Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123

Contents lists available at SciVerse ScienceDirect

Biocatalysis and Agricultural Biotechnology

journal homepage: www.elsevier.com/locate/bab

Progeny of tobacco mosaic virus-infected Nicotiana tabacum plants exhibit

trans-generational changes in metabolic profiles
Rupasri Mandal a,b, Palak Kathiria c, Nikolaos Psychogios a,b, Souhaila Bouatra a,b,
Ramanarayan Krishnamurthy a,b, David Wishart a,b, Igor Kovalchuk c,n
a
Departments of Computing Science, University of Alberta, Edmonton, AB, Canada T6G 2E8
b
Department of Biological Sciences, University of Alberta, Edmonton, AB, Canada T6G 2E8
c
Department of Biological Sciences, University of Lethbridge, Lethbridge, AB, Canada T1K 3M4

a r t i c l e i n f o abstract

Article history: Our previous work showed that infection of Nicotiana tabacum (tobacco) plants with tobacco mosaic
Received 18 October 2011 virus (TMV) results in transgenerational changes in plant phenotype. To test whether the progeny of
Received in revised form infected (PI) plants have changes in metabolites, we analyzed their metabolic profiles using NMR
21 December 2011
spectroscopy and gas chromatography–mass spectrometry. Our results showed that PI plants had
Accepted 23 January 2012
Available online 31 January 2012
higher levels of various amino acids such as alanine, valine, serine, threonine, leucine and glutamine as
well as various sugars, including sucrose, glucose and fructose. This work shows that viral infections
Keywords: can lead to trans-generational or heritable metabolic changes to plants.
Tobacco & 2012 Elsevier Ltd. All rights reserved.
Tobacco mosaic virus
Progeny of infected plants
Metabolites
NMR
GC–MS

1. Introduction to TMV. As a result, Big Havana cultivars, when infected with TMV,
exhibit classical features of an incompatible interaction (Whitham
Viral infections in plants are perhaps among the most well et al., 1994). In contrast, the compatible interaction is characterized
studied plant/pathogen interactions. Understanding how plant by viral spread from the site of infection, and it occurs if either the
viruses spread through plants and kill their hosts or become localized virus or the plant lacks the components for recognition (the Avr or R
leading to a chronic state of infection has important implications in gene, respectively). The tobacco cultivar SR1 does not have an active
the prevention, treatment and containment of many crop viruses. In N-gene and thus, if infected with TMV, it cannot restrict viral
particular, certain types of plant viral infections can result in viral movement (Whitham et al., 1994).
localization at the site of infection; this type of interaction is called The incompatible interaction results in what is termed as an
‘‘incompatible’’ and occurs when the virus carries an avirulence gene immediate local hypersensitive response (HR) which consists of
(Avr) and the plant carries a resistance (R) gene (Jones and Dangl, the activation of a salicylic acid-dependent pathway leading to
2006). The most common and well-described case of incompatible massive radical production, cell death and localization of viral
interaction is the infection of certain cultivars of Nicotiana tabacum progression (Loebenstein, 2009; Grant and Lamb, 2006). This is
(tobacco) plants with tobacco mosaic virus (TMV) (Whitham et al., followed by a systemic increase in pathogen tolerance, named
1994). The tobacco cultivar known as ‘‘Big Havana’’ carries in its systemic acquired resistance (SAR). However, the compatible
genome a specific resistance gene, the N-gene, that makes it resistant interaction does not lead to HR or SAR, but it does result in
an immediate plant response in the form of increased radical
production. Apparently, such response occurs regardless of
Abbreviations: Avr, avirulence gene; GC–MS, chromatography–mass spectro- whether the plant is resistant (incompatible) or sensitive (com-
metry; HR, hypersensitive response; NMR, nuclear magnetic resonance; PC, the patible) to bacterial pathogen (Grant et al., 2000). A second, larger
progeny of control plants; PI, the progeny of infected plants; R, resistance gene; burst of radicals occurs only if the plant has the required R gene.
SAR, systemic acquired resistance; TMV, tobacco mosaic virus It remains to be shown whether a similar phenomenon also takes
n
Correspondence to: Department of Biological Sciences, University of
Lethbridge, University Drive 4401, Lethbridge, T1K 3M4, AB, Canada.
place upon viral infection.
Tel.: þ1 403 329 2579; fax: þ 1 403 329 2242. Local infection of SR1 tobacco plants with TMV results in viral
E-mail address: igor.kovalchuk@uleth.ca (I. Kovalchuk). spread to non-infected tissues. This typically occurs within 48 h

1878-8181/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bcab.2012.01.004
116 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123

post infection, with the first symptoms appearing at 7–8 days had drastic changes in genome stability, stress tolerance and methy-
post infection (Kovalchuk et al., 2003; Ouko et al., 2010). Despite lation pattern, we hypothesized that they would also undergo
not having a TMV resistance gene, these SR1 cultivars are metabolic changes that allow them to tolerate pathogen infection.
generally able to mount a response to TMV infection. Since the To test the hypothesis, we checked levels of metabolites in control
infection of SR1 plants by TMV results in a compatible interaction, and infected plants and found the progeny of infected plants to have
it is possible that the local TMV infection results in a local a dramatically different metabolome as compared to the progeny of
transient increase in radical production (the first burst of radical non-infected plants.
production) leading to some sort of a system-wide pathogen
warning signal (Loebenstein, 2009; Grant and Lamb, 2006).
In plants that have an R-gene, the second radical burst may 2. Materials and methods
inactivate the initial warning signal. As a result, plants that do not
contain an appropriate R gene have the warning signal which 2.1. Infection of SR1 plants with TMV and generating the progeny of
systematically spreads throughout all plant tissues. The findings infected plants
of dramatic changes in the transcriptome of Arabidopsis plants
exposed to a compatible viral or bacterial pathogen support this The generation of a transgenic tobacco line #A of cultivar SR1
hypothesis (Love et al., 2005; De Vos et al., 2005). Tao et al. (2003) plants carrying in the genome a single copy of the luciferase (LUC)
actually suggested that a large part of the differences in tran- recombination substrate was described previously (Kovalchuk et al.,
scriptional changes between compatible and incompatible inter- 2003). For these experiments, the single leaves of twenty Nicotiana
actions upon infection by bacterial pathogens is quantitative tabacum cultivar SR1 plants were infected with 200 ng (100 ng/ml in
rather than qualitative. phosphate buffer) of the TMV strain U1. Twenty control plants were
As plants do not have a predetermined germline and plant mock-inoculated with phosphate buffer only. Carborundum was
cells are frequently totipotent, the changes triggered by a signal used as an abrasive to facilitate the viral infection. The experiments
from the local infection sites are likely to be transmitted to the were carried out in two independent rounds, with 10 plants from
next generation, thereby allowing the later progeny to be better each group in each round. The infected and mock-inoculated plants
prepared for similar pathogen infections (Boyko and Kovalchuk, were kept at 32 1C to allow the faster progression of the viral
2011). Our previous experiments showed that the local infection infection. At 24–32 h after infection, the infected leaves were cut,
of tobacco plants by a compatible pathogen TMV leads to a and the non-infected upper leaves were used to graft the naı̈ve non-
systemic increase in genome instability (Kovalchuk et al., 2003). infected plants. Tops of the plants were removed to promote
A more recent report showed that such infection caused an development of lateral shoots; plants were allowed to produce seeds.
increase in instability of N-gene-like R-gene loci in the progeny. Cutting leaves at 24–32 h post infection prevented the spread of
This type of a response was paralleled by global genome hyper- virus to non-infected tissue (as checked by PCR) and symptom
methylation and hypomethylation at the R-gene loci (Boyko et al., appearance at two weeks after infection (Kovalchuk et al., 2003).
2007). This is a remarkable observation as the infected plants The progeny of plants grafted with leaves from infected and
appeared to belong to the SR1 cultivar, a cultivar that does not non-infected plants were used for the analysis of metabolites. The
contain an active N-gene. The changes observed in the R-gene loci progeny plants were also tested for the presence of TMV, and no
that exhibit similarity to the N-gene could be an indication of a virus was found (data not shown).
transgenerational response to the infection by a compatible
pathogen. Our most recent work indicated that the progeny of 2.2. Infection of the progeny of infected and progeny of control
infected tobacco plants had an increase in the spontaneous non- plants with TMV
induced recombination frequency, the increased tolerance to TMV
and a higher level of certain phenolic compounds (Kathiria et al., To prepare material for metabolomic analysis, a total of twenty
2010). three-week-old plants corresponding to the progeny of infected
An increase in tolerance to viral and fungal pathogens together (10 plants) and control plants (10 plants) were used. The plants
with an increase in the level of phenolics in the progeny of infected were grown under controlled light and temperature conditions
plants suggest that metabolites may play an important role. The (16 h/light at 22 1C and 8 h/dark at 18 1C with illumination of
resistance of the extremophile shrub T. halophila, a distant relative of 300 mmol m  2 s  1). For the progeny of infected plants, we used
Arabidopsis, to temperature extremes is in part related to high steady- two independent lines corresponding to the progeny of infected
state levels of metabolites (Gong et al., 2005). Plants produce a plants from line #9 (PI#9) and the progeny of infected plants from
tremendous number of secondary compounds that allow them to line #19 (PI#19); for the progeny of control plants (PC), we used
respond to various abiotic and biotic stresses. A comprehensive five plants, each from two independent lines, PC#1 and PC#3. The
overview of various groups of secondary metabolites used by plants three-week-old plants were either infected with TMV (100 ng per
in their defense against pathogens has been recently published (Leiss each infected leaf; 200 ng total) or mock-treated. To reduce the
et al., 2011). Several methods for metabolome profiling are number of samples for metabolic profiling, we combined plants
known, including liquid chromatography–mass spectrometry (LC/ from PC#1 and PC#3. The samples (2 treated leaves from each of
MS), capillary electrophoresis–mass spectrometry (CE/MS), gas five infected plants per line) were collected 48 h after treatment;
chromatography–mass spectrometry (GC–MS) and nuclear magnetic the control samples were collected before infection. Three inde-
resonance spectroscopy (NMR), with the latter two being the most pendent repeats performed at intervals of two weeks each were
commonly used. NMR spectroscopy measures the resonances of made. Thus, we obtained 27 samples, 3 samples of PI#9, PI#19
magnetic nuclei such as 1H, 13C and 15N that interact with an external and PC plants at time 0, 3 samples of mock-treated leaves of PI#9,
magnetic field (Leiss et al., 2011). NMR requires minimal sample PI#19 and PC plants at 48 h and 3 samples of virus-infected PI#9,
preparation and allows the analysis of metabolites in crude extracts PI#19 and PC plants at 48 h.
with no restrictions relating to volatility or polarity (Leiss et al., 2011).
GC–MS is a combination of gas chromatography and mass spectro- 2.3. Metabolite extraction
metry that provides a substantially increased separation capacity,
chemical selectivity and sensitivity for metabolomics analyses (Kim One gram of tissue from each sample was placed into a 50 mL
et al., 2011). Given the fact that the progeny of TMV infected plants glass tube and quickly immersed in 3 mL isopropanol at 75 1C
 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123 117

(preheated) with 0.01% BHT (butyrated hydroxytoluene) (Sigma- 2.6. GC–MS compound identification and quantification
Aldrich) for 15 min. 1.5 mL chloroform and 0.6 mL water were
added, and the mixture was vortexed; then the solution was The identification of metabolites and the quantification of
agitated at room temperature for 1 h. The liquid extracts were metabolite concentrations are often much more tedious using
then transferred to a centrifuge tube. Four milliliters of the 2:1 GC–MS than NMR spectroscopy. To derivatize the metabolites for
chloroform:methanol solution with 0.01% BHT were again added GC–MS analysis, 40 mL of 20 mg/mL methoxyamine hydrochloride
to the leaves and shaken for 30 min. The liquid extracts were then (Sigma-Aldrich) in ACS grade pyridine was added to the water-
transferred back to the original centrifuge tube. soluble extracts of tobacco leaves, and the mixture was incubated
This extraction procedure was repeated until the leaves of at room temperature for 16 h. Then, 50 mL (N-Methyl-N-trifluor-
every sample became white (usually about 5 extractions were oacetamide) MSTFA (Thermo Scientific) with the 1% TMCS (Tri-
needed, including the one with isopropanol). One milliliter of 1 M methylchlorosilane) derivatization agent was added and
KCl was added to the combined extract, vortexed, centrifuged and incubated at 37 1C for 60 min on a hotplate. The samples were
the upper phase (containing water soluble metabolites) was vortexed twice throughout the incubation period to ensure
collected. A second extraction was performed with 2 mL of complete dissolution. The samples were refrigerated at 4 1C for
distilled water which was added to the lower phase, vortexed, no longer than 48 h prior to the analysis in order to avoid any
centrifuged and the upper phase was collected. All water soluble degradation of derivatized compounds.
extracts were combined and stored in the freezer at 20 1C until The derivatized extracts were analyzed using an Agilent 7890-
further analysis. 5975C GC–MS instrument operating in an Electron Impact (EI)
ionization mode. For GC–MS analysis, 2 mL of the derivatized
tobacco leaf solution were injected using a split/splitless injector
2.4. NMR sample preparation and spectroscopy
onto a HP-5MS capillary column (Agilent J&W Scientific,
30 m  250 mm  0.25 mm) with a split ratio of 5:1. The injector
The solution of tobacco leaf extracts (1 mL) was evaporated to
port temperature was held at 250 1C, and the helium carrier gas
dryness using a Speedvac concentrator. The dried samples
flow rate was set to 1 mL/min at the initial oven temperature of
(0.035 g) were reconstituted in 250 mL H2O. Fifty microliters of
70 1C. The oven temperature was increased at 1 1C/min to 76 1C,
50 mM NaH2PO4 buffer (pH 7), 35 mL of D2O and 15 mL of a
and then at 6.1 1C/min to 310 1C. The total run time was 45 min.
standard buffer solution (3.73 mM DSS (disodium-2,2-dimethyl-
The full scan mode of the quadrupole MS was used at a mass
2-silapentane-5-sulfonate and 0.47% NaN3 in H2O)) were added to
range of 50–500 m/z, with a solvent delay of 6 min. The MS ion
the samples. The sample was vortexed for 1 min and sonicated for
source temperature was 230 1C, and the Quadrupole temperature
30 min, and then transferred to a standard Shigemi microcell
was 150 1C. In GC–MS, a faster scan speed generally provides
NMR tube (matched with water) for the subsequent spectral
more data points across a chromatographic peak, but it tends to
analysis. All 1H-NMR spectra were collected on a 500 MHz Inova
lower the ion statistics. In contrast, a slower scan rate produces a
(Varian Inc., Palo Alto, CA) spectrometer equipped with a 5 mm
few scans across a peak and results in better spectra. The scan
HCN Z-gradient pulsed-field gradient (PFG) room-temperature
speed of our quadrupole MS was optimized over a number of
probe. 1H-NMR spectra were acquired at 25 1C using the first
samples. It has been found that a relatively slow scan rate of
transient of the tnnoesy-presaturation pulse sequence, which was
1.7 scans/s gives the best results; therefore, it was used through-
chosen for its high degree of quantitative accuracy. Spectra were
out all experiments.
collected with 4096 transients using a 4 s acquisition time and a
The AMDIS spectral deconvolution software (Version 2.62)
1 s recycle delay.
from NIST (National Institute of Standards and Technology) was
used to process the total ion chromatogram and the EI-MS
2.5. NMR compound identification and quantification spectra of each GC peak. After deconvolution, the purified mass
spectrum of each of the trimethylsilated metabolites was
All FIDs were zero-filled to 64k data points and subjected to identified using the NIST MS Search program (version 2.0d),
line broadening of 0.5 Hz. The singlet produced by the DSS methyl which was linked to the NIST mass spectral library (2005).
groups was used as an internal standard for chemical shift Retention Indices (RIs) were calculated using a C8–C20 alkane
referencing (set to 0 ppm) and for compound quantification. All mixture standard (Fluka, Sigma-Aldrich). Metabolites were iden-
1
H-NMR spectra were processed and analyzed using the Chenomx tified by matching the EI-MS spectra with those of reference
NMR Suite Professional software package version 5.0 (Chenomx compounds from the NIST library. In addition, safety checks were
Inc., Edmonton, AB). The Chenomx NMR Suite software allows for performed using additional published GC–MS libraries (Schauer
qualitative and quantitative analysis of an NMR spectrum by et al., 2005) for the rapid identification of metabolites in complex
manually fitting spectral signatures of metabolites from an inter- biological samples. Peaks having no match to the published
nal database of metabolite spectra to the spectrum of interest retention indices and/or no match to the AMDIS GC/MS spectral
(Weljie et al., 2006). Specifically, the spectral fitting for each library were identified by matching the experimental RI of each
metabolite was done using the standard Chenomx 500 MHz, pH metabolite with an in-house RI library (containing 312 TMS-
4–9 metabolite library. It has been previously shown that this derivatized metabolites) developed in our laboratory (Wishart
fitting procedure provides absolute concentration accuracies of et al., 2009). Thirty-five pure standards obtained from the Human
90% or better (Saude et al., 2004). Moreover, concentration data Metabolome Library (Wishart et al., 2007) were run through
was corrected for bandpass filter attenuation as previously the GC–MS (using the same protocol described above) to confirm
described (Saude et al., 2009). Each spectrum was processed their identity, retention indices and EI spectra; and they
and analyzed by at least two NMR spectroscopists to minimize were subsequently used for producing external five-point calibra-
compound misidentification and misquantification. We also used tion curves (for the quantification). Where peak baseline
sample spiking to confirm the identities of a number of com- resolution was not observed, peak deconvolution software was
pounds. Sample spiking involves adding 20–200 mM of the sus- used to separate peaks based on Gaussian shape recognition.
pected compound to the selected tobacco leaf samples and Using the mass spectral information obtained in this manner,
examining whether the relative NMR signal intensity changed peaks were successfully identified and peak areas successfully
as expected. quantified.
118 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123

2.7. Statistical treatment of data performed with non-stressed plants (at the 0 h time point), mock-
treated plants (at the 48 h time point) and TMV-infected plants
In all cases, either the mean and standard error or the standard (at the 48 h_V time point).
deviation was calculated. The statistical significance of the
experiment was confirmed either by the two-tailed paired Stu-
dent’s t-test with a ¼0.05 or a ¼ 0.1 (comparing data from two 3.2. NMR identification and quantification
treatments) or by the single-factor ANOVA (comparing data from
three or more treatments/time points). The statistical analysis As seen in Fig. 1, the NMR spectra from the tobacco leaf
was performed using the JMP 5.0 software (SAS Institute Inc.). samples are relatively simple. Typically, 90% of all visible peaks in
each NMR spectrum could be definitively assigned to a com-
pound, and more than 90% of the spectral area could be routinely
3. Results
fitted using the Chenomx spectral analysis software. As shown in
Fig. 1, most of the visible peaks are annotated with a compound
3.1. Experimental setup and generation of samples
name. Among the 27 tobacco leave samples, a total of 20
compounds (Table 1) were routinely identified and quantified,
For all the experiments described here, we used tobacco
with the most abundant compounds being malic acid (0.9 mM),
cultivar SR1 plants (Boyko et al., 2007). Single leaves of three
fructose (0.6 mM), glucose (0.5 mM), nicotine (0.5 mM) and cho-
week-old plants were infected with 200 ng (100 ng/leaf) of TMV,
line (0.3 mM). In general, the pattern was similar for all samples;
whereas control plants were mock-treated. The seeds from the
however, for a few samples, some metabolites (such as sucrose,
collected plants were used for future analysis. The plantlets from
the progeny of infected plants (PI) were tested for viral presence
Table 1
by Real-Time PCR and the plants were found to be virus-free (data Unique metabolites detected by NMR and GC–MS.
not shown). The progeny of the control plants (PC) were also
confirmed to be virus-free. NMR-specific GC–MS-specific Common: NMR/GC–MS
Based on previous studies (Boyko et al., 2007; Kathiria et al.,
2-Methylglutarate Butanoic acid 3-Hydroxybutyrate
2010), we hypothesized that PI plants would have different
3-Hydroxyisovalerate Propanoic acid 4-Aminobutyrate
metabolic expression profiles compared to PC plants. For the Choline Acetamide Alanine
analysis of metabolites, we selected two independent lines, PI#9 Formate Valine Fructose
and PI#19, from the progeny of infected plants and two lines from Fumarate Butylamine Glucose
Glutamate Leucine
the progeny of control plants. In our previous studies, line PI#9
Glutamine Malate
was analyzed in details as to changes in genome stability, its Methanol Serine
tolerance to various abiotic and biotic stresses, PR expression and Nicotine Threonine
callose deposition. In contrast, line PI#19 was only used for the Phenylalanine Sucrose
analysis of genome instability and response to TMV. To test the
The table shows a list of detectible metabolites as measured by NMR or GC–MS.
level of metabolites in PI and PC plants under non-induced and
‘NMR-specific’ shows the list of metabolites detected only by NMR. ‘GC–MS-
TMV-induced conditions, we infected three-week-old PI and PC specific’ shows the list of metabolites detected only by GS–MS. ‘NMR/GC–MS’
plants with 200 ng of TMV. Thus, the analysis of metabolites was shows the list of metabolites detected by both methods.

12

15 1
15

1515

15
15
Signal intensity

16
13
16

13 10 10
10 15 13
13 7
7
9 9 7 6
14 8
16 11 3
1
11 17 3 4 1 5 1
11 2 2
4

Fig. 1. A typical 500 MHz 1H-NMR spectrum of a tobacco leaf sample. The figure shows a typical image of the 500 MHz 1H-NMR spectrum of a tobacco leaf. This particular
image corresponds to the control sample taken at 0 h post infection. The numbers indicate the following compounds: 1. DSS (disodium-2,2-dimethyl-2-silapentane-5-
sulfonate; served as a Chemical Shape Indicator and an internal standard for the quantification); 2. Leucine; 3. 3-hydroxyisovalerate; 4. 2-methylglutarate; 5. Threonine; 6.
Alanine; 7. 4-aminobutyrate; 8. Glutamate; 9. Glutamine; 10. Malate; 11. Nicotine; 12. Choline; 14. Methanol; 15. Fructose; 16. Sucrose; 17. Phenylalanine. The details of
NMR are presented in Methods.
 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123 119

fructose, glucose and serine) could not be identified/quantified, lines) were observed in 15 out of 25 unique metabolites (Fig. 4,
but they were present in large quantities in other samples. Figs. 2S–5S). Metabolites were grouped into several different
groups: amino acids (alanine, glutamate, leucine, phenylalanine,
3.3. GC–MS identification and quantification serine, threonine and valine) (Fig. 4, Fig. 2S); organic acids
(butanoic acid, malate, fumarate, propanoic acid) (Fig. 4,
In our GC–MS spectra, most of the visible peaks could be Fig. 3S); sugars (fructose, glucose, sucrose) (Fig. 4, Fig. 3S); all
annotated with a compound name. In contrast to NMR where other detected metabolites (Fig. 4, Figs. 4S, 5S). Two general
peak coverage approached 90%, the fraction of the GC–MS peaks trends were observed for amino acids group: (a). there was a
that could be positively identified was approximately 70%; 15
different compounds were identified (Table 1).
A comparison of metabolite patterns detected by NMR and
GC–MS methods determined a common set of 10 compounds;
while 10 compounds, which could not be found by the GC–MS
methods, were detected by NMR (Table 1). Additionally, the GC–
MS analysis led to the detection of 5 compounds that NMR could
not detect (Table 1).

3.4. The progeny of infected plants exhibited substantial differences

in metabolite levels

The analysis of differences in metabolite levels between PC

and PI plants showed that both PI lines, PI#9 and PI#19, had the
higher total concentration of detected metabolites prior to infec-
tion (Fig. 1S). The total metabolite concentration was by  100%
higher in PI#9 and by 50% higher in PI#19 lines as compared in
PC lines (Fig. 1S; 0 h samples). A t-test indicated that the
concentration data for line PI#9 was significantly different,
whereas the data for line #19 was not (Po0.01 and P40.1,
respectively). A comparison of concentrations measured by NMR
and GC–MS methods showed a very high degree of similarity;
with six different metabolites showing identical concentrations
analyzed for all three groups in the 0 h samples (Fig. 2, Fig. 3). Fig. 3. A comparison of fold difference in the amounts of metabolites as measured
by NMR and GC–MS. Fold difference in the amounts of metabolites between the
The analysis of individual metabolites in the 0 h samples PI#9 (A) and PC or PI#19 (B) and PC samples collected from the non-infected
showed their concentrations were higher in PI plants as compared plants was analyzed. The comparison of fold differences is shown for NMR and
to PC plants, although significant differences (at least one of the PI GC–MS analyses. ‘‘3-HB’’ is 3-Hydroxybutyrate, ‘‘4-AB’’ is 4-Aminobutyrate.

Fig. 2. A comparison of the average metabolite concentrations in the non-infected PC, PI#9 and PI#19 plants as measured by NMR and GC–MS. The concentrations of
metabolites in the non-infected tissue of PC, PI#9 and PI#19 plants at the 0 h time point was compared for both NMR and GC–MS methods. ‘‘3-HB’’ is 3-Hydroxybutyrate,
whereas ‘‘4-AB’’ is 4-Aminobutyrate. (A) Shows the comparison of the total amounts of all metabolites. The asterisks show a significant difference between the treated and
control samples as measured for line PI#9 (Po 0.05). (B) Shows the comparison of six common metabolites as measured in the PC plants. (C) Shows the comparison of six
common metabolites as measured in the PI#9 plants. (D) Shows the comparison of six common metabolites as measured in the PI#19 plants.
120 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123

Fig. 4. Heat map showing the changes in metabolites as measured by NMR and GC–MS. Fold difference in the amounts of metabolites in the PC, PI#9 or PI#19 samples
collected from the non-infected plants at 0 h, mock-treated plants at 48 h and virus-infected plants at 48 h (48 h_V) was calculated by relating values of each metabolite to
the level of this metabolite in PC plants at 0 h (taken as ‘‘1’’). The data for the fold difference are shown as a heat map. Boxes showing significant differences are labeled
with asterisks: one—Po 0.05; two—P o 0.01; three—P o0.001.

higher level of most amino acids in PI#9 plants as compared to PC showed that most metabolite concentrations were similar. How-
plants at 0 h (the difference for serine and leucine was not ever, after infection, higher levels were still observed in 9 out of 26
significant) as well as in PI#19 plants as compared to PC plants, unique metabolites in at least one of the PI lines. At the same time,
although the differences were significant only for valine and the level of sugars (such as fructose, glucose and sucrose) dropped
glutamine; (b). there was an increase in the amount of amino substantially in PC, PI#9 and PI#19 plants (Fig. 4). In fact, the
acids in the samples of 48 h and 48 h_V in PC plants but not in PI concentrations of fructose and sucrose fell below detectable levels
plants, it was especially noticeable in the 48 h_V samples (Fig. 4, (Fig. 4).
Fig. 2S). These data showed that exposure to mechanical stress
(mock-treatment with carborundum) or exposure to TMV causes
an increase in amino acid amounts in PC plants, but not in PI 4. Discussion
plants. For sugars, we found higher levels of all sugars in PI plants
at 0 h and the amount of all sugars drastically dropped after mock In plants, metabolites play a major role in many resistance and
treatment (48 h) or viral infection (48 h_V) (Fig. 4; Fig. 3S). For stress responses. The identification and quantification of metabo-
organic acids group, higher levels were observed in PI plants at lites is of paramount importance for studies of the dynamics of
0 h and the amount of malate and fumarate dropped in all the metabolome, the analysis of fluxes in metabolic pathways and
samples at 48 h and 48 h_V (Fig. 4; Fig. 3S). In the group of the detection of the role of each metabolite following various
remaining metabolites, the amounts were higher for most meta- stimuli. Here, we tested and compared metabolic profiles of the
bolites in PI#9 line at 0 h and 48 h (Fig. 4; Figs. 4S and 5S). progeny of two different infected plants and the progeny of non-
It is interesting to note that the concentration of most metabo- treated control plants.
lites was found to be higher in line PI#9 as compared to line PI#19, GC–MS identified a lower number of metabolites as compared
with differences being most obvious for 4-aminobutyrate, 3-hydro- to NMR. This relatively low level of coverage is a common
xyisovalerate, 3-hydroxybutyrate, choline, threonine, alanine and problem in global or untargeted GC–MS metabolomics studies
fumarate (Fig. 4; Figs. 2S–5S). The analysis of changes in the (Schauer et al., 2005; Kopka, 2006). It is possible that some
infected 48 h_V samples and in the mock-inoculated 48 h samples of the unidentified peaks may be non-metabolic by-products of
 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123 121

chemical derivatization, or they may be degraded by-products of between soluble sugars (sucrose, fructose and glucose) and
derivatized metabolites. The NIST mass spectra library contains organic and amino acids is perturbed. A decrease in soluble sugar
only a relatively small portion of metabolically relevant com- contents in infected tissue has been observed in Chinese cabbage
pounds, and the lack of a comprehensive GC–MS library for plant infected by Turnip yellow mosaic virus (TYMV) and in squash
metabolites is likely an equally important reason for this incom- plants infected by Squash mosaic virus (SqMV). In contrast, sucrose
plete coverage. NMR and GC–MS identified slightly different type levels in Zucchini yellow mosaic virus (ZYMV)-infected marrow
of metabolites. The reasons for these differences could be mani- plants increased relative to levels in healthy controls (Handford
fold. Compounds might be identified by NMR but not in GC–MS, if and Carr, 2007). A metabolomic approach to study the effects
the metabolites of interest were either too volatile for GC–MS of viral infection on carbohydrate partitioning in A. thaliana was
detection, lost in sample preparation or eluted during the solvent carried out by feeding Col 0 plants with 13CO2 17 days after
delay. Compounds might be found by GC–MS but not by NMR, if inoculation with two strains of TMV known to cause mild or
their concentrations were below the NMR detection limits. In severe symptoms in tobacco (Handford et al., 2004). The results
general, GC–MS appears to have better sensitivity (sub-mM) demonstrated that the incorporation of 13C-label into carbohy-
compared to NMR spectroscopy (a few mM) (Wishart, 2008). In drates, amino acids and organic acids was 2–4 times higher in the
all cases, the existence of NMR detectable metabolites was virus-inoculated plants compared to the mock-inoculated control
explicitly checked in our GC–MS analyses and vice versa. (Handford et al., 2004). Thus, in a wide variety of plant–virus
The progeny of stressed plants had higher concentrations of interactions, there is a shift in carbon partitioning away from
sugars (only glucose was significantly different for PI#9 line) and soluble sugars towards organic and amino acids, although this is
amino acids (the differences for alanine, glutamine, phenylala- not necessarily the case in every system analyzed. Additionally, it
nine, threonine and valine) (Fig. 4; Figs. 2S and 3S). Sucrose, should be noted that a similar effect was also observed in the
glucose and fructose are the major carbohydrate storage com- mock-treated plants. Therefore, we assumed that mechanical
pounds in the majority of plants, compared to most other plant wounding caused by rubbing with carborundum may also lead
metabolites. Thus, carbohydrate accumulation has been, histori- to the same response.
cally, one of the most commonly used indicators for virus- At 0 h, high levels of the primary metabolites (such as sugars)
induced alterations in plant metabolism during both compatible and the tricarboxylic acid (TCA) cycle metabolites (particularly
and incompatible interactions. An abnormal accumulation or malate) were observed (Fig. 4; Fig. 3S). At 48 h, in the post-
disappearance of carbohydrates (especially sucrose), such as the infected leaves, a definitely different metabolome was observed
one observed in our studies (see Fig. 4), is diagnostic for (Fig. 4; Fig. 3S). The NMR spectra showed that alanine, glutamine
net alterations in the balance between those processes respon- and valine levels increased in infected tissue. On the contrary,
sible for creation and utilization of carbohydrate namely, photo- the infected leaves showed lower levels of sucrose, glucose,
synthesis and respiration. Assuming that a certain level of amino nicotine and malate when compared with control leaves (Fig. 4;
acids is required for stress response, it can be suggested that since Figs. 3S and 4S). The dicarboxylic acid malate has a multitude of
PI plants had higher levels of amino acids prior to stress treat- functions in plant metabolism and homeostasis. These range from
ment, they did not require an increase in amino acid levels, its most prominent roles in the mitochondrial TCA cycle, in
whereas PC plants did. Stress often results in substantial changes crassulacean acid metabolism (CAM) and C4 metabolism to its
in metabolite profiles. Elevated levels of sugar metabolites and roles as an osmoticum, as a regulator of pH homeostasis, as a
amino acids were found in plants exposed to a plethora of stress reducing equivalent that is shuttled between subcellular com-
conditions, such as temperature (Cook et al., 2004; Kaplan et al., partments (Martinoia et al., 2000). As an intermediate of the TCA
2007), water and salinity (Brosche et al., 2005; Cramer et al., cycle, malate is intimately associated with mitochondrial energy
2007; Kim et al., 2007), sulfur (Nikiforova et al., 2005), phos- metabolism and is also the origin of carbon skeletons exported
phorus (Hernandez et al., 2007), oxidative (Baxter et al., 2007) and from the mitochondrion in support of amino acid biosynthesis
heavy metals (Le Lay et al., 2006) as well as a combination of (Schneidereit et al., 2006).
multiple stresses (Rizhsky et al., 2004). The present results The differences in the metabolite levels observed between
indicating elevated levels of sugar metabolites in the progeny of non-exposed PI and PC plants almost disappeared if these plants
infected plants at 0 h compared to non-infected control plants were stressed. This finding is somewhat puzzling. We hypothe-
also agree with the previous findings. sized that transgenerational responses to viral infection include
Our metabolic profiling experiments showed that mock-treat- changes in levels of gene expression, resulting in higher endo-
ment (the 48 h sample) and TMV infection (the 48 h_V sample) of genous levels of metabolites. This response may be part of
PI or PC plants reduced the levels of glucose and fructose. the plant adaptation strategy; plants accumulate higher levels
Curiously, Sanchez et al. (2008) also observed a dramatic dose- of metabolites to be prepared for a possible encounter with a
dependent depletion of the level of fructose and glucose but pathogen. The higher endogenous levels of metabolites found in
not sucrose upon exposure of Arabidopsis to salt stress (Sanchez PI plants may be sufficient to counter stress, whereas in PC plants,
et al., 2008). A comparison of steady-state metabolite levels these metabolites need to accumulate further.
in Arabidopsis and a related extremophile shrub known as We found noticeable differences in the level of metabolites in
T. halophila showed higher amounts of sugars (e.g., sucrose, PI#9 plants as compared to PI#19 lines. In fact, in most of the
fructose and glucose), along with various organic and amino acids cases the differences between PI plants and PC plants was only
(e.g., proline and citric, malic and succinic acids) in the shrub even significant for PI#9 plants rather than PI#19 plants. We can
prior to stress exposure (Gong et al., 2005). It has been known for hypothesize that the reason for such differences could have been
many decades that changes in the accumulation of carbohydrates in the process of infection and grafting in the parental lines giving
frequently precede the appearance of virus symptoms (Handford rise to PI#9 and PI#19 plants (Kovalchuk et al., 2003). It is
and Carr, 2007). Earlier studies (Handford and Carr, 2007) have possible that the infection process in the original parental plants
also demonstrated that early in the infection process, even before infected with TMV was more effective in plant #9 as compared to
symptoms are discernible, virus infection alters both the starch plant #19, thus leading to more efficient signal passed to the
production during the day and its degradation and/or mobiliza- progeny (Kovalchuk et al., 2003). Our previous research showed
tion at night. Many studies carried out on a variety of host–virus that both PI#9 and PI#19 plants had hypermethylated genomes
systems have shown that partitioning of newly fixed carbon (Boyko et al., 2007). Analysis of methylation at resistance gene
122 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123

loci showed significant hypomethylation in PI#9 plants, whereas Cramer, G.R., Ergul, A., Grimplet, J., Tillett, R.L., Tattersall, E.A., Bohlman, M.C.,
PI#19 plants were not tested (Boyko et al., 2007). PI#9 plants also Vincent, D., Sonderegger, J., Evans, J., Osborne, C., Quilici, D., Schlauch, K.A.,
Schooley, D.A., Cushman, J.C., 2007. Water and salinity stress in grapevines:
showed higher tolerance to fungal pathogen as compared to PC early and late changes in transcript and metabolite profiles. Funct. Integrative
plants, whereas PI#9 plants were not tested. PI#9 plants showed Genomics 7, 111–134.
more profound resistance to TMV infection as compared to PI#19 De Vos, M., Van Oosten, V.R., Van Poecke, R.M., Van Pelt, J.A., Pozo, M.J., Mueller,
M.J., Buchala, A.J., Metraux, J.P., Van Loon, L.C., Dicke, M., Pieterse, C.M., 2005.
plants (Kathiria et al., 2010). Thus, it is highly likely that PI#9 Signal signature and transcriptome changes of Arabidopsis during pathogen
plants received a stronger transgenerational signal. and insect attack. Mol. Plant Microbe Interact. 18, 923–937.
In summary, the results obtained may indicate that TMV Grant, M., Lamb, C., 2006. Systemic immunity. Curr. Opinion Plant Biol. 9,
414–420.
infection leads to redirection of metabolic activity from a growth Grant, J.J., Yun, B.W., Loake, G.J., 2000. Oxidative burst and cognate redox
oriented phase to a stress response oriented phase. The elevated signalling reported by luciferase imaging: identification of a signal network
levels of metabolites in the progeny of infected plants at 0 h that functions independently of ethylene, SA and Me-JA but is dependent on
MAPKK activity. Plant J. 24, 569–582.
compared to control plants at 0 h may represent a role in stress
Gong, Q., Li, P., Ma, S., Indu Rupassara, S., Bohnert, H.J., 2005. Salinity stress
tolerance which warrants further investigation. The present adaptation competence in the extremophile Thellungiella halophila in compar-
results extend our previous findings that the progeny of infected ison with its relative Arabidopsis thaliana. Plant J. 44, 826–839.
plants exhibit changes in plant genome stability, methylation Hernandez, G., Ramirez, M., Valdes-Lopez, O., Tesfaye, M., Graham, M.A.,
Czechowski, T., Schlereth, A., Wandrey, M., Erban, A., Cheung, F., Wu, H.C.,
patterns, stress tolerance, the amount of phenolic compounds as Lara, M., Town, C.D., Kopka, J., Udvardi, M.K., Vance, C.P., 2007. Phosphorus
well as the concentration of various metabolites (Kathiria et al., stress in common bean: root transcript and metabolic responses. Plant Physiol.
2010). It is difficult to speculate why the progeny of stressed 144, 752–767.
Handford, M.G., Carr, J.P., 2007. A defect in carbohydrate metabolism ameliorates
plants has higher levels of metabolites. In the past, we showed symptom severity in virus-infected Arabidopsis thaliana. J. Gen. Virol. 88,
that the progeny of stressed plants, either treated with abiotic 337–341.
stress or infected with pathogen, underwent changes in methyla- Handford, M.G., Sicilia, F., Brandizzi, F., Chung, J.H., Dupree, P., 2004. Arabidopsis
thaliana expresses multiple Golgi-localised nucleotide-sugar transporters
tion patterns and gene expression. Therefore, we suggest that the related to GONST1. Mol. Genet. Genomics 272, 397–410.
stress-related signal, possibly small RNA, may establish a differ- Jones, J.D., Dangl, J.L., 2006. The plant immune system. Nature 444, 323–329.
ential pattern of methylation in the gametes of stressed plants, Kovalchuk, I., Kovalchuk, O., Kalck, V., Boyko, V., Filkowski, J., Heinlein, M.,
Hohn, B., 2003. Pathogen-induced systemic plant signal triggers DNA rearran-
allowing for differential patterns of gene expression and thus gements. Nature 423, 760–762.
different levels of various metabolites. It is not clear how specific Kathiria, P., Sidler, C., Golubov, A., Kalischuk, M., Kawchuk, L.M., Kovalchuk, I.,
these changes are, and in the future, it would be important to 2010. Tobacco mosaic virus infection results in an increase in recombination
frequency and resistance to viral, bacterial, and fungal pathogens in the
compare metabolite levels in seeds of plants exposed to various
progeny of infected tobacco plants. Plant Physiol. 153, 1859–1870.
abiotic and biotic stresses. Nevertheless, using both NMR and GC– Kim, S., Koo, I., Fang, A., Zhang, X., 2011. Smith–Waterman peak alignment for
MS, the current approach of metabolic profiling can provide a comprehensive two-dimensional gas chromatography–mass spectrometry.
detailed understanding of metabolic fluxes involved in the com- BMC Bioinformatics 12, 235.
Kopka, J., 2006. Current challenges and developments in GC–MS based metabolite
plex mechanisms regulating metabolic pathways in plants. profiling technology. J. Biotechnol. 124, 312–322.
Kaplan, F., Kopka, J., Sung, D.Y., Zhao, W., Popp, M., Porat, R., Guy, C.L., 2007.
Transcript and metabolite profiling during cold acclimation of Arabidopsis
reveals an intricate relationship of cold-regulated gene expression with
Acknowledgments modifications in metabolite content. Plant J. 50, 967–981.
Kim, J.K., Bamba, T., Harada, K., Fukusaki, E., Kobayashi, A., 2007. Time-course
DSW would like to acknowledge financial support from the metabolic profiling in Arabidopsis thaliana cell cultures after salt stress
treatment. J. Exp. Bot. 58, 415–424.
Genome Alberta and Alberta Advanced Education and Training. Loebenstein, G., 2009. Local lesions and induced resistance. Adv. Virus Res. 75,
IK would like to acknowledge financial support from the Alberta 73–117.
Agriculture Research Institute, NSERC Strategic and the Human Love, A.J., Yun, B.W., Laval, V., Loake, G.J., Milner, J.J., 2005. Cauliflower mosaic
virus, a compatible pathogen of Arabidopsis, engages three distinct defense-
Frontiers Science Program. We thank Valentina Titova for proof- signaling pathways and activates rapid systemic generation of reactive oxygen
reading the manuscript. species. Plant Physiol. 139, 935–948.
Leiss, K.A., Choi, Y.H., Verpoorte, R., Klinkhamer, P.G., 2011. An overview of
NMR-based metabolomics to identify secondary plant compounds involved
in host plant resistance. Phytochem. Rev. 10, 205–216.
Appendix A. Supporting information Le Lay, P., Isaure, M.P., Sarry, J.E., Kuhn, L., Fayard, B., Le Bail, J.L., Bastien, O.,
Garin, J., Roby, C., Bourguignon, J., 2006. Metabolomic, proteomic and biophy-
sical analyses of Arabidopsis thaliana cells exposed to a caesium stress.
Supplementary data associated with this article can be found
Influence of potassium supply, Biochimie 88, 1533–1547.
in the online version at doi:10.1016/j.bcab.2012.01.004. Martinoia, E., Massonneau, A., Frangne, N., 2000. Transport processes of solutes
across the vacuolar membrane of higher plants. Plant Cell Physiol 41,
1175–1186.
Nikiforova, V.J., Kopka, J., Tolstikov, V., Fiehn, O., Hopkins, L., Hawkesford, M.J.,
Hesse, H., Hoefgen, R., 2005. Systems rebalancing of metabolism in response to
References sulfur deprivation, as revealed by metabolome analysis of Arabidopsis plants.
Plant Physiol 138, 304–318.
Boyko, A., Kovalchuk, I., 2011. Genetic and epigenetic effects of plant–pathogen Ouko, M.O., Sambade, A., Brandner, K., Niehl, A., Pena, E., Ahad, A., Heinlein, M.,
interactions: an evolutionary perspective. Mol. Plant. Nick, P., 2010. Tobacco mutants with reduced microtubule dynamics are less
Boyko, A., Kathiria, P., Zemp, F.J., Yao, Y., Pogribny, I., Kovalchuk, I., 2007. susceptible to TMV. Plant J 62, 829–839.
Transgenerational changes in the genome stability and methylation in patho- Rizhsky, L., Liang, H., Shuman, J., Shulaev, V., Davletova, S., Mittler, R., 2004. When
gen-infected plants: (virus-induced plant genome instability). Nucleic Acids defense pathways collide. The response of Arabidopsis to a combination of
Res. 35, 1714–1725. drought and heat stress. Plant Physiol. 134, 1683–1696.
Brosche, M., Vinocur, B., Alatalo, E.R., Lamminmaki, A., Teichmann, T., Ottow, E.A., Saude, E.J., Lacy, P., Musat-Marcu, S., Mayes, D.C., Bagu, J., Man, S.F., Sykes, B.D.,
Djilianov, D., Afif, D., Bogeat-Triboulot, M.B., Altman, A., Polle, A., Dreyer, E., Moqbel, R., 2004. NMR analysis of neutrophil activation in sputum samples
Rudd, S., Paulin, L., Auvinen, P., Kangasjarvi, J., 2005. Gene expression and from patients with cystic fibrosis. Magn. Reson. Med. 52, 807–814.
metabolite profiling of Populus euphratica growing in the Negev desert. Saude, E.J., Obiefuna, I.P., Somorjai, R.L., Ajamian, F., Skappak, C., Ahmad, T.,
Genome Biol. 6, R101. Dolenko, B.K., Sykes, B.D., Moqbel, R., Adamko, D.J., 2009. Metabolomic
Baxter, C.J., Redestig, H., Schauer, N., Repsilber, D., Patil, K.R., Nielsen, J., Selbig, J., biomarkers in a model of asthma exacerbation: urine nuclear magnetic
Liu, J., Fernie, A.R., Sweetlove, L.J., 2007. The metabolic response of hetero- resonance. Am. J. Respir. Crit. Care Med. 179, 25–34.
trophic Arabidopsis cells to oxidative stress. Plant Physiol. 143, 312–325. Schauer, N., Steinhauser, D., Strelkov, S., Schomburg, D., Allison, G., Moritz, T.,
Cook, D., Fowler, S., Fiehn, O., Thomashow, M.F., 2004. A prominent role for the CBF Lundgren, K., Roessner-Tunali, U., Forbes, M.G., Willmitzer, L., Fernie, A.R.,
cold response pathway in configuring the low-temperature metabolome of Kopka, J., 2005. GC–MS libraries for the rapid identification of metabolites in
Arabidopsis. Proc. Natl. Acad. Sci. USA 101, 15243–15248. complex biological samples. FEBS Lett. 579, 1332–1337.
 R. Mandal et al. / Biocatalysis and Agricultural Biotechnology 1 (2012) 115–123 123

Sanchez, D.H., Siahpoosh, M.R., Roessner, U., Udvardi, M., Kopka, J., 2008. Plant Wishart, D.S., Knox, C., Guo, A.C., Eisner, R., Young, N., Gautam, B., Hau, D.D.,
metabolomics reveals conserved and divergent metabolic responses to sali- Psychogios, N., Dong, E., Bouatra, S., Mandal, R., Sinelnikov, I., Xia, J., Jia, L.,
nity. Physiol. Plant 132, 209–219. Cruz, J.A., Lim, E., Sobsey, C.A., Shrivastava, S., Huang, P., Liu, P., Fang, L.,
Schneidereit, J., Hausler, R.E., Fiene, G., Kaiser, W.M., Weber, A.P., 2006. Antisense Peng, J., Fradette, R., Cheng, D., Tzur, D., Clements, M., Lewis, A., De Souza, A.,
repression reveals a crucial role of the plastidic 2-oxoglutarate/malate trans- Zuniga, A., Dawe, M., Xiong, Y., Clive, D., Greiner, R., Nazyrova, A., Shaykhutdi-
locator DiT1 at the interface between carbon and nitrogen metabolism. Plant J. nov, R., Li, L., Vogel, H.J., Forsythe, I., 2009. HMDB: a knowledgebase for the
45, 206–224. human metabolome. Nucleic Acids Res. 37, D603–610.
Tao, Y., Xie, Z., Chen, W., Glazebrook, J., Chang, H.S., Han, B., Zhu, T., Zou, G., Wishart, D.S., Tzur, D., Knox, C., Eisner, R., Guo, A.C., Young, N., Cheng, D., Jewell, K.,
Katagiri, F., 2003. Quantitative nature of Arabidopsis responses during compa- Arndt, D., Sawhney, S., Fung, C., Nikolai, L., Lewis, M., Coutouly, M.A., Forsythe, I.,
tible and incompatible interactions with the bacterial pathogen Pseudomonas Tang, P., Shrivastava, S., Jeroncic, K., Stothard, P., Amegbey, G., Block, D.,
syringae. Plant Cell 15, 317–330. Hau, D.D., Wagner, J., Miniaci, J., Clements, M., Gebremedhin, M., Guo, N.,
Whitham, S., Dinesh-Kumar, S.P., Choi, D., Hehl, R., Corr, C., Baker, B., 1994. The Zhang, Y., Duggan, G.E., Macinnis, G.D., Weljie, A.M., Dowlatabadi, R., Bamforth, F.,
product of the tobacco mosaic virus resistance gene N: similarity to toll and Clive, D., Greiner, R., Li, L., Marrie, T., Sykes, B.D., Vogel, H.J., Querengesser, L.,
the interleukin-1 receptor. Cell 78, 1101–1115. 2007. HMDB: the human metabolome database. Nucleic Acids Res. 35,
Weljie, A.M., Newton, J., Mercier, P., Carlson, E., Slupsky, C.M., 2006. Targeted D521–526.
profiling: quantitative analysis of 1H NMR metabolomics data. Anal. Chem. 78, Wishart, D.S., 2008. Applications of metabolomics in drug discovery and develop-
4430–4442. ment. Drugs R D 9, 307–322.