Expression of the Bacillus subtilis dinR and recA genes after DNA damage and during competence

The Bacillus subtilis dinR gene product is homologous to the LexA protein of Escherichia coli and regulates the expression of dinR and dinC. Using transcriptional fusions in the dinR and the recA genes, we have investigated the epistatic relationship between these two genes during the SOS response induced either by DNA damage or by competence. The results show that after DNA damage, induction of the expression of both recA and dinR is dependent on the activity of the DinR and RecA proteins. A RecA-dependent activity on DinR is proposed as the initial event in the induction of the SOS network. In contrast, the competence-related induction of dinR and recA appears to involve two distinct mechanisms. While one mechanism corresponds to the classical regulation of the SOS response, the other appears to involve an activating factor. Moreover, this factor is active in cells in which competence is prevented by a mutation in the regulatory gene comA.

In Bacillus subtilis, an SOS system is induced when cells are exposed to conditions that damage DNA or undergo the competence state. Several phenomena that occur during the SOS response caused by DNA damage have been described elsewhere: overexpression of the din and recA genes (13), induction of DNA damage reparation (12), increase in the rate of mutagenesis (26), Weigle reactivation of prophages (5), induction of the lytic cycle of prophages (24), induction of the SP1 DNA methylase (25), and development of cell filamentation (12). To date, three damage-inducible genes (din) have been identified: dinA (uvrA), dinB, and dinC (10,11), as well as a uvrB gene (2,18). Recently, a new din gene, dinR, was identified for B. subtilis. The dinRl mutant was found to be deficient in both recombination and DNA repair. The DinR protein shares significant amino acid sequence similarities (47.3%) with the LexA protein of Escherichia coli (19). LexA is the repressor of the SOS gene network, and like LexA, DinR appears to be a regulator of the expression of other din genes (19,22). In E. coli, induction of the SOS system results from the reversible activation of RecA, leading to the cleavage of the repressor of the SOS genes (22). The amino acid sequences in the three regions known to be required for the cleavage of E. coli LexA are highly conserved in DinR, suggesting that DinR could undergo similar RecA-mediated cleavage. E. coli and B. subtilis RecA proteins have several regions of similarity (21), and B. subtilis RecA is immunoreactive with antibodies raised against E. coli RecA. Furthermore, the B. subtilis RecA protein can facilitate E. coli LexA cleavage in vitro (14), and the E. coli RecA protein expressed on a plasmid in B. subtilis complements, at least partially, the recA4 (previously recE4) mutant for recombination, induction of din genes, and Weigle reactivation (13,16). However, the E. coli RecA protein is unable to restore prophage induction in the recA4 mutant of B. subtilis (13), and the B. subtilis RecA cannot promote the cleavage of the XCI repressor molecule in vitro (14).
An additional regulation of the SOS system occurs when B. subtilis cells develop competence, a physiological state in * Corresponding author. which the cells are able to bind and take up exogenous DNA.
Competence is expressed after exponential growth and is subject to three types of regulation: nutritional, growth-stage specific, and cell-type specific (for a review, see reference 4). Several genes that regulate the development of competence have been identified. The comA and comP genes belong to the family of two-component regulators (23) and encode a response regulator and a histidine kinase protein, respectively. The regulatory com genes are expressed throughout the B. subtilis growth cycle. To express late competence genes in addition to the identified early com gene products, a limiting transcription factor is necessary. A DNA fragment upstream of the promoter region of the comDE, comC, and comG genes appears to bind this competence transcription factor (CTF) (4,17). The SOS system appears to be partially derepressed during competence, as judged by overexpression of the din (including dinR) and recA genes (11,15,19). The induction of the dinA, dinB, and dinC genes is observed only in the presence of wild-type RecA activity. During competence, the increased synthesis of RecA does not require the presence of a functional RecA protein; in the recA4 mutant, the competence-related stimulation of recA expression is observed, whereas overexpression of recA following DNA damage is inhibited (15). Overexpression of dinR is also observed during competence, even in the presence of the mutant DinRl protein (19). These results indicate that another factor might be involved in the specific induction of the SOS response.
Within the promoter regions of several din genes, there are conserved sequences that have been proposed as recognition elements for the binding of a presumably common SOSspecific regulatory protein. These sequences have also been detected upstream of the recA, recM, and dinR genes (3,19). In addition, a sequence that resembles the putative CTF binding element was found upstream of the recA gene (4). This finding raised the possibility that, in addition to the classical SOS regulation, a dual regulation of recA involved the binding of CTF and the subsequent displacement of the repressor of the SOS genes, leading to induction of recA expression during competence (4).
We have investigated the epistatic relationship between dinR and recA by measuring (i) expression of recA in 3172 RAYMOND-DENISE AND GUILLEN wild-type, merodiploid, and mutant dinR backgrounds and (ii) expression of dinR in both wild-type and recA backgrounds. These experiments were conducted after induction of the SOS response either by DNA-damaging agents or when cells developed competence. The results presented here indicate that, after DNA damage, the expression of recA and dinR is dependent on the wild-type activity of both DinR and RecA proteins. When cells become competent, in addition to this regulation, an unidentified factor seems to be necessary for the expression of dinR and recA.

MATERIALS AND METHODS
Bacterial strains, plasmids, and growth media. B. subtilis strains were derivatives of wild-type strain 168 and are listed in Table 1. Construction of the dinR merodiploid strain that contains both the dinR-lacZ fusion and a wild-type copy of dinR has been previously reported (19). The recA::cat mutation from the strain YB886-recA::cat was transduced into the dinRI strain M0534 to create the strain M0569. The strain M0556 was constructed by Campbell-like recombination between the pB16 plasmid, which carries the dinR-lacZ fusion and the cat gene, and the chromosome of the wildtype strain. Because strain M0556 was resistant to chloramphenicol, we performed its transformation with a recA:: aphA3 null allele (Kmr; recently constructed by F. Kuntz). The resulting strain, M0579, contains the dinRl-lacZ fusion along with an intact copy of the dinR gene and the recA::aphA3 mutation. Chromosomal DNA from QB4444, a strain which carries the recA::aphA3 mutation, was used to transform competent M0556 cells to kanamycin resistance (KMr), thereby generating the strain M0579. We renamed recA::cat and recA::aphA3 the recA2 allele of recA; in these constructions, the cat or aphA3 gene was inserted into the ClaI site of recA. To introduce the recA: :ylE fusion into the wild-type and dinRI strains, we used the strain BG225, which carries the recA::xylE gene fusion ectopically inte-grated onto the chromosome. The chromosomal DNA purified from BG225 (recA::xyE) was mixed (1:1) with DNA purified from M0506 (Rif); Rif' recA::ylE recombinants were then obtained by congression of these two markers into the wild-type strain and strain M0556. The Rif' recombinants were selected in rich medium in the presence of rifamycin and then sprayed with a 0.5 M solution of catechol to reveal the activity of the xylE gene product, catechol 2,3-oxygenase (catO2ase) (27). By a similar method, the wild-type strain was also transformed for dinR::Tn9171ac and recA::xyIE; in this case, DNA from BG225 (recA::xylE) was mixed with DNA purified from M0534 (dinRl:: Tn9171ac Eryr), and recombinants were selected on medium containing erythromycin and sprayed with the catechol solution. The strains M0572 (recA+ recA::ylE rfin-486), M0573 (dinR::Tn9171ac recA+ recA: :yIE), and M0574 (dinR+ dinR::Tn9171ac recA+ recA::ylE rfin-486) were subsequently purified by streaking them on rich medium in the presence of the appropriate antibiotic. The comA124 mutation from BD1626 (9) was moved into the strain M0534 by transformation. Plasmid pB16 has been previously described (19) and contains the BglII promoter proximal clone of the dinR::lacZ fusion from strain M0542. Erythromycin at 5 ,ug/ml, lincomycin at 20 ,ug/ml, mitomycin (MC) at 50 or 150 ng/ml, rifamycin at 5 ,ug/ml, kanamycin at 5 ,ug/ml, chloramphenicol at 5 ,ug/ml, and 5-bromo-4-chloro-3-indolyl-p-D-galactopyranoside (X-Gal) at 40 ,ug/ml were utilized.
Genetic and molecular procedures. Genetic transformation, transduction, development of competence, and chromosomal DNA purification were performed as described previously (6).
Enzymatic assays. ,-Galactosidase assays were conducted by using bacteria grown in Luria-Bertani or Schaeffer's medium; no significant difference in the 3-galactosidase activity between bacteria grown in each of the two media was observed. For the catO2ase assay, strains were grown in Luria-Bertani medium because the fusion has no activity in Schaeffer's medium. For measuring enzymatic activities during competence, a one-step protocol was used as described previously (1). In particular experiments, MC was added at 150 ng/ml during the exponential phase of growth. The level of ,B-galactosidase activity was determined as described previously (8). ,B-Galactosidase specific activities are expressed as units per milligram of protein. The amount of protein was determined from standard curves by relating the turbidity (optical density at 570 nm) to the protein concentration of the bacterial cultures. Preparation of samples from culture and determination of catO2ase activity were performed as described elsewhere (27), with minor modifications: the cells were washed in 20 mM potassium phosphate buffer (pH 7.2) and resuspended in AP buffer (100 mM potassium phosphate buffer [pH 7.5] and 10% acetone [vol/vol]). Whole cells were disrupted by 30 min of incubation with 50 ,ug of lysozyme per ml at 37°C and then stored at 4°C. The catO2ase activity was determined by following the increase in A375 due to the accumulation of 2-hydroxymuconic semialdehyde. Milliunits of activity are defined as described by Sala-Trepas and Evans (20).

RESULTS
Expression of dinR after DNA damage. The previously characterized dinRI mutant was deficient in both recombination and DNA repair, and the expression of the dinRI allele was noninducible by DNA damage. In the dinR merodiploid strain M0556, which carries an intact copy of the dinR gene coexistent with the dinR1::Tn9171ac mutation, recombination and DNA repair capacities were restored, and the expression of the dinR gene became inducible by DNA-damaging agents (19). To investigate the role of RecA on the expression of dinR, a recA2 null allele (recA::cat or recA::aphA3) was introduced into strains M0534 (dinRl) and M0556 (dinR+IdinRl), each of which contains a dinR-lacZ fusion. The double mutant strain M0569 (dinRl recA2) had a lower growth rate than the parental dinRl strain, which had a rate of growth equivalent to that of the wild-type strain. The doubling time of strain M0569 in rich medium was 35 min, whereas the doubling time of strain M0534 was 23 min. In addition, strain M0569 demonstrated a phenotype of extensive filamentation, which was not observed in the recA or dinR mutants. The f-galactosidase activities of cultures of M0534 (dinRl recA+), M0569 (dinRl recA2), M0556 (dinR+IdinRl recA+), and M0579 (dinR+IdinRl recA2) were measured, in the presence or absence of MC (Table 2). Increased expression of the dinR gene in the presence of MC was seen only in strain M0556 (dinR+IdinRl recA+4); P-galactosidase activity observed 3 h after the addition of MC was 4.5-fold greater than that observed in the absence of MC. In addition, the basal level of dinR-lacZ expression in this strain was threeto fourfold higher than that of the other strains. In contrast, no significant increase in the expression of the dinR-lacZ fusion in the presence versus the absence of MC was observed in strains M0534 (dinRI recA+), M0569 (dinRl recA2), and M0579 (dinR+IdinR1 recA2). Comparison of dinR expression in strain M0556 (dinR+IdinRl recA+) with its expression in strain M0579 (dinR+IdinRI recA2) indicates that the absence of a functional RecA protein prevents the induction of the dinR gene after DNA damage. These results suggest that a RecA-dependent activity is necessary for the induction of dinR expression following DNA damage.
Effect of DinR on recA expression following MC treatment. The DinR protein is a protein that regulates the expression of both dinR and dinC (19). Another important protein in the function of the SOS response in B. subtilis is RecA. To measure the effect of DinR on the transcription of the recA gene, a recA:xylE gene fusion was introduced into the wild-type, dinRI, and merodiploid dinR+IdinRl strains. All of the resultant strains are merodiploid for the recA gene, with the wild-type copy of recA at its normal position on the B. subtilis chromosome and the recA :xylE fusion integrated ectopically. Expression of the recA::xylE fusion was measured in the following strains: M0572 (wild type), M0573 (dinRI), and M0574 (dinR+IdinRl) ( Table 3). The levels of recA expression during growth were extremely different for the three strains. When the recA: :yE fusion was introduced into the dinRl mutant strain (MO573), the basal level of recA expression was only 10% of that of the wild-type strain (MO572). In the dinR merodiploid strain (MO574), the level of recA::xylE expression was approximately twofold higher than that of the fusion in the wild-type strain. In all of these strains, a 7to 10-fold increase of catO2ase activity was observed when cultures in rich medium entered the stationary phase of growth. A similar unexplained increase was observed by Gassel and Alonso when the recA::xylE fusion was carried on a multicopy plasmid (7).
In the wild-type strain grown in rich medium, expression of the recA::xylE fusion started to increase 1 h after the addition of MC. After 3 h of MC treatment, the recA::xy1E expression increased to 5.5-fold its expression in the untreated culture (Table 3 and Fig. 1). Induction of recA expression was also observed in the dinR merodiploid background (strain M0574), indicating that wild-type regulation of recA expression occurs in the merodiploid strain. In contrast, no induction of the recA::ylE fusion occurred in strain M0573 after MC treatment. These results suggest the following hypotheses. (i) DinR is the repressor for the damage-inducible genes and is inactivated after DNA damage, thereby leading to the expression of the SOS genes. In this case, the DinRl protein is a noninactivable mutant protein which acts as a super repressor. (ii) DinR is a positive regulator necessary for the induction of recA. Furthermore, the low level of expression of recA found in the dinRl mutant might explain the Rec-phenotype of this strain (19).
Effect of RecA on competence-induced dinR expression.
When B. subtilis cells reach competence, specific induction of the din and recA genes occurs (11,15). While induction of   (Table 4). In the dinR merodiploid strain M0556, a threefold increase in dinR expression was observed when the cells developed competence. In strain M0579 (dinR+/dinR1 recA2), a twofold increase of dinR expression was observed, but only 2 h after the end of exponential growth. In addition, the basal level of dinR expression (at 0.5 h before end of log-phase growth; T_0o5) was fivefold lower in strain M0579 than in strain M0556.
These results indicate that the RecA protein is (i) necessary for the normal timing of induction of dinR expression during competence and (ii) necessary for the maintenance of a basal level of DinR during growth. A competence-related increase of ,-galactosidase activity was also observed in strains M0534 (dinRI) and M0569 (dinRI recA2) ( Table 4), indicating that in these strains the augmentation of dinR expression occurs in the presence of the DinRl protein and in the absence of RecA. These results may indicate that the competence-related dinR induction occurs independently of any activity of RecA on DinR. In addition, dinR expression during vegetative growth (T_0o5) was lower in both strains M0534 and M0569 than in strain M0556. This suggests that the DinR protein also contributes  to the basal level of dinR expression. Such a contribution is evident when strains M0556 and M0534 are compared. Effect of DinR on the competence-related induction of recA. Previous work has shown that the recA gene is induced in competent cells (15). This induction occurs in the recA4 mutant strain, which encodes an SOS-inactive form of the RecA protein. To investigate whether the competence-related induction of recA is dependent on the activity of DinR, the expression of the recA IxylE fusion was measured during competence. In the dinR+ strain (MO572), the recA::xylE fusion was induced 10-fold 2 h after the end of exponential growth, whereas no induction of recA in the dinRI mutant was observed (Fig. 2). These results suggest that the competence-related overexpression of recA requires the presence of a wild-type DinR protein. Taking into account that only 10 to 20% of cells reach competence (4), the observed values (in terms of milliunits of catO2ase per milligram of protein) for competence-related recA induction are underestimates in these experiments.
Effect of comA on competence-related expression of dinR. Transcriptional activation of the dinR gene has been observed when B. subtilis becomes competent, even in the absence of any known DNA-damaging agent (19). The competence-related induction of dinR expression occurs in the dinRI mutant, thereby indicating that it is independent of the DinR wild-type protein. The induction of dinR could thus be controlled by comA, a major regulator of competence gene expression (9). To investigate the role of comA in competence-induced expression of dinR, the comA124 allele was introduced by transformation into strain M0534 (see Materials and Methods). ,-Galactosidase activities in the dinRI mutant (MO534) and in the dinRI-comA double mutant (MO551) were measured (Fig. 3). Comparable levels of expression of the dinR gene occurred in the two strains when the cells became competent, thereby indicating that dinR overexpression is independent of the comA gene product.

DISCUSSION
The development of the competence state in B. subtilis is accompanied by induction of the din and recA genes, the expression of which is also induced by DNA-damaging agents (11,15). Both the competence-and the DNA damage- We have previously reported the identification of dinR, a regulator of din gene expression, which encodes a protein homologous to the LexA protein of E. coli (19). The dinRI mutant, originally identified as deficient for homologous recombination, encodes a truncated form of the DinR protein which lacks 20% of the wild-type protein at the carboxy terminus. Expression of dinR is induced following DNA damage and when the cells become competent. In contrast to the DNA damage induction of dinR, its competence-related induction occurs in the dinRI mutant. In this work, we have investigated the epistatic relationship between the B. subtilis recA and dinR genes during both the SOS response and the development of competence.
Induction of a dinR-lacZ fusion following DNA damage was abolished in both recA2 and dinRI mutants, indicating that similar effects result from the presence of the DinRl protein (or absence of DinR) and from the absence of the active RecA protein. These observations support two hypotheses. (i) The amino acid motifs known to be necessary for the cleavage of the E. coli LexA protein are conserved in DinR, suggesting that the DinR protein could undergo a similar RecA-mediated cleavage (19). In this case, DinR normally acts as a repressor of the SOS genes and the DinRl protein is a noncleavable form of DinR. (ii) DinR is a positive effector necessary for the expression of the SOS genes, and the DinRl protein has lost this activity (discussed below).
Expression of the recA gene is highly influenced by the dinR allele present in the cell. In the presence of the dinRl mutation, recA expression was only 10% of that obtained in the presence of the dinR wild-type allele. This low rate of recA expression in the dinRI background might explain the Rec-phenotype of the DinRl mutant (19). In the dinR merodiploid strain, both DinR and DinRl are synthesized, and the level of recA::ylE expression was approximately twofold higher than that of the fusion in the wild-type strain. Thus, our results show that dinR, dinC, and recA are poorly expressed in the presence of the dinRl allele and that in the merodiploid strain the level of expression of recA is higher than it is in the wild-type strain. These results lead us to favor the first hypothesis, in which DinR acts as a repressor for damage-inducible genes such as recA, dinC, and dinR. The presence of the conserved SOS boxes in the regulatory regions of these genes and of dinA, dinB, and recM (3,19) suggests a possible interaction of DinR with these DNA sequences. From the results concerning dinR and recA expression during vegetative growth, we can conclude that a reciprocal regulatory effect allows a basal level of expression of these genes. A permanent low level of DinR expression appears to be necessary to ensure a basal quantity of RecA, which accounts for the induction of the SOS response. The presence of a DinRl protein leads to an SOS inductiondeficient phenotype. In contrast, RecA wild-type activity is needed to maintain the DinR constitutive level and to derepress dinR transcription following DNA damage. These results allow a comparison between the SOS responses of B. subtilis and E. coli (22), in which there is an interdependence of RecA and LexA in the regulation of the SOS network.
dinR expression, like that of the din and recA genes, is induced when the cells become competent. In this case, the RecA protein has no effect on dinR induction in the dinRI mutant, supporting the hypothesis that the DinRl protein is insensitive to RecA activity. Moreover, the absence of the RecA protein in the merodiploid strain was associated with a diminution of competence-related dinR induction. These results suggest that the DinR protein synthesized in the merodiploid strain can be modified by some RecA activity, leading to overexpression of the dinR gene. This mechanism is consistent with dependence on RecA for the induction of the other din genes during competence. Moreover, in the dinRI strain, when the cells reached competence, induction of recA expression was abolished whereas that of dinRI was still present. These results suggest that DinRl regulation of the recA promoter prevents recA induction. However, the activity of RecA on DinR does not seem to be the only way to activate dinR expression during competence. In the dinRl-recA2 double mutant, as in the dinRI strain, the competence-related overexpression of dinR is still observed. Thus, the RecA-dependent activity on DinR, necessary for SOS induction, could be bypassed by another mechanism during competence. The competence-related overexpression of recA depends on the presence of a DinR wild-type protein (Fig. 2); nevertheless, this induction is also observed in the recA4 mutant, which encodes an inactive form of the RecA protein (14). These results could also indicate that the RecA-dependent activity on DinR is not necessary for the induction of the recA gene; it is possible that the dinR and recA genes are also under the regulation of an unidentified activator that is independent of the SOS regulation. This putative factor would be able to displace DinR (or DinRl) from its target sequence without RecA-dependent activity, thereby permitting the transcription of the dinR or the recA gene. Overexpression of dinR during competence was observed in the dinRl comA double mutant, indicating that the competence-induced expression of dinR is independent of the regulator ComA. Thus, expression of the putative activator of dinR must also be independent of ComA. This putative factor is different from the CTF evoked as a regulator of the SOS competence phenomenon, since CTF is not synthesized in the comA124 mutant (4).
In conclusion, we propose the following hypothesis to VOL. 174, 1992 explain the double regulation of dinR and recA expression observed during competence. When competence is induced in the wild-type cell, a low level of expression of the dinR and recA genes occurs by a mechanism that involves the presence of an activating factor that displaces DinR from the dinR and recA promoters. This mechanism does not involve any activity of RecA on the DinR protein, thereby explaining the overproduction of DinR and RecA during competence in the dinRl and recA4 mutants. A competition between DinRl and the putative positive activating factor for binding to the recA promoter would account for the low level of transcription of recA in the dinRI mutant and for its Rec-phenotype. In parallel, the SOS signal might be produced by a limited rate of DNA replication in competent cells (4). At this time, the survival of the cell would be guaranteed by a derepression of the SOS functions which might be due to RecAmediated cleavage of DinR. Subsequently, full induction of the SOS system occurs, and the resultant increased quantity of RecA allows genetic recombination and derepression of the din genes.