INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The recognition of promoters by RNA polymerase (RNAP) requires an associated  specificity factor (9, 13). In Bacillus subtilis, the principle  subunit, ^A, can be substituted by any of a number of alternate  factors (11). Alternate  factors control the expression of stress responses (^B), flagellar motility, chemotaxis, and cell wall-related functions (^D), and the gene expression cascade leading to formation of a dormant endospore (^H, ^E, ^F, ^G, and ^K). Alternate  subunits may interact with a small fraction of the available core enzyme to activate transcription of a subset of genes or, as appears to happen during sporulation, become the predominant specificity factor present at a given time. The mechanisms acting to control  factor activity are remarkably diverse and include the production and regulation of specific anti- factors, proteolytic processing of active  factors from inactive precursors, and regulated turnover.

The sequencing of the B. subtilis genome has led to the identification of seven putative  factors of the extracytoplasmic function (ECF) subfamily (18). ECF  factors are found in a wide range of bacteria and control the uptake or secretion of specific molecules or ions and responses to a variety of extracellular stress signals (20). For example, ECF  factors regulate ferric citrate uptake and heat shock responses in Escherichia coli (2, 7), alginate and exotoxin secretion in Pseudomonas aeruginosa (14, 21), antibiotic production in Streptomyces antibioticus (17), hairpin protein secretion in Erwinia amylovora (25), and carotenoid biogenesis in Myxococcus xanthus (8). The roles of the seven B. subtilis ECF  factors, and the extent to which their functions may overlap, are largely unknown.

We have previously reported studies of one of the B. subtilis ECF  factors, ^X. Mutants lacking this protein have an increased sensitivity to elevated temperatures (15). We have determined the consensus sequence for recognition by holoenzyme containing ^X (E^X) and, using this information, were able to identify several target genes recognized by E^X in vivo and in vitro (16). In the course of this work, we noticed a similar promoter element preceding the gene encoding another putative ECF  factor designated ^W.

In this report, we demonstrate that the promoter element (P_W) preceding the sigW-ybbM operon is transcribed by E^W both in vivo and in vitro. In addition, E^W recognizes a subset of promoters that are partially dependent on ^X for expression, consistent with previous in vivo data suggesting that these regulons overlap (16).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. Bacterial strains and plasmids used in this study are described in Table 1. For routine DNA manipulations, E. coli Jm2r (4) was used as the host.

Growth conditions. E. coli was grown in 2×YT (22) liquid medium or on LB plates containing antibiotics as indicated. B. subtilis was grown in liquid culture in either 4×SG medium, Difco sporulation medium (DSM [12]), LB, or morpholinepropanesulfonic acid (MOPS)-buffered minimal medium (6). Since ^W-dependent promoter activity is maximal in rich medium, we routinely used 4×SG medium (modified from 2×SG [12]), which contains 32 g of Difco nutrient broth, 2 g of KCl, and 2.4 g of MgSO₄ · 7H₂O per liter (adjusted to pH 7.0). After autoclaving, sterile solutions were used to add 2 ml of 1 M Ca(NO₃)₂, 2 ml of 0.1 M MnCl₂, 2 ml of 1 mM FeSO₄, and 4 ml of 50% (wt/vol) glucose per liter. Cultures were grown at 37°C with vigorous shaking. For growth on plates, DSM containing 2% (wt/vol) glucose was used.

Construction of sigW and sigY mutants. Primers 143 (5'-GGGGTACCATGGAAATGATGATTAAAAAA-3') and 144 (5'-CGGGATCCTTAAAGATCCCTTAATTG-3') were used to amplify sigW from chromosomal DNA. The PCR product was digested with KpnI and BamHI (underlined sites) for cloning into pBKSII⁺ (Stratagene) to generate plasmid pKF84. pKF84 was digested with HindIII, and the region between codons 43 and 125 was replaced with a 1.6-kb erm gene (MLS^r) HindIII cassette isolated from plasmid pDG646 (10) to generate pKF88. In this plasmid, erm and sigW' are oriented in the same direction. To construct a sigW::erm mutant strain, pKF88 was used to transform B. subtilis to MLS^r to generate HB4246. The sigY mutant was constructed in a similar manner, using primers 141 (5'-GGGGTACCATGGATACACAAGAAGAACAG-3') and 142 (5'-CGGGATCCTTATTCATCATCCCACTCCT-3'). The amplified product was digested with KpnI and BamHI (sites underlined) and cloned into pBKSII⁺ to generate plasmid pKF83, and then the erm gene was inserted as a HindIII fragment to replace sigY codons 39 through 63 to generate pKF87. Transformation of JH642 with pKF87 generated the sigY mutant strain, HB4245.

Construction of P_W-cat-lacZ fusion. The region containing the putative ^W-dependent promoter element was amplified by PCR with chromosomal DNA as the template. The forward primer, 180 (5'-ACGAATAAGCTTCTACACCCTGCCAAA-3'), and reverse primer, 181 (5'-AATGGATCCTGGTCGCCTTTTTTGA-3'), amplify a 161-bp region of B. subtilis DNA. The PCR product was digested with BamHI and HindIII (underlined sites) and cloned into pJPM122 (23) to generate a cat-lacZ operon fusion in plasmid pXH23. This plasmid was linearized with ScaI and integrated into strain ZB307A (26) to generate strain HB7063. For transduction into wild-type and mutant backgrounds (Table 1), SP specialized transducing phage carrying the P_W-cat-lacZ operon fusion (SP7063) were recovered by heat induction and used to transduce recipient strains to neomycin resistance. Experiments were conducted in both CU1065 and JH642 wild-type backgrounds, and no differences were noted.

Overproduction and purification of ^W. The sigW PCR amplification product (as described above) was digested with NcoI and BamHI and cloned into pET16b (Novagen) to generate pKF86. The sequence of the cloned sigW gene was confirmed by DNA sequencing of pKF86. Transformation of pKF86 into E. coli BL21/DE3 (24) generates strain HE7123. For purification of ^W, HE7123 was grown to mid-logarithmic phase at 37°C in 1 liter of 2×YT medium supplemented with 0.4% (wt/vol) glucose and 100 µg of ampicillin per ml. Isopropyl--D-thiogalactopyranoside (IPTG) was added to 0.4 mM, and cells were harvested after further incubation for 3 h. After centrifugation, the cell paste was suspended in 20 ml of disruption buffer (50 mM Tris-HCl [pH 8.0], 2 mM EDTA, 0.1 mM dithiothreitol [DTT], 1 mM -mercaptoethanol, 233 mM NaCl, 10% glycerol) and lysed by sonication, and the inclusion bodies were recovered by centrifugation. The inclusion bodies were washed twice with 100 ml of TEDG buffer (10 mM Tris-HCl [pH 8.0], 0.1 mM EDTA, 1 mM DTT, 5% glycerol) containing 0.5% (vol/vol) Triton X-100 and 10 mM EDTA and then dissolved in 20 ml of TEDGX (TEDG containing 0.01% Triton X-100) containing 0.4% (wt/vol) Sarkosyl. The dissolved ^W was gradually diluted to 200 ml with TEDGX to allow renaturation of ^W, and the diluted ^W was dialyzed twice for 8 h against 10 vol of TEDGX at 4°C. The dialyzed material was centrifuged to remove any precipitate and loaded onto a 20-ml heparin-Sepharose CL-6B column equilibrated with TEDGX. After being washed with 200 ml of TEDGX and 0.2 M NaCl, ^W was eluted with TEDGX and 0.5 M NaCl. The peak fraction of ^W was further purified by chromatography on an FPLC (fast protein liquid chromatography) Superdex-75 column in TEDGX plus 0.2 M NaCl buffer, and the peak fractions were stored frozen at 80°C.

Runoff transcription assays. Runoff transcription assays were performed with PCR-amplified promoter fragments from pJPM122 subclones that had been tested for in vivo expression activity and verified by DNA sequencing. Reactions (25 µl) were performed and analyzed as described previously for ^X-dependent transcription (15) except that the buffer contained 110 mM NaCl and lacked added potassium glutamate. Each reaction mixture contained 0.5 pmol of DNA template, 2.5 pmol of B. subtilis RNAP, and, where indicated, 40 pmol of the designated  factor. Incubation was for 15 min at 37°C, and samples were analyzed by electrophoresis on an 8 M urea-6% polyacrylamide gels as described previously (15).

Primer extension assays. RNA was prepared from late-logarithmic-phase cells (approximately T₁) as described elsewhere (16). One hundred micrograms of total RNA was precipitated with 2 pmol of end-labeled reverse primer 181. The pellet was resuspended in 40 µl of hybridization buffer (60 mM NaCl, 50 mM Tris-HCl [pH 8.0], 10 mM DTT), heated to 95°C for 3 min, and cooled slowly to 40°C over 30 min. Then 60 µl of extension solution (60 mM NaCl, 50 mM Tris-HCl [pH 8.0], 13 mM DTT, 10 mM MgCl₂, 1 mM deoxynucleoside triphosphates, 20 U of avian myeloblastosis virus reverse transcriptase) was added, and the mixture was incubated at 37°C for 30 min. Nucleic acids were extracted with phenol-CHCl₃ (1:1, vol/vol), precipitated with ethanol, resuspended in 10 µl of double-distilled H₂O with 0.1 µg of DNase-free RNase per µl, and incubated at room temperature for 30 min; 10 µl of sequencing gel loading buffer was added, the reaction mixture was heated at 95°C for 3 min, and the nucleic acids were separated by electrophoresis on an 8 M urea-6% polyacrylamide gel and visualized by autoradiography. Double-stranded pXH23 DNA was sequenced by using phosphorylated primer 181, and the reaction products were electrophoresed adjacent to the primer extension products.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of P_W upstream of the sigW gene. In a previous study, we searched the B. subtilis genome for sequences resembling the sigX autoregulatory site, P_X (16). In the course of this analysis, we noticed an ECF-type promoter element upstream of the sigW gene (Fig. 1). Many ECF  factors recognize promoters with similar consensus elements, particularly in the 35 element, and autoregulation is quite common (20). We therefore hypothesized that this sequence might be a ^W-dependent promoter, P_W.

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Diagram of the sigW-ybbM region (16.6°; 195 kb) of the B. subtilis chromosome. The sequence of the region between trnL and sigW is shown. The inverted repeat predicted to terminate transcription of the trnL operon is illustrated. The bracket and arrow indicate the upstream boundary of the PW-containing fragment used in construction of the PW-cat-lacZ reporter. The 35 and 10 regions of PW, and the two observed transcription start points, are shown in boldface.

View larger version (21K): [in this window] [in a new window]   FIG. 2.   Expression of PW-cat-lacZ. Expression of -galactosidase (Miller Units) is plotted as a function of growth phase for cells grown in 4×SG. Time zero is the end of logarithmic-phase growth. The strains used contain SP7063 (PW-cat-lacZ) reporter phage in either wild-type (), sigW::erm (), sigX::spc (), or rsiX::pVA29 () mutant background.

View larger version (79K): [in this window] [in a new window]   FIG. 3.   Primer extension analysis of PW activity. RNA samples were prepared from late-logarithmic-phase cells of the wild-type strain CU1065 grown at 37°C (wt) or 15 min after a shift to 50°C [wt(hs)] and analyzed by reverse transcription. Samples were also prepared from the sigX::spc (X) and sigW::erm (W) mutants. The sequencing ladder on the left was generated by using the same reverse primer (primer 181) and double-stranded pXH23 DNA. The two start sites are A and G residues as shown in Fig. 1.

Expression of P_W in a sigX mutant. We noticed that P_W-cat-lacZ expression is derepressed in the sigX mutant strain as observed in plate assays, in the primer extension experiment (Fig. 3), and during growth in liquid culture (Fig. 2). This derepression occurs during late logarithmic phase, when ^X activity is normally the highest (15, 16). This finding suggests that ^X antagonizes the activity of ^W. However, since sigX is cotranscribed with a negative regulatory gene, rsiX (3, 15), it is also possible that the antagonism is due to RsiX and the observed derepression is due to polarity of the sigX::spc insertion. Since RsiX appears to function as an anti- factor (3), it could inhibit ^W activity directly. To test this idea, we compared expression of P_W-cat-lacZ in the wild type and an rsiX mutant (Fig. 2). In the rsiX mutant, expression of P_W-cat-lacZ no longer increases upon entry into stationary phase, suggesting that the increased activity of ^X in the absence of RsiX further antagonizes ^W. The opposite result would have been expected if RsiX was able to inhibit ^W directly.

Reconstitution and activity of ^W holoenzyme. We overproduced and purified ^W by using a T7 RNAP-based overexpression system. As found previously for several other B. subtilis  factors, ^W accumulates in inclusion bodies upon overproduction. Since ^W precipitated during renaturation from inclusion bodies dissolved in 6 M guanidine hydrochloride, we used Sarkosyl as a denaturant as originally described for E. coli ³² (5). ^W was about 90% pure after heparin-Sepharose chromatography and >95% pure after Superdex-75 chromatography (Fig. 4). ^W elutes during gel exclusion chromatography with an apparent molecular mass of 31.0 kDa, which is somewhat greater than the calculated monomer mass of 21.6 kDa. This is typical of  factors and has been interpreted as an indication of an asymmetric shape.

View larger version (75K): [in this window] [in a new window]   FIG. 4.   Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of fractions from the W purification. Lane M, molecular weight markers (sizes in kilodaltons are indicated on the left); lane 1, 100 µg of whole-cell lysate of HE7123 prior to induction with IPTG; lane 2, 100 µg of whole-cell lysate after 3 h of induction with IPTG; lane 3, 10 µg of peak fractions after heparin-Sepharose CL-6B chromatography; lane 4, 10 µg of peak fractions after FPLC Superdex-75 chromatography.

View larger version (80K): [in this window] [in a new window]   FIG. 5.   In vitro runoff transcription of PW and PX. PCR products containing either PW or PX were used as templates for runoff transcription reactions with core RNAP either alone () or after supplementation with either X (X) or W (W). Note that the core RNAP preparation has a low level of activity in the absence of added , suggesting that this preparation has some contaminating X as noted previously (15).

Overlapping promoter selectivity of ^W and ^X. In a previous study, we noted that a subset of putative ^X-dependent promoters were active in a sigX mutant but were, in some cases, inactive in a sigX sigW double mutant (16). We therefore suggested that these  factors might have overlapping promoter selectivity. To further define the extent to which the ^X and ^W regulons might overlap, we used in vitro runoff transcription assays to test the ability of E^W to recognize promoters identified originally as part of the E^X regulon. In these experiments, for the csbB and lytR promoters, E^W failed to stimulate transcription above the level detected in reactions with core RNAP alone. In contrast, both E^X and E^W were quite active with abh, divIC, ywbN, and yrhH templates (Fig. 6). Thus, both holoenzymes can recognize these promoter elements in vitro.

View larger version (31K): [in this window] [in a new window]   FIG. 6.   In vitro runoff transcription of EX and EW on six X-dependent promoters. PCR products containing the indicated promoter regions were amplified from the corresponding pJPM122 derivatives (16) and used as templates for runoff transcription reactions with core RNAP either alone () or after supplementation with either X (X) or W (W). Promoter sequences are shown in Table 2 together with the deduced transcription start sites. All runoff transcripts are between 110 and 180 nucleotides in length, allowing start sites to be assigned with an accuracy of ±2 nucleotides. The start sites for the in vitro EX transcripts have been previously determined at nucleotide resolution by primer extension mapping (16). Some of the transcript heterogeneity (most notably in the lytR reaction) is apparently due to different 3' ends resulting from variability in the PCR products since a single well-defined start point is observed in primer extension mapping experiments (data not shown and reference 16).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The B. subtilis genome encodes multiple ECF  factors of largely unknown function (18). As one approach to defining the physiological role of these proteins, we have sought to identify sets of target genes dependent on one or more ECF  factors for expression. Previously, we identified an autoregulatory ^X-dependent promoter and used saturation mutagenesis to define the bases relevant for ^X recognition in vivo. We then used the resulting consensus sequence to identify a set of eight promoters that were either partially or completely dependent on ^X for expression (16). The genes of the ^X regulon include lytR, a regulator of autolysins (19), and csbB, encoding a putative membrane-bound glucosyltransferase that is also part of the ^B regulon (1). Thus, ^X is postulated to control genes expressed in late logarithmic phase that act to modify the cell surface.

To define the physiological role of ^W, it will be necessary to identify both the target operons and the regulatory signals leading to their expression. As a first step toward defining ^W promoter selectivity, we have characterized an autoregulatory promoter (P_W) preceding the sigW-ybbM operon. P_W is the prototype for a large family of highly conserved ^W-dependent promoters: at least 12 additional promoters, similar in both the 35 and 10 elements to P_W, control the expression of more than two dozen, mostly membrane-localized proteins (our unpublished data). Our ongoing identification of target operons will facilitate physiological studies of the sigW mutant strain which is apparently unaffected in sporulation, competence, and most other post-exponential-phase functions.

P_W is completely dependent on ^W in vivo (Fig. 2) and is actively transcribed by E^W in vitro (Fig. 5). By primer extension analysis, we do not detect any additional promoters in the trnL-sigW intergenic region, suggesting that this promoter is primarily, if not exclusively, responsible for sigW transcription. The previously described sigX autoregulatory promoter is also transcribed with high specificity: P_X is dependent on ^X in vivo and is recognized in vitro by E^X but not by E^W (Fig. 5). Whereas expression of sigX (15) and ^X-controlled genes (16) is maximal in late-logarithmic-phase cells (T₂ to T₀), expression of sigW is highest in early stationary phase (approximately T₁ to T₃). Thus, the ^X and ^W regulons appear to be quite distinct both in the time of their optimal expression and in their constituent operons.

In contrast with the high selectivity of autoregulation, both lacZ reporter and start site mapping data indicate that some ^X-dependent promoters are still active in a sigX mutant and are therefore recognized by at least one additional  factor in vivo (16). To better define the sequence features that distinguish ^X- from ^W-dependent promoters, we have tested several ^X-dependent promoters for the ability to be recognized in vitro by E^W. Reconstituted E^W fails to stimulate transcription from the sigX, csbB, or lytR promoter in vitro (Fig. 5 and 6), consistent with the in vivo observation that transcripts from these sites are greatly reduced or eliminated in a sigX mutant strain (16). Thus, these promoters seem to be ^X specific. In contrast, both E^X and E^W recognize promoters for abh, divIC, yrhH, ywbN (Fig. 6), and rapD (data not shown). Therefore, these promoters have properties that allow them to be recognized by either holoenzyme, at least in vitro.

Overlapping holoenzyme selectivity is consistent with previous in vivo primer extension data demonstrating the persistence of transcripts from these sites even in a sigX mutant (16). Most surprising, ywbN and yrhH transcripts were not detected in wild-type cells, were easily detected in sigX mutant cells, but were not detected in sigX sigW double-mutant cells. It now seems likely that transcripts were detected in the sigX mutant because of increased ^W activity during late logarithmic phase, as noted above for P_W (Fig. 2). Thus, ywbN and yrhH are members of the ^W regulon. As noted previously, however, analyses of lacZ reporter fusions suggest that these genes are also part of the ^X regulon (16). Transcription from the ^X-like promoter upstream of divIC is reduced less than twofold in a sigX mutant and is still detectable even in a sigX sigW double-mutant strain. We suspect that there is at least one additional holoenzyme that can recognize this site. Thus, while divIC is an active template for both E^X and E^W in vitro (Fig. 6), it is difficult to assess the relative contribution of these  factors to in vivo expression until the remaining holoenzyme forms allowing initiation at this site are identified. In summary, our in vitro transcription results (Fig. 6), together with previous in vivo data (16), suggest that the ^X and ^W regulons overlap, with the relative contributions of these two  factors depending on both growth phase and nutritional factors.

Expression of P_X, and some ^X-dependent genes, is elevated in a sigW mutant (16). Similarly, we now find that expression of sigW is derepressed in late logarithmic phase in a sigX mutant and repressed in an rsiX mutant (Fig. 2). The basis of this mutually antagonistic behavior is not clear. It is possible that when the ^X holoenzyme is active, it binds and recognizes P_W but fails to initiate transcription efficiently, thereby preventing autoactivation of this site during late logarithmic phase. Then, when ^W become active in early stationary phase, it may bind and competitively inhibit expression from those promoter sites exclusively dependent on ^X (e.g., preceding csbB and lytR) while extending expression from those sites able to be recognized by either holoenzyme. In fact, the csbB and lytR promoters, which apparently cannot be recognized by ^W, are the most dramatically induced in a sigW mutant as assessed on X-Gal plates (16). Alternatively, the ^X and ^W regulons may be partially redundant in function and respond to an overlapping set of signals. By this model, expression of one regulon may decrease the need to express the other.

A comparison of promoter sites recognized by either ^X or ^W, or both, suggests that two positions in the 10 region may be key to understanding holoenzyme selectivity (Table 2). Mutational analysis of the sigX promoter identified a 10 consensus of CGwC (where w represents A or T). In contrast, promoters exclusively dependent on ^W (P_W and other characterized ^W-dependent promoters [our unpublished data]) contain a 10 region of CGTA. The ability of ^W to transcribe some ^X-dependent promoters (Table 2) demonstrates that this holoenzyme can recognize CGTC in addition to CGTA. Thus, the sequence specificity for ^W seems to be CGT(A or C), although this has not been directly tested by mutagenesis. This suggests a simple model for the partially overlapping recognition observed both in vitro (Fig. 6) and in vivo (16): those promoters that have a 10 element of CGAC are dependent on ^X, those containing CGTC can, at least in principle, be recognized by either ^X or ^W, and those containing CGTA are exclusively dependent on ^W (Table 2).

View this table: [in this window] [in a new window]   TABLE 2.   Comparison of promoter sequences recognized in vitro by EX, EX and EW, or EW alone

It is likely that other features of the promoter, such as preferred spacer length and the 35 region, may also contribute to selectivity. However, deletion of one base from the P_X spacer region reduced but did not eliminate activity, and two ^X-dependent promoters, csbB and lytR, have a shorter spacer region (16). E^W also tolerates at least a 1-bp variation in spacer length. Both P_X and P_W contain a highly conserved AAC motif in the 35 region, similar to promoters recognized by diverse ECF  factors, and there is no apparent correlation between sequence in this region and holoenzyme selectivity.