|Home | About | Journals | Submit | Contact Us | Français|
The host restriction factor TRIM5α provides intrinsic defense against retroviral infections in mammalian cells. TRIM5α blocks infection by targeting the viral capsid after entry but prior to completion of reverse transcription, but whether this interaction directly alters the structure of the viral capsid is unknown. A previous study reported that rhesus macaque TRIM5α protein stably associates with cylindrical complexes formed by assembly of recombinant HIV-1 CA-NC protein in vitro and that restriction leads to accelerated HIV-1 uncoating in target cells. To gain further insight into the mechanism of TRIM5α-dependent restriction, we examined the structural effects of TRIM5 proteins on preassembled CA-NC complexes by electron microscopy. Incubation of assembled complexes with lysate of cells expressing the restrictive rhesus TRIM5α protein resulted in marked disruption of the normal cylindrical structure of the complexes. In contrast, incubation with lysate of control cells or cells expressing comparable levels of the nonrestrictive human TRIM5α protein had little effect on the complexes. Incubation with lysate of cells expressing the TRIMCyp restriction factor also disrupted the cylinders. The effect of TRIMCyp was prevented by the addition of cyclosporine, which inhibits binding of TRIMCyp to the HIV-1 capsid. Thus, disruption of CA-NC cylinders by TRIM5α and TRIMCyp was correlated with the specificity of restriction. Collectively, these results suggest that TRIM5α-dependent restriction of HIV-1 infection results from structural perturbation of the viral capsid leading to aberrant HIV-1 uncoating in target cells.
In mammalian cells, intrinsic immunity has evolved as a defense against retroviral infection. Several antiviral systems mediated by restriction factors are important in establishing species-specific barriers against viral infection. One factor responsible for restriction in human and nonhuman primate cells is the alpha isoform of tripartite motif protein 5 (TRIM5α). TRIM5α was first identified as the protein responsible for host restriction of HIV-1 in rhesus macaque cells and was later shown to account for restriction of N-tropic murine leukemia virus (N-MLV) in human cells (5, 14, 16, 18, 32, 42, 53).
Localized in structures known as cytoplasmic bodies (42), TRIM5α proteins are dimeric in form (19). Like many other members of the tripartite motif family of proteins, TRIM5α contains RING, B-box, coiled-coil, and B30.2/SPRY domains (27, 33). Each domain contributes to the restriction activity of TRIM5α. The RING domain has intrinsic E3 ubiquitin ligase activity (51, 52) and is required for restriction activity (17, 29, 30, 42). The B-box domain appears to promote higher-ordered assembly of TRIM5α and is essential for restriction (8, 17, 20, 29). TRIM5α dimerization is a function of the coiled-coil domain and is necessary for capsid binding and restriction (17, 23, 29). The B30.2/SPRY domain is required for recognition of the incoming viral capsid, and sequence variation in this domain determines the specificity of retroviral restriction (16, 18, 25, 28, 29, 32, 40, 41, 44, 53, 54). For example, rhesus macaque TRIM5α inhibits HIV-1, whereas the human TRIM5α restricts N-MLV, equine infectious anemia virus (EIAV), and feline immunodeficiency virus (FIV) (15, 16, 18, 32, 36, 42, 44, 53, 56).
Another restriction factor, TRIMCyp, was originally identified in owl monkey cells. Like TRIM5α, TRIMCyp contains RING, B-box, and coiled-coil domains but with cyclophilin A (CypA) substituted for the B30.2/SPRY domain (26, 37). CypA directly binds the HIV-1 CA protein, and this interaction is inhibited by the addition of cyclosporine (CsA) (12, 22, 46). The replacement of the B30.2/SPRY domain by CypA confirmed that this region of TRIM5α is primarily responsible for recognition of the viral capsid. Restriction by TRIMCyp involves direct binding to CA (47): either introduction of a single amino acid substitution (G89V) into HIV-1 CA that ablates CypA binding (55) or addition of CsA renders owl monkey cells susceptible to HIV-1 infection (47). A TRIMCyp protein has also been identified in other primates, indicating that resistance to retroviral infection can be advantageous for the species (6, 21, 48, 49).
Although the detailed mechanism of restriction by TRIM5 proteins is unknown, several lines of evidence argue that TRIM5α binds directly to the retroviral capsid. A glutathione S-transferase (GST)-TRIM5α fusion protein was reported to bind N-MLV CA presented by detergent-stripped virions in vitro (38). GST-TRIM5α associated with N-MLV CA but not with the CA of B-tropic MLV (B-MLV), corresponding to the known viral specificity of restriction by the human TRIM5α protein. Additionally, a purified recombinant TRIM5α protein containing a replacement of the RING domain by that of TRIM5-21R bound to synthetic assemblies of HIV-1 CA-NC protein in vitro (19). Recognition of the viral capsid appears to occur prior to uncoating in target cells: by analyzing the effects of mutant HIV-1 particles containing unstable capsids, our lab has shown that the ability of the virus to saturate restriction is dependent on an intact or partially intact viral capsid (10, 39). This interpretation is also supported by the observation that expression of CA in target cells fails to abrogate restriction, suggesting that CA must be in an assembled form for stable TRIM5α association (9). TRIM5α-related proteins from different species can cosediment with HIV-1 CA-NC complexes assembled in vitro, and this association is correlated with the specificity of restriction (45). More recently, recombinant TRIM5-21R was reported to bind directly to assembled HIV-1 CA-NC complexes by an interaction dependent on an intact SPRY domain (19). Collectively, these results suggest a model involving recognition of an intact, polymeric viral capsid by oligomeric TRIM5α.
TRIM5α inhibits HIV-1 infection by targeting the viral capsid following entry, leading to abortive reverse transcription (2-4, 24, 42). The detailed molecular mechanism of TRIM5α restriction is not known. The available evidence suggests that association of the proteins with the viral capsid leads to either premature uncoating or degradation of the viral capsid. Under normal circumstances, TRIM5α inhibits the accumulation of viral reverse transcripts in the cytoplasm (42) and appears to accelerate HIV-1 capsid disassembly in target cells (31, 43). However, addition of proteasome inhibitors results in completion of reverse transcription, yet the cells remain resistant to infection (1, 7), suggesting that host accessory factors are involved in the process of restriction.
In the present study, we tested the hypothesis that TRIM5α disrupts the structure of the viral capsid. We examined the structural effects of TRIM5α and TRIMCyp proteins on preassembled tubular complexes of HIV-1 CA-NC protein by transmission electron microscopy. Analysis of human and rhesus macaque TRIM5α and TRIMCyp proteins revealed a correlation of the specificity of restriction with the structural disruption of the complexes.
Highly HIV-1-restrictive 293T cell lines expressing rhesus macaque and human TRIM5α proteins were generated by stable transduction with pLPCX vectors (42). Cells expressing TRIMCyp were generated as previously described (34). Rhesus macaque and human TRIM5α proteins contained a C-terminal hemagglutinin (HA) tag, and the TRIMCyp protein contained a Myc-His tag. The TRIMCyp-expressing line was expanded from a single-cell clone. All cell lines were cultured in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin and streptomycin, and puromycin (2.5 μg/ml).
Recombinant HIV-1 CA-NC protein was purified from Escherichia coli as described previously (13). Cylindrical complexes were formed by incubating 300 μM CA-NC protein with 60 μM (TG)50 DNA oligonucleotide (Integrated DNA Technologies) in a solution containing 50 mM Tris-HCl and 500 mM NaCl (pH 8.0) overnight at 4°C as described previously (43). Cell lysate was generated from 293T cell lines 48 h after seeding 2 × 106 cells in each of 20 100-mm dishes and culture for 3 days. The cells were detached with a solution of phosphate-buffered saline (PBS) containing 5 mM EDTA and resuspended in 2 ml of lysis buffer (50 mM Tris-HCl, 5 mM MgCl2, 0.5 mM EDTA [pH 7.5], 1 mM dithiothreitol [DTT], 1:100 mammalian protease inhibitor [Sigma]). Cells were lysed on ice with a ball-bearing homogenizer (18-μm gap, 10 passes) on ice, and lysates were clarified by centrifugation at 13,000 × g for 12 min at 4°C in an Eppendorf microcentrifuge. Aliquots of the supernatants were and stored at −80°C. Protein concentrations in the lysates were quantified by the bicinchoninic acid (BCA) assay (Pierce Chemical Co.). Volumes of cell lysate containing 200 μg of total cell protein were incubated with 5 or 10 μl of HIV-1 CA-NC assembly complexes in reaction volumes of 80 μl. The final reaction mixtures contained 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 0.5 mM EDTA, and 1 mM DTT. The binding reaction mixture was incubated for 1 h at room temperature with gentle agitation every 15 min. After incubation, each reaction mixture was layered onto a 1 ml 70% sucrose-PBS cushion in a 1.5-ml Beckman microcentrifuge tube and centrifuged in a Beckman TLA 55 rotor (45,000 × g, 1 h, 4°C). The sucrose cushions were removed by aspiration, and pellets were dissolved in SDS-PAGE sample loading buffer and subjected to electrophoresis on 4 to 20% acrylamide gradient gels containing SDS (Bio-Rad). Proteins were transferred electrophoretically to nitrocellulose, and the membranes were probed with anti-HA rat monoclonal antibody (clone 3F10; Roche Diagnostic), anti-c-Myc mouse monoclonal antibody (clone 9E10, IgG1; Roche Diagnostic), or anti-HIV-1 CA mouse monoclonal antibody (clone 183-H12-5C; NIH AIDS Research and Reference Reagent Program), followed by Alexa Fluor 680-conjugated goat anti-rat IgG secondary antibody (Molecular Probes). Bands were visualized by infrared detection with the Odyssey imaging system (Li-Cor).
Quantities of cell lysate containing 50 μg of total cell protein were incubated with HIV-1 CA-NC assembly complexes in the binding assay as described above. Following incubation, a 15-μl droplet of the reaction was applied directly to a carbon type-A, 300-mesh copper grid (Ted Pella, Inc.) for 1 min. Bound complexes were negatively stained with a 15-μl droplet of 2% uranyl acetate for 1 min, followed by a wash with a 15-μl droplet of H2O and air drying. After each incubation and wash, most of the liquid was removed from the grid by wicking with a torn edge of filter paper. Grids were examined with a Philips CM-12 transmission electron microscope, and images acquired with an AMT digital acquisition camera at a magnification of ×53,000.
The relative abundance of cylinders on the EM grids was determined by quantifying the average area of the grid occupied by cylindrical structures by the area-from-the-point-Count method of Russ and Dehoff (35). For each reaction, 20 random fields were captured to represent the sample. A 27- by 27-point grid was superimposed on each image using the Adobe Photoshop CS/Fovea Pro plug-in. The area fraction was calculated as the number of point hits on the cylindrical HIV-1 CA-NC complexes divided by the total number of points on the grid (729 points). The significance of the results of multiple experiments was determined by applying the Wilcoxon signed-rank test.
To test the hypothesis that TRIM5α alters the regular structure of the viral capsid, lysates of 293T cells stably expressing TRIM proteins were incubated with preassembled tubular HIV-1 CA-NC complexes formed by incubation of the recombinant protein with a (TG)50 oligonucleotide. A control (mock) reaction mixture contained only the HIV-1 CA-NC complexes with no cell extract (Fig. (Fig.11 A). Numerous tubular complexes were observed, consistent with the established ability of CA-NC protein to assemble into a cylindrical hexameric lattice. When the complexes were incubated with control 293T lysate (no TRIM5α) or with human TRIM5α, tubular complexes were still evident. In contrast, incubation with cell lysate containing rhesus TRIM5α resulted in an apparent loss of the tubular complexes. The abundance of tubular complexes in each sample was determined as the fractional area of each grid occupied by tubes. The average values from six independent experiments were as follows: mock (no lysate), 2.76%; 293T, 1.42%; human TRIM5α, 1.47%; and rhesus TRIM5α, 0.13% (Fig. (Fig.1B).1B). Statistical analysis demonstrated that the effect of the lysate containing human TRIM5α on the number of the HIV-1 CA-NC complexes was not significantly different from that of the control lysate. In contrast, incubation with lysate containing rhesus TRIM5α reaction significantly reduced the number of observed tubular complexes compared to that in the other reactions. Some darkly stained web-like material was also observed in all samples. In the case of the mock, 293T, and human TRIM5α reactions, the cylinders were sometimes found trapped within this material. Assays of the binding of the TRIM5α proteins to the CA-NC complexes revealed a correlation of the loss of tubular structures to the specificity of TRIM5α binding to the complexes, as shown by immunoblotting of the pelleted reaction mixtures for TRIM5α (Fig. (Fig.1C).1C). There was no apparent effect of the lysate on the levels of pelleted CA-NC protein (Fig. (Fig.1C),1C), indicating that the incubation with the cell lysate had not dissolved the complexes, as previously reported (43). Analysis of the cell lysates demonstrated equivalent quantities of rhesus and human TRIM5α proteins, confirming that the specificity was not an artifact of differential expression of the TRIM5 proteins (Fig. (Fig.1D).1D). Sedimentation of rhesus TRIM5α was dependent on the addition of CA-NC tubes (Fig. (Fig.1E1E).
To further test the connection between structural effects of TRIM5α on CA-NC complexes and restriction of HIV-1 infection, we tested the effects of lysate of 293T cells expressing the owl monkey restriction factor TRIMCyp. To control for nonspecific effects unrelated to TRIMCyp binding, we included a control reaction mixture containing added cyclosporine to prevent the binding of TRIMCyp to CA. Each reaction mixture was negatively stained on a carbon copper grid and visualized by electron microscopy. Incubation of the complexes with 293T lysate without TRIMCyp had little or no effect on the complexes (Fig. (Fig.22 A). In contrast, incubation with lysate of TRIMCyp-expressing cells resulted in a marked reduction in tubular HIV-1 CA-NC complexes detected on the grid. CsA appeared to prevent the effect of TRIMCyp; reaction mixtures containing cyclosporine contained nearly as many cylinders as the reaction mixture containing control 293T lysate. Quantification of the relative number of tubular structures in each sample revealed an 88% reduction in the number of cylinders in the reaction mixture with TRIMCyp (Fig. (Fig.2B).2B). Addition of CsA reduced this effect to 31%. Statistical analysis revealed no significant difference in tubular structures in reaction mixtures with the control 293T lysate and those containing both TRIMCyp and CsA. As expected, CsA prevented the binding of TRIMCyp to CA-NC tubes (Fig. (Fig.2C).2C). Finally, CA-NC complexes assembled from a CA-NC protein containing the G89V substitution in CA were unaffected by TRIMCyp (Fig. (Fig.3).3). The G89V mutation allows HIV-1 to escape restriction by owl monkey TRIMCyp by preventing binding to CA, adding further support to the link between the specificity of restriction and the loss of detectable tubular CA-NC structures in the reaction mixtures.
Collectively, our data indicate that tubular complexes of recombinant HIV-1 CA-NC protein with lysate of restrictive cells undergo structural disruption when incubated with lysate of human cells expressing restrictive TRIM5 proteins. The complexes were not solubilized, as the TRIM5α proteins had no effect on the levels of CA-NC protein detected in the pellets upon centrifugation. Both rhesus TRIM5α and TRIMCyp reduced the number of observed tubular structures. In contrast, human TRIM5α, which restricts HIV-1 only weakly, or cell lysate lacking TRIM5α did not significantly alter the number of detectable CA-NC tubes in the reaction mixtures. Similarly, addition of CsA, an inhibitor of the binding of TRIMCyp to HIV-1 CA, prevented the loss of tubes mediated by TRIMCyp, as did the G89V mutation in CA, which allows HIV-1 to evade restriction by TRIMCyp. Collectively, our results show a clear correlation of the specificity of restriction by TRIM5 factors with a loss of tubular structures in the reaction mixtures.
How does the binding of TRIM5 restriction factors to CA result in a loss of detectable tubular complexes in the reaction mixtures containing cell lysate but without solubilizing the CA-NC complexes? We propose that the restriction factor, perhaps in combination with accessory factors present in the cell lysate, disrupts the regular hexameric lattice that comprises the assembled tubular complex. In the case of CA-NC tubes, which are stabilized by the (TG)50 oligonucleotide used to promote their assembly, the structural disruption would not necessarily lead to solubilization, due to stabilization of the complex via binding NC domains of multiple subunits to each oligonucleotide molecule. Recent studies with a purified recombinant protein consisting of the rhesus macaque TRIM5α protein with a substitution of the RING domain of TRIM21 showed that the protein binds to CA-NC tubes in vitro (19, 43). Expression of the TRIM5-21R protein in human cells resulted in resistance to HIV-1 infection, suggesting that the protein is a functional restriction factor. However, examination of the binding reaction mixtures by transmission electron microscopy after negative staining revealed a shortening of the tubes as the only apparent structural alteration (19). We propose that cells contain cofactors for restriction that recognize TRIM5α when bound to a viral capsid. One possibility is that the RING domain becomes active as a ubiquitin ligase when TRIM5α is assembled onto the viral capsid. This could result in autoubiquitylation, an activity which TRIM5α and the related protein TRIM5δ exhibit (51, 52), leading to recruitment of the proteasome, subsequent degradation of TRIM5α, and concomitant dissolution of the capsid. This premature uncoating would impair reverse transcription, as has been observed for HIV-1 mutants with intrinsically unstable capsids (11). Consistent with this model, TRIM5α is degraded by a proteasome-dependent mechanism upon encounter of a restriction-sensitive core in the target cell (34). Proteasome activity is also required for the early block to retroviral reverse transcription mediated by most restrictive alleles of TRIM5α (50). However, despite the attractiveness of this model, we observed no protection of the CA-NC complexes from the structural effect of rhesus macaque TRIM5α by the proteasome inhibitor MG-132 (data not shown). Thus, the structural effects of TRIM5α on the capsid may be dependent on a host factor unrelated to the ubiquitin-proteasome pathway or may require a specific modification of TRIM5α that does not take place in the insect cells in which the TRIM5-21R protein was produced. Reconstitution of the structural disruption of CA-NC tubes by recombinant TRIM5 proteins in combination with cell extracts may facilitate the identification of host accessory factors involved in restriction.
We thank Jing Zhou and Chris Rold for technical assistance and Joseph Sodroski for TRIM5α expression plasmids. Imaging studies were performed in part through the use of the VUMC Research EM Resource. We also thank Jay Jerome in the Vanderbilt Cell Imaging Resource for technical assistance and advice and the Vanderbilt Institute for Clinical and Translation Research Biostatistics Clinic for statistical advice. The monoclonal antibody 183-H12-5C was obtained from Bruce Chesebro via the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID.
This work was supported by NIH grants AI75423 and AI076121. L.R.B. was supported by NIH grants T32AI007474 and F32AI071791. The VUMC Research EM Resource is supported in part by NIH grants DK20539, CA68485, and DK58404.
Published ahead of print on 21 April 2010.