J Biol Chem, Vol. 275, Issue 19, 14031-14037, May 12, 2000

Two Glyceraldehyde-3-phosphate Dehydrogenases with Opposite Physiological Roles in a Nonphotosynthetic Bacterium^*

Sabine Fillinger^§, Sandrine Boschi-Muller^¶, Saïd Azza^¶, Etienne Dervyn, Guy Branlant^¶, and Stéphane Aymerich^**

From the  Génétique Moléculaire et Cellulaire, INRA-CNRS (URA1925), 78850 Thiverval-Grignon, France, ^¶ UMR7567-CNRS-UHP-Maturation des ARN et Enzymologie Moléculaire, Faculté des Sciences, Bld des Aiguillettes, BP239, 54506 Vandoeuvre-les-Nançy, France, and the  Génétique Microbienne, INRA, 78352 Jouy-en-Josas, France

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Bacillus subtilis possesses two similar putative phosphorylating glyceraldehyde-3-phosphate dehydrogenase (GAPDH) encoding genes, gap (renamed gapA) and gapB. A gapA mutant was unable to grow on glycolytic carbon sources, although it developed as well as the wild-type strain on gluconeogenic carbon sources. A gapB mutant showed the opposite phenotype. Purified GapB showed a 50-fold higher GAPDHase activity with NADP⁺ than with NAD⁺, with K_m values of 0.86 and 5.7 mM, respectively. lacZ reporter gene fusions revealed that the gapB gene is transcribed during gluconeogenesis and repressed during glycolysis. Conversely, gapA transcription is 5-fold higher under glycolytic conditions than during gluconeogenesis. GAPDH activity assays in crude extracts of wild-type and mutant strains confirmed this differential expression pattern at the enzymatic level. Genetic analyses demonstrated that gapA transcription is repressed by the yvbQ (renamed cggR) gene product and indirectly stimulated by CcpA. Thus, the same enzymatic step is catalyzed in B. subtilis by two enzymes specialized, through the regulation of their synthesis and their enzymatic characteristics, either in catabolism (GapA) or in anabolism (GapB). Such a dual enzymatic system for this step of the central carbon metabolism is described for the first time in a nonphotosynthetic eubacterium, but genomic analyses suggest that it could be a widespread feature.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Glycolysis is the main pathway for degradation of carbohydrates and is found in nearly all groups of organisms. The formation of the final product of glycolysis, pyruvate, from glucose is achieved by nine enzymatic steps, most of which function in the reverse direction during gluconeogenesis. The phosphorylating NAD-dependent glyceraldehyde-3-phosphate dehydrogenase (GAPDH)¹ occupies a pivotal role in the Embden-Meyerhoff pathway not only in glycolysis but also in gluconeogenesis because of the reversibility of the oxidation of glyceraldehyde 3-phosphate (G3P) into 1,3-diphosphoglycerate (1,3dPG). In plants, two distinct types of phosphorylating GAPDH co-exist: (i) a strictly NAD-dependent cytoplasmic GAPDH involved in glycolysis and gluconeogenesis; and (ii) a chloroplastic GAPDH, which is involved in photosynthetic CO₂ assimilation and exhibits a dual coenzyme specificity with a preference for NADP (1, 2).

Recently, two gap genes, named gap1 and gap2, have been characterized in the cyanobacterium Synechocystis sp. PPC 6803. The NAD-dependent enzyme Gap1 was reported to be essential for glycolytic glucose breakdown, whereas the enzyme Gap2, which exhibits dual coenzyme specificity, was shown to be operative in the photosynthetic Calvin cycle and in nonphotosynthetic gluconeogenesis (3). Thus, at least in some photo-autotrophic bacterial species (in which the photosynthetic Calvin cycle and glycolysis/gluconeogenesis function in the same cellular compartment, in contrast to what happens in land plants and algae) two distinct GAPDHs, a strictly NAD-dependent one and the photosynthetic one, catalyze the two opposing directions of the reaction.

The primary structure of GAPDH is highly conserved in bacteria and eukarya. Their active form is tetrameric. Different GAPDH functions can be correlated to differences in GAPDH structure, which favor either NAD or NADP binding and then a catabolic rather than an anabolic carbon flow, and vice versa. Indeed, it has been shown from studies combining structural, enzymatic, and protein engineering approaches that both the presence of Asp at position 32 and Leu-Pro at positions 187-188, the latter being located at the subunit interfaces of the tetramer, excluded any activity with NADP⁺, whereas the presence of Gly or Ala at position 32 and Ala-Ser at position 187-188 favored an NADP-dependent activity (4, 5). Removing steric hindrance and/or electrostatic interactions from position 32 or introducing side chains of amino acids at positions 187-188 that stabilize the 2'-ribose phosphate binding can confer specificity for NADP⁺ and thus discriminate against NAD⁺. Thus, the presence of such signatures is a good indicator of efficient NADP⁺ binding and therefore of the involvement of the GAPDH in anabolic carbon flux.

The analysis of the genome sequence of Bacillus subtilis (6) revealed the presence of two putative gap genes, the previously described gap (renamed gapA) and gapB. The deduced amino acid sequences indicate different signatures at positions 32 and 187 (see Fig. 2), which suggested different physiological roles of both GAPDHs. To verify this hypothesis, both genes have been inactivated and the resulting mutant strains were tested for the utilization of various carbon sources. The enzymatic properties of GapB were also studied and compared with those of a mutant GapA from Bacillus stearothermophilus in which the amino acids present in GapB from B. subtilis at positions 32 and 187 were introduced. Taken together, our data reveal that the reversible oxidative phosphorylation of G3P into 1,3dPG is catalyzed by two different proteins in B. subtilis.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Bacterial Strains and Culture Conditions-- The B. subtilis strains used are listed in Table I.

Growth tests were performed at 37 °C with vigourous shaking in MM (7) mineral medium (the strain 168CA is not able to use the citrate present in the MM as carbon source for growth) supplemented with tryptophan (0.005%) and carbon sources (25 mM). Cultures for -galactosidase assays were performed in CQT medium (C mineral medium (8) supplemented with tryptophan 0.005% and glutamine 0.15%) supplemented with 1% glucose or glycerol (v/v) or in CQTHC medium (CQT medium plus casein hydrolysate 0.05%) supplemented with 1% succinate plus 1% glutamate or 1% proline. For GAPDH activity assays in crude extracts, strains were cultivated in LB medium (DIFCO, Detroit, MI) supplemented with proline (25 mM) and IPTG (1 mM) until the middle of the exponential phase of growth. These precultures were diluted 50-fold into LB medium supplemented with IPTG (1 mM) and glucose (25 mM) or proline (25 mM) and incubated at 37 °C until the A₅₉₅ reached 0.5-1.0.

Assay of GAPDH Activity in Cell-free Extracts-- Bacterial cells were harvested by centrifugation at 4 °C and washed twice in 0.2% ice-cold KCl. Cells were then concentrated 50-fold in extraction buffer (45 mM Tris, 15 mM tricarballylic acid, pH 8.2, 20% (v/v) glycerol, 50 mM MgCl₂, 1 mM dithiothreitol) and stored at 20 °C. After the addition of 1 mM phenylmethylsulfonyl fluoride and lysozyme at 100 µg/ml, cell lysis was achieved by incubation at 37 °C during 10 min and then sonication at 4 °C. Cell-free extracts were obtained by centrifugation at 12,000 × g, 4 °C during 10 min. The supernatant was used for the enzymatic assay in reaction buffer at 20 °C (125 mM triethanolamine, 5 mM L-cysteine, 20 mM potassium arsenate, 50 mM K₂HPO₄, pH 9.2) with G3P (4 mM) and NAD⁺ (2 mM) or NADP⁺ (5 mM). Reduction of NAD(P)⁺ was monitored spectrophotometrically at 340 nm. Protein concentration of the extracts were estimated relatively by absortion at 280 nm. Activities were expressed as nmol of NAD(P)⁺ reduced/min and per A₂₈₀.

Plasmid Constructions-- Plasmids pMUT2gapA and pSC11 designed for gapA and yvbQ disruption, respectively, were obtained by cloning an internal fragment of the corresponding ORF synthesized by PCR (oligonucleotides listed in Table II) into the plasmid pMUTIN2mcs (9) between the HindIII and the BamHI sites. The plasmid pMUT2gapB, designed for gapB deletion, was derived from pMUTIN2mcs by inserting two PCR-generated fragments (Table II) corresponding to the DNA regions upstream and downstream from gapB between the HindIII and BamHI sites. Plasmids pSF111 to pSF114 were constructed by insertion of PCR-generated (Table II) EcoRI-HindIII-digested fragments between the EcoRI and HindIII sites of the plasmid pDG1661 (10). Plasmid pSF120, used for complementation by yvbQ, was generated by cloning a yvbQ-containing EcoRI-HindIII PCR fragment (Table II) into pDG1662 (10) linearized by EcoRI and HindIII. For gapB overexpression in Escherichia coli, the gapB coding sequence was amplified by PCR using 5'- and 3'-primers (Table II) that contain a NcoI or a HindIII site, respectively, at their extremity and cloned between the NcoI and HindIII sites of the plasmid pT7-7 (Life Technologies, Inc.) resulting in plasmid pT7GapB1.

-Galactosidase Assay-- -Galactosidase activities were measured using the method of Miller (11) on extracts prepared by lysozyme treatment and centrifugation. Protein concentration was determined using the Bio-Rad protein assay reagent (Bio-Rad). One unit of -galactosidase activity is defined as the amount of enzyme that produces 1 nmol of O-nitrophenol/min at 20 °C/mg of protein.

GAPDH Assays and Kinetics-- Initial rate measurements were carried out at 25 °C on a Kontron Uvikon 933 spectrophotometer by following the absorbance of NAD(P)H at 340 nm. The experimental conditions for the enzymatic assays were 40 mM triethanolamine, pH 9.2, and 10 mM PIPES, pH 7, for the forward and the reverse reactions, respectively. The turnover number (k_cat) was calculated using a molar extinction coefficient at 280 nm of 1.53 × 10⁵ M¹·cm¹ and 1.17.10⁵ M¹ cm for the apoenzymes of B. subtilis and B. stearothermophilus, respectively. The k_cat is expressed per site (N). The initial rate data of the forward reaction were fitted to the Michaelis-Menten relationship using least squares analysis to determine k_cat and K_m. All cofactor K_m values were determined at saturating concentration of G3P and P_i.

Production and Purification of the Mutant GAPDHs of B. stearothermophilus-- Site-directed mutageneses (12) were performed on a pBluescriptII-derived plasmid containing the gapA gene of B. stearothermophilus. Purification of mutant GAPDHs was performed as described previously for other B. stearothermophilus mutants (13). All the mutants were isolated as apo form, as judged by the ratio A₂₈₀/A₂₆₀ of 2.

Production and Purification of GapB of B. subtilis-- The enzyme was produced in E. coli strain BL21(DE3)pLysS transformed with plasmid pT7GapB1. The overexpression of GapB was performed by the addition of 0.4 mM IPTG in the culture medium at 0.8 A₆₀₀. After 3 h of induction, cultures were harvested by centrifugation and resuspended in buffer A (50 mM phosphate, pH 8.2) containing 5 mM -mercaptoethanol. After sonication, B. subtilis GapB was purified by ammonium sulfate fractionation (50-70%) and exclusion size chromatography on ACA 34 resin at pH 8.2 (buffer A). Purified fractions were then pooled and applied to a Q-Sepharose column equilibrated with buffer A followed by a linear gradient of KCl (0-1 M) using a fast protein liquid chromatography system (Amersham Pharmacia Biotech). The B. subtilis GapB was eluted at 300 mM KCl. At this stage, the protein was pure as revealed by electrophoresis on 10% SDS-polyacrylamide gel followed by Coomassie Blue staining. The subunit molecular weight of GapB was verified by mass spectrometry. The protein was isolated as apo form, as judged by the ratio A₂₈₀/A₂₆₀ of 1.8. GapB concentration was estimated spectrophotometrically using an extinction coefficient of 1.53 × 10³ M¹ cm¹, as deduced from the Scopes method (14). Purified enzyme was stored at 4 °C in the presence of 1 mM -mercaptoethanol and of 70% ammonium sulfate.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

gapA and gapB Mutants Have Opposite Phenotypes-- To investigate the physiological roles of the two B. subtilis GAPDH-encoding genes, we constructed the corresponding gapA and gapB mutant strains, GM1501 and GM1500, respectively. The pMUTIN2mcs-derived plasmids (9) used for these constructions allowed, simultaneously with the disruption or the deletion of the target gene, (i) the creation of a transcriptional fusion of the 5'-part of the target gene to the lacZ gene; and (ii) the control of the expression of downstream gene(s) putatively belonging to the same operon via the IPTG-inducible Pspac promoter, to reduce possible polar effects (Fig. 1, A and C). The GM1501 gapA insertional mutant grew as well as the parental strain on LB only when IPTG was added. This suggested that the downstream glycolytic genes, pgk, tpi, pgm, and eno, are significantly transcribed from a promoter located upstream of gapA. Conversely, the growth rate of the GM1500 gapB mutant strain on LB was not affected by the presence of IPTG.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   A, schematic representation of the gapA disruption in the strain GM1501. B, growth curves of the GM1501 strain in MM glucose (open triangles) and in MM asparagine (open circles) compared with those of the wild-type strain in the same media (closed triangles and circles, respectively). C, schematic representation of the gapB deletion in the strain GM1500. D, growth curves of the GM1500 strain in MM glucose (open triangles) and in MM asparagine (open circles) compared with those of the wild-type strain in the same media (closed triangles and circles, respectively).

Growth tests of both mutant strains, GM1500 and GM1501, revealed opposite phenotypes as shown in Fig. 1, B and D. GM1500 could not grow in MM medium containing asparagine as a sole carbon source but presented the same growth rate as the 168CA parent in glucose-containing MM medium. The phenotype of GM1500 was identical when IPTG was added to the growth medium (data not shown), indicating that only the gapB inactivation was responsible for the growth defect. The opposite phenotype was observed for the gapA mutant strain, GM1501 (strongly reduced growth in glucose medium and normal growth in asparagine medium (both evaluated in the presence of IPTG)). Similar results were obtained when glycerol or glucitol were used instead of glucose, and when proline or succinate plus glutamate was used instead of asparagine (data not shown). Thus gapA is required for glycolysis, whereas gapB is required for gluconeogenesis.

GAPDH Activity in gapA and gapB Mutant Strains-- Both mutant and wild-type strains were cultivated in LB medium supplemented with glucose, representing glycolytic physiological conditions, and LB medium supplemented with proline, representing gluconeogenic conditions, to ensure significant growth of the three strains in both media. GAPDH activity was measured by the oxydation of G3P to 1,3dPG using either NAD⁺ or NADP⁺ as coenzyme. Extracts from the 168CA strain showed roughly 5 times higher NAD-dependent activity under glycolytic conditions than under gluconeogenic conditions (Table III). NADP-dependent activity could be detected only under gluconeogenic growth conditions. This latter activity completely disappeared in the GM1500 (gapB) strain, and no NAD-dependent GAPDH-activity could be detected in the GM1501 (gapA) strain. These results suggest that GapA and GapB proteins have different cofactor specificities, NAD⁺ for GapA and NADP⁺ for GapB. In glucose-containing medium the gapA encoded NAD-dependent activity is higher than in proline medium, whereas NADP-dependent GAPDH activity probably encoded by gapB can only be detected under gluconeogenic growth conditions.

GapA and GapB Exhibit Different Cofactor Specificity Sequence Signatures-- Amino acid sequences of bacterial GAPDHs were aligned (Fig. 2). Based on the known cofactor and substrate signatures, three types of phosphorylating GAPDHs can be distinguished. The first two types, named GapA and erythrose-4-phosphate dehydrogenase, possess a typical NAD⁺ signature with Asp or Glu at position 32 and amino acids at position 187-188, which are known to prevent an efficient binding of NADP⁺ (4, 5). But they differ in substrate specificity, with typical signatures at position 179 and around position 208, which favor the binding of either G3P for GapA or erythrose 4-phosphate for erythrose-4-phosphate dehydrogenase (15). The third type, named GapB, which is G3P-specific (3), contains a cofactor signature that supports a dual coenzyme specificity with a preference for NADP⁺ (5). Indeed, this signature either resembles that of chloroplastic GAPDH, with Asp at position 32 but with amino acids at position 187-188 that stabilize NADP⁺ binding, or presents no Asp at position 32 but amino acids at positions 32 and/or 187-188, which could stabilize NADP⁺ binding.

View larger version (54K): [in this window] [in a new window]   Fig. 2.   Alignment of the sequence signatures of GAPDH deduced from complete or unfinished genome sequences relevant for cofactor and substrate specificity predictions. Only bacterial species that possess at least two GAPDHs were considered (the GapA sequence of Yersinia pestis being not completed is not indicated), except for the GapA reference sequence of B. stearothermophilus. The consensus sequence of chloroplastic GAPDHs is also indicated. Source of sequences: GAPDH amino acid sequences of B. stearothermophilus, E. coli, H. pylori, Salmonella typhimurium, Anabaena variabilis, and Synechocystis were obtained from Swissprot; those of Synechococcus and Gloeobacter violaceus were from GenBankTM/EBI; those of B. subtilis, N. gonorrhoeae, N. meningitidis, Pseudomonas aeroginosa, Vibrio cholerae, and Y. pestis were from PEDANT directly; and that of Shewanella putrefaciens was from TIGR. E4PDH, erythrose 4-phosphate dehydrogenase.

In this context, GapA from B. subtilis has the typical signature of an NAD-dependent GAPDH with Asp-32, Leu-187, and Pro-188. This signature and its overall 85% sequence identity with the B. stearothermophilus GapA suggest that GapA from B. subtilis has enzymatic characteristics very similar to those of the well studied GapA from B. stearothermophilus (5, 16). In contrast, the B. subtilis GapB, with Ala and Asn at position 32 and 187, respectively, probably presents a dual cofactor signature.

Purification and Enzymatic Characterization of GapB-- The B. subtilis GapB protein was overproduced in E. coli, purified, and separated from the E. coli GapA protein by taking advantage of its higher hydrophobicity and its lower isoelectric point. GapB showed activity with G3P in the forward direction but only in the presence of P_i, whereas 1,3dPG was the substrate in the reverse direction (data not shown). This demonstrated that the GAPDH activity was of the phosphorylating type. The kinetic parameters determined for GapB, in the forward direction (k_cat and K_m for cofactors), are summarized in Table IV. The apparent affinity is about 7-fold higher for NADP⁺ than for NAD⁺ with a 50-fold higher catalytic efficiency. These results demonstrated that B. subtilis GapB has a strong preference for NADP⁺ as a cofactor. Mutations at the two signature positions, 32 and 187, in the B. stearothermophilus GapA protein were shown to increase NADP-dependent GAPDH activity and to decrease NAD-dependent GAPDH activity (Table IV). Whereas the wild-type B. stearothermophilus GapA protein had no detectable activity with NADP⁺ as cofactor, the D32A/L187N double mutant GapA protein clearly showed a catalytic efficiency with NADP⁺ similar to that observed with the B. subtilis GapB protein and higher than that measured with NAD⁺.

The gapA and gapB Genes Are Transcriptionally Regulated-- As the levels of GAPDH activities encoded by gapA and gapB are modulated in response to the growth conditions, we looked for a transcriptional regulation of the expression of these genes. A series of transcriptional fusions to the lacZ gene were constructed. As gapA is predicted to be the second gene of an operon, three different reporter constructs have been designed (Fig. 3A) to analyze its transcription pattern and to test whether its expression is driven from the region located immediately upstream of the gapA ORF or from the region located between the araE terminator and the initiation codon of the yvbQ ORF, the first gene of the putative gapA operon. These fusions, carried by plasmids pSF111 to pSF113, as well as the fusion of the gapB promoter region with lacZ carried by the plasmid pSF114 (Fig. 3), were integrated as single copy at the amyE locus of B. subtilis 168CA strain, and the -galactosidase activities of the resulting strains, GM1511 to GM1514, respectively, were measured during growth in glucose or succinate plus glutamate minimal medium.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   A, schematic representation of the DNA fragments (striped bars) transcriptionnally fused to lacZ in the pSF111 to pSF114 series of plasmids used to generate the GM1511 to GM1514 series of reporter strains. B, growth curve (closed circles) and -galactosidase synthesis (open triangles) of GM1511 and GM1514 strains in minimal CQT glucose medium or in minimal CQTHC succinate plus glutamate medium. -Galactosidase activities are measured as described under "Experimental Procedures."

In GM1511, lacZ expression was found to be roughly constant during the exponential and the beginning of the stationary phases of growth, but 4-fold higher in glucose than in succinate plus glutamate medium (Fig. 3B). Similar patterns of lacZ expression were found in GM1512 with a 4-5-fold stimulation of the expression in glucose medium (data not shown). On the other hand, no expression was detected in GM1513 (data not shown). These results indicated that yvbQ and gapA form an operon and that gapA transcription is driven from a promoter located between araE and yvbQ. The -galactosidase synthesis in GM1514, reflecting gapB expression, remained also roughly constant through the exponential and the beginning of the stationary phase of growth but was at least 50-fold higher in succinate plus glutamate than in glucose medium. Very similar patterns of expression were found when glycerol and proline were used instead of glucose and succinate plus glutamate, respectively (data not shown). Thus, we concluded that gapA and gapB expressions are transcriptionally regulated in an opposite manner; under glycolytic conditions, gapA expression is stimulated, whereas that of gapB is nearly completely repressed. Conversely, during gluconeogenesis, gapB transcription is strongly activated, whereas gapA is expressed at a basal level.

yvbQ Represses gapA Expression during Gluconeogenesis-- Searches for similarities revealed that the deduced product of yvbQ, located upstream of gapA, belongs to the SorC/DeoR family of transcriptional regulators (data not shown). Its similarity is particularly strong with three ORFs, Urf1 (17), YgaP (18), and ORF1' (19), linked to gap genes in Bacillus megaterium, Lactobacillus delbrueckii, and Clostridium acetobutylicum, respectively. The functions of these ORFs have not yet been identified. To test a possible role of yvbQ in the regulation of gapA transcription, we constructed the strain BFS1080, which carries a transcriptional fusion between the promoter region of the yvbQ-gapA operon and the lacZ gene, and is disrupted for yvbQ. Then, either the pSF120 plasmid, a pDG1662-derived vector expressing the wild-type yvbQ gene, or the parental pDG1662 plasmid was integrated into BFS1080 at the amyE locus creating the GM1520 or the GM1521 strains, respectively.

Although lacZ was expressed constitutively at a high level in GM1521 (yvbQ), it was expressed at a low level in the presence of succinate plus glutamate and 6-fold stimulated in the presence of glucose in GM1520 (yvbQ⁺) (Table V). A similar stimulation of lacZ expression was observed when the strain GM1520 was grown in minimal medium containing both glucose and succinate plus glutamate. The lacZ expression pattern in GM1520 and GM1521 remained the same when glycerol was used in place of glucose or proline in place of succinate plus glutamate (Table V). These results demonstrated that the yvbQ gene product is a repressor of the yvbQ-gapA operon and that its activity is inhibited by the presence of glycolytic carbon sources.

An Indirect Role of CcpA in Regulation of gapA Expression-- CcpA is a key regulator of carbon flow in B. subtilis, acting both as a negative regulator of numerous carbon utilization genes and as a positive regulator of genes involved in the excretion of byproducts when rapidly metabolizable carbon sources are in excess (20, 21). To investigate a possible role of ccpA in the regulation of the gapA operon, a ccpA deletion was introduced in the GM1520 (yvbQ⁺) and GM1521 (yvbQ) strains, respectively, and -galactosidase activity in the resulting strains, GM1530 and GM1531 was measured. The stimulation of lacZ expression in the presence of glucose observed in GM1520 was not found in its ccpA derivative, GM1530 (Table V). However, using glycerol as sole carbon source, no significant difference could be detected between both strains. The expression of the lacZ reporter gene was found to be high and constitutive in all the media tested in GM1531 (yvbQ, ccpA) as it is in GM1521 (yvbQ), demonstrating an epistatic effect of the yvbQ mutation over the deletion of the ccpA gene. Therefore, ccpA is required for stimulating the transcription of the gapA operon by glucose but not by glycerol. Moreover, this role of ccpA is indirect and probably mediated by YvbQ.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The complete genome sequence of B. subtilis revealed the co-existence of two putative GAPDH encoding genes, gapA (formerly gap, (22)) and gapB (6). To study the physiological role of these genes, we inactivated both of them independently. The phenotypical analysis of the mutants revealed their opposite physiological functions; gapA is required for glycolysis, whereas gapB is necessary for gluconeogenesis (Fig. 1). Enzymatic assays on crude extracts of the gapA and gapB mutants (Table III) demonstrated that gapA encodes a NAD-dependent GAPDH, whereas gapB encodes a NADP-dependent GAPDH. Moreover these experiments suggested that their synthesis and/or their activity are modulated in response to the physiological conditions.

Indeed, we were able to show that transcription of the two gap genes is inversely regulated. Under glycolytic growth conditions, gapA expression is stimulated 4-5-fold, although that of gapB is nearly completely switched off (Fig. 3). By contrast, growth on Krebs cycle intermediates or amino acids strongly induces gapB expression and maintains gapA transcription at a basal level. Transcriptional regulation of both genes estimated through the expression of lacZ reporter gene fusions at an ectopic locus accounts for the differences of GAPDH activity measured in extracts from cultures grown under glycolytic or gluconeogenic conditions (Table III), suggesting that the regulation mainly acts at the transcriptional level. Our data indicate that gapA and yvbQ are cotranscribed from a promoter located upstream of yvbQ (Fig. 3). The pgk-eno operon starts 320 base pairs downstream of gapA. The gapA gene is frequently clustered with other glycolytic genes in bacteria (17, 18, 23), but it is also encountered as an individual gene or transcription unit, e.g. in E. coli (24), in Synechocystis (25), and in Streptomyces aureofaciens (26). In B. subtilis, the polar effect of the gapA disruption (Fig. 1) suggests that significant transcription of the pgk-eno genes may initiate upstream of yvbQ despite the presence of a predicted transcription terminator just downstream of gapA. Indeed, our preliminary transcription analysis² of this region suggested the existence of a transcript corresponding to the pgk-eno operon and of a longer one extending from yvbQ to eno. This cotranscription would ensure the coordinated expression of all central glycolytic enzymes under glycolytic growth conditions.

Our genetic analyses demonstrated that the first gene of the operon, yvbQ, regulates glucose stimulation of gapA (and possibly of the downstream genes). The yvbQ gene product acts as a repressor, the activity of which would be inhibited under glycolytic growth conditions (Table V). We propose to rename this gene cggR for central glycolytic gene Regulator. CggR belongs to the SorC/DeoR family of transcriptional regulators, and homologues of unknown function of this protein are found upstream of gap genes in B. megaterium, C. acetobutylicum, or L. delbrueckii. It is therefore probable that also in these latter organisms gapA expression is regulated via the cggR-like genes. We also found that the general carbon catabolite repressor protein CcpA is necessary for the stimulation of gapA transcription by glucose (Table V), in agreement with the results of Tobisch et al. (27). However, our genetic analysis suggested that CcpA indirectly regulates gapA transcription; the signal recognized by CggR would be altered in the ccpA mutant context when glucose, but not glycerol, is the sole carbon source utilized for growth.

We are currently investigating the mechanisms involved in the regulation of gapB expression. A ccpA deletion only partially abolished the repression of gapB transcription in the presence of glucose.³ Thus an additional CcpA-independent mechanism for carbon catabolite repression must be considered for gapB regulation.

Our results clearly showed that purified GapB presents a strong preference and a better catalytic efficiency for NADP⁺ compared with NAD⁺. Introducing the predicted signature responsible of the NADP⁺ binding preference into a strictly NAD⁺-dependent GAPDH converts the latter into a dual cofactor enzyme with a marked preference for NADP⁺ (Table IV). Interestingly, the protein signature involved in the NADP⁺ preference differs from that of the chloroplastic GAPDHs (which also present dual coenzyme specificity with a preference for NADP⁺) with Ala instead of Asp at position 32 and Asn instead of Leu at position 187 (Fig. 2).

Catabolic reactions lead to the formation of reduced coenzymes, which are reoxidized by the oxidative phosphorylation in combination with ATP-production. Only NADH enters the oxidative phosphorylation pathway. The oxidized form of NAD⁺ appears to be predominant over NADH by 50-1000-fold in the bacterial cell depending on the environmental redox state (28-30), whereas NADPH appears to be more abundant than NADP⁺ (31). Redox potentials of both co-enzymes are similar. As expected, NADPH is indeed the preferred cofactor of those enzymes catalyzing anabolic reductions. From this point of view, our discovery of a NADP-dependent GAPDH specialized for the gluconeogenesis in B. subtilis is not surprising, especially when one considers that the catalysis of the reduction of 1,3dPG into G3P by the canonical NAD-dependent GAPDH is thermodynamically not favorable because of the in vivo NADH/NAD ratio (29).

In many eukaryotic organisms, more than a single GAPDH has been identified, e.g. TDH1-TDH3 in Saccharomyces cerevisiae. In plants, two types of GAPDHs coexist with distinct physiological roles and distinct cellular localizations (see Introduction). Among bacteria, four cyanobacterial species, which are photosynthetic bacteria, possess at least two putative GAPDHs with different cofactor specificity (Fig. 2). In one of these cyanobacterium species, Synechocystis, distinct roles have indeed been demonstrated for these GAPDH. The NAD⁺-dependent GAPDH is required for glycolysis, whereas the dual NADP⁺/NAD⁺-dependent GAPDH, operative in the anabolic Calvin cycle, is also active during gluconeogenesis (3).

Different situations have been described in Archaea; in the hyperthermophilic Pyrococcus furiosus and Thermotoga tenax, the GAPDH (NADPH-dependent) enzyme appears to be active mainly during gluconeogenesis, whereas either glyceraldehyde-3-phosphate ferredoxin oxidoreductase or nonphosphorylating glyceraldehyde-3-phosphate dehydrogenase enzymes catalyze in one step the unidirectional glycolytic conversion of G3P into 3-phosphoglycerate (32-34). With the current knowledge, a thermodynamic advantage of such a variation of the Embden-Meyerhoff pathway is not obvious (34). Apparently, Archaea possess an energetically less efficient glycolysis variant but in which the interconversion of G3P and 3-phosphoglycerate developed into a novel regulatory site of the glucose metabolism.

Up to now, GAPDHs of nonphotosynthetic eubacteria were supposed to have dual functions in catalyzing the reaction in both directions. The situation in B. subtilis that we describe in this report is the first example among nonphotosynthetic bacteria of the coexistence of a NAD-dependent and of a NADPH-dependent GAPDH specialized in glycolysis and in gluconeogenesis, respectively. This situation preserves the high energetic efficiency of the Embden-Meyerhoff glycolysis but also gives rise to a fourth regulatory checkpoint for the central carbon metabolism.

Presently, three other nonphotosynthetic bacterial species, Nesseiria meningitidis, Nesseiria gonorrhoae, and Helicobacter pylori, can be predicted on the basis of genome sequence analyses (Fig. 2) to possess like B. subtilis a NADP-dependent GAPDH in addition to a canonical NAD-dependent GAPDH; they probably share the GAPDH specialization identified in B. subtilis.

	ACKNOWLEDGEMENTS

We thank Evelyne Habermacher for technical help with site-directed mutagenesis and Andrée Lepingle for DNA sequencing. We thank Maryvonne Arnaud and Georges Rapoport for their kind permission to cite unpublished results. We thank Nic Lindley, Richard d'Ari, Josef Deutscher, Dominique Le Coq, Christophe d'Enfert, and Claude Gaillardin for helpful comments on the manuscript.

	FOOTNOTES

^* This work was supported by the EU Biotechnology Program Grant BIO-4CT95-0278.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ Present address: Unité de Physiologie Cellulaire, Institut Pasteur, 25-28, rue du Docteur Roux, 75724 Paris cedex 15, France.

^** To whom correspondence should be addressed. Tel.: 33 (0)1 30 81 54 49; Fax: 33 (0)1 30 81 54 57; E-mail: stef@platon.grignon.inra.fr.

² S. Fillinger and S. Aymerich, unpublished observation.

³ M. Arnaud and G. Rapoport, personal communication.

	ABBREVIATIONS

The abbreviations used are: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; G3P, glyceraldehyde 3-phosphate; 1, 3dPG, 1,3-diphosphoglycerate; IPTG, isopropyl-1-thio--D-galactopyranoside; ORF, open reading frame; PCR, polymerase chain reaction; PIPES, 1,4-piperazinediethanesulfonic acid.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES