Dunn 2008

Home Search Collections Journals About Contact us My IOPscience
Current trends and future requirements for the mass spectrometric investigation of microbial,
mammalian and plant metabolomes
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2008 Phys. Biol. 5 011001
(http://iopscience.iop.org/1478-3975/5/1/011001)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 130.220.71.22
The article was downloaded on 23/09/2012 at 08:45
Please note that terms and conditions apply.

IOP PUBLISHING PHYSICAL BIOLOGY
Phys. Biol. 5 (2008) 011001 (24pp) doi:10.1088/1478-3975/5/1/011001
REVIEW ARTICLE
Current trends and future requirements

for the mass spectrometric investigation
of microbial, mammalian and plant
metabolomes
Warwick B Dunn
Manchester Centre for Integrative Systems Biology, School of Chemistry, University of Manchester,
Manchester Interdisciplinary Biocentre, 131 Princess Street, Manchester, M1 7DN, UK
E-mail: warwick.dunn@manchester.ac.uk
Received 19 December 2007

Accepted for publication 22 January 2008
Published 20 February 2008
Online at stacks.iop.org/PhysBio/5/011001
Abstract
The functional levels of biological cells or organisms can be separated into the genome,
transcriptome, proteome and metabolome. Of these the metabolome offers specific advantages
to the investigation of the phenotype of biological systems. The investigation of the
metabolome (metabolomics) has only recently appeared as a mainstream scientific discipline
and is currently developing rapidly for the study of microbial, plant and mammalian
metabolomes. The metabolome pipeline or workflow encompasses the processes of sample
collection and preparation, collection of analytical data, raw data pre-processing, data analysis
and data storage. Of these processes the collection of analytical data will be discussed in this
review with specific interest shown in the application of mass spectrometry in the
metabolomics pipeline. The current developments in mass spectrometry platforms (GC–MS,
LC–MS, DIMS and imaging MS) and applications of specific interest will be highlighted. The
current limitations of these platforms and applications will be discussed with areas requiring
further development also highlighted. These include the detectable coverage of the
metabolome, the identification of metabolites and the process of converting raw data to
biological knowledge.
Introduction of biological systems, and the interactions of these different

levels and as such have provided insight into the functional
‘The process of scientific discovery is, in effect, a continual importance of genes. These studies allow an understanding
flight from wonder’. Albert Einstein (1879–1955) of the phenotype and genotype–phenotype relationships
Great advances have been observed in the biological and provide the capability to employ mathematical and
sciences during the previous 150 years. The sequencing of experimental descriptions to model the complex operation of
numerous genomes (see http://www.genomenewsnetwork.org biological systems (systems biology [1–3]). Of these different
for further information) in the previous two decades has functional levels, the metabolome has been employed in the
instigated great advances in the biological sciences, but investigation of microbial [4–6], plant [7–9], environmental
highlighted our limited understanding regarding the ‘building [10, 11] and mammalian systems. The latter system has been
blocks’ of biological systems. The post-genomic era studied for investigations of health and nutrition, [12–15] and
has progressed by focusing on the study of the different for the detection of biomarkers indicative of disease, drug
functional levels (transcriptome, proteome and metabolome) toxicity or drug efficacy [16–21].
1478-3975/08/011001+24$30.00 1 © 2008 IOP Publishing Ltd Printed in the UK

Phys. Biol. 5 (2008) 011001 Review Article
Innovations in analytical technologies have preceded the final downstream product of the genome, the metabolome is
many advances in scientific knowledge [22]. The invention closest to the function or phenotype of the cell. The application
of the mass spectrometer over 100 years ago heralded new of metabolic control analysis (MCA) [30] demonstrates that
capabilities to determine atomic and molecular masses and changes in concentrations (though not necessarily fluxes)
relative abundances of elemental isotopes [23]. More recently, of metabolites can be observed even when alterations in
Sanger’s development of automated gene sequencers [24] has the concentrations of proteins or transcripts are small or
enabled the sequencing of more than 180 genomes. The data- not detectable. Therefore the metabolome is a highly
driven strategies applied in post-genomic sciences require the discriminatory system to study changes in biology, which
observation of hundreds or thousands of transcripts, proteins has been shown experimentally [31, 32]. Finally, the
(or associated peptides) and metabolites in any single study. metabolome is a high-throughput strategy as costs per analysis
Mass spectrometry is an analytical platform that has played a are low compared to proteomics and transcriptomics [33–35].
key role in the progression of proteomics and metabolomics However, as with transciptomic and proteomic technologies,
into mainstream science providing multi-analyte detection the economic investment to purchase the high specification
with high sensitivities and specificities when combined with analytical instruments employed is high.
high-resolution chromatographic systems. In metabolomics, The study of the complete metabolome, sometimes
mass spectrometry is applied in many studies. In the referred to as metabolomics [35–38], is not achievable because
period January 1998 to November 2007 inclusively, 503 of of its complexity and size and instead a number of different
1360 publications (as defined as present in title, keyword strategies are employed as shown in table 1 [34, 35, 38, 39].
or abstract) cited in the ISI Web of Knowledge employed Definitions and their application within the metabolomics
mass spectrometry as the single analytical platform or in community are currently not clearly defined, are evolving and
combination with other platforms. can differ from group to group. A typical example is the
In this review, the role of mass spectrometry platforms interchange of the terms metabolomics and metabonomics.
including chromatography–mass spectrometry (XC–MS), Metabolites contained within the cell (endometabolome)
direct infusion mass spectrometry (DIMS) and imaging MS or alternatively those metabolites in the extracellular
will be described with relevant and recent illustrations of environment (exometabolome) are studied, either in situ
their application. The advantages and limitations of different (imaging) or more typically after sampling and extraction
mass spectrometry platforms for metabolomic applications procedures. The exometabolome has otherwise been described
will be highlighted with specific examples and some of the as the metabolic footprint [40], as this metabolome reflects the
requirements for future developments will be discussed. influence of the intracellular metabolic network on its external
environment, by the uptake of extracellular metabolites and
An overview of metabolomics secretion of intracellular metabolites. The metabolomes’
complexity is highlighted in the large variability of chemical
Metabolomics is focused on the holistic and data-driven and physical properties [38, 41] and the wide concentration
study of the metabolism of both endogenous and exogenous ranges to be detected (molar to sub pM). The study of the
metabolites present in biological systems. Metabolism is intracellular metabolome is technically demanding because
derived from the Greek word metabolè, meaning change, metabolism is in a rapid and constant flux and therefore
which is appropriate as metabolism operates with a constant any sampling strategy requires the rapid arrest of metabolic
and rapid flux. The metabolome is defined as the total activity, described as quenching, typically using a rapid
quantitative collection of low molecular weight compounds increase or decrease in temperature to halt enzyme activity.
(metabolites) present in cell or organism which participate An automated system to rapidly sample (less than 1 s)
in metabolic reactions required for growth, maintenance and microbial cultures has recently been introduced and minimizes
normal function [25, 26]. In reality, those metabolites the variability and time constraints of manual sampling [42].
consumed from the external environment (food nutrients, The subsequent release of metabolites into a suitable medium
drugs, alcohol) and from co-habiting organisms such as gut for analysis employs heat or cold and occasionally acid or
microflora should also be included in this definition. base to facilitate cell lysis. When appropriately applied this
Being composed of many hundreds or thousands of small provides a snapshot on metabolism [43–45]. Sampling of
molecular weight organic and inorganic species [27, 28], the metabolic footprint is less technically demanding and
generally of mass less than 1500 Da, the metabolome was more appropriate for high-throughput studies, providing a
initially defined in the late 1990s [25, 29]. Metabolites picture of metabolism over a cumulative time period. Both
can be of endogenous origin, biochemically synthesized or serum/plasma and urine can be described as metabolic
catabolized within the cell or organism, or be of exogenous footprints and provide a cumulative picture of mammalian
derivation as is the case for pharmaceuticals and food nutrients cellular metabolism for the complete organism. Once the
consumed by humans. The study of the metabolome, metabolites are extracted from the biological system there
compared to transcriptome and proteome, offers some distinct are a range of analytical platforms that can be applied
advantages. The metabolome is more tractable, generally a in metabolomics [38]. These include chromatography–
ten-fold difference in the number of metabolites compared to mass spectrometry platforms [gas chromatography (GC),
genes [for example, S. cerevisiae 584 metabolites [28] and liquid chromatography (LC) and capillary electrophoresis
more than 6600 genes (see http://www.yeastgenome.org)]. As (CE) interfaced to mass spectrometers], direct infusion mass
2
Table 1. Terms commonly employed in metabolomics and related disciplines.

(a) Metabolomics Initially defined as the non-biased quantification and identification of all metabolites present in a biological
system. The term is now employed to define the study of the metabolome using one or more of the strategies
defined below.
(b) Metabolic profiling Detection of a wide range of metabolites, generally related by metabolite class(es), employing single
or multiple analytical platforms so as to obtain as wide a detectable coverage of the metabolome as is
achievable. This strategy can also be described as metabolite profiling or untargeted analysis and provides
metabolite identifications where possible. Relative changes in response (correlated to changes in metabolite
concentration) are used to define metabolic differences. Metabolic profiling is normally applied in an inductive
experimental strategy [50] where the metabolites of biological interest are not known a priori.
(c) Metabolic fingerprinting High-throughput collection of a global snapshot, or fingerprint, of a crude sample with minimal sample
preparation. Identification and quantification is limited and the strategy is employed as a tool for discrimination
of samples from different biological origins or status. Also defined as the study of the intracellular metabolome.
(d) Targeted analyses Absolute quantification and identification of a single or highly-related small set of metabolites after significant
sample preparation to separate metabolites from sample matrix.
(e) Metabolic footprinting Analysis of the extracellular metabolome of an organism, composed of metabolites not consumed from
the organism’s environment and those metabolites secreted from the intracellular volume. In the example of
microbial systems the organism’s environment is the growth medium. The strategy is high-throughput as there
is no requirement for quenching of metabolism and extraction as is the case for intracellular metabolomes.
(f) Metabonomics The quantitative measurement of the dynamic multiparametric metabolic response of living systems to
pathophysiological stimuli or genetic modification [192].
spectrometry (DIMS), Fourier transform-infrared (FT-IR) important [55]. Ensuring all possibilities of bias are eliminated
and Raman spectroscopies and nuclear magnetic resonance or described is also essential [56, 57] but are not always
spectroscopy (NMR). No single analytical platform can observed (for example, differences in the distribution of ages
provide the detection of all metabolites and hence a multi- between two sample classes [58]). A number of reporting
platform approach is advantageous so as to provide maximal standards are currently available and from interactions with
coverage of the metabolome [46–49]. the community are subject to revision. These standards have
Metabolomics is a data-driven, or inductive, scientific been developed by the metabolomics standards initiative (MSI)
discipline compared to the hypothesis-driven strategies to enable other researchers to document research accurately
employed in traditional molecular biology [50]. Many studies (see [59] and nine other articles in the same issue). These
originate from an area of minimal biological knowledge. For are not guidelines on how to perform investigations but rather
example, in disease biomarker investigations in which the a common mechanism to describe experimental work so as
availability of aetiological information is limited, detectable to allow others to evaluate or repeat the work, similar to
differences in the phenotype can be expressed at multiple the reporting standards applied in transcriptomics [60] and
sites within the metabolic network [51, 52] or biological proteomics [61].
system [53]. To obtain a greater biological understanding the
experiment should be designed in a valid and robust manner
so as to obtain a large volume of reproducible metabolic Mass spectrometry platforms applied in
information, generally employing metabolic profiling. In metabolomics
many cases, this is a search for the proverbial needle(s) in the
metabolic haystack. Interrogation of the avalanches of data Mass spectrometry is a mature and well-developed analytical
produced can identify those metabolites that are differentially platform applied in many scientific disciplines. In simple
present in diverse sample classes (for example, healthy and terms the platform is involved in the measurement of the
diseased humans or a gene mutant and associated wild type mass-to-charge ratio of elemental or molecular species via
for eukaryotes). Once metabolites of biological interest are an experimental workflow. The mass spectrometer has
deduced, further targeted experiments can be employed to test operated in a similar workflow since its invention in 1912,
and further refine biological knowledge. as described in figure 1, though additions to the toolbox
The metabolomics experiment is operated by a are regularly observed. For example, the development of
multidisciplinary team of scientists in a workflow or pipeline, electrospray ionization and matrix-assisted laser desorption
from sample collection and preparation through analytical ionization (MALDI) has provided great impetus to higher-
operations to processing of raw data and interrogation of value biological experiments [62–64]. In metabolomics the
processed data [54]. The design of the metabolomics development of, or improvement in, mass analysers, including
experiment is critical to produce valid and reproducible time of flight (TOF) [65, 66], Fourier transform ion cyclotron
data from which biological knowledge can be deduced. All resonance (FTICR) [67] and Orbitrap [68], has provided
steps of the pipeline should be appropriately validated and the ability to acquire more accurate and specific biological
issues including sample numbers, sample preparation, choice information at greater volumes in relatively high-throughput
of analytical technology and data processing strategies are experiments.
3
GAS +
+
LIQUID +
SOLID
SAMPLE MASS DETECTION

ION SOURCE PC
INTRODUCTION ANALYSER SYSTEM
Figure 1. Schematic of mass spectrometer operation. Samples are introduced via an inlet system (including chromatographic systems) to an
ion source operating at atmospheric pressures (ESI, APCI) or vacuum pressures (EI, CI) where positively or negatively charged ions are
created. Ions are subsequently separated according to their mass-to-charge (m/z) ratio in the mass analyser prior to detection either
physically at a detector or as an image current of ion orbital frequencies. A PC is used to acquire data and also to control all instrument
parameters.
Mass spectrometers operate in a four-step process as [41, 69, 70]. A detailed discussion is beyond the scope of
shown in figure 1 [69]. Following sample introduction ions this review and instead will focus on specific ion sources and
are generated in an ion source, either operating at atmospheric mass analysers which have revolutionized metabolomics, and
or vacuum pressures. The creation of charged species is science in general, many of which are described in table 2. The
necessary to enable ion manipulation, a requirement for the introduction of the electrospray ionization source, originally
separation of ions according to mass-to-charge (m/z) ratio. described in the 1960s [62, 64] and commercially available
Subsequently separation, in space or time, of ions based on in the 1990s, has advanced many scientific disciplines.
their m/z ratio in a mass analyser occurs followed by the Its availability along with the complementary technique of
detection of ions either physically at a detector as an ion current atmospheric pressure chemical ionization (APCI) enables the
or by the detection of orbital frequencies as image currents. sampling of high liquid solvent flow rates (10–1000 µl min−1)
The range of ion sources and mass analysers currently available at atmospheric pressures from LC–MS applications with high
is detailed in table 2 and further information can be found sensitivity and without the requirement of solvent vapour
elsewhere [41, 69, 70]. Further information on the advantages overloading the vacuum system of the mass spectrometer.
and limitations of a number of mass analysers will be described Prior to the routine application of electrospray ionization
in this review. sources, other methods were available but were of limited
Mass spectrometry provides a number of advantages routine application [71].
for its application in metabolomic investigations [41, 70].
The platform is sensitive allowing the detection of µM
concentrations in typical metabolic profiling applications. Developments in mass analyser technologies
This allows the detection of many metabolite classes at Three mass analysers have provided significant advantages
physiological concentrations including amino and organic in metabolomics, compared to quadrupole and 3D ion traps.
acids, fatty acids, sugars, sugar phosphates, bile acids, lipids
The time of flight (TOF) mass analyser detects all ions
and nucleotide bases. A second advantage is the ability
simultaneously, rather than scanning mass ranges as is the case
to identify metabolites either through the measurement of
for many quadrupole instruments, and hence detects all ions
molecular mass (indicative of the molecular formula) to
continuously providing a greater sensitivity. However, the
a high mass accuracy (typically parts per million, ppm)
recent commercial introduction of fast scanning quadrupole
or by collection of fragmentation mass spectra (indicative
instruments with an improved sensitivity has provided results
of molecular structure). This allows the identification of
metabolites not currently described in metabolomic databases comparable with TOF instruments with respect to sensitivity
or enables more confident identifications. The characterization and scan/acquisition time. TOF mass analysers can acquire
of novel metabolites and reproducible identification of data at rates greater than 100 Hz, though generally higher
previously characterized metabolites across large sample acquisition rates correlate with a reduced sensitivity. High
sets is an important experimental requirement so as to acquisition rates allow suitable sampling of data points across
convert analytical data to biological information. Currently chromatographic peaks (>30 points is appropriate). Gas
many metabolites detected in metabolic profiling studies chromatographic peaks of typical width at a baseline of 3 s
of microbial, plant and mammalian systems have not been when detected with an acquisition rate of 10 Hz are suitable and
characterized and this is a major limitation in metabolomics. this allows deconvolution software to operate appropriately for
The interfacing of chromatographic systems with mass closely eluting peaks. In GC–MS generally four or more data
spectrometers increases the biological information obtained, points should separate peak apexes so as to be detected as
and spatial introduction of metabolites to mass spectrometry separate metabolite peaks by deconvolution software. TOF
platforms enhances sensitivity and the ability to identify mass analysers also allow a route to metabolite identification
metabolites in these complex biological systems. by the application of accurate mass measurements of the
A wide array of different sample introduction systems, molecular ion, with typical mass accuracies of less than 5 ppm
ion sources and mass analysers are employed in metabolomics [49, 72–74]. This is suitable for preliminary identifications
4
Table 2. Summary of the range of ion sources and mass analysers currently employed in metabolomic investigations, including a brief
description of their operation.
Ion sources
Atmospheric Employed in LC–MS applications for ionization of polar and semi-polar metabolites at atmospheric pressures.
pressure chemical Operates by passing the LC eluent through a heated glass tube (up to 500 ◦ C) which produces explosive vaporization
ionization (APCI) to generate a gaseous collection of solvent and metabolite molecules. A corona discharge needle provides ionization
of the solvent molecules (which are in a large excess) followed by ion or charge transfer to metabolite molecules.
Degradation of molecules can occur at the high temperatures used, though minimal fragmentation of molecular ions
is observed. The technique is described as soft-ionization.
Atmosphereic Employed in LC–MS applications for ionization of non-polar metabolites at atmospheric pressures and acts as a
pressure photo complementary tool to ESI and APCI. Not commonly available from many instrument manufacturers. Electron
ionization (APPI) ejection from molecules is produced by photons emitted from discharge UV lamps.
Chemical Employed in GC–MS applications for ionization of polar and non-polar metabolites at vacuum pressures. Chemical
ionization (CI) reagent gases (methane, ammonia and others) are introduced into an EI source, in large excess to metabolite
molecules, and are ionized by electron bombardment. Ion or charge transfer creates ionization of metabolites. This
is a soft-ionization technique producing minimal fragmentation of the molecular ion, tandem mass spectrometry
can be used to dissociate the molecular ion.
Electron impact Employed in GC–MS applications for ionization of polar and non-polar metabolites at vacuum pressures.
(EI) Bombardment of the metabolite molecules by electrons creates positively charged ions. The process imparts
significant energy to the molecular ion which is released by covalent bond fission to produce a mass spectrum which
is characteristic of the metabolite’s chemical structure. Fragmentation patterns are highly reproducible and can
be used to deduce metabolite structure. Libraries of these mass spectra are also available for the identification of
metabolites.
Electrospray Operates by passing the LC eluent through a capillary held at high voltages (2–5 kV). Ion formation occurs in
ionization (ESI) the capillary followed by nebulization and desolvation to provide transfer of ions from liquid to gas phase and
introduction to the mass spectrometer vacuum. The source acts as an electrochemical cell. Minimal fragmentation
of the molecular ion is observed, fragmentation can be produced in-source or by tandem mass spectrometry. The
technique is described as soft-ionization.
Mass analysers
Fourier transform Expensive mass analyser that attains the highest mass resolution (R > 100 000) and mass accuracy (less than 1
ion cyclotron ppm) currently available. Ions orbit in a cell operating at ultra-high vacuums (10−10 atmospheres) and in a high
resonance (FTICR) magnetic field strength (>7 Tesla). The orbital frequency, dependent on m/z, is detected as an image current and
is converted from time to frequency domains with a Fourier transform.
Linear A low mass resolution and cheap mass analyser. DC and RF potentials are applied to four parallel rods, each
quadrupole (Q) opposite pair being electrically connected. These potentials are varied so as to provide a mass filter where ions
of a chosen mass have trajectories of amplitude less than half the radius of the quadrupoles and traverse the mass
analyser whereas ions of lower and higher mass are lost by collisions with the rods because their amplitudes are too
great. The DC and RF potentials are varied (while keeping their ratio constant) so as to provide a mass scan where
ions of increasing mass traverse the mass analyser.
Orbitrap A new introduction which operates in a similar manner to FTICR, though without the requirement of high magnetic
strength fields and at a lower purchase cost. Another high mass resolution (up to 100 000) and mass accuracy
(<1 ppm) instrument. A coaxial inner spindle-like electrode is surrounded by two outer barrel-like electrodes with
an electric field applied between electrodes. Ions orbit the central electrode in axial and radial directions, through a
balance of electrostatic and centrifugal forces and the orbital frequency is detected as an image current by the outer
electrodes. A Fourier transform is employed to convert it from the time to frequency domain. The instrument is
marketed as a hybrid instrument, a linear ion trap is coupled to the Orbitrap so as to collect ions before periodic
introduction into the Orbitrap mass analyser.
Quadrupole ion A low mass resolution analyser that provides ion storage which allows MSn experiments to be performed to provide
trap (QIT) and structural information. The QIT is constructed of a ring and two cap electrodes and operates with the application
linear ion trap of DC and RF potentials to constrict ions in stable oscillatory orbits. A helium bath gas is used to stabilize ion
(LIT) populations. Detection occurs by destabilization of orbits and ion ejection to a detector. Linear ion traps apply a
2D quadrupole field rather than 3D field as applied in QIT. These are physically larger and allow trapping of larger
ion populations providing greater sensitivity.
Quadrupole-time A hybrid instrument allowing tandem mass spectrometry experiments with detection of product ions performed at
of flight (Q-TOF) high mass resolutions and accuracies. Constructed of quadrupole and TOF mass analysers separated by a higher
pressure collision cell used to provide collisionally induced dissociation (CID) of selected ions.
Time of flight A simple system which measures the flight time from source to detector in a vacuum, which is dependent on m/z,
(TOF) lower masses are detected first. The use of ion mirrors (reflectrons) focuses ions of the same m/z but different
kinetic energies which increases mass resolution (4000–15 000 FWHM) and allows accurate mass measurements.
Triple quadrupole Mass analyser commonly applied for MS/MS experiments for structural characterization or selective and sensitive
(QQQ) targeted analysis of limited number of metabolites. Two quadrupoles (Q1 and Q3) are separated by a quadrupole
(collision cell) operating at a higher pressure. Ions accelerated from Q1 to collision cell undergo collision-induced
dissociation (CID) followed by mass analysis in Q3. A range of MS/MS experiments are possible including product
ion scanning, neutral loss scanning and single/multiple reaction monitoring (SRM/MRM).
5
(a)
(b)
1 1 .5 0
100
1 .4 2
95
90
85
0 .7 1
80
7 .7 1
75
7 .2 2
70 5 .5 8
65
60 5 .3 5
55
Relative Abundance
1 2 .2 6
50 4 .0 6
45 1 1 .1 6
40
2 0 .8 6
35
3 .8 3
2 0 .8 1
30
1 2 .7 3 1 4 .8 8
25 1 5 .1 4
1 0 .8 6 1 5 .5 3
20
6 .8 1 8 .9 9 1 5 .6 8
15 2 .8 3 1 4 .7 7
9 .9 6 1 4 .4 3
10
1 6 .6 7
5 1 9 .7 8 2 0 .4 4
0
0 2 4 6 8 10 12 14 16 18 20
Time (min)
Figure 2. Graphical representation of GC–MS. LC–MS and DIMS data. (A) GC–MS single ion chromatogram for serum depicting the
retention time (abscissa) and ion intensity (ordinate) for m/z 73, chosen as a characteristic ion observed in the mass spectrum of
trimethylsilyl-derivatized metabolites. The chromatogram shows a high chromatographic resolution with more than 250 metabolite peaks
detected. (B) UPLC–MS total ion chromatogram for serum in negative ion mode depicting the retention time (abscissa) and ion intensity
(ordinate). The chromatogram also shows the high chromatographic resolution achievable with sub-2µm-particle LC systems with more
than 700 metabolite peaks detected. (C) DIMS mass spectrum for serum depicting m/z ratio (abscissa) and ion intensity (ordinate). The
mass spectrum is information rich with more than 500 separate ions, attributable to metabolites, detected. The mass spectrum shows the
wide range of metabolite masses in the metabolome.
though sub-parts per million (ppm) mass accuracies are using accurate mass measurements alone, chromatographic
not even adequate for unique identification of a molecular separation is also required.
formula [75]. Multiple structural isomers all have the same The hybrid quadrupole-time-of-flight (Q-TOF)
molecular formula as is observed for hexose sugars including instrument provides two routes to identify metabolites,
glucose and fructose and these isomers cannot be distinguished accurate mass measurements of the molecular ion and tandem
6
Figure 2. (Continued.)
mass spectrometry (MS/MS) to provide accurate mass spindle-shaped central electrode at ultra-high vacuums though
measurements of the precursor (molecular) and product ions. without the requirement of high magnetic fields. Instead the
This can be performed routinely in a process referred to as application of electric fields and centrifugal forces is used
MSE, where one experiment collects the molecular ion data to confine ion orbital frequencies and an image current is
followed in the next experiment with the collection of the detected by two bell-shaped outer electrodes. Ions orbit around
accurate masses of product ions [73]. This strategy normally the central electrode with stable trajectories and also along
employs data-dependent experiments where the precursor ion the central electrode with harmonic oscillations and it is the
for MS/MS analysis is chosen during the experiment in an frequency of these oscillations which define the m/z ratio
automated manner with no human intervention and therefore of ions. Mass resolutions of 100 000 are achievable, though
no de novo knowledge of which metabolites are expected to scan time is positively correlated to mass resolution, higher
be present is required. Stitching these data together requires resolutions require a longer scan time. For example, a mass
powerful software. The ability of the TOF mass analyser resolution of 30 000 requires a scan time of 0.4 s, not dissimilar
to acquire all data simultaneously allows no requirement to those employed for TOF instruments in many metabolomics
for de novo knowledge of the metabolites to be detected, as applications, though with a higher mass resolution achievable
would be required for triple quadrupole (QQQ) instruments with the Orbitrap instrument. Sub-ppm mass accuracy is
employing multiple reaction monitoring (MRM) experiments also readily achievable, and sub-3ppm mass accuracies are
in tandem mass spectrometry [76]. Further developments of generally observed over wider concentration ranges than for
Q-TOF instruments to increase mass accuracy, sensitivity and TOF mass analysers whose mass accuracy is highly dependent
dynamic range have been reported recently [77] and these on suitable ion populations and therefore the concentration
may increase their application in metabolomics as a cheaper of the metabolite. The mass accuracy for the Orbitrap
alternative to those higher mass accuracy and mass resolution is concentration independent and this provides a distinct
instruments described in the next paragraph. advantage over TOF instruments. In the author’s experience
Fourier transform ion cyclotron resonance (FTICR) mass the external mass calibration is stable over a number of days.
spectrometers provide the highest achievable mass resolution For both TOF and Orbitrap instruments mass accuracies can
and mass accuracy. Sub-ppm mass measurement accuracy be increased with internal mass calibration. Here single or
and mass resolutions greater than 100 000 allow, in theory, the multiple internal mass calibrants are continuously monitored
detection and identification of a greater number of metabolites and used to compensate for small and short-term drifts of
[78]. These mass analysers operate at ultra-high vacuum the external mass calibration. These mass calibrants can
(<10−9 atmospheres) and determine the orbital frequency be present as background ions or introduced via a separate
of ions perpendicular to a magnetic field of high strength electrospray source (for example, Waters Lockspray) and the
(generally greater than 7 Tesla). The time-domain data are application of a number of masses to extend over the large
converted to the frequency domain by the application of mass ranges of interest is recommended. The commercial
Fourier transform algorithms. Included in this class of mass Orbitrap instrument is coupled with a linear ion trap to produce
spectrometers is the first novel mass analyser introduced to the hybrid LTQ-Orbitrap, two mass spectrometers interfaced,
the scientific community for 25 years, the Orbitrap [68]. which allows broader functionality. As is observed with the
The instrument again performs mass determination by the Q-TOF operated in MSE mode [73], information regarding
detection of orbital frequencies, in this instrument ions orbit a molecular and product ion spectra can be obtained. However,
7
multiple scans can be obtained simultaneously rather than columns, the study of highly complex samples as shown in
sequentially and hence a higher sampling rate can be achieved, figure 2(a).
whilst maintaining sensitivity, across chromatographic peaks. Gas chromatography allows the direct investigation
Accurate mass measurements of the molecular ions are of volatile metabolites or after appropriate chemical
acquired in the Orbitrap in parallel to MS/MS mass spectra derivatization the analysis of semi-volatile metabolites. As
being collected in the linear ion trap. Obtaining accurate a general rule metabolites of molecular mass less than 350 Da
mass measurement for MS/MS spectra requires sequential are detectable. As examples, cholesterol, C2–C24 fatty acids
acquisitions in the Orbitrap and a longer experiment time, and disaccharides can all be detected. Sufficient volatility
which is not always suitable for chromatography with peak is required so as to produce a gaseous sample that can
widths of between 3 and 10 s. Two mass resolutions can traverse a GC column at temperatures of up to 320 ◦ C. Lower
be employed, high (30 000) for accurate mass determination volatility compounds with insufficient volatility will collect in
of the complete mass spectrum and low (7500) for accurate the injector or on the top of the GC column.
mass measurements of the product ions in MS/MS mode. A range of volatile metabolites have been studied directly
These mass resolutions combined provide 20 data points by GC–MS [86–91]. Other metabolites classes of semi-
across chromatographic peaks with baseline widths of 10 s. volatility are detectable after further sample preparation
In the authors viewpoint, the current benchmark in LC–MS processes to increase volatility. Chemical derivatization,
for metabolomics is the combination of the LTQ-Orbitrap generally a two-stage process for polar metabolites [92–94]
with liquid chromatography systems operating with sub-2µm and a one-stage process for non-polar metabolites [84], is
particles. commonly applied. For polar metabolites the chemical
transformation of oximation is performed to convert carbonyl
Mass spectrometry platforms in metabolomics moieties to an oxime as carbonyl groups react slowly and
in a reversible manner with trimethylsilyl reagents described
Five different, and complementary, mass spectrometry in step 2. Subsequently, the replacement of exchangeable
platforms are employed in metabolomics investigations hydrogens with trimethylsilyl groups reduces intra- and inter-
[38, 70, 79]. The combination of chromatographic separations molecular hydrogen bonding and increases volatility [93].
with mass spectrometry detection is a powerful tool for These processes allow the detection of many metabolite
detection, quantification and identification of a wide array of classes including amino and organic acids, fatty acids
metabolites, though at the expense of analysis time, typically and some lipids, sugars, sugar alcohols and phosphates,
10–60 min. Alternatively, direct infusion mass spectrometry amines, amides and thiol containing metabolites. However,
(DIMS) enables rapid and high-throughput screening of many this extra processing step can introduce greater technical
hundreds of samples as a method to classify samples into variability and complexity to the data, a single metabolite
separate groupings, though with limited quantification and can produce multiple derivatized metabolite peaks. This
metabolite identification capabilities. Recently, the in situ process is generally a manual process and associated errors
and spatial imaging of biological tissues and cellular systems and increased variabilities are observed, and sample stability
has been observed. Each of these strategies will be discussed can be an issue. As an alternative automated systems
below. Capillary electrophoresis is employed in the analysis of are becoming commercially available, using GERSTEL
microbial and plant metabolomes though will not be discussed autosampler systems (www.anatune.co.uk). These provide
in further detail here [80–83]. ‘just-in-time’ sample preparation for samples shown to provide
instability during longer time periods between the completion
(a) Gas chromatography–mass spectrometry (GC–MS) and of derivatization and analysis [95]. The presence of water
comprehensive GCxGC-MS. is detrimental to GC–MS analyses influencing the stability
Gas chromatography and its interfacing with mass of trimethylsilyl-derivatized samples (trimethylsilylation is
spectrometers is a well-developed and robust tool applied a reversible esterification reaction) and also degrading
in analytical sciences [84]. The first applications of GC– instrument performance. Therefore samples are dried using a
MS were described in the 1950s and 1960s [65, 85], flow of nitrogen gas and heat, though lyophilization (drying
employing TOF mass analysers. The platform has under vacuum and reduced heat) is recommended to provide
significantly developed in the following 50 years. Columns greater sample stability. Lyophilized samples stored for a
have evolved from wide bore packed columns to hollow number of weeks before analysis should be lyophilized again
capillary columns with higher chromatographic resolving as these samples can absorb moisture during storage.
power. These columns require lower gas flow rates and from In most metabolomic applications, gas chromatography
which all of the gas flow can be introduced directly to the operates by introduction of a liquid sample into a heated
mass spectrometer, providing higher sensitivities. Modern injector, where rapid vaporization and mixing with the carrier
GC columns operate at higher temperatures and provide gas occurs, followed by spatial separation of metabolites on
significantly higher chromatographic resolution. The mass the GC column and subsequent detection [41, 70, 96]. An
spectrometer as a detection system has also improved in aliquot of, or all of, the vaporized sample is introduced onto
its capabilities. Increased detection efficiency (including the top of the chromatography column. The temperature of the
instrument electronics) and the development of mass analysers column is regulated by its location in a temperature-controlled
have enabled, combined with the introduction of capillary oven, with an initial temperature more than 30 ◦ C lower than
8
the boiling point of the solvent. This allows refocusing of GC–MS has been one of the major analytical drivers
the sample at the top of the column by condensation with the in the early developments of metabolomics. Thirty years
solvent and improved chromatographic peak shape and width. before definitions of the metabolome and metabolomics were
Gas chromatography columns are silica capillaries of reported, pioneering work by Horning and Horning employed
length 10–60 m and internal diameters of 100–500 µm, GC–MS to obtain multi-component metabolic profiles of
coated externally with an imide layer to reduce column mammalian fluids and described its applicability to define
fragility and internally with a liquid stationary phase of normal and different pathological states [98]. Further work
thickness 10–50 µm. The stationary phase is of siloxane was described by Linus Pauling who employed GC–MS
or similar chemical composition with varying percentages of to study the headspace vapour of urine and breath [99]
different chemical moieties. ‘DB1’ columns contain 100% and Sauter at BASF used the platform to obtain metabolic
methyl groups, ‘DB5’ columns contain a 95/5 mix of methyl profiles of plants suitable to define the mode of action
and phenyl groups and ‘DB17’ columns contain a 50/50 mix of herbicides [100]. However, the major development of
of methyl/phenyl groups [84]. Typically, DB5 or DB17 metabolomics has been observed since the year 2000 with
stationary phases are employed in metabolomics. two pioneering publications from the plant metabolomics
Spatial separations of metabolites occurs by the community, only possible because of advances in GC–MS
equilibration of metabolite molecules between the gas platforms and instrumentation [92, 101]. These employed
phase and the liquid stationary phase with separations quadrupole instruments to analyse polar extracts of potatoes,
generally optimized through the application of appropriate with the detection of over 300 metabolite peaks and with
oven temperatures and stationary phase composition. For later developments in instrumentation and deconvolution
metabolomics, where the boiling points of derivatized software increased the number of peaks detected to greater
metabolites are diverse, a temperature program is employed than 1000 [102]. Although instrumentation has improved
to optimize chromatographic separations starting from since the early work of Horning to increase sensitivity and
temperatures lower than the solvent boiling point (typically chromatographic resolution, the introduction of deconvolution
pyridine) and rising to temperatures of 300–320 ◦ C. The software was also required to allow reporting of these
temperature limit is defined by the stability of the stationary numbers of peaks present in these complex samples. From
phases and metabolites at elevated temperatures. Further these early developments, the technique has become highly
detailed information regarding gas chromatography and the developed because of high sensitivity, high chromatographic
two strategies for ion formation (electron impact and chemical resolution, wide range of detectable metabolite classes and
ionization) are available elsewhere [41, 69, 84]. The advantage the ability to identify metabolites through the production of
of electron impact ionization (EI) is that of reproducible mass spectral/retention index libraries or by comparison to
fragmentation of the molecular ion which can be employed commercially available libraries [103, 104]. Certainly the
for the identification or characterization of a metabolite. production of metabolite-specific libraries have provided a
However, trimethylsilyl-derivatized metabolites, commonly greater ability to identify metabolites, not yet achievable
employed in metabolomics, show a weak molecular ion but in LC–MS applications. Since the early developments and
a stronger M-15 ion characteristic of the loss of a methyl validation of the analytical platform and its application a wide
group (a stable losing group) from the molecular ion, and range of applications have been observed in microbial, plant
hence defining the molecular ion and its mass can be difficult in and mammalian metabolomics. A range of these are described
identification processes. The process of ionization is relatively in table 3 though because of the higher numbers of applications
universal and is not dependent on metabolite class. As an not all examples can be detailed. A number of applications
alternative, chemical ionization (CI) can be applied, which is are worthy of further discussions.
a ‘soft’ ionization tool producing minimal fragmentation of the Following initial developments in the 1960s and 1970s,
molecular ion [72]. These strategies provide the most efficient applications focused on the diagnosis of inborn errors of
methods for metabolite identifications currently available, metabolism by the determination of discriminatory patterns
mainly through the use of mass spectral/retention index of amino and organic acids in urine [105, 106]. These studies
libraries. The between-instrument reproducibility of electron involve urease treatment to provide enzymatic conversion
impact ionization and gas chromatographic retention times of urea to ammonia and carbon dioxide, whose relatively
are higher than those observed for liquid chromatography high concentration (approximately 2%) can be detrimental
retention times and collision-induced dissociation MS/MS for GC–MS analyses. Although urease treatment has
mass spectra. been performed in more recent metabolic profiling studies
Recent investigations have highlighted the importance of [107–109], investigations have highlighted that this treatment
optimization of analytical platforms. O’Hagan and colleagues can alter the range and concentration of metabolites detected
have employed a closed-loop, multiobjective strategy, also [107]. Individual researchers have to decide whether to lose
referred to as the ‘robot chromatographer’, for the optimization information relating to metabolites which co-elute with the
of GC–MS and comprehensive GCxGC-MS instrumentation. broad urea peak or alter the range and responses for other
Heuristic methods of evolutionary computing are applied metabolites detected.
to navigate optimization search spaces effectively and have A study by Atherton and colleagues has highlighted
shown fold differences in the number of metabolites detected the advantage of employing multiple analytical platforms to
compared to optimization by humans [94, 97]. investigate a range of different biological tissues to obtain a
9
Table 3. Typical examples of applications of mass spectrometry in metabolomics.

Analytical Typical examples as observed
Sample type or area of investigation platform in the scientific literature
Microbial systems GC–MS [5]
Crude fungal extracts for taxonomy DIMS [193–195]
Bacterial cell identification DIMS [196]
Saccharomyces cerevisiae GC–MS [1, 31, 197, 198]
DIMS [1, 31, 199–201]
Chlamydomonas reinhardtii GC–MS [202]
Corynebacterium glutamicum GC–MS [203]
Plant systems
Potato GC–MS [101, 102]
DIMS [204, 205]
Arabidoposis thaliana GC–MS [206]
LC–MS [207, 208]
Tomato GC–MS [88, 185, 209]
LC–MS [209, 210]
DIMS [211]
Medicago truncatula GC–MS [212]
Apple GC–MS [213]
Lotus japonicus GC–MS [51]
Tobacco GC–MS [214]
Herbal medicine and traditional Chinese medicine GC–MS [215]
LC–MS [216]
Cucurbita maxima phloem GC–MS [217]
Ginger LC–MS [218]
Lupin roots LC–MS [219]
Brachypodium distachyon DIMS [220]
Pharbitis leaf sap DIMS [221]
Ryegrass DIMS [222]
Substantial equivalence of genetically modified plants GC–MS [223, 224]
DIMS [204]
Olive oil adulteration DIMS [225]
Mammalian diseases
Inborn errors of metabolism GC–MS [105, 106]
DIMS [226–229]
Kidney GC–MS [107]
Liver GC–MS [58]
LC–MS [230]
Heart failure GC–MS [52]
LC–MS [76]
Preeclampsia GC–MS [111]
Cancers GC–MS [231]
LC–MS [232]
Huntington’s disease GC–MS [109]
Lesch–Nyhan disease GC–MS [108]
Hepatitis B LC–MS [230]
Intestinal fistulas LC–MS [233]
Animals and animal models of disease LC–MS [53, 128, 234, 235]
Drug discovery and development GC–MS and LC–MS [16, 74, 141, 236–240]
greater picture of metabolism. The study showed that the preeclampsia and those obtained from healthy and matched
absence of the PPAR-alpha gene in mice produced metabolic controls. Three metabolites showed a clear discrimination
profiles similar to those observed in diabetes, dyslipidemia and between the two classes when their responses were plotted
metabolic syndrome [53]. Of specific interest is the range of together, with sensitivity and selectivity of 100 and 98%,
critical p-values used in many studies. A number of studies respectively [111]. Although, serum, plasma and urine are
have described very low p-values including the study of heart common mammalian biofluids analysed, other exotic biofluids
failure (2 metabolites with p < 10−15 and ROC area >0.90) can also be analysed including CSF [112] and sweat [113].
[52] and an investigation of metabolic differences associated Castrillo et al recently applied a systems biology
with strenuous exercise (34 metabolites with p < 0.00001) approach, analysing the intracellular metabolome, metabolic
[110]. In a recent study by Kenny et al GC–TOF–MS was footprint, proteome and transciptome to investigate the
employed to compare the plasma obtained from patients with differences associated with growth rate for four nutrient
10
limiting chemostats (nitrogen, carbon, phosphorus and sulfur). from the spleens of obese and lean mice [117]. The
This study highlighted a number of specific metabolism- report showed the influence that this platform can provide an
related differences including the coupling of carbon and improved chromatographic resolution, sensitivity and number
nitrogen fluxes in amino acid biosynthesis [1]. In a separate of metabolite peaks detected. A latter publication from the
study Pope and colleagues applied both genomic and metabolic same authors described a range of univariate data analysis
footprinting approach, employing GC–MS and DIMS, to strategies applied to the same dataset [117]. The complexity
study the phenotypes and genotype–phenotype relationship of the four-dimensional data (retention time 1, retention
of industrial brewing yeasts. The study highlighted the greater time 2, ion current, m/z) has led to the investigation of
discriminatory power of the metabolic footprint compared to the Fisher ratio and parallel factor analysis (PARAFAC),
a range of genome profiling techniques [31]. either singularly or combined, in data reduction and feature
A more recent introduction to the metabolomics toolbox is selection strategies. Fisher ratio is a feature selection tool
that of comprehensive GCxGC-MS [114]. Although GC–MS to define two-dimensional chromatographic peaks depicting
provides high chromatographic resolution, the complexity of biological significance and PARAFAC provides a fully-
many sample types requires a platform that can provide higher deconvoluted mass spectrum as an aid to convert raw data to
chromatographic separation efficiencies. This is accomplished biological significance. These methods have been described
with comprehensive GCxGC-MS by separations performed on for the analysis of data produced from the investigation
two columns of orthogonal properties. Typically, column 1 is of intracellular extracts of fermenting and respiring yeasts
of similar length to those employed in GC–MS (10–60 m) [118] and of urine provided by pregnant and non-pregnant
and column 2 is much shorter (1–2 m). A ‘non-polar’ and females [119]. Multivariate analyses, specifically principal
a ‘polar’ column are generally applied as columns 1 and 2, component analysis (PCA), have also been described for the
respectively, providing separations based in a generalized discrimination of data produced from the analysis of basil,
manner on volatility on column 1 and polarity on column 2, peppermint and stevia [120]. Targeted analyses of specific
though the reversed configuration can also be used. Therefore metabolite classes have also been described for amino acids
two metabolites of similar volatility but differing polarity can [121] and fatty acids [122].
be separated, which was improbable employing GC–MS. It
must always be remembered that these are generalized rules (b) Liquid chromatography–mass spectrometry.
applicable within a sample class but are less generalized LC–MS is a more recent introduction to analytical sciences
in multi-class samples as observed in metabolic profiling than gas chromatography and is highly complementary to
applications. Samples are automatically transferred between GC–MS applications. In metabolomics applications LC–
columns 1 and 2 through a modulator which acts to focus MS is applicable for the analysis of a wider range of
aliquots of eluent exiting column 1 (of temporal length 2–10 s) molecular weights extending from low molecular weight
and release these focused aliquots onto column 2. species that are detectable by GC–MS to molecular weights
Cryomodulation with liquid nitrogen cooled nitrogen jets greater than 600 Da including phospholipids, conjugated
to focus and trap column eluent and heated nitrogen jets bile acids, glycosides and sugars. However, matrix effects
to release focused eluent are common. Other methods of and their influence on the accuracy of results are observed
modulation are also being investigated [115]. Peak widths to a greater degree in LC–MS, compared to GC–MS,
are reduced from 2–6 s on column 1 to less than 0.3 s on applications. Liquid chromatography encompasses a range of
column 2, providing an improvement in S/N. However, this systems including high performance liquid chromatography,
improvement is partly offset by the requirement of higher capillary liquid chromatography and sub-2µm-particle liquid
acquisition rates which can result in a reduction in sensitivity. chromatography. Recent advances have increased LC–MS
These chromatographic improvements allow a greater number applicability, including the introduction of atmospheric
of metabolite peaks to be detected and a greater biological pressure ion sources, mass analysers capable of high mass
overview. However, as with all deconvolution software, resolution and accurate mass measurements and columns of
the software validity and reproducibility is highly dependent lower internal diameter and smaller stationary phase particles
on the reproducibility of the chromatography which can be which improve sensitivity and chromatographic resolution as
influenced by the dimensions of column 2. Rapid transfer shown in figure 2(b) [123].
through column 2 is required and hence this column is short Liquid chromatography operates by the passage of a
and the stationary phase thickness is small. Overloading of liquid mobile phase through a stainless steel column packed
column 2 is possible resulting in broadening of peak widths with the stationary phase, silica or related small diameter
and a reduction in the accuracy and precision of deconvolution particles to which the stationary phase is chemically bonded.
software, though higher film thicknesses can be applied (Paul The full system employs high-performance pumps capable
Begley, personal communication). It is hoped that two- of pumping liquid mobile phases through the high-resistance
dimensional LC–MS will become more routinely available system, a mixing system for combing multiple mobile phases,
in the future [116]. the column and the detector. The most commonly applied
A limited number of applications have been observed for mass spectrometers employ atmospheric pressure ion sources
comprehensive GCxGC-MS. Shellie and colleagues described which act as an efficient tool to create ions from a vaporized
the first application of comprehensive GCxGC–TOF–MS in liquid without overloading the vacuum system of the mass
metabolomics for the analysis of tissue extracts obtained spectrometer. Vaporization is performed at atmospheric
11
pressure, with ions and minimal solvent vapour transferred low abundance metabolites [132]. HILIC columns offer a
into the vacuum region of the mass spectrometer. Electrospray complementary metabolic profile providing greater retention
is the most popular of these and operates in principle as an of hydrophilic metabolites and have been discussed for the
electrochemical cell, where ions are formed in a metal needle analysis of plant extracts and mammalian urine [107, 133,
set at high voltages followed by nebulization from a Taylor 134]. HILIC columns are hydrated silica phases with gradient
cone at the needle exit and desolvation and droplet fission to elutions operating from high organic to high aqueous with the
release ions which are subsequently transferred into the mass greater retention of hydrophilic, compared to hydrophobic,
spectrometer [41, 124]. Electrospray ionization operates at metabolites. Applications include the analysis of urine
µl min−1 flow rates but lower flow rates (nl min−1) can be [107, 133], water soluble metabolites of E. coli [135] and
used with a nanoelectrospray source where the efficiency of ion highly polar metabolites present in phloem exudates, including
release is higher because of the smaller diameters of droplets oligosaccharides, glycosides, amino sugars, amino acids and
[125]. However, nanoelectrospray ionization has not been sugar nucleotides [134]. Greater coverage of the metabolome
routinely applied to metabolomics as of yet. The technique, is achieved with the application of both reversed and HILIC
using chip-based nanospray infusion, has been applied with phase columns. As an alternative for the analysis of the
DIMS for the analysis of deproteinated plasma samples metabolome where many metabolites are hydrophilic and
with MS/MS and accurate mass measurements employed for elute with no or minimal retention on reversed phase column,
metabolite identification. The metabolic discrimination of ten the use of ion-pair reagents containing non-polar moieties is
male and ten female subjects was achieved with this strategy becoming more popular and these provide the ability to retain
[126]. A second alternative is atmospheric pressure chemical hydrophilic metabolites on a hydrophobic stationary phase.
ionization (APCI), where the liquid flow is passed through The mechanism operates by the non-covalent interaction of
a heated glass tube where explosive vaporization occurs to the ion-pair reagent with the metabolite which is present
create solvent droplets and vapour followed by ionization of in an ionized form and allows greater retention of polar
solvent vapour via a corona discharge and subsequent charge or metabolites on reversed phase columns by the interaction of the
ion transfer to metabolite molecules. Electrospray and APCI hydrophobic moiety of the ion-pair reagent with the stationary
are complementary techniques providing a wider detectable phase. Applications include the study of microbial metabolites
coverage of the metabolome, both for polar (ESI) and non- [136] including those of glycolysis, pentose phosphate and
polar (APCI) metabolites, when combined than when operated TCA cycle metabolites [137].
separately. Instruments that allow simultaneous operation Analytical columns with internal diameters of 1–4.6 mm
with both modes are available. Other ionization methods are generally applied, though lower internal diameter columns
available, though rarely applied, include atmospheric pressure can provide greater sensitivity and smaller peak widths.
photoionization (APPI) [127]. To maximize metabolome A separate strategy is to use capillaries operating as
coverage, detection in both positive and negative ion modes monolithic columns which have chromatographic resolutions
is recommended, increasing the analysis time per sample. approaching that of GC applications. These columns have
Certain metabolite classes are expected to be preferentially been described for plant metabolomics [138–140]. The
detected in positive ion mode (for example, amines and amino greatest technological advance has arisen with the application
acids) and other classes in negative ion mode (for example, of sub-2µm chromatographic particles which provide high
sugars and lipids). chromatographic resolutions, narrower peak widths and higher
A wider range of chromatographic phases, and therefore sensitivities [123, 141]. These systems are commonly referred
separation mechanisms, are available when compared to to as ultra performance liquid chromatography (UPLC).
GC–MS. These can be viewed in any column manufacturers The use of sub-2µm particles allows the application of a
catalogue and include reversed phase (C18, C8, C4, phenyl, wider range of mobile phase velocities while maintaining
cyano), normal phase (silica), HILIC and ion exchange. low plate height and also allowing shorter analysis times.
However, a limited range have been regularly applied These instruments operate at up to 15 000 psi and specially
in metabolomics. Reversed phase C18 and C8 columns designed instruments are required to operate at these
are suitable for the analysis of hydrophobic metabolites pressures compared to pressures of up to 6000 psi for high
with a gradient elution operating from high aqueous to performance liquid chromatography (HPLC) [142]. A number
high organic solvent phases. Metabolites suitable for of manufacturers apply this technology including Waters
detection include aromatics, lipids, phospholipids and bile (Acquity UPLC R
), Agilent (1200 Series Rapid Resolution
acids. Metabolite classes investigated include lipids [128], System) and ThermoFisher Scientific (AccelaTM ). Plumb
unconjugated and taurine or glycine conjugated bile acids has described other areas of pioneering work in LC–MS
[129] including those detected in the investigation of gut-based including the application of MSE for the sequential and on-
microbiome–mammalian metabolic interactions [46] and line determination of molecular and CID mass spectra [73] and
plasma phospholipids for the determination of nephropathy the operation of columns at elevated temperatures of 90 ◦ C to
[130] and type II diabetes melititus [131]. Phospholipids increase chromatographic resolution and sensitivity [143]. A
are present in high abundance in mammalian serum and similar approach to elevated temperatures has been described
plasma and can prevent the detection of low abundance for 2D LC–MS studies [144].
and co-eluting metabolites. A method for phospholipid A range of different column chemistries and column
capture has been described to improve the detection of dimensions have been applied in metabolomics as described
12
above. The applicability of capillary HPLC and its advantages employing tandem mass spectrometry, is that of neonatal
compared to traditional HPLC was described by Grainger screening for inherited disorders of metabolism where
for the analysis of rat urine where diurnal variation was 100 000s of sample have been analysed across the world [150].
detected with an increased sensitivity and higher number of As there are no chromatographic separations of
detectable metabolites [145]. Lenz has applied microbore metabolites in DIMS, the ability to identify metabolites is
UPLC to study the hepatotoxic effect of pravastatin as limited. However, the application of high mass resolution and
determined in metabolic profiles of urine and showed the high mass accuracy mass analysers is useful in providing a
ability to profile the metabolic trajectory as toxicity develops higher degree of metabolite identification via accurate mass
[128]. Matrix effects or ionization suppression can limit the measurements and MS/MS experiments. TOF [126], FTICR
reproducibility and accuracy of relative quantification studies [78, 151–154] and Orbitrap [155] mass spectrometers have
in metabolomics and a recent study has shown that this effect is all employed this strategy of accurate mass measurement and
negligible assuming no or small changes in the sample matrix some have developed libraries and databases to enhance the
[146], also shown in DIMS applications [147]. Although ability to identify metabolites by accurate mass, including
rarely applied compared to GC–MS, derivatization to improve the Arabidopsis specific-metabolite relationship database,
chromatographic retention, metabolite stability or detection KNApSAcK [153]. This application showed its applicability
efficiency has been observed [148, 149]. Other applications as a number of flavanoid glycosides were identified by the
of LC–MS in metabolomics are described in table 3, though combination of accurate mass and MS/MS measurements. A
because of the higher numbers of applications not all uses can recent publication by Southam and colleagues has provided
be detailed. an improved approach to higher mass accuracy in FTICR–MS
instruments [156]. Dynamic range and mass accuracy can
be reduced in these instruments at the onset of space-charge
(c) Direct injection mass spectrometry (DIMS).
effects and therefore a small ion population is preferential for
The application of chromatographic separations prior to
high mass accuracy but not necessarily for high sensitivity. The
mass spectrometry has been discussed above and provides a
researchers have developed and validated an approach where
greater ability to identify and quantify metabolites (through
small mass ranges are collected sequentially and ‘stitched’
the combined application of retention time, accurate mass
together to construct the complete mass spectrum of interest.
and fragmentation mass spectra) and potentially to detect a
Sub-ppm mass accuracies were observed. The application
greater number of metabolites. However, chromatographic
was applied to the analysis of fish liver extracts by DIMS
separations are time consuming and an alternative strategy for screening purposes, though could also be applied using
is to introduce the sample directly to an electrospray LC–MS applications for metabolite identification of pooled
mass spectrometer for a short time period. The two- samples to enable a greater degree of metabolite identification.
dimensional mass spectrum is used as a fingerprint for DIMS is applicable for isotopomer-based metabolic
each sample as shown in figure 2(c). DIMS is commonly investigations. Lane and colleagues have employed a
performed using electrospray instruments or less commonly combination of FTICR–MS and NMR to apply stable
with nanoelectrospray instruments [126] and is applied as a isotope tracer studies and provide positional isotopomer
high-throughput technology for sample classification. Sample distributions which are relevant to the biochemical production
introduction can be performed via a syringe pump or in of metabolites from multiple biosynthesis pathways [157].
an automated manner using autosamplers to introduce the This approach of stable isotope labelling has also been applied
sample as a plug into a flowing stream supplied by flow to produce 13C-labelled extracts of S. cerevisiae that were
injection or liquid chromatography pumping systems. A employed as internal standards for relative quantification
range of mass spectrometers can be used though instruments and eliminates the problems of ionization suppression/matrix
operating with high mass resolution and mass accuracy provide effects observed in LC–MS applications [158].
a greater information content by the detection and preliminary The introduction of novel sampling and ionization systems
identification of a greater number of metabolites. High mass is currently being observed, including desorption electrospray
accuracy provides a greater possibility to identify metabolites, ionization (DESI) [159] and extractive electrospray ionization
though care should always be taken when using only one (EESI) [160]. These are systems which are applicable outside
characteristic of a metabolite in its identification. More the laboratory and employ charged liquid sprays from an
than one metabolite (structural isomers) have identical or electrospray source to sample biological systems directly
similar accurate masses. For example, glucose, fructose and without any/minimal sample preparation and at ambient
other hexose sugars all have the same molecular formula conditions. EESI has shown its applicability in the detection
and accurate mass. A separate strategy is to use tandem of meat spoilage, E. coli contamination of spinach and
mass spectrometry for more targeted analyses in multiple contaminants on the surface of skin [160]. DESI has a
reaction monitoring (MRM) mode. Here, identification number of applications including the detection of drugs of
is more definitive by applying the structural correlation abuse in urine [161] and has been compared with NMR for
between precursor and product ion to obtain greater specificity. the detection of inborn errors of metabolism [162]. Other
A higher sensitivity is also achieved compared to single ionization techniques have been applied for rapid analysis
quadrupole instruments and absolute quantification is possible of biological sample extracts including MALDI [163–166]
with the application of isotopic analogues of the analytes as and matrix-free laser desorption ionization (LDI), especially
internal standards. The most common application of DIMS, desorption/ionization on silicon (DIOS) [167, 168].
13
(d) Imaging mass spectrometry. Greater metabolome coverage. The large size of
The technologies and applications described above all require metabolomes investigated and the wide variability of physical
physical removal and ‘pooling’ of metabolites from the intact and chemical properties and the concentrations of metabolites
biological sample (either extraction or desorption). The ensure that no single analytical platform is suitable for
spatial context of the metabolite is lost. The ability to the detection of all metabolites in a biological system.
provide spatial mapping of all metabolites simultaneously Only in the previous 2 years have researchers started to
and correlate these changes to physical and temporal changes revise their views and employ multiple sample preparation
in concentrations can provide a greater oversight and methodologies and analytical platforms to probe the biological
understanding of metabolism. Any technology employed system, traditionally a researcher employed a single analytical
for imaging of metabolites requires the ability to release platform. Today, a growing number of researchers employ
metabolites from a biological matrix with minimal ion a range of analytical platforms of which studies combining
suppression, a good spatial resolution, provide minimal the advantages of mass spectrometry and nuclear magnetic
destruction of the biological tissue outside the area of analysis resonance spectroscopy (NMR) are the most powerful. MS
and allow rapid rastering across the sample so as to collect and NMR are complementary tools [48, 49, 53]. MS provides
data from multiple areas in a relatively high-throughput sensitive detection and the ability to identify metabolites
manner. A range of technologies are currently being assessed though generally at the expense of analysis time and absolute
as this is an emerging discipline, including secondary ion quantification in metabolic profiling studies is currently not
mass spectrometry (SIMS) [169] and matrix-assisted laser feasible. NMR can provide a more rapid analysis time, is
desorption ionization (MALDI) [170]. A recent review has non-destructive and allows metabolite identification. Absolute
described the theory and applications in detail [171]. A quantification is limited in metabolic profiling using MS
recent introduction is that of nanostructure-initiator mass platforms for a number of reasons including the dependence
spectrometry (NIMS), which provides good spatial resolution of response on a number of variables. NMR can provide
though with minimal molecular ion fragmentation compared absolute quantification as few variables influence the measured
to SIMS and MALDI, where significant fragmentation can be response. One important difference between MS and NMR is
observed. Hence more informative data can be collected with the reproducibility observed in short- and long-term studies,
NIMS [172]. where NMR is considered more reproducible than MS,
mainly because of differences in the operation of the two
Current limitations and future outlooks for metabolomics platforms. For example, contamination of MS instrumentation
and subsequent changes in sensitivity are observed because
The role of metabolomics in the study of biological systems the sample is introduced into the mass spectrometer. In NMR,
is increasing. However, focusing on a single functional samples remain in a tube and therefore contamination is not a
level is a reductionist approach and the investigation of all problem. Many of these multi-platform applications employ a
functional levels and their associated and complex interactions combination of gas chromatography, liquid chromatography
is a more appropriate strategy to understand the complex and NMR. The development of LC–NMR–MS platforms
workings of biological systems, termed systems biology is important [175–177]. Only the routine application of
[2, 3]. In most systems biology studies the cyclic roles of multiple analytical platforms and divergent methods of sample
theory, computational modelling and validation of models preparation will enable greater metabolic understanding of
by the acquisition of experimental data are performed. many poorly understood metabolomes, though at higher
Here metabolomics, proteomics and transcriptomics play a experimental costs.
key role in the validation of developed models. Other
systems biology studies have separately assessed or integrated Development of capabilities to improve metabolite
metabolomic, proteomic and transcriptomic data to investigate identification. A central role of the metabolomics
interactions in a qualitative or semi-quantitative approach experiment is to collect analytical data and convert these
[1, 51, 173, 174]. to biological knowledge. The chemical identification of
Metabolomics is playing a key role in these systems- metabolites is essential in this process, without identification
wide studies and also in phenotyping studies. However, the significance of any biological change is not apparent.
the technique is an emerging and continuously developing A large number of investigations observed in the scientific
scientific discipline. As with any emerging science there literature describe many of the metabolites of biological
are always limitations which require resolution, including interest as unidentified. A number of guidelines are available
in the application of mass spectrometry for metabolomic regarding criteria to define the chemical identity of a
investigations. Four areas of specific interest are metabolite [178]. For high specificity two orthogonal
properties should be employed. For example, retention
(i) greater metabolome coverage; time depicts a physical property characteristic (volatility,
(ii) the development of enhanced capabilities to improve hydrophobicity) and accurate mass or fragmentation mass
metabolite identification; spectrum depicts a characteristic of the structure of the
(iii) experimental reproducibility for long-term studies; metabolite.
(iv) deconvolution and feature selection strategies for There are a number of approaches to define or characterize
chromatography–mass spectrometry data. a metabolite:
14
(a) Previously detected and characterized endogenous are dependent on response and hence metabolite concentration
metabolite present in a metabolite library/database that and so a suitable error range in RT/RI should be used.
can be used to identify the metabolite in all samples. Each Generally, a match factor greater than 70% is employed
metabolite should be represented with two orthogonal for a mass spectral match but care must be taken in that
properties. these matches are biased to ions of greater response in the
(b) Endogenous metabolites previously detected and included mass spectrum, in GC–MS mass 73 and 147 are common
in metabolite libraries/databases, but as yet not high intensity ions in trimethylsilyl-derivatized metabolites
chemically characterized. This allows integration of and identification is generally focused on lower intensity
data across multiple samples by matching of one or two ions of higher specificity. The high reproducibility of
orthogonal properties (metabolite x detected in sample electron impact mass spectra and retention indices allow wide
one is referred to as metabolite x in sample, 1, 2, 100, ranging applications and transfer of these libraries between
1000, 10 000). This provides the ability to combine laboratories, a cost-effective process in metabolomics. Mass
datasets but with limited biological information. spectral/retention index libraries are also required, but not
(c) Endogenous metabolites that are present in metabolite currently available, for comprehensive GCxGC-MS.
libraries/databases but are not currently detected. The In LC–MS applications these libraries are much less
information source regarding their possible presence is as developed. Currently, massbank.jp is the single database of
described in (d). mass spectra acquired on a range of LC–MS and GC–MS
(d) Endogenous metabolites that are highlighted from instruments available to search (www.massbank.jp). Both
bibliographic data and reconstruction of metabolic electron impact mass spectra (GC–MS) and MSn mass
networks that are not present in libraries and hence are not spectra (LC–MS) are available. There are a number of
identifiable in a definitive manner in samples. For humans specific exogenous metabolite class libraries commercially
there are a number of bibliographic sources including available but no libraries extending over a wide range
metabolic network reconstructions [27, 179] and the of endogenous metabolites as is observed for GC–MS.
Human Metabolome Database (http://www.hmdb.ca/). Retention time and accurate mass are generally employed
(e) Exogenous metabolites arising from the environment for metabolite identification, though MS/MS or MSn will
(food, drugs, pollutants). These are generally not provide greater specificity. A separate issue is that of
present in commonly available libraries, but can chromatographic reproducibility between columns of the
be available from other bibliographic sources or same stationary phase available from different manufacturers
databases (see http://www.hmdb.ca/, http://drugbank.ca/ and the wider range of chromatography columns available.
and http://hmdb.med.ualberta.ca/foodb). Two reversed phase C18 columns available from different
The accurate chemical identification of metabolites across manufacturers chromatographically act differently for a
multiple samples requires the development of mass spectral number of reasons, including percentage carbon coverage
libraries. The construction of mass spectral libraries and and the chromatographic retention characteristics of other
metabolite databases is necessary in today’s laboratories so functional groups present on a column. These can include the
as to be able to define the metabolome more comprehensively. silanol groups present on the silica particles onto which the
Although there are commercially available GC–MS libraries stationary phase is bound and which can remain chemically
(NIST/EPA/NIH and Wiley) these are not developed with unaltered or changed to reduce the polar influence of the
the central objective of including endogenous (or exogenous) phase (for example, endcapping). Reproducibility of retention
metabolites. These libraries are not specific (for example, indices between GC columns from different manufactures
metabolites may be included but not in the chemically allows transfer of libraries between GC columns. The
derivatized form in which they are detected) and are therefore production of reproducible fragmentation mass spectra on
not widely applicable in metabolomics. In GC–MS, a different instruments is also difficult and is a major limitation
number of research groups are developing their own metabolite to the development of these libraries. The application of
mass spectral/retention index libraries. Currently, the most collision-induced dissociation (CID) is significantly more
appropriate library available is that of the Golm database variable across different instruments when compared to the
which employs both the mass spectrum and retention index higher reproducibility of fragmentation patterns observed with
to define a metabolite [103]. Retention indices are n-alkanes electron impact ionization acquired on different instruments.
(typically C10–C30) added to the sample and used to normalize A single electron energy is used in GC–MS (70 eV) but a range
for variations in retention times observed across multiple of collision energies can be employed in MS/MS experiments.
samples. These retention indices increase the reproducibility The degree of fragmentation is dependent on metabolite
of chromatographic alignment. Alternatively, a relative structure and hence the precursor ion of one metabolite may
retention time marker can be used, which provides a relative extensively fragment at a given collision energy whereas a
retention time (RRT) compared to a single chromatographic separate metabolite may not produce significant fragmentation
peak, with the reference peak generally defined with a RRT of the molecular ion at the same collision energy. It
of 1.0. Retention indices are recommended as these employ is recommended to use multiple collision energies in the
multiple markers and should therefore be more reproducible construction of MS/MS libraries (John Halket, personal
for changes in different regions of, but not all of, the retention communication). CID can be performed at multiple collision
time window. It should be noted that retention times/indices energies for the off-line identification of metabolites but this
15
is not suitable for the on-line identification of metabolites spectrometry instrumentation, changes in retention time and
in liquid chromatographic applications. Work is ongoing mass accuracy and the required maintenance of these systems).
regarding the preparation of mass spectral libraries suitable Variability is observed in the detectable variables of retention
across multiple manufacturer’s platforms. time, accurate mass and sensitivity. Variability can be
Further development of LC/MS and GC/MS mass observed as drift or sudden step changes in these detected
spectral libraries and their applicability and availability variables. It is therefore advisable to divide long-term studies
across many research laboratories is required, but not yet into smaller experiments to reduce this long-term variability.
possible. The costs to develop these libraries are high In most investigations a metabolic profiling approach
and require the purchase or synthesis of authentic standards is employed which is semi-quantitative to assess relative
which are analysed using the same analytical methods as for differences in the response of metabolites between different
metabolomic investigations. A cumulative effort in the field classes and therefore changes in sensitivity across time can
is required. Many primary metabolites are well characterized influence the validity of results. As a minimum requirement to
and available commercially, others are not commonly known overcome this problem, samples from multiple classes should
though can be derived from two sources shown below: be randomized in their order of analysis. However, ensuring
that changes in sensitivity do not occur is more important.
(a) genome-scale reconstructions of metabolic networks; Although it would be thought that internal standards would
(b) characterization of unidentified metabolites detected compensate for these changes in sensitivity, in reality this
in current metabolomic studies using other strategies is not the case (Ian Wilson, personal communication), an
including metabolite isolation and characterization by isotopic analogue of each metabolite would be required and
mass spectrometry and NMR. this is not experimentally feasible, both because not all
However, the majority of these metabolites are not metabolites are identified but also that the cost of this strategy
commercially available and therefore only ‘possible’ or is prohibitive. In long-term studies in the pharmaceutical
probable’ identifications can be obtained for those metabolites industry absolute quantification is employed to minimize the
for which authentic standards are not commercially available. influence of response variability over time, however this is
The development of libraries and databases requires for a limited number of chemicals and this approach is not
appropriate notation of metabolites with a primary identifier feasible in a data-driven strategy when unknown metabolites
that is recognizable to all users. However, all metabolites would be present in the samples and authentic standards are
not available for all. A reduction in biological information
generally have a number of synonyms (see a range of databases
can be obtained at the expense of absolute quantification in
including KEGG and the Human Metabolome Database for
some applications [185]. Further developments are required
examples) all descriptive to some but not all users. There is
for longer-term studies and are under investigation in a number
only one single identifier which if applied robustly is a unique
of laboratories worldwide.
identifier, isomeric SMILE strings [180]. The greater the
number of different identifiers described the more applicable a
library or database will be, so as to be able to integrate multiple Deconvolution and feature selection strategies for
databases and metabolic network or systems biology chromatography–mass spectrometry data. Metabolomics is
models. The application of INCHI (http://www.inchi.info/), routinely applied as a high-throughput discipline, providing
ChEBI (http://www.ebi.ac.uk/chebi/) and KEGG the analysis of tens to hundreds of samples per day, each
analysis providing multi-dimensional data detailing
(http://www.genome.jp/kegg/) identifiers to provide
information regarding the concentration and identification
metabolite names is recommended within the metabolomics
of metabolites. Storage of data in an appropriate format
community. However, ontologies are currently in the early
to allow rapid and appropriate interrogation is performed
stages of development [181].
with suitably designed databases, of which MeMo [186]
is the first designed for mass spectrometry/metabolomics
Experimental reproducibility for long-term studies. In the applications. In systems biology, the interrogation of data
early development of all scientific disciplines an assessment from multiple omic databases and its subsequent integration
of reproducibility related to sample collection, sample will be required and is currently being developed, including
preparation, analytical platform performance and data analysis at The Manchester Centre for Integrative Systems Biology
is required. One important principle is that any variability (MCISB, http://www.mcisb.org/).
associated with the combination of all these processes is Chromatography–mass spectrometry platforms can
lower than the biological variability observed for inter- and produce vast volumes of data. Typically raw data files of
intra-class distributions. There are a number of recent size 50 MB for each sample are typical, though file sizes of
publications that have assessed these different protocols 0.2–1 GB are observed for comprehensive GCxGC-MS data.
and their reproducibility [101, 182–184]. However, these Raw three- or four-dimensional data (retention time, mass, ion
are generally assessed over a short time period and as count or retention time 1, retention time 2, mass, ion count)
the time period of an investigation increases so will the require conversion into biological information and generally
associated technical variability. This increase in variability two different methods of data reduction can be applied:
is produced because of multiple replications of manual (a) Reduction into a two-dimensional data matrix of
procedures and changes in the operation of analytical platforms chromatographic peaks and peak areas applying an
(the contamination of chromatographic columns and mass in-silico conversion termed deconvolution.
16
Table 4. The range of deconvolution and feature selection software currently available.
Software Availability Further information
AnalyserPro Available from SpectralWorks http://www.spectralworks.com/analyzerpro.asp
ChromaTOF Available from LECO http://www.leco.com/index3.htm
COMSPARI [241] Freely available http://www.biomechanic.org/comspari/
Genespring Available from Agilent Technologies http://www.chem.agilent.com
MarkerLynxTM Available from Waters http://www.waters.com
MarkerviewTM Available from MDS/Applied Biosystems http://www.mdssciex.com/
MathDAMP [242] Freely available http://mathdamp.iab.keio.ac.jp/
MetAlign Freely available http://www.metalign.wur.nl/UK
MET-IDEA [243] Freely available http://www.noble.org/PlantBio/MS/MET-IDEA/index.html
MSFACTS [244] Freely available http://www.noble.org/PlantBio/MS/MSFACTs/MSFACTs.html
Mzmine [245, 246] Freely available http://mzmine.sourceforge.net/
SIEVETM Available from ThermoFisher Scientific http://www.thermofisher.com/global/en/home.asp
XCMS [247] Freely available http://metlin.scripps.edu/download/
(b) Feature selection method without deconvolution to reproducibility and deconvolution accuracy at the expense of
highlight chromatographic regions and associated biological information. A comparison of GC–MS specific
collection of ions relating to differences detected between deconvolution software packages has been performed recently
different sample classes [187].
A range of commercial and freely available software
Deconvolution software packages all operate by
packages are currently available as shown in table 4. In the
employing a characteristic of the metabolite that changes as
authors experience the analysis of data can be instrument,
it elutes from the column and is detected. With GC–MS data
sample and software dependent. Many of the deconvolution
an increase and subsequent decrease of a collection of ions,
software packages are designed for either GC–MS or LC–MS
which are highly correlated in response, is employed as a
data and were developed generally using data from one or a
characteristic of a peak, its mass spectrum and integration
limited number of manufacturers’ instruments. Data collected
under a single quantification ion employed to define the on different instruments are variable. Currently, no single
identity of a peak and its area. For LC–MS there are no deconvolution software is universally applicable for either
or a minimal number of fragment ions and hence a single ion GC–MS or LC–MS data and it is recommended to acquire
is chosen as a characteristic of a peak, and integration under data composed of multiple technical or biological replicates
this ion is employed to define the area of a peak. The use of of the same sample and employ these data to assess different
accurate masses is advisable to provide greater specificity with software and its applicability for the researcher’s data.
a larger degree of biological information and a higher number Feature selection methods do not require in-silico
of detected peaks. Of interest is that co-eluting peaks can be deconvolution of the three- or four-dimensional data and the
classified as peaks in LC–MS data meaning that a requirement associated problems that can arise from variation in noise, peak
of ‘perfect’ chromatography providing spatial separation of shape, retention time and chromatographic resolution. Instead
all metabolites is not essential, though is recommended by the the raw data are interrogated directly, either in a univariate or
author. The application of multiple ions and the more complex multivariate approach to define regions of a three-dimensional
mass spectrum observed for each metabolite peak in GC–MS space in which differences are observed. The associated level
data means that each peak requires a unique retention time, of discrimination and significant mass(es) associated with
each peak apex has to be separated by a number of data points, these differences are also reported. A number of researchers
dependent on the software package. have applied this strategy though again this can be platform,
The author highly recommends the collection of technical instrument and data dependent [118, 119, 188–191]. This type
replicate data from an identical sample to assess the most of data analysis is applicable in comparative metabolomics to
appropriate parameter values to apply in deconvolution define biological differences but is less applicable in long-
software so as to provide a greater understanding of the term studies where definition of metabolites, using retention
complex operations of the software. Simple changes to one time and accurate mass or fragmentation mass spectrum, is
parameter can influence the reproducibility and validity of required.
data and reduce biological knowledge. Unfortunately, the All commercial mass spectrometry software provides
variability observed in metabolomic analyses with respect to raw data in manufacturer-specific formats which requires
noise, peak shape, retention time, chromatographic resolution conversion to a common format for further data processing
and mass accuracy can all influence the efficiency of on software external to the manufacturer’s. At present this
chromatographic deconvolution and their application to data is the NetCDF (network common data format), a binary data
reduction within and between analytical batches. A simple format which is machine independent and which is applicable
example has been described by Joachim Kopka who showed to allow direct introduction into most deconvolution software
a correlation in retention index changes with metabolite packages. This enables data transfer but not the transfer of
concentration. Compromises are sometimes required instrument metadata. Currently mzXML and mzData are two
in experimental procedures to improve chromatographic separate formats that provide the transfer of both data and
17
metadata, though efforts in the proteomics field are enabling [8] Kopka J, Fernie A, Weckwerth W, Gibon Y and Stitt M 2004
the combination of these two formats into a single format, Metabolite profiling in plant biology: platforms and
destinations Genome Biol. 5 109
mzML (see http://www.psidev.info/index.php?q=node/257),
[9] Weckwerth W and Morgenthal K 2005 Metabolomics: from
which can allow the direct introduction of data and metadata pattern recognition to biological interpretation Drug
into proteomic and subsequently metabolomic databases. Discov. Today 10 1551–8
Currently, only a limited number of manufacturers provide [10] Lin C Y, Viant M R and Tjeerdema R S 2006 Metabolomics:
this data format. methodologies and applications in the environmental
sciences J. Pestic. Sci. 31 245–51
[11] Viant M R 2007 Metabolomics of aquatic organisms: the new
Summary ‘omics’ on the block Mar. Ecol.-Prog. Ser. 332 301–6
[12] Fava F, Lovegrove J A, Gitau R, Jackson K G and Tuohy K
Metabolomics has taken great strides in the previous M 2006 The gut microbiota and lipid metabolism:
30 years in the development and validation of sample implications for human health and coronary heart disease
Curr. Med. Chem. 13 3005–21
collection processes, analytical methodologies, analytical [13] German J B, Watkins S M and Fay L B 2005 Metabolomics
platforms, data analysis processes and data storage. These in practice: emerging knowledge to guide future dietetic
developments have provided data and a greater biological advice toward individualized health J. Am. Diet. Assoc.
understanding of a wide range of biological systems including 105 1425–32
microbial, plant and mammalian systems. Mass spectrometry [14] Goodacre R 2007 Metabolomics of a superorganism J. Nutr.
137 259S–66S
is one analytical platform that has contributed to these
[15] van der Greef J, Stroobant P and van der Heijden R 2004 The
advances with, among others, improvements in the number role of analytical sciences medical systems biology Curr.
of metabolites detected and the number of these metabolites Opin. Chem. Biol. 8 559–65
that have been identified. However, researchers in the [16] Ackermann B L, Hale J E and Duffin K L 2006 The role of
field are only at the start of a long journey and further mass spectrometry in biomarker discovery and
developments or refinements in all of the processes involved in measurement Curr. Drug Metab. 7 525–39
[17] Chen C, Gonzalez F J and Idle J R 2007 LC–MS-based
the metabolomics workflow are required. Presently, specific metabolomics in drug metabolism Drug Metab. Rev.
attention should be directed to the design of experiments, the 39 581–97
identification of metabolites and the storage and interrogation [18] Dieterle F, Schlotterbeck G T, Ross A, Niederhauser U and
of data. Senn H 2006 Application of metabonomics in a compound
ranking study in early drug development revealing
drug-induced excretion of choline into urine Chem. Res.
Acknowledgments Toxicol. 19 1175–81
[19] Kell D B 2006 Systems biology, metabolic modelling and
WD wishes to thank the BBSRC and EPSRC for financial metabolomics in drug discovery and development Drug
support of The Manchester Centre for Integrative Systems Discov. Today 11 1085–92
[20] Lindon J C, Holmes E and Nicholson J K 2006
Biology. Many colleagues have contributed to this manuscript Metabonomics techniques and applications to
with interesting debates and discussions. Special thanks go pharmaceutical research & development Pharm. Res.
to Douglas Kell, David Broadhurst, Marie Brown, Kathleen 23 1075–88
Carroll, Catherine Winder, Paul Begley, Sue McIntyre and Roy [21] van der Greef J et al 2007 The art and practice of systems
Goodacre. For continued assistance I also thank ThermoFisher biology in medicine: mapping patterns of relationships
J. Proteome Res. 6 1540–59
Scientific, LECO, Waters and Anatune Ltd.
[22] Kell D B 2006 Metabolomics, modelling and machine
learning in systems biology—towards an understanding of
References the languages of cells FEBS J. 273 873–94
[23] Griffiths I W and Thomson J J 1997 The centenary of his
[1] Castrillo J I et al 2007 Growth control of the eukaryote cell: discovery of the electron and of his invention of mass
a systems biology study in yeast J. Biol. 6 4 spectrometry Rapid Commun. Mass Spectrom. 11 3–16
[2] Palsson B O 2006 Systems Biology: Properties of [24] Sanger F, Nicklen S and Coulson A R 1977 DNA sequencing
Reconstructed Networks (Cambridge: Cambridge with chain-terminating inhibitors Proc. Natl Acad. Sci.
University Press) USA 74 5463–7
[3] Westerhoff H V and Palsson B O 2004 The evolution of [25] Oliver S G, Winson M K, Kell D B and Baganz F 1998
molecular biology into systems biology Nat. Biotechnol. Systematic functional analysis of the yeast genome Trends
22 1249–52 Biotechnol. 16 373–8
[4] Khoo S H G and Al-Rubeai M 2007 Metabolomics as a [26] Beecher C W W 2004 The human metabolome Metabolic
complementary tool in cell culture Biotechnol. Appl. Profiling: Its Role in Biomarker Discovery and Gene
Biochem. 47 71–84 Function Analysis ed G G Harrigan (London: Kluwer)
[5] Koek M M, Muilwijk B, van der Werf M J and Hankemeier T p 311
2006 Microbial metabolomics with gas chromatography/ [27] Duarte N C et al 2007 Global reconstruction of the human
mass spectrometry Anal. Chem. 78 1272–81 metabolic network based on genomic and bibliomic data
[6] Mashego M R, Rumbold K, De Mey M, Vandamme E, Proc. Natl Acad. Sci. USA 104 1777–82
Soetaert W and Heijnen J J 2007 Microbial metabolomics: [28] Forster J, Famili I, Fu P, Palsson B O and Nielsen J 2003
past, present and future methodologies Biotechnol. Lett. Genome-scale reconstruction of the Saccharomyces
29 1–16 cerevisiae metabolic network Genome Res. 13 244–53
[7] Hall R D 2006 Plant metabolomics: from holistic hope, to [29] Tweeddale H, Notley-McRobb L and Ferenci T 1998 Effect
hype, to hot topic New Phytol. 169 453–68 of slow growth on metabolism of Escherichia coli, as
18
revealed by global metabolite pool (‘Metabolome’) [49] Williams R et al 2006 A multi-analytical platform approach
analysis J. Bacteriol. 180 5109–16 to the metabonomic analysis of plasma from normal and
[30] Kell D B and Mendes P 1999 Snapshots of Zucker (FA/FA) obese rats Mol. Biosyst. 2 174–83
systems—metabolic control analysis and biotechnology in [50] Kell D B and Oliver S G 2004 Here is the evidence, now what
the post-genomic era Technological and Medical is the hypothesis? The complementary roles of inductive
Implications of Metabolic Control Analysis and hypothesis-driven science in the post-genomic era
ed A Cornish-Bowden and M L Cárdenas (Dordrecht: Bioessays 26 99–105
Kluwer) pp 3–25 [51] Colebatch G et al 2004 Global changes in transcription
[31] Pope G A et al 2007 Metabolic footprinting as a tool for orchestrate metabolic differentiation during symbiotic
discriminating between brewing yeasts Yeast nitrogen fixation in Lotus japonicus Plant J. 39 487–512
24 667–79 [52] Dunn W et al 2007 Serum metabolomics reveals many novel
[32] Urbanczyk-Wochniak E et al 2003 Parallel analysis of metabolic markers of heart failure, including
transcript and metabolic profiles: a new approach in pseudouridine and 2-oxoglutarate Metabolomics
systems biology EMBO Rep. 4 989–93 3 413–26
[33] Dunn W B and Ellis D I 2005 Metabolomics: current [53] Atherton H J et al 2006 A combined H-1-NMR
analytical platforms and methodologies Trends Anal Chem. spectroscopy- and mass spectrometry-based metabolomic
24 285–94 study of the PPAR-alpha null mutant mouse defines
[34] Goodacre R, Vaidyanathan S, Dunn W B, Harrigan G G and profound systemic changes in metabolism linked to the
Kell D B 2004 Metabolomics by numbers: acquiring and metabolic syndrome Physiol. Genomics 27 178–86
understanding global metabolite data Trends Biotechnol. [54] Brown M, Dunn W B, Ellis D I, Goodacre R, Handl J,
22 245–52 Knowles J D, O’Hagan S, Spasic I and Kell D B 2005 A
[35] Fiehn O 2002 Metabolomics—the link between genotypes metabolome pipeline: from concept to data to knowledge
and phenotypes Plant Mol. Biol. 48 155–71 Metabolomics 1 39–51
[36] Bino R J et al 2004 Potential of metabolomics as a functional [55] Broadhurst D I and Kell D B 2006 Statistical strategies for
genomics tool Trends Plant Sci. 9 418–25 avoiding false discoveries in metabolomics and related
experiments Metabolomics 2 171–96
[37] Dixon R A et al 2006 Perspective—applications of
[56] Ioannidis J P A 2005 Why most published research findings
metabolomics in agriculture J. Agric. Food Chem.
are false PLoS Med. 2 696–701
54 8984–94
[57] Ransohoff D F 2005 Opinion—bias as a threat to the validity
[38] Dunn W B, Bailey N J C and Johnson H E 2005 Measuring of cancer molecular-marker research Nat. Rev. Cancer
the metabolome: current analytical technologies Analyst 5 142–9
130 606–25
[58] Yu K et al 2007 A metabonomic investigation on the
[39] Sumner L W, Mendes P and Dixon R A 2003 Plant biochemical perturbation in liver failure patients caused by
metabolomics: large-scale phytochemistry in the hepatitis b virus J. Proteome Res. 6 2413–9
functional genomics era Phytochemistry 62 817–36 [59] Fiehn O et al 2007 The metabolomics standards initiative
[40] Kell D B, Brown M, Davey H M, Dunn W B, Spasic I and (MSI) Metabolomics 3 175–8
Oliver S G 2005 Metabolic footprinting and systems [60] Ball C A and Brazma A 2006 MGED standards: work in
biology: the medium is the message Nat. Rev. Microbiol. progress Omics 10 138–44
3 557–65 [61] Taylor C F, Hermjakob H, Julian R K, Garavelli J S,
[41] Villas-Boas S G, Mas S, Akesson M, Smedsgaard J and Aebersold R and Apweiler R 2006 The work of the
Nielsen J 2005 Mass spectrometry in metabolome analysis Human Proteome Organisation’s Proteomics Standards
Mass. Spectrom. Rev. 24 613–46 Initiative (HUPO PSI) Omics 10 145–51
[42] Noh K, Gronke K, Luo B, Takors R, Oldiges M and [62] Dole M, Cox H L and Gieniec J 1973 Electrospray
Wiechert W 2007 Metabolic flux analysis at ultra short mass-spectroscopy Adv. Chem. Ser. 73–84
time scale: isotopically non-stationary C-13 labeling [63] Karas M, Bachmann D and Hillenkamp F 1985 Influence of
experiments J. Biotechnol. 129 249–67 the wavelength in high-irradiance ultraviolet-laser
[43] Maharjan R P and Ferenci T 2003 Global metabolite analysis: desorption mass-spectrometry of organic-molecules Anal.
the influence of extraction methodology on metabolome Chem. 57 2935–9
profiles of Escherichia coli Anal. Biochem. 313 145–54 [64] Yamashita M and Fenn J B 1984 Electrospray
[44] Villas-Boas S G and Bruheim P 2007 Cold glycerol-saline: ion-source—another variation on the free-jet theme
the promising quenching solution for accurate intracellular J. Phys. Chem. 88 4451–9
metabolite analysis of microbial cells Anal. Biochem. [65] Gohlke R S 1959 Time-of-flight mass spectrometry and
370 87–97 gas–liquid partition chromatography Anal. Chem.
[45] Villas-Boas S G, Hojer-Pedersen J, Akesson M, 31 535–41
Smedsgaard J and Nielsen J 2005 Global metabolite [66] McFadden W H, Day J C, Teranishi R and Black D R 1963
analysis of yeast: evaluation of sample preparation Use of capillary gas chromatography with a time-of-flight
methods Yeast 22 1155–69 mass spectrometer J. Food Sci. 28 316
[46] Martin F P J et al 2007 A top–down systems biology view of [67] Anderson J S, Vartanian V H and Laude D A 1994 Evolution
microbiome–mammalian metabolic interactions in a of trapped ion cells in Fourier-transform ion-cyclotron
mouse model Mol. Syst. Biol. 3 resonance mass-spectrometry Trends Anal. Chem.
[47] Ramadan Z, Zhang P F, Jacobs D M, Tavazzi I and Kochhar S 13 234–9
2007 An NMR- and MS-based metabonomic investigation [68] Hu Q Z, Noll R J, Li H Y, Makarov A, Hardman M and
of saliva metabolic changes in feline odontoclastic Cooks R G 2005 The Orbitrap: a new mass spectrometer
resorptive lesions (FORL)-diseased cats Metabolomics J. Mass Spectrom. 40 430–43
3 113–9 [69] Gross J H Mass Spectrometry: A Textbook 1st edn (Berlin:
[48] van der Werf M J, Overkamp K M, Muilwijk B, Coulier L Springer) 2004
and Hankemeier T 2007 Microbial metabolomics: toward [70] Dettmer K, Aronov P A and Hammock B D 2007 Mass
a platform with full metabolome coverage Anal. Biochem. spectrometry-based metabolomics Mass Spectrom. Rev.
370 17–25 26 51–78
19
[71] Vandergreef J and Niessen W M A 1992 Hyphenated methods inoculated with dry and soft rot pathogens Am. J. Potato
in mass-spectrometry Int. J. Mass Spectrom. Ion Process Res. 82 1–8
118 857–73 [90] Verdonk J C, de Vos C H R, Verhoeven H A, Haring M A,
[72] Major H J, Williams R, Wilson A J and Wilson I D 2006 A van Tunen A J and Schuurink R C 2003 Regulation of
metabonomic analysis of plasma from Zucker rat strains floral scent production in petunia revealed by targeted
using gas chromatography/mass spectrometry and pattern metabolomics Phytochemistry 62 997–1008
recognition Rapid Commun. Mass Spectrom. 20 3295–302 [91] Vikram A, Hamzehzarghani H and Kushalappa A C 2005
[73] Plumb R S et al 2006 UPLIC/MSE: a new approach for Volatile metabolites from the headspace of onion bulbs
generating molecular fragment information for biomarker inoculated with postharvest pathogens as a tool for disease
structure elucidation Rapid Commun. Mass Spectrom. discrimination Can. J. Plant Pathol.-Rev. Can.
20 1989–94 Phytopathol. 27 194–203
[74] Plumb R S, Stumpf C L, Granger J H, Castro-Perez J, [92] Fiehn O, Kopka J, Dormann P, Altmann T, Trethewey R N
Haselden J N and Dear G J 2003 Use of liquid and Willmitzer L 2000 Metabolite profiling for plant
chromatography/time-of-flight mass spectrometry and functional genomics Nat. Biotechnol. 18 1157–61
multivariate statistical analysis shows promise for the [93] Halket J M and Zaikin V G 2003 Derivatization in mass
detection of drug metabolites in biological fluids Rapid spectrometry: 1. Silylation Eur. J. Mass Spectrom. 9 1–21
Commun. Mass Spectrom. 17 2632–8 [94] O’Hagan S, Dunn W B, Brown M, Knowles J D and Kell D B
[75] Kind T and Fiehn O 2006 Metabolomic database annotations 2005 Closed-loop, multiobjective optimization of
via query of elemental compositions: mass accuracy is analytical instrumentation: gas chromatography/time-of-
insufficient even at less than 1 ppm Bioinformatics 7 254 flight mass spectrometry of the metabolomes of human
[76] Sabatine M S et al 2005 Metabolomic identification of novel serum and of yeast fermentations Anal. Chem. 77 290–303
biomarkers of myocardial ischemia Circulation [95] Noctor G, Bergot G L, Mauve C, Thominet D,
112 3868–75 Lelarge-Trouverie C and Prioul J L 2007 A comparative
[77] Weaver P J, Laures A M F and Wolff J C 2007 Investigation study of amino acid measurement in leaf extracts by gas
of the advanced functionalities of a hybrid quadrupole chromatography–time of flight–mass spectrometry and
orthogonal acceleration time-of-flight mass spectrometer high performance liquid chromatography with
Rapid Commun. Mass Spectrom. 21 2415–21 fluorescence detection Metabolomics 3 161–74
[78] Brown S C, Kruppa G and Dasseux J L 2005 Metabolomics [96] Kopka J 2006 Current challenges and developments in
applications of FT-ICR mass spectrometry Mass Spectrom. GC–MS based metabolite profiling technology J.
Rev. 24 223–31 Biotechnol. 124 312–22
[79] Lenz E M and Wilson I D 2007 Analytical strategies in [97] Hagan S O et al 2007 Closed-loop, multiobjective
metabonomics J. Proteome Res. 6 443–58 optimization of two-dimensional gas
chromatography/mass spectrometry for serum
[80] Gao P et al 2007 Investigation on response of the metabolites
metabolomics Anal. Chem. 79 464–76
in tricarboxylic acid cycle of Escherichi coli and
[98] Horning E C 1968 Use of combined gas–liquid
Pseudomonas aeruginosa to antibiotic perturbation by
chromatography and mass spectrometry for clinical
capillary electrophoresis J. Pharm. Biomed. Anal.
problems Clin. Chem. 14 777
44 180–7
[99] Pauling L, Robinson A B, Teranish R and Cary P 1971
[81] Ramautar R, Demirci A and de Jong G J 2006 Capillary Quantitative analysis of urine vapor and breath by
electrophoresis in metabolomics Trends Anal. Chem. gas–liquid partition chromatography Proc. Natl Acad. Sci.
25 455–66 USA 68 2374
[82] Saito N et al 2006 Metabolomics approach for enzyme [100] Sauter H, Lauer M and Fritsch H 1991 Metabolic profiling
discovery J. Proteome Res. 5 1979–87 of plants—a new diagnostic-technique ACS Symp. Ser.
[83] Soga T, Ohashi Y, Ueno Y, Naraoka H, Tomita M and 443 288–99
Nishioka T 2003 Quantitative metabolome analysis using [101] Roessner U, Wagner C, Kopka J, Trethewey R N and
capillary electrophoresis mass spectrometry J. Proteome Willmitzer L 2000 Simultaneous analysis of metabolites in
Res. 2 488–94 potato tuber by gas chromatography–mass spectrometry
[84] Robards K, Haddad P R and Jackson P E Principles and Plant J. 23 131–42
Practice of Modern Chromatographic Methods (London: [102] Weckwerth W, Loureiro M E, Wenzel K and Fiehn O 2004
Academic) 1997 Differential metabolic networks unravel the effects of
[85] Gohlke R S 1962 Time-of-flight mass silent plant phenotypes Proc. Natl Acad. Sci. USA
spectrometry—application to capillary column gas 101 7809–14
chromatography Anal. Chem. 34 1332–3 [103] Kopka J et al 2005 GMD@CSB.DB: the Golm Metabolome
[86] Dixon S J et al 2007 Pattern recognition of gas Database Bioinformatics 21 1635–8
chromatography mass spectrometry of human volatiles in [104] Schauer N et al 2005 GC–MS libraries for the rapid
sweat to distinguish the sex of subjects and determine identification of metabolites in complex biological
potential discriminatory marker peaks Chemometrics samples FEBS Lett. 579 1332–7
Intell. Lab. Syst. 87 161–72 [105] Halket J M, Przyborowska A, Stein S E, Mallard W G,
[87] Farag M A, Ryu C M, Sumner L W and Pare P W 2006 Down S and Chalmers R A 1999 Deconvolution gas
GC–MS SPME profiling of rhizobacterial volatiles reveals chromatography mass spectrometry of urinary organic
prospective inducers of growth promotion and induced acids—potential for pattern recognition and automated
systemic resistance in plants Phytochemistry 67 2262–8 identification of metabolic disorders Rapid Commun. Mass
[88] Kant M R, Ament K, Sabelis M W, Haring M A and Spectrom. 13 279–84
Schuurink R C 2004 Differential timing of spider [106] Kuhara T 2007 Noninvasive human metabolome analysis for
mite-induced direct and indirect defenses in tomato plants differential diagnosis of inborn errors of metabolism
Plant Physiol. 135 483–95 J. Chromatogr. B 855 42–50
[89] Lui L H, Vikram A, Abu-Nada Y, Kushalappa A C, [107] Kind T, Tolstikov V, Fiehn O and Weiss R H 2007 A
Raghavan G S V and Al-Mughrabi K 2005 Volatile comprehensive urinary metabolomic approach for
metabolic profiling for discrimination of potato tubers identifying kidney cancer Anal. Biochem. 363 185–95
20
[108] Ohdoi C, Nyhan W L and Kuhara T 2003 Chemical diagnosis metabolomics profiling of plasma Anal.
of Lesch–Nyhan syndrome using gas chromatography– Chem. 77 7255–64
mass spectrometry detection J. Chromatogr. B [127] Raffaelli A and Saba A 2003 Atmospheric pressure
792 123–30 photoionization mass spectrometry Mass Spectrom. Rev.
[109] Underwood B R et al 2006 Huntington disease patients and 22 318–31
transgenic mice have similar pro-catabolic serum [128] Lenz E M et al 2007 The application of microbore
metabolite profiles Brain 129 877–86 UPLC/oa–TOF–MS and H-1 NMR spectroscopy to the
[110] Pohjanen E et al 2007 A multivariate screening strategy for metabonomic analysis of rat urine following the
investigating metabolic effects of strenuous physical intravenous administration of pravastatin J. Pharm
exercise in human serum J. Proteome Res. 6 2113–20 Biomed. Anal. 44 845–52
[111] Kenny L C et al 2005 Novel biomarkers for preeclampsia [129] Wang Y and Griffiths W J 2007 Modern methods of bile acid
detected using metabolomics and machine learning analysis by mass spectrometry: a view into the
Metabolomics 1 227–34 metabolome Curr. Anal. Chem. 3 103–26
[112] Obata T, Nagakura T, Maeda H, Yamashita K and [130] Jia L W, Wang C, Kong H W, Cai Z W and Xu G W 2006
Maekawa K 1999 Simultaneous assay of prostaglandins Plasma phospholipid metabolic profiling and biomarkers
and thromboxane in the cerebrospinal fluid by gas of mouse IgA nephropathy Metabolomics 2 95–104
chromatography–mass spectrometry-selected ion [131] Wang C et al 2005 Plasma phospholipid metabolic profiling
monitoring J. Chromatogr. B 731 73–81 and biomarkers of type 2 diabetes mellitus based on
[113] Dixon S J, Brereton R G, Soini H A, Novotny M V and high-performance liquid chromatography/electrospray
Penn D J 2006 An automated method for peak detection mass spectrometry and multivariate statistical analysis
and matching in large gas chromatography–mass Anal. Chem. 77 4108–16
spectrometry data sets J. Chemometr. 20 325–40 [132] Want E J, Smith C A, Qin C A, VanHorne K C and Siuzdak G
[114] Beens J and Brinkman U A T 2005 Comprehensive 2006 Phospholipid capture combined with non-linear
two-dimensional gas chromatography—a powerful and chromatographic correction for improved serum
versatile technique Analyst 130 123–7 metabolite profiling Metabolomics 2 145–54
[115] Seeley J V, Micyus N J, Bandurski S V, Seeley S K and [133] Idborg H, Zamani L, Edlund P O, Schuppe-Koistinen I and
McCurry J D 2007 Microfluidic Deans switch for Jacobsson S P 2005 Metabolic fingerprinting of rat urine
comprehensive two-dimensional gas chromatography by LC/MS: Part 1. Analysis by hydrophilic interaction
Anal. Chem. 79 1840–7 liquid chromatography–electrospray ionization mass
[116] Stoll D R, Li X P, Wang X O, Carr P W, Porter S E G and spectrometry J. Chromatogr. B 828 9–13
Rutan S C 2007 Fast, comprehensive two-dimensional [134] Tolstikov V V and Fiehn O 2002 Analysis of highly polar
liquid chromatography J. Chromatogr. A 1168 3–43 compounds of plant origin: combination of hydrophilic
[117] Shellie R A et al 2005 Statistical methods for comparing interaction chromatography and electrospray ion trap mass
comprehensive two-dimensional gas chromatography– spectrometry Anal. Biochem. 301 298–307
time-of-flight mass spectrometry results: metabolomic [135] Bajad S U, Lu W Y, Kimball E H, Yuan J, Peterson C and
analysis of mouse tissue extracts J. Chromatogr. Rabinowitz J D 2006 Separation and quantitation of water
A 1086 83–90 soluble cellular metabolites by hydrophilic interaction
[118] Mohler R E, Dombek K M, Hoggard J C, Pierce K M, chromatography–tandem mass spectrometry J.
Young E T and Synovec R E 2007 Comprehensive Chromatogr. A 1125 76–88
analysis of yeast metabolite GCxGC-TOFMS data: [136] Coulier L, Bas R, Jespersen S, Verheij E, van der Werf M J
combining discovery-mode and deconvolution and Hankemeier T 2006 Simultaneous quantitative
chemometric software Analyst 132 756–67 analysis of metabolites using ion-pair liquid
[119] Pierce K M et al 2006 Fisher ratio method applied to chromatography–electrospray ionization mass
third-order separation data to identify significant chemical spectrometry Anal. Chem. 78 6573–82
components of metabolite extracts Anal. Chem. [137] Luo B, Groenke K, Takors R, Wandrey C and Oldiges M
78 5068–75 2007 Simultaneous determination of multiple intracellular
[120] Pierce K M, Hope J L, Hoggard J C and Synovec R E 2006 metabolites in glycolysis, pentose phosphate pathway and
Principal component analysis based method to discover tricarboxylic acid cycle by liquid chromatography–mass
chemical differences in comprehensive two-dimensional spectrometry J. Chromatogr. A 1147 153–64
gas chromatography with time-of-flight mass spectrometry [138] Maruska A and Kornysova O 2006 Application of monolithic
(GCxGC-TOFMS) separations of metabolites in plant (continuous bed) chromatographic columns in
samples Talanta 70 797–804 phytochemical analysis J. Chromatogr. A 1112 319–30
[121] Mayadunne R, Nguyen T T and Marriott P J 2005 Amino [139] Shen Y F et al 2005 Automated 20 kpsi RPLC–MS and
acid analysis by using comprehensive two-dimensional MS/MS with chromatographic peak capacities of
gas chromatography Anal. Bioanal. Chem. 382 836–47 1000–1500 and capabilities in proteomics and
[122] Vlaeminck B, Harynuk J, Korkiasaari K, Fievez V and metabolomics Anal. Chem. 77 3090–100
Marriott P J 2006 Comprehensive two-dimensional gas [140] Tolstikov V V, Lommen A, Nakanishi K, Tanaka N and
chromatography (GCxGC) for the analysis of fatty acids Fiehn O 2003 Monolithic silica-based capillary
(FA) in milk J. Anim. Sci. 84 294–5 reversed-phase liquid chromatography/electrospray mass
[123] Swartz M E 2005 UPLC (TM): an introduction and review spectrometry for plant metabolomics Anal. Chem.
J. Liq. Chromatogr. Relat. Technol. 28 1253–63 75 6737–40
[124] Gaskell S J 1997 Electrospray: principles and practice [141] Plumb R S and Wilson I D 2004 High throughput and high
J. Mass Spectrom. 32 677–88 sensitivity LC/MS–OA–TOF and UPLC/TOF–MS for the
[125] Ding J et al 2007 Capillary LC coupled with high-mass identification of biomarkers of toxicity and disease using a
measurement accuracy mass spectrometry for metabolic metabonomics approach Abstr. Pap. Am. Chem. Soc.
profiling Anal. Chem. 79 6081–93 228 U164-U
[126] Boernsen K O, Gatzek S and Imbert G 2005 Controlled [142] Wilson I D et al 2005 High resolution ‘ultra performance’
protein precipitation in combination with chip-based liquid chromatography coupled to oa-TOF mass
nanospray infusion mass spectrometry. An approach for spectrometry as a tool for differential metabolic pathway
21
profiling in functional genomic studies J. Proteome Res. [159] Cooks R G, Ouyang Z, Takats Z and Wiseman J M 2006
4 591–8 Ambient mass spectrometry Science 311 1566–70
[143] Plumb R S et al 2006 Generation of ultrahigh peak capacity [160] Chen H W, Wortmann A and Zenobi R 2007 Neutral
LC separations via elevated temperatures and high linear desorption sampling coupled to extractive electrospray
mobile-phase velocities Anal. Chem. 78 7278–83 ionization mass spectrometry for rapid differentiation of
[144] Stoll D R, Cohen J D and Carr P W 2006 Fast, bilosamples by metabolomic fingerprinting J. Mass
comprehensive online two-dimensional high performance Spectrom. 42 1123–35
liquid chromatography through the use of high [161] Kauppila T J, Talaty N, Kuuranne T, Kotiaho T, Kostiainen R
temperature ultra-fast gradient elution reversed-phase and Cooks R G 2007 Rapid analysis of metabolites and
liquid chromatography J. Chromatogr. A 1122 123–37 drugs of abuse from urine samples by desorption
[145] Granger J, Plumb R, Castro-Perez J and Wilson I D 2005 electrospray ionization–mass spectrometry Analyst
Metabonomic studies comparing capillary and 132 868–75
conventional HPLC-oa-TOF MS for the analysis of urine [162] Pan Z Z et al 2007 Principal component analysis of urine
from Zucker obese rats Chromatographia 61 375–80 metabolites detected by NMR and DESI-MS in patients
[146] Bottcher C, von Roepenack-Lahaye E, Willscher E, Scheel D with inborn errors of metabolism Anal. Bioanal. Chem.
and Clemens S 2007 Evaluation of matrix effects in 387 539–49
metabolite profiling based on capillary liquid [163] Fraser P D, Enfissi E M A, Goodfellow M, Eguchi T and
chromatography electrospray ionization quadrupole Bramley P M 2007 Metabolite profiling of plant
time-of-flight mass spectrometry Anal. Chem. 79 1507–13 carotenoids using the matrix-assisted laser desorption
[147] Dunn W B, Overy S and Quick W P 2005 Evaluation of ionization time-of-flight mass spectrometry Plant J.
automated electrospray–TOF mass spectrometry for 49 552–64
metabolic fingerprinting of the plant metabolome [164] Vaidyanathan S, Gaskell S and Goodacre R 2006
Metabolomics 1 137–48 Matrix-suppressed laser desorption/ionisation mass
[148] Al-Dirbashi O Y, Santa T, Al-Qahtani K, Al-Amoudi M and spectrometry and its suitability for metabolome analyses
Rashed M S 2007 Analysis of organic acid markers Rapid Commun. Mass Spectrom. 20 1192–8
relevant to inherited metabolic diseases by [165] Vaidyanathan S and Goodacre R 2007 Quantitative detection
ultra-performance liquid chromatography/tandem mass of metabolites using matrix-assisted laser desorption/
spectrometry as benzofurazan derivatives Rapid Commun. ionization mass spectrometry with 9-aminoacridine as the
Mass Spectrom. 21 1984–90 matrix Rapid Commun. Mass Spectrom. 21 2072–8
[149] Barry S J et al 2003 Use of S-pentafluorophenyl [166] Sun G, Yang K, Zhao Z D, Guan S P, Han X L and
tris(2,4,6-trimethoxyphenyl) phosphonium acetate Gross R W 2007 Shotgun metabolomics approach for the
bromide and (4-hydrazino-4-oxobutyl) [tris(2,4,6- analysis of negatively charged water-soluble cellular
trimethoxyphenyl)]phosphonium bromide for the metabolites from mouse heart tissue Anal. Chem.
derivatization of alcohols, aldehydes and ketones for 79 6629–40
detection by liquid chromatography/electrospray mass [167] Vaidyanathan S et al 2007 Laser desorption/ionization mass
spectrometry Rapid Commun. Mass Spectrom. 17 484–97 spectrometry on porous silicon for metabolome analyses:
[150] Ellis D I, Dunn W B, Griffin J L, Allwood J W and influence of surface oxidation Rapid Commun. Mass
Goodacre R T 2007 Metabolic fingerprinting as a Spectrom. 21 2157–66
diagnostic tool Pharmacogenomics. 8 1243–66 [168] Kraj A, Dylag T, Gorecka-Drzazga A, Bargiel S, Dziuban J
[151] Breitling R, Ritchie S, Goodenowe D, Stewart M L and and Silberring J 2003 Desorption/ionization on silicon for
Barrett M P 2006 Ab initio prediction of metabolic small molecules: a promising alternative to MALDI TOF
networks using Fourier transform mass spectrometry data Acta Biochim. Pol. 50 783–7
Metabolomics 2 155–64 [169] Touboul D, Roy S, Germain D P, Chaminade P, Brunelle A
[152] Goodenowe D 2003 Comprehensive metabolomics using and Laprevote O 2007 MALDI-TOF and
FTMS technology: theory and applications Abstr. Pap. Am. cluster-TOF-SIMS imaging of Fabry disease biomarkers
Chem. Soc. 226 U151–U2 Int. J. Mass Spectrom. 260 158–65
[153] Oikawa A et al 2006 Clarification of pathway-specific [170] Burrell M M, Earnshaw C J and Clench M R 2007 Imaging
inhibition by Fourier transform ion cyclotron matrix assisted laser desorption ionization mass
resonance/mass spectrometry-based metabolic spectrometry: a technique to map plant metabolites
phenotyping studies Plant Physiol. 142 398–413 within tissues at high spatial resolution J. Exp. Bot.
[154] Tohge T et al 2005 Functional genomics by integrated 58 757–63
analysis of metabolome and transcriptome of Arabidopsis [171] McDonnell L A and Heeren R M A 2007 Imaging mass
plants over-expressing an MYB transcription factor Plant spectrometry Mass Spectrom. Rev. 26 606–43
J. 42 218–35 [172] Northen T R et al 2007 Clathrate nanostructures for mass
[155] Breitling R, Pitt A R and Barrett M P 2006 Precision spectrometry Nature 449 1033-U3
mapping of the metabolome Trends Biotechnol. 24 543–8 [173] Hirai M Y et al 2007 Omics-based identification of
[156] Southam A D, Payne T G, Cooper H J, Arvanitis T N and Arabidopsis Myb transcription factors regulating aliphatic
Viant M R 2007 Dynamic range and mass accuracy of glucosinolate biosynthesis Proc. Natl Acad. Sci. USA
wide-scan direct infusion nanoelectrospray Fourier 104 6478–83
transform ion cyclotron resonance mass [174] Lafaye A et al 2005 Combined proteome and
spectrometry-based metabolomics increased by the metabolite-profiling analyses reveal surprising insights
spectral stitching method Anal. Chem. 79 4595–602 into yeast sulfur metabolism J. Biol. Chem. 280 24723–30
[157] Lane A N, Fan T W and Higashi R M 2008 Isotopomer-based [175] Corcoran O and Spraul M 2003 LC–NMR–MS in drug
metabolomic analysis by NMR and mass spectrometry discovery Drug Discov. Today 8 624–31
Methods Cell Biol. 84 541–88 [176] Godejohann M 2007 Hydrophilic interaction chromatography
[158] Mashego M R et al 2004 MIRACLE: mass isotopomer ratio coupled to nuclear magnetic resonance spectroscopy and
analysis of U-C-13-labeled extracts. A new method for mass spectroscopy—a new approach for the separation and
accurate quantification of changes in concentrations of identification of extremely polar analytes in bodyfluids
intracellular metabolites Biotechnol. Bioeng. 85 620–8 J. Chromatogr. A 1156 87–93
22
[177] Shockcor J P, Unger S E, Savina P, Nicholson J K and [196] Vaidyanathan S, Kell D B and Goodacre R 2002
Lindon J C 2000 Application of directly coupled Flow-injection electrospray ionization mass spectrometry
LC–NMR–M S to the structural elucidation of metabolites of crude cell extracts for high-throughput bacterial
of the HIV-1 reverse-transcriptase inhibitor B W935U83 identification J. Am. Soc. Mass Spectrom. 13 118–28
J. Chromatogr. B 748 269–79 [197] Martins A M, Camacho D, Shuman J, Sha W, Mendes P and
[178] Sumner L W et al 2007 Proposed minimum reporting Shulaev V 2004 A systems biology study of two distinct
standards for chemical analysis Metabolomics. 3 211–21 growth phases of Saccharomyces cerevisiae cultures Curr.
[179] Ma H W et al 2007 The Edinburgh human metabolic network Genomics 5 649–63
reconstruction and its functional analysis Mol. Syst. [198] Villas-Boas S G, Akesson M and Nielsen J 2005 Biosynthesis
Biol. 3 135 of glyoxylate from glycine in Saccharomyces cerevisiae
[180] Weininger D 1988 Smiles, a chemical language and FEMS Yeast Res. 5 703–9
information-system:1. Introduction to methodology and [199] Allen J et al 2003 High-throughput classification of yeast
encoding rules J. Chem. Inf. Comput. Sci. 28 31–6 mutants for functional genomics using metabolic
[181] Sansone S-A et al 2007 Metabolomics standards initiative: footprinting Nat. Biotechnol. 21 692–6
ontology working group work in progress Metabolomics. [200] Allen J, Davey H M, Broadhurst D, Rowland J J, Oliver S G
3 249–56 and Kell D B 2004 Discrimination of modes of action of
[182] Gika H G, Theodoridis G A, Wingate J E and Wilson I D antifungal substances by use of metabolic footprinting
2007 Within-day reproducibility of an HPLC–MS-based Appl. Environ. Microbiol. 70 6157–65
method for metabonomic analysis: application to human [201] Mas S, Villas-Boas S G, Hansen M E, Akesson M and
urine J. Proteome Res. 6 3291–303 Nielsen J 2007 A comparison of direct infusion MS and
[183] Sangster T P, Wingate J E, Burton L, Teichert F and GC–MS for metabolic footprinting of yeast mutants
Wilson I D 2007 Investigation of analytical variation in Biotechnol. Bioeng. 96 1014–22
metabonomic analysis using liquid chromatography/mass [202] Bolling C and Fiehn O 2005 Metabolite profiling of
spectrometry Rapid Commun. Mass Spectrom. Chlamydomonas reinhardtii under nutrient deprivation
21 2965–70 Plant Physiol. 139 1995–2005
[184] Sangster T, Major H, Plumb R, Wilson A J and Wilson I D [203] Strelkov S, von Elstermann M and Schomburg D 2004
2006 A pragmatic and readily implemented quality control Comprehensive analysis of metabolites in
strategy for HPLC–MS and GC–MS-based metabonomic Corynebacterium glutamicum by gas chromatography/
analysis Analyst 131 1075–8 mass spectrometry Biol. Chem. 385 853–61
[185] Carrari F and Fernie A R 2006 Metabolic regulation [204] Beckmann M, Enot D P, Overy D P and Draper J 2007
underlying tomato fruit development J. Exp. Bot. Representation, comparison, and interpretation of
57 1883–97 metabolome fingerprint data for total composition analysis
[186] Spasic I et al 2006 MeMo: a hybrid SQL/XML approach to and quality trait investigation in potato cultivars J. Agric.
metabolomic data management for functional genomics Food Chem. 55 3444–51
BMC Bioinformatics 7 [205] Enot D P, Beckmann M and Draper J 2007 Detecting a
[187] Lu H, Dunn W B, Shen H, Kell D B and Liang Y 2008 difference—assessing generalisability when modelling
Comparative evaluation of software for deconvolution of metabolome fingerprint data in longer term studies of
metabolomics data based on GC–TOF–MS Trends Anal. genetically modified plants Metabolomics 3 335–47
Chem. at press [206] Gullberg J, Jonsson P, Nordstrom A, Sjostrom M and Moritz
[188] Jonsson P et al 2004 A strategy for identifying differences in T 2004 Design of experiments: an efficient strategy to
large series of metabolomic samples analyzed by GC/MS identify factors influencing extraction and derivatization of
Anal. Chem. 76 1738–45 Arabidopsis thaliana samples in metabolomic studies with
[189] Shen P, Kang Y F and Cheng Y 2005 Pattern feature gas chromatography/mass spectrometry Anal. Biochem.
discovery for metabonomics of breast cancer and 331 283–95
HPLC/MS/MS analysis of characteristic metabolites [207] Boccard J et al 2007 Multivariate data analysis of rapid
Chem. J. Chin. Univ.-Chin. 26 1798–802 LC–TOF/MS experiments from Arabidopsis thaliana
[190] Styczynski M P, Moxley J F, Tong L V, Walther J L, stressed by wounding Chemometrics Intell. Lab Syst.
Jensen K L and Stephanopoulos G N 2007 Systematic 86 189–97
identification of conserved metabolites in GC/MS data for [208] Gibon Y et al 2006 Integration of metabolite with transcript
metabolomics and biomarker discovery Anal. Chem. and enzyme activity profiling during diurnal cycles in
79 966–73 Arabidopsis rosettes Genome Biol. 7
[191] Thysell E et al 2007 Reliable profile detection in comparative [209] Bino R J et al 2005 The light-hyperresponsive high
metabolomics Omics 11 209–24 pigment-2(dg) mutation of tomato: alterations in the fruit
[192] Nicholson J K, Lindon J C and Holmes E 1999 metabolome New Phytol. 166 427–38
‘Metabonomics’: understanding the metabolic responses [210] Moco S et al 2006 A liquid chromatography–mass
of living systems to pathophysiological stimuli via spectrometry-based metabolome database for tomato Plant
multivariate statistical analysis of biological NMR Physiol. 141 1205–18
spectroscopic data Xenobiotica 29 1181–9 [211] Overy S A et al 2005 Application of metabolite profiling to
[193] Hansen M A E and Smedsgaard J 2007 Automated work-flow the identification of traits in a population of tomato
for processing high-resolution direct infusion electrospray introgression lines J. Exp. Bot. 56 287–96
ionization mass spectral fingerprints Metabolomics [212] Broeckling C D et al 2005 Metabolic profiling of Medicago
3 41–54 truncatula cell cultures reveals the effects of biotic and
[194] Smedsgaard J and Frisvad J C 1996 Using direct electrospray abiotic elicitors on metabolism J. Exp. Bot. 56 323–36
mass spectrometry in taxonomy and secondary metabolite [213] Vikram A, Prithiviraj B, Hamzehzarghani H and Kushalappa
profiling of crude fungal extracts J. Microbiol. Methods A 2004 Volatile metabolite profiling to discriminate
25 5–17 diseases of McIntosh apple inoculated with fungal
[195] Smedsgaard J and Nielsen J 2005 Metabolite profiling of pathogens J. Sci. Food Agric. 84 1333–40
fungi and yeast: from phenotype to metabolome by MS [214] Dauwe R et al 2007 Molecular phenotyping of
and informatics J. Exp. Bot. 56 273–86 lignin-modified tobacco reveals associated changes in
23
cell-wall metabolism, primary metabolism, stress ovarian carcinomas and ovarian borderline tumors Cancer
metabolism and photorespiration Plant J. 52 263–85 Res. 66 10795–804
[215] Chang W T et al 2006 Effects of growth conditions and [232] Zhao X J, Wang W Z, Wang J S, Yang J and Xu G W 2006
processing on Rehmannia glutinosa using fingerprint Urinary profiling investigation of metabolites with cis-diol
strategy Planta Med. 72 458–67 structure from cancer patients based on UPLC–MS and
[216] Chan E C Y, Yap S L, Lau A J, Leow P C, Toh D F and HPLC–MS as well as multivariate statistical analysis
Koh H L 2007 Ultra-performance liquid chromatography/ J. Sep. Sci. 29 2444–51
time-of-flight mass spectrometry based metabolomics of [233] Yin P Y, Zhao X J, Li Q R, Wang J S, Li J S and Xu G W
raw and steamed Panax notoginseng Rapid Commun. 2006 Metabonomics study of intestinal fistulas based on
Mass Spectrom. 21 519–28 ultraperformance liquid chromatography coupled with
[217] Fiehn O 2003 Metabolic networks of Cucurbita maxima Q-TOF mass spectrometry (UPLC/Q-TOF MS)
phloem Phytochemistry 62 875–86 J. Proteome Res. 5 2135–43
[218] Jiang H L, Somogyi A, Timmermann B N and Gang D R [234] Plumb R S et al 2006 The detection of phenotypic differences
2006 Instrument dependence of electrospray ionization in the metabolic plasma profile of three strains of Zucker
and tandem mass spectrometric fragmentation of the rats at 20 weeks of age using ultra-performance liquid
gingerols Rapid Commun. Mass Spectrom. 20 3089–100 chromatography/orthogonal acceleration time-of-flight
[219] Kachlicki P, Marczak L, Kerhoas L, Einhorn J and mass spectrometry Rapid Commun. Mass Spectrom.
Stobiecki M 2005 Profiling isoflavone conjugates in root 20 2800–6
extracts of lupine species with LC/ESI/MSn systems [235] Williams R E, Lenz E M, Lowden J S, Rantalainen M and
J. Mass Spectrom. 40 1088–103 Wilson I D 2005 The metabonomics of aging and
[220] Allwood J W, Ellis D I, Heald J K, Goodacre R and development in the rat: an investigation into the effect of
Mur L A J 2006 Metabolomic approaches reveal that age on the profile of endogenous metabolites in the urine
phosphatidic and phosphatidyl glycerol phospholipids are of male rats using H-1 NMR and HPLC–TOF MS Mol.
major discriminatory non-polar metabolites in responses Biosyst. 1 166–75
by Brachypodium distachyon to challenge by [236] Bajad S and Shulaev V 2007 Highly-parallel metabolomics
Magnaporthe grisea Plant J. 46 351–68 approaches using LC–MS2 for pharmaceutical and
[221] Kaderbhai N N, Broadhurst D I, Ellis D I, Goodacre R and environmental analysis Trends Anal. Chem. 26 625–36
Kell D B 2003 Functional genomics via metabolic [237] Dixon S P, Pitfield I D and Perrett D 2006 Comprehensive
footprinting: monitoring metabolite secretion by multi-dimensional liquid chromatographic separation in
Escherichia coli tryptophan metabolism mutants using biomedical and pharmaceutical analysis: a review Biomed.
FT-IR and direct injection electrospray mass spectrometry Chromatogr. 20 508–29
Comp. Funct. Genom. 4 376–91 [238] Guttman A, Varoglu M and Khandurina J 2004
[222] Koulman A, Tapper B A, Fraser K, Cao M S, Lane G A and Multidimensional separations in the pharmaceutical arena
Rasmussen S 2007 High-throughput direct-infusion ion Drug Discov. Today 9 136–44
trap mass spectrometry: a new method for metabolomics [239] Ni Y et al 2007 Metabolic profiling using combined GC–MS
Rapid Commun. Mass Spectrom. 21 421–8 and LC–MS provides a systems understanding of
[223] Baker J M et al 2006 A metabolomic study of substantial aristolochic acid-induced nephrotoxicity in rat FEBS Lett.
equivalence of field-grown genetically modified wheat 581 707–11
Plant Biotechnol. J. 4 381–92 [240] Pohjanen E et al 2006 Statistical multivariate metabolite
[224] Catchpole G S et al 2005 Hierarchical metabolomics profiling for aiding biomarker pattern detection and
demonstrates substantial compositional similarity between mechanistic interpretations in GC/MS based
genetically modified and conventional potato crops Proc. metabolomics Metabolomics 2 257–68
Natl Acad. Sci. USA 102 14458–62 [241] Katz J E, Dumlao D S, Clarke S and Hau J 2004 A new
[225] Goodacre R, Vaidyanathan S, Bianchi G and Kell D B 2002 technique (COMSPARI) to facilitate the identification of
Metabolic profiling using direct infusion electrospray minor compounds in complex mixtures by GC/MS and
ionisation mass spectrometry for the characterisation of LC/MS: tools for the visualization of matched datasets
olive oils Analyst 127 1457–62 J. Am. Soc. Mass Spectrom. 15 580–4
[226] Chace D H and Kalas T A 2005 A biochemical perspective [242] Baran R et al 2006 MathDAMP: a package for differential
on the use of tandem mass spectrometry for newborn analysis of metabolite profiles BMC Bioinformatics 7 530
screening and clinical testing Clin. Biochem. 38 296–309 [243] Broeckling C D, Reddy I R, Duran A L, Zhao X C and
[227] Naylor E W and Chace D H 1999 Automated tandem mass Sumner L W 2006 MET-IDEA: data extraction tool for
spectrometry for mass newborn screening for disorders in mass spectrometry-based metabolomics Anal. Chem.
fatty acid, organic acid, and amino acid metabolism J. 78 4334–41
Child Neurol. 14 S4–8 [244] Duran A L, Yang J, Wang L J and Sumner L W 2003
[228] Rashed M S 2001 Clinical applications of tandem mass Metabolomics spectral formatting, alignment and
spectrometry: ten years of diagnosis and screening for conversion tools (MSFACTs) Bioinformatics 19 2283–93
inherited metabolic diseases J. Chromatogr. B 758 27–48 [245] Katajamaa M, Miettinen J and Oresic M 2006 MZmine:
[229] Shigematsu Y et al 1999 Modifications in electrospray toolbox for processing and visualization of mass
tandem mass spectrometry for a neonatal-screening pilot spectrometry based molecular profile data Bioinformatics
study in Japan J. Chromatogr. B 731 97–103 22 634–6
[230] Yang J et al 2006 High performance liquid [246] Katajamaa M and Oresic M 2005 Processing methods for
chromatography–mass spectrometry for metabonomics: differential analysis of LC/MS profile data BMC
potential biomarkers for acute deterioration of liver Bioinformatics 6 179
function in chronic hepatitis B J. Proteome Res. [247] Smith C A, Want E J, O’Maille G, Abagyan R and Siuzdak G
5 554–61 2006 XCMS: processing mass spectrometry data for
[231] Denkert C et al 2006 Mass spectrometry-based metabolic metabolite profiling using nonlinear peak alignment,
profiling reveals different metabolite patterns in invasive matching, and identification Anal. Chem. 78 779–87
24

Dunn 2008

Uploaded by

Copyright:

Available Formats

Dunn 2008

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

Dunn 2008

Uploaded by

Copyright:

Available Formats

Home Search Collections Journals About Contact us My IOPscience

mammalian and plant metabolomes

2008 Phys. Biol. 5 011001

Please note that terms and conditions apply.

Current trends and future requirements

Received 19 December 2007

Introduction of biological systems, and the interactions of these different

1478-3975/08/011001+24$30.00 1 © 2008 IOP Publishing Ltd Printed in the UK

Table 1. Terms commonly employed in metabolomics and related disciplines.

SAMPLE MASS DETECTION

Table 3. Typical examples of applications of mass spectrometry in metabolomics.

You might also like