Proc. Natl Acad. Sci. USA Vol. 78, No. 8, pp. 4786-4790, August 1981 Biochemistry

Polarity of heteroduplex formation promoted by Escherichia coli recA protein* (genetic recombination/homologous pairing/strand transfer/joint molecules) ROGER KAHN, RICHARD P. CUNNINGHAM, CHANCHAL DASGUPrA, AND CHARLES M. RADDING The Departments of Human Genetics and Molecular Biophysics and Biochemistry, Yale University School of Medicine, New Haven, Connecticut 06510 Communicated by Aaron B. Lerner, April 24, 1981

ABSTRACT When recA protein pairs circular single strands tron microscopy, we observed that the product of this reaction with linear duplex DNA, the circular strand displaces its homolog is a joint molecule in which the circular single strand has dis- from only one end of the duplex molecule and rapidly creates het- placed its homolog, starting from one end of the duplex mole- eroduplexjoints that are thousands ofbase pairs long [DasGupta, cule (ref. 8 and this paper). In 15-30 min, some heteroduplex C., Shibata, T., Cunningham, R. P. & Radding, C. M. (1980) Cell joints become thousands ofnucleotides long (8). The rapid cre- 22, 437-446]. To examine this apparently polar reaction, we pre- ation ofthese long heteroduplex joints, and the absence ofmol- pared chimeric duplex fragments ofDNA that had M13 nucleotide ecules in which circular single strands invaded the duplex DNA sequences at one end and G4 sequences at the other. Circular sin- from both ends (unpublished data) led us to infer that recA pro- gle strands homologous to M13 DNA paired with a chimeric frag- ment when M13 sequences were located at the 3' end ofthe com- tein actively drives the formation ofheteroduplex DNA in one plementary strand but did not pair when the M13 sequences were direction. The experiments described in this paper were de- located at the 5' end. Likewise circular single-stranded G4 DNA signed to examine this apparent polarity. paired with chimeric fragments only when G4 sequences were lo- METHODS cated at the 3' end ofthe complementary strand. To confirm these . E. coli I was purified by John Wil- observations, we prepared fd DNA labeled only at the 5' or 3' end liams (11) according to the procedure ofLehman and Nussbaum of the plus strand, and we examined the susceptibility ofthese la- (12). E. coli exonuclease VII was a gift ofJohn Chase (13). recA beled ends to digestion by whenjoint molecules were protein (3, 14) and A exonuclease (15) were purified as de- formed. Eighty percent of the 5' label in joint molecules became sensitive to exonuclease VII. Displacement of that 5' end by recA scribed. Bacterial alkaline was a gift of Jan Chle- protein was concerted because it did not occur in the absence of bowsky. Polynucleotide kinase was purchased from Boehringer. single-stranded DNA or in the presence of heterologous single E. coli polymerase I (large fragment) and Sau 96 strands. By contrast, only a small fraction of the 3' label became I were purchased from New England BioLabs. Endonuclease sensitive to exonuclease VII or exonuclease I. These observations Hpa I was purchased from New England BioLabs and Bethesda show that recA protein forms heteroduplex joints in a concerted Research Laboratories (Rockville, MD). Endonuclease BamHI and polarized way. was purchased from Bethesda Research Laboratories. Preparation of DNA. Single-stranded and replicative form The product of the recA gene ofEscherichia coli is a DNA-de- DNA from phages fd, 4X174, and G4 were prepared as de- pendent ATPase that regulates a pathway of inducible repair scribed (7, 16). Superhelical DNA from M13 Goril was pre- by means of a specific protease activity (see ref. 1 for a review) pared according to the procedures used for fd DNA (7). and that also has a remarkably versatile ability to promote the Duplex DNA 32P-Labeled at the 5' End of the Plus Strand. homologous pairing of DNA molecules. With a few exceptions, We cut form I fd [3H]DNA at a single site by digesting it with recA protein can stably pair all manner of structural variants of Hpa I and labeled the 5' ends of this DNA by treating it with DNA molecules: complementary single strands with each other, bacterial followed by polynucleotide ki- single strands with duplex DNA, and partially single-stranded nase essentially as described by Maxam and Gilbert (17). To DNA with duplex DNA (2-9). The rule that governs stable pair- verify that the 32p label was present only at the ends, we cleaved ing appears to be that one molecule must be at least partially an aliquot ofthis DNA with BamHI which cuts Hpa I-linearized single-stranded and that afree end must exist somewhere in one fd DNA into three fragments: two end fragments and an internal of the two molecules (4, 8). We have experimentally distin- fragment 3425 base pairs long. The internal fragment, separated guished two steps in the formation of stable joint molecules by by gel electrophoresis, contained less than 1% of the total 32p recA protein: (i) synapsis, which appears to require single- recovered in the three fragments. stranded DNA but does not require a free end, and (ii) strand The linear DNA with 32p label at both ends was cleaved into transfer, which requires a homologous free end and produces two fragments by Sau 96 I: a large fragment (5725 base pairs) classical heteroduplex joints in which a strand from each pa- labeled at the 5' end ofthe plus strand and a small one (683 base rental molecule is helically interwound with its complement pairs) labeled at the 5' end ofthe minus strand. The fragments (8, 10). were separated by centrifugation through a 5-20% gradient of From an experimental point of view, one of the most inter- sucrose. According to the approximate specific activity of the esting pairings promoted by recA protein is that ofcircular sin- [y-32P]ATP used to label the ends, we estimated that 30% of gle strands with linear duplex DNA. This reaction is rapid and 5' ends of plus strands were phosphorylated. efficient; its substrates and products are well defined. By elec- Duplex DNA 32P-Labeled at the 3' End ofthe Plus Strand. This DNA, which also consisted ofthe large fragment produced I with Sau 96 I and The publication costs ofthis article were defrayed in part by page charge by cutting fd form [3H]DNA both Hpa I, payment. This article must therefore be hereby marked "advertise- ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. * This is paper no. 12 in a series. Paper no. 11 is ref. 11. 4786 Downloaded by guest on September 29, 2021 Biochemistry: Kahn et aL Proc. Natl. Acad. Sci. USA 78 (1981) 4787

0~~~ ~ ~ ~ ~ ~ ~ ~ ~ 0~ A~~~~~~~~~ 0~~~~~~~~~6~~~~ 'ZO I I~~~~~~~~~3

4.~~~~~ ~ ~ ~~~~~~~~~~~*~

4!~ ~ ~

Fig3.BndC)Cicuarsigl-srade G DA:nDN i jin mleuls it te mal ramet. D)Cicuarsigl-srade f jin molcueiththlrgfrgmntArow, ppren edsof etroupex egon frme b prtil rasfr o astandfrm helinardule moleculemoifedbyKegsrtothesingle-strandede a. 19. circle.h cndtinsusdweeReaction products wereNAat0.spread fr.t/melectroni 5%microscopyomaid,by the.1Mformamidemonumactae,000%technique of Davis etytchomal. (18) a (extratedfrmhorsheart, spred ontoa hypohase o doubl-distiled wter. Te grid were otary hadowe with ungste (20) t an agle o 80.TheDNAwasexamined with a Philips EM2O1 electron mirsoe

was labeled by a method to be described elsewhere. Other cular single strands as linear duplex molecules, 80% of the du- methods were as described (7, 8, 10, 14). plex DNA appeared in joint molecules in 10 min (Figs. 1 and 2). When we heated deproteinized samples for 4 min at 41'C RESULTS prior to assayingjoint molecules, we found nearly as much prod- Time Course ofFormation ofJoint Molecules from Circular uct as in the unheated samples (Fig. 2). At 500C there was a Single-Stranded DNA and Linear Duplex DNA by recA Pro- significant loss ofjoint molecules, but the fraction ofjoint mol- tein. In a reaction mixture containing about twice as -many cir- ecules that survived heating at 50'C increased from 27% at 5 min to 91% at 30 min. This increase in thermal stability of the products of the reaction as a function of the duration of the re- action is consistent with a progressive increase in the average length of heteroduplex joints. 80 As in the formation ofjoint molecules by single-stranded frag- ments and duplex DNA (21), the kinetics of formation of the joint molecules described here resemble Michaelis-Menten -@ 60 kinetics (unpublished data), which suggests that the mechanism u 1.0 offorming joint molecules from single-stranded circles and lin- 0 ear duplex DNA is related to the mechanism offorming D loops 40 (3, 21, 22). 0.5 Specific Pairing of Circular Single-Stranded DNA with the 3' End ofIts Complementary Strand. Circular single-stranded 20 DNA isolated from phage particles consists of viral or plus strands that can pair only with the minus strand of the duplex DNA. Therefore, if the growth of heteroduplex joints is uni- directional, the single-stranded DNA should pair with only one 10 20 30 end of a duplex fragment and should displace a specific end of Time, min the plus strand from the duplex DNA. To look for pairing with. FIG. 2. Time course and thermal stability ofjoint molecules made a specific end of the duplex DNA we used Hpa I to cleave by recA protein from circular single-stranded and linear duplex DNA. M13Goril DNA into two fragments that had M13 DNA at one Reaction mixtures contained 4 yM linear duplex OX174 [3H]DNA, 4.8 end and G4 DNA at the other (Fig. 3). Hpa I cuts the two strands AM circular single-stranded 4OX174 DNA, 2.4 pM recA protein, 31 mM at the same position, creating blunt ends without single- Tris-HOl (pH 7.5), 3 mM dithiothreitol, 6.8 mM MgCl2, 2 mM sper- stranded extensions (27). The large fragment, which was 6783 midine, 1.3 mM ATP, 0.6 mM EDTA, bovine serum albumin at.97 ,g/ ml, and 2% (vol/vol) glycerol. After incubation at 370C forthe indicated base pairs long, had sequences from M13 at the 3' end of the periods, joint molecules were assayed by the D-loop assay as described complementary strand, whereas the small fragment, which was by DasGupta et al. (8). Prior to filtration through nitrocellulose, sam- 2040 base pairs long, had sequences from G4 at the 3' end, a ples that had been quenched with EDTA and deproteinized with Sar- conclusion based on the published nucleotide sequences kosyl were diluted in 1.5 M NaCl/0.15 M Na citrate and incubated for (24-26) and on the orientation of the inserted G4 fragment in 4 min at 000 (), 41°C (x), and 500C (o). In the control (A), heterologous M13Goril (23). The large difference in size ofthe two fragments single-stranded fd DNA replaced X174 DNA. Control samples treated made it easy to them electron at 0°0, 410C, or 500C prior to filtration gave the same values. (Inset) identify by microscopy (Fig. LA). Fraction of joint molecules that survived heating for 4 min at In one tube we mixed the preparation of M13Goril DNA 50°C-i.e., the ratio of a sample treated at 500C to a sample treated containing both Hpa I fragments with circular single-stranded at 000 as a function of the period of reaction at 37°C. G4 DNA, and in another tube we mixed them with circular sin- Downloaded by guest on September 29, 2021 '4788 Biochemistry: Kahn et, al. Proc. NatL Acad. Sci. USA 78 (1981)

75

50 x r-4 25 ° 40364° x x 04 u ¢C.)! 6783 2040 150 Z

5' 5Q12(+) ______.5'(+)11204 18361 cqw 100 X 5571 1 1012 I120418361 50 FIG. 3. Map of DNA from the chimeric phage'M13Goril and of the products produced by cleavage with Hpa I according to the-data of Kaguni and Ray (23), Godson et al. (24, 25), and van Wezenbeek.et al. 10 20 30 10 20 30 (26). The numbers represent base pairs. 5' (+), 5' end of the plus Fraction strand. FIG. 4. Concerted displacement of the 5' end of- duplex DNA by gle-stranded fd DNA, which is 97% homologous to M13 DNA recA protein in the presence of homologous single-stranded DNA. Dis- (26). Fd DNA paired only with the large fragments, whereas placement is revealed' by the action of exonuclease VII on 3P label situated exclusively at the 5' end of the plus strand in duplex fd [3H]- the G4 DNA paired only with the small fragments (Fig. 1; Table DNA. Reaction mixtures contained 4pM linearduplexfdDNA, 4.8pM 1). The yield of joint molecules in this experiment was low, circular single-stranded fd DNA, 2.4 pM recAprotein, 31 mM Tris HCl possibly because we did not repurify the fragments of duplex (pH 7.5), 3 mM dithiothreitol, 6.8. mM MgCl2, 2.0 mM spermidine, 1.3 DNA after we prepared them by restriction cleavage of form mM ATP, 0.6 mM EDTA, bovine serum albumin at 97 Ig/ml, and 2% I DNA. This experiment, however, was controlled internally: glycerol. After 15 min of incubation at 3700 we diluted the mixture large fragments and small fragments were present together in 1:125 by adding a solution containing a saturating amount of exonu- clease VII (6 units/ml) in 240 mM Tris HCI/50 mM potassium phos- the reaction mixtures, and circular single-stranded fd DNA phate/40 mM EDTA/50 mM 2-mercaptoethanol/0.7 M potassium made as many joint molecules with the large fragment as G4 chloride. The final mixture had a pH of 7.9. After 10 min at 37TC, we DNA did with the small fragment. loaded the mixture onto a gradient of 5-20% sucrose in 1.5 M NaCl/ The specific pairing of fd DNA with the large fragment and 10. mM Tris'HCl, pH 7.5/1 mM EDTA. Sedimentation was for 3.5 hr ofG4 DNA with the small fragment indicates that the formation at 50,000 rpm at 400 in a Beckman SW 50.1 rotor. (A) Reaction with ofheteroduplex joints is polarized and shows that the incoming recA protein but no added exonuclease VII. (B) Complete reaction con- taining both recA'protein and exonuclease VII. (C) Circular single- strand pairs only with the 3' end ofthe complementary strand. stranded 4X174 DNA in place.of homologous single-strandedfd DNA. Concerted Displacement ofthe 5' End ofPlus Strands from (D) recA protein omitted. x 5'-Terminal 32p label; *, 3H label uni- Joint Molecules. The above observations imply that the for- formly distributed. *mation ofajoint molecule results in specific displacement ofthe 5' end of the plus strand. To test- this inference, we prepared strand by digestion with Sau 96 I, which cut 683 base pairs from two duplex substrates, one labeled with 32P at the 5' end ofthe the appropriate end. The rest ofthe molecule, consistingof5725 plus strand and the other labeled at the 3' end ofthe plus strand. base pairs labeled at the 5' end ofthe plus strand, was reisolated To prepare the 5'-labeled substrates, we cleaved form I fd DNA by sedimentation in sucrose. at a single site with Hpa I and phosphorylated the 5' ends by The pairing ofthis substrate with circular single-stranded fd using [y-32P]ATP. We removed the labeled 5' end ofthe minus DNA produced material that sedimented faster than the linear DNA in a discrete peak with a long leading shoulder (Fig. 4A). Table 1. Polarized formation of joint molecules: Pairing at the 3' The faster sedimenting material did not appear when heterol- end of the complementary strand ogous DNA replaced single-stranded fd DNA (Fig. 4C), when No. of molecules recA protein was omitted (Fig. 4D), or when single-stranded recA, protein recA protein DNA was omitted (data not shown). To look for displacement fd ssDNA G4 ssDNA fd ssDNA of the 5'-labeled end, we used E. coli exonuclease VII, which acts specifically on single-stranded DNA at both 5' and 3' ends Large fragment 325 323 210 (13). When the 'product of the complete reaction was treated Small fragment 284 257 159 with exonuclease VII fbr 10 min, most of the 32P label was ssDNA 421 266 160 cleaved from the material in the faster sedimenting peak and Joint molecules: appeared near the top of the gradient (Fig. 4B). Some 32p was Largefiagment 19 0 0 also removed from the slower sedimenting material. None of Smallfiragment 0 22 0 the 3P label was susceptible to exonuclease VII when joint Reaction mixtures (20.5 1l.) contained 31 mM Tris'HCl (pH 7.5), 12.5 molecules were not formed because heterologous single strands mM MgCl2, 1.8mM dithiothreitol, bovine serum albumin at 88 I.Lg/ were used (Fig. 4C) or recA protein was omitted (Fig. 4D) or ml, 1.2 mM ATP, 8.8 ,uM fragments from Ml3Goril form I DNA single-stranded DNA was omitted (data not shown). We con- cleaved by endonuclease Hpa I (see Fig. 3), 4.4 IAM circular single- stranded phage DNA, and 2.0 pM recA protein., After incubation for clude that the formation ofjoint molecules made the 5' end of 30 min at 370C the reactions were quenched by the addition of 20 Al the viral strand susceptible to exonuclease VII. of 25 mM EDTA and spread for electron microscopy. ss, Single- Lack of Displacement ofthe 3' End of the Plus Strand. A stranded. substrate labeled specifically at the 3' end ofthe plus strand was Downloaded by guest on September 29, 2021 Biochemistry: Kahn et al. Proc. NatL Acad. Sci. USA 78 (1981) 4789

prepared by a procedure similar to that used to make the 5' Table 2. Lack of displacement of the 3' end of the labeled substrate. We first cleaved fd form I DNA with Sau 96 complementary strand I, which makes a single staggered cut, leaving 5' extensions 3 -32P(+) strand duplex fd [3H]DNA three nucleotides long. The 3' ends were then labeled by using E. coli polymerase I and [a-32P]dCTP plus unlabeled deoxynu- After exonuclease I cleoside triphosphates. Hpa I removed the unwanted label at Before exonuclease from 15 to 21 min: the 3' end of the minus strand by cutting 683 base pairs from I: Assay D, % Acid-soluble, % the appropriate end. The large fragment was purified by sedi- Reaction 3H 32P 3H 32P mentation in sucrose. a. Complete 75 78 2 18 Label at the 3' end ofthe plus strand was resistant to removal b. Heterologous by exonuclease VII whetherjoint molecules were formed or not ssDNA 4 2 0 6 (Fig. 5 A-C). As a positive control (Fig. 5D), we partially di- c. NossDNA 3 2 0 6 gested the 3'-labeled substrate with A exonuclease to make sin- d. 3'-Tailed DNA - - 25 99 gle-stranded 3' ends whose average length was 560 nucleotides; This experiment was identical to that in Fig. 5 except that, instead most ofthe 3' label in this material was removed by exonuclease of exonuclease VII, exonuclease I at 230 units/ml was added at 15 min, VII. with no other reagents. We measured the fraction of joint molecules The peak ofmore rapidly sedimenting material shown in Fig. formed at 15 min (assay D) and the amount of label made acid soluble 5D is attributable to joint molecules that recA protein makes by exonuclease I in the ensuing 6 min. No detergent or heat treatment from DNA with single-stranded tails, a reaction that we will was used in assay D. Heterologous DNA was circular single-stranded describe elsewhere. 4X174 DNA. Tailed DNA was the substrate described inFig. 5D; it had 3' single-stranded tails 560 nucleotides long. This reaction mixture To confirm that recA protein does not displace the 3' end, contained recA protein andheterologous single-stranded circular DNA we repeated the same experiment with exonuclease I, which plus all the other reagents present in reactions a-c. degrades single-stranded DNA from its 3' end (12). We used enough exonuclease I to make the label at single-stranded 3' positive control (Table 2, reaction d), we conclude that the 3' ends completely acid soluble (Table 2, reaction d). In the com- plete reaction containing the 3'-labeled duplex substrate, ho- end ofthe plus strand was not displaced in mostjoint molecules. mologous circular single-stranded DNA, recA protein, and ex- DISCUSSION onuclease I, we observed that, after subtraction of the blank, 12% of the 3' ends became acid soluble. On the basis of the The pairing ofcircular single-stranded DNA with linear duplex complete digestion ofan authentic single-stranded 3' end in the DNA by recA protein is a rapid and efficient reaction that pro- duces long heteroduplex joints (Figs. 1 and 2; ref. 8). The in- crease in stability of these joint molecules as a function of time is consistent with a progressive growth in the average size of 15 heteroduplex joints. The formation ofthese heteroduplex joints is polar (Fig. 1; Table 1): there is a strong bias that favors the x pairing of the circular plus strand with the 3' end of its com- R 10 plementary strand. In the sample of41 joint molecules observed co by electron microscopy (Table 1), no joints were found at the 5 5' end of the complementary which indicates C- strand, that the Ill preference for pairing at the 3' end is greater than 40:1. By 6 preparing specific terminally labeled substrates, we showed x that only the 5' end of the plus strand is displaced from the E 15 duplex DNA in a concerted reaction that occurs only when ho- mologous single-stranded DNA is present. The latter data sup- -!; 10 port our previous inference on the concerted nature of strand transfer by recA protein (7, 8, 21) and agree with other obser- 5 vations (ref. 28; unpublished data). recA protein forms heteroduplex joints at the 3' end of the complementary strand of linear duplex DNA; 15-30 min later, 30 heteroduplex joints have grown to thousands of base pairs in Fraction length. How does the joint grow? recA protein might specifi- cally recognize the 3' end ofthe complementary strand, initiate FIG. 5. Lack of displacement of the 3' end of the plus strand in the formation ofa joint, and either block its dissociation or rap- joint molecules. Evidence of displacement was sought by examining idly promote reassociation. According to the formulation of the effect of exonuclease VII on 32P label situated exclusively at the Thompson et al. (29), in 30 min a random one-dimensional walk 3' end of the plus strand of linear duplex fd [3H]DNA. Reaction mix- starting from such a reflecting barrier could result in joints that tures and conditions of sedimentation were as in Fig. 4. (A) Reaction are with recA protein but no added exonuclease VII. (B) Complete reaction hundreds to thousands of nucleotides long, whether the containing both recA protein and exonuclease VII. (C) Complete re- jump time per base pair is 170 ,usec (29) or 12 ,usec (30). Al- action with circular single-stranded 4X174 DNA in place of homolo- ternatively, recA protein might form the joint and actively en- gous single-stranded fd DNA. (D) Positive control. Exonuclease VII large it in one direction-for example, by binding or acting at acting on the [3'-32P]DNA substrate that had been partially digested the site of strand exchange in a polar way. by A exonuclease to make 3' single-stranded ends that were 560 nu- In order to make a stable joint cleotides long, according to the fraction of DNA made acid soluble. molecule, recA protein must After digestion by A exonuclease, 98% of the 32p was still acid insol- put molecules in homologous alignment, and it must find the uble. This DNA was extracted with phenol and ether, but the acid sol- free end ofa DNA chain from which it can initiate a strand trans- uble 3H-labeled nucleotides were not removed. The conditions were fer. It might find the free end first by specifically binding to otherwise identical to those in C. ends and then searching for homology, or it might find homol- Downloaded by guest on September 29, 2021 4790 Biochemistry: Kahn et aL Proc. Natl. Acad. Sci. USA 78 (1981)

ogy first and then look for an end. Some observations favor the 10. Cunningham, R. P., Wu, A. M., Shibata, T., DasGupta, C. & latter explanation: (i) recA protein can find homology in the Radding, C. M. (1981) Cell 24, 213-223. absence of any end (8, 10), and (ii) recA protein can use any 11. Williams, J. G. K., Shibata, T. & Radding, C. M. (1981)J. Biol transfer-e.g., the Chem. 256, 7573-7582. physical variant ofan end to initiate a strand 12. Lehman, I. R. & Nussbaum, A. L. (1964) J. Biol. Chem. 239, end ofa single strand, the end of a duplex molecule, or the ends 2628-2636. present at nicks and gaps (7, 8). Because these observations rel- 13. Chase, J. W. & Richardson, C. C. (1974) J. Biol Chem. 249, egate the physical nature of an end to secondary importance, 4553-4561. they also favor the view that the polar formation ofheteroduplex 14. Shibata, T., Cunningham, R. P. & Radding, C. M. (1981)J. Biol. joints reported here does not result from the binding of recA Chem. 256, 7557-7564. 15. Radding, C. M. (1971) Methods Enzymol. 21, 273-280. protein to a specific end but rather results from a directional 16. Beattie, K. L., Wiegand, R. C. & Radding, C. M. (1977) J. Mol. action of recA protein analogous to that of DNA helicases (31). Biol 116, 783-803. While this paper was in preparation, Cox and Lehman (32) 17. Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol 65, kindly sent us their manuscript which reports that, in a system 499-560. containing both E. coli single-strand binding protein and recA 18. Davis, R. W., Simon, M. & Davidson, N. (1971) Methods En- protein, the extension ofheteroduplex joints requires the con- zymoL 21, 413-428. 19. Keegstra, W., Baas, P. D. & Jansz, H. S. (1979)J. MoL.BioL 135, tinued action of recA protein. 69-89. In meiotic recombination there appear to be special sites at 20. Williams, R. C. (1977) Proc. Natl Acad. Sci. USA 74, 2311-2315. which the formation ofheteroduplex DNA is initiated and from 21. Radding, C. M., Shibata, T., DasGupta, C., Cunningham, R. P. which the heteroduplex regions grow in a polar fashion (33-36). & Osber, L. (1980) Cold Spring Harbor Symp. Quant. Biol 45, There are many nucleic acid enzymes whose action is polar, 385-390. including polymerases, , and the ATP-dependent )el- 22. Shibata, T., Cunningham, R. P., DasGupta, -C. & Radding, C. M. (1979) Proc. Nati Acad. Sci. USA 76, 5100-5104. icases (31, 37, 38). A whole set ofsuch enzymes may determine 23. Kaguni, J. & Ray, D. S. (1979) J. MoL Biol. 135, 863-878. the polarity of recombination in any particular pathway. Be- 24. Godson, G. N., Fiddes, J. C., Barrell, B. G. & Sanger, F. (1978) cause recA protein is directly implicated in recombination (39, in The Single-Stranded DNA Phages, eds. Denhardt, D. T., 40) and has recombination-like activities in vitro (2-10), the Dressler, D. & Ray, D. S. (Cold Spring Harbor Laboratory, Cold observation that it formsjoint molecules in a polarfashion brings Spring Harbor, NY), pp. 51-86. us a step closer.to understanding the molecular basis ofpolarity 25. -Godson, G. N. (1978) in The' Single-Stranded DNA Phages, eds. Denhardt, D. T., Dressler, D. & Ray, D. S. (Cold Spring Harbor in genetic recombination. Laboratory, Cold Spring Harbor, NY), pp. 671-695. 26. van Wezenbeek, P. M. G. F., Hulsebos, T. J. M. & Schoenmak- ers, J. G. G. (1980) Gene 11, 129-148. We thank John Chase for a gift ofE. coli exonuclease VII, John Flory 27. Garfin, D. E. & Goodman, H. M. (1974) Biochem. Biophys. Res. for expert guidance in our electron microscopic studies, and Anna Wu Commun. 59, 108-116. and John Flory for helpful comments on our manuscript.This reasearch 28. West, S. C., Cassuto, E. & Howard-Flanders, P. (1981) Nature was sponsored by grants from the American Cancer Society and the (London), 290, 29-33. National Cancer Institute. 29. Thompson, B. J., Camien, M. & Warner, R. C. (1976) Proc. Nati Acad. Sci.. USA 73, 2299-2303. 30. Radding, C. M., Beattie, K. L., Holloman, W. K. & Wiegand, 1. Gottesman, S. (1981) Cell 23, 1-2. R. C. (1977)J. Mol Biol 116, 825-839. 2. Weinstock, G. M., McEntee, K. & Lehman, I. R. (1979) Proc. 31. Kuhn, B., Abdel-Monem, M. & Hoffmann-Berling, H. (1979) Natl Acad. Sci.. USA 76, 126-130. Cold Spring Harbor Symp. Quant. Biol 43, 63-67. 3. Shibata, T., DasGupta, C., Cunningham, R. P. & Radding. C. 32. Cox, M. M. & Lehman, I. R. (1981) Proc. Nati Acad. Sci. USA M. (1979) Proc. Natl Acad. Sci. USA 76, 1638-1642. 78, 3433-3437. 4. Shibata, T., DasGupta, C., Cunningham, R. P., Williams, J. G. 33. Lissouba, P. & Rizet, G. (1960) C. R. Hebd. Seances Acad. Sci. K., Osber, L. & Radding, C. M. (1981) J. Biol. Chem. 256, 250, 3408-3410. 7565-7572. 34. Whitehouse, W. L. K. (1973) Towards an Understanding of the 5. McEntee, K., Weinstock, G. M. & Lehman, I. R. (1979) Proc. Mechanism ofHeredity (St. Martin's, New York). Natl. Acad. Sci. USA 76, 2615-2619. 35. Fogel, S., Mortimer, R., Lusnak, K. & Tavares, F. (1979) Cold 6. Cassuto, E., West, S. C., Mursalim, J., Conlon, S. & Howard, Spring Harbor Symp. Quant. Biol 43, 1325-1341. Flanders, P. (1980) Proc. Nati Acad. Sci. USA 77, 3962-3966. 36. Rossignol, J.-L., Paquette, N. & Nicolas, A. (1979) Cold Spring 7. Cunningham, R. P., DasGupta, C., Shibata, T. & Radding, C. Harbor Symp. Quant. Biol 43, 1343-1352. M. (1980) Cell 20, 223-235. 37. Kornberg, A. (1980) DNA Replication (Freeman, San Francisco). 8. DasGupta, C., Shibata, T., Cunningham, R. P. & Radding, C. 38. Yarranton, G. T. & Gefter, M. L. (1979) Proc. Natl Acad. Sci. M. (1980) Cell 22, 437-446. USA 76, 1658-1662. 9. West, S. C., Cassuto, E. & Howard-Flanders, P. (1981) Proc. 39. Clark, A. J. (1973) Annu. Rev. Genet. 7, 67-86. NatL Acad. Sci. USA 78, 2100-2104. 40. Kobayashi, I. & Ikeda, H. (1978) Mol Gen. Genet. 166, 25-29. Downloaded by guest on September 29, 2021