Epitopes and Active Sites of the RecA Protein*

The RecA protein is indispensable for homologous genetic recombination in Escherichia coli. This protein alone promotes the ATP-dependent formation of ho- mologous joint molecules and their processing in u&o. Through the use of a set of anti-RecA protein mouse monoclonal IgGs, we have been attempting to divide the whole process into elementary steps to determine the basic functions of the protein. In order to correlate the basic functions with the active sites on the recA polypeptide, we located the epitopes for the anti-RecA protein-IgGs on the recA polypeptide by means of im- munoblotting experiments and an enzyme-linked immunosorbent assay involving isolated proteolytic poly- peptides or synthetic ones derived from various regions of the recA polypeptide. The epitopes for anti-RecA protein-IgGs ARM321 and ARM414, both of which are shown to inhibit the DNA-dependent ATP hydrol- ysis and the formation of homologous joints by the RecA protein, were found to be located between Thrss and Glu”’ and between G~u’~~ and Lys”‘, respectively, on the RecA polypeptide.

IgGs, we have been attempting to divide the whole process into elementary steps to determine the basic functions of the protein.
In order to correlate the basic functions with the active sites on the recA polypeptide, we located the epitopes for the anti-RecA protein-IgGs on the recA polypeptide by means of immunoblotting experiments and an enzyme-linked immunosorbent assay involving isolated proteolytic polypeptides or synthetic ones derived from various regions of the recA polypeptide.
The epitopes for anti-RecA protein-IgGs ARM321 and ARM414, both of which are shown to inhibit the DNA-dependent ATP hydrolysis and the formation of homologous joints by the RecA protein, were found to be located between Thrss and Glu"' and between G~u'~~ and Lys"', respectively, on the RecA polypeptide.
IgG ARM193 had been shown to interfere with the protein-protein interaction between two RecA protein molecules, and ARM191 had been suggested to inhibit the binding of doublestranded DNA to the RecA protein.
The epitopes for ARM193 and ARM191 were found to be located in a -go-amino acid region at the C terminus. These results suggest the locations of the active sites and a functional core on the RecA polypeptide.
"Homologous pairing," which includes searching for mutually homologous sequences between parental DNA molecules, and the formation of nascent homologous joints through intermolecular base pairing are the most critical and mysterious steps in homologous genetic recombination in vivo. The RecA protein was the first example of a class of proteins which promote homologous pairing of single-stranded and double-stranded DNAs in vitro (McEntee et al., 1979;Shibata et al., 1979), and has been studied extensively to understand the molecular mechanism underlying homologous pairing. Although the RecA protein is a relatively small single polypeptide species (-40 kDa) consisting of 352 amino acid residues (Horii et al., 1980;Sancar et al., 1980), it is involved in not only homologous pairing but also the processing of the nascent homologous joints formed through the former reac-* This research was supported by a grant for Life Sciences from the Agency of Science and Technology, Japan. The costs of publication of this article were defrayed in part by the payment of page charges.
This article must therefore be hereby marked "aduertise- tion. When double-stranded DNA has an end at or near the joint, the homologous pairing is followed by "strand exchange." Strand exchange is the unidirectional exchange of a parental strand of double-stranded DNA for an incoming strand of single-stranded DNA, and heteroduplex joints formed through homologous pairing are stabilized by this process (Cox et al., 1981;Kahn et al., 1981). When doublestranded DNA is of the closed-circle form and single-stranded DNA is much shorter than the former DNA, the joints are dissociated with the same strand polarity as that in the strand exchange: i.e. 5' to 3' with respect to the displaced strand (Shibata et al., 1982a, 198213;Wu et al., 1982). The formation and processing of the joint molecules includes various elementary processes which were identified in studies on various DNA substrates and ones involving analogues of substrates or cofactors, mutant RecA proteins and anti-RecA protein monoclonal IgGs. Other than the processes described above, the RecA protein removes the secondary structure of singlestranded DNA through the formation of a filamentous complex in the presence of ATP Flory et al., 1984), hydrolyzes ATP in a DNA-dependent manner (Ogawa et al., 1979;Roberts et al., 1979), extensively unwinds the double helix through a progressive reaction ("processive unwinding": Cunningham et al., 1979a("processive unwinding": Cunningham et al., , 1979bOhtani et al., 1982;Iwabuchi et al., 1983), and cleaves SOS-repressor proteins (Roberts et al., 1978).
In general, monoclonal IgGs are supposed to be useful for studying elementary functions involved in a complex reaction process promoted by a protein and for locating the active sites on the protein.
We constructed a set of mouse hybridoma clones which produce anti-RecA protein-IgGs (Makino et al., 1985) and have been characterizing them as to their inhibitory effects on the elementary functions of the RecA protein (Makino et al., 1985(Makino et al., , 1987). An anti-RecA protein-IgG, ARM193, severely inhibited the RecA protein-promoted processive unwinding and strand exchange, but affected the homologous pairing by the same protein only a little (Makino et al., 1987). Our previous studies suggested that ARM193 interferes with protein-protein interactions between RecA protein molecules (Makino et al., 1987;Ikawa et al., 1989). An IgG, ARM191, was shown to inhibit both homologous pairing and superhelical double-stranded DNA-dependent ATP hydrolysis by the RecA protein, but not the single-stranded DNA-dependent ATP hydrolysis by the protein (Makino et al., 1985 (Makino et al., 1985).
In the present study, we located the epitopes for the four anti-

Proteins and Antibodies
The RecA protein used in this study was Fraction V, which was prepared as described previously (Shibata et al., 1981(Shibata et al., , 1983. Anti-RecA protein mouse monoclonal IgGs were purified by affinity column chromatography on protein A-Sepharose from mouse ascitic fluid in which hybridoma cells had been grown. The procedure was described in detail previously (Makino et al., 1985; see also Nakagawa et al., 1988). The molecular masses of the recA polypeptide and the IgGs were assumed to be 40 and 160 kDa (Makino et al., 1985), respectively.
The amounts of the RecA protein and IgGs were estimated as described previously (Shibata et al., 1981;Makino et al., 1985). The amounts of the synthetic polypeptides and a fragment of the recA polypeptide (Fv 31-8-7) were estimated by their amino acid composition analysis. Horse heart myoglobin (16.9 kDa) and its fragments (2.5, 6.  (Fling and Gregerson, 1986). Immunoblotting experiments, which included the transfer of the polypeptide from the polyacrylamide gel to a membrane filter, blocking, treatment with a peroxidase-labeled anti-mouse IgG and coloring reactions, were performed as described previously (Nakagawa et al., 1988). Unless otherwise stated, in the present study we used a polyvinylidene difluoride membrane (Immobilon, pore size 0.22 pm; Millipore Ltd.) instead of the nitrocellulose filter used in the previous study. The polypeptide fragments were transferred from the gel to the polyvinylidene difluoride membrane at 36 V/8 cm for -4 h using 50 mM Tris, 0.19 M glycine, and 20% methanol in a Trans-Blot cell (Bio-Rad).
For estimation of the molecular masses of the polypeptide fragments from their mobilities on gel electrophoresis, the fragments were detected by an immunoblotting experiment and, if a sufficient amount was available, also by staining with Coomassie Brilliant Blue or silver stain (Merril et al., 1980). Then, the amino acid composition was analyzed with an automatic amino acid analyzer (Hitachi model 835).

RESULTS
Fragmentation of the RecA Polypeptide and Immunoblotting Analysis of the Fragments--We examined various proteolytic fragments of the recA polypeptide for their affinity toward anti-RecA protein-IgGs by means of immunoblotting experiments, followed by mapping of the fragments on the recA polypeptide.
In this study, we used for anti-RecA protein-IgGs, ARM321, ARM414, ARM191, and ARM193. The binding of these anti-RecA protein monoclonal IgGs to the immobilized RecA protein was compared by ELISA ( Fig. 1). Binding of ARM193 to the recA polypeptide was slightly weaker than that of ARM321, ARM414, and ARM191. The latter three anti-RecA protein-IgGs showed almost the same affinity toward the immobilized RecA protein (Fig. 1). The RecA protein that had been partially digested with an endopeptidase (endoproteinase Arg-C) was electrophoresed through polyacrylamide gel and then subjected to immunoblotting analysis using the anti-RecA protein-IgGs (Fig. 2). The profiles of the bands of the fragments of the RecA polypeptide which reacted with ARM321 and ARM414 were clearly different from each other and from that with ARM193 or ARM191.
This indicates that although the inhibitory effects of ARM321 and ARM414 on the in vitro activities of the RecA protein were shown to be identical (Makino et al., 1985), the epitopes for these IgGs are different from each other, and from those recognized by ARM193 and ARM191. The profile of the bands of fragments recognized by ARM193 and that in the case of ARM191 appeared to be similar but not identical. Some fragments were detected with one of them but not with the other, and vice versa (Fig. 2).
Fractionation and Mapping of the Fragments of the RecA Polypeptide which Reacted with anti-RecA Protein-IgGs-Then, we examined the binding of the anti-RecA protein-IgGs, ARM321, ARM414, ARM193 and ARM191, to the isolated fragments. The RecA protein was treated with various endopeptidases and the products were fractionated by HPLC. The material in each peak of the fragments on the chromatogram was collected and examined as to its reactivity with the IgGs by polyacrylamide gel electrophoresis and immunoblotting analyses. When more than one species of polypeptide was detected, the samples were repeatedly purified by HPLC until only a single peak was detected in the chromatogram. The purified samples contained a single species of polypeptide which was detected on immunoblotting and chemical staining. Each panel in Fig. 3 shows the profile on immunoblotting analysis of the purified proteolytic fragment which was the smallest in size among the fragments recognized by each of the IgGs tested. Four or five amino acid residues from the N terminus of each purified fragment were sequenced (Table I, fourth column) and located on the recA polypeptide (Table I, fifth column). The molecular masses of the fragments were determined by polyacrylamide gel electrophoresis under denaturing conditions using marker polypeptides (see "Materials and Methods"; Table I, third column). The C terminus of each fragment was deduced from its size and the specificity of the endopeptidase used for its preparation (Table I, sixth column). Thus, the position of the N terminus of each fragment was precisely determined, but that of the C terminus may not be accurate. The results of immunoblotting analyses, of which three cases are shown in Fig. 3, are summarized in Fig. 4. Considering the overlapping of fragments which did or not react with each IgG, we tentatively located the epitope for ARM321 between Hisg7 and G1ulZ3 of the recA polypeptide (residues are numbered from the N terminus of the recA polypeptide), that for ARM414 between GluZS3 and PhezS5, and those for ARM193 and ARM191 between Phe*" and G1u314. GhY was assumed to be the C terminus of Fv 31-8-7, which was found to be the smallest fragment recognized by ARM191 and ARM193 on electrophoresis under denaturing conditions. However, as described later, the true C terminus of this fragment was not G1u314 but G1uSs7. Antibody ARM321 showed weak cross-reaction with other fragments which did not overlap with the above region, but it showed the strongest cross-reaction with Fl 33-5 (ThrB9-Lys15*) and Fv 26-5-2 (His97-G1u"3) by immunoblotting experiment. As discussed in the following section, this antibody showed slight nonspecific binding to another portion of the RecA polypeptide.
Binding of Antibodies ARM321 and ARM414 to Synthetic  Thr-X-Ala-Phe F133-6 Lysyl endopeptidase 21.5 Ala-Glu-Ile-Glu   of the binding of ARM321 to P414 was apparently l/200 of that to P321. These characteristics of ARM321 appear to explain the weak cross-reaction with fragments which do not include residues 89-127. ELBA indicated that both antibodies ARM321 and ARM414 did not react with a C-terminal fragment including the region from Phez60 to GltY (Fv 31-8-7; Fig. 7).
The specific binding of antibody ARM321 to synthetic polypeptide P321, and that of ARM414 to P414 in solution were further tested by means of competition experiments. It had been shown that both ARM321 and ARM414 inhibit the single-stranded DNA-dependent hydrolysis of ATP by the RecA protein, when the RecA protein had been previously incubated with either of these antibodies (Makino et al. 1985 ase activity by itself (data not shown). As shown in Fig. 6, P321 and P414 specifically prevented the inhibition of ATPase activity of the RecA protein by ARM321 and ARM414, respectively.
In either case, the inhibition by the IgG was completely prevented when the synthetic polypeptide was added at 4-fold over the RecA protein. This indicates that the strength of the binding of either IgG to the relevant synthetic polypeptide and that to the whole recA polypeptide are of the same order of magnitude.
Based on the specific binding observed on ELISA and the specific competition described above, we conclude that the epitopes for ARM321 and ARM414 reside between Thrs9 and G~u'*~, and between G~u*~~ and LYS~~'~, respectively (Fig. 8).
Detailed Localization of the Epitopes for Antibodies ARM193 and ARMlSl-Judging from its behavior on gel electrophoresis, fragment Fv 31-8-7 was the smallest proteolytic fragment of the recA polypeptide which reacted with ARM191 or ARM193. The N terminus of Fv 31-8-7 was definitely located on amino acid sequence analysis, but the C terminus of the fragment was tentatively assigned as described in the previous section. This fragment size (65 amino acid residues) was close to the limit of the chemical synthesis in experience. Then, we selected another approach. We obtained an amount of Fv 31-8-7 by partial digestion of the RecA protein with V8 proteinase, followed by purification by HPLC. Both ARM191 and ARM193 showed cross-reaction with Fv 31-8-7 on ELISA (Fig. 7). The affinities of ARM191 and ARM193 toward Fv 31-8-7 were almost the same as and only slightly weaker than that of these IgGs toward the whole recA polypeptide, respectively (compare Fig. 7 with Fig. 1). As described in the preceding section, neither antibody ARM191 nor ARM193 reacted with synthetic fragments P321 and P414 (Fig. 5). These results indicate that both ARM191 and ARM193 specifically reacted with fragment Fv 31-8-7.
As described above, based on the apparent molecular mass, we tentatively located the C terminus of fragment Fv 31-8-7 at Glu314 (Table I). We precisely determined the site of the C terminus of fragment Fv 31-8-7 on the recA polypeptide by amino acid composition analysis. Amino acid composition analysis revealed that the above assignment was wrong. If fragment Fv 31-8-7 was assumed to have G1u314 at its C terminus, this fragment should have 1 Ser and 1 Pro, but the observed numbers were both 4 (assuming 8 Gly residues in this region). Other than these 2 residues, the expected numbers of Val(4 uerms 2), Asp plus Asn (12 uersu.s 5), and others did not agree with those observed. Since fragment Fv 31-8-7 was prepared through the use of V8 protease, the C-terminal residue of this fragment is likely to be Glu, and between G1u314 and the C terminus of the recA polypeptide, residues 318,320, 325, 343, 347, and 350 are Glu. Among them, G1u318, G1u3*", and G~L?*~ can be clearly eliminated as candidates for the Cterminal residue of this fragment, for similar reasons. When the C terminus of the fragment was assumed to be G~u~~~, the expected numbers of amino acid residues (assuming 9 Gly residues) and those observed showed the best agreement among the four candidates, including the C terminus of the recA polypeptide (Table II). The estimated molecular mass of the region from amino acids 260 and 347 is 1.6 times larger than the molecular mass observed upon gel electrophoresis. Preliminary analysis of the C-terminal amino acid residues by treatment of fragment Fv 31-8-7 with carboxypeptidase W (from wheat; Seikagaku Kogyo Co.) revealed that the Cterminal amino acid was Glu, and that Ala, Val, and Gly were released on the treatment. This result is only consistent with GIu347 as the C terminus of Fv 31-8-7 among candidate sites GIIY+'~, GltY, Glt?', and Phe352. Thus, we conclude that the epitopes for ARM191 and ARM193 reside between PheZ6' and G1u347 (Fig. 8).
The polypeptide fragment (Fv 31-8-7) was further digested with V8 proteinase, followed by fractionation by HPLC (Cosmosil 5C4-300). The materials in all detected peaks were subjected to immunoblotting after direct blotting of the samples. We found that none of them tested so far gave a positive signal for the reaction with either ARM191 or ARM193. This result may suggest that more than two discontinuous sites of Fv 31-8-7 are required for the binding of ARM191 or ARM193. However, it is also likely that subfragments of Fv 31-8-7 only weakly bound to the filter for the assay and thus were washed out during this assay. DISCUSSION It is likely that active sites distant from an epitope on the recA polypeptide are also affected by the binding of an IgG, but the localization of the epitopes for anti-RecA protein monoclonal IgGs provided some insight as to the loci of the active sites on the recA polypeptide.
The epitopes for anti-RecA protein monoclonal IgGs ARM321 and ARM414 were found to be located on the recA polypeptide between Thrs9 and G~u'*~ and between Glt? and Lys256, respectively (Fig. 8). The epitopes for IgGs ARM193 and ARM191 were found to be located in a go-amino acid  Numbers under the thick line, the number of amino acid residues. X amino acid residues which are not conserved in the RecA protein of P. aeruginosu (Sano and Kageyama, 1987).
region at the C terminus of the recA polypeptide (Fig. 8). It is unlikely that the epitopes for ARM193 and ARM191 are identical, since (i) the inhibitory effects of these IgGs are different from each other (Makino et al., 1985) and (ii) the profiles on immunoblotting analysis of these IgGs using the partial proteolytic digests of the recA polypeptide were not identical (Fig. 2).
IgG ARM321 was shown to inhibit single-or superhelical double-stranded DNA-dependent ATP hydrolysis and homologous pairing by the RecA protein (Makino et al., 1985). Our preliminary experiment revealed that this IgG inhibits the binding of ATP to the RecA protein in the activation cycle of the RecA protein in the homologous pairing and also the binding of single-stranded DNA to the protein.* It was shown that the modification of Cys'i6 does not affect the binding of ATP to the RecA protein but causes the inhibition of ATP hydrolysis (Kuramitsu et al., 1984). These observations suggest that some residues between amino acids 89 and 129 are closely related to the ATPase functions, which include the binding of ATP and catalysis of its hydrolysis. Like ARM321, IgG ARM414 was shown to inhibit singleor superhelical double-stranded DNA-dependent ATP hydrolysis and homologous pairing by the RecA protein (Makino et al., 1985). Our preliminary experiment revealed that this IgG also inhibits the binding of ATP to the RecA protein and the binding of DNA to the protein.' The amino acid sequence around the epitope for antibody ARM414 is highly conserved between the Escherichia coli RecA protein and the Pseudomonas aeruginosa RecA protein (Sano and Kageyama, 1987;see Fig. 8). This suggests that this region plays an essential role in the activities of the RecA protein. It was shown that Ty? is the site of affinity labeling by 8-azidoadenosine 5'-triphosphate (Knight and McEntee, 198513) and 5'-p-fluorosulfonylbenzoyladenosine (Knight and McEntee, 1985a), while as described above, IgG ARM414 inhibits the binding of ATP to the RecA protein. These observations suggest that the region around the epitope for ARM414 is also closely related to the ATPase functions.
It was suggested that the RecA protein has two kinds of DNA-binding site: a primary site, which is essential for all functions of the RecA protein, and a secondary site, which is required only for homologous pairing (Ikawa et al., 1989). Both ARM321 and ARM414 appear to interfere with the binding of single-stranded DNA to the free RecA protein. Unlike ARM321 and ARM414, both ARM191 and ARM193 did not affect the single-stranded DNA-dependent ATP hydrolysis by the RecA protein. These results suggest that the amino acid residues involved in the primary DNA-binding site are located near the epitopes for ARM321 and ARM414. It should also be considered that the binding of these IgGs changes the higher order structure of the RecA protein, preventing the binding of DNA and/or ATP to the protein.
Even in the presence of excess ARM193, the RecA protein promotes single-stranded DNA-dependent ATP hydrolysis and ATP-dependent homologous pairing (Makino et al., 1987). Thus, the binding of ARM193 to the RecA protein