Fumarate Regulation of Gene Expression in Escherichia coli by the DcuSR (dcuSR Genes) Two-Component Regulatory System

Abstract

In Escherichia coli the genes encoding the anaerobic fumarate respiratory system are transcriptionally regulated by C₄-dicarboxylates. The regulation is effected by a two-component regulatory system, DcuSR, consisting of a sensory histidine kinase (DcuS) and a response regulator (DcuR). DcuS and DcuR are encoded by the dcuSR genes (previously yjdHG) at 93.7 min on the calculated E. coli map. Inactivation of the dcuR and dcuS genes caused the loss of C₄-dicarboxylate-stimulated synthesis of fumarate reductase (frdABCD genes) and of the anaerobic fumarate-succinate antiporter DcuB (dcuB gene). DcuS is predicted to contain a large periplasmic domain as the supposed site for C₄-dicarboxylate sensing. Regulation by DcuR and DcuS responded to the presence of the C₄-dicarboxylates fumarate, succinate, malate, aspartate, tartrate, and maleate. Since maleate is not taken up by the bacteria under these conditions, the carboxylates presumably act from without. Genes of the aerobic C₄-dicarboxylate pathway encoding succinate dehydrogenase (sdhCDAB) and the aerobic succinate carrier (dctA) are only marginally or negatively regulated by the DcuSR system. The CitAB two-component regulatory system, which is highly similar to DcuSR, had no effect on C₄-dicarboxylate regulation of any of the genes.

In Escherichia coli the switch from aerobic to anaerobic metabolism is regulated at the transcriptional level in response to the presence of the electron acceptors O₂, nitrate, and fumarate (8, 9, 11, 25, 27, 28). This regulation ensures that in the presence of oxygen only aerobic metabolism and not anaerobic respiration or fermentation is functional. Under anoxic conditions, nitrate (and nitrite) represses the synthesis of the enzymes associated with fumarate respiration. The sensor-regulator systems controlling gene expression in response to O₂ and nitrate are known and have been studied in detail. Regulation by O₂ is effected by the two-component regulatory system ArcB/A (aerobic respiratory control) and by the cytoplasmic one-component regulator FNR (fumarate-nitrate reductase regulator) (8, 11, 27). Nitrate and nitrite regulate via two homologous two-component regulatory systems, NarX/L and NarP/Q (Nar is an acronym for nitrate reductase) (25).

Fumarate is also an important electron acceptor for respiration, and fumarate and related C₄-dicarboxylates are known to induce a variety of genes required for anaerobic fumarate metabolism, such as the structural genes for fumarate reductase (frdABCD) (8, 12), the proton-pumping NADH dehydrogenase I (nuoA to -N) (3, 26, 28), and dicarboxylate carriers (dcu genes) (7, 24, 29). In aerobic growth, synthesis of succinate dehydrogenase (sdhCDAB) is stimulated by the same substrates (18). Therefore, there is a large group of genes which should be transcriptionally regulated by fumarate or other C₄-dicarboxylates. For Rhizobium leguminosarum and Rhodobacter capsulatus, the two-component sensor-regulators, DctSR and DctBD, which control gene expression in response to C₄-dicarboxylates are known (10, 21). In the present study a two-component regulatory system was identified in E. coli. It is responsible for regulation of the genes of fumarate respiration, including fumarate reductase and a fumarate carrier (DcuB), in response to the presence of C₄-dicarboxylates.

MATERIALS AND METHODS

Bacterial strains and growth.

For genetic experiments the bacteria (Table 1) were grown aerobically in Luria Bertani broth (22). For expression studies the bacteria were grown in M9 mineral medium (15) supplemented with acid-hydrolyzed casein (1 g/liter) (26). Anaerobic growth was performed in gastight stoppered tubes under an atmosphere of N₂ (3). Aerobic growth was performed in flasks filled to 5% of the maximal volume with vigorous shaking. For anaerobic growth the carbon sources were added at 20 mM, and for aerobic growth the carbon sources were added at 10 mM. Cell densities were measured as the absorbance at 578 nm. Cells were harvested at an A₅₇₈ of 0.5 to 0.7. β-Galactosidase assays were performed according to Miller (15).

TABLE 1.

Strains of E. coli and plasmids used

Bacterial strain or plasmid	Genotype	Reference or source
E. coli K-12
MC4100	F−araD139 Δ(argF-lac)U169 rpsL150 relA1 flbB530 deoC1 ptsF25 rbsR	23
JC7623	recB21 recC22 sbcB15 leu his thr pro arg ara	16
IMW205	MC4100 but dcuR::Kanr	This study
IMW262	MC4100 but dcuS::Camr	This study
IMW220	MC4100 but citB::Spcr	This study
MC4100λJ100	MC4100 λ [Φ(frdA′-′lacZ)]	12
IMW206	MC4100 λ [Φ(frdA′-′lacZ)] but dcuR::Kanr	IMW205(P1) × MC4100λJ100
IMW216	MC4100 λ [Φ(frdA′-′lacZ)] but citB::Spcr	IMW220(P1) × MC4100λJ100
MC4100λPC33	MC4100 λ [Φ(sdhC′-′lacZ)]	18
IMW211	MC4100 λ [Φ(sdhC′-′lacZ)] but dcuR::Kanr	IMW205(P1) × MC4100λPC33
IMW33	MC4100 λ [Φ(nuo′-′lacZ)]	3
IMW207	MC4100 λ [Φ(nuo′-′lacZ)] but dcuR::Kanr	IMW205(P1) × IMW33
IMW237	MC4100 λ [Φ(dcuB′-′lacZ)]	This study
IMW238	MC4100 λ [Φ(dcuB′-′lacZ)] but dcuR::Kanr	IMW205(P1) × IMW237
IMW239	MC4100 λ [Φ(dcuB′-′lacZ)] but citB::Spcr	IMW220(P1) × IMW237
IMW240	MC4100 λ [Φ(dcuC′-′lacZ)]	This study
IMW241	MC4100 λ [Φ(dcuC′-′lacZ)] but dcuR::Kanr	IMW205(P1) × IMW240
IMW260	MC4100 λ [Φ(dcuB′-′lacZ)] but dcuS::Camr	IMW262(P1) × IMW237
IMW261	MC4100 λ [Φ(frdA′-′lacZ)] but dcuS::Camr	IMW262(P1) × MC4100λJ100
Plasmids
pJL28	′lacZ, protein fusion vector (Apr)	13
pJL29	′lacZ, protein fusion vector (Apr)	13
pMW75	pKS− but dcuR::Kanr	This study
pMW92	pKS− but citB::Spcr	This study
pMW108	pKS− but dcuS::Camr	This study
pMW99	pJL29 but dcuB′-′lacZ	This study
pMW98	pJL29 but dcuC′-′lacZ	This study
pMW103	pJL28 but dctA′-′lacZ	This study

Inactivation of dcuR (yjdG), dcuS (yjdH), and citB.

The genes were inactivated by replacing their central portions with resistance cassettes. The flanking regions upstream and downstream of the genes were amplified by PCR. The downstream region of dcuR was amplified with primers yjdG-Hin (5′-TGA CAT CAA GAC CGC CCG AAG CTT GCA AGG-3′) and yjdG-Eco (5′-GCG TCC AGT TTA CCG TTA CCG AAT TCA GGC-3′), generating a 848-bp fragment with flanking HindIII and EcoRI sites. The upstream region of dcuR was amplified with primers yjdG-Pst (5′-TGT TCG TTG GAG CTG CAG CCG TGG ATT AGC-3′) and yjdH-Xba (5′-CAG TGA AAG CCA GCT TCT AGA CAG CGG CAG-3′), producing a 815-bp fragment with flanking PstI and XbaI sites. The flanking region upstream of dcuS was amplified with primer YjdH-Eco (5′-CTC TCT GCG AAT TCT TTG TGC ATC-3′), introducing an EcoRI site, and primer YjdH-Bam-2 (5′-CTT CAG GAT CCG AGT AGC GAA GAC-3′), introducing a BamHI site, generating a 1,091-bp fragment. The downstream flanking region of dcuS was amplified with primer YjdH-Xba (5′-TGA GCG CCT CTA GAA AGC GGG AAG-3′), with a XbaI site, and primer YjdH-Bam-1 (5′-GGC GTT ATC ATC GGA TCC ATT TC-3′), with another BamHI site, generating a 1,020-bp fragment. The upstream region of citB was amplified with primers cri-Sac (5′-AAG ATG CTG GGG CTG AGC TCC-3′) and cri-Bam (5′-ATT CCG CAT GGA TCC CTG CC-3′), generating a 929-bp fragment with SacI and BamHI cloning sites. The downstream region of citB was amplified with primers cri-Hind (5′-ATG TTT AAA GCT TAT GCT CGC G-3′) and cri-Cla (5′ GAT CAT CGG TGT ATC GAT TTT TG-3′), producing a 918-bp fragment with HindIII and ClaI cloning sites. For each gene, the flanking regions were cloned into pKS⁻ (Stratagene). For the dcuR and citB genes the Kan^r and Spc^r resistance cassettes derived from pGS607 and pGS606, respectively (24), were cloned into the EcoRI-PstI and the BamHI sites, respectively, resulting in dcuR::Kan^r (pMW75) and citB::Spc^r (pMW92). For dcuS, the flanking regions were separated by a single BamHI site (pMW107). A Cam^r resistance cassette was amplified from pACYC184 (6) by PCR with primers CAMLIB2 (5′-CAA TAA CTG GAT CCA AAA AAT TAC GC-3′) and CAMREB (5′-ATA TCC TGG ATC CCA TAT TCT GC-3′), both introducing a BamHI site. The resistance cassette was then cloned into the BamHI site of pMW107, resulting in pMW108 (dcuS::Cam^r). Any possible terminating sequences downstream of the Cam^r resistance cassette were removed to enable transcription of dcuR located downstream of dcuS. The plasmids were transformed into E. coli JC7623 and were used for replacement of the intact genes by homologous recombination (16). Presence of the dcuR::Kan^r, dcuS::Cam^r, and citB::Spc^r alleles was confirmed by PCR of the genomic DNA with the corresponding primers, yielding fragments corresponding to the sizes of the inactivated genes. The inactivated genes were transferred to strains with suitable genetic backgrounds by P1 transduction (15).

Construction of protein fusions.

For creating protein fusions (dcuB′-′lacZ, dcuC′-′lacZ, and dctA′-′lacZ) plasmid pJL28 or its derivative pJL29 was used (13). The dcuB′-′lacZ fusion was obtained by cloning the 0.65-kb PCR fragment generated with primer dcuB-Bam (5′-AAG TTG GAT CCT AAA TAA CAT GTG TGA ACC-3′) and primer yjdG-Eco into the BamHI and EcoRI sites of pJL29, yielding pMW99. For the dcuC′-′lacZ fusion (pMW98), the dcuB promoter region was amplified with primers dcuC-Bam (5′-CCC CAA TAA GGA TCC CAA TG-3′) and dcuC-Eco (5′-CCA GCG GTG AAT TCC AGA CC-3′), and the 1.1-kb fragment was cloned into the BamHI and EcoRI sites of pJL29. The dctA′-′lacZ fusion (pMW103) was obtained by cloning the 0.5-kb PCR fragment generated with primers dctA-Bam (5′-CAG AGA GGG ATC CAT AGG GTG TCC-3′) and dctA-Eco (5′-CGC TGG ATG AAT TCG GCA TGG G-3′) into the respective restriction sites of pJL28. The dcuB′- and dcuC′-′lacZ fusions were transferred to the chromosome with phage λRZ5 (12, 17), and monolysogens were identified and used for further work (3).

RESULTS

Fumarate induction of dcuB and frdA depends on the dcuSR regulatory genes.

In a search for potential fumarate-responsive regulators, the E. coli data base was screened for gene products similar to the sensor-regulators DctRS and DctBD of R. capsulatus and Rhizobium leguminosarum, which stimulate the synthesis of the C₄-dicarboxylate carriers in response to C₄-dicarboxylates (10, 21). Both systems showed only low levels of similarity to two-component regulators of E. coli (<28% sequence identity). The genes for two of these systems, yjdHG and citAB, were located next to genes involved in anaerobic fumarate metabolism (Fig. 1). The yjdHG genes are in the dcuB fumB to lysU intergenic region at 93.7 min on the E. coli map (2). The dcuB fumB genes encode the anaerobically expressed fumarate carrier (dcuB) and fumarase (fumB) (1, 24). The citA citB (formerly criR) genes on the other hand are positioned at 14.1 min on the E. coli map between genes encoding an alternative fumarate carrier (dcuC) and the citC to citT gene cluster for anaerobic citrate metabolism (2, 20, 29). The citAB genes encode proteins homologous to the citrate sensor-regulators from Klebsiella pneumoniae (4, 5, 20). Anaerobic citrate metabolism of E. coli is related to C₄-dicarboxylate metabolism due to the production and excretion of succinate (5, 14).

FIG. 1.

Map positions and arrangement of the dcuSR (previously yjdHG) (A) and citAB (B) genes on the E. coli genome. The scale gives the DNA length in kilobases. The positions of the dcuSR and the citAB genes on the calculated E. coli map are shown. Data are from reference 2 and the E. coli data bank.

The genes for the putative response regulator yjdG and the sensor kinase yjdH were genetically inactivated by replacement with genes carrying resistance cassettes. The mutant strains were tested for fumarate regulation of the dcuB and frdA genes (Table 2). Expression of the genes was determined with lacZ fusions, and growth was performed under anaerobic conditions on glucose or glycerol plus dimethyl sulfoxide (DMSO). DMSO has to be included as an electron acceptor for growth on glycerol, which cannot be fermented by E. coli. In the wild type, the expression of dcuB was stimulated 5.6-fold or 10.9-fold after growth on glucose or glycerol, respectively, when fumarate was present in the medium. When DMSO was omitted from the glycerol medium, a similar stimulation was found with the addition of fumarate. The lower expression of dcuB during growth on glucose could be due to glucose repression. In the yjdG (dcuR) and yjdH (dcuS) mutants the expression of dcuB was decreased to background levels, and the expression was not stimulated by fumarate (Table 2). Therefore both genes are required for fumarate stimulation of dcuB expression.

TABLE 2.

Regulation of dcuB and frdA expression by fumarate, dcuR (formerly yjdG), and dcuS (formerly yjdH) under anaerobic conditions^a

Substrate(s)	Regulation of dcuB′-′lacZ expression by:			Regulation of frdA′-′lacZ expression by:
	WTb	dcuR	dcuS	WT	dcuR	dcuS
Glucose	8	5	2	150	130	130
Glucose + fumarate	45	3	1	300	155	150
Glycerol + DMSO	48	6	4	690	580	365
Glycerol + DMSO + fumarate	520	25	9	NDc	ND	ND
Glycerol + fumarate	560	7	18	1,010	400	430

^aExpression of dcuB and frdA is given in Miller units. Growth occurred in M9 medium with the indicated substrates. 

^bWT, wild type. 

^cND, not determined. 

Expression of frdA is stimulated by the presence of fumarate in the medium, too, but this stimulation is lower (about twofold [Table 2]). In the yjdG (dcuR) and yjdH (dcuS) mutants background expression of frdA was still high, but the fumarate-dependent stimulation was lost completely.

The citB gene encoding the response regulator of the second two-component system (CitAB) was inactivated, too. The inactivation of citB, however, had no effect on expression and fumarate stimulation of the dcuB and frdA genes (not shown). Therefore, the yjdHG genes, but not the citB gene, are required for fumarate stimulation of dcuB and frdA expression. For this reason the genes yjdH and yjdG were termed dcuS (sensor kinase) and dcuR (response regulator).

Genes regulated by DcuR: not all C₄-dicarboxylate-regulated genes respond to DcuSR.

Other genes which are transcriptionally regulated by C₄-dicarboxylates were tested in the same way for dcuR involvement (Table 3). The genes tested encode the proton-pumping NADH dehydrogenase I (nuoA to -N genes), an alternative anaerobic C₄-dicarboxylate carrier (dcuC gene), succinate dehydrogenase (sdhCDAB), and a C₄-dicarboxylate carrier for aerobic growth (dctA). The increase in the expression of the genes stimulated by fumarate or succinate was between 1.4- (nuoAB′-′lacZ) to 2.8-fold (dctA′-′lacZ) (Table 3). However, the fumarate- or succinate-dependent stimulation of nuoA, dcuC, and sdhC expression was not significantly affected in the dcuR mutant. Expression of the dctA′-′lacZ fusion was decreased in the dcuR mutant, but the succinate stimulation was retained and the increases were similar for the wild type (2.8-fold) and the mutant (3.2-fold). Thus, from the genes tested, only dcuB and frdA were clearly regulated by DcuR and DcuS. In the citB mutant neither of the genes was affected in C₄-dicarboxylate-stimulated expression (not shown).

TABLE 3.

Effects of C₄-dicarboxylates and DcuR on the expression of C₄-dicarboxylate-regulated genes

Gene fusiona	Substrate(s)c	β-Galactosidase activity (Miller units)
		WTd (dcuR+)	dcuR
nuoAB′-′lacZ	Glucose	95	96
	Glucose + fumarate	130	125
	Glycerol + fumarate	230	250
dcuC′-′lacZ	Glucose	52	52
	Glucose + fumarate	100	100
sdhC′-′lacZ	Glycerol + O2	3,580	3,680
	Succinate + O2	4,780	4,740
dctA′-′lacZb	Glycerol + O2	260	89
	Succinate + O2	730	286

^aMC4100 derivatives with dcuR⁺ and dcuR backgrounds given in Table 1. 

^bStrains MC4100pMW103 and IMW205pMW103, respectively. 

^cM9 medium was used with the indicated substrates. 

^dWT, wild type. 

C₄-dicarboxylates affecting regulation by DcuR.

The effects of various carboxylates on the expression of dcuB′-′lacZ were studied by including the respective substrates in the medium (Table 4). Growth was performed under anaerobic conditions in the presence of glycerol plus DMSO, which enables high expression of dcuB when suitable carboxylates are added (see Tables 2 and 4). Each of the C₄-dicarboxylates fumarate, succinate, malate, tartrate, aspartate, and maleate caused a strong induction of dcuB expression compared to growth with glycerol plus DMSO alone. Even with succinate and maleate, which are not metabolized under the respective (anaerobic) conditions, the induction amounted to at least 63% of the maximal induction found with fumarate. Most remarkably, maleate, which is not even taken up by the anaerobic Dcu carriers (24), induced the expression of dcuB strongly. For all the C₄-dicarboxylates the stimulation was completely lost in the dcuR mutant. Therefore neither uptake nor metabolism of the C₄-dicarboxylates is required for induction by the DcuSR system. The results suggest that the C₄-dicarboxylates bind to the sensor at the periplasmic aspect of the membrane and that the sensor is able to react with each of the C₄-dicarboxylates. Butyrate and acetate, on the other hand, had no stimulating effect. During anaerobic growth on glucose, the respective C₄-dicarboxylates and aspartate showed similar stimulating effects, but expression was generally lower, possibly due to glucose repression (not shown). Expression of frdA′-′lacZ responded in a similar way to the C₄-dicarboxylates (not shown).

TABLE 4.

Effectors for dcuR-dependent regulation of dcuB′-′lacZ expression during anaerobic growth with glycerol plus DMSO

Carboxylic acid in mediuma	dcuB′-′lacZ expression (Miller units)
	IMW237 (dcuR+)	IMW238 (dcuR)
None	48	6
Fumarate	537	7
Succinate	437	4
Malate	435	4
Tartrate	382	13
Aspartate	434	10
Maleate	337	19
Butyrate	18	5
Acetate	26	3

^aM9 medium was used as the growth medium. 

DISCUSSION

Physiology and significance of fumarate regulation in E. coli.

Transcriptional regulation by fumarate and other C₄-dicarboxylates plays an important role in E. coli. The DcuSR two-component regulators identified here apparently exert this fumarate regulation for the genes of fumarate respiration, that is, frdABCD and dcuB. The expression of fumB (encoding anaerobic fumarase B), which is located downstream of dcuB and is possibly expressed from the dcuB promoter (1, 24), could also be subject to DcuR regulation. Expression of other genes which are transcriptionally stimulated by C₄-dicarboxylates was not (nuoA to -N, sdhCDAB, and dcuC) or was only partially (dctA) dependent on DcuR. This indicates that DcuSR is required specifically for the regulation of the anaerobic fumarate respiratory pathway. The C₄-dicarboxylate regulation of other genes apparently is effected by a different system, and the CitA-CitB two-component regulatory system obviously does not serve this function either, as shown here.

DcuRS as a C₄-dicarboxylate-sensing two-component system.

The dcuSR genes, and the derived DcuS and DcuR proteins, show the typical properties of two-component regulatory systems. Both genes overlap by 4 bp, which is a strong indication for a joint transcription similar to that of the genes of other two-component regulators, which are mostly organized in one transcriptional unit. The DcuR protein contains a helix-turn-helix DNA-binding motif in the C-terminal half and an N-terminal receiver domain with a conserved aspartate residue (Asp56) as a potential phosphorylation site.

The DcuS protein contains the elements typical for sensory histidine kinases, and the arrangement is very similar to that found in the CitA protein of K. pneumoniae (4, 5) (Fig. 2). The CitA protein consists of an N-terminal sensory domain with two transmembrane helices which are separated by a long periplasmic domain of about 130 amino acids (5). The kinase domain is separated from the sensor domain by an extra domain of about 80 amino acids. The similarity of DcuS to CitA extends over the complete range, including the periplasmic and the extra domain. In the kinase domain the H, N, F, and G boxes, which are designated according to the characteristic amino acid residues (19), are present in an arrangement very similar to that of CitA. His349, which is supposed to be the phosphorylation site, is conserved in the H box.

FIG. 2.

Overview of the suggested domain structure of the sensor kinase DcuS. The positions of characteristic sequence features and of the domains are indicated by the numbers. The transmembrane helices (TM) were predicted from the sequence. The Q linkers (Q) and the signature segments of the kinase domain (H, N, G1, F, and G2) (19) were identified by sequence alignments. For the segments, the positions of the naming amino acid residue are given (drawn according to reference 5).

DcuS has significantly higher levels of similarity with the CitA citrate sensors of K. pneumoniae and E. coli than with the C₄-dicarboxylate sensors DctB and DctS of Rhizobium sp. strains and R. capsulatus (not shown). Both the citrate (CitA) and the C₄-dicarboxylate (DcuS, DctB, and DctS) sensors have similar N-terminal sensory domains consisting of two transmembrane helices and a long intervening periplasmic domain. The periplasmic domains of DctB and DctS, however, are about twice the size of the CitA or DcuS periplasmic domain, which presumably acts in ligand binding (5). Such a location of the sensory domain in the periplasm suggests sensing of the C₄-dicarboxylates from without. This is also supported from the functioning of maleate as a signal which apparently is not taken up by the bacteria (24). In agreement with the postulated fumarate sensing by DcuSR from without, the fumarate carrier (DcuB) shows a strong induction by fumarate (up to 10.9-fold), whereas fumarate reductase shows only a weak induction by fumarate (up to twofold). These different responses to external fumarate appear to be sensible since DcuB is required in particular when external fumarate is present. FrdA on the other hand is also required when internal fumarate is produced from intermediary metabolism.