R. L. White, Department of Chemistry, Dalhousie University, 1459 Oxford Street, Halifax, Nova Scotia, Canada, B3H 4R2 Fax: 902 494 1310 Tel: 902 494 6403 E-mail: email@example.com
The pathways of glutamate catabolism in the anaerobic bacterium Fusobacterium varium, grown on complex, undefined medium and chemically defined, minimal medium, were investigated using specifically labelled 13C-glutamate. The metabolic end-products acetate and butyrate were isolated from culture fluids and derivatized for analysis by nuclear magnetic resonance and mass spectrometry. On complex medium, labels from l-[1-13C]glutamate and l-[4-13C]glutamate were incorporated into C1 of acetate and equally into C1/C3 of butyrate, while label derived from l-[5-13C]glutamate was not incorporated. The isotopic incorporation results and the detection of glutamate mutase and 3-methylaspartate ammonia lyase in cell extracts are most consistent with the methylaspartate pathway, the best known route of glutamate catabolism in Clostridium species. When F. varium was grown on defined medium, label from l-[4-13C]glutamate was incorporated mainly into C4 of butyrate, demonstrating a major role for the hydroxyglutarate pathway. Upon addition of coenzyme B12 or cobalt ion to the defined medium in replicate experiments, isotope was located equally at C1/C3 of butyrate in accord with the methylaspartate pathway. Racemization of d-glutamate and subsequent degradation of l-glutamate via the methylaspartate pathway are supported by incorporation of label into C2 of acetate and equally into C2/C4 of butyrate from d-[3-13C]glutamate and the detection of a cofactor-independent glutamate racemase in cell extracts. Together the results demonstrate a major role for the methylaspartate pathway of glutamate catabolism in F. varium and substantial participation of the hydroxyglutarate pathway when coenzyme B12 is not available.
Glutamate is a valuable substrate for anaerobic bacteria, serving as a precursor for the biosynthesis of cellular constituents and as an energy source upon catabolism to acetate, butyrate, carbon dioxide and ammonia [1,2]. Of the five chemically distinct pathways documented for glutamate catabolism in anaerobic bacteria , three have been associated with Fusobacterium species. In Fusobacterium nucleatum, an anaerobe implicated in periodontal disease , the hydroxyglutarate pathway (Scheme 1, route B) is well established [3,5,6] and is usually reported as the pathway of glutamate catabolism in Fusobacterium species [1,2,7,8]. An initial study  attributed [1-14C]glutamate catabolism to the hydroxyglutarate pathway in Fusobacterium varium, an intestinal anaerobe [9,10] and a source of butyrate for intestinal health [11,12]. A subsequent report , however, suggested the occurrence of three pathways in F. varium, namely the methylaspartate [1–3,14] (Scheme 1, route A), the hydroxyglutarate [1–3] (Scheme 1, route B) and the aminobutyrate pathway [3,14] (Scheme 1, route C). The initial products of each pathway (acetate, acetyl-CoA and crotonyl-CoA) are linked to butyrate via the core acetate–butyrate pathway [15,16] consisting of seven enzyme-catalyzed steps: acetate kinase (EC 188.8.131.52), phosphate acetyltransferase (EC 184.108.40.206), acetyl-CoA C-acetyltransferase (EC 220.127.116.11), acetoacetyl-CoA reductase (EC 18.104.22.168), 3-hydroxybutyryl-CoA dehydratase (EC 22.214.171.124), butyryl-CoA dehydrogenase (EC 126.96.36.199) and butanoyl-CoA:acetate CoA-transferase (EC 188.8.131.52).
[ Catabolism of glutamate to acetate, butyrate and CO2 via the methylaspartate (A), hydroxyglutarate (B) and aminobutyrate (C) pathways. Only the key intermediates are shown, and all carbons are numbered according to their original position in glutamate. Two reaction arrows indicate two or more enzyme-catalysed steps. Steps in the methylaspartate pathway (A) are catalysed by glutamate mutase (a1; EC 184.108.40.206; methylaspartate mutase), methylaspartase (a2; EC 220.127.116.11; 3-methylaspartate ammonia-lyase), citramalate dehydratase (a3; EC 18.104.22.168; 2-methylmalate dehydratase), citramalate lyase (a4; EC 22.214.171.124), pyruvate:flavodoxin/ferredoxin oxidoreductase (a5; EC 126.96.36.199; pyruvate synthase) and butyryl-CoA:acetate CoA-transferase (a6; EC 188.8.131.52). The enzymes NAD-specific glutamate dehydrogenase (b1; EC 184.108.40.206), 2-hydroxyglutarate dehydrogenase (b2; EC 220.127.116.11), glutaconate CoA-transferase (b3; EC 18.104.22.168), (R)-2-hydroxyglutaryl-CoA dehydratase (b4; EC 4.2.1.-) and glutaconyl-CoA decarboxylase (b5; EC 22.214.171.124) mediate the hydroxyglutarate pathway (B), while enzymes in the aminobutyrate pathway (C) are glutamate decarboxylase (c1; EC 126.96.36.199), 4-aminobutyrate transaminase (c2; EC 188.8.131.52), 4-hydroxybutyrate dehydrogenase (c3; EC 184.108.40.206), 4-hydroxybutyrate CoA-transferase (c4) and 4-hydroxybutanoyl-CoA dehydratase (c5; EC 220.127.116.11). Core acetate–butyrate pathway enzymes [15,16] are indicated by d. In the Fusobacterium varium genome (http://www.ncbi.nlm.nih.gov/; identification number ACIE00000000), gene sequences have been assigned to the enzymes a1, a2, a5, b1–b5 and c5. ]
A precedent for metabolic heterogeneity within a bacterial genus is provided by the occurrence of the methylaspartate and hydroxyglutarate pathways (Scheme 1, routes A and B, respectively) in Clostridium species [1,2,6,17]; the majority of Clostridium species utilize the methylaspartate pathway and only one route was reported for each species [1–3]. By contrast, recent genome sequencing studies have uncovered genes from the methylaspartate and hydroxyglutarate pathways in both F. nucleatum [15,18] and F. varium (http://www.ncbi.nlm.nih.gov/; identification number ACIE00000000). The gene encoding glutamate mutase, however, was the only methylaspartate pathway gene annotated in F. nucleatum.
A proteomic study of F. varium [16,19] detected the expression of the first two enzymes in the methylaspartate pathway and two hydroxyglutarate pathway enzymes, NAD-specific glutamate dehydrogenase and (R)-2-hydroxyglutaryl-CoA dehydratase, when l-glutamate was supplied in the growth medium. Further identification of the genes encoding enzymes in the hydroxyglutarate pathway [i.e. NAD-specific glutamate dehydrogenase, 2-hydroxyglutarate dehydrogenase, glutaconic CoA-transferase, (R)-2-hydroxyglutaryl-CoA dehydratase and glutaconyl-CoA decarboxylase] using PCR methods  confirmed the capability of F. varium to use the hydroxyglutarate pathway. On the other hand, the purification of methylaspartase (3-methylaspartate ammonia-lyase) to homogeneity  and the detection of glutamate mutase and methylaspartase as major components of the proteome under all growth conditions investigated  suggest that the methylaspartate pathway plays a prominent role in glutamate catabolism in F. varium.
Despite the published results summarized above, whether one or two pathways are used for glutamate catabolism in F. varium is not clear. In general, gene transcription and enzyme expression must provide the required protein catalysts to convert the initial substrate to the pathway end-products (e.g. glutamate → acetate and butyrate). While the detection of an enzyme within a pathway* (by proteomics [16,19] or enzyme assay [13,21,22]) supports the occurrence of an individual step, it may not be definitive evidence for the operation of the entire pathway. A study of glutamate catabolism in the anaerobic thermophile Caloramator coolhaasii  supports the latter point. Whereas catalytic activities for two enzymes in the hydroxyglutarate pathway were detected, results obtained by analysing the catabolism of isotopically labelled glutamate† were consistent with only the methylaspartate pathway, leading the authors to conclude that the thermophilic anaerobe utilizes only the methylaspartate pathway for glutamate catabolism.
In the present investigation, incubation of F. varium with a series of 13C-glutamates specifically labelled at defined positions, followed by NMR spectroscopy to determine the positions labelled in the metabolic end-products, provided evidence consistent with the catabolism of glutamate by two pathways. Enzymes in the predominant pathway were detected in crude cell extracts, and a glutamate racemase utilized for the catabolism of d-glutamate was partially purified.
Pathway determination using labelled l-glutamate
Catabolism of glutamate via the chemically distinct, enzyme-catalyzed reactions in the aminobutyrate, hydroxyglutarate and methylaspartate pathways delivers C1–C5 of glutamate to individual carbons in the acetate and butyrate end-products in patterns that are characteristic for each pathway (Scheme 1) [5,6,17]. Accordingly, formation of crotonyl-CoA from glutamate via the 4-aminobutyrate and hydroxyglutarate pathways, respectively, would yield butyrate predominantly derived from C2–C5 and C1–C4 of glutamate, whereas formation of acetate and acetyl-CoA by cleavage of the C2–C3 bond in glutamate in the methylaspartate pathway, and the synthesis of butyrate from two acetyl units via the core acetate–butyrate pathway [15,16], would transfer C1–C4 of glutamate equally into two positions of butyrate as observed in several Clostridium species . Therefore, the three routes can be distinguished unambiguously by following the catabolism of specifically 13C-labelled glutamate and 13C-NMR spectroscopy [5,23].
Pathway determination: complex medium
In separate experiments, F. varium was cultured on peptone medium  containing l-[1-13C]glutamate, l-[4-13C]glutamate or l-[5-13C]glutamate. The acetate and butyrate end-products were isolated from culture fluids as their p-bromophenacyl esters and separated chromatographically prior to the complementary determination of isotopic incorporations by 13C-NMR spectroscopy and electron ionization mass spectrometry (EI-MS) . Similar total isotopic enrichments were measured by the two techniques (Table 1). Label from l-[5-13C]glutamate (Table 1, Expt 2) was not transferred to the carboxylate end-products, a prediction of the hydroxyglutarate and methylaspartate pathways (Scheme 1) but inconsistent with glutamate catabolism via the 4-aminobutyrate pathway. On the other hand, the catabolism of l-[1-13C]glutamate and l-[4-13C]glutamate led to the isolation of carboxylate end-products with 13C enrichments much above natural abundance, demonstrating a substrate–product relationship and the existence of a pathway for glutamate catabolism in F. varium. The dilution of label, from 40% initially to about 2% in acetate and 6% in butyrate, indicated the conversion of other carbon sources in the peptone medium (e.g. glucose ) to carboxylate end-products.
Table 1. Isotopic enrichments in acetic and butyric acids derived from [13C]glutamic acids. Note that the sum of the 13C enrichments observed by NMR at two carbons in butyrate is equivalent to the single isotopic enrichment (I1) determined by EI-MS. For the EI-MS analysis, the relative peak intensities of isotopically labelled (I1) and unlabelled (I0) molecular ions are given. The standard deviation calculated from replicate scans, typically six, is given in parentheses. In the NMR analysis, the standard deviation calculated from the natural abundance signals contributed by the derivatization reagent is given in parentheses.
13C enrichment (%) (EI-MS analysis)
13C enrichment (%) (NMR analysis)
a Peptone medium. b Defined medium. c Enrichments confirmed by 13C satellites in the 1H NMR spectra (acetate C2, 1JCH = 130 Hz; butyrate C4, 1JCH = 126 Hz; butyrate C3, 1JCH = 129 Hz). d Defined medium containing coenzyme B12. e Defined medium containing CoCl2. f Contained < 6.5%l-glutamate by HPLC.
Label from l-[1-13C]glutamate was specifically incorporated into C1 of acetate and either C1 or C3 of butyrate (Table 1, Expt 1). The equal 13C enrichment at two carbon atoms of butyrate is most consistent with degradation of glutamate to C2 units by the methylaspartate pathway (Scheme 1, route A) and synthesis of butyrate from two identical C2 units. The incorporation of label from l-[1-13C]glutamate is not predicted by the aminobutyrate pathway, while an unsymmetrical distribution of label at C1/C3 (mostly at C1 [5,6]) of butyrate would result from the hydroxyglutarate pathway.
Catabolism of l-[4-13C]glutamate also yielded C1-enriched acetate and C1- or C3-enriched butyrate (Table 1, Expt 3). The equal enrichment at C1/C3 of butyrate is again consistent with glutamate degradation by the methylaspartate pathway and butyrate synthesis from two C2 units. In this case, however, catabolism of l-[4-13C]glutamate via the aminobutyrate and hydroxyglutarate pathways would place 13C label at C2 of acetate and either C2 or C4, respectively, of butyrate (Scheme 1) [5,6,17]. The C2 and C4 signals in the 13C NMR spectrum of p-bromophenacyl butyrate, however, had the intensities expected for natural 13C abundances, indicating no 13C enrichment at these positions. In this NMR technique, enrichments as low as 0.2–0.3%13C can be detected . Given the observed 6% enrichments at C1/C3 and assuming enrichments just under the detection limit of 0.3% at C2 and C4, then < 5% of the labelled l-[4-13C]glutamate was catabolized via the aminobutyrate and hydroxyglutarate pathways and at least 95% via the methylaspartate pathway.
The butyrate samples isolated from the l-[1-13C]glutamate and l-[4-13C]glutamate experiments (Table 1, Expts 1 and 3) had similar isotopic enrichments of about 6%. By the methylaspartate pathway (Scheme 1), however, acetate and acetyl-CoA (the more direct precursor of butyrate) are derived from C1/C2 and C3/C4 of glutamate, respectively, and preferential incorporations of label into acetate from [1-14C]glutamate and butyrate from [4-14C]glutamate have been documented in several Clostridium species . A notable difference between the previous and the current experiments is the larger quantity of isotopically labelled glutamate (0.85 mmol) supplied to F. varium and the consequent greater availability of the 13C label over a larger portion of the culture period. As a result, the distribution of label in acetate and acetyl-CoA was facilitated by CoA transferases (e.g. a6), phosphate acetyltransferase and acetate kinase, and equivalent isotopic enrichments were attained in butyrate regardless of the initial position of 13C in glutamate.
Additional evidence for the methylaspartate pathway was obtained by incubating l-glutamate with cell extracts prepared by sonication of mid-log phase cells of F. varium. HPLC analysis of crude cell extract demonstrated that glutamate (initially 9 mm) was completely consumed within 10 min. However, the accumulation of another amino acid was not detected. When l-glutamate was added to crude cell extract containing o-phenanthroline, an inhibitor of citramalate dehydratase , formation of methylaspartate was detected by HPLC. Incubation of crude cell extract (plus o-phenanthroline) with l-[4-13C]glutamate‡ for 200 min led to the detection of 13C NMR signals corresponding to residual glutamate (36.2 p.p.m.), mesaconate (141.5 p.p.m., C-2), and 3-methylaspartate (42.7 p.p.m., C-3). The location of 13C enrichment in 3-methylaspartate indicates cleavage of the C2–C3 bond and formation of a bond between C2 and C4 as documented for the glutamate mutase catalysed reaction .
In a complementary experiment, 3-methylaspartate (0.36 mm) and glutamate (0.67 mm) were detected by HPLC after a 5-min incubation of partially purified cell extract with mesaconate (10 mm) and NH4Cl (25 mm). The 3-methylaspartate concentration reached a maximum of 2.1 mm at 15 min, whereas the glutamate concentration steadily increased to 2.6 mm at 30 min, indicating the formation of 3-methylaspartate and its subsequent conversion to glutamate. Equilibrium constants  favouring 3-methylaspartate over mesaconate, and glutamate over 3-methylaspartate, are consistent with the millimolar accumulation of the amino acids. When the experiment was repeated using a charcoal-treated, partially purified cell extract, more 3-methylaspartate (2.6 mm, 5 min; 3.0 mm, 15 min) and less glutamate (0.07 mm, 5 min; 1.2 mm, 30 min) accumulated. The reduced activity of glutamate mutase is attributed to charcoal adsorption of 5′-deoxyadenosylcobalamine (coenzyme B12) . HPLC analysis of a 3-h incubation mixture of mesaconate and Sephadex G-150 fractions containing methylaspartase revealed the formation of l-threo-(2S,3S)- and l-erythro-(2S,3R)-3-methylaspartate (retention times of 17.1 and 4.5 min, respectively), as previously observed .
The transformations observed in the cell extracts are consistent with the reactions catalysed by glutamate mutase and methylaspartase, the first two enzymes in the methylaspartate pathway.
Pathway determination: defined medium
While the results collected from F. varium grown on complex medium are only consistent with the methylaspartate pathway, two enzymes from each of the hydroxyglutarate and methylaspartate pathways were expressed by F. varium cultured on chemically defined minimal medium containing 40 mm l-glutamate . Whether one or both pathways are responsible for glutamate catabolism on defined medium was probed using l-[4-13C]glutamate. The defined culture conditions were identical to those used in the proteomic study , and the acetate and butyrate end-products were converted to their p-bromophenacyl derivatives  for 1H and 13C NMR analysis. In the three experiments on defined medium (Table 1, Expts 4–6), incorporations of 13C label were higher than those observed on complex medium (Table 1, Expts 1 and 3). On complex medium, glucose also is a major source of acetate and butyrate , while the dilution of isotopic label on defined medium is consistent with the catabolism of other amino acids to carboxylate end-products [1,2].
The 13C enrichment observed in both carbons of acetate and all four carbons of butyrate (Table 1, Expt 4) demonstrated that more than one pathway of glutamate catabolism was used by F. varium on defined medium. About 70% of the 13C label in butyrate was located at C4, the position predicted by the hydroxyglutarate pathway (Scheme 1, route B). Crotonyl-CoA, an intermediate in the hydroxyglutarate and core acetate–butyrate pathways, also would be enriched at C4. Cleavage of [4-13C]crotonyl-CoA to [2-13C]acetyl-CoA, and subsequent recombination of [2-13C]acetyl-CoA units, account for smaller enrichments at C2 of acetate and butyrate, and a minor contribution of 13C at C4 of butyrate. The labelling at C1 of acetate and C1/C3 of butyrate are consistent with catabolism of a small fraction of l-[4-13C]glutamate by the methylaspartate pathway (Scheme 1, route A).
The defined medium used in the above experiment, however, contained neither coenzyme B12, a necessary cofactor for the functioning of glutamate mutase and thus the methylaspartate pathway , nor cobalt ion, an essential constituent of coenzyme B12. When the isotopic experiment was repeated on defined medium supplemented with either coenzyme B12 or cobalt ions (Table 1, Expts 5 and 6), highly enriched acetate (C1) and butyrate (equally at C1/C3) were isolated. The positions of these high isotopic incorporations were consistent with the methylaspartate pathway. The small enrichment at C4 of butyrate shows a minor participation of the hydroxyglutarate pathway when coenzyme B12 is available.
Catabolism of d-glutamate
Previous investigations implicating the aminobutyrate and hydroxyglutarate pathway for glutamate catabolism in F. varium [8,13] employed racemic radioactively labelled glutamate. If chemically distinct pathways were utilized for the catabolism of d- and l-glutamate in F. varium, then different labelling patterns of the carboxylate end-products would be expected upon catabolism of the individual 13C-labelled stereoisomers (vide supra). The merit of this hypothesis is supported by the utilization of d-glutamate by F. varium , intestinal bacteria in rats  and extracts of Clostridium tetanomorphum , as well as reports of distinct pathways for the catabolism of glutamate enantiomers in rats  and lysine enantiomers in F. nucleatum  and Clostridium species [2,33].
When F. varium cultures were incubated separately with d- and l-glutamate (18 mm), HPLC analysis demonstrated that d- and l-glutamate were assimilated at different rates ; for example, after 10 h of incubation, 0.3 mm l-glutamate and 7 mm d-glutamate remained. The F. varium cells collected in the time-course experiment were extracted with 80% aqueous ethanol ; HPLC analysis showed no intracellular accumulation of glutamate, consistent with the degradation of both d- and l-glutamate. F. varium cultures supplemented with d-[3-13C]glutamate (Table 1, Expt 7) yielded 13C-enriched acetate (1.5% at C2) and butyrate (5.3% and 5.4% at C2 and C4, respectively). The isotopic incorporation patterns were consistent with catabolism of d-glutamate via the methylaspartate pathway. Similar to the incorporation of C4 via the methylaspartate pathway, the glutamate-mutase-catalyzed rearrangement of glutamate directs the incorporation of C3 to different positions in acetate and butyrate than the hydroxyglutarate and aminobutyrate pathways (Scheme 1). Consequently, the lack of 13C enrichment at C1 of acetate and C1/C3 of butyrate excludes significant participation of both the hydroxyglutarate and aminobutyrate pathways.
Detection of glutamate racemase
Separate incubations of crude cell extract with d-glutamate (9 mm initially; 1 mm after 2 h) and l-glutamate (9 mm initially; below detection limit at 10 min) showed a slower rate of d-glutamate degradation. No other amino acid accumulated under these conditions, but repetition of the experiments with the addition of o-phenanthroline (vide supra) led to the HPLC detection of 3-methylaspartate. Two explanations are possible for the slower rate of d-glutamate degradation: either d-glutamate is a poorer substrate for l-glutamate mutase or an additional rate-limiting enzyme-catalyzed step is needed for degradation of the d enantiomer.
d-Glutamate, however, is neither a substrate nor an inhibitor for the glutamate mutase purified from Clostridium cochlearium . Earlier investigations using partially purified glutamate mutase from C. tetanomorphum [29,36] attributed the small conversion observed for d-glutamate to the presence of a glutamate racemase.
While the presence of a racemase provides the simplest explanation for the transformation of d- to l-glutamate, an alternative conversion of d- to l-glutamate via 2-oxoglutarate is plausible. The required oxidation step may be catalysed by a d-glutamate dehydrogenase or a d-glutamate-specific transaminase (EC 18.104.22.168) that has been characterized in other bacteria . In crude cell extracts of F. varium, no d-glutamate dehydrogenase activity but a low level of l-glutamate dehydrogenase activity was detected by spectrophotometric assay . No glutamate formed by transamination in crude cell extracts incubated with 2-oxoglutarate and either dl-alanine, d-aspartate or dl-phenylalanine. When l-aspartate was incubated with partially purified cell extract, l-glutamate formation was detected by HPLC, indicating the presence of an aminotransferase for l amino acids. Thus, degradation of d-glutamate by dehydrogenation or transamination is not favoured, leaving racemization as the most likely possibility.
To probe for racemase activity, cell extracts were charcoal treated  to adsorb coenzyme B12 and inactivate glutamate mutase. When each enantiomer of glutamate (10 mm) was incubated with charcoal-treated, partially purified cell extract, HPLC determination of enantiomeric ratios demonstrated that the racemization of glutamate was complete after 3 h of incubation (Fig. 1). The total concentration of glutamate remained constant over this period. The rates of glutamate racemization measured in cell extracts prepared from cells grown in the presence of d- or l-glutamate (17 mm) were similar, suggesting that the racemase is constitutive and not induced by d-glutamate. Separate additions of the coenzymes pyridoxal phosphate, NAD+, NADP+ and FMN did not stimulate the rate of glutamate racemization, suggesting cofactor independence (further ruling out dehydrogenation and transamination). The glutamate racemase activity eluting from Sephadex G-150 corresponded to a molecular mass of approximately 20 kDa, somewhat smaller than the 29 kDa predicted from the annotated glutamate racemase gene (http://www.ncbi.nlm.nih.gov/; identification number ACIE00000000), but distinguishing it from the predicted 15 and 51 kDa subunits of glutamate mutase.
When d-glutamate (10 mm) was incubated with partially purified cell extract in D2O, the intensity of the 1H NMR signal for the alpha proton (3.72 p.p.m.) decreased and the coupling pattern for H3 simplified over time. At 48 h, about 85% exchange had occurred, and an l/d enantiomeric ratio of 0.7 was determined by HPLC. Thus glutamate underwent both solvent exchange and racemization during the incubation.
Experimental evidence collected over the past 40 years strongly supports the existence of alternative pathways for glutamate catabolism in anaerobic bacteria [1–3] and within four individual Fusobacterium species, including F. varium [13,16]. Of the three pathways proposed for glutamate degradation in F. varium, only one, the methylaspartate pathway, is consistent with the transfer of 13C label from glutamate to the specific carbons of acetate and butyrate enriched in the experiments conducted on complex medium. The equal 13C enrichments at two carbons of butyrate determined in these experiments are consistent with the results of previous 14C-labelling studies  in other anaerobes utilizing the methylaspartate pathway and attributed to the formation of butyrate from two acetyl units via the core acetate–butyrate pathway [15,16]. The ready utilization and transfer of label confirms glutamate as a major substrate for F. varium [24,38].
The first two enzymes in the methylaspartate pathway, glutamate mutase and methylaspartase, have been annotated in the F.varium genome and detected as major proteome constituents under all the culture conditions investigated [16,19]. Methylaspartase from F. varium has been purified to homogeneity and characterized . The catalytic activities of these enzymes led to the accumulation of intermediates of the methylaspartate pathway in cell extracts supplemented with substrate(s) and o-phenanthroline, an inhibitor of citramalate dehydratase.
Recent genomic annotations (http://www.ncbi.nlm.nih.gov/; identification number ACIE00000000), proteomic experiments  and PCR amplifications  have demonstrated that F. varium carries the genes associated with the hydroxyglutarate pathway but not the aminobutyrate pathway. In cell extracts of F. varium, catalytic activity was attributed to glutamate dehydrogenase and 2-hydroxyglutarate dehydrogenase , the first two enzymes in the hydroxyglutarate pathway. The expression of NAD-specific glutamate dehydrogenase and (R)-hydroxyglutaryl-CoA dehydratase, two enzymes in the hydroxyglutarate pathway, was detected only when F. varium was grown on defined medium containing l-glutamate at 40 mm . Under identical culture conditions in the present investigation, isotopic labelling demonstrated a major role for the hydroxyglutarate pathway and a minor role for the methylaspartate pathway. The latter contrasts with the prominent expression of glutamate mutase and methylaspartase by F. varium under these defined culture conditions . The prominence of the methylaspartate pathway was restored when coenzyme B12 or cobalt ion was included in the defined growth medium, suggesting that the glutamate mutase expressed in the absence of its essential coenzyme is catalytically inactive. With an inactive enzyme, the methylaspartate pathway would not function. The essential role of the coenzyme was not evident from the protein analysis conducted during the proteomic investigation . Furthermore, the similar isotopic results obtained when defined medium was supplemented with coenzyme B12 or cobalt ion indicate that F. varium is able to biosynthesize active coenzyme B12 when cobalt ions are available. The isotopic evidence clearly shows that availability of coenzyme B12 strongly influences the direction of glutamate catabolism in F. varium. The expression of enzymes in the hydroxyglutarate pathway may be influenced by a high glutamate concentration or another unidentified environmental factor , allowing F. varium to respond by shifting from one glutamate pathway to another.
In anaerobes, ATP is mainly formed by substrate level phosphorylation , particularly from acetyl phosphate derived from acetyl-CoA by phosphotransacetylation. In F. varium, acetyl-CoA also is linked to butyrate via a core acetate–butyrate pathway [15,16] used to regenerate NAD+. Glutamate catabolism by both the aminobutyrate and hydroxyglutarate pathways leads to crotonyl-CoA, an intermediate in the core acetate–butyrate pathway. Formation of crotonyl-CoA, however, requires the consumption of an ATP equivalent, and formation of acetyl-CoA requires oxidation of crotonyl-CoA. In the methylaspartate pathway, acetate and pyruvate are formed without NAD+/NADH dependent oxidation and reduction steps, and pyruvate is converted directly to acetyl-CoA without the need for an ATP equivalent. With fewer enzyme-catalyzed steps and a very small thermodynamic cost to form pyruvate , the methylaspartate pathway provides a shorter and therefore more efficient route for ATP generation from glutamate. On the other hand, the methylaspartate pathway does rely on the availability of coenzyme B12 , a product of a long biosynthetic pathway employing about 30 enzyme-catalyzed steps . Potential loss of the biosynthetic capabilities for coenzyme B12 provides a rationale for maintaining genes for the hydroxyglutarate pathway for the catabolism of glutamate in F. varium.
The methylaspartate pathway differs from the hydroxyglutarate pathway by converting glutamate to pyruvate, also a product of d-glucose catabolism by the Embden–Myerhof–Parnas pathway in F. varium . Upon decarboxylation catalysed by pyruvate:flavodoxin/ferredoxin oxidoreductase [3,41], electrons from pyruvate are transferred to ferredoxin. Subsequent transfer of the electrons from ferredoxin to H+ forming H2 (catalysed by ferredoxin hydrogenase (EC 22.214.171.124)), maintains redox balance without utilization of acetyl-CoA [41,42]. Production of hydrogen (H2) by F. varium  and annotations of the genes for these ferredoxin enzymes in the F. varium genome (http://www.ncbi.nlm.nih.gov/; identification number ACIE00000000) suggest that this route contributes to maintaining redox balance in F. varium. The recognition of coupling between ferredoxin reduction, crotonyl-CoA reduction and NADH oxidation in Clostridium species has provided a connection between hydrogen formation and the hydroxyglutarate pathway [43,44]. Formation of pyruvate by the methylaspartate pathway, however, allows F. varium to employ a common mode of energy conservation for both glutamate and glucose catabolism. Alternatively, non-oxidative cleavage of pyruvate to acetyl-CoA and formate catalysed by the pyruvate-formate lyase (EC 126.96.36.199) annotated in the F. varium genome (http://www.ncbi.nlm.nih.gov/; identification number ACIE00000000) would provide a redox advantage by avoiding the formation of reduced coenzyme upon conversion of pyruvate to acetyl-CoA.
The utilization of d-glutamate via a glutamate racemase expands the pool of possible energy substrates and provides F. varium with an environmental advantage. Other bacterial pathways starting with racemization as the first step are utilized for catabolism of ornithine  and mandelic acid , but racemases are present more commonly to provide d-amino acids as building blocks for bacterial cell walls . While the formation of a small amount of d-glutamate in an assay of glutamate mutase from C. tetanomorphum was attributed to a glutamate racemase contamination [29,36], a later study reported that no glutamate racemase activity was detected in Clostridium species . In F. varium, glutamate racemase may support both cell wall formation and substrate degradation; the recognition of d-glutamate catabolism in Lactobacillus arabinosus led to the initial characterization of glutamate racemase [49,50]. The lack of a cofactor requirement and the relatively small molecular mass of the F. varium enzyme are similar to the properties of known cofactor-independent glutamate racemases (EC 188.8.131.52) .
Heterogeneity in the 16S–23S rDNA internal transcribed spacer regions has been noted within Fusobacterium species , and metabolic heterogeneity for the metabolism of amino acids in Fusobacterium species is supported by the prominence of the methylaspartate pathway in F. varium and the hydroxyglutarate pathway in F. nucleatum [5,6,13], together with the use of 2,7-diaminopimelic acid or lanthionine, respectively, in cell wall peptidoglycan . Similarly, species within the genus Clostridium are known to utilize either the methylaspartate or the hydroxyglutarate pathway for glutamate catabolism , and a recent genome study of Clostridium sticklandii has revealed genes coding for proteins in three pathways of threonine catabolism . In F. varium, genes in two pathways of glutamate degradation are documented (http://www.ncbi.nlm.nih.gov/; identification number ACIE00000000) , and supported by the isotopic evidence collected in this investigation. On the other hand, enzymes in the methylaspartate and hydroxyglutarate pathways have been detected in extracts of Caloramator coolhaasii , but isotopic labelling studies are consistent with the catabolism of glutamate via only the methylaspartate pathway. Whether genetic information for multiple pathways is more widespread may be revealed as genomic sequencing is applied to other anaerobes.
Further information is required to define the parameters controlling the expression of hydroxyglutarate pathway genes in F. varium. The presence of a racemase that allows the utilization of both enantiomers of glutamate is further evidence of metabolic versatility in F. varium. Together, these metabolic capabilities provide F. varium with competitive advantages for adapting to genetic mutations and environmental changes in its natural anaerobic habitat.
Materials and general methods
l-[4-13C]Glutamic acid (95 atom%) was provided by Cambridge Isotope Laboratories (Woburn, MA, USA), and l-[1-13C]-, l-[5-13C]- and d-[3-13C]glutamic acid (all 99 atom%) were supplied by C/D/N Isotopes (Vaudreuil, QC, Canada). The d-[3-13C]glutamic acid contained < 6.5% of the l enantiomer (HPLC analysis). All other compounds and reagents were obtained from Sigma-Aldrich (Oakville, ON, Canada), unless indicated otherwise.
Sheep-blood agar (obtained as prepared plates from a local hospital laboratory) contained (g·L−1) tryptone, 14.0; peptone, 4.5; yeast extract, 4.5; NaCl, 5.0; agar, 12.5; and defibrinated sheep blood, 50 mL. BHI agar plates were prepared from autoclaved (120 °C, 20 min) brain heart infusion agar (52.0 g·L−1) (Fluka, Buchs, Switzerland). Peptone medium  contained (g·L−1) trypticase peptone (BBL Microbiology System, Cockeysville, MD, USA), 5.0; proteose peptone (Difco Laboratories, Detroit, MI, USA), 5.0; yeast extract (Difco), 5.0; d-glucose, 5.0; NaCl, 5.0; and l-cysteine HCl, 0.8.
Microorganism and culture conditions
Fusobacterium varium (NCTC 10560/ATCC 8501) was maintained by weekly subculture on sheep-blood agar or BHI agar. Suspensions of 48-h sheep-blood agar surface cultures in sterile peptone medium were used to initiate liquid peptone medium cultures. Sheep-blood agar inoculated from liquid cultures was incubated for 48 h to test for microbial contaminants. Cultures that showed growth under aerobic conditions or atypical morphology for F. varium under an anaerobic atmosphere were discarded. Microbial transfers were performed in air; all cultures were incubated at 37 °C in anaerobic jars containing palladium catalyst and an atmosphere of H2/CO2/N2 (5:10:85, v/v).
HPLC determination of amino acids
Pre-column o-phthalaldehyde derivatization of samples and reversed-phase separation of the fluorescent isoindole derivatives were carried out as described previously . Using gradient elution, glutamate and threo-3-methylaspartate had retention times of 1.2 and 2.8 min, respectively.
HPLC determination of amino acid stereoisomers
Fluorescent isoindole derivatives of amino acids were prepared using o-phthalaldehyde and N-acetyl-l-cysteine . Separations of stereoisomers were achieved on a reversed-phase column using individual binary gradients for glutamate  and 3-methylaspartate .
Determination of isotopic enrichments
NMR spectra were acquired on a Bruker AC 250F or a Bruker Avance 500 MHz spectrometer. The 13C enrichments of the acetate and butyrate carbons in the phenacyl derivatives were calculated from 13C NMR and EI-MS spectral peak intensities as described previously .
Metabolism of labelled glutamate by intact cells
Peptone medium (50 mL) supplemented with 13C-labelled d- or l-glutamic acid (17 mm, 125 mg; 40 atom%13C) was inoculated with F. varium from one 9-cm-diameter sheep-blood agar plate and incubated under anaerobic conditions for 48 h.
Chemically defined minimal medium (CDMM , 5 mL) supplemented with glucose (20 mm) was inoculated with a colony picked from BHI agar. After incubation under anaerobic conditions at 37 °C for 16 h, the cell suspension (0.2 mL) was used to inoculate fresh CDMM (20 mL) containing l-[4-13C]-glutamate (40 mm, 117 mg; 40 atom%13C) before incubation under anaerobic conditions at 37 °C for 24 h. In replicate experiments, CDMM was supplemented with coenzyme B12 (1 μm) or cobalt (II) chloride (1 μm) in addition to l-[4-13C]glutamate.
Derivatization of end-products
The supernatants from centrifugation (23 000 g, 20 min) were adjusted to pH 9.5 and lyophilized. As described previously , acetate and butyrate in the lyophilized residues were converted to p-bromophenacyl esters and chromatographically separated prior to NMR and mass spectral analysis. The isolated yields of p-bromophenacyl acetate and butyrate were 0.1–0.2 mmol and 0.1–0.5 mmol, respectively.
Experiments using cell extracts
Peptone medium (4 L) supplemented with l-glutamic acid (0.5 g·L−1) was inoculated with F. varium from three 9-cm-diameter sheep-blood agar plates and incubated under anaerobic conditions for 8.5 h. After centrifugation (8200 g, 20 min), the cell pellet (∼ 10 g) was washed twice with potassium phosphate buffer (50 mm, pH 7.4) containing dithiothreitol (DTT, 2 mm) and stored at −78 °C. The frozen cells were suspended in cold potassium phosphate buffer (23 mL, 50 mm, pH 7.4; DTT, 5 mm) and sonicated for 3 min in 15-s intervals with cooling in an ice–NaCl bath. The sonicated suspension (27 mL) was centrifuged (38 700 g, 20 min). In some experiments, the supernatant (crude cell extract) (15 mL) was stirred with charcoal (100 mg) for 30 min under N2 and clarified by centrifugation (38 700 g, 20 min).
All steps were carried out at 4 °C; centrifugation was performed at 30 000 g for 20 min. Protamine sulfate (8 mL, 1% solution) was added dropwise over 15 min to crude cell extract (15 mL) stirred under N2. Stirring was continued for 10 min, and the mixture was centrifuged. Solid (NH4)2SO4 was added to the supernatant (19.5 mL). The precipitate formed between 45% and 80% saturation (partially purified cell extract) was collected by centrifugation, resuspended in potassium phosphate buffer (4 mL, 50 mm, pH 7.4; DTT, 5 mm) and stored at −15 °C.
Molecular mass determination for glutamate racemase
Partially purified cell extract (4 mL) was applied to a Sephadex G-150 column (2.5 × 50 cm) pre-equilibrated in potassium phosphate buffer (50 mM, pH 7.4) containing DTT (2 mm) and dl-glutamate (1 mm). Fractions (3.8 mL) eluted with the same buffer were collected at 0.4 mL·min−1 flow rate and monitored at 254 and 280 nm. Individual fractions were assayed for glutamate racemase activity as described below. For molecular mass determination, the column was calibrated using mixtures of blue dextran (0.3 mg), alcohol dehydrogenase (4.4 mg), bovine serum albumin (1.8 mg), carbonic anhydrase (3.0 mg) and cytochrome c (1.1 mg) dissolved in the same buffer (1 mL).
Detection of enzyme activity
Crude or partially purified cell extract supplemented with substrate and cofactors was incubated under N2 at 30 °C. Samples (100 μL) taken from the incubation mixture were heated in a sand bath (130 °C, 30 s) to inactivate enzymes and centrifuged (15 400 g, 10 min). The supernatant was diluted with water and stored (−15 °C) for HPLC analysis.
Glutamate mutase and methylaspartase
(a) A mixture of crude cell extract (1.2 mL), potassium phosphate buffer (50 mm, pH 7.4) and either d- or l-glutamate (10 mm) in a total volume of 1.5 mL was incubated under N2, and samples were analysed by HPLC. For monitoring glutamate transformations by 13C NMR spectroscopy, l-[4-13C]glutamate (7.6 mm) was mixed with crude cell extract (700 μL), o-phenanthroline (0.5 mm) and D2O (100 μL) in a total volume of 950 μL.
(b) Mesaconate (10 mm) was incubated with non-intrinsic factor (300 units), o-phenanthroline (0.5 mm), NH4Cl (25 mm), potassium phosphate buffer (50 mm, pH 7.4) and partially purified cell extract (prepared from either crude or charcoal-treated crude extract, 480 μL), in a total volume of 1.2 mL. Samples were removed periodically to monitor the formation of methylaspartate and glutamate by HPLC.
(c) Absorbance (240 nm) changes  in reaction mixtures (1.0 mL) containing dl-threo-3-methylaspartate (10.8 mm), KCl (10 mm), MgCl2 (1 mm), ethanolamine (50 mm, pH 9.7) and partially purified cell extract (100 μL) were used to monitor mesaconate formation.
Absorbance (340 nm) changes  in a mixture of d- or l-glutamate (4 mm), NAD+ (1 mm), tris(hydroxymethyl)aminomethane-HCl (pH 8.5, 100 mm) and either crude (200 μL) or partially purified (100 μL) cell extract (total volume 1.0 mL) were used to monitor NADH formation.
Amino acid transaminases
A reaction mixture (0.5 mL) containing 2-oxoglutarate (4 mm), o-phenanthroline (0.5 mm), crude cell extract (375 μL) and either dl-alanine, d-aspartate, l-aspartate or dl-phenylalanine at 25 mm was incubated for 1 h and analysed by HPLC.
A reaction mixture (0.5 mL) containing charcoal-treated, partially purified cell extract (200 μL), o-phenanthroline (0.5 mm), either d- or l-glutamate (10 mm) and potassium phosphate buffer (50 mm, pH 7.4) was incubated. Samples were removed periodically for HPLC determinations of amino acids and amino acid stereoisomers. To test the effect of cofactors on racemase activity, the above incubation was repeated with the addition of non-intrinsic factor (50 units) and either flavin mononucleotide sodium salt (FMN, 1 mm), NAD+ (1 mm), NADP+ (1 mm) or pyridoxal phosphate (1 mm).
For the detection of enzyme-catalysed isotope exchange by 1H NMR spectroscopy, partially purified cell extract was prepared by dissolving the (NH4)2SO4 precipitate in potassium phosphate buffer (71 mm, pH 7.4; DTT, 1.2 mm) in D2O. 1H NMR spectra were recorded at 0, 15 and 48 h.
The detection of all enzymes in a multistep pathway is a technical challenge, complicated by the need for specific enzyme assays, as well as the potential insolubility and instability (e.g. oxygen sensitivity) of the enzymes. The availability of a convenient spectrophotometric assay and the stability of methylaspartase  have contributed to the use of this enzyme as a marker for the methylaspartate pathway in bacteria [21,22]. In proteomics, the enzyme must be within the protein subset selected for identification and be present above the detection limit of the analytical technique if it is to be identified.
Isotopic labelling is a well-established method that not only demonstrates substrate–product connections via a multistep sequence of enzyme-catalyzed reactions, but also can distinguish between alternative pathways. In the latter context, the labelling pattern of metabolic end-products derived by catabolism of isotopically labelled glutamate (Scheme 1) is characteristic of the aminobutyrate, hydroxyglutarate and methylaspartate pathways [5,6,17,23].
Upon the expected conversion of glutamate to 3-methylaspartate and mesaconate, C4 of glutamate would change from a secondary sp3 carbon to a tertiary sp3 carbon and then to an sp2 carbon, providing large chemical shift differences between the substrate and the corresponding carbons in the products.
We thank the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Innovations in Chemistry Fund, Dalhousie University, for financial support, and the Ministry of Health and Medical Education, Iran, for partial support to M.R. Also, we are indebted to NMR-3 for NMR spectra; Drs J.S. Grossert and J.-Y. Kim for collecting mass spectra; Drs H.N. Shah and S.E. Gharbia for helpful advice and assistance with initial experiments; and the Department of Applied Oral Sciences, Dalhousie University, for providing facilities to culture microorganisms.