Journal of Chromatography B, 871 (2008) 236–242
Contents lists available at ScienceDirect
Journal of Chromatography B
journal homepage: www.elsevier.com/locate/chromb
Analytical strategies for LC–MS-based targeted metabolomics?
Wenyun Lu, Bryson D. Bennett, Joshua D. Rabinowitz∗
Lewis-Sigler Institute for Integrative Genomics and Department of Chemistry, Princeton University,
Princeton, NJ 08544, United States
a r t i c l ei n f o
Received 8 February 2008
Accepted 22 April 2008
Available online 29 April 2008
Hydrophilic interaction chromatography
Heated electrospray ionization
Multiple reaction monitoring (MRM)
a b s t r a c t
Recent advances in mass spectrometry are enabling improved analysis of endogenous metabolites. Here
we discuss several issues relevant to developing liquid chromatography–electrospray ionization-mass
spectrometry methods for targeted metabolomics (i.e., quantitative analysis of dozens to hundreds of
specific metabolites). Sample preparation and liquid chromatography approaches are discussed, with an
unheated ESI. Applicability to targeted metabolomics of triple quadrupole mass spectrometry operating
in multiple reaction monitoring (MRM) mode and high mass resolution full scan mass spectrometry (e.g.,
time-of-flight, Orbitrap) are described. We suggest that both are viable solutions, with MRM preferred
targeted and untargeted metabolomics.
© 2008 Elsevier B.V. All rights reserved.
Metabolomics, in its most ambitious global form, tries to com-
prehensively analyze all known and unknown metabolites in a
given biological sample . Targeted metabolomics has the more
modest goal of quantitating selected metabolites, most typically
dozens to hundreds of known compounds. This requires the ability
to differentiate the targeted analytes from other interfering com-
magnetic resonance (NMR) spectrum, mass-to-charge ratio on a
mass spectrometer (MS), retention time in chromatography, or a
combination thereof. NMR has several notable advantages relative
to MS—it can deal with the biofluids without the need for sample
preparation and it produces signals that correlate directly and lin-
early with compound abundance . However, NMR has relatively
generally be detected . On the other hand, mass spectrometry,
Metabolite Profiling”, guest edited by Georgios Theodoridis and Ian D. Wilson.
∗Corresponding author at: Lewis-Sigler Institute for Integrative Genomics, 241
Carl Icahn Laboratory, Washington Road, Princeton University, Princeton, NJ 08544,
United States. Tel.: +1 609 258 8985; fax: +1 609 258 3565.
E-mail address: email@example.com (J.D. Rabinowitz).
when combined with effective sample preparation and chromato-
graphic separation, has high sensitivity and specificity, as well as
good dynamic range [4–7].
There has been tremendous progress in mass spectrometry-
based metabolomics recently, leaving researchers with a variety
of choices for chromatographic separation, ionization, and mass
spectrometric analysis. Separations may be achieved by gas chro-
matography (GC) , capillary electrophoresis (CE) , or liquid
chromatography (LC) , with LC approaches continuously evolv-
ing (e.g., to include capillary monolithic chromatography [11–14]
and ultra performance liquid chromatography [13,15–18]). In con-
junction with liquid chromatography, ionization may be achieved
using electrospray ionization (ESI), atmosphere pressure chem-
ical ionization (APCI), or atmospheric pressure photoionization
(APPI) . Mass spectrometer options include quadrupoles and
ion traps which offer good sensitivity but limited resolving power
, or higher mass resolution instruments such as time-of-
flight (TOF) [21,22], Fourier transform ion cyclotron resonance
(FTICR) [23,24] or Orbitrap [25,26]. The mass spectrometer can
also be arranged in a tandem configuration, such as a triple
quadrupole mass spectrometer. Different types of analyzer can
also be combined to form a hybrid mass spectrometer , such
as a quadrupole-TOF (Q-TOF) instrument or an ion trap-Orbitrap
(currently commercially available solely as the LTQ-Orbitrap from
Thermo Fischer Scientific).
1570-0232/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
W. Lu et al. / J. Chromatogr. B 871 (2008) 236–242
This brief review focuses on targeted metabolomics using liq-
uid chromatography–electrospray ionization-mass spectrometry
(LC–ESI-MS). We do not intend to give a comprehensive review
of all aspects of metabolomics (upon which there are already a
number of excellent reviews [5,27–32]), nor do we directly address
issues relating to absolute (rather than relative) metabolite quan-
titation (although the concepts described herein, when combined
with either an external standard curve or isotope-labeled internal
standards, can be used to provide absolute quantitation [33–35]).
issues: the choices of sample preparation procedure, chromato-
graphic separation methods, electrospray ionization approach
(heated versus unheated), and mass spectrometry technology.
2. Basics of sample preparation
While urine is sufficiently rich in metabolites to enable direct
analysis (often after diluting in a volume of water selected to main-
tain a fixed salt concentration across samples), in most biological
samples, metabolites of interest comprise a small minority of the
starting material. Accordingly, prior to metabolomic analysis, some
enrichment for metabolites is generally desirable. This processing
step should ideally quantitatively retain metabolites of interest,
which may span a wide range of chemical properties. It should also
degradation or interconversion. For serum or plasma, precipitation
of protein using cold methanol is a simple but effective method
For analysis of metabolites from cells or tissues, a concern is
logically important metabolites, such as ATP or glutamine, can be
on the order of a few seconds (and sometimes less), and seemingly
to dramatic alterations in intracellular metabolite levels. Accord-
ingly, rapid cooling of the cells, e.g., by mixing of cultured cells
with cold organic solvent , or grabbing of tissue samples using
liquid nitrogen-temperature tongs [38,39] is commonly employed
to stop metabolism. Alternatively, rapid quenching of enzymatic
activity via heat or acid denaturation can be used, if the analytes of
interest are adequately heat or acid stable .
Once metabolism is quenched, metabolites must be released
from cells. The most appropriate means of extraction depends
on both the sample type and the targeted metabolites of great-
est interest. For example, extraction of nucleotide triphosphates
from Escherichia coli or yeast, but not mammalian cells, is
enhanced in mixtures of acetonitrile:methanol:water relative to
methanol:water is adequate and results in cleaner samples than
acetonitrile:methanol:water, which less fully precipitates protein
and lipids. For tissue samples, many protocols involve freezing of
Once an extract is obtained, care must be taken to preserve its
integrity prior to LC–MS analysis. For cell or tissue samples, which
this can be a major concern. In our laboratory, we typically store
extracts at 4◦C and analyze them within 24h of their generation,
and yet more rapidly if particularly labile species like folates are
3. Chromatographic separation
Metabolomics deals with a diversity of small molecules that
differ greatly in their physical and chemical properties of size,
polarity/hydrophobicity, and charge. While no single chromato-
graphic method is ideal for all classes of metabolites, we have
found that two methods—one for positive ionization mode and
one for negative ionization mode (described below), provide a
reasonable breadth of coverage. The route by which we arrived
at this approach exemplifies some of the tradeoffs in targeted
Our initial approach was a reversed-phase chromatography
method, on a C18 Fusion-RP (Phenomenex) column with acidic
mobile phase . This method worked adequately for ∼100 com-
pounds in positive ion mode. However, many polar compounds
did not retain on this column, eluting near the void volume, and
nucleotide triphosphate compounds like ATP did not elute as well
different chromatography approaches involving seven different
column chemistries . This study identified hydrophilic inter-
action chromatography (HILIC) [43–47] on an aminopropyl column
lites including amino acids, nucleosides, nucleotides, coenzyme A
derivatives, carboxylic acids, and sugar phosphates. By separating
compounds into a 40-min positive ion run and a 50-min negative
ion run, a total of ∼140 compounds can be readily measured with
a total running time of 90min, using only one mass spectrometer.
The same column chemistry can also be used to analyze additional
classes of molecules, such as folates, which themselves include >50
chemical species . This method is a reasonable alternative for
those occasions where limited mass spectrometry resources are
Recent reports have demonstrated that reversed-phase chro-
matography with an amine ion-pairing agent is a useful method
for separation of a broad range of negatively charged metabo-
lites, including nucleotides, sugar phosphates, and carboxylic acids
[49–52]. These methods utilize a volatile cationic compound,
such as tributylamine  or hexylamine , to form ion pairs
with negatively charged analytes, improving retention and sepa-
ration on a C18 column. We performed a systematic comparison
between our HILIC method and a variant of the reversed-phase,
ion-pairing method of Luo et al. , using both compound stan-
dards and cellular extracts under identical mass spectrometry
conditions. The ion-pairing chromatography in general offered
better separation and higher signal for negatively charged metabo-
lites (Fig. 1). This improved sensitivity may in part be due to
improved separation leading to reduced ion suppression by co-
eluting compounds. We found that this method generally did not
work well in positive ionization mode, due to poor retention of
amine-containing compounds and ion suppression effects by trib-
Based on this study, we arrived at the dual chromatogra-
phy method approach of HILIC chromatography in conjunction
with positive mode ionization, and reversed-phase chromatog-
raphy with tributylamine as an ion-pairing agent in conjunction
with negative mode ionization, running on two separate LC–MS
systems. The LC conditions for the positive mode have been
previously reported : an aminopropyl column with acetoni-
trile and pH 9.45 aqueous buffer as the mobile phases and a
running time of 40min. The negative mode LC method uses a
Synergi Hydro column (4?m particle size, 150mm×2mm, from
Phenomenex, Torrance, CA), with solvent A being 10mM tributy-
lamine+15mM acetic acid in 97:3 water:methanol, and solvent
B being 100% methanol. The flow rate is 200?L/min and run-
ning time is 50min. The gradient is t=0min, 0% B; t=5min,
0% B; t=10min, 20% B; t=20min, 20% B; t=35min, 65% B;
t=38min, 95% B; t=42min, 95% B; t=43min, 0% B; t=50min,
0% B. This dual chromatography methodology enables quantita-
tion of approximately 250 water-soluble metabolites of validated
W. Lu et al. / J. Chromatogr. B 871 (2008) 236–242
Fig. 1. Chromatographic traces for selected metabolites in a cellular extract
comparing the performance for the HILIC (top panel) and reversed-phase ion-
pairing chromatographic (bottom panel) methods in negative ionization mode,
under identical mass spectrometry conditions. Selected metabolites are as follows:
glucose-6-phosphate and unresolved isomers thereof (a), glycerol-3-phosphate (b),
orotate (c), NAD+(d), pantothenate (e), succinate (f), malate (g), UDP-d-glucose
(h), fructose-1,6-biphosphate (i), phosphoenolpyruvate (j), NADH (k), and ATP (l).
The LC condition for the HILIC mode (aminopropyl column with basic pH mobile
the ion-pairing chromatography (reversed-phase Synergi Hydro column with acidic
mobile phase, running time 50min) can be found in the text. Note that metabolites
a, b, f, and g are not separated in HILIC method while they are well separated in
ion-pairing chromatography method.
identity including amino acids and derivatives, sugar phosphates,
nucleotides, coenzyme A and derivatives, and carboxylic acids,
with the number of quantifiable compound largely limited by the
availability of standards, which are used for confirmation of tar-
geted compound identity based on retention time and (for MS/MS
approaches) mass spectrometry fragmentation pattern. Due to
recent acquisition of several hundred additional standards from
the Human Metabolome Database , we anticipate generating
methods covering a yet greater number of validated known com-
In contrast to the success in coupling ion-pairing chromatog-
raphy with negative ionization mode, there have been a limited
number of studies exploring the possibility of coupling ion pair-
ing with positive ionization mode. Because amino ion-pairing
agents work well only with negatively charged compounds,
a different kind of ion-pairing agent is needed for positively
charged metabolites. Volatile perfluorinated acids such as tri-
fluoroacetic acid (TFA) have been used in the separation of
peptides . Recently Hsieh and Duncan reported the use of
two similar ion-pairing agents, heptafluorobutyric acid and non-
afluoropentanoic acid in the mobile phase to improve retention
and separation on a reversed-phase column, using APCI for ion-
ization . Unfortunately, these volatile perfluorinated acids
may cause ion suppression in ESI . More work is needed to
optimize ion-pairing agents for positively charged metabolites in
A variety of additional alternative analytical techniques
for metabolomics have been reported in the literature. Many
researchers used multiple analytical platforms in parallel in order
to detect as many compounds as possible. For examples, Sabatine
et al. used three different chromatographic modes for the analysis
of plasma samples . Amino acids and amines were separated
on a reversed-phase column at mobile phase pH 4, sugars and
nucleotides by normal phase chromatography at mobile phase pH
11, and organic acids by reversed-phase chromatography at pH 6. A
through 6 analytical methods for each sample. More recently, van
der Werf et al. developed a comprehensive metabolic platform
that utilizes three GC–MS methods and three LC–MS methods .
When applying this platform to the analysis of the metabolome of
E. coli, more than 400 compounds were detected.
In addition to using multiple chromatographic methods in
parallel, two-dimensional chromatography approaches have also
been described. In these methods the two chromatographic
separations occur serially, and the chromatographic modes are
orthogonal—they separate based upon different characteristics,
in a manner similar to a 2D-nanoLC method, now widely used
in proteomics for the separation of peptides . Kennedy and
colleagues developed a two-dimensional liquid chromatography
(2D-LC) method for metabolomics . In this approach, the sam-
ples were first separated on a strong anion exchange column,
with fractions released to the reversed-phase for further separa-
tion by pulses of incrementally higher ionic strength. When this
method was applied to the analysis of islets of Langerhans, roughly
400 peaks were detected [60,61]. The main disadvantage of this
approach is the long analysis time, as each fraction from the ion
exchange column must be separated sequentially via reversed-
Other notable developments in LC include the use of monolithic
capillary columns [11–14], high temperature LC [62–64], and ultra
performance liquid chromatography (UPLC, i.e., pressure >400bar
to drive flow through columns packed with <2?m diameter parti-
cles [13,15–18]). Recently, Guillarme et al. systematically evaluated
each of these three approaches compared to conventional LC .
They found that each of the approaches provided at least compara-
ble quantitative precision and accuracy to conventional LC while
also expediting analysis, with the greatest gains obtained with
On occasions in which metabolites of interests are at very low
abundance or have poor ionization, it is often possible to increase
the signal by derivatizing the compound(s) of interest, prior to
running the sample by LC–MS. Our laboratory routinely applies
several derivatization procedures for this purpose. These include
the derivatization of thiol compounds such as cysteine and homo-
cysteine which usually do not ionize very well, using S-methyl
methanethiosulfonate (MMTS) . The derivatized compounds
give substantially improved signal using the HILIC method in pos-
itive ionization mode. Similarly, reaction of amino acids with the
benzylchloroformate gives improved retention in reversed-phase
mode and negative mode ionization. Many other derivatizations
have been reported in the literature [29,58,66,67] and could be
used to enhance the signal for specific classes of compounds at
the expense of potentially impairing quantitative reliability (e.g.,
due to incomplete reactions or side reactions). For enhancing
class-specific signals, a particularly promising approach involves
and analyzed .
W. Lu et al. / J. Chromatogr. B 871 (2008) 236–242
4. Unheated electrospray versus heated electrospray
Metabolites eluting from the chromatographic column enter
the source region of the mass spectrometer where they are ion-
ized by electrospray ionization. While ESI source designs can vary
in many respects, one important distinction is between unheated
and heated ESI. Thermo Fisher Scientific’s TSQ Quantum series of
an unheated or a heated ESI source (HESI). In the unheated ESI
source, the sample solution exits the tip of a metal needle to form
a fine spray, with nitrogen at ambient temperature as the sheath
gas and auxiliary gas to assist ionization. The HESI source is simi-
lar, but includes a heating device in the ESI probe body that heats
the auxiliary gas to temperatures between 200 and 600◦C. When
the coaxial flow of heated nitrogen gas intersects with the charged
droplets emitted from the metal needle, these droplets evaporate
more rapidly than using ambient nitrogen, thereby increasing the
API 4000 instrument also includes a heating element.
Recently, we performed a small experiment to investigate
the utility of HESI for targeted quantitative metabolomics. We
tested both an E. coli cellular extract and 20 metabolite stan-
dards on two different mass spectrometers, Thermo Scientific TSQ
Quantum Ultra instruments with unheated or heated ESI sources
and an AppliedBiosystems API 4000 instrument with Turbo VTM
mode (glycine, serine, proline, glutamine, glutamate, methionine,
arginine, AMP, IMP, and riboflavin), and 10 compounds in neg-
ative ionization mode (threonine, succinate, malate, orotic acid,
phosphoenolpyruvate, glycerol-3-phosphate, 3-phosphoglycerate,
citrate, glucose-6-phosphate, and NAD+). The standards were
diluted to concentrations of 5, 50, 500 and 5000ng/mL. The LC
conditions were as previously reported for the HILIC aminopropyl
column, with same LC gradient applied for both positive and neg-
ative modes . Compounds were detected in multiple reaction
concentrations enabled approximate evaluation of sensitivity (i.e.,
limit of detection defined as the lowest concentration at which the
pound standards at 500ng/mL, as well as a cellular extract sample,
were injected four times to evaluate reproducibility, defined as the
from the four injections.
In general, HESI provided enhanced sensitivity compared with
the unheated ESI source on the TSQ Quantum Ultra. Most of the 20
also increased in most cases, the overall signal-to-noise ratios were
a 25-fold increase in terms of absolute ion counts for HESI versus
unheated ESI for a methionine standard at 50ng/mL, accompanied
by a less dramatic increase in noise. This increase in signal was
also seen for cell extract. The sensitivity benefit of a heated electro-
spray source was also evident in API 4000 instrument, which had
excellent overall performance.
in terms of the electrospray ionization mechanism. Electrospray
involves solvent molecules in the initial droplets evaporating
quickly to expose the charged analyte of interest . Ionization
surface tension and expedites solvent evaporation, improving neb-
ulization and desolvation, and increasing ion release from droplets
ization mode (a) and NAD+in a biological sample in negative mode (b), on identical
TSQ Quantum Ultra (Thermo Fisher Scientific) triple quadrupole instruments but
with different ion sources, the unheated electrospray source (solid line) and the
heated electrospray source (HESI, dashed line). The insets show the correspond-
ing noise/background region. The LC method was HILIC as reported previously .
The source conditions for the unheated ESI were as follows: spray voltage +3500 or
−3000V, sheath gas 30psi, auxiliary gas 10psi. The source conditions for the HESI
were as follows: spray voltage +2555 or −2555V, vaporizer temperature 230◦C,
sheath gas 22 psi, auxiliary gas 37psi.
compound-dependent. Not every tested compound showed a large
improvement in signal. This is illustrated in Fig. 2b, where we com-
NAD+from a cellular extract. For this compound, HESI does not
improve signal-to-noise, perhaps due to heat-induced decompo-
sition of some of the NAD+initially present. If certain compounds
decompose as a result of HESI, this may introduce quantitative arti-
facts. Thus, caution remains warranted in application of HESI to
labile metabolites until further testing is conducted.
5. Multiple reaction monitoring versus high resolution full
The triple quadrupole mass spectrometer has been the standard
workhorse in the quantitation of small molecules and metabolites,
as it offers good sensitivity, reproducibility, and a broad dynamic
range. For metabolomics, it is typically used in multiple reaction
monitoring mode, where the collision energy and product ion
mass-to-charge ratio are pre-optimized for each analyte of inter-
est to give the best signal. Instrumentally, three quadrupoles are
arranged in series. The first quadrupole selects the parent ion of
ion. When multiple metabolites are being analyzed, this process is
scan takes a certain amount of time (the scan or dwell time), each
W. Lu et al. / J. Chromatogr. B 871 (2008) 236–242
analyte is only monitored intermittently. Chromatographic peak
width divided by the time to cycle through the programmed set of
MRM scan events determines the number of data points available
to define a MRM chromatogram peak.
Current triple quadrupole mass spectrometers, such as Thermo
Fisher Scientific’s TSQ Quantum Ultra instrument and Applied-
Biosystems’ API 4000 instrument, typically work adequately with
a scan time of ∼50ms. For the simultaneous measurements of ∼50
which provides (for a chromatography peak width of 30s) ∼12
data points across the peak, enough to give reliable peak quanti-
tation . With narrower chromatographic peaks, such as those
produced by a UPLC, or a larger number of compound MRMs,
the time between data points will decrease the reliability of peak
quantitation. One possible solution to this problem is to reduce
the scan time; however, this can compromise signal-to-noise and
reproducibility. Indeed, preliminary testing on our instrument sug-
gests that while it is possible to reduce the scan time to 10ms,
the reproducibility is significantly worse than with a 50ms scan.
Although the effect of faster scanning may be instrument and
analyte-dependent, some decrement in performance is expected
is roughly proportional to the scan time. Less scan time means
fewer ions are detected, thus higher variation of the signal, with
the signal-to-noise ratio theoretically proportional to the square
root of scan time. This is a particularly serious problem for low
abundance ions. When using MRM, a more promising approach is
dividing the chromatography run into time segments, with MRMs
One alternative approach to using a triple quadrupole MS in
MRM mode is to perform full scan MS. Full scans on a low reso-
lution instrument, such as a single quadrupole mass spectrometer
(typical resolving power of 1000, as measured by m/?m, where
?m is full width at half maximum), while a viable option, are gen-
erally suboptimal for low molecular weight metabolites because of
potential interference from solvent ions, adducts, metabolites with
the signal from metabolite of interest in background noise impedes
an instrument with high mass resolution mitigates many of these
concerns, as mass-based separation will generally be achieved for
all compounds except isomers.
power (>500,000) [23,24,72]. However, time-of-flight (resolving
power 10,000) [21,22] and Thermo Fisher Scientific’s Orbitrap
(resolving power 100,000) [25,26] offer the advantages of lower
cost, greater ease of maintenance and use, and easier coupling to
LC by ESI. In addition, Orbitrap offers excellent mass accuracy, in
the range of 1–2ppm. It is typically coupled with a linear ion trap
to enable determination of the fragmentation pattern of ions of
lite identification. While the mass accuracy of TOF instruments has
historically been in the 5–10ppm range, technological advances in
recent years have shown that TOF can achieve a mass accuracy of
1–2ppm when internally calibrated , rendering them increas-
ingly competitive with Orbitrap on this dimension.
In order to determine what utility TOF-MS has in the field of tar-
geted metabolomics, we compared the performance of an Agilent
6220 TOF to a triple quadrupole instrument (Thermo Scientific TSQ
Quantum Ultra equipped with unheated ESI source) operating in
MRM mode. As before, 20 compound standards (10 in positive ion
and 5000ng/mL) and a cellular extract were run on both machines.
We evaluated the instrument sensitivity by comparing the signal-
to-noise ratio for the same compound at same concentration, and
the limit of detection (LOD) defined as the lowest concentration at
500ng/mL were run four times to evaluate reproducibility, defined
four injections. For the triple quadrupole MS, while only 10 MRM
MRM experiment, thereby somewhat penalizing the performance
with a scan time of 1s were performed.
The results showed that the TOF has good sensitivity and repro-
ducibility, and a reasonable dynamic range. All the standards had
four injections). Additionally, the TOF sensitivity was comparable
to the triple quadrupole with an unheated ESI source (the more
optimized triple quadrupole conditions of fewer MRM scans and
heated ESI source would presumably result in a substantial sen-
sitivity advantage for the triple quadrupole instrument, although
a direct comparison was not conducted). All compounds showed
a linear response over two to three orders of magnitude on the
TOF, as shown for glutamate in Fig. 3. These results indicated that
TOF, while not necessarily providing fully comparable quantitative
performance to MRM scanning, may have adequate quantitative
performance for many targeted metabolomics applications.
The scan time for TOF can be from 0.05 to 1s. The Orbitrap scan
time is mass resolution related, with a resolution of 7500 at a scan
time of 0.1s, 30,000 at a scan time of 0.4s, and 100,000 at a scan
time of 1.9s . Full scan MS approaches (TOF, Orbitrap) thus can
benefit from a relatively long scan time, compared to the need
for fast scanning to accommodate a large number of MRMs on a
triple quadrupole instrument. Longer scan times facilitate reliable
Fig. 3. Linearity test results for glutamate standards on an Agilent TOF instrument
(a) and a triple quadrupole instrument (b). Both show a linear response from 5 to
5000ng/mL with a R2>0.999. The LC method was HILIC as reported previously .
W. Lu et al. / J. Chromatogr. B 871 (2008) 236–242
quantitation, as it allows for detection of a greater number of ions.
Additionally, the high mass accuracy can effectively separate the
targeted compound from background ions having the same nom-
inal mass but a different exact mass. Overall, these features make
full scan approaches a promising alternative solution for targeted
metabolomics, as full scan approaches avoid the problems asso-
ciated with an excessive numbers of MRM scans and the need to
optimize collision energies and product ions for each analyte of
interest. Most importantly, full scan approaches provide the oppor-
tunity to marry targeted and untargeted metabolomics, where the
chemical identity of the metabolites is required for the former but
not for the latter. Many metabolites are currently unidentified or
not available as standards to purchase. One can envision mining
full scan MS data for both (1) quantitative information on targeted
analytes of interest and (2) other unanticipated peaks that show
large or highly statistically significant changes across biological
The biological utility of targeted metabolomic data ultimately
depends on the quality of the sample being analyzed, the choice of
the metabolites to target, and the selection of analytical modalities
to employ. Careful quenching and extraction is critical for cell and
tissue samples. Effective chromatographic separation is essential
when analyzing complex biological samples by LC–MS. A combi-
nation of HILIC chromatography in positive ionization mode and
reversed-phase ion-pairing chromatography in negative ionization
mode provides good coverage for a broad range of metabolites. For
certain classes of metabolites, derivatization facilitates separation
and ionization. In analyzing dozens of metabolites at once (up to
several hundred in a properly segmented LC run), multiple reac-
technique. For less targeted analyses (involving yet larger number
of analytes), and for marrying targeted with global metabolomics,
use of high mass resolution, high mass accuracy full scan MS holds
promise going forward.
This research was supported by NIH grant GM071508 for Center
of Quantitative Biology at Princeton University. Additional support
came from Beckman Foundation, NSF DDDAS grant CNS-0540181,
American Heart Association grant 0635188N, NSF Career Award
MCB-0643859, and NIH grant AI078063 (to J.D.R.). We thank Kath-
leen Anderson and Josef Ruzicka from Thermo Fisher Scientific,
Bob Walker and Jim Lau from Agilent, and James A. Ferguson
Sanders and the reviewers for comments on the manuscript.
 W.J. Griffiths, K. Karua, M. Hornshaw, G. Woffendin, Y. Wanga, Eur. J. Mass
Spectrom. 13 (2007) 45.
 I.A. Lewis, S.C. Schommer, B. Hodis, K.A. Robb, M. Tonelli, W.M. Westler, M.R.
Suissman, J.L. Markley, Anal. Chem. 79 (2007) 9385.
 E.J. Want, B.F. Cravatt, G. Siuzdak, Chembiochem 6 (2005) 1941.
 E. Kimball, J.D. Rabinowitz, Anal. Biochem. 358 (2006) 273.
 E.J. Want, A. Nordstrom, H. Morita, G. Siuzdak, J. Proteom. Res. 6 (2007) 459.
 J.D. Rabinowitz, Exp. Rev. Proteom. 4 (2007) 187.
 J.D. Rabinowitz, E. Kimball, Anal. Chem. 79 (2007) 6167.
 J. Lisec, N. Schauer, J. Kopka, L. Willmitzer, A.R. Fernie, Nat. Protoc. 1 (2006) 387.
 M.R.N. Monton, T. Soga, J. Chromatogr. A 1168 (2007) 237.
 S.U. Bajad, W.Y. Lu, E.H. Kimball, J. Yuan, C. Peterson, J.D. Rabinowitz, J. Chro-
matogr. A 1125 (2006) 76.
 V.V. Tolstikov, A. Lommen, K. Nakanishi, N. Tanaka, O. Fiehn, Anal. Chem. 75
 A. Maruska, O. Kornysova, J. Chromatogr. A 1112 (2006) 319.
 D. Guillarme, D.T.T. Nguyen, S. Rudaz, J.L. Veuthey, J. Chromatogr. A 1149 (2007)
 K. Horie, T. Ikegami, K. Hosoya, N. Saad, O. Fiehn, N. Tanaka, J. Chromatogr. A
1164 (2007) 198.
R.S. Plumb, J. Proteome Res. 4 (2005) 591.
 D.T.T. Nguyen, D. Guillarme, S. Rudaz, J.L. Veuthey, J. Sep. Sci. 29 (2006) 1836.
 M. Walles, C. Gauvin, P.E. Morin, R. Panetta, J. Ducharme, J. Sep. Sci. 30 (2007)
 K. Yu, D. Little, R. Plumb, B. Smith, Rapid Commun. Mass Spectrom. 20 (2006)
 S.S. Cai, L.C. Short, J.A. Syage, M. Potvin, J.M. Curtis, J. Chromatogr. A 1173 (2007)
 G.L. Glish, D.J. Burinsky, J. Am. Soc. Mass Spectrom. 19 (2008) 161.
 S. Lacorte, A.R. Fernandez-Albaz, Mass Spectrom. Rev. 25 (2006) 866.
18 (2007) 1612.
 S.D.H. Shi, C.L. Hendrickson, A.G. Marshall, Proc. Natl. Acad. Sci. U.S.A. 95 (1998)
ferty, J. Am. Chem. Soc. 121 (1999) 806.
 Q.Z. Hu, R.J. Noll, H.Y. Li, A. Makarov, M. Hardman, R.G. Cooks, J. Mass Spectrom.
40 (2005) 430.
 A. Makarov, E. Denisov, A. Kholomeev, W. Baischun, O. Lange, K. Strupat, S.
Horning, Anal. Chem. 78 (2006) 2113.
 W.B. Dunn, D.I. Ellis, Trends Anal. Chem. 24 (2005) 285.
 M. Oldiges, S. Lutz, S. Pflug, K. Schroer, N. Stein, C. Wiendahl, Appl. Microbiol.
Biotechnol. 76 (2007) 495.
 K. Dettmer, P.A. Aronov, B.D. Hammock, Mass Spectrom. Rev. 26 (2007) 51.
 S. Bajad, V. Shulaev, Trends Anal. Chem. 26 (2007) 625.
 M. Jemal, Y.Q. Xia, Curr. Drug Metab. 7 (2006) 491.
 M.R. Mashego, K. Rumbold, M. De Mey, E. Vandamme, W. Soetaert, J.J. Heijnen,
Biotechnol. Lett. 29 (2007) 1.
 D. Pigini, A.M. Cialdella, P. Faranda, G. Tranfo, Rapid Commun. Mass Spectrom.
20 (2006) 1013.
 P. Manini, R. Andreoli, W.M.A. Niessen, J. Chromatogr. A 1058 (2004) 21.
 M. Rychlik, S. Asam, Anal. Bioanal. Chem. 390 (2008) 617.
 E.J. Want, G. O’Maille, C.A. Smith, T.R. Brandon, W. Uritboonthai, C. Qin, S.A.
Trauger, G. Siuzdak, Anal. Chem. 78 (2006) 743.
 T.P. Ikeda, A.E. Shauger, S. Kustu, J. Mol. Biol. 259 (1996) 589.
30 (1970) 2049.
 R.C. Jackson, M.S. Lui, T.J. Boritzki, H.P. Morris, G. Weber, Cancer Res. 40 (1980)
 J. Schaub, C. Schiesling, M. Reuss, M. Dauner, Biotechnol. Progr. 22 (2006) 1434.
 J.D. Rabinowitz, B.D. Bennett, unpublished data.
 W.Y. Lu, E. Kimball, J.D. Rabinowitz, J. Am. Soc. Mass Spectrom. 17 (2006) 37.
 V.V. Tolstikov, O. Fiehn, Anal. Biochem. 301 (2002) 298.
 H. Idborg, L. Zamani, P.O. Edlund, I. Schuppe-Koistinen, S.P. Jacobsson, J. Chro-
matogr. B 828 (2005) 9.
 P. Hemstrom, K. Irgum, J. Sep. Sci. 29 (2006) 1784.
 S. Cubbon, T. Bradbury, J. Wilson, J. Thomas-Oates, Anal. Chem. 79 (2007) 8911.
 Y. Guo, S. Gaiki, J. Chromatogr. A 1074 (2005) 71.
 W.Y. Lu, Y.K. Kwon, J.D. Rabinowitz, J. Am. Soc. Mass Spectrom. 18 (2007) 898.
 J.H.J. Huck, E.A. Struys, N.M. Verhoeven, C. Jakobs, M.S. Van der Knaap, Clin.
Chem. 49 (2003) 1375.
 L. Coulier, R. Bas, S. Jespersen, E. Verheij, M.J. van der Werf, T. Hankemeier, Anal.
Chem. 78 (2006) 6573.
 B. Luo, K. Groenke, R. Takors, C. Wandrey, M. Oldiges, J. Chromatogr. A 1147
 R. Tuytten, F. Lemiere, W. Van Dongen, E.L. Esmans, H. Slegers, Rapid Commun.
Mass Spectrom. 16 (2002) 1205.
 D.S. Wishart, D. Tzur, C. Knox, R. Eisner, A.C. Guo, N. Young, D. Cheng, K. Jewell,
D. Arndt, S. Sawhney, C. Fung, L. Nikolai, M. Lewis, M.A. Coutouly, I. Forsythe, P.
Tang, S. Shrivastava, K. Jeroncic, P. Stothard, G. Amegbey, D. Block, D.D. Hau, J.
G.D. MacInnis, A.M. Weljie, R. Dowlatabadi, F. Bamforth, D. Clive, R. Greiner, L.
Li, T. Marrie, B.D. Sykes, H.J. Vogel, L. Querengesser, Nucleic Acids Res. 35 (2007)
 M. Shibue, C.T. Mant, R.S. Hodges, J. Chromatogr. A 1080 (2005) 68.
 Y. Hsieh, C.J.G. Duncan, Rapid Commun. Mass Spectrom. 21 (2007) 573.
 S.A. Gustavsson, J. Samskog, K.E. Markides, B. Langstrom, J. Chromatogr. A 937
F.P. Roth, R.E. Gerszten, Circulation 112 (2005) 3868.
 M.J. van der Werf, K.M. Overkamp, B. Muilwijk, L. Coulier, T. Hankemeier, Anal.
Biochem. 370 (2007) 17.
 E. Nagele, M. Vollmer, P. Horth, C. Vad, Exp. Rev. Proteom. 1 (2004) 37.
 J.L. Edwards, R.L. Edwards, K.R. Reid, R.T. Kennedy, J. Chromatogr. A 1172 (2007)
 Q. Ni, K.R. Reid, C.F. Burant, R.T. Kennedy, Anal. Chem. 80 (2008) 3539.
 B.W. Yan, J.H. Zhao, J.S. Brown, J. Blackwell, P.W. Carr, Anal. Chem. 72 (2000)
 D. Guillarme, S. Heinisch, J.L. Rocca, J. Chromatogr. A 1052 (2004) 39.
 S. Heinisch, G. Puy, M.P. Barrioulet, J.L. Rocca, J. Chromatogr. A 1118 (2006) 234.
242 Download full-text
W. Lu et al. / J. Chromatogr. B 871 (2008) 236–242
 A.R. Karala, L.W. Ruddock, Antioxid. Redox Signal. 9 (2007) 527.
 M.R. Shortreed, S.M. Lamos, B.L. Frey, M.F. Phillips, M. Patel, P.J. Belshaw, L.M.
Smith, Anal. Chem. 78 (2006) 6398.
 S.M. Lamos, M.R. Shortreed, B.L. Frey, P.J. Belshaw, L.M. Smith, Anal. Chem. 79
 E.E. Carlson, B.F. Cravatt, Nat. Methods 4 (2007) 429.
 S. Nguyen, J.B. Fenn, Proc. Natl. Acad. Sci. U.S.A. 104 (2007) 1111.
 M.G. Ikonomou, A.T. Blades, P. Kebarle, Anal. Chem. 63 (1991) 1989.
 M.G. Ikonomou, P. Kebarle, J. Am. Soc. Mass Spectrom. 5 (1994) 791.
 B. Domon, R. Aebersold, Science 312 (2006) 212.