|Home | About | Journals | Submit | Contact Us | Français|
The mammalian tripartite motif protein, TRIM5α, recognizes retroviral capsids entering the cytoplasm and blocks virus infection. Depending on the particular TRIM5α protein and retrovirus, complete disruption of the TRIM5α RING domain decreases virus-restricting activity to various degrees. TRIM5α exhibits RING domain-dependent E3 ubiquitin ligase activity, but the specific role of this activity in viral restriction is unknown. We created a panel of African green monkey TRIM5α (TRIM5αAGM) mutants, many of which are specifically altered in RING domain E3 ubiquitin ligase function, and characterized the phenotypes of these mutants with respect to restriction of simian and human immunodeficiency viruses (SIVmac and HIV-1, respectively). TRIM5αAGM ubiquitin ligase activity was essential for both the accelerated disassembly of SIVmac capsids and the disruption of reverse transcription. The levels of SIVmac particulate capsids in the cytosol of target cells expressing the TRIM5α variants strongly correlated with the levels of viral late reverse transcripts. RING-mediated ubiquitylation and B30.2(SPRY) domain-determined capsid binding independently contributed to the potency of SIVmac restriction by TRIM5αAGM. In contrast, TRIM5α proteins attenuated in RING ubiquitin ligase function still accelerated HIV-1 capsid disassembly, inhibited reverse transcription, and blocked infection. Replacement of the helix-4/5 loop in the SIVmac capsid with the corresponding region of the HIV-1 capsid diminished the dependence of restriction on TRIM5α RING function. Thus, ubiquitylation mediated by the RING domain of TRIM5αAGM is essential for blocking SIVmac infection at the stage of capsid uncoating.
Following entry into the cells of certain mammalian species, some retroviruses encounter blocks prior to reverse transcription (5, 22, 25, 52, 71, 79). For example, human immunodeficiency virus type 1 (HIV-1) infection is blocked at this stage in the cells of Old World monkeys (25, 52, 71). The simian immunodeficiency virus of macaques (SIVmac) encounters similar early blocks in the cells of New World monkeys and African green monkeys (25). These early postentry blocks are mediated by TRIM5α, and interspecies variation in TRIM5α accounts for species-specific patterns of viral restriction (23, 32, 62, 73, 75, 88). Differences in susceptibility to TRIM5α are determined by the retroviral capsid protein (8, 58).
TRIM5α is a member of the tripartite motif (TRIM) family of proteins, which contain RING, B-box, and coiled-coil (RBCC) domains (50, 65). The α isoform of TRIM5 also has a carboxy-terminal B30.2(SPRY) domain, which mediates specific capsid recognition and is important for antiretroviral activity (10, 41, 53, 57, 60, 63, 67, 74, 76, 89). The coiled-coil and B-box2 domains contribute to the avidity of TRIM5α for the retroviral capsid by mediating the dimerization and higher-order association, respectively, of TRIM5α proteins (11, 27, 31, 38, 43, 44, 51). Some primate species express a TRIMCyp protein, in which the RBCC domains of TRIM5 are fused to cyclophilin A (6, 7, 45, 54, 56, 68, 81, 83). TRIMCyp proteins block the infection of a subset of retroviruses whose capsid proteins bind cyclophilin A (6, 7, 12, 37, 45, 46, 54, 56, 68, 69, 80, 81, 83). In cells expressing potently restricting TRIM5α or TRIMCyp proteins, particulate retroviral capsids are converted to soluble capsid proteins more rapidly than in control cells (10, 11, 13, 28, 64, 77). Thus, potent TRIM5α and TRIMCyp proteins promote premature, accelerated capsid disassembly.
The zinc-binding RING finger domain is the signature of a class of E3 ubiquitin ligases involved in proteasome-mediated protein degradation and other aspects of protein regulation (3, 18, 26, 50). Ubiquitin (Ub) ligation has been shown to be important for the ability of some TRIM proteins to mediate their effects on a target protein; in some cases, this leads to the proteasome-dependent turnover of the targeted protein, but in other cases, modulation of protein function or even stabilization of the protein results (19, 30, 39, 50, 55). TRIM5α possesses E3 ubiquitin ligase activity and can ubiquitylate itself (87). Deletion of the TRIM5α/TRIMCyp RING domain or the disruption of RING domain folding by alteration of the zinc-binding cysteine residues results in various effects on retroviral inhibition, depending on the TRIM5 protein and the restricted virus (12, 29, 48, 60, 75). The SIVmac-restricting activity of African green monkey TRIM5α (TRIM5αAGM) is severely attenuated by alteration of RING domain cysteines (48). However, because RING domain changes can affect TRIM5α steady-state levels, intracellular localization, higher-order self-association, or capsid binding (14, 29, 34; X. Li and J. Sodroski, unpublished observations), and because alteration of RING domain cysteines likely disrupts multiple functions, mechanistic interpretation of this observation is complicated. Nonetheless, these cysteine changes in the TRIM5αAGM RING domain or complete deletion of the rhesus macaque TRIM5α (TRIM5αrh) RING domain only partially attenuated HIV-1-inhibiting activity (29, 60). Moreover, TRIM5αAGM and TRIM5αrh efficiently restrict HIV-1 infection in the presence of proteasome inhibitors (2, 61, 66, 71, 84). Furthermore, owl monkey TRIMCyp blocks HIV-1 infection and human TRIM5α blocks N-tropic murine leukemia virus (N-MLV) infection in cells expressing a temperature-sensitive ubiquitin-activating (E1) enzyme, even at the nonpermissive temperature (61). Thus, while the integrity of the RING domain is important for TRIM5α-mediated restriction of some retroviruses (48), the precise mechanistic role of the RING domain, ubiquitylation, or the proteasome in retroviral restriction is still uncertain.
Here, we investigate the basis for the dramatic difference in SIVmac restriction mediated by the TRIM5α proteins from two African green monkey subspecies, both of which restrict HIV-1 infection. A RING domain alteration was found to be responsible for the difference in SIVmac restriction potency. Subsequent study of SIVmac restriction by a panel of RING mutants implicated E3 ubiquitin ligase activity in TRIM5αAGM-mediated premature disassembly of the SIVmac capsid and disruption of reverse transcription. TRIM5α RING-mediated ubiquitin ligase function and B30.2(SPRY) domain-mediated capsid binding independently contributed to SIVmac restriction. In contrast, the E3 ubiquitin ligase activity of the RING domain exerted little impact on the effect of TRIM5αAGM on HIV-1 capsid disassembly, reverse transcription, or infection. Replacement of the helix-4/5 loop of the SIVmac capsid with the corresponding region from the HIV-1 capsid diminished the requirement for TRIM5αAGM ubiquitin ligase activity in the restriction process.
HeLa, HEK293T, and Cf2Th cells were obtained from the American Type Culture Collection and maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics. To generate stable cell lines expressing TRIM5α variants, recombinant retrovirus vectors were produced, using the pLPCX vector plasmid and a packaging system (pVPack-GP and pVPack-VSV-G; Stratagene), as described previously (75). The retroviral vectors were used to transduce either HeLa or Cf2Th cells, followed by selection in either 1 μg/ml or 5 μg/ml puromycin, respectively. Once the cell lines were generated, the level of TRIM5α expressed stably in the cells was determined by Western blotting with an antibody (Roche) directed against the hemagglutinin (HA) epitope tag on the TRIM5α proteins (see below).
The TRIM5 genes from the tantalus and pygerythrus subspecies of African green monkeys were originally cloned by PCR amplification from COS-1 and Vero cells, respectively, as previously reported (73). All TRIM5 genes were cloned by using the EcoRI and ClaI sites in pLPCX; the encoded TRIM5α proteins have an HA tag at their C termini. All mutations were introduced by using overlap PCR and were confirmed by sequencing at the Dana-Farber Cancer Institute (DFCI) Molecular Biology Core Facility. The chimeras between TRIM5αAGM(Tan) and TRIM5αAGM(Pyg) or between TRIM5αAGM(Tan), TRIM5αAGM(Pyg), and TRIM5αrh were made by swapping the corresponding gene fragments generated by combinations of EcoRI, BstXI, SphI, BamHI, and ClaI.
SIVmac(HIV 4/5) contains an SIVmac capsid with the helix-4/5 loop replaced by that of HIV-1. To generate the SIVmac(HIV 4/5) proviral clone, a PstI-SpeI fragment from the HIV-1 gag-pol gene was used to replace the corresponding fragment of the SIVmac-ΔNefΔEnv-GFP plasmid.
Recombinant green fluorescent protein (GFP)-expressing viruses [SIVmac-GFP, HIV-1-GFP, and SIVmac(HIV-1 H4/5)-GFP] pseudotyped with the vesicular stomatitis virus (VSV) G glycoprotein were generated by transient transfection of 293T cells, using either the calcium phosphate method or Lipofectamine 2000, as described previously (25, 73, 75). For the single-round infection assay, virus stocks were serially diluted 2-fold and used to infect 24-well plates seeded with 2 × 104 cells the day before. After 48 to 60 h of incubation at 37°C, each well was washed with phosphate-buffered saline (PBS), and the cells were detached and fixed with 3.7% paraformaldehyde. The percentage of GFP-positive cells was analyzed by fluorescence-activated cell sorting (FACS).
The relative levels of SIVmac infection in cells expressing different TRIM5α variants were calculated using the data from three independent single-round infections. In making the comparison, we used the lowest dose of virus that yielded >99% GFP positivity in the control target cells that had been transduced with the empty LPCX vector. At this virus dose, the percentage of GFP-positive cells observed in the target cells expressing a particular TRIM5α variant was divided by the percentage of GFP-positive cells observed in the control LPCX-transduced cells to derive the relative SIVmac infection level.
TRIM5α-expressing Cf2Th cell lines were seeded in 6-well plates a day in advance. Cells were lysed with MPER lysis buffer (Pierce) containing 1× protease inhibitor cocktail (Roche). After centrifugation at 15,000 × g for 10 min at 4°C, the soluble fractions were retrieved, and a Bradford assay (Bio-Rad) was used to measure the concentration of total protein in the lysates.
For cross-linking TRIM5α, the same amount of protein from each lysate was mixed with increasing amounts (0, 0.5, and 2 mM) of ethylene glycol bis-[sulfosuccinimidylsuccinate] (sulfo-EGS) (Pierce) and incubated at room temperature for 30 min. The reactions were stopped with 1 M Tris-HCl, pH 8.0. The cross-linked TRIM5α proteins were analyzed by Western blotting with an anti-HA antibody.
To measure the half-lives of TRIM5α variants, TRIM5α-expressing Cf2Th cells were incubated in medium containing cycloheximide at a final concentration of 100 μg/ml. Cells were harvested at regular intervals, washed with PBS, and kept on ice until the final batch of cells was harvested. The cells were then lysed in MPER lysis buffer, as described above. The same amount of protein from each lysate was analyzed by SDS-PAGE and Western blotting to detect the HA-tagged TRIM5α protein.
The HIV-1 and SIVmac capsid-nucleocapsid (CA-NC) fusion proteins were purified from IPTG (isopropyl-β-d-thiogalactopyranoside)-induced Escherichia coli as previously described (20). The assay measuring the binding of TRIM5α variants to the assembled HIV-1 CA-NC complexes was performed as previously described (77). To assemble SIVmac CA-NC complexes, 0.6 mM purified protein was mixed with 16 μM (TG)50 DNA oligonucleotide in 50 mM Tris-HCl, pH 7.0, and 1 M NaCl and incubated at 37°C for more than 1 day. The resulting SIVmac CA-NC complexes were negatively stained and examined under the electron microscope; these studies confirmed that the diameters and lengths of the cylindrical CA-NC complexes are similar to those of previously reported HIV-1 CA-NC complexes (20) (see Fig. S3 in the supplemental material).
Approximately 2 × 106 HEK293T cells were transfected to express the different TRIM5α proteins transiently and then harvested 24 h later. The cells were lysed in 0.2 ml of hypotonic lysis buffer (10 mM Tris-HCl, pH 7.0, 10 mM KCl, and 1 mM EDTA), and a concentrated NaCl solution was added to achieve a final concentration of 200 mM. After a brief centrifugation at 4°C to remove large cell debris, the supernatant was spun at 110,000 × g for 1 h at 4°C in a Beckman ultracentrifuge. After this preclearing spin, the retrieved supernatant was brought up to a final volume of 200 μl, with the supernatant prepared in parallel from cells transfected with the empty pLPCX plasmid (if necessary). To these mixtures, the same amount of SIVmac CA-NC complexes was added and incubated at 30°C for 1 h. After incubation, 20 μl of the mixture was saved for further analysis (“input”). The rest of the mixture was layered onto a 60% (wt/vol) sucrose cushion (prepared in 1× PBS) and centrifuged at 110,000 × g for 1 h at 4°C in a swinging-bucket rotor. The pellet was resuspended in SDS sample buffer and subjected to SDS-PAGE. The gels were Western blotted for the bound TRIM5α proteins with an anti-HA antibody and stained with Coomassie brilliant blue to visualize the SIVmac CA-NC proteins. Control experiments indicated that, in the absence of added SIVmac CA-NC complexes, no TRIM5α proteins were detected in the pellets (data not shown).
The E3 ubiquitin ligase activity of a TRIM5α variant was assessed by measuring self-ubiquitylation in vitro. Either transfected HEK293T cells transiently expressing TRIM5α or Cf2Th cells stably expressing TRIM5α were used in this assay. Cells were harvested, washed with PBS, and then lysed in either RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and 1× protease inhibitor cocktail) or NP-40 buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% NP-40, and 1× protease inhibitor cocktail). After cell debris was removed by brief centrifugation, 5 μg of anti-HA antibody (Sigma) was added to the soluble fraction and incubated at 4°C for 1 h or overnight. Magnetic protein G beads (Invitrogen), preequilibrated with the lysis buffer, were used to precipitate the complex of TRIM5α and anti-HA antibody, washed twice with the same lysis buffer and once with HEPES buffer (50 mM HEPES, pH 8.0, 200 mM NaCl, and 0.5 mM dithiothreitol [DTT]), and resuspended in 10 μl HEPES buffer. The protein G-bead suspension was mixed with 0.44 μg E1, 0.34 μg UbcH5a, and 20 nM myc-ubiquitin (final concentration) in either the presence or absence of the Energy regeneration system (1× Energy regeneration solution and 1 mM ATP, final concentration) in a reaction volume of 20 μl. UbcH5a was used as the E2 enzyme for all in vitro ubiquitylation assays, except when the different E2 enzymes were screened in the experiments shown in Fig. S2C in the supplemental material. All of the components were purchased from Boston Biochem. The mixture was incubated on a 37°C shaker for different lengths of time. The reactions were stopped by adding SDS sample buffer, and the proteins were resolved on SDS-polyacrylamide gels. The unmodified, monoubiquitylated, and polyubiquitylated TRIM5α proteins were detected using an anti-HA antibody. Control experiments demonstrated that the ratio of polyubiquitylated TRIM5α protein to the unmodified form was a reliable indicator of TRIM5 RING E3 ubiquitin ligase activity (data not shown). The anti-myc Western blot was used to verify that the in vitro ubiquitylation reaction was successful. Because the anti-myc Western blot detects the ubiquitin-charged E1 enzyme and other potential cellular contaminants, the anti-HA Western blot was used for quantification of the E3 ubiquitin ligase activity of the TRIM5α protein of interest. In two or three independent experiments, the amounts of unmodified and polyubiquitylated TRIM5α proteins were measured using a densitometer. After subtracting the background polyubiquitylated TRIM5α signal observed in the absence of the added Energy regeneration mixture, the ratio of polyubiquitylated to unmodified TRIM5α protein for each TRIM5α variant was normalized to that observed for wild-type TRIM5αAGM(Tan). The relative E3 ubiquitin ligase activity is expressed as a percentage of the activity of wild-type TRIM5αAGM(Tan), which is set at 100%.
One day prior to infection, 5 × 105 Cf2Th cells were seeded in 6-well plates. The next day, the cell medium was replaced by 1 ml fresh medium containing recombinant viruses, prepared as described above. Each virus stock used for measurement of viral cDNA production was quantitated by an in vitro reverse transcriptase (RT) assay described previously (75), and 5 × 104 cpm of RT activity was used for infection. The virus stocks were all pretreated with 60 units of turbo-DNase I (Ambion) at 37°C for 2 h prior to incubation with cells. As a negative control, cells were treated with 25 μM azidothymidine (AZT) for 1 h at 37°C prior to and during infection. For measurement of the early reverse transcripts (minus-strand strong-stop cDNA), cells were incubated with virus for 3 h; for measurement of the late reverse transcripts, a 10- to 12-hour incubation was used. After 37°C incubation in cell culture media with viruses, total DNA was extracted with the DNeasy kit (Qiagen). One-hundred nanograms of total DNA was used as a template for the PCRs, which were conducted in a final volume of 50 μl under the following conditions: 95°C for 3 min, followed by 32 cycles of 95°C for 30 s, 55°C for 30 s, and 68°C for 30 s. The reaction products were analyzed on a 1.5% agarose gel. The primer pairs were as follows: for minus-strand strong-stop cDNA, forward, 5′-AGTCGCTCTGCGGAGAGGCTG-3′, and reverse, 5′-TGCTAGGGATTTTCCTGCTTCGGTTT-3′; for late reverse transcripts, forward, 5′-CACTAGCAGGTAGAGCCTGGGTGT-3′, and reverse, 5′-GTCATCCCACTGGGAAGTTTGAGC-3′ (modified from reference 21). The band intensities were measured with Fluor-Chem FC2 (Alpha Innotech). The relative level of late reverse transcripts was expressed as a percentage of the level observed in control cells transduced with the empty LPCX vector.
The Cf2Th cells expressing different TRIM5α variants were seeded in T75 flasks. On the following day, the flask was incubated at 4°C for 15 min, and the cell medium was replaced by 6 ml cold cell culture medium containing 5 × 106 RT cpm units of recombinant, VSV G-pseudotyped viruses. In some experiments, viruses prepared without the VSV G glycoprotein were studied as controls. After incubation at 4°C for 20 min, the cell-virus mixture was shifted to 37°C. After incubation at 37°C for 6 to 10 h, the cells were washed three times with cold PBS and then treated with pronase (7 mg/ml in Dulbecco's modified Eagle's medium [DMEM]; Roche) for 5 min on ice. The cells were harvested, washed three times with cold PBS, resuspended in 250 μl hypolysis buffer (10 mM Tris-Cl, pH 8.0, 10 mM KCl2, 1 mM EDTA), and lysed with a motorized pestle (Sigma) for 1 min. After centrifugation at 1,500 × g for 3 min, the supernatant was transferred to a new prechilled tube and mixed with 250 μl cold hypolysis buffer. Fifty microliters of this mixture was saved as “input” for analysis by Western blotting. The remaining mixture was adjusted to 300 mM NaCl, incubated at room temperature for 15 min, and then loaded onto a 50% (wt/vol) sucrose cushion in PBS. The particulate viral capsids were pelleted by centrifugation at 30,000 rpm for 2 h in a Beckman SW41 rotor at 4°C. From the top of each tube, 100 μl of supernatant was saved for analysis, and the pellet was resuspended in 60 μl 1× sample buffer. Equal volumes of the input, supernatant, and pellet samples were analyzed by Western blotting with an anti-SIV serum (NIH AIDS Research and Reference Reagent Program) and peroxidase-conjugated anti-human antibody (Pierce). Western blots were exposed to film for time periods that allowed detection of signals but avoided saturation of the film. The intensity of each p27 capsid band was measured by densitometry. The relative amount of pelletable capsid was expressed as a percentage of that seen in the control target cells transduced with the empty LPCX vector.
Correlations between measurements of relative TRIM5α E3 ubiquitin ligase activity, the levels of late reverse transcription products, the amount of pelletable capsid, and the level of SIVmac infection were tested using Prism software (GraphPad Software, Inc.). Spearman rank correlation analysis and linear regression were used to obtain the Spearman rank correlation coefficient (rS) and the P value.
The TRIM5α variants included in each analysis were as follows: (i) for the relative level of SIVmac infection versus E3 ubiquitin ligase activity, Tan, Pyg, E24N, E24D, E24Q, P29G, P29F, L19A, E20A, and L26A; (ii) for late reverse transcripts versus E3 ubiquitin ligase activity, Tan, Pyg, E24N, E24D, E24Q, P29G, P29F, L19A, E20A, L26A, RtVp, RpVp, RtVt, and RpVt; and (iii) for pelletable capsid versus late reverse transcripts and for E3 ubiquitin ligase activity versus pelletable capsids, Tan, Pyg, E24Q, P29F, L19A, E20A, L26A, RtVp, RpVp, RtVt, and RpVt.
The TRIM5α proteins, TRIM5αAGM(Tan) and TRIM5αAGM(Pyg), from the tantalus and pygerythrus subspecies of African green monkeys, respectively, both restrict HIV-1 infection potently (10). However, only TRIM5αAGM(Tan) efficiently inhibits SIVmac infection. There are eight amino acid differences between TRIM5αAGM(Tan) and TRIM5αAGM(Pyg); five of these involve amino acid residues in the B30.2(SPRY) domain (Fig. 1A). Chimeras between these TRIM5αAGM proteins were created and tested for SIVmac-restricting ability (Fig. 1A to C). These studies revealed that the TRIM5αAGM(Tan) B30.2(SPRY) domain did not determine the potency of TRIM5αAGM(Tan) in restricting SIVmac. Therefore, a subset of the 3 amino acid residues in the RING, coiled-coil, and linker 2 regions that differ between TRIM5αAGM(Tan) and TRIM5αAGM(Pyg) must contribute to SIVmac restriction potency. Each of these residues was individually changed in both TRIM5αAGM proteins to the amino acid found in the other TRIM5αAGM protein. HeLa cells stably expressing these TRIM5αAGM variants were challenged with SIVmac and HIV-1 (Fig. 1D; see Fig. S1B in the supplemental material). The results indicated that phenylalanine 34 in the RING domain is necessary and sufficient to account for the potent SIVmac-restricting ability of TRIM5αAGM(Tan). Leucine 34 in TRIM5αAGM(Pyg), which is less commonly found in TRIM5 or its close relatives (74), compromises the ability of the protein to restrict SIVmac infection. The parental TRIM5αAGM(Pyg) protein and all the mutants containing leucine 34 were efficiently expressed (Fig. 1B) and potently blocked HIV-1 infection (Fig. 1D; see Fig. S1B in the supplemental material). We conclude that the TRIM5αAGM(Pyg) protein fails to restrict SIVmac infection because of the identity of RING domain residue 34.
To assess whether leucine 34 influences the global conformation of the TRIM5αAGM protein, other known properties of TRIM5 proteins were examined. The parental TRIM5αAGM proteins and residue 34 mutants were similar with respect to dimerization efficiency and binding to HIV-1 CA-NC complexes (Fig. 1E and F). TRIM5αAGM(Pyg) exhibited a lower turnover rate than TRIM5αAGM(Tan), but leucine 34 in the RING domain contributed only slightly to this difference (Fig. 1G). The ability to dimerize, bind HIV-1 capsids, and restrict HIV-1 infection supports the structural and functional integrity of these TRIM5αAGM variants.
The structure of the human TRIM5 RING domain has been solved (1); like other RING/PHD-like/U-box domains, the TRIM5 RING domain coordinates two zinc ions by virtue of a cross-brace motif (1, 85, 94, 95). Two loops (loop 1 and loop 2) that flank the central α-helix contribute to the putative site of interaction with the E2 ubiquitin-conjugating enzyme (1, 85, 94, 95). Residue 34 is buried beneath the interface of these loops, and changes in residue 34 could potentially alter the conformation of either loop. Although the RING domain amino acid residues predicted to contact E2 are invariant in TRIM5 proteins, there are species-specific differences among the loop 1 and loop 2 sequences; loop 2 has more divergent sequences than loop 1 (Fig. 2A) (74). To investigate the possible functional implications of these differences, the loop 1 and loop 2 sequences of TRIM5αAGM(Tan) were replaced by those of TRIM5α proteins from other primates. Replacement of the TRIM5αAGM(Tan) loop 1 with those of TRIM5α from humans, rhesus macaques, or squirrel monkeys resulted in severe decreases in SIVmac restriction, despite efficient expression of all three mutant proteins (Fig. 2B; see Fig. S1A in the supplemental material). These mutant proteins all potently restricted HIV-1 infection (see Fig. S1C in the supplemental material). In contrast to the phenotypes observed for loop1-substituted TRIM5αAGM(Tan) proteins, TRIM5αAGM(Tan) with the human TRIM5 loop 2 inhibited SIVmac infection as efficiently as the parental TRIM5αAGM(Tan) protein (Fig. 2C). Apparently, restriction of SIVmac infection by TRIM5αAGM(Tan) is sensitive to changes in loop 1 of the RING domain.
The loop 1 regions of human and AGM TRIM5α RING domains differ in sequence only at residues 24 and 29 (Fig. 2A), both of which are surface exposed in the TRIM5 RING structure (1). Each of these residues in TRIM5αAGM(Tan) was altered to multiple other amino acids. Cells expressing these mutant proteins were challenged with SIVmac. The SIVmac-restricting ability of TRIM5αAGM(Tan) tolerated conservative changes in glutamic acid 24 but was completely eliminated by less conservative substitutions at this position (Fig. 2D). None of the TRIM5αAGM(Tan) proteins with changes in proline 29 restricted SIVmac infection (Fig. 2E). The sensitivity of TRIM5αAGM proline 29 to alteration may result from its proximity to the zinc-coordinating cysteine 30. These results support the importance of loop 1 of the RING domain to the SIVmac-restricting ability of TRIM5αAGM(Tan).
Amino acid residues in loop 1 that are conserved in the TRIM5 proteins of different species are predicted to contribute to binding the E2 ubiquitin-conjugating enzyme, based on analogy with other RING/PHD-like/U-box domains (1, 85, 93, 94). To investigate the potential importance of E2 interaction for TRIM5αAGM(Tan) function, two loop 1 residues, leucine 19 and glutamic acid 20, which are predicted to reside within or near, respectively, the E2-binding site (1, 84, 85, 93), were altered to alanine. As a control, leucine 26, which is in loop 1 but is not predicted to reside near the E2-binding site (1, 85, 94, 95), was also changed to alanine. Cells expressing the wild-type and mutant TRIM5αAGM(Tan) proteins, as well as TRIM5αAGM(Pyg), were challenged with SIVmac and HIV-1. The TRIM5αAGM(Tan) L19A and E20A proteins exhibited minimal anti-SIVmac activity (Fig. 2F and Table 1). The TRIM5αAGM(Tan) L26A mutant was nearly as effective as the wild-type TRIM5αAGM(Tan) protein in restricting SIVmac infection (Fig. 2F). However, all three loop 1 TRIM5αAGM(Tan) mutants efficiently restricted HIV-1 infection (see Fig. S1C in the supplemental material). Apparently, specific residues in loop 1 of the TRIM5αAGM(Tan) RING domain are more critical for restriction of SIVmac than for restriction of HIV-1.
Purified rhesus monkey TRIM5α protein has been shown to exhibit E3 ubiquitin ligase activity in vitro, resulting in self-ubiquitylation (58). We established an in vitro assay to examine the E3 ubiquitin ligase activities of TRIM5α variants. In this assay, the transiently expressed HA-tagged TRIM5α protein was immunoprecipitated by an anti-HA antibody and subsequently mixed with purified recombinant E1, E2, and myc-ubiquitin. As the E3-interacting regions of the E2 ubiquitin-conjugating enzymes are well conserved across vertebrate species (91), human reagents were employed in these assays. With increasing times of incubation at 37°C, more slowly migrating forms of TRIM5αAGM(Tan) were observed; concomitant incorporation of the myc-ubiquitin into proteins is consistent with TRIM5αAGM(Tan) polyubiquitylation (see Fig. S2A in the supplemental material). To examine the possibility that other coprecipitated cellular E3 ligase activities might mediate this ubiquitylation activity, we washed the immunoprecipitated TRIM5αAGM(Tan) with buffers containing increasing concentrations (from 0 to 1 M) of salt and then performed the in vitro ubiquitylation assay (see Fig. S2B in the supplemental material). The lack of apparent effect of salt washing on polyubiquitylation activity is consistent with the observed E3 ligase activity being mediated by TRIM5αAGM(Tan) itself. This interpretation is strengthened by our observation that changes in the TRIM5αAGM(Tan) RING domain dramatically influenced the degree of observed self-ubiquitylation (see below). Of the different E2 enzymes tested in the in vitro ubiquitylation assay, only UbcH5 functioned as an E2 ubiquitin-conjugating enzyme in support of both TRIM5αrh and TRIM5αAGM(Tan) polyubiquitylation (data not shown; see Fig. S2C in the supplemental material). No TRIM5α polyubiquitylation was observed in the absence of added E2 enzyme or an energy source. These results are consistent with those previously reported for TRIM5 self-ubiquitylation (87). In additional studies using wild-type and mutant ubiquitin proteins, TRIM5αrh polyubiquitylation occurred regardless of the position of the available lysine on the ubiquitin molecule (see Fig. S2D in the supplemental material). In addition, TRIM5αrh was found to have at least 5 sites of autoubiquitylation (see Fig. S2E in the supplemental material).
The E3 ubiquitin ligase activities of the TRIM5α variants studied here were compared in the above-mentioned in vitro assay. Of note, wild-type TRIM5αAGM(Tan) functioned as an efficient E3 ubiquitin ligase, whereas the wild-type TRIM5αAGM(Pyg) exhibited slower kinetics and a lower efficiency of polyubiquitylation (Fig. 3A). As expected from the predicted E2-interacting residues of TRIM5, L19A and E20A changes both greatly reduced the E3 ligase activity of TRIM5αAGM(Tan), but the E24Q and L26A mutants exhibited E3 ubiquitin ligase activities comparable to that of wild-type TRIM5αAGM(Tan) (Fig. 3B and C). Substantially less self-ubiquitylation was observed for the TRIM5αAGM(Tan) P29F and F34L mutants (Fig. 3C and data not shown). For the TRIM5αAGM(Tan) variants mentioned above, a strong correlation between in vitro E3 ubiquitin ligase activity and SIVmac restriction potency was observed (Fig. 3D) (rs = 0.9893; two-tailed P < 0.0001).
To determine if any of the TRIM5αAGM RING changes affected recognition of the SIVmac capsid, we assembled SIVmac CA-NC complexes in vitro (see Fig. S3 in the supplemental material, electron micrograph), and measured the binding of the TRIM5α proteins to the complexes. TRIM5αAGM(Tan) bound SIVmac CA-NC complexes with an efficiency comparable to that of TRIM5αrh (Table 1). Unexpectedly, the binding of the TRIM5αAGM(Pyg) protein to SIVmac CA-NC complexes was significantly better than that of the TRIM5αAGM(Tan) protein (Fig. 4 and Table 1). The binding of the TRIM5αAGM(Tan) F34L, P29F, L19A, and E20A mutants to the CA-NC complexes was comparable to that of the wild-type TRIM5αAGM(Tan) protein. The TRIM5αAGM(Pyg) L34F mutant bound CA-NC complexes comparably to the wild-type TRIM5αAGM(Pyg) protein. Thus, these changes in the TRIM5αAGM RING domain, which exert dramatic effects on both restriction of SIVmac infection and E3 ligase activity, do not detectably affect the interaction of the mutant TRIM5αAGM proteins with the SIVmac capsid.
To assess the relative contributions of RING-mediated ubiquitin ligase activity and capsid binding to TRIM5α restriction, we sought a system in which each of these components could be individually modulated. Because TRIM5αrh is a weak inhibitor of SIVmac infection, we used TRIM5αrh as a backbone to modify ubiquitin ligase activity (by RING substitution) and SIVmac capsid-binding ability [by B30.2(SPRY) substitution]. Thus, segments of the TRIM5αrh RING domain and/or the B30.2(SPRY) domain were replaced by the equivalent regions from TRIM5αAGM(Tan) or TRIM5αAGM(Pyg) (Fig. 5A). In the Rh-Rt and Rh-Rp chimeras, the RING domain of TRIM5αrh is replaced by the RING domains of TRIM5αAGM(Tan) and TRIM5αAGM(Pyg), respectively. In another pair of constructs [Rh-loop 1(Tan) and Rh-loop 2(Tan)], we replaced loop 1 and loop 2, respectively, of the TRIM5αrh RING domain with the equivalent loop from TRIM5αAGM(Tan). To modulate capsid binding, some of the RING domain chimeras were further modified by the replacement of the TRIM5αrh B30.2(SPRY) domain v1 variable region with that of TRIM5αAGM(Tan) or TRIM5αAGM(Pyg) (Rh-RtVp, Rh-RtVt, Rh-RpVp, and Rh-RpVt in Fig. 5A).
The chimeric proteins and the parental TRIM5α proteins were stably expressed in canine Cf2Th cells, which were challenged with SIVmac. The Rh-Rt protein with the TRIM5αAGM(Tan) RING domain inhibited SIVmac infection better than either wild-type TRIM5αrh or the Rh-Rp protein with the TRIM5αAGM(Pyg) RING domain (Fig. 5B, left, and Table 1). However, all three TRIM5α variants restricted HIV-1 infection equivalently (Fig. 5B, right). These results imply that the TRIM5αrh and TRIM5αAGM(Pyg) RING domains are not as functionally active as the TRIM5αAGM(Tan) RING domain and that TRIM5α RING function is more important for restriction of SIVmac than of HIV-1. Replacement of the TRIM5αrh RING domain loop 1 sequence with that of TRIM5αAGM(Tan) was sufficient to achieve enhanced restriction of SIVmac infection [Rh-loop 1(Tan) in Table 1]. As only amino acid residue 29 differs between the loop 1 sequences of TRIM5αrh and TRIM5αAGM(Tan), we conclude that the gain in SIVmac-restricting activity results from the replacement of histidine 29 in TRIM5αrh with proline. In contrast, the Rh-loop 2(Tan) protein, in which loop 2 of TRIM5αrh was replaced by that of TRIM5αAGM(Tan), did not efficiently inhibit SIVmac infection (Table 1). The E3 ubiquitin ligase activity of the TRIM5αrh-TRIM5αAGM chimeras was examined using the in vitro self-ubiquitylation assay. Of note, the Rh-Rt protein exhibited much greater efficiency than the Rh-Rp protein in this assay (Fig. 3C), supporting the correlation between E3 ligase activity and SIVmac restriction. Also consistent with this correlation is the observation that the potently restricting Rh-loop 1(Tan) chimera, with only the TRIM5αAGM(Tan) RING loop 1 region, exhibited more ubiquitin ligase activity than the poorly restricting Rh-loop 2(Tan) protein (Fig. 3C). No significant difference in the turnover rates of the TRIM5αrh, Rh-Rp, and Rh-Rt proteins was observed (see Fig. S4 in the supplemental material). These results support a role for the E3 ubiquitin ligase activity of the TRIM5α RING domain in restricting SIVmac infection.
The Rh-RtVp, Rh-RtVt, Rh-RpVp, and Rh-RpVt chimeras contain RING domains and B30.2(SPRY) domain v1 regions from TRIM5αAGM(Tan) or TRIM5αAGM(Pyg) (Fig. 5A). All of these chimeric proteins inhibited SIVmac infection more efficiently than TRIM5αAGM(Pyg) and TRIM5αrh (Fig. 5C and Table 1). Several independent experiments resulted in the same order of anti-SIVmac activity: Rh-RtVp > Rh-RtVt = Rh-RpVp > Rh-RpVt. For these four chimeras, both the RING domain and B30.2(SPRY) domain sequences contribute to the potency of SIVmac-restricting ability. The Rh-RtVp and Rh-RtVt chimeras with the TRIM5αAGM(Tan) RING domain inhibited SIVmac infection more efficiently than the matched proteins (Rh-RpVp and Rh-RpVt, respectively) with the RING domain of TRIM5αAGM(Pyg). Thus, we tested the hypothesis that the greater SIVmac restriction potency of the chimeric proteins with the TRIM5αAGM(Tan) RING domain was associated with higher E3 ubiquitin ligase activity. The Rh-RtVp and the Rh-RtVt chimeras containing the TRIM5αAGM(Tan) RING domain exhibited dramatically more self-ubiquitylation in the in vitro assay than the Rh-RpVp and Rh-RpVt chimeras with the TRIM5αAGM(Pyg) RING domain (Fig. 5D). TRIM5α E3 ubiquitin ligase activity was not apparently influenced by the sequence of the B30.2(SPRY) v1 region (Fig. 5D, compare Rh-RtVp with Rh-RtVt, and also Rh-RpVp with Rh-RpVt). These results are consistent with our hypothesis that superior E3 ubiquitin ligase activity contributes to the potency of SIVmac restriction by the chimeric TRIM5α proteins with the RING domain of TRIM5αAGM(Tan).
The B30.2(SPRY) domain v1 variable region is longer in TRIM5αAGM than in TRIM5αrh due to a tandem duplication of 20 amino acid residues (73, 74). In the Rh-RtVp, Rh-RtVt, Rh-RpVp, and Rh-RpVt chimeras, the v1 region from either TRIM5αAGM(Tan) or TRIM5αAGM(Pyg) substitutes for that of TRIM5αrh. The v1 sequences of TRIM5αAGM(Tan) differ by 5 amino acid residues from those of TRIM5αAGM(Pyg) (Fig. 1A). The Rh-RtVp and Rh-RpVp chimeras with the B30.2(SPRY) v1 region from TRIM5αAGM(Pyg) inhibited SIVmac infection more efficiently than the matched proteins (Rh-RtVt and Rh-RpVt, respectively) with the TRIM5αAGM(Tan) B30.2(SPRY) v1 region (Fig. 5C). We tested the hypothesis that differences in the efficiency of capsid binding determined the observed differences in SIVmac restriction potency by measuring the binding of these TRIM5α variants to SIVmac CA-NC complexes. As described above, the binding of the TRIM5αAGM(Pyg) protein to SIVmac capsid complexes was significantly better than that of the TRIM5αAGM(Tan) protein (Fig. 5E and Table 1). Both chimeric proteins (Rh-RtVp and Rh-RpVp) with the B30.2(SPRY) v1 region from TRIM5αAGM(Pyg) bound SIVmac capsid complexes more efficiently than the chimeric proteins (Rh-RtVt and Rh-RpVt) with the TRIM5αAGM(Tan) v1 region (Fig. 5E and Table 1). The SIVmac capsid-binding abilities of the Rh-RtVp and Rh-RpVp chimeras were almost equivalent to that of TRIM5αAGM(Pyg). Differences in the RING domains of the TRIM5α chimeras had no apparent effect on the efficiency of binding to the SIVmac CA-NC complexes. Thus, the v1 variable region of the B30.2(SPRY) domain is the major determinant of differences in SIVmac capsid-binding ability among these TRIM5α variants.
From these data, we conclude that both the RING domain, with associated E3 ubiquitin ligase activity, and the B30.2 v1 region, which determines capsid-binding affinity, independently contribute to the potency of SIVmac restriction. Thus, the Rh-RtVt chimera, which has potent E3 ubiquitin ligase activity but poor capsid-binding ability, achieved a level of SIVmac restriction similar to that of the Rh-RpVp chimera, which has poor E3 ubiquitin ligase activity but potent capsid-binding ability (Fig. 5C and Table 1).
The TRIM5α variants studied here exhibited defined differences in RING-mediated E3 ubiquitin ligase activity and, in some cases, in capsid-binding affinity, as well. This afforded an opportunity to examine the impacts of changes in TRIM5α ubiquitylation capacity and capsid binding on the reverse transcription process and the fate of the retroviral capsid in infected cells. The level of late SIVmac reverse transcripts following the infection of cells expressing TRIM5αAGM(Pyg) was comparable to that measured in cells transduced with the empty LPCX vector control (Fig. 6A). In contrast, the levels of SIVmac late reverse transcripts were dramatically reduced in target cells expressing TRIM5αAGM(Tan) or the functionally active TRIM5αAGM(Tan) E24Q and L26A mutants (Fig. 6A and Table 1). Efficient SIVmac reverse transcription was observed in cells expressing the TRIM5αAGM(Tan) L19A, E20A, and P29F mutants, which are defective in E3 ligase activity and SIVmac restriction (Fig. 2 and and33 and Table 1). Thus, for this group of TRIM5αAGM variants, both RING-mediated ubiquitylation and inhibition of viral reverse transcription correlated with SIVmac restriction potency.
The levels of SIVmac late reverse transcripts were also examined in target cells expressing the TRIM5αrh-TRIM5αAGM chimeric proteins (Fig. 6A and Table 1). The levels of SIVmac cDNA were lowest in the target cells expressing the chimeric proteins (Rh-RtVp and Rh-RtVt) containing the TRIM5αAGM(Tan) RING domain, which specifies robust E3 ubiquitin ligase activity. SIVmac reverse transcript levels were highest in cells expressing the Rh-RpVp protein and only slightly lower in cells expressing the Rh-RpVt protein. Of note, although the levels of SIVmac infection were nearly identical in target cells expressing the Rh-RtVt and Rh-RpVp proteins (Fig. 5C), which contain counterbalancing RING and B30.2(SPRY) domain functions, the levels of SIVmac reverse transcripts were markedly greater in the cells expressing the Rh-RpVp protein. Examination of minus-strand strong-stop cDNA corroborated the results observed for the late SIVmac reverse transcripts (Fig. 6B and data not shown), indicating that the TRIM5α-mediated blocks occur prior to or at the earliest stages of reverse transcription. For the entire panel of TRIM5α variants, inhibition of viral reverse transcription strongly correlated with the E3 ubiquitin ligase efficiency of the RING domain (Fig. 6C) (rS = 0.9510; two-tailed P < 0.0001).
A previous study suggested that approximately 10% of the TRIM5αAGM restriction of SIVmac infection could be relieved by treatment of the target cells with the proteasome inhibitor MG132 (48). We investigated the effects of inhibiting the proteasome on the reverse transcription block to SIVmac infection mediated by the TRIM5αrh-TRIM5αAGM chimeras. Late SIVmac reverse transcripts in target cells expressing the Rh-RtVp protein increased after treatment of the target cells with the proteasome inhibitor MG132 or ALLN (Fig. 6D). The proteasome inhibitor-mediated increase in Rh-RtVp late reverse transcripts was evident by 1 h after infection (Fig. 6E). Treatment with proteasome inhibitors minimally affected the levels of SIVmac late reverse transcripts in target cells expressing the Rh-RpVp protein or control cells transduced with the empty LPCX vector (Fig. 6D). By titrating the amount of input DNA in the assay, ALLN treatment was shown to result in only a 3-fold increase in the levels of SIVmac reverse transcripts in the Rh-RpVp-expressing or LPCX-transduced cells (Fig. 6F). Thus, proteasome inhibition specifically diminishes the ability of a TRIM5α protein with potent RING-associated E3 ubiquitin ligase activity to mediate decreases in SIVmac cDNA following infection.
The effects of proteasome inhibition on the restriction of SIVmac infection by the Rh-RtVp and Rh-RpVp chimeric proteins were examined. Treatment of the control LPCX-transduced cells with the proteasome inhibitor MG132 resulted in a modest decrease in SIVmac infection, perhaps due to the effects of the treatment on cell viability (Fig. 6G). MG132 treatment increased the efficiency with which SIVmac infected cells expressing the Rh-RtVp protein but did not affect SIVmac infection of the Rh-RpVp-expressing cells. Thus, proteasome inhibition partially relieves SIVmac restriction by a TRIM5α protein with potent RING-associated E3 ubiquitin ligase activity, but not by a matched TRIM5α protein with poor RING E3 ubiquitin ligase function.
To investigate the impact of TRIM5 RING function on capsid disassembly, the amount of particulate SIVmac capsids in the cytoplasm of SIVmac-challenged cells expressing the TRIM5α variants was measured (77). Minimal amounts of SIVmac capsid proteins were detected in the cytosol of cells exposed to a recombinant SIVmac without envelope glycoproteins (Fig. 7A). The amount of particulate SIVmac capsids detected in the cytosol of cells expressing a TRIM5α protein from spider monkeys (a New World monkey) was less than that detected in cells expressing TRIM5αAGM(Tan), which in turn was less than that seen in TRIM5αrh-expressing cells (Fig. 7A). This order corresponds to the level of SIVmac infection observed in the cells expressing the TRIM5α proteins from these three monkey species (17) (Table 1). The level of cytosolic particulate SIVmac capsids was lower in infected cells expressing TRIM5αAGM(Tan) than in TRIM5αAGM(Pyg)-expressing cells (Fig. 7B). The level of cytosolic particulate SIVmac capsids was lower in cells expressing the restricting TRIM5αAGM(Tan) L26A mutant than in cells expressing the poorly restricting L19A mutant (Fig. 7B).
The levels of cytosolic particulate SIVmac capsids in infected cells expressing the TRIM5αrh-TRIM5αAGM chimeric proteins were also examined. A high level of particulate SIVmac capsids was observed in the cells expressing the Rh-RpVp and Rh-RpVt proteins (Fig. 7C). Lower levels of particulate SIVmac capsids, comparable to those seen in TRIM5αAGM(Tan)-expressing cells, were observed in cells expressing the Rh-RtVt and Rh-RtVp chimeras (Fig. 7C). The observed levels of particulate cytoplasmic SIVmac capsids inversely correlated with the E3 ubiquitin ligase activities of the TRIM5α variants expressed in the target cells (Fig. 7E) (rS = 0.7467 and P = 0.0013). A strong correlation was observed between the amounts of particulate SIVmac capsids in the cytosol and the levels of late SIVmac reverse transcripts (Fig. 7F) (rS = 0.8289 and P = 0.0003). In experiments where the levels of capsid proteins in the input sample were well matched, decreases in pelletable capsid in cells expressing a restricting TRIM5α protein were accompanied by an increase in capsid protein in the supernatant. These data support the contribution of TRIM5α RING E3 ubiquitin ligase function to the premature disassembly of the SIVmac capsid in infected cells. Moreover, TRIM5α-mediated decreases in the amounts of particulate SIVmac capsids are strongly associated with blocks in reverse transcription.
The effect of proteasome inhibition by MG132 treatment of the Rh-RtVp- and Rh-RpVp-expressing cells on the fate of the SIVmac capsid was examined. MG132 treatment resulted in little change in the levels of particulate SIVmac capsids in the LPCX control cells and in the Rh-RpVp-expressing cells (Fig. 7D). In contrast, the low level of particulate capsids in Rh-RtVp-expressing cells was restored to much higher levels by MG132 treatment. These effects of MG132 treatment on the level of particulate cytosolic capsids mirrored the MG132 effects on reverse transcription shown in Fig. 6D. Thus, proteasome inhibition specifically diminishes the ability of a TRIM5α protein with potent RING-associated E3 ubiquitin ligase activity to accelerate the disassembly of SIVmac capsids and to disrupt reverse transcription in infected cells.
Changes in the helix-4/5 loop (the “cyclophilin-binding loop”) of the HIV-1 capsid can influence sensitivity to TRIM5α-mediated restriction (4, 24, 33, 35, 36, 47, 49, 59, 72, 78, 80, 92, 93). To investigate whether the helix-4/5 loop might modulate viral sensitivity to TRIM5α RING changes, we made and tested SIVmac(HIV 4/5), which is identical to SIVmac except that the helix-4/5 loop of its capsid has been replaced by that of HIV-1. SIVmac(HIV 4/5) infection was less efficient than that of SIVmac, and was potently inhibited in cells expressing the Rh-RtVp and Rh-RpVp proteins (Fig. 8A). The ratios of particulate:soluble capsid proteins in the cytosol and the levels of viral late reverse transcripts were decreased for SIVmac, HIV-1, and SIVmac(HIV 4/5) in cells expressing the Rh-RtVp protein compared with those in control cells transduced with the empty LPCX vector (Fig. 8B and C). In cells expressing the Rh-RpVp protein, with diminished RING E3 ubiquitin ligase function, the levels of SIVmac particulate capsids and late reverse transcripts were similar to those observed in the LPCX-transduced control cells (Fig. 8B, left, and C). In contrast, the levels of SIVmac(HIV 4/5) late reverse transcripts were reduced in the Rh-RpVp-expressing cells compared with those in the LPCX control cells (Fig. 8C). Moreover, the ratio of particulate:soluble capsid proteins following SIVmac(HIV 4/5) infection of the Rh-RpVp-expressing cells was approximately 5.3-fold lower than that seen after infection of the LPCX control cells (Fig. 8B, left). Treatment of the target cells with cyclosporine, which blocks cyclophilin A interaction with the HIV-1 helix-4/5 loop, did not change the level of SIVmac(HIV 4/5) reverse transcripts in cells expressing the Rh-RtVp and Rh-RpVp proteins (see Fig. S5 in the supplemental material). Smaller amounts of HIV-1 particulate capsids were observed in the target cells expressing TRIM5αAGM(Tan) and all of the TRIM5αrh/TRIM5αAGM chimeras than in the LPCX-transduced control cells (Fig. 8B, right). Thus, the helix-4/5 loop of the capsid protein can influence the sensitivity of the early blocks in retroviral infection to altered RING function of the restricting TRIM5α protein. Apparently, these effects are independent of the binding of cyclophilin A to the helix-4/5 loop.
Deletion of the TRIM5α RING domain or disruption of RING domain folding by alteration of the zinc-coordinating cysteine residues has been shown to result in varying effects on retroviral inhibition, depending on the TRIM5 protein and the restricted virus (12, 29, 48, 60, 75). In this study, by using a panel of RING mutants with intact zinc-binding residues, we show that the E3 ubiquitin ligase function of TRIM5αAGM is important for the restriction of SIVmac infection. The relative anti-SIVmac inhibitory activities of TRIM5α variants that differed in RING domain sequences strongly correlated with the E3 ubiquitin ligase activity measured in vitro. The TRIM5αAGM RING variants were shown to dimerize, bind HIV-1 and SIVmac capsid complexes, and restrict HIV-1 infection, supporting their structural integrity. Thus, the RING-mediated E3 ubiquitin ligase activity represents an “effector function” for TRIM5α-mediated SIVmac restriction, contributing additively with capsid binding to determine restriction potency. The Rh-RtVt and Rh-RpVp chimeric proteins provide an example of how equivalent levels of SIVmac restriction can be achieved by combining robust RING and weak B30.2(SPRY) functions, or weak RING and robust B30.2(SPRY) functions, respectively.
The RING-associated E3 ubiquitin ligase activity of TRIM5α inversely correlated with the levels of the SIVmac particulate capsids in the cytosol of the infected cells and with SIVmac cDNA synthesis. Moreover, the levels of particulate cytosolic capsids and viral cDNAs correlated. As the capsid is the direct binding target of TRIM5α (31, 38, 77), our data suggest a model in which the ubiquitin ligase activity of TRIM5α leads to accelerated disassembly of cytosolic SIVmac capsids. As has been seen for unstable HIV-1 capsid mutants (17, 40, 42), premature uncoating of the SIVmac capsids by TRIM5α directly or indirectly leads to disruption of reverse transcription.
Our results indicate the importance of the E3 ubiquitin ligase activity of TRIM5αAGM to SIVmac restriction. What might be the functionally important target of TRIM5α-mediated ubiquitylation, and what are the consequences of ubiquitylation of the potential target? RING-mediated E3 ubiquitin ligases simultaneously bind both the E2 enzyme partner and substrate to transfer the ubiquitin moiety from E2 to substrate; a few TRIM family members are known to transfer ubiquitin to their ligands, which are directly recognized by the B30.2(SPRY) domains (19, 30, 39, 50, 59). Based on these examples, one might expect the ubiquitylation target of TRIM5α to be the ligand of the B30.2(SPRY) domain, i.e., the retroviral capsid. However, no ubiquitylation of the SIVmac capsid protein in either particulate or soluble form was observed in the fate-of-capsid assay. Similarly, fate-of-capsid studies conducted with N-MLV restricted by human TRIM5α also allowed visualization of the solubilized capsid proteins that resulted from TRIM5α action; no modified form of the capsid protein was detected (10, 64, 77). Thus, if the targeted capsid is ubiquitylated, only a very small fraction of the total protein population is modified. Another possibility is that TRIM5α autoubiquitylation contributes to retroviral restriction. It has been reported that infection of cells with a retrovirus susceptible to TRIM5α restriction leads to a decrease in the level of TRIM5α protein in the cell (66). The aggregation of TRIM5α proteins on the capsid surface could favor ubiquitylation and proteasome-mediated turnover of a fraction of the TRIM5α protein in the cell, potentially leading to faster uncoating of the bound capsid. However, considering that the ubiquitin chains are also able to induce conformational changes and recruit several cellular factors, depending on the type of Ub-Ub chain linkages, further work is required to substantiate a role for TRIM5α autoubiquitylation in SIVmac restriction. Finally, host cell factors could be ubiquitylated by TRIM5α. The identification of such host cell proteins and dissection of their roles in early postentry steps in SIVmac infection are worthy goals of future studies.
Proteasome inhibition resulted in increases in the levels of SIVmac particulate cytosolic capsids and reverse transcripts in infected cells expressing particular restricting TRIM5α proteins. These effects were observed for a TRIM5α protein, Rh-RtVp, with potent RING E3 ubiquitin ligase function and were not seen in cells expressing the matched Rh-RpVp TRIM5α protein with diminished RING activity. The phenotypic similarity of proteasome inhibition and mutagenic inactivation of the TRIM5α RING E3 ubiquitin ligase supports the involvement of ubiquitin in pre-reverse transcription blocks to SIVmac infection. Previous studies suggested that, although proteasome activity is not required for TRIM5α antiviral activity per se (2, 48, 61, 66, 77), treatment with proteasome inhibitors resulted in higher levels of HIV-1 and murine leukemia virus reverse transcripts and particulate capsids in TRIM5α-expressing target cells (2, 10, 12, 84). Interpretation of results with proteasome inhibitors, however, is complicated by the observation that the levels of retroviral cDNA and particulate capsids in the cytosol can be increased by proteasome inhibition even in cells not expressing a restricting TRIM5α protein (Fig. 6F and and7D)7D) (10, 12, 16, 70, 82). Moreover, proteasome inhibition causes the redistribution of TRIM5α into large cytoplasmic aggregates (12, 14), potentially introducing nonphysiologic artifacts into the restriction mechanism. Finally, proteasome inhibitors not only block the proteasome-mediated degradation pathway, but can also deplete the pool of free ubiquitin in cells by impeding the recycling of ubiquitin from protein conjugates (9, 86). Thus, although our results clearly implicate ubiquitylation in TRIM5α-mediated pre-reverse transcription blocks of SIVmac infection, it is formally possible that ubiquitin ligation contributes to this restriction in a manner that is independent of proteasome-mediated degradation.
RING-mediated E3 ubiquitin ligase activity was apparently more important to SIVmac inhibition than to HIV-1 inhibition by TRIM5αAGM. In a recent report (48), Maegawa and colleagues also observed that the phenotypes of TRIM5α mutants with changes in RING domain cysteines depended on the restricted virus. Although further work will be needed to understand the basis for the SIVmac-HIV-1 difference, we found that the helix-4/5 loop on the capsid influenced the requirement for RING-mediated E3 ubiquitin ligase activity. As changes in this loop have been shown to modulate HIV-1 capsid stability and the rate of capsid disassembly in infected cells (15, 89–91), quantitative or qualitative differences between the subunit interactions in the HIV-1 and SIVmac capsids may alter the requirements for the E3 ubiquitin ligase function of TRIM5α.
Two TRIM5α variants, Rh-RpVp and Rh-RpVt, with poor RING domain function restricted SIVmac infection moderately, even though corresponding decreases in the levels of viral reverse transcripts or particulate capsids were not detected. Apparently, SIVmac restriction by these two TRIM5α proteins occurs after reverse transcription and without detectable increases in the rate of capsid disassembly. Rh-RpVp blocked SIVmac infection more potently than Rh-RpVt, indicating that the strength of this apparent post-reverse transcription block is influenced by capsid-binding affinity. TRIM protein variants have been reported to block retroviral infection after viral cDNA synthesis (90, 92). The availability of well-characterized TRIM5α mutants should allow further exploration of the mechanism and relevance of these blocks.
We thank Yvette McLaughlin and Elizabeth Carpelan for manuscript preparation. We thank Keith Reimann and the National Institutes of Health AIDS Research and Reference Reagent Program for the gift of serum from an SIVmac251-infected monkey.
We acknowledge the support of the National Institutes of Health (AI063987, AI076094, and a Center for AIDS Research Award, AI06354), the International AIDS Vaccine Initiative, and the late William F. McCarty-Cooper.
†Supplemental material for this article may be found at http://jvi.asm.org/.
Published ahead of print on 15 June 2011.