinct peaks are not observed for your two
control populations (in sample c).
10. The goal of each of the following steps in this section is to
gauge your ability to sort mixtures of the control populations.
Begin with the 50:50 mixture. Establish gating (threshold
fluorescence values) for a sort such that only the highly fluores-
cent cells are kept (see Fig. 3c). You might wish to try a few
different gating settings to vary the stringency of your selection.
Collect a minimum of several thousand positive cells for each

sort, and store samples on ice. Be sure to carefully label all col-
lection tubes.
11. Repeat the above step, this time sorting to keep only the low-
fluorescence cells (see Fig. 3c).
12. Move on to the mixtures containing 95 % of one control and
only 5 % of the other. Sort as described in step 10, keeping
only the rare population of cells in each instance (Fig. 3d, e).
Adjusting the stringency of the gating settings for these popu-
lations can be particularly informative of the gating that will be
required to get the desired purity for future sorts.
13. For mixtures c, d, and e, as well as for each of the sorted samples
described above, plate equal dilutions on two sets of LB agar
plates: one set containing the antibiotic to which the FP-positive
control plasmid is resistant and the other set containing the
antibiotic for FP-negative selection. Aim to plate several
hundred total bacteria per plate; plating multiple dilutions is
advised.
14. Count the total number of colonies on all plates from step 13,
and record the CFU numbers.
15. Determine the percentage of each control that has been recov-
ered from each sort. Comparing these figures to the initial
populations, the efficiency of each sort can be ascertained. For
populations with clearly distinct, non-overlapping fluorescence
intensity peaks (such as those shown in Fig. 3) it should be
possible to separate the cells with >95 % purity using conserva-
tive gating.
16. Optional: It is strongly recommended that the process
described above be repeated using the FP-repressed control
plasmid (described in the opening of this section) in place of
the FP-negative control. In this instance, the two populations
are likely to have more similar levels of fluorescence, and mix-
tures may not yield two completely distinct histogram peaks.
This process should provide excellent insight into appropriate
gating settings for the selection steps. If these populations can-
not be separated from one another, the design of the library,
the choice of the FP, or the inhibitory element used to control
expression might need to be altered prior to beginning the
selection.
3.3 Artificial A thorough description of the possibilities and considerations for

Riboswitch Selection the design and construction of a library of “riboswitch candidates”
would provide subject matter for a lengthy manuscript and is
beyond the scope of this chapter. To create a large library, a high-
efficiency cloning method in conjunction with highly competent
host cells will be required. Once the design and molecular cloning
steps have been sorted out, single colonies containing individual

library sequences should be pooled and stored for future use. As a
general guide, we suggest a protocol along the following lines:
Transform your plasmid DNA library into your chosen host strain,
and plate the transformants to yield ~1,000 colonies per 10 cm LB
agar plate. Add 2–3 mL of medium to each plate, and scrape the
colonies off the agar and into the liquid. Mix well, and transfer a
portion to a common pool with the bacteria from other plates.
This pool should be mixed well, diluted to an appropriate concen-
tration, and multiple aliquots should be stored in LB + 15 % glyc-
erol (or another storage solution) at −80 °C for future use. For
large library sizes, several of these pools can be created, stored, and
grown separately (to maintain maximum diversity) and combined
immediately prior to selection (step 6 below).
The following protocol assumes a selection for a riboswitch
that activates the expression of the downstream gene(s) in response
to high ligand levels. With obvious adjustments, the same method-
ology can be used to identify riboswitches that repress the expres-
sion of the regulated gene(s) at high target molecule concentrations.
The method requires two iterations of FACS for each selection
cycle: one to identify clones that yield relatively low levels of fluo-
rescence when the target molecule is present at low levels (or
absent) and the other to identify clones with relatively high FP
expression when the target molecule is present at high concentra-
tions. The order of these selection steps can be reversed from that
described below (see Note 6).
1. Inoculate 2 mL starter cultures for host cells carrying the
FP-positive control plasmid and those carrying the FP-negative
control plasmid (see Subheading 3.2). Allow these cultures to
grow to stationary phase overnight. These populations will be
required as controls to configure the FACS settings.
2. Thaw an aliquot of your host cells carrying the plasmid library
(see above) on ice for ~10–15 min.
3. Dilute the library significantly into the growth medium selected
for your experiment wherein the target molecule is present at
low levels (or absent). Under these conditions, the riboswitches
are expected to be in the unbound state with the expression of
the FP repressed. Grow the culture at 37 °C with shaking until
late log phase. [The dilution factor of the starting library
should be such that the population has undergone many
generations and the culture can grow at least ~5–8 h before
reaching late log phase (see Note 7)].
4. Concurrent with step 3, subculture the two control popula-
tions that are described in step 1 and grow to late log phase in
the same growth medium used to culture the library.
5. Collect each of the cultures described in steps 3 and 4 and

place on ice for 10 min.
6. Pellet a fraction of each of the cell populations by centrifuga-
tion at 5,000 × g for 10 min at 4 °C. The number of cells
that need to be collected depends on the size of the library
(see Note 8). Wash 3× with an equal volume of ice-cold DPBS
and resuspend in DPBS at a final concentration of ~106–107
cells/mL (see Note 3). Bring samples to the flow cytometer.
7. Using the FP-positive and the FP-negative populations, estab-
lish instrument settings for the flow cytometer to create an
appropriate forward and side scatter window for your bacteria
and to create an optimal spread in fluorescence between the
two controls.
8. Using the settings established in step 7, examine the fluores-
cence spread of your library population. Note the spread of the
population as well as the fluorescence intensity of the library
relative to those of the positive and negative controls.
9. Establish the gating settings for sorting your library to keep
the cells with a relatively low fluorescence value and to discard
cells with a relatively high fluorescence value. This is a crucial
and difficult decision that should be guided by the desired
characteristics of the riboswitches you seek, the observations
noted in step 8 and your experiences from the experiments
described in Subheading 3.2. If the gating is not stringent
enough the enrichment will be poor, and if the gating is too
stringent desirable sequences will be lost. Unless the fluores-
cence distribution of the cells dictates otherwise, the fraction
of the population of cells that is kept at any given FACS step
should be small. This fraction should not drop below ~0.5 %,
however, since FACS is disposed to yielding a small portion of
erroneously processed particles that will likely dominate the
extremities of the spectrum.
10. Sort the desired number of cells (see Note 8), keeping your
samples on ice both before and after sorting.
11. When sorting is complete, dilute the surviving population into
a rich growth medium and grow for several hours at 37 °C
with shaking to revitalize cells after the stress of sorting and to
recover cell numbers.
12. As with the starting library, this pool should be mixed well,
diluted to an appropriate concentration, and multiple aliquots
should be stored in LB + 15 % glycerol (or another storage
solution) at −80 °C for future use. This population can be
described as Round 1A.
13. Repeat the entire above procedure using the Round 1A
population as the starting library. For this iteration, grow the
cells under conditions that result in a relatively high target

molecule concentration. Riboswitches are now expected to be
in the bound state, which activates gene expression. Therefore
the population with a relatively high level of fluorescence
should be isolated and the low-fluorescence cells discarded.
Collect, grow, and store the isolated population as described
above. This population is dubbed Round 1B.
3.4 Identifying The two-step FACS selection described in Subheading 3.3 can be
and Characterizing repeated any number of times prior to checking the progress of the
Individual selection and seeking individual riboswitch candidates. For most
Riboswitches projects, two to three rounds are likely to be sufficient; however,
this will depend on factors such as the frequency of desirable
sequences in the library, the size of the library, and the extent of
purification that is achieved at each FACS step. Importantly, the
pool of clones that remains after each FACS step must be saved
(as described in the previous section), and so it is always possible to
return to any stage in the selection process. One addition that can
help improve the final outcome of your project is to identify
the most active riboswitch sequences from a given selection
(populations 2A, 2B, 3A, #B, etc) and then to build a second
library based on these sequences with random residues introduced
at key locations [9]. The process of identifying and improving
upon promising sequences can allow you to more shrewdly survey
the almost limitless sequence space that exists in a riboswitch-sized
oligonucleotide library.
Even after your library has been significantly enriched, it will
still contain a large number of unique sequences with varying levels
of activity. The procedure described below allows for the identifica-
tion of the most active individual sequences from your final, most
enriched population.
1. Thaw an aliquot of the population of cells that you wish to
survey on ice for 10–15 min.
2. Create several serial tenfold dilutions of the thawed cells in LB,
plate onto LB agar plates containing the appropriate antibiotic,
and incubate overnight at 37 °C. The goal of this step is to
generate plates containing large numbers of well-separated,
single colonies.
3. Fill a 96-well plate with 200 μL of LB plus the appropriate
antibiotic in each well. Inoculate each well with a single colony
from step 2. Set aside a few wells and inoculate them with cells
carrying the FP-positive and the FP-negative control vectors to
serve as controls for subsequent fluorescence assays. Grow
overnight (~16 h) with shaking at 37 °C.
4. The following day, fill a 96-well storage plate (ABgene) with
100 μL per well of LB + 30 % glycerol. Using a multichannel
pipette, add 100 μL per well of the plate from step 3 to the
corresponding well of the storage plate and mix well. Store

indefinitely at −80 °C. Repeat steps 3 and 4 to create multiple
plates in order to screen larger numbers of clones. Alternate
media can be used to create the frozen stocks if required (see
Note 9).
5. Prepare duplicate 96-well plates, one containing 250 μL of the
medium that creates low target metabolite levels and the other
containing 250 μL of the high target metabolite concentration
medium.
6. Pin the frozen stock plate prepared in step 4 to inoculate each of
the duplicate plates from step 5 and grow for several hours
(~5–8 h is suggested for most situations) at 37 °C with shaking.
7. Measure the OD600 for the plate using a plate-based
spectrophotometer.
8. Using a multichannel pipette, transfer 100 μL of sample from
each plate to a 96-well half-area flat-bottom NBS microplate.
9. Analyze the fluorescence of the duplicate plates using a TECAN
Safire fluorimeter or a similar instrument. The excitation/
emission settings should mirror those that are optimal for the
FP you have chosen.
10. Normalize the fluorescence values for cell growth by dividing
each fluorescence value by the corresponding OD600 value for
the same sample.
11. Examine the normalized fluorescence values from your control
wells. The values should be similar between replicates and
between duplicate plates.
12. Calculate activation ratios for each well by dividing the
corrected fluorescence values from the plate grown in high
levels of the target molecule to those from the plate grown in
low levels of the target molecule. An activation ratio of 1
indicates that the target molecule does not influence FP expres-
sion. Higher activation ratios indicate greater levels of (possible)
riboswitch activity.
13. Identify a handful of clones that represent the most active
sequences observed. Streak these sequences out on fresh LB
agar plates from the frozen stock. Assay these clones using a
similar methodology to that described above but on a lower
throughput scale using multiple replicates of each sample in
order to confirm the observed activity.
14. Purify plasmid DNA from the most active clones, and sequence
the relevant region of these vectors using standard techniques.
15. These sequences can be used as the basis to create a new library
(as described in the introduction to this section) or can be
further analyzed as described below.
16. A number of experiments will be required to confirm that the

regulation observed for these samples is due to a riboswitch
mechanism, to characterize the regulatory activity, and to sort
out the mechanism of regulation. Some of these experiments
may include the following:
– Creating point mutants that are known to disrupt the
binding capabilities of the aptamer: This mutant should
lose regulatory activity, which would imply that the regula-
tion is indeed due to the riboswitch binding its ligand.
– Assaying for regulatory activity over a broad range of tar-
get molecule levels to generate a dose–response curve:
This will indicate the dynamic range of the riboswitch. To
ensure the specificity of your riboswitch, this can also be
repeated using a similar concentration range of a close
chemical derivative of the target molecule.
– A myriad of experiments can be performed to help identify
the level at which the riboswitch is regulating the down-
stream gene (transcription, translation, RNA processing).
These can include mutagenesis, in vitro transcription or
translation, analyzing transcript levels from cells, and
checking for regulation of co-transcribed genes with dis-
tinct ribosome-binding sites.
– In-line probing [26] has been a very successful technique for
monitoring the structural changes undergone by RNA in
response to binding a ligand. Comparing the RNA degrada-
tion patterns at different target molecule concentrations
using this technique can be instrumental in identifying the
residues within the riboswitch that undergo significant struc-
tural changes in response to target binding. This information
is key for piecing together a detailed molecular mechanism.
4 Notes
1. Throughout this document, late log phase is suggested as the

end point for key culture steps. For many of these same steps,
the amount of time that the culture is allowed to grow prior to
collection is also an important factor (see Note 7), which can
create difficulties, particularly for conditions/strains that sup-
port fast growth. To accommodate both of these stipulations,
the volume of the culture and the dilution factor of the starter
culture are variables that can be manipulated. Alternately,
another subculture step can be introduced part way through
the growth stage to dilute the bacteria and allow for additional
time and growth to occur before stationary phase begins.
2. OD600 measurements are a useful means of gauging the relative
concentration of bacteria in a culture medium. The OD600
can be used to approximate the number of bacteria/mL.

To accomplish this, plate several dilutions of the bacterial
culture at a given OD600 and count the colony-forming units
(CFU) at each dilution. As an approximate guide, for normal
E. coli cultures growing in LB, an OD600 of 1.0 is often
estimated to indicate ~109 bacteria/mL.
3. Speak with the operator of the FACS instrument you will be
using regarding sample preparation. Samples should generally
be diluted to ~106 to 107 cells/mL using 1× DPBS (or a similar
solution), but we strongly suggest optimizing the sort rates to
best suit the sorting speed and purity required for your individ-
ual setup and experimental needs. Immediately prior to sorting,
samples can be strained through a polystyrene cell strainer cap
(see Subheading 2.2) to reduce issues with aggregation, however
this is not generally required for bacterial cells.
4. In flow cytometry, the measures of forward and side scatter are
useful parameters to identify particles with the approximate
size and makeup of those that you intend to sort. Forward
scatter profiles are typically proportional to the diameter of the
cell, whereas side scatter profiles highlight the internal com-
plexity or granularity of a cell. Prior to sorting, it is important
that the instrument has demonstrated that it is capable of
identifying the particles (bacterial cells) that you wish to sort.
The flow cytometer’s settings should be adjusted such that
only particles with the correct forward/side scatter profile are
examined and sorted.
5. The fluorescence profiles of the positive and negative controls
will, in most cases, represent the highest and lowest levels of
fluorescence achievable using your current setup. If there is
not a generous spread in the fluorescence values of these popu-
lations (~2 orders of magnitude or more), the FP is not a dom-
inant enough source of fluorescence under these conditions
and the selection methodology described will not be possible.
To overcome the weak signal various changes can be applied,
for example increasing expression of the FP (by increasing the
plasmid copy number and the strength of the promoter or by
altering the RBS region to a more efficient sequence) or select-
ing an alternate FP that produces a stronger signal or that
incurs less interference from autofluorescence.
6. The order of the two FACS steps (to select for low FP expres-
sion in the absence of the ligand and for high expression in the
presence of the ligand) can be switched. One consideration is
that most FPs are very stable and will persist in the population
for many hours of growth. When selecting for highly fluores-
cent cells first, one must be sure to dilute this population
significantly and allow many generations to pass before select-
ing for cells with a low level of fluorescence to avoid desirable
sequences from being lost. If multiple cycles of the two-step
FACS protocol are used, this is a consideration regardless of

the order of the steps in a given cycle.
7. It is important for sufficient time to pass and for many growth
generations to occur during the culture phase prior to FACS.
One reason for this is to allow for the dilution of the FP and of
the target molecule from the frozen stock. It is important that
the target molecule equilibrates at the desired levels created by
the growth medium conditions and for these levels to dictate
the regulatory decisions of any riboswitches in the population.
Following these regulatory decisions, the FP must have time to
mature and accumulate. This is essential in order for the
fluorescence level of cells in the population to reflect any
riboswitch activity of the individual plasmids they harbor.
8. The number of cells that should be sorted depends on the size
of the library and the limitation imposed by sorting speeds (see
Note 3). To comfortably survey your library, ~100 cells should
be sorted per sequence in the library. For example, for a library
size of 105 sequences, it might be possible to sort, 107 cells in
a single session, which would provide good library coverage. If
multiple cycles of selection are conducted, somewhat lower
numbers of cells can be sorted for later rounds when the library
diversity has presumably declined. For very large library sizes
that push the limits of this method, the ratio of cells sorted to
library size can be lowered; however, it should be noted that as
this ratio drops, the probability of stochastically losing desir-
able clones increases. Such large libraries will also require a
different strategy from that described in the introduction of
Subheading 3.3 to select and pool the library. Regardless
of the number of cells that you intend to sort, always bring a
generous excess of sample with you beyond this amount.
9. Just as for the FACS experiments, it is also important when
screening individual clones to ensure that the levels of the tar-
get molecule and the FP from the previous step do not affect
the present experiment (see Note 7). This can be a challenge in
the screening step due to the low volumes involved in the
96-well plate format. Accordingly, if growth in LB creates high
levels of the target metabolite or the FP that affect these assays,
individual clones should be grown and stored in alternate
media that mitigate these effects.
Acknowledgments
The riboswitch work in the Li Lab has been supported by the

Natural Science and Engineering Research Council of Canada
(NSERC).
References
1. Roth A, Breaker RR (2009) The structural and metabolite binding kinetics operate an FMN
functional diversity of metabolite-binding ribo- riboswitch. Mol Cell 18:49–60
switches. Annu Rev Biochem 78:305–334 17. Wickiser JK, Cheah MT, Breaker RR et al
2. Fowler CC, Brown ED, Li Y (2010) Using a (2005) The kinetics of ligand binding by an
riboswitch sensor to examine coenzyme B(12) adenine-sensing riboswitch. Biochemistry 44:
metabolism and transport in E. coli. Chem Biol 13404–13414
17:756–765 18. Gilbert SD, Stoddard CD, Wise SJ et al
3. Topp S, Gallivan JP (2007) Guiding bacteria (2006) Thermodynamic and kinetic charac-
with small molecules and RNA. J Am Chem terization of ligand binding to the purine
Soc 129:6807–6811 riboswitch aptamer domain. J Mol Biol 359:
4. Blount KF, Breaker RR (2006) Riboswitches as 754–768
antibacterial drug targets. Nat Biotechnol 24: 19. Trausch JJ, Ceres P, Reyes FE, Batey RT
1558–1564 (2011) The structure of a tetrahydrofolate-
5. Sudarsan N, Hammond MC, Block KF et al sensing riboswitch reveals two ligand binding
(2006) Tandem riboswitch architectures sites in a single aptamer. Structure 19:
exhibit complex gene control functions. 1413–1423
Science 314:300–304 20. Wachsmuth M, Findeiß S, Weissheimer N et al
6. Jin Y, Watt RM, Danchin A, Huang J (2009) (2013) De novo design of a synthetic ribo-
Use of a riboswitch-controlled conditional switch that regulates transcription termination.
hypomorphic mutation to uncover a role for Nucleic Acids Res 41:2541–2551
the essential csrA gene in bacterial autoaggre- 21. Suess B, Fink B, Berens C et al (2004) A the-
gation. J Biol Chem 284:28738–28745 ophylline responsive riboswitch based on helix
7. Djordjevic M (2007) SELEX experiments: new slipping controls gene expression in vivo.
prospects, applications and data analysis in Nucleic Acids Res 32:1610–1614
inferring regulatory pathways. Biomol Eng 24: 22. Win MN, Smolke CD (2007) A modular and
179–189 extensible RNA-based gene-regulatory platform
8. Stoltenburg R, Reinemann C, Strehlitz B for engineering cellular function. Proc Natl
(2007) SELEX—a (r)evolutionary method to Acad Sci U S A 104:14283–14288
generate high-affinity nucleic acid ligands. 23. Wieland M, Hartig JS (2008) Improved apta-
Biomol Eng 24:381–403 zyme design and in vivo screening enable
9. Fowler CC, Brown ED, Li Y (2008) A FACS- riboswitching in bacteria. Angew Chem Int Ed
based approach to engineering artificial ribo- 47:2604–2607
switches. ChemBioChem 9:1906–1911 24. Ogawa A, Maeda M (2007) Aptazyme-based
10. Lynch SA, Gallivan JP (2009) A flow cytometry- riboswitches as label-free and detector-free
based screen for synthetic riboswitches. Nucleic sensors for cofactors. Bioorg Med Chem Lett
Acids Res 37:184–192 17:3156–3160
11. Lynch SA, Desai SK, Sajja HK et al (2007) 25. Beisel CL, Smolke CD (2009) Design princi-
A high-throughput screen for synthetic ribo- ples for riboswitch function. PLoS Comput Biol
switches reveals mechanistic insights into their 5:e1000363
function. Chem Biol 14:173–184 26. Regulski EE, Breaker RR (2008) In-line prob-
12. Topp S, Gallivan JP (2008) Random walks to ing analysis of riboswitches. Methods Mol Biol
synthetic riboswitches—a high-throughput 419:53–67
selection based on cell motility. ChemBioChem 27. Tsien RY (1998) The green fluorescent protein.
9:210–213 Annu Rev Biochem 67:509–544
13. Muranaka N, Abe K, Yokobayashi Y (2009) 28. Shaner NC, Steinbach PA, Tsien RY (2005)
Mechanism-guided library design and dual A guide to choosing fluorescent proteins. Nat
genetic selection of synthetic OFF riboswitches. Methods 2:905–909
ChemBioChem 10:2375–2381 29. Telford WG, Hawley T, Subach F et al (2012)
14. Shapiro HM (2003) Practical flow cytometry, Flow cytometry of fluorescent proteins.
4th edition. Wiley-Liss, New York, NY Methods 57:318–330
15. Macey MG (2007) Flow cytometry: principles 30. Chudakov D, Matz M, Lukyanov S et al (2010)
and applications. Humana Press, Totowa, NJ Fluorescent proteins and their applications in
16. Wickiser JK, Winkler WC, Breaker RR et al imaging living cells and tissues. Physiol Rev
(2005) The speed of RNA transcription and 90:1103–1163
Chapter 6
FRET-Based Optical Assay for Selection of Artificial

Riboswitches
Svetlana V. Harbaugh, Molly E. Chapleau, Yaroslav G. Chushak,
Morley O. Stone, and Nancy Kelley-Loughnane
Abstract
Artificial riboswitches are engineered to regulate gene expression in response to a variety of non-endogenous
small molecules and, therefore, can be useful tools to reprogram cellular behavior for different applica-
tions. A new synthetic riboswitch can be created by linking an in vitro-selected aptamer with a randomized
expression platform followed by in vivo selection and screening. Here, we describe an in vivo selection and
screening technique to discover artificial riboswitches in E. coli cells that is based on TEV protease–FRET
substrate reporter system.
Key words Aptamer, Riboswitch, In vivo selection, TEV protease, FRET-based fusion protein
1 Introduction
The ability to create artificial riboswitches that control gene expression

in response to desired small molecules has important implications
for the development of chemical and synthetic biology. Riboswitches
can be powerful tools and have been used to study microbial genet-
ics [1], to identify new antimicrobial targets in a variety of bacterial
species [2, 3], and to develop logic gates for reprogramming cell
behavior and engineering metabolic pathways [4–6]. New ribo-
switches can be created by linking an in vitro-selected aptamer that
binds a target molecule with a randomized expression platform
followed by in vivo selection and screening. While methods to
discover aptamers that selectively recognize and bind a desired
compound are well established [7, 8], the next challenge has been
to develop and optimize a method for converting these aptamers
into artificial riboswitches that are active in cells. Since the starting
point for in vivo selection is large libraries of riboswitches, there are
some key requirements for genetic (in vivo) selection in order to
77
78 Svetlana V. Harbaugh et al.
result in functional switches. Ideally, a method for discovering

riboswitches should be rapid, sensitive, inexpensive, and of high
throughput. The method should have the ability to quantitatively
distinguish the switches with the best performance characteristics
from those switches that only weakly activate or repress gene
expression. Since engineered riboswitches ultimately control gene
expression, the efficiency of the selection technique generally depends
on the choice of reporter gene system.
We have developed a FRET-based optical assay for monitoring
riboswitch activation in E. coli cells [9]. This assay is based on
expression of two genes, tobacco etch virus (TEV) protease [10]
and its optical engineered protein substrate, within the same cell.
The protein substrate is a fluorescence resonance energy transfer
(FRET) construct between two fluorescent protein conjugates
coupled with a peptide linker containing a TEV protease cleavage
site (17 amino acid peptide sequence). The first protein in the pair
is enhanced green fluorescent protein (eGFP), which acts as the
electron energy donor; the second is a non-fluorescent mutant of
yellow fluorescent protein called resonance energy-accepting chro-
moprotein (REACh), which acts as an electron energy acceptor
[11]. Use of this type of FRET pair eliminates acceptor fluores-
cence, and therefore little to no fluorescence is observed prior to
cleavage [9].
In order to evaluate the performance of the reporter system we
placed a previously developed theophylline synthetic riboswitch
[12] upstream of the TEV protease gene. Activation of the ribo-
switch initiated translation of TEV protease, which cleaved the
FRET linker, allowing detection of eGFP fluorescence (Fig. 1).
Previously, Gallivan and co-workers showed that riboswitch incor-
poration can significantly reduce the expression of downstream
genes; for instance, in the theophylline synthetic riboswitch, ana-
lyte induces translation of only ~16 % of the transcripts while ~84 %
remain inactive [12]. In this case, a highly sensitive enzymatic
reporter system is advantageous for riboswitch discovery, since
only a small amount of enzyme is required for cleavage of the
substrate and achievement of a measurable optical signal. Using
our reporter system, we were able to observe a detectable increase
in fluorescence intensity as early as 30 min post analyte addition
[9], when presumably only a small amount of TEV protease was
present in the system. In comparison with other enzymatic reporters
such as the widely used β-galactosidase and luciferase, our system
does not require addition of substrate; fusion protein substrate can
be directly expressed in cells. Moreover, we have demonstrated
superiority of TEV protease–FRET substrate system over direct
coupling of the riboswitch with fluorescent protein in terms of sen-
sitivity. When the eGFP gene was placed downstream of theophyl-
line riboswitch, we observed a very modest increase (~1.4-fold) in
fluorescence intensity of cells in response to analyte [9]. In contrast,
riboswitch activation of TEV protease gene expression followed by
FRET-Based Riboswitch Selection 79
Fig. 1 TEV protease–FRET substrate reporter system for monitoring riboswitch performance in E. coli cells.
In the absence of ligand, extensive pairing of riboswitch with ribosome-binding site (RBS) prevents translation
of TEV protease. Ligand binding induces conformational change of the riboswitch, unpairing RBS and initiating
translation of the TEV protease mRNA. The produced protease cleaves the FRET-based protein substrate, yield-
ing an increase in fluorescent emission at 510 nm upon excitation at 457 nm
cleavage of FRET protein resulted in an 11.3-fold increase in

fluorescence intensity in response to theophylline.
Here we describe a screening protocol that is based on TEV
protease–FRET substrate reporter system and was used to identify
artificial riboswitches that activate protein translation in E. coli cells
in the presence of 2,4-dinitrotoluene [13]. The in vivo selection
presented here focuses on isolation of synthetic riboswitches that
upregulate gene expression at the posttranscriptional level based
on the availability of the ribosome-binding site (RBS). However, it
is also possible to adapt this technique to discover riboswitches that
downregulate gene expression or that act at the transcriptional
level. The following procedure assumes that the user already has an
aptamer (from literature or from performing previously described
aptamer selection methods [7, 8]) that binds to the desired cell-
permeable and nontoxic compound. The first step in riboswitch
selection is construction of a riboswitch library by incorporating a
randomized expression platform (8–15 degenerate bases) between
the aptamer sequence and RBS coupled with the TEV protease-
encoding sequence. The next step is co-transformation of created
plasmid library and a plasmid encoding FRET-based TEV protease
substrate into E. coli cells. Then, all obtained clones must be visu-
ally screened for fluorescence increase in the presence of the
compound of interest. Colonies that appear to show increased
fluorescence in the presence of the desired compound are subjects
for further screening in liquid medium. Clones showing the highest

riboswitch performance in cell cultures can be further character-
ized by performing fluorescence intensity measurements in cell
lysates where background fluorescence of the medium is eliminated.
Candidate riboswitches identified by screening can be sequenced and
their secondary structures predicted using established methods [14].
2 Materials
Prepare all media and buffer solutions using ultrapure water

(prepared by purifying deionized water to obtain a sensitivity of
18 MΩ cm at 25 ºC). For antibiotic solutions and enzymatic
reactions, use nuclease-free water. Autoclave media solutions, and
filter-sterilize antibiotic and inducer solutions. Store media and
buffer solutions at room temperature. Store antibiotic and inducer
solutions at −20 ºC.
2.1 Construction of 1. MAX Efficiency® DH5α™ Competent Cells (Invitrogen,

Plasmid Containing Carlsbad, CA).
FRET-Based Protein 2. Plasmid containing YFP coding sequence (we have used
pET21a(+):YFP plasmid vector, created in the lab).
3. QuikChange II Site-Directed Mutagenesis Kit (Agilent
Technologies/Genomics).
4. Plasmid containing eGFP coding sequence (we have used
pIVEX2.3d:eGFP plasmid vector—a generous gift from
Dr. Vincent Noireaux from the University of Minnesota).
5. PfuUltra High-fidelity DNA Polymerase (Agilent Technologies/
Genomics).
6. NdeI, HindIII, and BamHI restriction enzymes and supple-
mented reaction buffers (Invitrogen, Carlsbad, CA).
7. T4 DNA Ligase and supplemented 5× T4 DNA Ligase Buffer
(Invitrogen, Carlsbad, CA).
8. E. coli expression vector (we have used pHWG640 plasmid
vector—a generous gift from Dr. Josef Altenbuchner from the
Institute of Industrial Genetics at the University of Stuttgart,
Germany).
9. UltraPure™ DNase/RNase-Free Distilled Water (Invitrogen,
Carlsbad, CA).
10. Ampicillin: 50 mg/mL stock solution in water, final concen-
tration 100 μg/mL of media for all procedures.
11. Chloramphenicol: 34 mg/mL stock solution in 100 % ethanol,
final concentration 25 μg/mL of media for all procedures.
12. LB agar: 35 g/L Difco LB Agar (Becton, Dickinson and
Company, Franklin Lakes, NJ).
Fig. 2 Plasmid map of pSAL:aptamer TEV used as a template to construct

riboswitch library. The plasmid contains aptamer sequence placed upstream of
TEV protease encoding gene under the control of ptac promoter, β-lactamase
encoding gene responsible for resistance to ampicillin
13. Sterile Petri dishes, 100 mm.

14. Agarose.
15. 1× Tris–acetate–EDTA (TAE) buffer prepared from TAE
50× solution.
16. 2-Log DNA Ladder (0.1–10.0 kb) and Gel Loading Dye, Blue
(6×) (New England Biolabs, Ipswich, MA).
17. SYBR® Safe DNA Gel Stain (Invitrogen, Carlsbad, CA).
18. MinElute PCR Purification Kit (Qiagen, Valencia, CA).
19. QIAEX II Gel Extraction Kit (Qiagen, Valencia, CA).
20. QIAprep Spin Miniprep Kit (Qiagen, Valencia, CA).
2.2 Construction 1. E. coli expression plasmid containing the aptamer of interest

of a Riboswitch coupled to TEV protease reporter gene. (Our construct of the
Library and FRET- aptamer upstream of TEV protease gene was made by GenScript
Based Selection (Piscataway, NJ). The construct was cloned into pSAL plasmid
vector (Fig. 2)—generous gift from Dr. Justin Gallivan from
Emory University, Atlanta, GA.)
2. Phusion DNA polymerase (New England Biolabs, Ipswich,
MA).
3. DpnI restriction endonuclease (Promega, Fitchburg, WI).
4. MinElute PCR Purification Kit (Qiagen, Valencia, CA).

5. T4 Polynucleotide Kinase and 1× Polynucleotide Kinase
Reaction Buffer (New England Biolabs, Ipswich, MA).
6. 1 mM Adenosine 5′-triphosphate disodium salt solution (ATP)
prepared by diluting 100 mM ATP solution.
7. MinElute Reaction Cleanup Kit (Qiagen, Valencia, CA).
8. Quick Ligation Kit (New England Biolabs, Ipswich, MA).
9. One Shot® TOP10 Chemically Competent E. coli cells
(Invitrogen, Carlsbad, CA).
10. Ligand of interest.
11. Rhamnose: 30 % stock solution in water, final concentration
0.4 % in media for all procedures.
12. Luria-Bertani (LB) media: 20 g/L Difco LB Broth (Becton,
Dickinson and Company, Franklin Lakes, NJ).
13. 24-well plates.
14. Absorbance and fluorescence microplate reader.
15. Dark Reader blue light transilluminator, amber screen, viewing
glasses (Clare Chemical Research, Dolores, CO).
16. Incubator with shaker.
2.3 Analysis 1. Baffled culture flasks, 500 mL.

of Individual 2. Lysis buffer: 100 mM Tris–HCl, pH 8.0, 150 mM NaCl,
Riboswitch Clones 1 mM MgCl2, 0.1 mg/mL of lysozyme (Sigma, St. Louis,
MO), 1 μL/mL of benzonase nuclease (Sigma, St. Louis, MO),
8 mM iodoacetamide (Sigma, St. Louis, MO).
3. Centrifuge.
3 Methods
3.1 Creation of 1. Perform sequential site-directed mutagenesis of YFP gene with

Plasmid Containing QuikChange II Site-Directed Mutagenesis Kit. To create a
Gene Encoding YFP mutant (Y145W), use E. coli cloning vector containing
FRET-Based Protein– YFP coding sequence as a template, forward (5′-CACAAGC
TEV Protease TGGAGTAC AACTGGAACAGCCACAACGT CTATATC-3′)
Substrate and reverse (5′-GATATAGACGTTGTGGCTG TTCCAGTT
GTACTCCAGCTTGTG-3′) primers. Perform PCR reaction,
treatment with DpnI endonuclease, and transformation into
competent E. coli cells according to the manufacturer’s proto-
col. Pick six to eight obtained clones for plasmid isolation
using QIAprep Spin Miniprep Kit. Verify sequences of purified
plasmids by DNA sequencing, and identify plasmids contain-
ing Y145W-YFP. To create a YFP mutant (Y145W/H148V),
use E. coli cloning vector containing Y145W-YFP coding

sequence as a template, forward (5′-CTGGAGTACAAC
TGGAACAGCGT GAACGTCTATATCATGGCCG-3′) and
reverse (5′-CGGCC ATGATATAGACGTTCACGCTGTTCC
AGTTGTAC TCCAG-3′) primers. Perform all manipulations
which are described above. Resulting plasmid vector should
contain sequence encoding a double mutant of YFP, Y145W/
H148V-YFP, called REACh.
2. Amplify the coding sequence of REACh (Y145W/H148V-
YFP) using forward (5′-GTGGATCCGAAAACCTGTAC
TTCCAGAGCGGCACCGTGAGCAAGGGCGAA-3′) and
reverse (5′-CGTAAGCTTTTAATGGTGATGGTGATGGTG
CTTGTACAGCTCGTCC-3′) primers. Amplify the coding
sequence of eGFP using forward (5′-CGTCATATGGTGA
GCAAGGGCGAGGAG-3′) and reverse (5′-CGTGGATCC
GGCCTT CTTGTACAGGCTCTTGTACAGCTCGTC-3′)
primers. (PCR will allow the addition of specific restriction
sites (underlined) to the ends of amplified products, a nucleo-
tide sequence (bold) encoding linker peptide and a nucleotide
sequence (italic) encoding 6× His-tag at the C-terminus of
REACh.) Perform both PCR reactions in 50 μL volume using
PfuUltra High-fidelity DNA polymerase according to the
manufacturer’s protocol. Analyze PCR-amplified products by
standard agarose gel electrophoresis (see Note 1). Purify PCR
products using MinElute PCR Purification Kit.
3. Digest the PCR products amplified from eGFP and REACh
and plasmid vector (pHWG640) with NdeI–BamHI, BamHI–
HindIII, and NdeI–HindIII restriction enzymes, respectively.
Set up restriction enzyme digestion reactions according to the
manufacturer’s protocol. Separate the digested fragments on
an agarose gel, and purify the DNA fragments using QIAEX II
Gel Extraction Kit.
4. Ligate digested PCR products amplified from eGFP and
REACh into linearized plasmid vector (pHWG640).
Ligation reaction (see Note 2):
Insert:vector molar ratio 5:1

Digested vector 3–20 fmol
Digested PCR product 1 15–100 fmol
Digested PCR product 2 15–100 fmol
Total DNA 0.01–0.1 μg
5× Ligase reaction buffer 2 μL
T4 DNA ligase 1 μL (1 unit)
(continued)
(continued)
Water To 10 μL final
Temperature 25 ºC
Time 3h
For optimal transformation, dilute the ligation reaction fivefold

with 1× TE buffer. (This method of construction introduces 17
amino acid linker peptide (SLYKKAGSENLYFQSGT), which
joins the C-terminus of eGFP to the N-terminus of REACh
and also places 6× His-tag at the C-terminus of REACh. The
linker contains a TEV protease cleavage site (underlined) and
flanking spacer arms to increase TEV protease accessibility.)
5. Transform E. coli DH5α competent cells with 10 ng/μL of
the ligation reaction mixture and grow overnight on LB-
agar plates supplemented with antibiotic (25 μg/mL of
chloramphenicol).
6. Pick the colonies from the plate. Transfer each colony into
5 mL of LB media supplemented with antibiotic and grow
overnight cultures. Collect cells; purify plasmids using QIAprep
Spin Miniprep Kit.
7. Verify sequences of purified plasmids by DNA sequencing. The
plasmid map of the resulting construct (pHWG640:eGFP-TL-
REAChHis) is shown in Fig. 3.
Fig. 3 Plasmid map of pHWG640:eGFP-TL-REACh. The plasmid contains gene

encoding fusion protein composed of eGFP- and REACh-encoding sequences
coupled with a linker (TL) containing TEV protease cleavage site. The fusion
protein gene placed downstream of Rhamnose-inducible promoter (rhaP).
The plasmid carries gene (cat) responsible for resistance to chloramphenicol
3.2 Construction The schematic illustration of the construction of a riboswitch

of a Riboswitch plasmid library is shown in Fig. 4.
Plasmid Library and
1. Design PCR primers for incorporation of 10–15 degenerate
an In Vivo Selection bases between the aptamer and a region of 5–7 constant bases
located immediately before the start codon of a reporter gene,
TEV protease-encoding sequence (see Note 3).
2. Incorporate randomized sequence (expression platform)
between the aptamer and the start codon of a TEV protease
gene by performing PCR reaction. Use E. coli expression
plasmid containing the aptamer of interest coupled to TEV
protease reporter gene as a template and primers designed in
step 1. Perform PCR reaction in 50 μL volume using Phusion
DNA polymerase:
Component Volume/reaction (μL) Final concentration

Nuclease-free water 29
Template DNA (10 ng/μL) 1 10 ng/reaction
Forward primer (5 pmol/μL) 2 0.2 μM
Reverse primer (5 pmol/μL) 2 0.2 μM
dNTP mixture (2.5 mM each) 4 0.2 μ
5× Phusion HF or GC buffer 10 1×
DMSO 1.5 3%
Phusion DNA polymerase 0.5 1.0 units/50 μL PCR
Set up thermocycling conditions according to the manufactur-

er’s protocol. Use 5 μL of the reaction mixture to analyze
PCR-amplified product by standard agarose gel electrophoresis
(see Note 1).
3. To digest the template plasmid, add 5 units of DpnI restriction
endonuclease to the PCR mixture and incubate at 37 ºC
for 60 min. Purify PCR product using MinElute PCR
Purification Kit.
4. To add 5′-phosphates to the PCR product, use T4
Polynucleotide Kinase. Use up to 300 pmol of 5′ termini in
50 μL reaction containing 1× T4 Polynucleotide Kinase Buffer,
1 mM ATP, and 10 units of T4 Polynucleotide Kinase. Incubate
at 37 ºC for 30 min (see Note 4). Purify the phosphorylated
DNA with MinElute Reaction Cleanup Kit.
Fig. 4 Construction of a riboswitch plasmid library. The library is constructed by

performing PCR using template plasmid with an aptamer sequence upstream of
a reporter gene and forward and reverse primers containing degenerate bases at
5′-ends. The resulting PCR product is treated with DpnI to digest the template
plasmid followed by phosphorylation of 5′-ends and ligation of linear DNA into
circular plasmid
5. Self-ligate linear DNA using Quick Ligation Kit:

Linear DNA 50 ng (adjust volume to 10 μL
with nuclease-free water)
2× quick ligation buffer 10 μL
Quick T4 DNA ligase 1 μL
Temperature 25 ºC
Time 5 min
Fig. 5 In vivo screening for functional riboswitches. (a) Obtained after transformation E. coli colonies are picked
and streaked on three LB-agar plates: without inducer and ligand; with inducer (expression of the fusion
protein-encoding gene) and without ligand (no riboswitch activation); with inducer (expression of the
fusion protein-encoding gene) and with ligand (riboswitch activation and TEV protease production). (b) Colonies
that show increased fluorescence in the presence of ligand are selected for screening in liquid medium
(an example is circled ). (c) The gene expression of the selected riboswitch clones is quantified in the presence
and absence of ligand
6. Transform E. coli TOP10 chemically competent cells with 5 ng

of the ligation mixture and 5 ng of E. coli expression vector
containing sequence encoding eGFP-TL-REACh protein (TL-
17 amino acid peptide sequence containing TEV protease
cleavage site). Plate the transformed cells on LB-agar plates
supplemented with appropriate antibiotics (ampicillin and
chloramphenicol) and grow overnight at 37 ºC (see Note 5).
7. Pick all identified colonies from the plate, and streak each
colony onto the same section of each of the three plates:
LB-agar plate supplemented with antibiotics, LB-agar plate
supplemented with antibiotics and inducer (0.4 % Rhamnose)
for eGFP-TL-REACh gene expression, and LB-agar plate
supplemented with antibiotics, inducer, and ligand (Fig. 5a).
(The number of plates will depend on the number of obtained
colonies. In our work, we have analyzed 1,000 clones.) Grow
cells overnight at 37 ºC.
8. Visualize fluorescent colonies using Dark Reader blue light
transilluminator and amber screen or viewing glasses. Pick colo-
nies that show increased fluorescence in the presence of ligand,
and inoculate them in 24-well plates with each well containing
1 mL of LB supplemented with appropriate antibiotics (Fig. 5b).
(The number of plates will depend on the number of colonies
assayed in step 7.)
9. Incubate the 24-well plates overnight with shaking (37 ºC,
215 rpm).
10. The following day, inoculate four 24-well plates for each plate
with overnight cultures with 20 μL of overnight culture. Each
well of 24-well plates should contain 1 mL of LB supplemented
with appropriate antibiotics.
11. Incubate the 24-well plates with new cultures for 2.5 h at
37 ºC with shaking (215 rpm) until appropriate absorbance
at 600 nm is reached (0.3–0.5 when corrected to 1-cm path
length cuvette). Add an appropriate amount of an inducer for
expression of eGFP-TL-REACh protein gene. (In our work,
gene encoding eGFP-TL-REACh protein was cloned in
pHWG640 plasmid vector under control of Rhamnose-
inducible promoter. For induction of the expression, we have
used 0.4 % Rhamnose.)
12. Incubate cultures for 30 min at 37 ºC with shaking (215 rpm).
Then, divide each group of four plates with the same cultures
on two sets of two. Add ligand to the first set of plates (this set
will contain cells with activated riboswitches). Add equivalent
volume of the solvent where ligand is dissolved to the second
set of plates (this set will contain cells with non-activated
riboswitch) (Fig. 5c). Measure the absorbance at 600 nm and
fluorescence at 510 nm (excitation 457 nm) of the cell cultures
in each plate.
13. Incubate the 24-well plates for 6 h at 37 ºC with shaking
(215 rpm). Measure the absorbance at 600 nm and fluores-
cence at 510 nm (excitation 457 nm) of the cell cultures in
each plate. To normalize the data, divide fluorescence intensity
by absorbance for each well.
14. Compare the ratios of the normalized data for cultures
grown in the presence of ligand to those grown in the absence
of ligand (the “activation ratio”) to identify functional
riboswitches.
3.3 Analysis of 1. Using 10 μL of selected culture, inoculate 10 mL of LB sup-

Individual Clones for plemented with appropriate antibiotics. Incubate overnight at
Riboswitch Activation 37 ºC with shaking.
2. The following day, using 1.5 mL of fresh overnight culture,
inoculate six new flasks containing 150 mL of LB supple-
mented with appropriate antibiotics.
3. Incubate flasks with new cultures at 37 ºC with shaking
(220 rpm) until absorbance at 600 nm is reached 0.4–0.5. Add
an appropriate amount of an inducer for expression of eGFP-
TL-REACh protein gene.
4. Incubate cultures for 30 min at 37 ºC with shaking (220 rpm).
Then, divide flasks into two sets of three. Add ligand to the
first set of flasks (this set will contain cells with activated
riboswitches). Add equivalent volume of the solvent (vehicle)
to the second set of flasks (this set will contain cells with
non-activated riboswitch). Measure the absorbance at 600 nm
and fluorescence at 510 nm (excitation 457 nm) of the cell
cultures in each flask (it will be time 0 h). Record the absor-
bance at 600 nm and fluorescence at 510 nm (excitation
457 nm) for each cell culture at time 1, 2, 3, 4, 5, and 6 h after
riboswitch activation (ligand addition). At each time point,
collect 10 mL of each culture for preparation of cellular lysates.
Harvest the cells by centrifugation at 3,200 rpm for 10 min at
4 ºC. Discard the supernatant. Store cell pellets at −80 ºC until
ready for processing.
5. For preparation of cellular lysates, thaw cell pellets on ice for
approximately 15–20 min, vortex, and resuspend in 0.4 mL of
lysis buffer. Incubate cell suspensions on ice for 30 min and
mix two to three times by gently swirling. Remove cell debris
by centrifugation at 13,200 rpm for 30 min at 4 ºC. Transfer
supernatant from each sample into new tubes, and measure
fluorescence at 510 nm (excitation 457 nm). Determine pro-
tein concentrations of the clarified cellular lysates using BCA
assay; follow the protocol provided by the manufacturer. To
normalize the data, divide fluorescence intensity by total pro-
tein concentration for each sample.
6. Compare the ratios of the normalized data for lysates from
cultures grown in the presence of ligand to those grown in the
absence of ligand (the “activation ratio”) to determine the
riboswitch performance at different time points (time course).
4 Notes
1. 1 % Agarose gel electrophoresis: Add 0.5 g agarose to 50 mL

1× TAE. Heat the solution to boiling in the microwave or on
the hotplate stirrer to dissolve the agarose. Cool the solution
to approximately 50 ºC, and add 3 μL of SYBR® Safe DNA Gel
Stain (Invitrogen, Carlsbad, CA). Pour the melted agarose
into a gel tray with the well comb in place. Cool gel at room
temperature for 20–30 min, until it is completely solidified.
Remove the comb, and place the tray with the agarose gel into
the electrophoresis box. Fill gel box with 1× TAE buffer until
the gel is covered. Load a 2-Log DNA Ladder (New England
Biolabs, Ipswich, MA) and samples mixed with gel loading dye
(New England Biolabs, Ipswich, MA). Run the gel at 80–130 V
until the dye line is approximately 75–80 % of the way down
the gel. Using any device that has blue light or UV excitation,
visualize the DNA fragments.
2. If the DNA fragments are too dilute, the reaction can be scaled
up to 20 μL ligation volume.
3. In our previous work, a riboswitch plasmid library was created

by incorporation of randomized expression platform between
aptamer and RBS upstream of TEV protease gene. However,
replacement of fixed RBS sequence with randomized allows to
optimize the strength of the RBS to achieve the high level of
reporter protein expression [15]. The constant region between
the RBS and the start codon (usually five to seven bases) is
important to provide suitable spacing for optimal ribosome
binding. The constant sequence in our library was “CAACAAG”
that was previously used by Gallivan and co-workers [15].
4. Since T4 Polynucleotide Kinase Buffer contains 1 mM dithio-
threitol for optimal activity of T4 Polynucleotide Kinase, fresh
buffer is required to perform a phosphorylation (the reduced
dithiothreitol in older buffer lowers activity). The phosphory-
lation efficiency can be improved by heating to 70 ºC for
5 min, then chilling on ice prior to kinase addition, and adding
PEG-8000 to 5 % (w/v).
5. In our work, we used 50 μL of transformation reaction per
each LB-agar plate.
Acknowledgement
This research was funded by the Air Force Office of Scientific

Research (AFOSR).
References
1. Jin Y, Watt RM, Danchin A, Huang JD (2009) dual genetic selection. J Am Chem Soc 130:
Use of a riboswitch-controlled conditional 16310–16315
hypomorphic mutation to uncover a role for 7. Tuerk C, Gold L (1990) Systematic evolution
the essential csrA gene in bacterial autoaggre- of ligands by exponential enrichment: RNA
gation. J Biol Chem 284:28738–28745 ligands to bacteriophage T4 DNA polymerase.
2. Lee ER, Blount KF, Breaker RR (2009) Science 249:505–510
Roseoflavin is a natural antibacterial compound 8. Roth A, Breaker RR (2004) Selection in vitro
that binds to FMN riboswitches and regulates of allosteric ribozymes. In: Sioud M (ed)
gene expression. RNA Biol 6:187–194 Ribozymes and siRNA protocols, methods in
3. Kim JN, Blount KF, Lim J, Link KH, Breaker R molecular biology, 2nd edn. Humana Press,
(2009) Design and antimicrobial action of Totowa, NJ, pp 145–164
purine analogues that bind guanine riboswitches. 9. Harbaugh S, Kelley-Loughnane N, Davidson
ACS Chem Biol 4:915–927 M, Narayanan L, Trott S, Chushak YG, Stone
4. Jise AM, Soukup GA, Breaker RR (2001) MO (2009) FRET-based optical assay for moni-
Cooperative binding of effectors by an toring riboswitch activation. Biomacromolecules
allosteric ribozyme. Nucleic Acids Res 29: 10:1055–1060
1631–1637 10. Van der Berg S, Löfdahl P-A, Härd T, Berglund
5. Win MN, Smolke CD (2008) Higher-order H (2006) Improved solubility of TEV protease by
cellular information processing with synthetic directed evolution. J Biotechnol 121:291–298
RNA devices. Science 322:456–460 11. Ganesan S, Ameer-beg SM, Ng TT, Vojnovic
6. Sharma V, Nomura Y, Yokobayashi Y (2008) B, Wouters FS (2006) A dark yellow fluorescent
Engineering complex riboswitch regulation by protein (YFP)-based resonance energy-accepting
chromoprotein (REACh) for Förster resonance Development of a 2,4-dinitrotoluene-responsive

energy transfer with GFP. Proc Natl Acad Sci synthetic riboswitch in E. coli cells. ACS Chem
U S A 103:4089–4094 Biol 8:234–241
12. Lynch SA, Desai SK, Sajja HK, Gallivan JP 14. Zuker M (2003) Mfold web server for nucleic
(2007) A high-throughput screen for synthetic acid folding and hybridization prediction.
riboswitches reveals mechanistic insights into Nucleic Acids Res 31:3406–3415
their function. Chem Biol 14:173–184 15. Lynch SA, Gallivan JP (2009) A flow cytome-
13. Davidson ME, Harbaugh SV, Chushak YG, try-based screen for synthetic riboswitches.
Stone MO, Kelley-Loughnane N (2013) Nucleic Acids Res 37:184–192
Chapter 7
Portable Two-Way Riboswitches: Design and Engineering

Ye Jin and Jian-Dong Huang
Abstract
Riboswitches are RNA-based regulatory devices that mediate ligand-dependent control of gene expression.
However, there has been limited success in rationally designing riboswitches. Moreover, most previous
riboswitches are confined to a particular gene and only perform one-way regulation. Here, we describe a
library screening strategy for efficient creation of ON riboswitches of lacI of Escherichia coli. An ON ribo-
switch of lacI is then integrated with the lac promoter, generating a hybrid device to achieve portable
sequential OFF-and-ON gene regulation.
Key words Riboswitch, LacI, LacP, Portable two-way regulation
1 Introduction
Inducible promoters are the major tools for specifically controlling

target gene expression. However, there are only limited numbers
of inducible promoters which, on some occasions, are insufficient
for independent control of multiple genes by addition of distinct
inducers. A promising alternative is riboswitch that is an RNA-based,
ligand-dependent genetic control element located in noncoding
regions of messenger RNAs. Unlike inducible promoters, these
RNA-based control elements are potentially numerous as a result
of high-throughput screen for ligand-specific aptamers (the sensor
domain of a riboswitch) [1–5].
Advancements have recently been made in the ability to
convert ligand-specific aptamers to functionally active riboswitches
by various random screening approaches [6–8] or rational design
[9–11]. However, library screening is labor intensive and time
consuming, and successful rational design is not guaranteed even
though a long list of design principles are followed due to incomplete
understanding of RNA-based gene regulation mechanisms. In
addition, a common disadvantage of riboswitches as well as induc-
ible promoters is that they only allow for one-way gene regulation.
93
94 Ye Jin and Jian-Dong Huang
Once activated, a target gene cannot be turned off unless promoter

inducers or riboswitch ligands are removed. The one-way regula-
tion mode makes the existing regulation systems unsuitable for
quick gene control in live animals or in vitro studies where intact
cell cultures are required. Thus, there is a great need for a two-way
regulation system, in which target gene can be switched off and on
in response to distinct effector molecules.
Here we solve the above questions by engineering an ON
theophylline-responsive riboswitch of the lacI gene and then inte-
grating it with the Lac-inducible promoter system (abbreviated as
LacP), in which the LacI repressor binds as a homotetramer to two
LacI-binding sites (LacIbs) positioned immediately downstream of
the lac promoter. First, we use a library screening strategy to create
ON theophylline-responsive riboswitches of the lacI gene that
encodes the LacI transcriptional regulator. Then, by combining an
ON riboswitch of LacI with two LacIbs, we establish a portable
hybrid device that acts as a versatile key for controlling gene expres-
sion without the need for further rational design or library
screening. Portability of this device has been verified previously
by applying it to rpoS (encoding RpoS, a master regulator of
acid resistance) and csrB (encoding a small noncoding RNA CsrB)
[12]. This portable device regulates target genes in a two-way
manner, switching off the targets in response to theophylline and
restoring the target expression in response to isopropyl β-D-1-
thiogalactopyranoside (IPTG).
2 Materials
All media and buffers are prepared using sterilized distilled deion-
ized water (ddH2O). All containers and pipette tips are sterilized
by heating (121 °C, 15 min) and stored at room temperature.
2.1 Incubation 1. Bacterial strain: E. coli strain MG1655 (F- lambda- ilvG- rfb-
of Bacteria 50 rph-1).
2. Growth medium: Luria-Bertani (LB) medium supplemented
with 2 mM magnesium sulfate. The antibiotics 50 μg/ml
ampicillin, 50 μg/ml kanamycin, and 12.5 μg/ml chloram-
phenicol are used for selection when appropriate.
3. Glycerol.
4. Temperature-controlled microplate shaker.
5. Temperature-controlled flask shaker.
6. Sterilized 96-well microtiter plate.
7. Spectrophotometer and cuvettes.
8. Flasks (500 ml).
9. Refrigerators at 4 °C and freezers at −20 and −80 °C.
Portable Two-Way Riboswitches 95
2.2 Recombineering 1. Ultrapure LB agar.

2. Ultrapure agarose (MB grade).
3. Micropulser electroporator.
4. 50-ml centrifuge tubes.
5. 1.5-ml microfuge tubes.
6. Controlled low-speed centrifuge (3,000 × g) at 4 °C.
7. Bench top centrifuge (max speed 16,000 × g).
8. Petri dishes (100 mm × 15 mm).
9. Pipettors of various volumes (10, 200, 1,000 μl).
10. Sterile micropulser electroporation cuvettes (0.1 cm gap)
(Bio-Rad) (see Note 1).
11. 42 °C water bath.
12. Ice buckets.
13. Plasmid isolation kit (Qiagen).
14. Gel extraction kit for DNA purification from agarose (Qiagen)
(see Note 2).
15. pCm: A plasmid carrying a self-driven chloramphenicol resis-
tance gene (cat) flanked by two loxP sites.
16. Materials for polymerase chain reaction (PCR): Standard Taq
polymerase, dNTP mixture, chimeric primers for PCR amplifi-
cation of recombineering substrates, PCR thermal cycler.
17. Agarose gel electrophoresis apparatus.
18. Materials required to express the Red functions: pSIM6 plasmid,
E. coli MG1655, ampicillin, chloramphenicol, distilled deion-
ized water at 4 °C.
2.3 X-Gal Screening 1. 20 mg/ml X-gal (5-bromo-4-chloro-3-indolyl-beta-D-galacto-

and Verification pyranoside) in DMSO, stored at −20 °C.
2. OFF plate: LB agar supplemented with 2 mM caffeine, 1 mM
magnesium sulfate, 60 μg/ml X-gal, and 12.5 μg/ml
chloramphenicol.
3. ON plate: LB agar supplemented with 2 mM theophylline,
1 mM magnesium sulfate, 60 μg/ml X-gal, and 12.5 μg/ml
chloramphenicol.
4. X-gal plate: LB agar supplemented with 60 μg/ml X-gal and
12.5 μg/ml chloramphenicol.
5. Beta-galactosidase activity kit (Thermo Scientific), stored at
−20 °C.
6. 100 mM Theophylline in ddH2O, stored at room temperature.
7. 100 mM Caffeine in ddH2O, stored at room temperature.
8. 1 M IPTG in ddH2O, stored at −20 °C.
2.4 Assessment 1. A computerized program for structural analysis of RNAs

of Riboswitches (RNAstructure) [13].
2. Reagents for acid resistance assay: Acidified LB medium
(pH 2.0), neutral LB.
3 Methods
To construct a riboswitch-LacP hybrid device, the first task is

engineering ligand-responsive riboswitches for the lacI gene which
encodes the LacI transcriptional regulator in the Lac system. To do
this, we establish a chromosomal riboswitch library, with the aid of
the recombineering technique, to screen for active riboswitches for
lacI. We use the theophylline-responsive aptamer as the sensor
domain of riboswitches. The aptamer is linked to a chlorampheni-
col resistance gene (cat) by ligation PCR, generating a riboswitch
cassette library containing a cat-aptamer-linker-random fragment
on the chromosome. The cat gene facilitates recombineering of
the cassette into the genome, and the 5-nt random sequence allows
construction of a riboswitch library. The cassette is integrated,
using the recombineering technique, into the genome of the E. coli
MG1655 strain right upstream of the ribosome-binding site (RBS)
of target genes, resulting in numerous mutant strains, each of
which carries a riboswitch candidate (functional or nonfunctional)
on the chromosome. In theory, the mutant library contains 45
riboswitch candidates. An effective ON riboswitch of lacI is
selected from the library by screening the color of colonies grown
on agar supplemented with X-gal and theophylline or caffeine
(an analogue of theophylline, serving as a negative control). Next,
the ON riboswitch of lacI is integrated with the LacP system,
generating a hybrid device that performs portable, two-way gene
control.
3.1 Construction 1. A cat fragment is PCR amplified using pCm as a template. The
of a Randomized resulting DNA product carries the cat gene to confer chloram-
Riboswitch Cassette phenicol resistance. Upstream of cat is a constitutive promoter
(Fig. 1) to drive expression of the chloramphenicol resistance gene.
The cat gene is followed by a 3′ untranslated region, which
serves as a linker between the cat gene and the downstream
theophylline-responsive aptamer (see Note 3).
2. The theophylline-responsive aptamer is synthesized by annealing
complementary primers. Sequence of the theophylline aptamer is
GAAATACCAGCATCGTCTTGATGCCCTTGGCAGTTC
(5′–3′). Design a forward primer GAAATACCAGCATCGT
CTTGATGCCCTTGGCAGTTC (5′–3′) and a reverse primer
GAACTGCCAAGGGCATCAAGACGATGCTGGTATTTC
(5′–3′) which is the reverse complement of the forward primers.
The two primers are denatured at 94 °C for 30 s and then annealed
Fig. 1 Construction of a theophylline-aptamer cassette. NNNNN denotes a 5-nt random sequence which is
added to the theophylline aptamer by PCR. The loxp-cat-loxp cassette is linked to the theophylline aptamer
followed by the 5-nt random sequence by fusion PCR. The resulting cassette is flanked with arms (45 nt)
homologous to the regions surrounding the ribosome-binding site of the lacI gene
at 55 °C for 30 s in a PCR thermal cycler, generating the double-

stranded aptamer DNA (see Note 4).
3. A 5-nt random sequence is then linked to 3′ of the theophyl-
line aptamer by PCR. To do this, a second reverse primer is
designed for the aptamer by adding five random nucleotides.
In companies like Techdragon, N denotes random nucleotides.
The aptamer DNA is PCR amplified using the above-described
forward primer and the second reverse primer, resulting in a
theophylline aptamer fragment followed by a 5-nt random
sequence (see Note 5).
4. The PCR-amplified cat cassette is linked to the aptamer with the
5-nt random sequence by ligation PCR. Primers are such designed
that the resulting PCR products are flanked with arms (45 nt)
homologous to the regions surrounding the RBS of the lacI
gene. Separate the PCR products with agarose electrophoresis,
and purify the products with gel extraction kit (see Note 2).
3.2 Construction 1. Chemically transform E. coli MG1655 with plasmid pSim6

of Randomized using conventional molecular techniques. pSim6 expresses the
Riboswitch Libraries λ recombination proteins at 42 °C.
for lacI 2. Incubate an isolated colony of MG1655 carrying pSim6 in
5 ml of LB plus 50 μg/ml ampicillin at 32 °C. Next morning,
dilute the overnight culture 1:500 into fresh LB medium and

incubate at 32 °C (see Note 6) with shaking at 220 rpm, until
OD600 reaches 0.4–0.6. This usually takes 2–2.5 h (see Note 7).
3. Incubate the bacteria for 15 min in a 42 °C water bath, with
shaking at 220 rpm. In this procedure, the λ recombination
proteins are induced from pSim6.
4. Immediately place the flask containing the bacteria on ice for
15 min (see Note 8).
5. Transfer the induced cultures to 50 ml centrifuge tubes and
centrifuge at 4 °C for 5 min at 3,000 × g. Promptly decant the
supernatant, and resuspend the bacterial pellet in 50 ml of
ddH2O (4 °C) (see Note 9).
6. Repeat the washing procedure two more times (see Note 10).
7. Resuspend the bacteria in ddH2O at 4 °C. For each 50 ml of
original cell suspension, use 200–300 μl of ddH2O. Keep the
washed bacteria (electrocompetent) on ice. The electrocompe-
tent bacteria should be used for electroporation within 30 min
(see Note 11).
8. Mix 1–5 μl of the purified PCR products (approximately
100 ng, salt-free) from Subheading 3.1, step 4, and the elec-
trocompetent cells in microcentrifuge tubes on ice (see Note 12).
9. Meanwhile, dry the electroporation cuvettes (see Note 13).
10. Transfer the mixtures to electroporation cuvettes (0.1-cm gap)
on ice, and perform electroporation at 1.8 kV (see Note 14).
11. Immediately add 1 ml of LB (room temperature) to the
cuvette, and transfer the bacterial suspension to a microcentri-
fuge tube. Incubate the bacteria at 32 °C with shaking at
220 rpm for 60 min (see Note 15).
3.3 Identification 1. Spin down the bacteria using a bench-top microcentrifuge at

of an ON Riboswitch 16,100 × g for 30 s. Spread the pellet on OFF plates. Incubate
of lacI (Fig. 2) the plates overnight at 32 °C.
2. Next morning, search for colonies that are bluer than others.
LacI is a repressor of lacZ, and therefore bluer color represents
the off status of lacI. As the random library contains 45
(=1,024) riboswitch candidates, it is recommended that at
least 1,000–2,000 colonies be checked for lacI expression
(see Note 16).
3. Individually streak the blue colonies on ON plates. Incubate the
plates overnight at 32 °C. As controls, the streaking is also con-
ducted on OFF plates for the same set of colonies (see Note 17).
4. Next morning, search for colonies that are whiter on theophyl-
line agar than on caffeine agar. The selected colonies carry ON
riboswitch that turns on lacI in response to theophylline.
Fig. 2 Creation of ON riboswitches for the lacI gene using a library screening-
based method. A random cassette mixture containing a theophylline-responsive
aptamer and five random bases is “recombineered” into the chromosome of
MG1655, generating a riboswitch library for lacI. Cells are plated on X-gal plates
without theophylline but with 2 mM caffeine. Relatively blue colonies are searched
for and transferred to X-gal plates with 2 mM theophylline. Those that turn into
white are considered to carry ON riboswitches of lacI. The blue arrow denotes
activating effects; the blunt red arrow denotes repressing effects. Reproduced
with permission from Oxford University Press [12] (color figure online)
5. Compare the effectiveness of the identified ON riboswitches of

lacI by beta-galactosidase assay in the presence of either
caffeine or theophylline, and select the riboswitch that displays
the widest regulation range.
6. Determine the sequence of the 5-nt random region of the
selected riboswitch, and predict the secondary structure of the
riboswitch using RNAstructure [13]. Choose the optimal
structure with the minimum free energy for the structure anal-
ysis. To predict riboswitch structures in the ligand-bound
form, utilize the “force pair” function of RNAstructure to
force the theophylline-binding pocket to form (Fig. 3).
3.4 Construction Although the above-described riboswitches fine-tune lacI in

of a Riboswitch-LacP response to theophylline, they have the following limitations. First,
Hybrid Device they are not applicable to genes other than lacI and therefore lack
portability. Second, they perform either ON or OFF function and
thus are “one-way” switches. We next demonstrate how to utilize
the ON riboswitch of lacI in combination with the LacI protein
and LacIbs to solve these problems. The transcriptional repressor
LacI inhibits lacZ transcription by binding to two LacIbs upstream
of lacZ. Inducible gene control built on binding of LacI to LacIbs
has been widely used for various non-lacZ genes in prokaryotic and
Fig. 3 Predicted structures of an ON riboswitch of lacI. In the non-ligand-bound state, there is a 9 bp stem
structure immediately upstream of the ribosome-binding site and is likely to block the gene translation. In the
theophylline-bound state, the long stem does not form and consequently permits the gene translation. The
random sequence is shown in red; the ribosome-binding site in blue; and the start codon in green. Reproduced
with permission from Oxford University Press [12] (color figure online)
eukaryotic cells [14–18], showing that the inducibility of the

LacI-responsive promoters are less likely to be affected by down-
stream targets than that of riboswitches. Thus, the ON-lacI-LacIbs
hybrid device should be capable of regulating any target gene when
integrated upstream of the RBS. Next we showcase the portability
and two-way control of this hybrid device by directly applying it to
the rpoS gene without adjusting the sequence of the device.
1. PCR amplify the ON-lacI-LacIbs cassette (approximately
2,500 bp) from the genome of the strain carrying the ON
riboswitch of lacI. The cassette is composed of the cat gene,
the ON-lacI riboswitch, the lacI gene, and the lacI–lacZ inter-
genic region containing the LacP with two LacIbs. The cat
gene facilitates integration of this cassette immediately
upstream of any target gene via recombineering (see Note 18).
2. Delete the lacI gene from MG1655 by recombineering,
generating a lacI null mutant. Design primers specific to the
loxP-cat-loxP cassette. Primers are preceded by 45-nt over-
hangs at the 5′ ends. The resulting PCR products encompass
the loxP-cat-loxP fragment flanked with homology (45 nt) to
the regions immediately flanking lacI. This fragment is inserted
into the genome of MG1655 harboring pSim6, replacing the
lacI gene, by recombineering. Recombinants are selected for
chloramphenicol resistance (encoded by the cat gene) and veri-
fied by the blue color of colonies on X-gal plates (see Note 19).
3. Insert the riboswitch cassette upstream of the RBS of the rpoS
gene (or any gene of interest) of the lacI null mutant by recom-
bineering, generating a new strain named ON-lacI-rpoS
(Fig. 4a). Verify the mutation by sequencing the 5′ region
upstream of rpoS.
Fig. 4 Gene control mediated by the hybrid riboswitch device. (a) Constitution of the riboswitch hybrid device.
The boxed is a portable riboswitch cassette ON-lacI-LacIbs (LacI-binding sites) integrated immediately
upstream of ribosome-binding site of the rpoS gene that is a master regulator of acid resistance (AR).
(b) Tunable and two-way control of rpoS via the ON-lacI-LacIbs hybrid device, as revealed by acid survival. The
strain ON-lacI-rpoS that carries the ON-lacI-LacIbs device upstream of the ribosome-binding site becomes
increasingly sensitive to acid with theophylline (0–1,600 μM). When increasing concentrations of IPTG are added
to cells grown in media with 1,600 μM theophylline, acid survival is increased and restored to wild-type levels
in the presence of 40 μM IPTG. Error bars, SD. Reproduced with permission from Oxford University Press [12]
3.5 Evaluation 1. RpoS is an alternative sigma factor of RNA polymerase, playing

of the Hybrid Device a critical role in survival of a diverse number of stresses such as
for Portable, Two-Way acid shock [19, 20]. Since rpoS is required for acid resistance,
Gene Control we assay survival of the strain ON-lacI-rpoS under acidic con-
ditions to test if the hybrid riboswitch device controls the rpoS
expression in response to the ligand theophylline.
2. Before the acid resistance assay, add 10 μl of the overnight
culture of the strain ON-lacI-rpoS grown into 2.5 ml of each
of the following fresh LB medium: (a) LB plus 2 mM caffeine,
(b) LB plus 2 mM theophylline, and (c) LB plus 2 mM
theophylline and 1 mM IPTG. Incubate the bacteria aerobi-
cally at 37 °C with shaking at 220 rpm.
3. To assay the acid resistance of bacteria, treat the overnight
cultures (in stationary phase) with acidic LB medium (pH 2.0)
for 2 h and then serially (1:10) dilute the bacteria in neutral LB
medium. Inoculate 20 μl of the dilutions on LB agar which has
been dried at 37 °C for 2 h to ensure that the drops are inocu-
lated to soak quickly into the agar. Meanwhile, serially dilute
and inoculate the bacteria that are not treated with acid. After
overnight culture, determine colony-forming units (CFU) of
acid-treated and untreated cells. Calculate acid survival (%)
with the following formula: Acid survival (%) = 100 × (CFU of
treated/CFU of untreated).
4. We have observed that the rpoS expression is ON in the absence
of theophylline and becomes OFF in the presence of theophyl-
line and that adding IPTG restores the rpoS expression of
bacteria grown in LB plus theophylline (Fig. 4b). These confirm
the portability and two-way control features of the ON-lacI-
LacIbs device (see Note 20).
5. Unlike rpoS, expression of many genes is not linked to a
quantifiable phenotype. In those cases, expression control by
the hybrid device can be evaluated by constructing lacZ trans-
lational fusions for target genes on the chromosome and then
measuring beta-galactosidase assay.
6. To construct a chromosomal lacZ translational fusion, insert
the loxP-cat-loxP selectable cassette immediately after the stop
codon of the lacZ gene on the MG1655 chromosome using
recombineering (as described above). Next, PCR amplify the
lacZ-loxP-cat-loxP cassette, and insert it (in frame) immedi-
ately prior to the stop codon of the target gene on the chromo-
some. The inserted lacZ fragment starts from the eighth codon
of lacZ gene and is co-transcribed and translated with the fused
genes. Quantify the expression of gene-lacZ fusions using a
beta-galactosidase assay kit (see Note 21).
4 Notes
1. Micropulser electroporation cuvettes (0.1 cm gap) (Bio-Rad)

can be reused although they are claimed to be disposable. After
use, rinse the cuvette surface and their gaps with ddH2O and
store in 75 % ethanol at room temperature.
2. DNA used for electroporation must be free of salt; otherwise,
the electric shock will be discharged very fast, resulting in a
small explosion. The explosion will kill most competent cells.
Therefore, the PCR products must be purified using the gel
extraction kit to remove the salts.
3. In our design, the riboswitch cassette is preceded by a self-driven
cat cassette. This cassette carries a strong constitutive promoter
to drive the expression of the downstream riboswitch. The cat
gene also makes it possible to select for clones with the ribo-
switch cassette on LB agar plus chloramphenicol. The cat gene
can be replaced by any other antibiotic resistance gene as long as
the drug resistance cassette carries a constitutive promoter.
4. In our design, we utilize the theophylline-responsive aptamer

as the sensing domain of riboswitches. The theophylline
aptamer can be replaced by any other ligand-specific aptamers.
5. The size of a riboswitch library is 4N where N is the number of
the randomized nucleotides fused to the ligand-specific aptamer.
6. All bacteria harboring pSim6 must be incubated at 32 °C or
lower temperature. Higher temperature will cause the loss of
pSim6.
7. If OD600 is higher than 0.6, the recombineering efficiency will
be reduced dramatically.
8. Keep bacteria at 0–4 °C during the entire process of prepara-
tion of electrocompetent bacteria for recombineering, as low
temperature is critical for recombineering efficiency.
9. In preparation of electrocompetent bacteria for recombineering,
mix bacterial cells by vigorously shaking the centrifuge tubes
after centrifugation. Add 30 ml of chilled ddH2O to the bacte-
rial pellets and shake until the pellets on the tube bottom
disappear. Then add chilled ddH2O to 50 ml. Do the shaking
quickly as the bacteria are exposed to room temperature during
this step. Immediately insert the entire tubes into ice after
shaking to keep the tubes cold.
10. In preparation of electrocompetent bacteria for recombineering,
bacteria must be washed three times with chilled ddH2O. Two-
time washing reduces recombineering efficiency significantly.
11. For recombineering, electrocompetent bacteria must be freshly
prepared. No successful cases of recombineering have ever been
reported with electrocompetent bacteria from frozen stock.
12. If more than 5 μl of DNA is used for recombineering, recom-
bineering efficiency will be reduced.
13. Water on cuvettes will cause small explosion in electroporation.
Before electroporation, dry by draining the cuvettes upside
down on a paper towel.
14. Bubbles in cuvette gaps will cause small explosion in electro-
poration. Avoid bubbles by gently transferring electrocompe-
tent bacteria to electroporation cuvettes and gently striking the
cuvettes on the bench if bubbles occur.
15. Add recovery LB medium (room temperature) to bacterial
cells immediately after electroporation. One-minute delay can
cause a threefold reduction in recombineering efficiency.
16. To ensure that each member of a riboswitch library is examined
for its regulatory capacity, no less than 4N bacterial clones
should be checked. N represents the number of the random-
ized nucleotides fused to the ligand-specific aptamer. We intro-
duce a 5-nt random sequence to construct a riboswitch library
for the lacI gene. However, the number of the randomized

nucleotides does not have to be five. If shorter random
sequence is employed, the library size is smaller; if longer
random sequence is used, more clones have to be screened due
to the larger size of the random library.
17. For library screening, incubate bacteria carrying the riboswitch
library on X-gal agar plate for 12 h. Longer incubation makes
it difficult to discriminate whiter colonies from bluer ones.
18. The LacP harbors a cAMP receptor protein (CRP)-binding
site, and therefore the downstream gene can be activated by the
cAMP–CRP complex. As cAMP–CRP is inactivated by glucose,
the ON-lacI-LacIbs cassette confers glucose (in addition to
theophylline) repression of gene expression.
19. The native lacI gene will mask the riboswitch regulation of the
second copy of lacI and, therefore, must be deleted.
20. The riboswitch hybrid device not only regulates gene expres-
sion in a two-way manner but also fine-tunes gene expression.
For example, with the strain ON-lacI-rpoS that carries the
hybrid device immediately upstream of rpoS, acid survival
(representing the expression levels of rpoS) decreases with
increasing concentrations of theophylline and increases with
increasing concentrations of IPTG.
21. The ON riboswitch of lacI regulates the lacI expression at the
translational level. However, the hybrid riboswitch device reg-
ulates its target expression at the transcriptional level as the
LacI is a transcriptional repressor. Thus, in addition to the
translational lacZ fusion described in Subheading 3, techniques
quantifying RNA levels such as real-time PCR and northern
blot can be used for evaluating the gene control by the hybrid
device.
References
1. Ellington AD, Szostak JW (1990) In vitro with SPR imaging. Nucleic Acids Res 34:
selection of RNA molecules that bind specific 6416–6424
ligands. Nature 346:818–822 6. Lynch SA, Desai SK, Sajja HK et al (2007)
2. Tuerk C, Gold L (1990) Systematic evolution A high-throughput screen for synthetic ribo-
of ligands by exponential enrichment: RNA switches reveals mechanistic insights into their
ligands to bacteriophage T4 DNA polymerase. function. Chem Biol 14:173–184
Science 249:505–510 7. Lynch SA, Gallivan JP (2009) A flow cytometry-
3. Chushak Y, Stone MO (2009) In silico selec- based screen for synthetic riboswitches. Nucleic
tion of RNA aptamers. Nucleic Acids Res Acids Res 37:184–192
37:e87 8. Muranaka N, Abe K, Yokobayashi Y (2009)
4. Collett JR, Cho EJ, Lee JF et al (2005) Mechanism-guided library design and dual
Functional RNA microarrays for high- genetic selection of synthetic OFF riboswitches.
throughput screening of antiprotein aptamers. Chembiochem 10:2375–2381
Anal Biochem 338:113–123 9. Suess B, Fink B, Berens C et al (2004) A the-
5. Li Y, Lee HJ, Corn RM (2006) Fabrication and ophylline responsive riboswitch based on helix
characterization of RNA aptamer microarrays slipping controls gene expression in vivo.
for the study of protein-aptamer interactions Nucleic Acids Res 32:1610–1614
10. Jin Y, Watt RM, Danchin A et al (2009) Use of a unit as a tool to study U6 function. RNA 4:
riboswitch-controlled conditional hypomorphic 231–238
mutation to uncover a role for the essential csrA 16. Grilly C, Stricker J, Pang WL et al (2007)
gene in bacterial autoaggregation. J Biol Chem A synthetic gene network for tuning protein
284:28738–28745 degradation in Saccharomyces cerevisiae. Mol
11. Win MN, Smolke CD (2007) A modular and Syst Biol 3:127
extensible RNA-based gene-regulatory platform 17. Cronin CA, Gluba W, Scrable H (2001) The
for engineering cellular function. Proc Natl lac operator-repressor system is functional in
Acad Sci U S A 104:14283–14288 the mouse. Genes Dev 15:1506–1517
12. Jin Y, Huang JD (2011) Engineering a porta- 18. Ellis T, Wang X, Collins JJ (2009) Diversity-
ble riboswitch-LacP hybrid device for two-way based, model-guided construction of synthetic
gene regulation. Nucleic Acids Res 39:e131 gene networks with predicted functions. Nat
13. Reuter JS, Mathews DH (2010) RNAstructure: Biotechnol 27:465–471
software for RNA secondary structure predic- 19. Small P, Blankenhorn D, Welty D et al (1994)
tion and analysis. BMC Bioinformatics 11:129 Acid and base resistance in Escherichia coli and
14. Studier FW, Moffatt BA (1986) Use of bacte- Shigella flexneri: role of rpoS and growth pH.
riophage T7 RNA polymerase to direct selective J Bacteriol 176:1729–1737
high-level expression of cloned genes. J Mol 20. Fang FC, Libby SJ, Buchmeier NA et al (1992)
Biol 189:113–130 The alternative sigma factor katF (rpoS) regu-
15. Luukkonen BG, Seraphin B (1998) Construction lates Salmonella virulence. Proc Natl Acad Sci
of an in vivo-regulated U6 snRNA transcription U S A 89:11978–11982
Chapter 8
Generation of Orthogonally Selective Bacterial

Riboswitches by Targeted Mutagenesis
and In Vivo Screening
Helen A. Vincent, Christopher J. Robinson, Ming-Cheng Wu,
Neil Dixon, and Jason Micklefield
Abstract
Riboswitches are naturally occurring RNA-based genetic switches that control gene expression in
response to the binding of small-molecule ligands, typically through modulation of transcription or
translation. Their simple mechanism of action and the expanding diversity of riboswitch classes make
them attractive targets for the development of novel gene expression tools. The essential first step in
realizing this potential is to generate artificial riboswitches that respond to nonnatural, synthetic ligands,
thereby avoiding disruption of normal cellular function. Here we describe a strategy for engineering
orthogonally selective riboswitches based on natural switches. The approach begins with saturation
mutagenesis of the ligand-binding pocket of a naturally occurring riboswitch to generate a library of
riboswitch mutants. These mutants are then screened in vivo against a synthetic compound library to
identify functional riboswitch–ligand combinations. Promising riboswitch–ligand pairs are then further
characterized both in vivo and in vitro. Using this method, a series of artificial riboswitches can be gen-
erated that are versatile synthetic biology tools for use in protein production, gene functional analysis,
metabolic engineering, and other biotechnological applications.
Key words Orthogonal riboswitches, Artificial riboswitches, Aptamer domain, Synthetic ligands,
Gene expression tools, In vivo screening, In vitro transcription, Isothermal titration calorimetry,
Synthetic biology
1 Introduction
Riboswitches are modular noncoding regions of mRNAs that func-

tion as genetic switches to turn genes either on or off (see refs. 1, 2
for recent reviews). They consist of an aptamer domain, which
binds a small-molecule ligand with both high affinity and selectivity,
and a regulatory expression platform. Ligand binding to the
aptamer domain induces a conformational change in the down-
stream expression platform, thereby modulating gene expression.
107
108 Helen A. Vincent et al.
The majority of riboswitches are found in the 5′ untranslated region

(UTR) of bacterial mRNAs, although thiamine pyrophosphate
(TPP)-dependent riboswitches have also been found in archaea,
fungi, and plants [3]. Over 20 classes of bacterial riboswitch have
been identified to-date, which collectively respond to a diverse array
of ligands including complex coenzymes (e.g., TPP, adenosylco-
balamin, flavin mononucleotide, S-adenosylmethionine); purines
and their derivatives (e.g., adenine, guanine, 2′-deoxyguanosine,
adenosine triphosphate, 7-aminomethyl-7-deazaguanine (preQ1));
amino acids (glycine, glutamine, lysine); sugars (glucosamine-
6-phosphate); and simple inorganic ions (Mg2+, F−) [1, 2]. Typically,
these ligands regulate genes required for their own biosynthesis,
catabolism, and/or transport with the mechanism of gene regula-
tion utilized being largely dependent upon the organism from
which the riboswitch originates. Riboswitches in Gram-negative
bacteria (e.g., Escherichia coli) typically operate at the translational
level, whereas transcriptional mechanisms dominate in Gram-
positive bacteria (e.g., Bacillus subtilis), and in eukaryotes alterna-
tive splicing is affected [2].
The relative simplicity of their protein-free mechanism of
action, with sensor and regulatory domains contained within a
single RNA molecule, makes riboswitches an attractive platform
for the development of new gene expression tools. In fact, with
their diversity in both ligand recognition and regulatory mecha-
nisms, natural riboswitches appear to represent a ready-made
gene expression toolkit that could be applied to the fields of
protein production, gene functional analysis, metabolic engi-
neering, and synthetic biology. However, since the majority of
riboswitches respond to essential metabolites (e.g., amino acids,
coenzymes, purines, and sugars) their use may be limited by
regulation of intracellular metabolite concentrations, and the
exogenous addition of these compounds is likely to negatively
impact normal cellular function. Therefore, there is a clear need
to develop artificial riboswitches that can regulate gene expres-
sion in response to synthetic nonnatural ligands, which do not
affect normal cellular function.
Like natural riboswitches, artificial riboswitches must bind
their ligands with high affinity and selectivity; ligand binding
should lead to a conformational change that results in modulation
of gene expression, and this regulation should be fast and dose-
dependent with a dynamic range at least equivalent to that exhib-
ited by natural switches. A number of strategies have now been
validated for the production of artificial riboswitches. The most
common approach is to take RNA aptamers, generated in vitro
(e.g., through SELEX protocols [4]), and connect them to an
expression platform, either by rational design [5–7] or through
selection/screening for in vivo functionality [5–9]. However,
Orthogonally Selective Bacterial Riboswitches 109
recently our laboratory has pioneered a novel strategy for the pro-
duction of artificial orthogonal riboswitches, which involves
reengineering the aptamer domain of natural riboswitches so that
they respond to synthetic ligands rather than the native metabo-
lites [10]. Applications such as the production of designer auxo-
trophs [11], bioremediation [12], development of gene expression
tools for pathogenic organisms [13, 14], and simultaneous control
of multiple genes within the same bacterial cell [15] are beginning
to be explored.
The major advantage of in vitro generation of aptamer
domains is that SELEX can, in theory, be used to produce
aptamers with high affinity and selectivity against any ligand of
choice. However, in practice, only a limited number of aptam-
ers have been applied to the development of riboswitch tech-
nologies in bacteria: the theophylline aptamer [5–9, 11, 13,
14, 16, 17] and, more recently, an atrazine aptamer [12]. This
is largely because the de novo development of aptamers in vitro
can be both costly and time consuming and, more critically,
does not guarantee in vivo functionality [18]. Functionality is
fundamentally dependent on the bioavailability of the ligand.
Even the validated synthetic ligands theophylline and atrazine
exhibit poor cell permeability, such that far higher concentra-
tions of ligand are required to elicit a response in vivo than
would be predicted from binding affinities determined in vitro
[11, 12, 18]. Also, when challenged with the intracellular
metabolome, the selectivity of the aptamer domain for the syn-
thetic ligand is often weaker than observed in vitro [19].
Finally, in vivo functionality is strongly dependent upon how
the aptamer and expression platform are linked, and this con-
nection is not easy to rationally design. Consequently, the
majority of functional in vitro-engineered artificial riboswitches
have only been isolated following further extensive in vivo
genetic selections or screens [5–9, 12].
To address the limited diversity of functional synthetic aptam-
ers, our laboratory has developed a novel strategy for reengineering
the aptamer domains of natural riboswitches to produce artificial
riboswitches that respond to diverse synthetic ligands [10]. We have
demonstrated the application of this approach with the adenine-
responsive add translational-ON switch from Vibrio vulnificus, pro-
ducing an orthogonally selective artificial riboswitch for inducible
gene expression in E. coli [10]. In this chapter, we present the pro-
tocol that we use to change the selectivity of translational ribo-
switches for use as gene expression tools in bacteria [10]. An
overview of this protocol is presented in Fig. 1. In principle, this
strategy is applicable to any riboswitch, for use in any bacterial
organism. However, for simplicity, the method presented here
focuses on E. coli, a common model organism in the laboratory.
A/C/G/U A/C/G/U
R CAT
A/C/G/U A/C/G/U
2. Site-saturation mutagenesis of
ligand binding pocket to generate
library of riboswitch mutants
- Chloramphenicol + Chloramphenicol
H NH2
S N N O
N
N
HN N
H N N O2N
2
O 4. Screen riboswitch mutants against
O H O
compound library in vivo
NH2
H2N N O HO N O
N N
N N N
O N NH
1. Select natural riboswitch H 2
O O O
for reengineering
3. Build synthetic compound
library
Synthetic ligand
R eGFP
Gene Expression
5. In vivo characterization of
functional mutant riboswitch-
6. In vitro characterization of synthetic ligand pairs (dose-
functional mutant riboswitch- dependence and Natural ligand
synthetic ligand pairs orthogonality)
(binding and structure) [Ligand]
Fig. 1 Schematic overview of the strategy to engineer orthogonally selective artificial riboswitches. Individual
steps are explained in detail in the subsequent protocol
In order to overcome the challenge of linking the aptamer

domain and expression platform in a functional conformation, our
approach starts with an entire natural riboswitch (the aptamer
domain and expression platform) for which in vivo functionality has
been validated and for which an X-ray crystal structure of the ligand-
bound aptamer domain is available. Based on the crystal structure,
the ligand-binding pocket is mutated so that it no longer accepts the
natural ligand. These riboswitch mutants are fused upstream of a
chloramphenicol resistance gene (CAT ) and then screened in the
presence of chloramphenicol against a library of suitable nonnatural
compounds for ligand-dependent growth (ON switches) or growth
inhibition (OFF switches). By starting with an in vivo functionality
screen, only those mutant riboswitch–ligand pairs with the potential
to be suitable for biological applications are carried forward. Selected
candidates are then cloned upstream of an enhanced green fluores-
cent protein (eGFP) gene to allow for characterization of ligand
selectivity/orthogonality and dose-dependent response in vivo.
Finally, in vitro characterization of ligand binding by biophysical
methods (e.g., isothermal titration calorimetry (ITC) and X-ray
crystallography) is carried out to confirm orthogonality and to
inform subsequent rounds of riboswitch–ligand design.
2 Materials
2.1 Construction 1. Standard reagents and equipment for molecular cloning.

of Mutant Riboswitch 2. A high-copy-number vector carrying an ampicillin resistance
Chloramphenicol- gene, e.g., pBluescript II KS(+) (Stratagene) (see Note 1).
Based Reporter
3. DNA fragments containing (1) the lac promoter/operator
Plasmids
(see Note 2), (2) the natural riboswitch, and (3) the CAT gene
(see Note 3).
4. QuikChange Site-directed Mutagenesis Kit (Stratagene).
5. Mutagenic primers.
2.2 Building a 1. Access to CAS SciFinder (American Chemical Society)

Screening Library (see Note 4).
of Nonnatural 2. Commercial databases, e.g., Aurora Fine Chemicals, TimTec,
Compounds Enamine, Bionet.
3. Chemical database management software, e.g., ChemFinder
(CambridgeSoft) (see Note 4).
4. Dimethyl sulfoxide (DMSO).
2.3 Screening of 1. 37 °C incubator for plates.

Mutant Riboswitches 2. 37 °C shaker incubator for tubes, flasks, and 96-well plates.
Against a Library
3. Anthos Zenyth 3100 plate reader (or a suitable alternative
of Nonnatural
capable of measuring absorbance at 620 nm).
Compounds
4. Sterile 90 mm Petri dishes.
5. Sterile 20 ml culture tubes.
6. Sterile 250 ml flasks.
7. Sterile non-treated polystyrene 96-well microplates with
400 μl wells with flat, clear bottoms (Nunc).
8. LB-agar: 1 % (w/v) bactotryptone, 0.5 % (w/v) yeast extract,
0.5 % (w/v) NaCl, and 2.5 % (w/v) agar in dH2O, sterilized
by autoclaving.
9. LB media: 1 % (w/v) tryptone, 0.5 % (w/v) yeast extract,
0.5 % (w/v) NaCl in dH2O, sterilized by autoclaving.
10. 5× M9 salts (1 l): 64 g Na2HPO4·7H2O, 15 g KH2PO4, 2.5 g
NaCl, and 5 g NH4Cl in dH2O, sterilized by autoclaving.
11. M9 minimal media (1 l): 1× M9 salts, 2 mM MgSO4, 0.1 mM
CaCl2, 0.4 % (w/v) glucose, 2 mg/ml casamino acids, and
0.1 mg/ml thiamine prepared in sterile dH2O.
12. 100 mg/ml ampicillin stock prepared in dH2O, filter sterilized,
and stored at −20 °C.
13. 34 mg/ml chloramphenicol stock prepared in 70 % ethanol,
filter sterilized, and stored at −20 °C.
14. 1 M isopropyl β-D-THIOGALACTOPYRANOSIDE (IPTG)

stock prepared in dH2O, filter sterilized, and stored at −20 °C.
15. E. coli TOP10F′ chemically competent cells (Life Technologies)
(see Note 5).
16. The parental and mutant riboswitch chloramphenicol reporter
plasmids (see Subheading 3.1).
17. DMSO.
18. 5 mM natural riboswitch ligand stock prepared in DMSO and
stored at −20 °C.
19. Screening library of nonnatural/synthetic compounds (see
Subheading 3.2).
2.4 In Vivo 1. Standard reagents and equipment for molecular cloning.

Characterization of 2. 37 °C incubator for plates.
Functional Mutant
3. 37 °C shaker incubator for tubes, flasks, and 96-well microplates.
Riboswitch–Synthetic
Ligand Pairs 4. Anthos Zenyth 3100 plate reader (or a suitable alternative
capable of measuring absorbance at 620 nm and fluorescence
with excitation at 485 nm and emission at 535 nm).
5. Sterile 90 mm Petri dishes.
6. Sterile 20 ml culture tubes.
7. Sterile non-treated polystyrene 96-well microplates with
400 μl wells with black sides and flat, clear bottoms (Greiner
Bio-One).
8. LB-agar (see Subheading 2.3, item 8).
9. LB media (see Subheading 2.3, item 9).
10. 5× M9 salts (see Subheading 2.3, item 10).
11. M9 minimal media (see Subheading 2.3, item 11).
12. 100 mg/ml ampicillin stock (see Subheading 2.3, item 12).
13. 34 mg/ml chloramphenicol stock (see Subheading 2.3, item 13).
14. 1 M IPTG stock (see Subheading 2.3, item 14).
15. A high-copy-number vector carrying an ampicillin resistance
gene, e.g., pBluescript II KS(+) (Stratagene).
16. DNA fragments containing (1) the lac promoter/operator (see
Note 2), (2) the natural riboswitch, and (3) the eGFP gene.
17. QuikChange Site-directed Mutagenesis Kit (Stratagene).
18. Mutagenic primers.
19. E. coli TOP10F′ chemically competent cells (Life Technologies)
(see Note 5).
20. DMSO.
21. 5 mM natural riboswitch ligand stock prepared in DMSO.

22. 5 mM synthetic ligand stocks prepared in DMSO from the
screening compound library (see Subheading 3.2).
2.5 Preparation of 1. PCR thermocycler.

DNA Template to 2. MaXtract high-density tubes, 1.5 ml (Qiagen).
Produce Aptamer RNA
3. 3 M sodium acetate (pH 5.2).
for In Vitro Biophysical
Characterization 4. 70 and 100 % ethanol solutions, chilled to −20 °C.
5. Phenol:chloroform:isoamyl alcohol mixture (25:24:1).
6. Chloroform.
7. Nuclease-free H2O (Ambion), not diethylpyrocarbonate
(DEPC) treated (see Note 6).
8. Phusion High-Fidelity DNA Polymerase, including supplied
5× Phusion HF buffer (New England BioLabs).
9. Deoxyribonucleoside 5′ triphosphate (dNTP) mix: 2 mM
each of dATP, dCTP, dGTP, and dTTP.
10. Suitable 5′ and 3′ DNA oligonucleotide primers, 100 μM each
in dH2O.
11. Single-stranded sense DNA oligonucleotide template, 100 μM
in dH2O.
2.6 Preparation of 1. 37 °C incubator.

Aptamer RNA by In 2. 12 ml Slide-A-Lyzer dialysis cassettes, 3,000 MWCO (Thermo
Vitro Transcription for Scientific).
In Vitro Biophysical
3. Disposable syringes and 18-guage needles.
Characterization
4. Vivaspin 6 centrifugal concentrators, 3,000 MWCO
(Sartorius).
5. Nuclease-free H2O (Ambion), not DEPC treated (see Note 6).
6. DEPC-treated dH2O.
7. 10× transcription buffer: 300 mM Tris-HCl (pH 8.0),
100 mM dithiothreitol (DTT), 34 mM spermidine, 0.1 %
Triton X-100 prepared in nuclease-free H2O and stored at
−20 °C.
8. 1 M MgCl2 prepared in nuclease-free H2O.
9. 1 M DTT prepared in nuclease-free H2O and stored at −20 °C.
10. Nucleoside 5′ triphosphate (NTP) mix: 25 mM each of ATP,
CTP, GTP, and UTP prepared in nuclease-free H2O and
stored at −20 °C.
11. 0.5 M EDTA (pH 8.0) prepared in nuclease-free H2O.
12. Riboswitch buffer: 100 mM KCl, 50 mM HEPES (pH 7.5),
20 mM MgCl2 in DEPC-treated H2O and stored at 4 °C.
13. DNase I, RNase-free (New England BioLabs).
14. Bacteriophage T7 RNA polymerase purified in-house or pur-

chased from Ambion (see Note 7).
15. Purified DNA template prepared in nuclease-free H2O (see
Subheading 3.5).
2.7 Purification of 1. ÄKTA FPLC system (GE Healthcare) at 4 °C.

Aptamer RNA for In 2. HiLoad 26/600 Superdex 200 PG size-exclusion column
Vitro Biophysical (GE Healthcare) (see Note 8).
Characterization
3. RNase-free disposable tubes for FPLC fraction collector.
4. Disposable syringes and 18-guage needles.
5. 3 ml Slide-A-Lyzer dialysis cassettes, 3,000 MWCO (Thermo
Scientific).
6. Vivaspin 6 centrifugal concentrators, 3,000 MWCO
(Sartorius).
7. 0.5 ml Amicon Ultra centrifugal concentrators, 3,000 MWCO
(Millipore).
8. 0.2 μm bottle-top filter units (Nalgene).
9. 0.2 μm syringe filter units (Millipore).
10. DEPC-treated dH2O.
11. Riboswitch buffer: 100 mM KCl, 50 mM HEPES (pH 7.5),
20 mM MgCl2 in DEPC-treated dH2O.
12. Crystallography buffer: 100 mM KCl, 50 mM HEPES
(pH 7.5), 5 mM MgCl2 in nuclease-free dH2O (0.2 μm fil-
tered and vacuum degassed).
13. 2.5 ml in vitro transcription reaction (see Subheading 3.6).
2.8 In Vitro 1. A VP-ITC (GE Healthcare) or other appropriate

Biophysical microcalorimeter.
Characterization 2. Origin software (OriginLab) or similar software capable of
of Aptamer–Ligand nonlinear curve fitting, including “Origin for ITC” custom
Interactions: software (GE Healthcare) for use with a VP-ITC
Isothermal Titration microcalorimeter.
Calorimetry 3. Thermovac degassing unit (MicroCal).
4. 2.5 ml long needle Hamilton filling-syringe.
5. Disposable syringe fitted with tubing.
6. KimWipes.
7. DEPC-treated H2O.
8. 12 % NaOH solution.
9. 5 % RNaseZap (Invitrogen) solution.
10. Dialysis buffer: 100 mM KCl, 50 mM HEPES (pH 7.5),

20 mM MgCl2 in DEPC-treated dH2O (see Subheading 3.7,
step 8).
11. 10 μM aptamer RNA (see Subheadings 3.5–3.7) in dialysis
buffer.
12. 100 μM ligand in dialysis buffer.
3 Methods
The following protocols assume that the user has a good working
knowledge of standard molecular cloning techniques. All in vitro
procedures (see Subheading 3.5 onwards) should be carried out in
an RNase-free environment (see Note 9), including the use of ster-
ile reagents and consumables. Gloves should be changed frequently
to avoid contamination from ribonucleases found on the hands
and in the environment. RNA samples should be kept on ice wher-
ever possible.
3.1 Construction of The X-ray crystal structures of aptamer domains in complex with
Mutant Riboswitch their cognate ligand have been determined for most classes of ribo-
Chloramphenicol- switch. In general, aptamer domains form tight binding pockets
Based Reporter that completely surround the ligand. With the notable exception
Plasmids of adenosylcobalamin riboswitches [20, 21], the majority of ribo-
switches appear to selectively bind the ligand through specific
hydrogen bonding interactions mediated by conserved nucleotides
within the ligand-binding pocket, with the binding of planar
ligands further stabilized through base-stacking interactions [2,
22]. When changing the specificity of natural riboswitch aptamers,
a good initial strategy is to specifically target the conserved nucleo-
tides responsible for forming the aptamer–ligand hydrogen bonds.
1. Select a suitable riboswitch for reengineering (see Note 10).
2. Identify key ligand-binding nucleotides from the X-ray crystal
structure of the aptamer–ligand complex, and select nucleo-
tides for mutagenesis (see Note 11).
3. Using established methods, construct a parental chloram-
phenicol reporter plasmid (Fig. 2a) by cloning the following
elements, in order, into a suitable high-copy-number vector
(see Note 1).
(a) The IPTG-inducible lac promoter and operator derived
from the 5′ UTR of the lacZ gene from E. coli (see Note 2).
(b) The natural riboswitch aptamer domain and expression
platform up to the native translational start site.
(c) The open reading frame of the CAT gene, which confers
chloramphenicol resistance (see Note 3).
a Chloramphenicol b Enhanced green

Riboswitch
Riboswitch resistance gene (CAT ) fluorescent protein
gene (eGFP)
Lac promoter/ Lac promoter/

operator operator
Chloramphenicol GFP
reporter plasmid reporter plasmid
Ampicillin Ampicillin
resistance gene resistance gene
Fig. 2 Schematic representation of the reporter plasmids. (a) Chloramphenicol reporter plasmid for in vivo
screening. (b) GFP reporter plasmid for quantitative in vivo characterization
The final cassette should be sequenced to ensure that the

correct construct has been successfully assembled.
4. Using the parental chloramphenicol reporter plasmid as a
template and the QuikChange Site-directed Mutagenesis Kit
(Stratagene) with appropriate mutagenic primers perform site-
directed mutagenesis to mutate each selected ligand-binding
nucleotide to all other nucleotide possibilities (see Note 12).
Final clones should be sequenced to ensure that the desired
mutations have been introduced.
3.2 Building a The choice of the screening library size and composition directly
Screening Library affects the probability of successfully identifying functional mutant
of Nonnatural riboswitch–synthetic ligand combinations. In practice, we have found
Compounds that a pool of 15 riboswitch mutants (based upon saturation muta-
genesis of two nucleotides in the ligand-binding pocket), screened
against a carefully selected library of no more than 100 compounds
(1,500 different riboswitch–ligand scenarios in total), should be suf-
ficient to find functional mutant–ligand pairs. When designing the
library it is important to consider that in order to be fit for purpose,
the compounds should have suitable physicochemical properties
(e.g., stable, at least modestly soluble in water, cell permeable) and
they should be accessible in terms of availability and cost.
1. Using the chemical structure of the natural ligand as the query,
use structural chemical research tools (e.g., CAS SciFinder)
and search commercial databases (e.g., Aurora Fine Chemicals,
TimTec, Enamine, Bionet) using chemical database software
(e.g., ChemFinder) to generate a list of analogues.
2. Filter out any compounds that are natural metabolites, since
these compounds would be expected to encounter similar
problems with regard to disrupting normal cellular function as

the natural ligand.
3. Filter out any compounds with a molecular weight of more
than twice that of the natural ligand since it is unlikely that the
ligand-binding pocket will be able to accommodate com-
pounds that vary dramatically in size from the natural ligand.
4. Filter out any compounds that do not contain hydrogen bond
acceptor or donor groups since hydrogen bonds are crucial for
the specific recognition of ligands by riboswitches (see Note 13).
5. Check for commercial availability of the refined library using
online chemical databases (e.g., ChemBuyersGuide,
ChemExper, eMolecules), and filter out any that are not gen-
erally kept in stock (see Note 14).
6. Decide upon a price threshold, and filter out compounds
above this threshold (see Note 15).
7. Purchase the remaining compounds.
8. Prepare 5 mM stocks of each compound in DMSO and store
at −20 °C.
3.3 Screening of To identify functional mutant riboswitch–ligand pairs, the parental

Mutant Riboswitches and mutant riboswitches are screened against each of the com-
Against a Library of pounds in the library, for compound-dependent changes in CAT
Nonnatural expression, by simply assaying growth in the presence and absence
Compounds of chloramphenicol. Functional mutant riboswitch–synthetic
ligand pairs should grow in the presence and absence of chloram-
phenicol (ON switches) or only in the absence of chloramphenicol
(OFF switches). Furthermore, to satisfy the requirement for
orthogonality, these responses should only be observed for the
synthetic ligand and not for the natural ligand (e.g., see Fig. 3).
The following screen should be carried out in duplicate to reduce
the possibility of selecting false positives.
1. Using established methods, transform the parental and mutant
riboswitch reporter plasmids into E. coli TOP10F′ (see Note
5) cells.
2. Streak each strain to single colonies on LB-agar, supplemented
with 100 μg/ml ampicillin, and incubate overnight at 37 °C.
3. Inoculate 5 ml of LB, supplemented with 100 μg/ml ampicil-
lin, with individual colonies from each strain and incubate at
37 °C overnight with shaking.
4. Dilute 2.5 ml of the overnight cultures 20-fold into 50 ml M9
minimal media, supplemented with 100 μg/ml ampicillin and
1 mM IPTG (to induce transcription of the chloramphenicol
reporter cassette), and incubate for 1 h at 37 °C with
shaking.
a ‘ON’ Switch b ‘OFF’ Switch
1 2 3 4 1 2 3 4
- Chloramphenicol
5 6 7 8 5 6 7 8
9 10 11 12 9 10 11 12
Well 1 = DMSO (no ligand control)

Well 2 = Natural ligand (orthogonality
control)
Well 3-Well N = Synthetic ligands
1 2 3 4 1 2 3 4
+ Chloramphenicol
5 6 7 8 5 6 7 8
9 10 11 12 9 10 11 12
Fig. 3 Schematic representation of the expected results of chloramphenicol screening. Shaded circles indicate
bacterial growth, and white circles indicate growth inhibition. (a) “ON” riboswitches. Wells 1 and 2 represent
the no ligand and natural ligand controls and should grow only in the absence of chloramphenicol. Wells 3–N
represent different synthetic ligands. Functional riboswitch–ligand pairs are identified by growth in both the
absence and presence of chloramphenicol, e.g., wells 3 and 6. (b) “OFF” riboswitches. Wells 1 and 2 represent
the no ligand and natural ligand controls and should grow in both the absence and presence of chlorampheni-
col. Wells 3–N represent different synthetic ligands. Functional riboswitch–ligand pairs are identified by growth
only in the absence of chloramphenicol, e.g., wells 5 and 11 (color figure online)
5. Divide this culture into two 25 ml aliquots, and add chlorampheni-

col to a final concentration of 170 μg/ml to one of the aliquots.
6. For each strain, prepare two identical compound library
screens in 96-well microplates containing:
(a) Two wells with 20 μl of DMSO (10 % final concentra-
tion) control.
(b) One well with 20 μl of 5 mM natural ligand (500 μM
final ligand concentration and 10 % final DMSO concen-
tration) control.
(c) N wells with 20 μl of different 5 mM ligand stocks
(500 μM final ligand concentration and 10 % final DMSO
concentration) from the screening compound library (see
Subheading 3.2).
7. To one of the DMSO wells in each screen add 180 μl of M9
minimal media as a blank (no ligand control).
8. To all other wells in the first screen add 180 μl of the induced
culture without chloramphenicol.
9. To all other wells in the second screen add 180 μl of the
induced culture supplemented with chloramphenicol.
10. Incubate at 37 °C for 4 h with shaking.
11. Measure the absorbance at 620 nm of each well using a plate
reader, and subtract the reading from the media-only well
from all other wells to assess the extent of bacterial growth.
3.4 In Vivo The unambiguous phenotype (bacterial growth) of the chloram-

Characterization of phenicol reporter is ideal for a qualitative initial screen or selection.
Functional Mutant However, it is less suitable for quantitatively characterizing the
Riboswitch–Synthetic dynamic range and dose-dependent response of riboswitch–ligand
Ligand Pairs pairs. Therefore, we use an eGFP-based screen to further charac-
terize candidates that are identified in the original chloramphenicol-
based screen (see Note 16). The following procedure should be
carried out in triplicate.
1. Using established methods, construct a parental eGFP reporter
plasmid (Fig. 2b) by cloning the following elements in order
into a suitable high-copy-number vector:
(a) The IPTG-inducible lac promoter and operator derived
from the 5′ UTR of the lacZ gene from E. coli (see Note 2).
(b) The natural riboswitch aptamer domain and expression
platform up to the native translational start site.
(c) The open reading frame of the eGFP gene.
The final cassette should be sequenced to ensure that the
correct construct has been successfully assembled.
2. Using the parental eGFP reporter plasmid as a template and
the QuikChange Site-directed Mutagenesis Kit (Stratagene)
with appropriate mutagenic primers perform site-directed
mutagenesis to introduce the functional mutations identified
in the initial chloramphenicol screen. Final clones should be
sequenced to ensure that the desired mutations have been
introduced.
3. Using established methods, transform the parental and mutant ribo-
switch reporter plasmids into E. coli TOP10F′ (see Note 5) cells.
4. Streak each strain to single colonies on LB-agar, supplemented
with 100 μg/ml ampicillin, and incubate overnight at 37 °C.
5. Inoculate 2.5 ml of LB, supplemented with 100 μg/ml ampi-
cillin, with individual colonies from each strain and incubate at
37 °C overnight with shaking.
Ligand concentrations (µM)
0 0.5 1 5 10 25 50 125 250 500
No cells
Natural Uninduced
cells
Ligand
Induced
cells
No cells
Synthetic Uninduced
Ligand cells
Induced
cells
Fig. 4 Example plate layout for quantitative in vivo characterization. Testing different ligand concentrations
allows the dose-dependent response to be assessed. Comparison of the natural ligand to synthetic ligands
confirms the selectivity of the synthetic ligand
6. Dilute 250 μl of the overnight cultures 20-fold into 5 ml of M9

minimal media, supplemented with 100 μg/ml ampicillin, with or
without 1 mM IPTG (required to induce expression of the eGFP
reporter cassette), and incubate for 1 h at 37 °C with shaking.
7. Dilute each ligand stock with DMSO to prepare working stocks
of, e.g., 0.005, 0.01, 0.05, 0.1, 0.25, 0.5, 1.25, 2.5, and 5 mM
(yielding final concentrations of 10 % DMSO and 0.5, 1, 5, 10,
25, 50, 125, 250, and 500 μM ligand) (see Note 17).
8. For each strain, prepare a ligand concentration screen in a
96-well microplate (see Fig. 4):
(a) Three series of ten concentrations of natural ligand (20 μl
of DMSO (0 μM ligand) and each of the working stocks).
(b) Three series of ten concentrations of synthetic ligand (20 μl
of DMSO (0 μM ligand) and each of the working stocks).
9. To one of each series add:
(a) 180 μl of M9 minimal media (to allow calculation of
background absorbance/fluorescence from the ligand).
(b) 180 μl of uninduced (without IPTG) cells (to allow cal-
culation of background fluorescence from the cells).
(c) 180 μl of induced (with IPTG) cells (the experiment).
10. Grow cells at 37 °C with shaking in an Anthos Zenyth 3100
plate reader for 5 h measuring the absorbance at 620 nm and
eGFP fluorescence (excitation 485 nm, emission 535 nm) at
6-min intervals (50 readings).
11. Calculate normalized fluorescence units as follows:

(a) Subtract the media-only absorbance/fluorescence read-
ings from the corresponding uninduced and induced cell
absorbance/fluorescence readings.
(b) Divide the adjusted fluorescence values by the adjusted
absorbance values to normalize for cell density.
(c) Subtract the normalized uninduced fluorescence from the
normalized induced fluorescence to account for basal
expression and background fluorescence.
12. Calculate induction/repression factors by dividing the final
normalized fluorescence values by the value for 0 μM ligand
(DMSO control).
13. Data can be plotted against time (e.g., Fig. 5, left panels) or,
for a single time point, against ligand concentration (e.g.,
Fig. 5, right panels).
a ‘ON’ Switch
Synthetic ligand
Gene Expression
Gene Expression
Synthetic ligand
Natural ligand Natural ligand
Time [Ligand]
b ‘OFF’ Switch
Natural ligand
Natural ligand
Gene Expression
Gene Expression
Synthetic ligand
Synthetic ligand
Time [Ligand]
Fig. 5 Schematic representation of expected results from quantitative in vivo characterization. Left panels
show gene expression plotted against time with different ligand concentrations plotted as individual series.
Right panels show gene expression plotted against ligand concentration for a single time point. (a) “ON” ribo-
switches. Gene expression increases with increasing time and increasing ligand concentration for the syn-
thetic ligand (dose-dependence). Gene expression is not affected by increasing time or increasing ligand
concentration for the natural ligand (orthogonality). (b) “OFF” riboswitches. Gene expression decreases with
increasing time and increasing ligand concentration (dose-dependence). Gene expression is not affected by
increasing time or increasing ligand concentration for the natural ligand (orthogonality)
3.5 Preparation of It may be desirable to characterize the new aptamer domain–syn-

DNA Template to thetic ligand interaction in vitro to demonstrate that the synthetic
Produce Aptamer RNA ligand binds to the aptamer domain directly (i.e., the ligand is not
for In Vitro Biophysical converted into another compound in vivo which subsequently
Characterization binds to the riboswitch), to confirm the selectivity of binding, and
to determine the binding affinity of the aptamer for the ligand. For
these studies it is necessary to produce the RNA corresponding to
the aptamer domain by in vitro transcription. Since most aptamer
domains are less than 100 nucleotides in length, it is possible to
generate the required double-stranded transcription template in a
PCR reaction using a single-stranded oligonucleotide as the tem-
plate. This oligonucleotide should begin with the T7 RNA poly-
merase promoter sequence (TAA TAC GAC TCA CTA TA)
immediately followed by two G residues (required for efficient
transcription) (see Note 18), which, in turn, are followed by the
desired aptamer sequence.
1. Design 5′ and 3′ oligonucleotide primers 15–20 nucleotides
in length with homology to the 5′ end of the template and
complementary to the 3′ end of the template, respectively.
2. Prepare the following 1 ml scale (see Note 19) PCR reaction:
(a) 200 μl of 5× HF buffer.
(b) 100 μl of dNTP mix.
(c) 5 μl of 100 μM 5′ primer.
(d) 5 μl of 100 μM 3′ primer.
(e) 1 μl of 100 μM single-stranded DNA template.
(f) 10 μl of Phusion High Fidelity DNA Polymerase (20 units).
(g) dH2O to 1 ml.
3. Using standard PCR conditions, perform 30 cycles with a suit-
able annealing temperature for the primers.
4. Purify the DNA by phenol:chloroform extraction using
MaXtract high-density tubes (see Note 20).
(a) Centrifuge two MaXtract tubes for 30 s at 13,000 × g.
(b) Add 500 μl of the PCR reaction and 500 μl of
phenol:chloroform:isoamyl alcohol (25:24:1) to each
tube, mix by inverting the tube several times (do not vor-
tex), and centrifuge for 5 min at 13,000 × g.
(c) Add 500 μl of chloroform to each tube, mix, and centri-
fuge for 5 min at 13,000 × g.
(d) Combine the upper aqueous layers from both tubes in a
fresh 15 ml tube.
(e) Add 1/9th volume (110 µl) of 3 M sodium acetate
(pH 5.2) and mix thoroughly.
(f) Add 2.5 volumes (2.775 ml) of chilled 100 % ethanol and
incubate at −20 °C overnight.
(g) Pellet the DNA by centrifugation at 20,000 × g for 20 min
at 4 °C.
(h) Decant the ethanol, and wash the pellet with 2 ml chilled
70 % ethanol.
(i) Decant the ethanol, and allow the pellet to air-dry.
(j) Resuspend in 200 µl of nuclease-free H2O, and determine
the concentration by measuring the absorbance at
260 nm.
3.6 Preparation of 1. Prepare the following 5 ml scale (see Note 19) in vitro tran-
Aptamer RNA by In scription reaction:
Vitro Transcription for (a) 500 μl of 10× transcription buffer.
In Vitro Biophysical
(b) 1.2 ml NTP mix.
Characterization
(c) 50 μl of 1 M DTT.
(d) 140 μl of 1 M MgCl2.
(e) 50–100 μg/ml (final concentration) double-stranded
DNA template (see Subheading 3.5).
(f) 0.25 mg/ml (final concentration) T7 RNA polymerase.
(g) Nuclease-free H2O to 5 ml.
2. Incubate at 37 °C for 6–8 h (see Note 21).
3. Add 500 μl of 0.5 M EDTA (pH 8.0) and incubate for a fur-
ther 30 min at 37 °C.
4. Using a syringe fitted with a needle, transfer the reaction to a
Slide-A-Lyzer dialysis cassette (3,000 MWCO, 12 ml capac-
ity), and dialyze overnight at 4 °C in 2 l of freshly prepared
riboswitch buffer.
5. Using a syringe fitted with a needle, transfer the dialyzed RNA
to a Vivaspin 6 centrifugal concentrator (3,000 MWCO) and
concentrate to a final volume of ~2 ml by centrifugation at
4,000 × g at 4 °C.
6. Add 200 μl (400 units) of DNase I and incubate for 30 min at
37 °C.
7. Store at 4 °C.
3.7 Purification of We purify our RNA transcripts (from T7 RNA polymerase, DNase
Aptamer RNA for In I, and prematurely terminated transcripts) by size-exclusion chro-
Vitro Biophysical matography. This intentionally avoids denaturation or
Characterization precipitation of the RNA which could trap the RNA in inactive,
nonnative conformations [23]. The following purification proto-
col is designed to produce RNA of suitable quality and quantity
for carrying out ITC and X-ray crystallography and assumes that
the user is familiar with operating an FPLC system since FPLC

training is beyond the scope of this chapter.
1. Attach a HiLoad 26/600 Superdex 200 size-exclusion col-
umn (320 ml column volume) to the FPLC system and equili-
brate with three column volumes of riboswitch buffer.
2. Attach a 5 ml sample loop, flush with 15 ml of buffer, and load
~2.2 ml RNA (see Subheading 3.6) into the loop.
3. Set the FPLC system to run isocratically at 1 ml/min, with
UV monitoring at 260 and 280 nm.
4. Inject the sample onto the column and elute with one column
volume riboswitch buffer, collecting 1.5 ml fractions.
5. Pool fractions containing full-length RNA transcript (see Note
22), transfer the RNA to a Vivaspin 6 centrifugal concentrator
(3,000 MWCO), and concentrate it to a final volume of ~2 ml
by centrifugation at 4,000 × g at 4 °C.
6. Transfer the RNA to a Slide-A-Lyzer dialysis cassette (3,000
MWCO, 3 ml capacity) using a syringe fitted with a needle
and dialyze overnight at 4 °C in 2 l of riboswitch buffer.
7. Recover the RNA, and determine the concentration by mea-
suring the absorbance at 260 nm.
8. The RNA is ready for ITC. Filter and degas the dialysis buffer
and store at 4 °C for cleaning the ITC sample cell, diluting the
RNA sample, and for making up ligand solutions.
9. For crystallography trials:
(a) Mix 1 ml of 100 μM RNA aptamer with 9 ml of 550 μM
syringe-filtered ligand solution, prepared in crystallogra-
phy buffer, and incubate at 4 °C for 30 min.
(b) Concentrate the aptamer–ligand complex to 200 μl
(~500 μM) using Vivaspin 6 and 0.5 ml Amicon Ultra
centrifugal concentrators.
(c) Use immediately or flash freeze in liquid nitrogen, and
store at −80 °C until needed.
3.8 In Vitro We typically characterize the aptamer–ligand interactions using

Biophysical ITC and X-ray crystallography (see Note 23). Detailed protocols
Characterization of for both of these techniques, with respect to riboswitches/RNA
Aptamer–Ligand [24–27], have been reported previously; therefore, we will not go
Interactions: into detail here. As an example, we present a brief outline of our
Isothermal Titration ITC method, focusing on how we design our experiments to
Calorimetry obtain useful binding data for our mutant aptamer–ligand pairs.
The following method assumes that the user is familiar with ITC,
as ITC training is beyond the scope of this chapter.
1. Before the first use, clean the sample cell as follows (see
Note 24):
(a) Fill the sample cell with 12 % NaOH and incubate at 65 °C
for 1 h.
(b) Wash the sample cell with DEPC-treated dH2O.
(c) Fill the sample cell with 5 % RNaseZap and incubate at
25 °C for 1 h.
(d) Wash the sample cell with DEPC-treated dH2O.
(e) Wash the sample cell thoroughly with dialysis buffer.
2. Before the first use, clean the titrant syringe and the Hamilton
syringe used to load the sample cell as follows:
(a) Fill with 5 % RNaseZap, and leave for 1 h.
(b) Wash with DEPC-treated dH2O.
(c) Wash with dialysis buffer.
3. Degas the aptamer RNA and ligand solutions using a degas-
sing unit.
4. Add 2 ml of dialysis buffer to the sample cell using a Hamilton
syringe (the blank).
5. Purge the titrant syringe twice and fill with 100 μM (see Note
25) ligand solution using a syringe fitted with tubing.
6. Dry the titrant syringe needle gently using a clean KimWipe, tak-
ing care not to touch the outlet pore at the base of the needle,
and carefully lower into the sample cell until it clicks into place.
7. Set the cell temperature to 25 °C, reference power to 5 μcal/s,
and stirrer speed to 310 rpm.
8. After a 60-s pre-titration delay, inject 2 μl of ligand over 4.8 s
followed by 24 × 12 μl injections of ligand, each over 28.8 s
with a 300-s delay between injections (ligand concentration
range of ~0.5 to ~12.5 μM when starting from a 100 μM
ligand stock solution).
9. Thoroughly wash the sample cell with dialysis buffer and then
load the sample cell with 2 ml of 10 µM (see Note 26) aptamer
RNA in dialysis buffer using a Hamilton syringe (the
experiment).
10. Carry out the same injection protocol as for ligand into buffer
(see step 8).
11. Fit the data using Origin software (see Note 27). A good data
set will have a clear sigmoidal curve (with c = 20–100) and an
appropriate stoichiometry (n = 1 for most riboswitch aptamer–
ligand interactions), and the first few injections should cover
at least 10 μcal each, with an average of 5 μcal per injection
over the course of the experiment (see Note 28).
4 Notes
1. The chosen vector should not require the CAT gene for plas-
mid maintenance.
2. We use the inducible lac promoter/operator to provide tighter
gene expression control when developing novel switches, but
alternative inducible or constitutive promoters could be
substituted.
3. The CAT gene could be substituted for alternative reporter
genes, e.g., tetA (allowing for selection or screening with tet-
racycline and/or NiCl2 [5, 28–30]), lacZ (allowing enzymatic
screening for β-galactosidase activity [6, 16]), GFP (allowing
for flow cytometry-based screening for fluorescence [8, 9]), or
cheZ (allowing for screening for motility [7, 12]).
4. CAS SciFinder and ChemFinder require a site licence.
5. The F′ episome carries the lacIq gene required for inducible
expression from the lac promoter.
6. Residual DEPC can inhibit the in vitro transcription reaction.
7. We purify T7 RNA polymerase in-house from an E. coli expres-
sion strain [31], since the quantities required for the
preparative-scale (5–20 ml) in vitro transcription reactions
that we routinely carry out can become costly if commercial
sources are used.
8. In our laboratory we keep a dedicated column for RNA
purification.
9. In our laboratory we have set aside an RNA bench space,
which includes a Labcare PCR cabinet and a dedicated set of
pipettes. Apparatus is frequently treated with RNaseZap
(Invitrogen) and washed with copious amounts of DEPC-
treated dH2O.
10. Riboswitch aptamer domains that bind small, uncharged, pla-
nar ligands with four or fewer conserved ligand-interacting
nucleotides are most suitable for mutagenesis. Riboswitches
that respond to more complex coenzymes (e.g., adenosylco-
balamin and flavin mononucleotide) are not suitable as alter-
native synthetic ligands are unavailable due to synthetic
inaccessibility.
11. In the case of riboswitches for which there is no X-ray crystal
structure available, in vitro mutagenesis analysis and/or
structural probing methods (e.g., in-line probing and enzy-
matic footprinting) could provide information about ligand-
binding nucleotides.
12. This approach has been validated and is feasible for small
ligands (e.g., purines), where initially up to two ligand-binding
nucleotides need to be mutated (4n − 1 = 15 mutants in total

if mutated to all other possibilities) to alter the ligand speci-
ficity of the aptamer. However, for binding pockets requiring
the mutation of more than two nucleotides simultaneously, a
library of mutants can be generated by site-saturated muta-
genesis PCR using the parental reporter plasmid as a tem-
plate and degenerate primers [32]. The resultant library
would then need to be subjected to a selection or a screen for
functional riboswitches and individual clones sequenced to
identify the mutations.
13. We typically filter out compounds with three or fewer hydro-
gen bond acceptors/donors.
14. The synthetic accessibility of the compounds should also be
considered in the eventuality that further derivatization and
optimization are necessary.
15. We typically set a maximum cost of £100 for 50 mg, as this is
sufficient for both in vivo and in vitro characterization, but
this can be revised up or down depending on the compounds
available.
16. Alternative reporters include lacZ [6, 16] and cheZ [7, 12].
17. At least ten ligand concentrations should be tested. The con-
centrations tested will depend upon the affinity of the aptamer
for the ligand, the kinetics of ligand binding, and the bioavail-
ability of the ligand.
18. The two G residues will be transcribed; therefore, it is impor-
tant to consider how these should be incorporated. Assuming
that the aptamer sequence does not begin GG, the required G
residues can be added to the 5′ end of the sequence. However,
this extension may prevent later crystallization of the RNA
transcript. Alternatively, the 5′ sequence can be modified,
along with any compensatory mutations to maintain second-
ary structure, to include the G residues.
19. The scale of the PCR/in vitro transcription reaction depends
upon how much RNA will be needed for biophysical charac-
terization. The scales suggested here are suitable for ITC and
X-ray crystallography, both of which typically require milli-
gram quantities of RNA.
20. See the manufacturer’s instructions for a more detailed
protocol.
21. The appearance of a white precipitate (magnesium pyrophos-
phate) indicates that the reaction is complete.
22. We avoid pooling shoulders or tails around the central RNA
peak, sacrificing recovery for purity, to ensure that the sample
is homogenous.
23. ITC has the advantage over many other in vitro methods in
that the interaction is monitored directly, in solution, without
immobilization or modification of either binding partner, and
it provides a complete biophysical characterization of the
interaction including binding affinity, binding stoichiometry,
and thermodynamic parameters (enthalpy change (ΔHobs),
entropy change (ΔS), and Gibbs free energy change (ΔG)).
However, it does require significant quantities of both aptamer
RNA and ligand (10–50 nmol in the sample cell, 2.5–250 nmol
in the titrant syringe) for each experimental run. Other bio-
chemical techniques can also be used, e.g., in-line probing
[33, 34], equilibrium dialysis [35], and selective 2′-hydroxyl
acylation and primer extension (SHAPE) [34].
24. Samples can become adsorbed onto the surface of the sample
cell, interfering with heat transfer and causing erratic, noisy
baselines.
25. The concentration of the ligand in the titrant syringe should
be 10–20 times the concentration of the aptamer RNA in the
sample cell.
26. To determine the lower and upper limits for the concentration
of the aptamer in the sample cell (Mcell ), we use the equation
c = (n ´ Mcell ) / K d
where c describes the sigmoidicity of the binding curve
(should be between 20 and 100 [36]), n is the stoichiometry
of the interaction (1 for most riboswitch aptamer–ligand inter-
actions), and Kd is the expected dissociation constant. If the
expected Kd is unknown, 10 μM aptamer is a reasonable con-
centration for pilot studies.
27. The baseline and integration limits for each injection should be
checked manually before attempting to fit the data since noise
can perturb the automatic fitting performed by the software.
28. The aptamer/ligand concentrations or injection settings can
be adjusted in subsequent experiments until these criteria are
satisfied.
References
1. Breaker RR (2011) Prospects for riboswitch ecule target. Curr Protoc Nucleic Acid Chem
discovery and analysis. Mol Cell 43:867–879 Chapter 9:Unit 9.5.1–5.23
2. Serganov A, Nudler E (2012) A decade of 5. Sharma V, Nomura Y, Yokobayashi Y (2008)
riboswitches. Cell 152:17–24 Engineering complex riboswitch regulation by
3. Barrick JE, Breaker RR (2007) The distribu- dual genetic selection. J Am Chem Soc
tions, mechanisms, and structures of 130:16310–16315
metabolite-binding riboswitches. Genome Biol 6. Lynch SA, Desai SK, Sajja HK et al (2007) A
8:R239 high-throughput screen for synthetic ribo-
4. Codrea V, Hayner M, Hall B et al (2010) In switches reveals mechanistic insights into their
vitro selection of RNA aptamers to a small mol- function. Chem Biol 14:173–184
7. Topp S, Gallivan JP (2008) Random walks to syn- 22. Serganov A, Patel DJ (2012) Molecular recog-
thetic riboswitches-a high-throughput selection nition and function of riboswitches. Curr Opin
based on cell motility. Chembiochem 9:210–213 Struct Biol 22:279–286
8. Fowler CC, Brown ED, Li Y (2008) A FACS- 23. Uhlenbeck OC (1995) Keeping RNA happy.
based approach to engineering artificial ribo- RNA 1:4–6
switches. Chembiochem 9:1906–1911 24. Gilbert SD, Batey RT (2009) Monitoring RNA-
9. Lynch SA, Gallivan JP (2009) A flow ligand interactions using isothermal titration
cytometry-based screen for synthetic ribo- calorimetry. Methods Mol Biol 540:97–114
switches. Nucleic Acids Res 37:184–192 25. Salim NN, Feig AL (2009) Isothermal titration
10. Dixon N, Duncan JN, Geerlings T et al (2010) calorimetry of RNA. Methods 47:198–205
Reengineering orthogonally selective riboswitches. 26. Pikovskaya O, Serganov AA, Polonskaia A et al
Proc Natl Acad Sci U S A 107:2830–2835 (2009) Preparation and crystallization of
11. Desai SK, Gallivan JP (2004) Genetic screens riboswitch-ligand complexes. Methods Mol
and selections for small molecules based on a Biol 540:115–128
synthetic riboswitch that activates protein trans- 27. Edwards AL, Garst AD, Batey RT (2009)
lation. J Am Chem Soc 126:13247–13254 Determining structures of RNA aptamers and
12. Sinha J, Reyes SJ, Gallivan JP (2010) riboswitches by X-ray crystallography. Methods
Reprogramming bacteria to seek and destroy Mol Biol 535:135–163
an herbicide. Nat Chem Biol 6:464–470 28. Nomura Y, Yokobayashi Y (2007) Reengineering
13. Topp S, Reynoso CMK, Seeliger JC et al a natural riboswitch by dual genetic selection. J
(2010) Synthetic riboswitches that induce gene Am Chem Soc 129:13814–13815
expression in diverse bacterial species. Appl 29. Muranaka N, Sharma V, Nomura Y et al (2009)
Environ Microbiol 76:7881–7884 An efficient platform for genetic selection and
14. Reynoso CMK, Miller MA, Bina JE et al screening of gene switches in Escherichia coli.
(2012) Riboswitches for intracellular study of Nucleic Acids Res 37:e39
genes involved in Francisella pathogenesis. 30. Muranaka N, Abe K, Yokobayashi Y (2009)
MBio 3:e00253–00212 Mechanism-guided library design and dual
15. Dixon N, Robinson CJ, Geerlings T et al genetic selection of synthetic OFF ribo-
(2012) Orthogonal riboswitches for tuneable switches. Chembiochem 10:2375–2381
coexpression in bacteria. Angew Chem Int Ed 31. He B, Rong M, Lyakhov D et al (1997) Rapid
51:3620–3624 mutagenesis and purification of phage RNA
16. Suess B, Fink B, Berens C et al (2004) A the- polymerases. Protein Expr Purif 9:142–151
ophylline responsive riboswitch based on helix 32. Chronopoulou EG, Labrou NE (2011) Site-
slipping controls gene expression in vivo. saturation mutagenesis: a powerful tool for
Nucleic Acids Res 32:1610–1614 structure-based design of combinatorial muta-
17. Wachsmuth M, Findeiß S, Weissheimer N et al tion libraries. Curr Protoc Protein Sci Chapter
(2013) De novo design of a synthetic ribo- 26:Unit 26.6
switch that regulates transcription termination. 33. Regulski EE, Breaker RR (2008) In-line prob-
Nucleic Acids Res 41:2541–2551 ing analysis of riboswitches. Methods Mol Biol
18. Link KH, Breaker RR (2009) Engineering 419:53–67
ligand-responsive gene-control elements: les- 34. Wakeman CA, Winkler WC (2009) Analysis of
sons learned from natural riboswitches. Gene the RNA backbone: structural analysis of ribo-
Ther 16:1189–1201 switches by in-line probing and selective
19. Carothers JM, Oestreich SC, Szostak JW 2′-hydroxyl acylation and primer extension.
(2006) Aptamers selected for higher-affinity Methods Mol Biol 540:173–191
binding are not more specific for the target 35. Mandal M, Breaker RR (2004) Adenine ribo-
ligand. J Am Chem Soc 128:7929–7937 switches and gene activation by disruption of a
20. Johnson JE Jr, Reyes FE, Polaski JT et al (2012) transcription terminator. Nat Struct Mol Biol
B12 cofactors directly stabilize an mRNA regu- 11:29–35
latory switch. Nature 492:133–137 36. Myszka DG, Abdiche YN, Arisaka F et al
21. Peselis A, Serganov A (2012) Structural (2003) The ABRF-MIRG‘02 study: assembly
insights into ligand binding and gene expres- state, thermodynamic, and kinetic analysis of
sion control by an adenosylcobalamin ribo- an enzyme/inhibitor interaction. J Biomol
switch. Nat Struct Mol Biol 19:1182–1184 Tech 14:247–269
Chapter 9
Dual Genetic Selection of Synthetic Riboswitches

in Escherichia coli
Yoko Nomura and Yohei Yokobayashi
Abstract
This chapter describes a genetic selection strategy to engineer synthetic riboswitches that can chemically
regulate gene expression in Escherichia coli. Riboswitch libraries are constructed by randomizing the nucle-
otides that potentially comprise an expression platform and fused to the hybrid selection/screening marker
tetA–gfpuv. Iterative ON and OFF selections are performed under appropriate conditions that favor the
survival or the growth of the cells harboring the desired riboswitches. After the selection, rapid screening
of individual riboswitch clones is performed by measuring GFPuv fluorescence without subcloning. This
optimized dual genetic selection strategy can be used to rapidly develop synthetic riboswitches without
detailed computational design or structural knowledge.
Key words Riboswitch, Aptamer, Gene regulation, RNA engineering, Translational regulation
1 Introduction
As synthetic biology progresses toward more practical applications,

we will inevitably face challenges to interface synthetic gene cir-
cuits with the complex chemical environment. For example, it may
be desirable to negatively modulate an enzyme level in response to
accumulation of an intermediate in metabolic engineering or acti-
vate a chemical degradation pathway in response to an environ-
mental pollutant [1]. Riboswitches are used extensively by many
bacteria to regulate endogenous gene expression [2], and they
have inspired researchers to engineer synthetic riboswitches [3].
In this chapter, we describe a strategy to rapidly engineer syn-
thetic riboswitches based on synthetic or natural RNA aptamers
using genetic selection in Escherichia coli (Fig. 1). The strategy
starts with the design and construction of a plasmid library of par-
tially randomized riboswitch mutants (>105) fused to the selection/
screening marker tetA–gfpuv fusion [4]. The E. coli cells trans-
formed with the riboswitch mutants are serially cultured in media
that favor the growth of either ON or OFF state in the presence or
131
132 Yoko Nomura and Yohei Yokobayashi
Riboswitch library Characterize individual riboswitch clones
Fluorescence
Aptamer NNN…NNN SD
tetA-gfpuv
- +
Clones
Characterize selected population

ON selection OFF selection
Fluorescence
(Tc +ligand) (NiCl2 -ligand)
- + - +
Ligand Ligand
Fig. 1 Dual genetic selection strategy. A riboswitch library is constructed in E. coli by partially randomizing an
appropriate region within the 5′ UTR. The library is subjected to one or more cycles of ON (using Tc: tetracy-
cline) and OFF (using NiCl2) selections to enrich the functional riboswitches. The selected population of E. coli
cells can be characterized as a population or as individual clones by measuring the fluorescence reporter gene
expression
the absence of the aptamer ligand. The tetracycline antiporter

(tetA) functions as a dual selection marker by enabling ON cells to
grow in the presence of tetracycline. Conversely, the OFF selection
is performed in the presence of NiCl2 because E. coli cells overex-
pressing tetA are more sensitive to Ni2+ compared to the cells that
do not [5, 6]. After the selection, the surviving cells can be charac-
terized for riboswitch activity as a population or cloned and indi-
vidually characterized using the fluorescence reporter (gfpuv)
without subcloning or retransformation. The salient feature of our
selection strategy is the use of a single selection marker (tetA) to
perform both ON and OFF selections as well as the fused fluores-
cence reporter (gfpuv) which enables rapid quantitative screening
of the selected cell populations and individual clones. This greatly
streamlines the selection as well as the subsequent screening pro-
cesses while minimizing the emergence of false positives that often
complicate genetic selection experiments.
2 Materials
2.1 Cell Culture 1. LB: 1 % bacto tryptone, 0.5 % yeast extract, 1 % NaCl in diH2O
Media and Strains supplemented with an appropriate antibiotic for plasmid main-
tenance. LB agar plates are prepared with 1.5 % w/v agar.
2. M9 minimal medium: Mix separately sterilized reagents in the
following order, thoroughly mixing after adding each reagent
to prepare 1 L (see Note 1): 777 mL of diH2O, 10 mL of 40 %
v/v glycerol, 0.1 mL of 1 M CaCl2, 2 mL of 1 M MgSO4,
10 mL of 10 % w/v casamino acids, 200 mL of 5× M9 salts
(64 g/L Na2HPO4·7H2O, 15 g/L KH2PO4, 2.5 g/L NaCl,
Dual Genetic Selection of Riboswitches 133
5 g/L NH4Cl), 1 mL of 1,000× filter-sterilized antibiotic for

plasmid maintenance.
3. Host E. coli strain for genetic selection: TOP10 (Life
Technologies) was used in our previous work [4]. In principle,
any strain that is compatible with the riboswitch’s aptamer
ligand and lacks tetracycline resistance should work. Competent
cells with a sufficient transformation efficiency to construct a
riboswitch library (>105 independent transformants) must be
prepared or purchased.
2.2 Equipment 1. UV transilluminator (320 and 360 nm).

2. Thermal cycler.
3. Incubator (37 °C for E. coli incubation).
4. Incubator shaker (37 °C for E. coli culture).
5. Microcentrifuge (plasmid miniprep and DNA purification).
6. Agarose gel electrophoresis apparatus (chamber and power
supply).
7. Microplate reader (for cell fluorescence and OD600
measurements).
2.3 Plasmids 1. A template plasmid with an aptamer ligand and tetA–gfpuv

and Primers selection marker (see Note 2).
2. PCR primers for library construction (sequence design consid-
erations are given in Subheading 3).
3. An appropriate sequencing primer for sequencing selected
clones.
4. (Optional) 5′ phosphorylated primers p-GFP-f (5′ATTGA
GTAAA GGAGA AGAAC TTTTC AC 3′) and p-ORF-r (5′
GGATC CAGCA GGTCG ACTTG CAT 3′) for tetA removal.
2.4 Reagents 1. High-fidelity DNA polymerase, e.g., Phusion DNA poly-

merase (Thermo Scientific).
2. dNTPs.
3. Tetracycline (see Note 3).
4. Nickel (II) chloride (NiCl2) (see Note 4).
5. Glycerol.
6. Restriction enzyme Dpn I (New England Biolabs).
7. Agarose gel electrophoresis reagents (buffer, agarose, dye,
etc.).
8. DNA Clean and Concentrator-5 (Zymo) or other silica col-
umn DNA purification kit.
9. Zyppy Plasmid Miniprep Kit (Zymo) or other plasmid mini-
prep kit.
10. Quick Ligation Kit (New England Biolabs) or other T4 DNA

ligase.
11. Phosphate-buffered saline (PBS).
3 Methods
3.1 Cell Culture All liquid cultures are grown in M9 minimal medium or LB
Conditions medium supplemented with an appropriate antibiotic to maintain
the plasmids and kept at 37 °C in an incubator shaker (~275 rpm).
All agar plates are incubated at 37 °C overnight after plating.
3.2 Design and 1. Successful genetic selection critically depends on the use of
Construction ON optimal selection pressures that favor the growth or the sur-
and OFF Controls vival of the desired phenotypes. The optimal selection condi-
tions depend on various parameters such as promoter strength,
plasmid copy number, and host genotype. Therefore, it is
important to carefully optimize the selection conditions prior
to the actual riboswitch selection. In order to optimize the
selection conditions, we recommend the construction of two
controls, one that represents an ON state and another that
represents an OFF state of the presumed riboswitches (Fig. 2a).
The controls should be constructed using the same plasmid
backbone and the promoter with which the researcher desires
to use the functional riboswitches. The ON control could be
any canonical 5′ UTR that allows efficient translation initiation
fused to the tetA–gfpuv marker gene (see Note 5). The OFF
control should be carefully designed so that it expresses a suf-
ficiently low level of TetA–GFPuv that is acceptable for the
desired riboswitches at their OFF state. One might consider
using a “blank” plasmid (no encoded tetA–gfpuv) as an OFF
control. In reality, however, most functional riboswitches (and
other protein-based switches) produce low levels of basal
expression even at OFF states. Using a blank plasmid as the
OFF control could result in the optimized selection condition
being too stringent such that no reasonably functional ribo-
switches would survive the selection. To obtain a “realistic”
OFF control, a stable hairpin loop may be engineered to
sequester the Shine–Dalgarno (SD) sequence to suppress
translation (Fig. 2a). Such sequestration of an SD region is
often observed in natural and engineered riboswitches.
2. After construction of the ON and OFF controls, transform
the plasmids into an appropriate selection host such as TOP10
(see Note 6).
3. Start overnight cultures of the control strains in 1.0 mL of M9
medium.
Fluorescence
ON control
5’ SD tetA-gfpuv 3’
- +
Ligand
Fluorescence
OFF control
SD
5’ tetA-gfpuv 3’
- +
b Ligand
ON
OFF
OD600
OD600
OFF ON
0 10 20 30 40 50 0 0.1 0.2 0.3 0.4 0.5

Tetracycline (mg/mL) NiCl2 (mM)
Fig. 2 Suggested ON and OFF control constructs to be used for optimization of

the selection conditions. (a) The ON and OFF controls should constitutively
express TetA–GFPuv marker at high and low levels, respectively, to mimic pre-
sumed outputs of the desired riboswitches. The ON control should possess a
canonical SD sequence without strong secondary structures. The OFF control
could contain a rationally designed stem–loop that sequesters the SD sequence.
Both constructs should be evaluated for cellular fluorescence to confirm the
desired expression levels. (b) An example of the survival profiles of the ON (filled
circles, black) and OFF (open circles, red ) controls under varying concentrations
of the selective reagents (tetracycline or NiCl2). The desirable selection condi-
tions are where the growth difference between the two phenotypes (ON and OFF)
is the greatest. In the depicted figure, 30 μg/mL tetracycline and 0.3 mM NiCl2
are likely to result in the best selection (color figure online)
4. Dilute the overnight cultures (e.g., ~100–200-fold) into fresh

1.0 mL of M9 medium (with and without aptamer ligand),
and culture the cells to mid to late log phase.
5. Harvest the cells by centrifugation, wash the cells once by
1.0 mL of PBS, and resuspend the cells in an appropriate vol-
ume of PBS (e.g., 0.2–1.0 mL).
6. Measure the OD600 and GFPuv fluorescence (excitation
395 nm; emission 509 nm) of each culture. Ensure that the
growth (OD600) and the GFPuv fluorescence normalized by
OD600 are not significantly affected by the ligand used. Also
verify that the expression levels of the ON and OFF controls
are acceptable for the intended application.
3.3 Optimization 1. Start overnight cultures of the ON and OFF control strains in
of the Selection 1.0 mL of M9 medium.
Conditions 2. Dilute the overnight cultures (e.g., 100-fold) into fresh 1.0 mL
of M9 medium containing varying concentrations of tetracy-
cline (e.g., 0, 10, 20, 30, 40, 50 μg/mL) or NiCl2 (0.1, 0.2,
0.3, 0.4, 0.5 mM). Depending on the type of the riboswitch
that is desired, the aptamer ligand should be added to either
tetracycline (riboswitch turns ON with the ligand) or NiCl2
(riboswitch turns OFF with the ligand) cultures. Culture the
cells for 24 h.
3. Measure OD600 of each culture using a microplate reader. The
ON and OFF controls should exhibit different sensitivity pro-
files to tetracycline and NiCl2 as illustrated in Fig. 2b. Identify
the optimal tetracycline and NiCl2 concentrations that allow
selective growth of the desired phenotype. For example, in
Fig. 2b, 30 μg/mL tetracycline and 0.3 mM NiCl2 exhibit the
best ON/OFF growth discrimination (see Note 7).
3.4 Library 1. Design and construct a riboswitch library by randomizing a

Construction part of the 5′ UTR that can function as an expression platform.
Which part(s) of the riboswitch to randomize may need some
considerations. An example from our previous work [4] shown
in Fig. 3 contains 15 consecutive randomized bases between
the thiamine pyrophosphate (TPP) aptamer and a canonical
SD sequence. This library yielded a number of riboswitches
that respond positively to the aptamer ligand but failed to yield
riboswitches that respond negatively [4]. It was necessary to
redesign the library to obtain riboswitches that repress gene
expression in response to the ligand [7]. Consequently, the
library design can influence the characteristics of the ribo-
switches that can be selected (no riboswitch can be obtained by
selection if it does not exist in the library) and thus deserves a
careful attention. A typical protocol for a library preparation is
described below.
2. Starting with a plasmid that contains an aptamer cloned in the
5′ UTR of tetA–gfpuv coding sequence, PCR amplify the
whole plasmid using primers that contain degenerate bases as
shown in Fig. 3 using a high-fidelity polymerase (e.g., Phusion
DNA polymerase). It is recommended that the primers be syn-
thesized with 5′ phosphate. Otherwise, the primers need to be
phosphorylated prior to PCR, or have the PCR product phos-
phorylated using T4 polynucleotide kinase.
3. Add 10 units of Dpn I per 50 μL PCR reaction and incubate at
37 °C for 3 h.
4. Purify the PCR product by column purification (e.g., DNA
Clean and Concentrator-5).
TPP aptamer C AGGAG CAAACT ATG CAAGTCG

GATCTGGATAATGCCAGCGTAGGGAAGC ·····C AGGAG CAAACT ATG CAAGTCGACC
CTAGACCTATTACGGTCGCATCCCTTCG ·····G TCCTC GTTTGA TAC GTTCAGCTGG
CTATTACGGTCGCATCCCTTCG-P 5’ SD Start codon
PCR/DpnI/Self-ligation
TPP aptamer SD Start codon

GATCTGGATAATGCCAGCGTAGGGAAGCNNN…NNNC AGGAG CAAACT ATG CAAGTCGACC
CTAGACCTATTACGGTCGCATCCCTTCGNNN…NNNG TCCTC GTTTGA TAC GTTCAGCTGG
Randomized expression platform (N15)
Transformation
Riboswitch library
Fig. 3 Library construction strategy. The illustration shows a strategy to randomize a short (15 base) region
between an aptamer (TPP) and a canonical SD sequence by whole plasmid PCR. The primers (blue ) are 5′
phosphorylated, and one of them contains degenerate bases (N) at its 5′ terminus. After PCR, the template
plasmid is digested by Dpn I and the PCR product is self-ligated using T4 DNA ligase. Transformation should
yield >105 colonies (color figure online)
5. Ligate the purified PCR product using T4 DNA ligase follow-

ing the manufacturer’s instructions (see Note 8).
6. Transform the ligation solution into competent E. coli host
cells (e.g., TOP10) and spread on LB agar plates. Incubate the
plates at 37 °C overnight. Each plate (10 cm) should have no
more than 3,000 colonies. Estimate the total number of colo-
nies (library size) which should be as large as possible, prefer-
ably >105.
7. Recover the library cells by overlaying ~1 mL of liquid M9
medium per plate and gently swirling the plate. Combine the
recovered cells, add glycerol to 20 % final volume, and store in
aliquots at −80 °C.
3.5 Dual Genetic 1. Prior to initiating selection, culture the library along with the
Selection ON and OFF controls as described in Subheading 3.3 with and
without the aptamer ligand except omitting the selective reagents
(tetracycline or NiCl2). Measure the GFPuv levels of each cul-
ture as described above. The average fluorescence of the naive
library can guide the selection strategy described below. In dual
genetic selection, it is desirable to eliminate as much nonfunc-
tional mutants as possible in the first selection to avoid further
dilution of the functional clones. Therefore, if the library popu-
lation is generally OFF (low fluorescence), it is advisable to
begin with the ON selection. If the library population is gener-
ally ON, the OFF selection should be performed first.
2. Start an overnight culture of the cell population to be selected
in M9 medium.
3. Dilute the overnight culture ~100-fold into fresh M9 medium
(1.0 mL) with or without the aptamer ligand (depending on
the desired riboswitch characteristics and ON or OFF selection
to be performed) and culture for ~8 h. This nonselective
growth is necessary to enable each riboswitch mutant to adapt
to the presence or the absence of the ligand.
4. Dilute the cells ~100-fold into fresh M9 medium containing
an appropriate concentration of the selective reagent (tetracy-
cline for ON selection and NiCl2 for OFF selection) and cul-
ture for an additional 24 h.
5. Wash the cells with M9 medium and store in 20 % glycerol for
additional rounds of selection and/or characterization.
6. Complete both ON and OFF selections as described above.
Depending on the library, additional cycles of ON/OFF selec-
tions may need to be performed to enrich the functional
riboswitches.
7. After one or more rounds of ON/OFF selection, evaluate the
selected cell population’s response to the aptamer ligand in
liquid culture as described in step 1 above. If the cells respond
to the aptamer ligand as a population, it strongly suggests that
the selection successfully enriched the desired riboswitches;
therefore, one should proceed to screening of individual clones
outlined below (3.6). If no response is observed, selection
conditions and/or library design may need to be further opti-
mized or additional cycles of selection may need to be per-
formed (see Note 9).
3.6 Screening of 1. Plate the selected cells on an LB agar plate to isolate single
Individual Clones colonies.
2. Pick and culture appropriate number (>10) of colonies in M9
medium overnight.
3. Dilute each clone ~100-fold into two M9 (1.0 mL) cultures

with and without the aptamer ligand. Grow to mid to late log
phase.
4. Harvest the cells by centrifugation (see Note 10), wash the
cells once with 1.0 mL of PBS, and resuspend the cells in an
appropriate volume of PBS (e.g., 0.2–1.0 mL).
5. Measure the OD600 and GFPuv fluorescence (excitation
395 nm, emission 509 nm) of each culture.
6. Identify functional riboswitch clones using the normalized
GFPuv fluorescence data.
7. Using a miniprep kit, isolate the plasmids of the functional
clones and analyze their sequences.
8. (Optional) Due to some toxicity exhibited by the overex-
pressed TetA, quantitative responses of the riboswitches to the
aptamer ligand concentration may vary when the riboswitches
are fused to other downstream genes [4]. If desired, majority
of the tetA coding sequence can be removed by PCR muta-
genesis, leaving GFPuv as the only reporter gene. We use two
universal primers p-GFP-f and p-ORF-r to amplify the whole
plasmid by PCR followed by self-ligation, similarly as described
above for library construction (Fig. 3).
4 Notes
1. The order of mixing is important to avoid precipitation of

CaCl2.
2. One of the plasmids described in our previous report [4] is
available from the authors upon request.
3. Prepare 20 mg/mL stock solution in 50 % ethanol in diH2O
prior to use.
4. Prepare 2 M stock solution in diH2O and pass through a
0.2 μm sterile syringe filter. Store in aliquots in the dark at
4 °C.
5. Excessively strong expression of TetA–GFPuv can inhibit cell
growth. Some tuning of the translation efficiency may be nec-
essary to moderate the expression level by manipulating the
ribosome binding sequence.
6. Due to the use of tetracycline for ON selection, tetracycline-
resistant strains cannot be used as host for selection. Other
host restrictions may exist depending on the aptamer ligand to
be used.
7. The actual optimal concentrations of the selection reagents
may vary depending on various factors. Cell culture conditions
such as dilution factor and incubation time may also be adjusted

according to each system.
8. Steps 5 and 6 should be performed in a small scale first to
optimize the transformation efficiency and colony counts on
each plate. As a negative control, a DNA solution without
ligase should be transformed to confirm that the template plas-
mid is digested.
9. Even if the cell population does not appear to respond to an
aptamer ligand, the selected cells may contain sufficiently
enriched riboswitches to be discovered by extensive screening
of individual clones.
10. For qualitative screening, it may be possible to rapidly screen a
large number of clones by visually observing the cell pellets
over a UV transilluminator (360 nm). Use appropriate eye
protection when using UV.
Acknowledgments
We thank the previous members of our group, in particular,

Dr. Norihito Muranaka and Dr. Vandana Sharma who contributed
to the development of the dual genetic selection methods. The
work was supported by National Science Foundation.
References
1. Sinha J, Reyes SJ, Gallivan JP (2010) screening of gene switches in Escherichia coli.
Reprogramming bacteria to seek and destroy an Nucleic Acids Res 37:e39
herbicide. Nat Chem Biol 6:464–470 5. Nomura Y, Yokobayashi Y (2007) Dual selec-
2. Winkler WC, Breaker RR (2005) Regulation of tion of a genetic switch by a single selection
bacterial gene expression by riboswitches. Annu marker. Biosystems 90:115–120
Rev Microbiol 59:487–517 6. Podolsky T, Fong ST, Lee BT (1996) Direct
3. Wittmann A, Suess B (2012) Engineered ribo- selection of tetracycline-sensitive Escherichia coli
switches: expanding researchers’ toolbox with cells using nickel salts. Plasmid 36:112–115
synthetic RNA regulators. FEBS Lett 586: 7. Muranaka N, Abe K, Yokobayashi Y (2009)
2076–2083 Mechanism-guided library design and dual
4. Muranaka N, Sharma V, Nomura Y et al (2009) genetic selection of synthetic OFF riboswitches.
An efficient platform for genetic selection and Chembiochem 10:2375–2381
Chapter 10
Nucleotide Kinase-Based Selection System

for Genetic Switches
Kohei Ike and Daisuke Umeno
Abstract
Ever-increasing repertories of RNA-based switching devices are enabling synthetic biologists to construct
compact, self-standing, and easy-to-integrate regulatory circuits. However, it is rather rare that the existing
RNA-based expression controllers happen to have the exact specification needed for particular applications
from the beginning. Evolutionary design of is powerful strategy for quickly tuning functions/specification
of genetic switches. Presented here are the steps required for rapid and efficient enrichment of genetic
switches with desired specification using recently developed nucleoside kinase-based dual selection system.
Here, the library of genetic switches, created by randomizing either the part or the entire sequence coding
switching components, is subjected to OFF (negative) selection and ON (positive) selection in various
conditions. The entire selection process is completed only by liquid handling, facilitating the parallel and
continuous operations of multiple selection projects. This automation-liable platform for genetic selection
of functional switches has potential applications for development of RNA-based biosensors, expression
controllers, and their integrated forms (genetic circuits).
Key words Thymidine kinase, Nucleoside analog, Genetic selection, Dual selection, Directed
evolution, Genetic switch
1 Introduction
Due to the limited prediction capability in sequence/function

relationship, rational design of the genetic switches with desired
specification remains a daunting challenge. Thus, the evolutionary
(or combinatorial) design has been the most successful and realis-
tic approach for the functional tuning of genetic switches. Here,
the pool of switch variants is prepared to create a diverse set of
switching properties, from which the variants with desired specifi-
cations are selected in a high-throughput manner. Because all
genetic switches and their assemblies (regulatory circuits) ulti-
mately turn ON or OFF the genetic expression, one can isolate
any type of genetic switches, once good selection platform for ON
and OFF states of gene expression is established. A variety of posi-
tive and negative selection markers are available. However, the use
141
142 Kohei Ike and Daisuke Umeno
of independent selection module for the ON/OFF state is in general

not recommended because it unnecessarily complicates the selec-
tion processes. Another issue is the frequent emergence of false
positives. This concern is especially prominent for cis-acting regu-
lators such as translation-controlling riboswitches. To circumvent
these problems, Yokobayashi and his colleagues have proposed the
single-gene dual selecting system for ON/OFF states of the
genetic switches [1] or its modified version [2]. This system uti-
lizes the fact that the expression of tetracycline/H+ antiporter
TetA confers host cells tetracycline resistance (ON selection) while
sensitizing cells to the toxic metal ions such as Ni (II) to host cells
(see Chapter 9).
Recently, we developed an alternative selection system using
herpes simplex virus thymidine kinase (hsvTK) as a dual selector
(Fig. 1) [3]. For selecting for the OFF state of the genetic switches,
it utilizes the highly efficient kinase activity on artificial nucleoside
called dP [3]. dP passes through the cellular membrane, phos-
phorylated by hsvTK and incorporated into the genomic DNA of
host cell. In duplex, it forms base pairs either with A and G,
thereby scrambling the genomic information of host cell. Thus,
the cells harboring genetic switches in OFF state are enriched by
>107-fold only by 5–15 min of exposure to low concentration
(10–100 nM) of dP (Fig. 1). By adding the cocktail of dT and
2′-deoxy-5-fluorouridine (5FdU), one can use hsvTK as efficient
selector for ON state of the genetic switches (Fig. 1). When 5FdU
is added, it is phosphorylated to yield 5F-dUMP, a potent inhibi-
tor of thymidine synthase (ThyA) [4]. The result is the blockage
of the de novo synthesis of thymidine. In this situation, growth of
E. coli is dependent on exogenously supplemented dT. With tdk
background, one can enrich the hsvTK-expressing cell, with the
efficiency of >106-fold, simply by culturing overnight in the pres-
ence of 5FdU and dT.
Here, we demonstrate the functional selection of quorum
sensing switches based on Vibrio fischeri LuxR/Plux system
(Fig. 2). This system is one of the most popular cell–cell commu-
nication devices in synthetic biology community [5–8]. The core
component is the LuxR sensory protein that specifically binds to
homoserine lactones (HSLs), the signaling molecules used in quo-
rum sensing system shared by diverse gram-negative bacteria, and
activates the gene expression under the control of Lux promoter
(Plux). Demonstrated here is the creation of LuxR library with
error-prone PCR, followed by OFF selection/ON selection to rap-
idly enrich the functional variants. Although this motif does not
includes the RNA-based switching mechanism, the exact proce-
dure shown here should be applicable to the engineering of RNA-
based switches including riboswitches, ribozyme switches, and
genetic controllers that employ RNA-binding protein.
Nucleotide Kinase-Based Selection System for Genetic Switches 143
Fig. 1 Dual selection of genetic switches using nucleoside kinase activity. The entire sequence or a part of the
genetic switch component is randomized either by error-prone PCR or by oligonucleotide-based randomiza-
tion. The resultant “switch library” is then placed on the condition to be at OFF state and then subjected to OFF
selection by adding nucleoside dP. The survivor pool is then placed on the condition to be ON, subjected to ON
selection by the growth in the presence of dT and 5FdU
2 Materials
2.1 Library Creation 1. Template plasmid (pTrcHis2-luxR) (see Notes 1 and 2).
(PCR Random
2. PCR primers (forward and reverse) (see Note 3):
Mutagenesis)
Primer 1: 5′-CAATCTGTGTGGGCACTCGAC-3′.
Primer 2: 5′-TACTGCCGCCAGGCAAATTC-3′.
3. Taq polymerase.
4. 10× Taq buffer.
5. 10× dNTP mixture: 2 mM each of dATP, dTTP, dCTP, dGTP.
Prepare 50 μL aliquots of this mixture (to avoid excessive
freeze/thaw cycles) and store at −20 °C.
Fig. 2 Model system: LuxR/Plux system. Sensory protein LuxR is constantly

expressed from Trc promoter. LuxR binds to pheromone molecule HSL in gram-
negative quorum sensing systems. The LuxR–HSL complex activates the gene
expression under Plux promoter. The reading frame of LuxR is randomized by
error-prone PCR to make LuxR library that is to be subjected to the ON/OFF
selection in various HSL concentrations
6. 10× MnCl2 solution: 10 μM prepared in distilled water and

stored at room temperature.
7. Nuclease-free water.
8. pTrcHis2-based vector (see Note 2) [7].
9. DNA clean and concentrator™-5 (Zymo Research Corporation,
Orange, CA, USA).
10. Restriction enzymes NcoI and HindIII.
11. T4 DNA Ligase (Invitrogen, high-conc.)
12. T4 DNA ligase buffer.
13. TAE: 40 mM Tris–acetate, and 1 mM ethylenediaminetet-
raacetic acid (EDTA).
14. 0.5 μg/mL Ethidium bromide in TAE.
15. Agarose gel: 0.7 % LE agarose in TAE and ethidium bromide.
16. Zymoclean™ Gel DNA Recovery Kit (Zymo Research
Corporation, Orange, CA, USA).
2.2 ON/OFF Selection 1. Selector plasmid (for details, see Notes 2 and 4).
2. E. coli JW1226 (KEIO collection, tdk strain) (see Note 5, [9]).
3. 50 mg/mL Carbenicillin (1,000× stock).
4. 30 mg/mL Chloramphenicol (1,000× stock).
5. Luria-Bertani (LB): 20 g of premixed LB medium (Invitrogen)
into 1 L of deionized water. Autoclave at 121 °C for 20 min.
6. SOC media: Add 20 g of bacto-tryptone, 5 g of yeast extract,
and 0.5 g of NaCl into 950 mL of deionized water. Add 10 mL
of a 250 mM solution of KCl. Adjust the volume of the solu-
tion to 1 L with deionized water. Autoclave at 121 °C for
20 min. Just before use, add 5 mL of a sterile solution of 2 M
MgCl2 and 20 mL of a sterile 1 M solution of glucose [10].
7. dP, 6-(β-D-2-Deoxyribofuranosyl)-3,4-dihydro-8H-pyrimido
[4,5-c][1,2]oxazin-7-one) [3, 11]: Prepare 1 μM–1 mM (typi-
cally 100 nM) dP solution dissolved in DMSO as 1,000× stock.
Sample stored at 4 °C.
8. OFF selection medium: LB containing 10–1,000 nM dP [3].
9. 10 mg/mL Thymidine in deionized water (1,000× stock):
Stock stored at 4 °C.
10. 1 mg/mL Adenosine in deionized water (1,000× Stock): Stock
stored at 4 °C.
11. 20 mg/mL 5FdU in deionized water (1,000× stock): Stored
at −20 °C.
12. Tryptone broth: Add 20 g of tryptone and 5 g of NaCl into
1 L of deionized water. Autoclave at 121 °C for 20 min.
13. ON selection medium: 10 μg/mL thymidine, 1 μg/mL ade-
nosine, and 20 μg/mL 5FdU [3, 4] in tryptone broth.
14. 10 mM stock solution of 3OC6-HSL in ethyl acetate acidified
with glacial acetic acid (0.001 % (v/v)): Stored at −20 °C.
2.3 Evaluating the 1. Reporter plasmid (for details, see Note 4).
Selected Switches 2. E. coli JW1226 (see Note 5, [9]).
3. Carbenicillin stock (see Subheading 2.2, item 3).
4. Chloramphenicol (see Subheading 2.2, item 4).
5. LB-agar (see Subheading 2.2, item 5).
6. SOC media (see Subheading 2.2, item 6).
7. NaCl solution: Add 9 g of NaCl into 1 L of deionized water.
Autoclave at 121 °C for 20 min.
8. Flow cytometer: MACS Quant VYB (Miltenyi Biotec).
9. Fluorescence plate reader: Fluoroskan Ascent® (Thermo Fisher
Scientific Inc.).
10. Plate reader: SpectraMax Plus384 (molecular devices).
3 Methods
3.1 Library Creation There are a variety of established methods for the construction of
random libraries [12]. Among them, error-prone PCR is the most
popular method that could apply for a region or the entire part of
the targeted gene. The simplest is the addition of Mn2+ into the
PCR, where manganese ion lowers the fidelity of the polymerase,
inserting random base substitutions into the target sequence [13].
1. For each PCR sample, add to tube 2 fmol of template DNA
(pTrcHis2-luxR), 5 μL of 10× Taq buffer, and 5 μL of 10×
dNTP mixture. 5 μL of 5 μM each primers (to be 25 pmol in
final concentration), 1 μL of Taq polymerase (5 U), 5 μL of
10× MnCl2 solution (to be 10–100 μM in final concentration),
and nuclease-free water to a final volume of 50 μL.
2. Mix the sample by pipetting, and confirm all the solution is on
the bottom.
3. Place the tubes in a thermal cycler, and run the following PCR
program: (1) 5-min initial denature at 94 °C; (2) 30-s denature
at 94 °C, 30-s annealing at the temperature designed for prim-
ers, and 1-min extension at 72 °C; repeat for 25 cycles; and (3)
10 min at 72 °C for final extension (see Notes 6 and 7).
4. Run the product for gel electrophoresis to estimate the yield of
full-length gene (see Notes 6 and 8).
5. Clean the PCR product by using a Zymo DNA clean and con-
centrator kit. In parallel, clean up the vector in the same way.
6. Digest both PCR product and the expression vector with
appropriate enzymes (NcoI/HindIII).
7. Gel-purify the digested samples using Zymoclean™ Gel DNA
Recovery Kit (see Note 9).
8. Treat the mixture of the purified insert (100 ng) and vector
(100 ng) with ligase buffer and T4 ligase for 2 h (400 units/
reaction to total volume of 10 μL) (see Note 10).
9. Optionally, concentrate/purify the ligation product by using a
Zymo DNA clean and concentrator kit (see Note 11).
10. Electroporate the ligation mixture (1 μL) into the competent
cell (40 μL) (see Note 12).
11. Add 1 mL of SOC media immediately after the electropora-
tion, and shake it for 1 h at 37 °C. After washing the transfor-
mants with fresh LB media, resuspend the cells into 10 mL of
LB media (see Notes 13 and 14).
12. After overnight shaking at 37 °C, miniprep the portion
(1–2 mL) of the above culture to obtain plasmid library
pTrcHis2-[luxR].
3.2 Selection of the 1. Add 100 ng (1 μL) of plasmid library (pTrcHis2-[luxR]) into
LuxR Variants with 40 μL of JW1226 cell harboring pAC-Plux-hsvtk (see Note 15).
Proper Switching 2. Add 1 mL of SOC media and gently shake for 1 h at 30 °C.
3. Collect the cells by centrifugation (4,000 rcf, 3 min), resus-
pend them into the 10 mL of LB media containing chloram-
phenicol (30 μg/mL) and carbenicillin (50 μg/mL), and allow
for overnight shaking at 37 °C.
4. OFF selection: Transfer about 106 cells into the 1 mL of OFF
selection medium containing 0–1,000 nM 3OC6-HSL and
leave in the shaking incubator (37 °C) for 2 h (see Notes 16
and 17).
5. Wash the cell by centrifugation (4,000 rcf, 3 min)/resuspen-
sion to fresh LB media containing the same concentration
of 3OC6-HSL at OFF selection. Repeat this process twice
(see Note 18).
6. Shake the culture for another 6 h at 37 °C (see Notes 19 and 20).
7. ON selection: From the above culture, take 10 μL (about 106
cells) to resuspend into 1 mL of ON selection medium con-
taining desired concentration of inducer (3OC6-HSL).
8. Shake the cell culture for another 20 h at 37 °C (see Note 21).
9. Miniprep the culture to collect the plasmid containing survivor
variants.
3.3 Characterization 1. Electroporate the library plasmids before and after the selec-
of the Survivor Pools tion into JW1226 harboring pAC-Plux-gfpuv.
and Isolation of 2. Add 1 mL of SOC and incubate at 37 °C for 1 h in shaker.
Switch Variants 3. After collecting the cells by centrifugation (4,000 rcf, 3 min),
resuspend them into the 10 mL of LB media containing chlor-
amphenicol (30 μg/mL) and carbenicillin (50 μg/mL), and
allow for overnight shaking at 37 °C.
4. Take 106 cells into fresh 500 μL of LB media containing vari-
ous concentrations (0–1 μM) of 3OC6-HSL and shake them
for another 12 h. Then proceed to step 5 or 6.
5. Population analysis (Fig. 3a): Wash the cell by saline solution.
Dilute the cell by about 1,000×. The diluted cell suspension is
subjected to flow cytometry (V2 channel, laser 405 nm, filter
525/50 nm) for the analysis of the pool.
6. Isolation of the switch variants: A fraction of transformant is
plated on LB-agar plate containing chloramphenicol (30 μg/
mL) and carbenicillin (50 μg/mL) to form colonies (see Note
20). Pick several to dozens of colonies from the plates, and
inoculate them into LB (2 mL) culture containing chloram-
phenicol (30 μg/mL) and carbenicillin (50 μg/mL) in deep-
well 96 plates, allowing the overnight growth at 37 °C.
Fig. 3 (a) Flow cytometric analysis of the library of HSL-induced genetic switches before and after the OFF and
ON selection in the presence of 10 and 1,000 nM 3OC6-HSL. Original variant mixture (library) was two-peaked
in any concentration of HSL, but after single round of OFF selection in the presence of 10 nM HSL and OFF
selection in the presence of 1,000 nM HSL, the properly responding LuxR/Plux systems were enriched.
(b) [HSL]-output transfer function of representative LuxR clones isolated from the survivor pool. Filled circle
represents the switching property of wild-type LuxR, while open circle, open triangle, and open squares rep-
resent that of LuxR variants
7. Each of the above cultures is inoculated to the LB plate con-

taining chloramphenicol (30 μg/mL), carbenicillin (50 μg/
mL), and various concentrations (0–10 μM) of 3OC6-HSL.
8. After overnight growth at 37 °C, select the variants for further
analysis. Analysis includes sequence analysis, input/output
transfer function in liquid medium (Fig. 3b), and cytometric
analysis of the switch variants. Upon necessity, one can deter-
mine the sensitivity (threshold concentration of input molecule
needed to turn on the switches), stringency (leakage level at
OFF condition), S/N ratio (maximum output intensity/
output intensity at OFF condition), cooperatively (Hill coeffi-
cient), signal selectivity, and transition time.
4 Notes
1. It is highly recommended to physically separate the region to

be randomized/mutagenized from selector genes. Thus, we
always use the double-plasmid system as is shown here. This is
mainly to minimize the accidental introduction of mutations
into the selector genes in the experimental processes for muta-

genesis and selection. Also, this format enables us to quickly
exchange the reporter/selector plasmids. For the functional
selection of cis-acting regulatory elements such as translational
regulator riboswitches, one can think of placing trans-acting
expression controllers such as LuxR under the direct control of
the riboswitches to be engineered, and the riboswitch region
should be mutagenized. This way, the selector/reporter plas-
mid described here can be used without further modification.
2. The gene coding for protein LuxR from Vibrio fischeri was sub-
cloned into pTrcHis2 vector (Invitrogen) to construct
pTrcHis2-luxR, which serves as template for PCR mutagene-
sis. LuxR gene used in this study has a sequence of luxR nearly
identical to the originally reported, except the amino acid sub-
stitution A4G (K2E) to create NcoI site in N terminal start
codon. The luxR also contains several non-synonymous sub-
stitutions A51G (K17K), T174C (I58I), and C624T (G208G).
3. PCR primers are designed to share the same or the similar
melting temperature. They are used at 5 μM and stored at
−20 °C. The primers for error-prone PCR are designed to
anneal to the outside of the reading frame of the gene so that
the entire reading frame is randomized. In this case, however,
regulator sequence (promoter and ribosome-binding site) is
also under mutagenesis. In the situation where the mutations
in the regulation sequence could lead to the false positives,
design the forward/reverse primer to anneal to the N/C ter-
minus of the reading frame, respectively. In this case, the prim-
ing sites of the target genes are virtually “masked” from the
mutagenesis.
4. Selector plasmid pAC-Plux-hsvtk was constructed by inserting
V. fischeri Lux promoter (Plux) and the gene for hsvTK into
pACYC184-based vector. The reporter plasmid pAC-Plux-
gfpuv was constructed simply by exchanging reading frame of
hsvTK of pAC-Plux-hsvtk with gfpuv (Clontech). The sequence
of the Lux promoter (Plux) is as follows:
5′- ACCTGTAGGATCGTACAG GT TTACG CAAGAAA
ATGGTTTGTTATAGTCGAATAAA-3′.
The sequence underlined is the binding site (operator site)
of LuxR–HSL complex, and letters in bold represent the
−35/−10 sequences (promoter regions).
5. For selecting ON state of the genetic switches, any tdk strains
could be used. In this work, we chose JW1226 (F− Δ(araD-
araB)567 ΔlacZ4787(::rrnB-3) λ− Δtdk-747::kan rph-1
Δ(rhaD-rhaB)568, hsdR514).
6. With a high amplification yield of error-prone PCR (>1,000-
fold), one can keep the library free from contamination by the
unamplified wild-type sequence.
7. Mutation rates of the library can be controlled by the concen-

tration of Mn2+ in a PCR reaction. Typical concentration range
for Mn2+ is from 0 to 100 μM. The higher concentration results
in the higher mutation rate of the resultant library. In the high
side (>200 μM), polymerase reaction is inhibited, resulting in
the low amount or no product.
8. In the given (fixed) concentration of Mn2+, the average muta-
tion rate of the library is proportional to the effective cycle
numbers of the PCR. The most convenient way to control the
cycle number is to change the amount of target DNA in the
PCR. In theory, tenfold decrease in template in error-prone
PCR results in log210 ~ca. 3.3-fold elevation in mutation rate,
given that final yield of PCR stays the same.
9. This step is important to remove the template plasmid. In
addition to this size fractionation, one can further eliminate
the template plasmid by the treatment of PCR product with
DpnI, which selectively digests plasmid-borne methylated
DNA, prior to the gel purification.
10. Addition of 10 mM ATP in the reaction mixture would increase
the ligation efficiency.
11. After purification, ligation can be stored at −20 °C without los-
ing the transforming efficiency (library size). In case where the
larger library is necessary, transform the (concentrated) liga-
tion into commercial super-competent cells and propagate the
transformant in a single mixed culture for 12 h. Miniprep this
mixed culture to prepare plasmid library ready for transform-
ing to screening strains.
12. To maximize the library size, use the ultrahigh competent cells
commercially available.
13. In case the density of competent cell is high, especially when
electroporation is applied, the subsequent culture should be of
relatively large volume (ca. 10 mL). Presence of too much
non-transformed cell (dead cell) hampers the normal propaga-
tion of transformant cell, resulting in the poor yield in plasmid
prep and in decreased library size.
14. Plate a portion (0.1–1 %) of the transformant mixture for col-
ony counting to evaluate the library size, which is an important
measure of genetic diversity.
15. Here again, plate a portion (1 %) of the transformant mixture
for colony counting in order to evaluate the effective selection
size (the number of variants subjected to the selection). For
the determination of selection size, the number of colonies
appeared on the plate is multiplied with 100.
16. E. coli cells (with and without tdk genotype) not expressing
hsvTK can tolerate up to 100 μM of dP without any detectable
change in cell viability and growth speed. However, we

observed ca. tenfold increase in mutation rate on the genomic
DNA, which was determined by rifampicin assay (unpublished
results).
17. In principle, one can start from ON selection and then move
on to OFF selection. We prefer to conduct OFF selection first
and then proceed to ON selection. This is firstly because OFF
selection is completed in a very short period of time and ON
selection requires overnight growth. Another and probably the
most important reason for this is that one cannot immediately
move on to the OFF selection after the ON selection; the tran-
sition from ON state to OFF state requires hours in proteomic
level due to the remainder of hsvTK (selector gene) when
expressed in the previous “ON” selection. Consequently, one
needs to wait for 4–6 h until the level of selector enzyme hsvTK
decreases to undetectable level before setting the media condi-
tion for OFF state. On the other hand, the transition from
OFF state to ON state is quick in proteomic level and free from
wrongly eliminating the proper switches (false negatives).
18. Washing should be repeated at least twice to minimize the dP
remaining in the media. Remaining dP could swipe off the
properly behaving switch variants by inducing the hsvTK at the
subsequent “ON” selection process.
19. To check whether OFF selection is properly conducted, you
can miniprep the rest of the OFF-selected culture at this point
and proceed for flow cytometric analysis. This also gives you
the backups for the possible failure in subsequent ON selec-
tion. With this “OFF-selected pool,” you can re-try the ON
selection in various different conditions.
20. In case you do not observe the elevation in cell density, this
could be due to the improper setup of the construct.
Frequently, genetic switches have the weak but certain level
of the leaky expression. Note that dP kinase activity is very
powerful in killing hsvTK-expressing cell, and even the mini-
mal level of leaky expression could be enough for cell death.
In such cases, basal expression construct should be down-
ward adjusted by changing either the promoter sequence or
ribosome-binding sequences.
21. In case you do not observe the increase in cell density, leave the
culture shaken for an additional 6–12 h. If the cell density
increases, then it means that the survivors (the cell harboring
switch variants properly in ON state) consist of the very minor
fraction of the library. If you do not observe the cell growth at
all, even after 24 h of shaking, it probably means that the initial
library did not contain the variants with desired specificity. You
should redesign the library.
References
1. Nomura Y, Yokobayashi Y (2007) Dual selec- 7. Tabor JJ, Salis HM, Simpson ZB et al (2009) A
tion of a genetic switch by a single selection synthetic genetic edge detection program. Cell
marker. Biosystems 90:115–120 137:1272–1281
2. Muranaka N, Sharma V, Nomura Y et al (2009) 8. Danino T, Mondragón-Palomino O, Tsimring
An efficient platform for genetic selection and L et al (2010) A synchronized quorum of
screening of gene switches in Escherichia coli. genetic clocks. Nature 463:326–330
Nucleic Acids Res 37:e39 9. Baba T, Ara T, Hasegawa M et al (2006)
3. Tashiro Y, Fukutomi H, Terakubo K et al Construction of Escherichia coli K-12 in-frame,
(2011) A nucleoside kinase as a dual selector single-gene knockout mutants: the Keio collec-
for genetic switches and circuits. Nucleic Acids tion. Mol Syst Biol 2:1–11
Res 39:e12 10. Sambrook J, Russell DW (2001) Molecular clon-
4. Yagila E, Rosnerb A (1971) Phosphorolysis of ing: a laboratory manual, 3rd edn. Cold Spring
5-fluoro-2′-deoxyuridine in Escherichia coli and Harbor Laboratory, Cold Spring Harbor, NY
its inhibition by nucleosides. J Bacteriol 108: 11. Negishi K, Loakes D, Schaaper RM (2002)
760–764 Saturation of DNA mismatch repair and error
5. Weiss R, Knight TF (2001) Engineered com- catastrophe by a base analogue in Escherichia
munications for microbial robotics, vol coli. Genetics 161:1363–1371
2054, Lecture Notes in Computer Science., 12. Arnold FH, Georgiou G (2003) Directed evo-
pp 1–16 lution library creation: methods and protocols.
6. Basu S, Gerchman Y, Collins CH et al (2005) Humana, Totowa, NJ
A synthetic multicellular system for pro- 13. Cirino PC, Mayer KM, Umeno D (2003)
grammed pattern formation. Nature 434: Generating mutant libraries using error-prone
1130–1134 PCR. Methods Mol Biol 231:3–9
Chapter 11
Measuring Riboswitch Activity In Vitro and in Artificial

Cells with Purified Transcription–Translation Machinery
Laura Martini and Sheref S. Mansy
Abstract
We present a simple method to measure the real-time activity of riboswitches with purified components in
vitro and inside of artificial cells. Typically, riboswitch activity is measured in vivo by exploiting β-galactosidase
encoding constructs with a putative riboswitch sequence in the untranslated region. Additional in vitro char-
acterization often makes use of in-line probing to explore conformational changes induced by ligand binding
to the mRNA or analyses of transcript lengths in the presence and absence of ligand. However, riboswitches
ultimately control protein levels and often times require accessory factors. Therefore, an in vitro system
capable of monitoring protein production with fully defined components that can be supplemented with
accessory factors would greatly aid riboswitch studies. Herein we present a system that is amenable to such
analyses. Further, since the described system can be easily reconstituted within compartments to build artifi-
cial, cellular mimics with sensing capability, protocols are provided for building sense-response systems within
water-in-oil emulsion compartments and lipid vesicles. Only standard laboratory equipment and commer-
cially available material are exploited for the described assays, including DNA, purified transcription–transla-
tion machinery, i.e., the PURE system, and a spectrofluorometer.
Key words Riboswitch, Transcription–translation, In vitro compartmentalization, Liposome,

Emulsion, Cell-free synthetic biology
1 Introduction
Riboswitches are genetically encoded control elements that

respond to small molecules through direct binding. Sensing is
mediated by an aptamer [1–4] sequence within the mRNA that
controls the conformation of the expression platform. Usually
ligand binding turns off gene expression; however, natural on-
riboswitches exist [5]. The induced conformational changes either
regulate transcription through terminator–anti-terminator activity,
translation by modulating the accessibility of the ribosome binding
153
154 Laura Martini and Sheref S. Mansy
site, mRNA processing, or splicing [6]. In addition to natural ribo-

switches, many riboswitches have been engineered by modifying
previously selected aptamer sequences [7] or by mutating natural
riboswitches to display new functionality [8]. Most of the charac-
terized natural riboswitches control transcription, whereas syn-
thetic riboswitches typically control translation.
Monitoring transcription in vitro is straight forward [9, 10]
thereby allowing for the characterization of riboswitches that alter
transcript length in a manner dependent upon the presence or
absence of ligand. However, riboswitches that control ribosome
binding site accessibility produce transcripts of the same length
regardless of the presence or absence of ligand, making methods
that quantify differences at the RNA level less insightful. Moreover,
riboswitches ultimately control protein synthesis, regardless of the
specific mechanism exploited. Therefore, more direct methods
that probe the influence of riboswitch activity on protein synthesis
are desirable. This is most often achieved by placing the riboswitch
in question within a genetic construct that encodes β-galactosidase,
a fluorescent protein [7, 11] or more recently, a protein involved
in motility [12] and monitoring the activity of the reporter protein
in Escherichia coli. In other words, the assay is carried out within
the cell and absorbance or fluorescence are quantified.
The advantage of such methods is that the activity of the ribo-
switch within a living cell is monitored, meaning that the measured
activity is not a result of imperfect in vitro approximations of in
vivo conditions. However, there are several limitations of such in-
cell assays. First, the influence of accessory proteins could easily be
missed, since their participation in sensing or transducing chemical
messages is largely uncontrolled in such experiments. Second, the
putative ligand either must be capable of crossing the membrane
(to allow for exogenous delivery) or easy to manipulate in terms of
concentration. For example, the activity of the flavin mononucleo-
tide (FMN) riboswitch was characterized at the transcriptional
level in vitro [13, 14], but the influence of FMN on protein syn-
thesis was not investigated, presumably due to the difficulty in
quantifying and modulating intracellular FMN concentrations.
Herein we present a simple method to characterize the influ-
ence of riboswitch activity on protein synthesis in vitro. Guidelines
for the design and assembly of the genetic construct, and the eval-
uation of in vitro riboswitch activity by monitoring the synthesis of
fluorescent protein with fully defined components are described
(Fig. 1). Importantly, this real-time fluorescence assay is amenable
to the screening of protein accessory factors and ligands, including
ligands that are metabolites, impermeable, or toxic. It should be
noted that the described protein synthesis assay does not replace
current methods that characterize transcriptional activity. The
investigation of both transcription and translation is needed in
order to fully define the mechanistic details of riboswitch activity.
Cell-free Riboswitch Activity 155
fluorescence
n time
atio
te riz
ac
ar
ch
ce
llu
lar
mi
mi
cs
Fig. 1 Cell-free systems for in vitro riboswitch characterization and the construction of artificial, cellular mim-
ics. The PURE system is used to characterize in real-time riboswitch activity through the expression of a
reporter protein either in vitro or inside of a compartment with dimensions similar to living cells
We additionally describe how this riboswitch controlled in

vitro transcription–translation system can be encapsulated within
compartments to build cellular mimics (Fig. 1). As opposed to the
majority of artificial cell studies that focus on self-replication, ribo-
switch sensing-based cellular mimics integrate more fully with the
environment and thus could potentially serve as a platform for
future technologies. The example described here uses water-in-oil
(w/o) emulsion droplets [15], vesicles [16], and a previously
reported theophylline riboswitch [7, 17, 18].
2 Materials
All solutions should be prepared using diethyl pyrocarbonate (DEPC)

treated water. All reagents are nuclease-free, molecular biology grade.
The theophylline riboswitch sequence used here is available from the
Registry of Standard Biological Parts (BBa_J89000).
2.1 Template 1. E. coli DH5α or similar laboratory, cloning strain.

Preparation 2. Commercial plasmid miniprep kit.
3. 25:24:1 Phenol–chloroform–isoamyl alcohol mixture (see
Note 1).
2.2 In Vitro 1. PURExpress in vitro protein synthesis kit (New England

Transcription Biolabs).
and Translation 2. Riboswitch ligand molecule (e.g., theophylline).
3. RNAse Inhibitor (RiboLock RNase Inhibitor, Fermentas).

4. Quartz ultra-micro cell cuvette (105.252-QS, Hellma).
5. QuantaMaster 40 UV–Vis Spectrofluorometer with a Peltier
temperature controlled single sample holder (Photon
Technology International) or a similar spectrofluorometer.
2.3 Emulsion 1. Mineral Oil.

Preparation 2. Span 80.
3. Tween 80.
4. Triton X-100.
5. 9 mm Teflon stir bar.
6. Magnetic stir plate.
7. Gilson Microman Positive displacement pipettes.
2.4 Liposome 1. 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC).

Preparation 2. Cholesterol.
3. N-(carbonyl-methoxypolyethyleneglycol 5000)-1,2-distearoyl-
sn-glycero-3-phosphoethanolamine (DSPE-PEG 5000,
NOF-Europe).
4. Rotary evaporator, e.g., Rotavapor R-210 with Vacuum Pump
V-700 (Buchi).
5. IKA T 10 basic ULTRA-TURRAX disperser with a 5 mm
diameter dispersing tool.
6. Mini-Extruder (Avanti Polar Lipids, Inc.).
7. Nucleopore Track-Etch Membrane 0.4 µm (Whatman).
8. Centrifugal evaporator, e.g., CentriVap Centrifugal Vacuum
Concentrator (Labconco).
9. Tris saline buffer; 50 mM Tris–HCl, 50 mM NaCl, pH 7.4
supplemented with 10 mg/mL Proteinase K (Fermentas).
2.5 Imaging 1. Zeiss Observer Z1 microscope (Carl Zeiss S.p.A.) or similar

fluorescence microscope.
3 Methods
3.1 DNA Template The DNA template can be either a circular plasmid or a linear PCR
Preparation product that contains a series of modular elements to allow for ribo-
switch controlled protein synthesis (Fig. 2). The sequence should
contain a transcriptional promoter, a sequence encoding the ribo-
switch that contains a ribosome binding site (preferably the natural
ribosome binding site sequence, if possible), a gene coding for a fluo-
rescent protein to act as a reporter, and a transcriptional terminator.
The transcriptional promoters for T7 and E. coli RNA polymerases are
typically used. However, since the activity of a riboswitch can depend
Fig. 2 The composition of a theophylline riboswitch for the real-time observation of riboswitch activity. The
construct shown here contains a standard T7 transcriptional promoter, a theophylline riboswitch that contains
a ribosome binding site, a gene that codes for A206K YPet (mYPet), and a T7 transcriptional terminator. Only
the start and stop codons of mYPet are shown. The riboswitch sequence is shown in bold
on the RNA polymerase [19], particularly for riboswitches that use a

terminator–anti-terminator mechanism, the choice of which promoter
to use can significantly impact in vitro riboswitch activity. Similarly, it
is advisable that the sequence chosen for the riboswitch portion of the
construct contains the ribosome binding site. Riboswitches are often
associated with inefficient ribosome binding sites, either due to a lack
of potential base-pairing interactions with the ribosome or because of
structural features of the riboswitch that obstruct ribosome binding
site—ribosome interaction. Further, since riboswitches typically do
not fully block protein synthesis in the off-state nor mediate robust
expression in the on-state, i.e., riboswitch control is leaky and gener-
ally mediates more subtle changes in expression [7], a ribosome bind-
ing site not tuned to the activity of the riboswitch could complicate
analyses. The reporter should be a protein that expresses well in vitro
and is easily detectable. We find the green fluorescent proteins super
folder GFP (sfGFP) and GFPmut3b and the yellow fluorescent pro-
teins YPet and Venus to be particularly good choices [20]. Finally,
incorporating a hairpin transcriptional terminator is advisable, even if
not absolutely required when using a linear PCR product as a tem-
plate, because structured RNA termini increase RNA stability and
thus protein yield [21].
1. The DNA template should contain from 5′ to 3′ a T7 pro-
moter followed by two GG nucleotides to enhance transcrip-
tion, a sequence encoding the riboswitch [17] and the RBS,
and the gene coding for the reporter protein followed by a
transcriptional terminator (Fig. 2).
2. The template should be amplified either by PCR or by
transforming a typical laboratory cloning strain of E. coli, such
as DH5α, and purified with a commercially available kit accord-
ing to the manufacturer’s instructions.
3. Subsequently, the DNA is phenol–chloroform extracted [22]
with an equal volume of Tris-buffered Phenol–Chloroform
(see Note 1).
4. The DNA is ethanol precipitated [23], resuspended in sterile
water, and stored at −20 °C.
3.2 In Vitro The major advantage of working in vitro, compared with in vivo, is
Transcription and that the system operates only with what is provided. In other
Translation Reaction words, in vitro activity cannot depend on unidentified cellular
components, because they are not present. To date, only E. coli
[24] and Thermus thermophilus [25] translation machinery have
been reconstituted in vitro from purified components. Of these
two, only the E. coli system, i.e., the PURE system, is commercially
available. It should be noted that in contrast to in vivo or cell-
extract conditions, reactions with purified transcription–translation
machinery do not contain nucleases. The lack of nucleases decreases
the amount of DNA template needed. Whether protein produc-
tion is more or less efficient with the PURE system in comparison
with cell-extract based systems depends on the specific folding
properties of the expressed protein.
1. The PURE system components should be aliquoted on ice and
stored in 0.2 mL microcentrifuge tubes. Convenient volumes
are 10 µL aliquots of solution A and 7.5 µL aliquots of solu-
tion B. Aliquots are stored at −80 °C.
2. Assemble the reaction components, except for the DNA tem-
plate, on ice following the manufacturer’s instructions.
Supplementary reagents can also be added, such as RNase
inhibitor (e.g., 20 U RiboLock RNase Inhibitor) or the ribo-
switch ligand (e.g., 0.5 mM theophylline).
3. The assembled reaction is transferred to a quartz cuvette and
incubated at 37 °C.
4. The reaction is initiated by the addition of the DNA template
and monitored by fluorescence spectroscopy for 6 h (Fig. 3)
(see Note 2). The DNA template concentration should be
screened. We used 250 ng of plasmid template in 25.5 µL
total reaction volume. If YPet is used as the reporter protein,
the excitation and emission wavelengths are 517 nm and
530 nm, respectively.
3.3 In Vitro Since cellular life is chemically distinct from the environment,
Compartmentalization efforts to mimic cells in the laboratory often times exploit w/o
emulsion droplets or vesicles to approximate the compartment of
the living, chemical system [26, 27]. However, even if cellular life
is distinct from the environment, life cannot exist in isolation and
must in some manner interface with the environment to survive
[28]. Since lipid vesicles are semipermeable and more similar to the
types of barriers found in biology, vesicles are better suited than
w/o emulsion droplets for the construction of cellular mimics.
Nevertheless, the encapsulation efficiency of w/o emulsions is
nearly 100 %, whereas encapsulation efficiency in vesicles is at best
30 % [29]. It is for this reason that the screening of compartmen-
talized reactions is carried out with w/o emulsions. Once optimal
conditions are identified, similar vesicle systems are setup.
500 ON
fluorescence (a.u) 400

RBS
300
200
100 OFF
0
0 100 200 300 400
time (min)
Fig. 3 In vitro theophylline riboswitch activity observed by measuring the expression of the reporter protein
mYPet. The presence of the ligand activates protein expression. A schematic representation of the mRNA in the
ligand-bound ON state and the uncomplexed OFF state are shown. Protein production is inhibited in this case
in the OFF state because the RBS is not available for base-pairing with the ribosome. The data were taken from
a previous in vitro theophylline riboswitch study [18]
3.3.1 Cell-Free The method described here is based on that of Davidson et al. [30]
Expression in w/o but has been scaled down to be compatible with small volume
Emulsion PURE system reactions. Positive displacement pipettes are used to
handle oil samples.
1. The oil phase that will be used for the w/o emulsion is assem-
bled in a 15 mL Falcon tube (see Note 3), in the following
order: 474.75 µL of mineral oil, 22.5 µL of span 80, 2.5 µL of
tween 80, and 0.25 µL of triton X-100.
2. The aqueous phase is first assembled in a microcentrifuge tube
by mixing the PURE system components on ice as described
above for the in vitro reactions. Note that the optimal DNA
template concentration may be different when transcription–
translation is performed in a compartment versus in vitro. We
used 500 ng of plasmid DNA in 25.5 μL of total aqueous vol-
ume (see Note 4) for expression in w/o emulsion droplets.
The theophylline concentration was also increased to 5 mM to
compensate for partitioning into the oil phase (see Note 5).
3. The 15 mL tube containing the oil phase is placed in a 250 mL
beaker filled with ice water on a magnetic stir plate. A teflon
stir bar is inserted in the oil phase and the oil is mixed by stir-
ring at maximum speed for 1 min.
4. The emulsion is formed by the drop-wise addition of the aque-
ous phase containing the PURE system reaction to the oil
phase over 1 min with continuous stirring. The emulsion is
then stirred for an additional 3 min (see Note 6).
5. Finally, the emulsion is transferred to a 2 mL microcentrifuge
tube and incubated at 37 °C for 6 h. 5 μL aliquots are removed
every hour for observation by fluorescence microscopy.
3.3.2 Cell-Free The freeze-dried empty liposome (FDEL) method, as described by

Expression in Vesicles Yomo and colleagues [16], is used with slight modification to build
the vesicles that house the transcription–translation reaction.
FDEL vesicles encapsulate macromolecular, hydrophilic compo-
nents relatively efficiently. A variety of lipid compositions can be
exploited. Here 12 μmol of 58:39:3 POPC–cholesterol–DSPE-
PEG 5000 is used.
1. Each lipid is dissolved in chloroform and mixed in a 5 mL
round-bottom flask.
2. The solution is subjected to rotary evaporation for 1 h. The
resulting thin lipid film is then hydrated with 1 mL of DEPC-
treated water and vortexed for 20 s or until a homogeneous
opaque solution is formed.
3. The vesicle solution is then transferred to a 2 mL microcentri-
fuge tube and disrupted with an IKA T 10 basic homogenizer
at high speed (level 4 setting) for 1 min (see Note 7).
4. Samples are extruded through 400 nm polycarbonate filters 11
times with an Avanti mini-extruder. 40 µL aliquots of the vesi-
cles are placed in 1.5 mL microcentrifuge tubes, frozen in liq-
uid nitrogen (see Note 8), and lyophilized with a centrifugal
evaporator overnight at 30 °C. A thin opaque lipid layer can be
observed at the bottom of the microcentrifuge tube after lyoph-
ilization. At this stage the samples can be stored at −20 °C.
5. A PURE system reaction is assembled on ice in a total volume of
22.1 µL, including 500 ng of the template plasmid (see Note 4).
20 U of RiboLock RNase inhibitor is added to the solution.
6. 10 µL of the assembled PURE system reaction is added to an
aliquot of FDEL vesicles on ice and incubated without agita-
tion for 2.5 h (see Note 9). The unused portion of the PURE
system reaction can be stored at −80 °C.
7. The hydrated liposomes are then diluted 20-fold in Tris saline
buffer supplemented with proteinase K in a 0.2 mL
microcentrifuge tube and incubated at 37 °C. The inclusion of
proteinase K is to degrade extravesicular proteins.
8. At this point the ligand to be sensed, e.g., 5 mM theophylline
(see Note 5) is added. Theophylline is capable of diffusing
across the membrane, binding directly to the mRNA, and acti-
vating translation thereby resulting in fluorescence. Control
reactions in the absence of ligand should result in no or signifi-
cantly reduced fluorescence.
9. The reactions are incubated at 37 °C for 6 h. 5 µL aliquots are
removed from the reaction every 1.5 h and visualized by fluo-
rescence microscopy.
3.4 Microscope Detecting activity inside of vesicles by fluorescence microscopy is

Sample Preparation more difficult than in vitro measurements with a spectrofluorome-
ter. First, encapsulation efficiency is low, particularly when over 80
different components need to be encapsulated within one vesicle in
order for protein synthesis to proceed [31] (see Notes 4 and 9).
Second, riboswitches typically have weak ribosome binding sites
and thus produce less protein than constructs typically exploited for
recombinant expression. Finally, fluorescent proteins photobleach
and require time to mature. We use a monomeric version of the yel-
low fluorescent protein YPet as a reporter because the characteris-
tics of YPet are more amenable to investigation by microscopy. YPet
is one of the brightest fluorescent proteins and is more photostable
than the majority of available fluorescent proteins [32]. The expres-
sion of monomeric YPet with the PURE system requires approxi-
mately 2 h to reach half maximal fluorescence [20].
1. 5 µL aliquots are removed from the reaction and spotted on a
clean glass slide. A cover slip is added.
2. The slide is then left for 2 min to rest on the bench. This step
helps decrease the number of rapidly moving vesicles.
3. The sample is then observed by bright field and epifluores-
cence with 100× magnification (Fig. 4). Care should be taken
to decrease photobleaching by decreasing exposure time.
4 Notes
1. It is preferable to avoid Phenol–Chloroform solutions that con-

tain ethylenediaminetetraacetic acid (EDTA), since the chela-
tion of metals by EDTA can interfere with enzyme activity.
2. A plate reader or a real-time PCR machine can be used in place
of a spectrofluorometer.
3. We also made w/o emulsions in 13 mL Sarstedt tubes with
stirring with x-shaped spinplus stir bars, as described by
Davison et al. [30]. The resulting emulsions were more homo-
geneous and more stable than the emulsions we obtained with
Falcon tubes and linear stir bars. However, we found the
Davidson et al. emulsion droplets to be smaller and thus more
difficult to observe by microscopy.
4. The PURE system instruction manual suggests screening
between 25 to 250 ng of template DNA for 25 μL reactions.
For the in vitro characterization of the theophylline riboswitch,
250 ng of plasmid template was found to be optimal. However,
better results were obtained with 500 ng DNA template for
compartmentalized reactions.
Fig. 4 Fluorescence microscopy images of riboswitch activity in compartments. Upper panels (a–c): activity
recorded in emulsion compartments. Middle panels (d–f): activity recorded in liposomes. Lower panels (g, h):
activity due to riboswitch sensing of the environment. Images are all epifluorescence, except for panels a, d, g,
which are bright-field. Panel b, e show the fluorescence resulting in compartments when the ligand is present.
The corresponding controls (i.e., in the absence of ligand) are also shown (c, f). In panel h, the activity of a ribo-
switch as a sensor element is shown. The data are from the characterization of a cell-free theophylline riboswitch
system and panels g and h are reproduced with permission from the Royal Society of Chemistry [18]
5. Since the oil–water partition coefficient of theophylline is low

[33], much more theophylline is required to activate the ribo-
switch in the presence of oil than in aqueous solution.
6. Stirring is an important parameter to consider when generat-
ing an emulsion. A constant stir force should be used and the
stir bar must be compatible with the tube holding the aqueous-oil
mixture. For example, the conical shape of a Falcon tube is not
compatible with x-shaped stir bars. The efficiency of mixing
can be qualitatively assed by eye by including in the aqueous
phase 1 mM HPTS (8-hydroxypyrene-1,3,6-trisulfonic acid)
and observing the distribution of green color throughout the
tube during stirring.
7. We also tested sonication as a possible disruption method.
Sonication at 70 A of amplitude for 5 min resulted in higher
dispersion. However, the number of liposomes observed by
microscopy was lower than that observed with the IKA T

homogenizer.
8. Freezing can also be performed with dry ice without any
appreciable difference in liposomes formation.
9. The slow vesicle formation process mediated by the natural
swelling method described herein results in fewer, but larger
vesicles that are easier to observe by microscopy than by other
methods that exploit vortexing. Additionally, compartment
size impacts protein synthesis efficiency with larger vesicles
being more compatible with protein synthesis [34].
References
1. Soukup GA, Breaker RR (1999) Relationship 12. Topp S, Gallivan JP (2007) Guiding bacteria
between internucleotide linkage geometry and with small molecules and RNA. J Am Chem
the stability of RNA. RNA 5:1308–1325 Soc 129:6807–6811
2. Nahvi A, Sudarsan N, Ebert MS, Zou X, Brown 13. Soukup GA, Breaker RR (1999) Engineering
KL, Braker RR (2002) Genetic control by a precision RNA molecular switches. Proc Natl
metabolite-binding mRNA. Chem Biol 9:1043 Acad Sci U S A 96:3584–3589
3. Wakeman CA, Winkler WC (2009) Analysis of 14. Wickiser JK, Winkler WC, Breaker RR,
the RNA backbone: structural analysis of ribo- Crothers DM (2005) The speed of RNA tran-
switches by in-line probing and selective scription and metabolite binding kinetics oper-
2′-hydroxyl acylation and primer extension. In: ate an FMN riboswitch. Mol Cell 18:49–60
Serganov A (ed) Riboswitches. Methods Mol 15. Tawfik DS, Griffiths AD (1998) Man-made
Biol, vol 540, pp 173–191 cell-like compartments for molecular evolu-
4. Ellington AD, Szostak JW (1990) In vitro tion. Nat Biotechnol 16:652–656
selection of RNA molecules that bind specific 16. Sunami T, Matsuura T, Suzuki H, Yomo T
ligands. Nature 346:818–822 (2010) Synthesis of functional proteins within
5. Serganov A, Yuan Y-R, Pikovskaya O, liposomes. Methods Mol Biol 607:243–256
Polonskaia A, Malinina L, Phan AT, Hobartner 17. Lynch SA, Gallivan JP (2009) A flow
C, Micura R, Breaker RR, Patel DJ (2004) cytometry-based screen for synthetic ribo-
Structural basis for discriminative regulation of switches. Nucleic Acids Res 37:184–192
gene expression by adenine- and guanine- 18. Martini L, Mansy SS (2011) Cell-like systems
sensing mRNAs. Chem Biol 11:1729–1741 with riboswitch controlled gene expression.
6. Serganov A, Nudler E (2013) A decade of Chem Commun 47:10734–10736
riboswitches. Cell 152:17–24 19. Lemay J-F, Desnoyers G, Blouin S, Heppell B,
7. Desai SK, Gallivan JP (2004) Genetic screens Bastet L, St-Pierre P, Massé E, Lafontaine DA
and selections for small molecules based on a (2011) Comparative study between transcrip-
synthetic riboswitch that activates protein trans- tionally- and translationally-acting adenine
lation. J Am Chem Soc 126:13247–13254 riboswitches reveals key differences in ribo-
8. Nomura Y, Yokobayashi Y (2007) switch regulatory mechanisms. PLoS Genet
Reengineering a natural riboswitch by dual 7:e1001278
genetic selection. J Am Chem Soc 20. Lentini R, Forlin M, Martini L, Del Bianco C,
129:13814–13815 Spencer AC, Torino D, Mansy SS (2013)
9. Artsimovitch I, Henkin TM (2009) In vitro Fluorescent proteins and in vitro genetic
approaches to analysis of transcription termina- organization for cell-free synthetic biology.
tion. Methods 47:37–43 ACS Synth Biol 2:482–489
10. Mironov A, Epshtein V, Nudler E (2009) 21. Aiba H, Hanamura A, Yamano H (1991)
Transcriptional approaches to riboswitch stud- Transcriptional terminator is a positive regula-
ies. IN: Serganov A (Ed) Riboswitches. tory element in the expression of the Escherichia
Methods Mol Biol, vol 540: 39-51, Humana coli crp gene. J Biol Chem 266:1721–1727
Press, NY, USA 22. Sambrook J, Russel DW (2001) Commonly
11. Muranaka N, Yokobayashi Y (2010) used techniques in molecular cloning. In:
Posttranscriptional signal integration of engi- Molecular cloning. (Ed) Jan Argentine Cold
neered riboswitches yields band-pass output. Spring Harbor Laboratory Press, NY, USA
Angew Chem Int Ed 49:4653–4655
23. Sambrook J, Russell DW (2006) Standard eth- liposomes: efficiency depends on interaction
anol precipitation of DNA in microcentrifuge between protein and phospholipid bilayer.
tubes. Cold Spring Harb Protoc. doi:10.1101/ BMC Biotechnol 2:9
pdb.prot4456 30. Davidson EA, Dlugosz PJ, Levy M, Ellington
24. Shimizu Y, Inoue A, Tomari Y, Suzuki T, AD (2009) Directed evolution of proteins
Yokogawa T, Nishikawa K, Ueda T (2001) in vitro using compartmentalization in
Cell-free translation reconstituted with purified emulsions. Curr Protoc Mol Biol. doi:
components. Nat Biotechnol 19:752–755 10.1002/0471142727.mb2406s87
25. Zhou Y, Asahara H, Gaucher EA, Chong S 31. Lazzerini-Ospri L, Stano P, Luisi P, Marangoni
(2012) Reconstitution of translation from R (2012) Characterization of the emergent
Thermus thermophilus reveals a minimal set of properties of a synthetic quasi-cellular system.
components sufficient for protein synthesis at BMC Bioinformat 13(Suppl 4):S9
high temperatures and functional conservation 32. Nguyen AW, Daugherty PS (2005)
of modern and ancient translation compo- Evolutionary optimization of fluorescent pro-
nents. Nucleic Acids Res 40:7932–7945 teins for intracellular FRET. Nat Biotechnol
26. Forlin M, Lentini R, Mansy SS (2012) Cellular 23:355–360
imitations. Curr Opin Chem Biol 16:586–592 33. Donoso P, O'Neill SC, Dilly KW, Negretti N,
27. Ichihashi N, Matsuura T, Kita H, Sunami T, Eisner DA (1994) Comparison of the effects of
Suzuki H, Yomo T (2010) Constructing par- caffeine and other methylxanthines on [Ca2+]i
tial models of cells. Cold Spring Harb Perspect in rat ventricular myocytes. Br J Pharmacol
Biol. doi:10.1101/cshperspect.a004945 111:455–458
28. Bianco CD, Mansy SS (2012) Non replicating 34. Pereira de Souza T, Stano P, Luisi PL (2009) The
protocells. Acc Chem Res 45:2125–2130 minimal size of liposome-based model cells brings
29. Colletier J-P, Chaize B, Winterhalter M, about a remarkably enhanced entrapment and
Fournier D (2002) Protein encapsulation in protein synthesis. Chembiochem 10:1056–1063
Chapter 12
Rational Design of Artificial ON-Riboswitches

Atsushi Ogawa
Abstract
Riboswitches are composed of two regions: one for binding to the ligand (the aptamer domain) and the
other for regulating the expression of the gene (the expression platform). In most riboswitches (both natu-
ral and artificial), a part of the aptamer domain required for ligand binding is directly involved in the regu-
lation of expression, so that it is difficult to design other ligand-responsive riboswitches based on these
riboswitches even by using artificial aptamers obtained through in vitro selection. This chapter describes a
method for rationally constructing a foundational ON-riboswitch, which is easily available for the design
of other ligand-dependent riboswitches, by introducing a new region (a modulator sequence: MS) in addi-
tion to the two basic regions. A facile method for preparing arbitrary molecule-dependent riboswitches
based on the foundational riboswitch is also presented.
Key words Riboswitch, IRES, Aptamer, Gene regulation, Biosensor, Cell-free translation
1 Introduction
Riboswitches are ligand-dependent and cis-regulatory RNAs, dis-

covered in nature in 2002 [1, 2]. Many types of natural ribo-
switches have been identified in the untranslated regions (UTRs)
of bacterial mRNAs, and some are present in plants and fungi [3–
5]. The riboswitch consists of two regions: “an aptamer domain”
which specifically binds to its ligand, and “an expression platform”
whose conformational change induced by aptamer–ligand complex
formation directly switches ON or OFF transcriptional termina-
tion or translational initiation (in some cases, mRNA cleavage or
splicing). There is however no definite borderline between these
two regions because the aptamer domain generally includes a part
of the expression platform: a part of a transcriptional terminator
(e.g., a stem-loop and poly U tail) and/or the complementary
sequence thereof in transcription-regulating riboswitches; an anti-
sequence to a translational initiator (TI: the ribosome binding site
(RBS) and start codon in the case of bacteria) and/or the comple-
mentary sequence thereof in translation-regulating ones [3–5].
165
166 Atsushi Ogawa
Fig. 1 An example of natural riboswitches: add adenine ON-riboswitch in Vibrio vulnificus. In the absence of
the ligand (adenine), the aTI hybridizes to the TI (the RBS and the start codon) to prevent the bacterial 30S
ribosome from binding to the RBS (OFF state, above). In the presence of the ligand, it leaves the TI to bind to
the aaTI and the ligand (dashed thin line), so that the ribosome can access the RBS to start translation (ON
state, below). Bold and thin lines represent mRNA and base-pair interactions, respectively
Although this part of the aptamer domain is involved in expression

regulation in the absence of the ligand to promote or inhibit the
expression, it is required for ligand binding, so that aptamer–ligand
complex formation makes it leave the expression platform core to
inversely inhibit or promote the expression, respectively (in the
case of OFF-riboswitches or ON-riboswitches, respectively).
Taking a translation-regulating bacterial add ON-riboswitch as an
example (Fig. 1) [6], the anti-TI sequence (aTI) corresponding to
the part described above hybridizes to the TI to form an aTI–TI
duplex in the absence of the ligand (OFF state), while it releases
Rational Design of Riboswitches 167
the TI and then binds to another partially complementary sequence

(anti-aTI: aaTI) in the presence of the ligand (ON state), because
aTI–aaTI duplex formation is required for ligand–aptamer com-
plex formation, and also because a part of aTI that does not hybrid-
ize to the aaTI directly contributes to ligand binding. Briefly, the
ligand biases the equilibrium between these two states toward the
ON state to upregulate gene expression.
Although the ligands of natural riboswitches are metabolites
(thiamine pyrophosphate, s-adenosyl methionine, purine, etc.)
related to the regulated gene, an arbitrary ligand-dependent
riboswitch can be artificially constructed by using an in vitro-
selected aptamer (see Note 1). However, artificial riboswitches
cannot be obtained only by converting an aptamer in a natural
riboswitch into an in vitro-selected one, because a part of the
natural aptamer required for ligand binding plays a role as a part
of the expression platform in the absence of the ligand, as
described above. The most powerful methods for obtaining arti-
ficial riboswitches are in vivo screening methods, with which
some groups have actually obtained bacterial artificial ribo-
switches in response to ligands that bacteria do not originally
have (see Note 2). In typical screening methods, an in vitro-
selected aptamer and some randomized nucleotides are inserted
before the TI, and then only mRNAs with the riboswitch activity
are selected based on their expression efficiencies with and with-
out the ligand. Thus, these methods allow us to obtain efficient
translation-regulating riboswitches. Nonetheless, the artificial
riboswitches selected through these methods generally have the
same feature as the natural riboswitches: a part of the region
used for ligand binding is also used for directly regulating the
expression in the absence of the ligand. Therefore, these artificial
riboswitches, just like natural ones, are not available as foundations
for rationally constructing other ligand-dependent riboswitches.
On the other hand, I have recently established a method for
rationally designing translation-regulating riboswitches [7, 8].
In the design strategy in this method, a “variable” modulator
sequence (MS) is used, separately from the aptamer domain and
the expression platform, for regulating aTI–aaTI duplex formation
in response to the ligand. Thanks to this new element, aTI and
aaTI can be designed so as not be involved directly with ligand
binding, meaning that they can be fixed for any aptamer to be
shared with other ligand-dependent riboswitches. Thus, although
optimization of these sequences is required for obtaining the first
riboswitch, it is not necessary for the designs of subsequent ribo-
switches. Even the design of the MS variable for each aptamer
sequence needs only a simple calculation with a computer. In other
words, based on the first riboswitch, we can easily design arbitrary
artificial riboswitches with only sequence information of the cor-
responding in vitro-selected aptamer.
168 Atsushi Ogawa
For the optimization of the first riboswitch (i.e., the foundational

riboswitch), cell-free translation systems are very useful because we
can easily synthesize many types of mRNAs in vitro, whereas much
time and work is required to prepare various transformants, espe-
cially in eukaryotic cells, wherein we also need to consider the mat-
uration of the mRNAs (see Note 3). Although there are some
differences in translation conditions between in vitro and in vivo,
artificial riboswitches optimized in vitro are also expected to work
well in vivo, with minor corrections. In addition, artificial ribo-
switches are very useful even in vitro: for example, they can be used
as label-free biosensors and logic gate elements [7, 8].
This chapter describes, as an example of how to use the method
for constructing artificial riboswitches, the rational design of
eukaryotic ON-riboswitches modulating internal ribosome entry
site (IRES)-mediated translation (i.e., TI = IRES) in wheat germ
extract (eukaryotic cell-free translation system) [9] (see Note 4).
2 Materials
2.1 General 1. Nuclease-free water (see Note 5).

2. Various sizes of nuclease-free microtubes and filter tips.
2.2 Preparation 1. DNA template with the IRES sequence chosen as described in
of DNA Templates Subheading 3.1.1 (see Note 6).
2.2.1 Amplification 2. Plasmid encoding the luciferase gene (see Note 7).
of DNA Templates 3. PrimeSTAR MAX DNA polymerase including dNTPs (Takara Bio,
Shiga, Japan) or other high-fidelity DNA polymerase with dNTPs.
4. Primers for the upstream part: various forward primers includ-
ing a variety of regulator sequences (see Note 8) and a reverse
primer with a unique restriction enzyme site (RES).
5. Primers for the downstream part: a forward primer with the
same RES as in the reverse primer for the upstream part and a
reverse primer (see Note 9).
6. Forward primer including the T7 promoter (5′-TAATACGA
CTCACTATA-3′) for the final PCR.
2.2.2 Digestion 1. The restriction enzyme whose recognition site is included in

and Ligation the primers described above (see Note 10).
2. Ligation high Ver.2 (TOYOBO, Osaka, Japan) or other liga-
tion kit.
2.2.3 Agarose 1. 1.0 % agarose gel containing 1 mg/L ethidium bromide

Gel Purification (see Note 11).
2. Running buffer: 1× Tris/acetate/EDTA (TAE) prepared from
50× solution.
3. 1-kb DNA ladder marker.

4. Gel loading dye for agarose gel electrophoresis.
5. Disposable knives for cutting the gel.
6. illustra GFX PCR DNA and Gel Band Purification Kit (GE
Healthcare) or a similar spin column kit for DNA
purification.
2.3 In Vitro 1. MEGAscript T7 Kit (Invitrogen) (see Note 12).

Transcription 2. RNeasy MinElute Cleanup Kit (QIAGEN) (see Note 13).
2.4 In Vitro 1. WEPRO1240 Expression Kit (CellFree Sciences, Yokohama,

Translation Japan) (see Note 14). The following preparation should be
performed beforehand by using materials attached to the kit:
Add 125 μL each of S1, S2, S3, and S4 one by one to 500 μL
of nuclease-free water to prepare 1 mL of 5× SUB-AMIX, and
aliquot it into 30-μL aliquots. Dilute 20 μL of 20 mg/mL
creatine kinase with 380 μL of nuclease-free water to 1 mg/
mL, and aliquot it into 6-μL aliquots. Aliquot 1 mL of
WEPRO (wheat germ extract) into 30-μL aliquots. Store
them at less than −80 °C until use.
2. 12-strip PCR tubes.
3. Theophylline and other ligands.
2.5 Luciferase Assay 1. Luciferase Assay System (Promega). It is better to aliquot

10 mL of a Luciferase Assay Reagent (prepared with the sys-
tem) into 1-mL aliquots and store them at less than −80 °C
beforehand.
2. 96-well microplates (black).
3. 12-channel pipette.
2.6 Calculation of 1. RNAstructure software (http://rna.urmc.rochester.edu/

Delta G Values RNAstructure.html).
2.7 Equipment 1. Thermal cycler for PCR, digestion, ligation, transcription, and
translation.
2. Agarose gel electrophoresis apparatus.
3. UV transilluminator for excitation of ethidium bromide bound
to DNA.
4. Incubator for dissolving of the gel slices at 50–60 °C.
5. Deep freezer.
6. UV spectrometer.
7. Microplate luminometer.
8. Centrifuge.
170 Atsushi Ogawa
3 Methods
This methods section is divided into two subsections: one is for the
design of riboswitches, and the other is for the actual experiments,
such as the construction of DNA templates, transcription, transla-
tion, and reporter protein assays. Experimental procedures are
common in any design step, and thus see the experimental subsec-
tion (Subheading 3.2) for preparing mRNAs and measuring their
riboswitch activities in each design step.
3.1 Design Methods An IRES that has been much studied (especially regarding its sec-
ondary and tertiary structure) and that works with higher transla-
3.1.1 Choice of an IRES
tion efficiency is a better choice than a less-studied IRES. One of
and a Reporter Gene
the best candidates is the intergenic IRES of the Plautia stali intes-
Downstream of the IRES
tine virus (PSIV), which meets these conditions and functions in
many types of eukaryotic translation systems [10–12] (see Note
15). Other IRESs could be used depending on the user’s circum-
stances. For a reporter gene, the firefly luciferase gene is recom-
mended due to its low background, low detection limit, and linear
response over a wide range (see Note 16).
3.1.2 General Design The following parts are required for mRNAs (in this order): a stable
of mRNAs stem-loop (5′-SL), a mimic start codon (optional), regulators (vari-
able), the IRES (see Note 17), the reporter gene (see Note 18), and
a 3′ untranslated region (UTR) (Fig. 2). The stable stem-loop
structure in the 5′-terminus is required for inhibiting 5′ terminus-
mediated translation of the reporter gene [13] (see Note 19). An
8-base stem with an ACAAC pentaloop (e.g., 5′-GGGAGAC
CACAACGGUUUCCC-3′) is sufficient for inhibition, but it
should start from 5′ G for efficient transcription by T7 RNA poly-
merase. A mimic gene in the 5′ UTR is also usable for the inhibi-
tion, because the eukaryotic ribosome tends to start translation at
the start codon that it encounters first (see Note 20). Regulators are
a set of several sequences involved in regulation of gene expression:
a modulator sequence (MS), an anti-anti-IRES (aaIRES), an
aptamer, and an anti-IRES (aIRES), the details of which are
described below. More than 250 nucleotides (nt) of the 3′ UTR are
needed to suppress mRNA degradation from the 3′ end during
translation.
3.1.3 Optimization The first step in rationally designing artificial riboswitches is to

of the IRES Length optimize the IRES length toward higher translation efficiency by
using mRNAs without regulators. This is an optional but impor-
tant step because the translation efficiency of riboswitches in the
ON state should be less than that of their original mRNA. If the
sequence has been already optimized for translation, this step can
be skipped. Otherwise, it is recommended that some regulator-free
Fig. 2 General design of mRNAs. SL stem-loop, mAUG mimic start codon, UTR
untranslated region, PK pseudoknot, I-dORF a 5′ part of the original downstream
ORF of the chosen IRES. The existence or nonexistence of parts in parentheses
depends on the circumstances of the experiment. The eukaryotic 40S ribosome
preferably binds to the IRES (not to the 5′ terminus) due to the 5′-SL
mRNAs with various lengths of the IRES (including or excluding

a terminal part of the original upstream and/or downstream ORF:
I-uORF and/or I-dORF, respectively) be prepared, and these
mRNAs should be translated to determine the optimal IRES
length. It is also advisable to prepare mRNAs without IRES and/
or 5′-SL to verify whether the 5′-SL actually inhibits 5′ terminus-
mediated translation (see Note 21).
3.1.4 Inhibition of The next step is to inhibit IRES-mediated translation by inserting

IRES-Mediated Translation an aIRES, an anti-sequence complementary to a part of the IRES,
by an aIRES into the regulator region (Fig. 3a). This is a very crucial step,
because lower false expression of protein in the OFF state leads to
higher switching efficiencies. The procedure for determining an
effective aIRES is as follows:
1. Choose some parts of the IRES as aIRES targets. The best
candidates for aIRES targets are important sequences for the
IRES function or structure such as stem-loops and pseu-
doknots. If the structure of the chosen IRES is unknown, pick
up as many aIRES targets as possible. Regarding the aIRES
length (equal to the target length), 8 nt is empirically long
enough to hybridize to its target when they are sterically close
in the antiparallel direction.
2. Prepare mRNAs with various aIRES sequences and check their
translation efficiencies to determine the most effective aIRES
(see Note 22). If the translation efficiency of the champion is
much higher than that of the IRES-free mRNA (see
Subheading 3.1.3), try other aIRES targets or other IRES.
3. Prepare mRNAs with various lengths of the best aIRES to
investigate the minimum length for efficiently inhibiting
IRES-mediated translation as if there are no IRES and the
172 Atsushi Ogawa
Fig. 3 Regulation of IRES-mediated translation by antisense sequences.

The aIRES disrupts the IRES conformation by hybridizing to a part of the IRES (a).
The aaIRES binds to the aIRES to restore the disrupted conformation to the active
state (b)
maximum length for hardly affecting IRES-mediated transla-

tion (see Note 23). The latter information is required for
designing riboswitches later.
3.1.5 Restoration of To rationally construct ON-riboswitches, a part for restoring

IRES-Mediated Translation IRES-mediated translation inhibited by the aIRES is required
by an aaIRES independently from an aptamer (Fig. 3b). It is much easier to iden-
tify an effective part: an aaIRES, an anti-sequence complementary
to aIRES. Nonetheless, it is recommended to confirm whether
IRES-mediated translation is restored by actually inserting the
aaIRES and some extra bases as a loop (e.g., 5′-ACCAC-3′) before
the aIRES. The same length of the aaIRES as that of the aIRES
should sufficiently restore translation (see Notes 24 and 25).
Fig. 4 Design of IRES-based ON-riboswitches. In the absence of the ligand, since the MS hybridizes to the 3’
end of the aaIRES (and the 5’ end of the aptamer), the aIRES binds to the IRES to inhibit IRES-mediated transla-
tion (OFF state, above). In the presence of the ligand, it releases the aptamer (due to aptamer–ligand complex
formation) and then leaves the aaIRES (due to destabilization of the MS-SL), so that the aIRES binds not to the
IRES but to the aaIRES to promote IRES-mediated translation (ON state, below)
3.1.6 Design and To construct the foundational IRES-based ON-riboswitch, which is

Optimization of the necessary and sufficient for designing various ligand–riboswitch
Foundational IRES-Based pairs, another two regulator parts in addition to the aIRES and the
ON-Riboswitch aaIRES are required (Fig. 4). One is of course an aptamer,
which is inserted between the aaIRES and the aIRES instead of
the intervening loop described above. A theophylline aptamer
(5′-AUACCAGCCGAAAGGCCCUUGGCAG-3′) is recommended
for the foundational riboswitch because it is well minimized and strongly
binds to the ligand (theophylline, Kd = 0.1 μM) (see Note 26) [14].
174 Atsushi Ogawa
The other is a modulator sequence (MS), which is inserted before the

aaIRES for ligand-dependently modulating aaIRES/aIRES duplex for-
mation. The MS is composed of two parts: a 5′ part that is complemen-
tary to the 5′ end of the aptamer (i.e., ….GUAU for the theophylline
aptamer), and a 3′ part that is complementary to the 3′ end of the
aaIRES. Since the 3′ part is the same sequence as the 5′ region of the
aIRES, it is automatically determined to be the 3’-shortened aIRES
with the pre-examined maximum length for hardly affecting the active
IRES structure (see Subheading 3.1.4) (see Note 27). Incidentally, it is
absolutely shorter than the aaIRES, so that the 5′ region of the aaIRES
should be a loop for a stem-loop (MS-SL) composed of the MS, the
aaIRES, and part of the aptamer bound to the MS (see Note 28). As for
the 5′ part of the MS, it is required to optimize the length toward
higher switching efficiency.
Therefore, all that one has to do in actual experiments is prepare
mRNAs with various lengths of the 5′ part of the MS and then mea-
sure their translation efficiencies without or with the ligand (see Note
29). Their switching efficiencies can be evaluated by the ratio of
translation efficiencies in the ON state to that in the OFF state (see
Note 30). The best mRNA with the highest switching efficiency
should be available as the foundational riboswitch (see Note 31).
3.1.7 Rational Design of Since the MS-SL in the foundational riboswitch with the highest
Other Ligand-Dependent switching efficiency has the almost-minimum free energy necessary
ON-Riboswitches for inhibiting IRES-mediated translation in the OFF state, any
riboswitch whose MS-SL has similar stability is expected to work
well as a riboswitch. Therefore, arbitrary ligand-dependent
ON-riboswitches can be easily designed as follows:
1. Calculate the free energy (delta G value) of the cropped MS-SL
part in the foundational riboswitch by using the RNAstructure
software [15].
2. On a computer, replace the aptamer and the 5′ part of the MS
in the foundational riboswitch with an aptamer for the ligand
of interest and various lengths of the sequence complementary
to the 5′ end of the chosen aptamer, respectively (see Note 26).
3. Calculate free energies of their MS-SL part and determine
which has the nearest value to the free energy of the MS-SL in
the foundational riboswitch (see Note 32).
4. Prepare actual mRNA with the most promising MS-SL (and
mRNAs with the second and third most promising MS-SL if
necessary), and confirm its switching efficiency in the presence
or absence of the ligand (see Notes 29 and 31).
3.2 Experimental DNA templates are constructed by preparing the upstream part
Methods including the IRES and the downstream part including the
luciferase gene separately and then joining them together
3.2.1 Construction of
(Fig. 5) (see Note 33).
DNA Templates for mRNA
Fig. 5 General procedures of the construction of DNA templates and their transcripts (mRNA)
1. PCR-amplify the upstream part including a 5′-SL, a mimic

start codon (optional), regulators, the IRES, and a unique
restriction enzyme site (RES) in a 20-μL reaction volume (see
Notes 8 and 34).
2. PCR-amplify the downstream part including the RES, the
luciferase gene (without the original start codon), and a 3′
UTR (see Note 35) in a 20-μL reaction volume (see Note 34).
3. Mix both parts and purify the mixture with a commercially
available spin column kit (see Notes 36 and 37).
4. Digest the purified DNAs with the restriction enzyme in a
20-μL reaction volume (see Note 34).
5. Purify the digested DNAs with the spin column (see Note 36)
or denaturation of the restriction enzyme with incubation at a
slightly high temperature (see Note 34).
6. Incubate a mixture of 20 μL of the resultant product and
10 μL of Ligation high Ver.2 at 16 °C for 30 min.
176 Atsushi Ogawa
7. Mix 8 μL of the ligation solution with a proper amount of gel

loading dye, and load the mixture on an agarose gel
(see Subheading 2.2.3, item 1).
8. Run gel electrophoresis for more than 30 min at 100 V and cut off
the band corresponding to the fused DNA between the upstream
and downstream part with a disposable knife (see Note 38)
9. Purify the fused DNA with the spin column (see Note 39).
10. PCR-amplify the gel-purified DNA with a forward primer
including the T7 promoter in a 20-μL reaction volume (final
PCR) (see Note 40).
11. Store the PCR products at less than −20 °C until use.
3.2.2 Preparation of The following protocol is for the preparation of 5′ cap- and 3′
mRNA (In Vitro poly(A)-free mRNAs. Both the 5′ cap and poyl(A) are not necessarily
Transcription) required for both efficient 5′ terminus- and IRES-mediated transla-
tion in the wheat germ extract described here (see Notes 19 and 35)
1. Transcribe the DNA templates with the MEGAscript T7 Kit
according to the manufacturer’s protocol. Typically, incubate
10 µL of the reaction solution containing 2–4 µL of the final
PCR products at 37 °C for 2–4 h.
2. Add 0.5 µL of DNase attached to the kit to the solution, and
then further incubate the mixture at 37 °C for 15 min.
3. Purify the transcribed mRNA with the RNeasy MinElute
Cleanup Kit. To concentrate mRNA sufficiently, use 14 μL of
nuclease-free water to elute mRNA in the final step. Typically,
approx. 50 μg of mRNA can be obtained.
4. Quantify the purified mRNA with a UV spectrometer and
adjust the concentration to 3 pmol/µL with nuclease-free
water.
5. Aliquot the mRNA into smaller volumes (e.g., 3.4 μL for three
times’ translation) and store them at less than −80 °C until use.
3.2.3 Cell-Free As many translation reactions as possible should be performed

Translation simultaneously to compare many types of mRNAs with the
regulator-free optimal mRNA (see Subheading 3.1.3), to measure
error bars, and to effectively use the materials in the cell-free trans-
lation kit. The following protocol is for 12 reactions. Note that the
protocol is different from that of the manufacturer.
1. Thaw 30 µL of 5× SUB-AMIX, 6 µL of 1 mg/mL creatine
kinase, and 30 µL of WEPRO prepared as described in
Subheading 2.4 and mRNA(s) prepared as described in
Subheading 3.2.2 on ice.
2. Put 1 µL of mRNA on the inner side surface of each PCR tube
(in a total of 12 tubes).
3. Put 1 µL of a ligand or nuclease-free water on a site next to the

mRNA spot so as not to mix them.
4. Mix 52.2 µL of nuclease-free water, 29 µL of WEPRO, 29 µL
of 5× SUB-AMIX, and 5.8 µL of 1 mg/mL creatine kinase.
5. Put 8 µL of the mixture on the site next to the ligand (or
water) spot so as not to mix them.
6. Centrifuge (lightly) the tubes simultaneously and mix them
thoroughly, and then centrifuge (lightly) them again.
7. Incubate the tubes at 26 °C for 1 h and then at 4 °C to stop
the translation.
3.2.4 Luciferase Assay To use the Luciferase Assay Reagent prepared as described in
Subheading 2.5 immediately after translation, start to thaw the ali-
quot on ice while protecting it from light right after starting the
translation in Subheading 3.2.3, step 7.
1. Add 100 µL of water to 10 µL of each translation solution to
dilute them 11-fold.
2. Put 5 µL of each diluted solution in a row (12 wells) of a
96-well microplate.
3. Aliquot 1 mL of the Luciferase Assay Reagent into 12 aliquots
(~80 µL) in new PCR tubes.
4. Add 75 µL of the reagent to each diluted translation solution
in the plate by using a 12-channel pipette, and mix them well
by pipetting three times.
5. Immediately after mixing, measure the luminescence values of
the mixtures with a luminometer.
4 Notes
1. See previous chapters for descriptions of the in vitro selection

of aptamers.
2. See previous chapters for descriptions of the in vivo screening
of artificial riboswitches.
3. Nonetheless, the design strategy described here (i.e., optimi-
zation of the foundational riboswitch and construction of
other riboswitches in reference to the foundational riboswitch)
should also be available in vivo.
4. An IRES functions as a receiver of the ribosome in eukaryotes
as well as an RBS in bacteria, so that bacterial riboswitches can
be constructed by reading “IRES” (precisely, a part targeted
by the aIRES) as “RBS (and the start codon)”. For the ratio-
nal design of OFF-riboswitches, refer also to my previous
report [8].
178 Atsushi Ogawa
5. Since a slight amount of contaminated nucleases and proteases

can degrade nucleic acids and proteins, respectively, nuclease-
free water should be used in all reactions (PCR, digestion,
ligation, transcription, and translation). In addition, the exper-
imenter should wear clean gloves and a mask to prevent nucle-
ase and protease contamination.
6. Commercial gene synthesis services are available for obtaining
the template.
7. Some plasmids encoding the luciferase gene are obtainable
from Promega.
8. If the forward primer is too long for PCR (more than 100 nt),
split it and perform subsequent PCRs.
9. This reverse primer is also used as the reverse primer in the
final PCR.
10. Various types of restriction enzymes are available for purchase
from New England Biolabs or Takara Bio.
11. Care should be taken when handling ethidium bromide, which
is thought to act as a mutagen.
12. A similar transcription kit or a hand-made transcription reac-
tion solution can be used instead.
13. Although mRNA can be also purified by phenol-chloroform
extraction and ethanol (or isopropanol) precipitation at low
cost, the use of the purification kit should give mRNAs with
higher purities.
14. For a eukaryotic cell-free translation system, the WEPRO1240
Expression Kit is recommended because the translation system
is very stable, like in vivo, and because the 5′ capping of
mRNAs is not necessary for efficient 5′ terminus-mediated
translation, although another eukaryotic translation system or
hand-made extract can be used instead.
15. The foundational riboswitch sequence for PSIV-IRES is
obtainable from my previous report [7].
16. Other reporter genes such as GFP and beta-galactosidase may
be available, but the assay methods for these proteins are
described elsewhere.
17. The IRES should include an I-dORF (a 5′ part of the original
downstream open reading frame), which generally increases
the IRES-mediated translation efficiency (see Subheading 3.1.3).
18. The start codon of the reporter gene can be removed because
the I-dORF is connected to the reporter gene as a part of the
ORF.
19. Even 5′ cap-free mRNA can be translated via 5′ terminus-
mediated translation in the wheat germ extract presented here,
depending on the 5′ terminal sequence of mRNA [13].
20. It should be noted that the ribosome tends to skip the start
codon that is too near to the 5′ terminus. To avoid this prob-
lem, a mimic start codon can also be inserted between regula-
tors or just before the IRES. The chosen IRES may originally
have a start codon(s) available as the mimic codon(s).
Incidentally, the PSIV-IRES has two mimic genes in itself.
Moreover, IRES-mediated translation via the PSIV-IRES can
start at a non-AUG codon, so that 5′ terminus-mediated
translation of the reporter gene should be completely sup-
pressed even if the ribosome skips the mimic start codons or
remains on mRNA after translation of the mimic genes. In this
case, the start codon of reporter gene must absolutely be
deleted (see Note 18).
21. If the inhibition effect is inadequate, strengthen the stem of
the 5′-SL by extending the length, or insert a mimic start
codon(s).
22. In the case of the PSIV-IRES, an aIRES to a 5′ side of the
pseudoknot III (PK-III) is the best [7].
23. In the latter case, aIRESs with the 3′ side shortened (or prefer-
ably replaced with the same length of nucleotides uninvolved
with the aIRES/IRES duplex) should be investigated.
Incidentally, for the PSIV-IRES, the former and the latter are
8 nt and 4 nt, respectively [7].
24. If the restoration efficiency is much lower, longer aaIRES may
work well, but a too-long aaIRES would lead to lower ON/
OFF switching efficiencies.
25. In terms of the PSIV-IRES, an 8-mer aaIRES for the comple-
mentary 8-mer aIRES shows an 80 % recovery of IRES-
mediated translation.
26. If an aptamer other than the theophylline aptamer is used, it
must be minimized beforehand, because it is required that the
ligand–aptamer interaction induces a release of the MS from
the aptamer.
27. Longer ones inhibit IRES-mediated translation even when the
aaIRES releases the MS and hybridizes to the aIRES in the
ON state, whereas shorter ones cannot prevent aaIRES/
aIRES duplex formation even in the OFF state.
28. If this loop length is very short (less than 3 or 4 nt), some
bases can be inserted between the MS and the aaIRES to sta-
bilize the stem-loop structure.
29. Use the maximum concentration of the ligand that does not
affect IRES-mediated translation, which should be investi-
gated beforehand with the regulator-free optimal mRNA
(see Subheading 3.1.3). To examine the ligand dependency,
use various concentrations of the ligand.
180 Atsushi Ogawa
30. Generally, the optimal length of the 5′ part of MS is the maxi-

mum length for satisfactorily inhibiting IRES-mediated trans-
lation in the OFF state.
31. If the highest switching efficiency is lower than 5 at the maxi-
mum concentration of the ligand (see Note 29), it is likely that
the aptamer sequence inhibits IRES-mediated translation or
the ligand–aptamer interaction is too weak at the maximum
ligand concentration. In this case, it is recommended that the
ligand–aptamer pair or the IRES be changed. Incidentally, the
switching efficiency of the foundational PSIV-IRES-based
riboswitch is approx. 10 at 1 mM theophylline [7].
32. To precisely adjust the free energy, it might be possible to
include a G–U wobble pair in the MS–aptamer duplex.
33. A “unique” restriction enzyme site (RES) is required for join-
ing these two parts. Since it remains in the fused ORF, it must
be inserted in order to keep the translational frame.
34. Refer to the manufacturers’ protocol for reaction conditions.
35. Although the 3′ UTR that the luciferase-encoding plasmid
originally has could be used as-is, the length should be more
than 250 nt for efficient translation.
36. Elute the PCR products with 20 μL of nuclease-free water.
37. In cases in which the luciferase-encoding plasmid has the T7
promoter, DpnI treatment or agarose gel purification is needed
to remove the plasmid. Otherwise, the plasmid sequence may
be amplified in the final PCR.
38. In cases in which the restriction enzyme recognizes a
palindromic sequence, five bands should be observed: two
reactants (the upstream part and the downstream part) and
three ligated products between two upstream parts, two
downstream parts, or each part. To cut off only the band cor-
responding to the ligated product between the upstream and
downstream part without contamination of others, separate
these bands completely with long electrophoresis. If this
operation is difficult, use restriction enzymes that recognize
asymmetric sequences.
39. Elute the ligated product with 50 μL of nuclease-free water
and then dilute it 10- to 100-fold with nuclease-free water, to
be used as the template in the final PCR.
40. Make sure that only a single clear band is satisfactorily ampli-
fied on an agarose gel. Incidentally, purification after the final
PCR is not necessary.
Acknowledgments
This work was supported by ‘Special Coordination Funds for

Promoting Science and Technology’ and ‘Grant-in-Aid for
Scientific Research on Innovative Areas’ from the Ministry of
Education, Culture, Sports, Science and Technology, Japan.
References
1. Winkler W, Nahvi A, Breaker RR (2002) 9. Madin K, Sawasaki T, Ogasawara T, Endo Y
Thiamine derivatives bind messenger RNAs (2000) A highly efficient and robust cell-free
directly to regulate bacterial gene expression. protein synthesis system prepared from wheat
Nature 419:952–956 embryos: plants apparently contain a suicide
2. Mironov AS, Gusarov I, Rafikov R et al (2002) system directed at ribosomes. Proc Natl Acad
Sensing small molecules by nascent RNA: a Sci U S A 97:559–564
mechanism to control transcription in bacteria. 10. Sasaki J, Nakashima N (1999) Translation initia-
Cell 111:747–756 tion at the CUU codon in mediated by the
3. Blount KF, Breaker RR (2006) Riboswitches as internal ribosome entry site of an insect picorna-
antibacterial drug targets. Nat Biotechnol like virus in vitro. J Virol 73:1219–1226
24:1558–1564 11. Shibuya N, Nishiyama T, Kanamori Y et al
4. Nudler E, Mironov AS (2004) The riboswitch (2003) Conditional rather than absolute
control of bacterial metabolism. Trends requirements of the capsid coding sequence for
Biochem Sci 29:11–17 initiation of methionine-independent transla-
5. Schwalbe H, Buck J, Fürtig B et al (2007) tion in Plautia stali intestine virus. J Virol
Structures of RNA switches: insight into 77:12002–12010
molecular recognition and tertiary structure. 12. Pfingsten JS, Costantino DA, Kieft JS (2006)
Angew Chem Int Ed 46:1212–1219 Structural basis for ribosome recruitment and
6. Serganov A, Yuan Y-R, Pikovskaya O et al manipulation by a viral IRES RNA. Science
(2004) Structural basis for discriminative regu- 314:1450–1454
lation of gene expression by adenine- and 13. Ogawa A (2009) Biofunction-assisted sensors
guanine-sensing mRNAs. Chem Biol based on a new method for converting apta-
11:1729–1741 zyme activity into reporter protein expression
7. Ogawa A (2011) Rational design of artificial with high efficiency in wheat germ extract.
riboswitches based on ligand-dependent mod- ChemBioChem 10:2465–2468
ulation of internal ribosome entry in wheat 14. Jenison RD, Gill SC, Pardi A, Polisky B (1994)
germ extract and their applications as label-free High-resolution molecular discrimination by
biosensors. RNA 17:478–488 RNA. Science 263:1425–1429
8. Ogawa A (2011) Rational construction of 15. Mathews DH, Disney MD, Childs JL et al
eukaryotic OFF-riboswitches that downregu- (2004) Incorporating chemical modification
late internal ribosome entry site-mediated constraints into a dynamic programming algo-
translation in response to their ligands. Bioorg rithm for prediction of RNA secondary structure.
Med Chem Lett 22:1639–1642 Proc Natl Acad Sci U S A 101:7287–7292
Chapter 13
Engineering Protein-Responsive mRNA Switch

in Mammalian Cells
Kei Endo and Hirohide Saito
Abstract
Engineering of translation provides an alternative regulatory layer for controlling transgene expression in
addition to transcriptional regulation. Synthetic mRNA switches that modulate translation of a target gene
of interest in response to an intracellular protein could be a key regulator to construct a genetic circuit.
Insertion of a protein binding RNA sequence in the 5′ UTR of mRNA would allow for the generation of
a protein-responsive RNA switch. Here we describe the design principle of the switch and methods for
tuning and analyzing its translational activity in mammalian cells.
Key words Riboswitch, Mammalian cell, Transfection, Reporter assay, RNA–protein interaction,
RNP, Synthetic biology, Translation, RNA module
1 Introduction
The function of riboswitches, monitoring intracellular conditions

and controlling synthesis of a specific protein, has attracted increas-
ing attention for their potential as synthetic regulators of various
biological processes [1–3]. In particular, synthetic switches in ani-
mals could be applied to various fields, such as medicine, food-
stuffs, and the environment [4–6]. Protein-responsive mRNA
switches control translation of an output protein in response to an
input protein expressed in the target cells. They provide an alterna-
tive regulatory layer of transgene expression in addition to tran-
scriptional regulation, and combination of these two layers has
successfully produced sophisticated gene circuits [7]. Ideally,
parameters of switches should be adjusted to desired levels. Thus,
the design principle of a protein-responsive RNA switch that
enables the level of the output protein to be tuned is increasingly
important.
In principle, translation in eukaryotic cells is initiated at the 5′
terminus of an mRNA. The 5′-cap-binding complex recruits the
183
184 Kei Endo and Hirohide Saito
pre-initiation complex containing the small ribosomal subunit,

followed by scanning to the first AUG codon and assembly of the
large ribosomal subunit [8]. This complicated, multistep process is
one of the most common targets of translational control [9]. Iron
regulatory protein (IRP) is a well-studied example of a transla-
tional regulatory protein [10]. It binds to the specific sequence
(iron-responsive element, IRE) in the 5′ UTR of mRNA encoding
the iron-storage protein, Ferritin, and represses synthesis of the
protein to maintain a concentration of Fe(III) in a cell. It is pro-
posed that IRP does not interfere with a specific function of trans-
lational apparatus nor does it recruit another regulatory protein,
but blocks translation based on steric hindrance [10]. Accordingly,
replacement of a specific RNA–protein (RNP) complex of IRP and
IRE with another pair produced synthetic protein-responsive
switches [11–14]. In addition to RNA–protein complexes, small
molecules that bind an artificial motif inserted into the 5′ UTR can
also block protein synthesis [15]. This is reminiscent of natural
riboswitches found in bacteria.
This chapter describes details in constructing, tuning, and
evaluating a protein-responsive RNA switch in mammalian cul-
tured cells. Furthermore, short introductions to the uses of the
switches are also provided.
2 Materials
All reagents are obtained from Life Technologies or otherwise

indicated in the text.
2.1 Plasmid 1. Expression vectors for mammalian cells, such as pIRES

Construction Bicistronic Expression Vector (Clontech) and pcDNA
Mammalian Expression Vector.
2. Oligodeoxynucleotides as primers to amplify DNA fragments
or a DNA fragment to be inserted into a plasmid vector.
3. Annealing buffer: 20 mM Tris–HCl, pH 7.6, 100 mM NaCl.
4. A thermal cycler and appropriate test tubes.
5. Enzymes for plasmid construction (ligase, alkaline phosphatase
and appropriate restriction enzymes). We use Ligation High
ver.2 (TOYOBO) and rAPid Alkaline Phosphatase (Roche).
6. Reagents for ethanol precipitation: Ethanol, 70 % ethanol, and
3 M potassium acetate, pH 5.2.
7. Competent E. coli cells, LB medium, and appropriate antibiot-
ics for transformation and prepping plasmids.
8. A kit for purification of PCR products. We use MinElute PCR
purification kit (QIAGEN).
Protein-Responsive mRNA Switch 185
2.2 Cell Culture 1. A kit for purification of transfection-grade plasmid DNA. In our
and DNA Transfection case, we use Plasmid Midi Kit (QIAGEN).
2. Mammalian cultured cell lines. We use HeLa cells in this
protocol.
3. Dulbecco’s modified Eagle medium (DMEM) supplemented
with 10 % fetal bovine serum and 1 % antibiotic antimycotic
solution (Sigma-Aldrich).
4. Phosphate buffered saline (PBS).
5. 0.25 % Trypsin–EDTA.
6. Equipment for cell counting. We use Countess Automated
Cell Counter.
7. Opti-MEM.
8. Lipofectamine 2000.
2.3 Flow Cytometry 1. Polystyrene tube with 35 μm cell strainer cap (BD Biosciences).
Analysis 2. A flow cytometer, such as BD Accuri C6 (BD Biosciences) and
FACSAria (BD Biosciences).
2.4 Apoptosis Assay Following reagents are also provided as a package (Pacific Blue
AnnexinV/SYTOX AADvanced Apoptosis kit, for flow cytometry).
1. Pacific Blue-conjugated Annexin V.
2. Annexin V binding buffer.
3. SYTOX AADvance.
3 Methods
3.1 Modular A protein-responsive mRNA OFF switch consists on an aptamer

Expression Vector for domain, which binds to a specific RNA-binding protein as an
Protein-Responsive input, and the open reading frame of an output protein. A newly
OFF Switch synthesized switch needs to be constructed in a mammalian expres-
sion vector to assess its activity in a cell. Changes not only in the
sequence of the aptamer but in configuration of the 5′ UTR mod-
ulate the activity of the switch (see Subheading 3.2). Thus, the
modular expression vector could help efficient construction, com-
parison and screening of new synthetic switches. This section pro-
vides information on the modular design of an expression vector
for switches (switch vector), and on a protocol for plasmid con-
struction when using it. An expression vector for an input protein
(trigger vector) is also described because it will be required to
evaluate the activity of the switch (see Note 1).
3.1.1 Switch Vector Figure 1b shows the modular design of a switch vector. The 5′
Expressing OFF Switch UTR of the switch has four specific sites of restriction enzymes.
mRNA Nhe I is placed just downstream of the CMV promoter and Age I
Fig. 1 Genetic configuration of plasmids expressing a protein-responsive RNA

switch and its trigger protein. (a) Schematic illustration of a protein-responsive
RNA switch. An input protein binds to an aptamer sequence placed in the 5′ UTR
of the switch and blocks translation of a gene of interest (GOI), which is the out-
put of the switch. (b) Configuration of a switch vector. (c) A fragment of an
aptamer to be inserted into the switch vector. BamH I and Age I are used to
replace an aptamer, and Bgl II and Age I are used to add an aptamer. (d)
Configuration of a trigger vector
is located upstream of the ORF. A sequence of an aptamer is

inserted between them with the BamH I and Bgl II sites at the 5′
end and 3′ end, respectively. As a result, the 5′ UTR is divided into
three parts; upstream of an aptamer (Nhe I/BamH I), the aptamer
(BamH I/Bgl II), and downstream of the aptamer (Bgl II/Age I).
In most cases, the length of an aptamer domain is relatively short
(<100 nucleotides), so here we show a protocol to insert a short
dsDNA derived from two synthetic oligodeoxynucleotides (oligo
DNAs) into the aptamer part in the switch vector (see Note 2).
In principle, plasmids are constructed according to general methods.
1. Design two oligo DNAs forming dsDNA as shown in Fig. 1c.
After annealing two oligo DNAs, the resulting dsDNA should
contain 5′-GATC and 5′-CCGG overhangs up- and down-
stream of the aptamer. This configuration is the same as a
product of BamH I and Age I digestion.
2. Mix 1 μL each of two oligo DNAs (100 μM) with 38 μL of
annealing buffer (see Note 3) in a PCR tube (f.c. 2.5 μM each).
3. Denature oligo DNAs by heating at 96 °C for 30 s and anneal
them by gradual cooling to 10 °C under the melting tempera-
ture of the annealing sequence using a thermal cycler.
4. Dilute the resulting dsDNA by 100-fold in DW (f.c. 25 nM).
5. Digest 250 ng of the switch vector (plasmid) with BamH I and

Age I. Do not add alkaline phosphatase because synthesized
oligo DNAs do not have 5′ phosphate groups.
6. Purify the digested plasmid by ethanol precipitation.
7. Resolve/resuspend the pellet in 15 μL of DW.
8. Mix 1 μL aliquot of the diluted dsDNA and 1.5 μL aliquot of
the resolved plasmid.
9. Incubate a mixture at 42 °C for 3 min to promote interaction
between the dsDNA and the plasmid, and cool it on ice.
10. Add 2.5 μL of Ligation High, which is a premix of ligation
reaction by T4 DNA ligase.
11. Incubate at 16 °C for 45 min reaction.
12. Transform E. coli with the product.
3.1.2 Trigger Vector The level of a transgene transiently expressed from a plasmid varies
Expressing an Input Protein among the transfected cells. Therefore, an indicator for transfec-
tion efficiency must be required to evaluate the activity of a syn-
thetic switch. To monitor the efficiency, a reporter protein
expressed by IRES in the same mRNA as the input protein can be
used as a control (Fig. 1d), because the ratio of gene expression
from two co-transfected plasmids is highly correlated in a popula-
tion (see Note 4). The trigger vector is derived from pIRES
Bicistronic Expression Vector.
3.2 Design and Both the position and the number of the aptamer motifs on the 5′
Tuning of OFF Switch UTR of mRNA are critical elements to control the activity of a
protein-responsive RNA switch in mammalian cells. Shortening
the distance from the 5′ end of mRNA to the aptamer and repeat-
ing the aptamer sequences in the 5′ UTR enhance the activity of
the OFF switch by the input protein [16]. In other words, engi-
neering mRNAs by modifying these features can tune the activity
of the switch composed of the same RNP pair (Fig. 2). This section
provides a strategy for tuning protein-responsive switches.
3.2.1 Spacing Between An aptamer located closer to the 5′ terminus of mRNA has more
the RNA Motif and the impact on repressing translation. The activity of the OFF switch
5′ End decreases as the distance of the aptamer from the 5′ end increases
[16]. Thus, spacing them could modify the activity in a continuous
manner. It is notable that a spacer sequence should not include
ATG triplets, because it is going to be inserted upstream of the
ORF. The following protocol describes a procedure for insertion
of a spacer DNA fragment upstream of an aptamer (Nhe I/BamH I)
in the switch vector (Fig. 1b). It is also possible to reduce the dis-
tance via PCR-based deletion methods [17, 18].
Fig. 2 Tuning the activity of the switch by modification of the copy number and
the position of the aptamer. (a) Schematic illustration of tuning strategy.
Increasing the distance of the aptamer from the 5′ end (l) and the number of the
aptamer (n) in the switch continuously decrease and discretely increase the
activity of the switch, respectively. (b) Tuning of the L7Ae-responsive switch
composed of a K-loop RNA motif (Kl) as an aptamer. Shown is the translational
efficiency (see also Subheading 3.3.4) of switches that contains one to four
times repeated Kl, the first one of which was placed at 18th, 67th, 120th, and
164th nucleotides (nt; l) from the 5′ end
1. Design a spacer sequence absent of ATG triplets (see Note 5).

2. Design a primer set to amplify the spacer. The 5′ end of the
forward and reverse primer should contain Nhe I and BamH I
restriction sites, respectively.
3. Amplify the spacer via PCR.
4. Add Dpn I to remove methylated plasmid DNA.
6. Purify the PCR product (e.g., QIAGEN, MinElute PCR puri-
fication kit).
7. Digest 200 ng of the DNA fragment with Nhe I and BamH I.
8. Digest 250 ng of the switch vector with Nhe I and BamH I,
together with alkaline phosphatase.
9. Purify the digested PCR fragment and plasmid, ligate them, and
transform E. coli with it according to commonly used methods.
3.2.2 Multiplying Tandem insertion of multiple RNA motifs into the 5′ UTR of
the RNA Motifs mRNA increases the activity of the OFF switch in a step-wise man-
ner. A DNA fragment, which contains multiple motifs, can be used
to replace an aptamer in a switch vector. Alternatively, it is possible
to reuse oligo DNAs described earlier (Subheading 3.2, Fig. 1c) if

a switch vector is digested with Bgl II and Age I (Fig. 1b). In this
case, the double-stranded oligo DNA is inserted downstream of
the slotted aptamer using compatible ends produced by BamH I
and Bgl II. The resulting plasmid has the same configuration as the
parental vector (a repeated aptamer sequence placed between
unique BamH I and Blg II sites) and enables repeating the same
procedure to concatenate aptamers, because neither BamH I nor
Blg II can cleave the hybrid sequence.
3.3 Assessment of a 1. Prepare transfection-grade plasmid DNAs (switch vectors and

Behavior of the Switch trigger vectors) by purification on anion-exchange resin
in Cultured according to the manufacturer’s protocol (e.g., QIAGEN,
Mammalian Cells Plasmid Midi Kit).
3.3.1 Preparation of the 2. Seed HeLa cells to a 24-well plate 24 h prior to transfection
Samples (see Note 6), as follows:
(a) Aspirate medium from a Petri dish containing HeLa cells
at 80 % confluence.
(b) Add 5 mL of PBS, wash cells and aspirate PBS.
(c) Add 1 mL of Trypsin–EDTA and incubate at 37 °C for
3 min.
(d) Add 9 mL of DMEM medium and detach cells by gently
pipetting up and down.
(e) Transfer cells to a 15-mL conical tube.
(f) Determine cell density using a cell counter.
(g) Dilute cells to a concentration of 100,000 cells/mL, and
seed 500 μL of cell suspension per well of a 24-well format
plate.
3.3.2 Cotransfection of 1. Mix 120 ng of a switch vector and 600 ng of a trigger vector
Plasmids into HeLa Cells in 60 μL of Opti-MEM.
2. Add 1.2 μL of Lipofectamine 2000 to 58.8 μL of
Opti-MEM.
3. Mix the two solutions above and incubate at room tempera-
ture for 20 min.
4. Apply 100 μL of the mixture per well and gently shake the
plate to mix.
5. Incubate cells at 37 °C for 4 h.
6. Replace the medium by aspiration and add 500 μL of fresh
DMEM containing serum and antibiotics.
3.3.3 Quantification of 1. Aspirate medium from the wells.

Reporter Fluorescent 2. Add 500 μL of PBS, wash cells and aspirate PBS.
Proteins Using a Flow
3. Add 100 μL of Trypsin–EDTA and incubate at 37 °C for 3 min.
Cytometer
a b c
Fig. 3 Plots from flow cytometry analysis. (a) Raw data of cells expressing either EGFP (upper; reporter) or
DsRed (lower; control) used for the fluorescence compensation (gray). Mock (red/dark gray) indicates untrans-
fected cells. Fluorescence of EGFP and DsRed is detected in GFP and PerCP channel, respectively. Fitting curve
and obtained coefficient are also shown. (b) Plots of compensated signals from the same experiments as
shown in (a). (c) Plots of cells expressing L7Ae-responsive RNA switch that outputs EGFP. DsRed was used as
a control. The upper and lower panel shows the output in the absence (w/o input) and presence (w/input) of
L7Ae, respectively
4. Add 200 μL of medium and gently rock the dish.

5. Immediately before measurement, gently mix cell suspension
by pipetting and pass cells through nylon mesh. We use 35 μm
cell strainer caps.
6. Measure fluorescence using a flow cytometer.
3.3.4 Post-measurement A broad emission spectrum of one fluorescent protein sometimes

Compensation of results in the signal spilling over into measurement of another
Fluorescent Signals (see Note 7). If necessary, perform the fluorescence compensa-
tion as follows.
1. Calculate the mean values for each fluorescence channel
(GFP-A and PerCP-A) from untransfected cells to evaluate
autofluorescence ([auto]GFP-A and [auto]PerCP-A).
2. Perform linear fitting to obtain a spillover coefficient (coef1) of
the reporter protein (EGFP) into the channel for the control
protein (PerCP-A) based on data from cells expressing only
EGFP, and the following equation (upper panel, Fig. 3a).

[PerCP - A ] = coef 1 [EGFP ] + [DsRed ] + [auto ]PerCP -A
3. Perform linear fitting to obtain the spillover coefficient (coef2)

of the control protein (DsRed) to the measurement of the
reporter protein (GFP-A) based on data from cells expressing
only DsRed, and on the following equation (lower panel,
Fig. 3a).

[GFP - A ] = [EGFP ] + coef 2 [DsRed ] + [auto ]GFP - A
4. Invert the spillover matrix (S) to obtain the compensation
matrix (C).
æ 1 coef 2 ö -1
Here, S = ç ÷ ,C = S
è coef 1 1 ø
From the two equations in steps 2 and 3,
æ [GFP - A ] ö æ [EGFP ] ö
çç ÷÷ = S çç ÷÷
è [PerCP - A ] ø è [ DsRed ] ø
æ [EGFP ] ö æ [GFP - A ] ö
Therefore, ç
ç [ DsRed ] ÷÷ = C çç [PerCP - A ] ÷÷
è ø è ø
5. Apply the compensation matrix (C) to all the other data to
obtain compensated signals from the reporter (EGFP) and the
control (DsRed).
3.3.5 Calculation of Cross comparing the activity of the switch with different configura-
Translational Efficiency tions or in different conditions requires some representative values.
It can be considered as the ratio of the translational efficiency of the
riboswitch in the presence and in the absence of the input protein.
When calculating the ratio, effects of the background fluorescence
and the efficiency of transfection should be taken into consider-
ation. Here it is shown one definition that represents translational
efficiency of the riboswitch as the average of the output from spe-
cific cells determined by the control signal (Fig. 4).
1. Collect cells expressing a specific range of the control fluores-
cent protein (e.g., 1,000 ± 100 a.u. of DsRed) (Fig. 4a).
2. Calculate the mean level of the reporter protein in the col-
lected cells.
3. Divide the mean value from cells in the presence of the input
protein by that in the absence of the input to obtain transla-
tional efficiency (Fig. 4b). Confirm that the choice of the spe-
cific range selected for the above steps does not significantly
change translational efficiency by varying the specific range as
shown in Fig. 4b.
3.4 Controlling Protein-responsive RNA switches can be employed for control-

Apoptotic Genes ling, in principle, any gene of interest. Induction of a particular
gene from a switch has potential to control cellular function, such
as apoptosis, in a manner responding to a specific protein.
Figure 5 shows an example of a system for switch-mediated
Fig. 4 A definition of translational efficiency. (a) An example for a specific window

of cells (1,000 ± 100 a.u. of DsRed) used to calculate translation efficiency. (b)
Comparison of translational efficiencies from ranged different windows (400–
1,600 a.u. of DsRed). Those of three different switches are shown. Kt, another
aptamer to L7Ae; Sp, a spacer sequence (164 nt). Choice of specific windows
does not change translational efficiency significantly
a b
Fig. 5 Cell death induced by a switch. (a) Schematic illustration of the system.
Bcl-xL protein produced from the switch blocks apoptosis induced by Bim-EL. In
the presence of input protein, the switch represses Bcl-xL expression and
induces apoptosis. (b) Induction of apoptosis was observed in an input protein
specific manner
induction of cell death [19]. In this case the switch which outputs
to an anti-apoptotic gene, Bcl-xL, is transferred into mammalian
cells, together with an apoptosis inducible gene, Bim-EL. Bcl-xL
interacts with Bim-EL and therefore blocks the function of
Bim-EL and cell death; in the presence of the input protein, the
expression of Bcl-xL is repressed and the function of Bim-EL
recovers, resulting in the induction of apoptosis.
A procedure for evaluating the induction of apoptosis in trans-
fected cells is as follows.
1. Prepare a switch vector with Bcl-xL as the output instead of
the reporter fluorescent protein described previously.
2. Cotransfect both the switch and trigger vectors as well as an
additional vector, which expresses Bim-EL (see Note 8).
3. Change the medium 4 h after the transfection.
4. Transfer media including detached cells from the well into a
1.5 mL-microcentrifuge tube 24 h after the transfection.
5. Add 100 μL of Trypsin–EDTA and incubate at 37 °C for
3 min.
6. Add 200 μL of medium and gently rock the dish.
7. Transfer the cell suspension to the same tube as in step 4.
8. Spin the suspension at 200 × g for 5 min at room temperature.
9. Aspirate supernatant and add 500 μL of PBS.
10. Spin again at 200 × g for 5 min at room temperature and aspi-
rate supernatant again.
11. Suspend cells with 53 μL of premixed staining reagent (Pacific
Blue Annexin V, 2.5 μL; SYTOX AADvance, 0.5 μL; binding
buffer, 50 μL).
12. Incubate samples in the dark for 30 min at room temperature.
13. Add 200 μL of binding buffer and keep on ice until analysis.
14. Perform the flow cytometry analysis.
15. Gate out untransfected cells based on the fluorescence of the
control protein.
16. Evaluate the ratio of Annexin V positive cells in populations
(Fig. 5b).
3.5 Dual Switch Analogous to transcriptional regulation, multiple transgenes need

Regulation to be controlled under a single protein in some cases. In addition,
it is desirable that expression levels from multiple genes can be
tuned simultaneously and specifically. Modification of cis-regulatory
elements (Subheading 3.2) in the mRNA is sufficient to produce
such switches. In fact we have created such a dual system whereby
one input protein can regulate two independent switches simulta-
neously. We have observed that addition and the configuration of
a b
Fig. 6 Simultaneous and specific control of two outputs by a single input. (a) Schematic illustration of the
system. (b) Translational efficiencies of co-transfected two distinct L7Ae-responsive switches. dKt defective
sequence of Kt
one switch did not affect the translational efficiency of the other
switch in the same cell (Fig. 6) [16].
1. Prepare another switch vector composed of the same aptamer
but a different reporter protein.
2. Transfect two switch vectors together with a trigger vector
(Fig. 6a).
3. Perform flow cytometry analysis with three fluorescent pro-
teins (two reporter and one control).
4. Evaluate translational efficiency of each switch (Fig. 6b).
3.6 Autofeedback Metabolite-binding riboswitches in bacteria monitor a physiological

Regulation condition and control specific genes in the pathway to give feedback
on a biosynthetic system [20]. Mammalian IRP also plays a central
role in iron homeostasis, controlling iron metabolic genes via iron-
dependent binding to a specific sequence in their mRNAs [21].
Similarly, and more simply, a protein-responsive riboswitch itself can
form a feedback loop because the same protein can be used for both
its input and output.
Figure 7 shows a configuration of translational autoregulation.
The genetic construct (Fig. 7a) can be derived from the two basic
constructs (Fig. 1b, d) by fusing an input protein in a trigger vec-
tor with an output gene (GOI) in a reporter vector. A fusion pro-
tein expressed from transcripts binds to its own mRNA and
represses further expression of the input–output fusion protein.
Modification in the interaction of an RNP pair, using modified
RNA motifs and/or variants of the input protein, can also tune the
behavior of the system [22]. Furthermore, input proteins under
autoregulation can be employed as a regulator for other mRNAs in
the same cells [22].
Fig. 7 Autofeedback regulation composed of protein-responsive switch. (a) Genetic configuration of the auto-
regulation. A GOI in a switch vector is fused with an input protein from a trigger vector. (b) Schematic illustra-
tion of translational negative feedback loop
4 Notes
1. In this protocol, input and output plasmids are transiently

transfected into cultured cells. Integration of these constructs
into the genome of host cell can produce cells that stably
express the switches and/or input proteins [19].
2. Alternatively, DNA fragments digested by BamH I (or Bgl II)
and Age I derived from a plasmid or a PCR fragment can be
inserted into the switch vector.
3. Reduce the concentration of sodium ions to hybridize the two
strands in more strict conditions.
4. It is possible to fuse the input protein with the control protein
for some purposes. For example, fluorescent protein (e.g.,
DsRed) can be fused with the input (e.g., L7Ae) to monitor its
expression level [19].
5. The average length of the 5′ UTRs in the human genome is
125 nt [23]. Too long or too short 5′ UTRs may reduce the
translation efficiency independent of the switch function.
6. In general, the level of transiently expressed transgene varies
widely. However, the ratio of the levels between two genes from
two different plasmids seems constant, provided that the two
plasmids were mixed with each other before the addition of a
transfection reagent. Untransfected cells and cells expressing
either of the two fluorescent proteins are needed to examine and
to correct for the spillover between the two (see Subheading 3.3.4).
7. The fluorescent spillover interferes with the quantification of the
signals. In the analysis of reporter fluorescent proteins, it is easy
to prepare cells expressing either of the two signals, and so, pos-
sible to compensate for the spillover mathematically after the
measurement. To circumvent the necessity for compensation,
an alternative would be to use fluorescent proteins whose signals

are more compatible and do not spill into each other. In addition
this alternative may improve the accuracy of measurement.
8. The ratio of Bcl-xL and Bim-EL proteins in a cell is critical for
apoptosis control. The efficiency of protein production from the
transferred plasmid is often affected by the configuration of the
vector, including the coding sequence of the protein, and is hard
to estimate. Thus, prior to evaluating the switches, it is required
to determine a proper ratio for transfection of the two plasmids.
References
1. Isaacs FJ, Dwyer DJ, Collins JJ (2006) RNA 12. Paraskeva E, Atzberger A, Hentze MW (1998)
synthetic biology. Nat Biotechnol 24: A translational repression assay procedure
545–554 (TRAP) for RNA-protein interactions in vivo.
2. Wieland M, Fussenegger M (2010) Ligand- Proc Natl Acad Sci U S A 95:951–956
dependent regulatory RNA parts for synthetic 13. Nie M, Htun H (2006) Different modes and
biology in eukaryotes. Curr Opin Biotechnol potencies of translational repression by
21:760–765 sequence-specific RNA-protein interaction at
3. Chang AL, Wolf JJ, Smolke CD (2012) the 5′-UTR. Nucleic Acids Res 34:5528–5540
Synthetic RNA switches as a tool for temporal 14. Saito H, Kobayashi T, Hara T et al (2010)
and spatial control over gene expression. Curr Synthetic translational regulation by an L7Ae-
Opin Biotechnol 23:679–688 kink-turn RNP switch. Nat Chem Biol 6:71–78
4. Nandagopal N, Elowitz MB (2011) Synthetic 15. Werstuck G, Green MR (1998) Controlling
biology: integrated gene circuits. Science gene expression in living cells through small mol-
333:1244–1248 ecule-RNA interactions. Science 282:296–298
5. Ruder WC, Lu T, Collins JJ (2011) Synthetic 16. Endo K, Stapleton JA, Hayashi K et al (2013)
biology moving into the clinic. Science Quantitative and simultaneous translational
333:1248–1252 control of distinct mammalian mRNAs. Nucleic
6. Wieland M, Fussenegger M (2012) Engineering Acids Res. 41:e135
molecular circuits using synthetic biology in 17. Imai Y, Matsushima Y, Sugimura T et al (1991)
mammalian cells. Annu Rev Chem Biomol Eng A simple and rapid method for generating a
3:209–234 deletion by PCR. Nucleic Acids Res 19:2785
7. Ausländer S, Ausländer D, Müller M et al 18. Hansson MD, Rzeznicka K, Rosenbäck M et al
(2012) Programmable single-cell mammalian (2008) PCR-mediated deletion of plasmid
biocomputers. Nature 487:123–127 DNA. Anal Biochem 375:373–375
8. Jackson RJ, Hellen CU, Pestova TV (2010) 19. Saito H, Fujita Y, Kashida S et al (2011)
The mechanism of eukaryotic translation initia- Synthetic human cell fate regulation by protein-
tion and principles of its regulation. Nat Rev driven RNA switches. Nat Commun 2:160
Mol Cell Biol 11:113–127 20. Smith AM, Fuchs RT, Grundy FJ et al (2010)
9. Sonenberg N, Hinnebusch AG (2009) Riboswitch RNAs: regulation of gene expres-
Regulation of translation initiation in eukary- sion by direct monitoring of a physiological
otes: mechanisms and biological targets. Cell signal. RNA Biol 7:104–110
136:731–745 21. Eisenstein RS (2000) Iron regulatory proteins
10. Hentze MW, Kühn LC (1996) Molecular con- and the molecular control of mammalian iron
trol of vertebrate iron metabolism: mRNA- metabolism. Annu Rev Nutr 20:627–662
based regulatory circuits operated by iron, 22. Stapleton JA, Endo K, Fujita Y et al (2012)
nitric oxide, and oxidative stress. Proc Natl Feedback control of protein expression in
Acad Sci U S A 93:8175–8182 mammalian cells by tunable synthetic transla-
11. Stripecke R, Oliveira CC, McCarthy JE et al tional inhibition. ACS Synth Biol 1:83–88
(1994) Proteins binding to 5′ untranslated 23. Suzuki Y, Ishihara D, Sasaki M et al (2000)
region sites: a general mechanism for transla- Statistical analysis of the 5′ untranslated region
tional regulation of mRNAs in human and of human mRNA using ‘Oligo-Capped’ cDNA
yeast cells. Mol Cell Biol 14:5898–5909 libraries. Genomics 64:286–297
Chapter 14
Guanine-Tethered Antisense Oligonucleotides

as Synthetic Riboregulators
Masaki Hagihara
Abstract
Regulation of gene expression by short oligonucleotides (antisense oligonucleotides), which can modulate
RNA structures and inhibit subsequent associations with the translation machinery, is a potential approach
for gene therapy. This chapter describes an alternative antisense strategy using guanine-tethered antisense
oligonucleotides (G-ASs) to introduce a DNA–RNA heteroquadruplex structure at a designated sequence
on RNA targets. The feasibility of using G-ASs to modulate RNA conformation may allow control of RNA
function by inducing biologically important quadruplex structures. This approach to manipulate quadru-
plex structures using G-ASs may expand the strategies for regulating RNA structures and the functions of
short oligonucleotide riboregulators.
Key words Antisense oligonucleotides, Guanine quadruplexes, DNA–RNA hybrids, Reverse

transcriptase stop assay, Circular dichroism analysis, UV melting analysis
1 Introduction
Natural riboswitches are folded RNA structures within mRNA that

sense metabolites and regulate protein biosynthesis in cells [1–3].
The coupling of ligand binding to structural changes in mRNA is
a central regulatory mechanism of riboswitches. Thus, develop-
ment of small molecules that modulate the RNA structure and
function has been a current topic of intensive research [4]. Such
designed ligand–RNA pairs would enable one to construct artificial
gene regulation systems, which may provide genetic tools for both
basic biological study and clinical research.
The function of RNAs depends largely on their tertiary struc-
tures, in which RNAs form active ternary structures through spe-
cific hydrogen bonds, such as Watson–Crick and Hoogsteen bonds.
Among a variety of RNA ternary structures, guanine quadruplexes
are a unique structure, in which four guanine bases engage in
Hoogsteen hydrogen bonds to build a guanine tetrad and incorporate
197
198 Masaki Hagihara
Fig. 1 Illustrations of the induced hybrid guanine quadruplex structure on mRNA templates by guanine-tethered
antisense oligonucleotides. A guanine-tethered antisense (G-AS, shown in gray ) hybridizes to a target RNA,
and contiguous guanines in both the RNA and the G-AS then associate to form a quadruplex. The drawing
shown in (a) and (c) show a 1:2 split G-quadruplex mode for the number of GGG sequences, and the drawing
shown in (b) uses a 2:2 split G-quadruplex mode
a metal ion, such as potassium and sodium, inside the tetrad [5, 6].
Recent reports have shown that quadruplex formation of mRNAs
plays a significant role in biological processes, including translation
and translocation [7, 8]. Thus, controlled introduction of quadru-
plex structures into mRNAs may provide opportunities for regula-
tion of biological processes.
Antisense oligonucleotides are relatively short single-stranded
DNAs designed to abrogate protein function by hybridizing
directly with oligonucleotides to exposed regions of the targeted
mRNA [9, 10]. A recently reported novel antisense strategy using
guanine-tethered antisense (G-AS) oligonucleotides confers an
aberrant DNA–RNA heteroquadruplex structure [11]. G-ASs
comprise two functionally independent domains; one is an anti-
sense domain that binds specifically to the target RNA in the same
way as conventional antisense oligonucleotides, and the other
involves contiguous guanine runs tethered at the 5′-end of the
antisense domain that is designed to assemble into heteroquadru-
plex structures together with a contiguous guanine run in the tar-
get RNAs (Fig. 1). The key concept of G-ASs is based on the
cooperative formation of DNA–RNA heteroquadruplex structures
with the assistance of hybridization of antisense sequences. The
combination of antisense recognition and self-assembly of contigu-
ous guanines affords predictable and sequence-specific DNA–RNA
heteroquadruplex structures, which effectively induce structural
changes in RNA templates. The induced quadruplex structures
produced effectively inhibit reverse transcription on the native
RNA sequences.
A reverse transcriptase-mediated enzymatic analysis (RTase
stop assay) [12] together with circular dichroism (CD) spectros-
copy [13, 14] and UV thermal melting experiments [15–17] using
the model assembly of RNA and G-AS oligonucleotides provides
information about the induced guanine quadruplex structures on
RNA templates in association with G-ASs. Detailed investigation
of DNA–RNA hybrid heteroquadruplex formation reveals that at
least one GGG sequence in the target RNA is the minimal sequence
for the introduction of a stable heteroquadruplex by G-ASs. This
feasible and effective approach to controlling RNA structures by
Guanine-Tethered Antisense Riboregulators 199
G-ASs opens the possibility of using G-ASs as synthetic riboregula-

tors for gene regulation. The following steps describe the design
guidelines for G-ASs and the characterization of the G-AS-induced
hybrid guanine quadruplex structures.
2 Materials
Great care should be taken to avoid inadvertently introducing con-

taminating RNases into the RNA sample during experiments.
Always wear clean gloves while handling RNA samples and use
RNase-free reagents to prevent contamination by nucleases from
the surface of the skin or from dusty laboratory equipment. Change
gloves often and keep tubes closed whenever possible. Prepare all
solutions using nuclease-free water and DNase- and RNase-free
molecular grade reagents, and store all reagents at room tempera-
ture unless otherwise indicated.
2.1 RTase Stop 1. Nuclease-free TE buffer: 10 mM Tris–HCl, pH 8.0, 1 mM

Assay Components EDTA, molecular biology grade.
2. 0.1× TE buffer (dilute nuclease-free TE buffer with nuclease-
free water).
3. 10 μM antisense oligonucleotides in 0.1× TE buffer (store at
–20 °C) (see Notes 1 and 2).
4. 3 μM RNA stock solution in 0.1× TE buffer (store at −20 °C)
5. 1 μM Texas Red-labeled cDNA primer in 0.1× TE buffer
(store at −20 °C, keep in a dark box to protect from light)
(see Note 5).
6. 100 units/mL ReverTra Ace reverse transcriptase (Toyobo,
Tokyo, Japan) (store at −20 °C) (see Note 6).
7. dNTP (10 mM each of dATP, dGTP, dCTP, and dTTP)
(Promega) (store at −20 °C).
8. Blue dextran stock: 30 mg/mL blue dextran (Sigma-Aldrich)
dissolved in 50 mM EDTA, pH 8.0 (store at −20 °C).
9. Reaction buffer (5×): 250 mM Tris–HCl, pH 8.0, 50 mM
dithiothreitol, 500 mM KCl (see Note 7).
10. Extension mixture: 24 μL of dNTP mixture, 10 μL of 150 mM
MgCl2, 6 μL of reverse transcriptase (add reverse transcriptase
just before use and mix well).
11. Precipitation buffer: 10 mL of 95 % ethanol, 1 mL of 3 M
sodium acetate, pH 5.3.
12. Loading solution: mix deionized formamide and blue dextran
stock in a ratio of 5:1 (store at −20 °C).
200 Masaki Hagihara
13. Standard thermal cycler (e.g., TaKaRa PCR Thermal Cycler

Dice).
14. Automated fluorescent DNA sequencer (e.g., Hitachi SQ5500
automated DNA sequencer).
2.2 CD Analysis 1. 100 μM oligoribonucleotide stock solutions in TE buffer

Components (store at −20 °C) (see Notes 8 and 9).
2. 100 µM antisense oligonucleotide stock solution in TE buffer
(store at −20 °C).
3. CD measurement buffer (2×): 20 mM lithium cacodylate,
pH 7.0 (see Note 10), 200 mM KCl.
4. Quartz cuvettes (10 mm path length).
5. CD spectrometer (e.g., Jasco 715 spectrometer).
6. Temperature controller (e.g., Eppendorf ThermoStat plus).
2.3 UV Melting 1. 100 μM oligoribonucleotide stock solutions in TE buffer

Analysis Components (store at −20 °C) (see Note 9).
2. 100 μM antisense oligonucleotide stock solution in TE buffer
(store at −20 °C).
3. UV melting measuring buffer (2×): 20 mM lithium cacodyl-
ate, pH 7.0 (see Note 11).
4. 1 M KCl, molecular biology grade.
5. Quartz cuvettes and caps.
6. UV–Vis spectrophotometer equipped with a cell changer
device (e.g., Shimadzu PharmaSpec UV-1700
spectrophotometer).
7. Temperature-control device (e.g., Shimadzu Peltier element-
controlled TMSPC-8 system).
3 Methods
3.1 RTase 1. Mix 6 μL of reaction buffer, 3 μL of cDNA primer, and 3 μL

Stop Assay of the RNA sample, and add 11 μL of water to 0.2 mL PCR
tubes.
2. Anneal cDNA primer and the RNA sample by heating tubes at
80 °C for 5 min and place tubes at room temperature for
10 min.
3. Add 3 μL of the G-AS to each reaction tube and place tubes at
room temperature for 15 min.
4. Immediately dispense 4 μL of extension mixture into each
reaction tube containing RNA–primer–G-AS mixture and
incubate tubes at 42 °C for 30 min in the thermal cycler.
Fig. 2 Guanine-tethered antisense sequences inhibit reverse transcription of an RNA template bearing two
runs of the GGG sequence. An RNA–AS heteroduplex sequence is shown in the figure. The bold line indicates
the AS hybridization site. The lane markers U, G, C, and A indicate the bases on the template RNA. A partial
genomic RNA sequence from the central polypurine tract of HIV-1 (nucleotides 4581–4756), which is involved
in nuclear import of HIV-1, was chosen as the target sequence
5. Transfer precisely 25 μL of reaction mixture to fresh 1.5 mL

tubes and add 75 μL of precipitation buffer to all tubes, vor-
tex, and centrifuge at 14,000 × g for 15 min at room
temperature.
6. Remove samples from the centrifuge and carefully remove
supernatants.
7. Wash pellet with 100 μL of 70 % ethanol and centrifuge at
14,000 × g for 3 min.
8. Remove supernatants and allow pellet to air dry for the mini-
mum time required to remove any residual ethanol. Overdrying
will make redissolving the pellet difficult.
9. Add 10 μL of loading solution per tube, resuspend by vortex-
ing several times, and centrifuge briefly.
10. Heat samples for 3 min at 90 °C and quickly move to ice.
11. Analyze 1–2 μL of the samples on a denaturing polyacrylamide
gel and analyze with a fluorescent DNA sequencer (Fig. 2)
(see Note 12).
3.2 CD Analysis 1. Determine the concentration of oligonucleotides by measur-

of DNA–RNA ing absorbance at 260 nm in a UV–Vis spectrophotometer
Quadruplexes and verify concentration of each oligonucleotide using the
extinction coefficient (Fig. 3a) (see Note 13).
2. Dilute RNA oligonucleotide and antisense DNA samples with
ultrapure water in a total volume of 175 μL, and then add
175 μL of CD measurement buffer to achieve a final concen-
tration of 5 μM for each strand.
3. Start refolding of secondary structure by heating to 90 °C for
5 min followed by slowly cooling to 25 °C (0.5 °C/min) in a
programmable temperature controller.
202 Masaki Hagihara
Fig. 3 Biophysical properties of a model assembly of RNA–DNA hybrids modified with runs of contiguous
guanines. (a) Oligonucleotide sequences used in these analyses. Guanine-tethered sequences are shown in
italic letters. (b) CD spectra of rnaG0 and dnaG0 (black line) and rnaG2 and dnaG2 (dotted line). (c, d) UV-melting
profile of DNA–RNA hybrids of rnaG2 and dnaG2 measured at (c) 260 nm and (d) 295 nm
4. Transfer refolded DNA–RNA samples to a quartz cuvette.

5. Record the CD spectra (Fig. 3b) (see Notes 14 and 15).
3.3 UV Melting 1. Determine the concentration of oligonucleotides by measur-

Analysis of DNA–RNA ing absorbance at 260 nm in a UV–Vis spectrophotometer
Quadruplexes and verify the concentration of oligonucleotides using the
extinction coefficient (Fig. 3a) (see Note 13).
2. Mix 50 µL of Tm-measuring buffer, 5 µL of 1 M KCl, 5 µL of
oligonucleotide stock solution, 5 µL of antisense
oligonucleotide stock solution, and ultrapure water to a final
volume of 100 µL.
3. Mix components in 0.2 mL PCR tubes, heat the tubes for

5 min at 90 °C, and cool the tubes to room temperature slowly
in a standard thermal cycler to maintain the proper secondary
structures of DNA–RNA hybrids (see Note 16).
4. Transfer samples to the quartz cuvette and overlay mineral oil
to limit evaporation if the cap does not fit perfectly on the
cuvette.
5. For a typical experiment, record absorbance changes at 295 nm
and 260 nm to examine the stability of the quadruplex and
duplex, respectively (see Note 17); use an integration time of
1 s and a temperature gradient of 0.5 °C/min.
6. Represent the absorbance profiles as a function of the tem-
perature (Fig. 3c, d) and check the quality of the data (see
Note 18).
7. Calculate the melting points of the duplex and the quadruplex
structures by determining the melting temperatures at the
half-maximum change of the signal (see Note 19).
4 Notes
1. The general principle for designing G-AS oligonucleotides

depends on the simple features of the guanine quadruplex for-
mation [18, 19] described below. (1) The formation of one
guanine quadruplex unit requires four consecutive GGG
repeats. (2) Guanine quadruplexes can be divided into two
parts; one possesses three GGG repeats, and another part pos-
sesses one GGG repeat, namely, the 3:1 split mode; or both
possess two GGG repeats, namely, the 2:2 split mode. (3) Two
single-stranded oligonucleotides having a total of four GGG
repeats can associate together to form a hybrid quadruplex
structure when brought into close proximity. To design G-ASs
for sequence-specific introduction of guanine quadruplex
structures, the association between four units of unpaired
GGG in close proximity and a covalent connection of the anti-
sense domain and the guanine-rich region are essential for
introducing stable hybrid guanine quadruplex structures on
the RNA templates. For example, when two adjacent GGG
sequences are found in the target RNAs, the design G-AS is
accomplished simply by modifying two runs of contiguous
GGG units at the 5′-terminal of the antisense region.
2. The stability of DNA–RNA heteroquadruplexes may depend
on both the sequence of the guanine tethers in the G-ASs and
the length of the antisense region. If generation of stable het-
eroquadruplexes is desired, the sequences of tethers and the
length of the antisense can be optimized.
204 Masaki Hagihara
3. Transcription by T7 RNA polymerase is generally better than

solid-phase chemical synthesis for producing RNA longer than
30 nucleotides. The DNA templates in the T7 RNA poly-
merase reaction can be either a linearized plasmid or PCR
products that contain the T7 promoter sequence.
4. The RTase stop assay is extremely sensitive to cleavage of the
RNA templates by contaminating nucleases. Thus, the use of
RNA templates of the highest purity is critical for successful
operation of the assay. Polyacrylamide gel purification of RNA
templates is strongly recommended.
5. The cDNA primer is designed to hybridize to a specific
sequence downstream of the consecutive guanine-rich
sequences in the target RNA. The optimum length of the
cDNA primer is 17–25 nucleotides.
6. Several commercially available reverse transcriptases have been
tested for their appropriateness in evaluating the stability of
guanine quadruplexes. Among the enzymes tested, ReverTra
Ace, an M-MLV reverse transcriptase genetically engineered
by the introduction of two point mutations in the polymerase
and RNase H domains, showed superior performance charac-
teristics for detecting guanine quadruplexes in RNA
templates.
7. The reaction buffer used here is designed to promote the for-
mation of DNA−RNA heteroquadruplexes. The concentra-
tion of potassium chloride will largely affect the stability of
quadruplex structures and can be optimized [20].
8. Measurement of the CD spectra of nucleic acids is operation-
ally similar to that of the UV–Vis absorbance spectrum.
However, sample degradation can lead to significant variations
in the resulting CD spectra, especially in the wavelength range
of 190–260 nm. Great care should be taken in the preparation
of samples.
9. The oligonucleotides should be ordered as HPLC purified for
greater purity. The purity and integrity of the oligonucleotides
should be verified by denaturing polyacrylamide gel
electrophoresis.
10. To prepare a 200 mL volume of 20 mM sodium cacodylate at
pH 7.0, dissolve 0.55 g of cacodylic acid in 150 mL of double-
distilled water. Adjust the pH of the solution using a 1 M
NaOH solution until the desired pH is reached. Adjust the
volume to 200 mL with RNase-free water.
11. For thermal melting analysis, the optimum buffer system
should have a temperature-independent pKa. A cacodylate
buffer (pKa = 6.3) or a phosphate buffer is preferred to the
more commonly used Tris, HEPES, and MOPS buffers [21].
12. (Anticipated results) In the RTase stop assay, an RTase pro-

ceeds along the RNA until the enzyme encounters stable a
guanine quadruplex structure. Inhibition of the RTase reac-
tion results in the production of truncated DNA transcripts,
which can be detected by DNA sequencing gel analyses.
13. To examine the formation of DNA–RNA heteroquadruplex
structures, it is critical to mix the DNA and RNA strands in an
equimolar concentration.
14. The following settings are used for CD measurements: resolu-
tion, 1 nm; bandwidth, 1.0 nm; sensitivity, 50 mdeg; response
time, 8 s; accumulation, 4 times; scan rate, 50 nm/min; and
temperature, 25 °C. Baseline spectra are recorded with 1× CD
measuring buffer in the cuvette and subtracted from the spec-
tra of the sample.
15. (Anticipated results) CD spectra of a DNA–RNA hybrid lack-
ing guanine regions (rnaG0 and dnaG0) exhibit a typical
A-form structure [22], showing a broad positive peak around
270 nm and a weak negative peak around 250 nm. By con-
trast, the spectra of the DNA–RNA hybrid involving two runs
of the GGG sequence in each strand (rnaG2 and dnaG2) show
a strong positive peak at 270 nm and a weak negative peak at
240 nm; this is assigned as a characteristic signature of a paral-
lel quadruplex formation [23].
16. This step also ensures removal of dissolved gas from the sam-
ples and avoidance of bubble formation, which will interfere
with UV measurements during melting analysis.
17. The formation of the DNA–RNA heteroguanine quadruplex
structure could be monitored by the absorbance changes at
295 nm, because hyperchromism is observed when consecu-
tive GGG sequences associate to form guanine quadruplexes
[16, 17]. In parallel, the absorbance profile at 260 nm should
be recorded to monitor the denaturation of DNA–RNA
duplex structures. Monitoring the thermal denaturation pro-
files by measuring CD spectra changes at 270 nm, the charac-
teristic signature of parallel quadruplex structures, works
equally well [24].
18. If the melting temperatures are too high to be determined
because of highly stable DNA–RNA hybrids, decreasing the
KCl content of the solution may allow a more precise Tm
determination. Decreasing the KCl content will destabilize the
DNA–RNA guanine quadruplex structures, leading to
decreased stability of DNA–RNA hybrids [20].
19. (Anticipated results) The DNA–RNA hybrid lacking the gua-
nine consecutive sequences (rnaG0 and dnaG0) shows only
duplex melting. The melting temperature of the hybrid with
quadruplexes (rnaG2 and dnaG2) measured at 260 nm is
206 Masaki Hagihara
almost identical to that of the quadruplex measured at 295 nm,

which suggests a cooperative melting of the duplex and the
quadruplex in the DNA–RNA hybrid.
20. (Caution) The cacodylate buffer used in these procedures
contains arsenic and can produce arsenic gas when mixed with
acids. The solution containing a cacodylate buffer should be
disposed of as toxic waste according to the institution’s
guidelines.
Acknowledgment
This work was supported by a grant from the Ichiro Kanehara

Foundation.
References
1. Roth A, Breaker RR (2009) The structural and transcriptase. J Am Chem Soc
functional diversity of metabolite-binding ribo- 132:11171–11178
switches. Annu Rev Biochem 78:305–334 12. Hagihara M, Yoneda K, Yabuuchi H et al
2. Serganov A, Patel DJ (2007) Ribozymes, ribo- (2010) A reverse transcriptase stop assay
switches and beyond: regulation of gene revealed diverse quadruplex formations in
expression without proteins. Nat Rev Genet UTRs in mRNA. Bioorg Med Chem Lett
8:776–790 20:2350–2353
3. Mandal M, Breaker RR (2004) Gene regula- 13. Maurizot JC (2000) Circular dichroism of
tion by riboswitches. Nat Rev Mol Cell Biol nucleic acids: nonclassical conformations and
5:451–463 modified oligonucleotides. In: Berova N,
4. Link KH, Breaker RR (2009) Engineering Nakanishi K, Woody RW (eds) Circular dichro-
ligand-responsive gene-control elements: les- ism: principles and applications, 2nd edn.
sons learned from natural riboswitches. Gene Wiley-VCH, New York, pp 719–739
Ther 16:1189–1201 14. Porumb H, Monnot M, Fermandjian S (2002)
5. Burge S, Parkinson GN, Hazel P et al (2006) Circular dichroism signatures of features simul-
Quadruplex DNA: sequence, topology and taneously present in structured guanine-rich
structure. Nucleic Acids Res 34:5402–5415 oligonucleotides: a combined spectroscopic
6. Lane AN, Chaires JB, Gray RD et al (2008) and electrophoretic approach. Electrophoresis
Stability and kinetics of G-quadruplex struc- 23:1013–1020
tures. Nucleic Acids Res 36:5482–5515 15. Sun XG, Cao EH, He YJ et al (1999)
7. Kumari S, Bugaut A, Huppert JL et al (2007) Spectroscopic comparison of different DNA
An RNA G-quadruplex in the 5′ UTR of the structures formed by oligonucleotides.
NRAS proto-oncogene modulates translation. J Biomol Struct Dyn 16:863–872
Nat Chem Biol 3:218–221 16. Mergny JL, Phan AT, Lacroix L (1998)
8. Arora A, Dutkiewicz M, Scaria V et al (2008) Following G-quartet formation by
Inhibition of translation in living eukaryotic UV-spectroscopy. FEBS Lett 435:74–78
cells by an RNA G-quadruplex motif. RNA 17. Mergny JL, Lacroix L (2009) UV melting of
14:1–7 G-Quadruplexes. In: Egli M, Herdewijn P,
9. Stein CA, Cheng YC (1993) Antisense oligo- Matusda A, Yogesh S (eds) Curr Protoc Nucleic
nucleotides as therapeutic agents—is the bullet Acid Chem, Wiley-VCH, New York, Chapter
really magical? Science 261:1004–1012 17: Unit 17.1
10. Eckstein F (2007) The versatility of oligonu- 18. Xiao Y, Pavlov V, Niazov T et al (2004)
cleotides as potential therapeutics. Expert Opin Catalytic beacons for the detection of DNA
Biol Ther 7:1021–1034 and telomerase activity. J Am Chem Soc
11. Hagihara M, Yamauchi L, Seo A et al (2010) 126:7430–7431
Antisense-induced Guanine Quadruplexes 19. Deng M, Zhang D, Zhou Y et al (2008) Highly
inhibit reverse transcription by HIV-1 reverse effective colorimetric and visual detection of
nucleic acids using an asymmetrically split per- and triplexes containing A-T or A-U base pairs.
oxidase DNAzyme. J Am Chem Soc Nucleic Acids Res 14:10071–10090
130:13095–13102 23. Jin R, Gaffney BL, Wang C et al (1992)
20. De Cian A, Guittat L, Kaiser M et al (2007) Thermodynamics and structure of a DNA tet-
Fluorescence-based melting assays for studying raplex: a spectroscopic and calorimetric study
quadruplex ligands. Methods 42:183–195 of the tetramolecular complexes of d(TG3T)
21. Fukada H, Takahashi K (1998) Enthalpy and and d(TG3T2G3T). Proc Natl Acad Sci U S A
heat capacity changes for the proton dissocia- 89:8832–8836
tion of various buffer components in 0.1 M 24. Benz A, Hartig JS (2008) Redesigned tetrads
potassium chloride. Proteins 33:159–166 with altered hydrogen bonding patterns enable
22. Steely HT Jr, Gray DM, Ratliff RL (1986) CD programming of quadruplex topologies. Chem
of homopolymer DNA-RNA hybrid duplexes Commun (Camb) 34:4010–4012
Chapter 15
In Vitro Selection of Allosteric Ribozymes that Sense

the Bacterial Second Messenger c-di-GMP
Kazuhiro Furukawa, Hongzhou Gu, and Ronald R. Breaker
Abstract
Recently, a number of study have shown the ligand-dependent allosteric ribozymes can be harnessed as
biosensors, high-throughput screening, and agents for the control of gene expression in vivo, called artifi-
cial riboswitches. In this chapter, we describe how in vitro selection can be used to create an allosteric
ribozyme that senses bacterial second messenger cyclic-di-GMP (c-di-GMP). A hammerhead ribozyme
was joined to a natural c-di-GMP class I riboswitch aptamer via communication modules. Both c-di-GMP-
activating and -inhibiting ribozyme can be obtained by this approach.
Key words Allosteric ribozyme, Aptamer, Hammerhead ribozyme, In vitro selection, Cyclic
di-GMP
1 Introduction
Allosteric ribozymes are RNA enzymes whose activity is modulated

by the binding of an effector molecule to an aptamer domain
[1, 2]. Most allosteric ribozymes are engineered by conjugating an
aptamer to a ribozyme through a communication module, that is,
ligand binding to the aptamer induces a conformational change,
which is transmitted to the ribozyme, the activity of which is modi-
fied. Our group [3–8] and others [9–13] have been applied
directed evolution strategies to engineer allosteric ribozymes or
artificial riboswitches. Also, our group recently found that in nature
cells use allosteric ribozymes, i.e., cyclic di-GMP class II (c-di-
GMP-II) riboswitch, to regulate gene expression [14]. In
Clostridium difficile strains, a c-di-GMP-II aptamer is found adja-
cent to a group I intron. Group I introns are ribozymes that cata-
lyze the splicing of the RNA molecule in which they are embedded.
In this case, the outcome of the splicing reaction catalyzed by the
intron is controlled by the natural allosteric ribozyme in response
to c-di-GMP levels.
209
210 Kazuhiro Furukawa et al.
c-di-GMP is a bacterial second messenger produced by the

GGDEF domain-containing diguanylate cyclase and regulates many
biological processes including the planktonic traits (motility) and
communal or sessile traits (synthesis of bacterial surface components
such as biofilm, exopolysaccharide, pili, and stalks) of several bacterial
groups by its intracellular concentration [15]. Due to the widespread
functions of c-di-GMP, the method to quantify the level of this second
messenger is required to understand a wide range of bacterial phe-
nomena. High-performance liquid chromatography (HPLC) is a con-
ventional method to monitor the concentration of bacterial
metabolites; however, it requires multiple semi-purification steps to
remove other metabolites that have similar chemical properties to the
target molecule. The allosteric ribozyme is an attractive alternative to
HPLC due to its high affinity and specificity to c-di-GMP. The natural
c-di-GMP-II riboswitch might be useful for quantitative analysis of
c-di-GMP. However, it is not a versatile approach because of low rate
constant of the group I intron and requirement of guanosine 5′-tri-
phosphate (GTP) for its catalytic activity.
We have used the method described below to create an engi-
neered allosteric ribozyme resulting from the fusion of a c-di-
GMP-I aptamer [16] to a hammerhead ribozyme for simple
monitoring of c-di-GMP levels (Fig. 1). Several engineered alloste-
ric ribozymes derived from the hammerhead ribozyme show equal
rate constant to the minimum ribozyme core, suggesting that they
could be used for rapid quantification of ligands in chemical or
biological samples. Moreover, allosteric ribozymes could be used
for high-throughput screening to facilitate the rapid discovery of
compounds that target the aptamer [17–19].
2 Materials
2.1 Construction 1. Chemically synthesized oligonucleotides for DNA templates

of the Starting Pool to prepare the RNA pool shown in Fig. 1a. Oligonucleotide 1:
5 ′ - T A A T A C G A C T C A C T A T A
GGC G TA G C C T G AT G A G N N N N -G C
ACAGGGCAAACCATTCGAAAGAGTGGGACGCAA
AGCCTCCGGC CTAAACCAGAAGACATGGTAG - 3 ′.
Oligonucleotide 2: 5′-GGCGCAGCTACGTGGCTTTCACC
ACGTTTC-GNNNNNGTAACCCCGCTACCTACCATGT
CTTCTGGTTTAG-3′. Underlined nucleotides are comple-
mentary each other at their 3′ ends. T7 promoter sequences
are shown in italic.
2. Chemically synthesized oligonucleotides for DNA templates to
prepare the RNA pool shown in Fig. 1b. Oligonucleotide 3:
5′-TAATACGACTCACTATAGGCACCTGATGAGNNNNN
N GCACAGG-GCAAACCATTCGAAAGAGTGGGAC
GCAAAGCCTCCGGCCTAAACCAGAAGACATGGTAG-3′.
Oligonucleotide 4: 5 ′-CTTAGGCACTACGTGGCT TTCACCA
Allosteric Ribozymes for c-di-GMP 211
Fig. 1 (a) Constructs used for the selection of allosteric ribozymes responding to c-di-GMP. The ribozyme
consists of a hammerhead ribozyme (HHR) joined to a class I c-di-GMP aptamer via communication module
(CM) consisting of random sequence (N) nucleotides. The three stems that form the HHR are designated I, II,
and III and the three stems that form the aptamer are labeled P1, P2, and P3. An arrowhead identifies the site
of hammerhead-mediated cleavage. (b) Modified construct for the selection of c-di-GMP-activated ribozyme
CGTTTCGNNNNNNNG-TAACCCCGCTACCTACCATG
TCTTCTGGTTTAG-3′. Underlined nucleotides are comple-
mentary each other at their 3′ ends. T7 promoter sequences are
shown in italic.
3. 5× reverse transcription buffer: 250 mM Tris–HCl, pH 8.3 at
23 °C, 375 mM KCl, 15 mM MgCl2.
4. 100 mM dithiothreitol (DTT).
5. 10× deoxynucleoside 5′-triphosphate (dNTP) mix: 2 mM each
of deoxyguanosine 5′-triphosphate (dGTP), deoxyadenosine
5′-triphosphate (dATP), deoxythymidine 5′-triphosphate
(dTTP), deoxycytidine 5′-triphosphate (dCTP).
6. Reverse transcriptase (e.g., SuperScript II, Invitrogen).
7. 10× transcription buffer: 500 mM HEPES, pH 7.5 at 23 °C,
150 mM MgCl2, 20 mM spermidine and 50 mM DTT.
8. 3 M sodium acetate, pH 5.2.
9. 100 % ethanol.
10. 5× nucleoside 5′-triphosphate (NTP) mix: 12.5 mM each of
guanosine 5′-triphosphate (GTP), adenosine 5′-triphos-
phate (ATP), thymidine 5′-triphosphate (TTP), cytidine
5′-triphosphate (CTP).
11. T7 RNA polymerase (T7 RNAP).

12. [α-32P] UTP, 3,000 Ci/mmol (PerkinElmer), and appropriate
equipment for handling 32P.
13. Crush-Soak buffer: 10 mM HEPES, pH 7.5 at 23 °C, 200 mM
NaCl, and 1 mM ethylenediaminetetraacetate (EDTA).
14. 2× denaturing gel-loading buffer: 32 % (w/v) sucrose, 0.16 %
(w/v) sodium dodecyl sulfate, 0.08 % (w/v) bromophenol
blue, 0.08 % (w/v) xylene cyanol, 7.25 M urea, 144 mM Tris–
HCl, 144 mM boric acid, and 200 mM EDTA.
15. Reagents and apparatus for denaturing 8 M urea polyacryl-
amide gel electrophoresis.
2.2 Selection 1. Cyclic-diguanosyl 5′-monophosphate (c-di-GMP, Biolog).

2. 2× selection buffer: 100 mM Tris–HCl, pH 7.5 at 23 °C,
200 mM NaCl.
3. 100 mM MgCl2.
4. X-ray film and reagents for developing the X-ray film.
5. PhosphorImager cassettes and PhosphorImager (e.g., Storm
PhosphorImager, GE HealthCare).
2.3 Reverse 1. Chemically synthesized oligonucleotides for the amplification

Transcription and of the cleaved products. Primer A:
Polymerase Chain 5′-TAATACGACTCACTATAGGCGTAGCCTGATGAG-3′.
Reaction Primer B: 5′-GGCGCAGCTACGTGGCTTTCACCACGTT
TCG-3′. Primer A’: 5′-TAATACGACTCACTATAGGCACCT
GATGAG-3′. Primer B’: 5′-CTTAGGCACTACGTGGCT
TTCACCACGTTTCG-3′.
2. Reverse Transcriptase (e.g., SuperScript II, Invitrogen).
3. 10× PCR buffer: 100 mM Tris–HCl, pH 8.3 at 23 °C, 400 mM
KCl, 15 mM MgCl2, and 0.1 % gelatin.
4. Taq DNA polymerase.
5. Reagent and apparatus for agarose gel electrophoresis.
3 Methods
3.1 General The construct used for the selection of allosteric ribozymes that
Selection Strategy respond to c-di-GMP was created by linking the P1 stem of the
c-di-GMP class I aptamer derived from Vibrio cholera [16] to stem
II of the hammerhead ribozyme via a communication module
(CM) (Fig. 1). The P1 stem of the aptamer was chosen for CM
since the stability of P1 stem is highly involved in the affinity
between the aptamer and the ligand. Also, there is an important
tertiary interaction between stems P2 and P3 [20, 21], so that
using these stems for CM might disrupt key structural contacts
within the aptamer. The CM also replaces a majority of the stem II

of the hammerhead ribozyme—a structural element that is a criti-
cal determinant of the ribozyme activity [22, 23]. The CM within
the resulting tripartite construct will provide a sampling of alterna-
tive stem II elements that might respond to c-di-GMP binding in
the adjacent aptamer domain and confer either positive or negative
allosteric control on the adjoining ribozyme domain (see Note 1).
This strategy has been applied to the selection of various allosteric
ribozymes (6–8). We have first used the construct shown in Fig. 1a
for both c-di-GMP-inhibited and -induced ribozymes. However,
the selection for the induced ribozymes was unsuccessful probably
because shorter CM (four and five randomized nucleotides) disfa-
vors allosteric induction due to steric clash between important
nucleotides of the aptamer and ribozyme. Therefore, we rede-
signed the construct by elongating randomized sequences
(Fig. 1b). This second construct allowed us to obtain the c-di-
GMP-induced ribozymes (see Note 2).
The outline of the selection scheme for the c-di-GMP-activated
allosteric ribozyme is shown in Fig. 2. The RNA population of the
initial generation (G0) was generated by primer extension of two
synthetic oligonucleotides with reverse transcriptase (Fig. 2a), fol-
lowed by transcription of about 20 pmol (10–20 copies) of the
resulting double-strand DNA (Fig. 2b, see Note 3). After the puri-
fication by denaturing PAGE, the full-length RNA precursors were
subjected to a negative selection in the absence of c-di-GMP
(Fig. 2c). RNAs that remained uncleaved during this reaction were
isolated by PAGE and subsequently subjected to a positive selec-
tion for self-cleavage in the presence of c-di-GMP (Fig. 2d). The
resulting cleavage products were isolated by PAGE and amplified
by reverse transcription-polymerase chain reaction (RT-PCR)
(Fig. 2e, see Note 4). The DNA from this stage were transcribed to
generate next generation of RNA population, which were then
subjected to additional rounds of selective amplification by repeat-
ing stages B to E in Fig. 2. The selection for the c-di-GMP-inhibited
allosteric ribozyme was done in a same way except for adding
c-di-GMP to the transcription and negative selection steps and
removing c-di-GMP from positive selection step.
3.2 Construction 1. To prepare the double-stranded DNA template for transcrip-

of the Starting Pool tion, incubate 40 pmol of each of the two oligonucleotides in
50 µL of 1× reverse transcription buffer supplemented with
10 mM DTT, 200 µM of the four dNTPs, and 8 U/µL reverse
transcriptase.
2. Incubate at 37 °C for 2 h.
3. Run a small aliquot out on a 2 % agarose gel to confirm that
double-strand DNA was synthesized and is the correct size.
Fig. 2 The selection scheme for the c-di-GMP-activated allosteric ribozyme. (A)
Two synthetic oligonucleotides are extended with reverse transcriptase to gener-
ate double-strand DNA encoding randomized sequence. (B) Transcription with T7
RNA polymerase. (C) The full-length RNAs (precursor) are purified by PAGE, and
subject to negative selection in the absence of c-di-GMP. The cleavage products
are removed by PAGE. (D) The precursor are subjected to positive selection in the
presence of c-di-GMP. (E) The PAGE-purified cleavage products are reverse tran-
scribed and amplified by PCR
4. Precipitate the double-stranded DNA by adding 0.1 vol of 3 M

sodium acetate and 2.5 vol of cold ethanol. Pellet the mixture by
centrifugation at 15,000 × g for 20 min.
5. Resuspend the DNA to 50 µL of 1× transcription buffer supple-
mented with 2 mM of the four NTPs, 1,000 units of T7 RNA
polymerase, and 10 µCi of [α-32P]UTP. For the c-di-GMP-
inhibited ribozyme selection, 100 µM c-di-GMP was added to
the mixture to favor isolation of ribozymes that are inactive when

bound to c-di-GMP.
6. Incubate at 37 °C for 2 h (see Note 5).
7. Add 2× denaturing loading buffer to the reaction mixture.
8. Separate full-length precursor RNAs from 3′ cleavage prod-
ucts, premature transcription products, and unincorporated
[α-32P]UTP by denaturing 8 % PAGE.
9. Use an autoradiogram of the gel to excise the gel band con-
taining the precursor RNAs.
10. Elute the RNA from the diced gel slices overnight at 4 °C
using crush soak buffer.
11. Transfer the solution containing the eluted RNAs into a fresh tube.
12. Concentrate the sample by ethanol precipitation and centri-
fuge as in step 4.
13. Resuspend the RNA in sterile deionized water.
3.3 Selection of 1. As a negative selection, incubate 300 pmol of an initial pool of

Allosterically Inhibited RNA precursors (internally 32P-labeled RNA) at 23 °C for 2 h in
Ribozymes 1× selection buffer in the presence of 10 mM MgCl2 and
100 µM c-di-GMP (see Notes 6 and 7).
2. To concentrate the RNAs from the reactions, precipitate with
ethanol and centrifuge as in Subheading 3.2, step 4.
3. Resuspend the RNAs in sterile deionized water and combine
with an equal volume of 2 × denaturing gel-loading buffer.
4. Separate the precursor RNAs that resisted cleavage during incu-
bation by denaturing 8 % PAGE. A cleavage marker (full-length
precursor) should be run in parallel (see Note 8).
5. Collect the precursor RNAs as in Subheading 3.2, steps 9–12.
6. Resuspend the resulting RNAs in the 1× selection buffer and
10 mM MgCl2 at 23 °C for appropriate time in the absence of
c-di-GMP (positive selection) (see Notes 9 and 10).
7. To concentrate the RNAs from the reactions, precipitate with
ethanol and centrifuge as in Subheading 3.2, step 4.
8. Resuspend the RNAs in sterile deionized water and combine
with an equal volume of 2 × denaturing gel-loading buffer.
9. Separate the cleaved products by denaturing 8 % PAGE.
A cleavage marker (e.g., cleaved products in the negative selec-
tion) should be run in parallel.
10. Image the resulting gel using a phosphorimager and calculate
the fraction of the RNAs that are cleaved (see Notes 11 and 12).
11. Collect the cleaved products as in Subheading 3.2, steps 9–13
(see Note 13).
3.4 Selection of Selection of ribozymes that undergo c-di-GMP induction was con-
Allosterically Inducted ducted using identical method as allosterically inhibited ribozymes
Ribozymes (Subheading 3.3), except that c-di-GMP was included in the posi-
tive selection reaction, but was excluded in both the transcription
and the negative selection reactions (see Note 14).
3.5 Reverse 1. Add 10 pmol of the 3′ primer (primer A or A′) to the resus-
Transcription and pended cleaved products.
Polymerase Chain 2. Reverse transcribe the recovered RNAs in 100 µL of 1× reverse
Reaction transcribe buffer supplemented with 10 mM DTT, 200 µM of
the four dNTPs, and 8 U/µL reverse transcriptase.
3. Incubate at 37 °C for 2 h.
4. Prepare PCR mixture by adding 5–10 µL of the reverse tran-
scription reaction and 20–50 pmol each of primers (primers A
and B for allosterically inhibited ribozymes, primers A′ and B′ for
allosterically inducted ribozymes) to a 100 µL PCR reaction
containing 1× PCR buffer, 200 µM each of the four dNTPs, and
1 U/µL Taq DNA polymerase (see Note 15).
5. Amplify full-length double-stranded DNAs using the following
PCR cycling parameters: segment 1: 94 °C for 1 min; segment
2: 94 °C for 30 s, 48 °C for 30 s, 72 °C for 30 s; segment 3:
72 °C for 5 min (see Note 16).
6. Analyze a small aliquot of the PCR reaction on a 2 % agarose gel
to check the amount and size of the PCR products.
7. Use the amplified DNAs as templates to transcribe the RNA
pool for the next round of selection, as in Subheading 3.2,
step 5. Save a portion of the PCR products.
8. Sequence the individual clones after several rounds of selection
4 Notes
1. A typical communication module ranges from a few to a dozen

nucleotides. The two constructs (Fig. 1) used in our study con-
tains 9 and 13 random nucleotides, respectively, in the domain of
the communication module. These should give us the anticipated
RNA libraries with great diversity (262,144 (49) and 67,108,864
(413) pieces of different RNA sequences, respectively).
2. When the hammerhead self-cleaving motif is fused with the large
c-di-GMP-I aptamer to create the c-di-GMP-dependent alloste-
ric ribozymes, there may be some steric hindrance between the
two moieties that could possibly affect the isolation of active
allosteric ribozymes. To minimize the potential steric clash, a
longer communication module and truncated stem I of the
hammerhead ribozyme was used as shown in Fig. 1b.
3. Considering that a typical run-off transcription produces

about 10–20 copies of each DNA sequence, including 20 pmol
of DNA template in the initial transcription step should give
us more than 200 pmol of the RNA library with guaranteed
sequence diversity.
4. When engineering the RNA constructs for allosteric selection,
make sure the anticipated difference in size between the full-
length RNA precursor and the cleaved RNA product is suffi-
cient (no less than eight nucleotides) to be resolved by
denaturing PAGE gels. Also ensure that there are enough
fixed sequences flanking the random domain (no less than 13
nucleotides) to allow the efficient PCR amplification of the
cleaved product with the appropriate primers.
5. A white precipitate or cloudy solution may be seen during the
course of in vitro transcription, resulting from the formation
of an insoluble compound between Mg2+ and inorganic pyro-
phosphate, which indicates that a large amount of RNA has
been produced. The insoluble compound can be easily
removed by PAGE purification.
6. To select for c-di-GMP-inhibited ribozymes, 100 μM c-di-
GMP was included in the mixture to ensure that the active
allosteric ribozymes did not cleave in the transcription step.
The starting concentration of c-di-GMP in the selection reac-
tion is influenced by its binding affinity to the aptamer. The
c-di-GMP-I aptamer used in this study has a reported KD value
in the picomolar range. Hence, it is very likely to recover all
possible c-di-GMP-dependent allosteric ribozymes by using
micromolar to nanomolar concentrations of c-di-GMP, which
is significantly above its KD value.
7. The order of addition of the components is critical for the
selection. In the negative selection for c-di-GMP-inhibited
ribozymes, RNA was firstly mixed with the assay buffer that
does not contain Mg2+. Then c-di-GMP was added and the
mixture was annealed to allow the RNAs to prefold in the
presence of 100 mM NaCl in the assay buffer. Last, 10 mM
Mg2+ (final concentration) was added to the mixture to initiate
the negative selection.
8. It is better to collect the cleaved product band in the negative
selection step, for use as a size marker in the following positive
selection step. However, one must pay attention not to
cross-contaminate the samples. To avoid the possible cross-
contamination during PAGE purification, it is suggested keep-
ing a one- or two-lane distance between the marker and the
selection product as the samples are loaded into the gel.
9. In the positive selection for c-di-GMP-inhibited ribozymes,
the same order was performed except that no c-di-GMP was
mixed with RNA. Also, the same order was applied to select
for c-di-GMP-induced ribozymes, except that c-di-GMP was

removed in the negative selection and included in the positive
selection.
10. In order to select for the most active effector-dependent ribo-
zymes, the incubation time for the positive selection reaction
was reduced after several rounds of selection. Usually in the
early rounds of selection, the RNA population was incubated
for relatively long time (15 min) to increase the recovery of
c-di-GMP-dependent self-cleaving RNAs. As more and more
rounds of selection were conducted, the duration of the incu-
bation time was gradually reduced to 1 min.
11. In the early (typically two to three) rounds of selection, the
fraction RNAs cleaved in both the positive and negative selec-
tion reactions should be at the similar level, because the effec-
tor-independent RNAs still dominate the early populations.
12. Usually the radioactive signal of the cleaved product is suffi-
cient to allow detection on an autoradiogram after an expo-
sure for 10–30 min. Using fresh [α-32P] UTP to internally
label RNA can enhance the radioactive signal of the cleaved
product for better detection.
13. The efficient recovery of cleaved RNAs from the positive selec-
tion step is important in the early rounds of selection because
the occurrence of c-di-GMP-dependent sequences is rare. To
avoid losing most molecules during the ethanol precipitation
step, glycogen (10 μg) was added as a carrier to help improve
the recovery yield.
14. In order to select for the effector-dependent ribozymes with the
best binding affinity, the concentration of c-di-GMP was
decreased as more rounds of selection were conducted. Initially
100 μM c-di-GMP was included in the selection to ensure the
maximal recovery of c-di-GMP-dependent self-cleaving RNAs.
After several rounds of selection, the concentration of c-di-
GMP was reduced to 0.3 μM to favor the isolation of the allo-
steric ribozymes that bind c-di-GMP tightly. However,
drastically decreasing the effector concentration may result in
the loss of the most sensitive effector-dependent ribozymes. It
is suggested that in the later rounds of selection, a 5- to 10-time
decrease in the effector concentration is performed each time.
15. During the first couple rounds of selection, it is recommended
to amplify all of the cDNA. This will increase the probability
that even the rarest of the active molecules will be amplified
and propagated in subsequent generations.
16. The number of cycles required for complete amplification
depends on the amount of input template DNA. For each
round of selection, the number of PCR cycles is estimated
from the quantity of input cDNA, which in turn is estimated
from the fraction of cleaved RNA in the positive selection step.

Over-amplifying cDNA with more cycles than needed will
generate undesired higher molecular weight products due to
mispriming.
17. To clone the PCR products, it is very convenient to work on a
TOPO TA cloning kit (Invitrogen) because this method does
not require restriction sites at both the two ends of the PCR
amplified product.
18. To generate DNA templates for the transcription of individual
clones for further analysis, either PCR amplifying the plasmid
DNA or chemically synthesizing the oligo-strands can be
conducted.
Acknowledgements
We thank members of the Breaker laboratory for helpful discus-

sions. RNA research in the Breaker laboratory is also supported by
the Howard Hughes Medical Institute.
References
1. Breaker RR (2004) Natural and engineered 10. Win MN, Smolke CD (2007) A modular and
nucleic acids as tools to explore biology. Nature extensible RNA-based gene-regulatory plat-
432:838–845 form for engineering cellular function. Proc
2. Famulok M (2005) Allosteric aptamers and Natl Acad Sci U S A 104:14283–14288
aptazymes as probes for screening approaches. 11. Ogawa A, Maeda M (2008) An artificial
Annu Rev Microbiol 59:487–517 aptazyme-based riboswitch and its cascading
3. Tang J, Breaker RR (1997) Rational design of system in E. coli. Chembiochem 9:206–209
allosteric ribozymes. Chem Biol 4:453–459 12. Robertson MP, Ellington AD (1999) In vitro
4. Tang J, Breaker RR (1998) Mechanism for selection of an allosteric ribozyme that trans-
allosteric inhibition of an ATP-sensitive ribo- duces analytes to amplicons. Nat Biotechnol
zyme. Nucleic Acid Res 26:4214–4221 17:62–66
5. Soukup GA, Breaker RR (1999) Design of 13. Najafi-Shoushtari SH, Famulok M (2005)
allosteric hammerhead ribozymes activated by Competitive regulation of modular allosteric
ligand-induced structure stabilization. aptazymes by a small molecule and oligonucle-
Structure 7:783–791 otide effector. RNA 11:1514–1520
6. Soukup GA, Breaker RR (1999) Engineering 14. Lee ER, Baker JL, Weinberg Z, Sudarsan N,
precision RNA molecular switches. Proc Natl Breaker RR (2010) An allosteric self-splicing
Acad Sci U S A 96:3584–3589 ribozyme triggered by a bacterial second mes-
7. Koizumi M, Soukup GA, Kerr JNQ, Breaker senger. Science 329:845–848
RR (1999) Allosteric selection of ribozymes 15. Hengge R (2009) Principles of c-di-GMP sig-
that respond to the second messengers cGMP naling in bacteria. Nat Rev Microbiol
and cAMP. Nat Struct Biol 6:1062–1071 7:263–273
8. Gu H, Furukawa K, Breaker RR (2012) 16. Sudarsan N, Lee ER, Weinberg Z, Moy RH,
Engineered allosteric ribozymes that sense the Kim JN, Link KH, Breaker RR (2008)
bacterial second messenger cyclic diguanosyl Riboswitches in eubacteria sense the second
5′-monophosphate. Anal Chem messenger cyclic di-GMP. Science
84:4935–4941 321:411–413
9. Wieland M, Benz A, Klauser B, Hartig JS 17. Mayer G, Famulok M (2006) High-
(2009) Artificial ribozyme switches containing throughput-compatible assay for glmS ribo-
natural riboswitch aptamer domains. Angew switch metabolite dependence. Chembiochem
Chem Int Ed 48:2715–2718 7:602–604
18. Lünse CE, Schmidt MS, Wittman V, Mayer G 21. Kulshina N, Baird NJ, Ferré-D’Amaré AR
(2011) Carba-sugars activate the glmS- (2009) Recognition of the bacterial second
riboswitch of Staphylococcus aureus. ACS Chem messenger cyclic diguanylate by its cognate
Biol 6:675–678 riboswitch. Nat Struct Mol Biol
19. Furukawa K, Gu H, Sudarsan N, Hayakawa Y, 16:1212–1217
Hyodo M, Breaker RR (2012) Identification of 22. Forster AC, Symons RH (1987) Self-
ligand analogues that control c-di-GMP ribo- cleavage of virusoid RNA is performed by
switches. ACS Chem Biol 7:1436–1443 the proposed 55-nucleotide active site. Cell
20. Smith KD, Lipchock SV, Ames TD, Wang J, 50:9–16
Breaker RR, Strobel SA (2009) Structural basis 23. Fedor MJ, Uhlenbeck OC (1992) Kinetics of
of ligand binding by a c-di-GMP riboswitch. intermolecular cleavage by hammerhead ribo-
Nat Struct Mol Biol 16:1218–1223 zymes. Biochemistry 31:12042–12054
Chapter 16
Dual-Selection for Evolution of In Vivo Functional

Aptazymes as Riboswitch Parts
Jonathan A. Goler, James M. Carothers, and Jay D. Keasling
Abstract
Both synthetic biology and metabolic engineering are aided by the development of genetic control parts.
One class of riboswitch parts that has great potential for sensing and regulation of protein levels is aptamer-
coupled ribozymes (aptazymes). These devices are comprised of an aptamer domain selected to bind a
particular ligand, a ribozyme domain, and a communication module that regulates the ribozyme activity
based on the state of the aptamer. We describe a broadly applicable method for coupling a novel, newly
selected aptamer to a ribozyme to generate functional aptazymes via in vitro and in vivo selection. To
illustrate this approach, we describe experimental procedures for selecting aptazymes assembled from
aptamers that bind p-amino-phenylalanine and a hammerhead ribozyme. Because this method uses selec-
tion, it does not rely on sequence-specific design and thus should be generalizable for the generation of in
vivo operational aptazymes that respond to any targeted molecules.
Key words In vitro selection, RNA aptamer, Ribozyme, Aptazyme, Riboswitch, Synthetic biology
1 Introduction
Synthetic biology in general, as well as construction of novel path-

ways in cells, needs control devices that can regulate cellular behav-
ior. Using the circuit metaphor, these devices are thought of as
switches or gates, while in the microbial chemical factory meta-
phor, the can be thought of as sensors and control valves. In both
cases, we are concerned with similar parameters: the input signal
level and the output range; additionally, response time may be an
important factor. Natural RNA aptamers are known to bind small
molecules in order to regulate several systems, including thiamine
synthesis, B12 transport, and the synthesis of several nucleotides
[1–9]. In addition, several examples of synthetic aptamer-based
gene regulatory riboswitches have been successfully demonstrated
in bacteria, yeast and mammalian cells [10]. Extensive work has
shown that ligand-responsive, self-cleaving ribozymes (aptazymes)
can be engineered by connecting ligand-binding aptamer domains
221
222 Jonathan A. Goler et al.
to the self-cleaving hammerhead ribozyme core structure through

secondary structures utilizing a helix-slipping mechanism, in which
three or more tracts of nucleotides interact differently based on the
physical presence of another molecule [11–14]. This effect changes
the three-dimensional structure of the aptazyme, which can cause
it to become either more or less active, in terms of the rate of phos-
phodiester bond self-cleavage. In this chapter, we describe a gener-
alizable dual-selection method for engineering functional
aptazymes that can be used as riboswitch parts from individual
aptamers and ribozymes.
To illustrate our method for rapidly generating novel apta-
zymes, we describe the selection of a composite aptazyme structure
as a sensor RNA component that is capable of detecting p-amino-
phenylalanine (pAF) in in vivo-like conditions. pAF is a non-
proteinogenic amino acid (MW = 180 g/mol) that is a precursor to
the pristinamycin family of antibiotics and other aromatic com-
pounds with potential industrial-relevance. It is cell-permeable and
can be produced via a 12 gene engineered pathway [15]. In the
current example, the pAF-R1-1 aptamer, previously selected to
bind the target ligand with micromolar dissociation constant in
physiological buffer conditions [16], is coupled to a Schistosoma
mansoni hammerhead ribozyme. In principle, the same methods
can be applied to engineer functional aptazymes from almost any
desired ligand-binding aptamer domain and any one of many dif-
ferent self-cleaving phosphodiester bond cleaving ribozymes [17].
The general outline of the first, in vitro, stage of the method is
based on work by Link et al. [18], in which they generated theoph-
ylline aptazymes from a combinatorial library comprised of the
ubiquitous Jenison theophylline aptamer [19] coupled to an
S. mansoni ribozyme. The combinatorial sequence library used to
generate the composite aptazymes structures had three parts of the
sequence bridging the aptamer and ribozyme domains random-
ized, with two 5-base tracts, and one 4-base tract. Although Link
et al. found that their aptazymes demonstrated tremendous (285-
fold) differences in cleavage rates in the presence versus the absence
of ligand in vitro, they were not able to dynamically regulate gene
expression in the in vivo context they employed. Later experiments
by Carothers et al. showed that model-driven approaches could be
employed to design a genetic control mechanism and sequence
context in which the Link aptazymes could be assembled into
functional aptazyme-Regulated Expression Devices (aREDS) with
quantitatively predictable genetic outputs in E. coli [15]. The dual-
selection method given here was developed in the course of gener-
ating pAF-responsive aptazyme parts.
Briefly, a combinatorial sequence library is designed where the
aptamer is attached to the ribozyme at the base-stem, via random-
ized tracts, and in vitro transcribed. Additionally, the size of a
library is important: transformation efficiency into E. coli is limited
Dual Selection of Aptazyme Parts 223
Fig. 1 Dual-selection for ligand-dependent self-cleaving aptazymes. (a) To generate aptazymes, a combinato-
rial sequence library is constructed from individual aptamer and ribozyme domains. In the illustration, a
sequence library comprises the pAF-R1-1 aptamer (kd = 19 μm) and S. mansoni hammerhead ribozyme
(kobs = 1.3 min−1), shown in secondary structure schematic, was constructed. 14 fully randomized bases within
stem I of the hammerhead ribozyme are colored and marked N1, N2, and N3. Aptazyme self-cleavage takes
place at the site indicated by the red triangle. PAGE is used to enrich the library pool for aptazymes with fast
kobs+ by isolating RNA that cleaves rapidly in the presence of ligand (dashed box at left) and for aptazymes with
slow kobs− by isolating RNA that does not cleave in the absence of ligand (dashed box at right). The complete
cycle is repeated 4–6 times. (b) Candidate aptazymes from the in vitro selection are subcloned into a genetic
device for aptazyme-regulated expression screening. A subsequent screen for ligand-dependent regulation of
reporter gene expression identifies aptazymes that catalyze self-cleavage reactions inside the cell
to approximately 109, so libraries larger than that cannot be

explored solely in vivo [20]. The first stage of (in vitro) selection is
then used to enrich for a pool of RNAs that preferentially cleave in
the presence of the ligand, but not in the absence of it. In the sec-
ond stage, the resulting molecules are then cloned into a genetic
device (Fig. 1) for aptazyme-regulated expression screening in
E. coli. Individual bacterial colonies are picked, grown and assayed
for differences in reporter gene expression as a function of exoge-
nous ligand added to the growth media. DNA from individual
colonies is sequenced to identify individual aptazyme parts
exhibiting ligand-dependent phosphodiester bond cleavage activ-

ity. Finally, the kinetics of ligand-dependent cleavage are assayed
for aptazymes showing function in vivo, using RNA transcribed in
vitro from the isolated sequences.
In the context of an aptamer-based ribozyme, the function of
the aptazyme is dependent on the kd of the aptamer, and the fold-
ing (kfold) and cleavage rates in the presence (kobs+ and absence kobs−)
of the ligand that collectively determine part function. The manner
in which the selections are conducted makes a large impact on the
functionality of the resulting sequences. To generate structures
meeting targeted performance criteria, pressure can be applied in
terms of time, which affects the folding and catalytic rate constants,
and ligand concentrations, which affects the apparent kd of the
aptazyme, to tune the in vitro selection stage. If the aptazymes do
not perform as needed, the selections can be biased to achieve dif-
ferent results, for example, the length of time the positive selec-
tions are run can be shortened and the length of the negative
selections can be lengthened to increase the aptazyme ligand-
activation ratios (kobs+/kobs−).
The co-transcriptional folding contexts in which aptazymes are
intended to be active are important for determining whether the
parts can be used to build genetic control devices functional in vivo
[15, 18, 21, 22]. At a minimum, this context includes the cellular
environment, pH, temperature, as well as the local flanking
sequences and identities of the promoter and RNA polymerase,
which affect transcription elongation rates, and thus the kinetics of
part folding. In light of these considerations, the first stage, in vitro
selection should be performed in physiological buffer conditions
[16]. Given local sequence constraints that may prevent aptazyme
folding in the context of an elongating transcript in vivo, it is
important to realize that the in vivo screen will enable the identifi-
cation of functional aptazyme parts, but will not tend to exhaus-
tively identify all of the active sequences in a given pool.
To assemble aptazyme component parts into genetic expres-
sion devices with quantitatively predictable functions, we previ-
ously utilized an underlying control mechanism that targets specific
RNA degradation pathways in the cell [23]. The aptazyme self-
cleavage reaction produces a 5′ hydroxyl-terminated downstream
signal that is degraded by a relatively slow RppH-independent
pathway, the effect of which is a predictable increase in mRNA
half-life. By developing a biochemical and biophysical model-
driven approach, we showed that aptazymes generated and charac-
terized in the ways described here could be assembled into aREDs
with the capacity to quantitatively program gene expression levels
in vivo. In principle there are many other mechanisms and contexts
in which aptazyme parts generated through dual in vitro and in
vivo selection could be employed to build sensors and genetic con-
trol devices for engineered circuits and pathways [24–26].
Selecting, rather than designing ligand-responsive aptazymes,

is a robust method for generating component parts that can be
used to engineer aptazyme-regulated expression devices (aREDs)
functional in vivo [15]. The use of a two-stage method allows one
to take advantage of the high sequence diversity that can be
screened with in vitro selection, while utilizing the in vivo selection
to rapidly screen the pool of enriched molecules for aptazyme
sequences that can undergo self-cleavage in a cellular environment.
The generality of the design and procedure should permit its appli-
cation in a wide range of potential applications where an aptamer
to an intermediate, signaling molecule, or protein can be isolated.
2 Materials
Prepare all solutions using RNAase-free water (fresh 18 MΩ pure

water from a Milli-Q system for example).
2.1 In Vitro Selection 1. In vitro Transcription Mix: 40 mM Tris–HCl, pH 7.8, 20 mM

each nucleotide triphosphate (NTP), 25 mM MgCl2, 2.5 mM
2.1.1 Solutions
Spermidine, 10 mM dithiothreitol (DTT), 600 μM Top Strand
and Buffers
ssDNA, 500 μM Template ssDNA, 50 units/μL T7 Polymerase,
10 units/μL Thermostable inorganic pyrophosphatase (TIPP,
New England Biolabs, Inc. USA), 200 units/μL RNAsin
(Promega, Inc. USA), 0.1 μCi 32P-ATP (where desired),
0.01 % Triton X-100.
2. 10× TBE Buffer: 108 g of Tris base, 55 g of boric acid, 40 mL
of 0.5 mM EDTA in 1 L of RNAse-free water (see Note 1).
Utilize 1× TBE in PAGE gels and the PAGE running buffer,
0.5× TBE in the electro-elution steps.
3. 2× Selection Buffer (2× SB): 130 mM K-Glutamine, pH 7.5,
15 mM NaCl, 10 mM DTT, 1 mM MgCl2.
2.1.2 PAGE Gel 1. APS solution: 25 mg of ammonium per sulfate in 100 mL of

RNAse-free water and vortex (see Note 2).
2. TEMED.
3. Acrylamide (19:1).
4. RNAse-free urea.
5. Clean razor blades.
6. TLC (thin-layer chromatography) plates.
7. UW wand.
8. Glass plates and plastic separators.
9. Binder clips.
10. 2× Loading buffer: 95 % formamide, 5 % 0.5 M EDTA.
2.1.3 Ethanol 1. 3 M RNAse-free KCl.

Precipitation
2. RNAse-free pure or 95 % ethanol.
3. Dry ice or a −20 °C freezer.
2.1.4 Equipment 1. Whatman Elutrap System.

2. NanoDrop Spectrophotometer (ThermoFisher, Inc. USA).
3. Ready-To-Go RT-PCR beads (GE Healthcare Inc., USA).
4. Plate Reader such as Tecan Safire.
5. Corning Costar plates (or other 96-well plates compatible with
your plate reader).
6. Phosphorimager Screen and Reader (such as a Typhoon 8600).
2.2 In Vivo Screen 1. LB media.

2. BL-21 cells (Life Technologies, Inc., USA).
3. Top-100 Chemically competent cells (Invitrogen).
4. Ice.
5. MOPS M-9 media.
6. LB-agar plates with kanamycin (Sigma).
7. 0.5 mM Isopropyl-d-1-thiogalactopyranoside (IPTG).
3 Methods
3.1 PAGE Gel Assemble the glass with spacers, prepare gel by adding acrylamide,
urea, and TBE (5 mL of 10× TBE, 12 mL of 20 % acrylamide, 24 g
of urea and water to 50 mL); add 50 μL of TEMED and 500 mL
of APS last and swirl to initiate the polymerization reaction
3.2 Running 1. Pre-run the gels using constant power (18–22 W) for
PAGE Gels 20–30 min to warm the gel a bit, the gel should be approxi-
mately 40–50 °C (see Note 4).
2. After the pre-run, clean the urea from the wells by utilizing a
sterile syringe. Withdraw buffer and expel firmly to flush out
the wells, you should see the urea being blown out.
3. Add loading buffer (95 % formamide, 5 % 0.5 M EDTA) to the
sample in a 1:1 ratio and gently mix.
4. Load the samples into the gel (see Note 5), actuate the power
and run the gel for 1–1.25 h.
5. Once the gel is complete, separate the gel from the glass plates
for UV shadowing using a fluorescent PEI-Cellulose TLC
plate (see Note 6).
6. Briefly illuminate the gel with a UV wand set to a short wave-

length (preferably 254 nm), using a marker to mark the band
locations.
7. Put aside the TLC plate and place the wrapped gel back on a
glass plate, excise the desired band with a clean razor blade.
8. If you are quantifying band intensity, expose the gel to a freshly
erased phosphor screen.
3.3 Stage One: In this illustration, we used the pAF-binding aptamer pAF-R1-1
In Vitro Aptazyme (kd = 30 μM at 2.5 mM Mg2+ [16], which was connected to the
Selection S. mansoni ribozyme via randomized regions as in Link et al. [18]
to construct the starting combinatorial sequence library. Oligos
were as follows (Integrated DNA Technologies): LSW-11F
5′-GCT AA TAC GAC TCA CTA TAG GCG AAA GCC GGC
GCG TCC TGG ATT CCA CNN NNN CAT GTC CCT ACC
ATA CGG GAT TGC CCA GCT TCG GCT GCC ATG CCG
GC, LSW-11Rc: 5′-AGC GCG TTT CGT CCT ATT TGG GAC
TCA TCA GCN NNN TGT ACC NNN NNC GTA GGC CGG
TTA CCG TTT GGC CGG CAT GGC AGC CG and primers
LSW-11Fp: 5′-GCT AAT ACG ACT CAC TAT AGG CGA AAG
CCG GCG CGT CCT GGA TTC CAC and LSW-11Rp: 5′-AGC
GCG TTT CGT CCT ATT TGG G (also see Note 7).
3.3.1 PAGE Purification The DNA oligos are PAGE-purified, primer-extended, and tran-
of Oligo DNA scribed in vitro, as detailed below.
1. First, solubilize the oligos in ultrapure water, add 1:1 2× load-
ing buffer and vortex.
2. Purify oligos on a 10–12 % PAGE gel, running gels as above.
3. Illuminate gels with UV light against a PEI-Cellulose TLC
plate, mark and excise bands with a razor blade.
4. Elute gel slices (Whatman Elutrap System) at 300 V for 30 min
to 2 h into a final volume between 200 and 400 μL (see Note 8).
5. Ethanol precipitate the eluate with 1/10 volume of 3 M KCl
and 2.5 volumes of 95 % ethanol, chill for 10 min on dry ice
(solution should become slightly whiter and more viscous),
spin for 15 min at 24,000 × g in a refrigerated centrifuge. Decant
using a pipette and air dry at room temperature or 37 °C.
6. Quantify DNA using a NanoDrop or other spectrophotometer.
3.3.2 In Vitro Selection 1. Prepare transcription mix according to the materials section
(see Note 9).
2. Add the template and top strand oligos (if needed) to the tran-
scription mix.
3. Run transcription at 20–25 °C for 15 min to 2 h.
4. Stop transcription reactions with an equal volume of 2× loading

buffer, mix and run out on an 8 % PAGE gel at 22 W for
1–1.25 h.
5. Excise the larger band (full length RNA), elute as above,
ethanol-precipitate, and dry the pellet.
6. Resuspend the pellet in water, flicking several times, and lightly
vortexing.
7. Add 1× Selection Buffer (SB) with or without the desired
ligand and incubate at 37 °C for 5–15 min.
8. In the positive selections (with ligand), the smaller (cleaved)
band is excised. For the negative selections, the larger
(uncleaved) fraction is excised.
9. Other purification, gel excision, electro-elution, precipitation,
are performed as above.
10. The resultant RNA is resuspended in ultrapure water and sub-
jected to 8–12 rounds of RT-PCR using Ready-To-Go beads
using the manufacturer’s protocol. After 8–12 rounds of PCR,
the reactions are visualized on a 2 % agarose gel; if there is no
band of the expected length, perform three to four more cycles
of basic PCR and recheck (see Note 10).
3.4 Stage Two: In After four to six rounds of in vitro aptazyme selection, the apta-
Vivo Aptazyme Screen zyme parts are screened for ligand-dependent self-cleavage in vivo.
1. The enriched, surviving, sequences are PCR amplified using
cloning adapters (For the pAF aptazyme selections these are
Adapter4A.F: 5′-GAT ATC ATC TCT AGA GGC GAA AGC
CGG CGC GTC CTG GAT TCC AC, Adapter4AR: 5′-GAC
CTT CTC GAG AGC GCG TTT CGT CCT ATT TGG G),
and cloned into prRED-T7-ref (Fig. 1), which harbors the
kanamycin (Kan) resistance gene, a T7 promoter, an XbaI-
BglII cloning site, and the gene encoding the dsRedExpress
Red Fluorescent Protein (RFP) [15] (see Note 11).
2. The resulting plasmids are then transformed initially into E. coli
TOP 10 chemically competent cells.
3. For the transformation, add a 2–3 μL of the enriched library
DNA to the cells.
4. Incubate on ice for 10 min.
5. Place aliquots in a 42 °C water bath for 15 s, then back on ice
for 2 min.
6. Incubate aliquots at 37 °C, shaking, for 75 min.
7. Plate cultures on LB-Kan plates and incubate overnight.
8. The next day, pick individual colonies into test tubes, each with
5 mL of LB medium containing 50 μg/mL kanamycin
(LB-Kan).
9. Incubate tubes at 37 °C, shaking, 12 h or overnight.

10. Use a standard plasmid prep to recover plasmids.
11. Re-transform each plasmid into E. coli BL-21 chemically com-
petent cells harboring an IPTG-induced T7 expression
cassette.
12. Individual colonies are picked and grown for 8 h to overnight
in LB-Kan.
13. To assay performance, these cultures are then used to inoculate
six tubes containing 5 mL of LB-Kan. Three of these tubes
have either none, or saturating concentrations, of the desired
ligand (in the current illustration, 0 and 5 mM pAF were used).
0.5 mM IPTG is added to all of the tubes to induce transcrip-
tion from the T7 promoter, and the cultures are incubated
with shaking.
14. Samples are then assayed at 12, 24, 36, and 48 h. 100 μL from
each culture is taken and added to a 96-well Costar plate and
assayed for optical density (OD380) and fluorescence (excita-
tion measured at 557 nm; emission measured at 579 nm) in a
plate reader.
15. Samples that exhibit a difference in expression ± ligand are then
plasmid-prepped, sequenced and re-transformed into fresh
BL-21 cells, and re-assayed to confirm the ligand-dependent
cleavage activity, one such resulting structure is shown in
Fig. 2.
16. To sequence samples, mini-prep the cultures that exhibit a dif-
ference in expression, you can then fold the sequence using an
RNA folding program like Vienna RNA, producing an image
such as Fig. 2.
3.5 Measuring 1. To confirm that the sequences exhibiting ligand-dependent

Aptazyme Kinetics differences in gene expression observed in the in vivo screen
In Vitro encode functional aptazymes, you must show that the presence
of the ligand significantly increases the cleavage rate constant
in vitro.
2. DNA from colonies exhibiting switching behavior is re-
amplified using the original selection primers with T7 pro-
moter and transcribed with 32 P-ATP. Transcribe sufficient
quantities for several experiments (200–500 μL) and purify as
above.
3. Resuspend purified RNA in water (100 μL).
4. Add RNA to selection buffer SB with or without the desired
target ligand (in triplicate) mix, and hold at the appropriate
temperature (e.g., 30, or 37 °C) for 1, 3, 5, 10, and 15 min,
stopping the reaction by adding 2× loading buffer and quickly
mixing.
Fig. 2 Winner from the example pAF aptazyme selection, W24. Secondary struc-
ture of the minimum free energy fold, as computed with Vienna RNAfold with
default settings, for the best performing pAF aptazyme from the example selec-
tion is shown with number and color coding as in Fig. 1
a 1.0 b
1.0
0.9
0.8
Fraction Uncleaved RNA
Fraction Uncleaved RNA

0.8
0.6
0.7
0.6
0.4
0.5
0.2
0.4
0.0
0.3
0 5 10 15 0 5 10 15
Time (Minutes) Time (Minutes)
Fig. 3 Measuring the kinetics of aptazyme cleavage in vitro. The kinetics of ligand-dependent cleavage are
assayed for aptazymes showing function in vivo, using RNA transcribed in vitro from the isolated sequences.
In this figure, results for two isolates from the pAF aptazyme selections are shown. The fraction of self-cleaved
aptazyme in the presence (open triangles) and absence (solid triangles) of the target ligand is plotted as a
function of time. The data are fitted to a first-order decay curve, overlaid
5. Load samples on a 10–12 % PAGE gel and run for 90 min.

6. Expose the gel to a phosphorimager screen for 20 min, and scan
on a Typhoon 8600 (50 μm resolution), or similar equipment.
7. Measure bands and analyze for volume (counts over an area).
8. Plot the fraction of aptazyme RNA cleaved as a function of
time and use nonlinear regression to calculate the aptazyme
cleavage rate constant [18]. An example is shown in Fig. 3.
To further confirm that functional ligand-responsive apta-
zymes have been isolated, the expression device plasmids can be
re-transformed into E. coli BL-21 cells, and re-assayed in triplicate
against a range of target ligand concentrations. For instance, Fig. 4
shows that the genetic device encoding the W24 aptazyme exhibits
clear, dose-dependent, differences in gene expression upon the
addition.
4 Notes
1. For TBE buffer, prepare the solution in a bottle, add 70 % of

the water to begin, add a RNAse-free stir bar, and put on a stir
plate with heating, get the stirrer rotating and slowly add the
dry chemicals, finally adding water to 1 L, and allow to dis-
solve. If you work in a general molecular biology lab, it is
important to keep a separate supply of chemicals to prevent
contamination with RNAse.
1.5
1.4
(relative to 0 mM pAF)
Fluorescence change
1.3
1.2
1.1
1.0
0 2 4 6 8 10
pAF concentration (mM)
Fig. 4 Aptazyme-regulated genetic expression. The genetic device encoding the

W24 aptazyme exhibits clear, dose-dependent, differences in gene expression
upon the addition of pAF to the media, confirming the isolation of functional
ligand-responsive aptazymes. In the example, isolate W24 was assayed in tripli-
cate, with RFP expression normalized to RFU/OD380 and plotted relative to a
negative control, as measured 48 h post-induction
2. APS solution should be prepared fresh or stored refrigerated

not more than 2 weeks. Prepare by adding 25 mg of ammo-
nium persulfate to 100 mL of ultrapure water, vortex until
dissolved.
3. For PAGE, create the gel utilizing glass plates, one 1–2 cm
longer than the other, with 1.5 mM spacers and comb (you
can utilize either a multichanneled comb to quantify and mea-
sure a series of samples, or a single wide comb for purification
of a large transcription. Clean the plates with RNAse away or
similar reagent, followed by a wash with running Milli-Q water.
To pour PAGE gels, assemble the glass with spaces, leave the
comb aside for now, the bottoms of the plates should be
aligned (The top has the differential height to allow the buffer
to flow to the channels but not spill). Secure the plates in place
using binder clips, ensuring that they are springy enough to
hold the spacers and glass together tightly. Carefully pour the
solution into the plate-rig, ensuring that no bubbles form; you
can shift the position or lightly tap the rig to allow the bubbles
to rise to the surface. Add the comb into the top of the gel.
Watch the edges of the rig to ensure that there are no leaks.
The gel should be ready in 30–45 min.
4. When running gels, we utilize constant power (as opposed to
voltage), from 18–22 W. It is important to ensure that the gel
does not overheat (the gel should be approximately 40–50 °C,

if it feels very hot, turn down the power), high temperature
could break the glass plates as well as cause bands to smear.
5. Utilize a long-tipped pipette to load the samples into the
desired wells of a PAGE gel, putting the solution at the bottom
of each well. You may choose to utilize dye (xylene cyanol and
bromphenol blue) with the sample or as tracking/ladder in a
separate well. The sample’s index of refraction is different from
the buffer, so you should be able to see it being injected into
the well; do not overfill the wells (you may see it flow over the
sides of the well) that will prevent a clean separation.
6. Once the gel is complete, remove the rig, remove the edges,
and very carefully separate the gel from the glass plates. The
first plate should separate, if it is difficult, use a tweezer to get
it started but proceed slowly to not damage the gel. Once the
first plate is removed, add a piece of saran wrap in its place,
ensuring that it is flush; you can then add a fluorescent PEI-
Cellulose TLC plate on top of the saran wrap, flip the gel over
and slowly remove the second plate. Replace with a piece of
saran wrap. If you are removing a band, find a dark place or
turn off the lights, briefly illuminate the gel with a UV wand
set to a short wavelength (preferably 254 nm), using a marker
to mark the band locations. Put aside the TLC plate and place
the wrapped gel back on a glass plate, excise the desired band
with a clean razor blade. If you are quantifying band intensity,
expose the gel to a freshly erased phosphor screen for 15 min
to 4 h, depending on the intensity of the 32P used, scan the
screen (we utilize a Typhoon 8600 with 50 μm resolution).
Utilize the included software package to measure the volume
(counts in a given area).
7. To evolve a new custom aptazyme, create oligos that encom-
pass the aptamer, the randomized section and ribozyme core,
and PCR primers that regenerate the otherwise-cleaved por-
tion of the ribozyme (Fig. 1). Since the pool is randomized,
the final structure cannot be elucidated directly, but it can be
biased towards forming the binding/cleavage domains by
ensuring that the aptamer and ribozyme base stems are con-
served by one to three bases (additional randomization may
actually result in increased activity differential ± ligand Order
DNA for the oligos from your preferred vendor. You will purify
the oligos yourself, so simply have them de-salted. Upon
receipt, solubilize the oligos in ultrapure water, add 1:1 2×
loading buffer and vortex. Purify your oligos on a 10–12 %
PAGE gel, running gels as above. Illuminate gels with UV
light against a PEI-Cellulose TLC plate, mark and excise bands
with a razor blade. Illuminate the gels in a dark room and min-
imize the time both you and the gel are exposed to UV, both
for safety and to minimize damage to the nucleic acid library.
8. When using an Elu-trap device to electroelute DNA or RNA

from gel slices, elution buffer should be prepared as 0.5× TBE
buffer, and fresh buffer utilized each time. Assemble the elu-
tion modules each time, and place gel slices close to the cap-
ture section. Fill the device with buffer such that the slices are
just covered by the buffer. Ensure that any unused modules are
blocked from the buffer, as they will “short out” the current.
Carefully draw the eluted material from the capture section
and aliquot into microcentrifuge tubes.
9. For in vitro transcription, pools and the first two rounds are
conducted in large volumes (1 mL) while remaining rounds
are 100–200 L volume. Running the transcription at 20–25 °C
will result in less mass of RNA being produced than at 37 °C,
but will yield more full-length transcripts. Longer transcription
times will result in a greater number of RNA molecules, but a
larger proportion of cleaved RNA molecules.
10. Stage One in vitro selection cycles may need to be repeated
approximately 4–6 times until the pool of evolved RNA mol-
ecules is dominated by sequences that undergo substantially
more cleavage in the presence than in the absence of ligand, as
judged by the relative sizes of the corresponding bands on the
selection PAGE gel. After each round, template DNA from
each step should be marked at stored at −80 °C.
11. We note that common subcloning techniques, such as SLIC
and CPEC, may not be successful for owing to the substantial
secondary structure present in the cognate single stranded-
DNAs [27].
12. Create the gel utilizing glass plates, one 1–2 cm longer than
the other, with 1.5 mM spacers and comb (you can utilize
either a multichanneled comb to quantify and measure a series
of samples, or a single wide comb for purification of a large
transcription. Clean the plates with RNAse away or similar
reagent, followed by a wash with running Milli-Q water.
Assemble the glass with the spacers, leave the comb aside for
now, the bottoms of the plates should be aligned (The top has
the differential height to allow the buffer to flow to the chan-
nels but not spill). Secure the plates in place using binder clips,
ensuring that they are springy enough to hold the spacers and
glass together tightly. Prepare gel by adding acrylamide, urea,
and TBE, add TEMED and APS last and swirl to initiate the
polymerization reaction. Carefully pour the solution into
the plate-rig, ensuring that no bubbles form; you can shift the
position or lightly tap the rig to allow the bubbles to rise to the
surface. Add the comb into the top of the gel. Watch the edges
of the rig to ensure that there are no leaks. The gel should be
ready in 30–45 min.
References
1. Vitreschak AG, Rodionov DA, Mironov AA, 14. Soukup GA, Breaker RR (1999) Engineering
Gelfand MS (2004) Riboswitches: the oldest precision RNA molecular switches. Proc Natl
mechanism for the regulation of gene expres- Acad Sci U S A 96:3584–3589
sion. Trends Genet 20:44–50 15. Carothers JM, Goler JA, Juminaga D, Keasling
2. Stormo GD, Ji Y (2001) Do mRNAs act as JD (2011) Model-driven engineering of RNA
direct sensors of small molecules to control devices to quantitatively program gene expres-
their expression? Proc Natl Acad Sci U S A sion. Science 334:1716–1719
98:9465–9467 16. Carothers JM, Goler JA, Kapoor R, Lara LD,
3. Patel DJ, Suri AK, Jiang F et al (1997) Keasling JD (2010) Selecting RNA aptamers
Structure, recognition and adaptive binding in for synthetic biology: investigating magnesium
RNA aptamer complexes. J Mol Biol 272: dependence and predicting binding affinity.
645–664 Nucleic Acids Res 38:2736–2747
4. Tuerk C (1997) Using the SELEX combinato- 17. Thompson KM, Syrett HA, Knudsen SM,
rial chemistry process to find high affinity Ellington AD (2002) Group I aptazymes as
nucleic acid ligands to target molecules. genetic regulatory switches. BMC Biotechnol
Methods Mol Biol 67:219–230 2:21
5. Klug SJ, Hüttenhofer A, Famulok M (2000) In 18. Link KH, Guo L, Ames TD et al (2007)
vitro selection of RNA aptamers that bind spe- Engineering high-speed allosteric hammerhead
cial elongation factor SelB, a protein with mul- ribozymes. Biol Chem 388:779–786
tiple RNA-binding sites, reveals one major 19. Jenison RD, Gill SC, Pardi A, Polisky B (1994)
interaction domain at the carboxyl terminus. High-resolution molecular discrimination by
RNA 5:1180–1190 RNA. Science 263:1425–1429
6. Lorsch JR, Szostak JW (1994) In vitro selec- 20. Dower WJ, Cwirla SE (1992) Guide to electro-
tion of RNA aptamers specific for cyanocobala- poration and electrofusion. Academic, San Diego
min. Biochemistry 33:973–982 21. Isambert H, Siggia E (2000) Modeling RNA
7. Winkler WC, Breaker RR (2003) Genetic con- folding paths with pseudo-knots: application to
trol by metabolite-binding riboswitches. hepatitis delta virus ribozyme. Proc Natl Acad
Chembiochem 4:1024–1032 Sci U S A 97:6515–6520
8. Nahvi A, Sudarsan N, Ebert MS et al (2002) 22. Xayaphoummine A, Viasnoff V, Harlepp S,
Genetic control by a metabolite binding Isambert H (2007) Encoding folding paths of
mRNA. Chem Biol 9:1043–1049 RNA switches. Nucleic Acids Res 35:614–622
9. Nahvi A, Barrick JE, Breaker RR (2004) 23. Deana A, Celesnik H, Belasco JG (2008) The
Coenzyme B 12 riboswitches are widespread bacterial enzyme RppH triggers messenger
genetic control elements in prokaryotes. RNA degradation by 5-prime pyrophosphate
Nucleic Acids Res 32:143–150 removal. Nature 451:355–358
10. Liu CC, Arkin AP (2010) The case for RNA. 24. Bayer TS, Smolke CD (2005) Programmable
Science 330:1185–1186 ligand-controlled riboregulators of eukaryotic
11. Hall B, Hesselberth JR, Ellington AD (2007) gene expression. Nat Biotechnol 23:337–343
Computational selection of nucleic acid bio- 25. Wieland M, Hartig JS (2008) Improved apta-
sensors via a slip structure model. Biosens zyme design and in vivo screening enable ribo-
Bioelectron 22:1939–1947 switching in bacteria. Angew Chem Int Ed
12. Lynch SA, Desai SK, Sajja HK, Gallivan JP 47:2604–2607
(2007) A high-throughput screen for synthetic 26. Weigand J, Sanchez M, Gunnesch EB et al
riboswitches reveals mechanistic insights into (2008) Screening for engineered neomycin
their function. Chem Biol 14:173–184 riboswitches that control translation initiation.
13. Suess B, Fink B, Berens C et al (2004) A the- RNA 14:89–97
ophylline responsive riboswitch based on helix 27. Hillson NJ, Rosengarten RD, Keasling JD
slipping controls gene expression in vivo. (2011) j5 DNA assembly design automation
Nucleic Acids Res 32:1610–1614 software. ACS Synth Biol 1:14–21
Chapter 17
In Vivo Screening for Aptazyme-Based Bacterial

Riboswitches
Charlotte Rehm and Jörg S. Hartig
Abstract
In many synthetic biology applications, modular and easily accessible tools for controlling gene expression
are required. In addition, in vivo biosensors and diagnostic devices will become more important in the
future to allow for noninvasive determination of protein, ion, or small molecule metabolite levels. In recent
years synthetic RNA-based switches have been developed to act as signal transducers to convert a binding
event of a small molecule (input) into a detectable output. Their modular design allows the development
of a variety of molecular switches to be used in biochemical assays or inside living cells. RNA switches
developed by our group are based on the Schistosoma mansoni hammerhead ribozyme, a self-cleaving RNA
sequence that can be inserted into any RNA of interest. Connection to an aptamer sensing a small mole-
cule renders the cleavage reaction ligand-dependent. In the past we have successfully designed and applied
such hammerhead aptazymes for the allosteric control of both bacterial and eukaryotic gene expression by
affecting transcription elongation, translation initiation, or mRNA stability. In order to yield functional
switches optimization of the connecting sequence between the aptamer and the HHR needs to be carried
out. We have therefore developed an in vivo screening protocol detailed in this chapter that allows the
identification of functional aptazymes in bacteria.
Key words Riboswitch, Allosteric ribozyme, Biosensor, Hammerhead aptazyme, In vivo screening,
Aptamer, Gene expression, Regulation
1 Introduction
Aptamers are short nucleic acids motifs which can specifically rec-
ognize small molecule interaction partners or even ions with high
specificity and affinity [1]. They were originally obtained by sys-
tematic evolution of ligands by exponential enrichment (SELEX)
[2, 3]. However, Breaker and coworkers later proofed the exis-
tence of RNA aptamers in the 5′ untranslated regions (5′ UTR) of
many bacterial messenger RNAs (mRNAs). There they form selec-
tive binding pockets for a respective metabolite and serve as sen-
sors of intracellular metabolite concentrations in regulatory RNA
elements called riboswitches [4–6]. Metabolite binding to the
237
238 Charlotte Rehm and Jörg S. Hartig
aptamer domain induces structural rearrangements in an adjoining

expression platform and consequently switches expression of the
downstream gene on or off by affecting transcription elongation or
translation initiation. The major advantages of the use of ribo-
switches in vivo are that protein factors are not required to induce
changes in gene expression and that the regulatory active device is
covalently attached to the encoded message.
Synthetic RNA-based devices containing aptamers have been
developed that process the binding of the aptamer ligand (input)
and convert it into a variety of detectable physical signals (output),
thereby acting as gene regulatory switches or biosensors, both in
biochemical assays or inside living cells [1, 7–9]. In our group we are
constructing artificial riboswitches using the Schistosoma mansoni
hammerhead ribozyme (HHR) as central RNA scaffold [10]. The
HHR is a short self-cleaving RNA motif formed by three helical
arms (stems I–III) surrounding a conserved catalytic core (Fig. 1b).
Fig. 1 The hammerhead ribozyme as catalytically active domain. (a) Reaction mechanism of the intramolecular
strand-cleavage reaction by in-line nucleophilic attack of the 2′ hydroxyl group. (b) Example of a hammerhead
aptazyme inserted into the 5′ untranslated region of a bacterial mRNA. The RBS is incorporated into stem I and
blocked from access by the ribosome, an aptamer is attached at stem III. Stabilizing tertiary interactions
between stem I and II are shown. The cleavage site is marked by the arrowhead. Variability of the randomiza-
tion site allows screening for activating as well as inactivating aptazymes
In Vivo Screening for Aptazyme-Based Bacterial Riboswitches 239
In the catalytically active conformation site-directed RNA strand

cleavage (in cis) occurs by 2′-hydroxyl group of C17 attacking the
phosphodiester, thus generating a 2′-3′-pentacyclic phosphodiester
and a free 5′-hydroxl group (Fig. 1a) [11]. The cleavage activity
strongly depends on the presence of divalent metals ions, e.g., Mg2+,
and is further enhanced by tertiary interactions between stem I and
II enabling efficient cleavage at low Mg2+ concentrations [12].
In order to achieve allosteric control of the cleavage reaction
aptamers can be attached to the HHR. Breaker and coworkers first
demonstrated allosteric control of a HHR by connecting an ATP-
aptamer with the ribozyme, creating a so-called aptazyme [13]. In
the past we have successfully constructed hammerhead aptazymes
(HHAz) by connecting theophylline- and thiamine pyrophosphate-
aptamers at stem III (Fig. 1b) and thereby rendering the cleavage
activity ligand-dependent [14, 15]. Variability of the connecting
sequences allows the construction of ON- as well OFF-switching
aptazymes. Artificial expression platforms can be attached at stem
I to create functional switches of gene expression. In further stud-
ies we and others have shown that is possible to attach the aptamer
at any of the three stems of the HHR [15–18]. However, preserv-
ing stem I and II interactions is advisable as to avoid special growth
condition, e.g., high Mg2+ concentrations. In particular, we have
constructed a very efficient artificial riboswitch in bacteria by
expanding the full-length HHR by addition of a helix to stem I
while at the same time maintaining stem I and II interactions, and
thereby preserving fast cleavage rates [19].
In bacteria sequestration of the ribosomal binding site has
been employed to switch gene expression (Fig. 1b) [14, 15, 20].
In case of ON-switches, activation of the HHR cleaves the mRNA.
This liberates the ribosomal binding site, which is now accessible
by the ribosome, so that translation can occur. Alternatively, mRNA
stability is regulated in eukaryotic cells. Ribozyme-mediated
detachment of the stabilizing 5′-cap structure leads to degradation
of the mRNA. In this case ON-switches are achieved by inhibiting
HHR cleavage upon ligand binding, so that translation can pro-
ceed [21, 22]. In addition to mRNAs other RNA classes can be
regulated: Attachment of a theophylline-dependent HHAz to the
5′ end of an E. coli tRNA has enabled us to switch the amino acid
identity in proteins post-transcriptionally [20, 23]. Incorporation
of a thiamine pyrophosphate-dependent HHAz into the E. coli
16S rRNA allowed control of the rRNA integrity; upon thiamine
addition to the medium cleavage of the 16S rRNA and subsequent
degradation occurs [24]. Yokobayashi and coworkers have
employed the design of our theophylline-HHAz to switch gene
expression in eukaryotes via regulation of RNAi. In their approach
a primary microRNA (pri-miRNA) hybridizes with an inhibitory
strand. Activation of the HHAz by theophylline frees the pri-
miRNA, which is then further processed by the intracellular RNAi
machinery to yield a functional miRNA [25].
Regarding non-small molecule inducers, our group has recently

developed HHR based genetic switches reacting to a physical stim-
ulus, namely a change in temperature, by attachment of a RNA
thermometer to the HHR scaffold [26]. Furthermore, we have
constructed a novel riboregulatory system in which allosteric con-
trol of the HHR is achieved by small trans-acting RNAs via forma-
tion of a RNA–RNA duplex structure [27].
Taken together, HHAzs have proven to be versatile tools
owing to their modular nature; input as well as output domains can
be varied and easily remodeled to fit one’s needs by standard
molecular biology methods. Incorporation of different aptamer
domains changes the ligand dependency of the aptazyme, modifi-
cations of the artificial expression platform enable the regulation of
a variety of RNA classes as outlined above. As the secondary struc-
ture of nucleic acids is primarily determined by Watson–Crick base
pairing, rational design can be applied to connect any given aptamer
with the HHR and thereby reprogram the allosteric regulation of
the encoded message or regulatory RNA [7]. However, Breaker
and coworkers have shown that the nucleotides in the sequence
connecting the aptamer and the ribozyme are crucial for relaying
the binding event to the catalytic domain, thereby enabling switch-
ing [28]. Optimization of the connecting sequence is necessary to
improve the performance of the HHAz for a given aptamer. In our
group we apply an in vivo screening protocol to detect suitable
switches from a pool of mutants carrying randomized connecting
sequences. Optimization of the connecting sequence can also be
performed in vitro [28]; however, our approach ensures that all
selected clones are already functional in vivo in bacteria.
To generate more complex switching pathways or biosensors it
is crucial that a greater number of sensor domains will be explored.
Herein, we describe the experimental details that allow reprogram-
ming the ligand-specificity of the HHAz in order to control mRNA
translation in E. coli. However, the general strategy can as well be
used to change ligand-dependency for the control of other RNA
classes. ON- as well as OFF-switching riboregulators can be identi-
fied using our protocol. First we will explain the construction of
mutant libraries, followed by our method for screening and initial
characterization of functional riboswitches.
2 Materials
All solutions are prepared using ultrapure water and analytical

grade reagents are used. Reagents are stored at room temperature
unless otherwise noted.
2.1 PCR 1. 3 M Na-acetate, pH 5.2: Dissolve 20.4 g CH3COONa-3H2O

in 50 mL H2O. Adjust to pH 5.2 with HCl.
2. 50, 70, and 99 % ethanol.
3. 0.5 M EDTA, pH 8.0: Dissolve 73.08 g EDTA and 30 g

NaOH pellets in 400 mL H2O, adjust pH 8.0 using NaOH,
adjust volume to 500 mL.
4. 5× TBE buffer: Dissolve 54 g Trizma base, 27.5 g boric acid,
20 mL 0.5 M EDTA (pH 8.0) in 1 L of H2O. Adjust to
pH 8.3, if needed, with concentrated HCl. Dilute to 0.5× in
H2O before use.
5. 6× agarose gel loading buffer: Dissolve 4 mL 50 mM Tris–
HCl, pH 7.6, 12 mL 100 % glycerol, 2.4 mL 0.5 M EDTA,
6 mg bromophenol blue, 6 mg xylene cyanol in 20 mL, pre-
pare aliquots of 1 mL. Can be stored at −20 °C for several
months.
6. Ready-to-use DNA size standard (e.g., GeneRuler 1 kb DNA
Ladder (Thermo Scientific)).
7. Agarose gels. For 0.8 % (w/v) dissolve 0.8 g agarose in 100 mL
0.5× TBE buffer by boiling in the microwave. Dissolved aga-
rose can be stored in an incubator at ≥70 °C for several days
(see Note 1).
8. DNA oligonucleotides as primers to be used in PCR
(see Subheading 3.1).
9. DpnI restriction enzyme and NEB buffer 4 (NEB).
10. Ethidium bromide gel staining solution. Dilute 200 μg in
400 mL H2O. Can be kept in the dark at room temperature
and reused (see Note 2).
11. Destaining solution: 400 mL. H2O. Can be kept at room tem-
perature and reused.
12. PCR cycler.
13. PCR template: e.g., eGFP expression vector (pET16b_eGFP_
wtHHR or pQE31-J06-wtHHR2, AG Hartig, University of
Konstanz).
14. Phusion Hot Start II High-Fidelity DNA polymerase, 5× HF
buffer and 100 % DMSO (NEB). Store at −20 °C.
15. 2 mM dNTP Mix. Store at −20 °C.
16. Quick Ligase, 2× Quick Ligation Reaction Buffer (NEB).
Store at −20 °C. Thaw Quick Ligation Reaction Buffer on ice.
17. DNA Gel Extraction Kit (e.g., Gel DNA Recovery Kit (Zymo
Research)).
18. DNA Purification Kit (e.g., DNA Clean and Concentrator-5
(Zymo Research)).
20. Agarose gel electrophoresis chamber and power supply.
21. Razor blade.
22. UV light table.
2.2 Cell Culture and 1. 96-deep-well plate and 96-well plate incubator (e.g., Heidolph
Screening Inkubator 1000 and Titramax 1000) (see Note 3).
2. Air-permeable adhesive seals for 96-well-plates.
3. 1,000× carbenicillin stock solution (100 mg/mL): Dissolve
1 g carbenicillin in 10 mL 50 % (v/v) ethanol. Can be stored
at −20 °C for several weeks.
4. Electro-competent E. coli (e.g., E. coli BL21 (DE3) gold,
Stratagene) (see Note 4). 80 μL aliquots frozen at −80 °C.
5. Electroporator (e.g., Eppendorf Electroporator 2510) and
electroporation cuvettes (e.g., Electroporation cuvettes,
0.1 cm (Bio-Rad)).
6. Fluorescence plate reader (e.g., TECAN M200).
7. LB-Carb-medium: For LB (Lennox) dissolve 10 g tryptone,
5 g yeast extract, 5 g NaCl in 1 L H2O, adjust to pH 7.0, if
necessary, and autoclave. Supplement with 1 mg/L carbenicil-
lin before use. Can be stored at 4 °C for up to 4 weeks.
8. LB-Carb-agar plates: Dissolve 10 g tryptone, 5 g yeast extract,
5 g NaCl, and 10 g agar-agar in 1 L H2O and autoclave. Let
cool until hand-warm, then supplement with 1 mg/L carbeni-
cillin. Pour plates in 15 cm petri dishes. Plates can be stored at
4 °C for up to 6 weeks.
9. Kit for plasmid isolation (e.g., Zyppy Plasmid Miniprep Kit).
10. SOC medium: Dissolve 20 g tryptone, 5 g yeast extract, 0.6 g
NaCl, 0.2 g KCl in 900 mL H2O. Adjust pH to 6.8–7.0 using
NaOH. Adjust volume to 960 mL and autoclave. Add 10 mL 1 M
MgCl2, 10 mL 1 M MgSO4, and 20 mL 1 M glucose from sterile
filtered stocks. Store 1 mL aliquots at −20 °C for several months.
11. Toothpicks sterilized by autoclaving.
12. Flat, transparent 96-well plates.
3 Methods
3.1 Construction of New aptamer sequences are incorporated into the HHR using
Randomized whole plasmid PCR as done for site-directed mutagenesis (Fig. 2).
Riboswitch Libraries Randomization of the connecting sequence is achieved by using
primers with unbiased random positions generated during solid
phase DNA synthesis using a 1:1:1:1 mixtures of nucleoside phos-
phoramidites. For successful ligation of the PCR product phos-
phorylation of one primer at the 5′ end is required. We recommend
purification of the long primers (≥45 nts) by denaturing PAGE, as
described by Sambrook and Russel in Molecular Cloning (see:
Preparation of Denaturing Polyacrylamide Gels and Purification of
Synthetic Oligonucleotides by Polyacrylamide Gel Electrophoresis).
Fig. 2 Library construction. Whole plasmid PCR is used for incorporation of new aptamer sequences using
primers carrying randomized sequences that will constitute the connecting sequence. After PCR the template
plasmid is degraded by DpnI digest. The purified PCR product is ligated yielding a library of randomized
riboswitches
1. Prepare the reaction mixture as follows and split the mixture

into 3 PCR tubes. The reaction can be scaled up, if necessary.
We recommend using no more than 30 μL per tube:
Starting concentration Material Volume Final concentration

5× HF buffer 16 1×
2 mM dNTP mix 8 200 μM
10 μM Forward primer 4.0 500 nM
10 μM Reverse primer 4.0 500 nM
10 ng/μl Template 1.6 16 ng
100 % (v/v) DMSO 2.4 3 % (v/v)
2 U/μl Phusion Hot Start DNA polymerase 0.8 1.6 U
H 2O 43.2
Total 80
2. Use the following conditions for the PCR with Phusion Hot
Start II High-Fidelity DNA polymerase (see Note 5).
Step Stage Temperature (°C) Time Go to step Repeat

1 Initial denaturation 98 30 s
2 Denaturation 98 10 s
3 Annealing 60 30 s
4 Extension 72 15–30 s/kb 2 24
5 Final extension 72 7 min
6 Cooling 4 Pause
3. Pool the PCR reaction mix into one 1.5 mL reaction tube. For
ethanol precipitation add 5 μL of 3 M Na-acetate, pH 5.2,
followed by 165 μL 99 % ethanol and mix by inverting.
4. Place samples at −80 °C for 20 min or −20 °C for a minimum
of 2 h or overnight.
5. Centrifuge samples in a tabletop centrifuge at 12,000 × g for
15 min at room temperature. Discard supernatant and wash
once with 70 % ethanol. Resuspend the pellet in 44 μL H2O.
6. For digestion of the template vector add 5 μL NEB buffer 4
and 1 μL DpnI and mix thoroughly. Incubate at 37 °C for
50 min followed by 10 min incubation at 80 °C for heat inac-
tivation of the enzyme.
7. In the meantime, pour a 0.8 % (w/v) TBE-agarose gel and let
it solidify. Once the sample is ready, add 10 μL of 6× agarose
gel loading buffer and mix. Load the sample in a long well and
load 2.5 μL DNA size standard in a small well to run along-
side. Run the gel at 10 V/cm for 75–90 min.
8. After the run, transfer the gel into ethidium bromide staining
solution for 10 min, then place the gel in destaining solution
for 10 min for improving contrast. As exposure to UV light
may introduce mutations to the DNA, cut the gel vertically
excising the marker and a small part of the band with your
sample. Visualize the DNA bands over a UV light table.
9. Excise the appropriate band in the correct size, piece the gel
back together and excise the corresponding band in the gel
that has not been exposed to UV light and purify the DNA
fragment using a DNA gel extraction kit according to manu-
facturer’s protocol.
10. Determine the DNA concentration by measuring A260.
11. For ligation of the PCR product combine 50 ng of the puri-
fied PCR product in 9 μL of water with 10 μL 2× Quick
Ligation Reaction buffer and add 1 μL Quick Ligase, mix well,
and briefly centrifuge samples. Incubate at 25 °C for 15 min.
12. Purify the DNA sample by using a DNA purification kit

according to the manufacturer’s instructions to remove the
ligation buffer for better transformation efficiency. Elute
ligated DNA using 6 μL H2O.
3.2 Screening for A pool of mutants carrying randomized connecting sequences is

Functional Aptazymes obtained by transformation of the randomized plasmid pool into
E. coli (Fig. 3). Single clone colonies can be picked and screened
for functional riboswitches in the presence and absence of the
respective ligand.
1. Thaw 80 μL aliquot of electro-competent E. coli on ice for
15 min. In the meantime, thaw 1 mL aliquot of SOC medium
and preheat to 37 °C, place LB-Carb-agar plates on 37 °C and
chill one electroporation cuvette on ice along with the puri-
fied, ligated DNA sample from the previous step.
2. Add 1.0 μL of the purified sample to 80 μL competent cells and
carefully mix on ice. Transfer the cells into the pre-chilled electro-
poration cuvette and transform by electroporation. Immediately
resuspend the cells in pre-warmed 1 mL SOC medium and trans-
fer to a fresh 1.5 mL reaction tube. Incubate for 1 h at 37 °C.
Plate the transformed cells on pre-warmed LB-agar plates and
incubate plates overnight at 37 °C (see Note 6).
3. Prepare 96 deep-well plates with 1 mL LB-medium supple-
mented with carbenicillin to be used as master plates. Save wells
in the first or last column of the 96 deep-well plate for control
cultures (see Note 7). Using sterilized toothpicks, pick single
colonies from the pool of mutants carrying randomized connect-
ing sequences. Seal with air-permeable adhesive cover. Incubate
plates at 37 °C overnight in the 96-well plate incubator.
4. For each master plate of the previous day (step 3), prepare
two additional destination plates to be used in the screening.
The medium added to every second column should be supple-
mented with the appropriate ligand as shown in Fig. 3.
5. Inoculate the destination plates with clones from the master
plates, so that each clone is tested with and without ligand (see
Note 8). Seal with air-permeable adhesive cover. Incubate the
destination plates in the 96-well plate incubator at 37 °C
overnight.
6. Transfer 100 μL of the out-grown culture into a black 96-well
plate and measure expression levels by determining eGFP fluo-
rescence (excitation wavelength λex = 488 nm, emission wave-
length λem = 535 nm) (see Note 9). Potential hits are identified by
comparing the fluorescence for each screened clone in the pres-
ence and absence of the ligand after subtraction of background
fluorescence. In addition measure OD600 for OD-correction.
7. Clones of interest are isolated by picking them from the mas-
ter plate and inoculating fresh LB cultures to be grown under
Fig. 3 Screening for functional riboswitches. The plasmid library of randomized riboswitches is transformed
into bacteria. Individual clones are picked to construct mutant libraries, which are screened for eGFP expres-
sion in the presence (white) or absence (dark grey) of the aptamer ligand. Controls for background fluores-
cence correction (B), positive control (+) and negative control (−) are included in the screening. Identified hits
are further characterized
vigorous shaking at 37 °C for 8 h for plasmid isolation and

preparation of glycerol stocks.
8. Plasmids are isolated using a plasmid isolation kit according to
manufacturer’s instruction. Integrity of the artificial riboswitch
construct and identity of the randomized connecting sequence

are checked by sequencing.
9. The riboswitches are further characterized by growing the
clones carrying a potential hit with increasing concentrations of
the ligand. Cultures are again prepared in 96-well plates as
detailed in step 4 using a gradient of increasing concentrations
of the ligand. Experiments are carried out in triplicate. Plates are
incubated at 37 °C overnight and eGFP expression measured to
confirm functional, concentration dependent riboswitches.
Pool coverage can be calculated by obtaining the oversampling
factor (Of) which is calculated as Of = T/V = − ln(1 − Pi) where T is the
number of clones actually screened, V the maximum number of dif-
ferent clones of a randomized sequence (sequence space) and Pi the
probability that a particular sequence occurs in the library [29]. Due
to the exponential relationship, high pool coverage is only achieved
when the theoretical sequence space is oversampled multiple times.
4 Notes
1. For a good separation and resolution of 5–10 kb PCR prod-

ucts, 0.8 % (w/v) agarose gels are sufficient. Higher percent-
ages of agarose (e.g.1.5 % (w/v)) can offer better resolution
for smaller products.
2. Ethidium bromide is toxic and mutagenic. Wear nitrile gloves
while working with it.
3. Alternatively, for larger screening purposes 384-deep-well plates
can be used for cell culture and fluorescence measurements.
4. For the preparation of electro-competent E. coli a detailed
protocol can be found at http://www.eppendorf.com.
5. Alternative to Phusion Hot Start polymerase, other polymer-
ases can be used. The protocol then needs to be adapted to the
appropriate buffer and conditions as required by manufactur-
er’s protocol. Use of a high-fidelity polymerase is advisable.
Primers should be designed to exhibit an annealing tempera-
ture of 60–63 °C; however, we suggest that a gradient PCR is
carried out to determine the optimal annealing temperature.
6. Dilutions of the transformed cells are plated in order to yield sin-
gle colonies. Dilutions will have to be determined experimentally
depending on the transformation efficiency of the electro-compe-
tent cells used.
7. As controls one can use cells transformed with a plasmid
expressing eGFP without any insert affecting the gene expres-
sion (positive control) and a negative control that bears an
inactivated hammerhead ribozyme variant by an A to G point
mutant in the catalytic core (Fig. 1b), masking the ribosomal

binding site permanently. Cells transformed with an empty
vector (e.g., pET16b or pGDR11) can be used to determine
background fluorescence. Note that all plasmids should trans-
fer the same antibiotic resistance. Since one master plate will
be distributed to two destination plates, we recommend using
the top half of the first column and the bottom half of last
column for controls, so that each destination plate will receive
a set of controls. We also recommend not using the wells in
the corners for screening purposes as growth of the cells will
suffer from evaporation of the medium.
8. Minimal medium (e.g., M9 or M63 medium) is recommended
for the screening process to receive full control over the ligand
concentration, especially when screening with naturally occur-
ring ligands. Membrane permeability also has to be considered
as not all ligands are readily taken up from the medium, while
others may not be stable in medium. In case of TPP switches,
we add thiamine to the medium, which is processed by the
intracellular thiamine kinase and thiamine phosphate kinase to
yield TPP [15, 23]. When using new ligands or media we rec-
ommend measuring growth curves for the bacteria, especially
to determine potential toxicity of a ligand.
9. It is recommended to wait until eGFP expression has reached
a plateau in outgrown cultures (at least 16 h). Use the fluores-
cence measured from cells transformed with an empty vector
to subtract background fluorescence.
References
1. Liu J, Cao Z, Lu Y (2009) Functional nucleic damental biochemical pathways in Bacillus sub-
acid sensors. Chem Rev 109:1948–1998 tilis and other bacteria. Cell 113:577–586
2. Tuerk C, Gold L (1990) Systematic evolution 7. Vinkenborg JL, Karnowski N, Famulok M
of ligands by exponential enrichment: RNA (2011) Aptamers for allosteric regulation. Nat
ligands to bacteriophage T4 DNA polymerase. Chem Biol 7:519–527
Science 249:505–510 8. Paige JS, Nguyen-Duc T, Song W, Jaffrey SR
3. Ellington AD, Szostak JW (1990) In vitro (2012) Fluorescence imaging of cellular
selection of RNA molecules that bind specific metabolites with RNA. Science 335:1194
ligands. Nature 346:818–822 9. Breaker RR (2002) Engineered allosteric ribo-
4. Winkler W, Nahvi A, Breaker RR (2002) zymes as biosensor components. Curr Opin
Thiamine derivatives bind messenger RNAs Biotechnol 13:31–39
directly to regulate bacterial gene expression. 10. Blount KF, Uhlenbeck OC (2002) The ham-
Nature 419:952–956 merhead ribozyme. Biochem Soc Trans
5. Winkler WC, Nahvi A, Sudarsan N, Barrick JE, 30:1119–1122
Breaker RR (2003) An mRNA structure that 11. Fedor MJ (2009) Comparative enzymology
controls gene expression by binding and structural biology of RNA self-cleavage.
S-adenosylmethionine. Nat Struct Biol Annu Rev Biophys 38:271–299
10:701–707 12. Khvorova A, Lescoute A, Westhof E, Jayasena
6. Mandal M, Boese B, Barrick JE, Winkler WC, SD (2003) Sequence elements outside the
Breaker RR (2003) Riboswitches control fun- hammerhead ribozyme catalytic core enable
intracellular activity. Nat Struct Biol switch for controlling mammalian gene expres-
10:708–712 sion. Mol BioSyst 6:807–814
13. Tang J, Breaker RR (1997) Rational design of 22. Ketzer P, Haas SF, Engelhardt S, Hartig JS,
allosteric ribozymes. Chem Biol 4:453–459 Nettelbeck DM (2012) Synthetic riboswitches
14. Wieland M, Hartig JS (2008) Improved apta- for external regulation of genes transferred by
zyme design and in vivo screening enable ribo- replication-deficient and oncolytic adenovi-
switching in bacteria. Angew Chem ruses. Nucleic Acids Res 40:e167
47:2604–2607 23. Berschneider B, Wieland M, Rubini M, Hartig
15. Wieland M, Benz A, Klauser B, Hartig JS JS (2009) Small-molecule-dependent regula-
(2009) Artificial ribozyme switches containing tion of transfer RNA in bacteria. Angew Chem
natural riboswitch aptamer domains. Angew 48:7564–7567
Chem 48:2715–2718 24. Wieland M, Berschneider B, Erlacher MD,
16. Ogawa A, Maeda M (2008) An artificial Hartig JS (2010) Aptazyme-mediated regula-
aptazyme-based riboswitch and its cascading tion of 16S ribosomal RNA. Chem Biol
system in E. coli. Chembiochem 9:206–209 17:236–242
17. Win MN, Smolke CD (2007) A modular and 25. Kumar D, Annna CI, Yokobayashi Y (2009)
extensible RNA-based gene-regulatory plat- Conditional RNA interference mediated by
form for engineering cellular function. Proc allosteric ribozyme. J Am Chem Soc
Natl Acad Sci U S A 104:14283–14288 131:13906–13907
18. Win MN, Smolke CD (2008) Higher-order 26. Saragliadis A, Krajewski SS, Rehm C,
cellular information processing with synthetic Narberhaus F, Hartig JS (2013) Thermozymes:
RNA devices. Science 322:456–460 Synthetic RNA thermometers based on ribo-
19. Wieland M, Gfell M, Hartig JS (2009) Expanded zyme activity. RNA Biol 10(6):1010–6
hammerhead ribozymes containing addressable 27. Klauser B, Hartig JS (2013) An engineered
three-way junctions. RNA 15:968–976 small RNA-mediated genetic switch based on a
20. Klauser B, Saragliadis A, Auslander S, Wieland ribozyme expression platform. Nucleic Acids
M, Berthold MR, Hartig JS (2012) Post- Res 41(10):5542–52
transcriptional Boolean computation by com- 28. Soukup GA, Breaker RR (1999) Engineering
bining aptazymes controlling mRNA precision RNA molecular switches. Proc Natl
translation initiation and tRNA activation. Mol Acad Sci U S A 96:3584–3589
BioSyst 8:2242–2248 29. Reetz MT, Kahakeaw D, Lohmer R (2008)
21. Auslander S, Ketzer P, Hartig JS (2010) A Addressing the numbers problem in directed
ligand-dependent hammerhead ribozyme evolution. Chembiochem 9:1797–1804
Chapter 18
Engineered Riboswitch as a Gene-Regulatory Platform

for Reducing Antibiotic Resistance
Libing Liu and Shu Wang
Abstract
Antibiotic resistance (AR), the ability of a microorganism to withstand the effects of antibiotics, is a growing
and increasingly serious global public health problem. Enzymatic activation of antibiotics though the
production of β-lactamase is one of the main mechanisms causing AR. Synthetic riboswitch containing
aptazyme is constructed in E. coli to regulate the expression of β-lactamase through small molecule–
aptamer interactions, which sharply reduces the antibiotic resistance of the engineered bacteria.
Key words Antibiotic resistance, Riboswitch, Theophylline-specific aptazyme, Plasmid construction,

Gene regulatory systems, β-Lactamase activity
1 Introduction
Riboswitches, which are based on small molecule–RNA interac-

tion, have attracted much attention due to their wide applications
in controllable gene expression [1–4]. Here, we present a method
for construction of a synthetic riboswitch containing theophylline-
specific aptazyme, which can regulate the genetically modified
expression of β-lactamase in response to theophylline. The theoph-
ylline can induce the conformation change of the riboswitch to
activate the ribozyme. The cleavage of active site results in the
decrease of β-lactamase gene expression via mRNA degradation
pathway. Through reducing the amount of β-lactamase, the antibi-
otic resistance can be ultimately regulated. The small molecule the-
ophylline can be used as a co-agent to preserve the efficacy of
traditional antibiotics, which provides a potential strategy to reduce
antibiotic resistance instead of discovering new antibiotics.
Resistance to β-lactam antibiotic agents in gram-negative E. coli
is primarily mediated by β-lactamase that can hydrolyze the β-lactam
ring to inactivate the drugs [5]. Our goal is to construct a ribo-
switch in E. coli for regulating the expression of β-lactamase so as
to reduce antibiotic resistance. The riboswitch of hammerhead
251
252 Libing Liu and Shu Wang
ribozyme is constructed inside the expression vector pBV220 [6].

The theophylline-specific aptazyme sequence is inserted at a down-
stream location of ampicillin-resistant (Ampr) gene in the plasmid
pBV220. The obtained new plasmid is designated as pBV220-Apt.
Precious evidence has shown that integration of the hammerhead
ribozyme into the 3′UTR of a reporter gene can suppress protein
translation in a theophylline-dependent fashion in Saccharomyces
cerevisiae [7]. In the absence of theophylline, the hammerhead
ribozyme is inactive and the β-lactamase is highly expressed. Upon
adding ampicillin the growth of E. coli is unaffected even with a
large amount. In the presence of theophylline, the ribozyme will be
activated via conformation change of the theophylline-specific
aptamer and the cleavage of active site results in the decrease of
β-lactamase gene expression by its mRNA degradation. In this case,
the β-lactamase expression is greatly inhibited and the antibiotic
resistance is sharply impaired. Under this circumstance E. coli cells
are efficiently inactivated by the addition of ampicillin. Therefore,
the growth of bacteria cells can be effectively controlled by theoph-
ylline in the presence of ampicillin, where the theophylline acts as a
co-agent to preserve the efficacy of traditional antibiotics.
2 Materials
Prepare all solutions using ultrapure water (by purifying deionized

water with Milli-Q water at 25 °C) and analytical grade reagents.
Prepare and store all reagents at room temperature (unless indi-
cated otherwise).
2.1 Construction of 1. 0.5 mg/mL pBV220 solution in water (see Note 1).
Plasmid pBV220-Apt 2. AvrII (NEB) (see Note 2).
3. XhoI (NEB) (see Note 2).
4. DNA purification kit (Tiangen).
5. 10 μM oligonucleotides of aptazyme template and primers in
ultrapure water (see Note 3). Aptazyme template: GCTGTCA
CCGGATGTGCTTTCCGGTCTGATGAGTCCGTGTTC
CT GATACCAGCATCGTCTTGATGCCCTTGGCAGCAG
TGGACGAGGACGAAACAGC.
Primer 1: ATCCTAGGGCTGTCACCGGATGTGC.
Primer 2: ACCTCGAGGCTGTTTCG TCCTCG. Store at
−20 °C.
6. dNTPs (TaKaRa) (see Note 4).
7. Pfu DNA polymerase (Tiangen) (see Note 5).
8. T4 DNA ligase (TaKaRa) (see Note 5).
9. DH5α competent cells (Tiangen) (see Note 6).
Aptazyme for Reducing Antibiotic Resistance 253
10. TOP10 competent cells (Tiangen) (see Note 6).

11. LB (Luria–Bertani) liquid medium: 10 g/L Tryptone, 5 g/L
Yeast extract, 5 g/L NaCl, pH 7.4 (see Note 7). Store at 4 °C.
12. LB solid/agar medium: 15 g/L agar in LB liquid medium (see
Note 7). Store at 4 °C.
13. 50 mg/mL ampicillin solution (see Note 8). Store at −20 °C.
2.2 Growth Curves of 1. 100 mM theophylline solution in ultrapure water (see Note 9).
E. coli TOP10 Cells Store at 4 °C.
2.3 Assay for 1. 100 mM nitrocefin solution in ultrapure water (see Note 9).
β-Lactamase Activity Store at 4 °C.
2.4 Equipment 1. Thermal cycler.

2. Shaking incubator.
3. Plate-reading spectrophotometer.
4. Centrifuge.
3 Methods
3.1 Construction of 1. Digest 1 μg of pBV220 by 1 unit of AvrII and XhoI at 37 °C

Plasmid pBV220-Apt for 1 h in a total reaction volume of 50 μl. And then purify the
Containing Aptazyme product by using the DNA purification kit (see Note 10). Store
Sequence it at −20 °C until use.
2. Generate the aptazyme sequence by PCR amplification using
the following conditions: 100 μM dNTPs, 0.4 μM primer 1,
0.4 μM primer 2, 5 μl of 10× buffer, 1 U of pfu DNA poly-
merase, 1 μl of 0.1 μM aptazyme template in a total reaction
volume of 50 μl; 35 rounds of 95 °C for 30 s, 50 °C for 60 s,
and 65 °C for 60 s (see Note 11). And then purify the PCR
product by using the DNA purification kit (see Note 10).
3. Digest 5 μl of the PCR product by 1 unit of AvrII and XhoI at
37 °C for 1 h in a total reaction volume of 50 μl (see Note 12).
And then purify the product by using the DNA purification kit
(see Note 10).
4. Ligate 3 μl each of digestion product of pBV220 and aptazyme
sequence with 1 μl of T4 DNA ligase in 1× ligation buffer in a
total volume of 20 μl at 4 °C for 16 h (Fig. 1) (see Note 13).
5. Mix 3 μl of ligation product and 100 μl of DH5α or TOP10
gently on ice for 5 min. And then put the mixture in 42 °C hot
water for 90 s (see Note 14). After that, add 500 μl of LB liquid
medium into the mixture and shake at 37 °C for 1 h. Spread
100 μl of the mixture on the LB solid plate with 50 μg/mL
ampicillin and keep them in the incubator at 37 °C for 12 h.
Fig. 1 Construction map of pBV220-Apt containing the aptazyme sequence and its electrophoretic analysis
(Line 2) and restriction enzymes (AvrII and XhoI) cutting (Line 1) (reproduced from ref. 6 with permission from
the Royal Society of Chemistry)
3.2 Growth Curves 1. Transfer a colony of E. coli TOP10 cells harboring the plasmid
of E. coli TOP10 Cell pBV220-Apt from the LB/agar plate containing 50 μg/mL
ampicillin into culture tubes containing 5 mL of LB media
supplemented with 50 μg/mL ampicillin. Then, incubate them
at 37 °C overnight in the shaking incubator (see Note 15).
2. Add 15 μL of the overnight bacteria liquid into culture tubes
containing 5 mL of fresh LB media supplemented with 50 μg/
mL ampicillin and appropriate concentrations of theophylline,
and shake the culture tubes at 37 °C with a speed of 200 rpm
in the shaking incubator (see Note 16).
3. Record the optical density at 600 nm for each concentration of
theophylline by using a plate-reading spectrophotometer at
1 h intervals over 12 h (Fig. 2) (see Note 17).
3.3 Assay for 1. Transfer two separate colonies of E. coli TOP10 cells harboring
β-Lactamase Activity pBV220 and pBV220-Apt plasmids into separate culture tubes
containing 5 mL of LB media. Then, incubate them at 37 °C
overnight in the shaking incubator (see Note 18).
2. Add 15 μL of the overnight culture into culture tubes contain-
ing 5 mL of fresh LB media supplemented with 50 μg/mL
ampicillin and appropriate concentrations (e.g., 0, 0.2, 0.4,
0.6, 0.8, 1.0 mM) of theophylline (see Note 19).
3. After the optical density at 600 nm reaches to the range of
0.5–0.6, collect 1 mL of bacterial cells by centrifugation with
3,000 × g for 4 min at 4 °C.
4. Resuspend the precipitation with 100 μL of pre-cooling puri-
fied water (see Note 20).
5. Freeze and thaw the suspension for 10 times by using liquid
nitrogen and 37 °C water.
6. Centrifuge it with 10,000 × g for 10 min.
Fig. 2 Growth curves of E. coli in the presence of varying concentrations of the-

ophylline. The error bars represent the standard deviation of three measurements
(reproduced from ref. 6 with permission from the Royal Society of Chemistry)
7. Collect the supernatant in a new tube and store on ice.

8. Use nitrocefin as the chromogenic cephalosporin substrate to
evaluate the β-lactamase activity [8]. Add 5 μL of the superna-
tant, 2 μL of nitrocefin, and 93 μL of purified water to each
well in a 96-well plate. Measure the change of the optical den-
sity at 492 nm in 10 min with a plate photometer at room
temperature (Fig. 3). All measurements are conducted in trip-
licate (see Note 21).
4 Notes
1. pBV220 is a kind of plasmid DNA and easily digested by

DNase. Therefore, it is strongly recommended that water is
sterilized at 121 °C and 102.9 kPa for 20 min before use.
pBV220 solution in water can be stored at −20 °C for a few
months. Plasmid pBV220 can be stored in ethanol for a longer
time. Avoid repeatedly freezing and thawing. Other expression
vector can also be used instead of pBV220.
2. Store at −20 °C before and after use. During experiments, it
should be placed on ice. We find that it is better to centrifuge
with 10,000 × g for 10 s before opening the lid of the tube.
3. Oligonucleotides of aptazyme template and primers are a
kind of DNA and easily digested by DNase. Thus, powder
and solution should be stored at −20 °C before and after use.
Fig. 3 Riboswitch mediated β-lactamase relative activity in E. coli TOP10 cells as

a function of varying concentrations of theophylline from 0 to 1.0 mM. The error
bars represent the standard deviation of three measurements (reproduced from
ref. 6 with permission from the Royal Society of Chemistry)
It is necessary to centrifuge with 10,000 × g for 10 s before

opening the lid of the tube. During experiments, all DNA
samples should be placed on ice.
4. We find it can be stored at 4 °C for 1 week or −20 °C for long
time. Avoid repeatedly freezing and thawing.
5. It should be stored at −20 °C. Avoid repeatedly freezing and
thawing.
6. It should be stored at −20 °C within 3 months or −80 °C for a
longer time. During experiments, it should be placed on ice.
DH5α and TOP10 competent cells should be reserved as indi-
vidual package, for the competent state will be lost after
thawing.
7. We find that solution pH is in the range of 7.0–7.4 after tryp-
tone, yeast extract, and NaCl are dissolved. It does not matter
about the pH deviation. Otherwise, pH can be adjusted by
using 1 N HCl. All media should be sterilized at 121 °C and
102.9 kPa for 20 min.
8. Ampicillin is easy to dissolve and easily loses the effectiveness in
water. It should be divided into small aliquots (e.g., 1 mL) and
stored at −20 °C after sterilizing with a 0.22 μm Millipore filter.
9. It should be divided into small aliquots (e.g., 1 mL) to avoid
contamination after sterilization by using a 0.22 μm Millipore
filter.
10. Follow the kit protocol to complete the purification steps.
11. Several reasons can account for PCR failure. First, PCR is very
sensitive to DNA contamination, which gives rise to nonspe-
cific DNA amplification. Exogenous contamination DNA can
be addressed by optimizing experimental protocols and proce-
dures. It is necessary to set up a separate PCR area for the
analysis or purification of PCR products, to use disposable
plasticware, and to thoroughly clean the work surface between
reaction facilities. Second, primer-design techniques are highly
important for improving PCR product yield and avoiding the
formation of false positive products. The use of alternative buf-
fer components or polymerase enzymes is beneficial.
12. Constant temperature is very critical for maintaining enzyme
activity. Volume of enzyme in the reaction cannot exceed 1/10
of the total. Digestion time can be extended approximately to
2 h if the ligation product cannot be obtained in the step 5 of
this section.
13. T4 DNA ligase easily loses the enzyme activity, so the ligation
reaction solution is usually kept at 4 °C. Ligation reaction
should be carried out at 4 °C for 16 h to obtain a ligation
product. We find that it is optimal to supplement additional
1 μl of T4 DNA ligase after 8 h.
14. It is necessary to incubate the competent E. coli cells for heat-
shocking at 42 °C for exact 90 s. This procedure can tran-
siently result in the formation of pores on the cell membrane,
which facilitates the entry process of exogenous DNA
(plasmid).
15. E. coli colonies on the LB plate contain both pBV220 and
pBV220-Apt. Therefore, we pick several single colonies and
send them to BGI Company for sequence to determine which
colony carries the contracted plasmid pBV220-Apt.
16. It is necessary to keep a proper shaking speed for E. coli growth.
We find the range of 150–220 rounds per minute suitable.
17. The optical density at 600 nm represents solution turbidity,
which reflects the bacterial growth condition.
18. Plasmid in E. coli TOP10 generally loses after cultivation of
generations because of unknown reasons. During experiments,
we find β-lactamase expresses higher in E. coli TOP10 than in
E. coli DH5α. Therefore, we use DH5α to store and amplify
the plasmid pBV220-Apt and use TOP10 to measure data.
19. The overnight culture is regarded as the bacterial seed for the
subsequent amplification. Mix it well before seeding.
20. Enzyme is a kind of protein in living cells and is easily digested.
Thus, the common operation is to keep the solution on ice or
low temperature after this step.
21. One unit of β-lactamase is defined as the amount which pro-
duces 1.0 mmol of product per minute.
Acknowledgements
This work was supported by the National Natural Science

Foundation of China (Nos. 20725308, 90913014, 20721061 and
TRR61) and the Major Research Plan of China (No.
2006CB932102).
References
1. Winkler WC, Nahvi A, Roth A, Collins JA, resistant Escherichia coli isolates from foods,
Breaker RR (2004) Control of gene expression humans, and healthy animals. Antimicrob
by a natural metabolite-responsive ribozyme. Agents Chemother 46:3156–3163
Nature 428:281–286 6. Feng X, Liu L, Duan X, Wang S (2011) An
2. Sharma V, Nomura Y, Yokobayashi Y (2008) engineered riboswitch as a potential gene-
Engineering complex riboswitch regulation by regulatory platform for reducing antibacterial
dual genetic selection. J Am Chem Soc 130: drug resistance. Chem Commun 47:
16310–16315 173–175
3. Bayer TS, Smolke CD (2005) Programmable 7. Win MN, Smolke CD (2007) A modular and
ligand-controlled riboregulators of eukaryotic extensible RNA-based gene-regulatory plat-
gene expression. Nat Biotechnol 23:337–343 form for engineering cellular function. Proc
4. Desai SK, Gallivan JP (2004) Genetic screens Natl Acad Sci U S A 104:14283–14288
and selections for small molecules based on a 8. Hedberg M, Lindqvist L, Bergman T, Nord CE
synthetic riboswitch that activates protein trans- (1995) Purification and characterization of a
lation. J Am Chem Soc 126:13247–13254 new β-lactamase from Bacteroides uniformis.
5. Brinas L, Zarazaga M, Saenz Y, Ruiz-Larrea F, Antimicrob Agents Chemother 39:
Torres C (2002) β-Lactamases in ampicillin- 1458–1461
Chapter 19
Construction of Ligand-Responsive MicroRNAs that

Operate Through Inhibition of Drosha Processing
Chase L. Beisel, Ryan J. Bloom, and Christina D. Smolke
Abstract
MicroRNAs (miRNAs) offer powerful tools for targeted gene silencing in almost all eukaryotes.
These tools have received considerable attention for their utility in both fundamental genetic studies and
as therapeutic agents. Rendering individual microRNAs responsive to endogenous or exogenously applied
molecules (or ligands) can improve the stringency of silencing and can mediate autonomous control. This
chapter describes the construction of ligand-responsive miRNAs that undergo reduced processing and
subsequent gene silencing when bound by the recognized ligand. Following a simple set of rules, the
engineered microRNAs can be readily modified to target different sequences and to bind different ligands.
Individual miRNAs also can be incorporated into the same transcript for tunable, multi-gene silencing.
Key words Aptamers, Drosha, Gene regulation, Mammalian cells, MicroRNA, Riboswitch, RNA
interference
1 Introduction
MiRNAs are small regulatory RNAs that carry out gene silencing
in most eukaryotes [1]. In mammalian cells, transcribed miRNAs
undergo sequential processing by the endoribonucleases Drosha
and Dicer to form a ~22-nucleotide “mature” miRNA [2]. The
mature miRNA serves as a guide to recruit the RNA-induced
silencing complex (RISC) to complementary mRNAs, which are
silenced by cleavage, inhibition of translation, or sequestration in
inaccessible intracellular bodies [3]. Many of the known targets of
natural miRNAs are involved in development, angiogenesis, and
cancer formation, underscoring the widespread importance of
these RNA regulators [4].
MiRNAs have also received considerable attention as tools for
targeted gene silencing [5, 6]. The mature miRNA sequence can be
altered to specifically silence different genes within the host organ-
ism or to access regulatory networks of endogenous miRNAs.
259
260 Chase L. Beisel et al.
One challenge impeding the implementation of miRNA-based

tools is ensuring temporal and spatial control of miRNA activity.
The use of inducible promoter systems offers one solution, although
these systems require the expression of heterologous proteins that
may trigger an immune response, that require large expression con-
structs, and that cannot be easily modified to accommodate differ-
ent inducing molecules.
A separate solution is integrating miRNAs with aptamers,
RNAs that specifically and tightly bind a selected molecule.
Procedures for the in vitro and in vivo selection of aptamers are
well established, resulting in aptamers against a large assortment of
ligands, including small molecules, metals, and proteins [7–9].
Furthermore, insights into the design of riboswitches have sug-
gested how to rationally combine aptamers and regulatory RNAs
for the conditional control of gene expression.
Here, we describe the rational integration of aptamers and
miRNAs for conditional gene silencing in eukaryotes. Ligand
binding to the aptamer inhibits Drosha processing, reducing the
production of mature miRNAs and subsequent gene silencing.
This design allows the aptamer and mature miRNA sequence to be
readily replaced, facilitating the broad use of conditional miRNAs
in applications ranging from genetic studies to the development of
“smart” therapeutics.
2 Materials
1. pCS1815 expression vector.

2. Standard molecular biology equipment and reagents.
3. Restriction enzymes AgeI, AsiSI, ClaI, and PacI.
4. E. coli JM110 (see Note 1).
5. Standard equipment for mammalian cell culture.
6. Fugene HD transfection reagent (Promega).
7. Dulbecco’s Modified Eagle Medium supplemented with 10%
FBS (Life Technologies).
8. Opti-MEM (Life Technologies).
9. 24-well cell culture plates (BD Biosciences).
10. Trypsin (Life Technologies).
11. HEK293 cell line.
12. HEK293 GFP cell line.
13. Flow cytometer with 488 nm laser with 525 nm and 610 nm
long pass filters and a 550 nm dichroic mirror (Beckman
Coulter).
Ligand-Responsive miRNAs 261
3 Methods
We first describe the design of ligand-responsive miRNAs that target

a desired gene and bind a specific ligand. As a model, we target the
gene encoding green fluorescent protein (GFP) and integrate an
aptamer that recognizes the small molecule theophylline [10]. Next,
we cover the cloning of designed microRNAs into an expression
vector from our laboratory, although other expression vectors can be
readily used (see Note 2). Finally, we introduce assays to evaluate the
activity of the designed miRNAs in mammalian cell culture.
The protocol below describes the expression of a miRNA that
silences the expression of chromosomally encoded GFP in trans.
However, there are two other possible configurations to regulate gene
expression with the ligand-responsive miRNA: cis regulation and a
combination of cis and trans regulation (see Note 3). Under cis regu-
lation, a ligand-responsive miRNA is placed in the 3′ untranslated
region of a non-targeted gene. Ligand-dependent processing of the
miRNA deactivates the transcript, silencing the upstream gene. Under
combined cis and trans regulation, a ligand-targeting miRNA is placed
in the 3′ untranslated region of a targeted gene. This configuration
imparts regulation through miRNA processing and miRNA-mediated
silencing, where both steps are inhibited by the ligand.
3.1 Design of 1. The ligand-responsive miRNA is comprised of two parts, a

Ligand-Responsive mature miRNA and an aptamer, grafted onto the base miRNA
MicroRNAs miR-30a (Fig. 1a). The mature miRNA is 22 nucleotides and
is complementary to the target mRNA (see Note 4). The
paired nucleotides must be complementary with the exception
of the first nucleotide on the 5′ end to preserve the secondary
structure of the miRNA. The example sequence in Fig. 1a
shows a potent sequence for silencing the expression of GFP
(top) and the expression of La (bottom).
2. The aptamer is placed at the base of the lower stem. Normally,
the unstructured base of the stem serves as a ruler for Drosha
cleavage [11]. When the base of the stem is structured, such as
through base pairing, Drosha does not effectively bind and
cleave the miRNA [11–13]. To exploit this feature of process-
ing by Drosha, the aptamer must be inserted such that it is
unstructured in the absence of ligand and tightly bound in the
presence of ligand. Fortunately, most aptamers contain stems
at the top and bottom of the binding region; these regions can
be removed in order to insert the aptamer into the miRNA.
The only limitation is that the aptamer must exhibit a minimal
bulge size in the absence of ligand (see Note 5). Below this
limit, Drosha processing is substantially inhibited [14]. If the
bulge of the selected aptamer is too small, the loop of the
aptamer can be included, as illustrated for the hypoxanthine
aptamer in Fig. 1b.
a b
1 2
5’ G U A AU
A
A U
A A A UAU
A G U GU
GUCCUG CAGUG GCG GCACAAGCUGGAGUACAACUAUA-GUG A GGCACGU ACC U
CGGGAC GUCAU CGC CGUGUUCGACCUCAUGUUGAUAU CACCG CCGUGCA UGG U U
G C C C G A GC
G C U G
3’ AU A C A A
2
1
NO YES
UAU GA CG
GGU AC
U G U
A A A G C GU AC CCA UG
AGCUG CC GCAUC UC UGGAAAUCAGAGAAGAUAAAA CA UG
AC U G
UAU
UCGAC GG CGUAG U U ACCUUUAGUCUCUUCUAUUUU GC
G C U C U
A UC
aptamer mature miRNA
Fig. 1 Design of ligand-responsive miRNAs. (a) MiRNAs undergo sequential processing by Drosha (circle 1) and
Dicer (circle 2), where one of the two final strands, the mature miRNA, is retained by RISC for the silencing of
complementary mRNAs. The design builds on miRNA-30a through the integration of a mature miRNA sequence
(white letters, dark gray box) and the binding domain of an aptamer (bolded letters, light gray box). The binding
domain of the aptamer is inserted below the lower stem of the miRNA. In the absence of ligand, the aptamer
will be unstructured, allowing Drosha processing. In the presence of ligand, the bound aptamer will be struc-
tured, inhibiting Drosha processing. The displayed design depicts the insertion of a mature miRNA targeting
the La gene in place of a mature miRNA targeting the GFP gene, and the insertion of the binding domain of the
theophylline aptamer in place of a large bulge. (b) The aptamer core must be of sufficient size to allow Drosha
processing in the absence of ligand. In cases where a small aptamer, such as the hypoxanthine aptamer as
shown, is selected, the aptamer binding domain and part of the loop can be integrated. Note that the binding
domain is placed adjacent to the lower stem to ensure that ligand binding inhibits Drosha processing
3.2 Construction of 1. Once the ligand-responsive miRNA is designed, the miRNA

Ligand-Responsive must be cloned into an expression vector. We use a construct
MicroRNAs (pCS1815, Fig. 2a) that can express the miRNA from the pol
II CMV promoter or the pol III U6 promoter. The construct
additionally expresses DsRed-Express as a transfection marker.
The plasmid can be isolated from E. coli cells using standard
plasmid preparation techniques. The miRNA is inserted into
the restriction sites AgeI and ClaI downstream of the CMV
promoter. The intervening CD19 gene serves as a surface
marker for transfection efficiency and for miRNA processing.
2. The miRNA is assembled into linear, double-stranded DNA
that can be cleaved with AgeI/ClaI restriction enzymes and
ligated into the plasmid (Fig. 2b). The linear DNA includes
four restriction sites adjacent to the stem below the aptamer:
AgeI/AsiSI on the 5′ end, and PacI/ClaI on the 3′ end. The
internal restriction sites allow the sequential insertion of mul-
tiple miRNAs as described below. The linear DNA encoding
the miRNA also contains four base pairs flanking the ends of
the restriction sites for efficient cleavage by AgeI and by ClaI.
a
bla U6 promoter
CMV promoter
pUC ori
pCS1815 CD19
SV40 polyA 8712 bp
AgeI
neoR ClaI
BGH polyA
SV40 early promoter
CMV promoter
BGH polyA
DsRed-Express
b
AgeI AsiSI
AATAACCGGTGCGATCGCGAACGGGTCCTGATACCAGCGTGAGCGAGCACAAGCTGGAGTACAACTATAGTGAAGCCA ...
TTATTGGCCACGCTAGCGCTTGCCCAGGACTATGGTCGCACTCGCTCGTGTTCGACCTCATGTTGATATCACTTCGGT
... CAGATGTATAGTTGTACTCCAGCTTGTGCCCGCCTACGCCCTTGGCAGCAGGGCCCGTTTTTAATTAAATCGATTATT
GTCTACATATCAACATGAGGTCGAACACGGGCGGATGCGGGAACCGTCGTCCCGGGCAAAAATTAATTTAGCTAATAA
PacI ClaI
c
AgeI AsiSI
AATAACCGGTGCGATCGCGA ... CCCGTTTaaaaaaaaaaaaaaaaaaccaaaaaaaaaaTTAATTATATT
TTATTGGCCACGCTAGCGCT GGGCAAAttttttttttttttttttggttttttttttAATTAATATAA
PacI
AsiSI
AATACGATCGCaaaaaaaaaaaaaaaaaaccaaaaaaaaaaGAACGGG ... TTTTAATTAAATCGATTATT
TTATGCTAGCGttttttttttttttttttggttttttttttCTTGCCC AAAATTAATTTAGCTAATAA
PacI ClaI
Fig. 2 Cloning designed miRNAs into the pCS1815 expression plasmid. (a) Plasmid map of pCS1815. Designed
miRNAs are cloned into the AgeI/ClaI restriction sites downstream of the gene encoding the CD19 surface
marker. (b) Sequence of linear double-stranded DNA encoding the designed miRNA. The DNA includes four
restriction sites for the insertion of the double-stranded DNA into the expression plasmid. The boxed nucleotides
indicate the oligonucleotide sequences needed to PCR amplify the full-length DNA. (c) Sequences for the
sequential insertion of multiple miRNAs. The A-rich spacer (lowercase letters) helps ensure processing of each
miRNA. The top sequence can be cleaved with AgeI/PacI and inserted into AgeI/AsiSI upstream of a miRNA in
pCS1815. The bottom sequence can be cleaved with AsiSI/ClaI and inserted downstream of a miRNA in pCS1815
3. The linear DNA is synthesized with two overlapping

oligonucleotides, which are boxed in Fig. 2b (see Note 6). The
two oligonucleotides can serve as primers and DNA templates
in a standard PCR, where six cycles are sufficient for the syn-
thesis of the linear DNA. The PCR-amplified DNA is then
column-purified to remove any remaining oligonucleotides
and unincorporated dNTPs.
4. The plasmid DNA and the linear DNA are digested with AgeI
and ClaI and column-purified to remove the small cleaved
fragments. The plasmid DNA can be further treated with a
phosphatase to reduce the frequency of ligation of partially
cleaved DNA. The phosphatase would need to be heat-inacti-

vated or removed by column purification prior to ligation.
5. The cleaved and purified plasmid DNA (200 ng) and linear
DNA (20 ng) can be ligated together using T4 DNA ligase
(100,000 U), incubated at 16 °C in a PCR machine for at least
10 min, and transformed into E. coli JM110 cells. We have
found that 50 μg/mL ampicillin is sufficient to select for the
plasmid in LB plates and in LB media. We recommend screen-
ing colonies by PCR using the primers GCTGTGACTTT
GGCTTATCTG and TCCAGGGTCAAGGAAGGCAC,
which should shift in size from ~ 200 bp to ~350 bp. One of
the primers additionally can be used to sequence the inserted
miRNA.
6. The restriction sites allow the sequential insertion of additional
miRNAs for tunable or combinatorial control of gene expres-
sion. The internal AsiSI and PacI restriction sites are isos-
chizomers whose sticky ends can be combined to form a
non-cleavable site. Therefore, miRNAs flanked by AgeI/AsisI
(5′ end) and PacI (3′ end) can be sequentially inserted into the
AgeI/AsiSI sites upstream of miRNAs in the plasmid.
Correspondingly, miRNAs flanked by AsiSI (5′end) and PacI/
ClaI (3′ end) can be sequentially inserted downstream of miR-
NAs in the plasmid. We recommend the sequences shown in
Fig. 2c. Note that these sequences include an A-rich spacer
that, in our experience, helps ensure processing of each
miRNA. The linear double-stranded DNA sequences for each
inserted miRNA can be generated as described above.
3.3 Measurement of 1. The cloned plasmid first is isolated from the JM110 cells. The
Ligand-Responsive purity of the DNA sample is a major factor in transfection effi-
MiRNA Activity in Cell ciency, with highly pure plasmid DNA lacking endotoxins
Culture resulting in the most reproducible, high efficiency transfec-
tions (see Note 7). The original pCS1815 plasmid or the same
plasmid containing a miRNA with a scrambled mature miRNA
sequence should be isolated; this plasmid will serve as a con-
trol when assessing the silencing of GFP expression.
2. HEK 293 cells stably expressing GFP are seeded in 24 well
plates at a density of 100,000 cells/well in Dulbecco’s
Modified Eagle Medium (DMEM) on Day 1. An exogenous
ligand can be added to the medium at this time or during
the transfection. HEK 293 cells lacking GFP can be addi-
tionally seeded to serve as a “DsRed-only” control for
color-compensation.
3. On Day 2, Fugene HD and Opti-MEM are allowed to equili-
brate to room temperature. Then purified plasmid (500 μg) is
mixed with 25 μL of Opti-MEM and 1.5 μL of Fugene HD.
This mixture is incubated at room temperature for 30 min
then added to a single well of cells and mixed with the medium
by gentle shaking. If multiple wells of cells are to be transfected

with the same plasmid (for experimental replicates or to test
different biological conditions) the mixture of transfection
reagents can be scaled accordingly. We recommend at least
two biological replicates for each condition tested as well as
single-color controls for both green and red and a no-color
sample. The transfected cells are then incubated for 48 h at
37 °C and 5 % CO2.
4. On Day 4, the cells are harvested by aspirating away the media
and adding 100 μL of trypsin to each well. Once the cells have
become detached (~15 min), the trypsin is neutralized by
adding 200 μL of DMEM and the medium is gently pipetted
up and down to break up clumps. The cells are then added to
a 96-shallow-well, round bottom plate for flow cytometry (see
Note 8).
5. A “GFP-only” control (untransfected HEK293 GFP cells) and
a “DsRed-only” control (HEK293 cells transfected with the
plasmid) are first run through the flow cytometer to set any
required compensation for the green and red signals. The
“GFP-only” control additionally is used to set a gate separating
DsRed-positive (transfected) and DsRed-negative (untrans-
fected) cells; all of these cells should be contained within the
DsRed-negative gate. When running the experimental samples,
the analysis will be restricted to DsRed-positive cells that have
received the plasmid containing the miRNAs.
6. Once all of the samples have been run on the flow cytometer, the
data are analyzed using appropriate analysis software, such as
FlowJo. The knockdown efficiency of a microRNA is determined
by dividing the median fluorescence of GFP in DsRed-positive
cells transfected with the tested plasmid by the GFP fluorescence
in DsRed-positive cells transfected with a control plasmid harbor-
ing a microRNA with a scrambled targeting sequence.
æ Median GFP fluorescence ( DsRed - positive cells with tested plasmid )ö

% Knockdown = 100% ç 1 - ÷
ç Median GFP fluorescence ( DsRed - negative cells with control plasmid ) ÷ø
è
The responsiveness of a miRNA to the ligand is measured

by the fold change. The fold change is calculated by dividing
the percent knockdown of a miRNA in cells cultured in
medium without or with the ligand:
1 - % Knockdown ( +ligand )
Fold change =
1 - % Knockdown ( -ligand )
The fold change is expected to be greater than one because

of reduced silencing in the presence of ligand.
4 Notes
1. The cloning strain must be dam−/dcm− for efficient cleavage of

the ClaI restriction site in pCS1815.
2. The designed miRNAs can be expressed from any expression
vector as long as the miRNA is encoded downstream of a polII
promoter (e.g., CMV, EF1α) or a polIII promoter (e.g., U6).
The miRNA also can be encoded as an independent transcript
or in the 3′ UTR of a gene. Note that the latter configuration
reduces the expression of the upstream gene [14].
3. The observed fold-change in gene expression in response to
ligand varies with the configuration. In our experience, the com-
bined cis and trans configuration exhibits the largest fold-change,
whereas the cis configuration exhibits the smallest fold-change.
4. The selected mature miRNA sequence has a major impact on
the potency of gene silencing. While there are excellent tools
available for the design of these sequences [15], we recom-
mend testing at least four different sequences.
5. In our experience, the bulge size should be a minimum of 13
nucleotides. The bulge can include intervening base pairs of
limited length.
6. In our experience, no additional purification steps are required
beyond the standard purification performed by the DNA syn-
thesis company. Alternatively, the designed miRNA can be
ordered from different gene synthesis companies as a single
piece of double-stranded DNA.
7. We recommend that plasmid DNA can be harvested using the
PureYield™ Plasmid Miniprep System, which yields highly
pure, endotoxin-free plasmid DNA.
8. Other fluorescence measurement techniques, such as fluores-
cence microscopy or a fluorescence microplate reader, can be
employed. In the case of a fluorescence microplate reader or
any other bulk measurement technique, plasmids expressing
GFP and DsRed-Express should be co-transfected so both
proteins will be expressed in the same cell. Then bulk GFP
fluorescence can be normalized to bulk DsRed-Express fluo-
rescence similar to previous work [16].
Acknowledgments
This work was supported by North Carolina State University

(start-up funds to CLB), the National Science Foundation (fellow-
ship to RJB), the National Institutes of Health (RC1GM091298),
and the Defense Advanced Research Projects Agency
(HR0011-11-2-0002).
References
1. Bartel DP (2009) MicroRNAs: target recognition 10. Jenison RD, Gill SC, Pardi A, Polisky B (1994)
and regulatory functions. Cell 136:215–233 High-resolution molecular discrimination by
2. Bartel DP (2004) MicroRNAs: genomics, bio- RNA. Science 263:1425–1429
genesis, mechanism, and function. Cell 11. Han J, Lee Y, Yeom KH, Nam JW, Heo I,
116:281–297 Rhee JK, Sohn SY, Cho Y, Zhang BT, Kim VN
3. Gu S, Kay MA (2010) How do miRNAs medi- (2006) Molecular basis for the recognition of
ate translational repression? Silence 1:11 primary microRNAs by the Drosha-DGCR8
4. Alvarez-Garcia I, Miska EA (2005) MicroRNA complex. Cell 125:887–901
functions in animal development and human 12. Zeng Y, Cullen BR (2005) Efficient processing
disease. Development 132:4653–4662 of primary microRNA hairpins by Drosha
5. Leisner M, Bleris L, Lohmueller J, Xie Z, requires flanking nonstructured RNA
Benenson Y (2010) Rationally designed logic sequences. J Biol Chem 280:27595–27603
integration of regulatory signals in mammalian 13. Zeng Y, Yi R, Cullen BR (2005) Recognition
cells. Nat Nanotechnol 5:666–670 and cleavage of primary microRNA precursors
6. Dickins RA, Hemann MT, Zilfou JT, Simpson by the nuclear processing enzyme Drosha.
DR, Ibarra I, Hannon GJ, Lowe SW (2005) EMBO J 24:138–148
Probing tumor phenotypes using stable and 14. Beisel CL, Chen YY, Culler SJ, Hoff KG,
regulated synthetic microRNA precursors. Nat Smolke CD (2011) Design of small molecule-
Genet 37:1289–1295 responsive microRNAs based on structural
7. Tuerk C, Gold L (1990) Systematic evolution requirements for Drosha processing. Nucleic
of ligands by exponential enrichment: RNA Acids Res 39:2981–2994
ligands to bacteriophage T4 DNA polymerase. 15. Petri S, Meister G (2013) siRNA design prin-
Science 249:505–510 ciples and off-target effects. Methods Mol Biol
8. Ellington AD, Szostak JW (1990) In vitro 986:59–71
selection of RNA molecules that bind specific 16. An CI, Trinh VB, Yokobayashi Y (2006)
ligands. Nature 346:818–822 Artificial control of gene expression in mam-
9. Hermann T, Patel DJ (2000) Adaptive recog- malian cells by modulating RNA interference
nition by nucleic acid aptamers. Science through aptamer-small molecule interaction.
287:820–825 RNA 12:710–716
Chapter 20
A Three-Dimensional Design Strategy

for a Protein-Responsive shRNA Switch
Shunnichi Kashida and Hirohide Saito
Abstract
We have recently developed synthetic short hairpin RNA (shRNA) switches that respond to intracellular
proteins and control the expression of target genes in mammalian cells (Kashida et al. Nucleic Acids Res
40:9369–9378, 2012; Saito et al. Nat Commun 2:160, 2011). Here, we describe a method for the three-
dimensional (3D) design of a protein-responsive shRNA switch that employs modeling software and
known 3D structures of RNA–protein complexes. We were able to predict the effect of steric hindrance
between the Dicer enzyme and shRNA-binding protein in silico by superimposing the 3D model of the
shRNA switch on Dicer. The function of the designed switch can be evaluated in vitro and in living cells.
Our expertise will help utilize the 3D structure of biomacromolecular complexes for the design of func-
tional genetic switches.
Key words 3D design, Riboswitch, RNP, shRNA, Dicer, Steric hindrance, Prediction of switch
efficiency, Synthetic biology, RNAi
1 Introduction
One of the main goals in synthetic biology is to control and rewire

gene regulatory networks to improve human health and to gener-
ate new drugs [1]. Synthetic RNA switches that control target
gene expression, depending on the intracellular environment, play
a crucial role in achieving these objectives [2–5]. In this chapter,
we focus on engineering an shRNA that responds to intracellular
proteins and control knockdown of target genes. In the mamma-
lian shRNA processing pathway, an RNase III-class endonuclease,
Dicer, recognizes shRNA at the end of its stem and cleaves the
shRNA at approximately the 22nd nucleotide from the 3′- and/or
5′-end to produce a functional double-stranded RNA (dsRNA) for
performing RNAi [6, 7]. This endonuclease does not discriminate
the loop region of the shRNA during its processing, and therefore,
the desired sequences can be incorporated into the loop region in
general [8].
269
270 Shunnichi Kashida and Hirohide Saito
Fig. 1 The 3D design and construction of protein-responsive shRNA switches. The function of the in silico-
designed RNA switches can be evaluated using biochemical assays in vitro and gene reporter assays in mam-
malian cultured cells
It has been previously reported that a small molecule (e.g.,

theophylline) responsive shRNA switch was constructed by intro-
ducing the theophylline-binding RNA motif (aptamer) into the
loop region of shRNA. This modification inhibits Dicer activity
when the drug is bound to the loop [9, 10]. In addition, naturally
occurring protein-responsive shRNA has been discovered in
human systems. For example, the Lin28 protein interacts with pre-
let-7 microRNA (miRNA) and inhibits the processing of pre-let-7
microRNA into mature let-7 miRNA by Dicer, which leads to the
de-repression of let-7 target genes, including oncogenes and cell-
cycle genes [11, 12].
To design and construct a variety of synthetic RNA switches,
synthetic biologists have utilized the primary and secondary struc-
tures of the RNA molecules [10, 13–15]. Alternatively, functional
switches have been screened from RNA library sets containing ran-
dom sequences [16, 17]. However, it is difficult to predict the
function of the generated RNA switches prior to experiments in
most cases. Thus, we aimed to develop a method for the 3D
molecular design of protein-responsive RNA switches with a pre-
dictable function. This method will contribute to the investigation
of the design principles of functional genetic switches composed of
biomolecular complexes [18].
Here, we describe the 3D design of protein-responsive shRNAs
in silico based on high-resolution structures of known RNA–pro-
tein (RNP) complexes. We evaluated the performance of these shR-
NAs in vitro and in mammalian cells (Fig. 1). The designed shRNA
senses the intracellular RNA-binding protein (trigger protein) and
inhibits the activity of Dicer via steric hindrance between Dicer and
the RNP complex (The model of a protein-responsive shRNA
switch is shown in Fig. 2). Thus, these synthetic shRNA devices can
control target gene expression in a predictable manner.
3D Molecular Design for shRNA Switches 271
Fig. 2 Schematic representation of the protein-responsive shRNA device mechanism and its bound protein
orientation. (a) In the absence of the trigger protein, Dicer can access and process the designed shRNA, and
the processed shRNA induces the knockdown of its target gene through RNAi (left). Dicer is inaccessible to
shRNA in the presence of the trigger protein that binds to the loop region of the shRNA because the RNP inter-
action faces Dicer and inhibits its access (right). The prevention of Dicer’s function causes the de-repression
of the gene knockdown. (b) Protein-responsive shRNA switches are designed in silico by connecting an RNP
motif to the corresponding dsRNA and Dicer. By specifying the position of the devices with reference to the
Dicer cleavage sites and by comparing the bound proteins’ locations, we observed that with a 1-bp (base pair)
insertion, the bound trigger protein on the shRNA switch is located approximately 2.6 Å more distant from the
Dicer cleavage sites and rotates approximately 30° in an anticlockwise direction around the axis of the dsRNA
2 Materials
2.1 3D RNP The 3D molecular structures of the RNP-binding motifs can be

Structures obtained from the Protein Data Bank (PDB, Worldwide Protein
Data Bank). We constructed the protein-responsive shRNA
switches using the four RNP molecular structures described in
Table 1. The structures from both X-ray and NMR structural anal-
yses are available. To generate effective shRNA switches, tight
interaction between RNP (e.g., less than a 50 nM dissociation con-
stant; Kd) is preferable. The RNA motif extracted from the paren-
tal RNP structure should have more than 1 bp at the 5′/3′-ends
because the region of the motif must be connected to and super-
imposed on the dsRNA region of the shRNA.
2.2 Software Discovery Studio (Accelrys) ver. 2.5-3.1 can be used for the mod-
and Operating eling. The operation of the software should be conducted in accor-
Environment for dance with the instructions of the software. In addition, more than
Macromolecular 6 GB of cache memory capacity is recommended to avoid stalling
Design the program during modeling.
2.3 Inhibition Assay For in vitro transcription, we used T7 RNA polymerase and buffer
of Human Dicer prepared in-house or used a MEGAshortscript™ T7 Kit (AM1354,
Cleavage for shRNA Invitrogen) with slight modifications (see Note 1).
In Vitro 1. Single-stranded template DNA (purchase from an oligonucle-
2.3.1 In Vitro otide synthesis company): 5′-(complementary strand of the
Transcription and shRNA switch) TATAGTGAGTCGTATTAGC-3′ (see also
Purification of shRNA Subheading 3.3 for the sequence design of the shRNA switch).
Switches 2. T7 annealing primer: 5′-GCTAATACGACTCACTATA-3′.
3. Transcription reaction mixture: 0.75 μM single-stranded tem-
plate DNA, 0.75 μM T7 annealing primer, 7.1 μg/mL Pyro
phosphatase (Roche), 50 μg/mL bovine serum albumin,
40 mM HEPES–KOH, 20 mM MgCl2, 5 mM DTT, 2 mM
Table 1
Features of the RNP structures used for shRNA switches
PDB ID Protein Kd protein organism RNA origin structural analysis method

1RLG L7Ae 0.46 nM Archaeoglobus fulgidus Ribosomal RNA X-ray structural analysis
1URN U1A <50 pM Homo sapiens U1snRNA stem loop II X-ray structural analysis
1AUD U1A <50 pM Homo sapiens 3'UTR of U1A mRNA NMR
1OOA NFκB p50 5.4 nM Mus musculus from in vitro selection X-ray structural analysis
PDB ID RNA sequence

1RLG GCUC UGACCGAAAGGCGUGAUGAGC
1URN AA UCCAUUGCACUCCGGAUUU
1AUD GGCAG AGUCCUUCGGGACAUUGCACCUGCC
1OOA CAUAC UUGAAACUGUAAGGUUGGCGUAUG
Spermidine, 4.8 mM ATP, 4.8 mM UTP, 4.8 mM CTP,

1.28 mM GTP, 12.8 mM GMP (see Note 2).
4. T7 RNA polymerase (prepared in-house).
5. 2 U/μL TURBO DNase (Ambion).
6. PD-10 column (GE Healthcare).
7. PD-10 buffer: 0.3 M potassium acetate, pH 6.0, 15 %
ethanol.
8. 15 % native gel: 15 % acrylamide–bisacrylamide (29:1), 0.1 %
ammonium persulfate, 0.05 % N,N,N′,N′-
tetramethylethylenediamine, 0.5× TBE buffer.
9. 5× loading buffer: 0.25 % BPB, 30 % glycerol.
10. 22-μm PES microfilter (Millex-GP).
11. Elution buffer: 0.5 M NaCl, 0.1 % SDS, 1 mM EDTA.
2.3.2 Expression and 1. Competent cell: BL21 (DE3) pLysS (Promega) or chemically
Purification of the Trigger competent E. coli cells suitable for transformation and protein
RNA Binding Proteins expression.
Using FPLC 2. Phosphate buffer A: 20 mM NaH2PO4, 20 mM Na2HPO4,
500 mM NaCl.
3. Phosphate buffer B: 20 mM NaH2PO4, 20 mM Na2HPO4,
500 mM NaCl, 1 M imidazole.
4. FPLC: AKTApurifier (GE Healthcare).
5. Prepacked column: HisTrap HP column (5 mL, GE
Healthcare).
6. 6× Sample buffer: Sample buffer solution with reducing
reagent (6×) for SDS-PAGE (Nacalai).
7. 2× dialysis buffer: 40 mM HEPES–KOH, pH 7.8, 230 mM
NaCl, 10.8 mM KCl, 1.6 mM MgCl2.
8. Dialysis cassette: Slide-A-Lyzer Dialysis Cassettes (Thermo
Scientific).
9. Sample concentrator: Vivaspin 6 concentrator (GE Healthcare).
2.3.3 Inhibition Assay of 1. Recombinant human Dicer: Recombinant human Dicer can be
Human Dicer Cleavage for purchased from Genlantis.inc for the experiments. Human
shRNA In Vitro Dicer purified in the laboratory can also be used for the
experiments.
2. Dicer reaction mixture: 1 μL of 10 mM ATP, 0.5 μL of 50 mM
MgCl2, 4 μL of Dicer Reaction Buffer (Genlantis), 1 μL of 0.5
unit/μL Recombinant Human Dicer Enzyme (Genlantis).
3. Binding buffer: Opti-MEM I (Life Technologies).
4. Gel staining buffer: 1× SYBR Green I and SYBR Green II in
0.5× TBE buffer.
2.4 Construction of 1. ShRNA cloning kit: BLOCK-iT™ Inducible H1 RNAi Entry

Plasmids and RNAi Vector Kit (Invitrogen).
Experiments in 2. ShRNA top strand (purchase from an oligonucleotide synthe-
Mammalian Cells sis company): 5′-CACC (strand of the shRNA switch without
2.4.1 shRNA Expressing
additional two nucleotides (Fig. 4, 3)) -3′ and ShRNA bottom
Plasmid
strand: 5′-AAAA (complementary strand of the shRNA switch
without additional two nucleotides (Fig. 4, 3)) -3′.
2.4.2 Trigger Protein pcDNA3.1, 4 and 5 (Life Technologies) or other expression plas-
Expression Plasmid mids suitable for expression in mammalian cells can be used.
2.4.3 EGFP Expression 1. Human cultured cells: Maintain 293FT cells (Life Technologies)
Control of the Designed using Dulbecco’s modified Eagle’s medium (Nacalai) contain-
shRNA Switches in ing 10 % fetal bovine serum (Cambrex), 2 mM L-glutamine
Cultured Mammalian Cells (Life Technologies), 0.1 mM MEM Nonessential Amino Acids
(Life Technologies), 1 mM sodium pyruvate (Sigma), 0.5 %
Penicillin-Streptomycin (Life Technologies) and 500 μg/mL
G418 sulfate at 37 °C under a 5 % CO2 concentration. HeLa
cells, which were also used, and other human cultured cells
successfully transfected with plasmids can also be used for the
experiment.
2. Lipofection reagent: Lipofectamine 2000 (Life Technologies).
3. Fluorescent microscopy: To observe EGFP expression in cells,
we used an IX-81 inverted microscope system (Olympus), a
470- to 495-nm filter for excitation and a 510- to 550-nm fil-
ter for emission.
4. Flow cytometry: To detect and quantify EGFP fluorescence,
we used a flow cytometer (FACS Aria, BD Bioscience), a 488-
nm semiconductor laser beam, and a 530/30-nm emission
detector.
3 Methods
The methods can be divided into three steps (Fig. 3) as follows.

Step 1: The three-dimensional design of protein-responsive shRNA
in silico based on the 3D molecular structures of the RNP motif
using modeling software. The steric hindrance between Dicer and
the trigger protein on the shRNA can be predicted from the model,
which determines the inhibition efficiency of Dicer cleavage.
Step 2: Evaluate the inhibition of Dicer cleavage using recombi-
nant human Dicer in vitro.
Step 3: Assess the function of the designed shRNA switches in
human cultured cells.
Fig. 3 Overview: 3D molecular design and assessment of the protein-responsive shRNA switches
3.1 Obtaining Construct protein-responsive shRNA switches using the specific

RNP Motifs RNP-binding pairs between a trigger protein and the correspond-
ing RNA sequence. Choose the RNP motifs from known RNP
complexes of the trigger proteins and their binding RNA.
3.2 dsRNA Sequence Obtain the dsRNA sequence that induces knockdown of the target
Targeting Specific gene through RNAi from literature reports or databases. Sequences
mRNA that start with “G” are preferable because the first “C” in the DNA
template is better suited for the efficient transcription of T7
RNA polymerase in vitro and RNA polymerase III in human cells.
If there are no known dsRNA sequences, construct and assess the
Fig. 4 Design steps of shRNA switch on the secondary structure
candidate dsRNA sequences using an siRNA design algorithm,

such as siExplorer, which is available at the URL given below:
http://rna.chem.t.u-tokyo.ac.jp/siexplorer.htm.
1. Input the mRNA sequence or NCBI RefSeq accession number
(NM_xxxx) and run the algorithm.
2. Pick the three or four sequences that start with “G” from the
results table and design the shRNA by connecting the dsRNA
with the appropriate loop (we used a 5′-GAAA-3′ loop or
5′-AGCAUAG-3′ loop).
3. Construct in vitro-transcribed shRNA or shRNA-expressing
plasmids.
4. Assess the RNAi efficiency in human cultured cells by
transfection.
5. After 24–48 h of transfection, collect the cells, extract the total
protein, and quantify the selected protein using western blot-
ting. Alternatively, analyze the knockdown efficiencies of the
target fluorescent protein by flow cytometry.
3.3 shRNA Switch Obtain the 3D molecular structure of the RNP motif composed of
Design on the the trigger protein and its binding RNA motif from the PDB. If
Secondary Structure the 3D structure is not available, the 3D molecular design step
(Fig. 3, Step 1) can be skipped. In this case, the shRNA switches
can be constructed based on the secondary structure as follows
(Fig. 4).
1. Choose the appropriate RNA motifs. Omit the unnecessary
sequences (sequences not having a positive effect on binding
affinity) and extract minimal RNA motifs.
2. Design defective RNA motifs that do not bind to the trigger
proteins by mutating or deleting the corresponding nucleo-
tides of the motif.
3. Connect the RNA motif with the corresponding length of
dsRNA (21–31 bp, also see Note 3) designed to code the tar-
get mRNA sequence to form the stem-loop structure of the
RNA (Fig. 4, 1–3).
4. Add two nucleotides (“AG” or “UU”) at the 3′ end of the
RNA (Fig. 4, 4) to help Dicer recognize the target dsRNA.
3.4 3D Molecular We describe the 3D molecular design method for protein-

Design of Protein- responsive shRNA switches using two examples: U1A-responsive
Responsive shRNA shRNA switches (Fig. 3, Step 1) and NFκB (p50)-responsive
Switches Using shRNA switches. U1A is one of the most abundantly expressed
Discovery Studio mRNA-processing and housekeeping proteins in mammalian cells
[19]. NFκB is one of the major transcription factors in eukaryotic
cells, and its abnormal expression has been linked to the prolifera-
tion of several types of tumor cells [20, 21]. For 3D modeling, the
Discovery Studio ver. 2.5-3.1 can be used.
1. Startup Discovery Studio and download the PDB file (.txt) of
the selected RNP motifs. The database stores the structure
from X-ray structural or NMR analyses (see Note 4 about
NMR structure). We obtained the N-terminal domain of U1A
(human) and its binding RNA motif in U1snRNA (U1A1-
RNA motif) from the PDB (ID: 1URN) and p50 (mouse
NFκB) bound to its aptamer RNA (p50-RNA motif) from the
PDB (ID: 1OOA) (see Note 5 for using the structure of heter-
ologous RNP). The PDB files from the X-ray structural analy-
sis contained atomic coordinates of the RNP crystal lattice
structure, which consists of the one, two, three, four or six
identical groups of the RNP (U1A, three groups; p50, two
groups), water, ions and other solvent molecules. The groups
are symmetrically arranged on the rotation axes.
2. Focus on one RNP group of the crystal lattice and exclude the
other groups (uncheck the boxes of the groups in the
“Hierarchy window”), and clear the water and solvent mole-
cules from the RNP group.
3. (Optional) Trim the a.a. sequence of the protein (or the pro-
tein domain) to be consistent with the sequence of the pro-
teins expressed from the plasmid.
4. Create 22–28 base pairs (bp) of A-form dsRNA. Use “create/
grow” at the 3′ function of the “Build and Edit Nucleic Acid”
tool and input the sequence from the start of the shRNA
switch sequence. The dsRNA consists of a sense strand (the
sequence list is indicated as “S” in the RNA duplex group of
the hierarchy window) and an antisense strand (indicated as
“A”), and the sense strand should correspond with the 5′-end
sequence of the stem in the shRNA switch.
5. Choose the RNA structures and change the display style of the
atoms from “None” to “Scaled ball and stick.”
6. Align the RNA motif and dsRNA three-dimensionally in
reference to the nucleotides of the stem end of the RNA motif
(U1A1-, 2 bp; p50-, 5 bp; italic RNA sequences in Table 1)
and those of the 3′-end of the sense strand.
7. Tether the corresponding atoms of the sugars, bases and

phosphate backbones in the reference nucleotides (>2 bp) of
the RNA motif and dsRNA using the “Add tether” function in
the superimpose protocol.
8. Superimpose the RNA motif on the dsRNA using the
“Superimpose by Tether” function in the superimpose proto-
col. The distance between the RNA motif and dsRNA are min-
imized using the least squares approximation polynomial and
the distance of the corresponding tethered atomic coordinate
pairs.
9. Delete the reference nucleotides of dsRNA that overlap with
those of the RNA motif.
10. To concatenate the sense strand, RNA motif and antisense
strand, connect the oxygen atom of the phosphate and the 5′
carbon atom (C5) of the ribose or connect the oxygen atom of
the C5 and the phosphorous atom at the ends of the sequences
with a single bond. If the oxygen atom of C5 and that of the
phosphate overlap with each other, delete either one of the
two and connect the C5 and phosphorus atoms with one oxy-
gen atom.
11. Drag and drop the list of the RNA motif nucleotides in the
“Hierarchy window” onto that of the sense strand “S.” Then,
drag and drop the list of the antisense strand “A” onto “S” to
concatenate the list of the sequences.
12. Add two nucleotides (“AG” or “UU”) at the 3′-end of the
concatenated RNA sequence using the function “Create/
Grow at 3′-”
13. Change the display style of the 22nd nucleotide from both
ends from “Scaled Ball and Stick” to “CPK” to highlight the
Dicer cleavage sites. Human Dicer recognizes the shRNA at
the end of its stem and cleaves it at the 22nd nucleotide from
the 3′- and/or 5′-end [6].
14. Choose the bound trigger protein, change the display style of
the atoms from “None” to “CPK” and add a surface to the
protein to visualize the occupied volume of the molecules.
15. Repeat steps 4–14 to generate the bp series of the shRNA
switch (U1A1-, 21–31 bp; p50-, 22–28 bp).
16. To overlay the 22-bp long switch and 23-bp variant, tether the
atoms of the base in the 3′ terminal nucleotide of the two.
Then, overlay the two switch variants using the “Superimpose
by Tether” function. Drag and drop the list of the 23-bp vari-
ant onto that of the 22-bp long switch. Then, hide the variant
by unchecking the group box.
17. Repeat step 16 to overlay 22 bp with the next bp variant to
overlay all of the bp variants individually.
3.5 Prediction of The entire structure of human Dicer has not been solved. For our
Steric Hindrance 3D model of Dicer, we used the RNase III catalytic domain and
Between Dicer and the peripheral structure of Giardia Dicer (PDB ID: 2QVW, 326–385,
Trigger Protein Bound and 401 C-terminal a.a.; see Note 6). In this section, we demon-
on the shRNA Switch strate how to display the catalytic domain of Dicer, highlight the
catalytic sites and superimpose the RNP on Dicer.
3.5.1 Displaying and 1. Download the PDB file of the crystal structure of Giardia
Modifying the Catalytic Dicer (PDB ID: 2QVW). The crystal structure has four sym-
Domain and Peripheral metric groups in the crystal lattice.
Structure of Giardia Dicer 2. Exclude three groups and focus on one group.
3. Trim the a.a. sequence of the protein without 326–385 and
401 of the C-terminal a.a by unchecking the checked boxes of
the nucleotides.
4. Change the display style of the Dicer protein and the four
metal ions from “None” to “CPK,” and highlight the metal
ions and 9 a.a. (E336, E340, D404, D407, E649, D653,
E684, D720, and E723) by changing the color of the atoms.
3.5.2 Superimpose the 1. Copy and paste the 3D molecular model of the protein-
Protein-Responsive shRNA responsive shRNA switch variants and bound trigger protein
Switch and Bound Trigger (designed in Subheading 3.4) on the modified Dicer structure
Protein on Dicer displayed in the window.
2. Arrange the positions and orientations of the switch and Dicer
three-dimensionally in reference to the two phosphates of the
22nd nucleotides, the highlighted metal ions and the 9 a.a. of
Dicer.
3. Choose the modified Dicer and add surfaces to the proteins.
4. Display the modified Dicer and each variant of the shRNA
switch with the trigger protein and observe the degree of ste-
ric hindrance between them. You can predict which trigger
protein of the variants will most strongly collide against Dicer
(see Note 7).
3.6 Inhibition Assay 1. Add 30 μL of T7 RNA polymerase to 670 μL of the transcrip-

of Human Dicer tion reaction mixture and mix gently by tapping. Do not spin
Cleavage for shRNA the mixture down because that may cause precipitation.
In Vitro Incubate for 3–4 h at 37 °C.
2. For DNA template degradation, add 10 μL of TURBO DNase
3.6.1 In Vitro
to the reaction and incubate for 20 min at 37 °C.
Transcription and
Purification of shRNA 3. Equilibrate a PD-10 column with 50 mL of PD-10 buffer.
Switches 4. To remove any proteins from the mixture via phenol-
chloroform extraction, add 700 μL (1× volume) of TE-saturated
phenol to the reaction and thoroughly vortex. Then, centri-
fuge the mixture at >20,000 × g at 4 °C for 5 min. Transfer the
upper layer to a new microtube.
5. Add 700 μL (1× volume) of chloroform to the tube and thor-

oughly vortex. Then, centrifuge the mixture at >20,000 × g at
4 °C for 5 min. Transfer the upper layer to a new microtube.
6. To remove nucleotide monomers, add the reaction to the pre-
viously equilibrated PD-10 column and wash the column with
3 mL of PD-10 buffer (for 55–80 nucleotides of RNA; the
buffer volume can be optimized for each RNA size).
7. Elute the shRNA switches with 1 mL of PD-10 buffer.
8. Add 1 mL of (1× volume) ethanol (−20 °C) to the elution, mix
thoroughly and incubate for 30 min at −20 °C.
9. Centrifuge the mixture at >20,000 × g at 4 °C for 20 min,
remove the supernatant and allow the pellet to dry.
10. Pre-run a 15 % native gel for polyacrylamide gel electrophore-
sis (PAGE) in 0.5× TBE buffer for 20 min at room
temperature.
11. Dissolve the pellet with 5× loading buffer and load the sample
into the polyacrylamide gel. Perform electrophoresis at room
temperature until the blue band of BPB travels to a distance of
one-third of the glass plate from the bottom.
12. Place the gel on a silica plate and expose it to UV (265 nm)
light to confirm the shRNA switch band, and then excise the
band.
13. To elute the shRNA switch, add the excised gel piece and
500 μL of elution buffer (0.5 M NaCl, 0.1 % SDS, 1 mM
EDTA) to a new tube and incubate at 37 °C overnight.
14. Filter the elution with a microfilter, add a 2.5 elution volume
of ethanol (−20 °C) to the elution, mix thoroughly and incu-
bate for 30 min at −20 °C.
15. Centrifuge the mixture at >20,000 × g at 4 °C for 20 min and
remove the supernatant. Then, add a 2.5 elution volume of
80 % ethanol to rinse the pellet.
16. Centrifuge the mixture at >20,000 × g at 4 °C for 5 min and
remove the supernatant. Allow the pellet to dry.
17. Dissolve the pellet with double-distilled water (D2W) and
adjust the concentration to approximately 10 μM.
3.6.2 Expression and The use of an affinity column and FPLC, such as the AKTA (GE
Purification of the Trigger Healthcare) system, is recommended for purification of the trigger
RNA Binding Proteins RNA binding proteins. For U1A protein purification, we modified
Using FPLC a protein expression and purification protocol that has been
reported previously [22].
1. Pre-culture competent cells transformed with the His-tagged
trigger protein expression plasmid with 10 mL of LB (+antibiotic
for selection) at 37 °C and 200 rpm (linear shake) overnight.
2. Transfer 7 mL of the pre-culture into 700 mL of LB (+antibi-

otic for selection) in flasks and culture at 37 °C and 180 rpm
(rotary shake) until the OD660 reading of the culture reaches
approximately 0.65 (requiring approximately 2.5 h).
3. Add 700 μL of 1 M IPTG (isopropylthio-β-galactoside) or the
suitable ligand for induction of T7 RNA polymerase expres-
sion in the culture (f.c. 1 mM) and culture for another 4 h.
4. Split the culture in two and collect the cells with centrifugation
at 4 °C and 4,170 × g for 10 min. The centrifuged pellets can
be stored at −80 °C.
5. Suspend one of the pellets with a mixture of 5 mL of phos-
phate buffer A and 25 μL of phosphate buffer B.
6. Sonicate the suspension for 15 s (interval of 30 s) 12 times on ice.
7. Centrifuge the suspension at >20,000 × g at 4 °C for 30 min
and collect the supernatant.
8. Filtrate the supernatant through a microfilter.
9. Perform FPLC purification with a prepacked column: Pomp
wash a 5 column volume (cv) of 5 mM imidazole buffer (mix-
ing phosphate buffer A:B = 199:1). Inject the supernatant from
the sample loop. Wash a 10 cv of 20 mM imidazole buffer.
Gradient elute a 10 cv of 20–200 mM imidazole buffer. Elute
a 10 cv of 500 mM imidazole buffer.
10. Add 2 μL of 6× Sample buffer to 10 μL of flow-through and
fractions whose absorbance at 280 nm are highly detectable by
FPLC and incubate at 95 °C for 5 min. After SDS-PAGE, stain
the gel with CBB and bleach with water. Check the band for
the desired protein.
11. Equilibrate a dialysis cassette with 2× dialysis buffer and add
the fraction containing the protein into the dialysis cassette
using a 10-mL syringe.
12. Dialyze with 600 mL of 2× dialysis buffer for 2 h at 4 °C,
change to fresh buffer and incubate for another 6 h.
13. Measure the absorbance of the protein at 280 nm and roughly
estimate the concentration. If the solution is sufficiently
concentrated (>40 μM), skip to step 16. If not, calculate how
many folds the volume must be reduced by to provide suffi-
cient concentration.
14. Equilibrate a sample concentrator with 2× dialysis buffer and
add the dialyzed protein solution into the sample concentrator
(up to 4 mL).
15. Concentrate the solution with centrifugation at 10,000 × g and
4 °C for 15 min. Check the volume and spin again until the
volume in the concentrator is reduced to the set volume.
16. Add the protein solution into the same volume of glycerol and
store at −30 °C.
3.6.3 The Binding 1. Mix 50 nM in vitro transcribed shRNA switch and 0 or

Affinities of the shRNA 0.1–2 μM purified trigger proteins (the concentration can be
Switches and the Purified adjusted and at least four concentration points are needed to
Trigger Proteins analyze the affinity) in the binding buffer to form the RNP
complex. The total volume of the mixture can be adjusted to
the loading volume for PAGE (we use 30 μL).
2. Incubate on ice for 30 min.
3. Pre-run a 15 % native gel (use an approximately 10-cm size
gel) in 0.5× TBE buffer for 10 min at 4 °C.
4. Add 5× loading buffer to the RNP solution and mix gently by
tapping.
5. Load the samples into the 15 % native gel and run PAGE for
40 min at 400 V and 4 °C.
6. Stain the gel with the gel staining buffer for 15 min at room
temperature.
7. Wash the gel with D2W a few times.
8. Image the gel with a fluorescent image analyzer to detect the
stained RNA.
9. Confirm that the shifted band can be observed from the cog-
nate RNP binding pair but not from the sample that contains
the defective RNA motif.
10. Determine the protein concentration using the Dicer cleavage
assay in Subheading 3.6.5 to a twofold higher concentration
than the concentration that totally shifted the band.
3.6.4 The Efficiency 1. Mix 0.5 μL of in vitro-transcribed shRNA switch with the Dicer
of In Vitro Human Dicer reaction mixture and water (up to 10 μL). Do not chill the
Cleavage for the shRNA mixture during assembly because this may cause precipitation.
Switches
2. Incubate the mixture at 37 °C for 15 h and stop the incubation
by adding 1 μL of Dicer Stop Solution to the reaction.
3. Conduct the same procedure detailed in steps 3–8 in
Subheading 3.6.3 on the incubated reaction.
4. Quantify the total fluorescent intensity of the band of cleaved
dsRNA and confirm that the cleavage efficiency by human
Dicer is the same among the base-pair variants and their nega-
tive controls (defective RNA motif) for the shRNA switches.
3.6.5 Inhibition Assay 1. Mix 0.5 μM in vitro-transcribed shRNA switch with the corre-
of Human Dicer Cleavage sponding trigger protein (The concentration of the protein was
for shRNA In Vitro determined at step 10 in Subheading 3.6.3 to form the RNP
complex). The total volume of the mixture is up to 3.5 μL.
2. (Optional) Incubate at room temperature for approximately
30 min. The incubation time will depend on the affinity
between the RNP complexes.
3. Mix the RNP complex with 1 μL of 10 mM ATP, 0.5 μL of

50 mM MgCl2, 4 μL of Dicer Reaction Buffer, 1 μL of
0.5 unit/μL Recombinant Human Dicer Enzyme, and water
(up to 10 μL).
4. Incubate the mixture at 37 °C for 15 h and stop the incubation
by adding 1 μL of Dicer Stop Solution to the reaction.
5. Conduct the same procedure as detailed in steps 3–8 in
Subheading 3.6.3 on the incubated reaction.
6. Quantify the total fluorescent intensity of the band of cleaved
dsRNA (Fig. 3, Step 2).
7. Calculate the relative inhibition efficiency values of Dicer cleav-
age using the ratio of the intensity in the presence of the trigger
protein to that in the absence of the trigger protein.
3.7 EGFP Expression The ShRNA cloning Kit can be used to construct the shRNA-
Control of the switch expressing plasmids. The plasmids should be constructed
Designed shRNA using the kit, and required DNA sequences should be synthesized
Switches in Cultured in accordance with the kit manual. The annealing steps in the kit
Mammalian Cells manual are modified as bellow:
3.7.1 Construction 1. Mix 200 μM of ShRNA top strand and 200 μM of ShRNA
of shRNA Expressing bottom strand in Oligo Annealing Buffer of the cloning kit.
Plasmids 2. Incubate the mixture using a thermal cycler: 95 °C for 4 min
and incubate at 80 °C for 1 min. Decreased 0.1 °C/s to the
melting temperature of the duplex of the top strand and bottom
strand. And incubate the mixture at the melting temperature for
5 min, followed by room temperature incubation for 10 min.
Prepare the negative control plasmid that contains a defective
RNA motif using the same procedure as indicated.
3.7.2 Trigger Protein The trigger protein expression plasmid is prepared as follows.
Expression Plasmid
1. Design the set of PCR primers for amplifying the region of the
gene that codes the trigger protein, whose amino acid (a.a.)
sequence corresponds with that of the protein in the RNP
complex derived from the PDB. Add each sequence of PCR
primer to the cutting site of a selected different restriction
enzyme.
2. Amplify the gene by PCR from the total cDNA or plasmid
using the designed primers.
3. Clone the gene into destination plasmids between the two dif-
ferent restriction enzyme sites to construct the trigger protein
expression plasmid.
4. Verify the sequence.
5. Prepare the negative control plasmid that contains a noncog-
nate protein sequence using the same procedure as indicated.
3.7.3 EGFP Expression 1. Seed 293FT cells (1.5 × 105 per well) or other chosen cultured
Control of the Designed cells on a 24-well plate and incubate at 37 °C in a CO2 incuba-
shRNA Switches in tor for 1 day.
Cultured Mammalian Cells
2. Co-transfect cells with the following three types of plasmids
using 2 μL of the Lipofection reagent: 0.3 μg of the bp variant
of protein-shRNA switches expressing plasmids or that of the
binding-defective control, 0.3 μg of the trigger protein expres-
sion plasmid or that of the binding-defective control and
0.3 μg of the EGFP expression plasmid.
3. Twenty-four hours after transfection, observe and image the
transfected cells using a fluorescent microscope (Fig. 3, Step 3).
4. Collect the medium of the 24-well plate and transfer to new
tubes.
5. Add 200 μL of 0.25 % trypsin-EDTA (1×) phenol red to each
well and incubate at 37 °C in a CO2 incubator for 2–5 min.
6. Add the collected medium to each well, resuspend in the
medium and filter the cell suspension with a 12 × 75 mm tube
and a cell strainer cap.
7. Quantify the mean intensity of the EGFP fluorescence of
30,000 cells per sample using a flow cytometer.
8. To exclude abnormal cells, select the cells whose FCS and SSC
signal scopes are identical to those of untreated cells to calcu-
late the mean intensities.
4 Notes
1. The MEGAshortscript™ T7 Kit can be used for the transcrip-

tion reaction by adding GMP and adjusting the ratio of NTP
concentrations as described below:
(a) Add 4 μL of D2W, 0.5 μL of 100 μM single-stranded tem-
plate DNA, 0.5 μL of 100 μM T7 annealing primer, 2 μL of
T7 10× reaction buffer, 1 μL of 75 mM GTP solution, 4 μL
of 75 mM GMP solution, 2 μL of 75 mM ATP solution,
2 μL of 75 mM CTP solution, and 2 μL of 75 mM UTP
solution to the tube and mix thoroughly with vortexing.
(b) Add 2 μL of T7 RNA polymerase and mix gently by tap-
ping. Do not spin the mixture down because this may
cause precipitation. Set the total reaction in a 20-μL
volume.
(c) Follow steps 3–17 in Subheading 3.6.1.
2. The GTP at the 5′-end of RNA causes interferon activation,
followed by nonspecific translational repression [23, 24].
To avoid the interferon activation and increase the efficiency
of Dicer processing [6], we introduced GMP at the 5′-end
of the RNA by adjusting the concentration ratio of GMP–

GTP from 4:1 to 10:1 during the transcription reaction.
3. We confirmed that the shRNA sequences that contain shorter
dsRNA regions (<22 bp) are not competent for Dicer process-
ing. We also observed that the longer dsRNA regions (>27 bp)
of the shRNA might cause lower transcription efficiency from
the shRNA-coding plasmids. To avoid the possible interferon
response of the longer dsRNA in the cellular assay, we think that
a range of 22–27-bp dsRNA is likely sufficient to generate opti-
mal shRNA switches. Changing the length of the dsRNA region
from 22 to 27 bp can rotate the RNA loop by approximately
150°, so the steric hindrance between Dicer and the RNP com-
plex can be optimized within this range in most cases [3].
4. The PDB files from the NMR analysis contain the RNP struc-
ture ensemble. Use the representative structure, which is dis-
played as the first alternative [25].
5. If the PDB does not offer the RNP structure of human, other
organisms may be used after verifying the a.a. sequence differ-
ences between the proteins of the two organisms (e.g., mouse
and human) and confirming that the differences are not critical
for the structure and binding affinity. We verified the a.a.
sequence differences between mouse and human p50 and con-
firmed that the six point mutations between these two
sequences were either on the terminal region or on the loop far
from the RNA-binding domain [3].
6. We used the RNase III catalytic domain and peripheral struc-
ture of Giardia Dicer because structural studies have indicated
that the crystal structure of Giardia Dicer is geometrically
similar to the nuclease core of human Dicer [26]. Two metal
ions are coordinated by 9 a.a. (E336, E340, D404, D407,
E649, D653, E684, D720, and E723) in each of the catalytic
sites of Dicer [27, 28]. The phosphodiester bonds of the 22nd
nucleotides from both the 5′-/3′-ends of the dsRNA are
cleaved between the two metal ions. We trimmed the Dicer
structure of 2QVW to use the 326–385 and 401 C-terminal
a.a. not only to use the domains whose structures are similar to
those of human but also to highlight the region at which the
steric hindrance between Dicer and the trigger protein bound
on the shRNA could occur.
7. Steric hindrance may occur not only between Dicer and the trig-
ger protein but also on the protein-stabilized RNA structural
motif because a particular RNA–protein interaction is known to
stabilize the RNA structural motif containing the protein-bind-
ing site. We have previously shown that both an RNA-binding
protein and the protein-stabilized RNA structural motif that has
the predicted steric hindrance with Dicer can inhibit RNAi [3].
References
1. Endy D (2005) Foundations for engineering ulation of internal ribosome entry in wheat
biology. Nature 438:449–453 germ extract and their applications as label-free
2. Kashida S, Inoue T, Saito H (2012) Three- biosensors. RNA 17:478–488
dimensionally designed protein-responsive 16. Sinha J, Reyes SJ, Gallivan JP (2010)
RNA devices for cell signaling regulation. Reprogramming bacteria to seek and destroy
Nucleic Acids Res 40:9369–9378 an herbicide. Nat Chem Biol 6:464–470
3. Saito H, Fujita Y, Kashida S et al (2011) 17. Muranaka N, Sharma V, Nomura Y et al (2009)
Synthetic human cell fate regulation by protein- An efficient platform for genetic selection and
driven RNA switches. Nat Commun 2:160 screening of gene switches in Escherichia coli.
4. Xie Z, Wroblewska L, Prochazka L et al (2011) Nucleic Acids Res 37:e39
Multi-input RNAi-based logic circuit for iden- 18. Liang JC, Bloom RJ, Smolke CD (2011)
tification of specific cancer cells. Science 333: Engineering biological systems with synthetic
1307–1311 RNA molecules. Mol Cell 43:915–926
5. Culler SJ, Hoff KG, Smolke CD (2010) 19. Oubridge C, Ito N, Evans PR et al (1994)
Reprogramming cellular behavior with RNA Crystal structure at 1.92 A resolution of the
controllers responsive to endogenous proteins. RNA-binding domain of the U1A spliceosomal
Science 330:1251–1255 protein complexed with an RNA hairpin.
6. Park J-E, Heo I, Tian Y et al (2011) Dicer rec- Nature 372:432–438
ognizes the 5′ end of RNA for efficient and 20. Galardi S, Mercatelli N, Farace MG et al (2011)
accurate processing. Nature 475:201–205 NF-kB and c-Jun induce the expression of the
7. Vermeulen A, Behlen L, Reynolds A et al oncogenic miR-221 and miR-222 in prostate
(2005) The contributions of dsRNA structure carcinoma and glioblastoma cells. Nucleic
to Dicer specificity and efficiency. RNA 11: Acids Res 39:3892–3902
674–682 21. Ben-Neriah Y, Karin M (2011) Inflammation
8. Tsutsumi A, Kawamata T, Izumi N et al (2011) meets cancer, with NF-κB as the matchmaker.
Recognition of the pre-miRNA structure by Nat Immunol 12:715–723
Drosophila Dicer-1. Nat Struct Mol Biol 22. Endoh T, Sisido M, Ohtsuki T (2008) Cellular
18:1153–1158 siRNA delivery mediated by a cell-permeant
9. An C-I, Trinh VB, Yokobayashi Y (2006) RNA-binding protein and photoinduced RNA
Artificial control of gene expression in mam- interference. Bioconjug Chem 19:1017–1024
malian cells by modulating RNA interference 23. Poeck H, Besch R, Maihoefer C et al (2008)
through aptamer–small molecule interaction. 5′-Triphosphate-siRNA: turning gene silencing
RNA 12:710–716 and Rig-I activation against melanoma. Nat
10. Beisel CL, Bayer TS, Hoff KG et al (2008) Med 14:1256–1263
Model-guided design of ligand-regulated 24. Kim D-H, Longo M, Han Y et al (2004)
RNAi for programmable control of gene Interferon induction by siRNAs and ssRNAs
expression. Mol Syst Biol 4:224 synthesized by phage polymerase. Nat
11. Rybak A, Fuchs H, Smirnova L et al (2008) A Biotechnol 22:321–325
feedback loop comprising lin-28 and let-7 con- 25. Sutcliffe MJ (1993) Representing an ensemble
trols pre-let-7 maturation during neural stem- of NMR-derived protein structures by a single
cell commitment. Nat Cell Biol 10:987–993 structure. Protein Sci 2:936–944
12. Nam Y, Chen C, Gregory RI et al (2011) 26. Takeshita D, Zenno S, Lee WC et al (2007)
Molecular basis for interaction of let-7 microR- Homodimeric structure and double-stranded
NAs with Lin28. Cell 147:1080–1091 RNA cleavage activity of the C-terminal RNase
13. Beisel CL, Smolke CD (2009) Design princi- III domain of human dicer. J Mol Biol 374:
ples for riboswitch function. PLoS Comput 106–120
Biol 5:e1000363 27. Macrae IJ, Li F, Zhou K et al (2006) Structure
14. Salis HM, Mirsky EA, Voigt CA (2009) of Dicer and mechanistic implications for
Automated design of synthetic ribosome bind- RNAi. Cold Spring Harb Symp Quant Biol 71:
ing sites to control protein expression. Nat 73–80
Biotechnol 27:946–950 28. Macrae IJ, Zhou K, Li F et al (2006) Structural
15. Ogawa A (2011) Rational design of artificial basis for double-stranded RNA processing by
riboswitches based on ligand-dependent mod- Dicer. Science 311:195–198
INDEX
A Phusion...........................................81, 85, 113, 122, 133,

136, 241, 243, 244
Allosteric ribozyme. See Aptazymes PrimeSTAR ................................................................168
Antibiotics Taq........................................................ 95, 143, 212, 216
ampicillin, ..................................25, 36, 80, 111, 253, 264 Drosha ......................................................................259–266
carbenicillin ........................................................ 145, 242 Dual genetic selection............................................... 131–152
chloramphenicol ..................................... 84, 94, 111, 145 Dual genetic selection marker
kanamycin............................................................. 94, 226 herpes simplex virus thymidine
Antisense ...........................................172, 197–206, 277, 278 kinase (hsvTK)........................ 142, 143, 149–151
Apoptosis.................................................. 185, 191–193, 196 tetracycline antiporter (TetA) ............................ 126, 132,
Aptamers 133, 139, 142
photoresponsive peptide ...............................................30 Dual switch regulation .............................................193–194
protein ..................................... 41–55, 185, 187, 194, 270
small-molecule dye .................................................17–27 E
theophylline ..................... 96–99, 173, 174, 261, 262, 270
Electrophoresis
thiamine pyrophosphate (TPP) .......................... 136, 239
agarose gel .................................. 83, 85, 89, 95, 133, 169,
Aptazymes
176, 212, 228, 241
cyclic di-GMP ............................................................209
polyacrylamide gel (PAGE) ............................. 34–36, 38,
p-Amino-phenylalanine (pAF) ................... 222, 227–231
39, 45, 47–49, 54, 204, 213–215, 217, 223,
theophylline ........................................ 222, 239, 251, 252
225–228, 231–234, 242, 273, 280–282
Artificial cells. See Liposomes; Water-in-oil emulsion
Electrophoretic mobility shift assay
Autofeedback regulation...........................................194–195
(EMSA) .......................................... 44, 50, 51, 53
B Electroporator ............................................................ 95, 242
EMSA. See Electrophoretic mobility shift assay (EMSA)
BCA assay. See Bicinchoninic acid (BCA) assay Error-prone PCR. See Mutagenesis
Bicinchoninic acid (BCA) assay .........................................89 Escherichia coli (E. coli).......................................20, 21, 26, 31,
Biosensors...................................... 41, 57, 141, 168, 209, 238 36, 59, 73, 78–85, 87, 94–97, 108, 109, 112, 115,
117, 119, 126, 131–140, 142, 145, 150, 154–158,
C
184, 187, 188, 222–223, 228, 229, 231, 239, 240,
CD. See Circular dichroism (CD) 242, 245, 247, 251–257, 260, 262, 264, 273
Cell-free translation. See In vitro translation Ethanol precipitation ..................................... 43, 45–47, 157,
Cell sorters. See Flow cytometer 184, 187, 215, 218, 226–228, 244
Circular dichroism (CD) ...................198, 200–202, 204, 205
Compound library .................................... 111–113, 116–119 F
FACS. See Fluorescence activated cell sorting (FACS)
D
Fast protein liquid chromatography
Dicer.......................... 259, 262, 269–274, 276, 278–283, 285 (FPLC) ............................114, 124, 273, 280–281
Dissociation constant (Kd) ....................... 5, 20, 26, 128, 173, Filter-binding assay .......................................... 44, 50–51, 55
217, 222, 272 Florescence microscope .....................156, 159–162, 274, 284
DNA polymerase Flow cytometer .................................................... 58, 63, 145,
ExTaq .....................................................................34, 43 148, 185, 189–190, 260, 274
Pfu ...................................................................... 252, 253 Fluorescence activated cell sorting
PfuUltra ..................................................................80, 83 (FACS) ....................................... 57–74, 185, 274
Pfx ........................................................................... 20, 22 Fluorescence resonance energy transfer (FRET) ..........77–90
DOI 10.1007/978-1-62703-755-6, © Springer Science+Business Media New York 2014
287
ARTIFICIAL RIBOSWITCHES: METHODS AND PROTOCOLS
288 Index
Fluorometer. See Spectrofluorometer Phosphorimager ................ 44, 49, 50, 55, 212, 215, 226, 231
FPLC. See Fast protein liquid chromatography (FPLC) Plate reader ...................................................20, 82, 111, 133,
Free energy .................................. 5, 6, 99, 128, 174, 180, 230 145, 226, 242, 253
FRET. See Fluorescence resonance energy transfer (FRET) Polymerase chain reaction (PCR) .................................. 1, 18,
32, 43, 81, 95, 113, 133, 142, 156, 168, 184, 200,
G 212, 228, 240, 253, 263, 283
Gel purification Promoters
agarose gel ............................................ 83, 168–169, 180 CMV promoter .................................. 185–186, 262, 266
polyacrylamide gel ....................... 215, 227, 228, 242, 280 lac promoter ........................... 94, 111, 112, 115, 119, 126
Lux promoter...................................................... 142, 149
I Rha promoter .........................................................84, 88
tac promoter .................................................................81
Internal ribosomal entry site (IRES) ........ 168, 170–180, 187
T7 promoter .........................................22, 25, 30–31, 42,
In vitro compartmentalization ..................................158–160
43, 122, 156, 157, 168, 176, 180, 204, 210–211,
In vitro selection
228, 229
aptamers ......................................... 17–27, 29–39, 41–55
Trc promoter............................................................... 144
aptazymes ........................................... 209–219, 223–228
U6 promoter ....................................................... 262, 266
In vitro transcription ..................................23, 42, 45, 47, 54,
Pull-down assay ................................................ 43–44, 49–50
72, 113–114, 122, 123, 126, 127, 155–156, 158,
PURE system ................................................... 155, 158–161
169, 176, 217, 222, 225, 234, 272–273, 276,
279–280, 282 Q
In vitro translation .................................................... 158, 176
In vivo screening. See In vivo selection Quadruplex ............................................... 197–199, 201–206
In vivo selection
R
aptazymes ........................................................... 224, 225
riboswitches ............................................... 57–74, 77–90, Rational design ...................................................... 52, 93, 94,
93–104, 107–128, 131–140, 237–248 108, 141, 165–181, 240
IPTG. See Isopropyl β-D-1-thiogalactopyranoside (IPTG) RBS. See Ribosome binding site (RBS)
IRES. See Internal ribosomal entry site (IRES) Real-time PCR............................................. 20, 26, 104, 161
Isopropyl β-D-1-thiogalactopyranoside Reporter protein
(IPTG) ...............................94, 95, 101, 102, 104, chloramphenicol acetyltransferase (CAT) ..................110
112, 115, 117, 119, 120, 226, 229, 281 fluorescent protein (FP)
Isothermal titration calorimetry (ITC) ............ 110, 114–115, CFP variant ..........................................................194
123–125, 127, 128 GFP variant ............................ 78, 110, 132, 147, 157,
ITC. See Isothermal titration calorimetry (ITC) 191, 246, 261, 274
RFP variant .......................................... 191, 228, 262
L YFP variant..................................................... 78, 157
Ligand molecule. See Aptamers; Aptazymes β-galactosidase (lacZ) ............................................ 99, 154
Liposomes ................................................ 156, 160, 162–163 LacI ..............................................................................94
β-lactamase ........................................................... 81, 252
M luciferase ..................................................................... 168
RpoS .............................................................................94
Mammalian cells ....................................... 64, 183–196, 221,
tobacco etch virus (TEV) protease .........................78–79
259–261, 270, 274, 277, 283–284
Reverse transcriptase-mediated enzymatic
MicroRNA (miRNA)............................... 239, 259–266, 270
analysis (RTase stop assay) ..................... 198–201,
Mutagenesis............................................72, 80, 82, 107–128,
204, 205
143–144, 149
Reverse transcription-PCR (RT-PCR) ....................... 22, 24,
N 26, 27, 30, 34–35, 38, 39, 42, 46–48, 52, 54–55,
198, 201, 211–213, 216, 226, 228
Negative selection ............................................ 19, 32–34, 42, Ribonucleoprotein (RNP) ....................................... 184, 187,
45–46, 67, 141, 213–218, 224, 228 194, 270–272, 274–277, 279, 282, 283, 285
Riboregulators .................................................. 197–206, 240
P
Ribosome binding site (RBS) ................................ 62, 72, 73,
PCR. See Polymerase chain reaction (PCR) 79, 90, 96, 97, 100, 101, 139, 149, 151, 153–154,
Phenol-chloroform extraction ...........122–123, 157, 178, 279 156, 157, 159, 161, 165, 166, 177, 238
ARTIFICIAL RIBOSWITCHES: METHODS AND PROTOCOLS
Index
289
Riboswitches Rosetta............................................................................4
in bacteria .....................................57–74, 77–90, 93–104, siExplorer ...................................................................276
107–128, 131–140, 221–234, 237–248, 251–258 Vienna RNA ..........................................................4, 229
in mammalian cells ............................................. 183–196 Spectrofluorometer. .............................................. ....156, 161
in PURE system ................................................. 153–163 Spectrometer .........................................43, 45, 169, 176, 200
in wheat germ extract ......................................... 165–181 SPR. See Surface plasmon resonance (SPR)
RISC. See RNA induced silencing complex (RISC) Surface plasmon resonance (SPR) .................... 30, 36–37, 51
RNAi. See RNA interference (RNAi) Systematic evolution of ligands by exponential enrichment
RNA induced silencing complex (RISC) ................. 259, 262 (SELEX). See In vitro selection, aptamers
RNA interference (RNAi) ....................................... 239, 269,
271, 274–276, 284, 285 T
RNP. See Ribonucleoprotein (RNP) Target molecule. See Aptamers; Aptazymes
RT-PCR. See Reverse transcription-PCR (RT-PCR) Thermal cycler ................................ 20, 43, 95, 113, 133, 146,
169, 184, 200, 253, 283
S Three-dimensional (3D) structure......................... 3–5, 8–11,
13, 222, 272, 274, 276
Secondary structures ............................................ 2–8, 13, 17, Transfection ..............................................185, 187, 189, 191,
19, 29, 80, 99, 127, 135, 201, 203, 222, 223, 230, 193–196, 260, 262, 264–266, 274, 276, 284
234, 240, 261, 270, 276 Transformation ..................................... 25, 58, 59, 65, 68, 79,
Sequencing ....................................... 1, 18, 29, 41, 58, 78, 96, 82, 84, 87, 90, 97, 117, 119, 133, 134, 137, 140,
116, 133, 141, 153, 165, 184, 196, 210, 222, 239, 147, 150, 157, 168, 184, 187, 188, 222, 228, 229,
252, 259, 269 231, 245–248, 264, 273, 280
Short hairpin RNA (shRNA) ...................................269–285 Transilluminator ..............................20, 82, 87, 133, 140, 169
shRNA. See Short hairpin RNA (shRNA)
Software U
AutoDockTools (ADT)................................................12
UV melting analysis...........................198, 200, 202–203, 205
AutoDock Vina ............................................................12
ChemFinder ....................................................... 111, 116 W
Discovery Studio ........................................................272
OpenBabel......................................................................4 Water-in-oil emulsion ......................................................155
RNAstructure ....................................................... 96, 169 Wheat germ extract .................................. 168, 169, 176, 178

(Methods in Molecular Biology) - Artificial Riboswitches - Methods and Protocols (2014, Humana Press)

Uploaded by

Copyright:

Available Formats

(Methods in Molecular Biology) - Artificial Riboswitches - Methods and Protocols (2014, Humana Press)

Uploaded by

Document Information

Original Title

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

(Methods in Molecular Biology) - Artificial Riboswitches - Methods and Protocols (2014, Humana Press)

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What topics are covered in the book?

What topics are covered in the book?

Methods in

Molecular Biology 1111

Atsushi Ogawa Editor

For further volumes:

Methods and Protocols

ISSN 1064-3745 ISSN 1940-6029 (electronic)

Library of Congress Control Number: 2013958315

© Springer Science+Business Media New York 2014

Printed on acid-free paper

Humana Press is a brand of Springer

Matsuyama, Ehime, Japan Atsushi Ogawa

1 Computational Design of RNA Libraries for In Vitro

14 Guanine-Tethered Antisense Oligonucleotides

CHASE L. BEISEL • Department of Chemical and Biomolecular Engineering,

YE JIN • Department of Biochemistry, Li Ka Shing Faculty of Medicine, University of Hong

Computational Design of RNA Libraries for In Vitro

Aptamers can be selected in vitro using an iterative process called

and have no potential to function as binders, as the majority has a

aptamers with binding affinity to a desired small-molecule target.

Patterned libraries were generated using Perl scripts. The secondary

The proposed method for generating starting pools of RNA

The using of patterned libraries increases the probability of folding

Fig. 2 Distribution of folding energies for 10,000 randomly chosen members of

3.1.1 Generation To generate random or patterned libraries in Perl script we use

open(temp_FILE, ">Seq1.temp") || die("Could not

3.2.1 Generation The command to run RNA 3D structure modeling in Rosetta is as

Next line is the description of different score fields:

The output file Seq1_Ct.pdbqt contains coordinates of

echo Processing receptor $r

1. Our experience with the prediction of 3D structures using

5. The biggest challenge in the modeling of RNA–small molecule

In Vitro Selection of RNA Aptamers

Bacteria and some eukaryotes use natural RNA aptamers for

Fig. 1 General scheme of in vitro selection

wide range of target ligands. RNA aptamers can be isolated from

Aptamer for BHQ-1

successful. In addition, positively charged functional groups

60-base random sequence. After 13 rounds of selection against a

2.1 Equipment 1. Thermal cycler.

2.2 Reagents 1. BHQ-1 carboxylic acid, BHQ-1 amine (Bioresearch Technologies).

14. RTS 100 E. coli HY Kit, 5 (Prime Inc.).

2.3 Supplies 1. Microtubes, 1.5 and 2.0 mL.

All solutions and buffers should be made with DEPC-treated

3.1 Preparation 1. Place 0.5 mL of EAH sepharose™ 4B resin in a Poly-Prep

Template DNA*1 10 μL

3.2.2 Preparation 1. Assemble the transcription reaction at room temperature in

Template DNA X μL

4. Precipitate the RNA by adding a half volume of 7.5 M

5. Collect the eluate in a 1.5 mL collection tube, and add 43 μL

where A is the increase in fluorescence at saturating aptamer

3.5 Real-Time Detailed protocol of the preparation of template DNAs is described

1. We use DMT-MM as a coupling reagent, but any other water-­

We are indebted to Professor M. Uesugi of Kyoto University for

Development of Photoswitchable RNA

Riboswitches are functional RNA structures usually positioned at 5′

1. We use DMT-MM as a coupling reagent, but any other water-