Characterization of ResDE-Dependent fnr Transcription in Bacillus subtilis

Abstract

The ResD-ResE signal transduction system is required for transcription of genes involved in aerobic and anaerobic respiration in Bacillus subtilis. Phosphorylated ResD (ResD∼P) interacts with target DNA to activate transcription. A strong sequence similarity was detected in promoter regions of some ResD-controlled genes including fnr and resA. Single-base substitutions in the fnr and resA promoters were performed to determine a ResD-binding sequence. DNase I footprinting analysis indicated that ResD∼P itself does not bind to fnr, but interaction of ResD∼P with the C-terminal domain of the α subunit (αCTD) of RNA polymerase (RNAP) facilitates cooperative binding of ResD∼P and RNAP, thereby increasing fnr transcription initiation. Consistent with this result, amino acid substitutions in αCTD, such as Y263A, K267A, A269I, or N290A, sharply reduced fnr transcription in vivo, and the K267A αCTD protein, unlike the wild-type protein, did not increase ResD∼P binding to the fnr promoter. Amino acid residues of αCTD required for ResD-dependent fnr transcription, with the exception of N290, which may interact with DNA, constitute a distinct surface, suggesting that these residues likely interact with ResD∼P.

The ResD-ResE signal transduction system is required for aerobic and anaerobic respiration in Bacillus subtilis (25, 30). ResE is a membrane-bound sensor histidine kinase that, upon autophosphorylation, donates a high-energy phosphate to its cognate response regulator, ResD. The resD and resE genes constitute an operon with the three upstream genes, resABC (30). ResA is a thiol-disulfide oxidoreductase involved in cytochrome c maturation (5, 6, 8), and ResB and ResC were also shown to play an essential role in cytochrome c synthesis (15). resD and resE are transcribed from a resDE-specific promoter and the resA operon promoter, the latter of which is dependent on ResD and ResE (30).

ResDE-controlled genes that are involved in anaerobic respiration include fnr encoding an anaerobic transcriptional regulator and nasDEF, which constitute an operon encoding subunits of nitrite reductase (21, 25). Fnr is essential for nitrate respiration, as expression of the respiratory nitrate reductase operon, narGHJI, is dependent on Fnr (7). In addition, expression of the flavohemoglobin gene hmp is highly induced by ResDE upon oxygen limitation (14). Transcription of fnr, nasD, and hmp is activated in vitro by ResD when the protein is phosphorylated by incubation with ResE and ATP (10), indicating that phosphorylated ResD (ResD∼P) directly interacts with the regulatory region of these genes.

Direct interaction of ResD∼P with ResD-controlled promoters, which was confirmed by DNase I or hydroxyl radical footprinting, in ctaA encoding heme A synthase (31), nasD (10, 24), hmp (10, 24), fnr (24), and yclJK encoding a two-component regulatory protein pair (11) has been reported. Binding of ResD∼P to fnr was observed either at a higher concentration of ResD∼P (24) or was not detected (10). ResD∼P bound to three distinct regions of ctaA (A1, positions −209 to −179 relative to the transcription start site; A2, −108 to −55; A3, −2 to +43) (31). The A1 site may be involved in ResD-dependent activation of the divergently transcribed ctaBCDEF operon encoding heme O synthase and the subunits of cytochrome caa₃ (17). A consensus ResD-binding site was proposed as TTTGTGAAT (consensus sequence a in Fig. 1A) by a sequence alignment of nasD, hmp, fnr, and ctaA (24, 31). Subsequently, a bioinformatics approach was used to compare ResD-binding sites in the nasD, hmp, fnr, yclJ, and ctaA promoters and TTGTN₆TTTNTN₂A (consensus sequence b in Fig. 1B) was proposed as a revised consensus ResD-binding sequence (11). The validity of the two proposed consensus sequences has pros and cons (Fig. 1A and B). For example, only a weak similarity (five of nine matches) to consensus sequence a was found in the ResD-binding region (−92 to −68) of yclJ (11), whereas the binding region contains a sequence with a perfect match to consensus sequence b. On the other hand, consensus sequence a, but not consensus sequence b, was identified in the ctaA3 region. We had previously shown that a region including −61 to −58 of the fnr promoter is indispensable for ResD-dependent activation (24); however, consensus sequence b in the fnr promoter does not include the sequence. Finally, although deletion analysis showed that a nasD region between −87 and −76 is critical for ResD-dependent activation, this region is not included either in consensus sequence a or consensus sequence b. Both consensus sequences share 5′ TTGT, and the major difference between these sequences is GAA adjacent to TTGT that was proposed only for consensus sequence a.

FIG. 1.

Comparison of proposed consensus ResD-binding sequences. (A) Consensus sequence a was proposed by a sequence alignment of experimentally detected ResD-binding regions (24, 31). Numbering is relative to the transcription start site. ctaA has three distinct ResD-binding regions (31). Bases identical to those in the consensus sequence are indicated by bold type, and the numbers in parentheses show the number of bases of the total number of bases that match those in the consensus sequence. An asterisk indicates the sequence on the complementary strand. (B) Consensus sequence b was defined by bioinformatics approach (11). Sequence similarity with consensus sequence b is not detected in ctaA3. (C) A newly proposed consensus ResD-binding sequence (consensus sequence c) for nasD, hmp, and fnr. The full consensus sequence is shown by a box above the sequence. Half-site a is TTGTGAAN₂, and half-site b is NTTTN₄A. The arrows show the orientation of each half-site. Nucleotides doubly underlined are the nucleotides protected by ResD∼P from attack by hydroxyl radicals, and a dotted double line shows protected nucleotides located on the opposite face of DNA helix (10).

Meanwhile, we noticed that there is a strong similarity among the promoter regions of ResD-controlled genes, namely, resA, fnr, and sboA encoding subtilosin (23, 33). As shown in Fig. 2A, a direct repeat (TTCA N₇ TTCA) is present in resA and sboA, and a single TTCA sequence is present in fnr. Interestingly, TGAA, which is complementary to TTCA, is the sequence present in consensus sequence a but not in consensus sequence b of ctaA1, ctaA2, ctaA3, and yclJ (Fig. 1A). As shown in Fig. 1A, ctaA1 carries the exact match to the TTCA N₇ TTCA sequence on the complementary strand, which is likely used as a ResD-binding sequence of the divergently transcribed ctaBCDEF. The direct repeat is also present in the ResD-binding consensus sequence a of yclJK identified by DNase I footprinting experiments (11) but on the noncoding strand (Fig. 1A), suggesting that the yclIH gene transcribed divergently from yclJK might be regulated by ResDE as well. Expression of lacZ fused to the yclI promoter region was activated upon oxygen limitation, and this activation required ResDE (C. S. Zuber and M. M. Nakano, unpublished results), indicating that yclIH also belongs to the ResDE regulon. These results suggested that TGAA (and TTCA) proposed as a part of consensus sequence a is indeed used as a ResD-binding site. In this study, we generated base substitutions in the fnr and resA promoters to determine whether the TTCA sequence is required for ResD-dependent control. Furthermore, we investigated how binding of ResD∼P to the fnr promoter facilitates transcription initiation. DNase I footprinting analysis and in vivo studies using the C-terminal domain of the α subunit (αCTD) alanine-scanning mutants indicated that interaction of ResD∼P with αCTD is required for activation of fnr transcription.

FIG. 2.

Mutational analysis of the fnr and resA promoters. (A) Comparison of the promoter regions of ResDE-dependent sboA, resA, and fnr. The positions of nucleotides are relative to the transcription start site. The conserved TTCA sequence is underlined. (B) Effects of base substitutions in the fnr promoter region on transcription. (C) Effects of base substitutions in the resA promoter region on transcription. Cells were grown anaerobically in 2× YT supplemented with 1% glucose and 0.2% KNO₃, and β-galactosidase activities were measured as described in Materials and Methods. Experiments were repeated two to six times, and the averages of maximal activities around T1 (1 h after the end of exponential growth) are shown with standard deviations.

MATERIALS AND METHODS

Strains and plasmids.

All B. subtilis strains used in this study are derivatives of B. subtilis JH642 and listed in Table 1. Construction of rpoA with alanine substitution at each residue in αCTD was previously described (32). Plasmids are also listed in Table 1.

TABLE 1.

Bacillus subtilis strains and plasmids

Strain or plasmid	Relevant feature	Reference
Strains
JH642	Parental strain (trpC2 pheA1)	J. A. Hoch
LAB2714	SPβc2Δ2::Tn917::pMMN350	This study
LAB2761	SPβc2Δ2::Tn917::pMMN380	This study
LAB2764	SPβc2Δ2::Tn917::pMMN351	This study
ORB3502	SPβc2Δ2::Tn917::pYZ32	This study
ORB3503	SPβc2Δ2::Tn917::pYZ33	This study
ORB3504	SPβc2Δ2::Tn917::pYZ34	This study
ORB3505	SPβc2Δ2::Tn917::pYZ35	This study
ORB5001	SPβc2Δ2::Tn917::pMMN580	This study
ORB5019	SPβc2Δ2::Tn917::pKM3	This study
ORB5090	SPβc2Δ2::Tn917::pKM4	This study
ORB5262	rpoA (K267A)	32
ORB6462	SPβc2Δ2::Tn917::pMMN650	This study
ORB6463	SPβc2Δ2::Tn917::pMMN652	This study
ORB6466	SPβc2Δ2::Tn917::pMMN651	This study
ORB6467	SPβc2Δ2::Tn917::pMMN653	This study
ORB6513	SPβc2Δ2::Tn917::pMMN649	This study
ORB6589	SPβc2Δ2::Tn917::pMMN658	This study
ORB6590	SPβc2Δ2::Tn917::pMMN659	This study
ORB6613	SPβc2Δ2::Tn917::pCSZ4	This study
ORB6614	SPβc2Δ2::Tn917::pCSZ5	This study
ORB6615	SPβc2Δ2::Tn917::pCSZ6	This study
ORB6637	SPβc2Δ2::Tn917::pMMN670	This study
Plasmids
pTKlac	Promoter-probe vector; Ampr Cmr	13
pCSZ4	pTKlac with resA promoter (−134 to +65, −48C to T)	This study
pCSZ5	pTKlac with resA promoter (−134 to +65, −60T to G)	This study
pCSZ6	pTKlac with resA promoter (−134 to +65, −59C to T)	This study
pHG45	pUC18 with fnr promoter (−169 to +96, −47T to C)	This study
pHG71	pUC18 with fnr promoter (−169 to +96, −48T to G)	This study
pKM3	pTKlac with fnr promoter (−62 to +265, −47T to C)	This study
pKM4	pTKlac with fnr promoter (−62 to +265, −45G to A)	This study
pMMN350	pTKlac with fnr promoter (−62 to +265, −59T to G)	This study
pMMN351	pTKlac with fnr promoter (−62 to +265, −57A to C)	This study
pMMN380	pTKlac with fnr promoter (−62 to +265, −48T to G)	This study
pMMN408	pUC18 with fnr promoter (−169 to +96)	24
pMMN580	pTKlac with fnr promoter (−62 to +265, −49G to T)	This study
pMMN649	pTKlac with resA promoter (−134 to +65)	This study
pMMN650	pTKlac with fnr promoter (−62 to +265, −55A to G)	This study
pMMN651	pTKlac with fnr promoter (−62 to +265, −46A to G)	This study
pMMN652	pTKlac with fnr promoter (−62 to +265, −49G to T, −47T to C, −45G to A)	This study
pMMN653	pTKlac with fnr promoter (−62 to +265)	This study
pMMN658	pTKlac with resA promoter (−134 to +65, −50T to G)	This study
pMMN659	pTKlac with resA promoter (−134 to +65, −47A to G)	This study
pMMN670	pTKlac with resA promoter (−134 to +65, −49T to G)	This study
pSN28	pTYB2 with rpoA	27
pSN64	pTYB4 with sigA	20
pSN106	pTYB2 with rpoA encoding αCTD	26
pYZ32	pTKlac with fnr promoter (−62 to +265, −60T to C)	This study
pYZ33	pTKlac with fnr promoter (−62 to +265, −58C to T)	This study
pYZ34	pTKlac with fnr promoter (−62 to +265, −56C to T)	This study
pYZ35	pTKlac with fnr promoter (−62 to +265, −54A to G)	This study
pZY18	pTYB2 with rpoA encoding αCTD (K267A)	This study

Measurement of β-galactosidase activity.

B. subtilis cells were grown anaerobically in 2× yeast extract-tryptone (2× YT) medium (22) supplemented with 1% glucose and 0.2% KNO₃ (the starting optical density at 600 nm was 0.02) with 5 μg/ml of chloramphenicol. Samples were taken at 1-h intervals, and β-galactosidase activity was measured as described previously (19). The activity reaches its maximum level around T1 (1 h after the end of exponential growth), which was shown in Fig. 2.

Mutational analysis of the fnr and resA promoters.

Base substitutions in the fnr promoter were generated by PCR amplification using an upstream mutagenic oligonucleotide (oligonucleotides listed in Table 2) and a downstream oligonucleotide, oMN97-3 with chromosomal DNA isolated from B. subtilis JH642 as the template. Base substitutions in the resA promoter were generated by two-step PCR amplification. Briefly, two PCR products were generated from pMMN649 using one of the complementary mutagenic oligonucleotides (Table 2) together with either oMN06-305 or oMN06-306. The two PCR products were annealed and extended, and the resulting product was used as the template in a second PCR with oMN06-305 and oMN06-306. The resultant PCR products carrying the fnr (positions −62 to +265 relative to the transcription start site) and resA (−134 to +65) promoters were digested with EcoRI and BamHI and inserted into pTKlac (13) which had been digested with the same enzymes, to generate transcriptional lacZ fusions. Each mutation was verified by sequencing. The fnr-lacZ and resA-lacZ constructs were integrated into the SPβ locus of JH642 chromosome as described previously (34).

TABLE 2.

Oligonucleotides

Oligonucleotide	Sequence (5′ to 3′)a	Use
oHG-2	CTCGAAGAAAGTCACGTTGT	DNase I footprinting
oHG-9	GGATGTATTGGCAGGAAAG	DNase I footprinting
oHG-65	CACAAGATTGTCAGTTTTTTCTC	pHG45
oHG-66	GAGAAAAAACTGACAATCTTGTG	pHG45
oHG-101	CACAAGATTGGTAGTTTTTTCTC	pHG71
oHG-102	GAGAAAAAACTACCAATCTTGTG	pHG71
oMN97-3	CGGGATCCGACGATATCATGCT	fnr downstream
oMN97-4	GGAATTCATGCACAAGATTGTTA	pMMN350
oMN97-5	GGAATTCATTCCCAAGATTGTTA	pMMN351
oMN97-6	GGAATTCATTCACAAGATTGGTAGTT	pMMN380
oMN98-24	GGAATTCAGAGGTGGCGTTA	pMMN408
oMN98-25	CGGGATCCAGCAATTCATAC	pMMN408
oMN99-83	GGAATTCACTCACAAGATTGTT	pYZ32
oMN99-84	GGAATTCATTTACAAGATTGTT	pYZ33
oMN99-85	GGAATTCATTCATAAGATTGTT	pYZ34
oMN99-86	GGAATTCATTCACAGGATTGTT	pYZ35
oMN04-256	GGAATTCATTCACAAGATTTTTAGTTTTTTCTC	pMMN580
oMN04-257	GGAATTCATTCACAAGATTGTCAGTTTTTTCTC	pKM3
oMN04-263	GGAATTCATTCACAAGATTGTTAATTTTTTCTC	pKM4
oMN06-305	GAGAATTCGATGCCAGAGAGTTA	pMMN649
oMN06-306	TTAGGATCCGGTCCGAATGAATAA	pMMN649
oMN06-307	GGAATTCATTCACAAGATTGTT	pMMN653
oMN06-308	GGAATTCATTCACGAGATTGTT	pMMN650
oMN06-309	GGAATTCATTCACAAGATTGTTGGTTTTTTCT	pMMN651
oMN06-310	GGAATTCATTCACAAGATTTTCAATTTTTTCT	pMMN652
oMN06-312	ATTTCACATAACCGTCAAAAAGTAAGA	pMMN658
oMN06-313	TTCACATAACCTTCGAAAAGTAAGAAA	pMMN659
oMN06-314	TCTTACTTTTTGACGGTTATGTGAAAT	pMMN658
oMN06-315	TTTCTTACTTTTCGAAGGTTATGTGAA	pMMN659
oMN06-324	TTTCACATAACCTTTAAAAAGTAAGAAA	pCSZ4
oMN06-325	TTTCTTACTTTTTAAAGGTTATGTGAAA	pCSZ4
oMN06-326	GCTTTCTAAATTGCACATAACCTTC	pCSZ5
oMN06-327	GAAGGTTATGTGCAATTTAGAAAGC	pCSZ5
oMN06-328	CGCTTTCTAAATTTTACATAACCTTCA	pCSZ6
oMN06-329	TGAAGGTTATGTAAAATTTAGAAAGCG	pCSZ6
oMN06-332	TTTCACATAACCTGCAAAAAGTAAGAAA	pMMN670
oMN06-333	TTTCTTACTTTTTGCAGGTTATGTGAAA	pMMN670
oSN03-88	GGAATTCCATATGGAAAAAGAAGAAGATCAAAAAG	pZY18
oSN03-89	ATCGTCTTTGCGAAGTCCGAGTC	pZY18

^aRestriction enzyme sites and base substitutions are underlined.

Protein purification.

Purification of ResD and ResE proteins were described elsewhere (10). RNA polymerase (RNAP) was purified from B. subtilis MH5646 that produces the RNAP β′ subunit fused to a 10× His tag by Ni²⁺-nitrilotriacetic acid chromatography as described previously (29), followed by HiQ column chromatography (26). σ^A, α, and αCTD (between residues 239 and 314) proteins were produced in Escherichia coli carrying pSN64, pSN28, and pSN106, respectively, and purified as described in previous papers (20, 26, 27). αCTD protein with the K267A mutation was produced using E. coli carrying pZY18 and purified by the procedure used for the wild-type αCTD purification (26). pZY18 was constructed in the same manner as pSN106 was (26), except chromosomal DNA isolated from B. subtilis ORB5262 (rpoA encoding αCTD with the K267A mutation [32]) was used as the template for PCR.

DNase I footprinting.

A fragment carrying the wild-type fnr promoter (positions −136 to +20) was amplified by PCR using oligonucleotides oHG-2 and oHG-9 with pMMN408 as the template. Mutant promoters (−47T to C and −48T to G) were amplified using the same oligonucleotides with pHG45 and pHG71, respectively, as the template. Two PCR products were generated from pMMN408 using oligonucleotide pairs, oMMN99-24/oHG66 and oMMN99-25/oHG65 where oHG65 and oHG66 are complementary mutagenic oligonucleotides. The second PCR was carried out using the first PCR products with oMN99-24 and oMMN99-25. The second PCR product digested with EcoRI and BamHI was cloned into pUC18 digested with the same enzymes to generate pHG45. pHG71 was generated similarly, except oHG101 and oHG102 were used for mutagenic oligonucleotides. To label the coding or noncoding strand, one of the primers was phosphorylated with T4 polynucleotide kinase and [γ-³²P]ATP. The labeled PCR products were separated on a nondenaturing polyacrylamide gel and purified as previously described (24), and DNase I footprinting was carried out as previously described (11). The dideoxynucleotide sequence ladder was obtained by using the Thermo Sequenase cycle sequencing kit (United States Biochemical) using the labeled primer and pMMN408 as the template.

RESULTS

Mutational analysis of the fnr promoter.

As described in the introduction, a high similarity was detected in a region between positions −61 and −45 of the resA, sboA, and fnr promoters (Fig. 2A). A direct repeat (TTCA N₇ TTCA) is present between −47 and −61 of the resA and sboA promoters. However, the sequence corresponding to the promoter-proximal TTCA repeat present in resA and sboA is GTTA in the fnr promoter. We speculated that a region of the direct repeat accommodates two ResD∼P monomers (or one dimer). Our previous deletion analysis of the fnr promoter indicated that the cis-acting sequence required for ResD-dependent fnr expression is located within a region downstream from −61 and a deletion to −58 severely reduced ResD-dependent transcription activation (24), which is in good agreement with the hypothesis that ResD∼P binds to the sequence between −60 and −46. In addition, the previous observation that ResD∼P weakly binds to the fnr promoter (10) might be explained by the less conserved promoter-proximal half-site (GTTA instead of TTCA).

In order to examine this possibility, we carried out mutational analysis of the fnr promoter. The result showed that any single base substitution in the TTCA sequence between −60 and −57 resulted in sharply reduced transcription, indicating that the TTCA sequence is important for ResD-dependent fnr transcription (Fig. 2B). The substitution of C at −56, at a position adjacent to the TTCA motif, moderately affected transcription. As described above, the resA and sboA promoters carry a downstream TTCA sequence. Therefore, we next determined whether generating a promoter-proximal TTCA sequence at the corresponding site of the fnr promoter increases transcription. fnr expression was slightly increased by the change of T at −47 to C that brings the sequence (GTCA) closer to the TTCA consensus sequence. The substitution of −48T to G (resulting in GGTA) nearly abolished fnr expression, which was also expected if the TTCA sequence is important for ResD binding. However, the change of G at −49 to the consensus T (resulting in TTTA) led to a decrease in transcription, and the substitution of −46A to G (resulting in GTTG) did not impair fnr expression, suggesting that generating the direct repeat does not increase fnr transcription. In fact, simultaneous substitutions of three nucleotides (−49G to T, −47T to C, and −45G to A), which create the consensus TTCA sequence, resulted in transcription slightly lower than the wild-type promoter, although the adverse effect of the −49G to T change appeared to be compensated by the changes in −47T and −45G (see Discussion). β-Galactosidase activity of the mutant promoters in the resDE mutant is similar (around 30 Miller units), indicating that the mutations do not affect basal level of transcription, which confirmed that the affected bases are important for ResD-dependent activation.

Mutational analysis of the resA promoter.

The results of base substitutions in the fnr promoter indicated that the single TTCA site, not TTCA repeats, is essential for ResD-dependent fnr transcription. As shown in Fig. 2A, the resA and sboA promoters contain the TTCA repeats. The next question was whether both TTCA sequences are important for ResD-dependent transcription of resA and sboA or whether the promoter-distal TTCA sequence is sufficient for transcription. The result obtained from mutational analysis of the resA promoter showed that the promoter-distal TTCA site is critical for transcription as shown by the substitution of T at −60 or C at −59, which greatly reduced promoter activity (Fig. 2C). In the downstream TTCA sequence, the second T (−49T), which corresponds to −48T in the fnr promoter, is the most critical residue, as is the case with fnr. The −48C to T change led to 60% reduction of transcription, which is consistent with the result that the −47T to C change in fnr moderately increased transcription. In contrast, the base substitution of T at −50 or A at −47 did not show any significant effect on resA transcription. These results support the conclusion obtained from the mutational analysis of the fnr promoter that TTCA (−60 to −57 of fnr) and −48T are essential for ResD-dependent transcription activation.

DNase I footprinting analysis of the fnr promoter.

In order to determine whether ResD∼P binds to the TTCA sequence in fnr, we carried out DNase I footprinting analysis of the wild-type and mutant fnr promoters. We used two mutant promoters with T−48G and T−47C, the nucleotide substitutions of which resulted in defective and slightly increased transcription, respectively (Fig. 2B). It was noticed that cleavage patterns of free DNA between positions −46 and −50 reproducibly changed when T at −48 was substituted with G (see Fig. 3 to 5). In the wild-type promoter, sensitivity to DNase I was similar among nucleotides at −47, −48, and −50 of the coding strand, whereas in the mutant promoter the position at −50 was hypersensitive to DNase I attack. On the noncoding strand of the wild-type promoter, cleavage sites were located at −46, −48, and −50. When T at −48 was substituted with G, residues −48 and −50 became resistant to DNase I, and a new hypersensitive site appeared at position −49. In the mutant T−47C promoter, a cleavage pattern around this region was similar to that of the wild-type DNA, except −46 of the coding strand became slightly more sensitive to cleavage.

FIG. 3.

DNase I footprinting analysis of the wild-type (wt) and mutant (T−47C and T−48G) promoters in the presence of ResD∼P and RNAP. The coding (A) and noncoding (B) strands of each promoter fragment were labeled as described in Materials and Methods. ResD was phosphorylated with 2 μM ResE. Two concentrations of ResD, 3 μM (+) or 6 μM (++), were used (−, none). B. subtilis RNAP (50 nM) was used with 50 nM purified σ^A. Regions of strong and weak protection from DNase I digestion are marked by solid and dotted lines, respectively. A box and an arrow show TTCA (positions −60 to −57) and T at −48, respectively, which are critical for fnr transcription. A hypersensitive site detected in the T−48G promoter is shown with an asterisk. Dideoxynucleotide sequencing reactions are also shown, and nucleotide positions are marked relative to the transcription start site.

FIG. 5.

DNase I footprinting analysis of the wild-type (wt) and mutant (T−47C and T−48G) promoters in the presence of ResD∼P and α. DNase I footprinting analysis was carried out in a manner similar to that described in the legend to Fig. 3, except that 3 μM α was used instead of RNAP. A box and an arrow show TTCA (−60 to −57) and T at −48, respectively. A hypersensitive site detected in the T−48G promoter is shown with an asterisk. Dideoxynucleotide sequencing reactions are also shown, and nucleotide positions are marked relative to the transcription start site.

Figure 3 indicated that ResD∼P alone did not bind to either the wild-type or mutant promoters (lanes 9, 14, and 19). Increasing the concentration of ResD∼P up to 12 μM did not protect the fnr promoter, although 4 μM ResD∼P fully protected nasD (data not shown). RNAP alone induced reproducible changes in cleavage patterns of the promoter DNA (lanes 6, 11, and 16), suggesting a distortion of DNA. In the presence of both ResD∼P and RNAP, the region between positions −63 and −40 was strongly protected in the wild-type and T−47C promoters, and a region downstream from −32 was also weakly protected (lanes 7, 8, 17, and 18). In contrast, the protection was not detected in the T−48G promoter (lanes 12 and 13), which is consistent with the in vivo result showing that the T−48G mutation impaired transcription.

The region protected in the presence of ResD∼P and RNAP included the critical T at −48 (shown by an arrow in Fig. 3) and the TTCA sequence (−60 to −57; marked with a box), which we proposed as the ResD-binding site. This result indicated that ResD∼P in the presence of RNAP binds to both the TTCA sequence and the sequence around −48 or that ResD∼P binds to TTCA, while a subunit of RNAP makes contact with T at −48. If the latter is the case, the C-terminal domain of the α subunit is a likely candidate, given the position of −48 with respect to the promoter DNA. Consistent with this notion, a region around −48 was strongly protected by purified α (Fig. 4A and B). Some weaker protection by α was observed in upstream and downstream regions of −48, which may be caused by nonspecific binding. When αCTD (amino acid residues 239 to 314) was used instead of α, the extended protection disappeared and only the region between −38 and −50 was protected (Fig. 4C and D). This result suggested that αCTD interacts with the region around −48; however, the substitution of T at −48 to G did not decrease the binding of either α or αCTD itself, suggesting that T at −48 is not involved in a direct contact with αCTD. A possible effect of the T−48G mutation on fnr transcription will be further discussed in the Discussion.

FIG. 4.

DNase I footprinting analysis of the wild-type (wt) and mutant (T−47C and T−48G) promoters in the presence of α or αCTD. The coding (A and C) and noncoding (B and D) strands of each promoter fragment were labeled as described in Materials and Methods. (A and B) Increased concentrations of α (0.75, 1.5, 3, and 6 μM) were used. −, none. (C and D) Increased concentrations of αCTD (12.5, 25, and 50 μM) were used. Regions of strong and weak protection from DNase I digestion are marked by solid and dotted lines, respectively. A box and an arrow show TTCA (positions −60 to −57) and T at −48, respectively. A hypersensitive site detected in the T−48G promoter is shown with an asterisk. Dideoxynucleotide sequencing reactions are also shown, and nucleotide positions are marked relative to the transcription start site.

When ResD∼P was present, the protection by α was further extended to a region including the TTCA sequence in the wild-type promoter and the T−47C promoter (Fig. 5). Consistent with the footprinting results with ResD∼P and RNAP (Fig. 3), no cofootprinting with ResD∼P and α in the T−48G promoter was observed. These results indicated that interaction between α and ResD∼P stabilizes ResD∼P binding to the fnr promoter and strongly suggested that ResD∼P makes contact with TTCA and the change of T at −48 to G adversely affects formation of the RNAP-ResD∼P-DNA ternary complex.

Effects of amino acid substitutions in αCTD on fnr expression.

The results of DNase I footprinting analysis described above strongly suggested that interaction of ResD∼P with α, very likely αCTD, is important in transcriptional activation of fnr. To examine this possibility, we took advantage of a recently constructed αCTD mutant library (32). This mutant library was constructed by replacing the wild-type rpoA with a mutant allele bearing an alanine codon substitution. B. subtilis strains carrying substitutions E255A, R261A, R268A, R289A, and G292A could not be isolated probably because these residues are essential for transcription of housekeeping genes. Residues 269, 278, and 301 are alanine in the wild-type α, and we substituted only residue 269 with isoleucine. Within the residues 251 to 314, substitutions Y263A, K267A, A269I, and N290A strongly reduced fnr expression (Fig. 6). I253A, E254A, V260A, L266A, and G309A had a moderate effect on transcription (30 to 50% of the wild-type expression). A previous study showed that substitutions of these residues had no significant effect on expression of rpsD encoding ribosomal S6, except that Y263C and Y263A led to 50% and 40% reduction in rpsD expression, respectively (32; H. Geng and M. M. Nakano, unpublished results), indicating that the effect of these substitutions in transcriptional activation is specific to fnr.

FIG. 6.

Effects of single alanine substitutions in αCTD on fnr-lacZ expression. The wild-type (WT) strain and each αCTD mutant were grown anaerobically in 2× YT supplemented with 1% glucose and 0.2% KNO₃, and β-galactosidase activities were measured as described in Materials and Methods. Experiments were repeated two or three times, and the average of maximal activities around T1 (1 h after the end of exponential growth) is expressed as percentage of the activity in cells carrying the wild-type rpoA gene. Black bars indicate substitutions that reduced the expression to less than 20% of the wild-type level, and gray bars indicate substitutions that reduced expression to 30 to 50%. Substitutions of residues with no data indicate those substitutions not obtained in B. subtilis (32).

K267 of αCTD is essential for interaction with ResD∼P.

One possible explanation of the reduced fnr expression in cells producing certain αCTD mutants is that the mutated residues are essential for interaction with ResD∼P. To examine this possibility, DNase I footprinting analysis of the wild-type fnr promoter was carried out using the αCTD (K267A) protein. As shown in Fig. 7, the mutant αCTD binds to the region between −38 and −50 with an affinity similar to that of the wild-type αCTD (lanes 6 to 8 and 12 to 14). In the presence of ResD∼P, the wild-type αCTD was able to bind to DNA at a concentration lower than the αCTD alone, and the protected region extended to the upstream region including the TTCA sequence (lanes 9 to 11), indicating that interaction between αCTD and ResD∼P stabilizes binding of both proteins to DNA. In contrast, ResD∼P did not significantly increase binding of the mutant αCTD to DNA, and the mutant αCTD did not stimulate binding of ResD∼P to the TTCA sequence (lanes 15 to 17). We concluded from these results that K267 of αCTD is essential for interaction with ResD∼P at the fnr promoter and that the interaction accelerates transcription initiation of fnr.

FIG. 7.

DNase I footprinting analysis of the wild-type fnr promoter in the presence of ResD∼P and the wild-type or K267A αCTD. The coding strand of the fnr promoter was labeled as described in Materials and Methods. ResD (6 μM) was phosphorylated with 2 μM ResE. An increased concentration (25, 50, and 100 μM) of the wild-type (wt) or K267A αCTD was included when indicated. The dotted line indicates the region protected by αCTD, and the solid line indicates the region protected in the presence of the wild-type αCTD and ResD∼P. The box and arrow show TTCA (−60 to −57) and T at −48, respectively. Dideoxynucleotide sequencing reactions are also shown, and nucleotide positions are marked relative to the transcription start site. −, none.

DISCUSSION

On the basis of the results described here, we propose that ResD∼P binds to the TTCA sequence in the fnr and resA promoters and likely the sequence in the sboA promoter, which supports the hypothesis that TGAA in the proposed consensus sequence a (Fig. 1A) is important for binding of ResD∼P to activate transcription of ctaA, ctaB, and yclJ. Given that either TTCA or TGAA is likely used as a ResD-binding site in divergently transcribed ctaA and ctaB as well as yclI and yclJ, we assume that ResD∼P could bind to DNA in both orientations. ResD∼P dissociation constants for the ctaA genes calculated by Zhang and Hulett were as follows: 8.9 nM for ctaA1, 200 nM for ctaA2, and 47 nM for ctaA3 (31). This may suggest that ResD∼P binds to a direct repeat (TTCA N₇ TTCA or TGAA N₇ TGAA present in ctaA1 and ctaA3) with an affinity higher than that to a single TGAA in ctaA2. TTCA N₇ TTCA is similar to a half-site of the PhoP-binding sequence [TT(A/T/C)ACA N_4-6 TT(A/T/C)ACA] that accommodates a PhoP dimer (12, 16). Previous work showed that resA expression is positively regulated by the ResDE and PhoPR signal transduction systems and that PhoP∼P binds to the resA promoter region that includes the direct repeat sequence shown in Fig. 2 (2), indicating that the same sequence (or overlapping sequence) of resA is used for binding of the two response regulators.

We have shown by hydroxyl radical footprinting that ResD∼P binds to five distinct regions of nasD (positions −90 to −45) and hmp (−80 to −40), and we have shown by deletion analysis that cis regions required for activation by ResD∼P reside downstream of −87 (nasD) or −67 (hmp). Within these regions, no TGAA sequence is identified, but a single TTCA sequence (−87 to −84 in nasD and −48 to −45 in hmp) is detected, suggesting that ResD∼P binds to and activates these promoters in a way different from cta and ycl that contain the direct repeat. In light of consensus sequence a and b, we propose TTGTGAAN₃TTTN₄A (Fig. 1C) as a consensus ResD box (consensus sequence c) for fnr, nasD, and hmp. Consensus sequence c is composed of two half-sites of 9 bases—site a (TTGTGAANN), which is a part of consensus sequence a, and site b (NTTTN₄A), which is similar to the 3′ end of consensus sequence b. Figure 1C shows a possible alignment of these sites in the nasD, hmp, and fnr promoters. Each regulatory sequence listed in Fig. 1C was shown to be sufficient for full activation by ResD∼P (24). Two consensus sequences (8 of 11 matches) are present in nasD, which likely accommodate two ResD∼P dimers. Upstream of the two putative ResD∼P dimer-binding sites, half-site a is present in the opposite orientation. The hmp and fnr promoters each contain a single full-site and one half-site a that is oriented oppositely to the full site. The half-site resides upstream (in fnr) or downstream (in hmp) of the full site. Site a, but not site b, is able to serve as a stand-alone half-site, suggesting that ResD∼P binds to site b only after site a is occupied by ResD∼P. We reexamined the result of base substitutions in the fnr promoter in the context of the newly proposed consensus ResD-binding sequence. Among the bases in fnr-2a (TTGTTAG), the first G (−49), and the third T (−48) in particular, are important for ResD-dependent transcription because the change from TTGTTAG to either TTTTTAG or TTGGTAG led to moderate and severe reduction of fnr transcription (Fig. 2A), respectively. This result is in good agreement with the proposed consensus sequence half-site a (TTGTGAAN₂). The change of A at −46 to G (resulting in TTGTTGG) and G at −45 to A (TTGTTAA) did not significantly decrease or increase transcription, indicating that the two A residues in the proposed half-site a are not critical. We showed that the adverse effect of the G−49T mutation was suppressed by the T−47C and G−45A mutations (Fig. 2B). We interpreted this result as meaning that the TTCA direct repeat does not increase fnr transcription; however, we now suggest another interpretation. The G−49T substitution (TTTT) impairs the function of the critical TTGT sequence in consensus sequence c, thereby adversely affecting fnr transcription. Introduction of the T−47C mutation in the G−49T promoter generates TTCA, and the resultant TTCA N₇ TTCA sequence now functions as a ResD-binding site like those found in cta and ycl. Further studies of base substitutions, including the first two T residues in the TTGTGAAN₂ sequence, are required to properly evaluate the putative consensus sequence c and to understand how differences in the half-site arrangement and orientation in various promoters affect binding and activation by ResD∼P.

This study also uncovered the critical role of αCTD for ResD-dependent fnr transcription. Transcriptional activators bind to specific DNA sequences and often interact with αCTD. This protein-protein interaction increases the affinity of RNAP to the promoter site to initiate transcription (reviewed in reference 4). Such an activator-dependent transcription is usually observed in promoters lacking the consensus −35 sequence as is the case with ResD-controlled promoters. The results of our footprinting analysis showed that ResD∼P alone does not bind to the fnr promoter, unlike the nasD or hmp promoter (10) and the cta (31) or ycl (11) promoter. Hence, a simple recruiting model in which ResD∼P binds to the ResD box and recruits RNAP to the promoter seems inapplicable to explain how ResD∼P activates fnr transcription. A ternary complex by ResD∼P, RNAP, and fnr is formed presumably through direct interaction between ResD∼P and αCTD. Screening of the αCTD alanine substitution library identified residues that are important for fnr activation by ResD∼P. Among these residues, E254, V260, Y263, K267, and A269 form a surface-exposed patch (Fig. 8). Residues D258, K271, A272, E273, and L289 in E. coli αCTD, which correspond to E254, K267, R268, A269, and M285, respectively, in B. subtilis αCTD, are proposed to interact with the Fis transcriptional activator (1, 18). Therefore, ResD∼P likely interacts with a surface of αCTD similar to the “273 determinant” that interacts with Fis. N290, which is important for ResD-dependent activation of fnr, is not a part of this surface patch proposed to interact with ResD∼P, and might be involved in interaction with DNA. In fact, the corresponding residue (N294) in E. coli αCTD was shown to interact with DNA (9). Our DNase I footprinting analysis demonstrated that αCTD carrying the K267A mutation binds to fnr DNA nearly as well as the wild-type αCTD does; however, the mutant αCTD, unlike the wild-type protein, does not facilitate binding of ResD∼P to the TTCA sequence, confirming our hypothesis that the surface including K267 interacts with ResD∼P. The residue K267, as well as Y263, C265, and L266, was shown to be required for optimal ComA-dependent activation of the srf operon encoding proteins that function in the control of competence development and in nonribosomal peptide synthesis (32). The C265A mutation that most severely affected srf expression had no effect on fnr expression. Conversely, A269I conferred severely reduced fnr expression but only moderately affected srf transcription. Therefore, it is likely that ResD∼P and ComA∼P interact with overlapping but distinct surfaces of αCTD.

FIG. 8.

Structure of B. subtilis αCTD (28) indicating the residues identified as important for ResD-dependent activation of fnr. Amino acid residues with more than twofold effect are dark gray.

The ResD-αCTD-DNA ternary complex was not formed with fnr carrying the T−48G mutation, suggesting that the in vivo defect in transcription caused by the mutation could be due to the lack of cooperative binding between ResD∼P and αCTD. Purified αCTD protects a region around −48, raising the possibility that αCTD might make direct contact with T at −48. However, the change of −48T to G did not weaken the binding of αCTD (Fig. 4), instead we observed DNase I-hypersensitive sites unique to T−48G DNA (Fig. 3 to 5). One could imagine that the substitution of T at −48 with G introduces a kink around −50, which results in interference between ResD∼P and RNAP, thus disrupting the ResD∼P-RNAP-DNA ternary complex.

Although αCTD alone binds to a region between −50 and −38, it remains unclear whether the protection of the region in the presence of αCTD and ResD∼P is caused by binding of αCTD, ResD∼P, or both because the sequence between −51 and −34 contains two half-sites that show similarity to consensus sequence c (Fig. 1C). One possibility is that ResD∼P and αCTD interact by binding to the same region but to different faces of the DNA helix as seen with BvgA and αCTD in E. coli (3). DNase I footprinting experiments in Fig. 3 showed that hypersensitive sites that were observed in the presence of RNAP disappeared when both RNAP and ResD∼P were present, indicating that the interaction with ResD∼P and αCTD remodels RNAP-promoter interaction to initiate transcription.