The central aromatic residue in loop L2 of RecA interacts with DNA. Quenching of the fluorescence of a tryptophan reporter inserted in L2 upon binding to DNA.

To determine the role of the central aromatic residue in one of the DNA binding domains in Escherichia coli RecA protein, we have constructed a protein in which a tryptophan fluorescence reporter is inserted in the place of phenylalanine residue 203 in loop L2, a putative DNA binding site, and measured its fluorescence. The modified protein is active both in vivo and in vitro. The binding of nucleotide cofactor (ATP or its analog adenosine 5'-O-3-thiotriphosphate) does not modify the fluorescence. By contrast, the binding of DNA, both in the absence and presence of cofactor, strongly decreases the fluorescence in intensity (40-65%) and shifts the emission peak from 344 to 337 nm. The change occurs both with single- and double-stranded DNA and also upon the binding of a second single-stranded DNA. The results indicate that the residue 203 is in fact close to the first and second DNA binding sites. However, the quenching is not total and depends only slightly on the nature of DNA bases, thus suggesting an indirect interaction with DNA bases.

To determine the role of the central aromatic residue in one of the DNA binding domains in Escherichia coli RecA protein, we have constructed a protein in which a tryptophan fluorescence reporter is inserted in the place of phenylalanine residue 203 in loop L2, a putative DNA binding site, and measured its fluorescence. The modified protein is active both in vivo and in vitro. The binding of nucleotide cofactor (ATP or its analog adenosine 5-O-3-thiotriphosphate) does not modify the fluorescence. By contrast, the binding of DNA, both in the absence and presence of cofactor, strongly decreases the fluorescence in intensity (40 -65%) and shifts the emission peak from 344 to 337 nm. The change occurs both with single-and double-stranded DNA and also upon the binding of a second single-stranded DNA. The results indicate that the residue 203 is in fact close to the first and second DNA binding sites. However, the quenching is not total and depends only slightly on the nature of DNA bases, thus suggesting an indirect interaction with DNA bases.

DNA repair is vital for both prokaryotic and eukaryotic cells.
RecA is an ubiquitous and multi-functional enzyme involved in various steps of DNA repair: regulation of synthesis of DNA repair proteins (SOS induction), promotion of homologous recombination, and mutagenesis (For reviews, see Refs. [1][2][3]. For these activities, RecA first binds to single-stranded DNA with a strong cooperativity and forms a filamentous complex in which RecA subunits are arranged in a helical manner around the DNA (4). This nucleofilament binds a second doublestranded DNA molecule for the strand exchange reaction (5) and interacts with repressors for the induction of the SOS system (6) and with UmuD and UmuDЈ proteins for mutagenesis (7,8).
The molecular structure of the RecA filament in the absence of DNA has been determined by x-ray crystallography (9). A lower resolution structure of the nucleofilament (4) and of the DNA in the complex (5) has also been determined. These studies have shown that the DNA strands lie near the axis of the RecA-DNA filament. Determination of the DNA binding sites in RecA would allow us to build a higher resolution model of the RecA-DNA complex. Despite various studies, however, the DNA binding site(s) of RecA has not yet been determined. From the comparison of sequence with the DNA binding domain of ssDNA 1 binding protein of the Ike phage, GP5 protein, the domain comprising residues 240 -310 has been proposed as a DNA binding site (10 -12). But this region is in the subunitsubunit interface in the crystal of the RecA filament (9). Story et al. (9) rejected this possibility and proposed that two disordered loops that could not be traced in the x-ray analysis were the DNA binding domains because these loops line the cavity running down the middle of the RecA filament. In fact, mutations in L2 loop affect the activities of RecA (13), although modifications in L1 loop are easily tolerated (14).
Limited proteolysis of RecA-DNA complexes in the presence of ATP␥S has shown that a peptide spanning loop L2 is protected from tryptic digestion (15). In addition, peptides derived from this region, including a 20-amino acid peptide from residues 193-212 can bind to ssDNA with a high affinity. Subsequent work has shown that such a peptide can also bind to dsDNA. In addition, a phenylalanine at position 203 in loop L2 that is highly conserved between prokaryotic RecAs and their eukaryotic homologues (16,17) plays a central role in the binding. Only aromatic acid substitutions allow the peptides to bind to both ss-and dsDNA. 2 With the purpose of examining the possibility that this central aromatic residue in loop L2 interacts with DNA, we inserted a tryptophan residue in this loop in the place of phenylalanine. An interaction with DNA could then be detected by a change in the tryptophan fluorescence. Our results show that the modified protein is active, and its fluorescence is strongly decreased upon binding to DNA.

EXPERIMENTAL PROCEDURES
Protein Purification-Wild type RecA protein was purified as described previously (19). The mutant gene encoding RecA F203W was obtained from the wt recA gene with the use of Muta-Gene Phagemide In Vitro Mutagenesis version 2 kit (Bio-Rad) in accordance with manufacturer's recommendations. Mutant RecA F203W protein was expressed in BLR(DEL3)pLys cells carrying pET9c plasmid (Novagen) with recA F203W gene; expression was induced by 0.4 mM isopropyl-␤-D-thiogalactopyranoside over a 3-h period. The cells grown in 1 liter of LB medium were collected by low speed centrifugation, resuspended in 35 ml of 50 mM Tris-HCl (pH 8.0), 2 mM EDTA; the suspension was frozen at Ϫ80°C, thawed in an ice-water bath, and sonicated. The crude lysate was cleared by centrifugation at 18,000 rpm for 1 h in a SS-34 Sorvall rotor. 10% solution of polymin P (pH 8.0) was added to the lysate dropwise to the final concentration of 0.5%, and the solution was stirred for 30 min. The precipitate was collected by centrifugation in SS-34 Sorvall rotor for 30 min at 12,000 rpm and resuspended in 25 ml of 20 mM Tris-HCl (pH 7.5), 150 mM ammonium sulfate, 10% glycerol, 1 mM * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Materials-Poly(dA) (lot no. CJ 7836103) and poly(dT) (lot no. 2067834021) were from Pharmacia. Poly(d⑀A) was prepared by modification of poly(dA) by chloroacetaldehyde (Merck), as described (20). ATP␥S was purchased from Boehringer Mannheim. Concentration of materials was estimated from UV absorption using the following absorption coefficients: ⑀ 280 ϭ 2. Fluorescence was measured in a FP-777 spectrofluorometer (Jasco) at 20.0°C. The excitation wavelength was set at 295 nm with a bandwidth of 3 nm for selective excitation of tryptophan residues. The emission was usually measured at 345 nm with a bandwidth of 5 nm. For the measurement of excitation spectra of poly(d⑀A) fluorescence, the fluorescence was observed at 450 nm to minimize the contribution of fluorescence from tryptophan residues. The bandwidths were 3 and 5 nm for excitation and emission, respectively. Fluorescence was measured in a 1.0 ϫ 0.4-cm quartz cell to minimize the inner filter effect. The spectra were yet corrected for inner filter effect (25). The spectra were also corrected for background and Raman scattering by subtracting the buffer signal. Spectra were averaged over at least two scans. The absorption was measured in a J-710 spectropolarimeter (Jasco) in UV absorption mode. The quartz cell was treated by silicon (Serva) to avoid the adsorption of protein to the wall.
Fluorescence polarization anisotropy was measured in a FP-777 spectrofluorometer with the aide of ADP-301 automatic anisotropy measurement apparatus (Jasco). The temperature was set at 20.0°C, and the excitation and emission wavelengths were set at 295 nm (bandwidth of 3 nm) and 345 nm (bandwidth of 5 nm), respectively. The signal was averaged over 25 measurements of 0.5 s.

RESULTS
Activity of RecA F203W -Using site-directed mutagenesis, we introduced single amino acid substitution Phe 3 Trp in loop L2 of the RecA protein. Mutant RecA F203W protein was tested in DNA binding assay, and its activity was compared with wt RecA (Fig. 1). One can see that mutant RecA F203W protein is able to bind both ss-and dsDNA at the same, or even higher affinity, compared to wt protein. These data are in good agreement with our results on synthetic peptides spanning loop L2 region; Trp-substituted peptide possesses higher affinity to both ss-and dsDNA than the wild type Phe-containing pep-tide. 2 The capability of the mutant and wt proteins to carry out the strand exchange reaction (26) and to form synaptic complexes (27) was also checked, and no significant difference between the two proteins was found (data not shown). To test recombinational activity of RecA F203W mutant in vivo, we examined it in trans-complementation assay. The mutant was entirely proficient in supporting plaque formation by mutant phage (red Ϫ gam Ϫ ) on a lawn of bacterial cells carrying a plasmid expressing RecA F203W . 3 We noted that the secondary structure of the modified protein was very similar to that of wt RecA (judged by circular dichroism, not shown).
Absence of a Large Change in Trp-203 Fluorescence upon Cofactor Binding- Fig. 2 shows fluorescence emission spectra of RecA F203W protein in the presence and absence of activator ATP␥S. The spectra are compared with those of wild type RecA. The intensity of RecA F203W fluorescence was about two times larger than that of wt RecA. This was certainly due to additional fluorescence from the tryptophan residue inserted at position 203 in place of the phenylalanine. The differential fluorescence spectrum (fluorescence of RecA F203W Ϫ fluorescence of wt RecA) was centered around 345 nm (Fig. 2b), indicating that Trp-203 was well exposed to the solvent (28).
No significant fluorescence change was detected upon the addition of ATP␥S both for wild type and modified RecA. A slight change was sometimes observed but was probably related to slight temperature fluctuations and/or adsorption of protein on the wall of the quartz cell. Since further addition of ATP␥S up to 100 M did not significantly alter the protein fluorescence (not shown), the absence of a fluorescence change was not due to a reduced affinity of the modified protein for ATP␥S. The addition of the physiological cofactor, ATP, up to 0.5 mM did not significantly modify the fluorescence of wt RecA or RecA F203W . The cofactor did not affect the fluorescence of Trp-203 and thus did not modify its environment.
The anisotropy of Trp-203 fluorescence was also not modified upon binding to ATP␥S. Thus the local motion of the residue was not affected by ATP␥S. The anisotropy of RecA F203W was 0.150 and as large as that of wt RecA (0.145) both in the presence and absence of ATP␥S. By subtraction of contribution from Trp-290 and Trp-308, which are also present in wt RecA, we got an anisotropy value r ϭ 0.15 for Trp-203. The value was large and indicated that the local motion of Trp-203 was restricted. The local motion was slower than the lifetime of tryptophan, which is in the order of nanoseconds.
Quenching of Trp-203 Fluorescence upon DNA Binding in the Absence of Nucleotide Cofactor- Fig. 3 shows the fluorescence change of RecA F203W upon the addition of poly(dA) or poly(dT) in absence of cofactor. There was a large decrease in the intensity of RecA F203W fluorescence in contrast to no significant change of wt RecA fluorescence. The signal was corrected for the increase of the inner filter effect upon the addition of polynucleotides. Since these fluorescence experiments were made with a selective excitation of tryptophan residues, the fluorescence change was certainly due to a quenching of Trp-203 upon the binding of ssDNA.
The fluorescence change was significantly smaller for poly(dA) (21% quenching) than poly(dT) (32% quenching). The former exhibits weaker binding affinity to RecA than poly(dT) (20). However, this difference was not due to an incomplete formation of RecA-poly(dA) complex because, as Fig. 4 shows, the fluorescence change was saturated about 5 bases of polynucleotide per RecA monomer for both poly(dA) and poly(dT). A similar binding stoichiometry was observed for wt RecA (29). Therefore, the slight difference between poly(dA) and poly(dT) may reflect a slight difference in the binding mode or the stability of the complexes. When we consider that the difference of fluorescence between RecA F203W and wt RecA simply corresponds to the fluorescence of Trp-203, the binding of poly(dA) and poly(dT) quench about 40 and 60% of Trp-203, respectively.
Quenching of Trp-203 Fluorescence upon DNA Binding in the Presence of Nucleotide Cofactor- Fig. 5 shows fluorescence change of RecA F203W upon the binding of poly(dA), poly(dT), or plasmid dsDNA in the presence of ATP␥S. We can observe a large decrease of the fluorescence of RecA F203W upon complex formation in contrast to no significant fluorescence change for wt RecA upon interaction with any DNA. The absence of change in tryptophanyl fluorescence of wt RecA upon DNA binding has been reported previously (25). The addition of any DNA with a ratio of 3 bases (or base pairs for dsDNA) per RecA, which should be just enough to saturate first site (site I), decreases the fluorescence of the protein about 20%. There was no significant difference for different DNAs. Poly(dA), poly(dT), and linearized ds plasmid DNA quenched the fluorescence to a similar extent. Further addition up to 9 base pairs per RecA of dsDNA did not any more change the fluorescence, while the addition of single-stranded polynucleotide further decreased the fluorescence (Fig. 5). We note that RecA bound only one chain of dsDNA with a stoichiometry of 3 base pairs per subunit but can bind up to three chains of ssDNA with a stoichiometry of 3 bases per subunit for each chain (29,30). The titration experiments indicate that RecA F203W binds to DNA with a similar binding stoichiometry as wt RecA (Fig. 4b). The fact that the addition of single-stranded polynucleotides exceeding the ratio of 3 bases per RecA further modifies the protein fluorescence is in contrast to the case with dsDNA and suggests that the binding of a second ssDNA (occupation of site II), as well as that of the first ssDNA (occupation of site I), decreases the fluorescence of Trp-203. The binding of the first DNA quenched about 40% of the Trp-203 fluorescence and that of the second DNA about 20%. The binding of a third DNA (occupation of site III) did not appear to affect the fluorescence (Fig. 4b). The same conclusion was obtained using heat-denatured calf thymus DNA as ssDNA. A significant blue shift in the emission spectra was observed upon the binding of two ssDNAs (not shown).
Absence of Efficient Energy Transfer between Trp-203 and DNA Bases-The results above indicate that Trp-203 could be close to the DNA. To gain more structural information, we have investigated the presence of singlet-singlet energy transfer from the tryptophan residue to the fluorescent nucleobase of poly(d⑀A), an analog of poly(dA). The presence of energy transfer has been examined by a change in shape of excitation spectra of poly(d⑀A) upon the binding of RecA as it was performed for a poly(d⑀A)-GP32 protein complex by Toulme and Helene (31). The emission spectrum of tryptophan fluorescence partially overlaps with the absorption spectrum of poly(d⑀A). Energy transfer, therefore, can occur and should be detected by enlargement of the excitation spectra of poly(d⑀A) in the tryptophan absorption region (around 280 nm). In Fig. 6, the excitation spectrum of poly(d⑀A) with RecA F203W is shown and compared with the spectrum of poly(d⑀A) without RecA and that with wt RecA. The fluorescence signal from the protein is subtracted. The fluorescence of poly(d⑀A) is strongly enhanced upon the binding of RecA F203W , although smaller than that by wt RecA (a 6.5-fold increase compared with a 9-fold increase by wt RecA).
To facilitate the comparison of spectral shape, we have computed the ratio of fluorescence of complex/fluorescence of free poly(d⑀A) at various wavelengths. The ratio is almost constant and independent of wavelength. There is not a large change in spectral shape upon the binding of RecA F203W . No significant change is observed upon the binding of wt RecA as reported previously (25). Since the two chromophores (Trp-203 and DNA bases) in the complex should be oriented (judged from their large anisotropy value), the absence of a large energy transfer between them could be due to their unfavorable relative orientation. The increase of poly(d⑀A) fluorescence by the modified RecA may occur by the mechanism described for wt RecA (20). Decrease in the mobility of the DNA bases upon RecA binding prevents the base-base collision and thus the dynamic quenching of poly(d⑀A) fluorescence. DISCUSSION By monitoring the fluorescence change of a tryptophan reporter residue inserted in loop L2 of RecA, we have investigated the role of this residue in the loop in its interactions with cofactor and DNA. Loop L2 in RecA crystal does not provide x-ray diffraction and is considered to be disordered or in several different conformations (9). This loop was proposed as a ssDNA binding site by Story et al. (9). Subsequent work has confirmed this proposal (15) and extended it by showing that this loop also binds to dsDNA. 2 The tryptophan reporter was inserted in the place of phenyl- alanine, another aromatic amino acid, and does not appear to significantly alter the activities in vitro or in vivo of RecA. The modified RecA enhances the fluorescence of the DNA analog poly(d⑀A) as does wt RecA. The binding stoichiometries are not affected by this substitution. These results also suggest that the structure of the DNA-RecA complex is not so much altered by the insertion of the tryptophan. Therefore, the conclusions obtained from this fluorescence study of this tryptophan reporter should reflect the role of Phe-203 in wt RecA. The fluorescence changes of the modified RecA can be also used, in future, for kinetic and thermodynamic analysis of RecA-DNA interaction.
Restricted Motion of Trp-203-The distance between the two ends (residue 194 and residue 210) of loop L2 in the crystal is only 0.7 nm, whereas loop L2 is 14 amino acid residues long. The loop could exist in various conformations, and residue 203, which is at the center of the loop, could be far from the ends of the loop. The fluorescence of Trp-203 estimated from the difference between the fluorescence of modified and wt RecA is centered around 345 nm and thus indicates that the residue is exposed to solvent. A large anisotropy value indicates that the local motion of Trp-203 is restricted. Thus, the loop should not be in fast motion, and the absence of diffraction from the loop in the x-ray analysis is probably due to the presence of various conformations.
No Interaction with Nucleotide Cofactor-Absence of a fluorescence change upon the binding of nucleotide cofactor indicates that the residue is far from the cofactor. Story and Steitz (32) proposed that Gln-194 in the loop may interact with ␥ phosphate of ATP, but we did not observe a significant fluorescence change of Trp-203. There may be a rather large distance between Phe-203 and the ␥ phosphate of ATP. The absence of fluorescence change upon cofactor binding also suggests that the residue may not be in the subunit-subunit interface because the cofactor is known to modify the helical organization of the protein (5,33) probably by changing subunit-subunit interactions. In fact, the fluorescence intensity of RecA F203W , as that of wt RecA, is simply proportional to the protein concentration from 10 nM to 2 M (not shown). The fluorescence is thus independent of association and dissociation of RecA subunits.
The Aromatic Residue in Loop L2 Is Probably Involved in DNA Binding-The large fluorescence change of Trp-203 upon DNA binding both in the presence and absence of cofactor suggests that this region could be a DNA binding domain, although the fluorescence change can also occur indirectly upon a conformation change induced by DNA binding. Site-directed mutagenesis made in L2 loop by Cazaux et al. (13) show that loop L2 is important for various activities of RecA. These results indicate that this residue is involved in DNA binding. However, the quenching of fluorescence is not total. There may not be strong stacking interaction of Trp-203 with DNA bases, although intercalation of aromatic residues between DNA bases has been proposed (34). The degree of quenching of Trp-203 fluorescence depends little upon the nature of the DNA bases. This, together with the absence of efficient energy transfer from Trp-203 to DNA bases, also suggests absence of a direct interaction between Trp-203 and the DNA bases.
Binding of Various DNA Occurs in the Same Domain-The fluorescence change occurs both in the presence and absence of cofactor. This indicates that DNA interacts with the same region of RecA in the presence and in the absence of cofactor, although the cofactor modifies the DNA binding mode of RecA. Our results also suggest that the binding of the second ssDNA occurs in the same part of RecA as that of the first ssDNA, although the fluorescence change upon the binding of the second DNA is smaller and not exactly the same. This suggests that site I and site II are very close to each other. Another interesting observation is that the fluorescence change promoted by dsDNA is similar to that by one ssDNA but smaller to that by two ssDNAs. One of the strands of dsDNA may interact with RecA in a similar way as the first ssDNA. By contrast, the second strand of dsDNA may not be interacting with this part of RecA. Such a model has been proposed previously (5). The binding of a third DNA may occur in a region different from loop L2. This is rather expected because the binding mode of a third ssDNA is clearly different from the binding mode of the first and second ssDNA (18,30).