Generality of the Shared Active Site among Yeast Family Site ...

THE JOURNAL OF BlomrcAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 269, No. 17, Issue of April 29, pp. 12789-12796, 1994 Printed in U.S.A.

Generality of the Shared Active Site among Yeast Family Site-specific Recombinases THE ARg SITE-SPECIFIC RECOMBINASE FOLLOWS THE Flp PARADIGM*

(Received for publication, November 8, 1993, and in revised form, January 26, 1994)

Sang Hwa Yang and Makkuni JayaramS From the Department of Microbiology, University of Texas at Austin, Austin, Texas 78712

Mutations of the invariant Int family tetrad residues, the RHR triad, and the active site tyrosine, within the Zygosaccharomyces rouxii site-specific recombinase ARg cause the same “step-arrest” phenotypes as they do in the Flp recombinase of Saccharomyces cereuisiae. In “half-site” recombinations, the ARg recombinase exhib- its catalytic complementation between an RHR triad mutant and an active site tyrosine mutant. Strand cutting by R follows the “trans” DNA cleavage rule. These results are best explained by the assembly of a functional active site from partial active sites harbored by the ARg monomers. Complementation tests using single and double step-arrest ARg mutants verify critical predictions of the “shared active site” model. A wild type monomer paired with an RHR tr iad-ww double mutant is a catalytically inactive combination. Pairwise combinations of a single or a double RHR mutant with R(Y358F) yield comparable levels of catalytic complementation. These results strongly imply conservation of the mechanism of active site assembly and the mode of substrate cleavage within the yeast family site-specific recombinases, and perhaps within the larger Int family recombinases.

The chemistry of site-specific recombination mediated by the Int family site-specific recombinases (Argos et al., 1986; Utatsu et al., 1987) involves two transesterification steps (Craig, 1988; Landy, 1989). The first step breaks the exchange site phosphodiester within the DNA substrate using the phenolate moiety of the active site tyrosine as the nucleophile. This reaction produces a phosphodiester between the 3”phosphate of DNA and the catalytic tyrosine. The other product is a free 5’-hydroxyl at the nick site. The second transesterification restores the 3‘-5’ DNA phosphodiester in the recombinant mode by using the 5‘-hydroxyl of the nicked partner substrate as the nucleophile and liberates the enzyme from its DNA linkage. The recombination reaction proceeds by exchange of a pair of strands, branch migration of the resulting Holliday junction and the resolution of this junction by exchange of the second pair of strands. The recombination system has to cope with the following mechanistic problems. 1) How to temporally coordi- nate the strand breakage and union events within the two substrates? 2) How to achieve strand joining between the bro-

Health. Partial support was provided by the Council for Tobacco Re- * This work was supported by a grant from the National Institutes of

search (U.S.A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed. Tel.: 512-471-0966; Fax: 512-471-5546; E-Mail: [email protected].

ken substrates rather than ligation within each substrate (re- versal of strand breakage)? 3) How to avoid partial reactions of recombination?

Recent results on the architecture of the active site of the Flp site-specific recombinase of Saccharomyces cereuisiae provide clues as to how temporal coordination of breakagelreunion, avoidance of partial reactions, and facilitation of strand joining in the recombinant mode may be accomplished (Chen et al., 1992a, 1993; Lee and Jayaram, 1993; Pan et al., 1993). The Flp protein monomer harbors a partial active site; a complete active site is assembled by sharing of critical amino acid residues between monomers. The shared active site precludes “cis” DNA cleavage, and directly leads to “trans” DNA cleavage. That is, a Flp monomer bound to a recombination site does not cleave the scissile phosphodiester adjacent to it, but rather cleaves the phosphodiester adjacent to a second bound Flp monomer. The function of a bound monomer is to activate the phosphodiester in cis for nucleophilic attack; the nucleophile itself is provided in trans by the partner Flp monomer in the form of the active site tyrosine. This cis activatiodtrans nucleophilic attack model can also be applied to the strand joining step. Here, the bound protein activates the phosphodiester between DNA and the catalytic tyrosine and the 5”hydroxyl of DNA provides the nucleophile. In the simplest form of the model, for a particular phosphodiester that is exchanged, it is the same Flp monomer that performs the cis activation in the cleavage and the exchange steps. The simplicity and parsimony of the model and its general implications for phosphoryl transfer suggest that its essential features are likely to be global to Int family recombinases.

We have tested the validity of the partial active site model and the cis activatiodtrans nucleophilic attack paradigm for the ARg site-specific recombinase of Zygosaccharomyces rouxii. The results uphold the model. We conclude that the mechanism of recombination proposed initially for Flp is general to the yeast family of site-specific recombinases. We are inclined to believe that the generality extends to the Int family (of which the yeast recombinases comprise a subfamily) as well.

MATERIALS AND METHODS Purification ofFlp and R-Purification of Flp and Flp variants have

been described previously (Parsons et al., 1990). These preparations were >95% pure as judged by the Coomassie Brilliant Blue staining of SDS-polyacrylamide gels in which they were fractionated. Fusion proteins between glutathione-S-transferase (GST)’ and Flp or R or their variants were expressed in Escherichia coli from the pGEX2T expression vector (Pharmacia LKB Biotechnology Inc.). In the hybrid protein, the carboxyl terminus of GST is fused to the amino terminus of the wild type or variant recombinase via a peptide bridge that contains thrombin cleavage site. Preparation of fusion proteins followed the procedures prescribed by Pharmacia. Extracts of E. coli cells induced with isopro-

The abbreviations used are: GST, glutathione-S-transferase; nt, nucleotides.

12789

12790 Common Reaction Mechanism in Yeast Family Site-specific Recombinases

TABLE I Reactivity of the step-arrest mutants of Arg toward the ARg recombination site

The ability of the ARg mutants to bind, cleave, and recombine substrates containing the AFtg target site was assayed. Symbols for approximately normal reactivity (+), little or no reactivity (-), and relatively weak reactivity (+/-) are based upon the activity of the wild type Arg protein (+) in similar assays. The list includes previously described mutants (Chen et al., 1992b; Lee et al., 1992) and those that were newly constructed and analyzed.

R207S R207P H3417P H3417L H317Q R32OQ R320G R320P R320K Y358F ~ ~~ ~~ ~

Full-site binding + + + + + + + + + + Full-site cleavage - - + + + - - + - Full-site recombination

+/ -

Half-site recombination - - - - - - - - +/ - - - - - - - - - - - -

pyl-1-thio-P-D-galactopyranoside were prepared by sonication in 20 IIIM sodium phosphate (pH 7.0), 150 IIIM NaCl, 1 IIIM EDTA, 10% glycerol (v/v) using a Tekmar Sonic disruptor. After removing cell debris by centrifugation at approximately 10,000 x g at 4 "C, the fusion protein was adsorbed to glutathione beads and recovered by centrifugation. The beads were washed twice with 50 IIIM Tris-HC1 (pH 7.5), 1 IIIM EDTA, 2.5 mM CaCl,, 150 IIIM NaCl and resuspended in the same buffer for cleavage with thrombin. Following thrombin treatment, the reaction mixture was centrifuged and the cleaved recombinase or variant was collected in the supernatant. The supernatant was diluted with an equal volume of 50 m~ sodium phosphate (pH 7.0), 150 IIIM NaCl, 1 m EDTA, 20% glycerol (vfv) and used in recombination assays immediately or stored at -70 "C prior to use. Electrophoresis in SDS-polyacrylamide gels indicated approximately 40% purity of the recombinase proteins as inferred from staining with Coomassie Brilliant Blue.

The GST fusion protein (without thrombin cleavage) was used in some experiments. Following adsorption of the hybrid protein to the beads, they were washed twice with 50 mM Tris-HC1 (pH 7.51, 1 m EDTA, 2.5 m CaCl,, 150 mM NaC1. They were then equilibrated with the elution buffer (20 m glutathione, 100 mM "is-HC1, pH 7.5, 125 m NaCI). The sample was spun briefly and the supernatant containing the R protein was recovered. The protein fraction was dialyzed against 50 mM Tris-HC1 (pH 7.9, 125 mM NaCl, 1 m EDTA, 10% glycerol prior to storage at -70 "C. The purity of the preparations, based on Coomassie Brilliant Blue staining, was approximately 3040%.

Half-recombination Sites-Oligodeoxynucleotides for construction of half-sites were synthesized in an Applied Biosystems DNA synthesizer (model 380A) using phosphoramidite chemistry (Beaucage and Caruth- ers, 1981). Normally 10-20 pmol each of the two appropriate oligodeoxynucleotide pairs were mixed in TE (10 m Tris-HC1, pH 7.8, at 23 "C; 1 IIIM EDTA, pH 8.01, heated to 65 "C for 10 min, and cooled slowly to room temperature.

Two synthetic half-site substrates were used. Their sequences were: 1)5'-ATCACTGTGGACGlTGA~AAAGAAtac-3'/5'-H0-taacgta~C- "TCATCAACGTCC; 2) 5'-HO-tacgtta'ITCTlTCATCAAG 3'/5'-AAT- TC'ITGATGAAAGAAtaa-3'. The sequences represented in bold letters correspond to the 12-base pair ARg protein-binding element. The sequences shown in italics correspond to the strand exchange region (spacer). Substrate 1 and 2 are arbitrarily designated as the lek and right half-sites, respectively. Cleavage by ARg occurs on the strands with the three nucleotide spacers (tac and taa) at the phosphate 5' to the spacer "t" and 3' to last nucleotide "A" of the binding element. The spacer 5'-OH groups on the noncleaved strands participate in the strand exchange event.

To label the 5' end of an oligonucleotide, it was phosphorylated by T4 polynucleotide kinase in presence of [-p3'P1ATP or unlabeled ATP. The unreacted ATP was removed by spin-dialysis on a G-25 column. Hybrid- ization to the partner oligodeoxynucleotide was done in TE.

Strand 7kansfer Reactions in Half-sites-The conditions for half-site strand transfer were essentially the same as those for Flp recombinations described previously (Parsons et al., 1990; Chen et al., 1992a). Reactions were terminated by the addition of SDS (0.1% final concentration) and treated with proteinase K (100 &reaction for 1 h at 37 "C). After chloroform-phenol extraction and ethanol precipitation, the DNA was fractionated by electrophoresis in 10% denaturing polyacrylamide gels (acrylamide to bis-acrylamide, 19:l). Since the reactions contained a radioactively labeled substrate, the strand exchange products were visualized by autoradiography.

Assay for Formation of Covalent Recombinase-DNA Complexes-The half-site substrate in the assay was 5' end-labeled with 32P on the cleavage strand. The spacer hydroxyl on the complementary strand was phosphorylated using unlabeled ATP and polynucleotide kinase. This treatment prevented the strand transfer reaction and caused the strand cleavage product to accumulate. Reactions were carried out under re-

combination conditions. The reactions were stopped by the addition of an equal volume of 250 m~ Tris-HC1 (pH 7.8), 4% SDS, 40% glycerol, and 300 m~ P-mercaptoethanol. Suitable aliquots were heated at 95 "C for 3 min and electrophoresed in 8% polyacrylamide (acrylamide to bis-acrylamide, 29:l) gels (Laemmli, 1970). After the run, gels were rinsed twice (10 min each) in distilled water with gentle shaking at room temperature and were dried. DNA-protein complexes were re- vealed by autoradiography.

General Methods-Restriction enzyme digestions, isolation of plas- mid DNA, and other miscellaneous procedures were done as described by Maniatis et al. (1982).

RESULTS

Complementation between Step-arrest Mutants ofARg Recom- bznase-The R protein from 2. rouxii is a typical Int family recombinase. It contains the invariant tetrad, Argo', His3l', AI-$", and QP5', that is the hallmark of the Int family (Argos et al., 1986; Utatsu et al., 1987; Abremski and Hoess, 1992). These residues align with Arglgl, His305, Argo', and Qr343 of the Flp recombinase of S. cereuisiae. lj~'~' is inferred to be the active site tyrosine of R. This residue provides the nucleophile during strand cleavage, causing its covalent linkage to DNA via the 3"phosphate at the break point. It thus becomes part of the target diester (ribose-3'-phosphotyrosyl diester) during the strand exchange step of recombination initiated by the 5'-hydroxyl of the nicked strand from the partner substrate.

Point mutations of the Int family tetrad within Flp are known to produce well-defined step-arrest phenotypes. Alter- ations ofArgls1 and Ar$08 of Flp lead to arrest of recombination at the strand cleavage step in full-sites and half-sites (Parsons et al., 1988, 1990; Chen et al., 1992b; Friesen and Sadowski, 1992; Lee et al., 1992; Pan and Sadowski, 1992; Serre et al., 1992). Substitutions at His305 permit normal cleavage in full- sites, but weak cleavage in half-sites; they block strand joining in both half-sites and full-sites. Mutations of "l?y943 abolish strand cleavage as expected. Each of the RHR triad (ArgIg1, His305, and Ar$08) point mutants of Flp can be efficiently complemented by the T y ? 4 3 mutant Flp(Y343F) in half-site recombination (Chen et al., 1992a; Pan et al., 1993).

Previous results had indicated that kgo7, His'l', and W5' of ARg protein play roles analogous to those of their counter- parts in Flp (Chen et al., 1992b; Lee et al., 1992). Difficulties in purifying Ar$20 mutants precluded reliable inferences regard- ing the potential catalytic role of this residue. Purification of Arg and Arg derivatives as fusions to glutathione-S-transferase and subsequent cleavage of the hybrid proteins have circum- vented this difficulty. The reactivity of the RHR triad and Y358 mutants towards full-site and half-site substrates are listed in Table I. In a half-site strand transfer reaction (Serre et al., 1992; see Fig. 2 A , for example), which involves one strand cleavage and one strand joining steps, each of these mutants was virtually inactive (Fig. IA, lanes 4-7). Among the 4 mutants shown in Fig. IA, only R(H317L) yielded a trace amount of recombinase activity (less than 1% of wild type activity; compare lane 5 to lanes 2 and 3). Furthermore, pairwise combinations of the RHR triad point mutants failed to complement

Common Reaction Mechanism in Yea

A I \

R207S H317L R320Q + + +

R H317L R320Q R207S

1 2 3 4 5 6 7 8 9 1 0 111213 141516

P

B Y358F Y358F Y358F

+ + + R207S R320Q H317L

1 2 3 4 5 6 7 8 9

* m u " 0 . . . .-

FIG. 1. Complementation between step-arrest mutants of ARg recombinase in half-site strand transfer. Strand transfer reactions were done with a half-site 5' end-labeled on the cleavage strand ( S ) and harboring a 5'-hydroxyl group on the spacer segment of the complementary strand. The cleavage strand contained 3 spacer nucleotides adjacent to the phosphodiester that was exchanged in the reaction. Follow- ing strand cleavage, the released 3-nt spacer segment would be expected to diffuse away from the reaction center. Strand transfer using the spacer hydroxyl within the cleaved half-site or that from a second half-site molecule would give rise to a hairpin or a linear recombinant, respectively. On a denaturing gel, both products would be represented by their common recombinant labeled strand (P). A, half-site reactions were carried out with wild type ARg (lanes 2 and 3), the indicated step-arrest mutants (lanes 4-7), and pairwise combinations of the RHR triad mutants (lanes 8-16). Lane 1 is a control in which no protein was added to the reaction. The wild type reaction contained approximately 4 and 6 pmol (lanes 2 and 3, respectively) ofARg/pmol of half-site. The reactions shown in lanes 4-7 contained approximately 10 pmol of the variant per pmol of substrate. No significant reaction with these mutants was observed when a range of molar ratios of protein and substrate (above and below 10) was tested. In the complementation reactions this ratio was (from leR to right) 4,6, and 8 for one protein and 8, 6, and 4 for its partner. B, complementation reactions with each of the RHR mutant shown in A paired with R(Y358F) are displayed (lanes 1-9). In these reactions R(Y358F) was maintained at approximately 6 pmoVpmol of half-site. The RHR mutant concentration was varied from 4 (left lane), to 6 (middle lane), and 8 (right lane) pmoVpmol of substrate.

each other (Fig. L4, lanes 8-16). However, each of the triad point mutants was complemented by a w5' mutant, R(Y358F) (Fig. 1B). The close correspondence of these results with those obtained with Flp variants substantiates the conserved roles of the tetrad residues in the catalytic steps of recombination. The product "P" represents the sum of the recombinants formed by

1st Family Site-specific Recombinases 12791 intramolecular strand transfer within the half-site substrate (S) and intermolecular strand transfer between two half-site molecules.

DNA Cleavage In mans by the Arg Protein-The mode of DNA cleavage by R was tested by using the experimental de- sign previously applied to Flp (Chen et al., 1992a; Fig. 2). Complementation tests were performed between one half-site substrate (say, left) bound by R(Y358F) and a second half-site (say, right) bound by an RHR mutant (in which the active site tyrosine, w5*, is intact). The size of a recombinant product unequivocally identifies the half-site that must be cleaved to generate it. In the experiments shown in Fig. 3, the 49- and 45-nt products (P,L and P,L, respectively) represent cleavage of the left-half site; the 37- and 41-nt products (P,R and P2R, respectively) correspond to cleavage of the right half-site. Of these, the 49- and 37-nt (P,L and P,R) products are hairpins resulting from intramolecular strand joining or linear recombinants produced by intermolecular self-cross (left X le% right X right). Note that both types of products contain identical labeled strands and are indistinguishable by electrophoresis in a denaturing gel. The 45- and 41-nt products (P,L and P,R) are the result of inter half-site (left X right) strand joining.

In a reaction containing a left half-site plus a right half-site, there was a significant under-representation of the 49-nt product (PILI, the result of the left half-site cleavage and intramolecular strand joining, or strand joining to a second left half-site (Fig. 3). This was observed in the wild type reaction (Fig. 3A, lane 1' ), or in reactions containing R(Y358F) paired with an RHR triad mutant (Fig. 3, A-C, lanes 2 ) . Furthermore, the right half-site was approximately 2-2.5-fold more active than the left half-site (compare, for example, the sum of P,R plus P2R to P,L plus P2L in lanes 1' and 2 of Fig. 3A ). We do not under- stand the reasons for this nonparity. One possible explanation rests on the asymmetry of the spacer sequence. This may, in some way, be responsible for differential cleavages of the two half-sites. In addition, the sequence of the spacer on the noncleaved strand of a half-site will determine the base-pairing potential at the strand joining end. This may, in turn, influence the efficiency of the strand exchange step. For inter-half-site (left X right) joining reaction, there is no difference between left and right cleavage as the spacer sequences are mutually complementary (refer to Fig. 2). For intra-half-site or inter- half-site self-cross (left X left) or (right X right), the situation is different. In both instances, the first two nucleotide positions can form identical base pairs regardless of left cleavage or right cleavage (see Fig. 2). In the intramolecular hairpin formation, the central three positions are Ei"CGT-3' for the right half-site (Fig. 2 A ) and 5'-ACG-3' for the left half-site (Fig. 2B). In the intermolecular self-reaction, the central mispaired positions would be 5'-CGT-3'/3'-TGC-5' for the right half-site and 5'- ACG-3'/3'-GCA-5' for the left half-site. Results with Flp indicate that the pairing potential at the three nt positions adjacent to the cleavage point influence the efficiency of strand joining? Finally, for a given amount of cleavage of a half-site (say, left), the relative concentration of its complementary partner (the right half-site) can significantly influence the type of recombinant formed: hairpin, left X left, and left X right (Serre et al., 1992). The above factors, singly or in combination, may be responsible for the unequal amounts of the various types of half-site recombinants produced in the complementation reactions shown in Fig. 3. This asymmetry in the extent of product formation, however, does not affect the interpretation of the results demonstrating the type of strand cleavage (cis or trans) mediated by R.

For every RHR mutant paired with R(Y358F1, recombination

* J. Lee and M. Jayaram, unpublished data.


SL SR 28 nt *-TAC 5’ H O - T A C G T T A “ A m 20 nt 24 flt w-”------)ATGCAAT-OH 5‘ AAT-* 20nt

FIG. 2. Predicted recombination products in a reaction containing two half-site substrates. The two substrates are the left half-site ( S L ) and the right half-site (SR) , each labeled on the

The left (A) and right ( B ) designation is cleavage strand at the 5’ end (asterisks).

arbitrary. The lengths of each strand within the substrates are indicated. Cleavage within SL followed by intramolecular strand exchange will give P,L (49 nt); intermolecular exchange using the unlabeled strand of SR will yield a recombinant in which the labeled strand is 45 nt (P,L). Similarly, cleavage within SR will give rise to the intramolecular product P,R (37 nt) and the intermolecular product P,R (41 nt).

I

Pi L * - TACGTTA ( 28+24-3 ) =49

P2L ( 28+20-3) =45

SL SR 28 nt * -TAc 5’ HO-TACGTTA - 20 nt 24 nt ww-w ATGCAAT-OH 5 AAT-* 20nt

.. - ATGCAATd- w * PI R

P2R (20+20-3 ) =37

( 24+20-3 ) =41

was the result of cleavage of the half-site to which R(Y358F) was bound (Fig. 3). When the right half-site was bound by R(Y358F) and the left half-site by the RHR mutant, it was the right half-site that was cleaved. The recombinant products were almost exclusively P,R + P,R at early time points and predominantly P,R + P,R at later time points (Fig. 3, A-C, lanes 3-6). When the protein partners were switched, the left half- site became the target of cleavage. Now, the predominant product was P,L; little of the PR products were formed. When the half-sites were not prebound by the individual mutants, recombination products resulted from cleavage of the left or right half-site, as expected for random binding of the two mutant proteins to the two half-sites (Fig. 3, A-C, lanes 2). The simplest conclusion then, based on the assumption that cleavage is executed by 15..‘“ during complementation, is that the ARg protein executes DNA cleavage in trans.

The results with the half-sites prebound with ARg variants does not absolutely rule out cis cleavage. For example, if, under the reaction conditions, a protein nucleophile other than the phenolate moiety of Tyr358 is efficiently utilized for strand cutting, it will not be possible to distinguish between cis and trans cleavage. The following strategy, based on the electrophoretic

mobility of the covalent DNA-recombinase product resulting from cleavage, resolves this dilemma (Fig. 4). The hybrid protein between ARg recombinase and GST constructed for facili- tating purification ofARg (which is larger than ARg by approximately 30 kDa) is active in strand cleavage and strand transfer. Hence, the covalent DNA adduct produced by the normal and the hybrid recombinases can be easily distinguished by their differential migration in an SDS-polyacrylamide gel. The results of fractionation of reactions in which a 5’ end-labeled half-site was incubated with wild type ARg or GST-ARg are shown in lanes 1 and 2, respectively, of Fig. 4A. The reason for the heterogeneity of the bands corresponding to the cleavage product is unclear. Presumably this is caused by mild proteolysis in the stored recombinase preparations, or by proteolysis occurring during the cleavage reaction. On the other hand, we cannot rule out some form of molecular aggregation, or conformational heterogeneity (for example, single stranded, double- stranded, or branched DNA) within the cleavage product. Nev- ertheless, the ARg.DNA complex (lane 1 ) and GST.ARg.DNA complex (lane 2) have quite distinct mobilities. Similar results were obtained with the Flp recombinase (Fig. 4B, lane 1 ) and the GST-Flp fusion recombinase (Fig. 4B, lane 2). In a reaction

Common Reaction Mechanism in Yeast Family Site-specific Recombinases 12793

A B

SL

SR

1 1' 2 3 4 5 6 7 8 910

Left: R207S Left: Y358F

Right: Y358F Right: R207S

C 3 a 0

""

"

1 2 3 4 5 6 7 8 910

Left: R320Q Left: Y.3581: Right: Y358F Right: R320Q

PlL P2 L P2 R PIR

1 2 3 4 5 6 7 8 9 10

lnft: 11.3 17L left: Y.3581: Right: Y.3581: l<ight: 113 171.

FIG. 3. l k a n s DNA cleavage during pairwise complementation between R(Y358F) and the RHR triad mutants. Reactions contained approximately equimolar amounts of the radioactively labeled left (SL) and right ( S R ) half-sites (see Fig. 2). The recombinant products resulting from cleavage of SL are P,L and P,L; those resulting from cleavage of the right SR are P,R and P,R. Complementation reactions of R(R207S), R(R320Q), and R(H317L), each paired with R(Y358F), are shown in A, B, and C, respectively. Lune 1 represents incubations of the substrates without added protein. Lune 1' inA is a 20-min wild type reaction at 6 pmol ofARg/pmol of half-site. Lune 2 is a reaction containing approximately equimolar amounts of R(Y358F) and the RHR point mutant (roughly 3 pmol each per pmol of half-site) added simultaneously to a mixture of the half-sites in the incubation buffer. Lunes 3-10 represent reactions in which each half-site was prebound by one of the two complementing protein partners. The protein-substrate configurations in the assay are indicated at the top. Experimentally, each half-site was incubated with the appropriate AFtg mutant on ice for 10 min and then mixed into a tube containing a large excess (50-fold) of the unlabeled half-sites. Reactions were transferred to 30 "C, and aliquots were sampled at 2, 5, 10, and 20 min. After proteinase K treatment and phenol-chloroform extraction (see "Materials and Methods"), samples were subjected to electrophoresis in 10% denaturing polyacrylamide gels. Products were identified by autoradiography.

containing the complementing pair GST-R(Y358F) and R(H317L) (Fig. 4A, lane 3) or the GST-Flp(Y343F) and Flp(H305L) pair (Fig. 4B, lane 3), it was the R(H317L) or the Flp(H305L) that became covalently linked to DNA. No trace of the larger complex, expected of GST-R(Y358F) or GST- Flp(Y343F), was detectable. Conversely when the complementary pair was GST-R(H317L) and R(Y358F) (Fig. 4A, lane 4 ) or GST-Flp(H305L) and Flp(Y343F) (Fig. 4 4 lane 4 ), the cleavage product was exclusively produced by the larger hybrid protein. Reactions with a second complementing pair of Arg variants, R(R207S) plus GST-R(Y358F) and GST-R(R207S) plus R(Y358F) (Fig. 4A, lanes 5 and 61, confirm these results.

The sum of the data shown in Figs. 3 and 4 is consistent with the trans cleavage model, and is inconsistent with the cis cleavage model. Furthermore, the finding that cleavage is executed

by the recombinase variant containing the active site tyrosine implies that, in the complementation reactions, it is the normal active site tyrosine that likely mediates strand cleavage.

The ARg Recombinase Assembles Its Active Site by Sharing the RHR Piad and the Catalytic arosine between Monomers- The trans mode of DNA cleavage by the Flp recombinase could be readily accounted for by the partial active site configuration of the Flp monomer (Chen et al., 1993). Assembly of active sites at the interface of two protein monomers has been documented in allosteric enzymes, a classic example of which is aspartate transcarbamylase (Wente and Schachman, 1987). If the ARg protein follows the Flp paradigm, the following rules must be obeyed (Fig. 5). 1) The combination of a wild type ARg monomer paired with an RHR-TY~~~* double mutant should not yield a functional active site. 2) Regardless of whether a monomer is


A

R C-R

+ ci B

0.

Flp

u. + m m d

L m rn d > 5

1 2 3 4 5 6 1 2 3 4

FIG. 4. Identification of the protein responsible for strand cleavage during complementation between pairs of steparrest recombinase mutants. Strand cleavage reactions were carried out under half-site recombination conditions. The substrate used was labeled at the 5' end on the cleavage strand, and was prevented from completing strand transfer by phosphorylating the 5'-hydroxyl of the spacer on the noncleaved strand. The covalent DNA-recombinase complex formed by strand cleavage was identified by electrophoresis in 10% SDS-polyacrylamide gel, followed by autoradiography. A, reactions with Arg and Arg variants; B , reactions with Flp and Flp variants. The proteins or protein pairs used are indicated above the appropriate lanes. The prefix "G" is used to identify the hybrid protein formed by fusing GST to the recombinase. CP, cleavage product.

A 6, lunes 5-8) bound to the same, but radioactively labeled, half-sites were mixed in the presence of an excess of the cold half-site. The extent of half-site recombination was measured after a 20-min incubation at 30 "C. As predicted by the shared active site model, recombinant product was not realized in the wild typeldouble mutant reaction (lunes 5-8); the wild type1

mutant yielded half-site recombination. The double mutant was capable of binding the ARg target site with approxi-

not shown). I t was important to ensure that the observed reaction with the FUR(Y358F) pair (lunes 9-12) was not due to the

',,, dissociation of wild type from the unlabeled half-site, followed A?, F'R.. 1 by its association with the labeled half-site. A control reaction

,,' was carried out exactly as above, except that the unlabeled

B

/'

C __ . mately the same affinity as wild type ARg and R(Y358F) (data

. .,I

FIG. 5. Predictions of the shared active-site model. A, the model half-site bound by wild type ARg was mixed with the labeled proposes that two functional active sites (RHRYRHRY) can be derived to which no ARg variant was prebound. The extent Of from two R monomers. One monomer provides the RGR (AI$"-H~~~'~- reaction seen in this assay ( h n e 1.3) would represent t,he extent

(RHRYIOHRF; RHRYIOORF; RHRYIOOO).

mutant at one, two, or all three RHR triad positions, it must yield one functional active site when paired with a w58 mutant.

The experimental test of the shared active site model is shown in Figs. 6 and 7. The following strategies were used. In one set of experiments, wild type bound to an unlabeled half- site and R(Y358F) (Fig. 6, lunes 9-12) or R(R207S, Y358F) (Fig.

R(R207S, R320Q) (lunes 10-13) was tested for recombination activity using a labeled half-site substrate. In each pairwise combination, catalytic complementation between R(358F) and the partner double mutant was observed.

DISCUSSION

The results of this study demonstrate that the partial active siteltrans-DNA cleavage model proposed for the S. cerevisiae site-specific recombinase Flp is applicable to the R recombinase

Common Reaction Mechanism in Yeast Family Site-specific Recombinases 12795 LL m

LL vi Oob + +

R $3 2 R207S, Y358F Y358F m o

1 2 3 4 5 6 7 8 9101112 13

w "0- P

FIG. 6. Functionality of the wild type Arg when paired with an RHR-W double mutant of Arg. The protocol for the assays was

prebound to the same, but unlabeled, half-site. The binding mixtures were combined into a tube containing a 50-100-fold excess of the unlabeled similar to that described in the legend for Fig. 3. Here, the mutant R was prebound to a radioactively labeled half-site, the wild type Arg was

half-site and incubated at 30 "C for 20 min. The labeled strand of the recombinant products ( P ) was separated from the substrate (S ) by electrophoresis in a 10% denaturing polyacrylamide gel. Lane designations: I , labeled half-site without added protein; 2 4 , reactions with wild type Arg, R(Y358F), and R(R207S, Y358F) (approximately 3-5 pmol of Arg/pmol of half-site); 54, reactions containing roughly 3 pmol of Arg and 1.5, 2.3, and 4 pmol (left to right) of R(R207S, Y358F) per pmol of half-site; 9-12, reactions with approximately 3 pmol ofArg and 1.5,2,3, and 4 pmol of Flp(Y358F)lpmol of half-site; 13, reactions in which wild type Arg prebound to unlabeled half-site was mixed with labeled half-site (not prebound by a mutant) and diluted instantaneously into 50-100-fold excess of unlabeled half-site.

1 2 3 4 5

of 2. rourii as well. Complementation tests between the RHR triad mutants paired with R(Y358F) reveal that each of the

Y358F Y358F triad mutants tested can assemble a functional active site in combination with R(Y358F). Furthermore, double RHR mutants yield levels of complementation comparable to those ob-

R207S. H317L R207S. R320Q tained with single RHR mutants. Finally, wild type ARg paired with an RHR-Tyr35R double mutant of ARg fails to yield a functional active site. The simplest explanation of these results is that the catal-ytically active unit ofARg is at least a dimer. One

+ +

P monomer provides the triad residues; the second monomer provides the active site tyrosine. Thus, a dimer of wild type ARg can potentially build two active sites. Catalytic complementation in the half-site reaction between a triad mutant and a Tyr358 mutant results from the ability of the mutant combination to assemble one functional active site. The trans mode of DNA cleavage by ARg, demonstrated by using prebound half- site-ARg complexes, follows neatly from the shared active site configuration.

In Flp recombination, a cis activatiodtrans nucleophilic attack mechanism has been put forward to unify the chemistry of the strand cleavage and strand union reaction (Lee and Ja- yaram, 1993). The architecture of theARg active site and DNA

6 7 8 9 10 1112 13 cleavage in trans by ARg conform to asimilar reaction mecha- FIG. 7. Complementation of the RHR triad double mutants of

Arg by R(Y358F). The substrate for the reaction ( S ) was the same 32P-labeled half-site as that used in the assays of Fig. 6. In the complementation assays, R(Y358F) and the indicated double or triple mutant were simultaneously added to the reaction mixture containing the substrate (approximately 0.1 pmol). Lane I is a mock reaction containing neither Arg nor an Arg variant. Lane 2 is a wild type R reaction containing 3 pmol ofArg/pmol of half-site. Lanes 3 5 are reactions with the individual mutants at a protein to half-site molar ratio of approximately 4-6. Lanes 6-13 represent reactions containing R(Y358F) and the indicated RHR double mutant. The amount of Flp(Y358F) was approximately 3-4 pmol/pmol of half-site in all reactions. The amounts of the mutant protein for each complementation set were (from left to right) approximately 1, 2, 4, and 6 pmoVpmol of half-site.

nism. According to this model, the bound ARg monomer activates the exchange site phosphodiester adjacent to it (activation in cis). The w58 from a second ARg monomer then delivers the nucleophile in trans to effect DNA breakage. This transesterification reaction results in the formation of a phosphodiester between the scissile phosphate and w58. In the simplest case, the same ARg monomer that was responsible for cis activation of the cleavage step activates the DNAphosphodi- ester. The 5'-hydroxyl of the nicked partner DNA (in a full-site reaction) or that from the bottom strand spacer of a half-site provides the nucleophile to execute strand joining. This transesterification step restores the DNA phosphodiester in the recombinant configuration.


The similarity in the active site organization of Flp and ARg and in their mode of DNA cleavage strongly suggests that these features of site-specific recombination are likely to be common to all of the known site-specific recombinases encoded by yeast plasmids (the yeast family recombinases) (Utatsu et al., 1987). Trans DNAcleavage can, in principle, lead the strand breakage step toward recombination rather than futile religation in the parental configuration. If the cleavage were to occur in cis, some conformational flexure must follow to suppress strand union within the same substrate and encourage strand union between partner substrates. The shared active site may also provide a safety mechanism by which partial reactions of recombination can be avoided by ensuring that the chemistry of recombination is not triggered till the complete recombination machine is assembled.

These simple, yet apparently clever solutions to problems of phosphoryl transfer in DNA may apply to other Int family recombinases, for example, A Int and the Cre protein of phage P1, which also make use of oligomeric protein assemblies to carry out catalysis. I t seems unlikely, given the chemical similarity of Int and Cre recombination to that carried out by Flp, that they would have devised independent solutions of their own. Previous results on strand cutting by the Int protein within half attachment sites have been interpreted to represent cis-cleavage (Kim et al., 1990). While this is the simplest interpretation of the data, it is not a unique interpretation. Trans cleavage can also accommodate the same results. In fact, recent complementation tests using step-arrest mutants of Int conform to the Flp paradigm of assembling an active site by sharing residues between at least two recombinase monomers (Han et al., 1993). The partial active site configuration of the Int monomer is suggestive of, but not proof for trans-DNA cleavage during A recombination. The implication is that the general rules of Flp recombination act not just on the yeast family recombinases, but may be global to the entire Int family. Some of the features of Flp and ARg recombination may extend

beyond the Int family and apply to other systems that utilize multisubunit protein machines to bring about phosphoryl transfer in nucleic acids.

Acknowledgments-We acknowledge Jehee Lee’s generosity in providing the results of experiments using the Flp protein. We are grateful to H. Araki and Y. Oshima (Osaka University, Japan) for initially providing ARg protein expression vectors. The experiments with GST- recombinase fusions were devised in response to Howard Nash’s criti- cism that previous complementation results do not unequivocally ex- clude cis DNA cleavage. We thank him for providing the challenge. We thank Belinda Gonzales (Data Processing Department, University of Texas, Austin) for preparing one of the figures.

REFERENCES

Abremski, K. E., and Hoess, R. H. (1992) Protein Eng. 5,87-91 Argos, P., Landy, A,, Abremski, K, Egan, J. B., Haggard-Ljungquist, E., Hoess, R.

H., Kahn, M. L., Kalionis, B., Narayana, S . V. L., Pierson, L. S., 111, Sternberg, N., and Leong, J. M. (1986) EMBO J. 6,433-440

Beaucage, S. L., and Caruthers, M. H. (1981) Tetrahedron Lett. 22, 1859-1862 Chen, J. W., Lee, J., and Jayaram, M. (1992a) Cell 6% 647458 Chen, J. W., Evans, B. R., Yang, S . H., Araki, H., Oshima, Y., and Jayaram, M.

Chen, J. W., Yang, S. H., and Jayaram, M. (1993) J. Biol. Chem. 268,14417-14425 Craig, N. L. (1988) Annu. Rev. Genet. 22.77-105 Friesen, H., and Sadowski, P. D. (1992) J. Mol. Biol. 226,313-326 Han, Y. W., Gumport, R. I., and Gardner, J. F. (1993) EMBO J. 12,45774584 Kim, S . , de Vargas, L. M., Nunes-Duby, S . E., and Landy, A. (1990) Cell 63,77%781 Laemmli, U. K (1970) Nature 286,68&685 Landy, A. (1989) Annu. Reu. Biochem. 68,913-949 Lee, J., and Jayaram, M. (1993) J. Biol. Chem. 268,17564-17579 Lee, J., Serre, M. C., Yang, S . H., Whang, I., Araki, H., Oshima, Y., and Jayaram,

Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular C1oning:A Labo- M. (1992) J. Mol. Biol. 228,1091-1103

Pan, H., and Sadowski, P. D. (1992) J. Biol. Chem. 267, 12397-12399 ratory Manual, Cold Spring Harbor Lahoratory, Cold Spring Harbor, NY

Pan, H., Luetke, K., and Sadowski, P. D. (1993) Mol. Cell. Biol. 13,31674175 Parsons, R. L., Prasad, P. V., Harsbey, R. M., and Jayaram, M. (1988) Mol. Cell.

Parsons, R. L., Evans, B. R., Zheng, L., and Jayaram, M. (1990) J. Biol. Chem. 265,

Serre. M. C.. Evans. B. R., Araki, H., Oshima, Y., and Jayaram, M. (1992) J. Mol.

(1992b) Mol. Cell. Biol. 12, 3757-3765

Biol. 8, 330S3310

45274533

Biol. 226, 621-642 Utatsu, J., Sakamoto, S., Imura, T., and lbh-E, A. (1987) J. Bacteriol. 169, 5537-

Wente, S. R., and Schachman, H. K. (1987) Proc. Natl. Acad. Sci. U. S. A. 84,3145 5545

Generality of the Shared Active Site among Yeast Family Site ...

Documents

Transcript of Generality of the Shared Active Site among Yeast Family Site ...