Multiple Integrase Functions Are Required to Form the Native Structure of the Human Immunodeficiency Virus Type I Intasome*

Mu-mediated polymerase chain reaction footprinting was used to investigate the protein-DNA structure of human immunodeficiency virus type I (HIV-I) preintegration complexes. Preintegration complexes were partially purified from cells after using an established coculture infection technique as well as a novel technique using cell-free supernatant from transfected cells as the source of virus. Footprinting revealed that bound proteins protected the terminal 200–250 base pairs of each viral end from nuclease attack. Bound proteins also caused strong transpositional enhancements near each end of HIV-I. In contrast, regions of viral DNA internal to the ends did not show evidence of strong protein binding. The end regions of preintegrative HIV-I apparently form a unique nucleoprotein structure, which we term the intasome to distinguish it from the greater preintegration complex. Our novel system also allowed us to analyze the structure and function of preintegration complexes isolated from cells infected with integrase mutant viruses. Complexes were derived from viruses defective for either integrase catalysis, integrase binding to the viral DNA substrate, or an unknown function in the carboxyl-terminal domain of the integrase protein. None of these mutant complexes supported detectable integration activity. Despite the presence of the mutant integrase proteins in purified samples, none of these nucleoprotein complexes displayed the native intasome structure detected in wild-type preintegration complexes. We conclude that multiple integrase functions are required to form the native structure of the HIV-I intasome in infected cells.

A pivotal step in the retroviral life cycle is forming the provirus, an integrated cDNA copy of the viral RNA genome.
The key viral players in integration are the trans-acting integrase (IN) 1 protein and the cis-acting DNA attachment site. Integration proceeds through three steps, the first two of which are known to require IN function. The linear ends of the cDNA are initially processed adjacent to phylogenetically conserved CA dinucleotides, resulting in a pair of recessed 3Ј ends. After nuclear localization, the exposed 3Ј-hydroxyls are joined to the 5Ј-phosphates of a double-stranded staggered cut in chromosomal DNA. The final step is DNA repair, wherein the singlestranded gaps at the sites of joining are sealed, resulting in the sequence duplication of the double-stranded cut flanking the integrated provirus (for a review, see Ref. 1).
In infected cells, integration is mediated by large nucleoprotein complexes known as preintegration complexes (PICs), which are derived from the cores of infecting virions (2). PICs isolated from infected cells can integrate their endogenous cDNA into an exogenously added target DNA in vitro (2)(3)(4)(5)(6)(7). Additionally, recombinant IN proteins purified after expression in bacteria can cut and join oligonucleotide attachment site DNA substrates (reviewed in Ref. 1). These latter in vitro assays have been invaluable for deciphering the structure and function of retroviral IN proteins. IN can be divided into three distinct functional domains: the amino-terminal, catalytic core, and carboxyl-terminal domains (reviewed in Ref. 8). The central domain contains the highly conserved D,D(35)E amino acid motif that comprises the IN active site (9,10).
Although simplified in vitro assays using purified IN proteins have provided essential knowledge toward understanding the overall integration process, these systems only partially mimic integration in vivo, because the predominant recombination products result from the insertion of only one viral DNA end into just one strand of target DNA. Virus replication requires integration of both DNA ends into both strands of target DNA; the single-ended activity typical of some in vitro systems would not yield a productive viral infection. Although altering the source of purified IN protein from bacterial to viral and/or modifying reaction conditions can increase the frequency of two-ended integration products (11)(12)(13), these systems still do not recapitulate the efficiency of two-ended integration activity displayed by PICs isolated from infected cells (14).
The discrepancy in reaction products catalyzed by PICs as compared with purified IN proteins suggests that efficient twoended integration activity might require higher-order proteinprotein and/or protein-DNA interactions specific to nucleoprotein complexes derived from infected cells. To begin to address this, we have used in vitro Mu-mediated polymerase chain reaction (MM-PCR) footprinting to analyze the protein-DNA structure of human immunodeficiency virus type I (HIV-I) PICs partially purified from infected cells. We have established an efficient HIV-I infection system initiated from transfected cell supernatant that yields active PICs. This has allowed us to analyze the structure and function of HIV-I PICs derived from a number of IN mutant viruses. Our results indicate that multiple IN functions are required to form the native protein-DNA structure of the HIV-I intasome.
* This work was funded in part by National Institutes of Health Grant AI39394, by funds from the G. Harold and Lelia Y. Mathers Foundation, by a gift from the Friends 10, and by the National Institutes of Health Intramural AIDS Targeted Antiviral Program. 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
Infections, PIC Isolation, and in Vitro Integration Assay-SupT1 cells (4 ϫ 10 7 ) were cocultivated with phorbol 12-myristate 13-acetatetreated MOLT IIIB cells in 20 ml of MOLT IIIB-conditioned medium (24-h supernatant of unstimulated cells seeded at 1 ϫ 10 6 cells/ml) at a ratio of 10:1. After 5 h, cells were washed twice in buffer K (20 mM HEPES, pH 7.5, 5 mM MgCl 2 , 150 mM KCl, 1 mM dithiothreitol, and 20 g/ml aprotinin) and lysed in 1 ml of buffer K-0.025% (w/v) digitonin. Crude cytoplasmic extract cleared by brief centrifugation was treated with RNase A (0.1 mg/ml) for 30 min at room temperature. To partially purify PICs, 1.5 ml of lysate was passed through a bovine serum albumin-coated Sepharose CL-4B (Amersham Pharmacia Biotech) spin column (12 ml) equilibrated in buffer K-0.025% digitonin. The eluate was further purified on a 10-ml, 10 -50% (w/v) Nycodenz gradient prepared in buffer K. Centrifugation was for 16 h at 274,000 ϫ g in a Beckman SW 41 Ti rotor at 4°C. Twelve fractions were collected from the top of the gradient.
293T cells were seeded at 5.8 ϫ 10 4 cells/cm 2 in a 14-cm plate 24 h before transfection. Cells were transfected with plasmid DNA (50 g) using calcium phosphate coprecipitation (28). Transfected cell supernatant was passed through 0.45 m filters and treated with DNase I (Promega; 2 units/ml) in the presence of 10 mM MgCl 2 for 45 min at 37°C. This step reduced the level of plasmid carry-over from transfection to the limit of detection by Southern blotting. For analyzing reverse transcription in cell lines, T cells (2 ϫ 10 7 cells) and HeLa-CD4 cells (4 ϫ 10 6 cells plated in a 10-cm plate 20 h before infection) were infected for 8 h with 10 ml of cell supernatant. Crude cytoplasmic extracts (0.5 ml) treated with RNase A were deproteinized by adding SDS, EDTA, and proteinase K to final concentrations of 0.5% (w/v), 6 mM, and 0.6 mg/ml, respectively. After incubating at 56°C for 1 h, DNA was recovered by phenol/chloroform extraction and precipitation with ethanol. To isolate PICs, C8166 cells (3 ϫ 10 7 ) were infected for 8 h with 20 ml of transfected cell supernatant. RNase A-treated cytoplasmic extract was loaded directly onto Nycodenz gradients.
PIC activity was assayed essentially as described previously (5). Briefly, 200 l of crude cytoplasmic extract, spin column eluate, or gradient fraction was incubated with 600 ng of linearized X174 DNA for 45 min at 37°C. Reactions were deproteinized, and DNA was recovered by precipitation with ethanol. DNA was electrophoresed through 0.6% agarose gels in Tris acetate-EDTA (28), transferred to GeneScreen Plus membrane (NEN Life Science Products), and probed with a LTRspecific riboprobe (HXBc2 nucleotides 8896 -9615). Integration activity was quantified as the percentage of cDNA substrate converted into product using either PhosphorImager analysis (Molecular Dynamics) or densitometry (IS-1000 Digital Imaging System).
Western Blotting-Gradient-purified C8166 cell extracts adjusted to 15% (w/v) glycerol were frozen in liquid N 2 and stored at Ϫ80°C. SDS was added to a final concentration of 0.25% to thawed samples (200 l), and proteins were recovered by methanol-chloroform-H 2 O extraction essentially as described previously (29). Briefly, samples were mixed with 4 volumes of methanol, followed by 2 volumes of chloroform. Two phases were separated after mixing 3 volumes of H 2 O and centrifuging at 9,000 ϫ g for 1 min. Proteins at the interface were precipitated using 3 volumes of methanol and spinning at 9,000 ϫ g for 2 min. Pelleted proteins were resuspended in 1ϫ sample buffer (12 mM Tris-HCl, pH 6.8, 5% glycerol, 0.4% SDS, 2.88 mM 2-mercaptoethanol, and 0.02% bromphenol blue), boiled for 10 min, and electrophoresed through 10% SDS-polyacrylamide gels. Proteins were transferred to Hybond-C extra membrane (Amersham Pharmacia Biotech), and IN was detected using monoclonal antibody 8E5 (30) with an ECL Western blotting kit (Amersham Pharmacia Biotech). Recombinant HIV-I IN protein was purified after expression in Escherichia coli as described previously (31).
MM-PCR Footprinting-Mu A protein was kindly provided by Dr. Michiyo Mizuuchi (National Institute of Diabetes and Digestive and Kidney Diseases). Two different naked DNA footprinting controls were generally used. In one, plasmid DNA was added to buffer K to a final concentration of 1.25 ng/ml. The other control was deproteinized PICs resuspended in buffer K. Mu transpososomes were assembled by mixing the annealed Mu end (40 nM) with Mu A transposase (232 nM) in 40 mM Tris-HCl, pH 8.0, 0.005% (w/v) bovine serum albumin, 30% (w/v) glycerol, 30% (v/v) dimethyl sulfoxide, 0.2% (v/v) Triton X-100, and 150 mM KCl. After a 10-min incubation at 30°C, the mixture was diluted 1:3 in the same buffer. Assembled Mu transpososomes (200 l) were mixed with an equal volume of gradient-purified PICs or naked DNA controls, and transposition was initiated by adding CaCl 2 to a final concentration of 10 mM. After 30 min at 30°C, reactions were stopped by adding SDS, EDTA, EGTA, and proteinase K to final concentrations of 0.5%, 6 mM, 10 mM, and 0.6 mg/ml, respectively. Deproteinized Mu transposition products were recovered by precipitation with ethanol.
DNA was dissolved in 50 l of H 2 O, of which 3 l was used in nested PCR. The first round (50 l) contained 20 mM Tris-HCl, pH 8.8, 10 mM KCl, 10 mM (NH 4 )SO 4 , 4 mM MgSO 4 , 0.1% Triton X-100, 0.4 M each of Mu25 and a viral-specific primer, 0.4 mM of each dNTP, 0.1 mg/ml bovine serum albumin, and 1 unit of Vent DNA polymerase (New England Biolabs). Twenty-five cycles were performed on a Perkin-Elmer GeneAmp PCR system with the following PCR profile: preheating at 96°C for 4 min, denaturation at 96°C for 15 s, annealing at 58°C for 1 min, and elongation at 74°C for 1 min. A final 10-min elongation at 74°C was then performed. An aliquot (2 l) was transferred for the second round (25 l; 25 cycles), which contained 20 mM Tris-HCl, pH 8.8, 10 mM KCl, 10 mM (NH 4 )SO 4 , 2 mM MgSO 4 , 0.1% Triton X-100, 0.2 M each of Mu25 and a nested viral primer, 5 ϫ 10 5 cpm of 5Ј end 32 P-labeled viral primer, 0.2 mM of each dNTP, 0.1 mg/ml bovine serum albumin, and 1 unit of Vent (exo Ϫ ) DNA polymerase (New England Biolabs). PCR was carried out essentially as described above, except that the elongation time was adjusted from 10 to 45 s, based on the length of the fragment being amplified. Aliquots of second-round PCRs were analyzed on 5% denaturing sequencing gels.

RESULTS
Active HIV-I PICs from Molecularly Cloned DNA-In this study, HIV-I PICs were isolated from acutely infected cells using one of two different tissue culture infection systems. In one system, HIV-I (HTLV-IIIB strain) infection was initiated by mixing uninfected SupT1 cells with chronically infected MOLT IIIB cells, essentially as described previously (5). Cells were lysed 5 h after infection, and cytoplasmic extract containing HIV-I PICs was purified by spin column chromatography. This step removed 80 -90% of the total protein present in the cell extract, yielding PICs that reproducibly supported a higher level of integration activity than those in the starting extract (Fig. 1A, lanes 1-4; data not shown). The column eluate was further purified on a Nycodenz gradient, which was fractionated and analyzed for HIV-I DNA content and PIC activity. DNA and integration activity co-sedimented to fractions 6, 7, and 8 (Fig. 1A, lanes 5-10). The integration activity of the crude cell extract varied from 10% to 50%, depending on the individual preparation.
One goal of this study was to analyze the structure and function of HIV-I PICs derived from IN mutant viruses. Thus, we set out to establish an infection technique using transfected cell supernatant as the source of cell-free virus. Preliminary experiments tested the ability of six different CD4-positive cell lines to support reverse transcription after infection with molecularly cloned HIV-I (NL4-3 strain). Of these six cell lines, C8166 T cells (16) supported the highest level of cDNA synthesis (Fig. 1B). A kinetic analysis revealed maximum levels of reverse transcription approximately 8 h after infecting C8166 cells (data not shown). Because HIV-I cDNA in the crude cytoplasmic extract of C8166 cells was lost during spin column chromatography (data not shown), these extracts were directly purified by Nycodenz gradient centrifugation. The gradients were fractionated and analyzed for HIV-I DNA content and PIC activity. Both cDNA and integration activity colocalized to fractions 6 -8 (Fig. 1C, lanes 2-4). PICs from C8166 cells routinely integrated more of their cDNA substrate into the target (30 -80% of substrate converted to product) as compared with PICs isolated from the coculture infection system.
MM-PCR Analysis of Wild-type HIV-I PICs-MM-PCR footprinting was developed to analyze the native protein-DNA structure of Moloney murine leukemia virus (MoMLV) PICs (32). In this footprinting technique, preassembled Mu transpososomes are the DNA cleavage reagent. Mu A transposase inserts Mu-end DNA at the site of cleavage, and, as in any protein-DNA footprinting technique, bound protein interferes with the ability of the cleavage reagent to cut the nucleic acid. The advantage of this coupled cutting and DNA joining technique over other footprinting methods is that the DNA cleavage reagent itself becomes a substrate for subsequent PCR amplification (Fig. 2). We thus applied this technique to analyze the native protein-DNA structure of HIV-I PICs isolated from cells using two different tissue culture infection systems.
Three different substrates, naked plasmid, deproteinized PICs, and native PICs, were generally analyzed. PICs partially purified after the coculture infection were initially tested. As predicted, deproteinized PICs supported a pattern of Mu transposition similar to the pattern detected using naked plasmid DNA (Fig. 3, A and C, compare lanes 2 to lanes 1). Native PICs, however, revealed dramatically different transpositional patterns. Bound proteins protected more than 100 base pairs from each LTR end (Fig. 3, A and C, lanes 3). In addition to these large footprinted regions, transpositional enhancements were detected near the very ends of HIV-I (Fig. 3, A and C, lanes 3). Both the footprinted and enhancement regions were characteristic of the viral DNA ends; internal regions of deproteinized and native PICs supported similar distributions of Mu transposition (Fig. 3B). However, subtle differences were detected (Fig. 3B), suggesting that proteins may loosely associate with internal regions of HIV-I. PICs partially purified from infected C8166 cells yielded the same overall results. These PICs generally displayed clearer footprinted regions (Fig. 3, D and E), perhaps due to their higher integration activity.
The internal boundaries of the footprints were next identified. Whereas the U3 footprint extended approximately 250 base pairs from the end of HIV-I, the U5 footprint extended about 200 base pairs (Fig. 4). PICs isolated from infected C8166 cells displayed sharper footprint boundaries than those isolated from the coculture infection ( Fig. 4; data not shown).
Multiple IN Functions Are Required to Form the HIV-I Intasome-Because IN is essential for retroviral DNA integration, we tested the effects of different IN mutations on the structure and function of HIV-I PICs. HIV-I IN mutants can be divided into two classes, class I and II, based on their effects on the virus replication cycle (33). Whereas class I mutants are blocked specifically at the integration step, class II mutants display reverse transcription and/or virus assembly defects. Thus, measurable quantities of PICs can only be derived from class I IN mutant viral infections. Most HIV-I deletions, as well as a number of point changes, unfortunately fall into class II (33).
HIV-I IN contains three functional domains, the amino-terminal, catalytic core, and carboxyl-terminal domains (8), and each domain contains at least one amino acid residue that is conserved among all retroviruses (33). The amino-terminal domain contains two conserved His residues and two conserved Cys residues. Viral mutants carrying substitutions of any one of these residues display the class II replication phenotype (33), therefore measurable quantities of these PICs were not recoverable (data not shown). The catalytic domain contains the conserved active site Asp and Glu residues of the D,D(35)E motif as well as residues such as Lys-159 (24,34) and Gln-62 (31,34,35) implicated in viral DNA end binding. Substituting any of the three active site residues yields the class I mutant viral phenotype, as do certain substitutions of Lys-159 (the double mutant K156E/K159E was used here) and Gln-62 (Q62K). The substitution of Glu for the conserved carboxylterminal domain residue Trp-235 (W235E) also yields the class I mutant viral phenotype (25). At present, the exact defect of the W235E mutant virus is unknown. We thus analyzed four fection with either wild-type or class I IN mutant viruses. Whereas wild-type PICs converted about 50% of the cDNA substrate to the integration product, none of the mutant viral PICs supported detectable integration activity (Fig. 5). The preparations were then centrifuged into Nycodenz gradients. Each mutant sedimented in Nycodenz to the same position as did wild-type, indicating that the nucleoprotein structure of each PIC was largely intact (data not shown). The gradient-purified samples were then subjected to MM-PCR footprinting. The results of this analysis showed that none of the mutant PICs displayed the pattern of protein-DNA footprinting and transpositional enhancements indicative of the wild-type HIV-I intasome (Fig. 6). To ensure that the lack of intasome structure for each of the mutants was not simply due to the absence of the IN protein, gradient-purified samples were analyzed by Western blotting using an anti-IN monoclonal antibody. The results of this experiment showed that each of the mutant PICs contained IN protein at a level comparable to that of wild-type (Fig. 7). DISCUSSION In this report, we describe the native protein-DNA structure of preintegrative HIV-I as detected by MM-PCR footprinting. The structure has two distinguishable characteristics. Firstly, large regions (200 -250 base pairs) of protein binding were detected at the ends of HIV-I. Secondly, enhanced regions of nuclease attack were detected near the very ends of the virus. The overall structure is quite similar to the one previously reported for MoMLV (32). Thus, similar intasome structures may be common to the preintegrative DNAs of a variety of retroviruses.  (38), suggesting that protein binding in native PICs may distort viral end regions and create hot spots for Mu insertion. Alternatively, it is possible that proteins bound near the ends of retroviral DNA physically interact with the Mu nuclease. Bacterially expressed MoMLV IN protein bound to recombinant viral DNA displayed the viral end footprinting and enhancement patterns associated with the endogenous intasome isolated from infected cells (39), suggesting that IN is solely responsible for the MoMLV structure as detected by MM-PCR footprinting. However, we have been unable to detect evidence for the HIV-I intasome structure using purified recombinant IN with either recombinant viral DNA or deproteinized cDNA substrates purified from infected cells (data not shown).
To study the role of IN in the formation of the HIV-I inta-some, we established a highly efficient infection system initiated with molecularly cloned virus. Numerous CD4-positive cell lines were screened for their ability to support reverse transcription after infection (Fig. 1B), and one T-cell line, C8166 (16), was not only found to support efficient DNA synthesis, but the resulting HIV-I PICs displayed efficient in vitro integration activity (Fig. 1C). As predicted, PICs isolated after infection with class I viral mutants defective for either IN catalysis (D116N), viral DNA end binding (Q62K and K156E/ K159E), or an unknown carboxyl-terminal function (W235E) did not display detectable levels of in vitro integration activity (Fig. 5). Despite containing their mutant IN proteins (Fig. 7), none of the mutant PICs displayed the protein-DNA footprints and viral end enhancements indicative of the wild-type HIV-I intasome (Fig. 6) (41). A region of the carboxyl terminus between residues 247 and 270 has been implicated in binding to the HIV-I attachment site (35,42). In contrast, the region between residues 213 and 246, where Trp-235 resides, cross-linked to the target DNA portion of an in vitro disintegration DNA substrate (42). It is therefore possible that the W235E mutant virus is defective for interacting with chromosomal DNA in infected cells. However, because the structure of W235E PICs purified from cytoplasmic extracts was indistinguishable from the structure of either the D116N or attachment site DNA binding mutants, we propose that the W235E virus is blocked at a step that precedes target DNA interaction in the nuclei of infected cells. Determining the precise defect of the W235E mutant may reveal higher-order protein-protein and/or protein-DNA interactions important for PIC activity that are dispensable for recombinant IN function in more simplified in vitro integration assays. We also plan to continue to analyze the structure and function of HIV-I IN and attachment site mutant PICs isolated from infected cells.