Homologous pairing in genetic recombination. Purification and characterization of Escherichia coli recA protein.

RecA protein, which is essential for genetic recombination in Escherichia coli, was extensively purified from a strain of E. coli which contained the recA gene cloned in a plasmid (Sancar, A., and Rupp, W. D. (1979) Proc. Natl. Acad. Sci. U. S. A. 76, 3144-3148). Using the DNA-dependent ATPase activity of recA protein as an assay, we obtained about 60 mg of purified recA protein from 100 g of cells. Ten micrograms or 1 microgram of the purified protein exhibited only one detectable band with Mr approximately = 40,000 upon sodium dodecyl sulfate-acrylamide gel electrophoresis. More than 99% of the ATPase activity of purified recA protein was dependent on single-stranded DNA. Purified recA protein had no detectable DNase, topoisomerase, or ligase activities. The enzyme was stable for a least a year when stored at 0-4 degrees C. The half-life of the ATPase activity of 25 microM recA protein was 37 min at 51 degrees C. Purified recA protein binds to single-stranded and double-stranded DNA, unwinds duplex DNA by a mechanism that is stimulated by single-stranded DNA or oligonucleotides, and pairs homologous single strands with duplex DNA.

discovered that recA+ function is required for the induction by mitomycin C or ultraviolet light of gene expression at no less than five loci. Until recently, the striking pleiotropy of recA mutations made it appear likely that recA played an indirect role in recombination; indeed, no published evidence indicated a direct role of the recA gene in recombination until Kobayashi and Ikeda (1978) demonstrated the effect of a thermosensitive recA mutant on molecular recombination in the absence of either RNA or protein synthesis. The discovery of the DNA-dependent ATPase activity of recA protein by Ogawa et al. (1979) and by Roberts et al. (1979) emphasized the interaction of recA protein with DNA (Gudas and Pardee, 1976;Satta et al., 1979). RecA protein was fiist purified by radiochemical methods (Ogawa et al., 1979) and as a protease that cleaved h repressor (Roberts et al., 1978). The discovery of the ATPase activity of recA protein (see above) and the cloning of the gene on a multicopied plasmid (McEntee and Epstein, 1977;Ogawa et al., 1979;Sancar and Rupp, 1979) further facilitated the purification of the protein Shibata et al., 1979a). Stimulated by the experiments of Holloman and Radding (1976), which suggested that the recA gene might play a role in the pairing of a single strand with duplex DNA to produce a D-loop , and by the experiments cited above, which indicated that recA protein might act directly in recombination, we purified the recA protein and found indeed that it catalyzed the formation of D-loops (Shibata et al., 1979a). The same discovery was made by McEntee et al. (1979). This paper describes the purification and properties of recA protein.

DNA
In this paper, the amounts of DNA or oligodeoxynucleotides are expressed in moles of nucleotide residues. Form I DNA' and single-stranded circular DNA of phages fd and +X174 were prepared as described by . As we described before , form I DNA sometimes gives a high background in the D-loop assay (see below), which can be eliminated by heating the preparation of form I DNA at 68 "C for 50-60 s in 10 mM Tris.HC1 (pH 7.5), 0.1 mM EDTA. Form IV DNA was prepared as described by Pulleyblank and Morgan (1975).
We prepared form I1 DNA by two methods. The fwst, digestion of form I DNA by S , nuclease (Wiegand et al., 1975), produced some form I11 DNA as well. The reaction mixture contained 0.65 mM form I DNA, 50 mM Na-acetate buffer (pH 5.0), 0.5 mM ZnS04, 100 mM NaCI, 1250 or 2500 units (Wiegand et al., 1975) of SI nuclease (Miles Biochemicals) in a total volume of 0.2 ml . After an incubation of 30 min at 37 "C, the sample was extracted twice with phenol and twice The duplex forms of DNA of the small DNA phages are designated as follows: form I, negatively superhelical DNA; form 11, circular duplex DNA with one or more interruptions in either strand; form 111, linear duplex DNA, form IV, closed circular duplex DNA that was fully relaxed under the conditions of closure. 7557 with ether. We evaporated the ether by carefully bubbling N, through the sample and then dialyzed overnight against 500 ml of 10 mM Tris. HCI (pH 7.5), 1 mM EDTA. As judged by gel electrophoresis and counting of radioactivity extracted from the bands of form I1 and form 111 DNA (see below), 1250 units of S1 nuclease (Wiegand et al., 1975) produced a preparation containing 70% form I1 and 30% form I11 DNA; 2500 units of S1 nuclease produced 62% form I1 and 38% form I11 DNA. Form I1 DNA, with a single nick and free of any form I11 DNA, was prepared by treatment of form I DNA with pancreatic DNase in the presence of ethidium bromide, according to Shortle et al. (1979). The reaction mixture (100 p l ) , which contained 50 mM Tris.HC1 (pH 7.5), 5 mM MgCL, 150 pg of ethidium bromide/ml, 540 p~ form I DNA, 2.0 pg of pancreatic DNase/ml (Worthington, 3200 units/mg), was incubated at 25 "C for 60 min. The reaction was terminated by the addition of %o volume of 1 M TrisaHCI (pH 9) and 1 volume of cold phenol saturated with 1.5 M NaC1, 0.15 M Na-citrate; the phases were mixed by agitation at 4°C. After centrifugation, the upper aqueous phase was collected, and remaining ethidium bromide was extracted 4 times by isoamyl alcohol. The solution of DNA was dialyzed against 10 mM Tris.HCI (pH 7.5), 0.1 mM EDTA. The fractions of form 11, form 111, and form I DNA were estimated from the radioactivities in the bands corresponding to each form after electrophoresis through a 1.4-2.2% agarose gel slab as described previously (Shibata et al., 1979a). In order to confirm that our preparations of form I1 DNA consisted principally of circular DNA with a single nick, we denatured it by incubation at pH 12.3 at room temperature for 3 min before gel electrophoresis. After denaturation, form I1 DNA with a single nick exhibited two discrete bands corresponding to circular and linear single-stranded DNA upon electrophoresis through 2.2% agarose (Matsumoto et al., 1979). A boiled DNase I limit digest of @X174 form I DNA was prepared as described previously (Cunningham et al., 1979). Fragments of single-stranded DNA were prepared by boiling 840 p~ single-stranded phage DNA (Cunningham et al., 1979) in 40 pl of 10 m~ Tris.HCI (pH 7.5), 0.1 mM EDTA for 10-15 min in a centrifuge tube (Eppendorf micro-test tubes, 1.5 ml), followed by quick chilling in ice water. The median chain length of the fragments was estimated by electrophoresis through a 2% agarose gel. A digest of 4x174 singlestranded DNA produced by endonuclease R Hue I11 provided fragments of known chain length which served as standards. The lengths covered the range from 70 to 1350 nucleotide residues (Blakesley and Wells, 1975;Horiuchi and Zinder, 1975;Sanger et al., 1977). Boiled single-stranded DNA consisted of fragments the size of which was broadly distributed. The estimated median chain length of the singlestranded fragments of fd was 600 nucleotide residues and that of the fragments of @X174 was 1200 nucleotide residues.

Strains of E. Coli K12
The strains used in this study were derived from KM4104; mtlA strA lysA argA A(lac),n (deletion of entire lac operon) A7 (srl-recA) A2134 (gal-bio) (McEntee, 1977) by Sancar and Rupp (1979) and generously provided by them. DR1453 contained plasmid pDR1453 which had the 6.8 kilobase pairs and 1.8 kilobase pairs PstI fragments inserted in pBR322; this plasmid made the strain Srl+ and Rec+. DR1461 contained plasmid pDR1461 which carried a part of the recA gene; the strain was recA-. DR4322 contained the parental plasmid, pBR322, and the strain was recA- (Sancar and Rupp, 1979).

Assay of ATPase Activity of RecA Protein
The standard reaction mixture for the assay of ATPase contained, in 18 p1,35 mM Tris. HCl (pH 7.5), 6.7 mM MgC12, 2 mM dithiothreitol, 100 pg of bovine serum albumin/ml, 1.44 mM [3H]ATP (total 26 nmol, stranded DNA of phage 4x174. The reaction mixture was incubated 7.7 Ci/mol; Amersham), recA protein plus or minus 50 p~ singleat 37 "c for 30 min in a well of a microtiter plate (Linbro/Titertek) floating on a water bath. The reaction was terminated by chilling to 0 " c and by addition of 12 pl of 25 mM EDTA containing 3 mM each of unlabeled ATP, ADP, and AMP as carrier. Strips (1 x 10 cm) of polyethyleneimine were prepared by scoring sheets of polyethyleneimine on a plastic film (10 X 10 cm, Polygram Cel300PE1, Macherey-Nagel Co.) in lines at intervals of 1 cm. Strips were soaked in 0.5 M LiCl2, 1 M formic acid, washed extensively with distilled water, and dried before use (Scott et al., 1977). An aliquot of 10 pl from each sample was spotted at 1.5-2 cm from the bottom, dried, and developed in a solvent of 0.5 M LiC12, 1 M formic acid. The spots of ATP, ADP, by ascending chromatography at room temperature for about 40 min and AMP (in this order from bottom to top) were located under illumination by a UV lamp (at 254 nm). Each spot was cut out and put in a vial for scintillation counting so that the layer of polyethyleneimine faced upward. (If the layer faced the bottom of the vial, the efficiency of counting 'H decreased 30%) Econofluor was added to the vials, and radioactivity was measured in a scintillation counter. One unit of ATPase is the amount that hydrolyzes 1 nmol of ATP under these conditions (Shibata et al., 1979a). The concentration of recA protein is expressed as moles of polypeptide of M , = 40,000.
Homologous Pairing Reaction, Unwinding of Duplex DNA, and the Formation of Complexes of RecA Protein and DNA The standard reaction mixture contained, in 20.5 pl, 31 mM Tris. HC1 (pH 7.5), 6.7 mM MgClz, 2.0 mM spermidine-HCI, 1.3 mM ATP, 1.8 mM dithiothreitol, 88 pg of bovine serum albumin/ml, 4.4, 8.8, or 17.6 p~ double-stranded fd or 4x174 ['HIDNA, 6 or 12 p~ singlestranded DNA (or fragments) of fd or @X174, various amounts of purified recA protein. To avoid precipitation of single-stranded DNA by spermidine (Christiansen and Baldwin, 1977), we incubated DNA, proteins, and ATP at 37 "C for 4 min in the presence of 1.2 mM MgCL before adding spermidine. To start the reaction, we added spermidine, increased the concentration of MgClz to 6.7 mM, and incubated the reaction mixture at 37°C. (Recently, we have altered the order of addition slightly, adding ATP together with spermidine and more Mgzi, after the preincubation of DNA with recA protein.) Microtiter plates were used as described above (see ATPase assay). For the unwinding of duplex DNA or the formation of complexes of DNA and recA protein, 0.5 mM adenosine 5'-0-(3-thiotriphosphate, (Boehringer Mannheim GmbH) replaced ATP, and spermidine was omitted (Cunningham et al., 1979;Shibata et al., 1979b).
Assay C-This assay measured the retention by nitrocellulose filters of complexes of protein and DNA (Shibata et al., 1979b). After DNA and recA protein were incubated at 37 "C for 30 min, we diluted the reaction mixture IO-fold with cold BD buffer which contained 31 mM Tris. HCI (pH 7.5), 6.7 mM MgC12, 1.8 mM dithiothreitol, 88 pg of bovine serum albumin/ml with or without NaCl. We filtered the sample at 1 m1/10 s through nitrocellulose filters (Millipore DAWP; pore size, 0.65 pm) which had been washed with 2 ml of BD buffer.
The filter was then washed with 1 ml of BD buffer, dried, and put in a vial with Econofluor (New England Nuclear). Radioactivity was counted in a scintillation counter.
Assay of D-loops-This assay measured the product of homologous pairing of double-stranded DNA with single-stranded or partially filters of %-labeled double-stranded DNA by virtue of its association single-stranded DNA. We measured the retention by nitrocellulose with single-stranded DNA Shibata et al., 1979aShibata et al., , 1980. After terminating the reaction by adding 2 volumes of 25 mM EDTA (pH 9.4), we treated the product with 0.5% Sarkosyl (NL97, Ciba-Geigy Co.) at 17'C for 5 min and then diluted the mixture about 15 times with 25 mM EDTA (pH 9.4). We took an aliquot of 50 pl to measure total counts and an aliquot of 200 pl for the nicking assay of Kuhnlein et al. (1976). Then we diluted 200 p, l of the mixture &fold with cold 1.5 M NaCI, 0.15 M Na-citrate, incubated at 41 "C for 4 min and immediately diluted it 7-fold with cold NaCl and Na-citrate. We filtered the sample at about 4 m1/10 s through a nitrocellulose filter (Sartorius membrane filter, SM11306; pore size, 0.45 pm) which had been washed with 2 ml of NaCl, Nacitrate. We washed the filter successively with 1.5, 1.5, and 5 ml of cold NaC1, Na-citrate. Radioactivity retained on the filter was counted as described above.
Assay of Unwinding of Duplex DNA-After form I1 DNA was Purification of RecA Protein 7559 incubated with recA protein and ATP-@' at 37 "C for 30 min in the mixture described above, the reaction mixture was chilled on ice and diluted 2-fold with cold BD buffer. We added KCI, (NHM04, and NAD to make their concentrations 25 mM, 10 mM, and 25 PM, respectively, and incubated the mixture at 18°C for 5 rnin. E. coli DNA ligase (0.3 unit, Miles Laboratories) was added to the sample, 50 pl total volume, and incubated for 90 min. The reaction was stopped by adding EDTA (28 mM) and sodium dodecyl sulfate (0.17%). We added proteinase K (0.2 mg/ml, Boehringer Mannheim GmbH) to the sample and incubated at 37OC for 15 min. Samples were subjected to electrophoresis through a 1% agarose gel slab (16 x 16 x 0.3 cm) in E buffer (Sharp et al., 1973) which contained 40 mM Tris. acetate buffer, 5 mM Na-acetate, 1 mM EDTA at pH 8.0. Current was applied at 35 V for 16-18 h, and the buffer was recirculated between the reservoirs, following which we stained the gels for 2 h in E buffer containing 0.5 pg of ethidium bromide/ml. We illuminated the gels from below with a short wavelength UV lamp (Ultra-Violet Products) and photographed them through a red filter (Kodak, Wratten 23A) with Polaroid Type 55 Land fiim.
Growth of Cells We inoculated 500 ml of K medium with cells of strain DR1453 which had grown on a 56-srl plate and incubated the culture at 37 "C overnight with aeration. On the next morning, the A5wnrn of the culture was 1 to 2. We diluted this culture with 2.2 liters of SLBH medium and incubated (2nd incubation) the culture in a Hi-Density Fermentor (Lab-Line Instruments No. 29500) at 30 "C with aeration (6 liters/min). The vessel of this fermentor was rotated at 250 rpm. Thirty minutes later, we added 20 ml of 50% (v/v) glycerol. When the "rn reached 1.8, we replaced air with oxygen (4 liters/min). Four to 4.5 h after the start of the second incubation, when the ASSO",,, was 5 to 8 and growth was still logarithmic, we added 140 mg of nalidixic acid (Sigma) dissolved in 50 ml of 0.03 N NaOH and incubated the culture for an additional hour. We collected the cells by centrifugation at 9000 rpm for 10 min in a Sorvall GS3 rotor at room temperature, resuspended the cells (32 to 38 g, wet weight) in about 45 ml of 50 mM Tris. HCI (pH 7.5) containing 10% (v/v) sucrose at room temperature, and froze the cell suspension quickly with ethanol-dry ice. We stored the frozen cells at -20°C. About 1 h before the addition of nalidixic acid, we sampled the culture (about 200 pl) for tests of Srl+ phenotype and UV sensitivity.
Purification of RecA Protein Cell-free extracts-The cell suspension, 65 g, was thawed in an icewater bath for about 12 h. To this suspension we added 1/100 volume each of 1 M dithiothreitol and 0.1 M EDTA and 1/20 volume of 1% lysozyme. We incubated the suspension at 0 "C for 30 min. After the incubation, we added 3.5 M KC1 solution and 8% Brij 58 (Sigma) to final concentrations of 0.2 M and 0.42%, respectively, and incubated again at 0°C for 30 min. We centrifuged the viscous cell lysate at 35,000 rpm for 60 min in a Type 45 Ti rotor (Beckman) and saved the supernatant (Table I; fraction I, 276 ml, 5.9 g of protein).
Polymin P Precipitation-Eleven ml of a 10% solution of Polymin P (pH 7.9) (Aldrich) were added to 276 ml of fraction I over a 15-min period, and the solution was stirred for 20 min more. The precipitate was collected by low speed centrifugation (15,000 rpm for 10 min in an SS34 Sorvall rotor) and resuspended in 65 ml of buffer A containing 0.5 M NaC1. After the suspension was stirred for 30 min, the precipitate was collected by low speed centrifugation. The precipitate was resuspended in 100 ml of buffer A containing 1 M NaCl and homogenized by means of a Teflon pestle over a 60-min period. The suspension was centrifuged and the supernatant, about 102 ml, was saved. Ammonium sulfate was added to 50% saturation over a 45-min period and the mixture was stirred for 30 min. The precipitate was collected by low speed centrifugation. This precipitate can be stored at -20 "C without loss of activity. The precipitate was dissolved in approximately 100 ml of buffer A containing 1 M NaCl and again precipitated by the addition of ammonium sulfate to 50% saturation. This process was repeated 2 times. The precipitate was finally dissolved in 12 ml of buffer B containing 50 mM potassium phosphate (pH 6.8) and dialyzed against two changes of 1 to 2 liters of buffer B. The precipitate that appeared during dialysis was removed by low speed centrifugation. The supernatant, 24 ml, containing 600 mg of protein, is fraction I1 (Table I).
Hydroxyapatite Chromatography-Fraction I1 was applied to a The abbreviation used is: ATPyS, adenosine 5'-0-(3-thiotriphosphate). previously had been equilibrated with buffer B (Fig. 1, A and B). The column was washed with 250 ml of buffer B and developed with a 3.6liter linear gradient (0.02 to 0.5 M) of potassium phosphate in buffer B at 60 d / h . RecA protein (DNA-dependent ATPase activity) was eluted at 0.06 M potassium phosphate. Protein in the pooled fraction (fraction 111, 111 ml, 113 mg of protein) was precipitated by ammonium sulfate at 75% of saturation, collected by low speed centrifugation, and dissolved in 7.4 ml of buffer A containing 0.3 M ammonium sulfate (fraction IIIa, 10 ml).
Sephacryl S200 Gel Filtration-Fraction IIIa was applied to a column (3.2 X 42 cm) of Sephacryl S200 (Pharmacia) which had been equilibrated with buffer A containing 0.3 M ammonium sulfate. The column was developed with the same buffer at 20 ml/h. Fractions with DNA-dependent ATPase but little DNA-independent activity were pooled and dialyzed against buffer B containing 1 mM EDTA (fraction IV, 28 ml, 64 mg of protein).
DEAE-cellulose Chromatography-Ten ml of fraction IV were applied to a column (I X 13.5 cm) of DEAE-cellulose (Whatman, DE52) previously equilibrated with buffer B containing 1 mM EDTA.
The column was washed with 15 ml of the equilibration buffer and developed with a 110-ml linear gradient of KC1 (0 to 0.5 M) in the equilibration buffer, at 12 ml/h (Fig. 1, C and D ) . Active fractions were pooled (between 0.23 and 0.26 M KC1) and dialyzed against 50 mM Tris. HCI (pH 7.5) containing 0.3 mM EDTA, 5 mM dithiothreitol, 10% (v/v) glycerol (fraction V, 10 ml, 12 mg of protein). Samples were stored on ice in a coldroom for at least a year without loss of activity.

Purification of RecA Protein
We purified recA protein from a strain of E . coli (DR1453) carrying the recA+ gene cloned on a plasmid (Sancar and Rupp, 1979). Cells had been treated with nalidixic acid in logarithmic phase of growth to induce extensive synthesis of recA protein (Inouye and Pardee, 1970;McEntee, 1977). To monitor the purification, we assayed the ATPase activity of recA protein, which requires single-stranded DNA as a cofactor (Ogawa et al., 1979;Roberts et al., 1979). We also examined fractions for the presence of a protein with M , = 40,000 by electrophoresis through an acrylamide gel containing 0.1% sodium dodecyl sulfate (Fig. 2). The presence of a distinct band corresponding to a protein of M, = 40,000 in fraction I1 was dependent both on the presence of the recA+ gene in the cells and on the inducing treatment with nalidixic acid. We could not detect the protein of M, = 40,000 in fraction I1 from induced cells of a strain containing only part of the recA gene (DR1461) or from a strain containing no recA+ gene (DR4322) (Sancar and Rupp, 1979).
The first chromatographic step in the purification was a column of hydroxyapatite (Fig. 1, A and B) which yielded two major peaks of protein and ATPase activity. We discarded the first peak which contained a significant amount of DNAindependent ATPase. The second peak contained one major step of purification were examined by electrophoresis through a slab (14 cm X 15 cm X 1.6 mm) of a 10% acrylamide gel containing 0.1% sodium dodecyl sulfate. Electrophoresis was at 20 mA for 5 h. Protein in the gel was stained by Coomassie brilliant blue. About 100 units of DNA-dependent ATPase activity of each fraction were loaded on channels c, d, e, and g. a, 14 pg of fraction I1 from hydroxyapatite column (Fig. 1). b, 60 pg of cell-free extract, fraction I. c, 23 pg of Polymin P-(NH&SOI, fraction 11. d, 14 pg of the pooled fractions from the second peak from the hydroxyapatite column (Fig. l), fraction 111. e, 9.6 pg of pooled fractions from the Sephacryl S200 column, fraction IV. f and g, 1 and 11 pg, respectively, of pooled fractions from the DEAE-cellulose column (Fig. l), fraction V. h, 10 pg of the fraction that contained the greatest activity of DNA-independent ATPase activity from the Sephacryl S200 column. i, 7 pg of fractions 13 and 14 from the DEAE-cellulose column (Fig. 1). appeared as faint bands on gel electrophoresis and reduced DNA-independent ATPase activity to trace levels. In fraction 111,89% of total ATPase activity was dependent on DNA and in fraction V, 99% was dependent (Table I). One or 11 pg of the protein in fraction V exhibited only one detectable band of protein by gel electrophoresis (Fig. 2, lanes f and g). The ratios of DNA-dependent ATPase to UV absorbance (Am) in the peak fractions (fraction 7 through 10) from DEAE-cellu- Under the conditions that we used to study the promotion of homologous pairing by recA protein (Shibata et al., 1979a), fraction V produced less than 0.2% acid-soluble material from either linear double-stranded or single-stranded DNA and caused the nicking of 2% or less of form I DNA. We detected neither topoisomerase nor ligase activity (Cunningham et al., 1979).
Since the major protein in fraction I1 was recA protein (Fig.  2, lane c), the specific activity of single-stranded DNA-dependent ATPase activity increased only 2-fold during further purification (Table I).
We could not detect the pairing activity of recA protein (see below) in fraction 11, but we could detect that activity in fraction I11 and later fractions. In fraction 11, the trapping of double-stranded DNA to a nitrocellulose filter by the D-loop assay was mostly independent of homologous single-stranded fragments.
From 100 g of induced cells, we got about 190 mg of recA protein in fraction I11 and 60 mg in fraction V (Table I).
Purified recA protein behaved as a single polypeptide with molecular weight of about 40,000 on electrophoresis through a 10% acrylamide gel containing 0.1% sodium dodecyl sulfate. The following proteins served as standards: bovine serum albumin ( M , = 67,000), ovalbumin (Mr = 45,000), and the a subunit of E. coli RNA polymerase (Mr = 39,000) (data not shown). On gel filtration through Sephacryl S200 (Pharmacia), recA protein exhibited a single peak near the void volume in a buffer containing 0.3 M (NH4)2S04, which suggests that the active protein is larger than the monomer of M , = 40,000 (see Ogawa et al., 1979). In 0.02 M K phosphate (pH 6.8), 2mercaptoethanol, 10% glycerol, the half-life of the ATPase activity of 25 p~ recA protein (fraction 111) was 37 min at 51 "C and 10-12 min at 52 "C. Similarly, when fraction IV was incubated for 5 min in 30 mM potassium phosphate buffer (pH 7.6) containing 10% glycerol, ATPase activity was not affected below 50"C, but was completely inactivated above 55°C. RecA protein was stable at 0 "C in phosphate buffer (pH 6.8) or Tris buffer (pH 7.5) containing 10% glycerol and either 5 mM 2-mercaptoethanol or 10 mM dithiothreitol, since no loss in DNA-dependent ATPase activity was detected during more than a year of storage in those buffers.
Activities of RecA Protein DNA-dependent ATPase- Ogawa et al. (1979) and Roberts et al. (1979) discovered that recA protein has ATPase activity that depends upon the presence of single-stranded DNA. In our standard assay for DNA-dependent ATPase (Scott et al., 1977;Shibata et al., 1979a), hydrolysis of ATP was linearly dependent on the concentration of recA protein, up to 2 p~ (Fig. 3).
Superhelical DNA (form I DNA) supported the ATPase activity of recA protein well in the presence of 1.2 mM MgC12 at pH 7.5, but relaxed double-stranded DNA, form I1 or form IV, did not work well under these conditions (Figs. 4 and 5).
Superhelical DNA was a good cofactor only when the concentration of MgC12 was 1 to 2 mM (Fig. 4), whereas singlestranded DNA was effective in supporting ATPase activity in Binding of RecA Protein to DNA-In order to assay rapidly the binding of recA protein to DNA, we sought conditions under which the retention of either single-stranded or doublestranded DNA by a nitrocellulose filter would depend upon the presence of recA protein. Two factors appeared to be important in the use of nitrocellulose filters to measure the binding of protein to DNA; these were the kind of filter and the ionic conditions (Table 11) (pore size, 0.45 pm) trapped 24 to 29% of single-stranded DNA in the absence of recA protein or added NaCl (Table 11, line la). The addition of NaCl to the buffer used during the filtration increased the retention of single-stranded DNA either by the Millipore filter DAWP (pore size, 0.65 pm) or by the Sartorius filter, SM11306 (Table 11, lines la and lb). In 1 M NaC1, SM11306 filters retained 100% of single-stranded DNA whereas DAWP filters retained only 13%. NaCl at 0.2 M also signifkantly increased the efficiency of trapping of single-stranded DNA by SM11306 filters but had no effect on the trapping by DAWP filters (Table 11, line 1). Neither SM11306 filters nor DAWP filters retained double-stranded DNA alone in the absence or in the presence of NaCl up to 1 M (Table 11, line 4).
Accordingly, we studied the binding of recA protein to DNA using principally the Millipore filter DAWP with or without 0.2 M NaC1. Under these conditions, recA protein caused the retention of some 20% of single-stranded DNA (Table 11, line 2) but no double-stranded DNA (Table 11, line 5). However, in the presence of the ATP analog, 5'-0-(3-thiotriphosphate), recA protein caused the retention of all of the single-stranded DNA (Table 11, line 3) and 8 to 16% of double-stranded DNA (Table 11, line 6). When we substituted the SM11306 filter for the DAWP filter, recA protein caused the retention of half of the double-stranded DNA in the presence of ATPyS (Table  11, line 6). The effect of recA protein on the retention of DNA by nitrocellulose cannot be attributed to a nonspecific effect of protein, since all of the reaction mixture in these experiments contained 88 pg of bovine serum albumin/ml. We conclude that recA protein binds both to single-stranded and double-stranded DNA in the presence of ATPyS. When added to recA protein and single-or double-stranded DNA, ATPyS increased the retention of DNA by nitrocellulose filters, either by enhancing the binding of DNA to recA protein or by changing the size or shape of the protein-DNA complexes. On the basis of increased stability of complexes, we have argued elsewhere (Shibata et al., 1979a) that ATPyS enhances the binding of recA protein to DNA.
In the presence of ATPyS, single-stranded DNA, whether homologous or not, increases the retention of double-stranded DNA by nitrocellulose filters (Table 11,  the rate of the reaction . '' Number in parentheses is the pore size. MgC12 concentration was 1.1 mM both in the reaction mixture and the buffer used for filtration (BD buffer). ss fragments, fragments of single-stranded DNA.
Pairing of Double-stranded DNA and Homologous Singlestranded Fragments-To study the role of recA protein in homologous pairing, we have used an assay originally designed to detect the uncatalyzed formation of D-loops by superhelical DNA and homologous single-stranded fragments . In addition to the formation of D-loops resulting from the homologous pairing of single-stranded fragments with duplex DNA, this assay reveals the pairing of several other substrates catalyzed by recA protein . The assay detects the retention by nitrocellulose filters of duplex DNA that has acquired a single-stranded region, such as a Dloop, or duplex DNA that has become attached to singlestranded DNA.
The formation of D-loops by recA protein occurs whether the double-stranded DNA is superhelical or not Cunningham et al., 1979) (Table 111). To improve the efficiency of the assay in detecting such a metastable product as a D-loop in nonsuperhelical DNA , we changed the conditions previously used for this assay. We deproteinized the product prior to Titration by treating it with 0.5% Sarkosyl a t 17 "C for 5 min (see McEntee et al., 1979) followed by heating at 41 "C for 4 min in 1.5 M NaC1, 0.15 M Na-citrate. We also changed the standard reaction mixture, since we observed anomalous kinetics in the formation of D-loops from form I DNA under the conditions used in previous studies . The standard reaction mixture used in the present experiments contained 6.7 mM MgC12, 2.0 mM spermidine, 1.3 mM ATP in addition to recA protein and DNA (see "Materials and Methods"). Under these conditions, the requirements for pairing single strands with form I1 DNA were much as seen before for form I DNA.
The pairing reaction required homologous DNA, ATP, recA protein, and divalent cation (Table 111). Spermidine, 2 mM, could only partly replace Mg" (Table 111, lines h-j;. The pairing reaction was inhibited by ATPyS ( Table I, line 8) a competitive inhibitor of the ATPase activity of recA protein (Shibata et al., 197913). In an accompanying paper (Shibata et al., 19811, we describe the characterization of D-loops formed by superhelical and nonsuperhelical DNA.

Requirements for homologous pairing of single-stranded DNA fragments and double-stranded DNA
For purposes of this table, the standard reaction mixture contained fd form I1 DNA and fragments of singIe-stranded fd DNA plus the other components described under "Materials and Methods." Variations from this standard mixture are as noted. Double-stranded DNA (4.4 PM) and single-stranded fragments of DNA (6 p~) were incubated with 2.2 PM recA protein (fraction V) in the standard reaction mixture at 37 "C for 20 min, and D-loops were assayed as described under "Materials and Methods." Fd form I1 DNA was 62% form I1 and 38% form 111; fd form I DNA contained more than 70% form I; +X174 form I1 DNA was free of detectable form I or form I11 DNA. Unwinding of Double-stranded DNA-Since recA protein will catalyze the formation of a D-loop with nonsuperhelical DNA, it must be capable of unwinding duplex DNA to some extent. Initially, we were unable to find direct evidence of unwinding. The key to the unwinding reaction was provided by the observation that heterologous single strands stimulate the interaction of recA protein with duplex DNA (Table 11; Shibata et al., 1979b). We observed extensive unwinding of duplex DNA when we added heterologous single strands or oligodeoxynucleotides to recA protein and form I1 DNA in the presence of ATPyS (Cunningham et al., 1979). Ligation of the form I1 DNA in the presence of these reagents produced negatively superhelical DNA, which we detected by isopycnic ssDNA m M dation of double-stranded DNA. The need to detach protein from the DNA prior to filtration through nitrocellulose imposes another limitation on the assay. As a function of the conditions for removing protein, some joint molecules do not survive the assay (Shibata et ~l . , 1981). Nonetheless, the assay, which is fast and reproducible, has lent itself well to the study of recA protein which pairs a number of topological variants of DNA thereby producing joint molecules that are sufficiently stable and that stick to nitrocellulose (Shibata et ~l . , 1979a(Shibata et ~l . , , 1981DasGupta et QL, 1980;Cunningham et ~l . , 1980).
The recA gene plays a central role in the metabolism of DNA in E. coli, most notably in repair and recombination (see Introduction). In a general way, the importance of having the purified gene product and functional assays is evident. More particularly, since recA protein catalyzes the homologous pairing of DNA molecules, it provides a critical reagent for the study of recombination in uitro, a reagent with which we can hope to discover both steps that lead up to homologous pairing and steps that resolve the intermediates produced. DNA that was not unwound at the time of closure. centrifugation in CsC1-ethidium bromide or by electrophoresis in agarose gels (Fig. 6). In these gels, unwound and ligated DNA appeared as a band at the same position as natural form I DNA. When DNA was ligated but not unwound, it appeared as a ladder-like set of bands between form I11 and form I1 DNA (Fig. 6).
Like the formation of D-loops (Shibata et ~l . , 1979a) no unwinding occurred until the concentration of recA protein exceeded some minimal amount. At a limiting concentration of recA protein, excess single-stranded DNA inhibited unwinding (Fig. 6, a-g). Optimal unwinding occurred when the ratio of single-stranded DNA to recA protein was 2.5 to 5 residues of nucleotide/molecule of recA protein (Fig. 6). Oligonucleotides could substitute for single-stranded DNA, but larger concentrations were needed to promote the same extent of unwinding (Fig. 6). DISCUSSION There are four potential assays for recA protein based on its ATPase (Ogawa et QL, 1979;Roberts et ~l . , 1979), protease (Roberts et ~l . , 1978), unwinding (Cunningham et ~l . , 1979), and pairing activities (Weinstock et QL, 1979;Shibata et ~l . , 1979a). Of these, the ATPase assay is the most suitable for purification of recA protein. Particularly in a strain in which the recA gene has been cloned on a plasmid, the recA protein is the principal ATPase activity that requires single-stranded DNA as a cofactor. The ATPase activity is readily assayed (Scott et ~l . , 1977) and is directly proportional to the amount of recA protein (Fig. 3). The pairing activity is not readily detectable at early stages in the isolation and purification, and it is not a linear function of the concentration of recA protein (Shibata et ~l . , 1979a;Shibata et ~l . , 1981). The assay for pairing does not specifically detect the formation of D-loops but rather detects the conversion of wholly duplex DNA to a partially single-stranded form. Controls are required to demonstrate in any particular instance that the assay is measuring the formation of a joint molecule. Since our preparations of recA protein lack nuclease activity, we have not encountered high assay values due to partial degra-