The Human TRIM5α Restriction Factor Mediates Accelerated Uncoating of the N-Tropic Murine Leukemia Virus Capsid

Abstract

The host cell factors TRIM5α_hu and Fv-1 restrict N-tropic murine leukemia virus (N-MLV) infection at an early postentry step before or after reverse transcription, respectively. Interestingly, the identity of residue 110 of the MLV capsid determines susceptibility to both TRIM5α_hu and Fv-1. In this study, we investigate the fate of the MLV capsid in cells expressing either the TRIM5α_hu or Fv-1 restriction factor. The expression of TRIM5α_hu, but not Fv-1, specifically promoted the premature conversion of particulate N-MLV capsids within infected cells to soluble capsid proteins. The TRIM5α_hu-mediated disassembly of particulate N-MLV capsids was dependent upon residue 110 of the viral capsid. Furthermore, the deletion or disruption of TRIM5α_hu domains necessary for potent N-MLV restriction completely abrogated the disappearance of particulate N-MLV capsids observed with wild-type TRIM5α_hu. These results suggest that premature disassembly of the viral capsid contributes to the restriction of N-MLV infection by TRIM5α_hu, but not by Fv-1.

Intracellular restriction factors can govern the species-specific tropism of a broad range of retroviruses. Early studies identified the Friend virus susceptibility factor 1 (Fv-1) locus as a mouse determinant governing the ability of mouse strains to be infected by murine leukemia viruses (MLV) and to subsequently develop leukemia (17). Two major alleles of Fv-1 have been identified: Fv-1ⁿ, which renders NIH mice resistant to B-tropic MLV (B-MLV) infection but susceptible to N-tropic (N-MLV) infection, and Fv-1^b, which renders BALB/c mice resistant to N-MLV but susceptible to B-MLV infection (4, 10). Fv-1 is a dominant, saturable restriction factor that inhibits MLV infection following reverse transcription of the viral RNA but prior to the integration of viral DNA into the cellular genome (14). The Fv-1 protein is related to retroviral capsid proteins (14). The viral determinant of susceptibility to Fv-1 restriction has been mapped to residue 110 of the MLV capsid (7, 16). Currently, the mechanism by which Fv-1 restricts MLV infection is unknown.

More recently, rhesus monkey TRIM5α (TRIM5α_rh) and human TRIM5α (TRIM5α_hu) were identified as intracellular restriction factors capable of blocking human immunodeficiency virus type 1 (HIV-1) and N-MLV infection, respectively, in the cells of these primate species (11, 15, 24, 33, 38). TRIM5α restricts HIV-1 and N-MLV at an early postentry step prior to reverse transcription (2, 3, 6, 12, 19, 30, 36). Interestingly, arginine 110 of the N-MLV capsid determines sensitivity to TRIM5α_hu-mediated restriction (24, 36). Replacing the arginine residue at position 110 of the N-MLV capsid with the corresponding glutamic acid residue from B-MLV generates a virus that can partially overcome TRIM5α_hu restriction. Conversely, replacement of glutamic acid 110 of the B-MLV capsid with arginine generates a virus that is susceptible to TRIM5α_hu restriction.

TRIM5α is a member of the tripartite motif family of proteins and contains RING, B-box 2, and coiled-coil (RBCC) domains (25). At least three isoforms (α, γ, and δ) of TRIM5 are generated by alternative splicing of the mRNA transcript. The longest of these isoforms, TRIM5α, contains a C-terminal B30.2/SPRY domain. The role of each of the TRIM5α domains in retroviral restriction has been studied. For instance, removal of the RING domain partially abrogates TRIM5α restriction of either HIV-1 or N-MLV (13, 21, 33). Deletion of the B-box 2 domain completely eliminates the restricting ability of TRIM5α; the precise function of this domain remains unknown (13, 21). The coiled-coil domain has been shown to facilitate the multimerization of TRIM5α, and its deletion completely abrogates HIV-1 and N-MLV restriction (1a, 18, 21). The TRIM5α B30.2 domain is the primary determinant for capsid recognition and restriction specificity (23, 26, 28, 35, 39). TRIM5γ, which lacks a B30.2 domain, and mutants of TRIM5α with deleted B30.2 domains do not block HIV-1 or N-MLV infection (21, 33).

The exact mechanism by which TRIM5α restricts N-MLV infection is currently unknown. To investigate how TRIM5α mediates the restriction of retroviruses, we developed a fate-of-capsid assay that allowed us to monitor the steady-state levels of particulate retroviral capsids within an infected cell (34). This assay discriminates between cytosolic capsids that are potentially on the infection pathway and capsids that are nonspecifically endocytosed into the cell but do not enter the cytosol. The assay further distinguishes particulate capsids from smaller moieties in the cytosol of infected cells. In this study, we utilized the fate-of-capsid assay to explore the effects of TRIM5α_hu and Fv-1 expression on the integrity of particulate MLV capsids within restricted cells.

MATERIALS AND METHODS

Vectors, cell lines, and viruses.

All cell lines were maintained in Dulbecco's modified Eagle's medium supplemented with 10% newborn calf serum and antibiotics. All stably transduced cell lines were generated using the pLPCX vector, selected using 3.0 μg/ml puromycin, and maintained in 1.0 μg/ml puromycin. All of the expressed TRIM proteins contain a hemagglutinin (HA) tag at the C terminus, with the exception of mouse 9230105E10, which contains an N-terminal HA tag. The C95A, ΔRING (Δ1-92), ΔB-box 2 (Δ95-127), ΔRING/ΔB-box 2 (Δ1-130), ΔCoiled-coil (Δ130-231) (ΔCC), and RBCC (Δ239-497) variants of TRIM5α_hu were generated in the pLPCX vector using PCR mutagenesis. The Fv-1ⁿ and Fv-1^b proteins are untagged; cDNA constructs were kindly provided by Jonathan P. Stoye. Recombinant MLV expressing green fluorescent protein (MLV-GFP) was prepared by calcium phosphate transfection of 293T cells with the pCIG Gag/Pol vector (5), an MLV vector expressing GFP, and pVPack-VSV-G at a weight ratio of 15:15:4, respectively. The pVPack-VSV-G plasmid expresses the vesicular stomatitis virus (VSV) G envelope glycoprotein, which mediates efficient entry into a variety of vertebrate cells (40). All MLV-GFP vectors were titrated onto 293T cells and normalized for infectivity based on a dilution of virus that yielded 50% infected cells.

A single-round infection assay using MLV-GFP.

MDTF cells (2 × 10⁴) stably transduced with either an empty LPCX vector or LPCX expressing HA-tagged TRIM5 or untagged Fv-1 were seeded onto a 24-well plate and incubated overnight at 37°C. Various doses of normalized N- or B-MLV-GFP were then diluted to 500 μl and added to the cells in the presence of 5.0 μg/ml Polybrene. After 16 h of incubation, the virus-containing medium was replaced with fresh medium and incubation continued for an additional 48 h. GFP-positive cells were then quantified using fluorescence-activated cell sorting (FACS).

Fate-of-capsid assay.

Recombinant MLV virus-like particles (VLPs) were produced by calcium phosphate cotransfection of 293T cells with pCIGN, pCIGB, pCIGNBNN, or pCIGBNBB vector (5), along with pVPack-VSV-G, at a weight ratio of 15:4.

Stably transduced NIH 3T3 (1.5 × 10⁶) or MDTF (1.5 × 10⁶) cells expressing the different TRIM5 or Fv-1 variants were seeded in 80-cm² flasks. The following day, the cells were incubated with 5 to 10 ml (approximately 2.5 × 10⁵ to 5.0 × 10⁵ transcriptase units) of N-, B-, NBNN-, or BNBB-MLV VLPs at 4°C for 30 min to allow viral attachment to the cells. The cells were then shifted to 37°C until they were harvested at various time points. For cells harvested at time points later than 4 h, the virus was removed after 4 h and replaced with fresh medium. The cells were washed three times using ice-cold phosphate-buffered saline (PBS) and detached by treatment with 1.0 ml of pronase (7.0 mg/ml in Dulbecco's modified Eagle's medium) for 5 min at 25°C. The cells were then washed three times with PBS. The cells were resuspended in 2.5 ml hypotonic lysis buffer (10 mM Tris-HCl, pH 8.0, 10 mM KCl, 1 mM EDTA, and one Complete protease inhibitor tablet) and incubated on ice for 15 min. The cells were lysed using 15 strokes in a 7.0-ml Dounce homogenizer with pestle B. Cellular debris was cleared by centrifugation for 3 min at 3,000 rpm. To allow assessment of the total input MLV p30 capsid protein, 100 μl of the cleared lysate was collected, made 1× in sodium dodecyl sulfate (SDS) sample buffer, and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and Western blotting (see below). Then, 2.0 ml of the cleared lysate was layered onto a 50% sucrose (weight per volume) cushion in 1× PBS and centrifuged at 125,000 × g for 2 h at 4°C in a Beckman SW41 rotor. Following centrifugation, 100 μl of the topmost portion of the supernatant was collected and made 1× in SDS sample buffer; this sample is referred to as the soluble p30 fraction. The pellet was resuspended in 50 μl 1× SDS sample buffer and is referred to as the particulate p30 fraction. All samples were then subjected to SDS-PAGE and Western blotting (see below).

Some pilot experiments were conducted using sucrose cushions in which the sucrose concentration ranged from 45 to 70% (weight per volume).

Immunoblotting.

HA-tagged TRIM5 variants were detected using the horseradish peroxidase-conjugated 3F10 antibody. The MLV p30 protein was detected using a rat monoclonal antibody purified from the R187 hybridoma cell line (American Type Culture Collection).

Analysis of the N-MLV genome.

Approximately 1.5 × 10⁶ MDTF cells stably transduced with either an empty LPCX vector or an LPCX vector expressing TRIM5α_hu were seeded in 80-cm² flasks. The following day, the cells were incubated with 5.0 ml (approximately 2.5 × 10⁵ to 5.0 × 10⁵ reverse transcriptase units) of N-MLV-GFP at 4°C for 30 min to allow viral attachment to the cells. The cells were then shifted to 37°C for 2 h and analyzed using the fate-of-capsid assay, as described above. The pellet was resuspended in either 50 μl 1× SDS sample buffer and subjected to SDS-PAGE and Western blotting (see above) or 250 μl RLT RNeasy lysis buffer, and total RNA was purified using an RNeasy Kit (QIAGEN). Reverse transcription (RT) PCR was performed using a QuantiTect SYBR Green RT-PCR kit (QIAGEN) and primers specific to the GFP gene in the vector (forward primer, GACGTAAACGGCCACAAGTT; reverse primer, GGTCTTGTAGTTGCCGTCGT). The reaction mixtures contained Quantitect SYBR Green, 500 nM of each primer, and 5.0 μl of purified RNA. The PCR conditions were 30 min at 50°C and 15 min at 95°C, followed by 40 cycles of 15 seconds at 95°C and 30 seconds at 72°C, on an MJ Research DNA Engine Opticon.

RESULTS

Establishment of an assay to investigate the postentry fate of MLV capsids in infected cells.

We previously utilized the fate-of-capsid assay to demonstrate that a decrease in the level of particulate HIV-1 capsids is observed in the cytosol of cells expressing a restricting TRIM5α protein (34). To investigate the fate of MLV capsids in infected cells, we developed an approach that would allow us to distinguish cytosolic MLV capsids that are potentially on the infection pathway from capsids that are nonspecifically endocytosed and that do not lead to infection of the cell. Because the density of endosomes is less than that of retroviral capsids, endosome-associated capsids can be separated from cytosolic capsids by density gradient centrifugation. To identify the appropriate sucrose concentrations that would allow discrimination between endosomal and cytosolic capsids, we incubated MDTF cells with N-MLV and B-MLV that lacked envelope glycoproteins or that were pseudotyped with the VSV G envelope glycoprotein. Although the viruses lacking envelope glycoproteins can attach to cells and be endocytosed, the capsids of these viruses do not enter the cytosol because virus-host cell membrane fusion does not occur. Figure 1A (left) shows that postnuclear lysates from MDTF cells contain similar levels of the N-MLV capsid protein, p30, following exposure to viruses with and without envelope glycoproteins. Sedimentation through 45 to 55% sucrose solutions allowed separation of particulate capsids from endosomal capsids (Fig. 1A, right). Density gradient sedimentation through solutions of 60% or higher concentrations of sucrose resulted in decreased recovery of cytosolic particulate N-MLV capsids. The density of a 60% (weight per volume) sucrose solution, 1.23 g/ml, corresponds exactly to the reported density of Moloney MLV capsids (1).

FIG. 1.

Assay to follow the fate of the MLV capsid within infected cells. (A) MDTF cells transduced with an empty LPCX control vector were infected with either VSV G-pseudotyped (Env+) N-MLV or envelope-deficient (Env−) N-MLV virus-like particles for 4 h. Cell lysates were then prepared, and an aliquot was used for Western blotting with an antibody directed against the MLV p30 capsid protein (left). The remaining cell lysates were analyzed on gradients formed with the indicated concentrations of sucrose (weight per volume). The pellets were Western blotted using an antibody directed against the MLV p30 capsid protein. (B) MDTF cells transduced with an empty LPCX control vector were infected with either VSV G-pseudotyped (Env+) or envelope-deficient (Env−) N-MLV or B-MLV virus-like particles for 4 h. Cell lysates were then prepared and analyzed on gradients formed with the indicated concentrations of sucrose (weight per volume). The total input and pellets were Western blotted using an antibody directed against the p30 capsid protein. (C) MDTF cells stably transduced with either an empty LPCX control vector or an LPCX vector expressing TRIM5α_hu were infected with either VSV G-pseudotyped (Env+) or envelope-deficient (Env−) N-MLV-GFP for 2 h and analyzed using the fate-of-capsid assay, as described in Materials and Methods. The total input, supernatants, and pellets were Western blotted using an antibody directed against the p30 capsid protein. (D) MDTF cells transduced with either an empty LPCX control vector or an LPCX vector expressing TRIM5α_hu were infected with either VSV G-pseudotyped (Env+) or envelope-deficient (Env−) N-MLV-GFP for 2 h and analyzed by the fate-of-capsid assay. The presence of viral RNA within the pelleted cores was detected by RT-PCR, as described in Materials and Methods. The amount of viral RNA in the pellet associated with the LPCX cells exposed to (Env+) N-MLV-GFP was standardized to 100% and set as a reference for comparison to the RNA levels associated with the LPCX-transduced cells exposed to (Env−) N-MLV-GFP and the TRIM5α_hu-expressing cells exposed to the (Env+) N-MLV-GFP. The error bars indicate standard deviations.

The results obtained with B-MLV differed from those seen for N-MLV. The amounts of MLV p30 capsid proteins in the lysates of MDTF cells exposed to VSV G-pseudotyped B-MLV virions were typically within threefold of those observed in cells exposed to the VSV G-pseudotyped N-MLV virions (Fig. 1B, left). However, unlike the cytosolic N-MLV capsids, the cytosolic B-MLV capsids did not sediment through 55% sucrose cushions (Fig. 1B, right). This result suggests that, under these conditions, cytosolic B-MLV capsids exhibit lower density or stability, or both, than N-MLV capsids. Based on these observations, we employed 50% (weight per volume) sucrose cushions for our subsequent analyses of the fates of both N-MLV and B-MLV capsids after infection.

To investigate the effect of TRIM5α_hu expression in the target cells on the fate of the N-MLV capsid, MDTF cells transduced with the empty LPCX vector or LPCX expressing TRIM5α_hu were incubated with N-MLV particles pseudotyped with VSV G glycoprotein or lacking envelope glycoproteins. Cell lysates cleared of nuclei and large debris were layered onto 50% sucrose cushions and centrifuged. The input lysates from the cells incubated with the virions lacking envelope glycoproteins, as well as with the VSV G-pseudotyped virions, contained detectable p30 capsid proteins (Fig. 1C). Particulate cytosolic capsids were readily detected 2 hours after infection of the control LPCX-transduced cells with VSV G-pseudotyped N-MLV. In comparison, by 2 hours after infection of TRIM5α_hu-expressing cells with VSV G-pseudotyped N-MLV, the amount of cytosolic capsid that sedimented through the 50% sucrose cushion was greatly reduced. The supernatants that did not enter the 50% sucrose cushion were analyzed for the presence of the N-MLV capsid protein. Multiple 100-μl samples were sequentially removed from the top of the supernatants and Western blotted for the p30 capsid protein. Figure 1C shows the results obtained for the uppermost layer of supernatants. The amount of the p30 capsid protein in this supernatant layer from the TRIM5α_hu-expressing cells was greater than that seen for the LPCX-transduced control cells. Similar results were obtained for the additional 100-μl samples taken from sequentially lower layers of the supernatants (data not shown). The endosomal layer that bands immediately above the 50% sucrose cushion demonstrated high levels of p30 capsid protein, as well as the unprocessed p65 Gag precursor protein, for both the LPCX-transduced and TRIM5α_hu-expressing cells (data not shown). This endosomal layer apparently contains immature virion particles, as well as mature virions that have not entered the cytosol. To avoid sampling this endosomal layer, we analyzed only the uppermost fraction of the supernatants in subsequent experiments.

We conclude that, in TRIM5α_hu-expressing cells compared with control cells, a reduction in the level of particulate capsids is accompanied by an increase in the amount of soluble capsid proteins in the cytosol. Note that, because only a fraction of the supernatant is sampled and analyzed, a quantitative conversion of capsid protein from the pellet to the supernatant is not expected. These results are consistent with a model in which TRIM5α_hu-mediated restriction of N-MLV infection is accompanied by an accelerated conversion of particulate capsid complexes to smaller subunits in the cytosol. The major fraction of N-MLV capsid proteins detectable in TRIM5α_hu-expressing cells did not exhibit evident changes in migration on SDS-polyacrylamide gels compared with capsid proteins derived from control cells. Thus, most N-MLV capsid proteins in TRIM5α_hu-expressing cells do not undergo extensive posttranslational modifications (e.g., ubiquitylation or SUMOylation) that differ from those occurring in the control cells.

To document that viral RNA is associated with the particulate capsids detected in the assay described above, the pellets from the sucrose density gradients were used for RNA extraction. The extracted RNA was reverse transcribed and quantified by real-time PCR. The amount of viral RNA detected by this method was proportional to the amount of p30 capsid protein pelleted from the lysates of TRIM5α_hu-expressing and LPCX control cells exposed to envelope-deficient and VSV G-pseudotyped N-MLV virions (compare Fig. 1D with the pellet p30 in C). Thus, the capsids that sediment in this assay are associated with viral RNA.

Association of human TRIM5α restriction with decreases in the amount of particulate N-MLV capsids in the cytosol.

To examine the relationship between TRIM5α_hu-mediated restriction and the levels of particulate N-MLV capsids in the cytosol of infected cells, we applied the fate-of-capsid assay to MDTF and NIH 3T3 cells expressing different TRIM proteins and exposed to N-MLV and B-MLV. MDTF and NIH 3T3 cells expressing TRIM5α_hu, TRIM5α_rh, and the mouse TRIM protein, 9230105E10, were studied (Fig. 2A). The mouse 9230105E10 protein is related to TRIM5 but is not orthologous and does not restrict N-MLV infection. As shown in Fig. 2B, the expression of TRIM5α_hu potently restricted N-MLV in MDTF cells, whereas only a modest level of restriction was observed in cells expressing TRIM5α_rh. The efficiency of N-MLV infection of cells expressing the murine TRIM protein, 9230105E10, was comparable to that of control cells transduced with the empty LPCX vector. B-MLV infection was unaffected by the expression of any of the TRIM variants.

FIG. 2.

Effect of the expression of a restricting TRIMα protein on the amounts of particulate N-MLV capsids in the cytosol of infected cells. (A) Steady-state expression levels of the TRIM proteins are shown. Lysates from MDTF or NIH 3T3 cells transduced with either an empty LPCX vector or LPCX vectors expressing HA-tagged TRIM proteins were subjected to Western blotting using an anti-HA antibody. The lysates were also blotted against β-actin to control for total protein. “Mouse” refers to the mouse 9230105E10 protein, used here as a negative control. (B) MDTF cells transduced with LPCX vectors expressing the indicated TRIM proteins were infected with various doses of N-MLV-GFP or B-MLV-GFP. GFP-positive cells were quantified by FACS analysis. (C) MDTF or NIH 3T3 cells stably transduced with either an empty LPCX control vector or an LPCX vector expressing the indicated TRIM protein were infected with VSV G-pseudotyped N- or B-MLV virus-like particles for 4 h and analyzed using the fate-of-capsid assay, as described in Materials and Methods. Total input, supernatants, and pellets were Western blotted using an antibody directed against the p30 capsid protein. “Mouse” refers to the mouse 9230105E10 protein.

Using the fate-of-capsid assay, we analyzed the amounts of particulate capsids in the cytosol of MDTF and NIH 3T3 cells expressing the various TRIM proteins. As shown in Fig. 2C, expression of the potently restricting TRIM5α_hu protein resulted in a dramatic reduction in the amount of particulate cytosolic N-MLV capsids compared to the amount in the empty LPCX control cells. Importantly, this decrease in particulate N-MLV capsid coincided with an increase in the amount of soluble p30 capsid protein within the supernatant. Only a slight reduction in the amount of particulate N-MLV capsids was observed in the cytoplasm of cells expressing the moderately restricting TRIM5α_rh protein. Expression of the nonrestricting mouse 9230105E10 TRIM protein had no effect on the amount of particulate cytosolic N-MLV capsids recovered compared with the amount in control cells. The amount of particulate B-MLV capsids recovered from the cytosol of the MDTF and NIH 3T3 cells was generally less than that in the N-MLV capsids, despite similar amounts of input capsid. This may be related to differences in the density and/or stability of N-MLV and B-MLV cytosolic capsids, as discussed above. Expression of the TRIM variants had no significant effect on the amounts of particulate B-MLV capsids. In some experiments in which the amount of pelletable B-MLV p30 was low, minor variation in pelletable p30 was observed, but this was not reproducible. We conclude that the expression of TRIM5α_hu results in restriction of N-MLV infection; accompanying this restriction are decreases in the levels of particulate N-MLV capsids and increases in the levels of soluble capsid protein in the cytosol.

Timing of the TRIM5α_hu effect on particulate cytosolic N-MLV capsids.

Studies examining the effects of cyclosporine A treatment have demonstrated that the owl monkey restriction factor, TRIMCyp, mediates HIV-1 restriction within 15 min of virus entry into the cell (22, 27). Furthermore, previous fate-of-capsid studies have demonstrated that TRIM5α_rh promotes the disappearance of particulate cytosolic HIV-1 capsids within 1 h after infection (34). These data suggest that TRIM proteins block retroviral infection shortly after virus entry into the cytoplasm. To determine how fast TRIM5α_hu decreases the levels of particulate cytosolic N-MLV capsids, MDTF cells were incubated with VSV-G-pseudotyped N-MLV at 4°C and then the temperature was raised to 37°C to allow virus entry to occur. At various times after the temperature increase, the fate-of-capsid assay was performed. As shown in Fig. 3, particulate N-MLV capsids were readily detectable in the cytosol of MDTF cells containing an empty LPCX vector within 15 min of the temperature increase. By contrast, the expression of TRIM5α_hu resulted in decreases in the amounts of particulate N-MLV capsids that were evident by 15 min after the temperature increase. TRIM5α_hu expression had no consistent effect on the steady-state levels of particulate B-MLV capsids. We conclude that TRIM5α_hu can promote the disappearance of particulate N-MLV capsids in the target cell cytoplasm within 15 min of infection.

FIG. 3.

Time course of TRIM5α_hu-induced loss of particulate N-MLV capsids in infected cells. MDTF cells transduced with either an empty LPCX vector or an LPCX vector expressing TRIM5α_hu were incubated at 4°C for 30 min with either N-MLV (A) or B-MLV (B) virus-like particles. The cells were then incubated at 37°C for the indicated times and analyzed using the fate-of-capsid assay. Total input, supernatants, and pellets were Western blotted using an antibody directed against the p30 capsid protein, as described in Materials and Methods.

Influence of residue 110 of the MLV capsid on TRIM5α_hu-mediated decreases in the amounts of particulate capsids.

N-MLV susceptibility to TRIM5α_hu restriction is influenced by arginine 110 of the viral capsid (24, 36). Replacement of this arginine with the corresponding residue of B-MLV, glutamic acid, creates NBNN-MLV, which partially escapes TRIM5α_hu restriction (24, 36). Conversely, replacing glutamic acid 110 of B-MLV with the arginine found in the N-MLV capsid generates BNBB-MLV, which is susceptible to TRIM5α_hu restriction (24, 36). To determine if there is a correlation between the residue at position 110 of the MLV capsid and the amount of particulate capsid observed in the cytosol of restricted cells, we infected control or TRIM5α_hu-expressing MDTF or NIH 3T3 cells with N-, B-, NBNN-, and BNBB-MLV and performed the fate-of-capsid assay. As demonstrated in Fig. 4, viruses susceptible to TRIM5α_hu restriction, N- and BNBB-MLV, exhibited levels of particulate capsids in cells expressing TRIM5α_hu significantly lower than those seen in cells transduced with the empty LPCX vector. By contrast, only a modest decrease in the steady-state level of particulate NBNN-MLV capsids was observed in TRIM5α_hu-expressing cells, consistent with the observation that NBNN-MLV can partially escape TRIM5α_hu restriction (24, 36). The expression of TRIM5α_hu in NIH 3T3 or MDTF cells did not result in a relative decrease in the amounts of particulate B-MLV capsids. Of interest, when the input amount of p30 capsid protein is taken into consideration, a generalized decrease in the amounts of particulate capsids was observed in all cell lines infected with B- and NBNN-MLV compared to N- and BNBB-MLV-infected cells. We conclude that the TRIM5α_hu-mediated decrease in the level of particulate MLV capsids in the cytosol is dependent upon the arginine at position 110 of the viral capsid.

FIG. 4.

Influence of MLV capsid residue 110 on the TRIM5α_hu-mediated loss of particulate N-MLV capsids. NIH 3T3 and MDTF cells transduced either with an empty LPCX vector or with a vector expressing TRIM5α_hu were infected with N-, B-, NBNN-, or BNBB-MLV virus-like particles for 4 h and analyzed using the fate-of-capsid assay, as described in Materials and Methods. Total input, supernatants, and pellets were Western blotted using an antibody directed against the p30 capsid protein. The amount of pelleted p30 was quantified and normalized against total input p30. The percentage of particulate p30 in TRIM5α_hu-expressing cells compared to that in the control LPCX cells is provided at the bottom of the figure for each virus.

TRIM5α_hu domains are required to promote decreases in the levels of particulate N-MLV capsids in the cytosol of cells.

Previous studies using deletion mutagenesis have identified the B-box 2, coiled-coil, and B30.2 domains as critical for the antiviral activity of TRIM5α_rh (13, 21). By contrast, deletion of the RING domain only partially reduces TRIM5α_rh antiviral activity (13, 21, 33). We wished to determine which TRIM5α_hu domains are necessary for N-MLV restriction and whether these correlate with the TRIM5α_hu domains required to mediate decreases in the level of particulate N-MLV capsids in the cytosol of infected cells. To that end, we generated a series of TRIM5α_hu deletion mutants and measured their abilities to restrict N- and B-MLV infection. As shown in Fig. 5A, MDTF cells expressing TRIM5α_hu are very resistant to N-MLV infection. Deletion of the RING domain (TRIM5α_hu ΔRING) reduced the N-MLV inhibition approximately 10-fold, suggesting that the RING domain potentiates, but is not absolutely required for, restriction. By contrast, deletion of the B-Box 2 (TRIM5α_hu ΔB-box), RING and B-box 2 (TRIM5α_hu ΔRING/ΔB-box), coiled-coil (TRIM5α_hu ΔCC), or B30.2 (TRIM5α_hu RBCC) domain resulted in TRIM5α_hu proteins that had no effect on N-MLV infection. B-MLV infection was unaffected by the expression of any of the TRIM5α_hu deletion variants. We conclude that the B-box 2, coiled-coil, and B30.2 domains of TRIM5α_hu are essential for its anti-N-MLV activity; the RING domain contributes significantly to the antiviral activity but is not absolutely essential.

FIG. 5.

Effects of deletion of the TRIM5α_hu domains on N-MLV restriction and levels of particulate N-MLV capsids in the cytosol of infected cells. (A) MDTF cells transduced with LPCX vectors expressing wild-type TRIM5α_hu or deletion mutants were infected with various doses of N-MLV-GFP or B-MLV-GFP. GFP-positive cells were quantified using FACS analysis. (B) Steady-state expression levels of TRIM5 variants are shown. Lysates from MDTF cells transduced with either an empty LPCX vector or LPCX vectors expressing the indicated HA-tagged TRIM5α_hu variants were subjected to Western blotting using an anti-HA antibody. The lysates were also blotted against β-actin to control for total protein. (C) MDTF cells transduced with either an empty LPCX vector or vectors expressing the indicated TRIM5α_hu variants were infected with VSV G-pseudotyped (Env+) N-MLV virus-like particles for 4 h and analyzed using the fate-of-capsid assay. Cells transduced with the empty LPCX vector were also incubated with N-MLV virus-like particles lacking envelope glycoproteins (Env−). Total input, supernatants, and pellets were Western blotted using an antibody directed against the MLV p30 capsid protein.

Western blot analysis demonstrated that all of the TRIM5α_hu deletion mutants tested were expressed at least as well as the wild-type TRIM5α_hu protein (Fig. 5B). As has been reported previously for TRIM5α_rh (8, 13, 21), deletion of the RING and/or B-box 2 domain of TRIM5α_hu resulted in increased steady-state levels of expression.

To identify which TRIM5α_hu domains are involved in mediating decreases in the levels of particulate cytosolic N-MLV capsids, we performed the fate-of-capsid assay using MDTF cells expressing the TRIM5α_hu deletion mutants. As observed previously, the expression of wild-type TRIM5α_hu greatly reduced the amount of particulate N-MLV capsids in the cytosol of MDTF cells (Fig. 5C). Deletion of any TRIM5α_hu domain severely compromised the ability of TRIM5α_hu to promote decreases in the level of particulate N-MLV capsids. An N-MLV vector lacking an envelope glycoprotein was included to ensure that endosomal p30 was not contaminating the cytosolic p30 fractions (34). The assay was repeated, using larger amounts (2.0 ml and 10.0 ml) of N-MLV VLPs, with similar results (data not shown). We conclude that deletion of any TRIM5α_hu domain necessary for potent restriction of N-MLV also compromises the ability of TRIM5α_hu to promote a decrease in the level of particulate N-MLV capsids in the cytosol of infected cells.

Deletion of the TRIM5α B-box 2 domain completely abrogates HIV-1 and N-MLV restriction yet has no effect on TRIM5α multimerization or interaction with the viral capsid (13, 18, 21, 34). These observations suggest that the B-box 2 domain may mediate an effector function critical for retroviral restriction. We examined the consequences of a single amino acid change within the TRIM5α_hu B-box 2 domain for N-MLV restriction and cytosolic levels of N-MLV capsid particles. Previously, we identified an alanine substitution within the TRIM5α_rh B-box 2 domain that completely abrogated HIV-1 restriction while maintaining wild-type TRIM5α_rh expression levels and subcellular localization (13). The analogous TRIM5α_hu variant [TRIM5α_hu(C95A)] was generated by substitution of alanine for cysteine 95 in the B-box 2 domain. As shown in Fig. 6A, TRIM5α_hu(C95A) was expressed in MDTF cells at a level at least as great as that of the wild-type TRIM5α_hu protein. Despite this efficient level of expression, TRIM5α_hu(C95A) was completely unable to restrict N-MLV infection (Fig. 6B). Furthermore, the expression of TRIM5α_hu(C95A) had no effect on the amount of particulate cytosolic N-MLV capsids compared with the levels observed in the control cells transduced with the empty LPCX vector (Fig. 6C). By contrast, expression of the wild-type TRIM5α_hu protein resulted in a decrease in the amount of particulate N-MLV capsids and a concomitant increase in the amount of soluble capsid proteins in the cytosol of infected cells. These data suggest that the conserved cysteine at position 95 of the B-box 2 domain is critical for both TRIM5α_hu-mediated processes: restriction of N-MLV infection and the conversion of particulate cytosolic N-MLV capsids into soluble forms of capsid.

FIG. 6.

Effect of a single amino acid change in the TRIM5α_hu B-box 2 domain on N-MLV restriction and the loss of particulate N-MLV capsids. (A) Steady-state expression levels of TRIM5 variants are shown. Lysates from MDTF cells transduced with either an empty LPCX vector or LPCX vectors expressing HA-tagged TRIM5α_hu or TRIM5α_hu(C95A) were subjected to Western blotting using an anti-HA antibody. The lysates were also blotted against β-actin to control for total protein. (B) MDTF cells transduced with either an empty LPCX vector or LPCX vectors expressing TRIM5α_hu or TRIM5α_hu(C95A) were infected with various doses of N-MLV-GFP or B-MLV-GFP. GFP-positive cells were quantified using FACS analysis. (C) MDTF cells expressing either an empty LPCX vector or LPCX vectors expressing TRIM5α_hu or TRIM5α_hu(C95A) were infected with VSV G-pseudotyped (Env+) N-MLV virus-like particles for 4 h and analyzed using the fate-of-capsid assay. The control cells transduced with the empty LPCX vector were also incubated with N-MLV virus-like particles without envelope glycoproteins (Env−). Total input, supernatants, and pellets were Western blotted using an antibody directed against the p30 capsid protein.

Effect of Fv-1 expression on the amount of particulate MLV capsid in the cytosol of infected cells.

Although TRIM5α and Fv-1 restrictions are both influenced by residue 110 of the MLV capsid, they target different steps of the MLV life cycle: TRIM5α restricts N-MLV prior to reverse transcription, whereas Fv-1 restricts N- or B-MLV following the completion of reverse transcription (24, 36). To determine whether Fv-1 restriction modulates the amount of particulate cytosolic MLV capsids, we first generated MDTF cells stably expressing Fv-1ⁿ or Fv-1^b and examined their abilities to support N- or B-MLV infection. As shown in Fig. 7A, the expression of Fv-1ⁿ potently restricted B-MLV infection but had no effect on N-MLV infection. Conversely, the expression of Fv-1^b potently restricted N-MLV infection but had no effect on B-MLV infection.

FIG. 7.

Effect of Fv-1 expression on the amount of particulate MLV capsids recovered in the fate-of-capsid assay. (A) MDTF cells transduced with LPCX vectors expressing either Fv-1ⁿ or Fv-1^b were infected with various doses of N-MLV-GFP or B-MLV-GFP. GFP-positive cells were quantified by FACS analysis. The experiment was repeated with comparable results; the data from a single experiment are shown. (B) MDTF cells expressing either an empty LPCX vector or LPCX vectors expressing TRIM5α_hu, Fv-1ⁿ, or Fv-1^b were incubated with VSV G-pseudotyped N-MLV or B-MLV virus-like particles for 4 h. The virus-containing medium was then replaced with fresh medium, and the cells were lysed 16 h later. Cell lysates were analyzed using the fate-of-capsid assay. Total input, supernatants, and pellets were Western blotted using an antibody directed against the p30 capsid protein.

The potency of Fv-1 restriction in MDTF cells allows examination of the effect of Fv-1 restriction on the level of particulate MLV capsids following infection. As shown in Fig. 7B, equivalent levels of particulate N- or B-MLV capsids were recovered after 16 h from the cytosol of MDTF cells transduced with an empty LPCX vector or vectors expressing Fv-1ⁿ or Fv-1^b. By contrast, expression of TRIM5α_hu in the MDTF cells resulted in a complete loss of particulate capsids from the cytosol, consistent with the results described above. As observed previously, compared with the amount of particulate N-MLV capsids, significantly less particulate B-MLV capsid protein was recovered after centrifugation of cytosolic lysates. Similar results were also obtained 4 h postinfection (data not shown). These results suggest that the expression of a restricting Fv-1 protein in MDTF cells exerts little or no effect on the amounts of particulate MLV capsids in the cytosol of infected cells.

DISCUSSION

In this study, we utilized a recently developed fate-of-capsid assay (24, 36) to demonstrate that TRIM5α_hu expression rapidly decreases the steady-state levels of particulate N-MLV capsids in the cytosol of the infected cell. The study of TRIM proteins from different species, TRIM5α_hu mutants, and MLV capsid variants revealed a close correlation between the restriction of virus infection and decreases in the amounts of particulate viral capsids in the cytosol. Hypothetically, two models might account for the TRIM5α_hu-associated decreases in the amounts of particulate N-MLV capsids observed in the fate-of-capsid assay: (i) TRIM5α_hu degrades the N-MLV capsid or (ii) TRIM5α_hu promotes the rapid disassembly of the N-MLV capsid. Several lines of evidence suggest that neither ubiquitin modification nor degradation of the retroviral capsid is an essential element of the TRIM5α restriction mechanism. First, treatment of TRIM5α_rh-expressing cells with proteasomal inhibitors does not relieve HIV-1 restriction (5a, 22, 34, 37). Moreover, N-MLV infection was efficiently blocked by TRIM5α_hu in a Chinese hamster E36 cell line expressing a temperature-sensitive E1 ubiquitin ligase, even at the nonpermissive temperature (22). In our study and a previous study (34), no modified forms or cleavage products of the N-MLV p30 capsid protein were observed specifically in TRIM5α_hu-expressing cells. Most importantly, in TRIM5α_hu-expressing cells compared with control cells, an increase in soluble p30 typically accompanied the observed decrease in particulate p30, suggesting that TRIM5α_hu promotes the conversion of the particulate viral core into slower-sedimenting forms of p30. Taken together, these data argue that ubiquitin ligation, proteasome degradation, and/or massive degradation of the viral core are not essential components of TRIM5α restriction.

Our data support a model in which TRIM5α_hu mediates N-MLV restriction by promoting the rapid disassembly of the viral capsid. Previous studies have hinted at the existence of an optimal stability of the retroviral core for HIV-1 infection (9, 29). Changes in the HIV-1 capsid protein that alter the stability of the viral core in either a positive or negative direction have deleterious effects on the efficiencies of early infection events (9). These data suggest that the retroviral core is a metastable structure that is stable enough to allow reverse transcription yet sufficiently labile to facilitate the uncoating necessary for subsequent steps in the viral life cycle. TRIM5α_hu may accelerate the normal uncoating process or promote aberrant disassembly of the N-MLV core. Either of these processes might result in aborted reverse transcription and infection. Premature disassembly of the retroviral capsid may expose other viral proteins and the genome to ubiquitin-dependent or independent degradation pathways. Wu et al. recently reported that treatment of TRIM5α_rh-expressing cells with proteasome inhibitors allowed HIV-1 to complete reverse transcription, even though the cells remained restricted for HIV-1 infection (37). If the viral cDNA detected in their study is on the infection pathway, the result suggests that the TRIM5α_rh restriction mechanism does not necessarily rely upon the degradation of the viral genome.

Interestingly, particulate B-MLV capsids were inefficiently recovered from the cellular cytosol by our fate-of-capsid assay. Moreover, the efficiency with which we recovered particulate MLV capsids was influenced by the identity of residue 110 of the MLV capsid (Fig. 4). One explanation for these observations is that N- and B-MLV have evolved relatively stable or unstable capsids, respectively. The intrinsic stability of the MLV capsid could govern viral susceptibility to intracellular restriction factors, possibly explaining why N-MLV and not B-MLV is targeted by a diverse group of TRIM molecules (3, 31, 32, 36). For example, a virus that has evolved a relatively stable capsid might be more susceptible to factors that promote rapid capsid destabilization than a virus that normally tolerates a less stable capsid. Alternatively, stable N-MLV capsids could provide a more suitable recognition surface for TRIM molecules. TRIM5α_hu has been reported to bind N-MLV, but not B-MLV, capsids (28). Furthermore, Shi and Aiken demonstrated that only stable HIV-1 virus-like particles are able to compete for TRIM5α (29). Destabilization of the HIV-1 core by mutagenesis abrogates the ability of these virus-like particles to saturate TRIM5α restriction, suggesting that the stability of the retroviral core can govern the efficiency of interaction with TRIM5α.

TRIM5α_rh-mediated restriction of HIV-1 infection involves multiple domains: the B30.2 domain specifies recognition of the incoming viral capsids, the coiled coil promotes the multimerization of TRIM5α molecules, and an effector region (RING and B-box 2 domains), although not needed for viral capsid binding, contributes to full restriction. Our deletion analysis of the TRIM5α_hu domains demonstrated that the B-box 2, coiled-coil, and B30.2 domains are essential for mediating N-MLV restriction. Furthermore, for this set of TRIM5α deletion mutants, we observed a correlation between restriction of infection and the ability to promote decreases in the levels of particulate N-MLV capsids in the cytosol of infected cells. Our assay for restriction of N-MLV infections may be more sensitive than the fate-of-capsid assay, based upon the observation that deletion of the TRIM5α_hu RING domain abrogated activity in the latter assay but only partially attenuated virus restriction.

A recent study has demonstrated that the overexpression of the Fv-1^b allele in human TE671 cells can compete with endogenous TRIM5α_hu (20). Indeed, the expression of Fv-1^b allows N-MLV to overcome TRIM5α_hu restriction and become blocked at a post-reverse transcription step, characteristic of Fv-1 restriction (20). These data suggest that both TRIM5α_hu and Fv-1 interact with the MLV capsid shortly after viral entry but restrict MLV infection either before or after reverse transcription, respectively. In this study, we demonstrated that the expression of TRIM5α_hu causes a rapid decrease in particulate N-MLV capsids within the cytoplasm as early as 15 min postinfection. Conversely, no change in the level of particulate MLV capsids was observed in the cytosol of Fv-1ⁿ- or Fv-1^b-expressing cells at either 4 or 16 h after infection. These data suggest that Fv-1 and TRIM5α_hu restrict MLV infection through two distinct mechanisms.

Further studies examining early postentry events and how these events are modulated by host factors will expand our knowledge of retroviral biology and the mechanism of TRIM5α restriction.