INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In Escherichia coli, several enzymes or pathways are able to convert malate to oxaloacetate. The NAD-dependent (cytoplasmic) malate dehydrogenase (MDH; EC 1.1.1.37) has always been considered to be the principal malate-oxidizing enzyme in the citric acid cycle (tricarboxylic acid [TCA] cycle) of this organism (9). Recently, a malate:quinone-oxidoreductase (MQO; EC 1.1.99.16) which is an essential enzyme in the TCA cycle of Corynebacterium glutamicum (reference 23 and accompanying paper) was described. MQO is a flavin adenine dinucleotide (FAD)- and lipid-dependent peripheral membrane protein catalyzing the oxidation of L-malate to oxaloacetate. The electrons are donated to the electron transfer chain at the level of quinones. This reaction is essentially irreversible (17). In C. glutamicum, MQO and the cytoplasmic NAD-dependent malate dehydrogenase, which is capable of reversible oxidation of malate to oxaloacetate, are active at the same time. Due to the different natures of their electron acceptors, the Gibbs standard free energies for the reactions catalyzed by MQO and MDH differ radically. For the MDH-dependent reaction, the standard free energy is equal to +28.6 kJ · mol¹, and for the MQO-dependent reaction, it is 55 or 18.9 kJ · mol¹, depending on whether the acceptor is ubiquinone or menaquinone, respectively (23). These figures suggest that, when both enzymes are active, under standard-state conditions, MQO should catalyze the oxidation of malate whereas MDH should catalyze the reduction of oxaloacetate. In C. glutamicum, it was, indeed, clearly demonstrated that MQO is the principal enzyme responsible for the oxidation of malate to oxaloacetate, whereas there were indications that MDH catalyzes oxaloacetate reduction (accompanying paper). MDH could be forced to function in the direction of malate oxidation only when the mqo gene was deleted and the NAD concentration in the cell was increased by addition of nicotinamide to the growth medium.

In E. coli, an MQO-like enzyme activity (at the time usually referred to as malate oxidase) was described before. This activity was thought to be induced to significant levels only in certain mutants lacking MDH activity (12). Because an MDH-negative mutant which still contained inactive MDH protein exhibited very high MQO activity, it was surmised that the expression of the gene encoding MQO was regulated by substrates or products of the MDH reaction (25). MQO was partially purified from an MDH-negative E. coli strain which overexpressed this enzyme (25). Properties of the partially purified protein were similar to those found for the purified protein of C. glutamicum (23). Initial studies by us, using a more sensitive MQO assay and optimized assay conditions (unpublished results), indicated that in contrast with previous observations MDH and MQO were active at the same time in E. coli. Because of this observation, and regarding the findings in C. glutamicum, a reinvestigation of the role of MDH and more detailed studies of the role of MQO in E. coli were deemed necessary.

The open reading frame (ORF) yojH (EMBL accession no. AE000310) found on the genome of E. coli has a high similarity (43.9% identical residues) to the mqo gene of C. glutamicum (23). This ORF (here called mqo) was used to overexpress MQO activity in E. coli. Furthermore, by using a construct of the mqo promoter region fused with the reporter gene lacZ, the regulation of expression was examined. To investigate the importance of MQO and MDH for the physiology of E. coli, and to investigate the supposed regulation of mqo expression by MDH, deletions of mqo and mdh were constructed on the chromosome. To assess the roles of MDH and MQO in malate oxidation, site-directed deletions of the genes coding for the malic enzymes and for phosphoenolpyruvate (PEP) synthase were constructed to block alternative metabolic routes leading from malate to oxaloacetate.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. E. coli DH5 (supE44 lacU169 80lacZM15 hsdR17 recA1 endA1 gyrA96 thi-1 relA1) (31) was used for standard genetic experiments. Measurement of MQO activity and gene replacements were carried out in E. coli K-12 strains DSM 5698 (ATCC 25404) (F ) and MC4100 [F araD139 (argF-lac) U169 rpsL150 flb5301 deoC1 ptsF25 rbsR] (4, 30). E. coli PC2 (parent, MC4100; fnr2) and PC35 (parent, MC4100; arcA Kan^r) were a generous gift from R. P. Gunsalus (7).

1

Construction of pECmqo. Oligonucleotides used are listed in Table 1. The mqo ORF (EMBL accession no. AE000310) was amplified by PCR from the E. coli MC4100 genome using YojH_01 as forward primer and YojH_02 as reverse primer. PCR conditions were a standard 30 cycles consisting of a temperature profile of 30 s at 94°C, 30 s at 60°C, and 2 min at 72°C. The YojH_01 primer is complementary to a region 101 bp in front of the +1 ATG start codon of the mqo ORF, and YojH_02 is located 92 bp following the TAA stop codon. The resulting PCR fragment was cloned in the vector pET324 (36), resulting in pECmqo for expression in E. coli. For expression in C. glutamicum, the fragment was cloned in pJCBW1, a derivative of pJC1 (8, 20) containing the lacZ complementation region and multiple cloning site from pSK+ (Stratagene), using EcoRV.

Construction of (mqo-lacZ). Oligonucleotides used are listed in Table 1. A construct was made in which the lacZ gene was fused to the promoter region of mqo. Since the mqo ORF has at least four possible start sites (ATG), and since it is unclear which one is used by the organism, the most probable start had to be determined by an educated guess. To this end, the hypothetical protein sequence of the mqo ORF was compared to that of MQO from C. glutamicum (23). The amino-terminal amino acid sequence of the purified MQO protein of C. glutamicum was SDSPK (D. Molenaar, unpublished results). This sequence can be recognized in the translated DNA sequence of the C. glutamicum mqo region and is preceded there by a methionine (see Fig. 5 in reference 23), which is apparently posttranslationally removed. The MQO protein from C. glutamicum therefore starts with the methionine immediately in front of a putative FAD binding site referred to in reference 23. The analogous methionine (ATG codon) in the mqo ORF, immediately in front of its putative FAD binding site, was defined as the most probable start of MQO of E. coli, and the corresponding hypothetical protein starts with the amino acid sequence MAAKA (see Fig. 5 in reference 23). The fusion of the mqo promoter region with the lacZ gene was constructed in such a way that the lacZ ORF starts with the ATG codon of the mqo ORF corresponding to this most probable start. A 681-bp region preceding this ATG codon of the mqo ORF was isolated using primer F_01 as forward primer and primer F_02 as reverse primer in reaction A. Since the most probable ATG is also the one furthest downstream in the sequence of possible ATG starts, all regulatory elements of the mqo promoter should be present in the construct, even if this ATG is not the true start codon. The lacZ ORF was isolated using the primer F_03 as forward primer and the primer F_04 as reverse primer in reaction B. The 3' region of primer F_02 and the 5' region of primer F_03 are complementary. In a crossover PCR, the products of reactions A and B are combined with the primers F_01 and F_04. The resulting PCR product is the desired fusion between the promoter region of the mqo ORF and the lacZ ORF starting with the most probable ATG start codon. The PCR product was digested with HindIII and EcoRI and inserted into the vector pBR322 (3), resulting in the plasmid pML2.

Construction of E. coli mdh, mqo, sfcA, b2463, and pps deletion strains. Oligonucleotides used are listed in Table 1. PCR conditions were chosen as described above. Chromosomal gene disruptions were carried out in E. coli MC4100 according to the method of Link et al. (19).

Preparation of membrane fragments. A sample of 5 × 10¹¹ cells (calculated by assuming that an optical density at 600 nm equal to 1 corresponds to approximately 5 × 10⁸ cells ml¹) was harvested and washed twice with 25 ml of ice-cold buffer A consisting of 50 mM HEPES, 10 mM K-acetate, 10 mM CaCl₂, and 5 mM MgCl₂ titrated with NaOH to pH 7.5. The washed cells were resuspended in 10 ml of buffer A and passed once through a French pressure cell at 10,000 lb/in² (69 MPa). Cell debris was removed by centrifuging it for 10 min at 10,000 × g. The supernatant was centrifuged for 30 min at 75,000 × g, and membranes were resuspended in buffer A to 10 to 15 mg of protein per ml.

Enzyme assays. Malate dehydrogenase activity was determined in cells grown overnight in LB. The cells were washed twice in 150 mM Tris-15 mM EDTA, titrated to pH 7.4 with acetic acid, and passed once through a French pressure cell at 10,000 lb/in² (69 MPa). After a low-speed spin at 2,500 × g, the cell extract was used for the assay. Malate dehydrogenase activity was assayed at pH 9.0 in the direction of malate oxidation as described in reference 32.

Competition experiments with wild type and mqo. The strain MC4100 and its derivative mqo were precultured overnight on LB, and the optical densities at 600 nm of these cultures were measured. The cultures were then mixed to obtain a 1:1 proportion of MC4100 and mqo cells. Of this mixture, 2 µl was used to inoculate 2 ml of NMM (26) with different carbon and energy sources. For each carbon and energy source, five cultures were inoculated and maintained in parallel throughout the experiment. The cultures were incubated in a shaker at 37°C for 24 h, and approximately 2 µl of each culture was transferred to 2 ml of fresh NMM. The optical density at 600 nm was measured, and the rest of the culture volume was kept frozen at 20°C. The transfer to fresh medium was repeated seven times. The number of generations after the last cultures had reached stationary phase was approximately 50 to 60. PCR was carried out as described above on 1 µl of a 50-fold-diluted sample of the culture to estimate the proportions of MC4100 and mqo using oligonucleotides Yojh_01 and Yojh_02. The linear quantitative character of the PCR method was confirmed by testing with defined mixtures of these strains.

Estimation of growth rates. Two growth curves were recorded by monitoring the optical density at 600 nm of cultures which had been inoculated with overnight cultures on the same medium from two different colonies. The specific growth rates were estimated by simultaneous least-squares fitting of both growth curves to exponential growth equations.

Protein determination. Protein was determined with bicinchoninic acid in the presence of 0.5% (wt/vol) sodium dodecyl sulfate, according to a protocol adapted from reference 34.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The gene mqo, formerly called yojH, encodes an MQO in E. coli. The high amino acid sequence similarity of MQO from C. glutamicum and the hypothetical protein encoded by the gene mqo (previously called yojH) of E. coli suggested that this gene encodes an MQO in E. coli (23). To test this possibility, the mqo gene was amplified by PCR. The PCR product containing the mqo ORF was inserted in a high-copy-number vector under the control of the trc (lac) promoter, resulting in plasmid pECmqo. This plasmid was introduced in E. coli DSM 5698 (wild-type K-12 strain). The resulting strain, when grown on LB, had an MQO activity of 32 nmol · min¹ · mg¹ of protein. The detection of activity in E. coli membranes with DCPIP as electron acceptor was absolutely dependent on the addition of a quinone and varied with the particular quinone added. The highest activity is found in the presence of ubiquinone 1 (results not shown). Addition of the inducer isopropyl--D-thiogalactopyranoside (IPTG) increased the activity to 176 nmol · min¹ · mg¹ of protein. Using a C. glutamicum-E. coli shuttle plasmid, pJCBW1, the mqo gene could also be expressed in the C. glutamicum DM22 mutant which lacks MQO activity (23). The addition of IPTG to these cells gave rise to MQO activity, albeit with lower specific activities than those found in membranes from C. glutamicum strain ATCC 13032 (wild type).

Effects of carbon substrates on (mqo-lacZ) expression. To determine the effect of the carbon source on mqo gene expression, an mqo-lacZ fusion (mqo-lacZ) was constructed on a plasmid (pML2). Strain MC4100/pML2, lacking the lacZ gene on the chromosome, was grown in NMM on different carbon sources to mid-logarithmic phase, and -galactosidase levels were determined. Expression was low on NMM with glucose and high when nonfermentable substrates were used as the carbon source (Fig. 1). Expression of mqo appears to be under catabolite control, similar to the regulation of mdh and the succinate dehydrogenase gene (sdh) (28, 29). Expression of (mqo-lacZ) in a arcA strain (PC35 [7]) was increased approximately two- to fourfold (Fig. 1). This shows that mqo, like many other genes encoding enzymes from the electron transfer chain, is regulated by the ArcA-ArcB two-component system. Deletion of fnr had no effect on expression of (mqo-lacZ), neither under aerobic conditions (Fig. 1) nor under anaerobic growth on glucose. The -galactosidase levels were approximately 10-fold lower under anaerobic conditions (data not shown).

View larger version (51K): [in this window] [in a new window]   FIG. 1.   Expression of (mqo-lacZ) from plasmid pML2 in E. coli MC4100 and in strains lacking ArcA, Fnr, or MDH. Cells were grown in NMM containing the indicated carbon source (0.5% [wt/vol]) to an optical density at 600 nm of 0.4. Hydrolysis of ortho-nitrophenyl--D-galactopyranoside (ONPG) was used as a measure of -galactosidase activity (31).

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Effect of the growth phase on MQO activity. E. coli MC4100 was grown in NMM using 0.5% (wt/vol) pyruvic acid as carbon source in a controlled batch under constant oxygen pressure. Membrane vesicles and cleared extracts were prepared from cells harvested at indicated positions. MQO and MDH were measured as described in the text.

Regulation of MQO activity in E. coli by ArcA, carbon source, and growth phase. Since expression of MQO in E. coli appears to be highly regulated by carbon source, membrane vesicles were prepared from cells grown on different carbon sources. The cells were harvested from cultures in the exponential growth phase, in the end-exponential phase, and in the stationary phase and from overnight cultures. High MQO activity could be measured in vesicles prepared from exponentially growing cells using pyruvate and acetate as carbon source (Table 2). Cells grown on glucose contained low MQO activities. The difference between MQO activity of cells grown on glucose and activity of cells grown on pyruvate is approximately fourfold and is in accordance with the results obtained for the expression of (mqo-lacZ) (Fig. 1). In cells grown on glucose, the MQO activity was fourfold higher in the arcA strain, in accordance with the (mqo-lacZ) promoter studies. Also, succinate dehydrogenase and MDH were derepressed in the arcA strain (results not shown and references 28 and 29). In cells growing on LB supplemented with 0.1% glucose, MQO activity can also be measured, in contrast to earlier observations (25).

Properties of mdh and mqo deletion strains. To investigate the role of MQO and MDH in E. coli, deletions of the mqo and mdh genes were generated. The deletion of mdh causes a severe decrease in growth rates on all substrates except glucose (Fig. 3). Strain mdh does not grow on acetate or malate. Strain mqo appears to have no growth defect on all substrates tested. Nevertheless, the double mutant mdh mqo grows more slowly on glucose, pyruvate, and lactate than does the mdh strain. This effect was confirmed in several independent growth experiments (data not shown). These results show that deletion of the mqo gene has a phenotype although this phenotype is seen only in combination with a deletion of the mdh gene. Overexpression of mqo from a plasmid in the mdh strain was deleterious for growth (Table 3). It should be mentioned that the mqo gene is also expressed from the plasmid pECmqo in the absence of the inducer IPTG, which in the mqo strain leads to a twofold-higher MQO activity than that present in the wild type. Also, a deleterious effect of both the empty vector pET324 and pECmqo on the growth rate of E. coli MC4100 can be observed, which is relieved by the addition of IPTG. For this reason, the rate of growth with the mqo gene relative to that without the mqo gene is displayed. Increasing overexpression of mqo by increasing inducer concentration led to a decrease of the growth rate.

View larger version (49K): [in this window] [in a new window]   FIG. 3.   Effect of mqo and mdh deletions on the specific growth rate of E. coli on various carbon sources. Cells were grown on NMM containing carbon sources (0.5% [wt/vol]) as indicated. Error bars represent estimated standard deviations.

Effects of sfcA, b2463, and pps deletions on growth of E. coli wild type and mqo. In view of the standard free energies of the MDH and MQO reactions, it would be likely that under physiological conditions (pH 7.0) MDH catalyzes the reduction of oxaloacetate and MQO oxidizes malate to oxaloacetate. The results presented above show that, in contrast to the deletion of the mdh gene, the deletion of the mqo gene in E. coli has no discernible effect on growth. This observation does not rule out the possibility that MQO catalyzes malate oxidation, since an alternative pathway for the conversion of malate to oxaloacetate, formed by the collective action of malic enzyme, PEP synthase, and PEP carboxylase (15) (see Fig. 4), might take over this function from MQO in a mqo strain. Furthermore, assuming that MDH catalyzes oxaloacetate reduction, we present an alternative explanation for the deleterious effect of an mdh deletion in the Discussion. Clearly, the alternative route for malate oxidation depicted in Fig. 4 had to be blocked to further investigate the function of MQO.

View larger version (21K): [in this window] [in a new window]   FIG. 4.   Alternative metabolic pathway (PEP shunt) for the conversion of malate to oxaloacetate in E. coli.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Earlier reports suggested that the level of MQO activity was down-regulated to very low levels by the presence of active MDH (12, 25). The increase in MQO activity in mutants lacking MDH activity could amount to a factor of 200 to 300 over the MQO activity found in the wild type. In the present study, it was shown that E. coli possesses significant MQO activity, even when MDH activity is present. Moreover, a defined deletion of the mdh gene did not lead to an increased MQO activity to the extent reported earlier. Also, the -galactosidase reporter levels in the mdh strain containing the mqo-lacZ fusion were only slightly increased compared to the levels in the wild type. These observations do not suggest a regulation of the mqo gene by active MDH.

The expression levels of the fusion protein in the arcA mutant clearly show that MQO activity is regulated by the ArcA-ArcB regulatory system. MDH expression is regulated by the same system (28). The activity of MQO also seems to be regulated both by carbon catabolite control and by the growth phase. The protein is active only during exponential growth, and when the cells reach stationary phase, it appears to be broken down very rapidly. In view of this intricate regulation, it is surprising that an mqo deletion in the wild type did not lead to an observable phenotype, and hence that no physiological function could be assigned to MQO.

In contrast, an mdh deletion induced severe growth defects. This observation per se does not imply that MDH oxidizes malate in the TCA cycle. The Gibbs standard free energy difference of the MDH reaction (+28.6 kJ · mol¹) is very unfavorable for malate oxidation. The NADH/NAD ratio in E. coli varies from 0.75 under conditions of low oxygen pressure to 0.075 under high oxygen pressure (10). Under low oxygen pressure, the equilibrium oxaloacetate/malate ratio of the MDH reaction would then be 2 × 10⁵. In practice, this would mean that the concentration of malate in the cell should be at least 5 to 50 mM, assuming that the concentration of oxaloacetate should be at least 0.1 to 1 µM to meet the requirements of oxaloacetate-consuming enzymes. Such malate concentrations may be difficult to attain.

An alternative route for the oxidation of malate to oxaloacetate, involving malic enzyme and pyruvate decarboxylase, has been suggested previously to exist in Bacillus subtilis and has been named pyruvate shunt (11). This organism possesses an MDH and lacks an MQO. No gene similar to mqo was found in its genome, nor could MQO activity be demonstrated in isolated membrane fragments (D. Molenaar, unpublished results). It was suggested that in B. subtilis malate dehydrogenase reduces oxaloacetate and that, therefore, the pyruvate shunt is necessary for the production of oxaloacetate (11).

Since E. coli does not possess a pyruvate carboxylase and assuming, as seems to be the case in C. glutamicum, that MQO catalyzes malate oxidation while MDH reduces oxaloacetate, the absence of a phenotype for the mqo deletion mutant in E. coli might be due to a PEP shunt taking over its function (Fig. 4). MDH in this case would function as a (down-)regulator of the oxaloacetate concentration, and the effect of an mdh deletion might then be due to increased intracellular oxaloacetate concentrations. Oxaloacetate is known to be a strong inhibitor of key metabolic enzymes, e.g., succinate dehydrogenase (1), malic enzyme (K_i, 10 µM) (33), or PEP synthase (400 µM) (5). This might also be the reason why overexpression of mqo from a plasmid causes growth inhibition (Table 3), because it would disturb the balance between MDH and MQO activity and lead to an increased oxaloacetate concentration. The results in Table 4, however, indicate that the reason for the absence of a phenotype in a strain lacking MQO is not that a PEP shunt takes over the (hypothetical) role of MQO in the conversion of malate to oxaloacetate. Furthermore, since the mqo pps and mqo sfcA b2463 strains lack the known alternative routes for conversion of malate to oxaloacetate, it has to be concluded that MDH oxidizes malate in E. coli. However, since the double mutant mqo mdh still grows on lactate, pyruvate, and glucose, an alternative route has to exist to complete the TCA cycle.

An interesting phenomenon still to discuss is that the additional deletion of the mqo gene has a negative effect on the growth of the mdh strain when growing on glucose, pyruvate, or lactate (Fig. 3). The simplest interpretation of these results is that, in the absence of MDH, MQO is capable of sustaining a low TCA cycle activity. This activity is just enough to support a growth rate comparable to that of the wild type during growth on glucose, and it sustains somewhat lower growth rates on lactate or pyruvate. The capacity of MQO is not sufficient for growth on acetate or C₄ compounds, when a very high TCA cycle activity is required for the generation of metabolic energy. The fact that MQO completely takes over the function of MDH in the mdh strain during growth on glucose is contradictory to the assertion by others that the TCA cycle is split during aerobic growth on glucose (13). Since MQO catalyzes irreversible oxidation of malate (17), it could not take over the function of an MDH which would reduce oxaloacetate in the reductive branch.

We are currently investigating possible reasons for the contrasting functions of MQO and MDH in E. coli and C. glutamicum. One of the obvious reasons to think of is the fact that there are significant differences in the specific activities of MQO and MDH and in the ratios of these activities between the two species. Compare, for example, the activities in Fig. 2 of this paper with those in Fig. 2 of the accompanying paper. Since the presence of both an MQO and MDH or their corresponding genes was observed in different eubacterial genera (6, 12, 16-18, 22, 23, 27), the results of these studies are of general importance for the physiology of the TCA cycle in bacteria.