A Metal-induced Conformational Change and Activation of HIV-1 Integrase*

Retroviral integrases are composed of three independently folding domains whose organization relevant to one another is largely unknown. As an approach to understanding its structure, we have investigated the effect of the required metal cofactor(s), Mn2+ or Mg2+, on the conformation of human immunodeficiency virus type 1 (HIV-1) integrase (IN) using monoclonal antibodies (mAbs) that are specific for each of these three domains. Upon the addition of increasing concentrations of the divalent cations to immobilized HIV-1 IN in ELISA assays, binding of mAbs specific for either the C-terminal domain or for an epitope in the catalytic core domain was lost, whereas binding of an N terminus-specific mAb was unaffected. Size exclusion chromatography of a nonaggregating derivative of HIV-1 IN showed that the oligomeric state of the protein did not change under conditions in which recognition of the core and C terminus-specific mAbs was lost. Preincubation with Mn2+ increased the resistance of HIV-1 IN to proteolytic digestion and produced a digestion pattern that was significantly different from that observed with the apoprotein. A derivative that lacked the N-terminal domain, IN(50–288), exhibited the same metal-dependent changes observed with the full-length protein, whereas the isolated catalytic core domain IN(50–212) did not. From this we conclude that the metal-induced conformational change comprises a reorganization of the core and C-terminal domains. Preincubation with Mn2+ increased the specific activity of HIV-1 IN 5-fold. Enzymatic activity was inhibited by the conformation-sensitive C terminus-specific mAb, but this inhibition was reduced greatly if the enzyme was first preincubated with metal ions. Thus, it appears that apo-HIV-1 IN exists predominantly in an inactive conformation that is converted into a catalytically competent form upon the addition of metal ions.

Integration of viral DNA into the host chromosome is a critical step in the life cycle of retroviruses, as it ensures efficient expression and perpetuation of the viral genome. This essential step is catalyzed by integrase (IN), 1 a viral enzyme that is both necessary and sufficient for the integration reaction in vitro (for a review, see Refs. 1 and 2). Integration has been studied in vitro with model DNA substrates that include the sequence at viral DNA ends. Using these substrates and purified integrases, two separate steps have been delineated: viral DNA end processing and its subsequent joining to a target DNA (3)(4)(5). The processing step comprises specific removal of the terminal dinucleotide from the 3Ј-end of the model substrates, thereby exposing a hydroxyl group (OH) that serves as the nucleophile in the subsequent joining step. The joining reaction is a concerted cleavage-ligation, which proceeds via a direct transesterification reaction involving processed viral DNA ends and a target DNA (6). A divalent metal cofactor (Mn 2ϩ or Mg 2ϩ ) is required for both reactions.
The retroviral integrases are made up of three independently folding domains (Fig. 1) as shown by spectroscopic and crystallographic studies (for a review, see Ref. 7). The N-terminal domain comprising approximately the first 50 amino acids, is characterized by the presence of an HHCC zinc finger-like motif. This domain is capable of binding zinc ions (8 -10), which probably facilitates its proper folding, thereby promoting oligomerization of HIV-1 IN. Interestingly, avian sarcoma virus (ASV) IN does not require the zinc finger domain for the processing and joining of viral DNA ends in vitro (11,12). The core (catalytic) domain includes the characteristic D,D(35)E motif; three essential acidic residues (Asp 64 , Asp 116 , and Glu 152 in HIV-1 IN), of which the last two are separated by 35 amino acids in the primary sequence. This motif is also conserved among retrotransposon integrases and some bacterial transposases (13,14). Although the isolated core domain cannot catalyze the processing or joining of viral DNA ends, it performs an apparent reversal of the joining reaction, referred to as disintegration, confirming that an active site resides in this domain (11,(15)(16)(17)(18). The C-terminal domain comprises the last approximately 80 amino acids. The amino acid sequence of this region is the most divergent among integrases. The C-terminal domains of ASV IN and HIV-1 IN have been shown to possess a nonspecific DNA binding capability (13, 19 -22) as well as determinants of multimerization (23,17). The fact that the core domain alone does not perform the bona fide reactions of the full-length integrases suggests that the termini contribute important structural components.
The structures of the core domains of HIV-1 (24) and ASV IN (25,26) show that both proteins fold with a similar architecture, which resembles that of several other nucleases and recombinases (27)(28)(29)(30). Both of these retroviral integrase core domains crystallize as dimers; this is consistent with previous biochemical experiments indicating that IN functions as a multimer, minimally a dimer (31)(32)(33). The proposal that the acidic residues of the D,D(35)E motif are involved in coordinating the required metal cofactor was supported by analysis of the crystal structures of Mn 2ϩ and Mg 2ϩ complexes of the ASV IN core domain (26). In these structures, a single divalent cation is bound between Asp 64 and Asp 121 . Two Zn 2ϩ ions are bound by the ASV IN core domain; one is coordinated by Asp 64 and Asp 121 and the other by Asp 64 and Glu 157 . 2 All three of these residues are components of the D,D(35)E motif.
Although the structural models of the core domains of HIV-1 and ASV IN and the C-terminal domain of HIV-1 IN (35,36) provide important new knowledge about these independently folding units, we still do not know how these domains are organized within the full-length monomeric IN or in a functional multimer. Recent studies suggest that metal ions can promote aggregation/oligomerization in HIV-1 IN and enhance enzyme activity (10,37,38,54); the biochemical basis of the metal effect, however, remains to be fully investigated. Furthermore, very little is known about potential dynamic changes in integrase that may occur during the processing and joining steps. As an initial approach to this question, we have used monoclonal antibodies (mAbs) and proteolytic enzymes as probes to investigate whether metal binding by HIV-1 IN induces conformational changes that affect the organization of these domains.

Materials
Hitrap heparin and Hitrap iminodiacetic acid (IDA) were purchased from Pharmacia Biotech Inc. High binding microtiter plates were procured from Corning-Costar. Goat anti-mouse alkaline phosphatase conjugate was obtained from Fisher. Proteinase K was purchased from Boehringer Mannheim. The enhanced chemiluminescent (ECL) immunoblotting kit was obtained from Amersham Corp.  (40), was constructed by oligonucleotide-directed mutagenesis following established methods (41). The F185K substitution was initially introduced into pET29bNY5IN. DNA from positive clones was subsequently introduced into Escherichia coli BL21(DE3) cells for protein expression. The C280S substitution was then introduced by a second round of oligonucleotide-directed mutagenesis to obtain the soluble double mutant, pET29bNY5IN F185K,C280S.

Methods
Construction of HIV-1 IN Deletion Mutants-Two primers were used in a polymerase chain reaction to amplify the sequences that encode residues of the core domain, amino acids 50 -212, or core and C-terminal domains, residues 50 -288 of HIV-1 IN. pET29bNY5IN F185K was used as the template DNA to include the solubility-promoting F185K substitution into these deletion constructs of HIV-1 IN. The correct construction of positive clones was verified by DNA sequencing.
Protein Purification-Bacterially expressed HIV-1 IN proteins were purified as follows. Typically, a 2-liter culture of the bacterial clone was grown to an A 600 of 0.7-0.8, and expression from the plasmid vector was then induced by the addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside. Cells were harvested 3 h postinduction by centrifugation at 10,000 ϫ g for 10 min at 4°C. Cells were lysed in 20 mM HEPES, pH 7.5, 5 mM imidazole, 2 M KCl by two passes through a French pressure cell at 18,000 p.s.i. at a concentration of 6 ml/g of wet cell paste. The lysate was subjected to centrifugation at 31,000 ϫ g for 30 min, and the soluble fraction was retained. The salt concentration of this fraction was adjusted to 1 M (binding buffer), and it was then applied to a nickelcharged IDA-agarose (Hitrap-IDA) column to which HIV-1 IN binds in the absence of a His tag. 3 The column was washed with 3 column volumes of binding buffer, and the protein was eluted with a linear gradient of imidazole. The IDA-purified protein was subsequently chromatographed on a heparin-agarose column (Hitrap-heparin) equilibrated in 20 mM HEPES, pH 7.5, 250 mM KCl, 1 mM dithiothreitol (binding buffer). The column was washed with 3 column volumes of binding buffer followed by 8 column volumes of a linear gradient of KCl (0. 25-1 M) to obtain a homogenous preparation that was stored in 20 mM HEPES, pH 7.5, 1 mM dithiothreitol, 500 mM KCl, and 40% (v/v) glycerol at Ϫ20°C.
The isolated catalytic core, IN)50 -212), was initially expressed as a His tag fusion protein. Following purification on a nickel-charged IDA column, the His tag was removed by digestion with thrombin, and the nonfused protein was recovered by following the manufacturer's (Novagen) suggested protocol.
Enzyme-linked Immunosorbent Assay (ELISA)-The ELISA protocol, in which HIV-1 IN is immobilized on the microtiter plates, has been described previously (42). Each well was coated with 1.5 nmol of protein in a total reaction volume of 50 l in TBS. A specified concentration of divalent cation or an ionic equivalent of monovalent cations was first added to the immobilized IN. The ionic equivalent of buffers was calculated from the equation, where M i represents molarity and Z i represents charge of the ion and was verified by measuring their conductivities. This was followed (within ϳ2-5 min) by the addition of comparable titers of the anti-HIV-1 IN mAbs (42). To determine whether the metal-induced change was reversible, 5 or 30 min after the addition of Mn 2ϩ , the wells were washed five times with 200 l of TBS or TBS containing 0.1 mM EDTA. After washing, incubation with anti-HIV-1 IN mAbs and all subsequent steps were identical to our standard ELISA assay. All buffers were treated with Chelex (Bio-Rad) to remove adventitious metals. Aggregation/Sedimentation Assay-Conditions described by Ellison et al. (37) were followed without modifications. Briefly, 100 nM HIV-1 IN was incubated with or without 5 mM MnCl 2 at room temperature for 5 min. The reactions were subjected to centrifugation at 10,000 ϫ g for 10 min at 4°C. The supernatant and pellet fractions were separated, and their protein content was determined by immunoblot analysis with anti-HIV-1 IN mAb 4.
Size Exclusion Chromatography-The oligomeric state of HIV-1 IN protein was analyzed on a tandem QC-PAK TSK GFC 300GL/200GL (15 cm ϫ 8 mm inner diameter) column in 20 mM Tris⅐HCl, pH 7.5, 500 mM KCl. Absorbance at 280 nm was followed on a Rainin high pressure liquid chromatography system at a flow rate of 0.9 ml/min. All samples were subjected to centrifugation at 10,000 ϫ g for 5 min in a microcentrifuge prior to injection. Two different sample volumes (20 and 400 l) at two concentrations (10 and 1 M, respectively) of the samples were analyzed in duplicate to access potential dilution effects upon injection. Comparable results were obtained with both concentrations. Metal preincubation was performed in 5 mM MnCl 2 for several hours (more than 2 h in all cases) prior to analysis. The protein samples that had been preincubated were also analyzed with 5 mM MnCl 2 in the running buffer. The size exclusion column was calibrated with the following globular protein markers (the molecular mass and mean retention time Ϯ S.D. of three independent determinations are reported): thyroglobulin (670 kDa, 7.45 Ϯ 0.03 min), ␥-globulin (158 kDa, 8.59 Ϯ 0.02 min), bovine serum albumin (68 kDa, 9.37 Ϯ 0.02 min), ovalbumin (44 kDa, 10.15 Ϯ 0.04 min), carbonic anhydrase (29 kDa, 11.33 Ϯ 0.12 min), myoglobulin (17 kDa, 11.77 Ϯ 0.05 min). The retention times of the protein standards were unchanged when analyzed with 5 mM MnCl 2 in the running buffer. A void volume eluting at 6.82 Ϯ 0.02 min was determined with blue dextran, while a total volume at 14.85 Ϯ 0.05 min was returned by paranitrophenyl phosphate.
Limited Proteolysis-A time course of limited proteolysis was carried out in a 75-l reaction containing 20 mM HEPES, pH 7.5, 100 mM NaCl, 10% (v/v) glycerol, and 10 g of IN at 37°C in the presence or absence of 10 mM MnCl 2 . Proteinase K at a final concentration of 4 ng/l, was used in all incubations. At appropriate times (indicated in the figures), a 15-l sample was removed and quenched by boiling at 95°C for 5 min with SDS-polyacrylamide gel electrophoresis loading buffer. The protein samples were subjected to electrophoresis in a 15% polyacrylamide gel and either silver-stained (Bio-Rad) or blotted onto a nitrocellulose membrane with a Sartorius blotting apparatus at 0.65 mA/cm 2 . Immunoblot analysis was performed with an enhanced chemiluminescent (ECL) kit from Amersham according to the manufacturer's instructions.
Integrase Assays-The activity of HIV-1 IN was determined as reported previously (18) using equimolar (1 M) amounts of enzyme and substrates in 20 mM HEPES, pH 7.5, 10% (v/v) glycerol, 10 mM MnCl 2 , 6.67% (v/v) Me 2 SO, and 50 mM KCl or NaCl at 37°C. In following the time course of the activity of HIV-1 IN (preincubated with or without 10 mM Mn 2ϩ for 5 min), a 5-l sample of the reaction was removed and quenched with an equal volume of loading buffer at the times indicated in Fig. 9. The quenched samples were subjected to electrophoresis in a 20% denaturing polyacrylamide gel. Quantitation was obtained after exposing the gel to an imaging plate, using a Fuji MacBAS 1000 imaging system. mAb inhibition of HIV-1 IN was conducted under normal assay conditions (18) except that the monoclonal antibodies at the concentrations indicated in the figure legends were included. Other modifications, if any, are also indicated in the figure legends.

Metal-dependent Changes in mAb Binding to Wild-type HIV-1 IN-
The requirement for a Mn 2ϩ or Mg 2ϩ metal cofactor for the activity of retroviral integrases is well documented (3,4,6). To investigate the possibility that divalent cations fulfill a structural as well as catalytic role, we first asked whether mAbs could serve as probes of conformational changes in HIV-1 IN. Such an approach has been employed successfully to identify and characterize conformational changes and reaction intermediates in a number of proteins, including human thrombospondin (43), tryptophan synthase (44), thrombin (45), and carboxypeptidase A (46).
The anti-HIV-1 IN mAbs used in our experiments were previously characterized (42); as illustrated in Fig. 1, mAb 17, mAb 4, and mAb 33 recognize epitopes in the N-terminal domain, the core (catalytic) domain, and the C-terminal domain of HIV-1 IN, respectively. In our assay ( Fig. 2A), HIV-1 IN was immobilized on the surface of the ELISA plates prior to the addition of metal ions. After a short incubation (2-5 min), mAb was added, and subsequent ELISA procedures were followed. Using this ELISA assay, we asked if the presence of the divalent ion, Mn 2ϩ , could affect the binding of each of the mAbs to immobilized HIV-1 IN. We found (Fig. 2B) that binding by mAb 17 was relatively unaffected by Mn 2ϩ , whereas binding of mAb 4 and mAb 33 decreased with increasing concentrations of Mn 2ϩ . A 50% reduction in HIV-1 IN binding by both mAb 4 and mAb 33 was observed at a concentration of approximately 1 mM. The loss of HIV-1 IN recognition by the core and C terminus-specific mAbs was not a result of increasing ionic strength, as equivalent ionic strength, maintained with increasing concentrations of monovalent Na ϩ ions, failed to elicit a similar response (Fig. 2C). K ϩ ions also failed to produce the observed phenomenon (data not shown). On the other hand, other divalent cations (i.e. Mg 2ϩ and Ca 2ϩ ) also caused a concentrationdependent decrease in the binding of mAb 4 and mAb 33 to HIV-1 IN without affecting the binding of mAb 17. However, the concentration of Ca 2ϩ required to achieve 50% reduction in mAb binding was very high, in the range of 50 -70 mM. The order of effectiveness in eliciting the metal-dependent changes was Mn 2ϩ Ͼ Mg 2ϩ Ͼ Ͼ Ca 2ϩ . The low solubility of Zn 2ϩ under our assay conditions prevented an accurate assessment of its effects (1 mM Zn 2ϩ had no effect on mAb 33 recognition of HIV-1 IN).
The Metal-induced Change in IN Conformation Is Reversible-The concentration dependence of the metal-induced conformational change in HIV-1 IN suggests a reversible thermodynamic phenomenon. To test this directly, the ELISA protocol was modified to include a wash step between applications of the Mn 2ϩ and the mAbs (Fig. 3A). The incubation with metal ions was carried out for 30 min to exclude slow time-dependent effects. The results (Fig. 3B) showed that introduction of the wash step (with or without EDTA) restored binding of mAb 33 to HIV-1 IN. Similar results were obtained with mAb 4 (not shown). Full recognition by mAb 4 and mAb 33 was also obtained after a 5-min incubation with Mn 2ϩ when followed by a  Fig. 1, using ELISA assays as described under "Experimental Procedures." B, the effect of increasing Mn 2ϩ concentration on mAb binding to HIV-1 IN was examined. C, an equivalent ionic strength was maintained with NaCl, and its effect on mAb binding was similarly assessed. The following symbols represent the indicated mAbs. Filled circles, mAb 17; filled squares, mAb 4; filled triangles, mAb 33.

FIG. 1. A linear model of HIV-1 IN showing the three major domains and their characteristic features or functions.
Horizontal arrows delimit the domain borders. The evolutionary conserved amino acids are shown within the domains with the residue numbers above them. The mAbs used in these studies are identified under the regions known to include their epitopes, depicted as rectangular boxes (42). wash (not shown), thus establishing the reversibility of the conformation change.
Metal-dependent Aggregation versus Specific Conformational Change-The fact that HIV-1 IN is immobilized on a microtiter plate makes it unlikely that protein oligomerization is the underlying cause of the observed metal-dependent change. However, we wished to distinguish this structural change from a metal-dependent aggregation phenomenon in HIV-1 IN reported by Ellison et al. (37). Thus, we constructed a mutant clone that expresses HIV-1 IN with the solubility-promoting substitutions F185K and C280S. As reported (17), we found this protein to be quite soluble (Ͼ10 mg/ml) and as enzymatically active as the wild-type protein in vitro. Following the protocol of Ellison et al. (37), we analyzed the sedimentation properties of both the wild-type and soluble HIV-1 IN proteins in the presence and absence of a divalent cation. We confirmed the finding that Mn 2ϩ can induce aggregation of wild-type HIV-1 IN. However, under identical experimental conditions, the soluble protein, IN F185K,C280S, did not exhibit this property (Fig. 4).
To address the question of aggregation directly, size exclusion chromatography was employed to examine the oligomeric state of wild-type and nonaggregating HIV-1 IN F185K,C280S proteins (purified and dialyzed in the presence of EDTA). was incubated and analyzed in the presence of 5 mM Mn 2ϩ (Fig.  5B), a retarded retention time of 10.08 was observed; this value corresponds to an apparent molecular mass of 52.3 kDa. As the total amount of protein injected is recovered in the eluted fraction, the simplest explanation for the observed retarded mobility in the presence of Mn 2ϩ is that the dimeric protein assumes an altered or a more compact conformation. A similar elution profile was observed if this protein was preincubated and analyzed with Mg 2ϩ (data not shown). When HIV-1 IN F185K,C280S was preincubated with Mn 2ϩ or Mg 2ϩ but analyzed in the absence of the divalent cation, elution profiles (retention times 9.85 and 9.84 min, respectively) identical to that of the apoprotein were obtained, confirming that the phenomenon is, indeed, reversible. From these results, we conclude that the addition of divalent cations does not cause a change in the oligomeric state of IN (F185K,C280S); rather, a reproducibly slower migrating species is observed.
The oligomeric state of wild-type HIV-1 IN was examined in a similar fashion. In the absence of Mn 2ϩ , wild-type HIV-1 IN was found to elute with a profile that suggests an equilibrium among higher order oligomers, tetramers, dimers, and a "tail" of monomers (Fig. 5C). The predominant species eluted with a retention time of 9.91 min, indicative of a protein of molecular mass 58.9 kDa, approximately that expected for a dimer. The tendency for wild-type HIV-1 IN to interact nonspecifically with several resins and chromatographic matrices 4 may account for the small difference in retention times for the main peaks of mutant and wild-type proteins. When incubated and analyzed in the presence of 5 mM MnCl 2 , ϳ40% of the wild type protein that eluted ahead of the putative dimer peak was missing (cf. Fig. 5, C and D). This loss can be accounted for by the amount pelleted upon centrifugation of the sample prior to injection. This result is consistent with the metal-induced aggregation phenomenon previously reported for wild-type HIV-1 IN (37) and illustrated in Fig. 4. As with HIV-1 IN F185K,C280S, the amount of putative dimeric species resolved by chromatography was the same in the absence or presence of Mn 2ϩ . In addition, the elution of the "dimeric" species of wildtype HIV-1 IN was also retarded, with a retention time of 10.33 min, which corresponds to a 44.4-kDa protein. Thus, apart from its tendency to aggregate in the presence of divalent metal ions (which appears to involve primarily the higher order oligomers), the results obtained with wild-type HIV-1 IN (Fig. 5, C and D) are similar to those observed with HIV-1 IN F185K,C280S (Fig. 5, A and B). As a control experiment, all of the protein markers used in calibrating the column were also analyzed in the presence of 5 mM MnCl 2 . All maintained identical retention times under these conditions. Thus, the retarded  Fig. 6 show that the metal-dependent differential binding observed with the wild-type enzyme can be reproduced with the soluble protein; Mn 2ϩ reduced the binding of mAb 4 and mAb 33 by 50% at ϳ1 mM concentration but did not affect the binding of mAb 17 (Fig. 6A). The mAbs bound to the soluble protein as efficiently as to the wild-type protein in the presence of increasing NaCl (Fig. 6B). From these results, we conclude that the conformational changes observed with wild-type HIV-1 IN and the soluble derivative are distinct from the previously reported metal-dependent aggregation phenomenon.
Metal-dependent Resistance to Proteolytic Digestion-To determine if a structural change could also be detected by an independent method, we subjected wild-type IN and IN F185K,C280S to limited digestion with the broadly specific proteinase K (47,48) and analyzed the products by electrophoresis and immunoblotting (Fig. 7). In the absence of Mn 2ϩ , HIV-1 IN was digested completely within 2 h under our standard IN assay conditions (Fig. 7A). On the other hand, virtually all of the HIV-1 IN could still be accounted for after a 2-h digestion in the presence of the divalent cation. Some fulllength IN was even detected after 18 h of incubation, implying that Mn 2ϩ stabilizes the enzyme against this type of proteolysis. All three of the domain-specific mAbs were used to probe the proteolytic digestion products of HIV-1 IN. The faster migrating product (Fig. 7A, dashed arrow) was detected by C terminus-specific mAb 33 (as shown) as well as the core-specific mAb 4, but not by N terminus-specific mAb 17. Thus, the N-terminal region appears to be most accessible to proteinase K. We also examined the activity of proteinase K on the catalytically and structurally related ASV IN with products visualized by polyclonal antibody (Fig. 7B). In this case, Mn 2ϩ had no significant effect on the sensitivity or pattern of digestion. Thus, it is clear that the activity of proteinase K is unaffected by the divalent cation. We conclude from this result that HIV-1 and ASV integrases differ in their response to metal binding.
To exclude the possibility of metal-induced aggregation as a basis for the increased resistance of wild-type HIV-1 IN to proteolysis, we also subjected the soluble, nonaggregating derivative IN F185K,C280S to a similar analysis. To visualize fragments not detected by immunoblotting, one-half of the reaction was also analyzed by silver staining. The mAb 33 im-   (Fig. 7C) shows an increased resistance to proteolysis upon the addition of Mn 2ϩ as was observed with the wild-type protein. In addition, a different pattern of proteolytic products was observed in the presence of Mn 2ϩ . This difference is most striking in the silver-stained gel (Fig. 7D). These analyses revealed three classes of fragments, denoted I, II, and III. Protein products in the group of larger fragments, band I, were significantly more resistant to proteolysis in the presence of Mn 2ϩ and appeared to be degraded at the same rate, suggesting that they all include the same resistance-determining elements of the protein. The intermediate sized fragments in band II are the most prominent products in the presence of Mn 2ϩ . The results illustrated in Fig. 7D and the analysis of products from time points earlier than those in the figure (i.e. within 10 min) showed little production of the band II fragment in the absence of Mn 2ϩ . Digestion of band II fragments appears to generate band III fragments. Immunoblotting with mAb 33 (Fig. 7C) shows that components of band I are recognized by this C terminus-specific antibody, whereas those in bands II and III are not. Immunoblotting was also performed using mAb 17 and mAb 4 (data not shown). The results showed that the N terminus-specific antibody recognized only the largest of the fragments in band I that were produced in the presence of Mn 2ϩ and none of the fragments produced in the absence of Mn 2ϩ . The catalytic core-specific mAb 4 recognized all of the IN-derived proteolytic fragments in bands I, II, and III and showed a metal-dependent pattern similar to that obtained with the silver stain. The different sensitivities to proteolysis and differential mAb recognitions suggest the induction of significantly altered protein conformation upon the addition of Mn 2ϩ . These results are consistent with both the ELISA and size exclusion chromatographic analyses.
The Metal-induced Conformational Change Involves the Core and C-terminal Domains-Both the ELISA and limited prote-olysis assays suggested that the N-terminal region was not required for the metal-induced conformational change. Therefore, we asked if the core domain alone or together with the C-terminal domain would exhibit this change. Two deletion mutants, encoding only the core domain, IN)50 -212)F185K, or the core plus the C-terminal domain, IN)50 -288)F185K, were constructed, and the corresponding proteins were prepared for analysis. Unlike the full-length protein, the addition of Mn 2ϩ to an isolated core domain did not affect its recognition by the core-specific mAb 4 (Fig. 8A). When subjected to proteolysis, only a slight increase in resistance of a faster migrating species was observed; this may reflect a local ordering of the core domain in the presence of the divalent cation (Fig. 8B). The almost identical pattern of proteolytic products observed in the presence and absence of the metal cofactor further confirmed that proteinase K was not directly affected by the metal ions.
In contrast to results with the core domain, the addition of metal to IN)50 -288)F185K diminished binding of mAb 33 (Fig.  8C). The concentration dependence of this effect resembled that observed with the full-length protein (cf. Fig. 2B). Inspection of the silver-stained gel of the resolved products of proteolysis (Fig. 8D) revealed a digestion pattern similar to that observed with the full-length protein with a transient build-up of band II and its apparent subsequent degradation to band III in the presence of Mn 2ϩ . Fragments corresponding to band I are not among the products of digestion, as IN)50 -288)F185K lacks the N-terminal domain. It is noteworthy that the intact IN)50 -288)F185K is relatively resistant to proteolysis even in the absence of Mn 2ϩ . We conclude from results with these two truncated proteins that HIV-1 IN undergoes a metal-induced structural reorganization that involves both the core and Cterminal domains.
Activation of HIV-1 Integrase by Preincubation with Mn 2ϩ -To investigate the functional role of the metal-induced conformational change(s) in HIV-1 IN, we compared the catalytic activity of protein preincubated for 5 min with Mn 2ϩ to a sample that had no prior exposure to metal ions. To circumvent potential aggregation effects (Ref. 36; Fig. 4), HIV-1 IN F185K,C280S was used for this experiment. Fig. 9 shows a time course for the processing and joining activities of IN preincubated with or without metal and assayed under identi-cal conditions. Preincubating with Mn 2ϩ resulted in an approximately 5-fold increase in processing activity when the early (5-min) time points for both samples were compared (Fig. 9C). The plateau phase of the time course reflects a steady state where production of the processed product is balanced by its utilization for joining. This is clearly observed with longer exposure (Fig. 9B), which shows increasing amounts of joined products after the 15-min time point where the plateau begins (Fig. 9A). This result also shows that joining is more efficient when the enzyme is preincubated with Mn 2ϩ . The lower specific activity of the protein that had no prior contact with metal is consistent with a rate-limiting requirement for a change in protein conformation. The yield of product ultimately reaches that with the more active protein with longer incubation times (not shown). mAb 33 Inhibits HIV-1 IN Activity by Preventing the Metalinduced Conformational Change(s)-As the presence of metal ions reduces the binding of mAb 33 to HIV-1 IN, we next asked if preincubation with metal ions could affect the ability of mAb 33 to inhibit enzyme activity. As illustrated in Fig. 10, mAb 33 inhibits the processing, joining, and disintegration activities of HIV-1 IN F185K,C280S, and the inhibition is concentrationdependent (e.g. Fig. 10A). A mAb of the same isotype but specific for the Ras oncoprotein had no effect on the processing (not shown), joining, or disintegration activities of HIV-1 IN F185K,C280S (Fig. 10, B and D). As expected, mAb 33 did not inhibit the disintegration activity of the core protein, IN-(50 -212), which lacks its epitope (data not shown).
The observation that mAb 33 inhibits the catalytic activities of HIV-1 IN in vitro suggested that the antibody may stabilize its "open" or inactive conformation. To test this hypothesis directly, we modified our ELISA assay by preincubating HIV-1 IN with the antibody for approximately 5 min prior to the addition of the metal cofactor (cf. Fig. 2A). Under these conditions, in which the usual order of addition of mAb and metal ion was reversed, a 30-fold higher concentration of Mn 2ϩ ions was required to achieve a comparable 50% reduction in mAb 33 binding (Fig. 11A). A similar stabilization of the "open" conformation was also observed with mAb 4 (data not shown).
If conformational change(s) play a significant role in HIV-1 IN catalysis, we hypothesized that preincubating the enzyme with the metal cofactor should also abolish or reduce significantly the extent of inhibition by mAb 33. Results illustrated in Fig. 11B show that when the enzyme is preincubated with Mn 2ϩ , there is only an ϳ10% decrease in activity upon the addition of mAb 33 (compare lanes 1 and 2). If the enzyme is preincubated with mAb 33 or if the metal and antibody are added simultaneously, the enzyme activity of IN is decreased by ϳ80% (compare lanes 1, 3, and 4). In contrast, no enzymatic activity was detected in the presence of mAb 17, regardless of the order of the addition of metal ions and antibody (Fig. 11B,  right part, lanes 2-5). This is consistent with our observation that recognition of HIV-1 IN by mAb 17 is unaffected by metal ions (see Figs. 2 and 6). These results suggest that mAb 33 inhibits HIV-1 IN by preventing the enzyme from adopting a metal-induced and enzymatically activating conformational change, while mAb 17 inhibits the enzyme by an alternate mechanism.

DISCUSSION
The interaction between an antibody and its cognate antigen is of such avidity that it is essentially irreversible under physiological conditions. Here, in an assay in which IN proteins are immobilized on the surface of an ELISA plate, we have shown that the addition of a divalent cation can prevent the binding of mAbs specific for the catalytic core or C-terminal domain of HIV-1 IN (Figs. 2, 3, and 6). This binding is unaffected at  Fig. 2. Filled symbols, Mn 2ϩ ; open symbols, Na ϩ equivalent ionic strength. Conditions for limited proteolysis are as described in Fig. 7. comparable ionic strengths maintained with monovalent cations. The fact that binding of an N-terminus-specific mAb was also unaffected by the presence of divalent cations indicates that this domain remains largely accessible under these conditions and does not appear to undergo any gross structural change. This view is supported by results from our limited proteolysis experiments (Fig. 7), which show that the N-terminal domain is one of the first regions of the protein to be digested by proteinase K in the presence or absence of metal ions. The ability of the N-terminally truncated protein IN)50 -288)F185K to exhibit the metal-induced conformational change lends further credence to this interpretation (Fig. 8, C and D). On the other hand, the isolated core domain exhibited no significant change in proteolytic sensitivity and no change in recognition by the same antibody (mAb 4) that showed decreased binding to the full-length IN in the presence of metal ions (Fig. 8, A and B). These results suggest that the metalinduced conformational change involves a reorganization of the core and C-terminal domains.
The use of a more soluble and uniform nonaggregating protein, HIV-1 IN F185K,C280S, allowed a clear distinction between a specific metal-induced conformational change and the metal-induced aggregation reported by Ellison et al. (37); the aggregation is eliminated by these substitutions, but the metalinduced conformational changes are not (Figs. 4 and 6). The two phenomena are further distinguished by the fact that the metal-induced conformational change described here is independent of the N-terminal "zinc finger" domain of HIV-1 IN (Fig. 8, C and D). This observation also distinguishes the conformational change described here from the recently reported enhancement of oligomerization of HIV-1 IN by zinc ions, which is dependent on the presence of the N-terminal domain (10). We have shown by size exclusion chromatography of the soluble, nonaggregating derivative IN F185K,C280S that preincubation with metal ions under conditions in which the conformational change is detected does not lead to a change in the oligomeric state of the protein (Fig. 5). Thus, the interpretation that we currently favor is that the change reflects specific rearrangements of regions within the core and C-terminal domains, possibly mediated by a flexible hinge or loop that connects the two. Our biochemical assay suggest that this rearrangement induces the formation of a more catalytically competent enzyme. From these results, we hypothesize that HIV-1 IN exists in at least two conformations: an "open" conformation in the absence of the metal cofactor and a more compact or "closed" state upon the addition of metal ions. As HIV-1 IN is dimeric under our assay conditions, we do not know if the metal-induced domain reorganization involves inter-or intramolecular interactions. It is possible that both types of interactions may contribute to the changes we detect.
Our enzymatic assays ( Fig. 9) indicate that preincubation with Mn 2ϩ relieves a slow step in the reaction catalyzed by HIV-1 IN. This activation of HIV-1 IN under conditions that induce the conformational change is consistent with a role for this structural reorganization in catalysis. Enhanced activation of HIV-1 IN by metal preincubation has been observed previously with the aggregating wild-type protein (37,38,54). Our results with the more soluble, nonaggregating protein suggest that the increased specific activity reflects the change in the conformation of the protein. The increased enzymatic activity with metal preincubation lasted beyond 10 min, ample time for the structural change to be induced by the metal introduced with the assay mixture. The basis for the extended effect of preincubation is unclear; however, other factors including the binding order of ligands may contribute to the lag observed without metal preincubation. Further support for the effect of metal-induced conformational change(s) on the catalytic properties of HIV-1 IN is provided by our inhibition assays (Figs. 10 and 11). Inhibition by the conformation-sensitive mAb 33 is reduced significantly when the enzyme is preincubated with Mn 2ϩ . These results suggest that mAb 33 inhibits HIV-1 IN by binding and stabilizing the more "open," inactive conformation of the enzyme, thereby preventing it from adopting the conformational change(s) necessary for catalysis. Because the FIG. 9. The effect of preincubation with Mn 2؉ on the enzymatic activities of HIV-1 IN F185K,C280S. Enzyme was preincubated with or without 10 mM Mn 2ϩ for 5 min. Then substrate or substrate plus 10 mM Mn 2ϩ was added to initiate the reactions. Samples were removed at the indicated times and analyzed on a sequencing gel as described under "Experimental Procedures." As illustrated in D, in the processing reaction, two nucleotides are removed from the 3Ј-end of the 21-mer duplex oligonucleotide substrate by the enzyme to generate a product shortened by 2 nucleotides. The joined products result from a nucleophilic attack by the 3Ј-hydroxyl (OH) of the Ϫ2 processed product on other substrate molecules that serve as target DNA surrogates. The nonspecific nature of the target site selection leads to a ladder of joined products that are longer than the substrate. The position of the Ϫ2 processed product is indicated in A. B shows a longer exposure of the autoradiogram necessary to detect joined products. Quantitation of the Ϫ2 cleavage product with a Fuji MacBAS 1000 imaging system is presented in C. Filled squares, products after preincubation with Mn 2ϩ ; open squares, products after preincubation without metal. epitope of mAb 33 lies within the nonspecific DNA binding domain of the C terminus, another possibility is that mAb 33 prevents binding of the substrate DNA to the enzyme. However, results from our ELISA assays indicate that the binding of substrate DNA and mAb 33 are not mutually exclusive. 4 It is perhaps significant that total inhibition of the enzymatic activity of HIV-1 IN F185K,C280S was not observed at an mAb 33:IN (monomer) molar ratio of 2:1 (Fig. 11). This could be explained by a thermodynamic equilibrium in which a proportion of HIV-1 IN may exist in the active, "closed" conformation under our assay conditions.
It is interesting to note that the epitope recognized by mAb 4 is located within a region that includes residues 141-153 (see Fig. 1). This region contains the conserved and catalytically essential Glu 152 , and it spans what appears to be a flexible loop in the crystal structures of both HIV-1 and ASV IN catalytic core domains. Because recognition of this epitope by mAb 4 is lost in the full-length protein or a derivative lacking the Nterminal domain in the presence of either Mn 2ϩ or Mg 2ϩ , it seems possible that this flexible loop region may be hidden or become stabilized or more ordered in the conformation induced by the binding of the divalent cation. A comparison of the crystal structures of the core domains of HIV-1 and ASV integrases reveals that they fold with a similar architecture. Although an x-ray crystal structure of a metal⅐HIV-1 IN complex is not currently available, metal complexes of the ASV IN catalytic core domain confirm that the D,D(35)E motif comprises metal-coordinating residues. As the N-terminal domain of HIV-1 IN is not required for the conformational change, it seems possible that the change is elicited by the binding of metal ions to the D,D(35)E residues in the active site. What, then, might be the role of the C terminus in the metal-dependent phenomenon? We speculate that binding of metal cofactors to the active site may promote an interaction between the nonspecific DNA binding domain of the C terminus and the core (catalytic) domain that is favorable for catalysis. Further work may reveal precisely how this is accomplished.
Our results show that Mn 2ϩ does not affect the sensitivity of ASV IN to protease digestion, suggesting that this protein may normally exist in a favorable conformation. Inspection of the structures of metal complexes of the ASV IN core domain shows an almost identical spatial positioning (root mean square deviation Ͻ1 Å) in the presence or absence of one or two bound metal ions in the active site. 2 Thus, full-length ASV IN appears to possess preformed metal coordinating pockets and may not were used in the assays. p refers to the assay components that were preincubated prior to initiation of the reaction. ϩ and Ϫ refer to components that were present or absent, respectively, in the assays. require a major conformational change to bind its metal cofactors. ASV IN is significantly more active than HIV-1 IN, on a mol/mol basis. The avian enzyme has also been shown to turn over in processing activity assays (31), while efforts to demonstrate turnover with HIV-1 IN have not been successful. 4 As the specific activity of ASV IN does not change upon preincubation with metal ions, 4 the requirement for HIV-1 IN to undergo a metal-induced conformational change may be responsible for some or all of these differences.
Except under certain conditions (49 -51), considerably more activity is observed with Mn 2ϩ than with Mg 2ϩ for the retroviral integrases in vitro. However, Mg 2ϩ is generally assumed to be the biologically relevant cofactor for these enzymes, because this cation is maintained at higher concentration in the cytoplasm of living cells (ϳ10 Ϫ3 M "free" Mg 2ϩ versus Ͻ10 Ϫ7 M Mn 2ϩ ). We estimate that ϳ25% of the HIV-1 IN protein could be induced to assume a catalytically competent conformation at the intracellular concentration of Mg 2ϩ . Thus, the changes observed in our in vitro studies could proceed under "physiological" conditions. IN is brought into cells and performs its critical functions as a part of large nucleoprotein complexes whose precise compositions, stoichiometries, and functions remain to be determined. Thus, it is possible that other viral as well as cellular components could affect its conformation and ability to bind the required metal cofactor.
Recent experiments with single chain variable fragments derived from mAbs 4 and 33 show that their intracellular expression can protect human T cells from HIV-1 infection (52). This is consistent with the hypothesis that HIV-1 IN is in an "open" conformation, either in the Gag-Pol precursor or in the mature protein introduced with infecting virions, and that later conformational changes similar to those detected in our in vitro assays are required for its activation. We note that in vivo modulation by a metal-induced conformational change has been reported to affect the function of the tumor-suppressor protein, p53 (34,53). Thus, the present studies provide a framework for the elucidation of further details of IN activation both in vitro and in vivo.