Functional Interactions of the HHCC Domain of Moloney Murine Leukemia Virus Integrase Revealed by Nonoverlapping Complementation and Zinc-Dependent Dimerization

Abstract

The retroviral integrase (IN) is required for the integration of viral DNA into the host genome. The N terminus of IN contains an HHCC zinc finger-like motif, which is conserved among all retroviruses. To study the function of the HHCC domain of Moloney murine leukemia virus IN, the first N-terminal 105 residues were expressed independently. This HHCC domain protein is found to complement a completely nonoverlapping construct lacking the HHCC domain for strand transfer, 3′ processing and coordinated disintegration reactions, revealing trans interactions among IN domains. The HHCC domain protein binds zinc at a 1:1 ratio and changes its conformation upon binding to zinc. The presence of zinc within the HHCC domain stimulates selective integration processes. Zinc promotes the dimerization of the HHCC domain and protects it from N-ethylmaleimide modification. These studies dissect and define the requirement for the HHCC domain, the exact function of which remains unknown.

An essential step of the retroviral life cycle is the integration of the reverse-transcribed viral DNA into the host genome. This two-step process is carried out by the viral enzyme, integrase (IN), encoded by the pol gene (for a review, see reference 5). First, two deoxynucleotides are removed from both 3′ termini of the viral long terminal repeats (LTRs), exposing the conserved 5′-CA-3′ at the recessed 3′ ends (3′ processing) (6, 29, 44, 54). The 3′ ends of the viral DNA are then joined to 5′ staggered sites in the target DNA in a concerted transesterification reaction (strand transfer) (14, 27, 43). Both the 3′ processing and strand transfer reactions are isoenergetic and occur without ATP (27). Integration is completed by repair of the 5′ overhang of the viral LTR and the single-strand gaps flanking the integrated viral DNA, presumably by host enzymes. Repair of these single-strand gaps creates the hallmark duplication of target DNA sequence flanking the retroviral integration site. For Moloney murine leukemia virus (M-MuLV), the direct duplication is 4 bp long.

In vitro assays have been developed to study the mechanism of retroviral integration by using purified enzymes and short oligodeoxynucleotide duplexes mimicking the viral LTR ends (14, 44). Purified IN is also able to carry out disintegration reactions with either Y oligomer (13) or crossbone substrates (12, 15), which contain single and double LTRs, respectively.

Sequence comparison and mutational analysis have identified three functional domains in the IN protein (9, 24, 36, 40, 45, 46, 63, 67). The first two domains are highly conserved among all retroviruses and retrotransposons. The N-terminal region is characterized by an HHCC zinc finger motif (8, 9, 36, 68). Mutational studies of the conserved cysteine or histidine residues produced varied results regarding the importance of the HHCC domain (9, 10, 17, 26, 40, 42, 45, 47, 53, 55, 61–63). The central region of IN is characterized by a D-D(35)-E motif. Mutation of the conserved aspartic or glutamic acid residues results in loss of all catalytic activities, indicating a role for these three residues in active site function (17, 26, 46, 47, 61, 62). The conserved residues in both the N terminus and the central core are also required for viral replication in vivo (25, 53, 56, 59, 65). The C-terminal region is less conserved and has been demonstrated to possess nonspecific DNA-binding activity (23, 45, 51, 52, 63, 66, 67).

There is strong evidence that the IN protein functions as a multimer (37). Within individual subdomains, homodimers have been identified. Dimers of the avian sarcoma virus and human immunodeficiency virus (HIV) core domains (7, 18), the HIV-1 C terminus (20, 50), and the HHCC region (68) have been identified. The association of these IN subdomains to form higher-ordered protein-DNA complexes is unknown. Complementation experiments employing different mutants of HIV-1 and M-MuLV IN support the multimerization of IN and provide an alternative approach in identifying protein-protein interactions (22, 24, 38, 60).

There is general agreement that the HHCC domain is essential for integration in vivo (25, 47, 53, 65); however, the function as assayed in vitro is less well defined. Although zinc finger domains frequently bind DNA, no evidence for DNA binding has been reported to date for the IN HHCC domain (45, 55); however, indirect effects have been noted (32, 53). Functions that might affect recognition (33, 45, 62) or positioning (39, 61) of the viral LTRs have been observed to be associated with the HHCC domain. Formation of a stable IN-LTR complex requires the HHCC region (22, 64). MuLV LTR substrates which lack the 5′ single-stranded (SS) tail require the presence of an HHCC domain for disintegration and coordinated disintegration reactions (15, 16). The N terminus of HIV-1 IN was found to cross-link to the target DNA (34, 35). The HHCC domain is also reported to be involved in protein-protein interactions and the process of multimerization (15, 22, 49, 68). Zinc is reported to promote multimerization of HIV-1 IN and to stimulate Mg²⁺-dependent 3′ processing and strand transfer activity (48, 49, 68).

Here, we report that the N terminus of M-MuLV IN containing the HHCC domain complements a completely nonoverlapping domain of M-MuLV IN containing the core and C terminus. This isolated HHCC domain binds zinc with a 1:1 ratio. Zinc binding induces a conformational change of the HHCC domain and stimulates the catalytic activity of M-MuLV IN in complementation assays.

MATERIALS AND METHODS

Materials.

Crude [γ-³²P]ATP (7,000 Ci/mmol) was purchased from ICN. [α-³²P]dATP and [α-³²P]TTP were purchased from Amersham. T4 polynucleotide kinase was obtained from New England Biolabs. Exonuclease-free Klenow fragment of DNA polymerase I was obtained from United States Biochemical. Ni²⁺-nitrilotriacetic acid agarose was purchased from Qiagen. All restriction enzymes were purchased from New England Biolabs. RecA protein was purchased from Promega. Glutathione S-transferase–HMGI-C was kindly provided by Kiran Chada (69).

Oligonucleotides.

DNA oligonucleotides were prepared by the University of Medicine and Dentistry of New Jersey Biochemistry Department Synthesis Facility and purified by electrophoresis on 20% denaturing polyacrylamide gels. Oligonucleotides used in this study are referred to by their synthesis numbers and were labeled and prepared as described elsewhere (15, 16, 40). Oligonucleotides 2783 (5′-GTCAGCGGGGGTCTTTCATT), 2784 (5′-GTCAGCGGGGGTCTTTCA), and 2785 (5′-AATGAAAGACCCCCGCTGAC) were used for 3′ processing and strand transfer reactions. Oligonucleotides 4166 (5′-AATGAAAGTTCTTTCACGCTAGTCCTTGGAC), 4167 (5′AATGAAAGTTCTTTCAAGCGAGTCCTTGGAC), 5354 (5′-TGAAAGTTCTTTCACGCTAGTCCTTGGAC), and 5355 (5′-TGAAAGTTCTTTCAAGCGAGTCCTTGGAC) were substrates for the coordinated disintegration reactions. Oligonucleotide 7440 (5′-ACCTGCGTAAGCAGGTAGACCGCAAGGTCTACTTTCGAATGCGAAAGT) was used for the disintegration reaction.

Construction of M-MuLV HHCC mutant.

Construction and expression of wild-type IN and mutants NΔ105 and CΔ232 were previously described (38, 40). M-MuLV HHCC mutant CΔ303 was constructed by PCR amplification with Vent DNA polymerase (New England Biolabs), with the plasmid CΔ232 as template and the T7 primer (5′-TAATACGACTCACTATAGGG) (Promega) as upstream primer. The downstream primer, 6351 (5′-CGGGATCCTAAGACTTGCTGGCGTTGAC), encodes a stop codon and a BamHI site adjacent to sequence complementary to the IN mRNA coding region (indicated in boldface; the first A is complementary to position 4925 in the MuLV RNA sequence [57]). The PCR product was digested with NdeI and BamHI and exchanged for the 1.2-kb NdeI and BamHI fragment from the construct CΔ232. The nucleotide sequence of the construct was verified by dideoxy sequencing with an AmpliCycle sequencing kit from Perkin-Elmer.

Purification of M-MuLV IN.

Recombinant M-MuLV IN (WT, NΔ105, and CΔ232) containing a hexahistidine tag was expressed in Escherichia coli BL21(DE3) (Novagen) and purified by Ni²⁺-nitriloacetate chromatography (Qiagen) as previously described (38, 40). CΔ303 was expressed and purified similarly. CΔ303/zinc was renatured in the presence of 10 μM ZnCl₂; zinc was omitted in the last step of renaturation. CΔ303 was further purified on a carboxymethyl Sepharose column and eluted with a 0 to 1 M NaCl gradient in buffer containing 10 mM Tris-HCl (pH 7.5), 2 mM dithiothreitol (DTT), and 5% glycerol. The peak fractions were pooled and concentrated with a Centricon-10 concentrator (Amicon). The concentrated sample was then applied onto a Superose-12 fast protein liquid chromatography column, and fractions were collected by monitoring absorption at 280 nm. To remove the hexahistidine tag from CΔ303, the peak fractions from carboxymethyl Sepharose were dialyzed in 50 mM Tris-HCl (pH 7.5)–50 mM NaCl–2.5 mM CaCl₂–2 mM DTT–0.1% Nonidet P-40 (NP-40)–4% glycerol and then cleaved by thrombin (3 U/mg of IN) (Sigma) at room temperature for 1 h. The cleaved HHCC mutant was separated from the tagged protein by further purification on a P11 column. The cleaved protein was eluted from the P11 column with a 0 to 1 M NaCl gradient in buffer containing 10 mM Tris-HCl (pH 7.5), 2 mM DTT, and 5% glycerol.

In vitro assays.

Strand transfer, 3′ processing, disintegration, and coordinated disintegration reactions were performed as previously described (15, 40). Typically, reactions contained 1 pmol of labeled substrate and 7.5 to 10 pmol of IN protein. Complementation assays were performed by titrating the HHCC finger domain protein against a fixed level of NΔ105. The buffer for strand transfer and 3′ processing reactions contained 20 mM morpholineethanesulfonic acid (MES; pH 6.2), 10 mM DTT, 10 mM MnCl₂, 10 mM KCl, and 10% glycerol. The buffer for disintegration reaction had 20 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES; pH 6.4), 5 mM CHAPS (3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate), 10 mM DTT, 25 mM MnCl₂, and 0.05% NP-40. The buffer for coordinated disintegration reactions included 20 mM PIPES (pH 6.4), 10 mM CHAPS, 10 mM DTT, 10 mM MnCl₂, 0.01% NP-40, 5% ethylene glycol, and 10 mM NaCl. Ratios of HHCC domain to NΔ105 varied from 0.1:1 to 10:1 as indicated in each figure. For the N-ethylmaleimide (NEM) protection assay, the proteins were treated with 10 mM NEM as described elsewhere (39). To label CΔ303 proteins with fluorescent maleimide, 2 μl of 50 mM N(1-pyrene) maleimide in dimethyl sulfoxide was added to 10 μl of CΔ303 protein sample at 40 pmol/μl and incubated on ice for 90 min. The reaction was stopped by adding 6 μl of sodium dodecyl sulfate-polyacrylamide gel electrophoresis sample buffer, and the mixture was heated for 2 min at 100°C and subjected to electrophoresis on a 15% acrylamide gel. The fluorescent picture was taken with a Vilber Lourmat (France) transilluminator and a Canon A-1 camera with a yellow filter (Agfa G2). Protein concentration was measured by the method of Bradford (3) (Bio-Rad).

Circular dichroism (CD) spectroscopy.

CD measurements were performed on an Aviv model 62D spectropolarimeter equipped with a thermally regulated cell holder. The CD far-UV spectra were analyzed at both 5 and 37°C from 250 to 200 nm, every 0.25 nm, with a 2-s collection time, in 1-mm rectangular quartz cuvettes. The content of secondary structures was calculated by using three different curve fitting programs: Lincomb (constrained least squares fit) (4), MLR (nonconstrained least squares fit) (4), and Selcon (41, 58) (for details of the programs, see the review in reference 31).

Zinc analysis by atomic absorption spectroscopy.

Zinc concentration in the protein was measured on a Perkin-Elmer model 3100 atomic absorption spectrometer, which was calibrated with a zinc standard curve made from a standard zinc solution (EM Science). The zinc concentration was measured by using the flame mode of the spectrometer. The protein concentrations of the samples were determined by a difference spectrum method (19, 28, 30), and the stoichiometry of zinc to HHCC domain was calculated.

RESULTS

Complementation of NΔ105 by a nonoverlapping HHCC domain.

The schematic representation of M-MuLV IN and IN mutants is shown in Fig. 1. Previous work in our laboratory has shown that the N-terminal deletion M-MuLV IN mutant NΔ105 (up to 40 pmol) produced a very low level of strand transfer at one preferential site, whereas no 3′ processing could be detected (Fig. 2A, lane 2) (references 16 and 38 and data not shown). NΔ105 could be complemented by another mutant, CΔ232, which has a region of 71 amino acids overlapping with NΔ105 (38). To determine if this overlapping region is essential for CΔ232’s ability to rescue NΔ105 for strand transfer reactions, a smaller nonoverlapping HHCC construct containing amino acid residues 1 to 105 (CΔ303) was generated. The ability of CΔ303 to complement NΔ105 for strand transfer and 3′ processing reactions was analyzed. For strand transfer reaction, a precleaved substrate, which lacks the two terminal bases lost during 3′ processing, thus exposing the 5′-CA-3′ end, was used (Fig. 1B). Addition of CΔ303 restores NΔ105’s ability to catalyze strand transfer reaction (Fig. 2A, lanes 10 to 14). The level of complementation achieved by CΔ303 is similar to that by CΔ232 (Fig. 2A, compare lanes 5 to 9 and lanes 10 to 14), indicating that the overlapping region from residues 106 to 176 is not required for the HHCC domain to complement NΔ105, at least for the strand transfer reaction. As a control, CΔ303 at 100 pmol had no strand transfer activity (data not shown). No complementing activity was detected when either RecA or glutathione S-transferase–HMGI-C protein was used to replace CΔ303 (data not shown).

FIG. 1.

Illustration of WT M-MuLV IN, mutant INs, and assays used in this study. (A) Schematic representation of WT M-MuLV IN and IN mutants used in this study. The name of each mutant is indicated to the left of each protein. (B) DNA substrates and assays for 3′ processing and strand transfer reactions. (C) DNA substrates and assay for coordinated disintegration reactions with crossbone substrates with 5′-ss overhang in the LTR (tailed crossbone substrates). (D) DNA substrates and assay for coordinated disintegration reaction with substrates without 5′-ss overhang in the LTR (untailed crossbone substrates). The asterisks indicate the ends labeled.

FIG. 2.

Complementation of NΔ105 by nonoverlapping HHCC domain protein. (A) Complementation of NΔ105 by CΔ232/CΔ303 for strand transfer reaction. Lane 1, control buffer incubation; lane 2, 10 pmol of NΔ105; lanes 3 and 4, 10 and 40 pmol of WT IN, respectively; lanes 5 to 9, 10 pmol of NΔ105 plus CΔ232; the ratio of CΔ232 to NΔ105 is indicated at the top of the panel; lanes 10 to 14, 10 pmol of NΔ105 plus CΔ303; the ratio of CΔ303 to NΔ105 is indicated at the top of the panel. (B) Complementation of NΔ105 by CΔ232/CΔ303 for 3′ processing reaction. The lanes are the same as in panel A. Both CΔ232 and CΔ303 were renatured in the presence of EDTA; the reaction buffer contained 10 mM MnCl₂.

In the 3′ processing reaction with a blunt-ended oligonucleotide substrate (Fig. 1B), CΔ303 complements NΔ105 to produce a “−2” product. Interestingly, the efficiency of CΔ303 is less than that of CΔ232 (compare lanes 5 to 9 and lanes 10 to 14 in Fig. 2B), though the same fractions were used for strand transfer and 3′ processing reactions. This implies that the 71-amino-acid overlapping region is stabilizing the interaction between CΔ232 and NΔ105 for 3′ processing.

Zinc stimulates HHCC domain’s ability to complement NΔ105.

Studies with HIV IN have shown that HIV IN binds zinc (8, 9, 32, 68), and zinc stimulates IN in strand transfer and 3′ processing reactions (48, 68). To see if zinc has any effect on M-MuLV IN, the HHCC domain protein CΔ303 was purified and refolded in the presence (CΔ303/zinc) or absence (CΔ303/EDTA) of zinc (see Materials and Methods for details) and assayed for complementation of NΔ105 in strand transfer, 3′ processing, and coordinated disintegration reactions.

In the strand transfer reaction, both the CΔ303 proteins, with or without zinc, are capable of complementing NΔ105 (compare lanes 5 to 9 with lanes 10 to 14 in Fig. 3A). There is a subtle but distinct difference in the pattern of strand transfer products between CΔ303/zinc and CΔ303/EDTA. CΔ303/EDTA complements NΔ105 to give rise to larger strand transfer products, which are barely detectable in the complementation by CΔ303/zinc (see asterisk in Fig. 3A). The pattern for CΔ303/zinc (Fig. 3A, lanes 10 to 14) resembles more that of WT IN (Fig. 3A, lane 4), which was purified without zinc.

FIG. 3.

Zinc stimulates CΔ303’s ability to complement NΔ105. (A) Complementation of NΔ105 by CΔ303/EDTA or CΔ303/zinc for strand transfer reaction. Lane 1, control buffer incubation; lane 2, 10 pmol of NΔ105; lanes 3 and 4, 10 and 40 pmol of WT IN, respectively; lanes 5 to 9, 10 pmol of NΔ105 plus CΔ303/EDTA; lanes 10 to 14, 10 pmol of NΔ105 plus CΔ303/zinc. Ratios of CΔ303/zinc or CΔ303/EDTA to NΔ105 are indicated at the top of each panel. Lanes in panels B to D are the same as in panel A. ∗, difference in strand transfer products. (B) Complementation of NΔ105 by CΔ303/EDTA or CΔ303/zinc for 3′ processing reaction. (C) Complementation of NΔ105 by CΔ303/EDTA or CΔ303/zinc for coordinated disintegration reaction with tailed crossbone substrates. (D) Complementation of NΔ105 by CΔ303/EDTA or CΔ303/zinc for coordinated disintegration reaction with untailed crossbone substrate. CΔ303 was renatured in zinc or EDTA as indicated at the top of each panel; all reaction buffers contained MnCl₂ as indicated in Materials and Methods. SD, single disintegration product; DD, double disintegration product.

For 3′ processing and coordinated disintegration reactions, the effect of zinc within the HHCC domain is more obvious. For 3′ processing, complementation of NΔ105 occurs with lower concentrations of CΔ303/zinc HHCC protein than of CΔ303/EDTA (compare lanes 5 to 9 with lanes 10 to 14 in Fig. 3B). CΔ303/zinc saturated NΔ105 at a ratio of 0.5 to 1 HHCC/NΔ105 subunit, whereas CΔ303/EDTA required minimally two HHCC/NΔ105 subunits for maximal activity.

In coordinated disintegration reactions with two half-crossbone substrates (Fig. 1C and D), four products have been previously identified, including single-end disintegration product (SD), a circular product resulting from disintegration of both LTR ends (DD), a small circular foldback product, and a hydrolysis product releasing the viral LTR (15). For coordinated disintegration reactions with the crossbone substrates containing 5′-ss overhang, the presence of zinc stimulates the production of the foldback and hydrolysis products (Fig. 3C, compare lanes 5 to 9 and lanes 10 to 14). In the coordinated disintegration reaction with crossbone substrates lacking the 5′-ss overhang, it was previously shown that the CΔ232 construct was capable of complementing NΔ105, yielding the double disintegration product (15). Interestingly, CΔ303/EDTA yielded little if any double disintegration product (lanes 5 to 9, Fig. 3D). However, CΔ303 renatured with zinc yielded increased levels of both single and double disintegration products (lanes 10 to 14 in Fig. 3D). This reaction required excess CΔ303 to NΔ105, with maximal stimulation detected at a 10:1 ratio.

The histidine tag was removed from CΔ303/zinc and analyzed for complementation with strand transfer, 3′ processing, and untailed coordinated disintegration reactions (data not shown). CΔ303/zinc without the tag maintained approximately 50 to 60% of its activity in all three assays compared to the mock-treated protein. Analysis of the protein by Coomassie blue staining indicated that more than 90% of the hexahistidine tag was removed by thrombin cleavage (data not shown), indicating that the hexahistidine tag is not contributing to the complementing activity.

Zinc binding of the HHCC domain protects the HHCC domain from NEM alkylation.

Previous work from our laboratory and others has shown that both M-MuLV IN (16, 40) and HIV IN (22) are sensitive to NEM modification. For M-MuLV IN, there are three cysteine residues in total: one in the core region (C209) and two in the HHCC region (C94 and C97). NEM modification of either NΔ105 or CΔ232 reduced the ability of the two IN mutants to complement each other (16). This was most clearly observed with a dumbbell single-end disintegration substrate lacking the 5′-ss tail in the viral LTR (untailed disintegration). With this substrate, the reaction is dependent on both the catalytic core and the HHCC domain (16) (Fig. 4A, lanes 3 to 5). To determine if zinc has any effect on the NEM sensitivity of the HHCC domain, both CΔ303/zinc and CΔ303/EDTA were treated with NEM and then used to complement NΔ105 in the untailed disintegration reaction. As shown in Fig. 4A, NΔ105 is active at a very low level for the untailed disintegration reaction (lane 3). Addition of either CΔ303/zinc or CΔ303/EDTA restores NΔ105’s ability to catalyze the disintegration reaction (lanes 4 and 5). NEM treatment of CΔ303/EDTA abolishes its ability to complement NΔ105 (lane 9), while NEM-treated CΔ303/zinc is still active in complementing NΔ105 (lane 8). CΔ303/zinc with the hexahistidine tag removed remained resistant to NEM modification (data not shown). The accessibility of the cysteines in both CΔ303/zinc and CΔ303/EDTA was further probed with N(1-pyrene) maleimide, which allows direct visualization of the conjugation of the maleimide to the cysteine by fluorescence analysis. When equivalent amounts of the two proteins, as detected by Coomassie blue staining (Fig. 4B, lanes 3 and 4), were treated with N(1-pyrene) maleimide, only CΔ303/EDTA could be fluorescently labeled (Fig. 4B, lane 2). CΔ303/zinc was not modified (Fig. 4B, lane 1). These results indicate that coordination of zinc by the HHCC domain makes the two cysteines inaccessible to NEM, thus protecting the HHCC domain from being alkylated.

FIG. 4.

Zinc binding protects the HHCC domain from NEM modification. (A) CΔ303/zinc remains active for complementation after NEM treatment. A disintegration assay with oligonucleotide 7440 as substrate was used to test the susceptibility of IN mutants to NEM modification. Lane 1, control buffer incubation; lane 2, WT IN; lane 3, NΔ105; lane 4, complementation of NΔ105 and CΔ303/zinc; lane 5, complementation of NΔ105 and CΔ303/EDTA; lane 6, NEM-modified WT IN; lane 7, NEM-modified NΔ105; lane 8, complementation of NΔ105 and NEM-modified CΔ303/zinc; lane 9, complementation of NΔ105 and NEM-modified CΔ303/EDTA. All NEM-treated proteins are underlined. (B) Only CΔ303/EDTA can be labeled by fluorescent maleimide. CΔ303/zinc and CΔ303/EDTA were modified with N(1-pyrene) maleimide as described in Materials and Methods, and the protein bands were analyzed for fluorescence (lanes 1 and 2) and by Coomassie blue staining (lanes 3 to 5). Lanes 1 and 3, CΔ303/zinc; lanes 2 and 4, CΔ303/EDTA; lane 5, protein molecular mass markers.

Zinc promotes dimerization of the HHCC domain.

To characterize the multimerization state of the HHCC domain in solution, CΔ303 was further purified after renaturation and applied to a Superose-12 sizing column (see Materials and Methods for details). CΔ303 refolded in the presence of zinc eluted as a single peak between 25- and 43-kDa protein markers (Fig. 5). The calculated molecular mass from the standard curve derived from the protein standard markers was 32 kDa, corresponding to a dimer of CΔ303; the molecular mass of a CΔ303 monomer is 14.5 kDa. In the absence of zinc, CΔ303/EDTA yielded no detectable dimers. Rather, these proteins chromatographed heterogeneously and eluted as an aggregate in the void volume and as monomers. In one preparation, a low level of hexamers was detected (data not shown). This is consistent with the solubility of the two proteins. CΔ303/zinc could be concentrated to greater than 20 mg/ml, whereas CΔ303/EDTA precipitated from solution at concentrations greater than 2 mg/ml. The presence of histidine tag does not interfere with the multimerization of the HHCC domain since CΔ303/zinc lacking the histidine tag was eluted as a dimer as well (data not shown). Neither manganese nor magnesium could replace zinc to promote the dimerization of the HHCC domain (data not shown).

FIG. 5.

Fast protein liquid chromatography elution profile of the HHCC domain protein CΔ303/zinc. CΔ303/zinc was chromatographed on a Superose-12 column. The protein standard markers are as follows: 1, bovine serum albumin, 67 kDa; 2, ovalbumin, 43 kDa; 3, chymotrypsinogen A, 25 kDa; 4, RNase A, 13.7 kDa. The elution time of the protein standards is indicated on the panel. OD 280, optical density at 280 nm; AU, absorbance units.

Zinc binds to the HHCC domain at a 1:1 ratio and induces a conformational change of the HHCC domain.

To determine the stoichiometry of zinc bound to the HHCC domain of M-MuLV IN and the effects of zinc on protein conformation, both zinc concentration and CD studies were performed on CΔ303, renatured with zinc or EDTA. For zinc content measurement, CΔ303/zinc and CΔ303/EDTA were analyzed by atomic absorption spectroscopy. For CΔ303/zinc, the calculated average molar ratio of zinc to the HHCC domain is 1.1:1, and the ratio for CΔ303/EDTA is 0.1:1. The presence of histidine tag has no effect on the concentration of zinc, as CΔ303/zinc without the His tag gave similar results.

From the CD profile, both CΔ303 proteins (EDTA and zinc) are well structured overall, at either 5 or 37°C (Fig. 6). CΔ303/EDTA contains more than 30% helical structure at both 5 and 37°C. This is consistent with results from in vitro activity assays, indicating that the HHCC protein renatured in EDTA maintains the ability to complement the strand transfer reaction (Fig. 3A). In the presence of zinc, the helical content of CΔ303 increases, with a corresponding decrease in β sheet, turn, and random-coil contents (Table 1). Addition of excessive EDTA to CΔ303/zinc at 37°C (Fig. 6B) would strip away zinc from the HHCC domain and result in a conformation similar to that of CΔ303/EDTA, which contains approximately 34% α-helix, 18% β sheet, 25% turn, and 24% random coil (Table 1). At 5°C, the CΔ303/zinc is generally stable despite the addition of excess EDTA by maintaining a helical content of 49% (Table 1).

FIG. 6.

CD spectra of the HHCC domain proteins. CD spectra of CΔ303/zinc, CΔ303/EDTA, and CΔ303/zinc plus 5 mM EDTA at 5°C (A) and 37°C (B) are shown.

TABLE 1.

Calculated content of secondary structures of the HHCC domain^a

Temp	Value (%)
	Mutant	α helix	β sheet	Turn	Random
5°C	CΔ303/zinc	53.2	7.8	19.6	18.8
	CΔ303/EDTA	33.4	16.9	26.3	23.0
	CΔ303/zinc + EDTA	49.3	11.4	27.1	15.3
37°C	CΔ303/zinc	52.7	15.0	19.5	16.1
	CΔ303/EDTA	37.5	18.7	26.3	22.2
	CΔ303/zinc + EDTA	33.8	17.6	24.9	23.7

^aThe value is derived from the CD spectra with three different curve fitting programs (see Materials and Methods for details). Only the calculations from the Selcon program are shown here. The other two programs gave similar results. 

DISCUSSION

Previous work in our laboratory has shown that M-MuLV IN mutant NΔ105 lacking the HHCC domain can be rescued by another mutant, CΔ232, lacking part of the core region and the C terminus (38). Our current study shows that the overlapping region between NΔ105 and CΔ232 was not required for the complementation. A new M-MuLV HHCC construct, CΔ303, which has only the complete HHCC domain and no common region with the NΔ105, could complement NΔ105 for strand transfer reaction. This nonoverlapping complementation indicates that (i) the N-terminal region of M-MuLV IN containing the HHCC domain is a structurally independent domain and (ii) this HHCC domain is able to function in trans to complement the core and C terminus. The N-terminal amino acids 1 to 105 behaved like a dimer, indicating that the HHCC domain interacts with itself as well as other parts of the IN (16). The ability of CΔ232 to assist the 3′ processing reaction at a lower protein level than CΔ303 implies that the region of overlap could provide an additional protein-protein interaction.

Zinc finger motifs have been found in many of the transcription factors and DNA binding proteins (1, 2). The spacing and positions of the histidine and cysteines within the highly conserved HHCC domain of retroviral IN are unique and can form a novel zinc finger-like motif. To date, the HHCC domain is directly related only to protein multimerization; however, there is no direct evidence that the HHCC domain of retroviral IN binds DNA (45, 55). Though HIV-1 IN has been shown to structurally bind zinc at the ratio of 1:1 through the HHCC domain (8, 9, 32), zinc was found recently to stimulate magnesium-dependent activities including strand transfer and 3′ processing (48, 68). The results from the studies reported here demonstrate that incorporation of zinc into the HHCC domain of M-MuLV IN could stimulate the catalytic activities of M-MuLV IN, indicating that zinc binding is functionally important and biologically relevant. Zinc did not stimulate the overall strand transfer activity of M-MuLV IN, as for HIV IN. Instead, the range of target sites selected was limited in the presence of a zinc-coordinated HHCC region. This increased stringency may reflect the tightness of the assembled complex; zinc binding may stabilize the HHCC domain and the core-C terminus interactions (16).

NEM modification has been used to probe the role of the HHCC domain in in vitro catalytic activities (22, 39). For M-MuLV IN, both the N-terminal cysteines within the HHCC domain and one central cysteine were sensitive to NEM alkylation. Alkylation of the HHCC domain protein CΔ232 abolished most of its ability to complement NΔ105 for strand transfer and coordinated disintegration reactions (16). The results here demonstrate that zinc binding by the HHCC domain protected it from NEM modification, indicating that zinc coordination by the His and Cys residues is stable and that the two cysteine residues are fully occupied by zinc. The results can explain why there was residual activity in complementation between NΔ105 and NEM-modified CΔ232 for strand transfer reactions (16), where a trace amount of zinc coordinated by a small amount of CΔ232 could be resistant to NEM. HHCC constructs renatured in the presence of EDTA maintained a 0.1:1 molar ratio of zinc to protein.

Various evidence has indicated that viral INs act as oligomers. Structural studies of the catalytic domains of HIV IN and avian sarcoma virus IN and of the C terminus of HIV IN showed that they exist as dimers in solution (7, 18, 20, 50). Though an earlier report on the HHCC domain of HIV IN did not find dimerization of the HHCC domain in the presence of zinc (8), recent structural studies of the HIV-1 IN (11) showed that the HIV-1 IN amino acids 1 to 55 formed a dimer in solution. In contrast, the nuclear magnetic resonance structure of HIV-2 IN amino acids 1 to 55 was not a dimer in solution (21). Although the individual domains have been identified as dimers, one key question which remains unanswered is how the components assemble into an active multimer. Complementation studies of M-MuLV and HIV-1 IN also suggest that IN functions as a multimer (22, 24, 38, 60). Recent studies start to reveal the role of the HHCC domain in the multimerization process of IN. Our results here demonstrate that the HHCC domain of M-MuLV IN formed a dimer in solution and that zinc was required for the dimerization. These results are consistent with that of HIV-1 IN. However, since the HHCC domain of M-MuLV IN is much larger than that of HIV IN (105 versus 55 amino acids), we cannot exclude the possibility that the additional sequences in M-MuLV IN are involved in the dimerization of the HHCC domain of M-MuLV IN. Studies are currently under way to determine regions within the N-terminal 105 amino acids of M-MuLV IN required for dimerization. For HIV IN, it has been shown that zinc can promote tetramerization of HIV IN, while the predominant form of IN is a dimer without zinc (49, 68). In light of the zinc-dependent dimerization of the HHCC domain, our current working model is that the HHCC domains bridge two sets of existing dimers formed through the central core and the C terminus, resulting in formation of an IN tetramer. The mechanism of how the dimerization of the HHCC domain is coupled to tetramerization and which oligomeric form is active remain to be determined.

Previous studies have shown that the HHCC domain of HIV-1 IN can bind zinc at a 1:1 molar ratio and changes its conformation upon zinc binding (8, 32, 68). The stoichiometry of zinc against the M-MuLV IN HHCC domain was 1:1, indicating that there is no zinc in the interface of the zinc finger dimer. By the inductively coupled plasma mass spectroscopy method, no other metal was found substituting for zinc at stoichiometric levels (unpublished results). The CD study of the M-MuLV IN amino acids 1 to 105 indicated that the addition of zinc induced conformational changes with increased alpha-helical content, as found for HIV IN (8, 68). This indicates that at least some portion of the HHCC domain has to adopt helix structure in order for the histidine and cysteine residues to coordinate the zinc. However, there are some differences between M-MuLV IN amino acids 1 to 105 (CΔ303) and HIV IN amino acids 1 to 55. From our CD study, the M-MuLV IN amino acids 1 to 105 were well structured even in the absence of zinc, as indicated by the helical content of CΔ303/EDTA. The HIV IN amino acids 1 to 55 were highly disordered without zinc (8, 68). This could be due to the fact that M-MuLV IN has a much larger N terminus than does HIV IN. M-MuLV IN has 50 amino acid residues N-terminal to the HHCC zinc finger motif. These extra 50 residues may help the HHCC motif fold and stabilize the domain, even in the absence of zinc.

The finding that CΔ303/EDTA in the absence of stoichiometric zinc could complement NΔ105 for both strand transfer and 3′ processing reactions raises the possibility that the N terminus of M-MuLV IN could be further divided into two separate domains: the N-terminal domain consisting of the first 50 amino acids and the zinc finger domain containing the conserved HHCC residues. Preliminary evidence defining these separate functions is presented in the comparison of the unimolecular (dumbbell in Fig. 4) and biomolecular disintegration (coordinated disintegration) (Fig. 3D) reactions with untailed substrates. In the absence of the 5′-ss tail, both reactions required an HHCC N-terminal domain. For the unimolecular reaction, either the zinc- or EDTA-renatured protein sufficed, and yet treatment with NEM inactivated the EDTA-renatured sample. The bimolecular disintegration requires substrate assembly. In the absence of the 5′-ss tail, the HHCC construct in the absence of zinc could only minimally catalyze a single-end disintegration. The presence of zinc may therefore assist in the dimerization of the protein scaffold, thereby associating with the DNA substrates. The reactions greatly stimulated by zinc, 3′ processing, foldback, and hydrolysis, may require a tightly assembled complex, inducing strain in the substrates and assisting the nucleophilic attack to release the LTR. Future studies are aimed at determining the function of the N-terminal 50 residues.