INTRODUCTION

The SOS regulatory system, as characterized in the bacterium Escherichia coli, comprises a set of DNA repair and cellular survival genes that are coordinately induced by DNA damage (1, 2, 3). The expression of these DNA damage-inducible (din) genes, or SOS genes, is controlled by two proteins (which are themselves products of SOS genes): the LexA protein, which represses SOS gene expression, and the RecA protein, which is activated by a metabolic signal to cause the proteolytic inactivation of LexA.

LexA binds to the operator sequences of SOS genes and inhibits their transcription in vitro (3, 4, 5). Analyses of recA and lexA operator mutants (6), quantitative footprinting studies of LexA binding to the recA operator (7, 8), and a comparison of the upstream sequences of the known SOS genes (2) has led to the identification of the so-called SOS box, CTGTN₈ACAG, which appears to be the principal recognition site for a LexA dimer. Induction of the SOS genes following DNA damage occurs when RecA, activated by an inducing signal, promotes the proteolytic destruction of LexA. Corresponding to the requirement for an activating signal, RecA is activated for repressor (LexA and certain phage repressors) cleavage in vitro when it binds single-stranded DNA and nucleoside triphosphate to form a ternary complex (9, 10, 11). There is strong evidence that for certain DNA damaging treatments, the inducing signal is single-stranded DNA exposed by either the processing or replication of damaged DNA (12, 13).

RecA-mediated cleavage of the 22.3 kDa LexA protein occurs at a single site between Ala-84 and Gly-85 (14). This site is part of a flexible hinge region that separates two functional domains (15), the amino-terminal domain that binds to the SOS operator and the carboxyl-terminal domain that is involved in dimerization (16, 17). A significant feature of the cleavage reaction promoted by RecA derives from the observation that LexA, in the absence of RecA, undergoes autodigestion at alkaline pH at the same Ala-Gly bond that is cleaved in the RecA-promoted reaction (18). Thus, contrary to early descriptions of RecA as a highly specific protease, RecA functions as a coprotease to facilitate the cleavage catalyzed by repressor functional groups. Kinetic analysis of the autodigestion reaction indicates that repressor cleavage is intramolecular and involves the deprotonation of an amino acid residue with a pK_a of 9.8 (19). A model for the cleavage mechanism, similar to that for serine proteases, has been proposed in which Ser-119 is activated as a nucleophile by Lys-156 (20, 21, 22).

The bacterium Bacillus subtilis responds to agents that damage DNA or block DNA replication by inducing an SOS-like system (23, 24), and several components of the B. subtilis SOS-like regulon have been identified. The B. subtilis RecA protein, which cross-reacts strongly with E. coli RecA antibody, is induced by mitomycin C, UV radiation, and nalidixic acid (25, 26). B. subtilis RecA promotes the cleavage of E. coli LexA repressor in vitro (25) and in vivo (27, 28). A set of B. subtilis din genes was identified whose induction by DNA damaging treatments is blocked in recA mutants (29). Inspection of the DNA sequences upstream of the B. subtilis din genes revealed a consensus sequence, 5-GAACN₄GTTC-3, overlapping the dinA, dinB, dinC, and recA promoters (30). This sequence was proposed as a possible binding site for the B. subtilis SOS repressor and has since been found overlapping the promoters of the recM13, dnaX (31), and dinR (32) genes. Using the din promoter regions to search for the putative SOS repressor, a 23-kDa DNA binding protein was purified that binds specifically to the consensus site and whose binding activity was destroyed by activated B. subtilis RecA protein in vivo and in vitro (33).

The B. subtilis dinR gene encodes a 22.8-kDa polypeptide having 34% identity and 47% similarity with the E. coli LexA protein (32). Thus, it has been considered the likely candidate for the B. subtilis lexA analogue. We report here the cloning, overexpression, and purification of the DinR protein, and we show that it has all of the activities of E. coli LexA in vitro: specific binding to din promoters and inhibition of RNA polymerase binding and transcription, autodigestion at high pH, and RecA-mediated cleavage. We also show that the dinR gene codes for the protein previously identified as the SOS repressor (33).

EXPERIMENTAL PROCEDURES

Plasmid pET21a and BL21(DE3) E. coli cells were purchased from Novagen. The B. subtilis RecA (25) protein was purified as described previously. Polyclonal B. subtilis DinR antiserum was prepared by the subcutaneous injection of New Zealand White rabbits with DinR protein of greater than 95% purity. Initially 100 µg of DinR protein suspended in 1 ml of complete Freund's adjuvant was injected, and two boosters of 100 µg of DinR suspended in 1 ml of incomplete Freund's adjuvant were injected at 3-week intervals. The antiserum was collected 10 days after the third injection. Oligonucleotides were synthesized on a Milligen Cyclone Plus DNA synthesizer. Restriction enzymes, DNA-modifying enzymes, and Taq polymerase were purchased from New England Biolabs Inc. and Promega and used as recommended by the manufacturers. Affinity-purified goat anti-rabbit horseradish peroxidase conjugate and heparin-agarose were purchased from Bio-Rad. Mitomycin C was purchased from Sigma. Nitrocellulose filters were purchased from Schleicher & Schuell. RNA polymerase was provided by Leendert Hamoen (University of Groningen). The dinR deletion strain 8G5 (pLGW3) was provided by Bert Jan Hajema (University of Groningen). ATPS¹ was from Boehringer Mannheim. Radiolabeled nucleotides were from DuPont NEN. Amino acid analyses were performed by Carol Gross (University of Kentucky).

The lysates used in this study were prepared from bacterial strains grown in LB medium at 37 °C and harvested in late log phase (A₆₀₀ = 0.8). Pelleted cells were resuspended in 5 ml of lysis buffer (20 mM Tris, pH 7.5, 10% (w/v) sucrose, 1 mM dithiothreitol, 0.1 mM EDTA) per liter of bacterial culture. After addition of lysozyme to 0.2 mg/ml, cells were incubated on ice for 30 min, sonicated for 1 min, and incubated at 37 °C for 15 min. Debris was removed by centrifugation at 100,000 × g for 45 min at 4 °C, and the supernatant was either used for mobility shift assays or Western analysis as described below.

The coding portion of dinR was amplified from YB886 (34) chromosomal DNA using a 5-primer modified such that an NdeI restriction site was introduced at the initial ATG codon (GAGGTGCGACATATGACGAAGCTATC); the 3-primer (CCCAAAAGTCAGGTCGACCCTG) corresponded to a site downstream of the structural gene (+715) modified to contain a SalI restriction site. After digestion with NdeI and SalI, the polymerase chain reaction product was ligated in NdeI-SalI-digested pET21a, and the ligation mixture was used to transform DH5 cells. Plasmid DNA obtained from DH5 transformants was then transformed into competent BL21(DE3) cells.

E. coli strain BL21(DE3) containing pET21a-dinR was grown in 1 liter of LB broth containing carbenicillin (50 µg/ml) with shaking until A₆₀₀ = 0.6. The culture was then induced with 10 ml of IPTG (100 mM) and grown for an additional 3 h. Cells were harvested by centrifugation at 4 °C, 5000 × g for 20 min, and then resuspended in 5 ml of 20 mM Tris, pH 7.5, 10% (w/v) sucrose, 1 mM EDTA. Cells were lysed with lysozyme (0.2 mg/ml) by incubation on ice for 30 min followed by a 15-min incubation at 37 °C. After 3 × of freeze-thawing and sonication, debris was removed by centrifugation at 18,000 × g for 15 min, and the supernatant was used for further purification. The supernatant (0.5-2 ml) was filtered and applied to a 5 ml-heparin-agarose column, equilibrated in 20 mM Tris, pH 7.5, 10 mM NaCl, and subjected to FPLC chromatography. Following elution with a 40-ml linear NaCl gradient (10 mM to 1 M), fractions containing DinR protein were pooled, concentrated 10-fold by centrifugation in a Centricon 10 unit (Amicon), and diluted to the original volume in 20 mM Tris, pH 7.5, 10 mM NaCl; concentration and dilution was then repeated twice. A sample was then applied to a Mono S column and chromatographed as described for the heparin-agarose column, and fractions containing DinR were pooled and dialyzed against 20 mM sodium phosphate, pH 7.5, 200 mM NaCl and then concentrated by centrifugation in a Centricon 10 unit.

DNA probes corresponding to the dinC (30 to +69), dinB (155 to +70), and recA (77 to +170) promoter regions were prepared as described previously (33). Crude extract or purified fractions were incubated with one of the radiolabeled DNA fragments for 30 min at 25 °C in the following incubation buffer (2): 12 mM HEPES-NaOH, pH 7.9, 4 mM Tris-Cl, pH 7.9, 12% glycerol, 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 2 µg poly(dI-dC)·poly(dI-dC), and 0.3 mg/ml bovine serum albumin. This incubation mixture (10 µl) was loaded on a 4% (acrylamide:bisacrylamide ratio of 80:1) non-denaturing polyacrylamide gel, and electrophoresis was begun immediately. The buffer within the gel and the running buffer were both 25 mM Tris-Cl, pH 8.5, 250 mM glycine, and 1 mM EDTA. Samples were electrophoresed at 25 mA, and the dried gel was subjected to autoradiography.

B. subtilis strains were treated with 1.0 µg/ml mitomycin C, and cell samples were analyzed by immunoblot transfer as described previously (19). Nitrocellulose membranes containing electrophoretically transferred proteins were incubated with B. subtilis DinR antiserum (diluted 1:1000), followed by peroxidase-conjugated anti-rabbit antibody (diluted 1:3000) essentially as described previously (21).

DNA containing the dinC promoter region was obtained by EcoRI-ClaI digestion of cesium chloride-purified pPL603-din17 DNA. The 199-base pair dinC fragment (130 to +69) thus released was then gel purified and radiolabeled at either restriction site. DNase footprinting experiments were performed using reagents from the Promega core footprinting kit as per manufacturer instructions. Samples (50 µl) for hydroxyl radical footprinting were incubated as described for mobility shift assays, without glycerol in the binding buffer, followed by addition of 6 µl of a freshly made mixture of (NH₄)₂Fe(SO₄)₂·6H₂O) (0.93 µM), EDTA (1.9 mM), H₂O₂ (1.1%), and sodium ascorbate (37 mM). After incubation for an empirically determined time period, reactions were stopped by the addition of 44 µl of stop solution (0.7 M sodium acetate, pH 5.2, yeast tRNA (0.14 mg/ml), 73 mM thiourea). Samples were prepared for electrophoresis by phenol extraction, ethanol precipitation, and resuspension in 4 µl of loading buffer (35). Samples were subjected to electrophoresis in a 6% acrylamide-urea gel and autoradiography.

In vitro transcription was analyzed using the "run-off" assay as described (36). Reaction mixtures contained 40 mM Tris, pH 8.0, 0.5 mM EDTA, 1 mM dithiothreitol, 5 mM MgCl₂, 100 mM KCl, bovine serum albumin (0.1 mg/ml), and indicated amounts of B. subtilis RNA polymerase, restriction-digested plasmid DNA, and DinR protein. After incubation at 37 °C for 15 min, a nucleotide mixture containing [-³²P]CTP was added to give final concentrations of 0.1 mM ATP, 0.1 mM GTP, and 0.1 mM UTP. After incubation at 37 °C for 1 min, heparin (0.025%) was added, and the reaction solution was incubated for an additional 10 min before adding CTP (0.1 mM). Reaction was incubated at 37 °C for 10 min after CTP addition and then stopped by the addition of stop solution (0.05% bromphenol blue and 0.05% xylene blue in formamide). Samples were subjected to electrophoresis in a 9% acrylamide-urea gel and autoradiography. Template DNA was a HindIII-ClaI fragment from plasmid pBT61 containing the recA promoter region and ending at +292 of the recA structural gene; the recA run-off transcript was identified by comparison with ethidium bromide-stained molecular weight markers (not shown) run alongside the transcription reactions.

Reaction mixtures, typically containing 10 µM DinR in 100 mM boric acid buffer adjusted to the appropriate pH, were incubated at 37 °C. After selected times, aliquots of the reaction were removed, added to 6 × SDS-PAGE sample buffer to stop the reaction, and then subjected to SDS-PAGE (13%). After staining gels with Coomassie Brilliant Blue, the cleavage activity was analyzed by scanning uncleaved DinR and cleavage fragments using an Alpha Innotech IS-1000 digital imaging system. Cleavage of DinR by B. subtilis RecA protein was assayed as described previously (25). Reaction mixtures contained 10 mM Tris, pH 7.9, 15% glycerol, 0.5 mM dithiothreitol, 0.5 mM EDTA, 60 mM NaCl, 30 mM MgCl₂, and indicated amounts of RecA, DinR, nucleoside triphosphate, and x174 single-stranded DNA. The reactions were then subjected to SDS-PAGE (13%) and stained with Coomassie Brilliant Blue.

RESULTS

Using the polymerase chain reaction, we produced the dinR gene with unique NdeI and SalI sites located at the beginning and end of the structural gene, respectively (see "Experimental Procedures"). We inserted the promoterless NdeI/SalI-ended dinR gene in plasmid pET21a such that dinR expression is controlled by bacteriophage T7 transcription and translation signals. The T7 promoter also contains the lac operator to minimize expression in the absence of an inducer. After transfer to an E. coli host containing a chromosomal copy of the T7 RNA polymerase, under the control of lacUV5 promoter, the DinR protein was induced by IPTG to comprise more than 40% of the total cellular protein (Fig. 1).

We found that the DinR protein binds tightly to heparin-agarose, which is commercially available as prepacked columns from Bio-Rad or Pharmacia, either of which can be used with an FPLC system. Crude extract from lysed cells was applied directly to a heparin-agarose column and eluted with an NaCl gradient. As illustrated in Fig. 1, DinR protein is purified to near homogeneity by heparin-agarose chromatography, eluting at about 250 mM NaCl. Subsequent chromatography on either Mono S or Superose 12 FPLC columns does not provide any further purification as judged by the specific activity of the protein (DNA binding activity per total protein).

Fig. 1 shows that DinR migrates in an SDS-polyacrylamide gel with an apparent M_r of about 27,000; however, amino acid analysis of the purified protein indicates an M_r of 22,823, consistent with the primary structure predicted from the DNA sequence (32). It is noteworthy that the M_r determined by SDS-PAGE for B. subtilis RecA is also significantly larger than the actual size (25). This behavior is not observed for the E. coli RecA and LexA proteins that are about the same size as the corresponding B. subtilis proteins.

The putative SOS repressor, previously purified by affinity chromatography, binds specifically to the SOS boxes (GAACN₄GTTC) of the B. subtilis dinA, dinB, dinC, and recA genes (33). We conducted mobility shift assays with purified DinR protein and the promoter regions of B. subtilis dinB, dinC, and recA genes. Fig. 2A shows that the DinR protein retards the mobility of DNA fragments containing recA (77 to +34), dinB (155 to +70), and dinC (130 to +69) promoter regions. To determine if additional cellular proteins (not present in the E. coli strain from which DinR was purified) might be involved in din promoter binding in B. subtilis, we ran samples containing purified DinR alongside samples containing crude extract from YB1015 (recA4) B. subtilis cells (it was previously shown that the putative SOS repressor produces mobility shifts with din promoters that are identical to those produced by these crude extracts (33)). In every case, DinR causes mobility shifts that are visually indistinguishable from those produced by crude extract. By contrast, a B. subtilis strain, 8G5(pLGW3), in which the dinR gene has been deleted, does not cause a detectable shift (Fig. 2B).

The mobility shifts caused by DinR with each of the din promoters are abolished by the addition of excess unlabeled DNA containing any of the other din promoters. Fig. 2A shows mobility shift assays using the radiolabeled recA promoter fragment and subsaturating amounts of DinR. In the presence of a 20-fold molar excess of either dinB or dinC promoter DNA, the recA promoter fragment shift is not detectable; however, a 20-fold molar excess of the recA promoter region lacking the SOS box has no inhibitory effect.

The dinC promoter contains two SOS boxes that results in two shifted bands at sub-saturating concentrations of either crude extract or purified repressor (33). Fig. 2B shows that purified DinR causes two distinct bands at subsaturating concentrations that are visually indistinguishible from those produced by B. subtilis crude extract or repressor that has been affinity purified from B. subtilis; at higher concentrations of DinR, only the larger complex is formed.

We conducted DNase and hydroxyl radical footprinting analyses on the binding of DinR to the dinC operator (Fig. 3). The DNase results show protection of the two consensus sequences as well as hypersensitivity to DNase digestion at sites flanking these sequences. Hydroxyl radical footprinting indicates specific contacts along the DNA backbone; all of the sites protected by hydroxyl radical coincide with the DNase footprints.

To determine where DinR interacts with the DNA closely enough to preclude penetration by hydroxyl radicals, we mapped graphically the protected DNA sites on a DNA model using SYBYL software (Tripos Associates Inc.) on a Silicon Graphics Indigo 2 workstation. The modeling results indicate that all of the protected sites are on one face of the helix as shown for one of the two dinC boxes in Fig. 4. Inspection of the protected face reveals that the base pairs corresponding to the entire consensus site are accessible in two adjacent major grooves of the DNA.

To confirm that DinR protein acts as a transcriptional repressor we tested the ability of DinR to prevent binding of B. subtilis RNA polymerase to din promoters and inhibit their transcription in vitro. Fig. 5A shows a mobility shift assay using the recA promoter fragment and B. subtilis RNA polymerase in the absence and presence of DinR protein. The addition of RNA polymerase holoenzyme retards significantly the mobility of the recA promoter fragment. When a subsaturating amount of DinR is included, the amount of RNA polymerase bound to the DNA is reduced, and a shift corresponding to the DinR-DNA complex appears. Thus, DinR inhibits the binding of RNA polymerase to the promoter. The fact that no other shifts are visible indicates that the two proteins do not bind simultaneously to the promoter fragment. Consistent with its ability to inhibit binding of RNA polymerase to the recA promoter, DinR protein inhibits the in vitro transcription of DNA from the recA promoter (Fig. 5B).

Although the degree of sequence homology is limited, the LexA catalytic residues (Ser-119 and Lys-156) are conserved in DinR (Ser-127 and Lys-165), suggesting that the cleavage mechanism has been conserved. We conducted autodigestion assays with DinR over a range of pH values essentially as described by Little (18). Fig. 6 shows typical autodigestion assays in which DinR was incubated at 37 °C. The reaction is first order in repressor concentration, and the apparent rate constants are independent of initial repressor concentration. Amino acid sequence analyses of the cleaved protein fragments revealed the cleavage site to be the specific peptide bond between Ala-91 and Gly-92 (corresponding to the LexA site between Ala-84 and Gly-85).

Fig. 7 shows the pH dependence of the autodigestion reaction. As with E. coli LexA, the pH dependence suggests that the deprotonation of a single ionizable group is required for activity. Fitting the data to the equation for a single ionizable base gives values for the rate constant and the pK_a of the basic catalytic group, or Thus, DinR autodigestion appears to depend on a general base with a pK_a of about 10.3, and the reaction has a k_max of 1.8 × 10⁴ s¹. The corresponding values for LexA at 37 °C are 9.8 and 2.5 × 10³ s¹, respectively (19). It is noteworthy that the pK_a for  repressor autodigestion is also 9.8, while the rate is 40-fold slower than for LexA (19). Although we have not made direct comparisons of LexA and DinR autodigestion reactions under identical conditions, the 10-fold slower rate and the difference in pK_a values seem significant.

According to the E. coli model, RecA protein, when activated by an inducing signal, promotes the cleavage of LexA repressor at physiological pH. There is evidence that in both E. coli and B. subtilis, the inducing signal is single-stranded DNA generated by the processing of damaged DNA (12, 13, 28). Correspondingly, the RecA-mediated cleavage reaction in vitro requires that RecA be activated by binding single-stranded DNA and nucleoside triphosphate (9, 10, 25). Fig. 8 shows that B. subtilis RecA, in the presence of single-stranded DNA and either dATP or ATPS promotes the cleavage of DinR protein. Analyses of cleavage rates by densitometric scanning of stained gels gives first order rate constants of 2.3 × 10⁴ s¹ for ATPS and 2.0 × 10⁴ s¹ for dATP at a RecA concentration of 2 µM (Table I). As with the autodigestion reaction, these values are about 10-fold lower than the corresponding reaction with LexA and E. coli RecA (37). Moreover, the rate of B. subtilis RecA-mediated DinR cleavage is about 3-fold lower than the cleavage of LexA by B. subtilis RecA (25). Under otherwise identical conditions, there is no detectable cleavage when DinR is incubated without either RecA or a nucleoside triphosphate. There is also no detectable cleavage when RecA alone is incubated with DinR (data not shown). Surprisingly, we detected a low level of cleavage activity (k = 6.3 × 10⁵ s¹) when RecA and dATP are incubated with DinR in the absence of single-stranded DNA.

Table I. First order rate constantsa for cleavage of B. subtilis DinR and E. coli LexA at 37 °C High pH E. coli RecA + ATPS B. subtilis RecA + ATPS B. subtilis RecA + dATP LexA 2.5  × 103b 3.1  × 103c 6.9  × 103d 6.1  × 103d DinR 1.8  × 104 NDe 2.3  × 104 2.0  × 104 a  All rate constants are relative to repressor concentration and given in units of s1. b  Taken from Ref. 19. c  Value for 1 µM RecA taken from Ref. 37. d  Values for 0.8 µM RecA taken from Ref. 25. e  ND, not determined.

We used Western analysis to examine the level of DinR over time in cells treated with mitomycin C at 1.0 µg/ml. Fig. 9 shows the results of an experiment in which equivalent amounts of cells were removed at the indicated time intervals after DNA-damaging treatment and subjected to Western analysis with DinR and LexA antisera (which do not show any detectable crossreactivity between the two proteins). In Fig. 9, the left-hand set corresponds to wild-type (YB886) cells, and the right-hand set corresponds to YB886 cells containing a lexA plasmid (28). In both cases, the level of DinR begins to decrease after a 5-min lag and does not return to its original level for at least 90 min; we were unable to detect any cleavage products by Western analysis of these crude extracts. Although we have preliminary evidence that the initial concentrations of DinR and LexA are about the same, direct comparisons of band intensities are not useful because the signal produced with LexA antisera is stronger than that with DinR antisera. We have not yet conducted a thorough quantitative kinetic analysis; however, qualitative comparisons between the rates of DinR and LexA disappearance in the same cells are certainly reasonable. By visual inspection of band intensities, the rate of LexA cleavage in B. subtilis cells appears to be faster than the rate of DinR cleavage between 5 and 15 min. By contrast, the rate of DinR disappearance is apparently greater than the rate for LexA disappearance between 15 and 30 min. This analysis may, however, be complicated by the fact that DinR expression is controlled by its own promoter and subject to autoregulation; LexA synthesis is driven by the cat promoter and not subject to autoregulation (28).

DISCUSSION

We have shown that the dinR gene codes for the B. subtilis SOS repressor and that the DinR protein is the functional analogue of E. coli LexA in every respect. This report presents definitive evidence that the SOS repressor has been functionally conserved in B. subtilis. Although lexA-like genes have been characterized from a variety of enterobacteria (37, 38), this is the first characterization of a LexA-like protein whose structure has diverged significantly from the E. coli protein. As such, this is an important step in the continued characterization of an SOS DNA repair system from a phylogenetically distant organism.

The B. subtilis SOS box does not resemble the E. coli SOS box in either sequence or spacing between the two half sites (by contrast, the SOS boxes identified in other bacterial species are identical to the E. coli consensus site (38)). Although we do not yet have any definitive evidence for the stoichiometry of DinR binding to the SOS operator, the palindromic nature of the site suggests that at least two monomers bind to each 8-base pair site. Since LexA binds its operator site as a dimer (8, 16), it is reasonable to assume that DinR also binds as a dimer. Analyses of DinR binding to various din operators reveal sigmoidal binding curves (indicative of highly cooperative binding), suggesting that the protein binds as a multimer.² Moreover, the extent of retardation in mobility shift experiments is similar to that with E. coli LexA and operator fragments of similar sizes suggesting that, like LexA, DinR binds its operator as a dimer (16, 33). However, we have not ruled out the possibility that it binds as a larger complex.

The rate constants determined for DinR cleavage at 37 °C are about the same for reactions promoted either by B. subtilis RecA or by alkaline pH and, in each case, about 10-fold lower than the corresponding rate constants for LexA cleavage (19, 39). This suggests that the rate of RecA-promoted cleavage (in both E. coli and B. subtilis) is primarily dependent on the intrinsic rate constant associated with autodigestion and not influenced significantly by the specific interaction between RecA and repressor (although it is possible that the in vitro reaction conditions are not optimal for DinR cleavage). On the other hand, the rate of E. coli LexA cleavage by B. subtilis RecA is about 3-fold slower than the rate promoted by E. coli RecA (25, 39), indicating that the interaction between RecA and repressor is not unimportant in determining the cleavage rate. It is possible that the reduced rate of LexA cleavage with B. subtilis RecA simply reflects a lower affinity for the foreign repressor; alternatively, the foreign RecA may not be as effective in stabilizing the catalytically active repressor conformation.

The reduced rate of SOS repressor cleavage in B. subtilis relative to E. coli is consistent with studies showing that the rates of repressor inactivation and SOS gene induction are slower in B. subtilis (26, 33) than in E. coli (13). Although we have not yet determined the in vivo cleavage rate, we conducted a qualitative comparison between the rates of DinR and LexA disappearance in the same cells. After a short lag period, the rate of LexA cleavage in B. subtilis cells (which is slower than the cleavage rate in induced E. coli cells (13)) appears to be faster than the rate of DinR cleavage between 5 and 15 min, whereas the rate of DinR disappearance is greater between 15 and 30 min. Although a more quantitative analysis is needed, a reasonable interpretation of this discrepancy is that B. subtilis RecA has a higher affinity for the B. subtilis repressor than for the E. coli repressor. In any case, our results are consistent with previous studies indicating that the time course for SOS induction in B. subtilis is considerably slower than it is in E. coli. Moreover, the evidence presented here suggests that differences in the rate of SOS induction may ultimately be due to a lower rate constant for repressor autodigestion.

A significant physiological difference between E. coli and B. subtilis makes the investigation of SOS regulation in B. subtilis particularly interesting. B. subtilis naturally differentiates to a state of competence, which is characterized by the ability of the cell to bind and take up exogenous DNA (42); incoming DNA can be integrated into the host chromosome via general genetic recombination, which presumably requires the DNA strand exchange activity of the RecA protein (25). Corresponding to this requirement, competence development in B. subtilis is accompanied by the induction of the recA gene as well as other SOS genes (43, 44). However, the induction of the recA gene in recA competent cells indicates that RecA is induced by a competence-specific mechanism that does not require activated RecA protein; by contrast, induction of other SOS genes during competence development requires RecA (43). In light of the results presented here, we infer that RecA can be induced in competent cells despite an inability to inactivate the DinR repressor. Since it has been shown that the recA gene is transcribed in recA competent cells from the same promoter used during DNA damage induction (45), recA induction during competence development must involve the displacement of DinR repressor from the recA operator. The likely candidate for this is the ComK protein, a transcriptional activator induced during competence development that binds to a site overlapping the recA operator in vitro (42, 46). If this is the case, we predict that ComK should be able to displace DinR from the recA promoter, but not other din promoters.

How are the other B. subtilis SOS genes induced during competence development in the absence of an inducing signal generated by DNA damage? Since RecA is required for induction, the simplest explanation is that RecA causes DinR cleavage. It has been suggested (43) that RecA could be activated in competent cells by single-stranded gaps present in chromosomal DNA (47) or by incoming donor DNA molecules that are single-stranded (42). Alternatively, our detection of DinR cleavage activity with dATP, in the absence of single-stranded DNA, suggests that B. subtilis RecA may be activated for DinR cleavage in competent cells by an altogether different metabolic signal.