Synthesis of the ς^D Protein Is Not Sufficient To Trigger Expression of Motility Functions in Bacillus subtilis

Abstract

The gene encoding ς^D, sigD, is transcribed from two promoter regions, the fla/che promoter region in front of the fla/che operon and P_sigD directly in front of sigD. If ς^D is translated from transcripts originating from P_sigD, the cell is unable to express motility functions but synthesizes autolysins. Therefore, one function of the additional promoter is to allow the cell to express autolysins without expressing motility functions as well.

Gene expression in bacteria is regulated primarily at the level of transcription. The specificity of transcription initiation, which rests on interactions between the RNA polymerase and the promoters with which it makes contact, is determined by sigma factors. Although most genes are preceded by promoters which are recognized by the main vegetative sigma factor, a subset of genes is preceded by promoters that are recognized by alternative sigmas. These genes are often part of regulons that encode products that are not necessary for normal vegetative bacterial growth but are important under specific circumstances. For example, in Bacillus subtilis most of the genes whose products are necessary for sporulation are transcribed by alternative sigma factors (see references 6, 10, and 20 for recent reviews). ς^H is the first alternative sigma factor in this regulatory cascade (10). Gene expression in the forespore is mediated by the activities of ς^F and ς^G, and transcription in the mother cell is dependent on ς^E and ς^K (20).

Sporulation is not the only response of B. subtilis to the stationary phase. The onset of stationary phase induces, at least, the synthesis and activity of two additional sigma factors that are not necessary for sporulation, ς^B and ς^D (10). The ς^B regulon is activated in response to different kinds of stress and stationary phase (2, 3). The proteins encoded by ς^B-dependent genes serve different physiological functions. It has been proposed that expression of these genes is advantageous during starvation and stress (11), but sigB mutants survive most of the stressful conditions as efficiently as the wild type does (2).

The B. subtilis transcription factor ς^D is necessary for the expression of genes required for bacterial motility and autolysins (12, 18, 22). Its counterpart in Escherichia coli, ς^F, serves a similar but more specialized function. With a single exception (13), all genes transcribed by E. coli ς^F are necessary for motility. Indeed, ς^F-negative E. coli strains are partially complemented by a functional copy of the B. subtilis sigD gene (5). The genes of the E. coli ς^F regulon are organized in several different regulatory groups (see reference 21 for a recent review). One major regulator, beside ς^F itself, is FlgM, an anti-sigma factor which binds to ς^F and blocks its activity (28).

In B. subtilis, transcription of the two different groups of genes which are dependent on ς^D containing RNA polymerase is influenced by several transition state regulators (19, 24, 30, 31). However, it is not clear whether these effects are due to a direct interaction of these proteins with ς^D-dependent promoters or are brought about by the physiological side effects of mutations in the genes encoding these regulators. One of the major regulators of ς^D activity is FlgM (4, 9, 26), the counterpart of the E. coli FlgM protein. In its absence, several operons encoding proteins necessary for the late flagellar synthesis are overproduced. Based on genetic data and due to the homology to E. coli FlgM, it is believed that B. subtilis FlgM acts as an anti-sigma factor.

The sigD gene is transcribed by the activity of at least two promoter regions (1, 7). The biological importance of the second promoter, present directly in front of the sigD gene, is not clear. We show that, at least under specific environmental conditions, this promoter is responsible for sigD transcription. If P_sigD is the sole promoter to transcribe sigD, the transcription of motility functions is impaired. Therefore, we propose that one function of P_sigD and P_sigD-derived transcripts is to allow autolysin expression under circumstances which are not favorable for the expression of motility functions.

Transcription of P_sigD and ς^D-dependent genes in minimal medium.

We were interested in determining whether temporal regulation of P_sigD is identical in minimal medium and in complex medium. To that end, we constructed plasmid pSD26 (Table 1) by inserting the 120-bp EcoRI-Sau3AI fragment from pSD5 (1) into EcoRI-SnaBI-digested pDH32M (15). All cloning experiments were done with E. coli DH5α as the host. The resulting plasmid harbors a 96-bp P_sigD-encoding fragment of the fla/che operon (positions 1104 to 1200 in reference 23) as a transcriptional fusion with the lacZ gene. The plasmid was linearized and integrated into the amyE gene of B. subtilis 168 to give B. subtilis SD26. In the resulting strain lacZ transcription was dependent solely on P_sigD activity. In addition, we constructed pSD28 by ligating the 0.9-kbp ClaI fragment from pSD26 into pKL2 (32). To obtain pSD28N, pSD28 was digested with PacI and EcoRI and ligated to the 0.5-kb PacI-EcoRI fragment from pSD5. This plasmid, which also encodes a transcriptional fusion of sigD to lacZ, was integrated into the fla/che operon of B. subtilis 168 to give B. subtilis SD28N. In this strain, lacZ expression was driven by P_sigD activity and the activity of the promoters directing the expression of the fla/che operon. We grew B. subtilis SD26 and SD28N in MOPSO minimal medium (16) with different concentrations of glucose as the carbon source. At glucose concentrations less than 0.2%, the C source was the growth-limiting factor (Fig. 1). Samples were removed at different growth stages, and β-galactosidase assays were done as described previously (27). Each experiment was repeated at least three times. Only the results of a single experiment are given. In MOPSO minimal medium, both P_sigD-dependent and fla/che-dependent β-galactosidase expression at the beginning of stationary phase were repressed in the presence of 0.2% glucose. At glucose concentrations of 0.1% and lower, β-galactosidase activity was induced at the onset of stationary phase, presumably due to the depletion of glucose. P_sigD-dependent activity of B. subtilis SD26 exceeded the fla/che-dependent activity of B. subtilis SD28N if the glucose concentration was less than 0.1%. This seems rather surprising since, as mentioned above, P_sigD drives LacZ expression in B. subtilis SD26 whereas the β-galactosidase synthesis determined in B. subtilis SD28N is driven by P_sigD plus the activity of the promoters in front of the fla/che operon. Under all other conditions tested, the β-galactosidase activity of B. subtilis SD28N exceeded the activity of B. subtilis SD26 (data not shown). A possible explanation of the higher levels of β-galactosidase in B. subtilis SD26 than B. subtilis SD28N in minimal medium is the site of integration of the constructs in each strain. The P_sigD-lacZ fusion was integrated into the amyE locus. amyE is next to the origin of replication, whereas the fla/che operon is near the terminus of the chromosome (17). If the chromosome is replicating, the majority of the cells contain two copies of amyE and therefore of P_sigD-lacZ but only a single copy of the fla/che operon. The fact that LacZ expression determined in B. subtilis SD28N did not exceed the expression determined in B. subtilis SD26 indicated that P_sigD activity was far higher than the activity of the promoter region in front of the operon. This implied that the synthesis of the proteins encoded by the upper part of the fla/che operon was reduced or even abolished.

TABLE 1.

Strains and plasmids used during this study

Strain or plasmid	Genotype or relevant marker	Source or reference
Strains
E. coli DH5α	hsdR17 endA1 recA1 gyrA96 thi relA1 supE44 φ80dlacZΔM15 Δ(lacZ-argF)U169	BRL
B. subtilis 168	trpC2	BGSC
B. subtilis SD26	trpC2 amyE::(cat sigD-lacZ)	This work
B. subtilis SD28N	trpC2 cat sigD (N-term)-lacZ	This work
B. subtilis HY5S	trpC2 pHY5S	This work
B. subtilis FLI	trpC2 pks::(Pfli-lacZ cat)	This work
B. subtilis MOT	trpC2 pks::(PmotAB-lacZ cat)	This work
B. subtilis HY5S	trpC2 pHY5S, fla/che::(spec neo T4t)	This work
SD28teo/SD30teo
B. subtilis FLI	trpC2 pks::(Pfli-lacZ) fla/che::(spec neo T4t)	This work
SD28neo/SD30teo
B. subtilis MOT	trpC2, pks::(PmotAB-lacZ), fla/che::(spec neo T4t)	This work
SD28teo/SD30teo
B. subtilis DH32M	trpC2 amyE::(lacZ cat)	This work
Plasmids
pBEST501	bla neo	14
pHY5S	bla tet PcwlB-lacZ	18
pDH32M	bla neo	15
pHP45Ω	bla spec T4t	29
pSD26	bla amyE::(cat sigD-lacZ)	This work
pSD28N/30N	bla cat sigD (N-term)-lacZ	This work
pSD28T/30T	bla cat spec T4t::sigD (N-term)-lacZ	This work
pSD28teo/30teo	bla neo spec T4t::sigD (N-term)	This work

FIG. 1.

P_sigD and fla/che-dependent β-galactosidase expression in minimal medium with different amounts of glucose as the carbon source. Shaded symbols indicate growth of B. subtilis SD26 with the indicated amount of glucose; solid symbols indicate β-galactosidase activity of B. subtilis SD26; and open symbols indicate β-galactosidase activity of B. subtilis SD28N. Circles, 0.2% glucose; squares, 0.1% glucose; triangles, 0.05% glucose; inverted triangles; 0.02% glucose. Activity units are those of Miller (25).

Induction of motility functions is absent in minimal medium.

Some of the proteins encoded by the fla/che operon are necessary for the basal structure of the flagellum. If this structure is missing, the flagellum cannot be assembled. Therefore, the synthesis of proteins necessary for motility which are encoded by genes and operons beside fla/che would be a waste of energy. We therefore asked whether operons which are under direct control of ς^D are expressed in MOPSO minimal medium. We constructed B. subtilis MOT and B. subtilis FLI strains encoding fusions of lacZ to P_motAB and P_fli, respectively, by transforming B. subtilis 168 with chromosomal DNA from B. subtilis P_mot-171 and P_fliD-224 obtained from A. Galizzi (8). In addition, we transformed B. subtilis 168 with plasmid pHY5S, obtained from J. Sekiguchi, to give B. subtilis HY5S. This strain harbors a fusion of the promoter of cwlB, encoding the major autolysin, to lacZ. The strains were grown in MOPSO minimal medium with 0.05% glucose. The results of this experiment are depicted in Fig. 2. Whereas transcription of the P_cwlB-dependent lacZ gene was induced at the onset of stationary phase, P_fli- and P_motAB-dependent transcription was not. Therefore, it was clear that in MOPSO medium with limiting amounts of glucose, the induction of ς^D synthesis is not a sufficient signal for the induction of transcription of motility genes but a signal for elevated autolysin synthesis. Mutations in the putative anti-sigma factor FlgM restore hag and motAB transcription in strains with an interrupted fla/che operon (26). Since FlgM has an identical effect on autolysin genes and motility genes (26), it seems unlikely that FlgM is the regulator responsible for this effect.

FIG. 2.

Expression of P_motAB-lacZ (■), P_fli-lacZ (▾), and P_cwlB-lacZ (●) in minimal medium with 0.05% glucose as the carbon source. β-Galactosidase activity is plotted against OD₆₀₀ of the culture. Activity units are those of Miller (25).

Transcription of ς^D-dependent genes is affected by insertions within the fla/che operon.

Whereas ς^D synthesis in minimal medium seemed to be dependent mainly on sigD transcription starting from P_sigD, at least two different classes of transcripts, originating either from P_sigD or from the fla/che promoter region, are used as a template for ς^D synthesis in complex medium (1). The results described above indicated that P_sigD-driven sigD expression induces transcription of autolysin genes but is not sufficient to allow transcription of ς^D-dependent motility genes. If this is indeed the case, the temporal regulation of P_motAB, P_fli, and P_cwlB should be different as well. P_cwlB has to become active as soon as P_sigD-dependent ς^D synthesis commences, whereas P_fli and P_motAB activity must parallel the P_fla/che-dependent ς^D synthesis. We therefore grew B. subtilis MOT, FLI, HY5S, SD26, and SD28N in parallel in NB medium (Oxoid, Basingstoke, United Kingdom). The cells were grown at 42°C, a temperature which allows a better differentiation between P_sigD- and fla/che-dependent ς^D expression (32a). Figure 3 shows the result of this experiment. The maximal β-galactosidase activity of each single strain was taken as 100%. In accordance with previous data (1), P_sigD activity was induced at an optical density at 600 nm (OD₆₀₀) of about 0.5 and reached maximal values at an OD₆₀₀ of about 0.6. The temporal regulation of cwlB-dependent LacZ expression was similar, but that of P_motAB-, P_fli-, and fla/che-dependent β-galactosidase expression was different. P_motAB- and P_fli-dependent activity did not commence until the OD₆₀₀ reached 0.7. Maximal activity of fla/che-, P_motAB-, and P_fli-dependent β-galactosidase activity was obtained at an OD₆₀₀ of about 1. This result is in accordance with the model proposed above. To test the model in vivo, we created an artificial situation where fla/che transcription is blocked in front of sigD. To do so, we constructed plasmids pSD28teo and pSD30teo. First, we created plasmid pSD30N by digesting pSD28N with PacI and BamHI. The protruding ends were trimmed with T4 DNA polymerase, and the plasmid was religated. In pSD30N, P_sigD was completely removed. pSD28T and pSD30T were obtained by inserting the Ω element of pHP45Ω (29), which harbors two transcriptional terminators, as a BamHI fragment into BglII-digested pSD28N or as a SmaI fragment into XbaI-digested pSD30N to give pSD28T and pSD30T, respectively. The lacZ and cat genes were deleted by digesting the plasmids with SnaBI. The fragments which code for the 5′ end of the sigD gene were ligated to a 1-kb fragment, created by a SmaI-SnaBI double digest of pBEST501 (14), to give pSD28teo and pSD30teo, respectively. The resulting plasmids confer neomycin resistance to both E. coli and B. subtilis. Integration of pSD28teo into the B. subtilis chromosome inserted the two transcription terminators between orfB and P_sigD. In this strain, P_sigD was the only promoter to transcribe sigD. Integration of pSD30teo inserted the terminators between P_sigD and sigD. Therefore, sigD expression was most probably completely abolished. These plasmids were integrated into the fla/che operon of B. subtilis strains which harbored either the P_fli, P_motAB, or P_cwlB fusions to lacZ. The β-galactosidase activity of the respective strains grown in NB medium was determined. Integration of pSD30teo resulted in a loss of β-galactosidase activity (data not shown). Therefore, it seemed unlikely that readthrough from the fla/che operon occurred. P_cwlB induction during mid-log phase was not changed by the insertion of pSD28teo, but the same integration had a striking effect on P_motAB and P_fli; the activity of both promoters was severely reduced (Fig. 4). It is not possible to explain this result by the lack of transcription of any gene. In B. subtilis SD28teo, all genes encoded by the native fla/che operon are transcribed either by the activity of the fla/che promoter region or by the activity of P_sigD as soon as P_sigD is induced (1). However, during mid-log phase, transcription of the motAB and the fli genes was not activated whereas P_cwlB activity was induced. The results are in agreement with our model. Transcription of the entire fla/che operon as a single transcript is necessary to allow ς^D to transcribe genes encoding motility functions, whereas P_sigD activity is sufficient only to drive autolysin expression. We are unable to provide any mechanistic explanation for this phenomenon; nevertheless, this observation is in agreement with some recently published data. Complementation of a sigD-negative B. subtilis strain with a plasmid-located copy of sigD does not restore the motility-negative phenotype of the strain but makes it autolysin positive (33). Mutations which cause a severe reduction of sigD expression also cause a reduced expression of autolysins and motility functions. Overexpression of ς^D restores autolysin synthesis but does not influence motility functions (31). Taken together, our results and the results from other groups indicate that transcription of sigD alone allows the expression of autolysins but not the expression of motility functions. Therefore, one of the functions of P_sigD is to allow high-level expression of autolysins under circumstances which are not favorable for expression of motility genes. The molecular mechanism which allows this differentiation between the two classes of ς^D-dependent genes is under investigation.

FIG. 3.

Relative β-galactosidase activity of different B. subtilis strains plotted against OD₆₀₀. Bacteria were grown in NB medium at 42°C, a temperature which allows a better distinction between P_sigD- and P_fla/che-dependent β-galactosidase activity (data not shown). The maximal activity obtained during growth for each single strain was taken as 100%. ●, B. subtilis SD26 (P_sigD-lacZ); ▾, B. subtilis SD28T (fla/che-lacZ); ○, B. subtilis HY5S (P_cwlB-lacZ); □, B. subtilis MOT (P_motAB-lacZ); ▿, B. subtilis FLI (P_fli-lacZ).

FIG. 4.

Influence of an integration into the fla/che operon on ς^D-dependent gene expression. P_motAB-lacZ (squares), P_fli-lacZ (inverted triangles), and P_cwlB-lacZ (circles) activity was determined in a B. subtilis 168 background (solid symbols) and in a B. subtilis SD28teo background (open symbols). Shaded triangles indicate growth of B. subtilis SD28teo/P_fli-lacZ. Activity units are those of Miller (25).