Mutations in a Conserved Motif Inhibit Single-stranded DNA Binding and Recombination Mediator Activities of Bacteriophage T4 UvsY Protein*

The UvsY recombination mediator protein is critical for homologous recombination in bacteriophage T4. UvsY uses both protein-protein and protein-DNA interactions to mediate the assembly of the T4 UvsX recombinase onto single-stranded (ss) DNA, forming presynaptic filaments that initiate DNA strand exchange. UvsY helps UvsX compete with Gp32, the T4 ssDNA-binding protein, for binding sites on ssDNA, in part by destabilizing Gp32-ssDNA interactions, and in part by stabilizing UvsX-ssDNA interactions. The relative contributions of UvsY-ssDNA, UvsY-Gp32, UvsY-UvsX, and UvsY-UvsY interactions to these processes are only partially understood. The goal of this study was to isolate mutant forms of UvsY protein that are specifically defective in UvsY-ssDNA interactions, so that the contribution of this activity to recombination processes could be assessed independent of other factors. A conserved motif of UvsY found in other DNA-binding proteins was targeted for mutagenesis. Two missense mutants of UvsY were isolated in which ssDNA binding activity is compromised. These mutants retain self-association activity, and form stable associations with UvsX and Gp32 proteins in patterns similar to wild-type UvsY. Both mutants are partially, but not totally, defective in stimulating UvsX-catalyzed recombination functions including ssDNA-dependent ATP hydrolysis and DNA strand exchange. The data are consistent with a model in which UvsY plays bipartite roles in presynaptic filament assembly. Its protein-ssDNA interactions are suggested to moderate the destabilization of Gp32-ssDNA, whereas its protein-protein contacts induce a conformational change of the UvsX protein, giving UvsX a higher affinity for the ssDNA and allowing it to compete more effectively with Gp32 for binding sites.

These effects are in agreement with UvsX and UvsY being essential for initiating phage recombination-dependent replication and for mediating recombinational DNA repair.
The 43-kDa UvsX protein, functionally homologous to the E. coli RecA protein and to Rad51 in Saccharomyces cerevisiae, is an ATP-dependent DNA strand transferase (6,7). Like RecA and Rad51, UvsX binds cooperatively to ssDNA, 1 forming presynaptic filaments that catalyze DNA strand exchange. The formation of UvsX presynaptic filaments is assisted by the UvsY protein, the T4 recombination mediator protein (RMP), and by Gp32, the T4 ssDNA-binding protein.
In the currently accepted model of T4 presynaptic filament assembly (8), Gp32 pre-coats the ssDNA, removing inhibitory secondary structure from the lattice. UvsX must then displace Gp32 from the ssDNA, but direct displacement is thermodynamically unfavorable and kinetically slow. Instead, UvsY protein mediates the formation of UvsX filaments on Gp32-saturated ssDNA, helping UvsX to displace Gp32 in the process. The mechanism of recombination mediation by UvsY is the focus of this study and of previous work by our laboratory and others (9 -16).
UvsY is the prototype of a class of proteins referred to as recombination mediator proteins or RMPs (17). The common function of these proteins is to overcome thermodynamic and/or kinetic barriers imposed by ssDNA-binding proteins and to mediate the specific assembly of a recombinase-ssDNA complex. Other examples of RMPs include the human and S. cerevisiae Rad52 proteins, the S. cerevisiae Rad55/57 protein dimer, and the E. coli RecO/R protein complex. The 15.8-kDa UvsY protein lacks enzymatic activities of its own, but stimulates the enzymatic activities of UvsX. In vitro, the addition of relatively low concentrations of UvsY generally overcomes the inhibitory effects of high Gp32, high salt, and/or low UvsX concentrations, conditions regularly encountered in vivo, in UvsX-catalyzed ssDNA-dependent ATP hydrolysis, DNA strand exchange, and recombination-dependent replication reactions (8 -13, 18). Biochemical properties of UvsY include non-cooperative and sequence nonspecific binding to ssDNA, as well as specific interactions with T4 recombination proteins UvsX, Gp32, Gp46, and Gp47 (8, 13, 15, 19 -21). 2 In addition, UvsY forms stable hexamers in solution and binds to ssDNA in this form (14). The ssDNA contacts multiple subunits of the UvsY hexamer, suggesting that the lattice is wrapped. 3 Wrapping or other distortions of ssDNA structure are proposed to explain the observed destabilization of Gp32-ssDNA interactions brought about by UvsY (16). Lowering of Gp32-ssDNA affinity and/or cooperativity within a tripartite UvsY-Gp32-ssDNA intermediate may facilitate the local displacement of Gp32 from the lattice by incoming UvsX, resulting in a UvsYmediated nucleation event for presynaptic filament formation (8,16,17).
Whereas data suggest that UvsY-ssDNA interactions are sufficient to destabilize Gp32-ssDNA complexes (16), other results suggest that UvsY-Gp32, UvsY-UvsX, and UvsY-UvsY interactions are also required for efficient presynapsis and for recombination functions of the presynaptic filament (13,21). In addition to stabilizing the presynaptic filament, these contacts are also thought to play critical roles in guiding the requisite proteins into the proper orientations to enable synapsis to occur. Evidence suggests that UvsY-Gp32 interactions are required to load UvsX onto Gp32-coated DNA (21), whereas UvsY-UvsX interactions contribute significantly to the stabilization of the UvsX-ssDNA polymer (10). To further explore the roles of protein-ssDNA and protein-protein interactions in presynapsis, we have begun to study mutant forms of UvsY that are defective in heteroprotein binding, ssDNA binding, or self-association. Limited chymotrypsinolysis digests the 137amino acid UvsY protein into two putative domains: an aminoterminal fragment or domain containing the first 101 residues, and a carboxyl-terminal fragment or domain containing the remaining 36 residues (13). The NH 2 -terminal fragment retains weak ssDNA binding activity, but is completely deficient in self-association (hexamerization) and in UvsY-Gp32 and UvsY-UvsX interactions (13,24). The decrease in ssDNA binding affinity appears to result from the loss of intrahexamer synergism or cooperativity in this monomeric form of UvsY, not from disruption of the ssDNA binding site itself (24). Despite the loss of protein-protein contacts, the NH 2 -terminal fragment of UvsY retains the ability to stimulate the ssDNA-dependent ATPase and DNA strand exchange activities of UvsX, but at dramatically lower levels than full-length UvsY (13). Thus both protein-ssDNA and protein-protein interactions of UvsY appear to be important for assembling UvsX-ssDNA presynaptic filaments and for stimulating the activities of UvsX.
To isolate the specific contributions of UvsY-ssDNA interactions to UvsY function, it is necessary to construct a UvsY species containing point mutations that would impair its interaction with ssDNA without significantly altering its other associative properties. The UvsY protein contains an LKARLDY sequence motif at positions 57-63. This motif appears to be conserved in a number of DNA repair proteins including the S. cerevisiae Rad3 DNA helicase and the human ERCC2 protein, a putative DNA helicase involved in excision DNA repair and defective in one complementation group of Xeroderma pigmentosum (25). Its location in the N-domain plus alternating basic and hydrophobic residues suggested that it could be important for UvsY-ssDNA interactions. This paper describes the construction and characterization of two site-directed UvsY mutant proteins, UvsY K58A and UvsY K58A,R60A , which contain single and double missense mutations, respectively, within the LKARLDY motif. As predicted, these mutants are defective in ssDNA binding, however, they retain self-and heteroproteinassociation activities similar to wild-type UvsY. We demonstrate that both mutants are partially but not totally defective in stimulating UvsX-catalyzed reactions, which has interesting implications for the mechanism by which UvsY mediates the assembly and function of T4 presynaptic filaments.

MATERIALS AND METHODS
Reagents, Enzymes, and Nucleic Acids-Chemicals, biochemicals, and commercial enzymes were purchased from Sigma unless specifically noted. All reagents were analytical grade, and solutions were made with de-ionized, glass-distilled water. Radiolabeled [␥-32 P]ATP was purchased from PerkinElmer Life Sciences. The pET3a vector was purchased from Novagen. Restriction endonucleases were purchased from New England Biolabs. Taq polymerase and PCR reagents were purchased from PerkinElmer Life Sciences. Oligonucleotides were purchased from Operon. Circular single-stranded DNA from the bacteriophage M13mp19 was isolated by extraction from purified phage particles (26). Supercoiled M13mp19 dsDNA (RFI) was isolated from phage-infected E. coli cells as described (26). RFI DNA was digested with the EcoRI restriction endonuclease to produce linear M13mp19 DNA (RFIII), which was labeled with [ 32 P] at its 5Ј ends as described (26). The concentrations of ssDNA and dsDNA were determined by the absorbance at 260 nm, using conversion factors of 36 and 50 g/ml/A 260 , respectively. All DNA concentrations are expressed in units of micromoles of nucleotide residues per liter. Etheno-modified ssDNA (⑀DNA) was synthesized by treating either poly(dA) lattices with an average chain length of 310 nucleotide residues (Amersham Biosciences), or M13mp19 ssDNA circles with chloroacetaldehyde as described (27,28).
The T4 UvsX (43 kDa), Gp32 (34 kDa), and wild-type UvsY (16 kDa) proteins were purified and stored as described (15,29). All T4 protein stock solutions used in this study were judged to be Ͼ95% pure based on analyses of Coomassie Blue-stained SDS-polyacrylamide gels. In addition, all protein solutions were found to be nuclease-free according to the criteria described by Sweezy and Morrical (15). All protein concentrations were determined by the absorbance at 280 nm, using molar extinction coefficients of 69,760 M Ϫ1 cm Ϫ1 for UvsX, 19,180 M Ϫ1 cm Ϫ1 for UvsY, and 41,360 M Ϫ1 cm Ϫ1 for Gp32 (30).
Construction of UvsY Missense Mutants-UvsY missense mutants were constructed using a three-step PCR site-directed mutagenesis strategy (31). The overall scheme involves amplifying the gene in two portions, one of which contains a targeted mutation. Denaturing the two products and annealing them at their region of overlap, and then performing a polymerase fill-in reaction creates a template of the gene containing the mutation. This new template is subsequently amplified by PCR, gel purified, and cloned into a suitable expression vector.
The forward primer introduces an NdeI restriction site containing the start codon onto the 5Ј end of the resultant PCR product. A second PCR amplifies the 3Ј end of the UvsY gene (nucleotide positions 150 -410) and introduces either one or two codon changes at sites described here. To construct the double mutant UvsY K58A,R60A , a forward primer was used that contains an altered sequence such that the codons for lysine 58 and arginine 60 are both changed to code for alanine. The primer is: forward (42-mer), primer 3a, 5Ј-GCACAGAAAAAAGTTGC-TCTTGCAGCTAGATTAGACTACTAC-3Ј. To construct the single mutant UvsY K58A , a forward primer was used such that the codon for lysine 58 is changed to code for alanine. This primer is: forward (42mer), primer 3b, 5Ј-GCACAGAAAAAAGTTGCTCTTAAAGCT-GCATTAGACTACTAC-3Ј.
The same reverse primer was used to amplify the 3Ј end of the UvsY gene containing both the single and double mutations. It introduces a BamHI site onto the 3Ј end of the UvsY gene, immediately following the stop codon. It has the following sequence: reverse (31-mer), primer 4: 5Ј-GCGCGCGGATCCCCTGACGCTTAAGATTATG-3Ј.
The UvsY K58A single and UvsY K58A,R60A double mutants were cloned using identical methods. After the 5Ј end of the UvsY gene had been PCR-amplified by standard procedures (26) with primers 1 and 2 and the 3Ј end had been amplified with either primer 3a or primer 3b for the upstream end, and primer 4 for the downstream end, each resulting fragment was gel-purified. The two pools of fragments were mixed and added to standard PCR buffer containing free nucleotides, and then subjected to 94°C for 30 s, followed by 50°C for 30 s. These conditions cause the respective fragments to denature and then re-anneal. Because of the overlapping regions in the two separate fragments, some of the annealed species are composites, with one strand from the UvsY-5Ј end pool and one strand from the UvsY-3Ј end pool. After the annealing step, Taq polymerase was added to the reaction to allow nucleotide fill-in of the un-annealed overhangs. A unique fragment is generated that contains a complete sequence of the UvsY gene, with point mutations mapping to codons 58 and/or 60, and with 5Ј-NdeI and 3Ј-BamHI sites. By adding primers 1 and 4 to the reaction, this mutated UvsY gene fragment can be PCR-amplified by standard methods. The final PCR products containing mutated UvsY genes were ligated into the pET3a expression vector to form the constructs designated pUvsY K58A,R60A or pUvsY K58A , respectively. Their sequences were verified by DNA sequencing, using the dideoxy method (26) and an ABI automated sequencer.
Expression and Purification of UvsY Mutants-The pUvsY K58A,R60A and pUvsY K58A plasmids were transformed into competent BL21(DE3) E. coli cells. Single colonies of pUvsY K58A,R60A -or pUvsY K58A -harboring transformants grown on LB/ampicillin plates were selected and grown at 37°C in liquid LB media, containing 100 g/ml ampicillin, to a cell density of A 600 ϭ 0.6, then induced by adding 1 mM isopropyl-1-thio-␤-D-galactopyranoside and continuing growth for 3 h at 37°C. For the UvsY K58A,R60A preparation, 19 g of induced cells were harvested from a 6-liter culture and resuspended in 100 ml of lysis buffer (20 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 5 mM 2-mercaptoethanol (BME), 1 M NaCl, 10 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride). DNase I was added to a final concentration of 20 g/ml. The cells were sonicated with continuous pulses of 6 ϫ 3 min at 4°C. All subsequent steps were performed at 4°C. The cell lysate was centrifuged for 30 min at 10,000 rpm in a Sorvall SS-34 rotor, and then centrifuged for 2 h at 30,000 rpm in a Beckman Ti-45 rotor. The supernatant was recovered and dialyzed exhaustively against 2 ϫ 4 liters of PC-100 buffer (20 mM Tris-HCl, pH 7.4, 2.2 mM EDTA, 1 mM BME, 100 mM NaCl, and 10% (w/v) glycerol). The dialysate was then loaded onto a Whatman P-11 phosphocellulose column (bed volume ϭ 180 ml) that had been pre-equilibrated with PC-100 buffer. A 1,200-ml linear gradient from PC-100 to PC-1000 (100 to 1000 mM NaCl) was run. UvsY K58A,R60A eluted with a peak at ϳ325 mM NaCl. Fractions were analyzed by Coomassie Blue staining of SDS-polyacrylamide gels, and UvsY K58A,R60A -containing fractions were pooled. The PC pool was dialyzed against 2 ϫ 2 liters of DC-100 buffer (20 mM Tris-HCl, pH 8.1, 5 mM EDTA, 1 mM BME, 100 mM NaCl, and 10% (w/v) glycerol), and then loaded onto a 50-ml ssDNA-cellulose (DC) column that had been prepared as described (32), and pre-equilibrated in the same buffer. The DC column was washed with DC-100 buffer, and then eluted with a step gradient of DC-200 (same as DC-100, except the NaCl concentration is 200 mM), followed by DC-600 (NaCl ϭ 600 mM). UvsY K58A,R60A did not adhere to the DC column, but instead eluted in the load flow-through. The DC flow-through fraction was loaded directly onto a 40-ml hydroxyapatite column (HAP) that had been pre-equilibrated in HAP-100 buffer (100 mM potassium phosphate, pH 7.4, 1 mM EDTA, 1 mM BME, and 10% (w/v) glycerol). The HAP column was washed with HAP-100 buffer, and then eluted with a 400-ml gradient from HAP-100 to HAP-900 (potassium phosphate concentration ϭ 900 mM). UvsY K58A,R60A eluted with a peak at ϳ460 mM potassium phosphate. Fractions were analyzed by Coomassie Blue staining of SDS-polyacrylamide gels and the peak fractions were pooled and dialyzed against 2 ϫ 2 liters of UvsY prestorage buffer (20 mM Tris-HCl, pH 7.4, 0.2 mM EDTA, 1 mM BME, 100 mM NaCl, and 20% (w/v) glycerol), and then against 2 ϫ 2 liters of UvsY storage buffer (20 mM Tris-HCl, pH 7.4, 0.2 mM EDTA, 1 mM BME, 100 mM NaCl, and 65% (w/v) glycerol). The final yield was ϳ66 mg of UvsY K58A,R60A . The protein was Ͼ99% homogeneous and both nuclease-and ATPase-free, as determined by methods described for native UvsY (15). Concentration was determined by the absorbance at 280 nm, using a molar extinction coefficient of 19,180 M Ϫ1 cm Ϫ1 for UvsY K58A,R60A . This extinction coefficient was calculated using the method of Gill and von Hippel (30).
The UvsY K58A single mutant protein was expressed as described for the double mutant protein, but was subjected to additional purification steps after the HAP column. The UvsY K58A -containing HAP column fractions were pooled and dialyzed into DEAE-50 buffer (20 mM Tris-HCl, pH 8.1, 5 mM EDTA, 1 mM BME, 50 mM NaCl, and 10% (w/v) glycerol) and loaded onto a 40-ml DEAE-cellulose column that had been pre-equilibrated in the same buffer. The DEAE column was washed with DEAE-50. The UvsY K58A did not adhere to the DEAE-cellulose column, but instead came off in the wash flow-through. Two minor contaminants were removed with this extra purification step, as demonstrated by SDS-PAGE. The wash flow-through pool, containing UvsY K58A , was concentrated via an additional HAP column, prepared and run as described above, and then dialyzed into UvsY prestorage and storage buffers using similar procedures as described for the double mutant. The final yield of Ͼ95% homogeneous, nuclease-and ATPasefree UvsY K58A protein was ϳ50 mg from 6 liters of culture.
Analytical Ultracentrifugation-Sedimentation velocity experiments were performed in a Beckman Optima XL-I analytical ultracentrifuge as described (14). UvsY wild-type, UvsY K58A,R60A , and UvsY K58A were each dialyzed into analytical ultracentrifugation buffers containing 20 mM Tris-HCl (pH 7.4), 1 mM MgCl 2 , and NaCl ranging from 0.18 to 1.0 M as indicated in each buffers name: e.g. AnU-0.5 contains 20 mM Tris-HCl (pH 7.4), 1 mM MgCl 2 , and 0.5 M NaCl. All velocity runs were performed at 20°C, in a Beckman An60-Ti 4-hole rotor, spinning at a speed of 42,000 rpm, with cells containing 12-mm double-sector charcoal-filled Epon centerpieces and sapphire windows. The initial protein loading concentrations ranged from 0.8 to 1.4 mg/ml. Protein sedimentation was monitored using absorbance and interference optics. All data were analyzed using the van Holde and Weischet method (33).
Protein Affinity Chromatography Experiments-UvsX-, Gp32-, and bovine serum albumin-agarose columns were prepared by covalently coupling the proteins to Bio-Rad Affi-Gel 10 beads as described (34). In each case, total immobilized protein was determined to be 1.5 to 2.0 mg of protein/ml of bed volume. 2-ml columns were poured and equilibrated with running buffer RB-50 (20 mM Tris-HCl, pH 8.1, 1 mM EDTA, 5 mM MgCl 2 , 50 mM NaCl, and 10% (w/v) glycerol). All chromatography steps were conducted at 4°C. 100-g quantities of UvsY wild-type, UvsY K58A,R60A , or UvsY K58A were dialyzed into RB-50 and loaded onto the columns by gravity flow. The flow-through was collected and protein was quantified. The columns were washed with 4 ml of RB-50, and then eluted with successive 4-ml steps of buffers RB-200, RB-400, RB-600, and RB-2000 (identical to RB-50 except with [NaCl] ϭ 200, 400, 600, and 2000 mM, respectively). Fractions of 200 l were collected and assayed for protein content by the Bradford assay and SDS-PAGE. Exact NaCl concentrations at which proteins eluted were determined by solution conductivity compared against NaCl standards.
⑀DNA Binding Studies-Fluorescence experiments were carried out in an SLM8000 fluorimeter as described (15,27). Data in each titration were corrected for the effects of sample dilution, intrinsic protein fluorescence, inner filter effects, and/or baseline fluorescence of ⑀DNA depending on the type of experiment (15,27,35). Titrations of UvsY species into blank solutions containing no ⑀DNA were used to identify fluorescence changes because of protein addition. This signal change was a linear function of protein concentration in all experiments. Data were not corrected for photobleaching of the ⑀DNA because photobleaching was found to be negligible over the time period of all experiments. As a general precaution against photobleaching, samples were only exposed to the light source during the 10 -15-s data acquisition interval following each addition and equilibration of titrant.
Salt-back titration experiments (salt added incrementally to preformed ⑀DNA/UvsY mutant protein mixtures (36)) were carried out at 25°C in buffer containing 20 mM Tris-HCl (pH 7.4), 1 mM MgCl 2 , and 5-75 mM NaCl. The starting concentrations of the ⑀DNA and UvsY mutant protein were 7.5 and 2.25 M, respectively, representing a 1.2-fold molar excess of UvsY mutant protein with respect to binding sites, assuming a UvsY binding site size of n ϭ 4 nucleotide residues per monomer (15). The starting volume was 2.0 ml in each experiment. The titrant consisted of concentrated NaCl, added in 1.2-l aliquots at 1-min intervals with constant stirring. The excitation and emission wavelengths for ⑀DNA were 300 and 405 nm, respectively. The bandpass was 5 nm. All data points were obtained as an average of 5-10 readings of the fluorimeter, set on continuous mode.
ATPase Assays-Rates of UvsX-catalyzed ssDNA-dependent ATP hydrolysis were determined by a coupled spectrophotometric assay, using a slight modification of the procedure of Morrical et al. (37). ATPase time courses were recorded on a Hitachi U-2000 spectrophotometer equipped with a water-jacketed cuvette holder to maintain a constant temperature of 37°C. Final reaction volumes of 700 l were contained in 1-ml quartz cuvettes of 1-cm path length. All reactions contained 20 mM Tris acetate (pH 7.4), 90 mM potassium acetate (KOAc), 10 mM magnesium acetate, 2 mM ATP, 6 units/ml pyruvate kinase, 6 units/ml lactate dehydrogenase, 2.3 mM phosphoenolpyruvate, and 0.23 mM NADH. In addition, all reactions contained 4.5 M (nucleotides) M13mp19 ssDNA, plus UvsX (10 g/ml), Gp32 (50 g/ml), and UvsY wild-type (0 -5 g/ml), UvsY K58A,R60A (0 -5 g/ml), or UvsY K58A (0 -5 g/ml) as indicated. Identical reaction buffer conditions were maintained between experiments by adding protein storage buffers to reactions in appropriate amounts. All reaction components except UvsX protein were preincubated for 5 min at 37°C, and then reactions were initiated by adding UvsX. Reaction velocities were calculated from the change in absorbance at 380 nm as described (37) using the linear portions of time courses.
DNA Strand Exchange Assays-We used an assay based on the protocol of Formosa and Alberts (7). Reaction mixtures contained 20 mM Tris acetate (pH 7.4), 90 mM potassium acetate, 10 mM magnesium acetate, 100 g/ml bovine serum albumin, 1 mM dithiothreitol, 2 mM ATP, 10 mM creatine phosphate, 10 g/ml creatine phosphokinase, 4.5 M M13mp19 ssDNA, 4.5 M 5Ј-32 P-labeled M13mp19-RFIII DNA, 35 g/ml Gp32, variable amounts of UvsX (0 -50 g/ml) and variable amounts of UvsY (0 -5 g/ml), UvsY K58A,R60A (0 -5 g/ml), or UvsY K58A (0 -5 g/ml). Constant salt conditions were maintained by adding protein storage buffers to reactions in appropriate amounts. Reactions were performed at 37°C. Each reaction was preincubated for 2 min and then initiated by the addition of ATP. 15-l aliquots were removed at the time points indicated in the figure legends, and added to a stopping solution yielding final concentrations of 380 g/ml proteinase K, 20 mM EDTA, 0.4% SDS, and 0.04% bromphenol blue. Stopped samples were placed on ice for 10 min and then loaded onto 13 ϫ 15-cm neutral 1.0% agarose gels. Gels were run for 3 h at 100 V, then dried onto Whatman DE-81 paper for autoradiography. K58A and UvsY K58A,R60A Mutant Proteins-UvsY K58A and UvsY K58A,R60A mutant proteins were generated, overexpressed, and purified as described under "Materials and Methods." Results of overexpression and purification are shown in Fig. 1. The mutant proteins were purified to Ͼ95 and Ͼ99% homogeneity, respectively, and were free of any detectable nuclease or ATPase contamination. The final purified fractions of UvsY K58A and UvsY K58A,R60A shown in Fig. 1 were used for all of the solution studies reported here. Note that both UvsY K58A and UvsY K58A,R60A were passed over ssDNA-cellulose affinity columns in the course of purification; neither mutant protein bound to ssDNA-cellulose in buffer containing 100 mM NaCl, whereas wild-type UvsY binds quantitatively to ssDNA-cellulose at the same salt concentration, and requires 600 mM NaCl for elution (15) (see Table II). The ssDNA-cellulose results indicate major ssDNA-binding defects in UvsY K58A and UvsY K58A,R60A , which we explore in greater detail in a later section.

Expression and Purification of UvsY
Self-Association Properties of UvsY K58A and UvsY K58A,R60A Mutant Proteins-Wild-type UvsY protein exists predominantly as a stable 6.0 S hexamer in analytical ultracentrifugation buffers containing NaCl concentrations greater than 0.2 M, and forms progressively larger oligomers as salt concentration decreases below the 0.2 M threshold (14). Conversely, a truncated form of UvsY protein lacking the COOH-terminal 36 amino acid residues sediments as a 2.1 S monomer at all salt concentrations examined (24). The self-association properties of missense mutants UvsY K58A and UvsY K58A,R60A were examined by sedimentation velocity. Fig. 2 shows the integral distribution of s 20,w for both mutant proteins as a function of NaCl concentration. The data demonstrate that UvsY K58A and UvsY K58A,R60A both resemble wild-type UvsY in their capacity to oligomerize. Both mutants sediment at ϳ6 S at higher salt concentrations, indicative of the stable hexameric form observed with wild-type UvsY (Table I and Ref. 14). Neither mutant shows any tendency to dissociate into quaternary structures smaller than hexamers under the wide range of solution conditions tested here (0.18 -1.0 M NaCl in analytical ultracentrifugation buffer; Fig. 2). Therefore it is reasonable to conclude that the K58A and K58A,R60A missense mutations, unlike COOH-terminal truncation (24), do not destabilize the fundamental hexameric quaternary structure of the UvsY protein. There are, however, differences observed in the sedimentation properties of UvsY K58A and UvsY K58A,R60A compared with wild-type UvsY (Table I). Both mutants appear to form higher order associations more readily than wild-type, as evidenced by the protein concentration dependence of s 20,w observed for UvsY K58A and UvsY K58A,R60A in salt concentrations ranging from 0.3 to 1.0 M NaCl (Fig. 2). This trend, which may indicate some degree of polydispersity on the part of UvsY K58A and UvsY K58A,R60A , stands in contrast to wild-type UvsY, which is highly monodisperse under equivalent conditions (Table I and Ref. 14). Therefore the K58A and K58A/R60A missense mutations, while preserving the hexamerization properties of UvsY, may also render the protein more susceptible to aggregation. Both UvsY K58A and UvsY K58A,R60A Retain Heteroprotein Interactions with Gp32 and UvsX-Protein affinity chromatography experiments were performed to examine interactions between the UvsY K58A and UvsY K58A,R60A mutants, respectively, and either UvsX or Gp32, depending on the experiment. Results are summarized in Table II. When either UvsY K58A or UvsY K58A,R60A was passed over a column containing Gp32 immobilized on agarose beads, it bound quantitatively, requiring a salt concentration of 200 mM NaCl to elute. When either mutant protein was passed over a UvsX-agarose column, it also bound quantitatively and required 200 mM NaCl for elution. However, when either UvsY K58A or UvsY K58A,R60A was passed over a bovine serum albumin-agarose control column, neither was retained, and each respective protein was recovered quantitatively in the flow-through fraction. In separate experiments, wild-type UvsY protein also bound to the same Gp32and UvsX-agarose columns quantitatively, and could be eluted off the UvsX column with 600 mM NaCl, and the Gp32 column with 400 mM NaCl. This agrees with the previously published values of 450 and 360 mM NaCl for elution of UvsY from UvsXand Gp32-agarose columns, respectively (7). The data indicate that both UvsY K58A and UvsY K58A,R60A mutants retain the ability to bind to UvsX and Gp32 proteins specifically, albeit with somewhat reduced stabilities compared with wild-type UvsY (Table II).
Etheno-DNA Binding Properties of UvsY K58A and UvsY K58A,R60A Mutants-The ssDNA-binding defects of UvsY K58A and UvsY K58A,R60A , first revealed by ssDNAcellulose affinity chromatography (see above), were examined in greater detail using etheno-DNA (⑀DNA) fluorescence enhancement assays. Previous studies established that both wildtype and COOH-terminal truncated forms of UvsY enhance the fluorescence of ⑀DNA (15,24), allowing binding to be detected and quantified by this method. Here, we compared the abilities of UvsY K58A,R60A , UvsY K58A , and wild-type UvsY to enhance ⑀DNA fluorescence. Protein-⑀DNA interactions were studied as a function of NaCl concentration, using salt-back titration profiles as indicators of relative binding affinity. The experiments depicted in Fig. 3 used a poly(d⑀A) lattice with an average chain length of 310 nucleotide residues. Similar results were obtained with random-sequence ⑀DNA derived from M13mp19 ssDNA circles (data not shown). The salt-back assay consisted of preincubating ⑀DNA with a slight excess of native UvsY, UvsY K58A , or UvsY K58A,R60A (1.2-fold saturating, assuming a binding site size of n ϭ 4 nucleotide residues per monomer, Ref. 15) in a low-ionic strength buffer containing 5 mM NaCl, then titrating each starting mixture with increasing concentrations of NaCl. Results are shown in Fig. 3.
The double mutant UvsY K58A,R60A exhibited no detectable increase in ⑀DNA fluorescence, relative to ⑀DNA-only control, under any salt conditions examined (5-75 mM NaCl in 20 mM Tris-HCl (pH 7.4), 1 mM MgCl 2 ) (Fig. 3). Thus UvsY K58A,R60A appears to be devoid of ssDNA binding activity by two different criteria: ⑀DNA and ssDNA-cellulose methods (see Table II). UvsY K58A enhanced the fluorescence of ⑀DNA at NaCl concentrations Ͻ25-30 mM (Fig. 3), indicating very weak, residual ssDNA binding activity for this single-mutant form. The apparent salt midpoint for dissociation of UvsY K58A -⑀DNA complexes was ϳ15 mM NaCl, and the complexes appeared to be completely dissociated by [NaCl] ϭ 30 mM, thus explaining why no binding of this mutant to ssDNA-cellulose was observed in buffer containing 100 mM NaCl (Table II). UvsY K58A binding to ⑀DNA appears to be even weaker than that of COOH-terminal truncated forms such as UvsY*, which exhibits ⑀DNA saltdissociation midpoints of 65 mM NaCl and 145 mM KOAc, respectively, and affinity for ssDNA generally 10 4 -fold lower than wild-type UvsY (24). Even allowing for some anion-specific stabilization of UvsY K58A -⑀DNA interactions (i.e. improved binding in acetate versus chloride as seen with UvsY and UvsY*), it is questionable whether UvsY K58A could form any productive complexes with ssDNA at all under buffer conditions used in ATPase and strand exchange assays described in subsequent sections.
Control experiments with wild-type UvsY validate the results of our ⑀DNA binding studies. Wild-type UvsY induces a large enhancement of ⑀DNA fluorescence at low NaCl concentrations (Fig. 3), consistent with previous results (15). Titration of wild-type UvsY-⑀DNA complexes with NaCl causes the magnitude of fluorescence enhancement to decrease (Fig. 3); eventually a plateau is reached in which an ϳ3-fold level of fluorescence enhancement is maintained up to an NaCl concentration of ϳ300 mM (data not shown). Further titration leads to UvsY-⑀DNA complex dissociation at a salt midpoint of ϳ450 mM NaCl (data not shown). The behavior of wild-type UvsY-⑀DNA complexes in this study parallels that observed by Sweezy and Morrical (15,16). Therefore the poor to non-existent binding of UvsY K58A and UvsY K58A,R60A , respectively, to ⑀DNA reflects true ssDNA-binding defects in these mutant proteins, and not a failure of our fluorescence assays to detect protein-⑀DNA complex formation.
UvsY K58A and UvsY K58A,R60A Are Partially Defective in Stimulating UvsX-catalyzed ssDNA-dependent ATPase Activity-Gp32 protein inhibits UvsX-catalyzed ssDNA-dependent ATP hydrolysis by competing with UvsX for binding sites on ssDNA (7,9,38). Wild-type UvsY overcomes this inhibition by nucleating UvsX filament formation onto Gp32-ssDNA complexes, and presumably by helping UvsX to displace Gp32 from the ssDNA (9,12,39). We tested the abilities of UvsY K58A and UvsY K58A,R60A mutants to stimulate UvsX-catalyzed ATPase activity in the presence of Gp32-saturated ssDNA. Results are summarized in Table III. Reactions were performed under conditions in which the ssDNA-dependent ATPase activity of UvsX is absolutely dependent on functional UvsY protein, i.e. subsaturating UvsX (0.2-fold) and saturating Gp32 (2.3-fold) with respect to binding sites on ssDNA (assuming n ϭ 4 and 7 nucleotide residues for UvsX and Gp32, respectively (36, 40 -42)). Under these conditions, reactions lacking UvsY had velocities indistinguishable from background, whereas the addition of wild-type UvsY (1.2-fold molar excess with respect to  (14). b Data derived from Fig. 2A, this study. c Data derived from Fig. 2B, this study. a For bovine serum albumin-agarose control column, FT denotes protein eluted in the 50 mM NaCl flow-through fraction (RB-50 buffer).
b For ssDNA-cellulose affinity column, FT denotes protein eluted in the 100 mM NaCl flow-through fraction (DC-100 buffer). UvsX) produced a high level of UvsX-catalyzed ATPase activity (Table III). With an identical concentration of either UvsY K58A or UvsY K58A,R60A replacing wild-type UvsY in the reaction, the ATPase velocity was approximately one-third of that obtained with wild-type UvsY (Table III). Thus, the two UvsY mutants are only partially defective in facilitating UvsX-ssDNA complex assembly on Gp32-saturated ssDNA. UvsY K58A and UvsY K58A,R60A retain significant and approximately equal abilities to promote UvsX-catalyzed ssDNA-dependent ATP hydrolysis under these conditions. Note that control reactions lacking UvsX enzyme had velocities indistinguishable from background, demonstrating that all three UvsY species were free of contaminating ATPase activity (Table III).
Effects of UvsY K58A and UvsY K58A,R60A Mutants on UvsXcatalyzed DNA Strand Exchange-We compared the abilities of wild-type UvsY, UvsY K58A , and UvsY K58A,R60A to stimulate UvsX-catalyzed DNA strand exchange reactions. First, conditions were established in which DNA strand exchange is co-dependent on UvsX and UvsY. Fig. 4 (lanes 1-4) shows that at a UvsX concentration of 50 g/ml (1.2 M), which is 100% satu-rating with respect to binding sites on ssDNA (24), DNA strand exchange occurs readily in the absence of UvsY protein. Reactions are characterized by the formation of DNA networks that do not enter the gel, a reaction outcome typical of UvsX-catalyzed strand exchange (7,10,18). Each reaction sample was treated with Proteinase K prior to gel loading to ensure that the aggregates that arose were results of the strand exchange and not from nonspecific aggregation of DNA by binding proteins (13,43). Networks appear only 5 min after reaction initiation, the earliest time point examined (Fig. 4, lane 2). In contrast to these results, no strand exchange products are observed over a 40-min time course when the UvsX concentration is lowered to 10 g/ml (0.23 M), which is only 20% saturating with respect to binding sites on ssDNA (Fig. 4, lanes [5][6][7][8][9]. Under these restrictive conditions, the addition of wildtype UvsY protein (5 g/ml or 0.31 M, a slight molar excess with respect to UvsX) activates the DNA strand exchange reaction, and the majority of 32 P-labeled RFIII substrate DNA is rapidly converted to well bound networks (Fig. 4, lanes  10 -14). Smaller D-loop intermediates, which do enter the gel but run slower than the 32 P-linear dsDNA substrate, are also evident early in the reaction but are rapidly converted into networks (Fig. 4, lanes 10 -14). Substitution of an equal concentration of UvsY K58A,R60A double mutant for wild-type UvsY under otherwise identical conditions causes a severe defect in DNA strand exchange (Fig. 4, lanes 20 -24). The reaction is not abolished, but formation of the network products is largely suppressed over the 40-min reaction time course, whereas Dloop intermediates accumulate at a greatly reduced rate. Otherwise identical experiments performed with UvsY K58A reveal that this single mutant is also partially defective in activating DNA strand exchange (Fig. 4, lanes 15-19). The defect is not as severe as that observed with UvsY K58A,R60A , but nevertheless, results in a lag time of about 10 min before network products begin to form, relative to reactions with wild-type UvsY. Thus the UvsY K58A and UvsY K58A,R60A mutants affect strand exchange differentially under restrictive conditions in which the reaction is co-dependent on UvsX and UvsY.

TABLE III
Velocities of UvsX-catalyzed ssDNA-dependent ATP hydrolysis in the presence/absence of UvsY wild-type and mutant proteins Spectrophotometric ATPase assays were performed as described under "Materials and Methods." Complete reactions contained 2 mM ATP, 4.5 M (nucleotides) M13mp19 ssDNA, 10 g/ml (0.23 M) UvsX, and 50 g/ml (1.5 M) Gp32. UvsY wild-type or mutant proteins, when present, were at a concentration of 5 g/ml (0.31 M). All other reaction components and conditions were as described under "Materials and Methods." Controls lacked UvsX protein, but were otherwise identical to "complete reactions" in the same row of the These studies further elucidate the details of the mechanism used by the T4 UvsY recombination mediator protein to facilitate the assembly of phage presynaptic filaments. Our attempts to pinpoint residues critical for ssDNA binding in the UvsY protein led us to the LKARLDY motif at positions 57-63 by several lines of reasoning. 1) This short motif is conserved in a number of DNA-binding proteins (25); 2) it is located in the NH 2 -terminal portion of UvsY protein, where previous studies showed the protein-ssDNA interactions to reside (13,24); and 3) it contains alternating large, polar, positively charged amino acid residues, consistent with evidence that the UvsY-ssDNA interactions are highly electrostatic in nature (15). As predicted, site-directed mutagenesis of basic residues within the LKARLDY sequence interrupts the ability of UvsY to bind ssDNA (Figs. 1 and 3; Table II). The disruption of protein-ssDNA interactions appears to be total with the UvsY K58A,R60A double mutant, whereas single mutant UvsY K58A retains very weak ssDNA binding activity observable only at very low salt concentrations (Fig. 3). This evidence establishes the conserved LKARLDY motif as a likely site of ssDNA contacts within UvsY protein.
Analysis by circular dichroism (data not shown) suggests that the two LKARLDY motif mutant proteins have slightly increased ␣-helicity relative to wild-type UvsY. This effect is most noticeable in the double mutant, UvsY K58A,R60A . Further experiments will be needed to establish whether the loss of ssDNA binding activity associated with these mutations is caused by a loss of direct electrostatic contacts with ssDNA, or by changes in local structure that disrupt the ssDNA binding site. In either case, it is clear that the single K58A and double K58A,R60A mutations do not disrupt the global fold of the UvsY protein, because both mutants retain self-and heteroprotein-association properties similar to wild-type ( Fig. 2 and Tables I and II). These results are consistent with our previous observation that residues in the COOH-terminal domain of UvsY (residues 102-137) are essential for UvsY-UvsX, UvsY-Gp32, and UvsY-UvsY interactions (13,24).
Having met our goal of isolating species of the UvsY protein that are deficient in ssDNA binding but otherwise functionally and structurally intact, we next determined how the loss of ssDNA binding influences the fundamental roles of UvsY in recombination mediation. Under conditions in which UvsXcatalyzed ssDNA-dependent ATP hydrolysis is co-dependent on UvsY, both the single (UvsY K58A ) and double (UvsY K58A,R60A ) mutant proteins exhibit a two-thirds reduction, compared with wild-type, in the ability to activate this reaction (Table III), indicating a partial defect in mediated presynaptic filament assembly. Likewise, both UvsY mutants show greatly reduced abilities to mediate UvsX-catalyzed DNA strand exchange reactions (Fig. 4). In this case the strand exchange defect is more severe with the double than with the single mutant form of UvsY (Fig. 4), suggesting that the residual, extremely weak ssDNA binding ability observed with UvsY K58A (Fig. 3) might facilitate some function in promoting DNA strand exchange that is not available in UvsY K58A, R60A . Surprisingly, neither mutant is totally defective in facilitating UvsX-catalyzed ssDNAdependent ATPase and DNA strand exchange reactions.
Based on previous biochemical studies of UvsY properties (13)(14)(15)(16)24), we proposed a mechanistic model for UvsY-mediated assembly of the T4 presynaptic filament, which is shown in Fig. 5A. This model includes a critical role for UvsY-ssDNA interactions in destabilizing Gp32-ssDNA interactions, which is necessary for nucleating UvsX-ssDNA filament assembly. The destabilization of Gp32-ssDNA occurs independently of UvsY-Gp32 protein-protein interactions (16). The proposed mechanism of destabilization involves wrapping of ssDNA around hexameric UvsY, which would disrupt cooperative interactions between neighboring Gp32 molecules. Based on this model, one might reasonably predict that mutations abrogating UvsY-ssDNA interactions would completely inhibit UvsY mediator function, but results of our current study show that this is not the case. Instead, the data indicate that UvsY heteroprotein interactions alone are sufficient to promote an attenuated level of presynaptic filament assembly and to provide partial activation of UvsX-catalyzed reactions. Because UvsY-Gp32 interactions are not critical for the Gp32-ssDNA destabilization step of presynapsis (Ref. 16, see Fig. 5A), we propose that the partial mediator activities observed with UvsY K58A and UvsY K58A,R60A mutants arise predominantly through UvsY-UvsX interactions, as shown schematically in Fig. 5B. Here, binding of UvsY K58A or UvsY K58A,R60A to UvsX induces a high affinity ssDNA binding conformation of the recombinase. The high affinity UvsY mutant-UvsX complex competes more effectively with Gp32 for binding sites on the ssDNA than does UvsX alone, allowing the formation of a limited number of nucleation sites for presynaptic filaments (Fig. 5B).
The models in Fig. 5, A and B, predict that the wild-type UvsY protein uses at least two different mechanistic effects to mediate presynaptic filament assembly. The first effect is destabilization of Gp32-ssDNA interactions, an effect dependent on UvsY-ssDNA interactions (16). This idea is consistent with the observation that the isolated ssDNA-binding domain of UvsY can still weakly stimulate UvsX-catalyzed reactions in the presence of ssDNA-saturating concentrations of Gp32 (13). The second effect is induction of high affinity UvsX-ssDNA interactions, an effect dependent on UvsY-UvsX interactions. The latter idea is supported by the observation that the salt stability of T4 presynaptic filaments is increased in the presence of UvsY (10), and by fluorescence anisotropy data suggesting that UvsY and the nucleotide analog ATP␥S (which stabilizes UvsX-ssDNA complexes; Ref. 40) induce similar conformational changes in UvsX protein. 4 Other effects are possible and are not necessarily mutually exclusive with the models shown in Fig. 5, A and B. The role of UvsY-Gp32 interactions, although poorly understood, could be important for maintaining appropriate spatial organization of the four macromolecular components, UvsY, Gp32, UvsX, and ssDNA, during the early stages of presynaptic filament assembly. This idea is presented schematically in Fig. 5C. Such an effect could help to explain the partial activities of UvsY mutants described in this study, i.e. Gp32 could recruit and orient the UvsY mutant into a suboptimal but nucleation-competent complex, even though the ssDNA binding activity of the mutant is compromised. This could, in turn, explain why the UvsY K58A single mutant performs better than the UvsY K58A,R60A double mutant in strand exchange reactions (Fig. 4); the residual ssDNA binding activity of UvsY K58A might make it that much easier to form a reasonably ordered nucleation complex in response to interactions with Gp32 and/or UvsX. The scheme in Fig. 5C might also explain the observation that UvsY cannot assemble UvsX onto ssDNA covered with Gp32-A, a truncated form of Gp32 lacking the domain for protein-protein interactions with UvsY and UvsX (21). Although UvsY destabilizes Gp32-A-ssDNA complexes (16), presumably Gp32-A fails to properly orient UvsY molecules for subsequent downstream steps in presynapsis and strand exchange.
The involvement of the LKARLDY motif of the UvsY protein in ssDNA binding suggests that this sequence element might be important for polynucleotide binding in other proteins as well. In addition to its conservation in the human ERCC2 nucleotide excision repair protein and the yeast Rad3 helicase (25), close matches appear in other DNA repair proteins including the UvrB excision repair protein of Thermus thermophilus (44), the RecB protein (exonuclease V ␤-subunit) of Pseudomonas syringae (Protein Data Bank accession number AAL79572), and the ERCC2/TFIIH orthologs of various eukaryotes (22, 23, 25, 45, 46 -49). Each of these proteins is known to bind to DNA. Because mutagenesis of the LKARLDY motif in T4 UvsY protein affects recombination functions, then similar mutations in other proteins conceivably could alter DNA recombination/repair outcomes and thereby impact genomic stability.
The close functional conservation of RMPs from bacteriophage to humans clearly demonstrates their importance in homologous recombination and recombinational DNA repair (17). Studies of the role of the T4 RMP, UvsY protein, have illuminated some of the biochemical trickery used to remodel protein-ssDNA complexes and to assemble enzymatically active 4 J. Farb and S. Morrical, unpublished data.
ssDNA-binding conformation of UvsX via protein-protein interactions. The high affinity form of UvsX competes more effectively with Gp32 for binding sites on ssDNA, allowing an attenuated level of filament nucleation to occur. C, UvsY LKARLDY-motif mutants lacking ssDNA binding activity could also act as adapters between UvsX and Gp32-ssDNA, using protein-protein interactions to tether UvsX to the complex. The tethered UvsX could displace Gp32 from the ssDNA locally by adopting a high affinity conformation (as in B) and/or via high local concentration effects.  (17) and Bleuit et al. (8). UvsY destabilizes Gp32-ssDNA interactions, an effect mediated predominantly by UvsY-ssDNA interactions. UvsY then recruits UvsX recombinase via protein-protein interactions, nucleating presynaptic filament assembly with concomitant expulsion of Gp32 from the complex. B, UvsY LKARLDY-motif mutants lacking ssDNA binding activity may still promote presynaptic filament assembly by inducing a high affinity presynaptic filaments from inactive precursors. Through continued physical and biochemical studies of UvsY and its mutants, we hope to further refine a rigorous model of UvsY structure and function, one that will hopefully shed light on recombination mechanisms used by many organisms.