Early stages in RecA protein-catalyzed pairing

J. Mol. Biol. (1992) 228, 409-420

Early Stages in RecA Protein-catalyzed Pairing Analysis of Coaggregate Formation and Non-homologous DNA Contacts Janet M. Pinsince and Jack D. Griffith? Lineberger Comprehensive Cancer Center and Department of Microbiology and Immunology University of North Carolina, Chapel Hill, NC 27599, U.S.A. (Received 13 April

1992; accepted 31 July

1992)

RecA protein will catalyze the in vitro pairing of homologous DNA molecules. To further explore the events involved in the search for homology, we have applied a nitrocellulose filter binding assay to follow pairing, and a sedimentation assay to follow the generation of aggregates (termed coaggregates) formed between RecA-complexed single-stranded (ss) DNA and double stranded (ds) DNA. Electron microscopy (EM) was used to visualize the structures involved. RecA protein promoted the pairing of circular Ml3 ssDNA and linear M13mp7 dsDNA efficiently in the absence of coaggregates. Indeed, pairing of homologous ss- and dsDNAs involved coaggregate formation only if the dsDNA was circular. For DNAs cont,aining only a few hundred base-pairs of homology, for example pUC7 dsDNA and MlSmp7 ssDNA, pairing and joint formation was observed if the dsDNA was superhelical but not if it was topologically relaxed or linear with the homology internal to an end of the dsDNA. The effect of non-covalently attached heterologous dsDNA on the RecA-promoted joining of Ml3 ssDNA and linear M13mp7 dsDNA (with non-Ml3 sequences at both ends) was found to depend on the topology and concentration of the heterologous DNA. A tenfold excess of superhelical pBR322 DNA strongly inhibited pairing. However, addition of relaxed or linear pBR322 DNA to the pairing reaction had little effect. As seen by EM, superhelical pBR322 DNA inhibited joint formation by excluding the homologous dsDNA from the coaggregates. EM also revealed heterologous DNA interactions presumably involved in the search for homology. Here the use of EM has provided a direct visualization of the form and architecture of coaggregates revealing a dense interweaving of presynaptic filaments and dsDNA. Keywords:

RecA protein; electron microscopy; coaggregates; strand exchange

1. Introduction RecA protein plays a central role in the major recombination in pathway of homologous Escherichia coli. In vitro, purified RecA protein will promote simple homologous pairing and strand transfer reactions (McEntee et al., 1979; Shibata et al., 1979). These reactions can be divided into three experimentally separable stages: presynaptic filament formation, joint formation, and strand transfer. In the first stage, RecA protein binds and assembles onto single-stranded DNA (ssDNA$) to t Author to whom all correspondence should be addressed. $ Abbreviations used: ssDh’A, daDNA, single-stranded and double-stranded DNA, respectively; EM, electron microscopy; SSB, single strand DNA binding protein; bp. base-pair(s).

form a helical nucleoprotein (presynaptic) filament whose assembly requires a single strand DNA binding protein (SSB), Mg’+ and ATP (Cox & Lehman, 1982; Griffith et al., 1984; Menetski & Kowalczykowski, 1985; Thresher et al., 1988). The second stage, orchestrated by the presynaptic filament, involves a search for homologous sequences in double-stranded DNA (dsDNA), resulting in the formation of a synapsis or joint. If homologous DNA ends are present, the reaction proceeds to the final stage, the transfer of DNA strands and the resolution of the products (Cox & Lehman, 1982; Christiansen & Griffith, 1986; Register et al., 1987; for reviews, see Cox & Lehman, 1987; Griffith & & Harris, 1988; Radding, 1989; Eggleston Kowalczykowski, 1991). As the molecular mechanisms of homologous recombination catalyzed by a variety

of homologous

pairing

and strand

transfer

409 0022-2836/92/220409-12

$08.00/O

0

1992 Academic

Press Limited

410

J. M. Pinsince

proteins are being elucidated, the basic mechanism of the search for homology remains one of the least understood aspects of this process. In vitro, two different kinds of joints promoted by RecA protein, termed plectonemic and paranemic, have been described. Plectonemic joints require a free homologous DNA end, and strands from the two DNAs are paired in a Watson-Crick fashion making these joints stable to deproteinization. Plectonemic joints are the immediate precursors to strand transfer. Paranemic joints were first identified as intermediates in RecA protein-mediated homologous pairings (Bianchi et al., 1983; Riddles & Lehman, 1985) that required the continuous presence of RecA protein for their stability. The dsDNA within paranemic joints appears to be partially unwound (Wu et al., 1983; Christiansen & Griffith, 1986; Schutte & Cox, 1988), and the nature of the interaction between the sbrands of the t,wo DNAs in these joints is poorly understood. An organized interaction between DNA strands within paranemic joints may only encompass 200 to 250 base-pairs (Bortner & Griffith, 1990; Umlauf et al.. 1990): however, a looser association may extend over thousands of base-pairs. Structural interactions between DNA molecules engaged in a search for homology which occur prior t’o paranemic joining are not well understood. Tt has been observed that DNA involved in the early stages of pairing is often present in an aggregate comprised of ssDNA, dsDNA and RecA protein so large that it can be sedimented from solution by low speed centrifugation (Tsang et aZ., 1985). Termed coaggregates, they may facilitate the efficient pairing of DNA molecules. It has been proposed that coaggregates are an obligatory intermediate in joint formation (Gonda & Radding, 1986); however. this conclusion was based on the analysis of only one pair of DNA templates: circular ssDNA and homologous linear dsDNA. When the rate and extent of DNA pairing by the bacteriophage T4 UvsX protein was optimized, coaggregates could not be detected (Harris &. Griffith, 1988). Further, in electron microscopic (EM) studies of RecA proteinmediated joining in t’his laborat’ory, we have observed efficient joining of DNA molecules without the presence of large aggregates. Finally, kinetic evidence indicates that ternary complexes, if formed during the search for homology, are weakly bound and short-lived (Julin et al., 1986). Thus, although the formation of coaggregates as an early step in joining remains an attractive vehicle for facilitating the search for homology, the required role of coaggregates remains unproven. Although it, is clear that non-homologous contacts between the pairings DNAs must be transiently made during the search for homology, evidence about’ such contacts is scanty, and the physical problem appears complicated by the fact that the ssDNA lies within a sheath of RecA protein and may be in an extended conformation which may not match the helical repeat of the dsDNA (Tsang et al., 1986). Using a slowly hydrolyzed ATP analog

and J. D. Gr
(ATP$) to force RecA protein to bind t,o dsDNA. Muller et al. (1990) observed that RecA-dsDNA filaments could bind heterologous SSDSA, Takahashi et al. (1989) used flow linear dichroism and etheno-modified ssDNA to search for heterologous DNA contacts, and found evidence for interaction of two heterologous ssDNAs. Although encouraging, these studies have not utilized the struct,ures: premost physiologically relevant synaptic filaments formed in the presence of ATP along unmodified ssDNA, and protein-free duplex DNA. An additional physical problem that must be considered is that while the in vitro pairing reactions are normally conducted using DNA molecules that, are nearly lOOo/o homologous, in the cell. t’he search for homology must proceed rapidly in the presence of large amounts of attached and unattached non homologous DNA sequences. How the presence of large amounts of non-homologous DNA effects the formation of coaggregates or the early events of strand exchange is poorly understood. Honigberg it al. (1986), however. have shown that a search can bc> successfully conducted in the presence of a 1006fold excess of non-homologous DNA. To learn more about the nature of t#he rarl,v events in pairing, it would be useful to develop reaction conditions free of coaggregate formation. These conditions could then be used to invest’igat’e how the search for homology is influenced by non covalently attached non-homologous dsDNA. and to search for evidence of heterologous contacts between DNA molecules. Three assays could hi employed to monitor pairing reactions catalyzed by RecA protein: a nitrocellulose fllt’er binding assay to follow the progress of joint formation, a sedimrntation assay to follow coaggregate formation. ant-l EM to directly visualize the interactions between presynaptic filaments and dsDNA. Using these assays and a combinat,ion of different DNA substrates. we report that coaggregate formation was easily de&ted during pairing but that thrl presence of coaggregat,es was dependent on thNA to a pairing reaction was found to inhibit joint formation but t)he extent of inhibition was dependent on the con and topology of thr heterologous centrat’ion dsDNA. As visualized by EM, the inhibition of pairing by haterologous dsl>NA resulted from a direct interaction between the presynapt,ir filaments and the heterologous dsI>NA. Finally. we have directly observed an interaction het.ween prrtsynaptic filaments and non-homologous dsl>Nr\. an int,eraction that may represent the first step in thr search for homology.

2. Materials (a) Protrins

and Methods and IjSrl

s

SSB protein (Chasr rt nl.. 1980). ItwA pdrin ((:rilfith & Shores. 1985) and Ml4 d&type (MlSwt~) and M13mp7

RecA Protein-catalyzed ss- and dsDNA (Register & Griffith, 1986) were purified as described. Plasmid DNA (pBR322) was purified using an alkaline lysis method (Maniatis et al., 1982). Restriction enzymes were purchased from GIBCO BRL (BarnHI and P&I) and ru’ew England Biolabs (NdeI, Sea1 and SspI). Kicked Ml3 dsDNA was obtained from a preparation of dsDNA that nicked. superhelical was >80% Topoisomerase I was obtained from GIBCO BRL and used according to t’he manufacturer’s directions. (1)) RecA protein-facilitated

joint formation

Ml 3 ssDNA at a concentration of 2 pg/ml (6.2 PM) in 20 mi?-Hepes, @l mM-EDTA (pH 7..5), was heated to 65°C for 5 min then cooled to 37°C. Magnesium acetate and ATP were added to concentrations of 12 mM and 3 mM, respectively. A protein mixture containing 80 pg/ml (2.1 PM) RecA protein, 60 mM-NaCl, 12 mM-magnesium acetate. 40 mw-phosphocreatine, 8 pg phosphocreatine kinase/ml and 3 mM-ATP in 20 mM-Hepes, @l mM-EDTA (pH 7.5) was incubated at 37°C for 10 min then combined with an equal volume of the Ml3 ssDPu’A mixture. Sfter 5 min at 37”C, SSB protein was added to a final concentration of 3 pg/ml (cl7 PM) and the incubation continued for 10 min to complete the formation of presynaptic filaments. NA. (c) Filter binding assay Joint formation between ssDNA and [3H]dsDKA was evaluated by nitrocellulose filter binding. Portions of reactions were diluted in an equal volume of cold reaction buffer (20 mw-Hepes, @l mM-EDTA, 30 m&i-Pu’aCl, 12 mw-magnesium acetate, pH 7.5). Samples were immediately applied to nitrocellulose membranes (Schleicher & Schuell BA85), soaked in 1 mM-ATP. Kitrocellulose filters were rinsed with 5 ml of cold reaction buffer. dried under a heat lamp and counted by liquid scintillation counting. Control reactions indicated that the background amount of dsDNA retained, independent of joint’ format,ion, was approximately 5:/, of the total concentration of dsDPu’A. All experimental results were reduced by this value. This assay measures both paranemic and plectonemic joints and differs from the filter binding assay described by Bianchi et al. (1983); here the modifications were designed to minimize the disruption of weak associations occurring between DKA molecules during joint formation. A comparison between the two filter binding protocols indicated that a higher and more consistent level of joint formation was detected using the modifications described here. (d) (Joaggregate assay The presence of coaggregates was determined by the method of Honigberg et al. (1986). In brief, radiolabeled dsDNA was added to a 50 ~1 joint formation reaction. The sample was centrifuged in an Eppendorf microfuge for 3 min. Three 15 ~1 portions were removed and the remaining 5 ~1 and pellet were resuspended in 200 ~1 of water. The radioactivity in each of the 3 supernatant portions and the pellet fraction was measured by liquid scintillation counting.

411

Pairing (e) Electron microscopy

Samples to be visualized by EM were removed from the reactions and fixed with glutaraldehyde as detailed in text. The fixed samples were then diluted with 20 m&I-Hepes, @l mM-EDTA (pH 7.5) and absorbed to thin, glow-charged carbon films, washed, dehydrated with ethanol and rotary shadowed with tungsten as described (Griffith & Christiansen, 1978). Micrographs were taken on a Philips EM 400TLG.

3. Results (a) Pairing of circular ssDNA and linear dsDNA occurs eficiently in the absence of coaggregates The pairing of Ml3 wild-type (wt) ssDNA with linear M13rqp7 dsDNA cleaved at the center of the 840 bp lac insert can only proceed to the first stage of pairing due to the non-homologous sequences at the ends of the dsD?$A. Here circular Ml3wt ssDNA was assembled with R&A protein into presynaptic filaments utilizing an SSB protein-mediated pathway (Materials and Methods). The reaction 3 mM-ATP, mixture included 30 miWNaC1, 12 mM-Mg2+ and @2 PM-SSB and an ATP regeneration system. These conditions have been shown to be optimal for facilitating the assembly of RecA protein onto ssDNA (Thresher et al., 1988). Once the filaments linear presynaptic were formed, [3H]M13mp7 dsDNA, cleaved in the polylinker region, was added to the reaction mixture to initiate joint formation. Portions of the reaction mixture were removed at various times and analyzed for joint and coaggregate formation as described below. Using a nitrocellulose filter binding assay optimized here to detect both paranemic and plectonemic joining (Materials and Methods), the extent of total joint formation was followed over a 60 minute period. In this assay, 3H-labeled dsDNA is retained on the filters only if it is either complexed by RecA protein or joined with presynaptic filaments. The extent of joint formation reached a maximum of 70 to 80% of the dsDNA engaged in joints with the ssDNA between 10 and 20 minutes (Fig. 1). Application of the sedimentation assay (Materials and Methods) indicated that the linear [3H]M13mp7 dsDNA did not pellet during centrifugation, showing that the conditions used for joint formation here were coaggregate-free (Fig. 1). Similar results were observed with a substrate pair capable of stable plectonemic joint formation, circular M13mp7 ssDNA and linear 13H]M13mp7 dsDNA (Fig. 1). The pairing of circular M13wt ssDNA and linear M13mp7 dsDNA was examined by EM. Only free, individual presynaptic filaments and joint complexes comprised of a single presynaptic filament joined to a dsDNA molecule were observed (Fig. 2), with no large aggregates apparent. The majority of the joint complexes evaluated appeared to have incorporated most of the dsDNA into the nucleoprotein filament leaving the ends of the dsDNA exposed. These results show that efficient joint formation can occur in the absence of coaggre-

412

J. M. Pinsince

loo

and J. D. GrifJith

Table 1

3zi

80

8 &

60

8 8

Comparison

of coaggregates formation substrate pairs

hetween six

t

.E

time (minutes) Figure 1. Kinetics of joint and coaggregate formation with circular ssDNA and linear dsDNA. Two DNA pairings: circular M13wt ssDNA paired with linear M13mp7 dsDNA (having non-homologous ends: 0, l ), and circular M13mp7 ssDNA paired with linear M13mp7

dsDNA (having homologous ends: 0, n ) were assayed for joint and coaggregate formation. RecA protein was assembled onto ssDNA using an SSB protein-mediated pathway,

and the reaction

3H-labeled binding

dsDNA;

initiated

joints

assay and coaggregates

(Methods and Materials).

by addition

were followed

of the

by a filter

by a sedimentation

Six different substrate pairs that are competent f’or joint formation were assayed for coaggregate formation using the sedimentation assay and the DNA and protein concentrations indicated in Fig. 1: A, circular M13mp7 ssDNA paired with linear M13mp7 dsDNA; B, circular M13wt ssDNA paired with linear M13mp7; (‘. linear M13mp7 ssDNA paired with superheliral M13wt dsDNA: D, circular M13wt ssDNA paired with nicked M13wt dsDNA; E, circular M13mp7 ssDNA paired with superhelical M13mp7 with superhelical

dsDNA; and F. circular M13mp7 dsDNA.

M13wt

ssDNA

paired

assay

The joining reactions contained

3 PM-ssDNA, 1 PM-RecA protein. 0.2 PM-SSB protein and 3 PM-dsDNA, while the coaggregate reactions contained 123 PM-ssDNA, 4.2 PM-RecA protein, @7 PM-SSB protein

and 12.3 PM-dsDPu’A.

gates and are in agreement with previous studies of pairing by these templates (Christiansen & Griffith, 1986; Bortner & Griffith, 1990). Bortner & Griffith (1990) reported structural differences in the paranemic joints formed between DNA molecules which were dependent on the dsDNA topology, suggesting that an examination of other DNA substrate pairs would be important. (b) Pairing of circular ssDNA and superhelical dsDNA involves the rapid formation of coaggregates To examine the dependence of joint formation on the topology of the dsDNA, four DNA pairs were compared: linear M13mp7 ssDNA paired with superhelical Ml3wt dsDNA; circular Ml3wt ssDNA paired with nicked M13wt dsDNA; circular M13mp7 ssDNA paired with superhelical M13mp7 dsDNA; and circular Ml3wt ssDNA paired with superhelical M13mp7 dsDNA (Fig. 3). Each pair can only form paranemic joints due to the absence of free homologous DNA ends. As measured by filter binding, the total number of joints formed between linear M13mp7 ssDNA and superhelical [3H]M13wt dsDNA increased rapidly with time until a maximum level was attained at 10 to 20 minutes, and this was maintained to 60 minutes (Fig. 3). For the other DNA pairings (circular ssDNA paired with either nicked or superhelical dsDNA), the number of joints increased during the first five minutes and then leveled off.

Coaggregate formation was evaluated during these incubations. For each substrate pair, portions from the reactions were assayed for coaggregat,es and their formation was found to be time-dependent (Table 1). R#egardless of whether the ssI)NA was linear or circular, approximately 3096 of the superhelical dsDNA was incorporated into coaggregates at early times. The amount of dsDNA in the coaggregates decreased after five minutes and COW tinued to decrease with continued incubation. When nicked dsDNA was used instead of superhelical dsDNA, at most 157” of the dsDNA pelleted in the coaggregate assay. indicating that coaggregates had formed but at a lower level. To examine the structures which sedimented in the coaggregate assays, pairing react,ions wert carried out using circular M13wt ssDNA and M13mp7 dsDNA (80% superhelical. 2O?h nicked). Following a 20 minute incubation, the samples wertb fixed with 2.6% (v/v) glutaraldehyde for five minutes at 37”C, and directly applied t,o the carbon foil support and processed for EM. As shown (Fig. 4), very large aggregates of presynaptic filamenth and dsDNA were observed. Much of the dsDXA contained in the aggregate appeared t,o be relaxed. indicative of the unwinding associated wit,h joint formation (Christiansen c1:Griffith: 1986). Somewhat less than half of the presynaptic filamentJs and dsDNA molecules in the reaction was contained within these networks with the remainder seen as free molecules. These EM observations are in good agreement with the sedimentation assay data (Table I), which indicated that approximately 3Oy,, of the dsDNA in the reaction was incorporated into coaggregates. Thus, it appears that coaggregat,c formation as well as paranemic joint formation depends on DNA tJopology.

RecA Protein-catalyzed

Pairing

413

Figul re 2. Visualization of joining by circular SSDNA and linear dsDNA with non-homologous ends. Circular Ml .3wt ssDNA and linear M13mp7 dsDNA (with non-homologous ends) were incubated for 20 min with RecA protein n as describe ?d for Fig. 1. The sample was fixed by the addition of glutaraldehyde to 2.6% for 5 min at 37”. The sample was then dil luted. directly applied to the EM support, washed, dried and rotary shadowcast with tungsten. As shown he]re in reverse contrast, individual presynaptic filaments and joint complexes were observed in the absence of aggregate !s or networl
(c) The pairing of limited homology substrates depends on the topology of the dsDNA In the pairing of fully homologous DNA molecules described above, coaggregate and paranemic joint formation was seen to depend on the topology of the dsDNA. Here DNA molecules possessing limited homology were examined to determine whether the location of the homologous sequences relative to the DNA ends as well as the topology of the dsDNA would influence joint formation. To examine

this, circular

M13mp7

ssDNA

was

paired with pUC7 dsDNA which shares 436 bp of homology located in the polylinker region of both DNAs.

When pairing reactions were carried out using either superhelical pUC7 DNA or linear pUC7 DNA (cleaved in the center of the polylinker to produce homologous ends) and circular M13mp7 ssDNA, efficient joining was observed by filter binding (Fig. 5A). In contrast, joint formation was not detected between these DNA substrates if the pUC7 DNA

had been topologically-relaxed

with topoisomerase I

(Fig. 5A). With these DNA assays detected coaggregates

pairs, sedimentation only in the reactions

containing

superhelical

pUC7 DNA.

Coaggregates

were absent if the pUC7 DNA was cleaved in the polylinker region or if it was topologically relaxed

(data not shown). When linear pUC7 DNA was paired with circular M13mp7 ssDNA, the ability of the molecules to pair was found to be depend on the location of the homologous sequences relative to the DNA ends. Linearizing pUC7 DNA with PstI or BamHI produces DNA ends with homology to the ssDNA allowing plectonemic joints to form. As shown in Figure 5A, circular M13mp7 ssDNA paired with linear pUC7 DNA possessing homologous ends formed joints efficiently but without detectable coaggregate formation (data not shown). Other cleavages of pUC7 DNA yielded linear dsDNA with ends lacking homology to M13mp7 DNA. These cleavages placed the homologous DNA segment 744,

J. M. Pinsince and J. D. Gri@h

414

60

0’

0

2'0

time

.

4'0

6b

(minutes)

Figure 3. Comparison of pairing between 4 substrate pairs blocked from plectonemic joining. Four different substrate pairs that are blocked from plectonemic joining were assayed for joint formation using the filter binding assay and the DNA and protein concentrations indicated for Fig. 1: linear M13mp7 ssDNA paired with superhelical Ml3wt dsDNA (a); circular M13wt ssDNA paired with nicked M13wt dsDNA (0); circular M13mp7 ssDNA paired with superhelical M13mp7 dsDNA (0); and circular M13wt ssDNA paired with superhelical M13mp7 dsDXA (0).

420, or 52 bp from the nearest end (BcaI, XspT and NdeI cleavages, respectively). For these three linear pUC7 DNAs, no joint formation was observed with M13mp7 ssDNA as assayed by filter binding (Fig. 5B). The sedimentation assay also failed to detect coaggregate formation bet’ween the presynaptic filaments and linear pUC7 DNA containing internal homology (data not shown). These results show that joining (paranemic or plectonemic) requires either dsDNA molecules with homologous ends or a superhelical topology, and that coaggregate formation requires superhelical dsDNA.

attached heterologous inhibits joint formation in a topology-dependent mwnner

(d) Non-covalently

DNA

In the experiments described above, the nonhomologous sequences were covalently attached t’o the dsDNA. Here we describe the effect of noncovalently attached heterologous dsDNA on joint formation. Circular M13wt ssDNA was paired with linear M13mp7 dsDNA, a combination which showed efficient joint formation in the absence of coaggregates (Fig. 1). To this reaction, increasing amounts of superhelical pBR322 DNA were added prior to the addition of linear M13mp7 dsDNA. Joint formation was inhibited as assayed by filter binding with the extent of inhibition depending on the concentration of superhelical pBR322 DNA (Fig. 6A). When topologicallg relaxed, covalently

Figure 4. Visualization of coaggregates formed brtween homologous DNA molecules. Circular Ml3wt ssDXA and a mixture of 80% superhelical and 20% nicked c,ircular M13mp7 dsDPu’A were incubated with RecA protein in a reaction that included 0.2 PM-SSB protein. 1 PM-RwX protein. 3 ~IV-~SDK’A and 3 PM-dsDNA as described for Fig, 1. Following a 20 min incubation, the sample was fixed and prepared for EM as described for Fig. P. Networks of presynaptie filaments and dsDSA comprised > 300/, of the material present. Shown in reverse c%ontrast’. The bar represents 0.2 pM.

closed pBR322 DNA was used instead, no inhibition of joint formation was observed, even with 10 to 15 t,imes as much pBR322 DNA as the homologous linear M13mp7 dsDNA (Fig. 6B). Linear pBR322 DNA also produced no inhibition of joint formation (Fig. 6C) with a loo-fold excess of linear pBR322 DNA reducing the total number of joints by less than 15% as contrasted to the reaction lacking any pBR322 DNA. If the heterologous supertwisted DNA was added prior to the addition of the homologous dsDNA, no difference in the outcome was observed (data not shown). Coaggregates were not observed when circular Ml3wt ssDNA was paired with linear M13mpi dsDNA in the absence of heterologous DNA as described above. However, the observation that this pairing was inhibited by the addition of superhelical pBR322 DNA led us to examine the structures formed in these reactions by EM. Incubations wet-c prepared using circular M13wt ssDNA, linear M13mp7 dsDNA and a tenfold excess of pBR322 DNA. Portions were removed after 15 minut,es.

RecA Protein-catalyzed

0

20

40

60

20

40

60

Pairing

415

100-B 60_ 6040209

0

time

(minutes)

Figure 5. Joining between M13mp7 ssDNA and pUC7 dsDNA: 2 DNAs sharing limited homology. Ml3mp7 and pUC7 DNAs share 436 bp of homology in the Zac insert. Here joining between M13mp7 ssDNA and PUC7 DNA was followed using filter binding. The reactions included 12.3 PM-ssDNA, 4.2 PM-RecA protein, 97 PM-SSB protein, and 4.6 ~-[~H]pUc7 DNA. In A, circular M13mp7 ssDNA was paired with: superhelical pUC7 DNA (m); linear pUC7 DNA with homologous ends (a); or topologically relaxed pUC7 DNA ( q ). In B, 3 pairings were examined in which the homology between the DNAs was internal from the ends of the dsDNA. Circular M13mp7 ssDNA was paired with linear pUC7 dsDNA cleaved with EcuI to place the homology 744 bp from the nearest end (@), Sap1 to place the homology 420 bp from the nearest end (O), or Ndel placing the homology 52 bp from the nearest end (al).

fixed with 2.6% glutaraldehyde, and directly applied to the EM supports. Coaggregates of presynaptic filaments and dsDNA were present in abundance (Fig. 7), with the superhelical pBR322 DNA clearly incorporated into them. The pBR322 DNA in the networks appeared to be relatively supercoiled; however, the DNA could have lost 30 to 500/b of its superhelicity and still appeared relatively twisted as seen by EM. Only a few presynaptic filaments were free of the coaggregates,

60

20 0 0

20

time

40

60

(minutes)

Figure 6. Dependence of joint formation on the presence of non-covalently attached heterologous DNA. The pairing of circular M13wt ssDNA and linear M13mp7 dsDNA (with non-homologous ends) was followed using filter binding as described for Fig. 1. Increasing concentrations of pBR322 DNA were added to the mixture 10 min prior to the addition of the [3H]M13mp7 dsDNA. A, Superhelical p5R322 DNA added to 0 ,LJM (a), 15 PM (u), 30 PM (0) or 46 PM (0). B, Topologically relaxed pBR322 DNA was added to 0 PM (O), 30 PM (O), or 46 pM ( n ). C, Linear pBR322 DNA was added to 0 PM (O), 46~~ (m), 92p~ (m), 153~~ (0) or 307~~ (0).

J. M. Pinsince

Figure 7. Visualization

of coaggregates formed in the presence of homologous and non-homologous dsDKA. A pairing reaction containing Ml3wt. ssDNA (3 PM). linear M13mp7 dsD?iA with non-homologous ends (3 PM). and 30 M-pBR322 DNA, was prepared as described for Fig. 6. Following a 20 min incubation. the sample was fixed and prepared for EM as described for Fig. 2. This micrograph shows a coaggregate of presynaptic filaments that has incorporated the heterologous superhelical pBR322 DKA. The homologous linear dsDNA is apparently not) present in the network. Shown in reverse contrast at a 30” tilt. The bar represents @2 pm.

which appeared large and densely packed. These results appeared in conflict with a sedimentation assay in which the linear Ml3mp7 dsDNA was 3H-labeled that showed no involvement of the linear [ 3H]M13mp7 dsDNA in coaggregates. However. when the coaggregate assay was repeated using superhelical 13H]pBR322 DNA. it was found that superhelical the 40% of approximately [ 3H]pBR322 DNA sedimented. (e) Non-homologous contacts can be observed between poresynaptic jilaments and superhelical dsDNA

by EM

The results above revealed an interaction between presynaptic filaments containing Ml3wt ssDNA and superhelical pBR322 DNA that excluded the homologous linear M13mp7 dsDNA. This suggested that between non-homologous a direct interaction dsDNA and a presynaptic filament might also be observed in the absence of homologous dsDNA. Interactions between DNA molecules at a region of

and J. D. Gr
Figure 8. Visualization of c*oaggregat,es fortnrtl bc>twrrll non-homologous DNAs. Ml3wt ssDPu’A was a.ssemblrd into presynaptic: filaments with RecA protein as for Fig. i and supertwisted pBR322 dsDru’A then added for -11)tnin at 37°C’. The sample was then fixed and preparer1for EM as described for Fig. 2. This micrograph shows a coaggrrgate comprised of presynaptic filaments and the h&n)logous superhelical pBR322 Dh‘A. Shown in rtxvt‘rsc contrast. The bar represents @2 pm.

heterology could represent an early active stage of the search for homology. Here joint rearCons were prepared as described above but in the absence of homologous dsDNA. Using the nitrocellulose filter binding assay, approximately 40!/,, of the superhelical 13H]pBR322 DNA in the reaction mixture was retained on the filters, revealing a direct int,c:raction wit,h the presynapt,ic filament)s. When thr samples were examined by EM, aggregates of prrsynaptic filaments containing circular Ml3wt ssDNA and superhelical pBR322 DNA rrsembling coaggregates werfl seen (Fig. 8). Thesrl structures were less dense and morp loosely associated than those formed in the presence of homologous linear M13mp7 dsL)NA (Fig. 7). As abovr. no extensive unwinding of the superhelical hrterologous DNA observed within t,hese structures. To verify that the inhibition of joint formation by the superhelical pBR322 DNA was not due to its sequestering RecA protein from the reaction, incubations were conducted with increasing concentraThe RecA prot,rin tions of RecA protein. concentration was increased from 1 p~ to 2.1 PM. to 3.1 PM. or to 63 pM, and pairing between circular MlSwt,

ssI)NA

and linear

(3H1M13mp7

dsI)NA

was

RecA Protein-catalyzed then assayed by filter binding in the presence of superhelical pBR322 DNA. Little recovery (0 to 20%) from the inhibition of joint formation at the higher RecA protein concentrations was observed (data not shown). 4. Discussion In this study, three different experimental assays have been used to follow the early stages of RecA protein-promoted homologous pairing and strand transfer and its dependence on DNA topology and homologous or non-homologous dsDNA. EM was used to directly visualize the interactions of RecA protein filaments and dsDNA, a sedimentation assay was employed to follow the formation of large protein-DNA aggregates, and a nitrocellulose filter binding assay was used to monitor the formation of joint complexes. There were three major findings in this study. First, large aggregates formed between presynaptic filaments and homologous dsDNA could be visualized by EM. Their formation required the dsDNA to be circular (superhelical or nicked). DNA pairing occurred efficiently with linear dsDNA substrates in the absence of coaggregates showing that they are not absolutely required for etlicient pairing. Second, pairing between circular M13wt ssDNA and linear M13mp7 dsDNA was found to be strongly inhibited by the addition of heterologous dsDNA but only if the heterologous dsDNA was superhelical. The addition of up to a loo-fold excess of linear heterologous dsDNA produced no significant inhibition of joining between the homologous DNAs. Finally, a direct interaction between presynaptic filaments containing ssDNA and heterologous dsDNA was visualized by EM, but only if the dsDNA was superhelical. Each of the three assays utilized has limitations, but combined they provide valuable and overlapping information. To capture the more labile associations of presynaptic filaments and homologous or heterologous dsDNA, and to stabilize the large coaggregates so that they could be mounted for EM, we found it necessary to use a higher concentration of glutaraldehyde (2.6%) than the 0.6% routinely employed. However, in previous studies, it was demonstrated that 2.6% glutaraldehyde was needed to preserve labile interactions between RecA protein and dsDNA (Shaner et al., 1987; Register & Griffith, 1988), which had been detected by ot)her investigators utilizing different assays (Shibata et al., 1981, 1982). The concern that this high concentration of glutaraldehyde might create some of the aggregates observed by EM was offset by the fact that fixation with 2.6% glutaraldehyde did not produce aggregates of presynaptic filaments and dsDNA when the dsDNA was linear (Fig. 2). Additionally, the same amount of dsDNA was present in the large aggregates (as seen by EM) as the fraction of labeled dsDNA that sedimented (in the coaggregate assay) in which fixation was not employed (Fig. 3). Using the filter binding assay, we

417

Pairing

examined the effects of heterologous dsDNA on joint formation only to a lOO-fold excess of heterologous dsDNA over homologous dsDNA. In the future, it would be valuable to extend this range but this would require optimizing the pairing reactions at greatly reduced levels of the homologous DNA substrates. (a) Goaggregate formation The conditions necessary to observe coaggregate formation in RecA protein-catalyzed pairing reactions have been characterized, in particular the requirements for low salt and a large excess of RecA protein (Tsang et al., 1985). However, the dependence of coaggregate formation on DNA topology had not been investigated. Here we have used conditions that are optimal for RecA proteinmediated strand exchange with a variety of different DNA substrates: saturating but not excess RecA protein, and 30 mM-NaCl (Register et al., 1987). Using these conditions, coaggregates were observed as detected by the sedimentation assay and EM. In this study, coaggregates were observed only when the dsDNA was either superhelical or nicked-circular. Coaggregates were not detected with linear dsDNA even though this DNA formed joints to an equivalent extent to that shown by circular dsDNAs. Thus, under conditions optimal for strand exchange in vitro, coaggregates are not absolutely required for an efficient search for homology and joint formation. In previous studies, we observed that when circular ss- and dsDNAs were paired, that the dsDNA became intertwined about the ssDNA (Christiansen & Griffith, 1986; Bortner & Griffith, 1990), possibly due to the inability of the dsDNA to undergo free rotation during pairing (Bortner & Griffith, 1990). This entanglement may trap the presynaptic filaments and dsDNA molecules producing very large aggregates. These EM studies provide the first direct visualization of coaggregates. The dsDNA molecules appeared incorporated into aggregates formed from hundreds of presynaptic filaments. The number of dsDNA molecules and presynaptic filaments present was difficult to quantitate due to the density and size of the coaggregates but sedimentation data indicated that approximately 30% of the dsDNA was entrapped. In the coaggregates formed between homologous DNAs, the dsDNA appeared unwound. This is to be expected since the formation of paranemic (Christiansen & Griffith, 1986) or plectonemic joints by supercoiled DNA will lead to the relaxation of the dsDNA. We also observed that the general appearance of coaggregates differed depending on the nature of the dsDNA molecules involved (compare Figs 7 and 8). When superhelical pBR322 DNA was mixed with circular Ml3wt ssDNA and RecA protein, the resulting aggregates were less dense and more loosely associated than those involving homologous DNA molecules. Tt was not

418

J. M. Pinsince

apparent by EM that the superhelical pBR322 DNA incorporated into these coaggregates was unwound; however, up to 30% of the supercoils could have been lost without greatly altering its coiled appearance. In these studies and those of Radding and coworkers (Tsang et aE., 1985; Gonda & Radding, 1986), it has been observed that coaggregates appear early in pairing reactions. We have shown that the fraction of DNA in the coaggregates rapidly diminishes with time, even when the pairing was unable to proceed past the stage of paranemic joining. Why coaggregates would resolve if the DNAs are irreversibly intertwined is unclear. (b) Dependence of paranemic dsDNA

joint formation topology

on

Paranemic joints should be able to form between two DNAs sharing 400 to 500 bp of homology, such as the pairing of Ml3mp7 ssDNA with plJC7 DNA. It was surprising, therefore, that the filter binding assay detected paranemic joints between these DNA substrates only when the dsDNA was superhelical. Joining was not detected when the dsDNA was either relaxed circular, or linear with as little as 52 bp of heterologous DNA sequences at one end (Fig. 5). This observation suggests that there is a hierarchy of stability for paranemic joints dependent on the topology of the dsDNA. The dependence of pairing on dsDNA topology cannot be explained simply by the differential ability of RecA protein binding to these dsDNA molecules. Under the reaction conditions used in this study, RecA protein bound less than 5% of the E3H]dsDNA in the reaction (data not shown). By EM, no significant binding of RecA protein to dsDNA was observed. In our original study of paranemic joints (Christiansen & Griffith, 1986): we showed that paranemic joints form efficiently and are relatively superhelical dsDNA is used. stable when Furthermore. as a result of joining, the dsDNA relaxed due to a net appears topologically unwinding of the DNA within the region of the presynaptic filament that contains the dsDNA and ssDNA. The conclusion of that and other studies (Wu et al., 1983; Schutte & Cox, 1988) was that the length of the paranemic joint is determined by the point at which further growth would begin to induce positive supercoiling in the dsDNA. The observations here further support this model, suggesting that the pairing of linear dsDNA is less stable than negatively supercoiled DNA, and pairing with topologically relaxed dsDNA, which induces positive supercoiling, would likely provide the least stable interaction. (c) Detection

of early, non-homologous during pairing

contacts

The earliest events in homologous pairing must involve interactions between presynaptic filaments and heterologous dsDNA. Here, using EM and an aggressive fixation protocol, we have been able to

and J. D. Gr
visualize directly an interaction between heterologous DNA molecules (Fig. 8). This finding was supported by the detection of heterologous contacts using filter binding and coaggregate assays. These associations were observed only when the heterologous dsDNA was superhelical, and resulted in relatively loose aggregates of presynaptic filaments and dsDNA. The nature of the association between the DNA molecules is unclear. Tt has been demonstrated that extensive unwinding of the dsDNA in joints is homology-dependent (Schutte & Cox, 1988). Our observations argue that there must be a limited unwinding of the heterologous dsDNA, independent of homology, otherwise it is difficult to explain why these heterologous associations were not observed with non-superhelical dsDNA. (d) The effect of non-homologous

DNA

on the .Tearch

for hom,ology

In the cell, the search for homology must proceed in the presence of a large excess of non-homologous DNA sequences, and indeed, Honigberg et al. (1986) have shown in vitro that RecA protein is able to pair homologous DNA molecules in the presence of a, IOOO-fold excess of linear heterologous dsDNA. Here, in agreement with that study. we observed that pairing was relatively unaffected by non-covalently attached heterologous linear dsDNA at concentrations IOO-fold over the concentration of ssDNA. However, if the dsDNA was superhelical. a strong inhibition (950/,) of homologous pairing was observed using a tenfold excess of superhelical heterologous dsDNA. An examination of these mixtures by EM revealed that the inhibition was the result of the sequestering of presynaptic filaments into aggregates. The aggregates were similar in appearance to those observed when only homologous superhelical dsDNA and presynaptic filaments were present (compare Figs 2 and 7). The results of our EM observations and sedimentation assays indicated that the linear homologous dsDNA was actually excluded from the aggregates, thus accounting for the inhibition of joint. formation. These observations are in agreement with those of Julin et al. (1986) who observed that the addition of heterologous linear bacteriopha,ge T7 DNA to a homologous DNA pairing reaction did not, inhibit, joint formation. On the other hand. Gonda & Radding (1986) reported an inhibition of joining by heterologous linear dsDNA with a, concomitant exclusion of the homologous dsDNA from the coaggregates. However, in their study, coaggregatex were present prior to the addition of the heterologous DNA. Since the formation of coaggregates will likely depend not only on the dsDNA topology but on reaction conditions, these result,s are not in serious disagreement but rather construct a picture in which coaggregate formation when it occurs. can have a major influence on the pairing process. In the E. coli cell, the chromosome is considered to be under superhelical strain with the unconstrained tension being less than that’ of purified

RecA Protein-catalyzed

plasmid DNA (Jaworski et al., 1991; Zheng et al., 1991). To extrapolate the results of this study to pairing in vivo, it will be important to examine in more detail the dependence of paranemic pairing, coaggregate formation and other processes described here as a function of the superhelical strain of the homologous and heterologous DNA molecules. This work was supported by a grant from the National Institutes of Health (GM-31819) and a’ grant from the American Cancer Society (NP 583). .I’

References Bianchi. M., DasGupta, C. & Radding, C. M. (1983). Synapsis and the formation of paranemic joints by E. coli recA protein. Cell, 34, 931-939. Bortner, C. & Griffith, J. (1990). Three-stranded paranemic joints: architecture, topological constraints and movement. J. Mol. Biol. 215, 623-634. Chase. J. W., Whittier, R. F., Auerbach, J., Sancar, A. & Rupp, W. D. (1980). Amplification of single-strand binding protein of Escherichia coli. Nucl. Acids Res. 8, 3215-3227. Christiansen, G. & Griffith, J. (1986). Visualization of the paranemic joining of homologous DNA molecules coli. catalyzed by the recA protein of Escherichia Proc. Nat.

Acad. Sci., U.S.A.

83, 2066-2070.

Cox. M. M. $ Lehman. I. R. (1982). RecA proteinpromoted DNA strand exchange. Stable complexes of recA protein and single-stranded DNA formed in the presence of ATP and single-stranded DNA binding protein. J. Biol. Chem. 257, 8523-8532. Cox, M. & Lehman, I. R. (1987). Enzymes of general recombination. Annu. Rev. Biochem. 56, 229-262. Eggleston, A. K. & Kowalczykowski, S. C. (1991). An overview of homologous pairing and DNA strand exchange proteins. Biochimie, 73, 163-176. Gonda. D. K. & Radding, C. M. (1986). The mechanism of the search for homology promoted by recA protein. J. Biol. Chem. 261, 13087-13096. Griffith, ,J. D. & Christiansen, G. (1978). Electron microscopic visualization of chromatin and other DNAprotein complexes. Annu. Rev. Biophys. Bioeng. 7, 19-m35. Griffith, J. D. & Harris, L. D. (1988). DNA strand exchanges. Grit. Rev. Biochem. 23, S43-886 (suppl. 1). Griffith, J. D. & Shores, C. G. (1985). RecA protein rapidly crystallizes in the presence of spermidine: a valuable step in its purification and physical characterization. Biochemistry, 24, 158-162. Griffith, J. D.. Harris, L. D. & Register, J. (1984). Visualization of SSB-ssDNA complexes active in the assembly of stable RecA-DNA filaments. Cold Spring Ha,rbor Symp.

Quant.

Biol.

49, 553-559.

Harris, L. D. & Griffith, J. D. (1988). Formation of D-loops by the UvsX protein of T4 bacteriophage: a comparison of the reaction catalyzed in the presence or absence of gene 32 protein. Biochemistry, 27. 6954-6959.

Honigberg, S. M., Rao. B. J. & Radding, C. M. (1986). Ability of recA protein to promote a search for rare sequences in duplex DNA. Proc. Nat. Acad. Sci., U.S.A.

83, 9586-9590.

Jaworski, A., Higgins, N’. P., Wells, R. D. & Zacharias, W. (1991). Topoisomerase mutants and physiological

Pairing

419

conditions control supercoiling and Z-DNA formation in vivo. J. Biol. Chem. 266, 2576-2581. Julin, D. A., Riddles, P. W. & Lehman, I. R. (1986). On the mechanism of pairing of single and doublestranded DNA molecules by the recA and singlestranded DNA-binding proteins of Escherichia co&. J. Biol. Chem. 261, 1025-1030. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982). Molecular Cloning: A Laboratory Manual, pp. 88-91, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. McEntee, K., Weinstock, G. M. & Lehman I. R. (1979). Initiation of general recombination catalyzed in vitro co&. Proc. Nat. by the recA protein of Escherichia Acad. Sci., U.S.A. 76, 2615-2620. Menetski, J. P. & Kowalczykowski, S. C. (1985). Interaction of recA protein with single-stranded DNA. Qualitative aspects of binding affinity modulation by nucleotide cofactors. J. Mol. Biol. 181, 281-295. Muller, B., Koller, Th. & Stasiak, A. (1990). Characterization of the DNA binding activity of stable RecA-DNA complexes. Interaction between the two DNA binding sites within RecA helical filaments. J. Mol. Biol. 212, 97-112. Radding, C. M. (1989). Helical recA nucleoprotein filaments mediate homologous pairing and strand exchange. Biochim. Biophys. Acta, 1008, 131-145. Register, J. C., III & Griffith, J. D. (1986). RecA protein filaments can juxtapose DNA ends: an activity that may reflect a function in DNA repair. Proc. Nat. Acad. Sk.,

L’.S.A.

83, 624-628.

Register, J. C., III & Griffith, J. (1988). Direct visualization of recA protein binding to and unwinding duplex DNA following the D-loop cycle. J. BioZ. Chem. 263, 11029-11032. Register, J. C.. III, Christiansen, G. & Griffith, J. (1987). Electron microscopic visualization of the recA protein-mediated pairing and branch migration phases of DNA strand exchange. .I. Biol. Chem. 262, 12812-12820. Riddles, P. W. & Lehman, I. R. (1985). The formation of paranemic and plectonemic joints between DNA molecules by the recA and single-stranded DNA-binding protein of Escherichia coli. J. Biol. Chem. 269, 165-169. Schutte, B. C. & Cox? M. M. (1988). Homology-dependent underwinding of duplex DNA in recA protein generated paranemic complexes. Biochemi.stry, 27, 7886-7894. Shaner, S. L., Flory, J. & Radding, C. M. (1987). The distribution of Escherichia coli recA protein bound to duplex DNA with single-stranded ends. J. Biol. Chem. 262, 9220-9230.

Shibata. T., DasGupta, C., Cunningham. R. P. & Radding, C. M. (1979). Purified Escherichia coli recA protein catalyzes homologous pairing of superhelical DEA and single-stranded fragments. Proc. Nat. Acad. Ski., U.S.A. 76, 1638-1642. Shibata, T.. DasGupta, C., Cunningham, R. P., Williams, J. G. K., Osber, L. & Radding, C. M. (1981). Homologous pairing in genetic recombination: the pairing reaction catalyzed by Escherichia coli recA protein. J. Biol. Chem. 256, 7565-7572. Shibata, T., Ohtani, T., Iwabuchi, M. & Ando, T. (1982). D-loop cycle. J. Biol. Chem. 257, 13981-13986. Takahashi, M., Kubista, M. & ru’orden. B. (1989). Binding stoichiometry of RecA-DNA and structure complexes studies by flow linear dichroism and

420

J. &I. Pinsince

fluorescence spectroscopy. J. Mol. Biol. 205, 137-147. Thresher, R. J., Christiansen, G. & Griffith, J. D. (1988). Assembly of presynaptic filaments. Factors affecting the assembly of recA protein onto single-stranded DNA. J. Mol. Biol. 201, 101-113. Tsang, S. S.. Chow, S. A. & Radding, C. M. (1985). Networks of DNA and recA protein are intermediates in homologous pairing. Biochemistry, 24, 3226-3232. Tsang, 6. S., Muniyappa, K.. Azhderian. E., Gonda. D. K.. Radding, C. M.. Flory. J. & Chase, J. W. (1986). Intermediates in homologous pairing promoted by recA protein: isolation and characterization of active presynaptic complexes. J. Mol. Rio!. 185. 295-309. Edited

and J. D. Grifith Umlauf, S. W.. Cox, M. $1. & Inman. R. B. ( 1990). Triple-helical DNA pairing intermediates formed b) recA protein. ,I. Biol. Chem. 265, 16898~-16912. Wu. A. M., Bianchi, M., DasGupta, C. & Radding. C. M. (1983). Unwinding associated with synapsis of DNA molecules by recA protein. Proc. Nat. Acad. Sci.. C’.S.A. 80, 1256-1260. Zheng. G. X., Kochel. T.. Hoepfner. R. IV.. Timmons. S. E. & Sinden. R. R. (1991). Torsionally t,uned and Z-DNA probes for measuring cruciform unrestrained supercoiling at specific sites in DNA of living rells. J. Mol. Biol. 211, 107-122.

by P. van Hippel