Metabolite Fingerprinting in Transgenic Nicotiana tabacum Altered by theEscherichia coli Glutamate Dehydrogenase Gene

Production of homozygous lines for biochemical evaluations

We had previously generated r2 seed from a series ofindependently regenerated plants that showed a range of GDHactivity of 2–25 μmol min⁻¹mg⁻¹ protein[12]. Each line was an independent transformant, withgenetic architecture consistent with one or two copies of thegdhA transgene [15]. The mRNA abundance and GDHactivity were correlated via Northern hybridization. The mRNAwas of high abundance for the GDH10 that produced between 20and 23 μmol min⁻¹mg⁻¹ protein GDH activity.GDH10 line was selected for further analyses compared tovector and nontransgenic controls.

Analysis of glutamate fraction labeling

For comparison of glutamate fraction labeling we selectedGDH10, GUS, and BAR transgenic tobacco lines because only GDHis expected to be resistant to methionine sulfoximine (MSX),an inhibitor of photorespiratory ammonium assimilation.Comparisons (Tables 1, 2, 3, and4) did show organ specific differences.

In the roots, labeling of the fraction containing¹³N-glutamate (from 13NH4+ administered during a 15-minute period) was increased 2.2 fold in GDH10compared to BAR and GUS plants as a result of the introduced GDHactivity, representing 21% (dpm/dpm) of the ¹³NH₃ applied(Table 1). Treatment with MSX, an inhibitor ofglutamine synthetase, reduced glutamate fraction labeling 7 foldamong the GDH, GUS, and BAR transgenics suggesting the GS/GOGATcycle accounts for 86% of the labeling in the absence of MSX.However, in MSX-inhibited GDH10 roots, glutamate labelingremained 2.2 fold higher than GUS and BAR roots(Table 2). Therefore, GDH was not inhibited by MSX.As expected BAR did not inactivate MSX.

In leaves, both glutamate fraction labeling and totallabeling were decreased by 1.2 to 1.5 fold in GDH10 comparedto GUS and BAR control plants (Table 3). Thedecrease was not significant in this experiment or experimentswith leaf discs (data not shown). However, glutamate fractionlabeling in presence of MSX was decreased 10 fold in GUS andBAR plants but only 2.6 fold in the GDH10(Table 4). In addition, glutamine fractionlabeling in presence of MSX was decreased 10 fold in GUS andBAR plants but only 0.6 fold in the GDH10. Therefore,in MSX-inhibited leaves; GDH10 assimilated 5 fold more¹³N than GUS control and BAR plants. The GDH10line, in the presence of MSX, also left less 13NH4+ unincorporated (84% compared to 95%, Table 4)reflecting the contribution of the gdhA gene in NH4+ assimilation in MSX-inhibited leaves. Theglutamine labeling in MSX-treated GDH10 leaves was not relatedto incomplete inhibition of GS since the same degree oflabeling was observed in 1 cm³ leaf discs floating inlabeling solution (data not shown).

In GDH10 there was 2–3 fold more label in theglutamate fraction of both MSX-treated leaves (7.0% of theabsorbed 13NH4+, Table 4) and roots(3.2%, Table 2) than BAR leaves (2.9%,Table 4) and roots (1.3%,Table 2).However, in non-MSX-treated GDH10 transgenic leaves, comparedto the roots, the very high activity of GS, the larger poolsizes of glutamate and the greater flux through pathwaysinvolving glutamate may have resulted in less labeling by¹³N (Tables 1 and3). The amountof label in the glutamate and glutamine fractions that couldbe attributed to GDH activity was modest in roots, about2.3%. ((3.2–1.3) + (1.9–1.5)).However, in leaves, labeling was significantly greater, about11.9% ((7.0–2.9) + (8.5–1.7)).

Analysis of ion fingerprints and profiles

Experiments with tobacco [13,14, 15]and corn [17, 18]had indicated that the total soluble amino acid, ammonium, andcarbohydrate contents of GDH transgenic plants were eachincreased. The transgenic seedlings were shown to reproducethis phenotype (Table 5(a)).Ions were separated and characterized to infer the detectable metabolite complementusing four ionization protocols for FT-ICR-MS. There were2 012 ions detected within 2–4 ionization protocols (uniqueions and isotope ions were removed). Regardless of genotype,ion fingerprints of leaves and roots differed significantlyjudged by FT-ICR-MS. Qualitative differences (compounds onlydetected in one organ) approached 23% (462/2012). Ion masseswere validated by internal calibration with compounds of knownmass and concentrations. Among the ions common in roots andleaves, apparent quantitative differences were in the majority60% (929/1550). Quantitative differences were inferred frompeak areas and validated by internal calibration. However,many factors can interfere with peak detection so that theestimates of differences in quantity are not unequivocal andsome may be erroneous. Within that context some of the dataobserved were consistent with known organ-specific metabolismsin plants and some were not.

The metabolites we putatively inferred from ion masses thatwere altered in abundance by GDH activity are depicted inFigures 1, 2,3, 4,and listed in Table 5. The majority of the metabolites increasedor decreased in leaves and/or roots by less than 10 fold.Between 5% and 14% of detectable metabolites were altered inabundance. This portion of the database can beexamined at http://www.siu.edu/~pbgc/metabolite-profiles/GDH/.

In leaves, 98 (5%) of the ions detected were changed inabundance between GDH and non-GDH plants. Only 91 empirical formulas could be inferred because sevenwere equivocal. Forty-one matched the formulas and predictedmasses of the ions of compounds found in the databases wesearched. The masses of the remaining fifty unidentifiedmetabolite ions are available athttp://www.siu.edu/~pbgc/metabolite-profiles/GDHbut not discussed further here for brevity. The 41 putativelyidentified compounds were categorized as follows:11 amino acids, 2 sugars, 8 fatty acids, 6 compounds ofspecial nitrogen metabolism, 2 nucleic acid derivatives, 1 TCAcycle intermediate, 1 stress-related compound, and 10miscellaneous metabolites not of those classes(Figure 4, Table 5). Not all of thesecompounds are common metabolites. Some are compounds notpreviously detected in plant cells, possibly reflecting theanimal and microbial fauna present on tobacco samples. Someidentified compounds were not previously detected in vivopossibly reflecting ionization artifacts. However, forbrevity hereafter the metabolite putatively inferred from adetected ion will be referred to as just themetabolite.

In roots, there were 283 ions (14%) that changed in abundanceamong the 2 012 ion species repeatedly detected. Only 268empirical formulas could be inferred. Database searchesputatively identified only 117 of the 283 changed metabolites(Figure 3). Masses of the unidentified ions are available athttp://www.siu.edu/∼pbgc/metabolite-profiles/GDHbut not reported further here for brevity. Among the 117altered metabolites, there were 14 amino acids, 6 sugars, 34fatty acids, 15 compounds of special nitrogen metabolism, 2nucleic acid derivatives, 4 TCA cycle intermediates, 2stress-related compounds, and 40 metabolites not of thoseclasses (Figure 4, Table 5). Judged bythe correspondence of ion mass estimates, 90% of thecompounds that changed in abundance in leaves also increasedor decreased in the same way in roots. Only three metaboliteswere altered so that the increase in one organ was accompaniedby decrease in the other organ (63 and 75, 86 and 87, and 60and 67, Table 5).

Amino acids, precursors, and derivatives

In leaf extracts, consistent with previous reports ofincreased free amino acids [12], we found 6 amino acidsthat increased in abundance (1.3 to 4.5 fold,Figure 4a) in GDH plants. Arginine, phenylalanine,tryptophan, asparagine, glutamine, and histidine were inferredto be altered in abundance. Most of the known pathwayintermediates involved in the biosynthesis of protein aminoacids were detected, but were not altered in abundance. Fouramino acid derivatives changed in abundance(Table 5(a)), one decreased, and three increased. Thenonprotein amino acid ornithine increased 2.3 fold.

In roots, 9 amino acids appeared to be increased in abundancein GDH plants by 2 to 11 fold (Table 5(b)).Arginine, phenylalanine, tryptophan, asparagine, glutamine,histidine, proline, threonine, and valine were inferred to bealtered in abundance. The root increases in proline,threonine, and valine were not detected in leaves. Many of theknown pathway intermediates involved in the biosynthesis ofprotein amino acids were detected but not altered inabundance, except for the proline precursordelta-pyrroline-5-carboxylate (91, Table 5). Noamino acid was decreased in abundance in GDH plants but threeof the five amino acid derivatives that were changed decreased(Table 5(b)).

Sugars and derivatives

In leaves, consistent with a previous report of a modestincrease in free carbohydrates [12], two sugarderivatives appeared to be increased 1.5–2 fold(10,11, Table 5) in GDH plants. In roots(Table 5(d)) six sugars appeared to be increased by2.3–12.5 fold between GDH and non-GDH plants and included keyintermediates involved in the regeneration ofribulose-5-phosphate in the Benson-Calvin cycle. One sugarderivative increased in both leaves and roots (10,15, Table 5).

Fatty acids

In leaves the six fatty acids and two derivatives thatappeared to be changed in abundance all decreased in the GDHplants (18–25, Table 5). Two 16-carbon fattyacids (19,20, 16:0 and 16:1, Table 5) and two18-carbon plant membrane fatty acids were reduced (21,22,18:1, 18:2, Table 5). These changed fatty acidsare minor components of both plant cell membranes andchloroplast membranes. However, α-linolenic acid(18:3), the main constituent of both membranes was not alteredin abundance. Two rare unsaturated fatty acids (18 and 23;15:0 and 24:0) were significantly reduced(Table 5(e)). Two fatty acid derivatives alsodecreased (Table 5(f)).

In roots, seventeen of the eighteen fatty acids and twelve ofthe sixteen fatty acid derivatives that appeared to be changedabundance decreased in GDH plants (26–40, 42,43,Table 5). The decreased fatty acids included 5 ofthose that decreased in leaves (33--35,38,43,Table 5) but the 18:2 was not decreased. The decreasedfatty acid derivatives included both those that decreased inleaves. The four fatty acid derivatives that increased included3 di-enoates or tri-enoates of 18-C fatty acids that may bebiosynthetically related (45,48,51,55, Table 5). Noneof the common diacylglycerol lipids were detected by FT-ICR-MS sowhether decreases in fatty acids were reflected by decreases in lipids is not known.

Special nitrogen metabolism

Six metabolites that appeared to be increased inabundance between GDH and non-GDH plants in leafextracts were amines (1), alkaloids (2), and phenolics(3), three classes of products derived from specialnitrogen metabolism (Table 5(i)).

From four classes of special nitrogen metabolites sixteenappeared to be altered in abundance between GDH and non-GDHplants in roots. There were amines (2), alkaloids (7),phenolics (5), and isoprenoids (1) (Table 5(j)). Fiveincreased and nine decreased. Only N-caffeoylputrescine wasaltered in both organs. However, it increased in leaves butdecreased in roots.

Nucleic acids and derivatives

Only two derivatives of nucleic acids appeared to be increased2–3 fold in leaf extracts between GDH and non-GDHplants (Table 5(k)). Two compounds were increasedin roots more than 2 fold, including the common ribonucleotideuridine (Table 5(l)).

TCA cycle intermediates and derivatives

The monoethyl ester of fumaric acid was the solemetabolite identified that appeared to be changed byGDH in leaves (Table 5(m)). In roots, all fourmetabolites that changed increased in abundance(2.7–4 fold) including three TCA cycle intermediates,fumaric, malic, and citric acids (Table 5(n)).

Stress-related compounds

Only one member of this group appeared to be altered in leafextracts (Table 5(o)); two increased more than 2fold in roots (Table 5(p)).

Miscellaneous

Ten metabolites appeared to be altered in leaves and eightcontained nitrogen. Five of the compounds identified in leafextracts represent known drugs and cigarette toxins(Table 5(q)).

Forty metabolites appeared to be altered in roots andtwenty-two contained nitrogen. Among these are five drugs,five flavoring agents, four pesticides, three carcinogens, andfive toxins. There were two compounds that were alsocoordinately altered in leaves: N-nitrosopyrollidone andmenthyl acetoacetate.

Metabolite analysis

This study used metabolite analysis with FT-ICR-MS[5, 8] toassociate phenotype with biochemical changes resulting fromendogenous effects of ectopic glutamate synthesis intransgenic plants. The GDH plants were a suitable test forFT-ICR-MS because they have cell composition alterations thatresult from a specific biochemical alteration in awell-characterized pathway targeting the cellular glutamatepools [12].

The identification of ions andthe inference of a metabolite were relativelyinefficient, with less than half the ions detectedhaving known metabolites of corresponding masses. Therest of the ions may represent reactions occurringbefore sample quenching, multiple ionization effects,ion suppression effects, or ion fragmentation [9].Some of these ions may represent new metabolites notpreviously reported in plants. An estimate of theextent of artifact ions compared to new products will be a future goal.

Although not reliable or fully quantitative, the changedabundances inferred from ions detected by FT-ICR-MS thatappear to correspond to metabolites such as amino acids,sugars, and fatty acids largely agreed with quantitativespectrophotometer assays [12,17, 18,19] HPLCseparation of sixteen individual amino acids from the methanolsoluble, low molecular weight fraction of cell extracts showedeight were significantly changed. Four of the eight aminoacids had been inferred to increase in abundance by FT-ICR-MS;the remainder had not been detected as quality peaks. Threeamino acids, histidine, valine, and threonine, were notincreased as expected from FT-ICR-MS. The difference appearsto be due to interference by other ions [9].

Given the only partial agreement among threedifferent measurements of the amino acids, we concludethat the abundance of specific ions in total infusionmass spectra were significantly affected by combinedion suppression effects of all other components, pHand salinity of the solution, flow rate, tip opening,and electrospray current [9]. Further the samplesmay have differed in the rapidity of turnover ofintracellular metabolites, the rate at whichmetabolism was quenched and the time for whichmetabolites were separated from the cell debris [10].The evidence of reproducibility for some amino acidmeasurements may be related to handling samplessimultaneously [9, 10]. Samples analyzed separately either temporally or spatially will be more difficult to compare.

However, the exact masses alone are not sufficient to identifyspecific compounds unequivocally. Several compounds wereidentified that are not metabolites in plants (eg,alpha-tert.butoxycarbonyl-L-tryptophancompound no 4, Table 5(a), a syntheticintermediate in peptide synthesis); some artificialpesticide-like metabolites (119,126,127,133,138,Table 5); and metabolites found ininsects not plants (eg, no 114, a component of bee royaljelly). Therefore, data from FT-ICR-MS analysis should be usedas preliminary evidence to suggest further experiments[8]. In this publication we focused on amino acidmetabolism and the effect on central metabolism.

Among the effects detected, those altering amino acidmetabolism and fatty acid metabolism were most profound andappear to underlie a doubling of free amino acids and halvingof free fatty acid content [12,13, 18]. In comparison theeffects of GDH on carbohydrate metabolism were comparativelytrivial and may not solely underlie the increased contentreported [12]. The increases in three abundant organicacids may have contributed to the carbohydrate contentreported by spectrophotometer assays of reducing sugarcontent. In addition some of the unidentified ions may havebeen sugars or carbohydrates.

The majority of ions detected by FT-ICR-MS could not beidentified from their predicted formulas or mass. Incomparison about 30% of ions identified in plants by GC-MScould not be identified [4, 7,11]. The unidentified ionsdetected may represent novel constituents of tobacco leaves orroots [38] or ionization artifacts of MS[5, 6]. Different abundances couldalso be experimental artifacts. To reduce artifacts we usedpooled samples for each genotype from plants grown in an RCB ina growth chamber; accepted only those ions derived from twoionization methods of the four applied; and by repeating the entire experiment.

Masses of the unidentified metabolite ions are availableonline at http://www.siu.edu/∼pbgc/metabolite-profiles/GDH/Ntabacum/IONS1-4. htmlbut not discussed further here for brevity.

The concurrence between FT-ICR-MS and spectrophotometer data[12] appears to validate the use of the method formetabolite analyses. However, the informational content ofFT-ICR-MS is orders of magnitude greater than otherhigh-throughput methods [1]. FT-ICR-MS detected 2 012ions that were consistent across ionization methods from eachreplicated extract. In comparison, tandem MS required severalindependent extractions to identify 326 metabolites [4]or 88 metabolites [11]. However, there is no doubt thatGC-MS in a tandem format is a superior technique forunequivocal identification of ions and therefore metabolites[1]. In addition GC-MS is superior in that ionconcentrations can be derived. Both methods suffer from tuningartifacts among spatially separated runs that can only bepartly compensated for by internal standards. We conclude thatFT-ICR-MS will have a role in functional genomics where samplethroughput is more important than chemical identification andrelative quantifications, a situation analogous to thedecision to employ microarray or macroarray for analysis of the transcriptome.

Amino acid metabolism

Ammonium assimilation fluxes in roots showed that the introducedGDH contributed to total labeling of glutamate regardless ofGS inhibition, suggesting that the enzymes compete forNH4+. In leaves, the GDH reduced net ¹³NH₃ assimilation, possibly by suppression of GS activity [20,21]. This is consistent with the increase in leafNH4+ reported [12]. However, in the presence of MSX, GDH partially substituted GS by increasing glutamatelabeling (7% label incorporated compared to 2.9% forcontrols). It is possible that the higher Km of GDHfor NH4+ [22] and the 7–10 fold greater fluxes in nitrogen (resulting from photorespiration)[23] drive the NAD(P)H-dependent GDH reaction forward toproduce glutamate in large quantities during GS inhibition[20]. From labeling we conclude that the modificationsin transgenic plants are not the product of greatly increasedefficiency of nitrogen assimilation by GDH plants. Instead,the glutamate generated in the cytoplasm may result in alteredmetabolic fluxes and profiles.

Metabolite analysis apparently contradicts ¹³N fluxlabeling because the steady-state of extractable glutamate is notaltered between GDH and control transgenic roots or leaves.Short-term flux does not always predict steady-stateconcentration because plants have mechanisms for sensing nitrogenfluxes and maintaining homeostasis [24,25]. Flux away fromglutamate appears to equal the extra flux into glutamate as manymajor nitrogen sinks were increased and few decreased (in leaves19 increased and 4 decreased, in roots 29 increased and 17decreased). Since plant mRNA abundances did not change (data notshown), allosteric effectors of many enzymes may be involved[26]. The effects of GDH expression on phenotype may resultfrom the signaling effects of increased cytosolic glutamate seenin plants grown at low light intensities, nia andrbc mutants [27] and the status of certaininorganic N compounds [28, 29, 30].Metabolites derived from nitrate, such as ammonium, glutamine,and glutamate all may act as signals to report on organicN status [25]. Cytosolic glutamate may act directly as aligand to activate ion channels.

Metabolites that shared C skeletons were coordinately alteredin abundance in both roots and leaves in response to GDHactivity. Among the amino acids, 4 that derived fromalpha-ketoglutarate were coordinately changed in roots and 3in leaves. Ornithine, the nonprotein amino acid derived fromalpha-ketoglutarate, was also increased in roots. Alsochanged to the same extent in both organs were phosphoenolpyruvate derivatives, phenylalanine, and tryptophan. However,tyrosine, the only other amino acid originating fromphosphoenolpyruvate C skeletons, was not altered in abundance.Asparagine was the only amino acid, derived from oxaloacetate,changed in both leaves and roots but threonine increased4–5 fold in roots. Among amino acids derived frompyruvate C skeletons, valine was increased 4–5fold in roots but no changes in leucine were seen and alaninewas used as an internal standard and therefore changes couldnot be detected. Similarly, the amino acids derived from6-phosphoglycerate could not be detected. The pattern ofamino acid changes is similar to that in maize endosperm withthe opaque mutation [29] where endogenous GDH activity isincreased, but different from that caused by photosynthesis[31, 32]. Therefore, the metabolic alterations are GDH specific, not systemic, implying that the effect of GDH onmetabolism depends on the metabolism occurring in the cell.

Changes related to water deficit tolerance

Increases in sugar concentrations could also significantlyincrease the water deficit tolerance [33]. However, thesugars increased by GDH were complex sugars, not the monosaccharides ordisaccharides normally associated with tolerance. The notionof sugar sensing is also gaining momentum [25]. TheFT-ICR-MS assays would not detect polysaccharides over700 d, so again flux and steady state may differ.

None of the following compatible solutes were changed inabundance in either leaves or roots [33]: trigonelline,trehalose, dimethylsulfoniopropionate, glycerol, sorbitol,mannitol, choline-O-sulphate, beta alanine betaine,glycinebetaine, prolinebetaine, N-methyl-proline,hydroxyproline, hydroxyprolinebetaine, and pipecolic acid.However, since the association of water deficit tolerance withany single solute is imperfect, we expect the phenotype wasderived from a combination of increased compatible solutes.One or a few of the unidentified metabolites may alsoparticipate. Stomatal behavior, GS activity and resistance tophotooxidation may contribute to the tolerant phenotype[34]. Plant morphology does not appear to contribute asthe root to shoot ratios were not changed [13,14]. Themechanism by which water deficit tolerance is afforded remains to be unraveled.

Fatty acids

Oil and protein contents are inversely related toeach another, to carbohydrate content and to yieldin many crop plants. The synthetic pathways for fatty acid,protein, and carbohydrate compete for carbon skeletons[35]. Therefore, the increase in protein and sugarcaused by GDH was expected to cause the reduction in fattyacid content observed in leaves and more pronouncedly inroots. The 16-carbon and 18-carbon fatty acids changed werecommon constituents of the diacylglycerols in plant cellmembranes and chloroplast membranes. However, the mostabundant fatty acid in cell membranes, α-linolenic acid(18:3), was not altered in abundance in either leaves orroots. The other fatty acids unaltered in abundance are ofthe 16:3 class. These are mainly found in chloroplastmembranes, albeit in quantities far lower thanα-linolenic acid (18:3). Interestingly, only the fattyacids whose contribution to plant membrane composition isminor were reduced. The cells of GDH transgenics appear to beregulating closely the abundance of the most common fattyacids necessary for normal cellular function.

The TCA cycle intermediates that are increased by GDH,fumarate, malate, and citrate (Figure 1) may beassociated with the redirection of C away from fatty acidsynthesis and toward amino acid synthesis. Fumarate and malateare immediate precursors to pyruvate. Citrate is produced fromthe catabolism of acetyl-CoA.

Special nitrogen metabolism

Special nitrogen metabolites (amines, alkaloids, phenolics, andisoprenoids) may represent more than 50% of the compounds inplants in the 100–700 d range [36]. They providedefense against herbivores, microorganisms, or competingplants and color or scent to attract pollinating insects andseed-dispersing or fruit-dispersing animals. Their nitrogen is derivedfrom ammonium assimilation via the amino acids (the carbonskeletons may derive from many diverse pathways) so it wassurprising that none were decreased in leaves and only ninewere decreased in roots in GDH plants.

The abundance of just 2 amines (of the 48 detected)altered in response to GDH (Figure 4,Table 5, 60and 67, 66). Amines are products of arginine or ornithinemetabolism (that were affected by GDH) so the unchanged aminecontents were surprising. The amine N-caffeoylputrescine thatwas increased in shoots and decreased in roots by GDH (bytransport) accumulates during abiotic or biotic stress[37,38] and will stabilize histones, stabilizebiomembranes, inhibit viral replication, and regulate cellulargrowth [36]. Such changes directed by GDH may be usefulfor the economic production of plant secondary metabolites.

The alkaloids that were altered by GDH (9 of 34 detected)were mainly coumarin (68, 72, 73, Table 5) and quinone(61, 69, 71, 74, Table 5) derivatives. Alkaloids occur in about15% of plant taxa, including N tabacum [36].Most derive from amines that are synthesized from amino acids.They accumulate in tissues that are important for survival andreproduction providing chemical defense. Targets includeheart, liver, lung, kidney, CNS, and reproductive organs.Toxic alkaloids may have pharmacological uses at nontoxicdoses (eg, 62, 68, Table 5) [36].Scopoletin (62, Table 5)inhibits Escherichia coli O157, is antiviral, isanti-inflammatory (5 fold more than aspirin), and is an asthmatreatment [36, 39].Increasing leaf concentrations2–3 fold with GDH may be a useful approach tofinding new uses for the tobacco crop. Coumarin (68, Table 5), aperfumed liver and lung toxicant [40] was decreased 10fold by GDH in roots, potentially useful for the manipulationof diets based on root crops.

Some (8 of the 186 detected) phenolic compounds werealtered by GDH. The production of phenylpropanoids occurspredominantly from the amino acid phenylalanine [41].Quinones, monoterpenes, and modified side chains derive fromother pathways. The 2-fold increase in phenylalanine caused byGDH (Figure 3) may explain why phenolics are thepredominant class of special nitrogen metabolites increased inleaves (3) and roots (3) of GDH transgenics. Phenolics providemechanical support and barriers; insect attractant or repellents;antioxidants used in leather making; and flavor components inwines and herbal teas. Swainsonine (78, Table 5) isan inhibitor of mannosidase II, used as a cancer therapy[42]. The 5-fold increase in abundance could be useful.Nicotine (synthesized from ornithine), an animal stimulant andinsect repellent, was increased in roots (77,Table 5) but not in leaves [41]. Nicotine is synthesized in the roots and transported to the leaves soincreased synthesis may not produce a desirable outcome. Nopinone(80, Table 5) was the only isoprenoid affected by GDHbut it has no important pharmacological properties [41].

Nucleic acids

The synthesis of nucleic acids from glutamine is a majornitrogen sink in plant cells [36]. GDH did not alter theabundance of the common phosphonucleotides. Uridine wasincreased 4 fold. Uridine is a precursor of importantbiosynthetic compounds UMP, UDP, UTP and their glycosylderivatives. However, these compounds were not altered inabundance. Clearly the altered amino acid fluxes caused by GDHare being directed toward specific pathways and intermediates,leaving others unaffected.

Miscellaneous compounds

The 46 compounds we termed miscellaneous that were altered by GDHin tobacco included 29 that contain nitrogen but structurallycannot be classified with the special nitrogen metabolites[36]. The predominance of N-containing compounds suggeststhese alterations are directed by GDH-induced metabolism(Table 5(q), Table 5(r), andTable 5(s)). This group of compounds includes some ofmedicinal relevance(http://www.cieer.org/geirs/);an antihelmitic (98, Table 5); a tumorstatic thatbinds to nucleic acids (104, Table 5); a vitamin Cmetabolite that causes increased absorption, cellular uptake,accumulation and reduced excretion (96, Table 5); anootropic (a drug that enhances mental function; 99,Table 5); treatments for diabetes, high bloodpressure and arteriosclerosis (also found in bee royal jelly; 114,Table 5). An inhibitor of neutral sphingomyelinase(117, Table 5); a mycotoxin and an antitumor agent(134, Table 5); and a GABA uptake inhibitor (116,Table 5). Some constituents of cosmetics (135, 113,Table 5), flavoring agents (131, 124,Table 5), and a solvent (110, Table 5)were altered in GDH plants. Altering abundance of these compoundsmay provide alternate uses for the tobacco crop.

Some pesticide-like metabolites were altered by GDH althoughthe plants were not exposed to pesticides(119,126,127,133,138, Table 5). These compounds may be enzyme substratesoccurring naturally in plants that are structurally similar topesticides. Cataloging metabolites may lead to new leads forpesticidal chemical discovery [1].

Carcinogens and poisons were primarily altered in abundance inthe roots (100 and 131, 105, 107, 108, 111, 115, 128, 132, Table 5) consistent with the root synthesis ofthese compounds early in development and later translocationto the leaf [41]. The carcinogens and cigarettecomponents detected are specific to tobacco and most probablypart of its inherent secondary metabolism. Detection ofcarcinogens and poisons may serve to validate the use ofFT-ICR-MS and suggests applications in the association ofsmoking with cancer incidence.

Plant pigments, haem, and other porphyrins are major sinks forglutamate source pools in plants [27]. However, theglutamate flux perturbation caused by GDH does not alter theregulation or intermediates of pigment biosynthesis.

We conclude that GDH can be useful for plant metabolicengineering to increase or decrease the yield of a largenumber of chemical compounds. GDH may be a useful tool as thepharmaceutical industry discovers new plant-derived compoundsof therapeutic value.

The work presented here demonstrated that metabolite analysesby FT-ICR-MS provide a useful tool for the analysis of crypticphenotypes in transgenic plants. The analysis of data fromextracts without derivatization allows analysis of therelationships between various metabolites and the equivalenceof samples. If there are 40 000 different molecules amongall extant plant species in the range of 100–700 d[36], cataloging them by means other than FT-ICR-MS wouldbe a mammoth task [3, 4].Assuming there are 3–4 thousand differentmolecules in individual plant species in the range of100–700 d [4,11] we will have sampled about50%–60% (2 012) in two analyses (replicated). However, itis clear from our data that about 50% of the moleculesdetected are not in the databases we interrogated. Therefore,estimates of the chemical diversity of plant may be grosslyunderestimated. The development of a cell map and explorationof metabolic instantiations with that map will be impossiblewithout cataloging the consequences of metabolism accurately.

The sensitivity and resolution of FT-ICR-MS provides a usefulmethod for cataloging chemical diversity. Within the existinglimits, differences may be measured between samples comprisingmore than ten thousand cells. Therefore, the occurrence ofnovel compounds of biomedical significance in individuals,populations, species, and genera may be cataloged.

Gene manipulations and construction of plasmids

To examine the effects of NADPH-GDH in plants we used threelines, GDH10, GUS, and BAR, described previously[12, 13,14, 15,17]. GDH10 is a well-characterized independentline of Nicotiana tabacum var “Petite Havana SR1” thatexpresses the E coli gdhA gene. The line represents anearly regenerant and lacks noticeable variation from the wildtype under normal growth conditions. The gdhAgene inserted in GDH10 plants has an architecture andsegregation pattern consistent with a single site of insertion.The gene is under the control of the CaMV 35S promoter.Transcript abundances are equal when comparing roots and leaves.Enzyme activity is found in the cytoplasm but not plastids and isequal in roots and leaves. GUS is a well-characterizedindependent line of N tabacum var Petite Havana SR1 thatexpresses the modified gusA gene. The line represents anearly regenerant and lacks noticeable variation from the wildtype under normal growth conditions. The gusA geneinserted in GUS plants has an architecture andsegregation pattern consistent with a single site of insertion.The gene is under the control of the CaMV 35S promoter. Enzymeactivity is found in the cytoplasm but not plastids and is equalin roots and leaves. BAR is a well-characterized independent lineof N tabacum var Petite Havana SR1 that expresses theS hygroscopicus bar gene. The line represents an earlyregenerant and lacks noticeable variation from the wild typeunder normal growth conditions. The bar geneinserted in BAR plants has an architecture andsegregation pattern consistent with a single site of insertion.The gene is under the control of the CaMV 35S promoter. Enzymeactivity provides tolerance to phosphinothricin herbicide toroots, leaves, and cell culture derived from them. BAR and GUSwere chosen as adequate controls because they were notsignificantly different from wild-type SR1 across a wide range ofgrowth conditions, locations, and years [12,13, 14,15, 17].

Seeds of the lines described and clones used fortransformation are freely available on request andare being widely used for transformation of other plant species.

Plant material and growth conditions

Tobacco seeds were obtained from the seed stocks at theAgriculture Research Center, Southern Illinois University atCarbondale (Carbondale, Ill). Seeds were sown in 4-inch pots[14] containing a mixture of sand and soil (1:1).Seedlings were thinned to one plant per pot, watered daily,and grown on unshaded benches and in the Horticulture ResearchCenter, Southern Illinois University, from 9/99 to 9/03. Theconditions for the growth of plants for ¹³Nlabeling, and metabolomics are described in the comingsections. Seeds of each line used are available on request.

Preparation of cell free extracts and GDH assays

GDH assays were performed exactly as described [12]. Allpreparative steps were carried out at 4^°C. The specificactivity of aminating NADPH-GDH was quantified by measuringthe rate of oxidation of NADPH dependent on reductiveamination of alpha-ketoglutarate. Assays were performed at25^°C. The amount of protein in the extracts wasdetermined by Bradford assay.

Labeling of the glutamate pool by ¹³N

Three individual plants were fed 13NH4+ for 15 minutes via hydroponic solutions for root feeding and via excised stems for leaf feeding, then treated with liquid N, ground up, extracted with distiled water, filtered throughglass wool, and separated on an anion-exchange column (Dowex2X8-100) which retained glutamate. The eluate was washedthrough the column with another 10 mL of distiled waterand passed through a cation-exchange resin (Dowex50WX8-100) which bound NH4+. Glutamine came through in the eluate. The columns were washed with10 mL of 2M KCl to elute glutamate (anion-exchange column) and NH4+ (cation-exchange column). Eluates were collected in 20 mLscintillation vials and counted in a Canberra Packard gammacounter that was automatically corrected for decay time(¹³N has a half-life of 10 minutes). The percent labelincorporated was calculated using the following formula:[{(percent ¹³N as glutamate in the presence ofMSX by GDH10 line)-(percent ¹³N as glutamate inthe presence of MSX by non-GDH line)}/(percent¹³N as glutamate in the absence of MSX by the GDH10 line)].

Preparation of metabolite extracts for FT-ICR-MS assays

Three pooled leaf and root samples from each control andtransgenic genotype were used to remove spatial and geneticvariation not associated with GDH activity. About 100 mgof tissue was ground to which 1.0 mL of 50/50 (v/v)methanol/0.1% (w/v) formic acid was added [8]. Thesamples were homogenized and centrifuged. The supernatant wasused for the analyses. Each sample was mixed with a known andequal amount of a standard mix of serine, tetra-alanine,reserpine, Hewlett-Packard tuning mix, and theadrenocorticotrophic hormone fragment 4–10. Theseinternal calibration compounds produced 4–5 ionsof mass encompassing the range reported allowing for controlof spectra used for mass reports. The internal calibrationcompound peak area was used to detect nonbiological variationsin abundance reported allowing for control of spectra used forthe quantities reported. All analytes we purchased fromSigma-Aldrich (St Louis, Mo) and used without further purification.

FT-ICR-MS assays

Briefly, we used the Bruker Daltonics APEX III FT-ICR-MS equipped with a 7.0 Tesla magnet,electrospray, and APCI ionization sources [8]. Bothpositive and negative ionizations were carried out. Tips wereprepared as previously described [8]. For negativeionization, samples were introduced by capillary, diluted 1:19in 50% (v/v) methanol, 0.2% (v/v) formic acid, 49.8% (v/v)water. For positive ionization, samples were introduced bycapillary, diluted 1:19 in 50% (v/v) methanol, 0.2% (v/v)ammonium hydroxide, 49.8% (v/v) water. Flow rates were 5μL/min for electrospray and 100 μL/minfor APCI ionization sources. ESI, APCI, and ion transferconditions were optimized using a standard mix of serine,tetra-alanine, reserpine, Hewlett-Packard tuning mix, and theadrenocorticotrophic hormone fragment 4–10.Instrument conditions were optimized for ion intensity andbroadband accumulation over the mass range of 100–1000 d.One-megaword data files were acquired and asinm data transformation was performed prior to Fourier transform andmagnitude calculations.

Preparation of database searching for FT-MS assays