UvsY protein of bacteriophage T4 is an accessory protein for in vitro catalysis of strand exchange

J. Mol. Hiol. (1989) 206, 19-27

UvsY Protein of Bacteriophage T4 is an Accessory Protein for in Vitro Catalysis of Strand Exchange Lorelei D. Harris and Jack D. Griffith+ Lineberger Cancer Research Center, Curriculum in Genetics and Department of Microbiology and Immunology University of North Carolina Chapel Hill, NC 27514, U.S.A. (Received 13 July 1988, and in revised form

13 September 1988)

The UVSX and uvs Y genes are essential to genetic recombination, recombination-dependent DNA synthesis and to the repair of DNA damage in bacteriophage T4. Purified UvsX protein has been shown to catalyze strand exchange and D-loop formation in vitro, but the role of UvsY protein has been unclear. We report that UvsY protein enhances strand exchange by UvsX protein by interacting specifically with UvsX protein: gene 32 protein (gp32) is not necessary for this effect and UvsY protein has no similar effect on the RecA protein of E. co2i. UvsY protein, like UvsX protein, protects single-stranded DNA from digestion by nucleases, but, unlike UvsX protein, shows no ability to protect doublestranded DNA. UvsY protein enhances the rate of single-stranded-DNA-dependent ATP hydrolysis by UvsX protein, particularly in the presence of gp32 or high concentrations of salt, factors that otherwise reduce the ATPase activity of UvsX protein. The enhancement of ATP hydrolysis by UvsY protein is shown to result from the ability of UvsY protein to increase the affinity of UvsX protein for single-stranded DNA.

1. Introduction

resolved to genome-length units (Huberman, 1969; Kozinski, 1968; Broker $ Lehman, 1971; Broker, 1973; Broker & Doermann, 1975; Kozinski & Kosturko, 1976; Dannenberg & Mosig, 1983). In cells infected with uvsX or UVSY mutant phage, no T4 DNA concatamers are observed, and T4 DNA synthesis arrests at about the time when concatamer formation would normally begin (Dewey & Frankel, 197&Q; Wakem & Ebisuzaki, 1976; Cunningham & Berger, 1977; Melamede & Wallace, 1977, 1986u,b). Luder & Mosig (1982) provided the link between these defects in recombination and replication, when they showed that initiation of DNA replication in T4 occurs through at least two pathways. Primary replication occurs immediately after infection and initiates from a few defined replication origins. Secondary replication, however, is primed by the invasion of a 3’ end of a newly replicated DNA into a homologous region on another DNA, forming a D-loop, which is then extended by the DNA polymerase complex. Dannenberg & Mosig (1983) diagram how such repeated rounds of strand invasion, branch migration, and replication fork movement would yield concatameric replicative intermediates. These models place the uvsX and uvsY genes in positions of crucial importance for recombination, repair of DNA damage and replication.

The UVSX gene is the center of an intricate combination of genes that control replication and genetic recombination in bacteriophage T4. Phage having mutations in UVSX were originally isolated by Harm (1963) on the basis of their increased sensitivity to ultraviolet light. He went on to show that uvsX mutants operated a different repair pathway from that of the previously characterized denV gene, and that they did not undergo genetic recombination (Harm, 1963, 1964). Soon after, Boyle & Symmonds (1969) isolated phage with a similar mutation in a gene now called uvs Y. Additional mutants were found to be related, either on the basis of defects in recombination or because they exhibited repair deficiencies unrelated to the denV pathway (for a review, see Bernstein & Wallace, 1983). Repeatedly, the conclusion of separate studies has been that the uvsX and uvsY genes are epistatic. Not only are they essential to normal recombination and repair of DNA damage, but UVSX and uvsY genes are required for late DNA replication. Normally, T4 produces long concatamers of DNA that are subsequently t Author to whom correspondence should be sent. 19 0022-28:16/89/050019-09

$03.00/O

0

1989 Academic

Press Limited

20

L. II. Harris

Recently, several groups have purified and studied the UvsX protein. It, forms helical nucleoprotein filaments with both single-stranded (SST) and double-stranded (ds) DNA, catalyzes ssDNAdependent ATP hydrolysis, strand exchange and D-loop formation; its behaviour in this regard is very similar to the RecA protein of Escherichia coli (Yonesaki & Minagawa, 1985; Yonesaki et al., 1985; Griffith & Formosa, 1985; Formosa & Alberts, 1986a; Hinton & Nossal, 1986; Harris & Griffith, 1987). [JvsX protein catalyzes all of these reactions in vitro, without the addition of UvsY protein. Purified UvsY protein does not hydrolyze ATP in t,he presence or absence of ssDNA. and. in previous studies, has shown no effect on the strand exchange activity of UvsX protein (Yonesaki et al.. 1985; Formosa &, Alberts, 1986a). More recent, studies, however, indicate t,hat UvsY protein acts as an accessory protein in joint formation (Minagawa et al., 1988; T. Kodadek & R. M. Alberts, personal c*ommunication). UvsY protein allowed very low concentrations of UvsX protein to catalyze strand exchange, while decreasing the salt, sensitivit,y of the reaction, yet it showed no eflect on strand exchange catalyzed by RecA protein (Kodadek B Alberts, personal communication). In previous work, we studied the abilit,y of c:vsX prot,ein to catalyze D-loop formation and &and exchange, both in the absence and in the presence of gene 3;! protein (gp32). a ssl)l\;A-binding protein that enhances the efficiency of these reactions (Harris 8r Griffith. 1987, 1988). In these studies, we have used techniques for monitoring simple strand exchange reactions with parallel assays of electron microscopy, electrophoresis, nitrocellulose filterof ATP hydrolysis. In binding and measurements this paper. we apply these same techniques to the st#udy of DNA strand exchange reactions carried out m the presenctb of UvsX. I:vsY and gp32. (:onsistent with the predictions from genetic studies, we find that IJvsY protein interacts specifically with I:vsX proeein t,o enhance st#rattd exchange catalyzed by CvsX protein in vitro. This work adds to a growing body of evidence that recaombination in T4 bacteriophage is catalyzed hi a large complex of multiply interacting proteins.

2. Materials and Methods (a) Proteins,

DIVAS and huffm

SSB. RecA, UvsX, Uvs‘L’ and gp32 proteins were prepared as described (Chase et al., 1980: Formosa & Alberts, 1986a; Yonesaki et al., 1985; Griffith & Shores, 1985; Bittner et aZ., 1979). Ml3 ssDNA and dsDNA and pBR322 DNAs were prepared as described (Modrich & Zabel, 1976), including linearization of the DNA with BumHI enzyme. dsDNA was radiolabeled with [32P]dCTP (Amersham), using T4 DNA polymerase (IBI) first to excise nucleotides and then to synthesize a f Abbreviations used: ssDNA, single-stranded DNA; dsDNA, double-stranded DNA; gp32, gene 32 protein; SSB protein, single-stranded DNA-binding protein.

and J. D. Qrifith

replacement) st)rand, as described by the vendor. All strand exchange, DNA binding, ATP hydrolysis and nuclease protection experiments were conducted in X,, buffer, (20 mM-Hepes (pH 7.5), 5 mM-ATP. 7 rnM magnesium acetate, 75 mw-potassium acetate, 20 mMphosphocreatine. 4 pg creatine phosphokinase/ml). (b) Strand mchange Circular Ml 3mp7 ssDNA and linear M13mp7 dsDNA were incubated at, 10 PM each (nucleotide concentration) in X,, buffer at 37°C for 15 min. Enzyme concentrations are given in the appropriate Figure legends. Strand exchange was stopped by adding a gel loading buffer that included SDS to give a final concentration of 1 y/o (w/v) SDS. Products were electrophoresed on a 0.7% agarose gel, after which the gel was dried on to Whatman 3MM paper and placed against Kodak X-ray tilm for autoradiography.

(c) Nuclease

protection

Proteins and DNAs were incubated together for 10 min at 37°C m X,, buffer at the concentrations listed in the Figure legends. DNase I and snake venom phosphodiesterase (Sigma) were added to 140 and 40 pg/ml, respectively, and incubation was continued for 3 min. Salmon sperm DNA was added to 200 pg/ml as carrier. along with 2 vol. ice-cold 10 “/b (v/v) trichloroacetic acid (TCA). The mixture was left on ice for 30 min, then was filtered through Whatman GF/C filters. The filters were rinsed 3 times with 1 ml cold 100/o TCA and once with 1 ml cold 9O”/b ethanol. then were dried under a heat’ lamp, placed in scintillation fluid, and processed in a scintillation counter to measure t)hr radactivr counts retained. The spectrophotometric assay for ATP hydrolysis used here was from Panuska & Goldthwait (1980) and Kreuzer & Jongoneel (1983). Preparation of samples for electron microscopy by surface spreading has been described (Chow C Broker. 1981; Harris & Griffith, 1987).

3. Results (a) ITvsY

protein

enhances strand protein

exchange by l!vsX

In studies of CJvsX or RecA proteins, the DNA substrates most commonly used t)o follow the progress of DNA strand transfer are a linear dsDNA and a homologous, circular ssDEA (for a review, see Griffith & Harris, 1988). A diagram of this reaction is shown in Figure 1. Synapsis and strand invasion align the (- ) strand of the dsDNA with the complement’ary (+) strand of the ssDNA. Branch migration completes the transfer of t,he (-) strand to the circular (+) strand, while displacing the linear (+) strand of the dsDNA. When this reaction is catalyzed by UvsX protein, however, a second round of strand invasion often begins before strand transfer is complete, so that the product is a branched network of multiply recombinant DNA molecules (Formosa & Alberts, 1986a; Kodadek et al., 1988). Small networks will penetrate an agarose gel, but larger, more complex networks cannot.

21

Uvs Y Protein is an Accessory Protein

At the end of the reaction, the DNA products were analyzed by agarose gel electrophoresis. Quantification

of the amount

synapsis 1 5’

Alberts

(+I

3’ t-1

(personal

products

communication),

we found

that

substrates

(0.7 VP~SUS1.5 PM-UvsX

protein).

We have

shown previously that, in the absence of any accessory proteins, UvsX protein must be added to a concentration of one UvsX monomer per three nucleotides to this reaction (Harris $ Griffith. 19X7, 1988). Here, in the presence of UvsY protein a,nd gp32, as little as one lJvsX monomer per 14 nucaleotides will catalyze strand exchange. and one monomer per seven nucleotides yields chse to

Figure 1. Schematic representation of a simple strand exchange. When a linear dsDNA is paired with a homologous, circular ssDEA, the (-) strand of the dsDNA pairs with the (+) strand of the ssDNA, displacing the (+) strand of the dsDNA, to yield nicked. circular dsDNA and linear ssDNA as products.

these DKA

in each of t#he

UvsX protein could catalyze strand exchange in the absence of CvsY protein, but that the addition of UvsY protein allowed for efficient strand exchange at lower concentrations of IJvsX protein (Figs 2 and 3). For example, in the presence of 1.5 PM-UvsY protein, half as much UvsX protein was necessary to convert 50°’ !. of the dsDNA to strand exchange

cl+=

In this study,

of [32P]dsDNA

products was accomplished with a scanning densitometer. In agreement with previous work by Kodadek &

lOO”;, recombinant from also clear containing ~‘vsY

were used in

molecxules in 15 minutes. It is Figure 2 that the reactions protein yielded larger networks

reaction mixtures containing UvsX protein, gp32, dsDNA and homologous linear [ 32P]M l3mpT M13mp7 ssDNA circles in X,, buffer, either with or

than the reactions with only lTvsX protein. For t,he quantification in Figure.3, any product that migrated slower than the linear dsl)NA was

without

considered

UvsT

protein

(see Materials

I

2

3

and Methods).

4

5

6

7

8

t’o be a recombinant,

9

IO

II

12

linear dsDNA

Figure 2. I!vsY protein increases strand exchange catalyzed by UvsX protein. RF111 at 10 PM each were incubated in X,, buffer with 1.3 PM-gp32, and increasing

M13mp7 ssDNA and [3H]M13mp7

concentrations of UvsX and UvsY proteins at 37°C for 15 min. The reaction was stopped with SDS and gel loading buffer was electrophoresed through an agarose gel and dried for autoradiography, as described in Materials and Methods. Lanes 1 to 3: 0 UvsX protein with 0, 0.75 and 1.5 PM-UvsY protein, respectively. Lanes 4 to 6: 0.7 PM-UvsX protein with 0, 0.75 and 1.5 PM-UvsY protein. respectively. Lanes 7 to 9: 1.4p~-UvsX protein with 0, 0.75 and 1.5 PM-IJvsY protein, respectively. Lanes 10 to 12: 2.1 ~M-I!vsX protein with 0, 0.75 and 1.5 PM-UvsY protein, respectively.

L. D. Hawk

22

and

J. 1,. (kifith

,-80 -

80 -

A

4 RecA (pd

I

2

I

0

uvsx (ptd

Figure 3. UvsY protein increases strand exchange catalyzed by UvsX protein. The autoradiogram in Fig. 2 was analyzed with a scanning densitometer to quantify the amount of strand exchange catalyzed by UvsX and

IJvsY proteins. UvsX protein concentrations are given in the graph. UvsY protein concentrations were 0 (O), 0.75 PM (+) and 1.5 PM (m).

We considered it possible that this enhancement of strand exchange could be the result of UvsY protein interacting with gp32, since UvsY protein binds to both IJvsX protein and gp32 on affinity columns (Formosa et al., 1983; Formosa & Alberts, 1984). To test this, we conducted the same reaction in the absence of gp32 (Fig. 4). Again, IJvsY protein allowed for catalysis of strand exchange at lower concentrations of UvsX protein, although more UvsX protein was necessary than in the reactions that included gp32. This implies that UvsY protein interacts specifically with UvsX protein in this reaction. Further evidence for this is offered by the fact that LJvsY protein had no effect on strand exchange catalyzed by RecA protein of

80 2 s jj

60

E $j 40 ,\” 20 i

Figure 4. gp32 is not necessary for UvsY protein to increase strand exchange catalyzed by UvsX protein. M13mp7 ssDNA and [‘H]M13mp7 RF111 were incubated at 10~~ each in X,, buffer at 37°C for 15 min with UvsX protein at the concentrations given. No gp32 was used in these reactions. Detection of the reaction products was accomplished as described in the legends for Figs 2 and 3, and in Materials and Methods. UsvY protein concentrations were 0 (a), 0.75 PM (+) and 1.5 PM (m)

Figure 5. UvsY protein has no effect on strand exchange catalyzed by RecA protein. Strand exchange was conducted in X,, buffer at 37°C for 15 min, using 10~~ each of M13mp7 ssDNA and [3H]M13mp7 RFITT with 0.5 PM-SSB protein and RecA protein in the concentrations shown. UvsY protein was added to OHM (a), 0.75pM (+) and 1.5pM (m). Detection of the reaction products was accomplished as described in the legends for Figs 2 and 3, and in Materials and Methods. Similar results were obtained when 0.9 PM-gp32 was substituted for SSB protein (data not shown).

E. coli, whether the reaction was supplemented with gp32 or with SSB protein (Fig. 5).

(b) Uvs Y protein

protects ssDNA, from nucleases

but not dsDNA,

The in vitro strand exchange reactions described above involve both ssDNA and dsDNA. IJvsX protein will bind to either, although it shows a preference for ssDNA (Yonesaki & Minagawa, 1985; Griffith & Formosa, 1985; Harris & Griffith, 1987). To examine the role of UvsY protein in these reactions, we studied the ability of UvsX and IJvsY proteins to protect DNA from digestion by nucleases. UvsX and UvsY proteins were incubated in increasing concentrations with either dsDNA or ssDNA in X,, buffer (see Materials and Methods). After a ten minute incubation, DNase I and snake venom phosphodiesterase were added to the reaction mixture for three minutes, followed by cold trichloroacetic acid to precipitate acid-insoluble DNA. Figure 6 shows that both UvsX and UvsY proteins protected ssDNA from nuclease digestion, but only UvsX protein protected dsDNA. ssDNA was completely protected by UvsX protein at a concentration of three nucleotides per protein monomer, which is in agreement with previous stoichiometric studies (Griffith & Formosa, 1985; Harris & Griffith, 1988). Similarly, ssDNA is completely protected from nuclease digestion when UvsY protein is present at three to four nucleotides per protein monomer. It seemed possible that UvsY protein might not bind to dsDNA directly, but would bind to dsDNA coated with IJvsX protein, since it binds to UvsX

23

Uvs Y Protein is an Accessory Protein

0

I

0

2

4

3

5

6

UVSX or UvsY (pm)

Figure 6. UvsX protein protects both ssDNA and dsDNA

from

nucleasr

digestion,

but

UvsY

UvsX or UvsY protein, as described in Materials and Methods. (0) 5 nM-]3H]M13mp7 ssDNA with a IJvsX at the concentrations shown; (+) 6 PMprotein [3H]M13mp7 ssDNA with UvsY protein at the concentrations shown; (m) 10 pM-[3H]M13mp7 dsDNA with UvsX protein at the concentrations shown; (0) 10 pM-[‘H]Ml3mp7 dsDNA with UvsY protein at the concentrations shown.

protein immobilized on a column matrix (Formosa al., 1983). To examine this, both proteins were incubated together with ssDNA and dsDNA, and the nuclease resistance was monitored. On ssDNA, the level of protection with the two proteins was but UvsY protein did not increase additive, protection of dsDNA by UvsX protein (Fig. 7). These results indicate that the site of interaction et

T

3 UVSY(/AM)

Figure 7. UvsX and IJvsY proteins both protect. ssDNA when added together, but UvsY protein does not increase UvsX protein’s protection of dsDNA. Nuclease protection experiments were conducted on either ds or ss DNA as described in Materials and Methods. (+) 10 PM[3H]M13mp7 ssDNA with UvsY protein at the concentrations shown; (0) same, but with 1.4 PM-UvsX protein added; (u) lOp~[~H]M13rnp7 dsDNA with UvsY protein at the concentrations shown; (0) same, but with 1-4 FM-CvsX protein added.

I

I

0-I

I

I

0.2

I

I

0.3

t

I

0.4

I

I

0.5

I

/

( .6

Number of UvsY monomers per UvsX monomer

protein

protects only ssl)Pu’A. Nuclease protection experiments were carried out with either ds or ss DNA and with either

0

T 0

Figure 8. UviY

protein increases UvsX protein’s ssDNA-dependent, but not its dsDNA-dependent, ATP hydrolysis activity. ATP hydrolysis experiments were

conducted with either ss or ds DNA in X,, buffer at 37°C as described in Materials and Methods. (+) 1 PM-Ml3mp7 ssDNA with 0.2 FM-UvsX and UvsY proteins added to give the GvsY : UvsX ratios shown; (0) 1 phr-M13mp7 ssDNA with 0.4 PM-UvsX and UvsY proteins added to give the UvsY : UvsX protein ratios shown; (m) 1 PMM13mp7 dsDNA with 0.2 PM-UvsX and UvsY proteins added to give the UvsY : UvsX protein ratios shown.

between UvsX and UvsY -proteins is likelv to be on ssDNA, and that UvsY protein does not bind UvsX protein-covered dsDNA, or that it does so in a way that provides no further nuclease protection.

(c) livs

increases the ssDNA -dependent ATPase rate of UvsX protein

Y protein

UvsX protein hydrolyzes ATP when bound to ssDNA (Yonesaki & Minagawa, 1985; Hinton & Nossal, 1986; Formosa & Alberts, 1986u; Harris & Griffith, 1988). A specific interaction between UvsX and UvsY proteins on ssDNA, therefore, might be expected to affect the rate of ATP hydrolysis by UvsX protein. When UvsY protein was added to UvsX protein and ssDNA in X,, buffer, the rate of hydrolysis by UvsX protein increased substantially, from 120 to 210 ATP molecules hydrolyzed per IJvsX monomer per minute (Fig. 8). UvsY protein exhibited no ATP hydrolysis activity on its own (data not shown). The maximum increase in ATP hydrolysis occurred at a ratio of between O-3 and O-4 UvsY protein monomers per one UvsX protein monomer (or at about 3 : 1, UvsX : IJvsY). UvsY protein showed no effect, on the rate of ATP hydrolysis by UvsX protein in the presence of dsDNA, although the basic rate is much lower; less than 15 molecules of ATP were hydrolyzed per UvsX protein monomer per minute, whether or not UvsY protein was present (Fig. 8). This difference in the rate of hydrolysis observed in the presence of UvsY protein was especially pronounced under conditions that. normally limit

“4

I,. I). Harris

and J. Lj. Gri;tfith

continued (Fig. 10).

hgdrolysis

at higher

salt c:oncrnt.rat,it,ns

(d) ITvs )I protein increases the affinity protein for SSLINA

0

I

400 Gene 32 protein

(nM)

Figure 9. UvsY protein prevents gp32 from lowering the rate of ATP hydrolysis by UvsX protein. ATP hydrolysis experiments were conducted in X,, buffer at 37°C as described in Materials and Methods, using 1 PMMl 3mp7 ssDNA and 0.14 PM-UvsX protein. with gp32 at the concentrations shown. (0) No UvsY protein; (+) 0.04 PM-I’vsY protein added.

ATT’ hydrolysis by UvsX protein. For example, gp32 reduces the ssDNA-dependent ATP hydrolysis of UvsX protein, presumably by competing for ssDNA-binding sites (Formosa & Albert+ 1986a; Harris & Griffith, 1988). When we included gp32 in reactions with both UvsX and UvsY proteins, however, no inhibition of ATP hydrolysis was seen (Fig. 9). In fact, gp32 allowed for a slight, but repeatable, increase in ATP hydrolysis. Similarly, although high concentrations (> 100 mM) of potassium acetate inhibit ATP hydrolysis by UvsX protein, the addition of UvsY protein allowed for

300

200

100 Potassium

acetate

The results described above could be interpreted to demonstrate that: (1) UvsY protein interacts with UvsX protein to increase its intrinsic rate of ATP hydrolysis; (2) UvsY protein causes a more rapid release of UvsX protein from ssDNA, concurrent with a higher rate of ATP hydrolysis; or (3) UvsY protein increases the affinity of ITvsX protein for ssDNA, allowing it t,o hydrolyze more ATP. These options were compared by examining the salt stability of DNA-protein filaments by electron microscopy. IJvsX protein was incubated with ssDNA and dsDNA with or without UvsY protein in X,, but with varying concentrations of buffer, potassium acetat’e. The protein-DNA complexes were fixed with glutaraldehyde, spread on a solution of formamide, and mounted for electron microscopy (see Materials and Methods). The extent of protein coverage was determined by measuring the length of protein-free DNA in elrrtron micrographs. Strand exchange by UvsX protein is conducted optimally in the presence of 75 to 100 mw potassium acetate (Formosa & Alberts, 1986~~; Harris & Griffith, 1987). Under these conditions, ssDNA appeared fully covered with protein, whether or not UvsY protein was present in the reaction. When the concentration of potassium acetate was increased to 200 mM, however, much of the UvsX protein was seen to dissociate from the ssDNA, and the DNA was entirely protein-free at 250 mm-potassium acetate (Table 1). In t)hr presence of I’vsY protein. however, much of the DNA remained at least partially protein-bound at these higher salt, concentrations (Table I). The in protein binding under high salt, decrease conditions paralleled the decrease in ATP hydrolysis described above. It might, be argued t,hat the DNA-protein complexes that are stable in 250 mMpotassium acetate contain only ITvsY protein, but t’he existence of a residual level of ATP hydrolysis demonstrates that both I’vsX and IJvsY proteins are present In agreement with the DNase protection experiments, the presence or absence of CvsY protein had no effect on the amount of dsDNA that was bound by LJrsX protein at varying potassium acetate concentrations (Table 1).

(mfd)

Figure 10. UvsY protein lessens the effect of potassium acetate on the rate of ATP hydrolysis by UvsX protein. ATP hydrolysis experiments were conducted in X,, buffer, trations,

of lT~vsX

but with varying potassium acetate concenat 37”C, as described in Materials and Methods, using 1 p(M-M13mp7 ssDPJA, 0.14 PM-CJvsX protein and either no UvsY protein (0) or 0.04 PM-UvsY protein (+). Potassium acetate was added at the concentrations given.

4. Discussion UvsY protein is shown to act as an accessory protein in in vitro strand exchange reactions. Its role is specific to UvsX protein: it shows no measurable effect on strand exchange catalyzed by RecA protein and gp32 is not required for it to increase strand exchange. Several lines of evidence

UvsY Protein is an Accessory Protein

UwY

Table 1 protein lessens the displacement of [JvsX protein from ssDNA, but not from dsDNA, caused by high potassium acetate concentrations

] Potassium acetate] (mM) A.

I:13x 7.5

n/o protein-free DNAs

0 3 38

22.5 250

100

13. I:anX and I:vsY on ssDNA 75 0 200

5

225

8

250

9

o,, protein-covered DNAs

Total OiODNA covered

100

loo

ii6 (53) 62 (32)

1 0

,52 20

0

0

0

0 91 (61)

100

4

60

92 (40) 91 (39)

0 0

37 35

100

100

0 0

61 0

loo 0

100 53

0

0

loo

on dsDNA

7s 150

0 0

200

100

I). l:v.sX and l:va Y on dsDNA 75 0 150 200

“/o partially-covered DNAs (avg. “/u of DNA) covered)

on ssnNL4

200

C’. l:wX

25

0

100

0

loo (61) 0 0 100 (53)

0

Either M13mp7 ssDNA at 5 PM or dsDNA at 10 PM was incubated in X,, buffer with 1.5 PM-UvsX protein and with or without 0.5 PM-UvsY protein for 15 min at 37 “C. Sample preparation and electron microscopy were conducted as described in Materials and Methods. Samples were viewed in a Philips EM400 microscope, and the number of protein-free, partially protein-covered and protein-covered DNAs were counted. The amount of protein coverage on partially covered DNA molecules was determined by measuring the protein-free DNA in electron micrographs with a Summapraphics digitizer coupled to an IBM AT computer

indicate that UvsX and UvsY proteins interact on ssDNA. UvsY protein shows no protection of dsDNA from nuclease digestion, does not effect the dsDNA-dependent ATPase activity of UvsX protein, and does not prevent UvsX protein from dissociating from dsDNA in high salt concentrations. It does, however, protect ssDNA from nuclease digestion, both on its own and in the presence of UvsX protein. Unlike other ssDNAbinding proteins, however, it does not seem to displace UvsX protein from ssDNA. Even at high concentrations of UvsY protein, UvsX protein’s ATPase activity is enhanced, contrary to what one would expect if the two proteins were competing for ssDNA-binding sites. Challenge of UvsX protein’s ssDNA-binding activity with high salt concentrations or with gp32 shows, again, that UvsY protein increases UvsX protein’s af-linity for ssDNA. Minagawa et al. (1988) offer additional evidence that UvsY protein acts as an accessory to UvsX protein, through studies that show that the transformation of E. coli recA - cells with plasmidborne copies of the UVSX gene increased the survival of ultraviolet irradiated cells, but that introduction of the UWSY gene into these cells had no effect. However, when the cells were transformed with both the UVSX and uvusY genes, the survival rate was much higher than with the UVSX gene alone. The observations in this study suggest a model in

which UvsX protein has two binding modes for ssDNA, one more stable than the other, and that a transition between the two states can be induced by UvsY protein. Such a transition could conceivably be accomplished in several ways. UvsX protein might, in a treadmilling-type model. polymerize unidirectionally along ssDNA, forming a helical tract, which grows at the front end, while simultaneously depolymerizing at a slower rate from the rear (Formosa & Alberts, 19866; Griffith & Harris, 1988; Kodadek et al., 1988). One could imagine that UvsY protein could “cap” the back ends of UvsX protein tracts, preventing depolymerization. Most models of depolymerization, however, rely on ATP hydrolysis as the mechanism to release UvsX protein from the DNA. This view would be inconsistent with a “capping” model, which would require that UvsY protein would increase ATP hydrolysis and yet prevent UvsX prot,ein from leaving the ssDNA. In addition. our data indicate that ATP hydrolysis by UvsX protein is enhanced most at a 3 : 1 ratio of UvsX to UvsY. This suggests an alternative model, in which UvsY protein intermixes with, and is incorporated int,o the UvsX protein helix, with a defined number of UvsY protein molecules per helical turn. It should be noted that separate studies by Kodadek & Alberts (personal communication) found a 1 : 1 ratio of UvsX : UvsY was optimal for strand exchange.

L. I). IIarris

%A

and ,I. b. Uri$ith

Klectron microscopic visualizat,ion of CvsS and VvsY proteins on ssl)NA has not. revealed any striking differences in the appearance of nucleo-

dsUNA visually

filaments. which arc from the ssl)NAPprotein

protein filaments formed by UvsX protein alone. or with LJvsX and CvsY proteins together (Harris & Grifith, unpublished results). However. when one considers the size of IJvsY, this is not surprising. If

It becomes clear t’hat predict’ how proteins will placed in t)he context of catalyze. These studies, growing understanding of

the

systems

filaments

contain

three

I’vsX

monomers

(3 x44,000 M, = 132,000 AI,) for each I’vs\ monomer (16,000 M,) then the total molecular weight of the protein filament would only increase by 12%. Even if the full manifested on the diameter

effect of t’hat 12:\, was of the filament,, and not

spread somehow between a combination of changes in both filament’ length and diameter. the diameter would only increase from 14 nm to 14.X ntn (based on measurements by Griffith & Formosa (1985) and assuming that the 12%) size increase is distribut,ed across the area of a circle with a 7 nm radius). An understanding of the location of IJvsY protein within the nucleoprotein filament might well be accomplished

in future

electron

microscopic

studies

by using antibodies to UvsY protein. The nature of all of these protein--protein and protein-DNA binding activities is complex. All three of the proteins

studied

here have been shown

t)o bind t’o each of the others and to bind to ssDNA. Protein affinity columns yield the following list of interactions, given in order of increasing affinity. with the sodium chloride concentration necessary to

disrupt binding in parentheses: l’vsX-gp32 (0%): IJvsX-L’vsY (0.45); LJvsX-UvsX (0.45); I:vsYgp32 (0%); gp32-gp32 (0%) (Formosa et al.. 1983; Formosa & Alberts, 1984). The affinity of each of the three proteins for ssDNA-cellulose columns are: UvsX,

0.3; UvsY,

Alberts,

0.6; and gp32,

198&z; Rittner

conditions

for

2.0 (Formosa

&

et ccl., 1979). Although

affinity

chromatography

are

the very

different from those used in in vitro strand exchange reactions, the results of the chromatography experiments do reveal trends and likely hierarchies of proteinqrotein

and protein-DNA

Several interesting

observations

IisGng.

Although

binds as tightly added

neither

ITvsX

Interactions.

arise from such a nor LTvsY protein

as gp32 to ssDNA, when they were

t’ogether

they

withstood

competitive

dis-

placement by gp32. It might be argued that’ LTvsY protein simply bound up gp32, preventing it from displacing UvsX would not explain

protein from ssDNA, but that how ITvsY protein also allowed

CvsX protein to bind to ssDNA in higher potassium acetat)e concentrations. A direct UvsXI’vsY interaction is more in agreement with the data. On afinity

columns,

UvsY

protein

appears

to

bind more tightly to ssDNA than it does to UvsX protein, and yet here it did not displace UvsX protein but actually increased its affinity for ssDNA. These observations, again, imply that lJvsY protein induces a transition in the ssI)NAbinding mode of LJvsX protein. Tt is puzzling that lJvsY protein binds to UvsX protein and apparently interacts with the UvsX--ssDNA filament, but has no apparent, interaction with UvsX-

that

protein-prot,ein

appear

iridist~itiguistlat,le filaments.

one cannot nec*essarily interact until t,hey are the

reactions

t’hat

they

then, contributr t)o a many ditl’erent protein

to associate

and protein--DNA

together

in large

complcx~~s.

This work was suppolted by a grant from the NTH (GM31819) and a NIH t,raining grant award to I,. D.H. (GMO7092).

References Kernstein, C. B. & Wallace, S. S. (1983). In Haeteriophagc! Td (Mathews, C. K., Kutter, E. M., Mosig, G. dt Herget, P. B., eds), pp. 138--151, American Society for Microbiology, Washington, D.C. Hittner, M., Burke, R. 1,. & Alberts. B. M. (1979). J. Biol. Chem. 254, 9565-9572.

Boyle, J. M. & Symonds. N. 439. Broker. T. R. (1973). J. Mol. Broker, T. R. & Doermann, Genet. 9, 213-244. Broker, T. R. & Lehman, I.

(1969). M&at.

Rrs. 8. 431

81. l-16. A. H. (1975). Annu.

Biol.

Rer.

R. (1971). J. Mol. Biol. 60.

131-149.

Clhase, ,J. W., Whittier, R. F., Auerbach, J., Sancar, A. & Rupp, W. D. (1980). Nucl. Acids Res. 8, 3215 3227. (:how, 1,. T. & Broker. T. R. (1981). In Electron Microscopy in Biology (GrifEth, J. I)., ed.), vol. I. pp. 139. 188, John Wiley and Sons, New York. Cunningham, R. 1’. & Berger, H. (1977). I’irologq, 80. 67-82. Dannenberg. R. & Mosig, G. (1983). J. I’irol. 45, 813-831. Dewey, M. J. & Frankel, F. R. (1975a). Virology, 68. 387~m 401. Dewey, M. ,J. & Frankel, F. R. (19756). Virology, 68. 402 417. Formosa. T. & Alberta. B. M. (1984). (‘old Spring ffarbor Symp. Qua&. Biol. 49, 363-370. Formosa. T. & Alberts, B. M. (1986a). J. Biol. (‘hem. 261. 6107 6118. Formosa, T. &, Alberts, H. M. (19686). Cell, 47, 793-806. Formosa. T.. Burke, R. L. C Alberts, B. M. (1983). Proc. ,Vat. Acad. Sci., f1.S.A.

80, 2442-2446.

Griffith. ,J. & Formosa. T. (1985). J. uiol. (Sm. 260, 4484-449 1. Griffith. *J. & Harris, I,. D. (1988). (‘RP (‘rit. Rev. Biochem. 23, S43-886. Griffith. ,I. & Shores, (‘. G. (1985). Biochemistry, 24, 158 162. Harm, W. (1963). Virology, 19, 6G7 I. Harm, b’. (1964). Mutat. Res. 1, 344-354. Harris, I,. I). & Griffith, .I. (1987). ,J. Hiol. (Them. 262, 9285.-9292. Harris, L. D. & Griffith, ,I. II. (1988). Biochemistry, 27. 6954-6959. Hinton, I). M. & Nossal, N. G. (1986). J. Biol. Chem. 261, 5663-5673. Huberman, J. A. (1969). (‘old Spring Harbor Symp. Quant.

Kodadek,

Biol.

33. 509-524.

T., Wong, M. L. & Alberts, B. M. (1988). J. Biol. Ghem. 263, 9427-9436. Kozinski. A. W. (1968). Cold Spring Harbor flymp. Quant. Biol. 33, 375-391.

l”vs Y Protein

is an Accessory

Kozinski, A. J. & Kosturko, L. D. (1976). J. Birol. 17, 8Olm-804. Kreuzer, K. N. & Jongoneel? C. V. (1983). Methods Enzymol. 100, 144-160. Luder. A. & Mosig, G. (1982). hoc. Nat. Acad. Sci., 1:S.A. 79, 1101-l 105. Melamede. R qJ. & Wallace, S. S. (1977). -1. Yirol. 24, 2&40. Melamede, R. J. & Wallace. S. 8. (198Oa). Mol. Den. Genet. 177, 501-509. Melamede. R. J. & Wallace, S. S. (1980b). Mol. Gen. Gm64. 179, 327-330.

Protein

27

Minagawa, T., Fujisawa, H., Yonesaki, T. dt Ryo, Y. (1988). Mol. Gen. Genet. 211, 35Cb-356. Modrich, P. & Zabel, D. (1976). J. Riol. (Ihem. 251, 5866 5874. Panuska, J. R. & Goldthwait, D. A. (1980). J. Hiol. Chem. 255, 5208-5214. Wakem, L. P. & Ebisuzaki, K. (1976). I’irolqy, 73. 155164. Yonesaki. T. & Minagawa, T. (1985). EMHO J. 4. 33213327. Yonesaki. T., Rye, Y., Minagawa, T. & Takahashi, H. (1985). Eur. J. Biochrm. 148, 127.-134.

Edited by P. von Hippel