INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Following reverse transcription, retroviral integrase (IN) catalyzes two endonuclease events that are necessary for insertion of each end of the newly synthesized viral DNA into an infected cell's chromosomal DNA. Initially, IN places a nick after invariant CA bases typically found two nucleotides from the 3' ends of blunt-ended linear viral DNA (33, 56); this sequence-specific reaction is referred to as processing (11, 34). Subsequently, IN inserts the nucleophilic 3'-OH group at each recessed end of the processed viral DNA into staggered sites on host DNA; this cleavage reaction, which is independent of the host DNA sequence, is referred to as DNA joining, strand transfer, or integration (11, 21, 32). Both processing and DNA joining can be mimicked in vitro, using short oligonucleotide substrates designed to resemble the U3 or U5 ends of viral DNA (11, 32, 33). Data from these and related assays have mapped the active site of the enzyme to the central region of IN (8, 13, 18, 43, 45, 47, 64, 68), which contains three highly conserved acidic amino acids that form a D,D-35-E motif (41). However, despite extensive deletion and mutagenesis studies, the portions of the protein that interact with viral DNA and host DNA have not been determined. It had been suggested that the N-terminal region, which has a conserved H-H-C-C motif that binds zinc (6, 8, 41), is responsible for specific interactions with viral DNA (8, 31, 43, 45, 52, 67). In contrast, the C-terminal region, which binds DNA nonspecifically (19, 49, 57, 68, 71, 72), has been an attractive location for sequence-independent interactions with host DNA. In this report, we present further evidence that both of these predictions are wrong.

The ability to form functional chimeras between appropriate wild-type proteins is a powerful and complementary approach to define functional domains. The human immunodeficiency virus type 1 (HIV-1) and visna virus integrases are particularly well suited for domain swapping experiments because they exhibit approximately 30% amino acid identity in each region, show optimal activity under identical in vitro reaction conditions, and yield distinguishable results in reliable assays (36). We previously described the first set of chimeric lentivirus integrases, constructed by exchange of the N-terminal, central, and C-terminal regions of the HIV-1 and visna virus proteins (36). The regions chosen for making the initial chimeras were defined by HIV-1 amino acid residues 1 to 49, 50 to 186, and 187 to 288 (with the caveat discussed below), based on the ability of these regions to functionally complement each other (17) and the ability of an isolated protein fragment representing residues 50 to 186 to catalyze some polynucleotidyl transfer reactions (8). Results obtained with these six chimeric integrases (summarized in Fig. 1, sets b to d) permitted three important observations to be made (36). First, the N-terminal region of IN does not contribute to viral DNA specificity, since a protein with the N-terminal region of HIV-1 IN and the central and C-terminal regions of visna virus IN exhibited the viral DNA specificity of visna virus IN. Second, target site selection with a viral DNA terminus as the nucleophile did not map to regions of IN defined by these boundaries, since chimeric proteins gave novel patterns of strand transfer products. Third, an additional endonuclease activity of IN was discovered. We termed this activity nonspecific alcoholysis because IN was shown to use small nucleophiles such as glycerol to attack any DNA sequence at multiple sites (38). Significantly, target site preferences for nonspecific alcoholysis, which is the most robust activity exhibited by several integrases, clearly mapped to the central region of IN (36, 38).

View larger version (39K): [in this window] [in a new window]   FIG. 1.   HIV-1/visna virus chimeric IN proteins. A linear representation of the IN protein is shown at the top, with the relative positions of the conserved H-H-C-C and D,D-35-E motifs indicated. The numbers above and below the linear map denote positions in the HIV-1 IN and visna virus IN proteins, respectively, that define domains utilized to form the chimeric proteins; note that amino acids 186 to 190 of HIV-1 IN are identical to residues 188 to 192 of visna virus IN. Proteins are presented schematically, with solid bars denoting fragments derived from HIV-1 and stippled bars indicating visna virus sequences. Wild-type and chimeric proteins are grouped in pairs a to g, as indicated under "Description." Sets b to d are our original six chimeras, named with three uppercase letters. Sets e to f are new chimeras, named with two uppercase letters and two lowercase letters; the second and third letters in this series can be considered an extended core region. The results of various oligonucleotide-based IN assays are summarized at the right, where "Preferred Viral U3 Substrate" refers to processing. , inactive in that assay; +, active for strand transfer but at too low a level to display a reproducibly discernible pattern.

Since target site selection for nonspecific alcoholysis maps to the central portion of IN, this region must participate in positioning nonviral DNA for nucleophilic attack (the limits for this domain are best defined by residues 50 and 190 because five consecutive amino acids, starting with HIV-1 IN residue 186, are identical in the two wild-type proteins). However, the ability to map target site selection to residues 50 to 190 only when IN utilizes small nucleophiles such as glycerol to attack DNA, but not when the larger viral DNA end is used as the nucleophile, suggests that residues outside numbers 50 to 190 are involved in recognizing or positioning the viral DNA end. For these and other reasons, we expanded the set of HIV-1/visna virus chimeric integrases by producing six new proteins in which the central region would now be defined by amino acid residues 50 to 212. The results map specificity during viral DNA processing, for the first time, to the central region of IN.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cloning of IN sequences. The HIV-1_HXB2 IN and visna virus IN coding regions were cloned into plasmid pQE-30 and expressed in Escherichia coli M15[pREP4] (Qiagen, Inc., Chatsworth, Calif.), as described previously (35, 36). Chimeric DNA was produced by the overlap extension PCR method (25), with all overlap primers beginning 3' to a T to minimize the introduction of errors by any nontemplated addition of a 3' A (62). Target sequences for amplification were the wild-type HIV-1 IN and visna virus IN cloned into pQE-30. Cassettes encoding HIV-1 IN amino acids 1 to 49 and 1 to 186 were amplified as previously described (36), and cassettes encoding residues 1 to 212, 50 to 212, 187 to 212, and 213 to 288 (Fig. 1) were amplified by the following pairs of primers, respectively: H1 and V3H2*, V1H2 and V3H2*, V2H3 and V3H2*, and V2H3* and H3. Analogous cassettes encoding visna virus IN amino acids 1 to 51 and 1 to 188 were produced as previously described (36), and cassettes encoding residues 1 to 214, 52 to 214, 189 to 214, and 215 to 281 (Fig. 1) were amplified by primer pairs V1 and H3V2*, H1V2 and H3V2*, H2V3 and H3V2*, and H2V3* and V3, respectively. Note that these crossovers correspond to DNA sequences; the resulting protein regions can be described slightly differently because amino acids 186 to 190 of HIV-1 IN are identical to residues 188 to 192 of visna virus IN (36). Sequences for H1, H3, V1, and V3, the outermost primer pairs for HIV-1 IN and visna virus IN sequences, respectively, and for overlap primers that are not named with an asterisk have been published previously (36). Sequences of the new overlap primers were as follows (HIV-1 sequences are in uppercase and visna virus sequences are in lowercase letters): V3H2*, 5'gttttgatttTTCTTTAGTTTGTATGTCTG3'; V2H3*, 5'acagcaacaaagtTTACAAAAACAAATTACAAAAATTC3'; H3V2*, 5'GTTTTTGTAAactttgttgctgtattctttg3'; and H2V3*, 5'ACAAACTAAAGAAaaatcaaaacaagaaaaaattcg3'. Products of the expected length were purified and appropriately mixed in a second round of PCR using the necessary outermost primers to yield full-length chimeras. These products were digested with BglII and SstI and ligated into the BamHI and SstI sites of pQE-30. M15[pREP4] was transformed with the ligation reaction mixtures, and colonies resistant to ampicillin and kanamycin were screened for the presence of insert DNA by restriction endonuclease digestion of plasmid DNA and for induction of IN by isopropyl--D-thiogalactopyranoside (IPTG), as described previously (35). The entire IN coding sequence of each construct was determined by dideoxy sequencing to ensure that expressed proteins had the correct amino acid sequence. A total of 31 clones were sequenced to obtain the complete set of 14 proteins shown in Fig. 1.

Expression and purification of integrases. Proteins were expressed by using the QIAexpress System Type IV construct (Qiagen) and carried 16 extra N-terminal amino acids: Met-Arg-Gly-Ser-(His)₆-Gly-Ser-Ile-Glu-Gly-Arg. Culturing, induction by IPTG, and purification of polyhistidine-tagged proteins from the pellets of 250-ml cultures were performed as described previously (36). Because initial purification of the VVhv protein (the nomenclature will be described later) yielded a relatively dilute product, a slightly more concentrated preparation that was purified in the presence of 10 mM 3[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), a zwitterionic detergent, was utilized for all experiments described herein. Purifications were monitored by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), using 12.5% separation gels and 3% stacking gels (acrylamide to methylenebisacrylamide ratio, 37.5:1). Protein concentrations were measured by comparison to Coomassie blue-stained standards, with quantitation by laser densitometry, as described previously (36).

Oligonucleotides. Sequences of terminal viral DNA 18-mer oligonucleotides and substrates used for the disintegration and alcoholysis assays have been published previously (36, 38). All oligodeoxynucleotides used as assay substrates were gel purified following synthesis and again after being 5'-end labeled with [-³²P]ATP by T4 polynucleotide kinase, as described previously (36). Sequence-specific markers for gel analysis were produced by the 3'-to-5' exonuclease activity of snake venom phosphodiesterase (Sigma, St. Louis, Mo.) on 5'-radiolabeled oligonucleotides, as described previously (33).

Processing, strand transfer, alcoholysis, and disintegration assays. Double-stranded DNA substrates were prepared by annealing the labeled strand with a fourfold excess of unlabeled complementary oligonucleotide, and the disintegration assay substrate was prepared and then purified on a native 15% polyacrylamide gel, both as described previously (35). Standard 10-µl reaction mixtures contained 0.5 to 1.0 pmol of double-stranded DNA or 0.06 pmol of the disintegration assay substrate, 25 mM Tris-HCl (pH 8.0), 10 mM dithiothreitol, 10 mM MnCl₂, and 0.5 to 1.5 µl of IN or protein storage buffer. Alcoholysis assays were conducted in the presence of 40% 1,2-ethanediol (ethylene glycol), which is less viscous than glycerol but yields the same pattern of cleavage products (38). Reaction mixtures were incubated for 90 to 180 min at 37°C, and then reactions were stopped by adding 10 µl of loading buffer (95% formamide, 20 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol) and heating at 95°C for 5 min. Aliquots were loaded onto 20% polyacrylamide (acrylamide to methylenebisacrylamide ratio, 19:1)-7 M urea denaturing gels; this was followed by electrophoresis at 75 W until the bromophenol blue dye had migrated 14 to 22 cm. Wet gels were autoradiographed at 80°C. The radioactivity of bands in wet gels was quantified with a Betascope (Betagen, Waltham, Mass.).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Expansion of the set of purified chimeras between HIV-1 and visna virus integrases. Our initial set of chimeric HIV-1/visna virus integrases had crossovers after HIV-1 IN positions 49 and 190 (the alternative designation of the second crossover was discussed earlier), corresponding to visna virus IN positions 51 and 192, respectively (Fig. 1, sets b to d). We referred to the wild-type proteins as HHH and VVV and to the six chimeras as VHH, HVV, HVH, VHV, HHV, and VVH, where the three letters represent (from left to right) the N-terminal, central, and C-terminal regions, respectively, and H (for HIV-1) or V (for visna virus) indicates the source of that region. As mentioned above, analysis of the activities of these proteins in various oligonucleotide-based assays suggested that residues outside numbers 50 to 190 are involved in recognizing or positioning the viral DNA ends. Since the study of the crystallization of the core region of HIV-1 IN utilized residues 50 to 212 to define this region (15), and because this extended core region shows greater activity in some in vitro assays (8), we decided to create additional chimeric integrases utilizing similar boundaries. In particular, six new chimeric integrases were produced (Fig. 1, sets e to g). The new chimeras all have a crossover after HIV-1 IN position 212, corresponding to visna virus IN position 214 (henceforth, only HIV-1 IN position numbers will be used in this report). To be consistent with our prior nomenclature, the new chimeras are named with uppercase letters for the regions defined by residues 1 to 49 and 50 to 190. In place of the third uppercase letter, two lowercase letters are now used for regions 191 to 212 and 213 to 288, although conceptually residues 191 to 212 should be considered part of an extended central region.

View larger version (63K): [in this window] [in a new window]   FIG. 2.   SDS-PAGE demonstrating purification of new chimeric integrases. Three microliters of each purified IN (indicated above the lanes; nomenclature as in Results) were heated in sample buffer and separated by SDS-PAGE, and the gel was stained with Coomassie blue. Proteins are paired (a, e, f, and g) as in Fig. 1. Molecular mass markers are in lane 1 (sizes are shown in kilodaltons at the left) and represent 200 ng per band.

Demonstration that each new chimeric IN has enzymatic activity. To assess whether the active site in the central region of the protein was maintained in a functional conformation, each chimera was tested for "disintegration" activity (shown schematically in Fig. 3), an in vitro phenomenon that reflects the ability of IN to reverse the strand transfer reaction (10). Of the processing, strand transfer, and disintegration activities of IN, disintegration has the least-stringent requirements for reaction conditions (30) and substrate DNA sequence (10, 59) and is detected even when the N-terminal and C-terminal regions of IN have been deleted (8, 68). The substrate used for this assay is a Y-shaped complex that represents the predicted immediate product of integration of a viral DNA terminus into host DNA. Reversal of the integration reaction yields the processed viral DNA end and a target strand in which the nick has been sealed. Incorporation of a radioactive phosphate at the 5' end of the nonviral 16-mer in the substrate permits monitoring of this result by detection of a radioactive 31-mer product. This product presumably forms by IN-catalyzed nucleophilic attack of the juxtaposed 3'-OH of the 16-mer at a site just 3' to the invariant CA of the viral sequence. Each of the six new chimeric integrases was active in this assay, as demonstrated by the appearance of a 31-mer product, whether the viral sequences were derived from HIV-1 (Fig. 3, lanes 4 to 9) or visna virus (data not shown). The doublet nature of the 31-mer product on this gel is likely due to residual secondary structure (10), and bands migrating between the 16- and 31-mer positions may represent breakdown products or reintegration events (67, 68). Although this assay is not able to distinguish between the sources of IN (36), it does demonstrate the integrity of the active site in the catalytic central region of each new chimeric protein.

View larger version (39K): [in this window] [in a new window]   FIG. 3.   Disintegration activity of new chimeric integrases. (Left) The substrate is a four-oligonucleotide complex representing the predicted immediate adduct of the HIV-1 U5 DNA end (thick lines) integrated into host DNA (thin lines). IN-mediated cleavage after the CA releases the viral DNA end, with concomitant joining of the juxtaposed 5'-radiolabeled 16-mer to the 15 nucleotides beyond the CA to yield a radiolabeled 31-mer product (asterisks denote 32P labels). The substrate was incubated with protein buffer or purified integrases under standard conditions for 90 min and analyzed as described in Materials and Methods. (Right) An autoradiogram from a denaturing polyacrylamide gel is shown. The sizes of the labeled component of the substrate and the product are indicated (in nucleotides) at the right, aligned with the complexes at the left from which they were derived. Proteins are paired (a, e, f, and g) as in Fig. 1. The two wild-type integrases and all six new chimeric integrases were active in this assay.

Specific processing activity of new chimeras on terminal viral DNA sequences. IN specifically removes two nucleotides from the 3' end of oligonucleotide substrates designed to resemble the termini of viral DNA (shown schematically in Fig. 4) (33). Of the two ends of HIV-1 or visna virus DNA, the HIV-1 U5 terminus provides the most susceptible in vitro substrate for processing by either wild-type IN (36). Our preparations of HIV-1 IN and visna virus IN process approximately 40 and 50% of this substrate, respectively (Fig. 4A, lanes 3 and 4), as determined using a highly reproducible and conservative method to quantify the extent of specific cleavage (37). Thus, assays with the HIV-1 U5 substrate can indicate whether a chimeric protein has any processing activity. In contrast, comparison of processing activities on the two U3 substrates best distinguishes between the HIV-1 and visna virus enzymes (36, 37). In particular, each wild-type IN preferentially cleaves oligonucleotide substrates derived from its own U3 substrate (Fig. 4B and C, lanes 3 and 4).

View larger version (42K): [in this window] [in a new window]   FIG. 4.   Processing activity of new chimeric integrases. (Top) A schematic of the site-specific 3'-end processing reaction is shown. Cleavage of blunt-ended viral DNA after the invariant CA (shown in boldface) converts a radiolabeled 18-mer to a labeled 16-mer (asterisks denote 32P labels). Duplex oligonucleotide 18-mer substrates derived from the U5 or U3 termini of HIV-1 DNA or the U3 terminus of visna virus DNA and 5' labeled on the plus or minus strand, as indicated in panels A to C, were incubated with protein buffer or purified integrases under standard conditions for 90 min and analyzed as described in Materials and Methods. The region of the autoradiogram extending down to 12-mers is shown to demonstrate the specificity of the reactions. Proteins are paired (a, e, f, and g) as in Fig. 1. Biologically relevant specific cleavage products two nucleotides shorter than the substrate are indicated by arrows. Sequence-specific markers are included in lanes 1, and their sizes (in nucleotides) are indicated at the left.

View larger version (66K): [in this window] [in a new window]   FIG. 5.   Duplicate processing reactions for informative proteins. Duplex oligonucleotide 18-mer substrates derived from the U3 termini of HIV-1 (A) or visna virus (B) DNA, which best distinguish between the wild-type enzymes, were 5' radiolabeled on the minus strand and incubated with protein buffer or selected integrases. Details are described in the legend to Fig. 4.

Target site selection using viral DNA termini as the nucleophile to nick DNA. IN catalyzes insertion of the recessed 3'-OH of processed viral DNA ends into various sites along other oligonucleotides that act as surrogates for host DNA to yield a set of labeled integration products that are longer than the substrate (shown schematically in Fig. 6) (11, 32). The patterns of insertion site preferences in this assay, as demonstrated by the autoradiographic locations and intensities of the longer products, distinguish between HIV-1 IN and visna virus IN (36). Thus, purified proteins were tested for strand transfer activity on a preprocessed duplex oligonucleotide substrate derived from the visna virus U3 end, which has been shown to be readily susceptible to both enzymes (36). To facilitate comparisons of the strand transfer patterns, the two wild-type enzymes and the complete set of 12 chimeric proteins were assayed in parallel. The distinctive patterns produced by the wild-type integrases are evident (Fig. 6, lanes 2 and 3). Of the 12 chimeras, 9 had activity in this assay, but only the HVV IN (lane 5) and the HVvh IN (lane 10) had levels of activity comparable to that of wild-type. To display the patterns produced by the other chimeras, severalfold more counts per minute were loaded in the other gel lanes.

View larger version (40K): [in this window] [in a new window]   FIG. 6.   Strand transfer activity of chimeric integrases. A duplex oligonucleotide derived from the visna virus U3 end, but preprocessed by omission of the final two nucleotides from the minus strand, was used as the substrate for the complete set of 14 purified integrases during 3-h incubations, and reactions were analyzed as described in Materials and Methods. An autoradiogram from the upper region of a denaturing polyacrylamide gel is shown. Insertion of the recessed 3' terminus into various sites along other oligonucleotides (shown as thinner lines in the scheme on the left) yield longer radiolabeled products (asterisks denote 32P labels). The positions of the labeled 16-mer component of the substrate and the labeled strands of longer products are indicated to the left of the gel. Proteins are paired (a to g) as in Fig. 1. The different patterns produced by HHH IN and VVV IN are demonstrated in lanes 2 and 3. Nine of the 12 chimeras had activity in this assay, but ~5- to 20-fold more counts per minute were loaded in lanes 4, 6 to 9, and 11 to 15 to display the patterns produced by less-active chimeras.

Target site selection using a small alcohol as the nucleophile to nick DNA. We recently described a fourth enzymatic activity of several retroviral integrases, which we named nonspecific alcoholysis (36, 38). IN was shown to use small nucleophiles, such as glycerol, ethylene glycol, and propylene glycol, to attack nonviral DNA sequences at multiple sites, with concomitant joining of alcohol groups from these nucleophiles to newly exposed 5' phosphate groups at sites of DNA nicking (shown schematically at the top of Fig. 7). Although every site in DNA substrates except those close to the ends can be nicked by IN in this assay, different integrases exhibit different target site preferences for nonspecific alcoholysis. In particular, HIV-1 IN prefers to nick at sites 17, 15, 11, and 8 nucleotides from the 5' end of this particular substrate, whereas visna virus IN preferentially cleaves to produce 19-, 15-, and 5-mer products (Fig. 7, lanes 3 and 4). Using our first set of six chimeras, we had previously mapped the outer limits of the region that determines target site selection for this activity to residues 50 and 190 (36).

View larger version (33K): [in this window] [in a new window]   FIG. 7.   Nonspecific alcoholysis activity of the new chimeric integrases. A schematic of the reaction is shown at the top. IN catalyzes attack by nucleophilic OH groups of various alcohols (ROH) that nick and join to newly exposed 5' phosphate groups at sites of DNA cleavage (asterisks denote 32P labels). Nicks occur at every site except those close to DNA ends, with a reproducible pattern of preferential sites that is a function of the target DNA sequence and source of IN. A 5'-labeled 23-mer of nonviral sequence was annealed to a complementary oligonucleotide, incubated for 90 min with the wild-type proteins or the new chimeras in the presence of 40% ethylene glycol, and analyzed as described in Materials and Methods. A sequence-specific oligonucleotide ladder (as markers) is in lane 1, and sizes (in nucleotides) are indicated at the left. Proteins are paired (a, e, f, and g) as in Fig. 1. Equal volumes of reaction mixtures were loaded so that relative intensities reflect efficiencies of the different proteins. Five of the six chimeras created patterns that segregated clearly to the HHH pattern (chimeras VHhv, HHhv, and HHvh, in lanes 6, 7, and 9, respectively) or to the VVV pattern (chimeras HVvh and VVvh, in lanes 5 and 8, respectively). Chimera VVhv (lane 10) was repeatedly inactive in this assay.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Integrase catalyzes insertion of a double-stranded DNA copy of the retroviral genome into host cell chromosomal DNA, contributes to the pathogenesis of AIDS, and is an attractive target for specific antiretroviral therapy. Identification of the regions of IN responsible for viral DNA specificity and for binding target DNA during viral DNA insertion is critical for understanding the organization of the protein, modeling its mechanism of action, and designing inhibitors of its activity. The results described in this report, obtained with an expanded set of functional chimeric lentivirus integrases, have further refined our understanding of the organization of the IN protein.

The region of IN responsible for viral DNA specificity. Despite early predictions, several lines of evidence indicate that the N-terminal region of IN does not contribute to viral DNA specificity. We previously reported that a chimeric protein with the N-terminal region of HIV-1 IN but the central and C-terminal regions of visna virus preferred the visna virus U3 substrate rather than the HIV-1 U3 substrate in processing assays (36); these two substrates unequivocally distinguish between the wild-type enzymes (36, 37). Similarly, a protein with the N-terminal region of HIV-1 IN but the central and C-terminal regions of human foamy virus IN was reported to process a human foamy virus DNA terminus but not an HIV-1 DNA end (53). Moreover, Rous sarcoma virus, HIV-1, and feline immunodeficiency virus IN proteins with various peptide replacements of the N-terminal region exhibit residual viral DNA processing and strand transfer activity (8, 9, 17, 41, 61).

The region of IN that interacts with host DNA. Although retroviral integration sites do not exhibit any sequence consensus, in vitro assays reveal preferential sites for viral DNA insertion (9, 55, 60, 63, 70). Nonspecific alcoholysis resembles viral DNA insertion in several respects, including the use of multiple sites, lack of a target site consensus, readily demonstrable cleavage site preferences that are a function of the integrase used, and avoidance of the ends of DNA targets (38). Thus, this activity offers a way to examine interactions between portions of IN and target DNA (36, 38). Our prior set of chimeric integrases mapped the selection of preferred alcoholysis sites to HIV-1 IN residues 50 to 190 (36), and the new chimeras have confirmed that result (Fig. 7). We showed elsewhere that an isolated protein fragment representing the central domain of HIV-1 IN (from residues 50 to 212) could catalyze nonspecific alcoholysis with a pattern of cleavage site preferences identical to that of the full-length protein (38). The same is true for a fragment representing residues 50 to 190 (40). Thus, the central region of IN is sufficient to position nonviral DNA for nucleophilic attack. Given the mechanistic similarities between nonspecific alcoholysis and insertion of viral DNA ends, these data suggest that the central region of IN positions host DNA for retroviral integration. A role for the central region in interactions with host DNA is supported by the finding that mutations of Asn-120 alter the integration site preferences of HIV-2 IN (64).

A new working model for the functional organization of IN. We propose an organization of functional domains on the IN protein that is different from prior models (Fig. 8). The key features are as follows. (i) Newly synthesized viral DNA initially binds in a nonspecific manner to the C-terminal region of IN. This suggestion is based on the nonspecific DNA binding properties of the C-terminal region, the fact that viral DNA is the only DNA present within the preintegration complex, and the ability to map target site selection for small alcohols but not for viral DNA ends to the central region of IN. (ii) Amino acids within the central region of IN (defined by residues 50 to 212) provide specificity by recognizing the viral DNA termini for the two endonuclease events that occur during integration (when viral DNA acts as the specific target for processing and then as the specific nucleophile during insertion). This new assignment is a direct result of the processing data presented here. Whether residues 191 to 212 are responsible for this function, as we initially hypothesized, awaits further evidence. Although Fig. 8 highlights only the final four viral DNA base pairs, we have shown that nucleotides in the fifth and sixth positions play dominant roles in viral DNA recognition (37) and likely interact directly with the part of IN that provides specificity for viral DNA. (iii) Host DNA binds in the central region of IN, as suggested by nonspecific alcoholysis assays. The limits of this site are defined by residues 50 and 190, based on the mapping analysis using chimeras as well as the activity of an isolated protein fragment of IN (40). (iv) Amino acids in the carboxyl half of the active site participate in positioning the nucleophile for catalysis, since residues near Asp-116 and Glu-152 influence the selection of different nucleophiles during processing (65).

View larger version (63K): [in this window] [in a new window]   FIG. 8.   Model for the functional organization of the IN protein. Protein monomers A and B (shaded ovals) are shown as a dimer, in parallel orientations, with a linear map of amino acid positions at the extreme left, numbered from the N terminus to the C terminus. Initially (left), viral DNA is bound nonspecifically by the C-terminal region of IN. Amino acids in the central region (perhaps residues 191 to 212) recognize the terminal six positions of the viral DNA sequence (only the final four positions are shown) and position the CA bases for attack by a water molecule that is positioned by the C-terminal portion of the active site (defined by the three invariant acidic amino acids in the central region). The terminal 3'-OH of the viral DNA itself can be used instead of water to release cyclic (GT) dinucleotides (21). In the nucleus (right), host DNA is bound by other portions of the central region of IN. Following a conformational change in the protein, each processed viral DNA end is positioned by the C-terminal portion of the active site, with the aid of the residues responsible for viral DNA recognition and perhaps the C-terminal region, to attack the host DNA. During nonspecific alcoholysis, small alcohols can be used by the active site without requiring residues outside of numbers 50 to 190. The model would accommodate actions occurring in trans or involving multimers larger than dimers.