A Mutation in the 3-Phosphoglycerate Kinase Gene Allows Anaerobic Growth of Bacillus subtilis in the Absence of ResE Kinase

Abstract

The Bacillus subtilis ResD-ResE two-component signal transduction system is essential for aerobic and anaerobic respiration. A spontaneous suppressor mutant that expresses ResD-controlled genes and grows anaerobically in the absence of the ResE histidine kinase was isolated. In addition, aerobic expression of ResD-controlled genes in the suppressed strain was constitutive and occurred at a much higher level than that observed in the wild-type strain. The suppressing mutation, which mapped to pgk, the gene encoding 3-phosphoglycerate kinase, failed to suppress a resD mutation, suggesting that the suppressing mutation creates a pathway for phosphorylation of the response regulator, ResD, which is independent of the cognate sensor kinase, ResE. The pgk-1 mutant exhibited very low but measurable 3-phosphoglycerate kinase activity compared to the wild-type strain. The results suggest that accumulation of a glycolytic intermediate, probably 1,3-diphosphoglycerate, is responsible for the observed effect of the pgk-1 mutation on anaerobiosis of resE mutant cells.

Many bacteria including Bacillus subtilis utilize two-component signal transduction systems to sense and respond to a variety of environmental changes. Most but not all response regulators of the two-component family function as transcriptional regulators of genes that must be activated or repressed to achieve an efficient adaptive response. The ResD-ResE two-component signal transduction system in B. subtilis plays an important role in aerobic and anaerobic respiration (33, 42) as well as in the activation of the phosphate (Pho) regulon (41). The combination of the response regulator, ResD, and the cognate histidine kinase, ResE, is required for one of the two major induction pathways for Pho regulation, i.e., AbrB-independent activation of the phoPR operon (41). PhoP and PhoR constitute a two-component regulatory system in which PhoP, a response regulator, receives phosphate from the sensor kinase, PhoR, and then activates the transcription of Pho regulon genes (16). ResD and ResE are also indispensable for transcriptional activation of genes involved in aerobic respiration, such as ctaA (required for heme A synthesis), the petCBD operon (encoding subunits of the cytochrome bf complex), and resABC (42). The essential resABC operon encodes proteins similar to cytochrome c biogenesis proteins (42). The resDE genes, located downstream of resABC, are transcribed primarily from the resA promoter. Thus, the ResDE signal transduction system forms an autoregulatory circuit (42). In addition to its role in aerobic respiration, ResDE positively regulates anaerobically induced genes. These include the anaerobic regulator gene fnr (33), the nitrite reductase genes nasDEF (27), and the flavohemoglobin gene hmp (20). The other mode of anaerobic energy production, fermentation, was shown to be only moderately affected by a resD mutation, and no significant effect of a resE mutation compared to an otherwise wild-type strain was observed (26).

Important questions about how ResD functions in gene regulation remain unanswered. First, it is not known if phosphorylated ResD activates some or all of the genes listed above by direct interaction with their regulatory regions. Second, it is not known why some ResDE-controlled genes are induced by oxygen limitation while others are expressed under aerobic conditions. Answers to these questions will give insights into how the ResDE signal transduction pathway functions in respiration and how B. subtilis cells choose between aerobic and anaerobic respiration pathways. In studies reported herein, we isolated a mutant strain of B. subtilis that grows anaerobically in the absence of ResE. Our results suggest that accumulation of 1,3-diphosphoglycerate, a glycolytic intermediate, may lead to ResE-independent activation of ResD.

MATERIALS AND METHODS

Bacterial strains.

The B. subtilis strains and plasmids used in this study are listed in Table 1. The construction of ΔresE and ΔresDE mutations was reported previously (26, 33, 42).

TABLE 1.

B. subtilis strains and plasmids used in this study

Strain or plasmid	Genotype or relevant phenotype	Reference
B. subtilis strains
JH642	trpC2 pheA1	J. A. Hoch
BD1853	met leu his ΔcomP neo	D. Dubnau
MH5201	trpC2 pheA1 amyE::pES17 (resA-lacZ) cat	42
ZB307A	SPβc2del2::Tn917::pSK1Δ6	48
OKB105	pheA1 sfp	30
LAB364	trpC2 pheA1 SPβc2del2::Tn917::pXL5	This study
LAB849	trpC2 pheA1 ΔcomP::neo	This study
LAB950	trpC2 pheA1 ΔcomP::neo SPβc2del2::Tn917::pXL5	This study
LAB2135	trpC2 pheA1 ΔresDE::tet	33
LAB2234	trpC2 pheA1 ΔresE::spc	26
LAB2252	trpC2 pheA1 SPβc2del2::Tn917::pMMN288	33
LAB2266	trpC2 pheA1 ΔresE::spc SPβc2del2::Tn917::pMMN288	This study
LAB2527	trpC2 pheA1 ΔresE::spc pgk-1a	This study
LAB2536	trpC2 pheA1 ΔresE::spc pgk-1 SPβc2del2::Tn917::pMMN288	This study
LAB2537	trpC2 pheA1 amyE::pES17 (resA-lacZ) cat	This study
LAB2540	trpC2 pheA1 ΔresE::spc amyE::pES17 (resA-lacZ) cat	This study
LAB2543	trpC2 pheA1 ΔresE::spc pgk-1 amyE::pES17 (resA-lacZ) cat	This study
LAB2545	trpC2 pheA1 ΔresDE::cat pgk-1	This study
LAB2555	trpC2 pheA1 ΔresDE::cat pgk-1 SPβc2del2::Tn917::pMMN288	This study
LAB2556	trpC2 pheA1 ΔresDE::cat pgk-1 amyE::pES17 (resA-lacZ) cat	This study
LAB2779	trpC2 pheA1 ΔcomP::neo pgk-1	This study
LAB2780	trpC2 pheA1 ΔcomP::neo SPβc2del2::Tn917::pXL5	This study
LAB2846	trpC2 pheA1 ΔresE::spc SPβc2del2::Tn917::pMMN387	This study
LAB2847	trpC2 pheA1 ΔresE::spc SPβc2del2::Tn917::pMMN388	This study
LAB2848	trpC2 pheA1 ΔresE::spc pgk-1 SPβc2del2::Tn917::pMMN387	This study
LAB2849	trpC2 pheA1 ΔresE::spc pgk-1 SPβc2del2::Tn917::pMMN388	This study
LAB2864	trpC2 pheA1 ΔresE::spc pgk-1 amyE::pES17 (resA-lacZ) cat SPβc2del2::Tn917::pMMN388	This study
LAB2898	trpC2 pheA1 thr::pMMN403 (fnr-lacZ) erm	This study
LAB2901	trpC2 pheA1 ΔresE::spc thr::pMMN403 (fnr-lacZ) erm	This study
LAB2902	trpC2 pheA1 ΔresE::spc pgk-1 thrC2::pMMN403 (fnr-lacZ) erm	This study
LAB2904	trpC2 pheA1 ΔresE::spc pgk-1 thrC2::pMMN403 (fnr-lacZ) erm SPβc2del2::Tn917::pMMN388	This study
LAB3014	trpC2 pheA1 gapA::neo	This study
LAB3015	trpC2 pheA1 gapB::erm	This study
LAB3023	trpC2 pheA1 gapA::neo SPβc2del2::Tn917::pMMN288	This study
LAB3024	trpC2 pheA1 gapB::erm SPβc2del2::Tn917::pMMN288	This study
LAB3034	pheA1 pgk-1 SPβc2del2::Tn917::pMMN288	This study
LAB3044	trpC2 pheA1 ΔresE::spc pgk-1 gapB::erm	This study
LAB3045	trpC2 pheA1 ΔresE::spc pgk-1 gapA::neo	This study
ORB3060	trpC2 pheA1 ΔresE::spc pgk-1 gapB::erm SPβc2del2::Tn917::pMMN288	This study
ORB3061	trpC2 pheA1 ΔresE::spc pgk-1 gapA::neo SPβc2del2::Tn917::pMMN288	This study
ORB3071	trpC2 pheA1 Δpgk::neo	This study
ORB3072	SPβc2del2::Tn917::pSK1 Δpgk::neo	This study
ORB3074	pheA1 Δpgk	This study
ORB3075	trpC2 pheA1 ΔresE::spc pgk-1 Δpgm::erm	This study
ORB3076	pheA1 ΔresE::spc Δpgk	This study
ORB3077	trpC2 pheA1 Δpgm::erm	This study
MAB113	pheA1 pgk-1	This study
MAB117	pheA1 pgk-1 amyE::pES17 (resA-lacZ) cat	This study
MAB128	pheA1 pgk-1 pPS34 integrated near pgk	This study
MAB129	pheA1 pPS34 integrated near pgk	This study
Plasmids
pDG782	Ampr Nmr	13
pDG792	Ampr Nmr	13
pDG793	Ampr Ermr, promoterless lacZ	14
pJDC9	Ermr	5
pUC18tet	Ampr Tetr, pUC18 with tet	F. M. Hulett
pMMN7	Ampr Cmr	32
pMMN13	Ampr Cmr	29
pMMN288	Ampr Cmr, fnr-lacZ	33
pMMN326	Ampr Ermr, pPS34 carrying pgk region with pgk-1	This study
pMMN327	Ampr Ermr, pPS34 carrying pgk region	This study
pMMN329	Ampr Ermr, pPS34 carrying gapA region	This study
pMMN387	Ampr Cmr, pMMN13 carrying pgk region with pgk-1	This study
pMMN388	Ampr Cmr, pMMN13 carrying pgk region	This study
pMMN402	Ampr Ermr, pMMN327 with pgk in-frame deletion	This study
pMMN403	Ampr Ermr, fnr-lacZ	This study
pMMN409	Ampr, pUC18 with gapB	This study
pMMN410	Ampr Ermr, pMMN409 with erm in gapB	This study
pMMN412	Ampr Ermr Tetr, pMMN410 with tet	This study
pMMN414	Ermr, pJDC9 with gapA	This study
pMMN416	Ermr Nmr, pMMN414 with neo in gapA	This study
pMMN421	Ampr Ermr Nmr, pMMN327 with neo in pgk	This study
pPS2019	Ampr Ermr, pUC19 carrying pgm (erm in pgm) and eno	21
pPS34	Ampr Ermr, pBluescript SK(−) with erm	37
pXL5	Ampr Cmr, srfA-lacZ	31
pYZ13	Ampr Ermr Cmr, pPS2019 with cat	This study

^aThe notations pgk-1 and sre-1 are equivalent (see the text). 

Culture conditions.

Aerobic and anaerobic cultures were grown under conditions described previously (34), except that aerobic cultures were grown in baffled flasks. Antibiotics were added to the following final concentrations: ampicillin, 25 μg/ml; chloramphenicol, 5 μg/ml; erythromycin plus lincomycin, 1 and 25 μg/ml, respectively; tetracycline, 12.5 μg/ml; spectinomycin, 75 μg/ml; and neomycin, 5 μg/ml.

Measurement of growth.

The aerobic growth phenotype on agar plates was detected with TSS minimal agar (11) supplemented with 0.5% glucose or 0.5% glucose plus 0.5% pyruvate. Cells were grown overnight on DS agar medium (30) supplemented with 0.5% glucose and 0.5% pyruvate. A single colony of each strain was then streaked onto TSS plates. A growth assay in liquid medium was performed as follows. Cells grown aerobically overnight in DS liquid medium supplemented with 0.5% glucose and 0.5% pyruvate were transferred to DS medium or DS medium supplemented with 0.5% glucose, 0.5% pyruvate, or both (for aerobic growth). For anaerobic growth, cells were transferred to DS medium supplemented with 0.1% glucose and 0.2% KNO₃. The starting optical density at 600 nm was 0.02, and growth was monitored by measuring the optical density at 600 nm.

Isolation of the sre-1 mutation.

To isolate suppressors of a resE mutation, LAB2234 (ΔresE::spc) was cultured overnight in 2× YT with 1% glucose and spectinomycin. A total of 10⁷ to 10⁸ cells were plated onto each LB agar plate, which was supplemented with 1% glucose, 0.2% KNO₃, and spectinomycin. The agar plates were incubated in an anaerobic jar under an atmosphere of H₂ and CO₂. Colonies formed under anaerobic conditions were streaked to isolate a single clone from which chromosomal DNA was prepared. The chromosomal DNA was used to transform JH642, and transformants were selected for spectinomycin resistance (Spc^r) conferred by the resE::spc mutation. A sre-1 transformant that grew anaerobically was isolated by cotransformation with the unlinked resE marker. This backcross was repeated twice, and a strain LAB2527 (ΔresE sre-1 [suppressor of resE]) was isolated after the second transformation.

A strain (LAB2545) that carries the ΔresDE::tet and sre-1 mutations was constructed as follows. LAB2527 (ΔresE::spc sre-1) cells were transformed with chromosomal DNA from LAB2135 (ΔresDE::tet), and a tetracycline-resistant (Tet^r) and Spc^s transformant, in which the resE deletion was replaced by the resDE deletion, was chosen. The presence of sre-1 in LAB2545 was confirmed by transforming LAB2545 cells with LAB2234 (ΔresE::spc) chromosomal DNA and by determining that the Spc^r Tet^s transformants grow anaerobically.

A sre-1 resE⁺ strain was constructed by cotransformation. LAB2527 cells were transformed with chromosomal DNA from OKB105 (pheA1) (30), and trp⁺ transformants were selected. The strain MAB113 (resE⁺ sre-1) was isolated by screening the transformants for Spc^s, indicating that the resE mutation was replaced with resE⁺ DNA from OKB105. When MAB113 was transformed with LAB2234 (ΔresE::spc) chromosomal DNA, Spc^r transformants grew anaerobically, confirming that MAB113 had retained the sre-1 mutation.

Construction of lacZ fusions and measurement of β-galactosidase activity.

Two fnr-lacZ fusions were used to examine fnr expression. The construction of the SPβ phage-borne fnr-lacZ fusion (SPβ::pMMN288) was reported previously (33), and the phage lysate was used to transduce various mutants. For the complementation analysis described below, fnr-lacZ that was inserted into the thrC locus was used instead of the phage-borne lacZ fusion, since the SPβ prophage locus was used for introduction of pgk in order to create the pgk/pgk-1 diploid strain. An EcoRI-HindIII fragment from pMMN288 that carries the fnr promoter was inserted into pDG793 (14) digested with EcoRI and HindIII. pDG793 has a spoVG Shine-Dalgarno sequence and a promoterless lacZ gene and is used to introduce a lacZ fusion at the thrC locus of the B. subtilis chromosome. The resultant plasmid, pMMN403, was used to transform JH642 with selection for erythromycin resistance (Erm^r). To obtain transformants resulting from a double-crossover recombination, threonine auxotrophs were sought and one (LAB2898) was chosen for further experiments. Various mutants carrying fnr-lacZ at thrC were constructed by transforming the mutants with chromosomal DNA isolated from LAB2898 and selecting for Erm^r.

A resA-lacZ fusion integrated at the amyE locus (42) was used to measure resA expression. MH5201 chromosomal DNA was used in transformation experiments to construct mutant strains carrying resA-lacZ.

To study srfA transcription, a previously constructed srfA-lacZ transcriptional fusion was used (31). The srfA promoter region contains the cis sequence required for ComP-ComA regulation. A comP deletion from BD1853 was introduced into MAB113 (sre-1) to generate strain LAB2779. ΔcomP (LAB950) and ΔcomP sre-1 (LAB2780) strains carrying srfA-lacZ were constructed by transducing LAB849 (ΔcomP) and LAB2779 (ΔcomP sre-1) strains with the SPβ phage carrying the srfA-lacZ fusion.

B. subtilis strains bearing lacZ fusions were grown aerobically or anaerobically in DS medium (30) supplemented with 0.1% glucose, 0.2% KNO₃, and appropriate antibiotics (the starting optical density at 600 nm was 0.02). Strains defective in nitrate respiration were able to grow anaerobically by fermentation. For example, ΔresE and ΔresDE mutants grew to an optical density at 600 nm of around 0.15 and 0.4, respectively, while wild-type strains grew to OD₆₀₀ = 1.0 to 1.2 under these culture conditions. β-Galactosidase activities were determined by the method of Miller (25) for culture samples taken at the indicated times during growth.

Cloning of sre.

The sre (pgk) gene (both sre-1 and sre⁺ alleles) was isolated as follows. Plasmid libraries were constructed by inserting B. subtilis chromosomal DNA fragments (approximately 0.45 kb) into the EcoRV site of pPS34 (erythromycin resistance plasmid; a derivative of pBluescript SK⁻ (Stratagene, La Jolla, Calif.) (37). JH642 cells were transformed with the plasmid libraries, and Erm^r transformants were obtained by integration of the plasmids into chromosomal DNA through homologous recombination. Chromosomal DNA prepared from the pooled transformants was used to transform MAB113 (sre-1) with selection for Erm^r. Loss of the sre-1 phenotype was observed as a change from the pale-colony phenotype of sre-1 mutants on DS agar medium (see Results). Seven Erm^r transformants exhibiting the sre⁺ phenotype were chosen, and chromosomal DNA prepared from these strains was used to retransform MAB113 to measure the linkage between the integrated plasmid and sre-1 in each case. Strains MAB128 (sre-1) and MAB129 (sre⁺) were derived by transformation with chromosomal DNA showing tight (up to 98%) linkage. Chromosomal DNA from MAB129 was digested with HindIII and then subjected to ligation at a low concentration of DNA to allow self-ligation. The ligation mixture was used to transform Escherichia coli DH5α competent cells. A resulting plasmid, pMMN327, carried a B. subtilis chromosomal region of about 2 kb from the sequence flanking the integrated plasmid. pMMN329, which was isolated by self-ligation of SacI-digested chromosomal DNA, carried a segment about 5 kb from the opposite flanking region. MAB128 chromosomal DNA digested with HindIII was used to isolate plasmid pMMN326 harboring the identical chromosomal DNA to that in pMMN327 but containing the sre-1 allele.

DNA sequencing of sre-1 and sre⁺.

B. subtilis chromosomal DNA in pMMN326 and pMMN327, which carry sre-1 and sre⁺, respectively, was sequenced by using reverse and T7 primers that hybridize to DNA located adjacent to the inserts. PCR primers were subsequently designed according to the sequences obtained above. These were used to amplify a series of segments of ca. 600 bp with overlaps of about 50 bp, which together covered the entire length of the sequence of interest. Either the −21M13 or the M13 reverse-primer sequence was added to the 5′ end of each primer used to carry out the PCRs, resulting in amplified products containing a −21M13 annealing site on one side and the M13 reverse-primer annealing site on the other. The PCR products were purified by polyethylene glycol precipitation, and sequencing was carried out with the dye primer cycle-sequencing kit of Perkin-Elmer Applied Biosystems. By this procedure, it was possible to sequence both strands of each fragment. The sequences adjacent to the cloned ends of the target DNA were amplified with the primer pair that anneals to the vector sequence, resulting in a product with the entire length of the fragment bearing the two dye primer-annealing sites derived from the vector sequences. This fragment was sequenced by the same method as the other 600-bp fragments.

Complementation analysis.

B. subtilis chromosomal DNA which carries pgk with or without the sre-1 mutation was isolated by digesting pMMN326 and pMMN327 with BamHI and HindIII. The isolated DNA fragments were inserted into an integration plasmid, pMMN13 (29), to generate pMMN387 (derived form pMMN326) and pMMN388 (from pMMN327). SPβ phages carrying pMMN387 and pMMN388 were constructed as previously described (48), and each lysate was used to transduce LAB2234 and LAB2527 cells.

Construction of a pgk in-frame mutation.

A pgk in-frame mutation was constructed by congression as follows. pMMN327 propagated in E. coli dam strain GM119 was cleaved with HpaI and BclI and then subjected to a fill-in reaction with T4 DNA polymerase. The plasmid DNA under went intramolecular ligation at low DNA concentration and then was used to transform E. coli DH5α. The resultant plasmid, pMMN402, was subjected to DNA sequencing to confirm that the pgk deletion created was in frame. The pgk mutation in pMMN402 resulted in loss of an internal 42-codon region (amino acids 217 to 258) encoding part of the C-terminal domain that functions in nucleotide binding (3). Next, the internal BclI-HpaI fragment of pMMN327 was replaced by an Nm^r (neomycin resistance) cassette isolated from pDG782 (13). The resultant plasmid, pMMN421, was used to transform B. subtilis JH642 and ZB307A (trp⁺ phe⁺) to generate ORB3071 and ORB3072, respectively. ORB3071 was transformed with pMMN402 and ORB3072 chromosomal DNA, and trp⁺ transformants were selected. Nm^s colonies were screened among trp⁺ transformants, which indicated that the neo insertion in pgk was replaced by the in-frame deletion from pMMN402. The pgk in-frame deletion in ORB3074, thus obtained, was confirmed by sequencing a PCR product, encompassing the pgk gene, that was amplified with ORB3074 chromosomal DNA as template. The resE Δpgk strain (ORB3076) was constructed by generalized transduction of ORB3074 with PBS1 phage carrying the resE mutation of strain LAB2234.

Construction of gapA and gapB mutations.

DNA containing the 5′ end of gapA was amplified by PCR with JH642 chromosomal DNA and two oligonucleotides, oMN98-36 (5′GGAATTCAATATAAATATCT3′) and oMN98-37 (5′CGGGATCCAAGTCAACTAGA3′). The 770-bp PCR product was inserted into pJDC9 (5), which carries an Erm^r marker. The resultant plasmid, pMMN414, was digested with EcoRV, which cleaves at a site within gapA. An Nm^r gene isolated from pDG792 (13) was inserted into the EcoRV site of pMMN414 to generate pMMN416. pMMN416 was used to transform LAB2527 (ΔresE sre-1), and an Nm^r Erm^s transformant (LAB3045) resulting from double-crossover recombination was chosen as a gapA::neo strain. To create a gapB mutation, DNA carrying gapB was amplified by PCR with two oligonucleotides, oMN98-28 (5′GTACTGGCGAATTCGTTTTAAT3′) and oMN98-29 (5′CTGTGTTTAAGCTTAATTTGCA3′). The amplified fragment (1.2 kb) was inserted into pUC18 that was digested with EcoRI and HindIII to create pMMN409. pMMN409 was digested with AccI, treated with T4 DNA polymerase to make the restriction enzyme site blunt, and further digested with PstI. The released 420-bp internal fragment of gapB was replaced by an Erm^r fragment from pJDC9 (pMMN410). A tetracycline resistance (Tet^r) gene isolated from pUC18tet was inserted into the EcoRI site of pMMN410. The resultant plasmid, pMMN412, was used to transform LAB2527, and one of the resulting Erm^r Tet^s transformants was named LAB3044, in which the internal segment of the gapB gene was substituted with the Erm^r marker. The chromosomal structures of gapA and gapB were confirmed by PCR.

Construction of pgm mutants.

A plasmid, pPS2019 (21), obtained from Peter Setlow carries the pgm-eno region of B. subtilis chromosomal DNA in which an internal fragment of pgm was replaced by an erm gene. A chloramphenicol resistance (Cm^r) cassette from pMMN7 (32) was inserted into the BamHI site (located in the multiple-cloning site) of pPS2019. The resulting plasmid, pYZ13, was linearized with ScaI (at a site located in the vector plasmid) and used to transform LAB2527. Erm^r Cm^s transformants (ORB3075) were chosen to ensure a double-crossover recombination, and the replacement of the internal segment of pgm by erm was confirmed by PCR.

Preparation of crude extracts and enzyme assays.

B. subtilis cells were grown aerobically in DS medium supplemented with 0.1% glucose, 0.2% KNO₃, and appropriate antibiotics. Cells were harvested from late-exponential growth cultures (when an optical density at 600 nm was 0.8 to 1.0) to prepare the crude extracts. The cells were washed with a solution containing 50 mM potassium phosphate buffer (pH 7.4)–2 mM EDTA–2 mM 2-mercaptoethanol and stored overnight at −70°C. The thawed cells were resuspended in the same buffer and were passed twice through a chilled French pressure cell at 18,000 lb/in². The lysate was centrifuged at 21,000 × g at 4°C for 20 min, and the supernatant was used to measure glycolytic enzyme activities. Triose-phosphate isomerase (TPI), glyceraldehyde-3-phosphate dehydrogenase (GAP), 3-phosphoglycerate kinase (PGK), and phosphoglycerate mutase (PGM) were determined by the methods described by Maitra and Lobo (23) with slight modifications. All enzyme activities were assayed in a buffer containing 50 mM triethanolamine hydrochloride (pH 7.4) and 10 mM MgCl₂. TPI activity was measured in a reaction mixture containing 0.15 mM NADH, 0.4 mM glyceraldehyde-3-phosphate, and 2 U of α-glycerophosphate dehydrogenase. The reaction mixture for GAP activity contained 0.15 mM NADH, 5 mM cysteine, 1 mM 3-phosphoglycerate, 1 mM ATP, and 2 U of PGK. For the PGK assay, the same reaction mixture was used, except that it contained 2 U of GAP instead of 3-phosphoglycerate kinase. The reaction mixture for PGM activity contained 0.15 mM NADH, 1 mM 3-phosphoglycerate, 1 mM ADP, 1 mM MnCl₂, 1 U of enolase, 2 U of pyruvate kinase, and 2 units of lactate dehydrogenase. All enzymes and substrates were obtained from Sigma Chemical Co. The rate of disappearance of NADH was monitored for 10 min at 340 nm in a total volume of 1 ml at room temperature. NADH oxidase activity was measured in a cell lysate without substrates and enzymes, and each enzyme-catalyzed reaction rate was corrected by the reference to the control rate. NADH was quantified by using an extinction coefficient of 6,220 M⁻¹ cm⁻¹. The protein concentration in the extract was determined by the Bio-Rad protein assay.

Immunoblotting analysis of PGM and PGK.

Crude extracts (10 μg for PGM and 40 μg for PGK) prepared for the glycolytic enzyme assays were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (10% polyacrylamide). Proteins were transferred to a nitrocellulose membrane, which was then treated with either a 1/500 dilution of anti-PGM serum or a 1/1,000 dilution of anti-PGK serum obtained from Peter Setlow (21) and Lonnie Ingram (2), respectively. The PGK antibody was prepared against Zymomonas mobilis PGK. Antigen-antibody reactions were detected by binding of goat anti-rabbit immunoglobulin G coupled to alkaline phosphatase and incubating with 5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium.

RESULTS

Isolation of a sre-1 mutant.

A previous report showed that ResD and, to a lesser extent, ResE are required for anaerobic growth of B. subtilis with nitrate as an electron acceptor (42). The anaerobic growth rate and maximal cell culture density of resE mutants are higher than those of a resD mutant (26), although a mutation in either resD or resE equally repressed the transcription of ResDE-controlled genes during anaerobic growth (33). In the hope of elucidating the role of the ResD-ResE signal transduction system in anaerobic respiration, a search was undertaken for suppressor mutants of resD or resE that are able to grow anaerobically with nitrate (Ang⁺) in the absence of ResD or ResE. Although no suppressor mutant of resD was isolated, a suppressor mutation (in strain LAB2527) that allows the mutant to undergo anaerobic nitrate respiration without ResE was obtained at a frequency of 10⁻⁷ to 10⁻⁸ in strain LAB2234 (ΔresE::spc) as described in Materials and Methods. When JH642 was transformed with chromosomal DNA prepared from LAB2527, only 2% of Spc^r transformants showed an Ang⁺ phenotype, indicating that the suppressor mutation was not linked to resE.

To determine if the suppressor mutation also allows the mutant to undergo anaerobic growth without ResD we substituted the resE mutation in LAB2527 with a resDE deletion mutation as described in Materials and Methods. The resultant mutant, LAB2545, was unable to grow anaerobically in the presence of nitrate. The failure of LAB2545 to grow anaerobically was not caused by loss of the suppressor mutation, as shown by the following experiment. LAB2545 was transformed with chromosomal DNA prepared from LAB2234 (ΔresE). All Spc^r Tet^s transformants tested (ΔresE::spc) grew anaerobically, and all Spc^r Tet^r transformants (ΔresE::spc ΔresDE::tet) failed to grow under anaerobic conditions. (These two types of recombinants were obtained because the ΔresDE deletion did not overlap with the ΔresE deletion.) These results demonstrated that the suppressor mutation can bypass the requirement for ResE but not for ResD in nitrate respiration. The mutation was named sre-1 (for suppressor of resE).

Effect of sre-1 on fnr expression.

We had previously shown that ResD and ResE are essential for transcription of fnr upon oxygen limitation (33). As discussed in a previous paper (33), fnr is expressed at the same low level in resD or resE mutants. These results strongly supported the hypothesis that ResE is the major (if not the sole) kinase for ResD in fnr activation. To determine whether the sre-1 mutation allows the mutant to express fnr without ResE, the β-galactosidase activity of a transcriptional fnr-lacZ fusion was measured in a wild-type strain (LAB2252) and in the resE (LAB2266), resE sre-1 (LAB2536), and resDE sre-1 (LAB2555) mutants. Figure 1B shows that fnr expression was highly induced in the wild-type strain during anaerobiosis, but no induction was observed in the resDE and resE mutants, as reported previously (33). Introduction of sre-1 in the resE mutant restored anaerobic fnr expression to the same level as in the wild type. In contrast, fnr expression was very low in the resDE sre-1 mutant. The sre-1 mutation in an otherwise wild-type genetic background had no significant effect on anaerobic fnr expression (compare LAB3034 and LAB2252 in Fig. 1B). The effect of sre-1 on fnr expression is similar to its effect on anaerobic growth of the resE and the resDE mutants.

FIG. 1.

Expression of fnr-lacZ fusion. Cells were grown in DS medium supplemented with 0.1% glucose and 0.2% KNO₃ under aerobic (A and C) and anaerobic (B and D) conditions. Time zero indicates the onset of stationary phase. (A and B) Symbols: ○, LAB2252 (wild type); ●, LAB2266 (ΔresE); □, LAB2536 (ΔresE pgk-1); ■, LAB2555 (ΔresDE pgk-1); ▴, LAB3034 (pgk-1). (C and D) Symbols: ○, LAB2898 (wild type); ●, LAB2901 (ΔresE); □, LAB2902 (ΔresE pgk-1); ▵, LAB2904 (ΔresE pgk-1 SPβ::pgk⁺). pgk-1 is identical to sre-1. β-gal., β-galactosidase.

The expression of fnr was severely reduced during aerobic growth in the wild-type strain as well as in the mutant strains, except for strain LAB2536 (resE sre-1), which showed constitutive fnr expression to levels 10 times higher than those observed in wild-type cells (Fig. 1A). The derepression of aerobic fnr expression caused by the sre-1 mutation was observed only in the resE mutant strains. In the resE⁺ sre-1 strain (LAB3034), fnr-lacZ was repressed by oxygen repletion as in the wild-type strain.

Effect of sre-1 on resA expression.

Among the aerobically expressed genes whose induction is dependent on ResD and ResE is the res operon itself (42). To determine if sre-1 suppresses the defect in aerobic gene expression as a result of the absence of resE, the effect of sre-1 on resA expression was studied. Figure 2A shows that resA-lacZ expression under aerobic conditions was induced threefold during postexponential growth of the wild-type strain (LAB2537) and that little or no induction was observed in a resD mutant (data not shown), as previously found (42). The previous study showed that aerobic resA expression was also reduced by a resE mutation, although the effect was not as severe as that of a resD mutation (42). However, under our culture conditions, the level of resA expression was not reduced by a resE mutation. We suspect that this discrepancy may be due to differences in culture aeration as well as in culture media. In the resE sre-1 double mutant (LAB2543), resA expression was constitutive and much higher than that observed in wild-type cells (LAB2537). The high constitutive resA transcription under aerobic conditions was observed in the sre-1 mutant only in the absence of intact resE, as in the case of fnr expression (Fig. 1A).

FIG. 2.

Expression of resA-lacZ fusion. Cells were grown in DS medium supplemented with 0.1% glucose and 0.2% KNO₃ under aerobic (A) and anaerobic (B) conditions. Time zero indicates the onset of stationary phase. Symbols: ○, LAB2537 (wild type); ●, LAB2540 (ΔresE); □, LAB2543 (ΔresE pgk-1); ■, LAB2556 (ΔresDE pgk-1); ▴, MAB117 (pgk-1); ▵, LAB2864 (ΔresE pgk-1 SPβ::pgk⁺). pgk-1 is identical to sre-1. β-gal., β-galactosidase.

Maximal resA expression in wild-type cells was 10-fold higher during anaerobic growth than during aerobic growth and was strongly dependent on ResD and ResE. The sre-1 mutation restored anaerobic resA expression to the resE mutant. However, anaerobic expression was not significantly different in the resE sre-1 (LAB2543) and resE⁺ sre-1 (MAB117) strains, unlike the case for aerobic resA expression.

Since resA is induced by oxygen limitation, we asked whether another ResD-controlled gene, ctaA, was also induced under anaerobic conditions. CtaA is required for cytochrome aa₃ biogenesis, and its aerobic expression was previously shown to be dependent on the ResD-ResE pathway (42). Maximal expression of ctaA was 10-fold higher under anaerobic than aerobic conditions, and the effect of the sre-1 mutation on ctaA expression was similar to its effect on fnr and resA expression (data not shown).

Taken together, these results suggest that the sre-1 mutation causes expression of all ResD-controlled genes to be increased during anaerobic growth. The sre-1 mutation bypasses the requirement of ResE for anaerobic expression and derepresses expression in the absence of resE under aerobic conditions. These results suggest that ResE functions differently under aerobic and anaerobic conditions and that the differential effect of the sre-1 mutation reflects these different activities of ResE.

Identifying the sre-1 locus.

To identify the sre gene and the sre-1 mutation, chromosomal DNA carrying the sre locus was isolated by taking advantage of the characteristic colony morphology of sre-1 mutants as described in Materials and Methods. A plasmid insertion into the B. subtilis chromosome was isolated that was tightly linked to the sre locus. When chromosomal DNA carrying the plasmid integration was used to transform the sre-1 mutant, 67 to 98% of transformants selected for Erm^r (conferred by the plasmid) were sre-1⁺. A transformant with the sre-1 phenotype with this chromosomal DNA was named MAB128, and a sre-1⁺ transformant was named MAB129.

Two plasmids, pMMN327 and pMMN329, were isolated from MAB129 chromosomal DNA digested with HindIII or SacI, respectively, as described in Materials and Methods. pMMN327 carried DNA (2 kb) flanking one side of the chromosomally inserted plasmid DNA, and pMMN329 carried 5 kb of DNA from the other side. LAB2527 (ΔresE sre-1) cells were transformed with pMMN327 and pMMN329, and the transformants were examined for anaerobic growth. Ang⁻ clones were generated by transformation with pMMN327 but not with pMMN329, indicating that the sre gene resides in the 2-kb insert of pMMN327. MAB128 chromosomal DNA digested with HindIII was used to isolate plasmid pMMN326 that carries the identical chromosomal fragment to that in pMMN327 but containing the sre-1 allele.

Nucleotide sequence of sre.

The sre locus was subjected to nucleotide sequence analysis by using pMMN326 for the sre-1 allele and pMMN327 for the sre⁺ allele. The entire segments of B. subtilis DNA in pMMN326 and pMMN327 were sequenced to identify the sre-1 mutation. The results indicated that the segment contains the 3′ end of gap (the GAP gene), the entire pgk (the PGK gene), and the 5′ end of tpi (the TPI gene). Comparison of the inserts of pMMN326 and pMMN327 identified a single difference, an insertion of one adenine residue in pMMN326, in the stretch of seven adenine residues located in the N-terminal coding region of pgk (Fig. 3). The insertion apparently caused a frameshift in the pgk coding sequence, leading to production of a truncated pgk product of 20 amino acids. This result indicated that sre-1 is probably a null mutation of pgk; sre-1 was renamed pgk-1.

FIG. 3.

The pgk region of the B. subtilis chromosome. The ORFs in the region are shown by the various shaded boxes. P with arrow indicates putative promoters and the direction of transcription. The line above the boxes denotes the DNA fragment inserted in pMMN326, pMMN327, pMMN387, and pMMN388. The translation start site of pgk was shown by the underlined ATG. The pgk-1 (sre-1) mutation is an insertion of a single adenine, indicated as the boxed A, and the pgk-1 mutation creates a stop codon TGA (underlined). The downstream doubly underlined ATG might be the translation reinitiation site (see Discussion). Shown at the bottom is the part of glycolytic pathway that is relevant to this study. ENO, enolase; PEP, phosphoenolpyruvate.

Complementation analysis.

The pgk gene is part of a cluster of glycolytic enzyme genes (Fig. 3) in the order gap, pgk, tpi, pgm, and eno (the enolase gene). It was suggested that pgk, tpi, pgm, and eno are cotranscribed from an unidentified promoter located within the intergenic region between gap and pgk. It was also suggested that some pgk transcription may initiate upstream of gap (21). To assess the possibility that the sre-1 phenotype was caused by a polar effect on a downstream gene, complementation analysis was carried out. Plasmids pMMN326 and pMMN327 were digested with BamHI and HindIII, and the chromosomal DNA carrying pgk (pgk-1 and pgk⁺) was subcloned into an integration plasmid, pMMN13 (29), to generate plasmids pMMN387 and pMMN388, respectively. The B. subtilis chromosomal DNA in the plasmids contained the 3′ end of gap, intact pgk, and the 5′ end of tpi (Fig. 3). These plasmids were introduced into the SPβ prophage locus of the LAB2234 (ΔresE) and LAB2527 (ΔresE pgk-1) strains by a standard procedure described previously (48). LAB2234 cells lysogenized with SPβ carrying pMMN387 (LAB2846) or pMMN388 (LAB2847) were unable to grow anaerobically. LAB2527 cells lysogenized with SPβ carrying pMMN387 (LAB2848) still grew anaerobically like the parent strain, but the LAB2527 lysogen with pMMN388 (LAB2849) lost the ability to grow by nitrate respiration under anaerobic conditions (Table 2). This result indicates that the wild-type copy of pgk complemented the pgk-1 mutation and restored the ResE⁻ phenotype under anaerobic growth. This analysis indicates that pgk-1 is a recessive mutation and that the observed mutational effect on ResDE-controlled genes is not the result of a polar effect on downstream genes.

TABLE 2.

Growth rates of the wild type and mutant strains

Strain (genotype)	Doubling time (min) in DS medium with addition ofa:
	Aerobic growth supplement				Anaerobic growth supplement (glucose + nitrate)
	None	Glucose	Pyruvate	Glucose + pyruvate
LAB2234 (ΔresE)	47 (3.8)	28 (6.4)	44 (4.4)	32 (7.7)	240 (0.4)
LAB2527 (ΔresE pgk-1)b	63 (3.4)	40 (2.5)	37 (2.7)	35 (4.2)	90 (1.2)
LAB2849 (ΔresE pgk-1 SPβ::pgk)	47 (2.8)	30 (7.5)	42 (5.9)	30 (7.2)	240 (0.4)
ORB3076 (ΔresE Δpgk-1)	273 (2.8)	140 (3.1)	37 (1.4)	43 (3.3)	>360 (<0.15)

^aCells grown overnight aerobically in DS medium supplemented with 0.5% glucose and pyruvate were transferred to DS medium or to DS medium supplemented with 0.5% glucose, 0.5% pyruvate, or both. For anaerobic growth, cells were grown in DS medium supplemented with 0.1% glucose and 0.2% KNO₃. Values in parentheses are the optical density after 18 h of growth. 

^bThe pgk-1 mutation is the same as sre-1 (see the text). 

As shown in Fig. 1C and D, the wild-type pgk gene also complemented the mutational effect of pgk-1 on fnr expression. Aerobic (Fig. 1C) and anaerobic (Fig. 1D) fnr expression in LAB2904 (ΔresE pgk-1 SPβ::pgk⁺) was as low as that in LAB2901 (ΔresE), indicating that pgk⁺ is dominant to pgk-1 with respect to fnr expression.

The resE pgk-1 strain (LAB2543) exhibited high constitutive resA expression under aerobic conditions, as described above. The introduction of wild-type pgk (LAB2864) reduced the expression to a level similar to that observed in the ΔresE strain (LAB2540) (Fig. 2A), except that LAB2864 cells showed slightly higher resA expression than did LAB2540 during postexponential growth. Similarly, anaerobic resA expression in the resE pgk-1 mutant was restored to the level of the resE mutant by addition of a wild-type copy of pgk (Fig. 2B).

An in-frame mutation of pgk showed a partial sre-1 phenotype.

An in-frame deletion of pgk was constructed as described in Materials and Methods to determine if a null mutation confers the same phenotype as the pgk-1 mutation. Neither a resE Δpgk mutant (ORB3076) nor a Δpgk mutant (ORB3074) was able to grow anaerobically on Luria-Bertani agar supplemented with 1% glucose and 0.2% KNO₃. In fact, the growth defect observed in these strains was more severe than that of the LAB2234 (ΔresE) strain (Table 2). As discussed below, this result is due to the requirement of glycolysis for nitrate respiration in B. subtilis. Although the pgk in-frame deletion failed to suppress the defect in anaerobic growth of resE mutants, anaerobic fnr expression was recovered to the wild-type level when the pgk mutation was introduced into the resE mutant (Fig. 4). This result raises the possibility that the pgk-1 and Δpgk mutations have different phenotypes because the pgk-1 mutant has a low level of PGK activity rather than a complete absence of activity.

FIG. 4.

Expression of the fnr-lacZ fusion. Cells were grown anaerobically in DS medium supplemented with 0.1% glucose and 0.2% KNO₃ with (hatched box) or without (solid box) 0.1% pyruvate. Maximal activity from each culture is shown. Strains: LAB2252 (wild type); LAB2266 (ΔresE); LAB2536 (ΔresE pgk-1); LAB3023 (gapA); ORB3061 (ΔresE pgk-1 gapA); ORB3128 (ΔresE Δpgk). pgk-1 is identical to sre-1. β-gal., β-galactosidase.

Aerobic growth of pgk strains in different carbon sources.

Freese et al. showed that a pgk mutant, in which the glycolytic pathway is separated into two disconnected sections, is unable to grow on single carbon sources from either the upper or lower subdivision. The mutant grows if a carbon source from each subdivision is provided (12). Strains LAB2234 (ΔresE), LAB2527 (ΔresE pgk-1), LAB2849 (ΔresE pgk-1 SPβ::pgk⁺), and ORB3076 (ΔresE Δpgk) were streaked on a TSS agar plate supplemented with 0.5% glucose or with 0.5% glucose and 0.5% pyruvate as the sole carbon sources. After 2 days, only LAB2234 and LAB2849 grew on glucose while LAB2234, LAB2849, and LAB2527 (to a lesser extent) grew on glucose and pyruvate. After 3 days of incubation, LAB2527, but not ORB3076, grew on glucose as the sole carbon source and LAB2527 grew better than ORB3076 on the plate containing glucose and pyruvate. LAB2849 cells grew as well as the wild type on either plate (data not shown).

This aerobic growth phenotype was confirmed by experiments in DS liquid medium alone or supplemented with glucose or pyruvate or both (Table 2). In DS medium, LAB2234 and LAB2849 cultures exhibited an almost identical growth rate and similar maximal cell density without or with any carbon source. Both strains grew at a higher rate in the presence of glucose than in the presence of pyruvate. The pgk in-frame deletion mutant carrying the pgk gene at the SPβ locus exhibited similar aerobic and anaerobic growth characteristic to those of LAB2849 (data not shown). However, cultures of the pgk-1 mutant and, more drastically, the pgk in-frame deletion mutant showed a lower growth rate than the PGK⁺ strains in DS medium without any supplement or with glucose supplement. In contrast, addition of pyruvate increased the growth rate of these mutant cultures, but cell lysis was observed in the ORB3076 culture after prolonged incubation. To achieve a higher cell density as well as an increased growth rate, carbon sources from the upper (glucose) and lower (pyruvate) division were required. These results suggest, again, that the pgk-1 mutant has reduced PGK activity but has higher activity than the pgk in-frame deletion mutant.

Measurement of GAP, PGK, TPI, and PGM activities in various mutants.

Measurements of the activities of the glycolytic enzymes, GAP, PGK, TPI, and PGM, in various mutant strains showed that LAB2527 (ΔresE pgk-1) cells have very low PGK activity (0.5% of the wild-type level) (Table 3). No PGK activity was detected in ORB3076 cells carrying the in-frame deletion of pgk, indicating that the pgk-1 mutation has very low but significant PGK activity. Complementation of pgk in trans (as in LAB2849) led to increased PGK activity compared to the resE pgk-1 strain, but the activity was still lower than in the wild-type strain. This suggests that the promoter in front of pgk is not sufficient to drive expression of pgk to the level observed in wild-type cells. Western blot analysis was carried out to detect PGK protein from the wild-type and mutant strains. Since antibody against B. subtilis PGK is not available at present, we used antiserum prepared against Zymomonas mobilis PGK by Ingram’s laboratory (2). Sequence analysis showed that PGK from B. subtilis is 48% identical in amino acid sequence to PGK from Z. mobilis. A 42-kDa protein in LAB2234 reacted with anti-PGK, but we could not detect the protein from the other strains including LAB2849 (data not shown). This was not so surprising, considering that the antibody was prepared against non-B. subtilis PGK and the activity in LAB2849 was 20-fold lower than that in LAB2234.

TABLE 3.

Specific activities of GAP, PGK, TPI and PGM in B. subtilis strains

Strain (genotype)	Spec act (nmol of NADH oxidized/mg of protein/min)a
	GAP	PGK	TPI	PGM
LAB2234 (ΔresE)	163	97	4,888	18
LAB2527 (ΔresE pgk-1)b	361	0.5	8,023	39
LAB2849 (ΔresE pgk-1 SPβ::pgk	197	4.3	5,345	16
ORB3045 (ΔresE pgk-1 gapA)	<0.1	<0.1	7,045	43
ORB3076 (ΔresE Δpgk)	60	<0.1	3,424	11

^aCells were grown aerobically in DS medium with 0.2% KNO₃ and 0.1% glucose. To the cultures of ORB3045 and 3076, 0.1% pyruvate was also supplemented. The values are averages of two to three experiments. Standard deviation was less than 25%. 

^bThe pgk-1 mutation is the same as sre-1. 

LAB2527 cells had higher activities of GAP, TPI, and PGM than did LAB2234 cells, but these activities returned to the wild-type level when pgk was introduced in trans. This suggests that there is some regulatory communication among glycolytic enzymes but also demonstrates that the pgk-1 mutation is not polar on downstream genes. This was confirmed by Western blot analysis with anti-PGM. A 56-kDa protein reacting with anti-PGM was not detected in a pgm mutant (ORB3075) and was present in the other strains including the pgk-1 and the pgk in-frame deletion mutants (data not shown).

PGK and GAP activities were not detected in ORB3045 (resE pgk-1 gapA), but the same levels of TPI and PGM activity as those in the resE sre-1 (LAB2527) mutant were observed.

Effect of gapA and gapB mutations on the pgk-1 phenotype.

The phenotypes of pgk-1 and Δpgk mutants might be caused by accumulation of a metabolite(s) in the pgk mutants. The growth experiment with DS medium (Table 2) showed that the growth rate of the pgk-1 mutant, like the pgk in-frame deletion mutant, was improved by addition of a carbon source (pyruvate) from the lower division of the glycolytic pathway but not by one (glucose) from the upper division. This indicates that in DS medium, the flow of carbon is from the upper to lower subdivision as previously described (10, 12). Carbon flow from the upper to the lower division of glycolysis strongly suggests that 1,3-diphosphoglycerate accumulates when PGK activity is impaired. To test these possibilities, a mutation in the gap gene was introduced into the resE pgk-1 mutant. GAP catalyzes the oxidative phosphorylation of d-glyceraldehyde 3-phosphate to 1,3-diphosphoglycerate, which is a substrate for PGK. B. subtilis has two gap genes (19), one (gapA, previously called gap [43]) located upstream of pgk and the other (gapB) at a locus unlinked to any other glycolysis gene (19). We created a mutation in each gap gene as described in Materials and Methods. The gapA mutant, but not the gapB mutant, was unable to grow on TSS medium supplemented with glucose, indicating that the gapA product is the major glycolytic enzyme.

The gapA or gapB mutation was introduced into LAB2527 (ΔresE pgk-1) cells, and the SPβ phage carrying the fnr-lacZ fusion was used to transduce these strains. β-Galactosidase activity derived from fnr-lacZ expression in the resE pgk-1 gapB mutant (LAB3060) during anaerobic growth was indistinguishable from that in the wild-type (LAB2252) strain (data not shown). In contrast, the activity in the ORB3061 (ΔresE pgk-1 gapA) mutant was significantly reduced when the cells were grown anaerobically in DS medium supplemented with 0.1% glucose and 0.2% KNO₃ (Fig. 4). The straightforward interpretation of this result is that accumulation of 1,3-diphosphoglycerate contributes to the suppression of ΔresE by pgk-1. However, this result must be interpreted carefully for the following reason. We have previously found that certain glycolytic mutants are impaired for fnr transcription. Addition of pyruvate restores anaerobic growth and fnr expression to these glycolytic mutant strains (28). As shown in Fig. 4, the gapA single mutant also exhibited very low fnr expression and was unable to grow anaerobically with nitrate (data not shown). Addition of pyruvate (0.1%) restored fnr expression in the mutant to the level of wild-type cells grown with glucose alone. Addition of pyruvate had no significant effect on fnr expression in the resE mutant, indicating that the effect of pyruvate addition requires the intact resE gene. Interestingly, β-galactosidase activity in the resE pgk-1 mutant grown with pyruvate was threefold lower than the activity observed in the absence of pyruvate. This suggests that the effect of pgk-1 on fnr expression is alleviated by pyruvate, probably by reducing carbon flow from glucose and thus reducing the accumulation of 1,3-diphosphoglycerate. β-Galactosidase activity in the resE pgk-1 gapA mutant was not increased by the addition of pyruvate and was lower than the activity in either the resE pgk-1 mutant or gapA mutant. This observation is consistent with the hypothesis that accumulation of 1,3-diphosphoglycerate is responsible for derepressed fnr expression in the resE pgk-1 mutant.

Effect of pgk-1 on srfA-lacZ in the absence of comP.

To determine if the pgk-1 mutation generally affects two-component signal transduction systems, the pgk-1 mutation was introduced into a ΔcomP mutant carrying srfA-lacZ. ComP is the histidine kinase (46) required for the phosphorylation of ComA (45), a transcriptional activator of the srf operon (32). The srf operon codes for the enzyme complex that catalyzes the synthesis of a lipopeptide, surfactin, as well as for a small protein needed for genetic competence (6, 29). As expected, the expression of srfA was severely reduced by the comP mutation; introduction of pgk-1 had no significant effect on srfA-lacZ expression (data not shown), indicating that the effect of pgk-1 may be specific for the ResD-ResE pathway.

DISCUSSION

A reduction in PGK activity allows anaerobic growth and expression of ResD-controlled genes to occur without ResE. Because ResD is still required under these conditions, ResD is probably phosphorylated by another pathway independent of ResE. Histidine sensor kinases generally have both kinase and phosphatase activities. The phosphatase activity is highly regulated by sensory signals and thus plays an important role in regulation (for a review, see reference 40). The ResDE signal transduction pathway activates the expression of genes that are expressed aerobically but much more highly expressed under anaerobic conditions. One possible explanation is that while the kinase activity of ResE is active under both aerobic and anaerobic conditions, the phosphatase is more (or only) active during aerobic growth. As a result, ResD could be phosphorylated to a greater extent under anaerobic conditions, which would lead to anaerobic induction of ResD-controlled genes.

The fact that the resE pgk-1 double mutant exhibits unusually high aerobic expression of ResD-controlled genes (as shown in Fig. 1 and 2) suggests that these mutant cells are deficient in the normal ResE phosphatase activity. During aerobic growth of the resE pgk-1 mutant, ResD is phosphorylated by a pathway independent of ResE but the resulting ResD phosphate may not be dephosphorylated because ResE phosphatase is absent in this strain. Therefore, the resE pgk-1 mutant would accumulate a higher level of phosphorylated ResD than would a pgk-1 single mutant. Under anaerobic conditions, the expression of ResD-controlled genes, either in the resE pgk-1 or in the pgk-1 mutant, is high because ResE phosphatase activity appears to be down-regulated under conditions of oxygen limitation.

The pgk null mutant, unlike the pgk-1 mutant, is unable to grow anaerobically in the absence of ResE. The reason why the two mutants behave differently in anaerobic growth can probably be attributed to residual glycolysis in the pgk-1 mutant. The defect of many glycolytic mutants in anaerobic growth and fnr expression is overcome by addition of pyruvate, but growth and expression are still dependent on ResD and ResE. One possible explanation for these results is that glycolysis normally produces a signal leading to autophosphorylation of ResE and that metabolism of pyruvate may also generate this signal. One candidate for this signal might be NADH, and its accumulation might directly or indirectly trigger the ResD-ResE signal transduction pathway required for aerobic and anaerobic respiration. Among the glycolytic mutants, only the pgk null mutant showed significant fnr expression, and this mutant still failed to grow anaerobically. This suggests that glycolysis plays an additional role in anaerobic respiration besides stimulating fnr expression.

The residual PGK activity in the pgk-1 mutant might be due to translational reinitiation or frame-shifting at the adenine stretches where the pgk-1 lesion is located (Fig. 3). In fact, an AUG start codon which lies in frame within the pgk coding sequence is present 13 bases downstream of the UGA stop codon that was created by the pgk-1 mutation (Fig. 3). Previous studies have shown that a ribosome that traverses a stop codon can remain bound to a mRNA and reach a downstream initiation codon more than 40 nucleotides from the termination site (1). Although there is no Shine-Dalgarno sequence preceding the second AUG codon, it has been shown that reinitiation can occur in the absence of a Shine-Dalgarno sequence, albeit at low efficiency (39).

We concluded that B. subtilis gapA, but not gapB, encodes GAP, as deduced from the growth phenotypes of gap mutants and from enzyme assays. Interestingly, E. coli also has two gap genes, gapB upstream of pgk and gapA in an unlinked region of the chromosome. In contrast to the case in B. subtilis, E. coli gapA encodes GAP and the pgk-linked gapB (renamed epd) encodes an erythrose 4-phosphate dehydrogenase that functions in pyridoxal 5′-phosphate biosynthesis (47). Recently, Fillinger et al. showed that B. subtilis GapA is indeed the glycolytic enzyme and GapB is required for gluconeogenesis (9). This result strongly supports our hypothesis that accumulation of 1,3-diphosphoglycerate is responsible for the suppressor effect by pgk-1. 1,3-Diphosphoglycerate is a high-energy phosphodonor, and other high-energy phosphodonors, such as acetyl phosphate and carbamoyl phosphate, are known to phosphorylate response regulators in vitro in the absence of cognate kinases (8, 22). 1,3-Diphosphoglycerate is synthesized by incorporation of inorganic phosphate into glyceraldehyde-3-phosphate, and conversion of 1,3-diphosphoglycerate to 3-phosphoglycerate is coupled to ATP generation. This pathway is analogous to that involved in acetyl phosphate synthesis. Acetyl phosphate, synthesized from acetyl coenzyme A and inorganic phosphate, is converted to acetate with concomitant production of ATP. Involvement of acetyl phosphate in activation of response regulators in vivo was demonstrated by analysis of mutations that alter the level of acetyl phosphate (44) and by direct measurement of intracellular acetyl phosphate levels (24, 36). Measurement of intracellular levels of 1,3-diphosphoglycerate is difficult due to its high instability, as described previously (17). Direct phosphorylation of ResD (purified and provided by Marion Hulett) in vitro by 1,3-diphosphoglycerate was examined. 1,3-Diphosphoglycerate was synthesized from 3-phosphoglycerate, [γ-³²P] ATP, and PGK. The radioactive 1,3-diphosphoglycerate was not further purified, to avoid rapid decay. Phosphorylation of ResD was not observed with 1,3-diphosphoglycerate up to 11 mM (the amount was measured spectrophotometrically by NADH oxidation in the presence of GAP). The reaction was tested at pH 7 to 8, at room temperature and 37°C, and in the presence of 5 or 10 mM MgCl₂. Although we cannot rule out the possibility that the in vitro conditions we used are not suitable for direct phosphorylation of ResD (for example, the presence of ATP, ADP, and 3-phosphoglycerate may inhibit phosphotransfer of ResD), this negative result may suggest the involvement of a noncognate kinase for the activation of ResD. Several mechanisms that include the involvement of a noncognate kinase can be considered. For example, accumulation of 1,3-diphosphoglycerate may be a signal for autophosphorylation of the kinase or 1,3-diphosphoglycerate may be an effector that stimulates autophosphorylation of the kinase. In fact, a recent study showed that the in vivo activation of PhoB by acetyl phosphate, the response regulator for the Pho regulon in E. coli, requires the osmoregulatory sensor kinase EnvZ (18). We sought to test the possibility that another sensor kinase is involved in activation of ResD in the resE pgk-1 mutant. We chose PhoR as a primary candidate since ResE shows significantly high sequence similarity to PhoR (15, 38) and both PhoR and ResE belong to an EnvZ family of kinases (35) or group IIIA among B. subtilis kinases (7). Furthermore, recent studies by Hulett and coworkers showed that regulatory networks governed by PhoP-PhoR and ResD-ResE are tightly interwoven (4, 41). Introduction of phoR mutation in the resE pgk-1 mutants resulted in slower growth in DS medium supplemented with 0.1% glucose and 0.2% nitrate; however, the expression of fnr and resA in the resE pgk-1 phoR mutants was similar to that in congenic phoR⁺ strains (data not shown). This result indicates that if another kinase is involved in the activation of ResD in the resE pgk-1 strain, that kinase is not PhoR.

Although further studies are required to uncover the ResE-independent phosphorylation of ResD, this study suggests a possible connection between glycolysis and activation of ResD. It also provides suggestive evidence that ResE phosphatase activity plays a pivotal role in differential regulation of gene expression during aerobic and anaerobic growth. It will be important to elucidate how ResE phosphatase activity is regulated by input signals.