Modeling the Late Steps in HIV-1 Retroviral Integrase-catalyzed DNA Integration*

Model oligodeoxyribonucleotide substrates representing viral DNA integration intermediates with a gap and a two-nucleotide 5′ overhang were used to examine late steps in human immunodeficiency virus, type 1 (HIV-1) retroviral integrase (IN)-catalyzed DNA integration in vitro. HIV-1 or avian myeloblastosis virus reverse transcriptase (RT) were capable of quantitatively filling in the gap to create a nicked substrate but did not remove the 5′ overhang. HIV-1 IN also failed to remove the 5′ overhang with the gapped substrate. However, with a nicked substrate formed by RT, HIV-1 IN removed the overhang and covalently closed the nick in a disintegration-like reaction. The efficiency of this closure reaction was very low. Such closure was not stimulated by the addition of HMG-(I/Y), suggesting that this protein only acts during the early processing and joining reactions. Addition of Flap endonuclease-1, a nuclease known to remove 5′ overhangs, abolished the closure reaction catalyzed by IN. A series of base pair inversions, introduced into the HIV-1 U5 long terminal repeat sequence adjacent to and/or including the conserved CA dinucleotide, produced no or only a small decrease in the HIV-1 IN-dependent strand closure reaction. These same mutations caused a significant decrease in the efficiency of concerted DNA integration by a modified donor DNA in vitro, suggesting that recognition of the ends of the long terminal repeat sequence is required only in the early steps of DNA integration. Finally, a combination of HIV-1 RT, Flap endonuclease-1, and DNA ligase is capable of quantitatively forming covalently closed DNA with these model substrates. These results support the hypothesis that cellular enzyme(s) may catalyze the late steps of retroviral DNA integration.

Insertion of viral DNA into the host genome is an obligatory step in the retroviral life cycle and requires a concerted mechanism that brings the two LTR 1 ends of a single donor into a complex with IN and the host DNA. HIV IN is associated with DNA binding, specific DNA endonuclease activity, DNA joining, and disintegration (see Refs. 1 and 2). The mechanism for insertion of a donor DNA into an acceptor comprises IN processing the 3Ј ends of viral LTRs, which are subsequently involved in a nucleophilic attack of phosphodiester bonds in the host DNA to covalently join the two DNAs. These reactions are catalyzed by IN alone (3,4) but are stimulated by cellular HMG-(I/Y) (5,6) and the viral NC protein (7). During the strand transfer reaction, the host DNA is cut in a staggered manner to create small gaps between the 5Ј ends of the viral DNA and 3Ј ends of host DNA. The repair of these gaps at both ends produces the direct repeats that flank the inserted viral DNA. In the case of HIV-1, the size of these duplications is 5 base pairs (8,9). During the initial processing only the 3Ј ends of the LTRs are shortened by two nucleotides. The complementary nucleotides on the opposite LTR strand form a 5Ј twonucleotide overhang after the strand transfer reaction. These overhangs have to be removed after the gaps are repaired so that the covalent joining of the cellular and viral DNA can be completed. The precise enzymes responsible for these reactions are not known, although it was reported that if presented with a nicked substrate, wild type IN or its central domain (10) is capable of removing the 5Ј overhang and sealing the nick in a disintegration-like reaction (2,10,11).
Disintegration is a polynucleotidyl transfer reaction that is relatively independent of viral DNA sequence; DNA structure rather than sequence (12, 13) is the major determinant of the site of nucleophilic attack. Although the processing and joining steps require a full-length IN protein, disintegration can be performed by the isolated catalytic domain (14). This suggests that the catalytic domain includes the active site for polynucleotidyl transfer and is consistent with the crystal structure determined for this domain, which revealed a five-strand ␤-sheet and six ␣-helices similar to other polynucleotidyl transfer enzymes (15).
In this report we demonstrate with model substrates that the combination of RT and IN is sufficient to complete all of the late steps required for integration: filling in the gap, removing the 5Ј overhang, and joining the second host DNA strand to the viral DNA. However, this reaction takes place with very low efficiency. We have also examined the ability of cellular proteins such as PCNA, HMG-(I/Y), or Flap endonuclease-1 (FEN-1) (16) to work in concert with RT and/or IN to increase the efficiency of the reaction. None of these enzymes stimulated IN-dependent closure of a nicked substrate with an overhang, whereas the addition of FEN-1 prevented this reaction. As expected from their known specificities, FEN-1 in combination with RT and DNA ligase quantitatively produced strand closure of gapped substrates with 5Ј two-nucleotide overhangs.
Preparation of Model Substrates-Model HIV-1 gapped substrates (see Fig. 1B) were prepared by annealing strands A, B, and D. Nicked HIV-1 or ASV substrates were formed by substituting the respective strand C for D. For the HIV-1 mutant LTR substrates 4, 5, and 6 (see Fig. 1A), the respective mutated strands A and B were substituted for the wild type strands A and B. Oligos were labeled either at the 5Ј end (strands B or C) or at the 3Ј end (strand B). The 5Ј end was labeled with [␥-33 P]ATP and T4 polynuleotide kinase (10 units/50 pmol of oligo). The 3Ј end was labeled with [␣-32 P]ddATP and TdT as described by the manufacturer. Labeled oligos were purified by 20% polyacrylamide gel electrophoresis. The labeled strand was annealed with a 3-fold molar excess of unlabeled complimentary strands in 10 mM Tris-HCl, pH 8.0, by heating to 95°C for 5 min followed by slow cooling to room temperature. Annealing was confirmed by the change in migration of the labeled oligos as analyzed by electrophoresis in nondenaturing polyacrylamide gels. A 3-fold excess of the unlabeled complimentary strands resulted in Ͼ95% duplex formation (data not shown).
Preparation of Oligonucleotide Ladders-Size markers for gel analysis are prepared by partial digestion of 5Ј 33 P-labeled oligodeoxyribonucleotides (39-mer and 21-mer) with snake venom phosphodiesterase, as described (19).
Bacterial Strains and Growth Conditions-Escherichia coli DH5␣ (Life Technologies, Inc.) and MC1061/P3 (Invitrogen) strains were used for these studies. MC1061/P3 is a derivative of MC1061 containing the episome, P3, which can be selected by the presence of an encoded Kan r gene. In addition, P3 possesses amp (Am) and tet (Am) genes. The expression of these genes can be rescued by the supF amber suppressor tRNA. Under these conditions, MC1061/P3 can be selected for ampicillin, tetracycline, and kanamycin resistance.
Plasmid Constructions and Preparations-Plasmid pHHIV2, which was used in this study as a template to amplify donor DNA, is a variation of pBCSK ϩ in which a wild type HIV-1 donor DNA polymerase chain reaction product was inserted into pBCSK ϩ catalyzed by IN, resulting in the loss of 2 base pairs from the LTR ends. This plasmid was propagated in E. coli MC1061/P3 under the conditions described above. The integration acceptor was plasmid pBCSK ϩ (Stratagene, La Jolla, CA), which was propagated in E. coli DH5␣. Plasmids were purified with Qiaprep columns (Qiagen, Chatsworth, CA) according to manufacturer's instructions. The growth of DH5␣ containing pBCSK ϩ was selected for by addition of chloramphenicol (35 g/ml).
Preparation of Donor DNAs-Integration donor DNA were amplified by using thermostable Vent DNA polymerase and the primers listed above. 25 pmol of each primer and 50 ng of pHHIV2 DNA as the template were used for each polymerase chain reaction. Vent DNA polymerase was used according to manufacturer's instructions. A total of 20 rounds of amplification were performed in each reaction. The amplification conditions were 94°C for 2 min, 50°C for 1 min, and 72°C for 1 min for three rounds. This was followed by amplification conditions that used 94°C for 2 min, 57°C for 1 min, and 72°C for 45 s for 17 additional rounds. The resultant product donor DNA was isolated after electrophoresis through 2% agarose gels equilibrated with 0.5ϫ Tris borate-EDTA (6). The purified DNA (600 ng) was recovered using Qiaex-II resin (Qiagen) and then precipitated with ethanol. The recovered DNA was suspended in either TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA) or deionized distilled water. The integration donors, which were approximately 300 base pairs in length, were internally labeled during the polymerase chain reaction by the inclusion of [␣-32 P]dCTP. The final concentrations of deoxyribonucleoside triphosphates during amplification reactions were 0.25 mM each of unlabeled dATP, dGTP, and dTTP. The final dCTP concentration was 0.0502 mM (12 Ci/mmol, 0.6 mCi/ml).
Flap Endonuclease Assay-Overhang cleavage activity of FEN-1 was examined in a 15-l reaction containing 50 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 0.5 mM ␤ME, 145 g/ml BSA, and 1 pmol of substrate. Where indicated, 5 ng of FEN-1, 500 ng of PCNA, 5 pmol of IN, or 110 g of HIV-1 RT and 10 M deoxyribonucleotides were added. The reaction was terminated by the addition of 16 l of 95% formamide, 10 mM EDTA, 0.1 mg/ml bromphenol blue, and 0.1 mg/ml xylene cyanole. The tubes were heated to 95°C for 5 min and loaded onto a 20% denaturing polyacrylamide gel. The reaction products were visualized by autoradiography.
Gap Repair-The coupled gap repair by RT and overhang removal by FEN-1 was assayed in a 15-l reaction containing 50 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 0.5 mM ␤ME, 145 g/ml BSA, 10 M deoxyribonucleotides, and 1 pmol of substrate. Where indicated, 5 ng of FEN-1, T4 DNA ligase (400 units), and T4 DNA ligase buffer, HIV-1 RT (110 g), or AMV RT (2 units) were added. The reaction mixture was incubated at 37°C for 30 min. The reaction was terminated by the addition of 16 l of 95% formamide, 10 mM EDTA, 1 mg/ml bromphenol blue, and 1 mg/ml xylene cyanole. The reaction products were analyzed by electrophoresis using denaturing 20% polyacrylamide gels. Bands were detected by autoradiography.
IN Cleavage/Ligation Reactions-Cleavage/ligation activity of IN was examined in a 15-l reaction containing 50 mM Tris-HCl, pH 8.0, 10 mM MnCl 2 , 0.5 mM ␤ME, 145 g/ml BSA, 6% dimethyl sulfoxide, and 10 pmol of substrate and HIV-1 IN. HMG-(I/Y) (200 ng) was added as indicated. Reaction products were extracted with phenol-chloroform, and oligos were precipitated from ethanol. The products were analyzed by polyacrylamide gel electrophoresis as described above.
RT and IN Repair Assays-The repair of model substrates by viral enzymes was examined in a 15-l reaction containing 50 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 0.5 mM ␤ME, 145 g/ml BSA, 10 M dNTPs, 10 pmol of substrate, and HIV-1 RT or AMV RT. The reaction mixture was incubated for 30 min at 37°C followed by a 30-min drop dialysis against 50 mM Tris-HCl, pH 8.0, using floating nitrocellulose filters (type US, 0.025 m) as described (20). 10 l of the reaction was recovered after dialysis, to which was added 5 l containing 10 mM MgCl 2 , 0.5 mM ␤ME, 145 g/ml BSA, 6% dimethyl sulfoxide, and 10 pmol of HIV-1 IN as indicated. The final reaction mixture was incubated at 37°C for 30 min. The reaction was terminated by addition of 16 l of 95% formamide, 10 mM EDTA, 1 mg/ml bromphenol blue, 1 mg/ml xylene cyanole, and reaction products were analyzed by electrophoresis using 20% denaturing polyacrylamide gels as described above.
Integration Reactions-The concerted integration reaction conditions were similar to those described by Hindmarsh et al. (6). Briefly, 15 ng (0.15 pmol of ends) of donor DNA was mixed with 50 ng of acceptor DNA (0.02 pmol) and 175 ng of HIV-1 IN (2.5 pmol) in a 8.5-l preincubation reaction mixture containing, at final concentrations, 25 mM MOPS, pH 7.2, 23 mM NaCl, 10 mM dithiothreitol, 5% polyethylene glycol 8000, 10% dimethyl sulfoxide, 0.05% Nonidet P-40, 1% glycerol, 1.6 mM HEPES, pH 8.0, and 3.3 mM EDTA. The IN was diluted in a buffer containing 30% glycerol, 0.5 M NaCl, 50 mM HEPES, pH 8.0, 1 mM dithiothreitol, and 0.1 mM EDTA. Where specified 100 ng of HMG-(I/Y) was added to the reaction mixtures. The preincubation reaction mixtures were placed on ice overnight. The volume of each preincubation mixture was then increased to 10 l with the addition of MgCl 2 to a final concentration of 7.5 mM, and the integration assay mixture was incubated at 37°C for 2 h. The reactions were stopped by increasing the volume to 150 l by the addition of EDTA (final concentration, 4.25 mM), sodium dodecyl sulfate (final concentration, 0.44%), and proteinase K (final concentration, 0.06 mg/ml). After digestion for 60 min at 37°C, the reaction mixtures were extracted with phenol followed by phenol-chloroform-isoamyl alcohol (25:24:1 mixture). 15 l of 3 M sodium acetate, pH 5.2, was added along with 1 l of glycogen (10 mg/ml stock solution). The reaction products were precipitated by the addition of 450 l of 100% ethanol and washed twice with 70% ethanol prior to electrophoresis and autoradiography. The reaction products were separated on a 1% agarose gel run in 0.5ϫ Tris borate, EDTA, and ethidium bromide at 10 V/cm for 2 h. Following electrophoresis, gels were submerged in 5% trichloroacetic acid for 20 min or until the bromphenol blue dye turned bright yellow. After being washed with water, the gels were dried on DE-81 paper (Whatman) in a Bio-Rad slab gel dryer at 80°C for approximately 2 h under vacuum. The dried gels were exposed to autoradiographic film overnight at Ϫ80°C in a film cassette with GAFMED TA-3 or Kodak mid speed screens.
Cloning and Sequencing of Integrants-In all experiments, integration products pooled from several separate reactions were used directly for transformation of bacteria. The integration products were introduced into E. coli MC1061/P3 by electroporation, using a Bio-Rad electroporator with 0.1-cm electroporation cuvettes, 1.8-kV voltage, 25microfarad capacitance, and 200-ohm resistance. The P3 episome is maintained at a low copy number. Therefore, only 40 g/ml of ampicillin, 15 g/ml of kanamycin, or 10 g/ml of tetracycline were required for selection. Under these conditions, we detected no colonies after supF selection when the donor, acceptor, or donor and acceptor were electroporated into cells in the absence of IN. Plasmid DNAs were recovered from individual clones, and integration junctions were sequenced by using primers U3seq (for sequencing the U3 junction) and U5seq (for sequencing the U5 junction). Sequencing was performed using the Thermo-Sequenase kit (U.S. Biochemical Corp.).

RESULTS
Model Substrates-In the present study, we have employed a series of model HIV-1 and ASV oligodeoxyribonucleotide sub-strates to examine the action of viral and/or cellular enzymes in catalyzing the repair and strand closure reactions associated with integration of viral DNA. The substrates contain HIV-1 or ASV U5 LTR terminal sequences in which the conserved CA dinucleotide is covalently linked to an acceptor DNA sequence. Substrate 1 in Fig. 1A depicts a simplified HIV-1 integration intermediate with a gap in only one of two strands, representing a single viral-host DNA junction. For this substrate, the gap is five nucleotides in length. The U5 HIV-1 LTR upper strand also contains at its 5Ј end a two-nucleotide mismatch. To obtain fully repaired covalently closed DNA, the gap in substrate 1 has to be repaired, the 5Ј overhang has to be excised, and the resultant nick has to be sealed. Substrate 2 in Fig. 1A represents an HIV-1 DNA intermediate in which the gap is already repaired to form a nick but still contains the two-nucleotide 5Ј overhang. Substrate 3 is comparable with substrate 2 but contains the ASV instead of the HIV-1 U5 LTR sequence, and the gap is six rather than five nucleotides in length. We have also examined several HIV-1 substrate variants in which base pair inversions were introduced at LTR positions 3-7, adjacent to and/or including the conserved CA dinucleotide. The number system used here refers to the ends of the unintegrated LTRs where the CA dinucleotide is at positions 3 and 4. Substrate 4 (Fig. 1A) contains a 2-base pair substitution in the HIV-1 U5 LTR IN recognition sequences at positions 5 and 6. Substrate 5 and 6 contain base pair inversions at positions 4 -7 and 3 and 4, respectively. For ease of discussion, we refer to the bottom covalently linked LTR cell DNA strand (36 mer) as strand A (Fig. 1B). Strand B refers to the upper LTR strand with the two-nucleotide 5Ј overhang (21-mer). Strand D refers to the upper cell DNA strand (12mer) that serves as the primer to fill in the gap. Strand C (17-mer) represents the same strand as D, but to which repair of the gap has occurred to form a nick (Fig. 1B). By placing a radioactive label on the 5Ј or 3Ј ends of the different strands B-D, we can examine gap repair, overhang removal, or ligation reactions independent of one another. Removal of the Two-nucleotide 5Ј Overhang-To look for activities that would remove the two-nucleotide 5Ј overhang, the HIV-1 gapped substrate 1 (Fig. 1A) was prepared with a 5Ј 33 P end label on strand B. This substrate was incubated with different viral and/or cellular enzymes, and the gel electrophoresis analysis of the products is shown in Fig. 2. Incubation with purified HIV-1 RT in the presence or absence of deoxynucleotides (data not shown) or with IN (Fig. 2A, lane 7) failed to release the mismatched nucleotides. In contrast and as a control, the addition of FEN-1, which is known to remove 5Ј overhang strands from duplexes, resulted in the endonucleolytic release of the 5Ј 33 P label as a series of bands representing dimers to hexamers and/or larger products ( Fig. 2A, lane 2). The most abundant product was a tetranucleotide, even though the initial mismatch was only two nucleotides in length. The addition of PCNA to the reaction resulted in the release of larger 5Ј labeled fragments (Fig. 2A, lane 3), whereas the addition of HIV-1 RT and/or IN did not influence the removal of the overhang strand ( Fig. 2A, lanes 4 and 5). If unlabeled deoxynucleotides were added to the reaction with RT, a 22-mer 5Ј end labeled product was observed, because of a nontemplated addition reaction to the 3Ј end of the stand B ( Fig. 2A,  lanes 5 and 6).
If a nicked substrate with a 5Ј 33 P-labeled two-nucleotide overhang (Fig. 1A, Substrate 2) was substituted for substrate 1, FEN-1 released the 5Ј overhang, but the predominate product in this case was a dinucleotide. Some mononucleotides were detected (Fig. 2B, lanes 2 and 3). The presence of HIV-1 IN had no significant effect on FEN-1-mediated removal of the 5Ј mismatched sequences (Fig. 2B, lane 4). To confirm the size of FEN-1 released products as well as to follow the exonuclease activity of FEN-1, strand B was prepared with a 3Ј 32 P end label (22-mer). In the presence of FEN-1, the 22-mer was converted to a series of smaller oligos ranging in size from 20 to 17 nucleotides or smaller (Fig. 2C). A 21-mer was not detected. Although FEN-1 has the ability to release products greater than two nucleotides in length, we asked whether there was a transient formation of a nick that could be trapped by DNA ligase. Substrate 2 with a 3Ј end label on strand B (22-mer) was incubated with FEN-1 in the presence or absence of T4 DNA ligase (Fig. 3, lanes 2 and 3, respectively). In the presence of both FEN-1 and ligase (lane 2), there was a quantitative conversion of the 22-mer to a ligated 37-mer product that would represent a covalent fusion of strands B and C. This indicates that a nick had formed in all of these substrates that could be sealed by ligase.
It has been reported that integrase alone is able to remove the overhang and seal the break with a substrate that contained a 5Ј overhang at a nick via a "disintegration-like" reaction (2, 10 -11). We therefore incubated substrate 2 with a 3Ј 32 P end label on strand B (22-mer) and looked for the appearance of a 37-mer product. As shown in Fig. 4A (lane 3) a small percentage (1-2%) of the labeled 22-mer was converted to the expected covalently linked 37-mer product by the action of HIV-1 IN. If FEN-1 was added along with IN to this reaction, the 37-mer product was no longer detected (Fig. 4A, lane 2). We further explored the specificity of the above trans-esterification reaction by presenting HIV IN with a nicked substrate mimicking the host DNA-U5 LTR junction of an ASV integration intermediate (substrate 3) again with the label introduced at the 3Ј end of strand B. The results are shown in Fig. 4B. Because strand B is a 16-mer in this substrate, the expected size of the covalently closed product would be a 32-mer. As shown in Fig. 4B (lane 2), a 32-mer was detected. The extent of the reaction was similar to that observed with the HIV-1 model substrate shown in panel A. We confirmed that HIV-1 IN catalyzed both overhang removal and ligation by using substrate 3, which contained a 5Ј 33 P-end label introduced on strand C (18-mer). The expected product in this case would be a 31-mer, which was detected as shown in Fig. 4C (lane 2).
We further explored the specificity of the reaction by presenting HIV-1 IN with model nicked substrates mimicking the HIV-1 U5 integration intermediates but with base pair substitutions introduced at positions 3-7 in the LTR IN recognition  1-7) with PCNA (500 ng), HIV-1 IN (5 pmol), and/or HIV-1 RT (110 g) and deoxynucleotides with substrate 1 (1 pmol) 5Ј 33 P-labeled on the overhang strand B, as described under "Experimental Procedures." Substrates are described in Fig. 1 and products were analyzed by polyacrylamide gel electrophoresis as described under "Experimental Procedures." B, FEN-1 was incubated as in A but with substrate 2 (1 pmol) 5Ј 33 P-labeled on the overhang strand B. C, FEN-1 was incubated as in A but with substrate 2 (1 pmol) 3Ј 32 P-labeled on the overhang strand B. A oligodeoxyribonucleotide ladder was prepared with a 5Ј sequence as shown in Fig. 1A (substrates 4 -6). Incubation of the wild type substrate 2 with HIV-1 IN produces the expected 37-mer product (Fig. 5A, lane 2). Base pair inversions at positions 5 and 6 or positions 4 -7 had little detectable effect on formation of the 37-mer covalently linked product (Fig. 5A,  lanes 4 and 6). Base pair inversions at positions 3 and 4 resulted in a reduction of the amount of the 37-mer detected (Fig.  5A, lane 8) as previously noted (9, 10). As discussed below, all of these base pair substitutions adversely affect concerted DNA integration in vitro. Taken together, these results indicate that the formation of the covalently linked DNA product via a disintegration-like reaction is relatively independent of sequence.  (6). The donor DNA is radiolabeled, and integration is followed by incorporation of the small donor into the larger acceptor. The integration products formed with a wild type HIV-1 donor DNA is shown in Fig. 5B (lane 1). When the base pair substitutions are introduced into the U5 IN recognition sequence at positions 5 and 6, 4 -7, or 3 and 4, respectively, there is a progressive decrease in detectable integration products compared with wild type (Fig. 5B, lanes 2-4). This was similar to the effect of base pair inversions introduced into the ASV U3 LTR sequence (18). 2 The products from these reactions were introduced into bacteria containing a P3 plasmid with drug-resistant markers that included amber mutations in their coding sequence. Individual integrants were isolated and sequenced, and the results are summarized in Tables I and II. Using the wild type donor, integrants exhibit characteristics associated with integration in vivo. This includes the loss of the two base pairs from the ends of the LTRs and mostly five base pair duplications introduced into the acceptor DNA at the site of insertion (in Table I and Hindmarsh et al. (6)). For the mutated donors, no integrants were recovered from reactions that contained base pair substitutions at positions 4 -7 or 3 and 4. However, the number of integrants recovered from reactions with the base pair substitutions at U5 positions 5 and 6 was reduced by 40% compared with wild type. Among these integrants, approximately 18% were derived from a nonconcerted DNA integration mechanism that produced large deletions in the acceptor DNA (Table II). This is similar to what is observed when base pair inversions are introduced into comparable U5 positions of an ASV donor DNA substrate (18). 2 Effect of HMG-(I/Y) on HIV-1 Overhang Removal and Ligation Reactions-We have previously reported that HMG-(I/Y) stimulates HIV-1 concerted integration using a reconstituted assay (6) and confirmed this observation as shown in Fig. 6A. Because this cellular protein stimulates the initial processing and joining reactions, we asked whether it would also stimulate the overhang removal/closure reactions catalyzed by HIV-1 IN using substrate 2. As shown in Fig. 6B, the formation of the 37-mer product was unaffected by the presence of HMG-(I/Y) (lanes 3 and 4).

Effect of HIV-1 U5 Base Pair Inversion on Integration in
Gap Repair-We next examined the ability of HIV-1 or AMV RT to fill the gap in our model substrate, forming a nicked duplex with the 5Ј overhang. The assay used to follow the repair reaction depended upon the formation of a nicked substrate that could be converted to a 37-mer product in the presence of FEN-1 and T4 DNA ligase. This assay was used because AMV and HIV-1 RT are known carry out strand displacement DNA synthesis (22,23). When coupled with pausing of RT on template during reverse transcription, these activities would produce a heterogeneous population of different sized DNA products. This was, in fact, observed when substrate 1 Lane 1, control with no added enzymes. The marker ladder was prepared as described in the legend to Fig. 2. B, as in A except that the ASV-based substrate 3 was substituted for HIV-1 substrate 2. The covalently closed product for this substrate is a 32-mer indicated on the right. C, as in B except that the ASV-based substrate 3 is 5Ј 33 P-labeled in strand B. The covalently closed product for this substrate is a 31-mer indicated on the right. Stick figures representing each of the substrates are below each panel.  1 and 2), mutant substrate 4 (lanes 3 and 4), mutant substrate 5 ( lanes 5 and 6), and mutant substrate 6 (lanes 7 and 8). Each substrate is shown in Fig. 1A. The substrates (10 pmol) were incubated with (lanes 2, 4, 6, and 8) or without (lanes 1, 3, 5, and 7) HIV-1 IN (10 pmol) as described under "Experimental Procedures," and products were analyzed by gel electrophoresis as described in legend to was incubated with RT and labeled deoxynucleotides (data not shown). Strand B of substrate 1 was 3Ј end labeled with a [␣ 32 P]ddATP to form the 22-mer and then incubated with HIV-1 RT (Fig. 7, lanes 2-4). Using the coupled assay with FEN-1 and ligase, substantial amounts of the 37-mer product were detected (Fig. 7, lane 4). A similar result was obtained if AMV RT was substituted for HIV-1 RT (Fig. 7, lane 7). These results indicate that both HIV-1 and AMV RT can fill in the gap to form a nicked substrate with a 5Ј overhang equivalent to substrate 2. If FEN-1 was left out of the assay, no 37-mer product was detected (Fig. 7, lanes 3 and 6). In separate experiments, we observed that cellular DNA polymerase ␦, PCNA, and RF-C would also fill in the gap to form the nicked substrate (data not shown).
Coupled Gap Repair, Overhang Removal, and Ligation in the Presence of RT and HIV-1 IN-The activity of IN on a nicked model substrate suggested that IN could remove the 5Ј overhang if a DNA polymerase, such as RT, repaired the gap. The same substrate used in the experiment described in Fig. 7 was incubated with HIV-1 IN and RT in the presence of unlabeled deoxyribonucleotides to determine whether the combination of the two viral enzymes was sufficient to catalyze the final steps in integration without the need of cellular enzymes. As shown in Fig. 8 (lane 3), HIV-1 IN and RT were sufficient to form the 37-mer product. Some smaller sized DNA products were detected because of possible nuclease contamination of our HIV-1 RT preparation. The 37-mer, but not the smaller products, was detected if AMV RT was substituted for HIV-1 RT (Fig. 8, lane  4). DNA polymerase ␦ could also substitute for HIV-1 RT in the reaction (data not shown). Taken together, these results indicate that any DNA polymerase that can fill in the gap will create a nicked substrate that HIV-1 IN can act upon to remove the 5Ј overhang and covalently close the resultant nick via a disintegration-like reaction. DISCUSSION Integration in vivo and in vitro is dependent upon the sequences at the LTR ends of viral DNA (1,6,8,18,19,25). 2 In this report, we demonstrate that base pair substitutions at positions 3 and 4, 5 and 6, or 4 -7 of the HIV-1 U5 donor DNA end significantly decrease the efficiency of concerted DNA integration in vitro, as detected by gel electrophoresis analysis of reaction products. In fact, the defect caused by base pair inversions at U5 LTR positions 3 and 4 or 4 -7 was so severe that we were unable to recover integrants when the reaction products were introduced into bacteria. With the base pair inversions at U5 LTR positions 5 and 6, integrants were recovered, but at about 40% the efficiency of comparable reactions with the wild type donor DNA. Among these recovered integrants, approximately 18% were derived by a nonconcerted DNA integration mechanism that introduced deletions into the acceptor DNA (Table II). This is similar to what is observed when base pair inversions are introduced into comparable positions of an ASV donor DNA substrate. 2 Of the concerted DNA integrants sequenced, another 18% contained deletions in the donor DNA, 80% of which were in U3 and the remainder in U5. Of the deletions found in U3, three resulted from a loss of 4 base pairs and utilized the same CT dinucleotide. The other lost 13 base pairs and utilized an internal GA dinucleotide that was previously found in ASV integrants containing base pair inversions at the comparable positions 5 and/or 6 in ASV donor DNA substrates. These results agree with other reports (18) 2 that the highly conserved CA dinucleotide is not absolutely required for concerted DNA integration. Among the sequenced integrants, approximately 21% arose from multiple insertions of donor into donor DNA, which in turn were integrated into the acceptor DNA. These products were not observed with the ASV reconstituted system. Taken together, these results indicate that mutations introduced into positions 3-7 of the HIV-1 IN recognition sequence adversely affect the efficiency and mechanism of DNA integration catalyzed by HIV-1 IN.
To examine further the properties of the reconstituted integration reaction, we used a model DNA substrate with a fivenucleotide gap and a two-nucleotide 5Ј overhang to examine the ability of HIV-1 RT and IN, respectively, to fill in the gap, remove the overhang, and covalently close the DNA. The sequences used in this substrate represent a U5 HIV-1 IN recognition sequence linked to an acceptor DNA at a site previously used for integration in vitro in the above reconstituted system (6). To detect the covalently closed product, the radioactive label had to be placed on the 3Ј end of the overhang strand. The label was also introduced with a dideoxyribonucleotide to prevent end addition to the overhang strand. Such labeling allowed discrimination of a covalently linked product from one in which RT displaced the overhang strand and extended the primer to the end of the template. In the presence of deoxyribonucleotides, RT derived from either HIV-1 or AMV readily cacACGAT TTCCAgtg   13). To determine whether the same property characterized the closure reaction, we prepared a series of mutant HIV-1 substrates with base pair inversions at positions 3-7 in the U5 HIV-1 LTR end sequence adjacent to and including the conserved CA dinu-cleotide. As noted above, these changes dramatically reduced the integration efficiency of an HIV-1 donor DNA in the reconstituted in vitro concerted integration reaction. However, although base pair substitutions at the conserved CA dinucleotide reduced the rate of covalent closure, base pair inversions at positions 5 and 6 and 4 -7 had no effect on the reaction (Fig.  5A). These results are consistent with the findings of Kulkosky et al. (10) and indicate that the covalent closure of these substrates has the characteristics of a disintegration-like reaction.
Although the combination of RT and IN can repair the gap and remove the 5Ј overhang to form covalently closed DNA in vitro, this was accomplished with very low efficiency. The ratelimiting step is the disintegration-like closure reaction rather than gap repair. Disintegration is known to occur with low efficiency (2, 10) and, unlike the processing and joining reactions (5, 6), 2 is not stimulated by the addition of the cellular protein HMG-(I/Y) (Fig. 6). This is consistent with a mechanism whereby HMG-(I/Y) acts on the LTR ends to stimulate the processing and joining reactions. 2 During replication in its host cell, an infecting retrovirus catalyzes only a single DNA integration event. Thus, a high turnover rate may not be required for the final closure step. However, it is possible that other cellular proteins can affect the rate of this reaction in a natural infection. If as estimated from our in vitro experiments, that only 1-2% of the DNA integration intermediate substrates are closed by IN, then within a given population of infected cells, most integration events would be incomplete unless cellular enzymes were recruited to complete the process. The requirement for cellular enzymes is consistent with the observation that in certain repair-deficient cells (e.g. DNA-PK cs , Ku86, or XRCC4-deficient), retroviral DNA integration is sensed as DNA damage and induces apoptosis (24).
Results from DNA transfection experiments indicate that cellular enzymes can repair the integration intermediates produced by IN in vitro. For example, the plasmid integrants we isolate after transfection of bacterial cells with the products of IN-catalyzed in vitro integration reactions are covalently closed plasmids with the expected host-viral DNA junction sequences. In this report, we demonstrate that FEN-1 nuclease together with T4 DNA ligase can close a nick with a two-nucleotide 5Ј overhang in our model substrates very efficiently. Nevertheless, it is likely that one of the eukaryotic cellular DNA ligases would serve this function during a natural infection. The importance of XRCC4 to retroviral DNA integration (24) suggests that its binding partner, ligase IV, is the likely candidate. FEN-1 is a structure-specific nuclease that recognizes and cleaves the Y structure in a DNA flap (16, 26 -28). This enzyme is implicated in both DNA replication and mismatch repair (29,30). In yeast, the FEN-1 equivalent RAD27 participates in the single strand-annealing pathway of repair that requires processing of 5Ј ends (31). Here, we demonstrate that FEN-1 can recognize the model retroviral integration intermediate and efficiently remove the two-nucleotide 5Ј overhang in either a gapped or nicked substrate. We also observed that this nuclease cleaved several deoxyribonucleotides into the duplex structure (Fig. 2). However, a transient nick was formed that was sealed quantitatively by DNA ligase. Neither RT nor IN affected the FEN-1 catalyzed 5Ј overhang cleavage with these model substrates. PCNA, a protein involved in DNA replication was reported to stimulate FEN-1 activity (especially exonuclease) on flap substrates (21,32). However, we did not observe any significant stimulation of FEN-1-mediated cleavage in our system. This may be explained by the small size of the overhang used in our integration intermediate. Although we have demonstrated that FEN-1 can remove the 5Ј overhang, it is possible that other cellular nucleases, such as the FEN-1 ho- Although results from our in vitro reconstituted experiments do not rule out a direct role for retrovirus RT and IN in catalysis of the last steps of retroviral DNA integration, the low efficiency that we observed for this reaction does not support this simple mechanism. It seems more likely that host repair enzymes such as those described in this report are critical for completion of retroviral DNA integration in vivo.