Function of Nucleoside Triphosphate and Polynucleotide in Escherichia coli recA Protein-directed Cleavage of Phage X Repressor *

Escherichia coli recA protein catalyzes a specific proteolytic cleavage of repressors in vitro when it is activated by interaction with a single-stranded polynucleotide and a nucleoside triphosphate. The ATP analogue adenosine-5’-0-(3-thiotriphosphate) (ATPyS) satisfies the NTP requirement. We show here that despite its activity in repressor cleavage, ATPyS is hydrolyzed at a negligible rate by the recA protein DNA-dependent nucleoside triphosphatase activity. In the presence of DNA, ATPyS binds tightly to recA protein in a complex that can be detected because it is trapped by a nitrocellulose filter. One ATPyS molecule is bound per recA monomer. These results suggest that a ternary complex of mcA protein, DNA, and nucleoside triphosphate is the species active in repressor cleavage. The activation of recA protein by small, defined oligonucleotides in place of DNA is described and characterized.

Escherichia coli recA protein catalyzes a specific proteolytic cleavage of repressors in vitro when it is activated by interaction with a single-stranded polynucleotide and a nucleoside triphosphate. The ATP analogue adenosine-5'-0-(3-thiotriphosphate) (ATPyS) satisfies the N T P requirement. We show here that despite its activity in repressor cleavage, ATPyS is hydrolyzed at a negligible rate by the recA protein DNA-dependent nucleoside triphosphatase activity. In the presence of DNA, ATPyS binds tightly to recA protein in a complex that can be detected because it is trapped by a nitrocellulose filter. One ATPyS molecule is bound per recA monomer. These results suggest that a ternary complex of mcA protein, DNA, and nucleoside triphosphate is the species active in repressor cleavage. The activation of recA protein by small, defined oligonucleotides in place of DNA is described and characterized.
The protein encoded by the Escherichia coli recA gene has two activities that involve a polynucleotide and a nucleoside triphosphate such as ATP: it promotes the pairing of singlestranded DNA to its homologous sequence in a DNA duplex, an ATP-dependent reaction called "strand exchange" (1, 2); and it catalyzes a specific proteolytic cleavage of repressors, a reaction that requires both a nucleoside triphosphate and a polynucleotide (3-5). recA protein also has a DNA-dependent nucleoside triphosphatase activity (5, 6) These activities reflect various roles of recA function in certain cellular processes (sometimes called SOS functions) that promote the repair of damaged DNA (7,8). The ability of recA protein to destroy repressors by proteolytic cleavage underlies its regulatory role in directing the expression of genes that encode DNA repairrelated functions (3, 4, 9-11). A primary target of this proteolytic activity is the E. coli lexA gene product (9), which probably acts as a repressor of both the recA gene itself (9,10, 12) and other genes involved in DNA repair (11). recA protein also directs the cleavage of the immunity repressors of temperate bacteriophages such as A and P22 (4, 13), the critical event in recA-dependent prophage induction in response to DNA damage. The ability of recA protein to promote the pairing of DNA strands reflects its direct role in homologous recombination (14, 15), including recombinational repair of damaged DNA (16).
We present evidence here that recA protein is activated to cleave repressors by binding to single-stranded DNA in the presence of a nucleoside triphosphate. The ATP analogue ATP$' (17) substitutes for ATP in two reactions of recA protein: repressor cleavage (4, 5,9), and the partial unwinding of duplex DNA in the presence of single-stranded DNA (18), the latter a reaction that presumably represents the initiation of DNA strand exchange. Despite being active in these reactions, ATPyS is a potent inhibitor of both the complete DNA strand exchange reaction (2, 18) and the recA DNA-dependence ATPase activity (4). We show here that ATPyS binds in a tight complex with recA protein in the presence of singlestranded DNA, and is hydrolyzed only very slowly by the recA triphosphatase activity. These results support our presumption (4) that a ternary complex of recA protein, DNA, and nucleoside triphosphate is the active species in proteolytic cleavage, and probably also in the reaction that initiates DNA strand exchange. We also determine that further purification of recA protein does not identify any other macromolecular components required for repressor cleavage, and we show that small oligonucleotides efficiently support h repressor cleavage.

EXPERIMENTAL PROCEDURES'
k t e r i a l r were grown i n 100 l i t e r batches ~n d New B r u n r v i c k F e m t r a n fermentor a t 37" ln medium containing 5 gm NaCl, 10 gm yeast extract and 16 9" tryptone p r l i t e ? .   The concentPation o f e protein was c a l c u l a t e d from the measured extinction caefflcient,

RESULTS
X Repressor Cleavage Activity Is Associated with Extensively Purified recA Protein-Incubation of purified A repressor with purified recA protein, ATP, and polynucleotide results in the specific proteolytic cleavage of the repressor polypeptide (3, 4). recA protein is required for this cleavage reaction, because the cleavage activity co-purifies with recA protein, and mutations in the recA gene alter the repressor cleavage activity of purified recA protein in vitro (3, 33, 34). Since it is surprising that recA protein should be a protease in addition to its other activities, we extended our purification to several steps beyond the stage of apparent homogeneity for both recA protein and X repressor, to lessen the probability that a separate unidentified trace component is required for repressor cleavage. Fig. 1 shows that cleavage activity still copurifies with the recA polypeptide in the final stage of this extensive purification, sucrose gradient sedimentation. The specific activity of cleavage is uniform across this gradient and is identical with that of the preceding several purification steps. Thus we find no evidence that any protein except recA is required for cleavage, and we presume that the recA polypeptide contains the catalytic site for proteolysis. Since recA protein clearly is at least essential to the reaction, none of our conclusions in this paper (or elsewhere) would be affected if an unidentified minor component were involved.
ATP@ Is Hydrolyzed Very Slowly by recA Protein-In the presence of polynucleotide, recA protein catalyzes the hydrolysis of ATP to ADP and Pi (5, 6). The ATP analogue ATP+, which both promotes repressor cleavage and inhibits the ATPase activity of recA protein at micromolar concentrations, is itself hydrolyzed very slowly by recA protein.

TABLE I1
Rates of ATP and ATPyS hydrolysis Reaction mixtures of 25 pl contained 10 mM Tris-HC1, pH 7.5,0.25 mM EDTA, 0.5 mM dithiothreitol, 5% (w/v) sucrose, 0.06 M NaC1, and 20 p g / d of heat-denatured X DNA. Reaction mixtures in which ATP hydrolysis was measured contained in addition 10 mM MgCL, 5 mM [y3'P]ATP, and 1 pg of recA+ protein; they were incubated 60 min at 37 "C, followed by chilling on ice to stop the reaction. Reaction mixtures in which ATPyS hydrolysis was measured contained, in addition, 3.3 mM MgC12, 2.3 PM [I4C]ATPy-S and 0.86 pg of recA' protein, and were incubated 45 min at 37 "C, followed by chilling on ice. One-tenth of a volume of 10 mM Tris, pH 7.9, 20% (v/v) glycerol, 1 mM EDTA, 10 mM dithiothreitol, 2 mM CaC12,0.2 M KCI, and 1.5 mg/ ml of h repressor was added, and incubation was continued at 37 "C. At 5-min intervals, aliquots were removed and chilled, and ATPyS hydrolysis and repressor cleavage were assayed, from 25-and 50-pl portions, respectively. A, time course of repressor cleavage after different times of preincubation. The rate was determined by the slope of the line in the interval 5-15 min (except at 45 min when the interval 5-10 min was used). Because the cleavage reaction is inhibited by high salt (4), the decrease in rate of cleavage after the first interval could result from a slow response to the increase in salt concentration upon addition of repressor. B, extent of ATPyS hydrolysis and the rate of repressor cleavage. The ATPyS hydrolysis represents incubation both before and after addition of repressor, but repressor did not affect the rate of ATPyS hydrolysls. The rate of repressor cleavage is plotted against the time at which repressor was added.
it is possible that some species resulting from this hydrolysis would accumulate during incubation, causing the rate of repressor cleavage to increase continuously. In the experiment of Fig. 2, we examined ATPyS hydrolysis and repressor cleavage in the same reaction, by incubating recA protein and ATPyS together in conditions of the cleavage reaction and removing samples at intervals to measure both their ability to cleave repressor and the extent of ATPyS hydrolysis. Fig. 2 shows that the rate of repressor cleavage is independent of the time of preincubation of recA protein with ATPyS and with the extent of ATPyS hydrolysis. This experiment also shows that stoichiometric hydrolysis of ATPyS is not required for repressor cleavage: in the 15 min following addition of repressor after the 5-min preincubation, 8 X lo-" mol of repressor were cleaved while 0.6 X 10"' mol of ATPyS were hydrolyzed per 50-pl sample, more than 10 cleavage events per molecule of ATPyS hydrolyzed. We conclude that if any species dependent upon ATPyS hydrolysis is required for repressor cleavage, it exists in very small concentration and does not accumulate during the reaction. Since ATPyS promotes repressor cleavage more efficiently than do nucleoside triphosphates that are hydrolyzed much faster (34), it seems most likely that a complex of recA protein with the triphosphate is the active species in repressor cleavage, and that hydrolysis of the triphosphate is not required.
ATPyS Binds Tightly to recA Protein in the Presence of DNA-The finding that ATPyS at micromolar concentrations efficiently stimulates recA protein to cleave repressors, even though it is hydrolyzed extremely slowly, suggests that ATPyS forms a relatively stable complex with recA protein.
The experiment of Table I11 shows that radioactive ATPyS incubated with recA protein in the presence of DNA is retained by a nitrocellulose fiiter. Neither ATPyS itself nor ATPyS incubated with recA protein in the absence of DNA is bound in these conditions, although recA protein alone is retained by nitrocellulose (data not shown). Thus, incorporation of ATPyS into this complex requires both recA protein and polynucleotide. The DNA concentration dependence of ATPyS binding is similar to that of ATPase activity, in that bound ATP@ reaches a plateau at higher DNA concentrations (data not shown), and it is unlike the DNA concentration response of repressor cleavage, which is inhibited by higher than optimal DNA concentrations (see Ref. 4 and below). The binding of ATPyS is apparently irreversible, since 1) the halflife of the complex is of the order of hours, similar to the turnover number for ATPyS hydrolysis; and 2) bound radio-   Table 111, except that bovine serum albumin was omitted; the concentration of ["SIATPyS was varied as indicated; the concentration of recA4f1 protein was 85 pg/ml; either 20 pg/ml of heat-denatured h DNA or 10 pg/ml of ( d A ) 1 6 was present; and incubation was for 5 min. Because a significant fraction of the ATPyS is bound at low concentration, the unbound concentration is plotted. The ordinate is the ratio of moles of ATPyS bound to moles of recA protein in the filtered sample. 0, heat-denatured X DNA 0, (dA),R.

FIG. 5.
Single-stranded DNA dosage curve of recA proteindependent cleavage of X repressor. Reaction mixtures of 20 p1 were constructed as described in the legend to Fig. 1, except with 148 pg/ml of repressor, 51 pg/ml of recA441 protein, 0.020 M NaC1,0.040 M KCI, and heat-denatured h DNA as indicated. After incubation of 45 min at 37 "C, cleavage was measured as described.
active ATPyS is not exchanged for nonradioactive ATPyS added after the complex is f~r m e d . ~ We presume that the trappable species is a ternary complex of recA protein, polynucleotide, and ATPyS, although we have not measured bound DNA directly. Others have shown that DNA binds tightly to recA protein in the presence of ATPyS (2, 35).
There is evidence of three sorts that ATPyS is bound without hydrolysis or other covalent modification. 1) Direct examination by polyethyleneimine chromatography of material retained by the fiiter showed that more than 90% of the radioactivity was in ATPyS (data not shown). 2) We detect trapping of either ring-labeled ['*C]ATPyS or thiophosphatelabeled [35S]ATPyS. 3) ATPyS hydrolysis is so slow that little could occur during the time in which complex formation is assayed.
The Stoichiometry of ATPyS Binding-The binding of ATPyS to recA protein saturates at high concentrations of ATPyS, as shown below (Fig. 4). To determine the stoichi-' ' C. Roberts and J. W. Roberts, unpublished results. ometry of bound ATPyS and recA protein, we measured ATPyS bound with increasing concentrations of recA protein, using a saturating concentration of ATPyS (Fig. 3). The slope of this curve gives 0.9 molecules of ATPyS bound per recA protein monomer. It is a reasonable presumption that each monomer has one binding site for ATPyS (and for other NTP's), although this measurement obviously is consistent with some different configuration such as two ATP@ molecules bound to half of the recA monomers.
Both DNA and oligonucleotides such as (dA)16 support the binding of ATP@ to recA protein (Fig. 4) of ATPyS bound a t saturation in the presence of this oligonucleotide is identical with that with single-stranded DNA, so that the stoichiometry is an intrinsic property of the recA polypeptide. However, a higher concentration of ATPyS is required to saturate the binding supported by (dA)lG, suggesting that the oligonucleotide interacts with recA protein less efficiently than does DNA.
Stimulation of Proteolytic Activity of recA Protein by Oligonucleotides-The proteolytic activity of recA protein toward repressors, like its ATPase and ATPyS-binding activities, requires polynucleotide. Both long denatured DNA molecules, such as phage A DNA, and small oligonucleotides satisfy the requirement (4). In reactions using ATPyS, the maximum rate of repressor cleavage occurs at a concentration of DNA that is proportional to the recA protein concentration, suggesting that a stoichiometric complex between these components is the active complex in repressor cleavage (4). Higher than optimal concentrations of single-stranded h DNA inhibit cleavage, as shown in the experiment of Fig. 5. We have found two differences between the activity of DNA and the activity of a series of deoxyadenosine oligonucleotides. 1) Small oligonucleotides do not inhibit repressor cleavage at high concentration.
2) The smaller oligonucleotides promote cleavage only at much higher concentrations than are required of DNA, suggesting that they do not interact stoichiometrically with recA protein in these conditions. Fig. 6 shows that (dA)9, (dA)16, and (dA)19-24 are similar to h DNA in promoting repressor cleavage a t low concentrations, and thus presumably also bind efficiently to recA protein. In contrast, a higher concentration of (dA)* is required to saturate the cleavage rate, and (dA), shows only slight activity at much higher concentration. The oligonucleotide (dT)g is also active at somewhat higher concentrations than single-stranded DNA.
Although (dA) 19-24 inhibits cleavage at higher concentrations, (dA)16 and smaller oligonucleotides do not. This result shows that the inhibition of cleavage by high concentrations of DNA is separable from the stimulation, and is thus probably an interaction distinct from that which stimulates repressor cleavage. If recA protein is preincubated with a saturating concentration of (dA),6 in the presence of ATPyS, and excess single-stranded DNA is then added, repressor cleavage is inhibited (data not shown); thus, the site or sites responsible for inhibition are not irreversibly bound by excess oligonucleotide.

DISCUSSION
We have shown that recA protein binds ATPyS tightly in the presence of DNA, probably in a ternary complex of all three components. Presumably, an equivalent complex forms with natural triphosphates such as ATP before they are hydrolyzed. We have inferred that binding of recA protein into this initial complex, without hydrolysis of the NTP, invokes its proteolytic activity toward repressors. ATPyS stimulates recA protein to cleave repressors more efficiently than does ATP (4, 34), yet the rate of ATPyS hydrolysis by recA protein is several thousand-fold less than the rate of ATP hydrolysis. Furthermore, less than one ATPyS molecule is hydrolyzed for each repressor monomer cleaved, indicating that ATPyS hydrolysis is not directly coupled to repressor cleavage. ATPyS also stimulates recA protein to bind and unwind duplex DNA (18), a reaction that may represent the initiation of DNA strand exchange; we presume that ATPyS (or NTP) hydrolysis also is not essential to this initial reaction. Since ATPyS strongly inhibits both ATP hydrolysis and the completion of strand exchange catalyzed by recA protein, ATP hydrolysis probably is required for subsequent steps in strand exchange that involve the alignment of homologous sequences during strand pairing.
Besides their response to ATPyS, a second property common to the repressor cleavage reaction and the unwinding of duplex DNA by recA protein is that oligonucleotides satisfy the polynucleotide requirement for both (4,18). The affinity of recA protein for longer oligonucleotides is apparently high, because the longer deoxyadenosine oligonucleotides and single-stranded h DNA are active at similar low concentrations.
The activity of oligonucleotides suggests that these reactions may not require the extended polymerization of recA protein that occurs on a long polynucleotide strand (36).
Excess single-stranded DNA inhibits both repressor cleavage and DNA strand exchange (4, l), suggesting that the target for inhibition is recA protein, not h repressor. One explanation of this inhibition follows from the observation that binding of recA protein to (limiting) single-stranded DNA in the presence of NTP activates it to melt duplex DNA: excess single-stranded DNA could inhibit both strand exchange and cleavage by occupying secondary sites that would otherwise bind duplex DNA. Cleavage could be inhibited in several ways: because a recA protein-DNA network impenetrable by repressor is formed, or the cleavage site is directly obscured, or a structural change is induced in the enzyme. Smaller oligonucleotides might not inhibit at high concentrations either because they bind too weakly to occupy the secondary sites, or because they cannot form an extended network of recA protein and DNA.
Because an excess of smaller oligonucleotides does not inhibit repressor cleavage, the rate of repressor cleavage is proportional to the concentration of recA protein in these conditions. Thus it is convenient to use oligonucleotides to assay the cleavage activity of recA protein, as we have done in the experiment of Fig. 1.
Activation of recA protein to cleave repressors in vivo occurs when cellular DNA is damaged or its replication is interrupted, for example by ultraviolet irradiation. We have suggested that one pathway of activation is the binding of recA protein to single-stranded DNA in gaps that result from these treatments (4). Any single-stranded polynucleotide appears to promote repressor cleavage in vitro, even though the different activities of (dT)9 and the dA series suggest that there is some effect of base composition. We interpret this generalized activity of polynucleotide to represent the ability of recA protein to interact with any sequence of exposed single-stranded chromosomal DNA, simultaneously invoking repressor cleavage activity and engaging this DNA in strand exchange. The finding that small oligonucleotides also activate repressor cleavage is consistent with suggestions that degradation fragments of damaged DNA (37) also could provide a pathway of recA activation.