ABSTRACT
The N terminus of the capsid protein (CA) undergoes a considerable conformational change when the human immunodeficiency virus (HIV) protease cleaves it free from the Pr55Gag polyprotein. This rearrangement is thought to facilitate the establishment of specific CA-CA interactions that are required for the formation of the mature viral core. Substitution of amino acids that are critical for this refolding of the N terminus is generally detrimental to virus replication and mature virion core morphology. Here, we identify a conserved threonine in simian immunodeficiency virus (SIV) CA, T(47)CA, that is requisite for viral replication. Replacement of T(47)CA in the infectious viral clone SIVmac239 with amino acids with different hydrogen-bonding capabilities and analysis of the effects of these substitutions at key steps in the viral life cycle demonstrate that hydrogen bonding at this position is important for virus infectivity and virion release. In the HIV-based homology model of the mature SIV CA N terminus presented in this study, T(47)CA forms several hydrogen bonds with a proximal aspartate, D(50)CA. This model, coupled with strong phenotypic similarities between viral substitution mutants of each of these two residues in all of the virological assays described herein, indicates that hydrogen bonding between T(47)CA and D(50)CA is likely required for viral replication. As hydrogen bonding between these two residues is present in HIV CA as well, this interaction presents a potential target for antiviral drug design.
The retroviral Gag protein is necessary and sufficient to direct assembly and budding of viral particles (9, 19, 28). In human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV), the gag gene is expressed as a polyprotein, Pr55Gag, which contains (from N to C terminus) the matrix (MA), capsid (CA), p2, nucleocapsid, p1, and p6 proteins (25, 26, 33, 36). Pr55Gag is targeted to and interacts with the plasma membrane via a myristic acid moiety at the N terminus of the MA domain, in conjunction with several downstream basic residues (5, 38, 49, 60), and it is there that the virion assembles and buds from the cell (for a review, see reference 27). At an undetermined step in the assembly and budding process, the viral protease cleaves Pr55Gag into its constituent proteins. This cleavage results in a significant morphological change in the virion termed maturation (reviewed in reference 54). In the immature virion, a shell of Pr55Gag lines the inner leaflet of the lipid bilayer comprising the viral envelope. Upon maturation, MA remains associated primarily with the viral envelope, while CA collapses into a conical core surrounding nucleocapsid complexed with the genomic viral RNA (18, 40, 41).
HIV CA is a key component in several steps of the virus life cycle, particularly in virion assembly and maturation (also referred to as core formation). The CA protein consists of two largely alpha-helical domains connected by a flexible linker: an N-terminal domain (amino acids 1 to 145) and a C-terminal domain (amino acids 151 to 231) (2, 15, 16, 20, 37, 58). The N terminus of the HIV CA protein undergoes a structural rearrangement subsequent to proteolytic cleavage. In this rearrangement, the N-terminal proline of CA [HIV P(1)CA] folds back into the core of the protein to form a buried salt bridge (a hydrogen bond between two oppositely charged groups [35]) with the side chain of an aspartate in helix 3 [HIV D(51)CA], forming a β-hairpin structure in the first 13 amino acids of CA (15, 20, 37). As formation of the salt bridge and the β-hairpin requires the free amino group of P(1)CA that is generated upon Pr55Gag cleavage, these features are absent in the immature form of CA (20, 50, 55).
Strong sequence conservation of the HIV CA salt bridge amino acids suggests that the salt bridge and β-hairpin may be structurally conserved among the mature CA proteins of most retroviruses (55). In fact, these two structural elements present in mature HIV CA are also present in the other complete, innate mature CA structures reported thus far, those of human T-cell leukemia virus type 1 (8, 31) and Rous sarcoma virus (6, 32) CA. The conservation of the salt bridge and β-hairpin suggests that these structural features are important for CA function. A single role has been proposed for the N-terminal β-hairpin in mature CA: interactions between the β-hairpins of CA monomers may facilitate dimerization and core assembly (15, 20, 23, 55). Amino acid substitutions in HIV CA that disrupt the salt bridge, thereby destabilizing the β-hairpin, cause a variety of viral defects, including abnormal mature core morphology and loss of infectivity (13, 51, 55). Because the conformational change in the CA N terminus is essential for proper CA function in the virus life cycle, the interactions that facilitate the formation of the β-hairpin and the salt bridge in mature CA are important potential targets for antiviral drug design (50; reviewed in reference 29).
In this study, we identified a conserved threonine [T(47)CA] required for replication of SIVmac239. The results of virological assays using a panel of polar and nonpolar substitutions of T(47)CA in SIV CA demonstrate that hydrogen bonding at this position is critical for the proper function of CA in viral infectivity and virion release. In addition, strong phenotypic similarities were observed between alanine substitution mutants of T(47)CA and a proximal aspartate, D(50)CA. Using the coordinates of the HIV CA N-terminal domain (15), we constructed a homology model of SIV CA. In this model, T(47)CA and D(50)CA are directly hydrogen bonded to each other. This observation, coupled with the phenotypic similarities between T(47)CA and D(50)CA mutants, strongly suggests that hydrogen bonding between T(47)CA and D(50)CA is essential for viral replication. As hydrogen bonding between these two residues is present in HIV CA as well, this interaction may provide a novel target for inhibitors of HIV replication.
MATERIALS AND METHODS
Mutagenesis and cloning.SIVmac239 gag mutants were generated by using the cloned 5′ (p239SpSp5′) and 3′ (p239SpE3′) halves of the SIVmac239 genome in the vectors pBS+ and pBS−, respectively (AIDS Research and Reference Reagent Program) (30, 42). In brief, p239SpSp5′ was subjected to in vitro mutagenesis with the QuikChange mutagenesis kit (Stratagene, Cedar Creek, Tex.). Mutagenesis primers (Keystone Labs, Camarillo, Calif.) were designed according to the recommendations of the QuikChange manufacturer. The codon alterations resulting in amino acid substitutions in MA or CA were GGC→GCA for G(2)AMA, CCA→GCA for P(1)ACA, GAC→GCC for D(50)ACA, ACC→GCC for T(47)ACA, ACC→TGC for T(47)CCA, ACC→TCC for T(47)SCA, and ACC→GTC for T(47)VCA. To generate full-length, mutant SIVmac239 in pBS−, mutant p239SpSp5′ and wild-type p239SpE3′ were both digested with SphI (Invitrogen Corporation, Carlsbad, Calif.) and ligated. All clones were verified by sequencing the complete coding region of the viral genome.
Cell culture and reagents.CEM×174 cells, a human T-cell/B-cell hybrid line (48) (a generous gift from James Hoxie [University of Pennsylvania]), were cultured as described previously (45). African green monkey kidney COS-1 cells (American Type Culture Collection, Manassas, Va.) were cultured like human embryonic kidney 293T cells (American Type Culture Collection) as described previously (4), except with gentamicin (0.5 mg/ml) instead of penicillin-streptomycin.
Viral growth and infectivity.CEM×174 cells (5 × 106) were transfected with 12 μg of viral DNA by electroporation at 200 V and 960 μF using a Gene Pulser (Bio-Rad, Hercules, Calif.). Reverse transcriptase (RT) activity in cell-free supernatants from transfected cultures was analyzed at various times posttransfection by using a standard RT assay (7). The infectivity (LuSIV assay; see below) and 50% tissue culture infective dose (TCID50) were determined for viral stocks generated by transfection of 293T (human embryonic kidney) cells and COS-1 (African green monkey kidney) cells in 10-cm dishes with 7.5 μg of wild-type or mutant infectious viral DNA by using Lipofectamine and Plus reagents (Invitrogen), according to the manufacturer's recommendations. The 293T and COS-1 cell lines were chosen for this study because they are derived from species susceptible to HIV and SIV, respectively. For the LuSIV assay, 293T- and COS-1-derived virus stocks were normalized by RT activity and used to infect LuSIV cells as previously described (45). Numbers represent the fold induction of luciferase activity over background at 2 days postinfection. The TCID50 assay was performed as described previously (61).
Metabolic labeling.293T cells or COS-1 cells grown in 10-cm dishes were transfected as described above. The following day, cells were labeled with 55.5 μCi of Tran35S-label (ICN Pharmaceuticals, Inc., Costa Mesa, Calif.) per ml and incubated for 6 h at 37°C. Cells were lysed in modified radioimmunoprecipitation assay (RIPA) buffer as described previously (1) and sonicated. Cell supernatants were filtered through a Millex-HV 0.45-μm syringe-driven filter (Millipore, Bedford, Mass.), pelleted, and lysed with modified RIPA buffer (similarly prepared lysates are referred to as virus lysates). Virus lysates were immunoprecipitated with either 6.5 μg of immunoglobulin G (IgG)-purified SIV CA antiserum or 6.3 μg of IgG-purified SIV antiserum. Cell lysates (500 μg) were precleared and immunoprecipitated overnight at 4°C with 5 μg of IgG-purified rabbit SIV CA antiserum. After immunoprecipitation, lysates were rotated at 4°C with protein A-Sepharose 4 Fast Flow (Amersham Pharmacia, Piscataway, N.J.) for 1 h. Pelleted Sepharose beads were washed with RIPA buffer, and proteins were resolved on 12.5% Tris-HCl Criterion precast gels (Bio-Rad) that were fixed and processed for autoradiography. Band densitometry was performed with Molecular Dynamics ImageQuant version 5.2 software (Amersham Pharmacia).
Transmission electron microscopy (EM).293T cells were transfected with infectious viral DNA as described above. At 1 day posttransfection, the cells were pelleted and fixed at 4°C in 2.5% glutaraldehyde in Millonig's sodium phosphate buffer (pH 7.4) for 2 to 3 h and washed gently three times with Millonig's sodium phosphate buffer. Samples were then submitted to Electron Microscopy Bioservices (Monrovia, Md.) for analysis.
Virion morphology comparison.The proportion of mature wild-type, D(50)ACA, and T(47)ACA SIVmac239 virions with acentric cores was determined as follows. Mature virions in 50 photomicrographs of each virus sample were assessed for conical, centric, or acentric cores. Precautions were taken to minimize bias in this study: both the EM specialist and the individuals who analyzed the photomicrographs were blinded to the identity of each sample, and an outside individual randomly ordered and numbered the photomicrographs before analysis. To test the validity of the data, the reported proportions of acentric cores for each genotype were compared with the proportions obtained by an independent observer using a two-sided, two-sample test of proportions (46). The proportion of acentric virions in T(47)ACA photomicrographs was compared to the proportions for wild-type and D(50)ACA individually by using a one-sided test for difference of proportions (46).
Computer modeling.The structure of cyclophilin A in complex with the N terminus of HIV-1 CA (15) was used to create a homology model of the residues surrounding T(47)CA in SIVmac239 CA. To generate this model, a multiple amino acid alignment of the following Pr55Gag sequences (obtained from the National Center for Biotechnology Information protein database) was generated with ClustalW (www2.ebi.ac.uk/clustalw ) (52) and the BLOSUM matrix: SIVmac239 (accession no. AAA47632 ) (30), HIV-2 (accession no. FOLJG2 ) (24), HIV-1 (accession no. AAK08483 ), equine infectious anemia virus (accession no. AAK21111 ), feline immunodeficiency virus (accession no. Q05313 ) (39), visna virus (accession no. NP_040839 ) (3), caprine arthritis encephalitis virus (accession no. P33458 ) (47), and bovine immunodeficiency virus (accession no. NP_040562 ) (17). Using this alignment as a guide, the Protein Modeler tool of the program QUANTA (Molecular Simulations) was used to replace amino acids in the local HIV-1 CA structure to match the sequence of SIVmac239 CA. The side chain of each mutated amino acid was rotated to find the best rotamer and further refined by energy minimization by using conjugate gradients. The depicted models of HIV-1 and SIV CA were rendered with SETOR (12). Hydrogen bonding in HIV-1 CA or SIVmac239 CA was identified by using QUANTA and the coordinates published in reference 15 or our homology model, respectively. In this analysis, we used an upper limit for hydrogen bond length (the distance between donor and acceptor atoms, e.g., between N of donor residue and O of acceptor residue) of 3.3 Å.
RESULTS
Identification of a conserved threonine required for viral replication.In a previous study of the role of capsid phosphorylation in viral budding, we characterized a panel of CA substitution mutants in the infectious viral clone SIVmac239 (J. W. Roos, S. M. Rue, J. E. Clements, and S. A. Barber, submitted for publication). Phosphoamino acid analysis indicated that serine was the predominant phosphoamino acid, and a panel of mutants was created, replacing serines in consensus casein kinase II (CK2) and protein kinase C sites with alanine (Roos et al., submitted). Because threonine is also a potential site of phosphorylation by phosphoamino acid analysis (although to a lesser degree than serine [Roos et al., submitted]), and to ensure that all known potential consensus sites were included in the mutagenesis study, we substituted alanine for each of the two threonines in consensus CK2 sites, in positions 47 [T(47)ACA] and 208 [T(208)ACA] of CA. Equivalent amounts of proviral DNA from each of these mutants and wild-type SIVmac239 were transfected into CEM×174 cells, and virus-containing cell supernatants were subjected to an RT assay at various times posttransfection (Fig. 1A). The T(47)ACA substitution severely compromised viral replication, while the T(208)ACA substitution only modestly delayed replication.
Growth kinetics and infectivity of mutants with substitutions of SIVmac239 CA threonines in consensus CK2 sites. (A) Cell-free supernatants of CEM×174 cells transfected with infectious viral DNA were analyzed at various times posttransfection with a standard RT assay. (B) Virus derived from transfection of 293T cells was normalized according to RT activity and analyzed with the LuSIV assay. Induction values are relative to background at day 2 postinfection.
We next examined the infectivity of virions produced from 293T cells transfected with wild-type and mutant viral DNAs by the LuSIV assay (Fig. 1B). The LuSIV assay, developed in our laboratory (45), is based on a CEM×174 cell line stably transfected with a plasmid encoding luciferase under the control of the SIV long terminal repeat. Upon virus infection of this cell line, the viral Tat protein transactivates the long terminal repeat, inducing luciferase expression. Subsequent quantitation of luciferase activity provides a very sensitive measure of virus infectivity. While both substitution mutants were less infectious than wild-type virus in this assay, the infectivity of T(47)ACA virus was approximately fourfold lower than that of T(208)ACA virus.
To investigate whether either of the threonine substitutions affected virion morphology, we transfected 293T cells with wild-type or mutant infectious viral DNA and subjected them to transmission EM analysis. The T(208)ACA particles appeared to be morphologically wild type by EM (data not shown). In contrast, a predominance of the mature particles in photomicrographs of the T(47)ACA sample had acentric cores (core is collapsed and juxtaposed to the viral envelope [see below]).
Alignment of lentiviral Pr55Gag protein sequences indicated that T(47)CA is well conserved among lentiviruses (Fig. 2), underscoring the importance of this residue and suggesting that it may be similarly required for the replication of other members of this family of retroviruses. Interestingly, in all three virological analyses of the threonine substitution mutants, the T(47)ACA mutant bore a strong resemblance to previously characterized mutants with substitutions of the two residues that participate in the salt bridge in HIV CA, P(1)LCA and D(51)ACA. Like T(47)ACA SIV, P(1)LCA and D(51)ACA HIV do not replicate well, are noninfectious, and have largely acentric mature cores (13, 51, 55). The similarities between these mutants, in conjunction with the fact that T(47)CA is highly conserved (suggesting its functional importance), prompted us to analyze this CA threonine residue in additional experiments.
Sequence alignment of lentivirus CA proteins. Amino acid sequences for the aligned viruses were obtained from the National Center for Biotechnology Information Entrez protein database. Gag proteins were aligned by ClustalW (www2.ebi.ac.uk/clustalw ) (52) by using the BLOSUM matrix and visualized with BOXSHADE (www.ch.embnet.org/software/BOX_form.html ). Black shading indicates sequence identity; gray shading indicates sequence similarity. The solid arrow indicates the position of T(47)CA of SIV and homologous residues; the dashed arrow indicates the position of the aspartate or glutamate that participates in the salt bridge. The SIV sequence is representative of SIVmac, SIVsmm, SIVstm, and SIVagmver strains (34). EIAV, equine infectious anemia virus; FIV, feline immunodeficiency virus; VISNA, visna virus; CAEV, caprine arthritis encephalitis virus; BIV, bovine immunodeficiency virus.
Because the HIV P(1)CA and D(51)CA residues are both critical to the structure of the mature CA N terminus (15, 20, 37), we hypothesized that SIV T(47)CA may have an important structural role as well. As the structure of SIV CA has not been reported, we investigated the crystal structure of HIV CA (15) to visualize the HIV homologue of SIV T(47)CA, HIV T(48)CA. The most striking feature of this HIV CA structure was that T(48)CA participates in four hydrogen bonds with D(51)CA: between the T(48)CA N and the D(51)CA Oδ, between the D(51)CA N and the T(48)CA Oγ, between the D(51)CA N and the T(48)CA O, and between the T(48)CA Oγ and the D(51)CA Oδ. By extension, we considered the possibility that hydrogen bonding at T(47)CA is structurally important in SIV CA and therefore potentially required for proper CA function in the viral life cycle. We hypothesized that SIV T(47)CA hydrogen bonds with D(50)CA, in a manner similar to that observed with HIV.
To test the idea that hydrogen bonding at T(47)CA is required for viral replication, we constructed a panel of polar and nonpolar substitutions of this residue in the full-length viral clone SIVmac239. Two polar amino acids were substituted for T(47)CA in SIVmac239: cysteine [T(47)CCA] and serine [T(47)SCA]. Although both polar residues may form the required hydrogen bonds, these bonds will differ from those potentially formed by T(47)CA in wild-type SIV CA. Another nonpolar substitution [in addition to T(47)ACA] was also made, replacing T(47)CA with valine [T(47)VCA]. Of note, the valine side chain is very similar to that of threonine from a space-filling perspective. Both T(47)ACA and T(47)VCA would abolish hydrogen bonding of the side chain at position 47 in CA.
Because the HIV CA structure indicates that HIV D(51)CA and T(48)CA are hydrogen bonded to each other, we also substituted alanine for the SIV homologue of HIV D(51)CA [SIV D(50)ACA]. We used D(50)ACA as a reference for comparison with the T(47)CA substitution mutants in assays that examine critical steps of viral replication. If T(47)CA and D(50)CA truly are hydrogen bonded to each other in SIV CA, we would expect that T(47)CA and D(50)CA substitution mutants might have similar functional or morphological defects. By analogy to HIV, SIV D(50)CA may participate in a salt bridge with P(1)CA. Since no previous study has investigated whether the sequence homologues of either of the HIV salt bridge partners are critical for SIV replication, for completeness, we substituted alanine for SIV P(1)CA in SIVmac239 [P(1)ACA].
Finally, we changed the glycine at position 2 of the matrix domain to alanine in SIVmac239 [G(2)AMA]. This substitution prevents myristoylation of Pr55Gag and abolishes viral budding (5, 22, 38), providing a negative control for viral replication. To ensure that no other mutations had spontaneously occurred in the cloning process, SIVmac239 mutants were verified by sequencing the entire coding region of the viral genome.
Growth kinetics of SIVmac239 substitution mutants.To compare the growth kinetics of our panel of viral mutants, CEM×174 cells were transfected with equivalent amounts of infectious wild-type or mutant viral DNA, and RT activity in virus-containing cell supernatants was measured at various times posttransfection (Fig. 3). P(1)ACA virus consistently reached peak RT activity 6 days later than wild-type virus. In contrast, D(50)ACA virus replicated in only one of three independent transfections (Fig. 3), and in this instance it was delayed by 18 days relative to the wild type and never replicated to wild-type levels.
Growth kinetics of a panel of SIVmac239 CA mutants with polar and nonpolar substitutions. RT activity in cell-free supernatants of CEM×174 cells transfected with infectious viral DNA was assayed at various times posttransfection. The D(50)ACA growth curve is not typical (arrow) and is representative of only one of three independent transfections.
All of the T(47)CA substitution mutants exhibited severely compromised viral replication: T(47)ACA, T(47)CCA, and T(47)SCA mutants consistently reached peak RT activity 8 to 12 days later than the wild type and never approached wild-type replication levels. The T(47)VCA mutant replicated in only one (data not shown) of three independent transfections, was delayed by 20 days relative to the wild type, and never reached wild-type levels. The observation that all of the T(47)CA substitution mutations inhibited viral replication indicates that a threonine at this position is crucial for the proper function of Gag in the viral life cycle.
Nonpolar substitutions of T(47)CA exhibited greatly reduced infectivity.We next investigated the relative infectivities of virions produced from cells transfected with wild-type SIV and each of our viral mutants using two assays, a standard TCID50 assay (Fig. 4A and B) and the LuSIV assay described above (Fig. 4C). The results for each substitution mutant were consistent in both assays. In keeping with the phenotypes of mutants of the HIV salt bridge participant residues, HIV D(51)ACA and HIV P(1)LCA, D(50)ACA SIV was noninfectious and P(1)ACA SIV exhibited greatly reduced infectivity relative to wild-type virus (13, 51, 55) (Fig. 4). All of the substitutions of T(47)CA greatly compromised viral infectivity, although mutants with the polar substitutions T(47)SCA and T(47)CCA were more infectious than those with nonpolar substitutions [T(47)VCA and T(47)ACA]. The infectivity assay results were consistent for virus derived from both 293T cells (Fig. 4A and C) and COS-1 cells (Fig. 4B and data not shown [LuSIV]), demonstrating that these results are independent of cell type. Collectively, these data indicate that substitution of T(47)CA, D(50)CA, or P(1)CA compromises virion infectivity. Furthermore, nonpolar substitutions of T(47)CA are more detrimental to virus infectivity than polar substitutions at this position.
TCID50 and infectivity of wild-type and mutant SIVmac239. (A) TCID50 of virus derived from 293T cells at day 7 postinfection. (B) TCID50 of virus derived from COS-1 cells at day 7 postinfection. (C) LuSIV assay for infectivity of virus derived from 293T cells. Values are the fold induction over the background at day 2 postinfection. Results are representative of three independent experiments.
Virion release and incorporation of viral proteins.Since the attenuation of viral replication observed for some of the substitution mutants was severe, the observed replication defects could have been manifested not only in virus entry into cells (described above) but also in virion release and/or viral protein incorporation. To assess whether any of the CA substitution mutants were defective in the release of nascent virions, 293T cells were transfected with equivalent amounts of wild-type or mutant SIVmac239 infectious viral DNA and metabolically labeled with [35S]methionine-cysteine. Cell and virus lysates (see Materials and Methods) were immunoprecipitated with SIV CA antiserum and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by autoradiography (Fig. 5A). As 293T cells are nonpermissive to SIV infection, this assay reflects one round of virus production. The band just above CA in this assay is the CA-p2 cleavage intermediate, similar to that described for Rous sarcoma virus (57) and HIV-1 (53). These gels were subjected to densitometric analysis, and the relative virion release efficiencies for the substitution mutants and wild-type SIV were compared by dividing the total amount of Gag in the virus lysate by the sum of the total amount of Gag in the cell and virus lysates (Fig. 5B).
Virion release and analysis of viral proteins. (A) Virus lysates and cell lysates from metabolically labeled transfected 293T cells were immunoprecipitated with IgG-purified SIV CA antiserum, resolved by SDS-PAGE, and visualized by autoradiography. (B) Relative virion release efficiency was calculated as the total amount of Gag in the virus lysate divided by the sum of the total amount of Gag in the cell and virus lysates, based on densitometric analysis of the gels in panel A. (C) Virus lysates produced as for panel A were immunoprecipitated with IgG-purified SIV antiserum, resolved by SDS-PAGE, and visualized by autoradiography. Env, envelope; INT, integrase.
Consistent with previous studies (5, 22, 38), no Gag was detected in G(2)AMA virus lysates, indicating a severe defect in virion release. The overaccumulation of Pr55Gag observed in cells transfected with G(2)AMA SIV infectious viral DNA is consistent with the phenotype of HIV-1 G(2)AMA (22). While P(1)ACA virus released approximately wild-type levels of Gag, the D(50)ACA mutant released only about 30% as much Gag into the cell supernatant as wild-type virus did, indicating a defect in particle production. Mutants with the nonpolar substitutions T(47)ACA and T(47)VCA released approximately 90 and 60% as much virus into the cell supernatant as wild-type virus, respectively, while mutants with the polar substitutions T(47)CCA and T(47)SCA released approximately wild-type levels of Gag. Similar results were obtained for the panel of mutants when cell and virus lysates from transfected 293T or COS-1 cells were analyzed directly by Western blot analysis with SIV CA antiserum or SIV antiserum (data not shown). Based on the infectivity data (Fig. 4), the Gag that is released by the budding-deficient D(50)ACA, T(47)ACA and T(47)VCA mutants does not likely represent infectious virus.
The virion release defects observed for D(50)ACA, T(47)ACA, and T(47)VCA were somewhat unexpected, given that defective cell exit is not typical of substitutions in the CA N terminus (11, 43, 44, 56). However, our group has recently characterized several other substitution mutants in this region of CA that also exhibit release defects (Roos et al., submitted), and a mutant with a substitution for proline 34 of HIV-1 CA is similarly deficient in particle production (13). Nevertheless, the results of all three virion release assays indicate that the D(50)ACA mutant exhibits impaired virion release, while the P(1)ACA mutant does not. Furthermore, nonpolar substitutions of T(47)CA compromise virion release more than polar substitutions at this position.
To determine whether the CA substitution mutants exhibited a gross defect in viral protein incorporation, 293T cells transfected with wild-type or mutant infectious viral DNA were metabolically labeled with [35S]methionine-cysteine. Virus lysates were immunoprecipitated with SIV antiserum and resolved by SDS-PAGE (Fig. 5C). In keeping with the virion release data (Fig. 5A and B), the G(2)AMA mutant exhibited a severe release defect. As envelope, RT, and integrase were detected for all of the other mutants, none of these substitutions caused a dramatic defect in viral protein incorporation.
Transmission EM phenotypes of wild-type and mutant SIVmac239.We investigated whether capsid morphology was affected by the CA substitutions by transmission EM of 293T cells transfected with the mutant panel. In the wild-type sample (Fig. 6A), we observed budding, immature, and mature virus particles, consistent with normal lentiviral morphology. All three typical mature core phenotypes were present in this sample: centric (core is round and centered with respect to the viral envelope), conical (core is cone-shaped and centered), and acentric (described above). No intracellular or extracellular viral particles were observed in the G(2)AMA sample (data not shown), a finding that is in keeping with EM analysis of G(2)AMA HIV-1 (14).
Transmission electron micrographs of wild-type and CA mutant SIVmac239 virions. 293T cells transfected with wild-type (WT) or mutant infectious viral DNA were harvested 1 day posttransfection, ultrathin sectioned, and analyzed by transmission EM. Labeled arrows indicate virions representative of typical virion morphologies: Ce, centric; I, immature; A, acentric; Co, conical. Unlabeled arrows show tethered immature virions (G), doublet and triplet particles (J), tethered virion chains (K), and accumulation of Gag at the plasma membrane (L). Bars, 100 nm.
Substitution of alanine for HIV-1 D(51)CA [HIV D(51)ACA] (51, 55) or leucine for HIV P(1)CA [HIV P(1)LCA] (13) alters the core morphology of mature virions, such that the mature cores are predominantly acentric. The core morphology of substitution mutants of the SIV sequence homologues of either of these HIV-1 salt bridge participants has not, to our knowledge, been examined. Unlike P(1)LCA HIV (13), P(1)ACA SIV was morphologically wild type by EM (Fig. 6B). The differences between P(1)ACA SIV and P(1)LCA HIV are likely due to inherent differences in tolerance of substitutions at this position between HIV and SIV or to the fact that different substitutions were made. In contrast, consistent with the EM phenotype of D(51)ACA HIV, D(50)ACA SIV virions exhibited gross morphological abnormalities (Fig. 6C and D). In this sample, several budding particles emerged from the same point in the plasma membrane, giving the buds an uncharacteristic multilobed shape (Fig. 6C). Some of these buds had pinched off from the plasma membrane, forming oddly shaped, abnormally large immature particles. While relatively few mature virus particles were observed in cells transfected with the D(50)ACA mutant, all of them had acentric cores (see Fig. 6D for representative virions; see Table 1 and below).
Proportion of acentric mature cores observed for wild-type and capsid mutant SIVmac239 viruses
Consistent with our earlier studies, the EM morphology of T(47)ACA virus (Fig. 6E, F, and G) was strikingly reminiscent of that of the D(50)ACA mutant. Budding virions had a branched or multilobed appearance, as though several virions were exiting the cell from the same region of the plasma membrane (Fig. 6E). Importantly, there were some mature virions present in the T(47)ACA sample (Fig. 6F), but those also appeared to have mostly acentric cores. Some extracellular immature virions in the T(47)ACA photomicrographs were tethered together (Fig. 6G), similar to the tethered structures observed in late domain and deletion mutants of HIV-1 p6 (10, 21, 59).
In order to quantitatively determine whether T(47)ACA produces a higher proportion of acentric mature virions than wild-type virus, we compared the proportions of mature wild-type, D(50)ACA, and T(47)ACA SIVmac239 virions with acentric, centric, and conical mature cores (Table 1) by statistical analysis (see Materials and Methods). T(47)ACA virus had an intermediate proportion of acentric mature cores, which was significantly different from both the wild-type and D(50)ACA proportions. Therefore, the T(47)ACA substitution causes acentric virion core morphology, but to a lesser degree than the D(50)ACA substitution.
T(47)CCA had approximately wild-type proportions of budding, immature, and mature virions, but roughly 40% of the mature cores appeared to have condensed in an aberrant manner, as they had either two electron-dense core regions or acentric cores (Fig. 6H). Interestingly, the T(47)CCA mature core phenotype is similar to that reported for an HIV-1 p1-p6 cleavage site mutant (59). Few (∼10%) mature virions were observed in the T(47)SCA sample, and the majority of these had abnormal or acentric cores (Fig. 6I). Approximately 90% of the virions in this sample were budding or extracellular and immature, and many of these virions were connected as doublet or triplet particles (Fig. 6J) or tethered to the plasma membrane in chains (Fig. 6K), a morphology that resembles both T(47)ACA and HIV-1 p6 mutants (10, 21, 59). No extracellular mature virions were found in cells transfected with the T(47)VCA mutant. Approximately 95% of Gag in this sample accumulated at the plasma membrane in discrete patches (Fig. 6L). Only 5% of the T(47)VCA virus particles were extracellular (data not shown), and the majority of these were large, multilobed, and immature, similar to the immature particles in the D(50)ACA sample.
Molecular modeling of SIVmac239 CA.As mentioned before, T(48)CA and D(51)CA form four hydrogen bonds in HIV CA (15) (see Fig. 7A for a model of these residues in HIV CA). Because both T(47)ACA and D(50)ACA had virion infectivity and release defects, and because the virion morphologies of these two mutants were so similar, we hypothesized that the interaction between HIV T(48)CA and HIV D(51)CA is conserved in SIV. To address this question, we constructed a model of SIVmac239 CA by homology to the crystal structure of the HIV-1 CA N terminus (15) using the program QUANTA (Fig. 7B). Due to sequence constraints, we limited this model to the region surrounding T(47)CA. Significantly, in the model, T(47)CA and D(50)CA form the same four hydrogen bonds that exist between the corresponding HIV CA residues: between the T(47)CA N and the D(50)CA Oδ, between the D(50)CA N and the T(47)CA Oγ, between the D(50)CA N and the T(47)CA O, and between the T(47)CA Oγ and the D(50)CA Oδ.
Models of the HIV-1 and SIVmac239 CA proteins. Models were generated with QUANTA and rendered with SETOR (12). (A) Region surrounding HIV-1 T(48)CA modeled according to the reported crystal structure (15). (B) Region surrounding SIV T(47)CA modeled by homology to the HIV CA structure used for panel A. Color coding: blue, nitrogen; red, oxygen; gray, carbon.
DISCUSSION
In this study, we characterized the functional importance of hydrogen bonding at a conserved threonine in SIVmac239 CA, T(47)CA. Two observations from this study indicate that hydrogen bonding at T(47)CA is essential to virus replication. First, mutants with nonpolar substitutions of T(47)CA were more compromised in virion infectivity and release than polar substitution mutants, indicating that hydrogen bonding at T(47)CA is specifically required for early and late events in the SIV life cycle. Secondly, T(47)VCA was severely deficient in virion release and infectivity relative to wild-type virus. As threonine and valine are quite similar from a space-filling perspective, the deficiencies of T(47)VCA are the result of the loss of hydrogen bonding at this position, rather than an alteration in the shape of the amino acid side chain.
The phenotypic similarities between the D(50)ACA mutant and T(47)CA substitution mutants [particularly the T(47)ACA mutant] indicate that these two residues likely have similar functionally important structural roles in SIV CA. Taken together, the results of the mutagenesis study and the modeling experiment presented herein strongly suggest that hydrogen bonding between SIV T(47)CA and D(50)CA is present in SIV CA and is required for viral replication. Because this interaction may be critical to CA function in the virus life cycle, these results provide a solid rationale for our ongoing efforts to determine the structure of mature SIV CA.
Two results indicate that T(47)ACA SIV is phenotypically intermediate between D(50)ACA and WT SIV [i.e., substitution of D(50)CA is more detrimental to the virus than substitution of T(47)CA], suggesting that D(50)CA may have an additional important structural interaction in CA. First, the proportion of acentric mature cores observed in photomicrographs of T(47)ACA is significantly different from that of either the wild type or the D(50)ACA mutant. Second, the budding deficiency of T(47)ACA virus was not as extreme as that of D(50)ACA virus. One hypothesis to explain the intermediate phenotype of the T(47)ACA mutant is that hydrogen bonding between T(47)CA and D(50)CA may position D(50)CA throughout the rearrangement of the SIV CA N terminus so that it can optimally participate in another requisite interaction, such as a salt bridge as in HIV.
The nuclear magnetic resonance structure of HIV-1 Pr55Gag from MA through the N terminus of CA was recently reported (50). In this structure, HIV-1 T(48)CA hydrogen bonds with the aspartate that participates in the salt bridge [HIV D(51)CA] in the immature form of CA, although only one of the four hydrogen bonds present in mature CA, the hydrogen bond between the D(51)CA N and the T(48)CA Oγ, likely exists in immature CA; the heavy atom distances for the other three bonds in the majority of the reported nuclear magnetic resonance models are greater than 3.5 Å. The fact that hydrogen bonding between T(48)CA and D(51)CA exists in both the immature and the mature forms of HIV CA supports the idea that T(48)CA positions D(51)CA for a requisite interaction with another residue [likely P(1)CA] throughout the rearrangement of the CA N terminus. However, based on the reported structures of mature and immature HIV CA (15, 50), it is reasonable to conclude that the pattern of hydrogen bonding between T(48)CA and D(51)CA changes subsequent to the proteolytic release of the HIV CA N terminus. As such, the three additional hydrogen bonds present only in mature HIV CA may provide critical support to the salt bridge in its largely hydrophobic environment.
We conclude that hydrogen bonding between the SIV homologues of HIV T(48)CA and D(51)CA is likely essential for proper CA function in virion release, infectivity, and core formation, suggesting that the interaction between these residues may be critical in HIV as well. If our conclusions about the requirement for hydrogen bonding between these two residues in SIV replication hold true in HIV, the change in hydrogen bonding between HIV T(48)CA and D(51)CA during the rearrangement of the CA N terminus may be a target for inhibitors of virus replication, since the structural constraints at these sites limit the acquisition of viable viral mutations.
ACKNOWLEDGMENTS
We are extremely grateful to Patrick M. Tarwater (School of Public Health, University of Texas Health Science Center at Houston) for his extensive statistical analysis of the virion morphology comparison results. We thank John Bernbaum for his superb EM services, Justyna Dudaronek for constructing G(2)AMA SIVmac239, and Brandon Bullock, Laurie Queen, and David Herbst for excellent technical assistance. Finally, we thank Gary Ketner and Carolyn Machamer for critical reading of the manuscript. The following reagents were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: p239SpSp5′ and p239SpE3′ from Ronald Desrosiers.
This work was supported by NIH grants to J.E.C. (NS07392, NS35751, and NS23039).
Under a licensing agreement between Bayer AG and the Johns Hopkins University, J.E.C. is entitled to a share of a payment received by the university on sales of products embodying the technology described in this article. The terms of this agreement are being managed by the Johns Hopkins University in accordance with its conflict-of-interest policies.
FOOTNOTES
- Received 16 January 2003.
- Accepted 29 April 2003.
- Copyright © 2003 American Society for Microbiology
