Homologous pairing in genetic recombination: recA protein makes joint molecules of gapped circular DNA and closed circular DNA

Cell, Vol. 20. 223-235,

May 1980,

Copyright

0 1980 by MIT

Homologous Pairing in Genetic Recombination: recA Protein Makes Joint Molecules of Gapped Circular DNA and Closed Circular DNA Richard P. Cunningham, Chanchal DasGupta, Takehiko Shibata and Charles M. Radding Departments of Human Genetics and Molecular Biophysics and Biochemistry Yale University School of Medicine New Haven, Connecticut 06510

Summary The recA protein, which is essential for genetic recombination in E. coli, promotes the homologous pairing of double-stranded DNA and linear singlestranded DNA, thereby forming a three-stranded joint molecule called a D loop. Single-stranded DNA stimulates recA protein to unwind double-stranded DNA. By a presumably related mechanism, recA protein promoted the homologous pairing of two circular double-stranded molecules when one of them had a gap in one strand. The two molecules were joined at homologous sites by noncovalent bonds. The covalently closed molecule remained intact and was not topologically linked to the intact circular strand of the gapped substrate. Electron microscopy showed that molecules were usually linked at two or more nearby points. The junctions in most molecules were shorter than 300 nucleotides. Sometimes the region between two extreme points was separated into two arms, producing an ellipsoidal loop (called an eye loop). The junctions in these biparental joint molecules were frequently remote from the site of the gap. We infer that a free end of the interrupted strand crossed over to form a structure like a D loop which moved away from the gap by branch migration. Introduction According to current theories, breakage and reunion of DNA underlie general genetic recombination, and reunion occurs by means of a molecular splice, called a heteroduplex joint, in which one strand from each parent pairs with its complement. Which comes first, breakage or homologous pairing? If breakage occurs first, how are the breaks made at or near homologous sites? Breakage limited to homologous sites implies prior recognition. If, on the other hand, homologous pairing occurs first, what overcomes the energetic and topological barriers posed by the structure of duplex DNA? Recognition implies some disruption of the duplex structure. Investigators have postulated two solutions to this riddle. Because breakage by various means stimulates recombination, many have favored the view that breakage of one parental molecule exposes a single strand which pairs with its duplex partner and thereby provokes its cleavage (see Radding, 1978 for review). Other investigators have

suggested that molecules of duplex DNA may pair initially without the breakage of any phosphodiester bonds (Cross and Lieb, 1967; McGavin, 1971; Moore, 1974; Champoux, 1977; Kirkegaard and Wang, 1978; Wilson, 1979; Potter and Dressler, 1979a). A direct experimental approach to solving this riddle has been made possible by the purification of recA protein (Roberts, Roberts and Craig, 1978; Weinstock, McEntee and Lehman, 1979; Ogawa et al., 1979; Shibata et al., 1979a) which is essential for recombination in E. coli (Clark, 1973; Kobayashi and Ikeda, 1978). Weinstock et al. (1979) recently observed that recA protein promotes the pairing of complementary single strands, and Shibata et al. (1979a) found that recA protein promotes the homologous pairing of single strands with superhelical DNA, thereby forming a three-stranded joint molecule called a D loop. The same two laboratories reported the homologous pairing of single strands with nonsuperhelical duplex DNA (Cunningham et al., 1979; McEntee et al., 1979). Shibata et al. (197913) observed that single-stranded DNA promotes the binding of double-stranded DNA by recA protein, and Cunningham et al. (1979) found that in the resulting complexes all the double-stranded DNA is partially unwound. Moreover, the singlestranded DNA that stimulates the unwinding of duplex DNA need not be homologous nor have free ends. Because of these observations, we suggested that recA protein first brings together single-stranded and double-stranded DNA and then promotes a search for homology (Cunningham et al., 1979). The ability of single strands to stimulate unwinding, and the efficacy of circular single-stranded DNA, led us to the present experiments, which explore another reaction catalyzed by recA protein, that of closed circular doublestranded DNA and circular double-stranded DNA with a gap in one strand (gapped DNA). According to the experiments cited above, the single-stranded DNA in the gapped substrate should promote the formation of complexes with other double-stranded molecules and should unwind the latter. The outcome of such an interaction is particularly interesting because the gapped substrate resembles one of the natural substrates of recA protein (Rupp et al., 1971). West et al. (1980) have reported related observations on interactions of recA protein with gapped DNA and the formation of complexes of gapped DNA with intact DNA. Results Joint Molecules of Closed Circular and Gapped Circular Double-Stranded DNA To study the formation of D loops in vitro, we have used nitrocellulose filters under conditions that cause efficient retention of single-stranded DNA without retention of double-stranded DNA (Beattie, Wiegand

Cell 224

and Radding, 1977; Cunningham et al., 1979; Shibata et al., 1979a). When the formation of D loops was catalyzed by recA protein, we added salt and heated the reaction mixture prior to filtration to eliminate trapping of DNA due to formation of complexes with protein (see Experimental Procedures). We use the following designations for the different forms of double-stranded DNA from phages @Xl 74 and fd: form I, superhelical DNA; form II, nicked circular DNA; form III, linear DNA. In the absence of adventitious degradation, the retention of form I or form II DNA under these conditions reflects their attachment to singlestranded DNA or their conversion to a partially singlestranded form such as a D loop. Additional controls show that the retention of form I or form II DNA depends upon the presence of homologous singlestranded or partially single-stranded DNA, as well as that of recA protein (see below). We call this assay the D loop assay. Most of the observations reported here relate to the interactions that recA protein promotes between closed circular duplex DNA and circular duplex DNA that has a single gap in one strand. The latter we call gapped circular DNA or gapped DNA. When we incubated gapped circular 3”P-DNA with either form I or form II 3H-DNA and recA protein plus ATP and Mg++, a large fraction of the 3H-DNA was retained by nitrocellulose under the conditions of the D loop assay (Table 1, experiments 1 and 2). The gapped DNA required in this reaction must be homologous. Form I DNA of @Xl 74 did not make complexes with gapped fd DNA. As a positive control, we showed that recA protein promoted the formation of D loops from form I +X174 DNA and homologous singlestranded fragments, as expected (Table 1, experiment 3). As in the other pairing reactions promoted by recA protein (Shibata et al., 1979a; Weinstock et al., 1979) this reaction required ATP and Mg++ (Table : 1, experiment 1). When form II and form I DNA were both present in the same reaction, gapped DNA reacted preferentially with form I DNA. We made this observation in several ways. We examined the reaction mixture by agarose gel electrophoresis and extracted the labeled DNA from slices of the gel. In a reaction mixture which contained twice as much form II as form I DNA, most or all of the form I DNA was used up at the end of the reaction, whereas only a third of the form II DNA was used. When we examined the product by isopycnic centrifugation, the amount of product corresponded closely to the measured decrease in the amount of form I DNA (see Figures 2 and 4 below). Important information about the nature of the association of form I 3H-DNA and gapped 32P-DNA came from examining the integrity of the 3H-DNA by the assay of Kuhnlein, Penhoet and Linn (1976) for nicked circular DNA (the nicking assay). In this assay, circular DNA is treated briefly with alkali, neutralized, and

Table

1. Complexes

Detected

by Assay

for D Loops

DNA 32P 1.

a)

3H fd gapped

b)

2.

3.

4.

Omissions Additions

“

c)

“

d)

“

fd form ,/

I

‘/

fd gapped

b)

none

fd form

0.9

-Mg++

0.9

II

21 .o 2.0

a)

fd gapped

+X form I

b)

+X ss fragments

+X form

a)

fd form

fd form I /,

b)

“

0.8

-ATP

.c

II

96 ‘H-DNA Retained by Nitrocellulose 33.0

- recA protein

<‘

a)

or

2.3

I

40.0

2.9 ++X ss ments

frag-

1.6

Reaction mixtures of 20.5 ~1 contained: 31 mM Tris-HCI (pH 7.5), 25 mM Mg&, 1.8 mM dithiothreitol, 88 pg bovine serum albumin per ml, 1.2 mM ATP, 8.8 pM form I 3H-DNA, 8.8 gM fd gapped DNA or 11.7 PM fragments of single-stranded phage DNA and 2.4 pM recA protein. After incubation at 37’C for 30 min, the reactions were terminated by the addition of 300 pl of cold 25 mM EDTA (pH 7). 50 gl of the reaction mixture were spotted directly onto nitrocellulose filters to measure total counts. We added 200 ~1 to 3 ml of 1.5 M NaCI, 0.15 M Na citrate (pH 7). and heated at 50°C for 4 min before filtering through nitrocellulose filters to detect the formation of complexes of form I ‘H-DNA with single-stranded or partially single-stranded DNA. 4X ss fragments means fragments of single-stranded DNA of phage +X174 (see Experimental Procedures).

filtered through nitrocellulose. When a circular molecule with one or more interruptions is incubated in alkali, the strands separate and are retained by nitrocellulose. Since the strands of a closed circular molecule do not separate in alkali, these molecules rapidly renature on neutralization and pass through the filter. When gapped DNA and closed circular 3H-DNA formed complexes, there was no increase in the retention of’3H above that due to form II 3H-DNA initially present in’ the preparation (Table 2). Since most of the 3H-DNA in complexes came from form I DNA (see above), it follows that recA protein paired gapped DNA and form I DNA without nicking the form I DNA, without linking form I DNA covalently to either strand of the gapped DNA, and without linking form I DNA topologically to the intact circular strand of the gapped substrate. Any of these events would have increased the amount of 3H-DNA retained by nitrocellulose after denaturation and renaturation. In experiment 1, for example (Table 2) in which about 70% of the 3HDNA was covalently closed and 33% of the total 3HDNA formed complexes with gapped DNA (35% minus the blank), more than 70% of the 3H-DNA in complexes should have come from form I DNA; thus more than 23% (70% of 33%) of the total 3H-DNA should have been closed circular DNA present in complexes with gapped DNA. Had this form I DNA been nicked

Homologous 225

Table

Pairing

2. Integrity

in Genetic

of Form

Recombination

30

I DNA in the Complexes 3H-DNA

Gapped Form II 32PDNA

recA

Protein

In Complexes (D Loop Assay)

Nicked

-

+

2.4

32

b)

+

+

35.0

29

2a)

-

1 .o

27

2.3

30

la)

b)

-

+

C)

+

+

18.0

29

3a)

+

-

0.8

40

b)

-

+

2.2

38

c)

+

+

25.0

40

Reaction mixtures were as described in the legend to Table 1. At the end of the incubation at 37°C for 30 min in experiments 1 and 2, we added sodium dodecylsulfate, EDTA and proteinase K to achieve final concentrations of 0.4%, 20 mM and 200 as/ml. respectively, in a total volume of 30 ~1. After an additional incubation at 37’C for 30 min, 25 yl of the mixture were added to 0.3 ml cold 25 mM EDTA. We assayed total counts by putting 50 ~1 directly on nitrocellulose filters. We added 100 ~1 to 5 ml of cold 1.5 M NaCI, 0.15 M Na citrate (pH 7) and filtered it through nitrocellulose to measure the formation of complexes. To measure nicked DNA by the assay of Kuhnlein et al. (1976). we added 100 ~1 to an equal volume of 0.3 M potassium phosphate (pH 12.3), incubated for 2 min at 23”C, neutralized the mixture by adding 75 pl 1 M potassium phosphate (pH 4), added 5 ml of cold 1.5 M NaCI, 0.15 M Na citrate (pH 7), and filtered the final mixture through nitrocellulose. In experiment 3, after the initial incubation we added the reaction mixture directly to 0.3 ml of 25 mM EDTA and proceeded as described above.

or linked either covalently or topologically to the gapped DNA, we should have detected a significant increase in the amount of 3H-DNA retained by filters in the nicking assay. This conclusion is strengthened by similar observations on purified complexes where we could directly measure the amount of closed circular DNA in the complex (see below). The lack of topological linkage is consistent with earlier experiments which revealed no topoisomerase activity of recA protein under similar conditions (Cunningham et al., 1979). The complexes that were trapped by nitrocellulose were not dissociated by incubation with 0.46% sodium dodecylsulfate and 200 pg proteinase K per ml for 30 min at 37’ or 50°C. Figure 1 shows the stability of complexes heated in 0.15 M NaCI, 0.015 M Na citrate at pH 7.0. Complexes were largely stable at temperatures below the melting transition of the 3H-DNA from which they were made. All the foregoing observations, namely the requirement for homology, the lack of covalent or topological joining, the resistance to detergent or heat, and the dissociation by alkali, support the conclusion that the complexes are joint molecules in which gapped DNA and closed circular DNA are held together by heteroduplex joints whose stability is not very different from that of duplex DNA.

0’

I 40

1 60

TEMPERATURE

I 80

100

‘C

Figure 1. Thermal Stability of Complexes Formed by Gapped Duplex DNA and Homologous Circular Duplex DNA

Circular

In 61.5 ~1, the reaction mixture contained 31 mM Tris-HCI (pH 7.5), 25 mM MgClz. 1.8 mM dithiothreitol. 88 (Lg bovine serum albumin per ml, 1.2 mM ATP, 8.8 pM fd duplex ‘H-DNA (40% form I and 60% form II) and 8.8 PM gapped circular duplex DNA. Samples were incubated at 37’C for 30 min. The reaction was stopped by the addition of EDTA at 25 mM, and protein was removed from the DNA by incubation with 0.2% sodium dodecylsulfate and 200 pg of proteinase K per ml for 15 min at 37’C. The mixture was diluted with 10 mM Tris-HCI (pH 7.51, 1 mM EDTA to a final volume of 0.6 ml and stored on ice. Aliquots of 100 pl were removed and added to 400 ~1 of 0.19 M NaCI, 0.019 M Na citrate already incubated for 2 min at the desired temperature. Samples were incubated for 5 min more at the indicated temperatures, diluted in 7.5 ml of cold 1.5 M NaCI, 0.15 M Na citrate and filtered through nitrocellulose filters, (x) fd duplex DNA alone (40% form I, 60% form II); (0) the same preparation of fd duplex DNA incubated with gapped fd DNA in the reaction mixture described above.

Purification and Characterization of Dimeric Joint Molecules In the presence of an intercalating agent, a dimer consisting of a closed circular monomer and an open circular monomer should have a buoyant density between those of the two kinds of monomer. Accordingly, we centrifuged the product of the reaction in isopycnic gradients of CsCl and either propidium diiodide or ethidium bromide. In the absence of recA protein (Figure 2A), 61% of 3H-DNA was found in a peak corresponding to closed circular DNA, and 36% was found in a peak corresponding to open circular or linear DNA. The specific activity of the gapped circular 3’P-DNA was very low, but all the 32P was found in the appropriate peak of lighter density. When recA protein was present in the reaction, 18% of the 3H was found in complexes as measured by the D loop assay, and 18% of the 3H was found at densities between those of closed and open DNA (Figure 28). The relative amount of 3H in the position of form I DNA decreased 17%, which suggests that form I DNA contributed most of the 3H found at intermediate densities. In a similar experiment, we pooled fractions from

Cdl 226

r

A

61

%

3 %

36

%

---

m 4 x m

0

“/L.

. ,%.-x-r/ 10

20

I

Fraction No. Figure DNA

Fraction Figure 2. Isolation of Joint Circular DNA by lsopycnic Diiodide

No.

Molecules of Gapped DNA and Closed Centrifugation in CsCl and Propidium

Reaction mixtures of 20.5 pl contained 31 mM Tris-HCI (pH 7.5). 25 mM MgC12, 1.6 mM dithiothreitol, 88 fig bovine serum albumin per ml, 1.2 mM ATP. 8.8 pM fd form I ‘H-DNA and 8.8 pM fd gapped 32PDNA. The sample shown in (B) contained 2.4 pM recA protein; A had none. After 30 min incubation at 37°C. sodium dodecylsulfate, EDTA and proteinase K (EM Biochemical.9 were added to final concentrations of 0.4%, 20 mM and 200 pg/ml, respectively, in a total volume of 30 pl. After an additional 30 min at 37”C, the reaction mixtures were added to 3 ml of a solution containing 1.56 g CsCl per cc, 525 pg propidium diiodide per ml. 10 mM Tris-HCI (pH 8) and 1 mM EDTA. The solutions were centrifuged at 20°C for 38.5 hr at 40,000 rpm in an SW50.1 Spinco rotor. Fractions of the resulting gradients were collected from the bottom of the centrifuge tubes and counted in a scintillation fluid containing Triton X-100. The total recoveries of 3H-DNA in (A) and (B) differed by less than 10%. In parallel reactions. in the presence of recA protein 20% of the 3H-DNA was retained by a &trocellulose filter in high salt (D loop assay: see Experimental Procedures); in the absence of recA protein, only 3% was retained. (0) ‘H-DNA; (x) =P-DNA.

different parts of a gradient; we measured complexes by the D loop assay, nicked molecules by the nicking assay, and dimers or other structures by electron microscopy (Figure 3 and Table 3). More than 80% of the material of intermediate density was retained by nitrocellulose filters in the D loop assay, whereas little or none of the material in the flanking peaks was retained. Pool a, which should contain almost exclusively form I DNA, had a background of 13.5% nicked molecules, which may have been produced in the assay itself. Pool d, as expected, contained mostly nicked or linear DNA. In pool b, we found that 84% of

3. Purification

of Joint Molecules

of Gapped

DNA and Form I

The reaction was that described in the legend to Figure 28. scaled up 10 fold. The products were centrifuged in 7.4 ml of a solution containing 1.56 g CsCl per cc, 175 pg ethidium bromide per ml, 10 mM Tris-HCI (pH 8) and 1 mM EDTA. After centrifugation at 15°C for 46 hr at 40,000 rpm in a 75ti Spinco rotor, the resulting gradient was collected from the bottom of the centrifuge tube. Aliquots were counted in a scintillation fluid containing Triton X-100. The fractions indicated were pooled, extracted 5 times with isoamyl alcohol, and dialyzed against 10 mM Tris-HCI (pH 7.5) and 1 mM EDTA. Aliquots of these pools were analyzed by the D loop assay, by the nicking assay and by electron microscopy. In the initial reaction mixture, 24% of the 3H-DNA was retained by a nitrocellulose filter in the D loop assay.

the 3H-DNA was in complexes, and, even neglecting any correction for the background indicated by pool a, we found that at least 71% of ‘H-DNA in pool b was intact covalently closed circular DNA. Electron microscopy of pool b showed that 92 out of 171 identifiable molecules were figure 8 dimers consisting of two circular monomers united by a small joint (Table 3; see also below). The rest of the complexes in pool b may be represented by the aggregates which contain an unknown amount of DNA. We concluded that a fraction of the joint molecules trapped by nitrocellulose filters consisted of dimers that were enriched by banding in CsCl plus ethidium bromide or propidium diiodide. The banding of dimers at an intermediate density further confirmed that they were biparental, consisting of one closed circular and one open circular molecule. Pool c, however, which contained about two thirds of the material of intermediate density, consisted of larger aggregates than biparental dimers. The lighter density of material in pool c suggested that these aggregates contained relatively more gapped DNA

Homologous 227

Pairing

Table 3. lsopycnic Bromide

in Genetic

Banding

Recombination

of Joint

Molecules

in CsCl and Ethidium

Pool a

b

c

d

2.2

84

92

11

13.5

29

50

80

form I

28

11

form II

39

12

form Ill

8

1

92

IO

Complexes Nicked

(%)

molecules

Electron

microscopy

1%) (number

of

molecules):

dimers

(2 joined circles)

dimers

(1 linear,

aggregates

1 circular)

4 16

11

Analysis of pools indicated in Figure 3. Complexes were measured by the D loop assay and nicked molecules by the nicking assay (see Experimental Procedures).

than closed circular DNA. The aggregates were seen by electron microscopy, where a simple count of molecules underestimated the mass of DNA in aggregates (Table 31, and by gel electrophoresis, in which case the aggregates remained at the top of the gel (data not shown). These aggregates contained the bulk of material that was trapped by nitrocellulose filters in the D loop assay. The formation of dimers and aggregates had the same requirements. Observations made by three methods correlated perfectly: the formation of complexes measured by the D loop assay (Table 1). the formation of material that banded at an intermediate density in CsCl plus intercalating dyes (Figures 2 and 4, Table 3), and the formation of dimers estimated by electron microscopy (see below and Table 4). The fraction of 3H-DNA retained by filters corresponded to the fraction of 3H-DNA of intermediate density. The formation of stable joint molecules measured by all three methods required recA protein and a gap in one of the parental molecules (Tables 1 and 4, Figures 2 and 4). Homologous DNA was required for the formation of complexes measured by the D loop assay and by isopycnic centrifugation. Complexes observed by electron microscopy had crossovers at homologous sites (see below). Electron Microscopic Observations of Joints Our first examination of complexes by electron microscopy revealed a problem related to the action of recA protein, namely the formation of large aggregates, and two problems of a more general nature, namely the difficulty of distinguishing accidental overlaps from joints, and the difficulty of interpreting micrographs of constrained and twisted structures. For purposes of microscopy, we either ignored the aggregates or reduced their number of centrifugation in

Figure 4. Requirements for the Formation of Joint Molecules Detected by Their Intermediate Density in CsCl and Propidium Bromide Reaction mixtures contained 31 mM Tris-HCI (pH 7.5), 25 mM Mg&, 1.8 mM dithiothreitol, 1.2 mM ATP, 88 pg bovine serum albumin per ml, 2.4 PM recA protein and DNA as indicated below: (A) Positive and negative controls, 8.8 pM fd form I 3H-DNA and 4.4 pM fd gapped “P-DNA. (X) with recA protein and (0) without recA protein: (8) a heterologous combination, 8.8 pM @Xl 74 form I 3H-DNA and 4.4 PM fd gapped 32P-DNA; (C) Nicked DNA in place of gapped DNA, 8.8 fiM fd form I 3H-DNA and 8.8 pM fd form II 32P-DNA; (D) As in (C). plus heterologous single-stranded DNA and 11.7 pM fragments of unlabeled +X1 74 phage DNA. Samples were centrifuged in CsCl and propidium diiodide as described in the legend to Figure 2. The specific activity of ‘*P-DNA was very low; only counts due to ‘H are shown.

Table 4. The Role of Single-Stranded Molecules

DNA in the Formation

of Joint

Joint Molecules

Donor

recA Protein

DNA

By D Loop Assay (%)

By Electron Microscopy

Form

II

-

0.8

i/493

Form

II

+

2.0

o/339

Form II + heterologous fragments

+

1.6

o/405

Gapped

form II

-

1.8

2/202

Gapped

form

+

32.5

35/221

II

Reaction mixtures were as described in the legend to Table 1. By electron microscopy, we scored as joint molecules Figure 8 cleaved dimers like those illustrated in Figures 6A-6D, whether or not we could discern the structure of the junction. The denominators in the scores denote total identifiable duplex molecules, excluding any large aggregates or tangled structures.

CsCI-ethidium bromide (Table 3, pool b). We reduced the other problems by cutting the complexes with the restriction endonuclease Hpa I, which cuts doublestranded fd DNA only once (Beck et al., 1978). We could then clearly distinguish dimeric molecules that were joined at homologous sites. We collected micrographs of cleaved dimers with two pairs of apparently equal arms. For each pair of arms we plotted the length of the shorter arm versus the length of the longer one (Figure 5; see also Potter and Dressler,

Cell 228

0

0

.5 FRACTIONAL

Figure 5. Location Molecules

of Junctions

1.0 LENGTH

at Homologous

Sites in Dimeric

Joint

Dimers in unpurified samples of joint molecules were cleaved by restriction endonuclease Hpa I, as described in Figure 6. For each pair of arms of similar length, as in Figures 6A-6D. the distances from the ends to the junction were measured, and the length of the longer member was plotted on the ordinate versus the length of the other member on the abscissa. Similar results were obtained with samples of purified dimers. Symbols represent the types of junctions: (x) short eye loops (Figure 6A); (0) long eye-loops (Figure 66); (A) fused junctions (Figure 6D); and (0) point-junctions (Figure 60.

1979a). This plot reveals a population of molecules, not likely to contain many accidental overlaps, in which crossovers are evenly distributed relative to the ends defined by the Hpa I cut. At the joint in either intact or cut dimers, we distinguished three kinds of structure: the eye loop, an eyeshaped loop with two arms coming out of each end of the loop (Figures 6A, 6B and 6F); the fused junction, a linear structure with two arms coming out of each end (Figures 6D and 6E); and a point junction with no apparent fine structure (Figure 6C). In the dimers illustrated in Figures 6A, 6C, 6D and 6E, a singlestranded region was detectable in one of the two molecules, as indicated by the arrows, and was remote from the junction. Figure 6B illustrates a large eye loop in which the single-stranded region was adjacent to one junction. We detected single-stranded DNA in 44 molecules out of a sample of 57 dimers. In

Figure

6. The Structure

31 instances the single-stranded gap was remote from the junction. Of these, eighteen molecules were intact, which means that the remote location of the junction is not a selective effect that occurs only in molecules cut by endonuclease Hpa I. Figure 6F illustrates a dimer in which one molecule appeared relaxed and the other appeared supertwisted. Half of a small sample of purified dimers contained one supertwisted molecule (Table 5). The relative frequencies of these types of joint are shown in Table 5. All three types of joint were observed in purified or unpurified preparations and in molecules that either had or had not been cut by Hpa I. In a pooled sample of 74 molecules (Table 51, the fused junction was about twice as common as either eye loops or point junctions. In a preparation of D loops in cut molecules, we observed junctions of single and double-stranded DNA that have the appearance of fused junctions (Figure 7) and point junctions (not shown). These observations suggest that three-stranded joints can appear as fused junctions under some conditions. In a pooled sample of 51 joint molecules made from gapped DNA and duplex circular DNA, 40 of which had been cut by endonuclease Hpa I, we observed junctions varying in length from 50-l 600 nucleotides. Among the 51 joint molecules, 39 had junctions that were shorter than 300 nucleotides. The nonrandom distribution of the sizes of eye loops suggests that they were not caused exclusively by an adventitious overlap of molecules already joined at one site: out of a sample of 28 eye loops, 18 were between 123 and 365 nucleotides long. The Role of Single-Stranded DNA Since single strands or oligonucleotides stimulate recA protein to unwind duplex DNA, we wondered whether they might act purely as an effector to promote the homologous pairing of duplex molecules that lack any single-stranded regions. We examined this possibility by adding fragments of single-stranded +X174 DNA to form I and form II fd DNA plus recA protein. By using heterologous single-stranded DNA, we were able to study the possible stimulatory role of single strands without forming D loops. Neither by the D loop assay (Table 1, experiment 41, nor by electron microscopy (Table 41, nor by isopycnic centrifugation in CsCl plus ethidium bromide (Figure 4D) did we

of Joint Molecules

Joint molecules of gapped fd DNA and closed circular fd DNA, prepared as described in the legend to Table 1, were examined either directly or after purification by centrifugation in CsCl plus ethidium bromide as described in the legend to Figure 3. In the instances illustrated, each monomer of fd DNA was cleaved once by the restriction endonuclease Hpa I. The reaction mixture of 50 81 contained 10 mM Tris-HCI (pH 7.5). 10 mM MgCb, 20 mM KCI, 1 mM dithiothreitol, 0.13 PM DNA and 8 units of endonuclease Hpa I per ml. After incubation at 37°C for 2 min, EDTA was added at 25 mM and the DNA was spread immediately for electron microscopy (see Experimental procedures). (A). (B) and (C) are examples from unpurified samples: (D), (E) and (F) are from purified samples. (A) and (F) Small eye loops; (B) a large eye loop: (C) a point junction; (D) and (E) fused junctions. The pairs of arrows indicate the two ends of regions that we judge to be single-stranded on the basis of their thinner and less regular appearance.

Homologous 229

Pairing

in Genetic

Recombination

Cell 230

Table

5. Types

of Joints:

Relative

Eye

Fused Junctions

Point Junctions

Total

5

21

12

38

6

5

8

19

2

lo

15

36

23

74

Loops Unpurified

cut dimers

Purified

dimers

Purified

cut dimers

I:

Frequencies

Dimers were purified as indicated in Figure 3 and cut by restriction endonuclease Hpa I as described in the legend to Figure 6. In half the purified dimers. one molecule appeared relaxed and the other was supertwisted. as illustrated in Figure 6F.

detect the formation of any joint molecules. These experiments indicate that single-stranded DNA must be a part of one of the two homologous molecules that pair under the influence of recA protein. Previous observations on the formation of D loops by recA protein showed that the amount of recA protein required was proportional to the amount of single-stranded DNA and was unaffected by similar variations in the concentration of double-stranded DNA (Shibata et al., 1979b). More recently, we have found (Shibata et al., 1980) that the required amount of recA protein can be reduced by adding the E. coli helix destabilizing protein (recently named singlestrand binding protein; Meyer, Glassberg and Kornberg, 1979). These observations suggest that stoichiometric amounts of recA protein or single-strand binding protein are required to unfold single strands. In two experiments on the formation of joint molecules from gapped DNA and circular duplex DNA ,however, we found that the amount of recA protein required far exceeded the amount of single-stranded DNA present in the gaps. We obtained the largest yield of joint molecules, as measured by the D loop assay, when we used one molecule of recA protein per four nucleotide residues of total DNA in an experiment in which single-stranded DNA was only 4% of the total. Discussion In the presence of ATP and a divalent cation, recA protein catalyzed the homologous pairing of gapped circular DNA with closed circular duplex DNA about as well as it catalyzed the pairing of single-stranded fragments with duplex DNA (Table 1). On the other hand, by three independent methods, we did not detect any homologous pairing of nicked circular DNA with closed circular DNA, even in the presence of heterologous single strands, which cannot form D loops but which can stimulate recA protein to unwind duplex DNA (Tables 1 and 4, Figure 4) (Cunningham et al., 1979). Two of our methods, the D loop assay and isopycnic banding in CsCl and ethidium bromide, might have failed to detect the pairing of form II molecules with each other or form I molecules with

each other, but the observations by electron microscopy would have revealed any such pairings if they were stable. Since, in addition, the product of the reaction studied here contains both gapped DNA and intact circular DNA (Figure 3, Table 31, including many biparental dimers (Table 3, Figure 6), we infer that purified recA protein promotes homologous pairing if just one of two parental molecules is partially singlestranded. The biparental product made by recA protein from gapped DNA and closed circular DNA is a joint molecule in which the form I DNA is not linked by topological or covalent bonds to either strand of the gapped circular DNA (Figure 3, Tables 2 and 3). Figure 8 presents our ideas on the formation and structure of these joint molecules. Since the binding of single strands to recA protein stimulates the binding and partial unwinding of duplex DNA (Cunningham et al., 1979; Shibata et al., 1979b3), we presume that the initial interaction of gapped DNA with intact duplex DNA occurs by the same mechanisms, producing at the site of the gap a ternary complex such as we have postulated before (Shibata et al., 1979b) (Figure 8a). Only rarely, however, could homologous pairing result directly from the first such encounter. Either the two juxtaposed DNA molecules must move relative to each other, or the complex must dissociate and reform many times, or both, until a homologous match occurs. Since gapped DNA and closed circular DNA in the joint molecules are not linked topologically, the joint in structure II (Figure 8) cannot be the normal Watson-Crick structure consisting of strands interwound in a right-handed helix. Our experiments suggest, however, that the joints have the stability of Watson-Crick DNA (Figure 11, rather than the lesser stability of a “side-by-side” structure (Stettler et al., 1979). Moreover, in a sample of 44 dimeric joint molecules in which we could clearly distinguish a single-stranded gap in one member of each pair, the homologous junction was far removed from the site of the gap in 31 cases (see Figure 6 for examples). From these observations we infer that a free end of the gapped strand must cross over to form a structure like a D loop (Figure 8b) which can move by branch migration (Figure 8c; see also Radding et al., 1977). The idea that extensive branch migration may accompany the action of recA protein is supported by two other observations: recA protein can make very large D loops when the recipient duplex molecule is nicked circular DNA, and recA protein can make joint molecules from single-stranded circular DNA and linear duplex DNA in which the heteroduplex region may be several thousand bases long, or more (C. DasGupta and T. Shibata, unpublished observations). As to the postulated crossing over or crossing back of the interrupted strand (Figures 8b and 8~1, we have little evidence beyond that given by the observed structures and the absence of any topoisomerase activity

Homologous 231

Figure

Pairing

in Genetic

7. The Structure

Recombination

of D Loops

in Molecules

of Double-Stranded

fd DNA Cut Once by Restriction

Endonuclease

Hpa I

(A) and (B) Are from samples of D loops made by heating as described by Beattie et al. (1977): (C) and (D) are from D loops made by recA protein as described by Shibata et al. (1979a). Molecules were cut by endonuclease Hpa I as described in the legend to Figure 6.

of recA protein (Cunningham et al., 1979). Efforts to demonstrate the displacement of single-stranded DNA from a free end at the gap in the absence of a recipient duplex (Yarranton and Gefter, 1979; Duguet, Yarranand Hoffton and Gefter, 1979; Kuhn, Abdel-Monem mann-Berling, 1979a; Kuhn et al., 1979b) have so far been unsuccessful (our unpublished observations). If recA protein promotes branch migration as a natural consequence of making heteroduplex joints, however, it is not necessary to postulate any additional functions of the enzyme to explain steps b and c of Figure 8. In any event, these experiments suggest that recA protein can promote not only the receiving end of a strand transfer, viz the uptake of a single strand by duplex DNA (McEntee et al., 1979; Shibata et al., 1979a), but also the donor end, the displacement either directly or indirectly of a strand from a gapped donor molecule. The above model can account for the microscopic

structure that we have called an eye loop, and for point junctions, which could be very small eye loops. Fused junctions might be the same structures, representing instances in which the techniques of shadowing and electron microscopy failed to reveal the fine structure (see Figure 7). Alternatively, fused junctions may represent more complicated and more interesting structures, in which, for example, repetition of steps b and c (Figure 8) led to multiple crossovers. In Figure 8 we have indicated, by V-shaped marks, apparent heteroduplex regions that cannot be normal Watson-Crick structures according to this model. (If the D loop migrates to a nick in one of the two strands involved, or to the end of a linear molecule, the postulated unusual heteroduplex region could convert to a normal Watson-Crick structure.) These heteroduplex regions might nonetheless involve base pairing (Stettler et al., 1979) and could be involved in the search for homology, which presumably must be com-

Cell 232

(I)

1

a

(II) *-&JJ&

(III 1n

1

b

C +

(IV)fl Figure 8. Model for the Formation of Joint Circular DNA and Closed Circular DNA

Molecules

by Gapped

(a) The binding of recA protein to the single-stranded DNA in the gap stimulates the binding and unwinding of duplex DNA. The small ellipses in II. representing recA protein, are omitted elsewhere for the sake of simplicity. Before step b can occur, the molecules in II must be put in homologous alignment, or nearly so. either by reiterative dissociation and formation of structure II or by procession of one DNA molecule relative to the other. (b) Uptake of a free end of the interrupted strand produces a structure like a D loop. The displaced loop, which is complementary to the strand indicated, cannot form a normal Watson-Crick heteroduplex region for topological reasons, but nonetheless may form a paired structure denoted by the V-shaped symbols (Stettler et al., 1979). The cross-hatching in the opposite molecule represents a normal heteroduplex region. tc) Branch migration, possibly favored by recA protein, moves the joint away from the gap. If both boundaries of the D loop move as indicated, the end of the strand displaced from the gapped molecule will cross back to its original location.

pleted (Figure 8, II) before a junction can migrate away from the gap (Figure 8, III and IV). From experiments on several kinds of substrates, we have deduced a rule that may govern homologous pairing by recA protein in vitro: recA protein can make a stable joint molecule if one of the participants is single-stranded or partially single-stranded, and if either of the participants has a free end. The experiments reported here fit that rule; in this case both the single-stranded region and the free end belong to the same participant, namely the gapped molecule. In

another striking instance (C. DasGupta and T. Shibata, unpublished observations), recA protein pairs circular single-stranded DNA with a linear duplex molecule, in which case single-strandedness and free ends belong to different molecules. We believe that the requirement for single-stranded DNA reflects the special role that single-stranded DNA plays as an effector that stimulates recA protein to bind and unwind duplex DNA (Figure 8a) (Cunningham et al., 1979; Shibata et al., 1979b). We interpret the requirement for a free end as a topologic and thermodynamic one: in the absence of topoisomerase activity, a free end is necessary to form a heteroduplex region that has the normal Watson-Crick duplex structure. None of our observations, however, excludes the possibility that homologous recognition may first occur at sites that are removed from a free end, and that such an unstable junction may subsequently migrate to an end where a stable heteroduplex structure can be formed. This idea is closely related to the suggestion (see above and Figure 8, II) that homologous recognition may occur first in the gap between strands that are not topologically interwound. Benbow, Zuccarelli and Sinsheimer (1975) incubated circular double-stranded +X1 74 DNA in extracts of E. coli and recovered the DNA for examination by electron microscopy. Out of approximately 4000 molecules they found 26 “Figure 8” molecules when the extract came from a recA+ strain, but only one when the extract was from a recA- strain. Using a preparation of enzyme obtained by DEAE chromatography of an extract from Xenopus eggs, Benbow and Krauss (1977) observed the formation of a sizeable fraction of recombinant Figure 8 structures from partially homologous circular molecules. The junctions had the microscopic appearances that we have called point junctions and fused junctions. In some of the Figure 8 molecules, one of two monomers appeared supercoiled (see Figure 6F). Using crude extracts of E. coli as a source of enzyme, Potter and Dressler (1978, 1979a) observed the formation of Figure 8 molecules with junctions that are very similar in appearance to the fused junctions and eye loops described here. They partially purified an activity that joined DNA molecules, but which apparently was not recA protein, since it could be obtained from extracts of recA- cells (Potter and Dressler, 1979b). The joined molecules made by purified recA protein may well be related to those formed by extracts or fractions thereof (Benbow and Krauss, 1977; Potter and Dressler, 1978). The observations reported here suggest several new interpretations of the structures of junctions in Figure 8 molecules, specifically that one side of an eye loop may involve complementary strands that are not topologically linked, and that a fused junction may represent multiple crossovers produced by repetition of the process that yields eye loops (see Figure 8). The experiments described above show that recA protein will catalyze homologous pairing if just one of

Homologous 233

a

pair

Pairing

of

in Genetic

molecules

is

addition,

when

helical

DNA are

present,

gapped

molecule

pairs

the

These

both

phages

Benbow

involves

parental

Doniger

radiation

1978; of these

(Baker,

of

DNA

circular recA

protein

with vitro

the

and

super-

preferentially

superhelical

appear

to

DNA.

account

for

and Tessman,

DNA 1978).

requires

(Tess-

of the

Warner

In

phages

small

recA+

et al., 1974b), RF,

which

and preferentially largely form I DNA 1971). Ultraviolet ir-

is

and Tessman,

phage

after

single-stranded.

of recombination

(Zinder,

man, 1968;

rivation

in

aspects

Recombination

partially

nicked

observations

prominent

Recombination

prior

infection,

to

infection,

which

or thymine

produce

similar

depgapped

DNA molecules (Benbow, Zuccarelli and Sinsheimer, 1974a), stimulate recombination of S13 or $X1 74 (Tessman, 1968; Benbow et al., 1974a). The stimulation of recombination by irradiation of only one of the parental phages led Benbow et al. (1974a) to emphasize

the

causal

role

of the

interruption

or gap.

Zinder and co-workers showed that recombination of the filamentous phage fl involves very long heteroduplex regions (Enea and Zinder, 1976; Zinder, 1978). While the joints made by recA protein in the experiments described here are short (Figure 6), we have also observed the formation of very long heteroduplex regions by recA protein (C. DasGupta and T. Shibata, unpublished

observations;

see

above).

Observations on the role of single-stranded DNA in the action of recA protein (Cunningham et al., 1979; Shibata et al., 1979b; West et al., 1980; this work) also

provide

new

insight

into

sister-strand

exchange,

a mechanism of repair of ultraviolet damage for which the recA gene is essential (Rupp et al., 1971). Step a of Figure 8 suggests in part how a gap resulting from the blockage of DNA synthesis by thymine dimers may stimulate recA protein to promote the transfer of an undamaged strand from the other arm of newly replicated DNA. Experimental

Procedures

Phage and Bacteria @Xl 74am3 phage and E. coli C strain HF4704 (Benbow et al., 1971) were from our laboratory collection. Fd phage and E. coli strain K37 (Marvin and Hoffman-Berling, 1963; Lyons and Zinder, 1972) were obtained from W. Konigsberg. Media TPA medium contained 8.5 mM NaCI, 100 mM KCI, 12 mM NHJ&. 1 mM CaCl>. 2 mM MgS04, 100 mM Tris-HCI (pH 7.5), 0.2% glucose. 0.3% salt-free casein hydrolysate, 7.3 mM sodium pyruvate and 4 mM potassium phosphate (pH 7.5). This was supplemented with 5 pg thymine per ml for growth of HF4704 or 2 pg thymine per ml when 3H-thymidine was used to label phage DNA. When ‘*P-orthophosphate was added to label DNA, the concentration of potassium phosphate was reduced to 0.4 mM. M9 buffer contains 22 mM KH2P04, 42 mM NazHP04 and 19 mM NHICI. Preparation of +X174am3 Phage DNA E. coli strain HF4704 was grown in 1 I of TPA at 37’C to 4 X 10’ cells per ml. 5-10 +X174am3 phages per ceil and 10 mCi 32Porthophosphate (New England Nuclear, carrier free) were added. After 3 hr the cells were harvested by centrifugation and suspended

in 25 ml 0.1 M Na Borate (pH 9.3). The cells were disrupted by sonication, and cell debris was removed by centrifugation. Concentrated Tris-HCI (pH 7.5) and MgC12 were added to the supernatant to produce a final concentration of 100 mM. The supernatant solution was then treated with 10 pg/ml each of RNAase and DNAase for 15 min at 37°C. Debris was removed again by centrifugation at 12,000 x g for 30 min. The phage were purified by glass bead exclusion chromatography according to the method of Gschwender, Hailer and Hofschneider (1969), followed by centrifugation in CsCI. Solutions of CsCl were made in 50 mM Na borate (pH 9.3) at densities of 1.2. 1.3, 1.4 and 1 .5 g/cc. These were put into a centrifuge tube in successive layers consisting of 2 ml of 1.5 g/cc, 4 ml 1.4 g/cc, 4 ml 1.3 g/cc and 2 ml 1 .2 g/cc, from bottom to top. Approximately 26 ml of phage suspension was layered on the top and centrifuged for 3 hr at 25,000 rpm at 15°C in a Spinco SW27 rotor. The phages formed a band at the boundary between 1.3 g/cc and 1.4 g/cc. This band was collected from the side of the tube by means of a hypodermic syringe and dialyzed against 10 mM Tris-HCI (pH 7.5), 1 mM EDTA. An equal volume of phenol, equilibrated against the same buffer, was added. The mixture was mixed gently for a few minutes at room temperature and centrifuged at 12,000 X g for 10 min at 23°C. The aqueous layer was recovered and shaken with phenol twice more in the same fashion. Residual phenol was removed by adding an equal volume of ether to the aqueous phase and mixing. After centrifugation at 12,000 x g for 10 min at 23’C. the ether phase was removed. After two repetitions of the procedure, we evaporated the residual ether by carefully directing a stream of nitrogen at the recovered aqueous phase. Concentrated NaCl was added to a final concentration of 0.3 M. The DNA was precipitated by the addition of 2 vol of cold 95% ethanol and incubation at -20°C for at least 60 min. DNA was collected by centrifugation at 4°C for 30 min at 25.000 rpm in a Spinco SW27 rotor. It was dissolved in IO mM Tris-HCI (pH 7.51, 1 mM EDTA. Per liter of culture, this procedure yields 1 O-20 pmole of phage DNA with a specific activity of 2-4 x 1 O7 cpm/(lmole. Preparation of +X174am3 Replicative Form DNA E. coli HF4704 was grown and infected as described above. 3 min after infection, chloramphenicol was added to a final concentration of 50 pg/ml. At 15 min after infection, 5 mCi of either 3H-thymidine or 32P-orthophosphate were added. After 2 hr. the cells were harvested by centrifugation, washed once with cold M-9 buffer, and suspended in 12.5 ml of a cold solution of 50 mM Tris-HCI (pH 8). 10 mM EDTA, 10% sucrose. 0.6 ml of lysozyme (30 mg/ml) were added and the mixture was incubated on ice for 10 min. The cells were then lysed by the addition of 1.6 ml 10% Sarcosyl. with gentle mixing. After a IO min incubation on ice, the cell debris was removed by centrifugation at 35,000 rpm for 45 min in a Spinco 45ti rotor at 4°C. The resulting supernatant was extracted 3 times with redistilled phenol as described above. Nucleic acids were precipitated with ethanol in 0.3 M NaCl as described above and suspended in 2 ml 10 mM Tris-HCI (pH 7.5). 1 mM EDTA. A solution of 1 mg RNAase per ml was incubated at 75°C for 15 min, then 0.1 ml of this solution was incubated with the DNA for 30 min at 37’C. The DNA was layered onto a 36 ml 520% gradient of neutral sucrose and centrifuged at 27,000 rpm for 18 hr at 4’C. The gradient was collected from the bottom of the tube and the DNA was recovered from the appropriate fractions by precipitation with ethanol. The DNA was suspended in 10 mM Tris-HCI (pH 7.51, 1 mM EDTA and extensively dialyzed against the same buffer. More than 90% of the DNA obtained is covalently closed, as judged by gel electrophoresis and velocity sedimentation in alkaline sucrose. Per liter of culture. the yield is 2-4 pmole of DNA with a specific activity of 2-4 X 1 O7 cpm/pmole. Preparation of fd Replicative Form DNA The same procedure as described for $X1 74 replicative form DNA was used with the following modifications: E. coli K12 strain K37 was used as the host; the moi was 25; radioactive label was added 10 min after infection; and chloramphenicol was added at 15 mm after infection. When the radioactive label was 3H-thymidine, 400 mg/l of 2’-deoxyadenosine (10 mg/ml in TPA and 20% ethanol) were added simultaneously. Per liter of culture, the yield is approximately 1 pmole of DNA with a specific activity of 2-4 X 10’ cpmlpmole.

Cell 234

Preparation of fd Phage DNA E. coli K12 strain K37 was grown to 4 X 1 O8 cells per ml in 1 I TPA medium at 37’C. Phages, at an moi of 20-30, and 5 mCi 32Porthophosphate were added together. After 4-5 hr the cells were removed by centrifugation. The phages were concentrated with PEG 6000 and purified by CsCl gradient centrifugation as described by Yamamoto and Alberts (1970). The phages were then dialyzed against 0.15 M NaCI, $.015 M Na citrate at pH 8. DNA was extracted from purified phage at 48’C with redistilled phenol equilibrated against the same buffer (Salivar, Henry and Pratt, 1967). After extraction, the DNA was dialyzed extensively against 10 mM Tris-HCI (pH 7.5), 1 mM EDTA. The yield per liter of culture is 2-5 pmole of DNA with a specific activity of 2-4 X 10’ cpm/pmole.

microscopy, 95% of molecules in this preparation were circular, and the gaps were 1170 ? 625 bases, as estimated by measurement of 20 molecules. When a sample of this material was filtered through nitrocellulose in 1.5 M NaCI, 0.15 M Na citrate (pH 7). all the radioactivity was retained by the filter, consistent with the presence of single-stranded regions in all molecules. After denaturation and neutralization, the gapped DNA produced two major bands on electrophoresis through a 1.8% agarose gel, corresponding to a singlestranded circle and a linear single strand slightly shorter than full length. A minor band of full-length linear DNA was also observed. These findings support the conclusion that our substrate was a double-stranded circle with one intact circular strand and one strand with a single gap.

Enzymes RecA protein was purified as described by Shibata et al. (1979a). Exonuclease III (Weiss, 1976) and restriction endonuclease Hpa I (Beck et al., 1978) were purchased from New England BioLabs, proteinase K (Ebeling et al., 1974) from EM Laboratories Inc., and Sl nuclease (Wiegand et al., 1975) from Miles Biochemicals.

Electron Microscopy The DNA in 50 mM Tris-HCI (pH 8.51, 5 mM EDTA, 30% (v/v) formamide, 0.01% cytochrome C was spread on a hypophase of 10 mM Tris-HCI (pH 8.51, 1 mM EDTA. The sample was picked onto a 3.5% parlodion-coated grid, stained with uranyl acetate and rotaryshadowed with Pt/Pd (4:i). In some cases, the products of reactions were directly used for spreading. In others, joint molecules were purified by isopycnic centrifugation in CsCl and ethidium bromide after protein had been removed by treatment with sodium dodecylsulfate and proteinase K. Contour lengths of DNA molecules were measured with a Hewlett-Packard 9864 A Digitizer coupled to a 9825A calculator at a magnification of either 50,000 or 100,000.

Assays D loop Assay This assay measures the retention by nitrocellulose of doublestranded DNA by virtue of its attachment to single-stranded or partially single-stranded DNA. The details of this assay were as described by Beattie et al. (1977), except that we used the “Sartorius Membranfilter” type SM 11306, which has a pore size of 0.45 pm. In the present experiments, recA protein was removed by heating or by treatment with detergent and proteinase K as described in each experiment. Nicking Assay This assay measures the nicking of closed circular DNA as described by Kuhnlein et al. (1976). The conditions for filtration through nitrocellulose and the type of filter were the same as in the D loop assay described above. In our hands, this assay tended to overestimate the fraction of nicked circular or linear double-stranded DNA, as judged by examination of the same samples by agarose gel electrophoresis. Nonetheless the assay is rapid and convenient. It is also useful for examination of samples of DNA that for some reason, such as aggregation, are unsuitable for analysis by gel electrophoresis. Preparation of Circular Double-Stranded DNA with a Single Gap in One Strand (Gapped DNA) We made form II DNA containing a single nick per molecule by cleaving form I “P-DNA with Sl nuclease under the conditions described by Wiegand et al. (1975). By limiting cleavage of the form I DNA to approximately 50%, we eliminated significant formation of linear duplex molecules. Form II DNA was separated from form I DNA by isopycnic centrifugation in CsCl and ethidium bromide (see below). After removal of the ethidium bromide and CsCl by extraction with isoamyl alcohol and dialysis, we treated the form II DNA with E. coli exonuclease Ill (New England BioLabs) to produce gapped DNA. The reaction mixture contained 67 M Tris-HCI (pH 8). 5 mM MgClz, 1 mM dithiothreitol. 0.1 mM fd form II 32P-DNA and 10 units of exonuclease Ill per ml. We determined the amount of exonuclease Ill required by trials in which we measured the production of acid-soluble material and the conversion of form II DNA to a smaller molecule that migrates more rapidly on gel elkctrophoresis and that is more readily convertible to linear DNA by Sl nuclease. Under these conditions, exonuclease 111made 12.5% of the DNA acid-soluble, corresponding to a gap of approximately 1600 bases per molecule. After treatment with exonuclease III, the form II DNA was converted almost entirely to material that, as expected, migrated slightly more rapidly than form II DNA on electrophoresis through an agarose gel. Treatment of a sample of the gapped DNA with Sl nuclease converted all of it to material that migrated slightly more rapidly than form Ill linear DNA made from the original form I DNA. The preparation of gapped DNA was further purified by passage through a column of Sepharose 28 in 200 mM NaCI, 10 mM Tris-HCI (pH 7.5), 1 mM Na EDTA. The peak fractions were dialyzed against 10 mM Tris-HCI (pH 7.5), 1 mM EDTA. As judged by electron

Gel Electrophoresis To analyze double-stranded DNA substrates we used 1.2% agarose gels, and to analyze single-stranded or denatured double-stranded DNA substrates we used 1.8% agarose gels, 16 cm X 16 cm X 0.3 cm, and E buffer which contained 40 mM Tris-acetate, 5 mM sodium acetate and 1 mM EDTA (pH 8.0). Current was applied at 2 V/cm for 16-18 hr at room temperature, and the buffer was recirculated between reservoirs. Gels were stained for 2 hr in E buffer containing 0.5 pg ethidium bromide per ml. The gels were illuminated from below with a short-wave length ultraviolet lamp and were photographed with Polaroid type 55 film. Acknowledgments T. S. is a visiting fellow from The Institute of Physical and Chemical Research, Saitama, Japan. We gratefully acknowledge the expert technical assistance of Lynn Osber. and the aid and guidance of John Flory in our microscopic studies. This research was sponsored by grants from the American Cancer Society and the National Cancer Institute. This is the fifth paper of a series. Paper number 4 is Shibata et al. (1980). The costs of publication of this article was 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. Received

December

27, 1979

References Baker, R.. Doniger. J. and Tessman, progeny DNA in the two mechanisms Nature New Biol. 230, 23-25.

I. (1971). Roles of parental and of phage S13 recombination.

Beattie, K. L.. Wiegand, R. C. and Radding, C. M. (1977). Uptake of homologous single-stranded fragments by superhelical DNA. II. Characterization of the reaction. J. Mol. Biol. 176, 783-803. Beck, E.. Sommer. R., Auerswald. E. A., Kurz, C., Zink, B., Osterburg. G., Schaller, H.. Sugimoto, K., Sugisaki. H., Okamoto, T. and Takanami, M. (1978). Nucleotide sequence of bacteriophage fd DNA. Nucl. Acids Res. 5, 4495-4503. Benbow, R. M. and Krauss, M. R. (1977). Recombinant DNA formation in a cell-free system from Xenopus laevis eggs. Cell 72, 191204. Benbow.

R. M., Hutchinson.

C. A., Fabricant,

J. D. and Sinsheimer,

Homologous 235

Pairing

Ft. L. (1971). 558.

Genetic

in Genetic

Recombination

map of bacteriophage

@Xl 74. J. Viral.

7, 549-

Benbow, R. M.. Zuccarelli, A. J. and Sinsheimer. R. L. (1974a). A role for single-strand breaks in bacteriophage @Xl 74 genetic recombination. J. Mol. Biol. 88, 629-651. Benbow, R. M.. Zuccarelli, Recombinant DNA molecules Acad. Sci. USA 72, 235-239.

A. J. and Sinsheimer, R. L. (1975). of bacteriophage $X174. Proc. Nat.

Benbow R. M., Zuccarelli, A. J., Davis, G. C. and Sinsheimer, R. L. (1974b). Genetic recombination in bacteriophage @X174. J. Virol. 13, 898-907.

Biol. 43, 909-915. Potter, H. and Dressler, D. (1978). In vitro system from coli that catalyzes generalized genetic recombination. Acad. Sci. USA 75, 3698-3702.

Escherichia Proc. Nat.

Potter, H. and Dressler, D. (1979a). DNA recombination: in vivo and in vitro studies. Cold Spring Harbor Symp. Quant. Biol. 43, 969-985. Potter, H. and Dressier. D. (1979b). Biochemical detect formation of recombination intermediates Acad. Sci. USA 76, 1084-1088. Radding, mismatch

assay designed to in vitro. Proc. Nat.

C. M. (1978). Genetic recombination: strand repair. Ann. Rev. Biochem. 47, 847-880.

transfer

and

Champoux, J. J. (1977). Renaturation of complementary singlestranded DNA circles: complete rewinding facilitated by the DNA untwisting enzyme. Proc. Nat. Acad. Sci. USA 74, 5328-5332.

Radding. C. M., Beattie, K. L., Holloman. W. K. and Wiegand, R. C. (1977). Uptake of homologous single-stranded fragments by superhelical DNA. IV. Branch migration. J. Mol. Biol. 716, 825-839.

Clark, A. J. (1973). Recombination deficient other bacteria. Ann. Rev. Genet. 7, 67-86.

Roberts, J. W.. Roberts, C. W. and Craig, N. L. (1978). coli recA gene product inactivates phage h repressor. Acad Sci. USA 75, 4714-4718.

mutants

of E. coli and

Cross, R. A. and Lieb, M. (1967). Heat-inducible A phage. V. Induction of prophages with mutations in genes 0, P and R. Genetics 57, 549560. Cunningham, R. P., Shibata. T., DasGupta. C. and Radding, C. M. (1979). Homologous pairing in genetic recombination: single strands induce recA protein to unwind duplex DNA. Nature 287, 191-195, 282, 426.

Escherichia Proc. Nat.

Rupp, W. D., Wilde, C. E. Ill, Rena. D. L. and Howard-Flanders, (1971). Exchanges between DNA strands in ultraviolet-irradiated Escherichia coli. J. Mol. Biol. 67, 25-44.

P.

Salivar, W. 0.. Henry, T. J. and Pratt, D. (1967). Purification and properties of diploid particles of coliphage Ml 3. Virology 32, 41-51.

Duguet, M., Yarranton. G. and Gefter, M. (1979). The rep protein of Escherichia coli: interaction with DNA and other proteins. Cold Spring Harbor Symp. @ant. Biol. 43, 335-343.

Shibata, T., DasGupta, C.. Cunningham, R. P. and Radding, C. M. (1979a). Purified E. coli recA protein catalyzes homologous pairing of superhelical DNA and single-stranded fragments. Proc. Nat. Acad. Sci. USA 76, 1638-i 642.

Ebeling, W.. Hennrich. N., Klockow, M., Metz. H., Orth, H. D. and Lang, H. (1974). Proteinase K from tritirachium album limber. Eur. J. Biochem. 47, 91-97.

Shibata, T.. Cunningham, R. P., DasGupta. C. and Radding, C. M. (1979b). Homologous pairing in genetic recombination: complexes of recA protein and DNA. Proc. Nat. Acad. Sci. USA 76, 5100-5104.

Enea, V. and Zinder, N. D. (1976). tional intermediate in bacteriophage

Shibata, T.. DasGupta, C., Cunningham, R. P. and Radding, C. M. (1980). Homologous pairing in genetic recombination: formation of D-loops by combined action of recA protein and single-strand binding protein. Proc. Nat. Acad. Sci. USA, in press.

Heteroduplex DNA: a recombinafl. J. Mol. Biol. 707, 25-38.

Gschwender, H. H.. Hailer, W. and Hofschneider. P. H. (1969). Largescale preparations of viruses by steric chromatography on columns of controlled pore glass. Biochim. Biophys. Acta 790, 460-469. Kirkegaard, K. and Wang, J. C. (1978). Escherichia coli DNA topoisomerase I catalyzed linking of single-stranded rings of complementary base sequences. Nucl. Acids Res. 5, 381 l-3820. Kobayashi, I. and Ikeda, H. (1978). On the role of recA gene product in genetic recombination: an analysis by in vitro packaging of recombinant DNA molecules formed in the absence of protein synthesis. Mol. Gen. Genet. 766, 25-29. Kuhn, B., Abdel-Monem, helicases. Cold Spring

M. and Hoffmann-Berling, H. (1979a). Harbor Symp. &ant. Biol. 43, 63-67.

DNA

Kuhn, B., AbdeCMonem, M., Krell, H. and Hoffmann-Berling, H. (1979b). Evidence for two mechanisms for DNA unwinding catalyzed by DNA helicases. J. Biol. Chem. 254, 11343-l 1350.

Stettler. U. H., Weber, H., Keller, T. and Weissmann, C. (1979). Preparation and characterization of form V DNA, the duplex DNA resulting from association of complementary, circular single-stranded DNA. J. Mol. Biol. 131, 21-40. Tessman, I. (1968). Selective stimulation for genetic recombination of bacteriophage 482.

of one of the mechanisms S13. Science 767, 481-

Warner, R. C. and Tessman, I. (1978). Mechanism of genetic recombination. In The Single-Stranded DNA Phages, D. T. Denhardt, D. Dressler and D. S. Ray, eds. (New York: Cold Spring Harbor Laboratory), pp. 417-432. Weinstock. dependent coli. Proc.

G. M., McEntee, K. and Lehman, I. R. (1979). renaturation of DNA catalyzed by the recA protein Nat. Acad. Sci. USA 76, 126-I 30.

Kuhnlein, U., Penhoet. E. E. and Linn, S. (1976). An altered apurinic DNA endonuclease activity in group A and group D xeroderma pigmentosum fibroblasts. Proc. Nat. Acad. Sci. USA 73, 1169-i 173.

Weiss, B. (1976). Ill. J. Biol. Chem.

Lyons, L. B. and Zinder, filamentous bacteriophage

West, S. C., Cassuto, E.. Mursalim, J. and Howard-Flanders, (1980). Recognition of duplex DNA containing single-stranded gions by recA protein. Proc. Nat. Acad. Sci. USA, in press.

N. D. (1972). The genetic fl. Virology 49, 45-60.

map

of the

McEntee, K., Weinstock, G. M. and Lehman, I. R. (1979). Initiation of general recombination catalyzed in vitro by the recA protein of E. coli. Proc. Nat. Acad. Sci. USA 76, 2615-2619. McGavin, S. (1971). Models of specifically paired nucleic acid structures. J. Mol. Biol. 55, 293-298.

like (homologous)

Marvin, D. A. and Hoffmann-Berling, H. (1963). Physical and chemical properties of two new small bacteriophages. Nature 197, 517-518. Meyer, R. R., Glassberg, J. and Kornberg. A. (1979). An Escherichia coli mutant defective in single-strand binding protein is defective in DNA replication. Proc. Nat. Acad. Sci. USA 76, 1702-l 705. Moore, D. (I 974). Dynamic unwinding for genetic recombination. J. Theoret.

of DNA helices: a mechanism Biol. 43, 167-l 86.

Ogawa. T., Wabiko, H., Tsurimoto, T., Horii, T., Masukata, H. and Ogawa, H. (1979). Characteristics of purified recA protein and the regulation of its synthesis in viva. Cold Spring Harbor Symp. &ant.

Endonuclease II of Escherichia 257, 1896-1901.

ATPof E.

coli is exonuclease P. re-

Wiegand, R. C., Godson, G. N. and Radding, C. M. (1975). Specificity of the S, nuclease from Aspergillus oryzae. J. Biol. Chem. 250, 8848-8855. Wilson, J. H. (1979). Nick-free formation of reciprocal heteroduplexes: a simple solution to the topological problem. Proc. Nat. Acad. Sci. USA 76, 3641-3645. Yamamoto, K. R. and Albert% B. M. (1970). sedimentation in the presence of polyethylene cation of large-scale virus purification. Virology

Rapid bacteriophage glycol and its appli40, 734-744.

Yarranton. G. T. and Gefter. M. L. (1979). Enzyme catalyzed DNA unwinding: studies on Escherichia coli rep protein. Proc. Nat. Acad. Sci. USA 76, 1658-l 662. Zinder, N. D. (1978). Genetic recombination. In The Single-Stranded DNA Phages. D. T. Denhardt. D. Dressler and D. S. Ray, eds. (New York: Cold Spring Harbor Laboratory), pp. 403-415.