The Fumarate/Succinate Antiporter DcuB of Escherichia coli Is a Bifunctional Protein with Sites for Regulation of DcuS-dependent Gene Expression*

DcuB of Escherichia coli catalyzes C4-dicarboxylate/succinate antiport during growth by fumarate respiration. The expression of genes of fumarate respiration, including the genes for DcuB (dcuB) and fumarate reductase (frdABCD) is transcriptionally activated by C4-dicarboxylates via the DcuS-DcuR two-component system, comprising the sensor kinase DcuS, which contains a periplasmic sensing domain for C4-dicarboxylates. Deletion or inactivation of dcuB caused constitutive expression of DcuS-regulated genes in the absence of C4-dicarboxylates. The effect was specific for DcuB and not observed after inactivation of the homologous DcuA or the more distantly related DcuC transporter. Random and site-directed mutation identified three point mutations (T394I, D398N, and K353A) in DcuB that caused a similar derepression as dcuB deletion, whereas the transport activity of the DcuB mutants was retained. Constitutive expression in the dcuB mutants depended on the presence of a functional DcuS-DcuR two-component system. Mutation of residues E79A, R83A, and R127A of DcuB, on the other hand, inactivated growth by fumarate respiration and transport of [14C]succinate, whereas the expression of dcuB′-′lacZ was not affected. Therefore, the antiporter DcuB is a bifunctional protein and has a regulatory function that is independent from transport, and sites for transport and regulation can be differentiated.

The fumarate/succinate antiporter DcuB (dicarboxylate uptake) of Escherichia coli catalyzes the uptake of external C 4 -dicarboxylates like fumarate, L-malate, or aspartate (1)(2)(3). Fumarate is used as an electron acceptor in fumarate respiration by fumarate reductase, which carries the active site at the cytoplasmic side of the membrane (for reviews, see Refs. 4 -6). L-Malate and aspartate are converted to fumarate by fumarase and aspartase, respectively, and are then metabolized in the same way as fumarate. Fumarate respiration results in the generation of a proton potential and drives ATP synthesis and growth of the bacteria. The product of fumarate respiration, succinate, is not further catabolized in anaerobic growth and is excreted by DcuB. DcuB, therefore, is responsible for the substrate/product antiport of fumarate, L-malate, or aspartate against succinate. DcuB and fumarase B are encoded by the dcuB fumB gene cluster (1,7).
E. coli contains two further carriers for C 4 -dicarboxylates during anaerobic growth, DcuA and DcuC (1,8). DcuA shows high sequence identity (36%) and similarity (63%) to DcuB and constitutes with DcuB the DcuAB family of C 4 -dicarboxylate transporters. The dcuA gene encoding DcuA is expressed constitutively under aerobic and anaerobic conditions (7). DcuA has been suggested to serve as a backup of DcuB or as a C 4 -dicarboxylate transporter for anabolic and other purposes. DcuC, on the other hand, is produced only during anaerobic growth (8). Expression of dcuC is (in contrast to dcuB) not subject to catabolite repression by glucose. DcuC is used under conditions of glucose fermentation when succinate is formed and functions as a succinate efflux transporter.
The Dcu transporters have different roles in the metabolism of E. coli and catalyze in the bacteria antiport (DcuB), uptake (DcuA), and efflux (DcuC) (1,3,8). Each of the carriers can, however, operate in any of the three transport modes (1,8). The transporters, therefore, can replace each other in mutants that are deficient in one or two of the Dcu carriers. The corresponding single and double mutants retain high activities for antiport and uptake (as well as for efflux) of C 4 -dicarboxylates. Only the triple mutant loses the capacity for uptake and antiport of C 4 -dicarboxylates and for growth by fumarate respiration (8,9).
Expression of dcuB and of dcuC depends on gene activation by the FNR (fumarate nitrate reductase regulator) protein (7)(8)(9). FNR activates expression of genes of anaerobic respiration in the absence of O 2 . Expression of dcuB requires in addition transcriptional activation by the DcuS-DcuR (dicarboxylate uptake sensor and regulator) two-component system (10,11). The system consists of the membrane-bound sensor kinase DcuS, which is activated in the presence of C 4 -dicarboxylates (12,13). The presence of stimuli like fumarate, L-malate, or aspartate causes phosphorylation of DcuS and transfer of the phosphate group to the response regulator DcuR, which then stimulates the expression of the target genes (14,15). These include the genes for fumarate reductase (frdABCD), DcuB (dcuB), fumarase B (fumB), and of the aerobic C 4 -dicarboxylate carrier DctA (dctA). Expression of dcuA and dcuC, on the other hand, is transcriptionally independent of DcuS-DcuR. The sensor site of DcuS responding to the C 4 -dicarboxylates is located in a periplasmic domain and has been characterized by structural and mutational studies (13,16,17). Recently, the cytoplas-mic PAS 2 domain of DcuS has been studied structurally as a part of a membrane-embedded construct containing the periplasmic and cytoplasmic PAS domains together with the transmembrane domain (18). The study showed an important role for the cytoplasmic PAS domain in signal transduction by the sensor kinase DcuS.
Here deletion and point mutants of DcuB were generated for a more detailed understanding of DcuB function. It turned out that deletion and inactivation of DcuB has, in addition to the effects on transport of C 4 -dicarboxylates, unexpectedly also an effect on the function of the DcuS-DcuR regulatory system. The regulatory function was analyzed by mutagenesis. Regulatory competent sites could be identified and separated from transport essential sites, demonstrating that DcuB is a bifunctional protein with transport and regulatory function.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Media-The bacterial strains and plasmids used are given in Table 1. For genetic experiments, the bacteria were routinely grown in Luria-Bertani (LB) broth and on LB agar (19). For anaerobic expression studies, M9 medium or enriched M9 (eM9) medium (20,21), with acid-hydrolyzed casamino acids (0.1%) and tryptophan (0.005%), was used. Glucose (20 mM), glycerol (50 mM), DMSO (50 mM), and sodium fumarate (50 mM) were added as energy substrates, as indicated. For growth experiments, eM9 medium supplemented with 50 mM glycerol and 50 mM fumarate was used, and growth was monitored for at least 30 h. Bacteria for transport assays were grown in the same medium containing 50 mM DMSO in addition. All media were inoculated at 37°C with 1-5% (v/v) of an overnight culture grown under the same conditions and in the same medium. When required, antibiotics were added as follows: 50 g ml Ϫ1 kanamycin, 50 g ml Ϫ1 spectinomycin, 20 g ml Ϫ1 chloramphenicol, and 15 g ml Ϫ1 tetracycline.
Genetic Procedures-Phage P1 transduction and DNA isolation and manipulation followed standard procedures (19,22). Plasmid DNA and PCR products were isolated and purified using kits (Qiagen). For cloning of dcuB, a 1.9-kb fragment containing the dcuB gene and its promoter was obtained by PCR from genomic DNA of E. coli AN387 (23) using primers dcuB-EcoRI and dcuB-XhoI (supplemental Table 1). The PCR product was digested with EcoRI and XhoI and cloned into the low copy number plasmid pME6010 with orip15A as in pACYC (ϳ10 copies/cell), resulting in plasmid pMW228. For mutagenesis, the 1.9-kb fragment was cloned from pMW228 into pET28a by the EcoRI and XhoI restriction sites, generating plasmid pMW281. Mutated forms of dcuB were recloned into pME6010 for in vivo complementation tests. Plasmids encoding wild-type and mutated forms of DcuS (13) are derivatives of plasmid pET28a (Novagen) with an intermediate copy number of 40 -50 copies/cell. The fumB gene was amplified from chromosomal DNA of E. coli MG 1655 using oligonucleotide primers fumB-NdeI-for1 and fumB-EcoRI-rev2 (supplemental  Table 1). The 1.6-kbp fragment was cloned into pET28a behind the T7-regulated T7/lac promoter using NdeI and EcoRI restriction sites, resulting in plasmid pMW522. Deletion of dcuB and Complementation-The dcuB gene was deleted by the method of Datsenko and Wanner (24). A Cam resistance cassette with flanking FRT sequences was amplified from plasmid pKD3 with the 70-nt deletion primers dcuB-H1-P1 and dcuB-H2-P2 (supplemental Table 1). The deletion primers each contained a region of 20 nucleotides before the ATG codon of dcuB homologous to the plasmid, including an FRT site, and another homologous region of 50 nucleotides after the stop codon. The purified PCR product was used to transform strain MC4100 (25) containing the helper plasmid pKD46 (26) for efficient recombination. Recombinants carrying the resistance cassette in place of the dcuB gene were selected on LB agar plates with chloramphenicol (25 g ml Ϫ1 ). From the resulting insertion mutant IMW497, the helper plasmid was eliminated by a temperature shift from 30 to 37°C. To eliminate the chloramphenicol resistance cassette, IMW497 was transformed with pCP20 (26), which encodes an FLP recombinase. The dcuB deletion mutant IMW502 was tested for chloramphenicol sensitivity and the loss of the helper plasmid. The dcuB deletion was complemented by the dcuB gene under the control of its native promoter on plasmid pMW228. Electro-competent E. coli IMW503 or IMW505 were transformed with pMW228 or its derivatives (see below), and the transformants were selected for tetracycline resistance. For expression analysis or for growth experiments, cells were grown as described, and the influence of pMW228 or derivatives on expression and growth was determined.
Hydroxylamine Mutagenesis and Selection of Mutants-The dcuB gene was mutagenized with hydroxylamine (27). Plasmid pMW228 DNA (20 g in 100 l of water) encoding dcuB was mixed with 500 l of buffer (0.1 M sodium phosphate, pH 6.0, and 1 mM EDTA) and 400 l of 1 M hydroxylamine hydrochloride (pH 6.0). The mixture was incubated at 70°C for 10 h, and a 200-l aliquot was removed every 2 h. Reactions were terminated either by ethanol precipitation or by purification with a PCR purification kit (Qiagen) by elution in 10 l of water, followed by butanol precipitation. Mutagenized plasmids were electroporated into E. coli IMW505, and transformants were selected by growth on M9 minimal agar plates containing glycerol, DMSO, X-gal (20 -40 g ml Ϫ1 ), spectinomycin (25 g ml Ϫ1 ), chloramphenicol (10 g ml Ϫ1 ), and tetracycline (7.5 g ml Ϫ1 ). Mutant colonies with increased expression of the reporter dcuBЈ-ЈlacZ were blue in color compared with white colonies of strains with wild-type dcuB. Strains from blue colonies were tested for their ability to grow anaerobically in eM9 medium with fumarate plus glycerol. Growth was monitored at least for 30 h. Bacteria capable of growth on fumarate as the sole electron acceptor reached an A 578 of 0.7. Reporter gene expression was quantified by measuring ␤-galactosidase activity after anaerobic growth in eM9 minimal medium supplemented with glycerol and DMSO, with or without fumarate (10). Plasmid DNA was extracted from strains with increased ␤-galactosidase activity that grew on fumarate, and the dcuB gene was sequenced.
Site-directed Mutagenesis of DcuB-Site-directed mutagenesis was carried out with the QuikChange kit (Stratagene). The dcuB coding region, including the complete promoter, was amplified by PCR and cloned into pET28a (Novagen). The resulting plasmid, pMW281, served as the template for sitedirected mutagenesis by PCR with synthetic primers containing the desired mutation. The mutations were verified by DNA sequencing. Oligonucleotide primers used for mutagenesis are listed in supplemental Table 1. E. coli strains were transformed with plasmids by electroporation (28). For growth tests, transport assays, and ␤-galactosidase activity tests, the mutated dcuB gene, including the promoter, was cloned into pME6010, resulting in derivatives of plasmid pMW228. Screening for (blue) regulatory mutants was performed as described above on X-gal agar plates containing glycerol plus DMSO.
DcuB-specific Antiserum and Western Blotting of DcuB-Polyclonal antibodies were produced by GenScript Corp. in rabbits using peptide W 194 FRGKDLDKDEEFQ 207 C (sequence in italic type is derived from DcuB). The peptide is part of the large cytoplasmic loop located between transmembrane helices V and VI of DcuB, assuming a topology as for the homologous DcuA (29). Membranes of the bacteria were prepared by breaking the bacteria in a French press (83 bar, two passages) followed by low (10,000 ϫ g) and high speed (45,000 ϫ g) centrifugation. The membrane pellet from the ultracentrifugation was suspended in buffer. The proteins were separated by SDS gel electrophoresis (30), blotted onto nitrocellulose membranes, and detected by immunoblotting with the antiserum.
Transport Assays (Uptake of [ 14 C]Succinate)-Bacteria were grown anaerobically in eM9 medium supplemented with glycerol, DMSO, and fumarate to an A 578 of 0.6 -0.8. After harvesting and washing, cells were resuspended in anoxic buffer (100 mM sodium potassium phosphate, pH 7, 1 mM MgSO 4 ) to an A 578 of 6.6 -8.0 at 4°C. After activation with glucose (20 mM) for 5 min at 37°C, uptake of [ 14 C]succinate (33.4 Bq/nmol, 100 M) was tested by a filtration assay with membrane filters (mixed cellulose ester, 0.2 m, Whatman/Schleicher & Schuell catalog number 10401706) as described (1,8). The amount of substrate taken up was calculated from the bacterial dry weight (an A 578 of 1 corresponds to 281 mg dry weight ml Ϫ1 ) and the radioactivity after background subtraction. The transport is given in mol of [ 14 C]succinate min Ϫ1 (g dry weight) Ϫ1 taken up or units (g dry weight) Ϫ1 at 37°C.
␤-Galactosidase Assay-Expression of the dcuBЈ-ЈlacZ reporter gene fusions was determined by measuring the ␤-galactosidase activity (22) of exponentially growing cultures (A 578 of 0.5-0.8) at 37°C under anaerobic conditions in eM9 medium supplemented with glycerol and DMSO, with or without fumarate.

DcuB Is Required in Vivo for the Response of DcuSR to C 4 -dicarboxylates-After initial indications that inactivation of
Dcu transporters has an effect on the expression of DcuS-dependent genes, the effect was studied systematically by inactivating the dcuA, dcuB, or dcuC genes in a strain that carries a chromosomal dcuBЈ-ЈlacZ reporter gene fusion ( Table 2).
Expression of dcuB is a good indicator for the functional state of the DcuS-DcuR system in vivo when the bacteria are grown under anaerobic conditions on glycerol plus fumarate (10,11). DMSO was included in the medium as an alternative electron acceptor to support growth when no fumarate was included or when mutations inhibited growth by fumarate respiration. The expression of dcuBЈ-ЈlacZ is stimulated about 15-fold by fumarate or other C 4 -dicarboxylates in the strain with wild-type dcuA dcuB dcuC genes ( Table 2). When the dcuB gene was inactivated by insertional disruption or deletion, expression of dcuBЈ-ЈlacZ unexpectedly was fully induced without the need for C 4 -dicarboxylates. The addition of fumarate even caused some repression. The effect was specific for dcuB and not observed after inactivation of dcuA or dcuC. In mutants with two or three of the dcu genes inactivated, expression became C 4 -dicarboxylate-independent whenever dcuB was one of the inactivated genes. The fumarate-independent (or constitutive) expression of dcuBЈ-ЈlacZ in the dcuA dcuB dcuC-negative strain IMW505 (Table 2) or in the dcuB-negative strain IMW370 (not shown) was reversed by introducing dcuB on plasmid, whereas plasmid-encoded dcuA restored only anaerobic growth on fumarate but not fumarate regulation of dcuBЈ-ЈlacZ expression.
For the above mentioned growth experiments, the mineral medium was supplemented with low amounts of casamino acids (eM9 medium) to enhance anaerobic growth, which is low in mineral media. The same experiments were performed without casamino acids to exclude effects by components like aspartate or L-malate that can be present in low amounts in the casamino acids (and function as effectors of DcuS). Glucose was used under these conditions as the substrate to increase growth. In the experiments, deletion of dcuB had the same stimulating effect as the addition of fumarate or L-malate ( Table 3). The activity of dcuBЈ-ЈlacZ expression, however, was generally low due to glucose repression (7,10). The constitutive expression without fumarate was lost here, too, after complementation with dcuB.
The fumB gene encoding the anaerobic fumarase B is located downstream of dcuB. The fumB gene can be transcribed from promoters in front of fumB and of dcuB, since monocistronic fumB and bicistronic dcuB fumB transcripts were identified (7). It is not clear which is the major transcript of fumB expression. It was checked that the effect of dcuB deletion was not caused by loss of FumB due to polar effects of dcuB deletion on fumB. A strain that was negative for fumB only (E. coli JW4083) still required fumarate (or L-malate) for full expression of the dcuBЈ-ЈlacZ fusion (Table 3). On the other hand, introducing fumB on plasmid in the strain with the chromosomally inactivated dcuB (E. coli IMW503) did not relieve the constitutive (fumarateindependent) expression of dcuBЈ-ЈlacZ. In control experiments, the same fumB gene on plasmid was able to complement the fumB-deficient strain JW4083 when anaerobic consumption of L-malate was tested (not shown). Overall, the effect of dcuB deletion on dcuBЈ-ЈlacZ expression was specific for the loss of dcuB and independent from fumB or fumarase B.
A further target for transcriptional stimulation by DcuS-DcuR is the frdABCD operon, which is stimulated to a low but significant extent (1.6 -2-fold) by the addition of C 4 -dicarboxylates (10,11). Expression of frdA was stimulated 2.3-fold in the dcuB mutant when no fumarate was present and exceeded slightly the stimulation by the presence of C 4 -dicarboxylates in the wild type. Thus, inactivation of dcuB generally affects expression of DcuS-DcuR-regulated genes.
Fumarate-independent Induction in the dcuB Mutant Requires Intact DcuS-In a dcuB mutant carrying an inactivated dcuS (Table 4) or dcuR gene (not shown), expression of dcuBЈ-ЈlacZ was completely lost, demonstrating that the C 4 -dicarboxylate-independent induction in the dcuB mutant still depends on DcuS-DcuR. Sensing of the C 4 -dicarboxylates by DcuS occurs by the periplasmic domain and can be inactivated by mutating essential residues Arg 147 or His 110 of the C 4 -dicarboxylate site (13). In mutant strains, producing DcuS(R147A) or DcuS(H110A) instead of wild-type DcuS, expression of dcuBЈ-ЈlacZ was lost in the same way as in the dcuS deletion strain, irrespective of the absence or presence of dcuB (Table 4). Therefore, the activation of dcuBЈ-ЈlacZ expression by dcuB deletion requires intact DcuS. Expression of dcuBЈ-ЈlacZ in the dcuB-negative background depends also on anaerobic conditions and the presence of FNR (not shown), similar to the situation in a dcuB wild-type background (10, 11).

TABLE 2 Effect of the inactivation of the Dcu carriers on the expression of dcuB-lacZ and anaerobic growth on fumarate
The bacteria are derivaties of E. coli MC4100, containing a chromosomal dcuBЈ-ЈlacZ fusion. The bacteria were grown anaerobically in eM9 medium with glycerol plus DMSO, with or without fumarate. ␤-Galactosidase was determined in exponentially growing cultures (A 578 of 0.5-0.8). ϩ, growth comparable with wild type; Ϫ, no significant growth.  The Regulatory Effect of DcuB Is Independent from Its Transport Function-The transport activity of the dcu mutants was determined, and it was tested whether the constitutive expression in the dcuB negative strain was related to the loss of transport activity or to the loss of DcuB. As an indicator for the transport capacities, the uptake activity of the strains was determined, since each of the transporters is able to catalyze antiport, uptake, or efflux of C 4 -dicarboxylates (1,3,8). In the dcuA, dcuB, and dcuC single mutants, uptake of [ 14 C]succinate was only slightly impaired (maximally 26% decrease in activity compared with the wild type). Even in the double mutants, the activity decreased maximally by 51% in relation to the wild type, which is in agreement with earlier studies (1,8). None of the single or double mutants was significantly impaired in growth by fumarate respiration (Table 2), which shows that the residual transport activities are sufficient for fumarate/succinate antiport. Only in the triple mutant, the transport activity decreased to background levels (11% of wild type), which prevented also growth by fumarate respiration. Therefore, the fumarate-independent expression of dcuBЈ-ЈlacZ is related to the loss of the DcuB protein but not to the activity of C 4 -dicarboxylate transport.

Strain
Mutations in DcuB Conferring Fumarate-independent Expression of dcuBЈ-ЈlacZ-In order to identify sites or amino acid residues that are responsible for the regulatory effect of DcuB, the dcuB gene was mutagenized in vitro randomly with hydroxylamine, which introduces C/G 3 T/A transitions. An E. coli reporter strain with the dcuBЈ-ЈlacZ fusion and inactivated dcuA, dcuB, and dcuC genes was transformed with a plasmid encoding the mutated dcuB gene. The transformants were screened for blue colonies on X-gal agar after anaerobic growth in the absence of fumarate. The strain with wild-type dcuB on plasmid produced blue colonies only when fumarate was included in the agar. A small percentage of the colonies containing hydroxylamine-treated dcuB (Ͻ Ͻ1%) showed blue staining in the absence of fumarate, indicating dcuBЈ-ЈlacZ expression without C 4 -dicarboxylates. For differentiation of the desired deregulated from dcuB-deficient mutants (for which the same phenotype of increased ␤-galactosidase activity would be expected), the strains from the blue colonies were tested for anaerobic growth on fumarate. Five independent mutants were obtained from two mutagenesis experiments that were able to grow on glycerol plus fumarate but required no fumarate for dcuBЈ-ЈlacZ induction. The isolates contained two types of mutations in DcuB, mutation T394I (ACT 3 ATT transition, four independent isolates) and mutation D398N (GAT 3 AAT transition, one mutant). The same mutations generated by directed mutagenesis in cloned dcuB showed the same phenotype as the random mutants, confirming that mutations T394I and D398N of DcuB are responsible for the phenotype.
Mutations DcuB(T394I) and DcuB(D398N) Affect Regulation by DcuB but Not Transport-A strain of E. coli producing DcuB(T394I) or DcuB(D398N) from plasmid as the only Dcu carrier was able to grow on glycerol plus fumarate similar to wild-type DcuB supplied from plasmid, confirming that the mutant forms of DcuB are transport-active (Fig. 1). Both mutants retained full activity for the uptake of [ 14 C]succinate into the bacterial cells (Fig. 2), whereas in the parental dcuA dcuB dcuC-deficient strain, the transport activity was close to background levels. When the expression of dcuBЈ-ЈlacZ was measured, both mutants showed full induction of the reporter gene fusion without fumarate. The expression of dcuBЈ-ЈlacZ was even higher than in the wild type after fumarate induction, and the presence of fumarate caused some decrease in expression. The DcuB(T394I) and DcuB(D398N) mutants therefore have the same regulatory phenotype as the ⌬dcuB mutant, but C 4 -dicarboxylate transport is retained. This indicates that residues Thr 394 and Asp 398 play a specific role in the regulation of DcuS-DcuR function but not in transport by DcuB.
Residues Thr 394 and Asp 398 were replaced by various residues (Table 5). In the DcuB(T394S) and DcuB(T394N) mutants, dcuBЈ-ЈlacZ expression was very similar to that in the wild type and required fumarate for induction, and the constitutive phenotype of DcuB(T394I) was no longer observed. The DcuB(T394A) mutant showed partial expression of dcuBЈ-ЈlacZ without C 4 -dicarboxylates, and fumarate caused further increase of the expression to wild-type levels. Since  Requirement of DcuS and its C 4 -carboxylate binding site for fumarate-independent induction in the dcuB mutant. Expression of dcuBЈ-ЈlacZ, measured as ␤-galactosidase activity, was determined in the midexponential growth phase. The variants of DcuS were supplied on plasmid, since it has been shown earlier (13) that plasmidencoded dcuS gene can be used as a source of DcuS. DcuB(T394N) has wild-type activity, the hydroxyl group of Thr is not essential for the function of DcuB in sensing. Replacement of Thr by nonpolar amino acid residues (Ile and in part also Ala) appears to be responsible for the constitutive phenotype, whereas polar and hydrophilic side groups (Thr, Asn, or Ser) confer C 4 -dicarboxylate responsiveness to DcuB. The DcuB(D398E) and DcuB(D398A) mutants did not retain the constitutive phenotype of DcuB(D398N) and were fumarate-responsive, similar to wild-type DcuB (Table 5), which indicates that the constitutive phenotype of DcuB(D398N) is specific for the Asp/Asn exchange. Most other residues are apparently tolerated without significant functional change. Residue Asp 398 might be required for its negative charge or another chemical or structural property of the side group. Altogether, the data suggest that the regulatory mutations T394I and D398N specifically depend on the exchange of the Thr and Asp residues by Ile and Asn.

␤-Galactosidase
The mutant forms of DcuB were tested after complementation by plasmid-borne dcuB for their activity in the uptake of [ 14 C]succinate in a dcuA dcuB dcuC-negative transport test strain (Table 5). Strains producing DcuB with mutations T394I, T394S, T394N, D398N, and D398A showed transport activities similar to the corresponding wild-type DcuB strain. Only strains with mutations T394A and D398E were impaired for transport to some extent (decrease of wild-type activity by 17-40%), but none of the strains lost the transport activity completely. Thus, mutants DcuB(T394I) and DcuB(D398N) were regulation-incompetent but had wild-type phenotype with respect to anaerobic growth on fumarate and C 4 -dicarboxylate transport. The loss in regulatory competence therefore was specific not only for the site of mutation but also for the type of exchange (Thr 3 Ile and Asp 3 Asn replacements, respectively).
Identification of Regulatory Mutants of DcuB by Directed Mutation-DcuB contains a considerable number (60) of basic (Arg, Lys, and His) and acidic (Asp and Glu) residues (Fig. 3). Most of the residues are located in cytoplasmic or periplasmic loops that connect the transmembrane helices, when an arrangement of transmembrane helices similar to those in the related DcuA transporter (29) is assumed. From the residues, 37 were selected for directed mutation in addition to residues Asp 398 and Thr 394 that had been identified by random mutagnesis. The selected residues were mostly conserved in DcuB and DcuA or were specific for DcuB (Fig.  3). The basic or acidic residues are candidates for interaction with C 4 -dicarboxylate anions. The residues were mutated to Ala residues in cloned dcuB. The mutant forms were tested in vivo in a dcuA dcuB dcuC deletion strain IMW505 for fumarate-independent induction of dcuBЈ-ЈlacZ. To this end, the mutant strains were screened for blue colonies on X-gal indicator plates after anaerobic growth on glycerol plus DMSO without fumarate. Strains with mutations DcuB(D119A), DcuB(K353A), and DcuB(D405A) formed blue colonies similar to the DcuB(T394I) and DcuB(D398N) mutants. In the ␤-galactosidase reporter assay, the three

TABLE 5 Effect of mutations in Thr 394 and Asp 398 of DcuB on the C 4 -dicarboxylate-dependent regulation of dcuB-lacZ expression and on transport activity
The regulatory activity was measured as ␤-galactosidase activity using th dcuBЈ-ЈlacZ reporter fusion. Uptake of ͓ 14 C͔succinate was determined in the transport test strain IMW505 (dcuA ⌬dcuB dcuC) transformed with the same plasmid as for the expression studies. The bacteria were grown in eM9 medium with glycerol ϩ DMSO plus fumarate (anoxic conditions).

Strain (relevant genotype)
␤-Galactosidase (dcuB-lacZ) strains showed high C 4 -dicarboxylate independent induction of dcuBЈ-ЈlacZ during anaerobic growth (Table 6), whereas the other mutants had no effect on the expression of dcuBЈ-Јlac. The strain with DcuB(K353A) was fully induced in the absence of fumarate, like mutants DcuB(T394I) and DcuB (D398N). . Basic (Arg, Lys, and His) and acidic (Asp and Glu) residues are labeled by red and green background. Basic and acidic residues that were mutated are indicated (filled circles, green). Regulatory residues Thr 394 and Asp 398 that were identified by random mutagenesis are highlighted (f). The numbers on the right give the numbering of the corresponding amino acid residues. In addition some of the mutated residues are numbered according to their position. The supposed positions of transmembrane helices I-X are indicated by gray bars below the sequence, assuming a topology according to DcuA (29).
Mutants DcuB(D119A) and DcuB(D405A) were in addition deficient for growth on fumarate, raising the possibility that both properties are caused by the lack of DcuB or complete inactivation of DcuB. The strains were tested by immunoblotting for the presence of the DcuB protein (Fig. 4). The membrane fraction of the strain producing DcuB(D119A) contained about half the levels of DcuB compared with the strain with wild-type DcuB. This suggests that the loss of transport and of the regulatory effect is caused by functional inactivation of DcuB(D119A) and by decreased contents. When residue Asp 119 was replaced by an Asn residue, DcuB(D119N) was fully active in growth and regulatory competence (not shown). This indicates that residue Asp 119 is not essential for normal regulatory competence, but some amino acid replacements might affect regulatory and transport properties by structural changes. The amount of DcuB(D405A) in the membranes of the bacteria was very low (Fig. 4), suggesting that the regulatory and transport deficiency of this mutant is due to the low content of the carrier. The same low expression was observed in DcuB(D405N/V/E) mutants. Overall, the directed mutation of DcuB revealed DcuB(K353A) as a further regulatory mutation of DcuB that had lost regulatory competence similar to DcuB(T394I) and DcuB (D398N).
Mutants of DcuB That Are Deficient in Transport but Retain Regulatory Competence-All point mutants of DcuB were tested in a second approach for their capability to grow on glyc-

Site-directed mutants of DcuB affecting expression of DcuS-DcuR-dependent genes and DcuB-dependent transport (uptake) of ͓ 14 C͔succinate
The experiments were performed with strain E. coli IMW505 (dcuA dcuB dcuC), which carried dcuB-encoding plasmids as indicated. Anaerobic growth was determined on glycerol plus fumarate as in Fig. 5 (ϩϩϩ, growth rate similar to strain with wild-type dcuB on plasmid; ϩ, slow growth (class II); Ϫ, no growth (comparable with class III, or dcuB-negative strain). Transport (uptake of ͓ 14 C͔succinate) was determined with resting cells as described in the legend to Fig. 2 erol plus fumarate as the substrates under anaerobic conditions. Growth on glycerol-fumarate indicates function of the fumarate respiratory system, including intact fumarate/succinate antiport. The strains were classified according to their growth responses (Fig. 5). Most of the mutant strains grew with rates comparable with the wild type (class I mutants, 32 of 37 mutants). A small number (class III, three mutants) had lost growth similar to dcuA dcuB dcuC triple mutants, whereas class II mutants (two mutants) were capable of slow growth. When tested, in the class II mutants, the activity for succinate uptake was strongly decreased (e.g. mutant strain with DcuB(R83A) or decreased to background levels, as in the dcuA dcuB dcuC mutant (mutant with DcuB(E79A)) ( Table 6). The expression of dcuBЈ-ЈlacZ was repressed in the strains in the absence and stimulated in the presence of fumarate as in the wild type, demonstrating that the corresponding forms of DcuB lost the capacity for succinate uptake but still required fumarate for dcuBЈ-ЈlacZ induction. From the class III mutants the strains with DcuB(D119A) and DcuB(D405) were deficient in growth and uptake activity for [ 14 C]succinate. This is explained for mutant D405A by the low contents of DcuB (see Fig. 4) and for mutant D119A presumably by structural changes, as explained above. The class III mutant (DcuB(R127A)) showed an unexpected phenotype.
Despite a complete lack of growth on glycerol plus fumarate, the strain catalyzed uptake of [ 14 C]succinate with wild-type rates. Since other genes required for fumarate respiration (frdABCD) are not mutated, it is assumed that the mutant is able to catalyze uptake but has lost the capacity for antiport.

DcuB Is a Bifunctional Carrier Protein with Regulatory Function-
The C 4 -dicarboxylate/succinate antiporter DcuB is a bifunctional protein with a transport and a regulatory function. The regulatory function affects expression of DcuS-DcuR-regulated genes. The latter function was independent from its transport activity and from sites in DcuB that are essential for transport (residues Glu 79 , Arg 83 , and Arg 127 ). The regulatory function, on the other hand, required sites (Lys 353 , Thr 394 , and Asp 398 ) that were not essential for transport. The transport-specific residues are located in the N-terminal half of DcuB, whereas the regulatory residues concentrate in the other half close to the C-terminal end (Fig. 6). The topology model is hypothetical and based on the homology to DcuA (29). Topology models for DcuB differ considerably in the number and position of transmembrane helices, depending whether the model is based on that of the related DcuA carrier or in silico predictions, such as by the TMHMM server (32). Most of the regulation and transport competent residues are located in cytoplasmic or periplasmic loops. The topology of the residues is important for understanding their role in transport and regulation and should be clarified in future work. The transporter DcuB is due to the presence of regulatory and transport function, a bifunctional "trigger" enzyme or protein (for a review, see Ref. 31), which functions in metabolism and in controlling gene expression.
The sequence alignment of the C 4 -dicarboxylate transporters DcuA and DcuB (Fig. 3) demonstrates the significance of basic and acidic residues for the function of the transporters. The alignment shows 25 acidic (Asp or Glu) or basic (Arg, Lys, or His) amino acid residues that were conserved or exchanged by similar residues in most of the DcuB and DcuA proteins (at least 7 of 8 of the transporters). In addition, 17 further basic or acidic residues were specific and conserved only in DcuB and not in DcuA, including the regulatory Asp 398 residue. Some of the conserved basic residues of DcuB were replaced by acidic residues in DcuA, and vice versa. Remarkably, the essential transport-specific residues (Glu 79 , Arg 83 , and Asp 119 ) were  (29). The approximate positions of the residues that have no effect on the function of DcuB in regulation and anaerobic growth on fumarate are shown in gray. The positions of the amino acid residues that are important for regulation of DcuS-DcuR-dependent gene expression (Lys 353 , Thr 394 , and Asp 398 ) are indicated in red, and those affected by anaerobic growth on glycerol plus fumarate (E79A, R83A, and R127A) are shown in green. DcuB(D119A) and DcuB(D405) (blue circles) were produced in low amounts and had effects on transport and regulation. The position of peptide W 194 FRGKDLDKDEEFQ 207 C (sequence in italic type is derived from DcuB), which was used for production of polyclonal antiserum, is indicated (broken line, red). conserved in the DcuA and DcuB proteins. Residues Asp 398 and Lys 353 with regulatory function were conserved only in the DcuB proteins in agreement with the restriction of regulatory function to the latter. The regulatory residue Thr 394 is found in DcuB and DcuA, suggesting that it has an additional function that is required also in DcuA.
The acidic and basic amino acid residues that are essential for transport and regulation might be involved in the binding of the C 4 -dicarboxylates. Translocation of substrates by secondary transporters requires specific binding of the substrate at the binding site and in the translocation pathway of the transporter (33,34). The transport-essential residues might be part of the corresponding sites of DcuB for binding and translocation of the C 4 -dicarboxylates and of protons that are cosubstrates for transport (3). The residues that are essential for the regulatory function of DcuB would be suitable for binding of C 4 -dicarboxylates as well. Alternatively, it is feasible that the regulatory residues are involved in a specific interaction with a protein from the DcuS-DcuR two-component system that is required for exerting the regulatory effect of DcuB. The mode of interaction is not known, but it could involve protein-protein interaction between DcuB and the DcuS-DcuR system. Interestingly, the C 4 -dicarboxylate-and DcuS-DcuR-dependent stimulation of dctA expression became constitutive also in a mutant deficient in dctA (35), indicating that DctA might affect DcuS-DcuR-dependent expression during aerobic growth in a similar way as DcuB during anaerobic growth. It will be interesting to see whether an interaction between DcuB and DcuS-DcuR can be demonstrated. However, DcuB could not be isolated so far in significant amounts for in vitro studies, 3 whereas the DcuS-DcuR proteins can be isolated and reconstituted functionally (12,14,15).
Regulation by DcuB means that the DcuS-DcuR system has a second signal input site in addition to the DcuS sensor. What could be the physiological significance for dual regulation of the DcuS-DcuR system by the same stimulus? DcuS responds to C 4 -dicarboxylate concentration in the periplasm by the periplasmic sensing domain (13,16,36), which appears to be the primary sensory site of DcuS. The cytoplasmic PAS domain of DcuS plays an important role in signal transduction of the periplasmically derived stimulus to the kinase domain (18). The DcuB transporter, on the other hand, senses C 4 -dicarboxylate transport and metabolism. Metabolic flux might represent a more direct measure for the status of C 4 -dicarboxylate metabolism than concentration measurement in the periplasm. A transporter catalyzing the first step of metabolism, therefore, is in an optimal position to detect relevant metabolic conditions. Therefore, both sensory devices respond to alternative perspectives of the same regulatory signal.
Transporters as Sensory Devices for Histidine Kinases and Other Regulatory Systems-Transporters are able to participate in various modes in sensing and signal perception in transcriptional regulation, but there are only few systems where regulation by transporters has been studied at the molecular level. The sugar transporter of the glucose phosphotransferase sys-tem represents a signal input site for catabolite control in E. coli. Enzyme II (EII Gluc ) of the transport system controls the activity of adenylate cyclase, which then synthesizes the intramolecular signaling molecule cyclic AMP (for a review, see Ref. 37). It is feasible that DcuB interacts with DcuS-DcuR in order to control the sensitivity of the system. Regulation by direct interaction has been shown for the membrane-integrated sensor/regulator CadC, which controls the synthesis of lysine decarboxylase. The function of CadC is regulated by direct interaction with the lysine permease LysP (38).
There are also two-component systems responding to transmembrane carriers. The UhpB-UhpA two-component system of E. coli controls the expression of the glucose 6-phosphate uptake transporter UhpT, using the accessory membrane protein UhpC for sensing. UhpC is supposed to represent a former secondary carrier. UhpC stimulates the sensor kinase after binding of glucose 6-phosphate (39,40). A system similar to the DcuB/DcuS-DcuR system is represented by DctB-DctD of Rhizobium meliloti. The DctB-DctD two-component system controls the expression of dctA encoding the C 4 -dicarboxylate carrier DctA (41,42). DctA is assumed to affect the DctB sensor kinase by direct interaction when it is in the transport-inactive (or C 4 -dicarboxylate-deficient) form.
Transporters, and in particular secondary carriers, represent an interesting device for sensing of stimuli by histidine kinases, but their function in sensing and controlling the kinase activity is generally not clear. Therefore, studies on DcuB of E. coli might support our understanding of how secondary carriers can interact with and control the function of sensor kinases in two-component systems and how transporter proteins can function as additional sensory sites of such systems.