RecA Protein-promoted Cleavage of LexA Repressor in the Presence of ADP and Structural Analogues of Inorganic Phosphate, the Fluoride Complexes of Aluminum and Beryllium*

Complexes formed from A13+ or Be2+ and fluoride inhibit the single-stranded DNA-dependent ATPase activity of RecA protein. In contrast, poly(dT)-RecA-ADP complexes, which are inactive for cleavage of LexA protein, become fully active in the presence of AlF; or BeFS ions. These data suggest that fluoride complexes of aluminum and beryllium (called herein X) convert RecA-ADP complexes, which bind weakly to single-stranded DNA, into RecA-ADP-X complexes, which bind tightly to single-stranded DNA, the ADP-X moiety behaving as a nonhydrolyzable analogue of ATP. We propose that AlF; and BeF; ions act as analogues of inorganic phosphate by binding to the site of the y-phosphate of ATP on RecA-ADP complexes, hence mimicking the single-stranded DNA-RecA-ADP-PI transition state. We conclude that the elementary reaction that switches RecA protein from a high affinity single-stranded DNA binding state to a low affinity single-stranded DNA binding state is not ATP hydrolysis per 8e but Pi release. The RecA Escherichia coli essential in genetic recombination and in the induction of at least 17 genes which are negatively controlled by the LexA protein 1985). The latter activity of the RecA protein results from its capacity to promote cleavage of the LexA repressor RecA-promoted proteolysis is pg/ml bovine serum albumin, 2 pg/ml poly(dT), 0.5 mM ATP or ADP or dATP, 10 pM LexA protein, and 1 p~ RecA protein. LexA cleavage was determined by separating and staining the proteins in a 15% acrylamide-sodium dodecyl sulfate gel (Moreau and Roberts, 1984). The autodigestion of LexA protein was induced by a 10-fold dilu- tion of LexA protein at 20 pM in 100 mM Tris-HCI, pH 9.4, and 200 mM NaCl (Slilaty and Little, 1987); approximately half of the protein was cleaved after 2 h of incubation. All incubations were performed at 37 "C.

of bacteriophage repressors, such as the X c l repressor (Roberts and ; all these proteins share structural homologies (Nohmi et al., 1988). RecA-promoted proteolysis is not constitutively expressed in bacteria; RecA protein must be activated in order to mediate cleavage of proteins (Moreau et al., 1980a, Quillardet et al., 1982. The activation of RecA protein occurs when DNA replication is perturbed for example by the presence of DNA lesions generated by carcinogenic agents (Moreau et al., 1980b). However, it is still unclear exactly what constitutes the inducing signal for the activation of the RecA protein.
The biochemical evidence suggests that RecA protein is activated to cleave repressors when it forms a ternary complex with a polynucleotide, such as poly(dT), and a nucleoside *This work was supported in part by Grant BI16-145-F from Euratom, Association pour la Recherche sur le Cancer, Ligue Nationale contre le Cancer, and Fondation pour la Recherche Medicale. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. triphosphate, such as ATP or dATP (Craig and Roberts, 1980;Phizicky and Roberts, 1981;Cotterill et al., 1982). It is generally thought that the requirement in vitro of a polynucleotide for proteolysis reflects in vivo the generation of ssDNA' regions by inducing agents. The role of the second cofactor required for the activation of the RecA protein, the nucleoside triphosphate, is less clear. RecA protein exhibits a polynucleotide-dependent ATPase activity (Roberts et al., 1979). However, binding of ATP but not its hydrolysis is required for proteolysis since ATP-y-S, a nonhydrolyzable analogue of ATP, can substitute for ATP in RecA-dependent proteolytic processes (Craig and Roberts, 1981;Slilaty and Little, 1987). The hydrolysis of ATP to ADP and inorganic phosphate is, however, intricately coupled to the activity of RecA protein since ADP inhibits RecA-dependent processes (McEntee et al., 1981;Cox et al., 1983;Menetski and Kowalczykowski, 1987). It has been suggested, therefore, that RecA protein exists in two different states: a high affinity ssDNA binding state when ATP is bound to RecA protein, and a low affinity ssDNA binding state when ADP is bound to RecA protein (Menetski et al., 1988).
It is an important issue to understand how the elementary reactions of ATP hydrolysis, including the release of Pi, regulate the activities of the RecA protein by switching it from a high to a low affinity ssDNA binding state. In order to investigate these processes, we studied the effects of structural analogues of phosphate, the fluoride complexes of aluminum and beryllium, on two RecA-dependent reactions, namely ATP hydrolysis and cleavage of LexA protein. It has been proposed that AlF; and BeF; complex ions, which have the same tetrahedral geometry and bond length as inorganic phosphate, act by mimicking the y-phosphate of ATP or GTP, thereby converting various enzymes from a conformation normally observed with GDP or ADP to a conformation observed with nucleoside triphosphates (Sternweis and Gilman, 1982;Bigay et al., 1985;Lange et al., 1986;Robinson et al., 1986;Bigay et al., 1987;Carlier et al., 1988). These analogues may therefore be useful to probe the catalytic mechanism of nucleotidases involved in energy transduction.

MATERIALS AND METHODS
RecA and LexA proteins were purified as previously described (Moreau, 1987). RecA protein was stored in 20 mM Tris-HC1, pH 7.5, 0.1 mM EDTA, 1 mM dithiothreitol, and 10% (w/v) glycerol at -80 "C; its concentration was determined assuming an A% of 5.9. LexA protein was stored in 10 mM Pipes, pH 7.0, 0.1 mM EDTA, 1 mM dithiothreitol, 200 mM NaCl, and 10% (w/v) glycerol at -80 "C; its concentration was determined assuming an A% of 3.2. Poly(dT) was obtained from Pharmacia LKB Biotechnology Inc.; it was assumed that a solution of 50 pg/ml has an absorbance at 260 nm of 1.0. ATP The abbreviations used are: ssDNA, single-stranded DNA; Pipes, 1,4-piperazinediethanesulfonic acid.
(special quality) and ADP were from Boehringer Mannheim, dATP from Pharmacia LKB Biotechnology Inc., and [T-~'P]ATP from Amersham Corp. Bovine serum albumin (RNase-DNase free) was from Pharmacia LKB Biotechnology Inc. NaF (analytical grade) was from Merck, aluminum nitrate (gold label) and beryllium sulfate (gold label) from Aldrich.
Counts/min were corrected for by subtraction of counts/min found in a reaction in which RecA protein was omitted (approximately 1% of total radioactivity).
LexA cleavage was determined by separating and staining the proteins in a 15% acrylamide-sodium dodecyl sulfate gel (Moreau and Roberts, 1984).
The autodigestion of LexA protein was induced by a 10-fold dilution of LexA protein at 20 p M in 100 mM Tris-HCI, pH 9.4, and 200 mM NaCl (Slilaty and Little, 1987); approximately half of the protein was cleaved after 2 h of incubation.
All incubations were performed at 37 "C.

Inhibition of RecA Protein ATPase Activity by Fluoride
Complexes of Aluminum and Beryllium-It has been previously shown that in the presence of 1-10 mM fluoride, A13+ ions at micromolar concentrations inhibit the glucose-6-phosphatase (Lange et al., 1986) and the (Na + K)-ATPase (Robinson et al., 1986). We found that in the presence of 10 mM NaF, 17 p~ aluminum nitrate strongly inhibited the ATPase activity of the FkcA protein (Table I). NaF alone weakly inhibited RecA protein activity, whereas aluminum nitrate was totally ineffective in the absence of NaF. The inhibiting effect of NaF was not unexpected. Although plastic tubes and pipetting devices were used at all times to prevent etching by fluoride of aluminum from glassware, it is known that aluminum fluosilicates are common impurities in commercial preparations of NaF and ATP (Sternweis and Gilman, 1982).
By using a reaction mixture containing 1 I M RecA protein and 2.5 mM ATP, the rate of ATP hydrolysis was linear for 90 min, at which time approximately 50% ATP was hydrolyzed to ADP and Pi (Fig. 1). The rate of ATP hydrolysis then decreased rapidly as a consequence, probably, of the accumulation of ADP (Cox et at., 1983). In the presence of 10 mM NaF and 10 p~ aluminum nitrate, the hydrolysis of ATP  proceeded at a normal rate for 10 min and then was gradually inhibited with time; the rate of ATP hydrolysis was 90% inhibited after 60 min of incubation. The addition of 50 I.~M aluminum nitrate produced a complete inhibition after 10 min of incubation, while approximately 100 p~ ATP (4% of the input concentration) was hydrolyzed. These results indicate that fluoride complexes of aluminum, probably as AlF; (Lange et al., 1986), prevent the turnover of ATP on RecA protein.
The nonlinear kinetics of ATP hydrolysis observed in the presence of AIF; are consistent with a slow binding of the fluoride complexes to the RecA protein as compared to the other reactions involved in the hydrolysis of ATP.
In the case of activation of GTP-binding proteins, it has been shown that among various metal ions other than A13+, only Be2+ can form active complexes with fluoride, mainly as BeF3 (Sternweis and Gilman, 1982;Bigay et al., 1987). Indeed, we found that Be2' ions inhibit RecA protein ATPase activity in the presence of NaF (Fig. 21, but the extent of inhibition observed over the range of concentrations of beryllium sulfate tested (0-100 p M ) indicates that BeF: binds to RecA protein more weakly than does AlF;. AlF; Inhibits ATP Hydrolysis but Stimulates LexA Protein Cleavage-Does A I R inhibit all RecA protein activities like ADP, which decreases the binding affinity of RecA protein for polynucleotides? We can answer this question by testing for another activity of the RecA protein which can be expressed in the absence of ATP hydrolysis but which nevertheless requires the formation of a stable complex of RecA protein with ssDNA, i.e. cleavage of LexA protein (Slilaty and Little, 1987). Cleavage conditions differed from the standard ATPase assay conditions that were used for the kinetic measurements described above essentially in that salt concentration was increased in order to reduce the rate of LexA cleavage. As seen in Fig. 3, in the absence of aluminum fluoride, ATP hydrolysis and LexA cleavage proceeded respectively to 34% and approximately 50% completion in 15 min; both reactions were then essentially inhibited probably as a consequence of the accumulation of ADP (Fig. 3). In contrast, in the presence of AlF;, whereas ATP hydrolysis was rapidly inhibited, cleav-age of LexA protein could proceed to completion within 45 min of incubation. We can therefore conclude that in the presence of A1F; and ATP, RecA protein binds to poly(dT) and is activated to promote LexA protein cleavage although it no longer hydrolyzes ATP.
These results were not indicative of the nature of the nucleotide bound to the RecA molecules which promoted LexA cleavage. The nucleotide site on these RecA molecules could be either occupied by AlF; that could induce a ATPlike conformation of the protein, or it could be occupied by ATP whereas AlF; would inhibit ATP hydrolysis through its binding to another site, or it could be occupied by a ADP-AlFh complex that would behave as a nonhydrolyzable analogue of ATP. In order to test these different hypotheses, the cleavage of LexA protein was assayed in the presence of AlF; or ADP, or both.

Poly(dT)-RecA-ADP Complexes Cleave the L e d Protein in the Presence of A1F;"
No cleavage of LexA protein occurred in the presence of RecA protein and poly(dT) upon the addition either of ADP (Fig. 4) or of AlF; (10 mM NaF plus 10 p~ Al(NO&) (data not shown). Although dADP reduces the apparent affinity of X RecA protein for ssDNA less than ADP (Menetski et aZ., 1988), no cleavage was observed when dADP substituted for ADP (Fig. 4). In contrast, cleavage of the LexA protein reached nearly 100% completion within 60 min of incubation in the presence of both ADP and A1F; as well as in the presence of ATP and AlFh. Cleavage was slightly more efficient in the presence of dADP and AlF; than in the presence of ADP and AlF;, a result that is in good agreement with the hypothesis that dADP binds more tightly to RecA protein than does ADP (Kowalczykowski, 1986). No cleavage of LexA protein occurred, however, in the presence of AlF; and dADP when poly(dT) was omitted from the reaction mixture (Fig. 5), indicating that AlF; does not release RecA protein from the requirement of a polynucleotide to be active. Taken together, these results indicate that ssDNA-RecA-ADP-AlF; complexes are as active as ssDNA-RecA-ATP complexes for cleavage of the LexA protein. The final extent of the reaction could be even higher in the presence of either ADP or ATP and AIF; than in the presence of ATP alone since AlF; prevents the inhibition due to the accumulation of ADP. protein and 500 pM dADP, the concentrations of aluminum nitrate and NaF required for half-maximal rate of cleavage of LexA protein were approximately 2.5 p~ and 2.5 mM, respectively (Fig. 5). The functional range of NaF concentrations is consistent with AlF; being the active species among the various known complexes of A13' and fluoride (Lange et al., 1986). The rate of cleavage was maximal in the presence of 10 p~ aluminum nitrate, and no inhibition was detected when this concentration was increased up to 200 p~ (Fig. 5, and data not shown). A much higher concentration of beryllium sulfate than that of aluminum nitrate (800 p M uersw 10 pM) was required to support LexA cleavage in the presence of NaF and dADP (Fig. 6), indicating that BeF; has a lower apparent affinity for RecA-dADP complexes than does AlF;, which is in qualitative agreement with the ATPase results mentioned above. No cleavage could be detected in the presence of NaPi at 50 or 100 mM (Fig. 6).

DISCUSSION
We show here that AlF; ions, and to a lesser extent BeF; ions, alter the activities of poly(dT)-RecA-ATP complexes in different ways. On the one hand, they inhibit the hydrolysis of ATP, and on the other hand, they stimulate the cleavage of the LexA protein. Moreover, we found that poly(dT)-RecA-ADP (or dADP) complexes which are totally inefficient for the cleavage of the LexA protein, become fully active in the presence of 10 mM NaF and 10 p~ aluminum nitrate or 800 p~ beryllium sulfate. In contrast, Pi ions at concentrations up to 100 mM do not allow LexA protein cleavage. Taken together, these results may be accounted for by the hypothesis that AlF; and BeF; (called herein X) act as high affinity analogues of H2PO;, converting RecA-ADP complexes, in which RecA protein is in a low ssDNA binding affinity state, into RecA-ADP-X complexes, in which RecA protein is in a high affinity state because the ADP-X moiety behaves as a nonhydrolyzable derivative of ATP. A similar mechanism was previously proposed to explain the effects of fluoride complexes of aluminum and beryllium on various GTP-binding proteins (Sternweis and Gilman, 1982;Bigay et al., 1985;Bigay et al., 1987), a phosphatase (Lange et al., 1986), an ATPase (Robinson et al., 1986), tubulin , and actin (Combeau and Carlier, 1988). Free phosphate having a much weaker affinity for RecA-dADP complexes than BeF; or A1F; may account for the inability to detect exchange between "0-Pi and water during ATP hydrolysis in the presence of 6 mM Pi (cox et al., 1983).
Analogues Rectangles represent the conformation which binds ssDNA tightly, and triangles the conformation which binds ssDNA weakly. RecA-ATP binds more tightly to ssDNA than RecA-ADP, that is the equilibrium association constant LT is larger than LD. ATP hydrolysis occurs via the formation of a transient ssDNA-RecA-ADP-Pi complex ( X is then equivalent to Pi). Pi release is linked to the destabilization of the ssDNA-RecA complex. Analogues of Pi (AlF; or BeF; called herein X ) bind to RecA-ADP with an equilibrium association constant & and to ssDNA-RecA-ADP with a corresponding association constant K'x. Binding of X to RecA protein shifts equilibria toward the highly stable ssDNA-RecA-ADP-X complex, in which X mimics Pi in the transient ssDNA-RecA-ADP-Pi state, i.e. LDX > LD. Since = K'xLD, X has a higher affinity for ssDNA-RecA-ADP than for RecA-ADP ( K ' x > Kx). The known cooperativity involved in the binding of RecA protein to ssDNA is not featured for simplicity. This cooperativity may prevent the dissociation of RecA-ADP subunits from ssDNA as long as enough ATP is available in the medium to allow a rapid exchange of ADP for ATP. was as efficient in the presence of ATP as in the presence of ADP plus AlF;. However, the inhibition of ATP hydrolysis by Pi analogues was not instantaneous. In the presence of 1 pM RecA protein, 2.5 mM ATP, and 50 p M aluminum fluoride, 125 cycles of ATP hydrolysis occurred (5% ATP was hydrolyzed) before all RecA molecules were blocked in poly(dT)-RecA-ADP-X complexes. This indicates that at high ATP concentration, the rate of binding of AlF; was lower than the rate of exchange of ADP for ATP. This exchange probably occurs without the release of RecA protein from ssDNA since dissociation of ssDNA-RecA complexes apparently occurs cooperatively when approximately 50% of the input ATP has been hydrolyzed (Cox et al., 1983).
These results allow us to expand the ssDNA-dependent ATPase cycle of RecA protein proposed by Menetski and Kowalczykowski (1987) (Fig. 7). Upon binding of RecA-ATP complexes to ssDNA, ATP is hydrolyzed. The hydrolysis of ATP to ADP and Pi can be divided into two distinct steps.
First, cleavage of the y-phosphate leads to the kinetic intermediate RecA-ADP-Pi, which binds tightly to ssDNA. Second, Pi release in the medium leads to the RecA-ADP complex, which binds weakly to ssDNA. Exchange of ADP for ATP might occur either in this complex or on RecA-ADP subunits following their dissociation from ssDNA; the choice between these two pathways depends on the relative concentration of ATP and ADP in the medium (Cox et al., 1983;Menetski and Kowalczykowski, 1987). Restoration of the ssDNA-RecA-ATP complex allows then a new cycle of ATP hydrolysis to occur. According to this model, the elementary reaction that triggers a major conformation change of RecA protein leading to a large decrease in its affinity for ssDNA is not hydrolysis of ATP per se but Pi release. However, the ratio of RecA-ADP over RecA-ATP and RecA-ADP-Pi subunits that induce the dissociation of a ssDNA-RecA nucleoprotein filament is unknown at this point (Brenner et al., 1987).
RecA protein ATPase activity may represent another example of a coupled vectorial process (Jencks, 1980), similar to that observed with other enzymes involved in energy or signal transduction. In the case of actomyosin, Pi release is linked to the development of force (Hibberd et al., 1985); with rhodopsin, the regulation of the associated phosphodiesterase is exerted by Pi release from the complex transducin-GDP-Pi (Bigay et al., 1985;Bigay et ai., 1987); with actin and tubulin, Pi release following ATP or GTP hydrolysis is also associated to a structural change causing the destabilization of actin filaments or microtubules (Carlier and Pantaloni, 1988;Carlier et d . , 1988;Combeau and Carlier, 1988).
A number of factors such as ionic conditions, pH, and the presence of proteins such as Escherichia coli single-stranded DNA-binding protein are known to affect RecA protein activities, such as ATP hydrolysis, LexA protein cleavage, and strand exchange (McEntee et al., 1981;Cotterill and Fersht, 1983;Moreau and Roberts, 1984;Chow and Radding, 1985;Kowalczykowski and Krupp, 1987). It will be interesting to determine if these parameters affect the rate of Pi release and, hence, modulate RecA protein activities.