Metabolic profiling of Medicago truncatula cell cultures
reveals the effects of biotic and abiotic elicitors on
Corey D. Broeckling1, David V. Huhman1, Mohamed A. Farag1, Joel T. Smith2, Gregory D. May1,
Pedro Mendes3, Richard A. Dixon1and Lloyd W. Sumner1,*
1The Samuel Roberts Noble Foundation, Plant Biology, 2510 Sam Noble Parkway, Ardmore, OK 73401, USA
2Southeastern Oklahoma State University, Physical Sciences, Durant, OK 74701, USA
3Virginia Bioinformatics Institute, Virginia Polytechnic Institute and State University, Blacksburg, VA 24060, USA
Received 13 July 2004; Accepted 22 October 2004
GC-MS-based metabolite profiling was used to analyse
the response of Medicago truncatula cell cultures to
elicitation with methyl jasmonate (MeJa), yeast elicitor
(YE), or ultraviolet light (UV). Marked changes in the
levels of primary metabolites, including several amino
acids, organic acids, and carbohydrates, were observed
following elicitation with MeJa. A similar, but attenuated
whereas little response was observed following UV
elicitation. MeJa induced the accumulation of the triter-
pene b-amyrin, a precursor to the triterpene saponins,
and LC-MS analysis confirmed the accumulation of
triterpene saponins in MeJa-elicited samples. In addi-
tion,YEinduceda slight,butsignificant accumulationof
by LC-MS analyses. Correlation analyses of metabolite
and threonine biosynthetic pathway, and suggested
the induction of threonine aldolase activity, an enzyme
as yet uncharacterized from plants. Members of the
branched chain amino acid pathway accumulated in
exposure itself had little effect on primary metabolites,
ments, induced changes in several metabolites which
levels were lower in MJ- and YE-elicited samples com-
pared with control samples, suggesting that a portion of
the effects observed on the primary metabolic pool are
a consequence of fundamental metabolic repartitioning
of carbon resources rather than elicitor-specific induc-
tion. In addition, b-alanine levels were elevated in all
elicited samples, which, when viewed in the context of
other elicitation responses, suggests the altered metab-
olism of coenzyme A and its esters, which are essential
in secondary metabolism.
Key words: Elicitation, Medicago truncatula, metabolite pro-
filing, metabolomics, methyl jasmonate, primary metabolism,
Medicago truncatula is a rapidly developing model for the
study of legume biology, and is an excellent species for
fundamental studies on the unique secondary metabolism
of legumes (Dixon and Sumner, 2003). Two classes of sec-
ondary metabolites are of particular interest. Isoflavonoids
are a subclass of the phenylpropanoids which have been
attributed with health-promoting properties, and are nearly
exclusive to leguminous plants. The biosynthetic pathway
leading to the production of isoflavonoids can be elicited by
the application of yeast cell wall extract (Kessmann et al.,
1990) or pathogen attack and is relatively well understood,
with several of the enzymes characterized (Liu et al., 2002).
The phenylpropanoid pathway is also commonly induced
ical damage through the UV absorbent character of phenyl-
propanoids (Mazza et al., 2000). Another important class of
* To whom correspondence should be addressed. Fax: +1 580 224 6692. E-mail: email@example.com
Journal of Experimental Botany ª Society for Experimental Biology 2004; all rights reserved
Journal of Experimental Botany, Page 1 of 14
Metabolomics and Metabolic Profiling Special Issue
Advance Access published December 13, 2004
by guest on June 3, 2013
secondary metabolites from M. truncatula are the triterpene
1999), allelopathic (Waller et al., 1993), anti-microbial
(Osbourn, 2003), anti-insect (Tava and Odoardi, 1996), and
anti-palatability activity, as well as anti-nutritional (Oleszek
et al., 1999) effects. Saponins also exhibit anticholesterole-
mic, anti-cancer (Haridas et al., 2001), adjuvant (Behboudi
et al., 1999), and haemolytic (Oh et al., 2000) activity.
M. truncatula has a diverse saponin content (Huhman and
Sumner, 2002) which is regulated by jasmonates, as exo-
cultures results in the accumulation of saponins within 48 h
of elicitation (Suzuki et al., 2002). Whereas the enzymes
responsible for isoflavone biosynthesis are fairly well
characterized, those involved in the biosynthesis of triter-
pene saponins are still relatively unknown.
Secondary metabolites are nearly universally derived
from primary metabolic pathways. For example, flavonoids
throughout the plant kingdom, and more specifically iso-
flavonoids of legumes, are derived initially through the
phenylpropanoid pathway, originating from the protein
amino acid phenylalanine (Kessmann et al., 1990). Like-
wise, the triterpene saponins are derived from the cycliza-
to membrane phytosterols (Suzuki et al., 2002).
Only recently has the monitoring of metabolites grown
into an ‘omics’ level field (Trethewey et al., 1999). Gas
chromatography-mass spectrometry (GC-MS) has been
applied to examine the effects of genetic and environmental
manipulations (Roessner et al., 2001), to determine phloem
composition (Fiehn, 2003), for plant genotyping (Taylor
et al., 2002) and, recently, for detecting silent phenotypes
in transgenic potato (Weckwerth et al., 2004). GC-MS is
currently the most developed of the available analytical
tools, but other techniques are currently in use or being
developed (Sumner et al., 2003). The growth of this tech-
nology offers the opportunity to view the effect of elicit-
genomics project studying the effects of elicitation with
various biotic and abiotic elicitors on three biological levels
of function: the transcriptome, the proteome, and the
metabolome (VandenBosch and Stacey, 2003). This global
response to elicitation than previously available. The ap-
proach to the metabolomics portion of this study attempts
to cover a large portion of the metabolome, with primary
metabolites monitored by gas chromatography-mass spec-
trometry (GC-MS), lower abundant intermediates through
capillary electrophoresis, and secondary metabolites by
liquid chromatography-mass spectrometry (LC-MS).
The results of GC-MS based metabolite analyses reveal
the effect of elicitation on the accumulation of many pri-
mary metabolites and their interrelationships. In addition,
correlation analyses revealed more universal metabolite
relationships which are robust to elicitor-induced metabolic
reprogramming. The results suggest both elicitor-specific
changes in metabolite abundance and correlations as well
as a more generic response in which metabolites demon-
strate a similar trend regardless of the elicitor used.
Materials and methods
Cell cultures and elicitation
A total of four separate experiments were performed. Three, one for
each of the three elicitors, were highly detailed and included 21
sampling points over a 48 h period following elicitation (Table 1).
The first time-course examined elicitation with methyl jasmonate
(MeJa), the second with yeast elicitor (YE), and the third with UV-
light (UV). Each of the three time-courses contained, in addition to
the primary elicitor, 2–3 time points for each of the other elicitors,
allowing evaluation of the potential effect of cell culture passage
in monitoring the response. The fourth time-course contained fewer
time points, but all three elicitors were examined simultaneously in
order to allow a direct comparison of the cell culture responses to the
elicitors at the same passage and to validate the responses previously
demonstrated in the detailed time-courses using single elicitors.
Callus culture was initiated from M. truncatula roots, maintained
on modified Schenk and Hilderbrandt (1971, SH) agar plates (see
below) in the dark at 25 8C, and subcultured approximately every
4 weeks. Liquid subcultures were initiated in 40 ml modified SH
medium in a 125 ml Erlenmeyer flask from 5.0 g callus and
maintained in the dark at 22 8C with shaking at ;130 rpm. Liquid
media were composed of sucrose (30.0 g l?1), KNO3(2.525 g l?1),
MgSO4(370 mg l?1), NH4H2PO4(290 mg l?1), CaCl2(220 mg l?1),
myo-inositol (1.0 g l?1) MnSO4(8.925 mg l?1), H3BO4(5 mg l?1),
ZnSO4.7H2O (1.0 mg l?1), KI (1.0 mg l?1), FeSO4.7H2O (15.0 mg
l?1) Na2EDTA (20 mg l?1), thiamine.HCl (5 mg l?1), nicotinic acid
(5 mg l?1), pyridoxine.HCl (0.5 mg l?1), Na2MoO4(0.1 mg l?1),
CoCl2.6H2O (0.1 mg l?1), CuSO4.5H2O (0.2 mg l?1), kinetin (0.11
mg l?1), 2,4-D (0.45 mg l?1), and PCPA (1.87 mg l?1). Solid media
(for UV elicitation) additionally contained 8 g Bacto?agar l?1.
Cultures were transferred to 250 ml flasks and subcultured approx-
imately every 2 weeks until elicited.
Triplicate biological replicates were collected for both control and
elicited samples at each time point, with each replicate collected from
a separate culture flask. Thus, each elicitation time-course contained
126 culture flasks, in addition to 12–18 confirmatory samples of
elicitations (with controls) other than the primary for that time-course.
For MeJa elicitation, 2.5 ml of a 50 mM solution of methyl
jasmonate in ethanol was added to culture flasks to achieve a final
concentration of 500 lM. Control flasks received 2.5 ml ethanol.
MeJa elicitation was performed during the 9th passage. The YE time-
course was conducted during the 11th passage by adding 2.5 ml of a
5 mg ml?1aqueous solution of a yeast cell wall preparation for a final
concentration of 50 lg glucose equivalents ml?1(Schumacher et al.,
1987). UV elicitation was performed during the 12th passage.
Cultures were strained from culture media and spread onto 150 ml
plates containing ; 50 ml modified SH agar. Treatment plates were
irradiated in a UV box for 5.5 min at 8000 J m?2while control plates
received no exposure. Plates were then held on an illuminated shelf at
24 8C until harvested. At the time of harvest, the entire cell population
was collected by vacuum filtration, washed with 50 ml 25% MS salts,
divided into four 50 ml tubes, and flash-frozen in liquid N2.
Metabolite analysis of cell culture tissue
One 50 ml tube containing frozen tissue was lyophilized for 48–72 h
until dry, noting that the tissue was maintained in its frozen state
2 of 14 Broeckling et al.
by guest on June 3, 2013
through evaporative cooling during the lyophilization process. Dried
tissue was homogenized with a glass rod, and 6.0–6.05 mg of dried
tissue was weighed into a 4.0 ml glass vial. The dried tissue was
stored at ?80 8C until extraction. Chloroform (1.5 ml) containing
10 lg ml?1docosanol (internal standard) was added to dried tissue.
After equilibrating to room temperature, 1.5 ml of HPLC-grade water
containing 25 lg ml?1ribitol was added to the chloroform. The
sample was then vortexed, and incubated for a second 45 min period.
at 4 8C to separate the layers. One ml of each layer was collected
and transferred to individual 2.0 ml autosampler vials. The chloro-
form layer (non-polar) was dried under nitrogen and the aqueous
layer dried in a vacuum centrifuge at ambient temperature.
The non-polar layer was resuspended in 0.8 ml chloroform and
hydrolysed by adding0.5 ml 1.25 M HCl in MeOH and incubating for
4 h at 50 8C. Following hydrolysis, HCl and solvent were evaporated
under nitrogen. The sample was resuspended in 70 ll pyridine and
derivatized through the addition of 30 ll of a commercial derivatiza-
tion solution containing MSTFA+1%TMCS (Pierce Biotechnology,
Rockford, IL, USA) and incubation for 1 h at 50 8C. The sample was
equilibrated to room temperature, transferred to a 200 ll glass insert,
and analysed using an Agilent 6890 GC coupled to a 5973 MSD
scanning from m/z 50–650. Samples were injected at a 1:1 split ratio,
and the inlet and transfer line were held at 280 8C. Separation was
achieved with a temperature programme of 80 8C for 2 min, then
ramped at 5 8C min?1to 315 8C and held for 12 min, a 60 m DB-5MS
column (J&W Scientific, 0.25 mm ID, 0.25 lm film thickness) and a
constant flow of 1.0 ml min?1.
Dried polar extracts were methoximated in pyridine with 120 ll of
15.0 lg ll?1methoxyamine–HCl, briefly sonicated, and incubated
at 50 8C until the residue was resuspended. Metabolites were then
derivatized with 120 ll of MSTFA+1% TMCS for 1 h at 50 8C. The
sample was subsequently transferred to a 300 ll glass insert and
analysed by GC-MS using the same parameters as described for the
non-polar extracts, with the exception that the injection split ratio was
set to 15:1 for polar samples.
Secondary metabolites, including triterpene saponins, and isofla-
vanoids were analysed using liquid chromatography-electrospray
ionization mass spectrometry (LC-MS). Metabolites were extracted
in 1.8 ml of 80% MeOH containing 2 lg umbelliferone as an internal
standard for 10 h. Extracts (1.4 ml) were centrifuged at 3000 g for
60 min and the resulting supernatant was evaporated under nitrogen
to dryness. The residue was resuspended in 300 ll of 45% MeOH
(isoflavonoids) or 100 ll water (triterpene saponins) and the samples
were analysed by LC-MS.
An Agilent 1100 series II LC system (Agilent Technologies, Palo
Alto, CA) equipped with a photodiode array detector was coupled to
a Bruker Esquire ion-trap mass spectrometer equipped with an
from 200 nm to 600 nm. A reverse-phase, C18, 5 lm, 4.63250 mm
column (JT Baker, Phillipsburg, NJ) was used for separations. The
mobile phase consisted of eluent A (0.1% [v/v] CH3COOH/water)
and eluent B (acetonitrile), and separations achieved using a linear
gradient of 5–90% B (v/v) over 70 min. The flow rate was 0.8 ml
min?1, and the temperature of the column was maintained at 28 8C.
Both positive and negative ion mass spectra were acquired. Positive-
ion ESI was performed using an ion source voltage of 4.0 kV and
a capillary offset voltage of 86.0 V. Nebulization was aided with
a coaxial nitrogen sheath gas provided at a pressure of 60 psi.
Desolvation was aided using a counter current nitrogen flow set at
were recorded over the range 50–2200 m/z. The Bruker ion-trap mass
spectrometer (ITMS) was operated under an ion current control (ICC)
of approximately 10 000 with a maximum acquire time of 100 ms.
Database search and sequence alignment
Sequence data for threonine aldolase and serine hydroxymethyl-
transferase genes was collected from public databases linked through
KEGG for Arabidopsis thaliana, Saccharomyces cerevisiae, and
Escherichia coli. M. truncatula sequence data were isolated from
the TIGR M. truncatula gene index using TBLASTN against S.
cerevisiae amino acid sequence. Amino acid sequences were aligned
using Clustal W, the results of which are presented as a branch-length
Relative metabolite abundances were calculated using a custom
PERL script to extract peak areas of individual ions characteristic of
each component. Metabolites were identified through spectral and
retention time matching with authentic compounds prepared in an
identical manner. Identifications were further confirmed through
spectral matching against the National Institutes of Standards and
Technology (NIST) library. Peak areas were normalized by dividing
each peak area value by the mean peak area for that compound, with
each time-course treated independently. Correlation analyses were
performed with a custom PERL script executing Pearson’s correl-
ation formulas (Zar, 1999). Principal component analysis (PCA) was
performed on normalized datasets with Pirouette?(InfoMetrix,
Woodinville, WA) software. Cumulative GC-MS metabolite pro-
filing data is provided as supplementary materials (S1 and S2).
Analytical and biological reproducibility
The instrumental variation attributed to multiple chromato-
graphic analyses of the same sample was quantified to
for each sample ofthe MeJa time-course, and allpeaks from
Table 1. Design of elicitation experiments
‘X’ indicates sample collected for given timepoint/elicitor combination:
MJ, methyl jasmonate; YE, yeast elicitor; UV, ultraviolet light.
Time-course MJYEUV Mixed
ElicitorMJ YE UV MJ YE UV MJ YE UV MJ YE UV
Metabolomics of M. truncatula elicitation3 of 14
by guest on June 3, 2013
the polar extracts consistently above the limit of detection,
the majority of the metabolites and only 13% of the peaks
ents were those of lowest abundance (Fig. 1b) and peaks
with the highest total variation across the entire dataset also
tended to possess the higher injection variation (Fig. 1c).
Based on these results, single GC-MS analyses were per-
formed for nearly all subsequent samples, as the benefits of
performing multiple analyses (slightly greater accuracy in
peak areas) did not justify the additional resources (double
particularly for large datasets. However, biological tripli-
cates were still utilized and triplicate instrumental analyses
of individual biological replicates were performed on 4–6
samples throughouteach time-course to provide an estimate
of instrumental variability.
The analytical variation associated with the entire
sampling, extraction, and analysis procedure was also
quantified, as was the biological variance associated with
different cell culture replicates. The analytical coefficient of
variation (CV) was calculated using the internal standard
peak area and ranged from 4.6% to 7.8% for polar extracts.
For the calculation of biological variance, a list of approxi-
mately 120 components was compiled for comparative
analyses of polar extracts from all elicited time-courses.
This list was based on the consistent presence of these
metabolites in all time-course data. The median biological
CV (including elicitation responses and temporal trends)
ranged from 27.4% to 33.3% over each time-course (mean
biological variability values were approximately 10% to
15% higher than median values due to the influence of
a few exceptionally variable peaks). Approximately 40% of
quantified peaks have been identified (72 out of 169 for
polar and non-polar metabolites). Peak area values based on
individual representative ions for all metabolites used are
presented as supplementary files and can be located at
JXB online (S1 contains data on polar metabolites and
S2 contains data on non-polar metabolites).
The effect on primary metabolite pools was most dramatic
following elicitation with MeJa. Increased levels of several
amino acids, most notably valine, leucine, isoleucine, and
threonine, were observed over the 48 h period (Fig. 2; Table
2). In addition, succinic and fumaric acid demonstrated
similar trends. Phosphate accumulated to slightly higher
the non-protein amino acids c-aminobutyric acid (GABA)
and b-alanine. Sucrose demonstrated the opposite trend,
with decreased levels in elicited tissue relative to controls.
The triterpene b-amyrin accumulated in MeJa-elicited sam-
ples, and was the only identified non-polar metabolite to
demonstrate an elicitation response. Further, LC-MS analy-
sis revealed the accumulation of triterpene saponins after
40 h (Fig. 3), suggesting that the accumulation of b-amyrin
precedes increased saponin biosynthesis. The small but
following MeJa elicitation, and of shikimic acid following
YE elicitation discussed below, confirm that the M. trun-
catula cultures are responding in a similar manner to pre-
viously published reports (Suzuki et al., 2002).
Virtually all of the observed effects of elicitation were
quantitative rather than qualitative. Two peaks, however,
were only detected in extracts of MeJa-elicited tissue. The
first was identified as jasmonic acid, presumably arising
< 11-2 2-5>5
0 75150 225 300
Peak area (x1000)
0 0.25 0.5 0.751
Total peak CV
Fig.1. Duplicate GC-MS analyses were performed on each sample of the
polar extracts from the MeJa time-course. This dataset was analysed to
evaluate the importance of multiple instrumental analyses of individual
samples and guide the analytical approach for the remaining three time-
courses. (A) The average difference between replicate peak areas for each
of 249 peaks was calculated and tabulated as a percentage of the mean
peak area. The majority of the peaks differed by less than 2% peak area,
while less than 13% peaks differed by greater than 5%. (B) Deviation
between analyses was found to be highly dependent on peak area, with
the highest deviations demonstrated for the lowest abundance peaks. (C)
Instrumental variability was also related to the total variation in peak area
for the dataset. Peaks with high overall variability varied more between
analyses, either due to low peak areas, compound instability, or other
unknown factors. CV, Coefficient of variation; Diff CV, the difference
between the CV for all samples of the first injection and the CV for all
samples of the second injection.
4 of 14 Broeckling et al.
by guest on June 3, 2013
from hydrolysis of exogenously applied MeJa (Swiatek
et al., 2004). The second was an unidentified compound
eluting approximately 5 min later. The spectral character-
istics of the unknown compound are similar to those of
derivatized jasmonic acid, with characteristic fragment ions
with shifts of two m/z units in either direction. More
specifically, jasmonic acid possesses fragment ions at m/z
280, 180, and 148, while the unknown possesses fragment
ions at m/z 282, 178, and 146. The unknown peak is
believed to be an intermediate in the enzymatic degradation
or inactivation of jasmonic acid, as the second peak trails
jasmonic acid in abundance by at least 18 h (Fig. 4). The
putative molecular weight of the derivatized unidentified
compound is 399, 88 m/z, greater than that of derivatized
jasmonic acid. This difference can be explained by an
additional hydroxylation (addition of 16) with subsequent
TMS derivatization (addition of 72) and is consistent with
hydroxylated jasmonic acid. Hydroxylated JA was recently
identified as an intermediate in JA degradation in tobacco
(Swiatek et al., 2004).
Correlation analyses and the related ‘connectivity’ of
metabolites has recently been used to detect the metabolic
consequences of sucrose synthase isoform II suppression,
which fails to demonstrate a visible phenotype (Weckwerth
et al., 2004). A similar approach using correlation analyses
in a comparative fashion to reveal the effect of elicitors
on metabolite relationships was used on this dataset. The
dataset was divided into elicited and control samples, and
metabolite-to-metabolite correlation analyses were per-
formed on each dataset. The resulting correlation param-
This analysis utilized all time points for the estimation of
correlation parameters, which is probably an oversimplifi-
cation of the time-course nature of the data, but still
valuable for comparative purposes. Several amino acids
increased following elicitation while sucrose levels de-
creased. This trend is exemplified by the relationship
between b-alanine and sucrose, which changed from ab-
sent in control samples (r2=0.028) to negative (r2=0.796,
r=?0.892) following elicitation. In control extracts, threo-
nine and pyroglutamic acid were very poorly correlated
(r2=0.010), but positively correlated following elicitation
(r2=0.718). An additional example of altered correlation
parameters following elicitation was between the three
Time (h)Time (h)
010 2030 4050
010 203040 50
0 1020 30 4050
010 2030 40 50
Fig. 2. Responses of selected metabolites to elicitation with MeJa. Y-axis values represent relative peak areas after normalization to the mean peak area
for that compound. Many amino acids increased following elicitation while sucrose simultaneously decreased. Solid squares represent elicited sample
means (stand. dev. error bars) and open diamonds represent control sample means.
Metabolomics of M. truncatula elicitation5 of 14
by guest on June 3, 2013
amino acids serine, glycine, and threonine. Glycine is
biosynthetically linked to both serine and threonine by as
few as one enzymatic step in micro-organisms. There was
no correlation between glycine and threonine in unelicited
samples, but a clear relationship was observed in elicited
samples (Fig. 5).
The effect of YE on primary metabolism was more subtle
than that of MeJa, but several trends were observed, with
some similar to those following MeJa elicitation while
others were unique (Table 2). Phosphate levels increased
following exposure to YE, whereas sucrose decreased. b-
alanine levels were induced by YE, while other amino acids
which showed MeJa responsiveness failed to show a clear
response to YE. Shikimic acid, a precursor of the phenyl-
propanoid pathway, accumulated following YE elicitation,
as did citric acid and glucose-6-phosphate. Several end-
products of the phenylpropanoid pathway, of which shi-
kimic acid is a precursor, also accumulated with maximal
elicitation at either 10 h or over the 48 h period (Fig. 3).
Comparative correlation analysis of YE data yielded
fewer changes in metabolite relationships than did the MeJa
data. Valine and leucine were moderately correlated in con-
trol samples, and the strength of the correlation increased
following treatment with YE from r2=0.445 in controls to
r2=0.860. b-Alanine became negatively correlated with
sucrose, with r2increasing from 0.051 to 0.465 following
UV elicitation had less of an effect on primary metabolism
than either MeJa or YE. The procedure of transferring and
spreading the original suspension cell cultures onto agar
plates for UV exposure may have prevented observation of
an elicitation response at the level of primary metabolism.
In addition, the plates were maintained on an illuminated
shelf exposed to a diurnal cycle following elicitation with
strong UV exposure. The design of the experiment (see
Materials and methods) may have induced changes in the
cellular metabolic state which masked UV-elicitation ef-
fects at the level of primary metabolism. In fact, all UV
samples, elicited and unelicited, looked similar to MeJa-
elicited samples in many respects (see below).
A fourth, mixed time-course served as a means of validat-
ing and correlating responses observed in each of the three
more detailed individual time-courses. Although fewer time
points were analysed, all elicitations were performed on
cells from the same cell culture passage that were extracted
and analysed in one large batch. Using the mixed time-
course data, the elicitation responses were analysed using
principal component analysis (Fig. 6a). Both UV control
Table 2. Fold change in peak area following elicitation
and respective individual elicitor time-courses) significant (P <0.05 in
mixed and individual elicitor time-courses) response to at least one
the average of the 18, 24, and 36 h timepoints (values >1.0 represent
due to elicitation; blank cells, statistically insignificant in one of both
Mixed MJ Mixed YEMixed UV
Unknown MW=330 27.78
24.191.29 1.16 1.251.27
0.92 1.2527.91 1.16
0.860.89 1.45 1.56
32.59 1.23 1.27
37.26 1.17 1.18
0.50 0.69 0.76
6 of 14 Broeckling et al.
by guest on June 3, 2013
and UV-elicited, and MeJa-elicited, samples segregated
from the general cluster containing early time points from
all samples and YE and MeJa controls. UV control and
elicited samples were indistinguishable from each other in
PCAspace,and MeJa samplestrended in the samedirection
as the UV samples. However, the two groups were clearly
separated along the first principal component axis (Fig. 6a).
To identify the source of this similarity in elicitation
response, the UV elicited and control samples were pooled,
based on the lack of significant changes. The MeJa and YE
controls were also pooled to provide a reliable estimate of
control values. The similarity of the pooled UV samples to
the MeJa-elicited samples (and to a lesser extent the YE-
elicited samples) is reflected in the plots of individual
compounds (Fig. 6b).
Elicitor independent relationships
While many of the metabolite correlations were altered by
one or more elicitors, several relationships between metab-
olites were found to be robust to the effects of any of the
elicitors, such that the correlation parameters between
a metabolite pair were unaltered by elicitation procedures.
To explore these elicitor-independent relationships thor-
oughly, correlation analysis was applied to a composite
dataset compiled from all four elicited time-courses, with
each metabolite normalized to its intra-time-course mean.
The relationship between leucine and isoleucine was re-
markably conserved through the entire dataset (r2=0.941).
Valine was also highly correlated with both leucine and
isoleucine (r2=0.790 and r2=0.822, respectively), and
leucine, isoleucine, and valine correlated moderately with
threonine (r2=0.498, 0.599, and 0.458, respectively). Ala-
nine and pyroglutamic acid are correlated (r2=0.683),
despite the fact that there were no dramatic elicitor-induced
changes in levels of either to buttress the r2value. The
serine–threonine relationship previously discussed was
considerably stronger (r2=0.652) than either the glycine–
serine (r2=0.353) or glycine–threonine (r2=0.432) correl-
ations. Alanine was negatively correlated to fumarate with
r2=0.467, although a linear regression line is an overly
simplistic model for this relationship.
Biotic and abiotic elicitors are often applied in the exam-
ination of secondary metabolism and the responses of cul-
at various levels of detail in several species. The effect of
0 10 2030 40 50
0 1020 3040 50
0 10 20 30 40 50
0 10 2030 40 50
Relative peak area
Relative peak area
Relative peak area
Relative peak area
Fig. 3. Two triterpene saponins (Soyasapogenol B and E glycosides) accumulate at late time points following MeJa elicitation. The triterpene saponins
are derived ultimately through b-amyrin, which accumulates following MeJa elicitation (Table 2), although far less dramatically than the saponin end-
products. In addition, two isoflavone aglycones (Formononetin and Afrormosin) accumulate following YE elicitation with varying dynamics. The
isoflavones are derived through the phenylpropanoid pathway, and the phenylpropanoid precursor, shikimic acid, is demonstrated to accumulate subtly
following YE elicitation in a manner similar to b-amyrin following MeJa elicitation.
Metabolomics of M. truncatula elicitation7 of 14
by guest on June 3, 2013
elicitation on primary metabolite accumulation has largely
been overlooked. However, at the transcript level, fungal
elicitation altered expression of over 40 transcripts tested,
including representatives from the phenylpropanoid, pen-
tose phosphate, glycolytic, and fatty acid metabolic path-
ways (Batz et al., 1998), suggesting that the response to
elicitation is much more than the simple induction of
biosynthetic enzymes of secondary metabolism.
Primary metabolism provides critical substrates for sec-
ondary metabolic pathways. For example, the entry point
many essential cofactors and ligands involved in primary
metabolism are required for secondary metabolite biosyn-
metabolites. For example, Coenzyme A (CoA) is listed in
over 300 metabolic reactions in the KEGG metabolism
database (Kanehisa et al., 2004). CoA is an essential
component in both primary and secondary metabolic reac-
tions, and the regulation of enzymes utilizing CoA or its
thioesters is often affected by the induction of secondary
et al (1998) also demonstrated the induction of S-adenosyl-
methionine synthase transcription following fungal elicit-
ation, which serves as a methyl donor to furanocoumarins.
In this study, dramatic changes in accumulation patterns for
several metabolites and pathways of primary metabolism,
both distant and proximal to secondary metabolic branch
points, were observed and are discussed below.
Exposure of M. truncatula cells to MeJa, YE, or UV
resulted in decreased sucrose levels over 48 h, with the
simultaneous accumulation of several amino acids and
some organic acids. This pattern indicates altered carbohy-
drate metabolism following elicitation. A portion of the
diverted carbon is shifted toward secondary metabolism, as
revealed by increased triterpene saponin levels following
MeJa elicitation and increased isoflavonoid accumulation
following YE. Presumably, an additional portion of carbo-
hydrate is consumed for production of energy to support
secondary metabolite biosynthesis. In a similar fashion,
elicitation of parsley cell cultures with Phytophthora
megasperma extracts increased the rate of respiratory CO2
evolution, particularly through glycolysis and the oxidative
pentose phosphate pathways (Norman et al., 1994). The
authors proposed that this response served to supply sub-
strate for the synthesis of furanocoumarins. Although
parsley and M. truncatula synthesize differing classes of
induced reallocation of carbon toward secondary metab-
olism appears similar. However, in addition to secondary
metabolites, several primary metabolites, such as b-alanine,
GABA, and succinic acid are observed to accumulate
following MeJa elicitation. Accumulation of these metab-
olites cannot be explained by their ecological functions or
common catabolic phenomena such as protein degradation.
Negative correlations between amino acids and sucrose
have frequently been observed with the advent of global
0 10 203040 50
Rel. peak area
Fig. 4. Jasmonic acid (JA, solid diamonds) and a second unidentified
peak (open squares) were the only two qualitative differences following
any elicitation. Each peak was detected only in MeJa-elicited samples.
The mass spectrum of the unidentified peak is visually similar to that of
JA, and the dynamics of the accumulation trail that of JA by at least 18 h,
suggesting that the unidentified peak may be an intermediate in the
degradation of JA. The putative molecular weight and observed fragment
ions of the unidentified compound suggest a hydroxylation product of JA.
Fig. 5. Diagram demonstrating the effect of MeJa elicitation on the
relationship between glycine, serine and threonine. In control samples,
glycine is correlated to serine (r2=0.774), but not threonine. Following
elicitation, all three metabolites showed strong correlations between each
8 of 14 Broeckling et al.
by guest on June 3, 2013
0 162432 40
Rel peak area
Rel peak area
Rel peak area
Rel peak area
0 16 243240
0 16 24 3240
Fig. 6. A portion of the elicitation response seems to be common to MeJa and YE elicitation and exposure to light following UV elicitation. (a) Principal
component analysis of time points 15 min, 18 h, 24 h, and 36 h samples from the mixed-elicitor time-course. The 15 min samples and YE and MeJa-
unelicited samples from later time points are unseparated (unlabelled open diamonds), while the MeJa-elicited as well as the UV-elicited and control
samples (filled diamonds) drift from the centre of the PCA plot in the same direction. However, MeJa-elicited and UV samples are clearly distinguishable
along the first principal component axis. (b) The dynamics of several amino acids and sucrose nearly overlap for the pooled UV samples (open squares)
and MeJa-elicited samples (closed squares), while the YE-elicited samples (open triangles) are generally more similar to pooled MeJa and YE control
samples (closed diamonds).
Metabolomics of M. truncatula elicitation 9 of 14
by guest on June 3, 2013
metabolite profiling. To demonstrate the generality of the
phenomenon, sucrose levels in potato tubers were altered in
response to changes in light intensity, transgenic manipu-
lation of sucrosetransport from source leaves to sink tubers,
or direct alteration of sucrose delivered to the tuber through
a cut stolon (Roessner-Tunali et al., 2003). Although the
method used to alter sucrose levels had some effect on the
specific amino acid levels altered, the total amino acid
content was consistently negatively correlated with sucrose
Tunali et al., 2003). In addition, repression of a constitutive
sucrose transporter (SUT1) resulted in the increased ex-
pression of certain amino acid biosynthetic genes, including
aspartate aminotransferase, suggesting that increased amino
or decreased protein synthesis, but arise at least partially
through increased biosynthesis. Likewise, the addition of
sucrose to carrot cell suspension cultures resulted in a de-
crease in glutamate dehydrogenase activity and a resultant
drop in glutamate concentration (Robinson et al., 1992).
Non-protein amino acids and polyamines
The strongest inverse carbohydrate-to-amino acid relation-
ship observed was between sucrose and b-alanine. b-
Alanine is a non-protein amino acid which can serve as an
intermediate in coenzyme A synthesis through pantothenic
with the preferred biosynthetic route apparently being clade
specific. The exact synthetic mechanism in plants is un-
known, and a metabolic pathway based on A. thaliana
sequence data cannot be reconstructed between sucrose and
seedlings and characterized (Walsh et al., 2001) and is
include b-alanine and b-aminoisobutyric acid, affording the
enzyme a simultaneous biosynthetic function. However, no
b-aminoisobutyric acid was detected in this study, as might
be expected if ureidopropionase were responsible for the
b-aminoisobutyric acid is currently unknown, this cannot
account for the absence of this metabolite through its con-
version into an accumulating metabolite.
Recently, an additional biosynthetic route for the pro-
duction of b-alanine through degradation of the polyamines
spermidine and spermine was described in yeast (White
in the production of spermidine and 3-aminopropionalde-
hyde. The aldehyde is subsequently converted to b-alanine.
In this study, the polyamine putrescine was observed at
slightly increased levels following YE and MeJa elicitation,
but neither spermine nor spermidine were detected. Poly-
amine synthesis is MeJa-inducible, apparently through the
arginine decarboxylase pathway in tobacco (Biondi et al.,
2001) and barley (Walters et al., 2002). This pathway
proceeds through several steps to convert arginine through
agmatine to putrescine, and genes encoding several en-
zymes of this pathway have been cloned (Piotrowski et al.,
2003). However, over-expression of arginine decarboxyl-
ase, in an attempt to increase polyamine production in
tobacco, resulted in either a 10–20-fold accumulation of
agmatine without the accumulation of polyamines (Burtin
and Michael, 1997), or a slight accumulation of polyamines
which was correlated with a growth phenotype (Masgrau
et al., 1997). In A. thaliana, MeJa induced a local induction
of arginine decarboxylase, a transient accumulation of
putrescine, no effect on spermidine, and a subtle decrease
in spermine (Perez-Amador et al., 2002). The authors
suggest that putrescine may be converted to GABA through
4-aminobutanal (Flores and Filner, 1985) or that degrad-
the accumulation of the higher polyamines.
Increased polyamine biosynthesis without dramatic ac-
cumulation (Burtin and Michael, 1997; Masgrau et al.,
1997) would simultaneously increase the availability of
3-aminopropionaldehyde, a possible precursor to b-alanine
as well as 4-aminobutanaldehyde, an immediate precursor
to GABA. Both b-alanine and GABA were found at higher
concentrations in elicited compared with control cells fol-
lowing MeJa elicitation in this study. b-Alanine and the
in the formation of pantothenic acid (White et al., 2001)
which is subsequently converted to Coenzyme A (Kupke
et al., 2003). Valine, leucine, and isoleucine, which with
their intermediates including 2-oxoisovalerate comprise the
Assuming metabolite accumulation can be interpreted as
a metabolic imbalance which appears as a consequence of
increased flux through that metabolite, as was observed in
this study for shikimic acid and b-amyrin, then accumula-
tion of branched chain amino acids, putrescine, GABA, and
b-alanine, might collectively be interpreted as altered CoA
biosynthesis (Fig. 7). This suggestion represents an hy-
pothesis generated using metabolomics, rather than a con-
clusion based on experimental data, and will be pursued in
CoA serves as a carrier for organic acids including acetic
acid (utilized in fatty acid biosynthesis, glycolysis, citrate
cycle, amino acid synthesis etc.), malonic, sinapic, and
ferulic acid (intermediates in the phenylpropanoid pathway
leading to lignin and flavonoids), and 3-hydroxymethyl-
glutaric acid (an intermediate in sterol and terpenoid
biosynthesis). Thus, CoA is an essential cofactor, not
only for primary metabolism, but, also for the phenyl-
propanoid and triterpene saponin pathways which are
up-regulated in M. truncatula by the elicitors used in
this study. A tentative consensus (TC) sequence (TC78022)
10 of 14 Broeckling et al.
by guest on June 3, 2013
with similarity to the A. thaliana pantothenate kinase
(Kupke et al., 2003), which catalyses the initial step in
the incorporation of pantothenate to CoA, was found in the
TIGR M. truncatula database. This TC was most highly
expressed in a library constructed from YE-treated cell
cultures, further supporting the hypothesis that CoA bio-
synthesis may be inducible.
Branched-chain amino acids
A highly linear and precise correlation was observed
between the levels of leucine and isoleucine which was
robust to perturbation with elicitors. Valine, leucine, and
isoleucine are all produced by the same biosynthetic
pathway, the enzymes of which are plastid localized.
Although the biosynthetic pathways are similar for each
of these metabolites, the first enzymatic step uses different
precursors for each branched-chain amino acid. Leucine is
ultimately synthesized from acetyl-CoA through 2-oxoiso-
propylmalate, isoleucine from threonine through 2-oxobu-
tyrate, and valine from pyruvate through 2-oxoisovalerate.
The terminal step, converting 2-oxoacids to their corres-
ponding amino acids, is accomplished by two separate
enzymes in spinach chloroplasts (Hagelstein et al., 1997).
The first enzyme completes the synthesis of either leucine
or isoleucine, while the second functions in the synthesis of
valine. The dual function of the leucine/isoleucine amino-
transferase helps to explain the stronger correlation be-
tween leucine and isoleucine (r2=0.941) than between
either metabolite with valine (r2=0.790 and r2=0.822).
However, this scenario also implies a tightly regulated ratio
of 2-oxoisopropylmalate and 2-oxobutyrate, the 2-oxoacid
precursors. Although the mechanisms are not completely
understood, a complex co-ordination of negative feedback
as well as enzymatic capacity and specificity contribute to
this highly controlled regulation (Hagelstein et al., 1997).
Threonine dehydratase (threonine deaminase – TD),
which is induced by wounding or the addition of either
abscisic acid or MeJa in potato (Hildmann et al., 1992) and
tomato (Samach et al., 1995), converts threonine to 2-
oxobutanoic acid. This compound is used in the formation
of isoleucine, which accumulated following MeJa elicit-
ation. This establishes a link between threonine and the
branched chain amino acids made apparent through a mod-
erate correlational relationship (r2=0.45–0.60). Isoleucine,
biosynthetically the most proximal to threonine of the three
branched chain amino acids, was the most strongly correl-
ated to threonine.
Glycine, serine, and threonine metabolism
Comparative correlation analyses revealed altered relation-
ships between metabolites suggestive of the altered activity
levels of particular enzymatic functions, some yet to be
described from plants. For example, there was no correl-
ation between glycine and threonine in unelicited samples,
Leucine Isoleucine Valine
Branched-chain amino acids
Fig. 7. Proposed metabolic model of the elicitation response based on experimental results. With exposure to YE, MeJa, or plating and exposure to UV
light, sucrose levels (red box) decreased and concomitant increases in several other metabolites (yellow boxes) were observed. The subtle accumulation
of putrescine following YE and MeJa elicitation was observed, as well as the accumulation of branched chain amino acids and b-alanine. Polyamine
catabolism provides substrates for b-alanine and GABA synthesis, and b-alanine and 2-oxoisovaleric acid, an intermediate to valine, are incorporated
into pantothenate which is converted subsequently to CoA. Altered CoA synthesis and accumulation is not yet experimentally validated, as no method
exists to profile CoA organic acid esters broadly.
Metabolomics of M. truncatula elicitation11 of 14
by guest on June 3, 2013
but a clear relationship was evident in MeJa-elicited sam-
ples (Fig. 5). In yeast, glycine is biosynthetically linked to
serine by serine hydroxymethyltranferase (EC 184.108.40.206) and
to threonine by threonine aldolase (EC 220.127.116.11) (Woldman
and Appling, 2002). In plants, serine hydroxymethyl-
tranferase is well characterized (McClung et al., 2000);
however, threonine aldolase is yet to be characterized from
plants. The strength of the relationship between glycine and
threonine increased following MeJa elicitation. The sim-
plest explanation is that a threonine aldolase enzymatic
function is present and inducible by MeJa in M. truncatula
cell cultures, illustrating the value of correlation analyses
and metabolomics for the discovery of potentially novel
Queries of the M. truncatula EST datasets utilizing the
yeast amino acid sequence for threonine aldolase (TA)
revealed a TC sequence (TC77640) with 35% amino acid
identity (52% similarity) which was most highly expressed
in libraries from nodulated root, irradiated seedlings,
fungal-elicited cell cultures, and pathogen-infected whole
tissues. In addition, this TC contains a lysine residue
(Lys222) which is highly conserved in fungal (Lys199)
threonine aldolase (Monschau et al., 1998) and is essential
for pyridoxal 59-phosphate binding and catalytic activity
(Liu et al., 1997). TAs are very closely related to serine
hydroxymethyltransferases (SHMT), which share structural
and functional similarities (Contestabile et al., 2001). The
amino acid sequence for MtTC77640 was aligned with
various SHMT and TA sequences from yeast (Sc) E. coli
(Ec), M. truncatula (Mt), and Arabidopsis thaliana (At).
The putative MtTA sequence is similar to yeast and E. coli
threonine aldolase sequences and distinct from the SHMT
sequences (Fig. 8).
Alternatively, altered correlation parameters may be
due to a less direct effect than changes in immediate bio-
synthetic connectivity. Utilizing the pathway reconstruc-
tion tool (PathComp) in the KEGG database (Kanehisa
et al., 2004), enzyme functions were compiled to link
glycine to threonine based on genome sequence data from
Arabidopsis thaliana. This tool could identify no logical
path of less than 15 steps between these two metabolites,
suggesting the existence of either an unidentified mecha-
nism of co-regulation or a novel biosynthetic pathway.
In summary, this report presents a detailed study of
biotic and abiotic stimuli. Significant changes in the relative
abundance of multiple metabolites were observed and are
the result of genetic reprogramming of primary metabolism
in response to stress. Of specific interest are decreased
sucrose, increased branched-chain amino acids, and in-
creased b-alanine levels, suggestive of a generic stress
response. Further, these changes represent repartitioning of
carbon from primary metabolism, specifically sucrose, into
secondary metabolism such as the triterpene saponins and
isoflavonoids. It has been speculated that elevated branched
chain amino acids, putrescine, GABA, and b-alanine,
collectively represent altered CoA biosynthesis, integral
to the elicitation response and partially directed towards
secondary metabolism. In addition, the data support the
presence of a threonine aldolase in M. truncatula which has
currently not been characterized in any plant. The evidence
for both increased CoA metabolism and threonine aldolase
ever, these are credible examples of an ‘omics’ approach
successfully functioning as a discovery platform and pro-
ducing new hypotheses for future investigations. It is per-
ceived that these discovery hypotheses will continue to
this project are integrated.
Two files containing peak area for all metabolites from all
samples used in this study are available as supplemental
materials at JXB online. S1 contains data for polar metab-
olite profiles and S2 contains data for non-polar metabolite
profiles. The files are in a tab-delimited text format. In-
cluded for each analysis is the elicitor used, the time-course
in which the sample was taken (Time_crs), the time after
elicitation (Time (h)), treatment or control, the biological
replicate number (Biol_rep), the injection replicate number
(Inj_rep), the liquid extraction phase (P for polar; L for
lipid or non-polar), and area data for each metabolite.
Metabolite identifiers are encoded as follows: Metabolite-
ID_Retention-time (Extracted-ion). All analyses in this
Fig. 8. Dendrogram demonstrating the amino acid sequence similarity
between M. truncatula (Mt) TC77640, a putative threonine aldolase, and
threonine aldolase genes from yeast (Sc) and E. coli (Ec). The serine
hydroxymethyltransferase genes from A. thaliana (At), yeast, and E. coli
are included to demonstrate that the protein encoded by MtTC77640 is
distinct from the structurally and functionally similar SHMT enzyme
12 of 14 Broeckling et al.
by guest on June 3, 2013
publication were performed on datasets which have been
normalized as described above, S1 and S2 contain non-
The authors would like to thank all those involved in culturing,
elicitation, and harvesting of M. truncatula suspension cultures,
including: Lahoucine Achnine, Courtney Allen, Stacy Allen, Victor
Asirvatham, Naveed Aziz, Jack W Blount, Fang Chen, John Cooper,
Bettina Deavours, Anthony Duran, Patrick Fennell, Xian Zhi He,
Lisa Jackson, Parvathi Kota, Changjun Liu, Srinu Reddy, Gail
Shadle, Shashi Sharma, Hideyuki Suzuki, Ivone Torres-Jerez,
Bonnie Watson, and Deyu Xie, in addition to the authors. We also
thank Anthony Duran for custom PERL scripts used to extract and
analyse GC-MS data. We appreciate the constructive comments
provided by the anonymous reviewers of this manuscript. This
project was funded by a NSF Plant Genome Research Program
Award (DBI-0109732). Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the
author(s) and do not necessarily reflect the views of the National
Science Foundation. Additional personnel and instrumentation sup-
port was provided by The Samuel Roberts Noble Foundation.
Alex D, Bach TJ, Chye ML. 2000. Expression of Brassica juncea
3-hydroxy-3-methylglutaryl CoA synthase is developmentally
regulated and stress-responsive. The Plant Journal 22, 415–426.
Batz O, Logemann E, Reinold S, Hahlbrock K. 1998. Extensive
reprogramming of primary and secondary metabolism by fungal
elicitor or infection in parsley cells. Biological Chemistry 379,
Behboudi S, Morein B, Villacres-Eriksson M. 1999. Quillaja
saponin formulations that stimulate proinflammatory cytokins
elicit a potent acquired cell-mediated immunity. Scandinavian
Journal of Immunology 50, 371–377.
Biondi S, Scaramagli S, Capitani F, Maddalena Altamura M,
Torrigiani P. 2001. Methyl jasmonate upregulates biosynthetic
gene expression, oxidation and conjugation of polyamines, and
inhibits shoot formation in tobacco thin layers. Journal of
Experimental Botany 52, 231–242.
Burtin D, Michael AJ. 1997. Overexpression of arginine decarbox-
ylase in transgenic plants. Biochemical Journal 325, 331–337.
Contestabile R, Paiardini A, Pascarella S, di Salvo ML,
D’Aguanno S, Bossa F. 2001. L-threonine aldolase, serine hydro-
xymethyltransferase and fungal alanine racemase. A subgroup
of strictly related enzymes specialized for different functions.
European Journal of Biochemistry 268, 6508–6525.
Dixon RA, Sumner LW. 2003. Legume natural products: under-
standing and manipulating complex pathways for human and
animal health. Plant Physiology 131, 878–885.
Fiehn O. 2003. Metabolic networks of Cucurbita maxima phloem.
Phytochemistry 62, 875–886.
Flores HE, Filner P. 1985. Polyamine catabolism in higher plants:
characterization of pyrroline dehydrogenase. Plant Growth Regu-
lation 3, 277–291.
Hagelstein P, Sieve B, Klein M, Jans H, Schultz G. 1997. Leucine
sysnthesis in chloroplasts: leucine/isoleucine aminotransferase and
valine aminotransferase are different enzymes in spinach chloro-
plasts. Journal of Plant Physiology 150, 23–30.
Haridas V, Arntzen CJ, Gutterman JU. 2001. Avicins, a family
of triterpenoid saponins from Acacia victoriae (Bentham), inhibit
activation of nuclear factor-kappa B by inhibiting both its nuclear
localization and ability to bind DNA. Proceedings of the National
Academy of Sciences, USA 98, 11557–11562.
Hildmann T, Ebneth M, Pena-Cortes H, Sanchez-Serrano JJ,
Willmitzer L, Prat S. 1992. General roles of abscisic and
jasmonic acids in gene activation as a result of mechanical
wounding. The Plant Cell 4, 1157–1170.
Huhman D, Sumner L. 2002. Metabolic profiling of saponins in
Medicago sativa and Medicago truncatula using HPLC coupled to
an electrospray ion-trap mass spectrometer. Phytochemistry 59,
Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M. 2004.
The KEGG resource for deciphering the genome. Nucleic Acids
Research 32, D277–D280.
Kessmann H, Edwards R, Geno PW, Dixon RA. 1990. Stress
responses in alfalfa (Medicago sativa L.). V. Consitutive and
elicitor-induced accumulation of isoflavanoid conjugates in cell
suspension cultures. Plant Physiology 94, 227–232.
Kupke T, Hernandez-Acosta P, Culianez-Macia FA. 2003. 49-
Phosphopantetheine and coenzyme A biosynthesis in plants.
Journal of Biological Chemistry 278, 38229–38237.
Liu C-J, Blount JW, Steele CL, Dixon RA. 2002. Bottlenecks for
metabolic engineering of isoflavone glycoconjugates in Arabidop-
sis. Proceedings of the National Academy of Sciences, USA 99,
Liu JQ, Dairi T, Kataoka M, Shimizu S, Yamada H. 1997. L-allo-
threonine aldolase from Aeromonas jandaei DK-39: gene cloning,
nucleotide sequencing, and identification of the pyridoxal 59-
phosphate-binding lysine residue by site-directed mutagenesis.
Journal of Bacteriology 179, 3555–3560.
Logemann E, Tavernaro A, Schulz W, Somssich IE, Hahlbrock
K. 2000. UV light selectively coinduces supply pathways from
primary metabolism and flavonoid secondary product formation in
parsley. Proceedings of the National Academy of Sciences, USA
Masgrau C, Altabella T, Farras R, Flores D, Thompson AJ,
Besford RT, Tiburcio AF. 1997. Inducible overexpression of oat
arginine decarboxylase in transgenic tobacco plants. The Plant
Journal 11, 465–473.
Mazza CA, Boccalandro HE, Giordano CV, Battista D, Scopel
AL, Ballare CL. 2000. Functional significance and induction by
solar radiation of ultraviolet-absorbing sunscreens in field-grown
soybean crops. Plant Physiology 122, 117–126.
McClung CR, Hsu M, Painter JE, Gagne JM, Karlsberg SD,
Salome PA. 2000. Integrated temporal regulation of the photo-
respiratory pathway. Circadian regulation of two Arabidopsis
genes encoding serine hydroxymethyltransferase. Plant Physi-
ology 123, 381–392.
Monschau N, Sahm H, Stahmann K. 1998. Threonine aldolase
overexpression plus threonine supplementation enhanced ribofla-
vin production in Ashbya gossypii. Applied Environmental Micro-
biology 64, 4283–4290.
Norman EG, Walton AB, Turpin DH. 1994. Immediate activation
of respiration in Petroselinum crispum L. in response to the
Phytophthora megasperma f. sp. glycinea elicitor. Plant Physi-
ology 106, 1541–1546.
Oh S, Kinjo J, Shii Y, Ikeda T, Nohara T, Ahn K, Kim J, Lee H.
2000. Effects of triterpenoids from Pueraria lobata on immuno-
hemolysis: b-D-glucuronic acid plays an active role in anticomple-
mentary activity in vitro. Planta Medica 66, 506–510.
Oleszek W, Junkuszew M, Stochmal A. 1999. Determination and
toxicity of saponins from Amaranthus cruentus seeds. Journal of
Agricultural and Food Chemistry 47, 3685–3687.
Osbourn A. 2003. Saponins in cereals. Phytochemistry 62, 1–4.
Papadopoulou K, Melton RE, Leggett M, Daniels MJ, Osbourn
AE. 1999. Compromised disease resistance in saponin-deficient
Metabolomics of M. truncatula elicitation13 of 14
by guest on June 3, 2013
plants. Proceedings of the National Academy of Sciences, USA 96, Download full-text
Perez-Amador MA, Leon J, Green PJ, Carbonell J. 2002.
Induction of the arginine decarboxylase ADC2 gene provides
evidence for the involvement of polyamines in the wound response
in Arabidopsis. Plant Physiology 130, 1454–1463.
Piotrowski M, Janowitz T, Kneifel H. 2003. Plant C-N hydrolases
lase involved in polyamine biosynthesis. Journal of Biological
Chemistry 278, 1708–1712.
Robinson SA, Stewart GR, Phillips R. 1992. Regulation of
glutamate dehydrogenase activity in relation to carbon limitation
and protein catabolism in carrot cell suspension cultures. Plant
Physiology 98, 1190–1195.
Roessner U, Luedemann A, Brust D, Fiehn O, Linke T, Will-
mitzer L, Fernie AR. 2001. Metabolic profiling allows compre-
hensive phenotyping of genetically or environmentally modified
plant systems. The Plant Cell 13, 11–29.
Roessner-Tunali U, Urbanczyk-Wochniak E, Czechowski T,
Kolbe A, Willmitzer L, Fernie AR. 2003. De novo amino acid
biosynthesis in potato tubers is regulated by sucrose levels. Plant
Physiology 133, 683–962.
Samach A, Broday L, Hareven D, Lifschitz E. 1995. Expression of
an amino acid biosynthesis gene in tomato flowers: developmental
upregulation and MeJa response are parenchyma-specific and
mutually compatible. The Plant Journal 8, 391–406.
Schenk U, Hilderbrandt AC. 1971. Medium and techniques for
induction and growth of monocotyledonous and dicotyledonous
plant cell cultures. Canadian Journal of Botany 50, 199–204.
Schumacher H-M, Gundlack H, Fiedler F, Zenk MH. 1987.
Elicitation of benzophenathridine alkaloid synthesis in Eschscholt-
zia cell cultures. Plant Cell Reproduction 6, 410–413.
Sumner L, Mendes P, Dixon R. 2003. Plant metabolomics: large-
scale phytochemistry in the functional genomics era. Phytochem-
istry 62, 817–836.
Suzuki H, Achnine L, Xu R, Matsuda S, Dixon R. 2002. A
genomics approach to the early stages of triterpene saponin
biosynthesis in Medicago truncatula. The Plant Journal 32,
Swiatek A, Van Dongen W, Esmans EL, Van Onckelen H. 2004.
Metabolic fate of jasmonates in tobacco bright yellow-2 cells.
Plant Physiology 135, 161–172.
Tava A, Odoardi M. 1996. Saponins from Medicago spp.: chemical
characterization and biological activity against insects. Advances
in Experimental Medicine and Biology 405, 97–109.
Taylor J, King RD, Altmann T, Fiehn O. 2002. Application of
metabolomics to plant genotype discrimination using statistics and
machine learning. Bioinformatics 18, S241–S248.
Trethewey RN, Krotzky AJ, Willmitzer L. 1999. Metabolic
profiling: a rosetta stone for genomics? Current Opinion in
Biotechnology 2, 83–85.
VandenBosch KA, Stacey G. 2003. Summaries of legume genomics
projects from around the globe: community resources for crops and
models. Plant Physiology 131, 840–865.
Waller GR, Jurzysta M, Thorne RLZ. 1993. Allelopathic activity
of root saponins from alfalfa (Medicago sativa L.) on weeds and
wheat. Botanical Bulletin of Academia Sinica 34, 1–11.
Walsh TA, Green SB, Larrinua IM, Schmitzer PR. 2001.
Characterization of plant beta-ureidopropionase and functional
overexpression in Escherichia coli. Plant Physiology 125, 1001–
Walters D, Cowley T, Mitchell A. 2002. Methyl jasmonate alters
polyamine metabolism and induces systemic protection against
powdery mildew infection in barley seedlings. Journal of Experi-
mental Botany 53, 747–756.
Weckwerth W, Loureiro ME, Wenzel K, Fiehn O. 2004. Differ-
ential metabolic networks unravel the effects of silent plant
phenotypes. Proceedings of the National Academy of Sciences,
USA 101, 7809–7814.
White WH, Gunyuzlu PL, Toyn JH. 2001. Saccharomyces
cerevisiae is capable of de novo pantothenic acid biosynthesis
involving a novel pathway of b-alanine production from spermine.
Journal of Biological Chemistry 276, 10794–10800.
Woldman Y, Appling DR. 2002. A general method for determining
the contribution of split pathways in metabolite production in the
yeast Saccharomyces cerevisiae. Metabolomic Engineering 4,
Zar J. 1999. Biostatistical analysis. London: Prentice-Hall.
14 of 14 Broeckling et al.
by guest on June 3, 2013