Identification of ς^B-Dependent Genes in Bacillus subtilis Using a Promoter Consensus-Directed Search and Oligonucleotide Hybridization

Abstract

A consensus-directed search for ς^B promoters was used to locate potential candidates for new ς^B-dependent genes in Bacillus subtilis. Screening of those candidates by oligonucleotide hybridizations with total RNA from exponentially growing or ethanol-stressed cells of the wild type as well as a sigB mutant revealed 22 genes that required ς^B for induction by ethanol. Although almost 50% of the proteins encoded by the newly discovered ς^B-dependent stress genes seem to be membrane localized, biochemical functions have so far not been defined for any of the gene products. Allocation of the genes to the ς^B-dependent stress regulon may indicate a potential function in the establishment of a multiple stress resistance. AldY and YhdF show similarities to NAD(P)-dependent dehydrogenases and YdbP to thioredoxins, supporting our suggestion that ς^B-dependent proteins may be involved in the maintenance of the intracellular redox balance after stress.

In Bacillus subtilis, the alternative sigma factor ς^B tightly controls a large stationary-phase and stress regulon (7–10, 16, 17, 31). The functional characterization of members of the ς^B regulon led to the assumption that the proteins encoded are involved in the protection of DNA, membranes, and proteins against oxidative damage, which might represent an important component within the complex stress response (6, 14). Moreover, general stress proteins appear to contribute to survival of extreme environmental conditions such as severe heat or osmotic stress, repeated freezing and thawing, as well as acid or alkaline shock of starving B. subtilis (15, 32). In summary, the expression of the ς^B-dependent general stress regulon is expected to provide an unspecific, multiple and prospective stress resistance to nongrowing B. subtilis cells in anticipation of future stress (for a review, see reference 17).

The discovery and characterization of new ς^B-dependent genes will certainly improve our understanding of the physiological role of the entire ς^B regulon in B. subtilis. So far, members of the ς^B regulon have been defined mainly on the basis of transposon mutagenesis or identification of protein spots from two-dimensional protein gels (5, 10, 11, 31). In this study, we used the combination of a consensus promoter-based search for new ς^B targets and an oligonucleotide hybridization to detect new members of the general stress regulon of B. subtilis.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The B. subtilis wild-type strain 168 (3) and its isogenic sigB mutant ML6 (22) were cultivated in a shaking water bath at 37°C in a minimal medium with glucose as the carbon source previously described (29). Ethanol stress was imposed by adding ethanol to exponentially growing cells to a final concentration of 4% (vol/vol).

Analysis of transcription.

Total RNA of both B. subtilis strains was isolated by the acid phenol method of Majumdar et al. (25), with modifications as previously described (31). Decreasing amounts of total RNA were transferred onto a positively charged nylon membrane by slot blotting. Hybridizations at 50°C with digoxigenin-end-labeled oligonucleotides specific for the various genes (Table 1) and detections were performed as instructed by the manufacturer (Boehringer Mannheim). For 3′-end labeling, 100 pmol of each oligonucleotide was incubated for 1 h in a 20-μl reaction mixture containing 12.5 nmol of digoxigenin-11-ddUTP, 2.5 mM cobalt chloride, 1× reaction buffer and 12.5 U of terminal transferase. The whole volume was used for the hybridization. Primer extensions were performed with radiolabeled primers as previously described (33) (Table 1). A DNA-sequencing ladder was generated with the same primers, using PCR products as a template (Table 1).

TABLE 1.

Oligonucleotides used in this study

Gene-specific oligonucleotide or primer	Sequence (5′→3′)
Positive control oligonucleotides
ctc	CATGTCCTGAAGTACGGATATTCC
gspA	TTACCTCTCTCTCCTGATCC
trxA	CAAGGTCCGCACCAAGGAGC
Oligonucleotides specific for tested genes
aldY	TTCCGTCGTGCTCTTGGCCCACTC
lctE	CGCAAATGCATAACTGCTTCC
yabJ	CCGCAAACTGTTCCATATCCGCG
yacL	CAGCTTCTTGTCTTGAACCTCATG
ycnH	TGAGGAATCGTTTTGATCTCCTCG
ydaS	GGCGAGGCTCGGTCCCCATGTGCC
ydaT	CCCATTCCTTCGCTTTGCTTGTCGC
ydbP	ATATTCATGCGTGTGCAGTCTGGG
ydhK	AGAGATAACATCAGAATTCCCAGTGC
yfhK	ATTTGAGCCCAATCGGCATTTACG
yflA	TAGCGGCGGACCCCACAGCCAATC
yhaR	GCGTGATGGGCATCAGGCCGATTC
ydhF	TGGAGATATCAGCCCCCTCTTTAGA
yjbC	CCCTCTTGCATCCTTAGACACATAC
yjgB	GGCATTTCCCCTTTATAGGCGGTG
ykgA	GTGCTGGTCGTTAGCGGTTTTTAC
yotK	CATCTTGTTTAACAGTGTTTGAATTGAATG
yoxA	GCCATACAATGTTGGTGTGTCGTG
yoxC	CTTGTTTTGGATCACGGTAACTCC
ypuB	CTTAGCACCCAGTTTAACTTTTCTTG
yqhA	TACAAACTTCCAGCCATTGTTGCG
yqhQ	TCCGTTCTTCTGATGGCTGTAACG
yqhZ	GCTTCCAATTCACCAGATGCTTGGAG
yqiS	CACTTCCTCATCCTCAGCATGAGC
yqxL	TCTTGCCTTACTTCTTTCTTGAGG
yrvD	CGGCCATTCTGATCTTCCAAACAGG
ysdB	TTAGGATTCGCGACATATTTGACC
yvrE	CTGTCATCCCGAAGATGATACAGG
yxkO	GGCAATGGCCCGGTACGTCTCTTC
yycD	TGTCGGCGCTACATTCTCCACCTC
Primers for primer extension and generation of PCR products
yacL-PE	AATGAAGAACGCCTGAACTATTCG
yhdF-PE	TTCCTCGATAATCCTCGTCCTCTG
yjbC-PE	TCAATTGGAAAGTACTCGCTAAGC
yacL-forward	GGACAGAAGGGCATCTGATC
yacL-reverse	TCTTCTCCCTTTACAGCGCC
yhdF-forward	GTGGTGCGCCCATGTACTGG
yhdF-reverse	CACCACACAAAAAACCAGCTC
yjbC-forward	GCAGGAAGTTATTCCCGAGC
yjbC-reverse	CATCGTTGTTTTCCAAGACCTC

General methods.

The BSOrf homology search tool of the database of the B. subtilis genome sequencing project was used for deriving of the primers. Database searches were performed with the Blast program (2).

RESULTS AND DISCUSSION

Knowledge of the sequence of the entire B. subtilis genome (24) provides an excellent basis for a comprehensive analysis of gene expression by using global approaches such as two-dimensional protein electrophoresis (proteome analysis) (5), chip technology (transcriptome analysis) (13), and consensus sequence-based searches for members of individual regulons (20). In this report, we used the genome sequence for the identification of new ς^B-dependent genes.

After an initial manual evaluation, the whole genome was searched with the sequence pattern DGWKTNDN_12–15GGRWAW (D = A, G, T; W = A, T; K = G, T; R = A, G; N = A, C, G, T). Only targets deviating not more than three nucleotides from the consensus of ς^B-dependent promoters GTTTWWN_12–15GGGWAW and lying within 400 bp upstream of predicted open reading frames were considered for further analyses. The rather large deviation from the consensus was intentionally permitted so that we could identify even weak ς^B-dependent promoters.

To demonstrate that the presence of a putative ς^B promoter indeed conferred ς^B-dependent stress induction, total RNA from isogenic wild-type and sigB mutant cells was isolated before (control) and after ethanol stress (4% [vol/vol]) and hybridized with a digoxigenin-labeled oligonucleotide designed against the open reading frame downstream of the potential ς^B target.

Prior to the screening, oligonucleotide probes against known ς^B-dependent genes were used to prove the specificity of the hybridization assay. Rigorously ς^B-dependent genes such as gspA displayed a signal only after treatment of the wild type with ethanol; no signal was observed with RNA from growing cells or from the sigB mutant as the template (Fig. 1A; Table 2, group A) (4). ctc illustrates the pattern for genes the transcription of which is driven by the vegetative sigma factor ς^A in addition to ς^B (18). Basal-level expression was easily detected for such genes in the wild type and the sigB mutant prior to stress, whereas induction by ethanol stress absolutely required ς^B (Fig. 1A; Table 2, group A). Genes such as trxA, which are subject to a double or multiple control, may retain induction even in the sigB mutant albeit at a reduced level (Fig. 1A; Table 2, group A) (28).

FIG. 1.

Screening of potential targets for ς^B-dependent transcription by slot blot hybridization with digoxigenin-labeled oligonucleotides specific for the corresponding genes. The lanes contain decreasing amounts (3, 1.5, and 0.75 μg) of total RNA isolated from wild-type bacteria and sigB mutant cells before (controls 1 and 2, exponential growth) and 10 min after the imposition of ethanol stress. (A) Hybridization pattern of known ς^B-dependent genes. Whereas the stress induction of ctc and gspA strictly requires ς^B, induction of trxA can occur at both ς^B-dependent and ς^B-independent promoters. (B) New ς^B-dependent genes requiring ς^B for induction by ethanol stress. (C) Genes displaying induction by ethanol stress even in the absence of ς^B. (D) Examples of target genes that lack induction by ethanol stress despite displaying a putative ς^B promoter.

TABLE 2.

Defined ς^B promoters of the control genes and promoters of the presumable new ς^B-dependent genes found in this study

Group and promoter sequencea	Gene	Protein		Presumed function and/or similarity	Probably cotranscribed gene(s)
		Molecular mass (kDa)	pI
A (ςB-dependent genes [positive controls])
CGAGGTTTAAATCCTTATCGTTAT-GGGTATTGTTTGTAATAG-N33-ATG	ctc	22.0	4.24	Belongs to the L25P family of ribosomal proteins
ACGTGTTTATTTTTTTGAAAAA---GGGTATGTAACTTG-TAC-N23-ATG	gspA	33.5	5.19	Similar to lipopolysaccharide-1,3-galactosyltransferases and to lipopolysaccharide-1,2-glucosyltransferases
TCAGGTTTTAAAACAGCTCCGGCA-GGGCATGGTAAAGTACAT-N241-ATG	trxA	11.39	4.34	Thioredoxin
B (genes controlled by ςB)
ATTCGTTTACATATTGGCTTCAG--CGGAAATAGAAGAAGACA-N25-ATG	aldY	52.8	5.07	Probable aldehyde dehydrogenase, contains an aldehyde dehydrogenase signature (glutamic acid active site)
AAGTGTTTCGAAATGATCAGGAGC-GGATATTGATTGGTAAAT-N21-ATG	ydaS	8.64	12.22	Membrane protein, unknown function
ATGCGTTTTTATTTTTCACCTGC--GGGTACCATTTTTATAAA-N19-ATG	ydaT	16.5	6.38	Unknown	ydaS
GTTTGATTACAGCCAGAGCTGTCTGCAGGGAAACATTATGTTCTG-N193-ATG	ydbP	12.4	4.42	Similar to thioredoxins of A. fulgidus and S. cerevisiae
TTATGTTTGGCTTTGCAAACAAAG-GGGAATAGGACAAACTGC-N76-TTG	ydhK	22.5	7.81	Protein with 1 transmembrane domain, unknown function
ACATGTTTCACCAGCCTGTCAATCAGGGAATACCACTTATATC-N26-ATG	yfhK	18.7	10.37	Protein probably with signal sequence, similar to proteins with unknown function
AAAGGTTTATGTTTTTCCATCTAT-GGGAAATGATTCATAAAC-N17-ATG	yflA	50.6	10.4	Integral membrane protein, similar to amino acid carrier protein
TGGCGTTTATTCATTTCTGTCGTG-TGGTAACGTTCAGTATCA-N24-GTG	yhdF	31.4	5.9	Similar to glucose-1-dehydrogenase, belongs to the short-chain dehydrogenases/reductases family
ACATGTTTCGTTGCAAAAACACG--GGGAAACGGAATGGTAGA-N19-ATG	yjgB	20.8	6.68	Membrane lipoprotein, unknown function
AATGGTTTAATGATTTTCATGATGAGGGAATAATAAATGTATT-N25-ATG	ykgA	29.6	4.88	Function unknown, similar to streptococcal acid lipoprotein	ykgB
TCTAGTTTCTCTTTTTAAAAGAGTAGGGTATTGCAAACAACAG-N58-ATT	yotK	7.2	7.69	Unknown
AACGGTGTTTTTTTATTTGATAG--GGGAAAATATAAAATGGA-N64-ATG	yoxA	37.2	6.00	Similar to hypothetical proteins of B. subtilis	pbp
TTCTGATTAAAAAAACGGATACA--GGGTAATGACATAAGAAA-N12-ATG	yoxC	11.39	10.37	Protein probably with signal or transmembrane sequence, similar to the general stress protein YtxG	yoxB, yoaA
TGCTGATTATACAAAAAGTGGATT-GGGAATGATAAAAGAACA-N21-ATG	ypuB	7.23	4.24	Protein with 1 transmembrane domain, unknown function	ypuC
TTCGGTTTTTCACCTGTCCAGAAACTGGGCTAGCTGGATTCGAAC-N136-ATG	yqhA	31.8	5.17	Membrane protein with 1 transmembrane domain, similar to RsbR of B. subtilis
CTGCGAGTAAAATTTGAAAATAAC-GGGTATAATGCATGTAGG-N118-ATG	yqhQ	36.0	9.71	Integral membrane protein with unknown function	yqhP
ACTGGTTTAGTGACGCGGTTATT--GGGCAATTAAAGAATAAA-N25-ATG	yqxL	37.7	10.00	Membrane protein, similar to divalent cation transport proteins
CCATGTTTAAAACCAGTCTTGATT-GGGAAAGTTACCTCAATA-N21-ATG	yrvD	12.2	9.16	Membrane protein with unknown function	yrvE, apt
ATAAGTTTTTCAGCTTTTTAAAAA-GGGAAAATAAAAAAAACA-N24-ATG	ytkL	10.9	6.41	Similar to proteins with unknown function
AGTGGTTTGGACACCTCTTTGCC--GGGAATAACAATATATAG-N24-ATG	yvrE	33.2	4.65	Similar to the eukaryotic SMP-30/CGR1 protein family
TTTTGTTTGAAAAAGAAAAGGGAC-AGGAAAAATAGGAAAAGA-N19-ATG	yxkO	29.9	5.8	Membrane protein with unknown function, similar to hypothetical family of uncharacterized proteins (UPF0031)
GATCGTTTCGGACAGTAACAAGGC-GGGAAAAATGCAATAAAA-N19-ATG	yycD	7.4	4.81	Unknown
C (complex-regulated genes, ςB could be involved)
AATTGAATTGACCGGATCTTGGCC-TGGAAAACCAATGACTAA-N363-ATG	lctE	34.9	5.64	l-Lactate dehydrogenase	lctP
TTTCGGTTAAAACCTTATGAATAC-GGGTATATTAATGTTGGT-N74-ATG	yacL	40.8	5.01	Membrane protein, similar to hypothetical proteins	yacM, yacN, gltX
AACGGATTACTTTTGCTGACAGC--GGGAATTAACGGTAATAT-N153-ATG	ycnH	50.2	4.8	Similar to the succinate-semialdehyde dehydrogenase GabD of E. coli, contains an aldehyde dehydrogenase signature (cysteine active site)
GGCTGTTTAAACAAGAAGAAAATG-GGGTATATCTAAAAGTAT-N30-ATG	yjbC	23.1	5.22	Unknown	yjbD
TTCGGAATTTATATTACAAAAT---GGATAAATATGGCCTTGC-N143-ATG	yqiS	31.7	8.18	Protein with 1 transmembrane domain, similar to phosphate butyryltransferases, belongs to the phosphate acetyltransferase and butyryltransferase family	yqiT, yqiV, yqiW, bfmBAA, bfmBAB, bfmBB
ATACGACTATTTCACTTGAAAATC-GGGTATATGTTTTTACAG-N83-ATG	ysdB	15.6	9.85	Membrane protein with unknown function
D (noninducible genes preceded by a putative ςB promoter sequence)
ATGAGTTTAACGCAAATGTGGC---GGGAATCGGCGTCTTAGT-N157-ATG	yabJ	13.6	5.24	Unknown, similar to proteins with unknown function
AAAGGTTTATTTGGGGATGGCG---GGGAATGTATAGGAGAAA-N34-ATG	yhaR	29.5	6.27	Protein with 1 transmembrane domain, similar to 3-hydroxybutyryl-coenzyme A-dehydratase	yhaQ, yhaP
AGGTGTGTGAAAATATTCGGTAATAGGGTAAAAAAACCTTGAT-N86-ATG	yqhZ	14.8	4.8	Unknown, similar to NusB (transcription termination)	folD

^aBoldface sequences represent, from left to right, −35 region, −10 region, and transcriptional start site (+1). Start codons are indicated on the far right. 

Of the 28 genes induced by ethanol stress, 22 were induced only in the wild type and required ς^B for induction (Fig. 1B). The number of new ς^B-dependent genes may even exceed 22 since some of the promoters were located in front of operons containing probably one or two additional genes (e.g., ykgA/ykgB, ypuB/ypuC, yqhQ/yqhP, and yoxC/yoxB/yoaA). Although these oligonucleotide hybridizations clearly establish the ς^B dependency of the induction, they do not provide information about the precise location of the ς^B-dependent promoter. Therefore, primer extension experiments were performed for selected genes in order to map the 5′ ends of the corresponding transcripts and to ascertain that transcription initiated downstream of the potential ς^B-dependent promoters after the imposition of ethanol stress. Figure 2A displays the results obtained for yhdF as a representative example. The 5′ ends of yhdF (Fig. 2A) were barely detectable with RNA from exponentially growing bacteria, but the intensity of the signal dramatically increased after stress. In agreement with the ς^B dependency, the signal was not observed with RNA isolated from a sigB mutant.

FIG. 2.

Mapping of the 5′ ends of yhdF (A), yacL (B), and yjbC (C) during exponential growth (co), after exposure to ethanol stress (e) and salt stress (s), by primer extension analyses with RNA from wild-type (168) and sigB mutant (ML6) cells. The transcriptional start sites are marked with asterisks. The −35 and −10 regions (underlined), the 5′ ends of the messages, and the translational start sites are in boldface; putative ς^W-recognition sequences in front of yjbC are underlined.

The remaining six genes were induced by ethanol in the wild type as well as in the sigB mutant, indicating a more complex control (Fig. 1C). However, induction seemed to be more pronounced in the wild type than in the sigB mutant, implying an involvement of ς^B (Fig. 1C). Evidence for a role of ς^B in ethanol induction was obtained by primer extension experiments, two of which are displayed in Fig. 2B and C. One 5′ end of the yacL mRNA was mapped to a site which is preceded by a putative ς^B-dependent promoter. The signal was exclusively detected after ethanol stress in the wild type and was missing with RNA from the sigB mutant as template (Fig. 2B). A second ς^B-independent start site was mapped further downstream (data not shown).

Using a yjbC-specific primer, we mapped two 5′ ends of the yjbC mRNA. The intensity of both reverse transcripts increased after ethanol treatment (Fig. 2C). The sequence preceding the downstream signal resembled a typical ς^B-dependent promoter, and this start site was not used in the sigB mutant. The promoter sequence located in front of the upstream start site resembled ς^W-dependent promoters (19), and the intensity of the signal increased even in a sigB mutant after ethanol stress (Fig. 2C). ysdB, a second gene displaying partially ς^B-dependent ethanol induction, also possesses a ς^W-dependent promoter (21) and moreover contains fairly well conserved recognition sequences for ς^B (Table 2,C). ς^W is a new extracytoplasmic function sigma factor recently described by Huang et al. (19–21). The authors suggest that ς^W activates a large stationary-phase regulon that functions in detoxification or production of antimicrobial compounds. Our data indicate that in addition to entry into stationary phase, ethanol stress induces transcription at ς^W-dependent promoters (Fig. 1C and 2C) and that genes such as yjbC seem to be subject to a control by ς^B and ς^W.

The new presumably ς^B-dependent promoters as well as the defined promoters of the control genes are summarized in Table 2. An alignment of all the ς^B-dependent promoters currently available yields consensus sequences GTTTaa and GGG(A/T)A(A/T) for the −35 and −10 regions, respectively, which are separated by 13 to 15 nucleotides. Capital letters indicate bases which are conserved in more than 80% of the 58 promoters analyzed. Residues −36 (G), −33 (T), −15 (G), and −12 (A) seem to be particularly important since they are absolutely conserved in all 58 promoters. All of these residues exception the T at position −33 have previously been recognized as being critical for ς^B promoter activity (27, 30). Less than 10% of all known ς^B-dependent promoters deviate in more than two residues from this consensus sequence.

For most of the proteins encoded by the new ς^B-dependent genes, biochemical functions have not been determined although similarities to sequences found in databases suggested putative functions for some of them. aldY and yhdF encode gene products with similarities to aldehyde and glucose dehydrogenases, respectively. Among the gene products which are induced by ethanol stress in the sigB mutant are two more dehydrogenases: LctE, an l-lactate dehydrogenase, and YcnH, similar to succinate-semialdehyde dehydrogenases. These data appear to confirm our earlier suggestion that several general stress proteins with similarities to NAD- or NAD(P)-dependent dehydrogenases might be involved in the maintenance of the redox balance during stress (17). YdbP, which needed ς^B for induction, is similar to thioredoxins of Archaeoglobus fulgidus and Saccharomyces cerevisiae (44% identity in a 85-amino-acid overlap and 42% identity in a 75-amino-acid overlap, respectively) and could therefore also be involved in the protection against oxidative stress.

YqhA is highly similar to modulators of sigma factor activity such as RsbR and IspU of B. subtilis (1, 26). RsbR functions as a positive regulator of ς^B activity by interaction with the antagonist protein RsbS (1). YqhA and RsbR share a high degree of identity in their C-terminal portion, which also comprises a conserved phosphorylation site (1). Consequently, it would be attractive to analyze whether YqhA belongs to the family of antagonist proteins in the partner-switching regulatory mechanism and might be involved in conveying environmental signals in the ς^B signal transduction network.

Gaidenko and Price pointed out that ς^B-dependent stress proteins may be involved in the maintenance of the cell envelope integrity during stress (15). Supporting their hypothesis, many new ς^B-dependent genes described in this study seem to code for integral membrane proteins. The high proportion of membrane proteins found can partially be attributed to the fact that two-dimensional protein electrophoresis as one of the two approaches of identification of stress genes failed to detect this class of proteins due to their alkaline isoelectric point and solubilization problems (5). Induction of all of those genes by the stress sigma factor ς^B provides only a first hint for their involvement in stress protection. Knockout mutations in the genes need to be analyzed for resistance to oxidative, acid, alkaline, heat or salt stress in order to gather more information on their function.

Another interesting problem raised in this study is the question of why genes with presumable ς^B promoters, which are conserved in all the positions known to be critical for promoter recognition by ς^B (27, 30), still lacked stress induction (Fig. 1D). A potential reason for such a failure of induction can be the presence of operator elements blocked by repressor molecules. Recently, we obtained evidence that CtsR, a global repressor of class III heat stress genes (12, 23), prevents the ς^B-dependent induction of the clpC operon by glucose starvation (23). A detailed transcriptional analysis is needed to determine whether the genes in question are controlled by ς^B only under special circumstances or not at all.

Although the number of ς^B-dependent genes is now more than 80, the use of sophisticated approaches such as DNA chip technology and detailed analysis of genes subject to complex transcriptional regulation by several networks will help to define still undiscovered members of this important regulon.