Dual Promoters Are Responsible for Transcription Initiation of the fla/che Operon in Bacillus subtilis

Abstract

The fla/che region contains more than 30 genes required for flagellar synthesis and chemotaxis in Bacillus subtilis, including the gene for the flagellum-specific ς^D factor, sigD. Sequence and primer extension data demonstrate that a P_A promoter immediately upstream of flgB, henceforth referred to as the fla/che P_A, and the P_D-3 promoter are active in vivo. Transcription from the P_D-3 element is dependent on ς^D activity and is regulated by the flagellum-specific negative regulator, FlgM. In a strain containing a deletion of fla/che P_A (P_AΔ), ς^D protein was not detected, demonstrating that the fla/che P_A is necessary for wild-type expression of the sigD gene. Thus, sigD is part of the >26-kb fla/che operon. Consistent with a lack of detectable ς^D protein, the P_AΔ strain grows as long filaments and does not express a ς^D-dependent hag::lacZ reporter construct. These phenotypes are indicative of a lack of sigD expression or complete inhibition of ς^D activity by FlgM. However, ς^D activity is found in a double mutant containing the P_AΔ and a null mutation in flgM. The double mutant no longer grows as long filaments, and expression of hag::lacZ is partially restored. These data demonstrate that a low level of ς^D activity does exist in the P_AΔ mutant but can be detected only in the presence of a null mutation in flgM. Therefore, normal expression of sigD may also involve another promoter(s) within the fla/che operon.

Gene expression in Bacillus subtilis is principally regulated by alternate forms of RNA polymerase differing in composition by the association of alternate ς factors, which determine the promoter specificity of the resulting holoenzyme (25, 34, 44). The ς^D holoenzyme is directly responsible for transcription of the genes for flagellin (30), several hook-associated proteins (4), the MotA and MotB motor proteins (31), the CheV chemotaxis protein (7), the methyl-accepting chemotaxis proteins (12), and the FlgM negative regulator (33). The ς^D regulon in B. subtilis is therefore composed of genes encoding proteins for flagellar synthesis, motility, and chemotaxis.

The structural gene for the ς^D factor, sigD, is located at the distal end of the fla/che region of DNA (29). The fla/che region of DNA was originally isolated as two overlapping lambda clones, containing B. subtilis genomic DNA, that complemented chemotaxis mutations mapping between pyrD and thyA on the bacterial chromosome (39). The genes residing in this region of DNA have been sequenced, and many of the predicted protein products have been shown to be homologous to structural proteins that form the hook-basal body (HBB), as well as several chemotaxis proteins in the enteric bacteria Escherichia coli and Salmonella typhimurium (38). The homologous genes in these bacteria, however, are found in 13 different operons (21, 22).

In the enteric bacteria, a hierarchy of transcription of three classes of genes (21, 26) ensures that the expression of the gene encoding the major flagellar protein flagellin (hag) is tightly regulated and is dependent on the functional assembly of the HBB complex (14, 20). The master regulators FlhD and -C are class I gene products and are transcriptional activators that are required for expression of class II genes (24). Class II genes encode the structural proteins that form the HBB complex (36) and include the fliA locus, which encodes the alternate sigma factor ς^F, a homolog of ς^D (36). Class III genes possess ς^F-dependent promoters and are transcribed by this form of RNA polymerase (27). Furthermore, class III gene expression is regulated by FlgM and is dependent on the expression of all class II genes (15). FlgM has been described as an anti-sigma factor that inhibits ς^F binding to RNA polymerase (37). Once the HBB complex is functionally assembled, the FlgM regulator is specifically exported and ς^F is able to associate with RNA polymerase (14, 20) and initiate transcription of the class III genes, including hag.

A similar hierarchy of flagellar gene expression seems to exist in B. subtilis, although homologs to the class I master regulators have not been identified. The fla/che operon appears to be primarily a class II transcription unit. Mutations within the fla/che operon that disrupt HBB open reading frames eliminate flagellin synthesis (29, 46). This morphogenetic repression of class III gene expression is controlled by the FlgM homolog in B. subtilis, since the lack of this factor restores expression of the hag gene in strains lacking an HBB complex (32). Therefore, hag gene expression in B. subtilis is dependent upon the expression of the HBB genes and on ς^D RNA polymerase as it is in the enteric bacteria.

Interestingly, the genes encoding the HBB complex appear to be part of a single operon in B. subtilis. Sequence analyses of the region demonstrate the existence of consecutive open reading frames that appear to be translationally coupled (1, 38). Additionally, genetic analyses of this region suggest that it is transcribed from a single upstream promoter region, since insertions of heterologous DNA within fla/che result in decreased expression of downstream genes (29, 45, 46). Although much of the DNA that makes up the fla/che region has been cloned, sequenced, and characterized genetically, the transcription initiation signals have not been identified. Therefore, to better understand the molecular mechanisms that govern flagellar gene expression in B. subtilis, this study has centered on the characterization of the promoter region for the fla/che transcription unit. This work provides a foundation for studying the molecular machinery that governs expression of a very large transcription unit encoding flagellar and chemotaxis functions in B. subtilis which, in enteric bacteria, are encoded in more than 13 operons (21, 22).

MATERIALS AND METHODS

Bacterial strains, media, growth, and transformation.

The E. coli host for growth of recombinant plasmid was the strain Epicurian Coli XL1-Blue {recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F′ proAB lacI^qZΔM15 Tn10 (Tet^r)]} (Stratagene). For site-directed oligonucleotide mutagenesis, E. coli CJ236 [F⁻ dut-1 ung-1 thi-1 relA1/pCJ105 (Cm^r)] (Bio-Rad Laboratories), E. coli Epicurian Coli XLmutS {Δ(mcrA)183 Δ(mcrCB-hsdSMR-mrr)173 endA1 supE44 thi-1 gyrA96 relA1 lad mutS::Tn10 (Tet^r)[F′ proAB lacI^qZΔM15 Tn5 (Kan^r)]} (Stratagene), and the M13 helper phage R408 (Promega) were used. Transformation of competent E. coli cells was performed as described in the suppliers’ protocols. Electrocompetent E. coli CJ236 (Bio-Rad Laboratories) cells were transformed by electroporation with a Gene-Pulser. E. coli cells containing plasmids were selected by growth in LB (Luria-Bertani) broth supplemented with 50 μg of ampicillin (Sigma) per ml. In cases where a blue-white screen was used to identify transformants containing the desired recombinant plasmids, transformants were plated on LB agar with 32 μg of X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; Sigma) per ml, 32 μg of IPTG (isopropyl-β-d-thiogalactopyranoside; Sigma) per ml, and the appropriate antibiotic.

The B. subtilis strains used in this study are listed in Table 1. Strains were grown in 2× SG medium [2× nutrient broth, 1 mM Ca(NO₃)₂, 0.1 mM MnSO₄, 0.1% glucose] and tryptose blood agar base (TBAB) and LB plates. Transformation of plasmid or chromosomal DNA into B. subtilis was accomplished as previously described (5). Chloramphenicol-resistant (Cm^r) strains were maintained by adding 5 μg of chloramphenicol (Sigma) per ml to the appropriate medium, whereas spectinomycin-resistant (Sp^r) strains were maintained by adding 100 μg of spectinomycin (Sigma) per ml.

TABLE 1.

B. subtilis strains used in this study

Strain	Genotype	Source or derivation and referencea
LMB1	trpC2	E. Ferrari, I168b
LMB10	trpC2 sigD::pLM5 (Cmr)	M. J. Chamberlin, CB100b; 13
LMB24	trpC2 phe hag::lacZ (Cmr)	M. J. Chamberlin, CB25b; 32
LMB213	trpC2 flgM::mini-Tn10 (Spr)	Transform [LMB1:HB4229, Spr]; reference 6 and this study
LMB214	trpC2 fla/che PAΔ	Transform [LMB1:pWE4-int, Cmr]c; this study
LMB216	trpC2 fla/che PAΔ flgM::mini-Tn10 (Spr)	Transform [LMB214:LMB213, Spr]; this study
LMB219	trpC2 hag::lacZ (Cmr)	Transform [LMB1:LMB24, Cmr]; this study
LMB220	trpC2 fla/che PAΔ hag::lacZ (Cmr)	Transform [LMB214:LMB24, Cmr]; this study
LMB221	trpC2 flgM::mini-Tn10 (Spr) hag::lacZ (Cmr)	Transform [LMB219:LMB213, Spr]; this study
LMB222	trpC2 flgM::mini-Tn10 (Spr) fla/che PAΔ hag::lacZ (Cmr)	Transform [LMB220:LMB213, Spr]; this study

^aTransformation of [first (recipient) strain: with chromosomal DNA from second strain, and selecting for resistance indicated] is shown for some strains. 

^bPrevious designation of strain. 

^cThis strain was subsequently cured of the chloramphenicol resistance marker. 

Manipulations of DNA.

Plasmid DNA propagated in E. coli was purified by a modification of the alkaline lysis method (41), which included isopropanol and lithium chloride precipitations. B. subtilis chromosomal DNA was isolated as described previously (28). The recovered plasmid and chromosomal DNA was resuspended in TE (10 mM Tris-Cl [pH 8.0], 1 mM EDTA), and the yield and concentration were quantified by use of a spectrophotometer (UV-1201; Shimadzu).

Restriction digests for preparatory or diagnostic purposes were performed by standard procedures (41). Agarose gels were photographed with a GDS 7500 gel documentation system (UVP) or a Polaroid camera and Polaroid 667 black-and-white film. DNA fragments of interest were gel purified with the Prep-A-Gene DNA purification system (Bio-Rad Laboratories) under the manufacturer’s recommended conditions. The 5′ phosphates were removed from the digested vector DNA by using shrimp alkaline phosphatase (United States Biochemical), and the subsequent ligation was accomplished by using the Takara ligation kit (Panvera) as recommended by the manufacturer.

DNA sequencing and analysis.

Sequencing reactions were carried out with Sequenase version 2.0 (United States Biochemical), by using the following primers: PD3 (5′-CACCCTCAATATCCTTGTCG-3′), OWE1 (5′-CATAGAAAGACTTTCAA-CCCAGG-3′), and OWE2 (5′-GAGGGTTCTTTTTTTATTTC-3′). Primers were synthesized by Cruachem, Inc. (Dulles, Va.). The nucleotide sequences obtained were compiled and analyzed by using MacVector (Kodak).

Construction of plasmids.

The 2.6-kb PstI fragment from pAZ210 (5), which contains the fla/che promoter region, was subcloned into pGEM5Zf(+) (Promega) to yield the plasmid pWE1. pWE4, a derivative of pWE1, contains a 29-bp deletion of the fla/che P_A promoter (P_AΔ) created by oligonucleotide-directed mutagenesis (see below). The 2.6-kb PstI fragment from pWE4 containing the P_AΔ was subcloned into the B. subtilis integrational vector pJM102 (41) to generate the plasmid pWE4-int.

RNA extraction and primer extension analysis.

To determine if promoter consensus sequences identified by computer analyses are functional in vivo, total RNA was purified from the appropriate strains and subjected to primer extension analyses. B. subtilis LMB1 and LMB10 were grown to late log phase in 2× SG medium. The cells were harvested at two time points, T₀, marking the point when the B. subtilis culture breaks from logarithmic-phase growth, and T_0.5, marking 30 min after T₀. Total RNA was extracted by using the RNeasy Maxi RNA isolation kit (Qiagen) as recommended by the manufacturer except that cells were disrupted by sonication. Total RNA was precipitated and resuspended in double-distilled water. The integrity of the total RNA was verified by resolution on a formaldehyde-agarose gel under standard conditions (41).

Primer extension reactions were performed as described previously (33). Oligonucleotides PD3 and OWE3 (5′-AATATCCGCTCTGCTCAAGGCAT-3′; Cruachem, Inc.) were end labeled with ³²P to high specific activity (∼10⁸ cpm/μg). Primer extension reactions were resolved by electrophoresis alongside sequencing reactions of pWE1 which were generated by using the same oligonucleotide used in the primer extension reactions. Autoradiography was carried out at room temperature without an intensifying screen for 5 to 22 days. Exposed X-ray film was scanned with an Epson 2.01 scanner into Adobe Photoshop 3.0 (Adobe Systems, Inc.).

Deletion of the fla/che P_A.

To determine the importance of the fla/che P_A promoter in initiating transcription of sigD, the promoter sequence was deleted from the B. subtilis chromosome by using oligonucleotide-directed mutagenesis as described by others (19). Oligonucleotide OWE4 (5′-GGACATTTTTTTACACGAACTTCAGAATTCAAGCATATAGTTTTACAATTC-3′) lacks the 29 bp comprising the −10 and −35 P_A consensus sequences as well as the 17-bp spacer region for the fla/che P_A (see Fig. 1) and in their place bears an EcoRI restriction site (underlined in the sequence above). In the in vitro synthesis reaction, OWE4 was extended with T4 DNA polymerase (Promega), and the newly synthesized strand was joined with T4 DNA ligase (Promega). The plasmid containing the P_AΔ was identified by restriction map analysis and named pWE4. This plasmid was sequenced by use of an ABI Prism 377 automatic sequencer with the PD3, OWE1, and OWE2 primers; besides the P_AΔ, no new mutation was introduced as a result of the mutagenesis procedure.

FIG. 1.

fla/che promoter region. The intergenic region between codY and flgB has been defined as the fla/che promoter region. Two promoter elements were found to function in vivo: a ς^A-dependent promoter (fla/che P_A, present study) and P_D-3, a previously identified ς^D-dependent promoter (11). The −10 and −35 sequences for both of these promoters are underlined, and the +1 transcription initiation sites appear in boldface type. The 29 bp that were deleted in the fla/che P_AΔ strain and replaced with the EcoRI restriction site (GAATTC) are indicated by asterisks.

A fragment bearing the P_AΔ was then subcloned into the B. subtilis integrational vector pJM102 (40) under standard conditions. The resultant plasmid, pWE4-int, was transformed into a wild-type B. subtilis strain, LMB1. Transformant colonies were selected by growth on LB plates containing 5 μg of chloramphenicol per ml. Transformants generated from a double-crossover event, in which two deleted copies of the fla/che P_A replaced the wild-type sequence, were identified by single-colony PCR as described below. Loss of the plasmid sequences was accomplished by repeatedly growing the cells in PA (Penassay or antibiotic medium 3; Difco) broth without chloramphenicol and screening for Cm^s colonies by replica plating. This regimen allows for homologous recombination between directly repeated P_AΔ sequences, resulting in the loss of the Cm^r gene and thus antibiotic resistance. Curing of the inserted plasmid from the integrant strain was verified by single-colony PCR as described below.

Single-colony PCR.

To identify the B. subtilis cells harboring the double-crossover event and the P_AΔ, DNA from single colonies was subjected to PCR analysis. Cell pellets derived from saturated 2-ml cultures were lysed in 0.1 mg of lysozyme per ml at 37°C. Cell debris was collected by centrifugation in a microcentrifuge for 1 min, and the supernatant was collected for PCR amplification. One microliter of the supernatant collected was added to a 20-μl reaction mixture containing 1× Taq buffer, 2.5 mM MgCl₂, 0.18 mM deoxynucleoside triphosphates, 1 μM concentrations of oligonucleotide OWE7 (5′-GTGAGGACATTTTTTTACACG-3′) and OWE8 (5′-ACCCTCAATATCCTTGTCGAG-3′), and 0.3 U of Taq polymerase. PCRs were done in an MJR thermal cycler by using the following parameters: 1 min at 95°C, 1.5 min at 52°C, and 1 min at 72°C for 30 cycles. The PCR products were resolved on a 4% agarose gel and stained with 5 μg of ethidium bromide per ml. The gel was photographed with Polaroid 667 black-and-white film, and the photograph was scanned and the image was processed as described above.

Anti-ς^D immunoblot.

The level of ς^D protein was determined by immunoblot analysis. Strains to be analyzed were grown in 2× SG medium containing the appropriate antibiotic. Thirty minutes after the end of logarithmic growth (T_0.5), 20 ml of cell culture was collected by centrifugation and washed in ice-cold STE (150 mM NaCl, 10 mM Tris-Cl, 100 mM EDTA). Total protein was then extracted and quantified as described previously (28), and 50 μg of protein per sample was resolved by electrophoresis on a sodium dodecyl sulfate–12.5% polyacrylamide gel. The protein was electrophoretically transferred to nitrocellulose, and the filter was blocked, incubated in hybridization buffers, and washed as described previously except that the secondary goat anti-rabbit antibody was conjugated to alkaline phosphatase (29). Reactive proteins were visualized by using BCIP-NBT (5-bromo-4-chloro-3-indolylphosphate–nitroblue tetrazolium) substrate (Sigma) in distilled deionized water. The developed nitrocellulose filter was scanned, and the image was processed as described above.

Light microscopy.

Strains were grown in 2× SG medium, and samples were collected 30 min after the break from logarithmic growth (T_0.5). Cells were viewed by differential interference microscopy at ×1,250 with a Nikon Optiphot-2 scope with Nomarski optics and photographed with a Nikon Microflex UFX-DX attachment and Ektachrome black-and-white film for slides. Slides were subsequently scanned with a Polaroid Sprint Scan 35, and the image was processed as described above.

β-Galactosidase assay.

Flagellin gene expression was monitored by measuring the β-galactosidase activity produced in strains bearing a hag::lacZ translational fusion. Strains to be tested were grown in 2× SG sporulation medium containing 5 μg of chloramphenicol per ml when appropriate. Growth was monitored by spectrophotometer at 600 nm, and samples were collected at optical densities ranging from 0.2 to 3.0 and stored on ice. The β-galactosidase assay was conducted as described previously (8), except that β-galactosidase activity in the supernatant was measured at 420 nm. The calculation and manipulation of specific activity as Miller units, average values, and standard deviations were performed with Excel version 5.0 (Microsoft).

RESULTS

Analysis of the nucleotide sequence upstream of the fla/che region.

pAZ210 contains a 5.2-kb segment of B. subtilis DNA bearing flgB, the first gene of the fla/che region, and 1.8 kb of DNA upstream of this gene, including the region of DNA thought to be responsible for initiating transcription of the fla/che operon (46). Directed sequencing was used to obtain 620 nucleotides of sequence data upstream of flgB (Fig. 1). Computer analysis of the data generated two interesting findings.

First, the DNA sequence furthest upstream of the flgB translational start site was found to be identical to the 3′ terminus of the cod operon, an operon in B. subtilis that was recently characterized (42). The cod and fla/che operons are transcribed in the same direction and are separated by a 379-bp sequence. We have defined the intergenic region between codY (the last gene in the cod operon) and flgB (the first gene in the fla/che transcription unit) as the fla/che promoter region. It is possible, however, that one or more promoters within the cod operon, or read-through from this transcription unit, may contribute to fla/che expression.

Second, sequence analysis of this promoter region demonstrated that part of the fla/che promoter sequence is identical to a segment of B. subtilis DNA previously shown to contain a ς^D-dependent promoter (11). This particular ς^D-dependent promoter, originally and hereinafter called P_D-3, was shown to be a sequence of DNA specifically recognized by the ς^D holoenzyme in an in vitro transcription reaction (11). Whether this promoter was active in vivo, however, had not been determined; the nature of the gene(s) it potentially regulated was also unknown.

Primer extension analysis of fla/che promoter region.

Primer extension analyses reveal that transcription of the fla/che operon initiates from a P_A immediately upstream of flgB and from P_D-3. The P_A immediately upstream of flgB has therefore been renamed fla/che P_A. Expression from this promoter is found at T₀ and at T_0.5 in both the wild-type and sigD mutant strains (Fig. 2A, lanes 5 to 8). Primer extension products from P_D-3, however, are generated from RNA derived from the wild-type strain (Fig. 2B, lanes 5 and 6) but not from the sigD mutant (Fig. 2B, lanes 7 and 8). Therefore, transcription initiation from P_D-3 is dependent on the ς^D holoenzyme. Furthermore, there appears to be an increase in primer extension products produced from RNA isolated from a flgM mutant strain (Fig. 2B, lanes 9 and 10). This result demonstrates that P_D-3 is subject to regulation by the FlgM negative regulator; the absence of FlgM protein may allow for increased ς^D activity and thus increased transcription from P_D-3 in this strain.

FIG. 2.

Primer extension products from fla/che P_A and P_D-3. Total RNA from B. subtilis wild-type (LMB1), sigD null (LMB10), and flgM null (LMB213) strains that was isolated at the end of logarithmic growth (T₀) and 30 min later (T_0.5) was subjected to primer extension analysis. (A) Lanes 1 to 4, sequencing reactions of pWE1 (G, A, T, and C, respectively); lanes 5 to 8, primer extension products from total RNA, isolated by using primer OWE3, from B. subtilis LMB1 at T₀ (lane 5) and T_0.5 (lane 6) and from B. subtilis LMB10 at T₀ (lane 7) and T_0.5 (lane 8). (B) Lanes 1 to 4, sequencing reactions of pWE1 (G, A, T, and C, respectively); lanes 5 to 10, primer extension products from total RNA, isolated by using primer PD3, from B. subtilis LMB1 at T₀ (lane 5) and T_0.5 (lane 6), from B. subtilis LMB10 at T₀ (lane 7) and T_0.5 (lane 8), and from B. subtilis LMB213 at T₀ (lane 9) and T_0.5 (lane 10). The sequence given to the left of each panel corresponds to the nontemplate sequence, where +1 represents the nucleotide at which transcription initiates. Reactions were performed with 98 μg of LMB1 RNA isolated at T₀, 131 μg of LMB1 RNA at T_0.5, 153 μg of LMB10 RNA at T₀, 128 μg of LMB10 RNA at T_0.5, 84 μg of LMB213 RNA at T₀, and 92 μg of LMB213 RNA at T_0.5.

Construction of fla/che P_A deletion mutant.

After the fla/che P_A was shown to be active in vivo, it was deleted from the B. subtilis chromosome to determine its importance in promoting expression of sigD. The fla/che P_A was first deleted by oligonucleotide-directed site mutagenesis of the plasmid pWE1, resulting in plasmid pWE4. To introduce this deletion into the B. subtilis chromosome, the 2.6-kb PstI fragment containing the fla/che P_A deletion was subcloned into the integrational vector pJM102 (40), generating pWE4-int. pJM102 lacks a B. subtilis origin of replication and cannot be maintained as an episome. Concatemers of pWE4-int, however, can integrate into the fla/che promoter region by homologous recombination, by using either a single- or double-crossover mechanism (Fig. 3A, panels b and c). These integration events can be distinguished by the single-colony PCR method (see Materials and Methods). Only a strain bearing the double crossover can be used to obtain an exact deletion of the endogenous fla/che P_A without the introduction of plasmid sequences (43). This strain was grown under nonselective conditions to allow for homologous recombination of the duplicated regions and the concomitant loss of the intervening plasmid sequence (Fig. 3A, panel d). Transformants containing a double-crossover event were identified by using a PCR-based assay (Fig. 3B, lane 6), and strains bearing the exact deletion of fla/che P_A were verified by the same assay (Fig. 3B, lanes 7 and 8). Strains deleted for the endogenous fla/che P_A will henceforth be referred to as P_AΔ strains, and the deletion mutation will be referred to as P_AΔ.

FIG. 3.

PCR-based assay for the analysis of chromosome structure at the fla/che promoter region. (A) (a) The wild-type strain is sensitive to chloramphenicol, and its chromosome contains a single fla/che P_A, resulting in a 98-bp PCR product upon amplification. Introduction of pWE4-int gives rise to chloramphenicol-resistant transformants due to integration of plasmid sequences by homologous recombination. (b) DNA from a transformant generated by a single-crossover event (Campbell-like recombination) gives rise to PCR products of 98 and 75 bp. (c) DNA from a transformant generated by a double-crossover event with concatemers of pWE4-int (gene replacement) results in a single 75-bp product. (d) Once the strain bearing the double crossover is cured of chloramphenicol resistance, with a concomitant loss of the plasmid sequences, a single 75-bp fragment is produced. Half-arrows indicate primers used for PCR amplification. (B) Agarose gel of PCR products from wild-type (w.t.), transformant, and deleted strains. PCR-amplified products obtained from plasmid and chromosomal DNA with primers OWE7 and OWE8 were resolved by agarose electrophoresis. Lane 1, 10-bp ladder; lane 2, amplification of pWE1; lane 3, amplification of pWE4; lane 4, amplification of chromosomal DNA from LMB1; lane 5, amplification of chromosomal DNA from a chloramphenicol-resistant strain bearing a single crossover; lane 6, amplification of chromosomal DNA from a chloramphenicol-resistant strain bearing a double crossover; lanes 7 and 8, amplification of chromosomal DNA from chloramphenicol-sensitive strains bearing an exact deletion of fla/che P_A.

Characterization of fla/che P_A deletion strain. (i) Anti-ς^D immunoblot.

Computer analyses of sequences 5′ and 3′ of the sigD gene suggest that sigD is the penultimate gene in the fla/che region, >26 kb downstream of the fla/che P_A (4a). To determine the effect of the P_AΔ on sigD expression, total protein was isolated from the appropriate strains and analyzed by anti-ς^D immunoblot. While ς^D protein was found in a wild-type and an flgM null mutant strain, there was no detectable ς^D protein in either the fla/che P_AΔ strain or the sigD null mutant (Fig. 4, lanes 2 and 3). Interestingly, the level of ς^D protein found in the flgM null mutant is comparable to the level found in the wild-type strain, even though transcription from the P_D-3 promoter elements is dramatically increased in the mutant lacking the negative regulator (Fig. 2B, lanes 9 and 10).

FIG. 4.

Anti-ς^D immunoblot. The reactive ς^D protein was visualized in total protein extracts isolated from a wild-type and three mutant strains. Lane 1, wild-type strain (LMB1); lane 2, sigD null mutant (LMB10); lane 3, fla/che P_AΔ strain (LMB214); lane 4, flgM null mutant (LMB213). Fifty micrograms of total protein was loaded into each lane.

(ii) Microscopic observation.

Having demonstrated that there is no detectable ς^D protein in the P_AΔ strain, we examined cell morphology since the lack of ς^D has been shown to result in a filamentous phenotype (28). This is due to the fact that the genes encoding the major autolysins in B. subtilis responsible for hydrolyzing the cell wall after chromosomal replication are preceded by ς^D-dependent promoters (23). The lack of ς^D protein results in a significant decrease in autolysin production and thus a filamentous phenotype. The results obtained by microscopic observation are shown in Fig. 5. The wild-type strain grows as short rods (Fig. 5A), whereas the sigD null mutant grows as long filaments (Fig. 5B). The P_AΔ strain also grows as long filaments (Fig. 5C), which is consistent with a lack of detectable ς^D protein in this strain (Fig. 4, lane 3). This phenotype can also be explained by a complete inhibition of ς^D activity by FlgM (33). A double mutant bearing the P_AΔ and a null mutation in flgM (P_AΔ/flgM null strain) no longer grows as long filaments (Fig. 5D) but as short rods like the wild-type cells.

FIG. 5.

Microscopic observation of wild-type and mutant strains. (A) Wild-type strain (LMB1); (B) fla/che P_AΔ mutant (LMB214); (C) sigD null mutant (LMB10); (D) fla/che P_AΔ/flgM double mutant (LMB216). White scale bar = 5 μm.

(iii) hag::lacZ expression.

To further assess ς^D activity, we examined the expression of a hag::lacZ reporter construct in the wild-type, P_AΔ, and P_AΔ/flgM null strains. The hag gene, which encodes the flagellin protein in B. subtilis, is preceded by a strong ς^D-dependent promoter element that is at least partially responsible for its high level of expression (30). In fact, 7% of the normal amount of ς^D protein is sufficient to produce nearly 70% of the wild-type levels of flagellin protein (28). Expression of the hag::lacZ fusion, as measured by β-galactosidase assay, is therefore a sensitive measure of ς^D activity in the cell.

We found that expression of the hag::lacZ reporter is greatly reduced in a strain bearing the P_AΔ (Fig. 6B). The pattern of expression is nearly identical to that found for an isogenic strain lacking the hag::lacZ fusion and for a strain bearing the fusion as well as a null mutation in sigD (data not shown). The low levels of expression found early in logarithmic growth appears to result from an error introduced by very low cell density during this time. In the P_AΔ/flgM null double mutant, hag::lacZ expression is restored to approximately 50% of the wild-type level and exhibits an altered pattern (Fig. 6C). Specifically, hag::lacZ expression in the double mutant appears higher earlier in vegetative growth yet never displays the significant increase at the end of logarithmic growth that is found in the wild-type strain (Fig. 6A).

FIG. 6.

hag::lacZ expression in wild-type and mutant strains. β-Galactosidase activity was monitored in strains bearing a hag::lacZ reporter construct throughout growth in sporulation medium. For each panel, the left y axis is absorbance at 600 nm, the right y axis is β-galactosidase activity as expressed in Miller units, and the x axis is time expressed in hours. Symbols: ○, growth; ▪, β-galactosidase activity. (A) Wild-type strain (LMB219); (B) fla/che P_AΔ strain (LMB220); (C) fla/che P_AΔ/flgM double mutant (LMB222).

DISCUSSION

The 620-nucleotide fragment upstream of flgB, the first gene in the fla/che transcription unit, has been sequenced and characterized by primer extension and genetic analyses. Computer analysis of the sequence data obtained has allowed for (i) the demarcation of the putative limits of the fla/che promoter region as the intergenic region between the cod operon and flgB and (ii) the identification of a ς^D promoter consensus sequence, P_D-3, a previously described sequence of DNA shown to be recognized by the ς^D holoenzyme in vitro (11). However, whether this promoter was functional in vivo and the function of the gene(s) dependent on this element for transcription initiation were unknown. Primer extension data demonstrate that the P_A immediately upstream of flgB (fla/che P_A) and the P_D-3 promoter are active in vivo (Fig. 2A, lanes 5 to 8, and 2B, lanes 5 to 6). Expression from the P_D-3 element is absent in a sigD null mutant and is increased in a flgM null mutant (Fig. 2B, lanes 7 to 10), demonstrating that this promoter is ς^D dependent and is regulated by the flagellum-specific regulator, FlgM. The presence of two functional promoters upstream of a single transcription unit, recognized by different forms of the holoenzyme, is common in B. subtilis and allows for precise control of gene expression depending upon the availability and activation of the alternate sigma factors (25). Interestingly, the structural gene for ς^F, the enteric homolog of ς^D, has been shown to be transcribed from two overlapping promoters, one recognized by the major form of the holoenzyme and a second recognized by the ς^F holoenzyme (35). The spacing of the fla/che P_A and P_D-3 elements, however, differs from the enteric system in that the P_D-3 is 133 bp upstream of fla/che P_A.

The functional P_A promoter, identified above and referred to as fla/che P_A, was deleted to analyze its role in sigD expression and ς^D activity. In B. subtilis cells lacking the fla/che P_A (P_AΔ), no ς^D protein was detected, in contrast to ς^D levels in wild-type or flgM null mutant backgrounds (Fig. 4, lanes 1 and 4). Microscopic observation of the strain bearing the P_AΔ demonstrates that this mutant grows as long filaments, which is consistent with a lack of ς^D expression or activity (Fig. 5B). However, a mutant strain containing both the P_AΔ and a null mutation in flgM no longer grows as long filaments. This suggests that a low level of ς^D protein is present in the P_AΔ strain (which is not detectable by immunoblot) but activity of the remaining sigma factor is inhibited by the FlgM regulator. It is likely that deletion of the fla/che P_A results in a lack of HBB gene expression, preventing the functional assembly of the HBB complex, which would allow FlgM to inhibit ς^D activity. This inference is further supported by the lack of expression of a hag::lacZ reporter construct in the P_AΔ mutant, whereas expression is partially restored in a P_AΔ/flgM null double mutant (Fig. 6B and C).

These data demonstrate that the sigD structural gene, located >26 kb downstream of the fla/che P_A, is part of the fla/che operon, since the fla/che P_A is necessary for wild-type sigD expression. While the fla/che P_A is necessary for the production of detectable ς^D protein, expression of sigD may also involve another promoter(s) within the fla/che operon, since strains lacking the fla/che P_A and bearing a flgM null mutation have the ς^D function partially restored. In the absence of the FlgM regulator, low levels of ς^D protein produced from an internal promoter(s) could bind to RNA polymerase and initiate transcription from the P_D-3 promoter. Expression from the P_D-3 promoter would then result in increased expression of the fla/che transcription unit, including sigD. A minor ς^A-dependent promoter immediately upstream of sigD, P_sigD, has been identified recently, and deletion of this element results in a modest decrease in fla/che expression (2). Studies are in progress to determine the relative importance of P_D-3, fla/che P_A, and a putative promoter(s) within the fla/che operon in initiating transcription of sigD.

In this work, we have demonstrated that dual promoters recognized by the ς^A and ς^D holoenzymes initiate transcription of the fla/che operon, including sigD. In a previous study, however, it was concluded that the fla/che transcription unit (and sigD in particular) is not expressed from a ς^D-dependent promoter (3). This inference was based on the observation that ς^D protein levels in a flgM mutant are not elevated with respect to a wild-type control (3), since a lack of the FlgM regulator has been shown to increase the expression of ς^D-dependent genes 10- to 15-fold (6). We have also found the level of ς^D protein in the flgM mutant (Fig. 4, lane 4) to be comparable to the amount found in a wild-type strain, despite the significant increase in mRNA production from P_D-3 found in this strain (Fig. 2B, lanes 9 and 10). Our results suggest that either (i) P_D-3 drives transcription of fla/che including sigD and there is posttranscriptional regulation of ς^D expression, (ii) the P_D-3 transcript does not extend 26 kb and does not encode ς^D protein, or (iii) P_D-3 is a very weak promoter, in comparison to fla/che P_A, such that even increased ς^D activity cannot increase overall fla/che gene expression. We favor the first hypothesis and postulate that there is a general mechanism for posttranscriptional regulation of flagellar gene expression. As shown for sigD mRNA and ς^D protein in this study, it has been demonstrated that hag mRNA production in a flgM null mutant is significantly increased but flagellin protein levels in the cell are nearly identical to flagellin protein levels found in a wild-type strain (9).

Our working model for transcriptional regulation of fla/che is diagrammed in Fig. 7. Transcription of the fla/che operon, including sigD, is initiated from the fla/che P_A. The importance of P_sigD, the minor P_A immediately upstream of sigD (2), in allowing expression of ς^D protein under normal conditions is unclear. Expression of ς^D protein and its association with RNA polymerase allows for transcription from P_D-3, which may account for the rapid accumulation of ς^D protein during logarithmic growth (29a) and the rapid increase in ς^D activity found during this period of growth (10, 30). In fact, several alternate sigma factors are encoded in transcription units preceded by a cognate promoter sequence, allowing for a rapid accumulation of the factor (16–18). The dependence of wild-type fla/che expression on the P_D-3 element would also allow for a sensitive monitor of HBB assembly, since ς^D activity is intimately tied to FlgM regulation (32).

FIG. 7.

Model for fla/che expression. Transcription of the fla/che operon, including sigD, is initiated from the fla/che P_A. Once ς^D protein is synthesized, expression from P_D-3 is initiated and may account for the rapid accumulation of this transcription factor. The importance of the minor P_A immediately upstream of sigD under normal conditions is unclear.

Although the flagellar regulons in both B. subtilis and the enteric bacteria are highly homologous (38), the physical organizations of the genes that comprise the HBB complex and the alternate sigma factors are very different. The genes encoding the HBB and ς^D in B. subtilis are found within the fla/che operon (38). In contrast, the genes encoding the HBB in the enteric bacteria are found in 13 operons located throughout the bacterial chromosome (21, 22) and the gene for the ς^D homolog, ς^F, is found in yet another transcription unit (27). Our studies demonstrate that the precise expression of fla/che operon gene products is dependent on at least two forms of the holoenzyme and morphogenetic control by FlgM. Moreover, our data suggests that posttranscriptional regulation of the operon occurs.