INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Based upon sequence comparisons, it has been hypothesized that the Bacillus subtilis protein DinR is the functional homolog of the Escherichia coli SOS transcriptional repressor, LexA (20, 21). Indeed, recently published data has firmly established DinR as the transcriptional repressor of the SOS system in B. subtilis (5, 16, 27). Although it is only 34% identical to LexA, DinR demonstrates many biochemical and physical properties that are reminiscent of LexA. For example, like that of LexA, the deduced amino acid sequence of DinR predicts two distinct domains within the protein. DinR has most homology to LexA and other LexA-like proteins in its carboxyl-terminal domain (10, 27). This C-terminal domain is thought to be primarily responsible for the cooperative dimerization of the normally monomeric LexA protein at its target site in DNA (8, 22, 23, 25). The C-terminal domain also contains a conserved Ala-Gly cleavage site as well as the appropriately spaced serine and lysine residues that have been identified as critical for autodigestion (14). Indeed, like LexA, DinR undergoes RecA-independent autocatalysis at alkaline pH and RecA-mediated autocatalysis under more physiological conditions (16, 27). Such cleavage inactivates repressor function, thereby allowing DinR-regulated genes to be expressed.

Despite the notion that DinR displays transcriptional repressor activity that is comparable to that of LexA (16), there is in fact little homology between the amino-terminal DNA binding domains of the two proteins (10, 27). In addition to the obvious lack of primary sequence homology, the typical repressor-like, helix-turn-helix motif present in LexA is not immediately obvious in DinR. This disparity coincides with the appearance of completely distinct DNA binding sequences in the two repressors. In E. coli and many other gram-negative organisms, the SOS box is a region of 16 bp that displays dyad symmetry [5'-CTGT-(AT)₄-ACAG-3'] (13, 26). In several gram-positive bacteria (e.g., B. subtilis and Mycobacteria sp.) (2, 4, 17, 19), the binding site for DinR is thought to be the previously described Cheo box, a region of 12 bp with dyad symmetry (5'-GAAC-N₄-GTTC-3') but no homology to the gram-negative SOS box.

It has recently been suggested that the B. subtilis DinR protein should be renamed LexA (16). Given the huge differences in the recognition sites between the E. coli LexA protein and the gram-positive DinR-like proteins, however (4, 17, 19, 27; see below), we propose retaining the descriptive name DinR (damage inducible repressor) rather than renaming the protein LexA (originally defined as locus for X-ray sensitivity A [7]) and using the term DinR box to describe the binding site for DinR to avoid confusion between it and the commonly referred to SOS box of E. coli.

We have previously purified the B. subtilis DinR protein to homogeneity (27) and shown that it does bind to the proposed DinR binding site in the B. subtilis recA promoter region but does not bind to certain mutant sequences located within the previously identified Cheo box. The availability of the highly purified B. subtilis DinR protein has enabled us to extend these studies and perform a detailed molecular analysis of the B. subtilis DinR box. Indeed, by using a combination of gel electrophoretic mobility shift assays, hydroxyl radical footprint protection assays, and recA-lacZ transcriptional fusions, we have determined that certain bases within the previously defined Cheo box are more critical for binding than others. This data, together with computational analyses of potential binding sites in other gram-positive organisms, allows us to propose a new consensus DinR box, 5'-CGAACRNRYGTTYC-3'.

(This research was conducted by K. Winterling and J. Sun in partial fulfillment of the requirements for a Ph.D.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. The B. subtilis strain used in this study, YB886A, serves as a wild-type strain and is cured of all known prophages. E. coli DH5 (GIBCO-Life Technologies, Gaithersburg, Md.) and GBE180 (DH5 pcnB1) were used for routine cloning and maintenance of plasmids (27).

Media and growth conditions. B. subtilis strains were maintained on tryptose blood agar base medium, and liquid cultures were grown in antibiotic medium 3 (Difco Laboratories, Detroit, Mich.) or nutrient broth (Oxoid Ltd., Basingstoke, United Kingdom) with aeration at 37°C. E. coli strains were grown on Luria-Bertani agar or in Luria-Bertani broth. Ampicillin (50 µg/ml), chloramphenicol (20 µg/ml), kanamycin (30 µg/ml), and isopropyl--D-thiogalactopyranoside (IPTG) (1 mM) (Gold Biotechnology, Inc., St. Louis, Mo.) were added as required.

-Galactosidase assays. DNA damage-inducible promoter activity of the recA-lacZ and din-lacZ fusions was examined by measuring -galactosidase activity as previously described (27). Briefly, B. subtilis cultures were grown with aeration at 37°C in nutrient broth supplemented with 0.1% yeast extract. Cultures were grown to early exponential phase, when an aliquot was removed and the culture was divided in half. Mitomycin (0.5 µg/ml) was added to one-half of the culture, and 1-ml aliquots were taken from the induced and uninduced cultures after 90 min of additional incubation. After the absorbance (optical density at 595 nm) of each sample was spectrophotometrically measured, the cells were harvested and washed in 0.5 ml of 25 mM Tris-HCl (pH 7.4). The supernatant was decanted, and the pellet was placed in a dry ice-ethanol bath. The pellet was subsequently resuspended in 0.64 ml of Z buffer (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, and 50 mM -mercaptoethanol [pH 7.0]) (15), 0.16 ml of a lysozyme solution (2.5 mg/ml in Z buffer) was added, and the sample was incubated at 37°C for 5 min. Eight microliters of 10% Triton X-100 was added, and the samples were warmed to 30°C. -Galactosidase activity was calculated by the method of Miller (15).

Electrophoretic mobility shift assays. The exact location of each DinR box relative to the 35 and 10 promoter elements as well as the translational start site of each gene is shown in Fig. 1. The wild-type and mutant recA DinR boxes used in this assay were gel-purified synthetic oligonucleotides that varied in length from 60 to 67 bp. Complementary single-stranded oligonucleotides were annealed together by mixing equimolar amounts of each oligonucleotide, thus producing short regions of double-stranded DNA (27). In addition, oligonucleotides were similarly synthesized, purified, and annealed so that they corresponded to the wild-type promoter sequences of dinA and dinB. All of these probes were designed so that the DinR binding site was centered, with approximately 26 bp of wild-type sequence flanking each side.

View larger version (33K): [in this window] [in a new window]   FIG. 1.   Location of DinR boxes in the promoter regions of the B. subtilis recA, dinC, and dinR genes. The respective 35 and 10 promoter elements for each gene are in slightly larger, bold-faced letters. The DinR boxes are double underlined. The recA-DinR box is centered at 51 relative to the 35 promoter. The two dinC-DinR boxes are located at 24 and 53 relative to the 35 and 10 elements. The three dinR-DinR boxes are located at 39, 67, and 104 relative to the 35 promoter.

Determination of the oligomeric state of DinR in solution and DinR bound to target DNA. To determine the oligomeric state of DinR in solution, we compared its relative sedimentation in a glycerol gradient to that of the E. coli LexA protein. Highly purified DinR and LexA proteins (5 µg each) were loaded onto separate 5 to 30% linear glycerol gradients in buffer D [10 mM piperazine-N,N'-bis(2-ethanesulfonic acid) (PIPES)-NaOH (pH 7.0), 0.1 mM EDTA, 10% (vol/vol) glycerol, 200 mM NaCl]. Each gradient also contained 5 µg of BSA, 5 µg of ovalbumin, and 2 µg of cytochrome c as internal molecular weight standards. Ultracentrifugation was carried out for 26 h at 49,000 rpm with a SW60Ti rotor. Fractions (100 µl each) were collected, and proteins were separated in sodium dodecyl sulfate-polyacrylamide gel electrophoresis gradient gels containing 9 to 19% polyacrylamide. Proteins were subsequently visualized by staining the gel with silver.

Labeled DNA for hydroxyl radical footprint analysis. Approximately 15 µg of PCR primer, DINR01 (5'-GCGAAGCTTCTCATGATCATAACCTC-CAAC-3'), RECA01 (5'-GCGAAGCTTACATGATTTTCTGATACATTA-3'), or RECA02 (5'-CGCGAATTCCTTTTATGTTACACTACATA-3'), was 5' radiolabeled in separate reactions with T4 polynucleotide kinase. Briefly, 10 U of T4 polynucleotide kinase (New England Biolabs) was used to singly radioactively label each PCR primer in a 20-µl reaction mixture containing 15 µg of PCR primer (DINR01, RECA01, or RECA02), 1× T4 PNK reaction buffer (70 mM Tris · HCl [pH 7.6], 10 mM MgCl₂, 5 mM dithiothreitol), and 50 µCi of [-³²P]dATP at 6,000 Ci/mmol for 30 to 60 min at 37°C. Radioactive primers were used in standard PCRs to obtain the dinR and recA promoter sequences in 100-µl reaction mixtures containing 10 µl of labeled primer (from the labeling reaction), 1× Vent polymerase buffer, 6 µg of complementary unlabeled PCR primer, 270 ng of B. subtilis YB886A chromosomal DNA, 10 mM each deoxynucleoside triphosphate, and 2 U of Vent DNA polymerase (New England Biolabs). PCR DNA was precipitated and separated on a native 8% polyacrylamide gel made with 1× Tris-borate-EDTA. After separation, the wet gel was exposed to autoradiography film to identify the radioactive PCR product. The radioactive dinR and recA DNAs were gel purified by soaking a crushed gel slice in 700 µl of TE (10 mM Tris-Cl [pH 8.0], 1 mM EDTA) buffer overnight. The radioactive DNAs were filtered through series 8000 microcentrifuge filtration devices (Lida Manufacturing Corporation), precipitated, dissolved in TE buffer, and stored at 3,000 to 5,000 cpm/µl.

Hydroxyl radical footprint analysis. Equal volumes of labeled DNA (approximately 60,000 cpm) and DinR (diluted into glycerol-free 2× binding buffer [27]) were incubated at room temperature for 25 min. Labeled DNA from dinR-DinR or recA-DinR complexes was gel purified in the same manner as the labeled PCR products described above. In a typical 40-µl reaction mixture, 2 µl of 1 mM Fe-EDTA (50 µM final) and 2 µl of 20 mM sodium ascorbate (1 mM final) were pipetted onto the side of the reaction tube. Hydroxyl radical cleavage was initiated by adding 2 µl of a 1:250 dilution of 30% hydrogen peroxide solution into the existing drop (0.0075% final) and the DNA solution for 2.5 min. Cleavage was stopped by adding 1/10 volume of stop solution containing 50% glycerol and 10 mM EDTA. Protein-DNA complexes were separated by loading the cleavage reaction mixtures immediately onto a native 5% polyacrylamide-0.5× Tris-borate-EDTA gel.

G-specific reaction. In a typical 200-µl reaction mixture, approximately 20,000 cpm of singly labeled DNA was added to 20 µl of 10× G-specific reaction buffer (0.5 M sodium cacodylate-10 mM EDTA) and 160 µl of water. One microliter of straight dimethyl sulfate was added to the tube. The reaction mixture was mixed immediately and incubated for 1 min before being briefly spun in a microcentrifuge. Fifty microliters of stop solution was added (1.5 M sodium acetate, 1 M -mercaptoethanol, 0.004 µg of sonicated calf thymus DNA per ml), and the DNA was precipitated and dissolved in 90 µl of TE buffer. Ten microliters of piperidine was added and incubated at 90°C for 30 min. The DNA was dried to completion in a speed-vac concentrator. The dried DNA was dissolved twice in 20 µl of water and redried. The DNAs were finally dissolved in 40 to 50 µl of TE buffer and stored at 4°C.

Sequencing gel analysis of hydroxyl radical footprint. Approximately 5,000 cpm from each sample was placed into separate Eppendorf tubes and dried completely in a speed-vac concentrator. Dried DNAs were dissolved in 4 µl of formamide loading buffer (100% formamide, xylene cylanol, bromophenol blue) and heated to 95°C for 2 min. Samples were immediately placed on ice and loaded onto a 6% polyacrylamide-8 M urea sequencing gel. After analysis, gels were dried and exposed to autoradiography film or to a Molecular Dynamics PhosphorImager screen.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Regulation of damage-inducible genes in B. subtilis. All living organisms are exposed to a variety of synthetic and natural DNA-damaging agents. The differential regulation of a response for coping with such damage would allow cells to respond to the extent of DNA damage by inducing only proteins that are required to efficiently repair all of the damaged DNA. In E. coli, differential regulation of SOS genes occurs and is achieved, at least in part, by variation of the affinity of the transcriptional repressor, LexA, for its binding site (reference 13 and references therein). Analysis of the levels of -galactosidase produced by various din-lacZ fusions revealed that the basal level of B. subtilis din expression also varies considerably (Fig. 2). Obviously, such differences in expression could, in theory, be achieved by differential promoter activity or by alteration of the affinity of the repressor for its binding site. While previous studies have demonstrated that DinR binds to the Cheo box upstream of recA, dinB, dinC, and dinR (5, 16, 27), the relative affinity of DinR for each site is largely unknown. Interestingly, gel electrophoretic mobility shift assays of the dinA and dinB DinR boxes (Fig. 3) revealed that DinR binds to both boxes with roughly the same affinity and that these affinities are qualitatively similar to those of the recA-DinR box (27) and the dinR-DinR boxes (see below). Such observations suggest, therefore, that at least for the din genes assayed here, differential expression is achieved via differential promoter activities rather than differential operator affinities. This conclusion is also supported by our observation that all of the fusions were induced to the same extent (fourfold) under mild inducing conditions. If regulation was primarily at the level of operator binding, one might have expected the din-lacZ fusions to exhibit much more variable induction ratios.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Expression of B. subtilis dinA-lacZ, dinB-lacZ, dinC-lacZ, dinR-lacZ, and recA-lacZ transcriptional fusions. The various constructs were present in a single copy due to integration at the amyE locus. The SOS regulon was induced, where noted, by the addition of 0.5 µg of mitomycin per ml, cultures were harvested 90 min later, and the level of -galactosidase activity was determined. The data for the dinR-lacZ fusion was taken from the study of Haijema et al. (5), and that for the recA-lacZ fusion, taken from the study of Winterling et al. (27), was used for comparison.

View larger version (29K): [in this window] [in a new window]   FIG. 3.   Binding of DinR to the DinR box located upstream of dinA and dinB. Radiolabeled dinA and dinB promoter fragments (3.0 ng or ~3.4 nM) were incubated with various concentrations of DinR (nanomolar monomer) as indicated at the top of each lane. Reactions were performed at room temperature for 25 min. Protein-DNA complexes were separated in native polyacrylamide gels (5% acrylamide) and visualized by exposure to X-ray film.

Binding of DinR protein to the two DinR boxes upstream of dinC. Unlike many din genes, dinC contains two DinR boxes, located close to its promoter. The location of each of these sites makes them potential transcriptional repressor binding sites. (As noted above, for the sake of simplicity, we denoted these sites the 53 box and the 24 box (with respect to their locations near the 35 and 10 promoter elements, respectively) (Fig. 1). Indeed, DNase I and hydroxy radical footprinting experiments have shown that both sites are apparently protected in the presence of DinR (16). The relative affinity of DinR for each DinR box is, however, unknown. To determine the affinity of DinR for each of these sites, DNA mobility gel shift assays were performed with a probe containing only the 53 site, a probe containing only the 24 site, and a probe that contained both the 53 and the 24 sites (Fig. 4A). Experiments revealed that under these assay conditions, DinR binds specifically to each of the three probes (Fig. 4B). Visual inspection of the shifted complexes suggests, however, that there is no significant difference in the affinity for one probe over another. With the probe containing both DinR boxes, there was, however, a second, larger shifted complex detectable at higher concentrations of DinR. We suggest, therefore, that each DinR box serves equally well as a potential binding site but that at higher cellular concentrations there is a greater likelihood of DinR occupancy at both sites.

View larger version (23K): [in this window] [in a new window]   FIG. 4.   Binding of DinR to the two DinR boxes located upstream of dinC. (A) Nucleotide sequence of the region upstream of dinC. The two previously identified Cheo boxes are indicated in bold-faced type, and the positions of the three probes used in the in vitro binding studies are indicated below the sequence as P1 (24 DinR box), P2 (53 DinR box), and P3 (both DinR boxes). (B) The individually radiolabeled fragments (P1, P2, or P3; 3.0 ng or ~3.4 nM) were incubated with various concentrations of DinR (nanomolar monomer) as indicated at the top of each lane. Reactions were performed at room temperature for 25 min. Protein-DNA complexes were separated in native polyacrylamide gels (5% acrylamide) and visualized by exposure to X-ray film.

Self-regulation of dinR. E. coli LexA not only serves as the repressor of a number of unlinked genes in the SOS regulon but also acts as a negative regulator of its own synthesis. We presume, therefore, that like LexA, DinR regulates its own expression. Indeed, gel mobility shift assays with crude cell extracts suggest that this may be the case (5). Although dinR contains three potential binding sites, Haijema et al. have demonstrated that the DinR box, which we denote the 104 box, apparently plays no role in regulating DinR expression (5). The two remaining sites are located at 67 and 39, respectively (Fig. 1 and 5A). These sites have identical core Cheo box sequences and differ from each other only in the N₄ region. More importantly, however, they diverge from the consensus sequence, having a 3' T instead of C (GTTCGTTT). Thus, we were interested in determining if DinR bound preferentially to one site or equally well to both sites (as is the case for the two dinC DinR boxes). Indeed, the DNA mobility shift assays revealed that DinR binds specifically to all three probes (Fig. 5B), and analysis of the shifted complexes revealed that the affinities of DinR for the 39 and the 67 binding sites are qualitatively similar. As in the case of dinC, however, a second shifted complex is clearly detectable when DinR is incubated with the probe containing both putative binding sites, suggesting that both DinR binding sites can be occupied if there is enough DinR.

View larger version (25K): [in this window] [in a new window]   FIG. 5.   Binding of DinR to the two DinR boxes located upstream of dinR. (A) Nucleotide sequence of the region upstream of dinR. The two previously identified Cheo boxes are indicated in bold-faced type, and the positions of the three probes used in the in vitro binding studies are indicated below the sequence as P1 (39 DinR box), P2 (67 DinR box), and P3 (both DinR boxes). (B) The individually radiolabeled fragments (P1, P2, or P3; 3.0 ng or ~3.4 nM) were incubated with various concentrations of DinR (nanomolar monomer) as indicated at the top of each lane. Reactions were performed at room temperature for 25 min. Protein-DNA complexes were separated in native polyacrylamide gels (6% acrylamide) and visualized by exposure to X-ray film.

Detailed molecular analysis of the DinR box located upstream of recA. Based upon previous studies (2, 4, 16, 17, 19, 27), DinR or a DinR-like protein unequivocally recognizes and binds to specific sequences located upstream of a number of din genes. In the case of B. subtilis recA, this site is located upstream of the 35 promoter element (approximately at 51) (Fig. 1). By performing detailed mutational studies on this DinR binding site, we have been able to determine which of the bases within the formerly identified Cheo box are important for regulated expression of RecA (Fig. 6). As expected, certain base changes within the core Cheo box resulted in increased basal expression of recA-lacZ transcription in the absence of exogenous DNA damage. It seems very unlikely that these changes affect promoter activity per se, given the location of the DinR box upstream of the 35 region, and we interpret the data as reflecting specific changes in operator affinities. In general, lowest expression was achieved when the sequence matched the consensus Cheo box (2). In some instances, changes in the sequence resulted in constitutive expression (i.e., the most-5' G of the Cheo box to either A, T, or C), while certain other changes were accommodated (i.e., the CT transition in the 5' GAAC side of the Cheo box) (Fig. 6). Interestingly, substitutions in the two outermost nucleotides in the N4 region also appear to affect recA-lacZ expression, whereas those in the very middle of this region appear to have little effect (Fig. 6).

View larger version (47K): [in this window] [in a new window]   FIG. 6.   Expression of B. subtilis recA-lacZ transcriptional fusions. The effects of single base pair changes within the B. subtilis recA DinR box were determined by quantitating -galactosidase activities generated from recA-lacZ transcriptional fusions expressed in B. subtilis. The histogram represents the level of -galactosidase activity of each individual mutant construct. The effects of substitutions at each nucleotide in the DinR box were analyzed and are depicted as follows: G, ; A, &atyp0220;; T, ; C, . The consensus sequence that resulted in equal to or less than 17 Miller units of -galactosidase activity from the recA-lacZ fusion is given below the histogram. For comparison, the wild-type sequence for the recA DinR box is also listed.

Hydroxyl radical footprinting of the B. subtilis recA and dinR DinR boxes. Given that we and others (4, 27) have demonstrated that nucleotide changes in the core Cheo box abolish binding of DinR or a DinR-like protein, as measured by gel mobility shift assays, the assumption that the binding occurs at the Cheo box seems more than reasonable. Unfortunately, the results of gel mobility shift assays can often be misleading when binding is weak (see reference 1 for a detailed discussion). As a consequence, binding often needs to be analyzed by alternative methods, such as DNA-footprint analysis. Indeed, Miller et al. (16) and Durbach et al. (4) have used such an approach to show that the core Cheo box is protected from DNase I and hydroxyl radical cleavage. Nevertheless, we were encouraged to perform additional footprint analyses to firmly establish that DinR does bind to the DinR box proposed in this study. To this end, we performed hydroxyl radical protection assays in the presence and absence of DinR with the B. subtilis recA and dinR promoter/operator sequences (Fig. 7 and 8). As noted previously, DinR binds specifically to the recA promoter/operator region to form a single major protein DNA species as evidenced by nondenaturing gel mobility shift (27). When this species is analyzed by hydroxyl radical footprinting, a pattern of protection is observed which is consistent with a single high-affinity binding site. A single region of protection is observed on each DNA strand. Within each region, a stretch of 3 to 4 bases is most strongly protected by DinR. These strongly protected stretches are each flanked by two regions of weaker protection (Fig. 7A). When the protection pattern is plotted on a linear representation of the recA sequence (Fig. 7B), it is clear that these stretches of protection are separated by about 10 bp and staggered on opposite strands by about 2 to 4 bp, with a total of 30 to 31 bp being protected from cleavage. These results indicate a cross-groove pattern of protection most likely caused by a protein lying along one side of the DNA helix. In support of this interpretation, a plot of protected positions on a three-dimensional representation of the DNA double helix reveals that all protected sites lie along one face of the DNA (Fig. 7C), exactly as has been found for footprints of other bacterial repressor-DNA complexes (6). Furthermore, and perhaps more importantly, the center of the footprint coincides exactly with the center of the proposed DinR box (CGAATATGCGTTCG) found at this position.

View larger version (49K): [in this window] [in a new window]   View larger version (22K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   FIG. 7.   Footprinting analysis of DinR bound to the recA operator. (A) The coding and noncoding strands of a radiolabeled recA promoter fragment were incubated with increasing concentrations of purified DinR and subjected to hydroxyl radical cleavage (lanes 4 and 5 and 9 and 10, respectively). Labeled DNA was also cleaved in the absence of DinR (lanes 3 and 8). Each experiment was run alongside naked DNA (Control) and a G sequencing reaction (G-Rxn). The boxed regions denote the location of the DinR box on each strand. The 51 site is representative of the most intense region of protection as well as the center of each DinR box. (B) Nucleotide sequence of the recA operator and regions protected from hydroxyl radical cleavage. The sizes of the bars correspond to the intensities of protection. The center of the protected region is represented by a shaded circle that corresponds to the center of the DinR box. (C) Three-dimensional representation of the bases within the recA operator that are protected by DinR during hydroxyl radical cleavage, indicating that the protected sites lie on one face of the DNA.

View larger version (26K): [in this window] [in a new window]   FIG. 8.   Footprinting analysis of DinR bound to the dinR operator. The coding strand of a radiolabeled dinR promoter fragment was incubated with increasing concentrations of purified DinR and subjected to hydroxyl radical cleavage (lanes 4 and 5). Labeled DNA was also cleaved in the absence of DinR (lane 3). The experiment was run alongside naked DNA (Control) and a G sequencing reaction (G-Rxn). The boxed regions denote the locations of the DinR boxes. The 39, 67, and 104 sites are representative of the most intense regions of protection as well as the centers of the respective DinR boxes.

Determination of the oligomeric state of DinR. Sequence analysis of the DinR box shows two regions of dyad symmetry. This configuration appears to represent two potential half sites that could theoretically each be bound by a DinR monomer. LexA has previously been shown to exist primarily as a monomer in solution (22, 23) and dimerizes upon binding to the SOS box (8).

View larger version (54K): [in this window] [in a new window]   FIG. 9.   Determination of the oligomeric state of DinR in solution. A mixture of proteins containing DinR or LexA and standards was sedimented on a 5 to 30% glycerol gradient. Fractions (100 µl each) were collected, and proteins were separated by electrophoresis on 9 to 19% gradient polyacrylamide gels. Proteins were visualized by silver staining the gels. Every other fraction is shown sequentially, starting with the top fraction (more slowly sedimenting, smaller proteins) at the far left. (A) BSA (5 µg), ovalbumin (5 µg), LexA (5 µg), and cytochrome c (2 µg). (B) BSA (5 µg), ovalbumin (5 µg), DinR (5 µg), and cytochrome c (2 µg). Since LexA and DinR sediment identically, we assume that DinR, like LexA, is predominantly a monomer in solution.

View larger version (12K): [in this window] [in a new window]   FIG. 10.   Determination of the oligomeric state and molecular weight of DinR bound to the dinR and recA operators. The retardation coefficient (Kr) of each protein standard was plotted against the molecular weight of the protein to generate the standard curve. The standard curve and the Krs for DinR (A) and the DinR-DNA complexes (B) were used to determine their respective molecular weights. (A) Molecular weight standards: urease trimer, BSA dimer, BSA monomer, ovalbumin, carbonic anhydrase, and -lactalbumin (); DinR monomer (); and DinR dimer (). (B) Molecular weight standards: urease trimer, BSA dimer, BSA monomer, ovalbumin, carbonic anhydrase, and -lactalbumin (); DNA probe, (); DinR-dinR (); and DinR-recA (). Based upon these observations, we conclude that DinR binds to its operator sequence in a dimeric state (see Results for detailed description of calculations).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

A new consensus binding site for DinR. Based upon the combined transcriptional fusion studies, gel mobility shift assays, and hydroxyl radical footprinting, we have derived a new consensus binding site for DinR. Upon reanalysis of the previously identified B. subtilis din genes and those of (presumably damage inducible) recA and lexA/dinR from a variety of gram-positive bacterial species, we find that the 5' residue of the DinR box is generally cytosine and the 3' residue is generally guanine (Table 1). In addition, the N₄ region does not appear to accommodate all nucleotides and retain equal DinR binding efficiency (Fig. 6) (4, 27). Using such an approach, we derived a consensus DinR binding site of 5'-CGAACRNRYGTTCG-3'. Footprint analysis of several din genes reveals that this sequence is centrally located within a DNA region of 31 bp that is protected by DinR from hydroxyl radical cleavage (Fig. 7 and 8) (4), definitive evidence that this sequence is the one that is recognized and bound by DinR. Furthermore, we found that like E. coli LexA (8, 22, 23, 25), DinR is predominantly monomeric in solution but binds to its target DNA as a dimer (Fig. 10), presumably because each monomer binds cooperatively to one half site.

Affinity of DinR for each DinR box. One interesting feature of this study is our observation that all the DinR operator sequences appear to bind DinR protein with roughly equal affinities. A simple explanation is that the gel mobility shift assay used to quantitate binding is not sensitive enough to identify possible differences (1). Indeed, more quantitative experiments currently in progress will allow us to determine the dissociation constant of DinR for each DinR box. A finding of an equal affinity of DinR for each DinR box would contrast dramatically with the finding that E. coli uses differential binding of LexA to its SOS box to regulate damage-inducible gene expression. Unlike E. coli, however, in which SOS regulation appears to be straightforward, B. subtilis has at least four different modes of SOS induction (28), and additional factors, such as activator proteins (21), seem likely to provide the ancillary functions to induce the regulon under various environmental and developmental stimuli. The availability of an inducible SOS response is important and has been conserved in both gram-positive and gram-negative organisms, but the process of evolution has allowed the development of divergent regulatory mechanisms.