Repair of Gaps in Retroviral DNA Integration ... - Journal of Virology

Download PDF
2 downloads 0 Views 1MB Size Report
Comment
chemical approach to identify DNA repair proteins capable of repairing gapped ... IV, and XRCC4, is necessary for nonhomologous end joining of double-strand ...
JOURNAL OF VIROLOGY, Dec. 2000, p. 11191–11200 0022-538X/00/$04.00⫹0 Copyright © 2000, American Society for Microbiology. All Rights Reserved.

Vol. 74, No. 23

Repair of Gaps in Retroviral DNA Integration Intermediates KRISTINE E. YODER

AND

FREDERIC D. BUSHMAN*

Infectious Disease Laboratory, The Salk Institute, La Jolla, California 92037 Received 27 June 2000/Accepted 28 August 2000

Diverse mobile DNA elements are believed to pirate host cell enzymes to complete DNA transfer. Prominent examples are provided by retroviral cDNA integration and transposon insertion. These reactions initially involve the attachment of each element 3ⴕ DNA end to staggered sites in the host DNA by element-encoded integrase or transposase enzymes. Unfolding of such intermediates yields DNA gaps at each junction. It has been widely assumed that host DNA repair enzymes complete attachment of the remaining DNA ends, but the enzymes involved have not been identified for any system. We have synthesized DNA substrates containing the expected gap and 5ⴕ two-base flap structure present in retroviral integration intermediates and tested candidate enzymes for the ability to support repair in vitro. We find three required activities, two of which can be satisfied by multiple enzymes. These are a polymerase (polymerase beta, polymerase delta and its cofactor PCNA, or reverse transcriptase), a nuclease (flap endonuclease), and a ligase (ligase I, III, or IV and its cofactor XRCC4). A proposed pathway involving retroviral integrase and reverse transcriptase did not carry out repair under the conditions tested. In addition, prebinding of integrase protein to gapped DNA inhibited repair reactions, indicating that gap repair in vivo may require active disassembly of the integrase complex. Integration of the viral cDNA into the host chromosome is an essential part of the retrovirus life cycle (for recent reviews, see references 13 and 23). In vivo, integration is performed by a large complex of proteins and the viral cDNA called the preintegration complex (PIC). For the case of human immunodeficiency type 1 (HIV-1), proteins detected in PICs include the viral proteins integrase, reverse transcriptase (RT), matrix, and the host high-mobility group protein I(Y) (6, 18, 19, 35). The substrate for integration is the linear cDNA product of reverse transcription. Prior to integration, integrase protein removes two nucleotides from the viral cDNA 3⬘ ends, thereby preparing a defined substrate for the subsequent strand transfer reaction (7, 15, 27, 42). Integrase then catalyzes the covalent joining of the 3⬘ hydroxyl groups at each end of the viral cDNA to the host DNA (9, 15, 26). Joining in host DNA takes place on each strand at points separated by 4 to 6 bp. The spacing between points of joining is characteristic of the retrovirus involved. Unpairing of the host DNA between the points of cDNA joining produces DNA gaps at each host-virus DNA junction. Two bases from the 5⬘ cDNA ends remain unpaired and overhang as flaps. This gapped intermediate is then repaired, yielding short duplications of host sequence and completing formation of the provirus. Integration reactions with PICs in vitro yield the gapped intermediate but do not carry out subsequent repair (36). This is consistent with the idea that host DNA repair enzymes, which are not present in PIC preparations, complete provirus formation in vivo. Many transposable elements use related pathways for transposition, involving cleavage at the end of the element to liberate a 3⬘ hydroxyl and joining to staggered points in target DNA (16, 41; for a review, see reference 5). Like retroviruses, these mobile DNA elements must also repair gaps resulting in short duplications of target DNA sequence flanking the inserted element. To date genetic screens have not identified the enzymes directly responsible for repair of the gapped interme-

* Corresponding author. Mailing address: Infectious Disease Laboratory, The Salk Institute, 10010 N. Torrey Pines Rd., La Jolla, CA 29037. Phone: (858) 453-4100, ext. 1630. Fax: (858) 554-0341. E-mail: [email protected].

diates generated during transposon or retroviral integration. One possible explanation is that the mechanisms for repair of the gapped intermediate are redundant, and so single mutations do not yield clear defects in integration. Alternatively, the proteins involved may be essential for viability and so not easily recovered in genetic screens. We have chosen instead a biochemical approach to identify DNA repair proteins capable of repairing gapped integration intermediates in vitro. A simple mechanism of repair requires a polymerase to synthesize across the gap, a nuclease to remove the 5⬘ flap, and a ligase to join the strands (Fig. 1A). Repair pathways such as mismatch repair or base excision repair similarly require a polymerase, a nuclease, and a ligase to repair gaps generated at sites of DNA damage (for reviews, see references 20 and 34). For example, in one pathway of base excision repair, the damaged base is first removed by a glycosylase and then the backbone is cleaved by AP nuclease. Polymerase beta (Pol beta) then polymerases across the damaged region (21, 44). The reaction is completed by the mammalian 5⬘-to-3⬘ nuclease, flap endonuclease (FEN), and DNA ligase I (28, 29). It is estimated that in the course of 1 day each cell repairs 10,000 damaged sites (30). Thus, the host proteins involved in this repair are highly active, adding to their appeal as candidates for repair of integration intermediates. An alternative proposal holds that gap repair may be carried out entirely by viral proteins (Fig. 1B). Chow et al. found that in vitro, integrase could carry out a DNA splicing reaction, called disintegration, on suitable DNA substrates (12). They pointed out that if polymerization by RT across the gap produced such a substrate, then integrase might join the DNA strands by a disintegration reaction. It has even been proposed that integrase itself may have a DNA polymerase activity (1). Other factors have also been proposed to be involved in repair of retroviral integration intermediates, based on studies of infection of animal cells. Poly(ADP-ribose) polymerase (PARP), a proposed regulator of DNA repair, has been proposed to be involved in integration, based on studies of retroviral infection in the presence of PARP inhibitors (22). Another proposal arose from studies of retroviral integration in cell lines mutant in the genes encoding the dependent protein kinase-DNA DNA (DNA-PK) complex (17). This complex,

11191

11192

J. VIROL.

YODER AND BUSHMAN MATERIALS AND METHODS

FIG. 1. Candidate pathways for repair of integration intermediates. The action of integrase protein joins one DNA strand of the viral cDNA to target DNA at each cDNA end. Unfolding of this intermediate yields DNA gaps at each host-virus DNA junction, as shown at the top of each panel. The viral cDNA is shown by the darker line. 5⬘ DNA ends are shown by filled circles; hash marks indicate unpaired DNA bases. (A) Potential pathway for repair employing a polymerase, a 5⬘-to-3⬘ nuclease, and a ligase; (B) potential pathway for repair employing RT and integrase (12).

composed of DNA-PK catalytic subunit, Ku70, Ku80, ligase IV, and XRCC4, is necessary for nonhomologous end joining of double-strand breaks. Infection of cells mutant in this system led to an increase in apoptosis, which depended on the presence of an intact integrase gene. The apoptotic signal was attributed to the presence of unrepaired integration intermediates, though other sources of the signal have not been ruled out (14). Neither study provided specific candidates for the enzymes involved in the covalent chemistry of gap repair. In this study we used an in vitro assay to model potential repair pathways. Cellular and viral proteins were tested for the ability to repair model integration intermediates containing a five-base gap and two-base 5⬘ flap. We found that in vitro several polymerases, both host and viral, were capable of acting in concert with FEN and multiple host ligases to repair the gapped DNA substrate. The proposed pathway involving integrase and RT was not sufficient for repair under the conditions tested. Our data further indicated that the reported integraseassociated polymerase activity is unlikely to be involved. We found that added integrase protein blocks access of repair proteins to DNA gaps, indicating that there may be a requirement for a specific disassembly reaction to permit repair.

Proteins. All integrase and molluscum contagiosum virus topoisomerase (MCV topo) enzymes were purified after overexpression in Escherichia coli BL21/DE3. HIV integrase containing a hexahistidine (His) tag was purified as described elsewhere (8). HIV integrase lacking the His tag was purified as described in reference (2). MCV topo containing a His tag was purified as described in reference (25). MCV topo lacking a His tag was purified using SP-Sepharose and Mono S chromatography. Bacterially expressed human FEN1 was a gift from M. Park. Baculovirusexpressed human ligase I and ligase III beta were gifts from A. Tomkinson. Bacterially expressed human Pol beta was a gift from S. Wilson. Pol delta purified from calf thymus was generously provided by C. Tan. Bacterially expressed PCNA was a gift from B. Stillman. Baculovirus-coexpressed ligase IV and XRCC4 and baculovirus-coexpressed Ku70 and Ku80 were gifts from D. Ramsden. Rous sarcoma virus (RSV) integrase purified as described previously (53) was a gift from C. Hyde. Protein concentrations were confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) except for the case of Pol delta, which was too dilute to visualize; activity of all enzymes was confirmed by in vitro assays with appropriate DNA substrates. Repair assays in vitro. DNA oligonucleotides used are listed in Table 1. DNA oligonucleotides were synthesized on an ABI DNA synthesizer or purchased from Integrated DNA Technologies. Oligonucleotides were purified on 8% polyacrylamide DNA sequencing-type gels and eluted from gel slices (39). 5⬘ end labeling was accomplished by incubation with T4 polynucleotide kinase (New England Biolabs) and [␥-32P]dATP (Amersham Pharmacia). 3⬘ end labeling was carried out by incubation with terminal transferase (New England Biolabs) and [␣-32P]ddATP (Amersham Pharmacia). Labeled oligonucleotides were purified by G-25 spin columns (Boehringer Mannheim). For the gapped intermediate substrate, 5 pmol of labeled oligonucleotide was combined with 10 pmol of both partner oligonucleotides in the presence of 150 mM NaCl in a 100-␮l volume. The oligonucleotides were annealed by boiling and slow cooling to room temperature. For the RSV gapped dumbbell, 5-pmol aliquots of both oligonucleotides were combined and annealed as above except for being cooled to 4°C. The gapped intermediate substrate is composed of three oligonucleotides: the host sequence strand oKEY35, the HIV sequence oligonucleotide oKEY31, and the bridging strand oKEY36 (Table 1). The RSV gapped dumbbell substrate is composed of two oligonucleotides, oKEY40 and oKEY41. Two different RSV gapped dumbbell oligonucleotides were synthesized with complementary singlestranded regions so that the single-stranded gap could potentially anneal between molecules. The nicked substrate is composed of two oligonucleotides, oKEY48 and oKEY51 (note that one oligonucleotide forms a hairpin). Repair of the gapped intermediate substrate. Repair reactions were performed with 10 fmol of labeled gapped intermediate substrate in a mixture of 50 mM Tris HCl (pH 8.0), 10 mM MgCl2, 2 mM dithiothreitol, 0.2 mg of bovine serum albumin (BSA; New England Biolabs)/ml, and 2.5% glycerol in a 15-␮l total volume. Amounts of purified proteins used in reactions were 17 ng of Pol beta, 7 ng of FEN, 5 ng of ligase I, 6.4 ng of Pol delta, 20 or 800 ng of PCNA, 5 ng of HIV RT, 0.7 or 100 ng of HIV integrase. Reaction mixtures were incubated at 30°C for 90 min; reactions were stopped with addition of 15 ␮l of 80% formamide plus 10 mM EDTA. Samples were separated by SDS-PAGE on a 15% gel and analyzed by using a PhosphorImager. The 30°C reaction temperature was chosen to facilitate comparison with previous studies in the DNA repair field (e.g., reference 24). All assays were repeated with 2 to 12 independent substrate preparations. Quantitation of representative gels is shown. RSV gapped dumbbell assay. First, 0.1, 1, 10, 50, or 100 ng of RSV integrase or BSA (New England Biolabs) was preassembled with 10 fmol of RSV gapped dumbbell substrate in a mixture of 20 mM Tris HCl (pH 8.0), 2 mM ␤-mercaptoethanol, 5 mM MgCl2, 100 mM NaCl, 0.1 mg of BSA/ml, 0.1% Triton X-100, 20 ␮M dGTP, and 1 ␮l of [␣-32P]dCTP (Amersham Pharmacia) for 10 min at 30°C. Following preassembly, 17 ng of Pol beta was added. The reaction mixtures were incubated for 10 min at 30°C. Reactions were stopped with 5 ␮l of 0.5 M EDTA. Products were purified by G-25 spin columns (Boehringer Mannheim). An equal volume of 80% formamide plus 10 mM EDTA was added. Samples were separated on a 15% polyacrylamide gel and analyzed by using a PhosphorImager. Note that the formation of fold-back product seen in reactions with

TABLE 1. Oligonucleotides used in this study Name

oKEY35 ...............................5⬘ oKEY31 ...............................5⬘ oKEY36 ...............................5⬘ oKEY40 ...............................5⬘ oKEY41 ...............................5⬘ oKEY48 ...............................5⬘ oKEY51 ...............................5⬘

Sequence

ATTCGAGCTATCCTTGCGCG 3⬘ ACTGCTAGAGATTTTCCACACTGACTA 3⬘ TAGTCAGTGTGGAAAATCTCTAGCAGGCCCCGCGCAAGGATAGCTCGAAT 3⬘ AATGTAGTCTTATGCAGTTTTCTGCATAAGACTACAGGCGCCGAGTCAGCCGTTTTCGGCTGACTC 3⬘ AATCTAGTCTTATGCAGTTTTCTGCATAAGACTACAGGCGCCGAGTCAGCCGTTTTCGGCTGACTC 3⬘ CAGCAACGCAAGCTTG 3⬘ GGCTGCAGGTCGACGTCGACCTGCAGCCCAAGCTTGCGTTGCTG 3⬘

VOL. 74, 2000

REPAIR OF cDNA INTEGRATION INTERMEDIATES

11193

these substrates in the presence of Mn2⫹ (11, 49) did not take place in reactions with Mg2⫹, the counterion used here (data not shown). Assays of nicked substrates were identical to the assay of the RSV gapped dumbbell substrate except that oKEY48 was 5⬘ labeled and 20 ␮M dCTP was used instead of [␣-32P]dCTP.

RESULTS Substrate design. To test candidate factors for the ability to repair gapped integration intermediates, a model DNA substrate was synthesized by annealing three oligonucleotides (Fig. 2, top). This yielded a substrate with a five-base gap and a two-base 5⬘ flap. The two-base 5⬘ flap sequence and sequences to the right of the gap as drawn match one end of the HIV cDNA. The rest of the substrate sequence was chosen to mimic host target DNA. Repair should convert the gapped strand to a continuous product the length of the complementary strand. The model substrate was labeled on either end of the gapped strand (Fig. 2, asterisk). Assays with DNA labeled at the 5⬘ end of the top strand monitor both polymerase activity and repair of the substrate. However, some polymerases can carry out robust strand displacement synthesis, resulting in formation of full-length top strands indistinguishable from repair products. To distinguish repair from strand displacement synthesis, we also carried out experiments in which the HIV oligonucleotide was labeled at the 3⬘ end of the top strand. The 3⬘-labeled strand can be converted to a labeled full-length product only if it is joined by gap repair to the 5⬘ oligonucleotide. The sequence of the gapped region was chosen to allow the extent of polymerization to be controlled experimentally. The gap region contains only dCMP and dGMP residues (Fig. 2, top), allowing polymerization to be restricted to the five-base gap by addition of dGTP and dCTP only. Strand displacement synthesis partway into the adjacent duplex is made possible by further addition of dATP. Addition of all four deoxynucleoside triphosphates (dNTPs) places no restriction on polymerization. Gap repair in vitro directed by Pol beta, FEN, and ligase I. We first tested the DNA repair enzymes involved in the late steps of base excision repair, Pol beta, FEN, and ligase I, for the ability to repair the gapped intermediate substrate (29, 44). These repair proteins were added to 32P-labeled gapped substrates individually or in all possible combinations in the presence or absence of nucleotides. Reaction products were separated on denaturing polyacrylamide gels and visualized by phosphorimaging. Addition of 7 ng of FEN, 5 ng of ligase I, or 17 ng of Pol beta individually did not alter the 5⬘-labeled substrate (Fig. 2A, lanes 2 to 4). Addition of Pol beta and the four nucleotides resulted in elongation of the 5⬘-labeled substrate across the gap (lanes 5 to 7). Further polymerization requiring strand displacement synthesis was detected, including extension to the end of the template. Strand displacement synthesis was inefficient, as expected for Pol beta (43, 50). Polymerization was not influenced by the presence of FEN or ligase I added singly (lanes 6 and 7). In the presence of only dGTP, Pol beta added three nucleotides but cannot extend across the gap. Full-length DNA forms are not produced re-

FIG. 2. Repair of model gapped integration intermediates by Pol beta, FEN, and ligase I. The model gapped integration intermediate substrate is composed of three DNA oligonucleotides annealed to form a five-base gap and two-base 5⬘ flap (see Table 1 for sequences). The sequence to the right of the gap as drawn matches 27 nucleotides of HIV-1 cDNA sequence; sequence to the left of the HIV sequence is chosen to model target DNA. Addition of proteins or nucleotides is indicated by ⫹. Reaction conditions are as described in Materials and

Methods. Reaction products were denatured, separated by SDS-PAGE and visualized by phosphorimaging. 32P labels were attached to the substrates on the 5⬘ or 3⬘ end as indicated by the asterisks in the diagram beside each gel. (A) Repair of the 5⬘-labeled gapped substrate. Diagrammed markers indicate the unreacted substrate, the substrate extended by five nucleotides to the end of the gap, and the full-length product. (B) Repair of the 3⬘-labeled gapped substrate. Diagrammed markers indicate the beginning substrate, the substrate shortened by two nucleotides, and the full-length product.

11194

YODER AND BUSHMAN

gardless of the addition of FEN and ligase (Lane 10). The addition of Pol beta, dGTP, and dCTP allows polymerization across the five-base gap. Inclusion of FEN and ligase with Pol beta, dGTP, and dCTP allowed full-length product to accumulate (Fig. 2A, lane 11). Further addition of dATP and dTTP results in a limited amount of strand displacement synthesis by Pol beta and recovery of the full-length product (lanes 12 and 13). Because the substrate is labeled on the 5⬘ end, the full-length DNA strand could have been formed by repair or strand displacement synthesis across the full substrate. Conditions expected to support repair yielded more product, but some full-length product was seen in all reactions containing Pol beta and the four deoxynucleotides. To distinguish gap repair from strand displacement synthesis, we carried out reactions in which the gapped strand was 3⬘ labeled with 32P (Fig. 2B). In the presence of FEN only, slow rate of cleavage of the two nucleotide overhang could be detected (Fig. 2B, lane 2, and data not shown). Incubation with ligase I or Pol beta, with or without dNTPs, did not yield full-length product (lanes 3 to 7). The presence of Pol beta and dNTPs stimulated cleavage of the 5⬘ flap by FEN, either by polymerization across the gap to create a better FEN substrate or by direct interaction of the two proteins. Addition of enzymes in the absence of enough dNTPs to permit polymerase to traverse the gap also failed to yield full-length product (lanes 8 to 10). However, addition of Pol beta, dGTP and dCTP, FEN, and ligase I resulted in formation of full-length product (lane 11). Further addition of dATP and dTTP also allowed full-length product to accumulate (lanes 12 and 13). The reactions with only two or three nucleotides actually resulted in more efficient repair than the reaction with all four (14 to 21% of substrate converted to product for lanes 11 and 12, versus 4% for lane 13). It seems likely that polymerase stalling facilitates cleavage by FEN and subsequent ligation. These data indicate that Pol beta, FEN, and ligase I are sufficient for repair in the presence of sufficient dNTPs to permit polymerization across the gap. Polymerases supporting gap repair. Further polymerases were tested for the ability to substitute in the gap repair reaction. Long-patch base excision repair and gap repair in vivo may also be completed by the host Pol delta or Pol epsilon acting with FEN and ligase I (20, 21, 34). PCNA acts as a cofactor for Pol delta and stimulates its activity by tethering it to the DNA substrate (20, 30). Pol delta and PCNA were tested for the ability to participate in repair of the gapped intermediate substrate. In the presence of dGTP and dCTP, FEN, and ligase I, Pol delta also supported repair (Fig. 3). Addition of PCNA stimulated the reaction 10-fold (Fig. 3A, lanes 2 to 4; Fig. 3B, lanes 2 to 4; 1 to 10% conversion of substrate to the repair product). PCNA is not expected to affect Pol beta, and it stimulated only 2.3-fold in our repair reactions (Fig. 3A, lanes 5 to 7; Fig. 3B, lanes 5 to 7; 13% conversion in lane 5 to 30% conversion in lane 7). The viral RT copurifies with the PIC following reverse transcription, providing another candidate polymerase for repair. We therefore tested the ability of HIV RT to repair the gapped substrate in the presence of FEN and ligase I. Assays of HIV RT on the 5⬘-labeled substrate were complicated by the robust strand displacement synthesis activity of RT. Full-length products are seen in reactions containing just RT and the four nucleotides independent of FEN and ligase I (Fig. 4A, lanes 5 to 7). Full-length products are also seen in complete reactions (lanes 10 to 13). Analysis of the 3⬘-labeled intermediate substrate confirmed that the products seen with only RT and the four nucleotides

J. VIROL.

FIG. 3. Gap repair by Pol delta, PCNA, FEN, and ligase I. Repair reactions were performed with either Pol delta (lanes 2 to 4) or Pol beta (lanes 5 to 7). PCNA was added at 20 ng (lanes 3 and 6) and 800 ng (lanes 4 and 7). Labeling is as in Fig. 2. (A) Repair of 5⬘-labeled gapped substrate; (B) repair of 3⬘-labeled gapped substrate.

VOL. 74, 2000

FIG. 4. Gap repair by RT, FEN, and ligase I. Repair reactions were performed as described for Fig. 2 except using HIV RT to support polymerization. Labeling is as in Fig. 2. (A) Repair of 5⬘-labeled gapped substrate; (B) repair of 3⬘-labeled gapped substrate.

are due to strand displacement synthesis, since 3⬘-labeled fulllength products did not accumulate (Fig. 4B, lanes 4 to 6). However, in complete reactions with the 3⬘-labeled substrate, containing RT, FEN, ligase I, and at least dCTP and dGTP, full-length products did accumulate (lanes 10 to 12; 18% con-

REPAIR OF cDNA INTEGRATION INTERMEDIATES

11195

version in lane 10, 13% conversion in lane 11, and 9% conversion in lane 12). Evidently RT can also support gap repair in the presence of FEN, ligase I, and appropriate nucleotides. These reactions displayed remarkably poor fidelity of RT under the conditions tested. In Fig. 4A, lane 10, for example, polymerization is seen across the gap in the presence of dGTP only, potentially a result of incorporation of two dGTP residues opposite dGMP on the other chain. Alternatively, RT may be scavenging low levels of contaminating dNTPs from the dGTP preparation. However, such polymerization was not seen with the cellular polymerases (Fig. 2A, lane 10), supporting the view that incorporation by RT is more permissive, a point made by many previous studies (13). A previous report identified a polymerase activity associated with HIV integrase possibly involved in gap repair. Integrase was modified to contain a His tag, expressed in bacteria, and purified by nickel-chelating Sepharose (1). These authors identified a cofractionating polymerase activity that they proposed was contributed by the integrase polypeptide itself. We purified His-tagged integrase using nickel-chelating Sepharose and tested it for polymerase activity. Addition of such an integrase fraction to a 5⬘-labeled gapped intermediate substrate in the presence of labeled nucleotides yielded polymerization products (data not shown). Strand displacement synthesis was not seen, as previously described for integrase-associated polymerase activity (1). To test for bacterial polymerases cofractionating on nickel-chelating Sepharose, we assayed another protein, MCV topo, that was modified to contain a His tag, expressed, and purified in a similar fashion (25). We also detected a polymerase activity in these fractions that did not carry out strand displacement synthesis efficiently, as in the integrase preparation. Preparations of integrase and MCV topo lacking His tags purified through several steps including ion-exchange chromatography instead of nickel-chelating Sepharose did not yield polymerase activity though the enzymes retained high levels of the expected activities (data not shown). We conclude that our preparations of integrase or topoisomerase that were purified by nickel-chelating Sepharose contained a copurifying bacterial polymerase. Ligases supporting gap repair. Ligase I is a candidate for repair of the gapped integration intermediate in vivo because this protein is ubiquitously expressed and known to be involved in base excision repair. However, two other mammalian ligases, ligase III and ligase IV, are known to be involved in repair pathways (for reviews, see references 47 and 48). Ligase IV requires the cofactor XRCC4, while ligase III (beta form) does not require any cofactors. We tested ligase III and ligase IV/XRCC4 for the ability to support repair of the gapped intermediate substrate with FEN and Pol beta or RT. Ligases I, III, and IV/XRCC4 were all able to support repair of the gapped substrate (Fig. 5). Ligase IV and XRCC4 are involved in nonhomologous end joining, but no end-joining products were detected (data not shown). This is as expected from previous work, which indicated that low DNA concentrations such as those used here are not sufficient for end joining (37). Lack of stimulation by Ku70/80. In end joining assays, the Ku70/80 heterodimer is known to stimulate ligase IV/XRCC4 activity. Cells mutant in Ku70/80 have also been found to show altered responses to retroviral infection (17). We tested the ability of the Ku70/80 heterodimer to stimulate gap repair in the presence of Pol beta, nucleotides, FEN, and either ligase I or ligase IV/XRCC4 (Fig. 6). Addition of 1 ng of Ku70/80 heterodimer, a concentration sufficient for supporting end joining (37), had little effect on product formation (Fig. 6, lanes 3 to 5 and 7 to 9). At high concentrations (10 ng), ligation by both ligases was inhibited (lanes 5 and 9; also data not shown).

11196

YODER AND BUSHMAN

FIG. 5. Ligases I, III, and IV/XRCC4 all support gap repair. Repair reactions were performed with either Pol beta (lanes 2 to 4) or HIV RT (lanes 5 to 7) and ligase I (lanes 2 and 5), ligase III (lanes 3 and 6), and ligase IV/XRCC4 (lanes 4 and 7). Labeling is as in Fig. 2. (A) Repair of 5⬘-labeled gapped substrate; (B) repair of 3⬘-labeled gapped substrate. Percent conversion values for lanes 2 to 7 were, respectively, 25, 36, 53, 2, 2, and 3%.

J. VIROL.

FIG. 6. Inhibition of gap repair by Ku70/80. Repair reactions were conducted in the presence of either ligase I (lanes 2 to 5) or ligase IV (lanes 6 to 9). The Ku70/80 heterodimer was added in increasing concentrations of 100 pg (lanes 3 and 7), 1 ng (lanes 4 and 8), and 10 ng (lanes 5 and 9). Labeling is as in Fig. 2; each panel contains lanes from the same autoradiogram. (A) Repair of 5⬘labeled gapped substrate; (B) repair of 3⬘-labeled gapped substrate. Percent conversion values for lanes 2 to 9 were, respectively, 44, 40, 36, 2, 48, 57, 53, and 18%.

VOL. 74, 2000

Reactions with ligase IV were inhibited 2.7-fold, and reactions with ligase I were inhibited 22-fold. The effect is only on the ligation step of repair, since polymerization and FEN cleavage products accumulated. Thus, Ku70/80 did not detectably stimulate repair under the conditions tested. Lack of gap repair by RT and integrase. An alternative pathway for repair of the postintegration gaps involving RT and integrase has been suggested (Fig. 1B) (12). In this clever model, RT would first synthesize across the gap sequence; then integrase would catalyze the cleavage of the 5⬘ flap and ligation of the strands in a single-step reaction. This reaction is similar to the reverse reaction of integration, termed disintegration, which has been described for integrase in vitro (12). To test this pathway, RT and integrase were added to the gapped intermediate substrate (Fig. 7). The 5⬘-labeled substrate showed polymerization by RT in the presence or absence of integrase, due to the strong strand displacement activity of RT. Assays with the 3⬘-labeled substrate yielded no labeled full-length product, indicating that the combination of RT and integrase did not suffice for repair under these conditions (Fig. 7B). Positive control reactions with Pol beta, FEN, and ligase I run in parallel yielded the usual repair products (data not shown). Reaction mixtures containing Mn2⫹ instead of Mg2⫹ have been found to favor disintegration (reference 12 and our unpublished observations). In an effort to obtain repair by RT and integrase, reactions were also carried out with Mn2⫹. Repair was not detected under these conditions either (data not shown). These data do not support the idea that RT and integrase can carry out repair of the gapped integration intermediate. Integrase binding blocks access of polymerases to DNA gaps. Integrase binds stably to the viral cDNA ends in PICs (10, 35, 51), suggesting that integrase may remain bound to integration products as well. Thus, it may be necessary for integrase to be actively removed from integration products to permit repair. Such a specific disassembly step has been characterized for phage Mu, which carries out DNA transposition by a closely related pathway (31, 32). We have investigated whether integrase blocks the access of repair factors to the gapped intermediate following covalent joining. For these experiments we used the RSV integrase protein since it is more soluble than HIV integrase and potentially less prone to nonspecific DNA binding. To assess the ability of integrase to block repair, a new substrate having two arms of RSV sequence and two arms of sequence mimicking host target sequence was synthesized (Fig. 8A) (11, 49). Two different DNAs were synthesized with complementary single-stranded gaps, allowing formation of the annealed molecule in Fig. 8A. The termini of the DNA arms were connected by hairpins for convenience in preparation of the substrates. Blocking of polymerase by integrase can be assessed by prebinding integrase to the RSV gapped dumbbell substrate prior to addition of polymerase. The gapped dumbbell substrate was not labeled. Addition of 32P-labeled dCTP and a polymerase should reveal a product if the dCTP is incorporated into the substrate opposite the first dGMP residue in the gap. Labeled DNA products were denatured, separated by SDS-PAGE, and visualized by phosphorimaging. Pol beta directed incorporation into the gapped substrate as expected (Fig. 8B). The highly purified preparation of RSV integrase did not show detectable polymerase activity (data not shown). Titration of integrase into the polymerization reactions resulted in a 50-fold reduction of polymerization at the highest integrase concentration (Fig. 8B, lanes 2 to 6). Increasing concentrations of BSA did not inhibit the ability of Pol beta to incorporate labeled nucleotides (lanes 7 to 11). Thus, pre-

REPAIR OF cDNA INTEGRATION INTERMEDIATES

11197

FIG. 7. Lack of gap repair by HIV-1 RT and integrase. Reactions were carried out in the presence of HIV RT only (lanes 2 to 6) or in the presence of both HIV RT and HIV integrase (lanes 7 to 11). Labeling is as in Fig. 2. (A) Repair of 5⬘-labeled gapped substrate. (B) repair of 3⬘-labeled gapped substrate.

bound integrase can block access of a polymerase to the DNA gap. Addition of integrase and a polymerase simultaneously did not block access of the polymerase to the substrate (Fig. 7 and data not shown). To test the ability of integrase to block polymerase activity generally, the influence of integrase on polymerization at a

11198

YODER AND BUSHMAN

J. VIROL.

DNA nick was tested (Fig. 8C). For this experiment the 5⬘ end of the top strand was labeled, and extension products were visualized as slower-migrating DNA forms. Addition of Pol beta and dNTPs resulted in incorporation of one dNMP and weak incorporation of additional dNMPs. Addition of integrase (Fig. 8C, lanes 2 to 6) had no effect at low concentrations but resulted in about twofold reduction at the highest concentration tested, indicating preferential binding of integrase to the branched substrate (E. Johnson and F. D. Bushman, unpublished data). Similarly, addition of BSA showed little effect on incorporation (Fig. 8C, lanes 7 to 11). These data support the idea that repair of integration junctions in vivo may require a specific disassembly step to remove integrase. DISCUSSION The pathway of retroviral cDNA integration, like transposition of many transposons, involves gapped DNA intermediates. The element-encoded integrase or transposase carries out only the early DNA breaking and joining steps necessary for attachment of element DNA to target DNA. The host cell is forced to complete the reaction, since cellular DNA replication cannot proceed through the gapped intermediate. Here we describe studies that identify a set of host cell DNA repair proteins that are capable of processing integration intermediates in vitro to yield fully double stranded junctions. Initially, only one DNA strand from the element is attached to host DNA at each host-element junction. The intermediate also contains a 5⬘ unpaired flap derived from the cDNA end. Assays with this repair system indicate that integrase protein can block repair in vitro, indicating that a specific disassembly step may be required to allow access of repair enzymes. We find that any of several combinations of a polymerase, nuclease, and ligase suffice for repair of the gapped integration intermediate substrate. DNA repair pathways such as base excision repair and mismatch repair involve related gap repair steps (20, 34). Connection of Okazaki fragments generated during DNA synthesis also requires related activities (30). Work in the DNA repair field has indicated that these pathways are often redundant (e.g., reference 29). For example, it is thought that both Pol beta and Pol delta/PCNA can support base excision repair (21, 29). We find that these polymerases can perform repair in our reactions in vitro as well. Similarly, ligases I, III, and IV/XRCC4 all supported the ligation step. For DNA repair, nuclease requirements differ depending on the pathway involved. We have tested FEN only, since it possesses the required 5⬘-to-3⬘ nuclease activity. We note the recent identification of another eukaryotic 5⬘ exonuclease, Exo I, that may be involved in mismatch repair (40, 46, 52). It will be interesting to determine whether Exo I can also participate in repair of model gaps in vitro. FEN has also been implicated in repair of another DNA structure found in HIV, DNA flaps within the cDNA resulting from internal initiation of reverse

FIG. 8. Blocking of polymerase access to DNA gaps by added integrase. (A) A four-arm substrate was synthesized with two arms matching RSV cDNA sequences (bold lines) and two arms of mixed sequence recapitulating target DNA (thin lines). Two separate molecules with complementary gap sequences were studied to permit assembly of the four-armed structure. Addition of polymerase and a labeled nucleotide (asterisks) results in incorporation of radioactive nucleotides in the gap. Addition of integrase, in contrast, can potentially block access. (B) Polymerization on the RSV gapped dumbbell substrate. Labeled DNA products indicates incorporation of 32P-labeled dCTPs in the dumbbell substrate by Pol beta. RSV integrase was preincubated for 10 min with the

substrate in increasing amounts; lanes 2 to 6 contained 0.1, 1, 10, 50, and 100 ng, respectively. BSA was similarly preincubated with the substrate in lanes 7 to 11 in increasing amounts (0.1, 1, 10, 50, and 100 ng, respectively). The preincubation was followed by the addition of Pol beta and further incubation for 10 min. The amounts of relative incorporation in lanes 2 to 6, respectively, were 100, 108, 34, 15, and 2%. The amounts of relative incorporation in lanes 7 to 11, respectively, were 100, 107, 91, 321, and 108%. (C) Polymerization on a 5⬘-labeled nicked substrate. The diagrams at the left indicate the unreacted nicked substrate and a strand displacement synthesis product. Reaction compositions were the same as for panel B except for the DNA substrate. The relative percent conversion values for lanes 2 to 6 were, respectively, 100, 113, 115, 87, and 58%. The relative percent conversion values for lanes 7 to 11 were, respectively, 100, 88, 152, 127, and 126%.

VOL. 74, 2000

REPAIR OF cDNA INTEGRATION INTERMEDIATES

transcription (38). The redundancy of the activities required for repair, together with their requirement in cellular replication pathways, may explain why genetic methods have not identified these functions. Our findings suggest that repair of the integration intermediate involves action of polymerase prior to FEN nuclease. FEN cleaved the 5⬘ flap only inefficiently on an unrepaired substrate in which the flap was adjacent to the DNA gap. FEN acted much more efficiently when the flap protruded from continuous double-helical DNA, consistent with previous studies (33). Thus, we favor a model in which the polymerase acts first, polymerizing across the gap and thereby creating an efficient substrate for FEN. Pol beta is reported to function most efficiently when the DNA to the 5⬘ side of the gap is phosphorylated (43, 44), a modification lacking in our substrates. Our substrates do contain two nucleotide flaps on the 5⬘ side, possibly substituting for a 5⬘ phosphate. We do not favor the idea that the sequential action of RT and integrase repairs the DNA gaps (Fig. 1B), since we could not detect function of this pathway under any of the conditions tested. We do find that RT can carry out DNA synthesis across the gaps and complete repair in conjunction with FEN and a ligase; thus, it is a candidate for a repair factor. It will be important to retest the action of RT and integrase in reaction products made with PICs, where assembly may differ, but evidence to date indicates that this pathway is unlikely to be active. Studies of retroviral infection of cultured cells have led to the proposal that two other DNA repair functions, PARP and the DNA-PK complex, may play a role in repair of integration intermediates. The DNA-PK complex is known to be important for joining of double-stranded DNA breaks, but it is not thought to be involved in the covalent chemistry of gap repair. Ligase IV and XRCC4, components of the DNA-PK complex, are competent to complete gap repair in the presence of a polymerase and FEN nuclease, indicating a possible connection between the DNA-PK complex and reactions reported here. The heterodimer of Ku70/80 did not promote repair and in fact inhibited ligation at high concentrations. The relationship of the DNA-PK complex and PARP to the gap repair activities reported here will require further study. For example, the DNA-PK complex and/or PARP might provide scaffolding or regulatory functions for the gap repair enzymes. Repair proteins must first achieve access to the gaps in integration intermediates for repair to take place. We find that integrase is capable of inhibiting the ability of a polymerase to act on a model gap, suggesting that integrase must be removed prior to repair. Studies with PICs show that integrase is stably associated with the viral cDNA, bolstering the idea that integrase must be actively removed from integration intermediates. This is reminiscent of recent findings on transposition by bacteriophage Mu. The MuA transposase protein, which carries out reactions related to those of retroviral integration, remains tightly associated with transposition intermediates (3, 4, 45). MuA must be removed by action of the ClpX protein chaperone in an ATP-dependent fashion to permit completion of Mu transposition (31, 32). Our data suggest a similar model for repair of HIV integration intermediates. Further studies of repair of integration products made with PICs should help identify additional protein factors responsible for completing integration in vivo. ACKNOWLEDGMENTS We thank C. Hyde, S. Wilson, M. Park, B. Stillman, A. Tomkinson, C. Tan, and D. Ramsden for providing proteins and members of the Bushman laboratory for suggestions and comments on the manuscript.

11199

K.E.Y. was supported in part by the Chapman Foundation. This work was supported by NIH grants GM56553 and AI34786 to F.D.B., the James B. Pendleton Charitable Trust, the Berger Foundation, and Cornelia Mackey. F.D.B. is a Scholar of the Leukemia and Lymphoma Society of America. REFERENCES 1. Acel, A., B. E. Udashkin, M. A. Wainberg, and E. A. Faust. 1998. Efficient gap repair catalyzed in vitro by an intrinsic DNA polymerase activity of human immunodeficiency virus type 1 integrase. J. Virol. 72:2062–2071. 2. Allen, P., S. Worland, and L. Gold. 1995. Isolation of high-affinity RNA ligands to HIV-1 integrase from a random pool. Virology 209:327–336. 3. Baker, T. A., E. Kremenstova, and L. Luo. 1994. Complete transposition requires four active monomers in the Mu transposase tetramer. Genes Dev. 8:2416–2428. 4. Baker, T. A., M. Mizuuchi, H. Savilahti, and K. Mizuuchi. 1993. Division of labor among monomers within the Mu transposase tetramer. Cell 74:723– 733. 5. Berg, D. E., and M. M. Howe (ed.). 1989. Mobile DNA. American Society for Microbiology, Washington, D.C. 6. Bukrinsky, M. I., N. Sharova, T. L. McDonald, T. Pushkarskaya, G. W. Tarpley, and M. Stevenson. 1993. Association of integrase, matrix, and reverse transcriptase antigens of human immunodeficiency virus type 1 with viral nucleic acids following acute infection. Proc. Natl. Acad. Sci. USA 90:6125–6129. 7. Bushman, F. D., and R. Craigie. 1991. Activities of human immunodeficiency virus (HIV) integration protein in vitro: specific cleavage and integration of HIV DNA. Proc. Natl. Acad. Sci. USA 88:1339–1343. 8. Bushman, F. D., A. Engelman, I. Palmer, P. Wingfield, and R. Craigie. 1993. Domains of the integrase protein of human immunodeficiency virus type 1 responsible for polynucleotidyl transfer and zinc binding. Proc. Natl. Acad. Sci. USA 90:3428–3432. 9. Bushman, F. D., T. Fujiwara, and R. Craigie. 1990. Retroviral DNA integration directed by HIV integration protein in vitro. Science 249:1555–1558. 10. Chen, H., and A. Engelman. 1998. The barrier-to-autointegration protein is a host factor for HIV type 1 integration. Proc. Natl. Acad. Sci. USA 95: 15270–15274. 11. Chow, S. A., and P. O. Brown. 1994. Juxtaposition of two viral DNA ends in a bimolecular disintegration reaction mediated by multimers of human immunodeficiency virus type 1 or murine leukemia virus integrase. J. Virol. 68:7869–7878. 12. Chow, S. A., K. A. Vincent, V. Ellison, and P. O. Brown. 1992. Reversal of integration and DNA splicing mediated by integrase of human immunodeficiency virus. Science 255:723–726. 13. Coffin, J. M., S. H. Hughes, and H. E. Varmus. 1997. Retroviruses. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 14. Coffin, J. M., and N. Rosenberg. 1999. Retroviruses. Closing the joint. Nature 399:413–416. 15. Craigie, R., T. Fujiwara, and F. Bushman. 1990. The IN protein of Moloney murine leukemia virus processes the viral DNA ends and accomplishes their integration in vitro. Cell 62:829–837. 16. Craigie, R., and K. Mizuuchi. 1985. Mechanism of transposition of bacteriophage Mu: structure of a transposition intermediate. Cell 41:867–876. 17. Daniel, R., R. A. Katz, and A. M. Skalka. 1999. A role for DNA-PK in retroviral DNA integration. Science 284:644–647. 18. Farnet, C., and F. D. Bushman. 1997. HIV-1 cDNA integration: requirement of HMG I(Y) protein for function of preintegration complexes in vitro. Cell 88:1–20. 19. Farnet, C. M., and W. A. Haseltine. 1991. Determination of viral proteins present in the human immunodeficiency virus type 1 preintegration complex. J. Virol. 65:1910–1915. 20. Friedberg, E. C., G. C. Walker, and W. Siede. 1995. DNA repair and mutagenesis. ASM Press, Washington, D.C. 21. Frosina, G., P. Fortini, O. Rossi, F. Carrozzino, G. Raspaglio, L. S. Cox, D. P. Lane, A. Abbondandolo, and E. Dogliotti. 1996. Two pathways for base excision repair in mammalian cells. J. Biol. Chem. 271:9573–9578. 22. Gaken, J. A., M. Tavassoli, S.-U. Gan, S. Villian, I. Giddings, D. C. Darling, J. Galea-Lauri, M. G. Thomas, H. Abedi, V. Schreiber, H. Menissier de Murcia, M. K. L. Collins, S. Shall, and F. Farzaneh. 1996. Efficient retroviral infection of mammalian cells is blocked by inhibition of poly(ADP-ribose) polymerase activity. J. Virol. 70:3992–4000. 23. Hansen, M. S. T., S. Carteau, C. Hoffmann, L. Li, and F. Bushman. 1998. Retroviral cDNA integration: mechanism, applications and inhibition, p. 41–62. In J. K. Setlow (ed.), Genetic engineering. Principles and Methods, vol. 20. Plenum Press, New York, N.Y. 24. Harrington, J. J., and M. R. Lieber. 1994. The characterization of a mammalian DNA structure-specific endonuclease. EMBO J. 13:1235–1246. 25. Hwang, Y., B. Wang, and F. D. Bushman. 1998. Molluscum contagiosum virus topoisomerase: purification, activities, and response to inhibitors. J. Virol. 72:3401–3406. 26. Katz, R. A., G. Merkel, J. Kullkosky, J. Leis, and A. M. Skalka. 1990. The avian retroviral IN protein is both necessary and sufficient for integrative

11200

YODER AND BUSHMAN

recombination in vitro. Cell 63:87–95. 27. Katzman, M., R. A. Katz, A. M. Skalka, and J. Leis. 1989. The avian retroviral integration protein cleaves the terminal sequences of linear viral DNA at the in vivo sites of integration. J. Virol. 63:5319–5327. 28. Kim, K., S. Biade, and Y. Matsumoto. 1998. Involvement of flap endonuclease 1 in base excision DNA repair. J. Biol. Chem. 273:8842–8848. 29. Klungland, A., and T. Lindahl. 1997. Second pathway for completion of human DNA base excision-repair: reconstitution with purified proteins and requirement for DNase IV (FEN1). EMBO J. 16:3341–3348. 30. Kornberg, A., and T. Baker. 1991. DNA replication. W. H. Freeman and Company, New York, N.Y. 31. Levchenko, I., L. Luo, and T. A. Baker. 1995. Disassembly of the Mu transposase tetramer by the ClpX chaperone. Genes Dev. 9:2399–2408. 32. Levchenko, I., M. Yamauchi, and T. A. Baker. 1997. ClpX and MuB interact with overlapping regions of Mu transposase: implications for control of the transposition pathway. Genes Dev. 11:1561–1572. 33. Lieber, M. R. 1997. The FEN-1 family of structure-specific nucleases in eukaryotic DNA replication, recombination and repair. Bioessays 19:233– 240. 34. Lindahl, T., and R. D. Wood. 1999. Quality control by DNA repair. Science 286:1897–1905. 35. Miller, M. D., C. M. Farnet, and F. D. Bushman. 1997. Human immunodeficiency virus type 1 preintegration complexes: studies of organization and composition. J. Virol. 71:5382–5390. 36. Miller, M. D., B. Wang, and F. D. Bushman. 1995. Human immunodeficiency virus type 1 preintegration complexes containing discontinuous plus strands are competent to integrate in vitro. J. Virol. 69:3938–3944. 37. Ramsden, D. A., and M. Gellert. 1998. Ku protein stimulates DNA end joining by mammalian DNA ligases: a direct role for Ku in repair of DNA double-strand breaks. EMBO J. 17:609–614. 38. Rumbaugh, J. A., G. M. Fuentes, and R. A. Bambara. 1998. Processing of an HIV replication intermediate by the human DNA replication enzyme FEN1. J. Biol. Chem. 273:28740–28745. 39. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Press, Cold Spring Harbor, N.Y. 40. Schmutte, C., R. C. Marinescu, M. M. Sadoff, S. Guerrette, J. Overhauser, and R. Fishel. 1998. Human exonuclease I interacts with the mismatch repair protein hMSH2. Cancer Res. 58:4537–4542.

J. VIROL. 41. Shapiro, J. 1979. A molecular model for the transposition and replication of bacteriophage Mu and other transposable elements. Proc. Natl. Acad. Sci. USA 76:1933. 42. Sherman, P. A., and J. A. Fyfe. 1990. Human immunodeficiency virus integration protein expressed in Escherichia coli possesses selective DNA cleaving activity. Proc. Natl. Acad. Sci. USA 87:5119–5123. 43. Singhal, R. K., and S. H. Wilson. 1993. Short gap-filling synthesis by DNA polymerase beta is processive. J. Biol. Chem. 268:15906–15911. 44. Sobol, R. W., J. K. Horton, R. Kuhn, H. Gu, R. K. Singhal, R. Prasad, K. Rajewsky, and S. H. Wilson. 1996. Requirement of mammalian DNA polymerase beta in base excision repair. Nature 379:183–186. 45. Surette, M. G., S. J. Buch, and G. Chaconas. 1987. Transpososomes: stable protein-DNA complexes involved in the in vitro transposition of bacteriophage Mu DNA. Cell 49:235–262. 46. Tishkoff, D. X., N. S. Amin, C. S. Viars, K. C. Arden, and R. D. Kolodner. 1998. Identification of a human gene encoding a homologue of Saccharomyces cerevisiae EXO1, an exonuclease implicated in mismatch repair and recombination. Cancer Res. 58:5027–5031. 47. Tomkinson, A. E., and D. S. Levin. 1997. Mammalian DNA ligases. Bioessays 19:893–901. 48. Tomkinson, A. E., and Z. B. Mackey. 1998. Structure and function of mammalian DNA ligases. Mutat. Res. 407:1–9. 49. Vink, C., R. A. Lutzke, and R. H. Plasterk. 1994. Formation of a stable complex between the human immunodeficiency virus integrase protein and viral DNA. Nucleic Acids Res. 22:4103–4110. 50. Wang, T. S.-F., and D. Korn. 1980. Reactivity of KB cell deoxyribonucleic acid polymerases alpha and beta with nicked and gapped deoxyribonucleic acid. Biochemistry 19:1782–1790. 51. Wei, S.-Q., K. Mizuuchi, and R. Craigie. 1997. A large nucleoprotein assembly at the ends of the viral DNA mediates retroviral DNA integration. EMBO J. 16:7511–7520. 52. Wilson, D. M., J. P. Carney, M. A. Coleman, A. W. Adamson, M. Christensen, and J. E. Lamerdin. 1998. Hex1: a new human Rad2 nuclease family member with homology to yeast exonuclease 1. Nucleic Acids Res. 26:3762– 3768. 53. Yang, Z. N., T. C. Mueser, F. D. Bushman, and C. C. Hyde. 2000. Crystal structure of an active two-domain derivative of Rous sarcoma virus integrase. J. Mol. Biol. 296:535–548.