INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The -keto acid dehydrogenases constitute a family of multienzyme complexes which consists of the pyruvate, 2-oxoglutarate, acetoin, and branched-chain -keto acid dehydrogenase (BKDH) complexes. They each consist of three enzymes that together catalyze the oxidative decarboxylation of -keto acids with the concomitant reduction of NAD⁺ and the formation of the coenzyme A (CoA) adduct of the substrate. In Enterococcus faecalis a pyruvate dehydrogenase (PDH) complex and a BKDH complex have been identified. These enterococci lack the 2-oxoglutarate dehydrogenase complex (5), and the only evidence for an acetoin dehydrogenase complex is an acetoin dehydrogenase activity detected in crude extracts (9). The PDH complex of E. faecalis has been characterized in the purified form and in physiological studies and to some extent genetically (1, 22, 23). In this organism, which lacks a respiratory chain, the enzyme complex is important in the conversion of pyruvate to acetate, which yields an additional ATP, in comparison to lactate production. The conversion of the resulting acetyl-CoA to acetate and ATP is carried out by the combined actions of the phosphotransacetylase and acetate kinase.

The BKDH complex has been studied much less extensively. The BKDH complex purified from E. faecalis 10C1 catalyzed the oxidative decarboxylation of the branched-chain -keto acids -ketoisovalerate (KIV), -ketoisocaproate (KIC), and -keto--methylvalerate KMV (20). The reaction was dependent on NAD⁺, CoASH, and lipoic acid, reminiscent of the requirements of the PDH complex. The metabolic roles of the BKDH complex are diverse. BKDH complexes have been found to play roles in ATP generation in Pseudomonas putida (24), in the production of branched-chain fatty acids for membrane biosynthesis in Bacillus subtilis (28), in cell-cell signaling in Myxococcus xanthus (26), and in avermectin biosynthesis in Streptomyces avermitilis (6).

Recently, the E. faecalis gene cluster ptb-buk-bkdDABC was found to encode the E1 (bkdA), E1 (bkdB), E2 (bkdC), and E3 (bkdD) components of the BKDH complex (29). The activities of the enzymes encoded by the ptb-buk-bkdDABC gene cluster and the fact that they are coexpressed suggest that the enzymes constitute a functional pathway (29). We hypothesize that the E. faecalis bkd gene cluster is involved in the catabolism of the branched-chain -keto acids and that it generates ATP via substrate level phosphorylation in a system analogous to that of the PDH complex, phosphotransacetylase, and acetate kinase. A putative catabolite responsive element (CRE) was identified in the promoter region, suggesting a role of catabolite repression (CR) in the regulation of the bkd gene cluster. CR is a mechanism by which the expression of genes involved in the metabolism of a growth substrate is inhibited by the presence of a more readily metabolizable carbon source such as glucose (4). CR in B. subtilis and other gram-positive organisms utilizes a mechanism different from that in Escherichia coli. In B. subtilis it involves the cis-acting CRE sequence, catabolite control protein A (CcpA), and the regulatory form of HPr [HPr(ser-P)] of the phosphotransferase system (PTS) (21). Formation of HPr(ser-P) is stimulated by the glycolytic intermediate fructose-1,6-bisphosphate and is catalyzed by the Hpr kinase (12, 14).

Here we test these hypotheses. We confirm that the bkd gene products constitute a pathway for the breakdown of the branched-chain -keto acids, which results in the formation of ATP. In addition, it is shown that the branched-chain -keto acids and CR play a role in the regulation. We also identified the formation of intermediate 1,1-dihydroxy-4-methyl-2 pentanone, which acts as a temporary redox sink during the oxidative catabolism of KIC, one of the branched-chain -keto acids.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Growth of bacteria. E. coli was grown as previously described (29). E. faecalis was grown on either M17 medium (25) or, for the physiological studies, Evans medium (11). Evans medium is a minimal-salt medium in which a phosphate buffer was used (50 mM; pH 7.0) and which was supplemented with 0.5% (wt/vol) yeast extract. E. faecalis was grown in batch cultures either in Erlenmeyer flasks for the aerobic experiments or in controlled fermentors for the anaerobic experiments. In these experiments cultures were inoculated at an initial optical density at 600 nm (OD₆₀₀) of 0.01 and were monitored over a 24-h period with samples taken every hour.

DNA manipulation and amplification. Recombinant DNA procedures including plasmid and total DNA isolations, DNA-DNA hybridizations, and molecular cloning were performed essentially as described previously (29). For purification of plasmid DNA from gram-positive organisms the method of Anderson and McKay was employed (2).

RNA isolation and primer extension. Total RNA was isolated essentially as described previously (29). E. faecalis was grown in M17 medium containing 20 mM branched-chain amino acids in the presence or absence of glucose (25 mM). Cells were harvested during exponential growth at an OD₆₀₀ between 0.6 and 0.7. The 5' end of the bkd transcript was mapped using the oligonucleotide LipDH15 (5'-GTGAACCTCCTGCAATTGAAAC-3'), which was end labeled and used in primer extension reactions with 15 µg of total RNA. The reverse transcription products were resolved on a denaturing 6% polyacrylamide gel, using a DNA sequencing reaction generated with LipDH15 as the size standard.

Construction of a gusA promoter probe vector for E. faecalis. Plasmid pNZ273 is a broad-host-range plasmid that contains the replicon of pSH71 and the E. coli gusA gene encoding -glucuronidase (19). While pNZ273 was being used, it was discovered that a deletion event was occurring within the plasmid in E. faecalis. For this reason pNZ273 was abandoned as a reporter construct in E. faecalis. Plasmid pDL278 is also a broad-host-range vector that contains the pVA380-1 basic replicon and the pUC origin of replication and that replicates stably in enterococci (16). To generate a gusA reporter construct for E. faecalis, pNZ273 was cut with EcoRI, made blunt with Klenow polymerase, and religated upon itself, effectively removing the EcoRI site and generating pNZ273e. The gusA gene was excised from pNZ273e with BamHI and HindIII, and this 2.0-kb fragment was then ligated into similarly cut pDL278. The resulting plasmid, designated pDW100, contains six unique restriction sites upstream of the gusA gene.

View larger version (67K): [in this window] [in a new window]   FIG. 1.   (A) Identification of the transcriptional start site for the bkd operon. E. faecalis 10C1 was grown in M17 medium containing 20 mM concentrations of each of the branched-chain amino acids in the presence (+ Gluc) or absence ( Gluc) of glucose (25 mM). Total RNA was isolated from mid-log-phase cells, and 15 µg was used for each of the primer extension reactions. A sequencing reaction using the identical oligonucleotide was run in parallel. (B) Sequence of the 390-bp fragment, containing the bkd promoter, cloned into pDW100. +1, transcriptional start site as determined by primer extension analysis. The promoter 10 and 35 sites are in boldface. S/D, putative ribosome binding site. The CRE is underlined.

Assay for GUS. For the determination of -glucuronidase (GUS) activity, 10 ml of cells was harvested by centrifugation and resuspended in 1 ml of GUS assay buffer (50 mM NaH₂PO₄ [pH 7.0], 10 mM -mercaptoethanol, 1 mM EDTA, 0.1% Triton X-100). For the induction and repression studies cells were harvested during exponential growth at an OD₆₀₀ of 0.6 to 0.7. The cells were then disrupted by bead beating (1 min of beating followed by 1 min on ice, repeated three times). After centrifugation (15,000 × g for 15 min) the cell extract was assayed for activity. For the determination of GUS activity, 100 µl of cell extract was added to 1 ml of GUS assay buffer containing 1.25 mM p-nitro--D-glucuronic acid (X-Gluc; Sigma). The reaction was carried out at 37°C, and the increase in absorbance at 405 nm was measured.

Mass spectroscopy. Mass analyses were done on a Micromass Quattro triple-quadrupole mass spectrometer, with electrospray ionization in the negative mode. The samples were dissolved in acetonitrile-water (50/50) and directly infused into the ionization source with a Harvard apparatus syringe pump 22 at 5 µl/min. The capillary voltage was 3.5 kV, and the cone voltage was 50 V. The source temperature was 80°C. For fragmentation of KIC and intermediate X, the parent ion was selected with the first analyzer and fragmented with collisionally induced dissociation by introducing argon into the collision cell at a pressure of 0.28 Pa and by applying a collision energy of 30 eV. The fragments were detected by scanning the second analyzer from an m/z of 10 to 150 at 1 s/scan. Data were collected in the MCA mode. The instrument was calibrated across the acquisition range with the negative ions and clusters formed from sodium iodide.

Product analysis. In the physiological studies, 1-ml samples were taken at hourly intervals from the incubation mixtures and the cells were removed by centrifugation. To remove any protein present in the sample, 100 µl of 35% perchloric acid was added to the supernatant and the mixture was incubated on ice for 10 min, followed by the addition of 55 µl of 7 N KOH. The resulting precipitate was removed by centrifugation, and product concentrations in the supernatant were determined using an Aminex HPX87H organic acid analysis high-pressure liquid chromatography (HPLC) column as described by Snoep et al. (22). All samples were analyzed for pyruvate, glucose, acetate, formate, ethanol, lactate, fumarate, succinate, KIC, isovaleric acid, and intermediate X. On the basis of the product analysis carbon and redox balances could be calculated. In all experiments these balances were closed (i.e., from 90 to 110% recovery). The response factor of intermediate X was estimated on the basis of an assumed constant total concentration of KIC plus isovaleric acid plus intermediate X. The same response coefficient was used throughout the study. Samples for ATP analysis were mixed with 80°C phenol, and ATP was measured using a luciferin-luciferase ATP monitoring kit (LKB), essentially according to the manufacturer's recommendations. This method is described in more detail in reference 27.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

In an earlier study we demonstrated that the bkd gene cluster encodes the enzymes for the BKDH complex and two additional enzymes, which we identified as a branched-chain phosphotransacylase (Bct) and branched-chain acyl kinase (Bck). This homology led us to the suggestion that the branched-chain -keto acids are catabolized via a pathway similar to that for the aerobic breakdown of pyruvate to acetate (PDH complex, phosphotransacetylase, and acetate kinase). We now focused on the regulation of the gene cluster and on the metabolic role of this proposed pathway in E. faecalis.

Transcriptional analysis of the bkd gene cluster. A single, apparent mRNA 5' end was identified upstream of ptb by primer extension analysis (Fig. 1). Analysis of the upstream sequence identified the features common to the majority of lactic acid bacterial promoters identified, including the presence of 10 and 35 sites, a spacing of 16 to 18 bp between those sites, and the presence of an AT-rich region upstream of the 35 site (7). The transcriptional start site is a pyrimidine base rather than the more common purine (8). The CRE sequence (TGTATGCGCTTACA; nucleotides 752 to 765) was found to overlap the 35 region of the promoter and is identical, with the exception of 1 base, to the 14-bp consensus (TGWNANCGNTNWCA) (21). The relative amount of the primer extension product from cells grown in the presence of glucose was significantly lower than that from cells grown in the absence of glucose (Fig. 1). This suggests that bkd expression is repressed by glucose, which is consistent with a role for CR in the regulation.

View this table: [in this window] [in a new window]   TABLE 2.   GUS activities for E. faecalis OG1RF(pDW101) grown aerobically in the presence of the branched-chain amino acids and branched-chain -keto acids

Biomass increase of E. faecalis OG1RF due to KIC. We had shown previously that the -keto acids KIC, KIV, and KMV were utilized by E. faecalis and were converted to their corresponding free acids isovalerate, isobutyrate, and methylbutyrate and that this process is dependent on the bkd gene cluster (29). We now studied this process in greater detail. Since the presence of glucose was inhibitory to bkd expression, we studied the physiological role of the bkd gene cluster by using pyruvate as the primary free-energy source. E. faecalis OG1RF was grown aerobically in batch cultures (pH 7.0) with pyruvate (20 mM) as the free-energy source on minimal medium, phosphate buffered, supplemented with 0.5% yeast extract in the absence or presence of KIC (20 mM). In the absence of KIC growth was exponential, with a specific growth rate (µ) of 1.06 ± 0.01 h¹, until the cells ran out of pyruvate after approximately 5 h (Fig. 2). In the presence of KIC (20 mM) growth was slower (µ = 0.83 ± 0.02 h¹) but continued after the cells ran out of pyruvate (after approximately 8 h) and a higher final OD was reached (Fig. 2). To evaluate whether the BKDH complex is essential for KIC catabolism, the same growth experiments were also performed using OGBKD1 and OGBKD2, strains with mutated BkdA and BkdC, respectively (Table 1) (29). In the absence of KIC no difference between OG1RF and the mutant strains, both in final OD and growth rate (µ = 1.07 ± 0.02 h¹; Fig. 2) was observed. For both mutant strains very similar results were obtained, and for reasons of clarity only the results for OGBKD1 are shown. In the presence of KIC a decrease in growth rate was observed (µ = 0.49 ± 0.02 h¹) and the mutant strains did not reach an OD as high as that observed for the wild type (Fig. 2). The extent of growth inhibition was very dependent on the culture conditions.

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Aerobic growth of E. faecalis OG1RF and OGBKD1 with pyruvate as the primary free-energy source in the absence or presence of KIC. Solid symbols, OD600; open symbols, pyruvate concentration; squares, E. faecalis OG1RF without KIC; diamonds, OG1RF with KIC; triangles, OGBKD1 without KIC; circles, OGBKD1 with KIC. Results obtained with OGBKD2 were identical to those obtained with OGBKD1 (data not shown). Cells were grown in a phosphate-buffered minimal medium supplemented with 0.5% yeast extract, with pyruvate as the primary free-energy source. For precise experimental conditions see Materials and Methods.

KIC catabolism by E. faecalis. Product pattern analysis by HPLC showed that KIC was quantitatively converted into isovaleric acid (Fig. 3). However, during this conversion an (extracellular) intermediate, which we call intermediate X, was also observed. In an effort to identify intermediate X, it was partly purified via HPLC and both KIC [CH₃CH(CH₃)CH₂COCOOH] and intermediate X were subjected to negative-ionization electrospray mass spectroscopy. Anions of KIC and intermediate X were observed at m/z values of 129 and 131, respectively. For KIC, this agrees with the value for the deprotonated anion [M-H], as expected. Fragmentation of KIC resulted in the formation of two fragment anions at m/z values of 85 and 57, which can be explained as the decarboxylated fragment anion [CH₃CH(CH₃)CH₂CO] (CO₂ is not charged and is thus "invisible") and the further dissociation of the keto group to the anion [CH₃CH(CH₃)CH₂], respectively. Fragmentation of the intermediate resulted in the same m/z 85 fragment anion, indicating no changes in the C-2-to-C-5 part of the molecule. However the m/z 57 fragment anion was not evident, and a new fragment anion at an m/z of 45, which is interpreted as being the formate anion ([HCOO]), was observed. Taken together, the difference in m/z of 2 mass units between KIC and the intermediate molecular anions, the unchanged C-2-to-C-5 part, and the appearance of what appears to be a formate anion led us to identify intermediate X tentatively as 1,1-dihydroxy-4-methyl-2-pentanone (DMP). Importantly, conversion of KIC to the intermediate appears to involve a reduction. Distinct extracellular intermediates were also observed during the utilization of KIV and KMV by E. faecalis.

View larger version (12K): [in this window] [in a new window]   FIG. 3.   KIC consumption and isovaleric acid production in E. faecalis OG1RF grown on pyruvate under aerobic conditions. Cells were grown in a phosphate-buffered minimal medium supplemented with 0.5% yeast extract, with pyruvate as the primary free-energy source. During the batch fermentation samples were taken and analyzed by HPLC (fermentation conditions and time points are as described for Fig. 2). Diamonds, squares, and circles indicate the concentrations of KIC, DMP, and isovaleric acid, respectively.

View larger version (12K): [in this window] [in a new window]   FIG. 4.   ATP concentrations in E. faecalis OG1RF after depletion of pyruvate in the presence or absence of KIC. The concentrations of ATP in E. faecalis OG1RF in the absence of KIC (solid circles) and in the presence of KIC (open circles) were measured. Time zero corresponds to 5 or 7 h after the start of fermentation, as shown in Fig. 2, for cells grown in the absence or presence of KIC, respectively, and corresponds to the onset of ATP decrease due to low pyruvate concentrations. Cells were grown aerobically in a phosphate-buffered minimal medium supplemented with 0.5% yeast extract.

Activity of the BKDH complex under anaerobic conditions and involvement of the intermediate as a redox sink. Since the BKDH complex produces NADH during the reaction, it was of interest to see if the complex was active under anaerobic conditions and, if so, how the organism was able to maintain the redox balance. For these studies we again used 20 mM pyruvate as the primary free-energy source in the presence or absence of KIC (20 mM). In the absence of KIC all pyruvate was catabolized to acetate via pyruvate formate lyase (PFL). This was deduced from the equimolar amounts of acetate and formate produced. However, in the presence of KIC significant activity of the PDH complex was observed, as inferred from the higher concentrations of acetate (25 mM) than of formate (15 mM), indicating that 40% of the pyruvate was converted via the PDH complex and the remaining 60% via PFL. The activity of both the PDH and the BKDH complexes led to NADH formation, but none of the normal reduced compounds, such as lactate and ethanol, were formed in significant amounts (<1 mM). Interestingly, intermediate X accumulated relative to the amounts of the -keto acids that were utilized by the PDH and BKDH complexes (Fig. 5). Thus, at the end of the fermentation 25 mM acetate and 15 mM formate were formed, of which 10 mM acetate was formed via the PDH complex (i.e., 10 mM NADH surplus from PDH activity). In addition 5 mM isovalerate, and therefore 5 mM NADH, was also formed via BKDH complex. To compensate for this excess (15 mM) NADH generated by the actions of the PDH and BKDH complexes, a 16 mM concentration of intermediate X was formed. The reduced structure of intermediate X as determined via mass spectroscopy and the balancing of the redox equivalents led us to postulate that the formation of intermediate X can function as a redox sink for E. faecalis (Fig. 6).

View larger version (16K): [in this window] [in a new window]   FIG. 5.   Anaerobic growth and KIC catabolism of E. faecalis OG1RF with pyruvate as the primary free-energy source. Concentrations of KIC (diamonds), the intermediate DMP (squares), and isovaleric acid (circles) are shown as a function of time. Arrow 1, time point at which additional KIC was added to the fermentation mixture at a slow rate (concentration remained below the detection limit); arrow 2, time point at which the culture was switched to aerobic conditions. Cells were grown anaerobically in a phosphate-buffered minimal medium supplemented with 0.5% yeast extract with pyruvate as the primary free-energy source.

View larger version (12K): [in this window] [in a new window]   FIG. 6.   Analogy between pyruvate and KIC catabolism. Activity of the multienzyme complexes PDH and BKDH leads to the formation of the oxidized CoA adducts of the substrates, which can be further catabolized via a transacetylase and a kinase to the respective products acetate and isovalerate. Both -keto acids can also be reduced (pyruvate can be reduced to lactate, and KIC can be reduced to DMP), thereby reoxidizing the NADH formed during the activity of the multienzyme complexes. In lactic acid bacteria lactate can usually not be used as a substrate, in contrast to DMP, which can be readily oxidized in the presence of an electron acceptor such as oxygen or fumarate. LDH, lactate dehydrogenase; XDH, intermediate X dehydrogenase.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have shown that the addition of branched-chain -keto acids KIC, KMV, and KIV resulted in an increase in expression from the bkd promoter, with KMV having the highest level of induction at fourfold. Addition of the branched-chain -keto acids resulted in a roughly twofold-higher level of induction than was achieved with the corresponding amino acids. This suggests that the -keto acids are the inducers and not the amino acids. Strictly speaking we cannot exclude the possibility that the branched-chain amino acids are the actual inducers; via a reversible transamination the -keto acids could be converted to the branched-chain amino acids, and thereafter this could lead to the induction of the bkd gene cluster. However, the fact that the -keto acids are the substrates for the BKDH complex and the observation that the branched-chain amino acids were not utilized as substrates by E. faecalis strongly suggest that the -keto acids are the inducers. This differs from what was reported for the P. putida BKDH complex, in which the branched-chain amino acids and not the branched-chain -keto acids were the positive coeffector (18).

There is evidence to suggest that the mechanism of CR characterized in B. subtilis also pertains to E. faecalis. A putative CRE has been identified in the promoter region of the bkd gene cluster as well as in those of the glp operons of E. faecalis and Enterococcus casseliflavus (D. Parsonage, D. E. Ward, and A. Claiborne, unpublished data). Furthermore, a polyclonal antiserum to the Bacillus megaterium CcpA revealed the existence of anti-CcpA cross-reacting proteins in extracts of E. faecalis (15), and the bifunctional HPr kinase/phosphatase has also been recently identified (14). Analysis of the effects of various carbohydrates on the expression of the bkd-gusA fusion revealed that the sugars glucose, lactose, and fructose all inhibited transcription from the bkd promoter. The PTS substrate gluconate had no repressive effect on the expression of the bkd promoter. Gluconate catabolism is believed to occur via a hexose monophosphate shunt (HMS) pathway in the lactic acid bacteria (17). Utilization of the HMS pathway does not generate fructose-1,6-bisphosphate and therefore would not induce phosphorylation of the regulatory serine residue of HPr (12, 14). These results are consistent with the effect being mediated by the glycolytic intermediate fructose-1,6-bisphosphate. CR of the bkd gene cluster was also shown in vivo. The presence of glucose inhibited the formation of isovaleric acid from KIC. Whereas with pyruvate as the primary free-energy source all KIC is converted into isovaleric acid in 17 h (Fig. 3), with glucose the conversion is much slower, with less than 20% of the KIC being converted to isovaleric acid after 17 h (data not shown).

Earlier we postulated that the pathway encoded by the bkd gene cluster would be beneficial to the cell since it should be coupled to ATP production. With pyruvate as the free-energy source 1 mol of ATP should be formed per mol of acetate produced. For the cultures grown in the absence of KIC a final OD₆₀₀ of 0.91 was obtained. From these data a yield on ATP of 0.2 g (dry weight)/0.0195 mol of ATP = 10.2 g (dry weight)/mol of ATP is calculated (assuming an OD₆₀₀ of 0.91 corresponds to 0.2 g [dry weight]/liter). This yield is close to the ATP yield of 10.5 g (dry weight)/mol of ATP observed by Bauchop and Elsden (3). Assuming the same yields on ATP in the presence and absence of KIC, the additional increase in OD (from 0.91 to 1.35) of 0.44 units corresponds to 0.5 mol of ATP produced per mol of KIC consumed. In the proposed pathway 1 mol of ATP would be produced per mol of isovaleric acid formed, assuming that the NADH produced is reoxidized in an energetically neutral way. If KIC or the intermediate is imported actively (e.g., through proton symport), this would lower the apparent amount of ATP that is produced via the bkd gene cluster, unless the isovalerate, intermediate X, and KIC are transported at equal proton stoichiometries. That ATP is produced by the pathway is also suggested by the persistently high ATP levels that are observed in cells growing in the presence of KIC after depletion of pyruvate (Fig. 4). Despite the additional ATP produced, a decrease in growth rate was observed when KIC was present in the culture medium. This inhibitory effect on growth rate was strongest in the bkd mutants. An explanation for growth rate inhibition by the branched-chain -keto acids could be that the acids work as uncouplers. However, then a lower final OD (i.e., lower growth yield) would be expected, and this was not observed.

During the growth of E. faecalis in the presence of the branched-chain -keto acid KIC, KIV, or KMV, distinct intermediates were observed for each. Intermediate X, formed during the utilization of KIC, was purified and was tentatively identified as 1,1-dihydroxy-4-methyl-2-pentanone by mass spectroscopy. Gem-diol structures are known to be formed upon the hydration of aldehydes. They are stable only in the presence of water, and the equilibrium toward the carbonyl form is highly dependent on the structure of the hydrate. More than 99% of formaldehyde is in the hydrated form, while for acetone the hydrated form is negligible. Electron-withdrawing groups stabilize the gem-diol structure, and for the -keto aldehydes stable hydrates are formed. Thus the formation of DMP from the aldehyde would be spontaneous, but the mechanism for its formation from KIC is unknown to us. In absence of bacteria, i.e., in the medium without inoculation, no intermediate was formed in detectable concentrations overnight.

The -keto acid dehydrogenase complexes are known to be inhibited by high levels of NADH via the E3 subunit, consistent with their predominant role in aerobic catabolism. The PDH complex of E. faecalis is also active under anaerobic conditions when oxidized substrates (such as pyruvate) are used (23). When tested with respect to its redox properties the purified LipDH of the BKDH complex was found to be more resistant to overreduction than the LipDH of the PDH complex of E. coli but more sensitive than the LipDH of the PDH complex of E. faecalis (29). In addition to an inhibition of the activity of the enzyme complex by NADH, the necessity of the overall metabolism to be redox neutral imposes a stoichiometric constraint. Under aerobic conditions excess NADH can be readily oxidized via NADH oxidase, while under anaerobic conditions cells need to balance NADH turnover via internally generated electron acceptors. When cells are grown on pyruvate, anaerobic PDH complex activity is accompanied by lactate production (22). Reducing pyruvate via lactate dehydrogenase oxidizes the NADH formed during the oxidative decarboxylation of pyruvate. The substrate is used both as an electron donor and as an electron acceptor. Growing E. faecalis under anaerobic conditions on pyruvate in the presence of KIC led to much higher PDH complex activities than those for growth on pyruvate alone. Strikingly, the PDH complex activity was not balanced by an equally high lactate production (also no ethanol was formed), leading to an apparently unbalanced fermentation. On the basis of the structure of intermediate X we postulated that KIC could function as an electron acceptor in addition to being the substrate for the BKDH complex. Indeed a closed redox balance could be calculated assuming that 1 mol of NADH was oxidized per mol of intermediate formed (Fig. 6). The hypothesis that KIC can be reduced to an intermediate, thereby serving as a redox sink, was further strengthened by the observation that KIC was catabolized to intermediate X and isovalerate at a 1-to-1 stoichiometry when no other electron acceptor was present. The strongest indication in favor of our hypothesis was the subsequent conversion of intermediate X to isovalerate when an alternative electron acceptor was available, such as oxygen or fumarate. By using fumarate it was confirmed that 2 mol of NADH is formed per mol of intermediate X converted to isovalerate. Analysis of the flanking regions of the bkd operon did not reveal the presence of a putative dehydrogenase or reductase. Neither the orf1 nor orf8 (29) appears to encode the dehydrogenase. However, since the genome of E. faecalis is being sequenced (see the Institute for Genomic Research website at http://www.tigr.org), we have analyzed the sequence further downstream and identified an open reading frame encoding a putative dehydrogenase/reductase. Studies are under way to determine if this open reading frame encodes the DMP dehydrogenase.

Catabolism of the branched-chain -keto acids serves as a nice example of the versatility of microorganisms in coping with changes in the availability of carbon sources as well as the stoichiometric constraints. The analogy to pyruvate catabolism is strong, but with the branched-chain -keto acids the constraint is even more strict since apparently no equivalent to the PFL exists for the catabolism of these keto acids. In the absence of a suitable electron acceptor the only solution is to use the -keto acid itself as an electron sink. The intermediate thus formed can be used if the environmental conditions change such that an additional electron acceptor becomes available.