RecA Protein-dependent Proteolysis of Bacteriophage X Repressor CHARACTERIZATION OF THE REACTION AND STIMULATION BY DNA-BINDING PROTEINS;

RecA protein (the wild type form of the protein, Le. the product of the mcA+ gene), purified to homogeneity by a novel ATP elution step (Cox, M. M., McEntee, K., and Lehman, I. R. (1981) J. BioL Chem 266,4676-4678), cleaves bacteriophage X repressor in a reaction requiring a nucleoside triphosphate, single-stranded polynu- cleotide and divalent cation. (d,r)ATP, (d,r)UTP, rATP[yS], rUTP[yS], and rGTP[yS] serve as cofactors for recA protease activity, whereas (r,d)GTP, (r,d)CTP, and d’ITP do not. The reaction is inhibited by ADP, UDP, and dITP, all of which bind to recA protein. In the presence of ATP or other hydrolyzable nucleoside triphosphates, the rate of repressor cleavage is greatly enhanced by substituting Mn” ion for M&+, an effect which is correlated with a reduction in ATP hydrolysis. The polynucleotide requirement is satisfied by +X114 DNA, poly(dT), poly(dU), and poly(dC). Polyribonucle- otides and oligodeoxynucleotides are significantly less effective and duplex DNA inhibits the cleavage of X repressor in the presence of single-stranded DNA. A ratio of 2-3 single-stranded nucleotides/recA protein monomer is optimal for proteolysis and the reaction is

' In this paper "recA protein" refers to the wild type form of the protein, i.e. the product of the recA+ gene. these genes are controlled at the transcriptional level by the l e d protein, a repressor which is inactivated in response to DNA damage (4). Bacteriophage X is also induced in response to DNA damage, and the A repressor is inactivated analogously to the lexA repressor. Roberts et al. (5) first determined the mechanism of X repressor inactivation by showing that recA protein was a protease that cleaved the X repressor protein. Subsequently, recA protein-dependent proteolytic cleavage of the l e d repressor (4) and the bacteriophage P22 repressor (6) have been demonstrated.
In addition to its protease activity, the recA protein promotes the hybridization of SS2 DNA molecules (7) as well as formation of D-loops by hybridizing SS DNA to complementary regions within duplex DNA molecules (8,9). This biochemical versatility is consistent with genetic studies of recA mutants, which are defective in DNA recombination and repair (presumably requiring the hybridization function) as well as the induction of SOS functions and prophages (10). A special class of recA mutation, termed lexB, is defective in SOS and prophage induction activities but normal in DNA recombination (11), demonstrating that the protease and hybridization activities of recA protein are genetically and functionally distinct. Another interesting mutation in the recA gene, tif-1, is able to induce SOS functions and prophages in the absence of DNA damage (12).
The protease activity of the recA protein is unusual in its requirement for ATP (but not its hydrolysis) and SS polynucleotides (5,13,14). Remarkable substrate specscity is observed for the recA protease: cleavage of the X, lexA, and P22 repressors occurs at single sites within the polypeptides. The precise endopeptidase site has been determined in the case of X repressor (15). No cleavage of other proteins by recA endopeptidase has been observed in crude extracts or partially purified fractions (6, 13).
The requirement for a SS polynucleotide for in vitro protease activity is consistent with the notion that the inducing signal in vivo is SS DNA regions, generated as a result of DNA damage. It has been suggested that this SS DNA is produced as a replication fork encounters the DNA lesion, leaving postreplication gaps (16), or as a result of DNA degradation, thereby creating SS oligonucleotides (17). Several distinct genetic loci are implicated in induction of SOS functions and prophages (other than recA and lexA genes) including recB, recC, ssb (ZexC), and recF, and the products of the recB, recC, and ssb genes have been identified. The recB and recC genes encode exonuclease V, a multifunctional nuclease that presumably has a role in generating the inducer of SOS functions through DNA degradation (17). The product of the ssb (or ZexC) gene is a SS DNA-binding protein (SSB) that is The abbreviations used are: SS, single strand; DS, double strand; ATP[yS], adenosine-5"0-(3-thiotriphosphate); SSB, single strand DNA-binding protein.
required for DNA replication and repair of damage (18-20). Baluch et al. (21) have recently demonstrated that SSB is required in vivo for recA protein and phage X induction by ultraviolet irradiation.
Previous studies of the recA protein protease activity have concentrated on the tzfmutant form of recA protein. Here we present an analysis of the protease activity of the wild type recA protein which has been extensively characterized for DNA binding, nucleoside triphosphate hydrolysis, and hybridization activities. We have also investigated the effects of SSB on recA protein-dependent proteolysis of A repressor and find that SSB from wild type, but not from ZexCll3 mutant, cells stimulates X repressor cleavage by recA protein in the presence of excess single-stranded DNA.

EXPERIMENTAL PROCEDURES
Enzymes-The recA protein was purified by ATP elution from DNA cellulose as described (22) and was greater than 98% pure. X repressor (23) was generously provided by Dr. A. D. Kaiser. Homogeneous SSB was a gift of Robert Fuller. The lexC protein was as described (24). dnaC protein, dnaB protein, protein n, protein n', and rep protein (25) were from the laboratory of Dr. A. Kornberg. DNA polymerase I was provided by Stuart Scherer, p factor was provided by Ron Conaway (all of this department) and polynucleotide phosphorylase was provided by Dr. Michael Smith (University of British Columbia). DNA ligase was as described (26). Gene 32 protein of bacteriophage T4 was from Dr. Bruce Alberts (University of California, San Francisco). Pyruvate kinase was purchased from Sigma.
Nucleic Acids-Bacteriophage +X174 single-stranded DNA and double-stranded DNA of the plasmid pZ6b were prepared as described repressor, and recA protein as indicated were incubated at 37 "C in plastic Eppendorf tubes. Reactions were stopped by addition of 6 pl of a solution containing 5% sodium dodecyl sulfate, 4% dithiothreitol, 0.05% bromophenol blue, and 25% glycerol. The samples were applied to a 15% polyacrylamide gel (30) and electrophoresed at 25 to 50 mA for approximately 3 h. Gels were stained for 10 min in a solution containing 45.5% methanol, 0.91% acetic acid, and 2.5 mg/ml of Coomassie brilliant blue R250. Following destaining in 10% methanol, 7% acetic acid, the gels were scanned spectrophotometrically using a Quick Scan Jr. densitometer equipped with integrator and the extent of cleavage was calculated.

RESULTS
Characterization of the Protease Activity of the Wild Type recA Protein-Proteolytic cleavage of X repressor by recA protein required ATP or ATP[yS], a divalent cation, and a single-stranded polynucleotide, analogous to the tif-1 mutant enzyme requirement (14). With ATP as a cofactor, efficient repressor cleavage occurred only with Mn2+ ion and not with M e , Ca2+, or Zn2' ions ( Fig. la). This contrasts with the SS DNA-dependent ATPase activity of recA protein which is optimal with Mg2' ion and shows considerably reduced activity with Mn2+ ion (Fig. lb). The failure of Mg2+ ion to stimulate the protease activity is related to increased ATP hydrolysis since repressor cleavage was enhanced in the presence of M$+ ion by either raising the initial ATP concentration or including an ATP-regenerating system (pyruvate kinase/phosphoenolpyruvate; data not shown).
ATP [$], which also serves as a cofactor for the protease activity, is not appreciably hydrolyzed by recA protein (14, 31, 32), although in the presence of DNA it binds extremely tightly to the enzyme (32) and stabilizes recA proteineDNA complexes (33). With ATP[#], both M$+ and Ca2' ions stimulated the protease; other divalent cations were not tested. With 1 m~ ATP[@], optimal cleavage activity occurred at 3 m~ Mg2' ion and inhibition was observed above 5 mM Mg2+ ion (data not shown), similar to Mn2+ ion effects upon the reaction with ATP. The inhibition by M$+ ion was reversed by Ca2+ ion (data not shown).
Repressor cleavage was absolutely dependent on a singlestranded polynucleotide with either ATP (Mn) or ATP [$] (Mg) as cofactor. In both cases, optimum activity was observed at a ratio of about 3 single-stranded nucleotides/recA protein monomer (Fig. 2), similar to that reported previously for the tifmutant protein in the presence of ATP [yS] (14). At this stoichiometry the SS DNA is saturated with recA protein (34). DS DNA failed to stimulate proteolysis, even below pH loo t l      (Fig. 3) where DS DNA binding to recA protein and stimulation of ATP hydrolysis are optimal (27, 33). In fact, DS DNA inhibited the SS DNA-dependent cleavage of X repressor (Fig. 4). Under the conditions of this experiment, SS DNA stimulates binding of DS DNA to recA protein (31) and thus it appears likely that this inhibition is due to the interaction of DS DNA with recA protein rather than with X repressor. Proteolysis was sensitive to monovalent ions (Fig. 5). The reaction was inhibited by concentrations of spermidine greater than 1 mM; lower concentrations had no effect (data not shown). RecA protein-dependent proteolysis with either ATP or ATP[yS] as cofactor was completely abolished by 10 mM N-ethylmaleimide. This result is noteworthy in that the formation of stable recA protein. ATP[yS] complexes is not sensitive to N-ethylmaleimide (32).
The rate of proteolysis in the presence of ATP[yS] as cofactor was proportional to recA protein concentration and proceeded at a constant rate until all of the repressor was cleaved (Fig. 6). In addition, this experiment demonstrates turnover of the recA protein. Complete cleavage did not occur with ATP as a cofactor, presumably as a consequence of ATP hydrolysis. Proteolysis with either ATP (Mn) or ATP[@] as cofactor, however, showed the same initial rate, which corresponded to one cleavage event/recA monomer/l30 min at 37 "C. It should be noted that at the high concentrations of X repreasor used in these experiments the rate of repressor cleavage is probably limited in part by the dimerization of the repressor monomer (6).   for Proteolysis-In the presence of MnZ" ion, rATP and dATP were the preferred cofactors for the proteolytic reaction but significant activity was also observed with rUTP and dUTP; other nucleoside triphosphates, diphosphates, and monophosphates were not active as cofactors (Table I). This specificity parallels that of the NTPase activity of recA protein (27) and indicates that the same nucleoside triphosphate binding site is involved in both reactions. (r,d)GTP, (r,d)CTP, and (r,d)UTP were modest inhibitors of ATP-dependent proteolysis; however, dTTP, ADP, and UDP were more potent inhibitors (Table 11), similar to their effect on the ATPase activity of recA protein (35). dAMP and dTMP had no effect.

Specificity of the Nucleoside Triphosphate Requirement
As noted above, proteolysis with ATP[@] was as efficient as with ATP (Mn

Specificity of the Polynucleotide Requirement for
Proteolysis-Proteolysis by the recA protein was stimulated by deoxyribopolymers while ribopolymers had little effect (Table 111). Polydeoxypyrimidines were most active as cofactors. Furthermore, short oligonucleotides (dTlz) were not active as cofactors. Thus, proteolysis by the wild type recA protein shows a similar polynucleotide specificity as does its ATPase activity (27). However, this specificity differs from that of the tif mutant protein3 (14).
Substrate Specificity of the Protease-Several purified E. coli proteins were tested as substrates for the recA protein protease in the presence of ATP [yS]. No recA protein-dependent cleavage, as detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, was observed for any of the following purified proteins: SSB, dnaC protein, protein n, dnaB protein, rep protein, protein n', DNA polymerase I, DNA ligase, p factor, or polynucleotide phosphorylase. Furthermore, none of these proteins inhibited A repressor cleavage by recA protein. Functional assays of these proteins were not performed after incubation with recA protein; consequently, proteolytic cleavage events which alter the activity without appreciably changing the electrophoretic mobility of the protein would not have been detected.
Effect of the E. coli SSB on Proteolysis-As described previously, excess SS DNA inhibits recA protein-dependent proteolysis of X repressor ( Fig. 2; Ref. 14). Likewise, strand assimilation is inhibited by excess SS DNA and the inhibition is overcome by SSB (24, 36) which binds to the excess SS DNA. As shown in Fig. 7, SSB also overcame inhibition of the recA protease activity by excess SS DNA. The stimulatory effect of SSB was seen in both ATP-and ATP[yS]-dependent reactions. In both cases an optimum SSB concentration was observed, above which SSB inhibited the reaction. The gene 32 protein of bacteriophage T4 also overcame the SS DNA inhibition of recA protein-dependent proteolysis (Fig. 7c), similar to its effect on the assimilation reaction3 (36). However, SSB protein purified from the ZexC113 mutant of E. coli only weakly stimulated proteolysis (Fig. 7b). The mutant lexC binding protein also failed to stimulate the strand-assimilation reaction (24).
Unpublished results.
As shown in Fig. 8, the effect of SSB is to shift the DNA optimum to higher concentrations while having little effect on the maximum rate, which is determined by the recA protein concentration. Higher Concentrations of DNA were required to inhibit proteolysis in the presence of SSB but inhibition was observed, indicating that SSB does not alter the recA protein but exerts its effect by covering the DNA. The shift in the optimum in Fig. 8 indicates that about 10 nucleotides are bound/SSB monomer; thus, the DNA is nearly saturated with SSB (37).

Further Evidence for the Protease Activity of the recA+
Protein-Our results further support the conclusion that proteolytic cleavage of X repressor is catalyzed by recA protein.
Thus, homogeneous recA protein purified by a procedure that differs from those used previously (22) contains an endopeptidase activity that cleaves A repressor. The nucleoside triphosphate specificity that is observed for proteolysis is very similar to the S S DNA-dependent hydrolytic activity of recA protein. Preliminary characterization of mutant forms of the recA protein purified by this procedure is consistent with this view. The tif-1 mutant protein is altered in its polynucleotide requirements for both ATPase and repressor cleavage activities? The lexB30 mutant protein fails to cleave X repressor in vivo (11) and in vitro but is active for recombination in vivo and strand pairing in vitro3 (7,11).
Under the conditions described in this paper, recA protein cleaves an equimolar amount of X repressor i n about 2 h; in vivo, the cleavage of X repressor is complete in about 30 min (38). However, in our reactions the concentration of repressor is at least 50-fold higher than in vivo (39), whereas the recA protein concentration is comparable to that found in uninduced cells.5 In view of the fact that at high concentrations, repressor dimerization may inhibit its proteolysis by recA protein (6,40), the rate of cleavage that we observe in vitro is not inconsistent with the in vivo rate of X induction.

Inhibition by Single-stranded DNA and the Effect of
SSB-Single-stranded DNA is required stoichiometrically for both repressor cleavage (14) and strand assimilation (8) reactions and, when present in excess, inhibits both reactions (14, 24). Previously it was suggested that the inhibition of assimilation is due to competition between the excess S S DNA and DS DNA for binding to recA protein (24). However, it seems unlikely that a similar competition exists between the site for binding X repressor and S S DNA. The stoichiometric requirement for SS DNA in strand assimilation has also been attributed to a requirement for melting secondary structure in the S S DNA (36); however, here we report inhibition of X repressor cleavage by poly(dT), which lacks such secondary structure. High concentrations of SS DNA are not inhibitory to the recA protein per se since neither the ATPase activity nor the strand-reassociation activity is inhibited by excess SS  DNA (7, 35). In addition, it appears that high concentrations of oligonucleotides, unlike polynucleotides, are significantly less inhibitory (in Ref. 14, compare Table I and Fig. 6 The effect of SSB on the protease reaction appears to be to bind to the excess S S regions and abolish their inhibitory effect. The observation that high concentrations of SSB inhibit proteolysis suggests that recA protein binds poorly to SS DNA that is covered with SSB. The data of Figs. 7B and 8 are consistent with this notion. Optimal activity occurs when the SS DNA just saturates the recA protein and SSB. At lower SSB concentrations there is inhibition due to free SS regions while at higher SSB concentrations the inhibition would be due to the competition between SSB and recA protein for SS DNA. Thus, this model suggests that no interaction between SSB and recA protein is necessary for repressor cleavage in vitro. Consistent with this, we find that the gene 32 protein of bacteriophage T4 substitutes for SSB. A similar effect of this phage-coded protein has been reported for strand assimilation promoted by recA protein (36).
The observation that SS DNA covered with SSB is not an effector for the protease reaction in vitro has implications for the in vivo reaction. S S DNA at the replication fork is likely to be complexed with SSB and thus will not be a good effector for the induction of SOS functions by recA protein. However, DNA damage will generate additional SS DNA regions lowering the SSB:SS DNA ratio in vivo.
The increased amount of SS DNA regions could be bound with recA protein in order to stimulate the recA protease and derepress SOS functions including the recA gene. Thus the level of SSB determines a threshold value of SS DNA needed to activate recA protein. A prediction of this model is that strains containing plasmids that overproduce SSB w i l l show an altered dose response for recA protein induction. Implicitly assumed in this model is that the synthesis of SSB is not itself regulated by DNA damage, a possibility that has not been tested directly.
Recently Baluch et aZ. (21) demonstrated the ZexC113 and ssbl mutant strains are defective in their ability to synthesize recA protein and induce prophage A following ultraviolet treatment. In the case of ssbl mutant strains, ultraviolet induction of recA protein synthesis is blocked at 42 "C, a temperature at which the binding protein is nonfunctional in vivo (19) and in vitro (18). The ZexCll3 mutation prevents ultraviolet induction of recA protein synthesis and prophage A at all temperatures (21). Our observations are consistent with these results. Our results that excess SSB inhibits recA protein dependent proteolysis of phage A repressor is compatible with the idea that when SS DNA regions (gaps) are limiting, recA protein and SSB compete for these binding sites. A more detailed analysis of the interaction between recA protein and both wild type and mutant forms of SSB (ZexCl13 and ssbl) should elucidate the mechanism of these effects.