INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Escherichia coli can utilize C₄-dicarboxylates as a carbon and energy source under aerobic and anaerobic conditions (9, 50, 56). Anaerobically, the uptake, exchange, and efflux of C₄-dicarboxylates (fumarate, malate, maleate, and succinate) and L-aspartate are mediated by the three independent dicarboxylate uptake (Dcu) systems, DcuA, DcuB, and DcuC (9, 12, 13, 50, 56). These Dcu systems appear to be active solely under anaerobic conditions (9).

Aerobically, uptake of C₄-dicarboxylates is mediated by a secondary transporter and/or a binding-protein-dependent system, designated Dct (20, 24). The Dct system has an apparent K_m of 10 to 20 µM for C₄-dicarboxylates and is driven by the electrochemical proton gradient (15), and its activity is induced by succinate and is subject to catabolite repression (20, 27). The corresponding dctA mutants cannot utilize the C₄-dicarboxylates malate and fumarate but grow normally on the monocarboxylate lactate (27). Transport across the outer membrane may be mediated by a C₄-dicarboxylate-binding protein (Cbt; K_d for C₄-dicarboxylates of 30 to 50 µM) and a porin (3, 4, 25-30).

Three genetic loci (cbt at 16.6 min, dctA at 79.3 min, and dctB at 16.4 min) are involved in aerobic C₄-dicarboxylate transport (27). The nucleotide sequence of the 76- to 81.5-min region revealed a putative dctA gene (f428) encoding a protein (DctA) most closely resembling the DctA proteins of Sinorhizobium meliloti and Rhizobium leguminosarum (62 to 63% identity) that function as H⁺/C₄-dicarboxylate symporters (51). The DctA proteins are members of a family that includes the Na⁺/H⁺ glutamate symporters (GltP/GltT). A role for the putative dctA gene of E. coli in the utilization of C₄-dicarboxylates (and the cyclic monocarboxylate orotate) has been suggested by complementation studies with Salmonella typhimurium dctA or outA mutants (2, 51). The coding regions corresponding to the dctB (predicted to encode an inner membrane protein) and cbt (predicted to encode the binding protein) genes have yet to be identified (23).

In addition to the dctA (f428) gene, E. coli contains three apparently cotranscribed genes (o157-o424-o328 or orfQMP) at 79.9 min that encode products 23 to 32% identical to the DctPQM components of the periplasmic binding protein-dependent C₄-dicarboxylate transport system of Rhodobacter capsulatus (11, 46, 51). The orfQMP genes are apparently part of a large operon involved in pentose sugar metabolism (11, 42). This suggests that the orfQMP products form a pentose sugar transporter, although, given their similarity to the R. capsulatus DctPQM components, it is also possible that they transport C₄-dicarboxylates.

To investigate the potential roles of the dctA and orfQMP genes of E. coli in C₄-dicarboxylate transport, the corresponding genes were inactivated and the phenotypes of the resulting mutants were studied. The results showed that the dctA (f428) product is required for aerobic growth on malate and fumarate and mediates the transport of C₄-dicarboxylates. Interestingly, dctA mutants were still able to grow aerobically on succinate, indicating the presence of an uncharacterized transporter with specificity for succinate. In contrast, the orfQMP products play no apparent role in C₄-dicarboxylate utilization and transport. Transcript mapping and regulatory studies with a dctA-lacZ transcriptional fusion showed that the dctA gene is monocistronic, has a single transcriptional start site, and is activated by cyclic AMP receptor protein (CRP) in the absence of glucose, repressed by ArcA during anaerobiosis, and weakly activated by the recently identified DcuS-DcuR system (13, 57) in the presence of C₄-dicarboxylates. In addition, inactivation of dctA led to constitutive dctA-lacZ expression with respect to C₄-dicarboxylates, suggesting that DctA regulates its own synthesis through an interaction with DcuS in a manner similar to that proposed for DctA- and DctB-dependent regulation of dctA in S. meliloti and R. leguminosarum.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Subcloning the dctA (f428) and orfQMP genes. The dctA (f428) and orfQMP genes were subcloned from phages 605 and 578, respectively (21), by standard procedures (36). DNA was isolated from the liquid lysates as described by Miller (36). A 4.9-kb BglII-SalI restriction fragment containing dctA was subcloned from 605 into the BamHI and SalI sites of pSU18, generating plasmid pGS753 (Fig. 1 and Table 1). Similarly, a 5.9-kb HindIII-BamHI fragment, containing orfQMP, was isolated from 578 and inserted into pSU18, generating pGS754 (Fig. 1 and Table 1).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Restriction maps of the dctA-orfQMP region of the E. coli chromosome. The inserts cloned in 578, 605, pGS753, pGS754, pGS928, pDctA, pOrfQMP, pDctA::Sp and pOrfQMP::Ap are shown along with E. coli DNA (thick black lines) and the Apr and Spr resistance cassettes (open bars). Relevant restriction sites are indicated: B, BamHI; Bc, BclI; Bg, BglII; Bs, BsphI; C, ClaI; E, EcoRI; H, HindIII; Hp, HpaI; P, PstI; S, SalI; and Sm, SmaI. Restriction sites within vector DNA are denoted by a v, and hybrid restriction sites no longer recognized by the corresponding enzymes are in parentheses. Solid arrows indicate the positions and polarities of relevant structural genes. The hatched bar represents the DNA fragment used as a hybridization probe, and a strongly predicted stem-loop structure is also indicated. Coordinates are from reference 51.

Inactivation of dctA (f428) and orfQMP. A 1.7-kb fragment containing the putative dctA gene was PCR amplified from pGS753 by using Pfu DNA polymerase (Stratagene) and primers annealing ~300 bp upstream and ~100 bp downstream of the dctA coding region, and the 1.7-kb product was subcloned into the SmaI site of pSU18 to generate pDctA (Fig. 1). The cloned dctA gene was disrupted by inserting a 1.9-kb SmaI fragment containing the spc cassette of pHP45Spc into the HpaI site of pDctA, generating pDctA::Sp (Fig. 1). A similar strategy was used to disrupt the orfQMP operon. A 3.1-kb fragment containing the orfQMP genes was PCR amplified from plasmid pGS754 with Pfu and primers annealing ~250 bp upstream of orfQ and ~150 bp downstream of orfP. The 3.1-kb PCR product was subcloned into the SmaI site of pSU18, generating pOrfQMP (Fig. 1), and the operon was disrupted by inserting the amp cassette contained in the 2.6-kb SmaI fragment of pHP45Ap into the Klenow-infilled BsphI site of pGS754 to generate pOrfQMP::Ap (Fig. 1). The chromosomal dctA and orfQMP genes were replaced by the disrupted versions in pDctA::Sp and pOrfQMP::Ap by allelic exchange (37). This was achieved by transforming JC7623 (recBC sbcB) with pDctA::Sp or pOrfQMP::Ap and isolating potential dctA::spc and orfM::amp mutants by screening for Sp^r Cm^s or Ap^r Cm^s colonies. The successful disruption of the dctA and orfM genes was confirmed by PCR and Southern blot analyses (Fig. 2). PCR amplification and Southern blotting of the dctA and orfM genes gave the expected band sizes for the parental strain and representative Sp^r Cm^s or Ap^r Cm^s mutants (MDO1 and MDO2, respectively), indicating that the corresponding dctA or orfM genes had indeed been disrupted with the spc or amp cassette (Fig. 1 and 2). Finally, the dctA::spc and orfM::amp mutations of strains MDO1 and MDO2 were transferred to the wild-type strain, AN387, by phage P1-mediated transductions (36) to produce the mutants MDO800 (AN387, dctA::spc) and MDO900 (AN387, orfM::amp). The identities of the resulting mutants were confirmed by PCR analysis (results not shown).

View larger version (53K): [in this window] [in a new window]   FIG. 2.   PCR and Southern blot analysis of dctA::spc and orfM::amp mutants. PCR (A and B) and Southern blotting (C and D) were performed, as described in Materials and Methods, with chromosomal DNA from JC7623 (lanes 1), JC7623 dctA::spc (MDO1) (lanes 2), and JC7623 orfM::amp (MDO2) (lanes 3), together with PCR primers or hybridization probes specific for the dctA (A and C) and orfQMP (B and D) regions. The sizes of the PCR products and major hybridizing bands are shown. Analysis of MDO3 gave results similar to those of MDO2 (data not shown).

Southern blot analysis of dctA and orfQMP mutants. Chromosomal DNA was isolated from JC7623 and the Sp^r Cm^s and Ap^r Cm^s derivatives (34). Southern blotting was performed with 10 µg of chromosomal DNA digested with PstI (dctA analysis) or with HindIII and BamHI (orfQMP analysis), separated by agarose electrophoresis, and blotted onto nylon membranes (41). Hybridizations were carried out at 63°C, and for the dctA analysis the hybridization probe was the digoxigenin-labeled 4.9-kb insert of pGS753, whereas for the orfQMP analysis the hybridization probe was the digoxigenin-labeled 5.9-kb insert of pGS754.

[¹⁴C]fumarate and [¹⁴C]succinate uptake experiments. Cultures were grown aerobically to late exponential phase in 500 ml of Luria broth (L broth), and the bacteria were harvested by centrifugation (2,000 × g for 15 min at 4°C), resuspended in 100 ml of M9 salts solution (Sigma) at 4°C, centrifuged as before, and finally resuspended in 5 ml of M9 salts solution and kept on ice for up to 6 h before use. Uptake of [¹⁴C]fumarate or [¹⁴C]succinate (1.8 to 2.2 Gbq mmol¹; NEN) was measured in a stirred Clark-type oxygen electrode assembly at 37°C under aerobic conditions (11, 45, 46). Aliquots (5 µl) of cell suspension were added to 2 ml of M9 salts solution and equilibrated for 1 min before the addition of the radiolabeled substrate to a final concentration of 20 µM [¹⁴C]fumarate or [¹⁴C]succinate. Samples (0.1 ml) were taken after 20 s and thereafter at 30-s intervals, immediately added to 5 ml of stop buffer (64 mM phosphate buffer [pH 7.0], 10 mM fumarate, 0.2 mM sodium fluoroacetate), and filtered rapidly through Whatman GF/F filters. The filters were dried and assayed for radioactivity by scintillation counting. The total protein contents of the cell suspensions were determined by the Lowry et al. assay (32).

Construction of the dctA-lacZ transcriptional fusion. The 0.9-kb HpaI-HpaI f651'-dctA' fragment of pGS753 was subcloned into the SmaI site of pRS528 (49) to generate pGS928, carrying a dctA-lacZ transcriptional fusion (Fig. 1). The dctA-lacZ fusion was transferred to phage RS45 (49) by in vivo homologous recombination as described by Simons et al. (49), and the resulting Lac⁺ phage, RS45(dctA'-lacZYA), was used to create a monolysogenic derivative of MC4100, JRG3351 (lac dctA-lacZYA).

Growth media and conditions. Cultures were grown at 37°C either aerobically (50 ml in 250-ml conical flasks at 250 rpm) or anaerobically (10 ml in filled and sealed Bijou bottles) in L broth or M9 minimal salts (Sigma) with various carbon sources.

-Galactosidase measurements. -Galactosidase specific activities (expressed in micromoles of o-nitrophenyl--D-galactopyranoside [ONPG] per minute per milligram of protein) were determined for samples taken at 0.5- to 1-h intervals from two independent cultures. Each sample was assayed in duplicate, as described previously (12).

Northern hybridization and primer extension analysis. Total RNA was extracted by a method (12) based upon that described by Aiba et al. (1). The 0.46-kb ClaI-PstI dctA fragment of pDctA (Fig. 1) was used as the hybridization probe. Reverse transcriptase-mediated primer extension analysis was performed (12, 39) with two independent primers for each promoter (see reference 51 for coordinates): P1_dctA (5'-²²³⁵ATCGCTGTCAGGACCTGAAAGTAAAGGCTT²²⁰⁶-3') and P2_dctA (5'-²²¹¹TAGAAATGGCCAAGGAGAATACCAATGGCT²¹⁸²-3'). Sequence ladders were generated with the T7 Sequenase DNA sequencing kit (Amersham) together with pGS753 as template and the primers described above. A potential CRP site was identified by using the score-matrix searching option of the xnip program (53) and a score matrix derived from 25 experimentally determined CRP-binding sites.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Growth properties of the dctA and orfM mutants. The wild type and dctA mutant grew identically under aerobic conditions in L broth (data not shown) and minimal medium containing 0.4% glucose (Fig. 3A), 0.4% fructose, 40 mM lactate, 40 mM acetate, or 0.4% glycerol (data not shown) as the sole carbon source. However, the dctA mutant (MDO800) failed to grow with malate or fumarate as the sole carbon source and grew more slowly than the wild type with succinate (Fig. 3). The growth difference on succinate was fully complemented by pDctA (Fig. 3B), but the growth defects on fumarate and malate were only partially restored by pDctA (Fig. 3C and D). However, AN387(pDctA) grew at the same rate as MDO800(pDctA) in fumarate and malate minimal medium (Fig. 3C and D), showing that pDctA lowers the fumarate- and malate-dependent growth of the parental strain. No growth defects were detected for the orfM mutant, MDO900, under the same conditions.

View larger version (25K): [in this window] [in a new window]   FIG. 3.   Effects of the dctA::spc mutation on growth. Cultures were grown aerobically in M9 salts medium containing 0.4% glucose (A), 50 mM succinate (B), 50 mM malate (C), or 50 mM fumarate: AN387 (dctA+) (), MDO800 (dctA::spc) (), AN387(pDctA) (), and MDO800(pDctA) (). OD650 nm, optical density at 650 nm.

Fumarate and succinate uptake properties of the dctA mutant. To determine whether the dctA::spc mutation affects transport or some other metabolic function, [¹⁴C]fumarate uptake was compared in the wild type and dctA mutant, and in the corresponding pDctA (dctA⁺) transformants, after aerobic growth to late log phase in L broth (Fig. 4). High levels of fumarate transport activity (~56 pmol of fumarate/min/mg of protein with 20 µM fumarate) were observed for the wild type (AN387), and although no uptake activity was detected with the dctA mutant, it was fully restored in the pDctA transformant (Fig. 4). This shows that the growth defects of the dctA mutant are due to deficient C₄-dicarboxylate transport (Dct) activity and that the failure to transport fumarate is due to inactivation of the dctA gene. The Dct activity of AN387 was inhibited to below detectable levels by a 100-fold excess of unlabeled fumarate but was not affected by 100-fold excess lactate, pyruvate, or acetate (data not shown). This indicates that neither lactate, pyruvate, nor acetate is transported by the DctA system. Dct activity for AN387 in L broth was 19-fold higher during the stationary phase than the early log phase (not shown), probably due to stationary-phase induction of dctA transcription (see below).

View larger version (19K): [in this window] [in a new window]   FIG. 4.   Effects of the dctA::spc mutation on fumarate transport. Cultures were grown aerobically in L broth, harvested, washed in M9 salts solution, and then used to measure the rate of [2,3-14C]fumarate uptake, in duplicate, as described in Materials and Methods. The strains were AN387 (dctA+) (), MDO800 (dctA::spc) (), AN387(pDctA) (), and MDO800(pDctA) (). Standard deviations are shown.

Environmental factors affecting dctA expression. To provide a more complete understanding of the role of DctA in C₄-dicarboxylic acid transport, factors that regulate dctA expression were examined by using a strain harboring a single-copy dctA-lacZ transcriptional fusion: JRG3351 (MC4100/RS45[dctA'-lacZYA]). The fusion contains 0.43 kb of the upstream f651 gene, the 0.18-kb f651-dctA intergenic region, and 0.3 kb of the dctA coding region (Fig. 1). The activity of the dctA-lacZ fusion in single copy (see below) indicated that the dctA gene possesses an independent promoter and therefore, contrary to a previous suggestion (51), is not dependent on the upstream f651 gene for transcription.

View larger version (23K): [in this window] [in a new window]   FIG. 5.   Expression of a dctA-lacZ transcriptional fusion during aerobic growth in L broth. The -galactosidase activities (solid symbols) and culture densities (open symbols) of JRG3351 (dctA-lacZ) are shown after growth at 37°C in L broth ( and ), L broth with 50 mM succinate ( and ), and L broth with 1% glucose ( and ). Error bars represent standard deviations for two cultures, each assayed in duplicate. OD650 nm, optical density at 650 nm.

View larger version (19K): [in this window] [in a new window]   FIG. 6.   Effects of various carboxylates and the dcuS mutation on dctA-lacZ expression. Cultures of JRG3351 (dcuS+) (A) and JRG3984 (dcuS) (B) were grown aerobically to the postexponential phase in L broth supplemented with carboxylates (50 mM) as indicated. The -galactosidase activities are shown with standard deviations.

View larger version (16K): [in this window] [in a new window]   FIG. 7.   Expression of dctA-lacZ during aerobic (A) and anaerobic (B) growth. Growth of JRG3351 took place in L broth or M9 minimal medium with or without 50 mM succinate (Suc), malate (Mal), fumarate (Fum), or nitrate or 0.4% glycerol (Gly) or maltose (Malt), or 0.4% glucose (Glu) in minimal medium (M) and 1% glucose in L broth (L). -Galactosidase activities were assayed in duplicate with samples taken from duplicate cultures at 0.5- to 1-h intervals over the entire growth phase, but only those corresponding to early-logarithmic (open bars) and postexponential (solid bars) growth are shown. Standard deviations are indicated.

Effects of global regulators on dctA-lacZ expression. The above studies show that dctA expression is strongly affected by catabolite repression, repressed in the absence of oxygen, induced in the stationary phase, and induced ca. twofold in the presence of citrate and C₄-dicarboxylates. The roles of the global transcriptional regulators, ArcA, FNR, CRP, and RpoS, in the regulation of dctA transcription were investigated by studying the appropriate regulatory mutants. The postexponential anaerobic expression of the dctA-lacZ fusion under fumarate respiratory conditions was 18-fold higher in the arcA strain, JRG4011, than in the arcA⁺ parental strain, JRG3351 (Fig. 8A), but transformation of the arcA strain with the arcA-containing plasmid, pRB38, restored the anaerobic repression of dctA expression, resulting in expression levels similar to those of the wild type (Fig. 8A). In contrast, the fnr mutation of JRG4013 increased dctA expression by only twofold. This effect was reversed by supplying the multicopy fnr⁺ plasmid, pCH21 (Fig. 8A). Similar results were observed with cultures grown under fermentative conditions (L broth plus 0.4% fructose), where postexponential dctA expression was 12- and 2-fold higher in the arcA and fnr mutants (2.4 and 0.4 µmol/min/mg), respectively, than in the wild type (0.2 µmol/min/mg). These results indicate that the anaerobic repression of dctA is mediated primarily by ArcA, with a minor contribution from FNR that is likely to be related to the activation of arcA expression by FNR (6).

View larger version (28K): [in this window] [in a new window]   FIG. 8.   Effects of ArcA, FNR, cAMP, and CRP on dctA-lacZ expression. Cultures were grown anaerobically (A) or aerobically (B to D) at 37°C in L broth plus 0.4% glycerol and 50 mM fumarate (A), in L broth only (B and D), or in L broth with and without 1% glucose (Glu) and 5 mM cAMP (C). The strains were JRG3351 (wild type [wt]), JRG4011 (arcA), JRG4013 (fnr), and JRG4016 (crp) and the corresponding transformants containing plasmids pRB38 (arcA+), pCH21 (fnr+), and pGS279 (crp+). Note that JRG4017 (the corresponding crp+ control for JRG4016) gave very similar results to those obtained with JRG3351 (data not shown). Other details are as for Fig. 7.

Roles of DcuS-DcuR and DctA in the C₄-dicarboxylate-dependent regulation of dctA-lacZ expression. The possibility that the C₄-dicarboxylate-dependent induction of dctA expression is mediated by the recently discovered two-component C₄-dicarboxylate-responsive DcuS-DcuR system was tested by using a dcuS mutant, JRG3984. Under aerobic conditions in L broth, the dcuS mutation caused a twofold reduction in postexponential dctA expression and abolished dctA induction by C₄-dicarboxylates or citrate (Fig. 6B). These findings are consistent with the recent report that dctA is induced fourfold by C₄-dicarboxylates in a DcuS-DcuR-dependent manner (13). The dctA gene could thus be a member of the DcuSR regulon, along with the dcuB-fumB (specifying an anaerobic C₄-dicarboxylate transporter and the anaerobic fumarase) and frdABCD (encoding fumarate reductase) operons (13, 57). However, it should be noted that these results contradict those of Zientz et al. (57), who concluded that the C₄-dicarboxylate induction of dctA is DcuS-DcuR independent.

View larger version (24K): [in this window] [in a new window]   FIG. 9.   Effects of the dctA mutation on dctA-lacZ expression. Cultures of JRG3351 (dctA+) (open bars) or JRG4005 (dctA::spc) (solid bars) were grown aerobically at 37°C in M9 salts medium containing 0.4% glycerol with or without 50 mM malate, succinate, or maleate. Other details are as for Fig. 7.

Citrate induces dcuB expression in a DcuS-DcuR-dependent manner. Although the DcuS-DcuR system is known to respond to external C₄-dicarboxylates (aspartate, fumarate, malate, maleate, succinate, and tartrate) (12, 13, 57), it had not been shown to respond to citrate until tested here (Fig. 6). To determine whether other members of the DcuSR regulon are also induced by citrate in a DcuS-DcuR-dependent fashion, expression of a dcuB-lacZ transcriptional fusion was measured following anaerobic growth to the late log phase (25 h of growth to an optical density at 650 nm of ~0.4) in minimal medium containing 0.4% glycerol and 50 mM trimethylamine oxide, with and without 50 mM citrate. The expression of dcuB was induced ~20-fold by citrate in the dcuS⁺ strain (JRG3835) (0.59 ± 0.05 and 0.032 ± 0.01 µmol of ONPG/min/mg of protein in the presence and absence, respectively, of citrate) but not in the dcuS mutant, JRG3983 (0.009 ± 0.0004 and 0.012 ± 0.002 µmol/min/mg, respectively), and in the presence of citrate, dcuB expression was 66-fold greater in the dcuS⁺ strain than the dcuS mutant. As expected, these results show that dcuB (like dctA) is induced by citrate in a DcuS-DcuR-dependent manner. This confirms that the DcuS-DcuR system responds to citrate as well as to C₄-dicarboxylates.

Northern blot analysis of dctA transcription. Northern hybridization was performed to determine the size of the dctA transcript and to correlate its abundance with expression of the dctA-lacZ fusion (Fig. 10A). Total RNA was extracted from MC1000 (wild type), JRG1999 (crp), and a pGS279 (crp⁺) transformant of JRG1999, grown to stationary phase in L broth with and without 0.4% glucose, and then hybridized with a labeled dctA fragment (Fig. 10A). A single dctA-hybridizing (1,300-nucleotide [nt]) transcript was detected for the wild type grown in the absence of glucose (Fig. 10A). The size corresponds to that expected for a monocistronic dctA transcript initiating at the promoter identified in the f651-dctA intergenic region (see below) and terminating at an inverted repeat (bp 96402 to 96372) located 15 bp downstream of the dctA stop codon (bp 96417) (GenBank accession no. U00039). No dctA-hybridizing transcript was detected for the wild type grown with glucose or from the crp mutant grown in the absence of glucose. However, a single major hybridizing band corresponding to the 1,300-nt dctA transcript was observed in the pGS279-complemented crp strain grown in the absence of glucose. Thus, the dctA Northern blot analysis supports the conclusions derived from the studies with the dctA-lacZ fusion that dctA transcription is strongly activated by the cAMP-CRP complex. Also, contrary to an earlier prediction (51) that dctA forms an operon with the upstream f651 gene (encoding a serpin-like protein), dctA appears to be monocistronic (Fig. 1).

View larger version (29K): [in this window] [in a new window]   FIG. 10.   Analysis of the dctA transcript. (A) Northern blotting. Total RNA was extracted from MC1000 (wild type [wt]), JRG1999 (crp), and JRG1999 transformed with pGS279 (crp+). Strains were grown aerobically in L broth with or without 1% glucose (Glu). Following electrophoresis and capillary transfer, RNA was hybridized with a labeled dctA fragment. The arrow on the right indicates the specifically hybridizing dctA transcripts. The size of the dctA mRNA transcript (in kilonucleotides) is indicated. (B) Determination of the dctA transcriptional start site by reverse transcriptase-mediated primer extension. RNA was isolated from MC1000 grown aerobically in L broth. Lane P indicates the primer extension product with primer P2dctA. Similar results were obtained with primer P1dctA (results not shown). The sequencing ladder (lanes A, C, G, and T) was generated by using primer P2dctA and pGS753 as template. The corresponding nucleotide sequence, its complement, and the transcriptional start site (indicated by an arrow) are shown. (C) Nucleotide sequence of the dctA promoter region. Coordinates are from reference 51. The experimentally determined +1 site is boxed and labeled, as are the deduced 35 and 10 sites and the predicted CRP site. Residues matching the corresponding consensus sequence for the CRP, 35, 10, +1, and Shine-Dalgarno (S-D) sites are in bold and underlined. The positions of the predicted start codon of dctA and termination codon of f651 are in bold.

Transcriptional start site of the dctA gene. The 5' end of the dctA transcript was defined by primer extension analysis (Fig. 10B). A single primer extension product corresponding to a start site at G-97752, 51 bp upstream of the anticipated dctA translational start codon, was detected (Fig. 10C). It matches the start site determined for the dctA gene of Salmonella typhimurium (2) and is preceded by appropriately positioned 10 and 35 sites, which are separated by the optimal distance, 17 bp. The 35 site is relatively poor, and this would explain why dctA expression is weak in the absence of cAMP-CRP activation. There is a strongly predicted CRP-binding site at bp 81.5, which is consistent with the strong cAMP-CRP activation of dctA expression (Fig. 10C), and this indicates that dctA is expressed from a class I CRP-dependent promoter (5). Although ArcA binding-site consensus sequences have been proposed (8, 33), they do not allow the accurate prediction of ArcA-binding sites (47). Therefore, it is not possible to locate potential ArcA-binding sites by sequence analysis, although it is presumed that dctA is directly repressed by ArcA.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The coidentity of dctA and f428 was previously assumed because of the high degree (~63%) of amino acid sequence identity between the translation products of the E. coli f428 gene and the rhizobial dctA genes (51), together with the close locations of the E. coli f428 and dctA genes in the respective physical and linkage maps (19, 25, 51). The data presented here confirm that f428 does indeed correspond to the dctA gene encoding the aerobic C₄-dicarboxylate transporter of E. coli. The dctA::spc mutant was unable to grow aerobically on fumarate or malate and grew weakly on succinate, which is consistent with the phenotype reported by Kay and Kornberg (19). Anaerobic growth was not affected, nor was aerobic growth on other carbon sources, indicating that dctA is strictly concerned with aerobic C₄-dicarboxylate metabolism. The growth defects of the dctA mutant were complemented by a multicopy dctA⁺ plasmid, confirming that the mutant phenotype is indeed caused by the inactivation of the chromosomal dctA gene. No comparable effects were detected for the orfM::amp mutant, suggesting that the orfQMP genes are not required for C₄-dicarboxylate transport. The overexpressed orfP-encoded periplasmic-binding protein has also been shown not to bind C₄-dicarboxylates (11a). The orfQMP genes would appear to form part of a nine-gene cluster that is required for L-lyxose utilization (42, 51), and so these genes are likely to be involved in the transport of L-lyxose or some other pentose sugar.

The role of DctA in the aerobic transport of C₄-dicarboxylates was confirmed by the aerobic fumarate transport deficiency of the dctA mutant and its complementation by a dctA⁺ plasmid. Overall, the results show that DctA is probably the sole mediator of fumarate and malate transport under aerobic conditions and that there is at least one undefined transporter (SucT) mediating the aerobic uptake of succinate, albeit at relatively low rates. Possible candidates are the products of the ygjU and o463 genes at 70 and 39 min (respectively), which encode DctA homologues with 27 and 20% sequence identity relative to DctA, or the ygjE (a potential tartrate/succinate antiporter), citT/ybdS (a citrate/succinate antiporter), and ybhI (a potential tricarboxylate transporter) genes (37a, 38).

Previous work has shown that a dctA mutant of Salmonella typhimurium cannot use orotate, a cyclic monocarboxylate, for pyrimidine synthesis, but that the lesion could be complemented by a dctA⁺ plasmid encoding the E. coli DctA protein, which is 95% identical to the S. typhimurium protein (2). The S. typhimurium dctA mutant was reported to be unable to grow on solid medium containing fumarate and malate, and it grew weakly with succinate, but no transport studies or attempts to complement the nutritional phenotype were made (2), nor was the capacity of the mutant to undergo aerobic growth on other carboxylates and anaerobic growth on C₄-dicarboxylates tested. A role for E. coli dctA in C₄-dicarboxylate transport has also been implied in a recent study (4a) showing that the inability of an E. coli atp deletion strain to grow on C₄-dicarboxylates can be complemented by a dctA⁺ plasmid.

Expression of a dctA-lacZ transcriptional fusion was found to be subject to strong CRP-mediated catabolite repression and ArcA-mediated anaerobic repression. Despite an earlier report to the contrary (56), the present results clearly support the view that the DcuS-DcuR system is involved in the C₄-dicarboxylate-dependent regulation of dctA (13). In a dctA mutant, dctA-lacZ expression was constitutive with respect to the presence or absence of C₄-dicarboxylates, indicating a role for DctA in regulating its own synthesis. By analogy to the mechanism proposed for rhizobial dctA autoregulation (17, 18, 40, 55), it is suggested that in aerobically grown E. coli in the absence of exogenous C₄-dicarboxylates, DcuS is inhibited by inactive (nontransporting) DctA. Conversely, in the presence of C₄-dicarboxylates, DctA would be active (transporting) and the inhibition of DcuS would be relieved, thus allowing signal transduction from DcuR to DcuS and the consequent induction of dctA. Such a mechanism implies that DcuS activity is dictated by the status of DctA. However, anaerobically, DcuS appears to directly sense external C₄-dicarboxylates in way that is independent of DctA or the DcuA, DcuB, and DcuC proteins (13, 57), suggesting that DcuS has two modes of operation, one that acts anaerobically and one that acts aerobically. Switching between the anaerobic and aerobic modes could be mediated by the central PAS domain of DcuS. The proposed redox-sensing role of the PAS domain would be consistent with this suggestion (13).

Previous work had shown that the DcuS-DcuR system is responsive to external C₄-dicarboxylates, namely, aspartate, fumarate, malate, maleate, succinate, and tartrate (13, 57). In this study, citrate was also found to be a coeffector for the DcuS-DcuR system, indicating that DcuS has broader ligand specificity than was previously realized. It is likely that the induction of the DcuSR regulon by citrate is physiologically significant, since, although most E. coli strains cannot utilize citrate aerobically, under anaerobic conditions citrate can be converted to fumarate for respiratory purposes (38). The anaerobic utilization of citrate would require the expression of the fumarate reductase operon, frdABCD, and the anaerobic fumarase gene of the dcuB-fumB operon. Appropriately, these operons are members of the DcuSR regulon (13, 57) and would therefore be induced by external citrate.

The pattern of dctA regulation is largely consistent with the role of its product in the aerobic uptake of C₄-dicarboxylates. The strong CRP activation of dctA ensures that external C₄-dicarboxylates are taken up for catabolic purposes only in the absence of "more preferable" carbon and energy sources, such as glucose. The anaerobic repression of dctA by ArcA limits the function of DctA to aerobic conditions. This is desirable since DctA is thought to be a proton symporter and would therefore consume energy, whereas the anaerobic C₄-dicarboxylate transporters can act in substrate-product exchange (antiport) mode and would therefore be non-energy consuming, a property that is likely to be important under the relatively low-energy-yielding conditions of anaerobic respiration. The relatively weak induction of dctA expression by C₄-dicarboxylates (and citrate) is unlikely to have any major impact on the C₄-dicarboxylate metabolism. Interestingly, a dcuS mutant exhibits the same growth defects as the dctA mutant, namely, no growth on fumarate and malate, and weak growth on succinate (13). This suggests that the dcuS mutant is devoid of Dct activity, which in turn suggests that DcuS (or DcuR, since the dcuS mutation is likely to have a polar effect on dcuR) could have a direct effect on DctA transport activity. This possibility is supported by the observation that the dcuS mutant totally lacks Dct activity (7a), although dctA transcription is only halved. The physiological purpose of such a regulatory mechanism is unclear but could be related to the need to ensure that DctA does not mediate net export of C₄-dicarboxylates, which could otherwise occur when external C₄-dicarboxylate concentrations are low.

The regulatory features of the dctA gene are reminiscent of those of the genes encoding the aerobic CAC (citric acid cycle) enzymes. Like dctA, CAC enzymes are regulated at the level of transcription by CRP-mediated catabolite repression and by the oxygen/redox-responsive, two-component ArcBA system (7). This mode of regulation ensures that CAC activity is relatively low both anaerobically, when the ability of the cycle to supply energy is restricted, and aerobically during growth on glycolytic substrates (such as glucose), when the full cycle is unnecessary because sufficient energy can be derived through glycolysis. Since the role of DctA is to deliver external C₄-dicarboxylates for consumption via the aerobically operating CAC, it is physiologically appropriate that dctA regulation should largely match that of the aerobic CAC enzymes.