The Antiviral Spectra of TRIM5α Orthologues and Human TRIM Family Proteins against Lentiviral Production

Background Rhesus monkey TRIM5α (TRIM5αrh) recognizes the incoming HIV-1 core through its C-terminal B30.2(PRYSPRY) domain and promotes its premature disassembly or degradation before reverse transcription. Previously, we have shown that TRIM5αrh blocks HIV-1 production through the N-terminal RBCC domain by the recognition of Gag polyproteins. Although all TRIM family proteins have RBCC domains, it remains elusive whether they possess similar late-restriction activities. Methodology/Principal Findings We examined the antiviral spectra of TRIM5α orthologues and human TRIM family members which have a genetic locus proximal to human TRIM5α (TRIM5αhu), against primate lentiviral production. When HIV-1 virus-like particles (VLPs) were generated in the presence of TRIM5α proteins, rhesus, African green and cynomolgus monkey TRIM5α (TRIM5αag and TRIM5αcy), but not TRIM5αhu, were efficiently incorporated into VLPs, suggesting an interaction between HIV-1 Gag and TRIM5α proteins. TRIM5αrh potently restricted the viral production of HIV-1 groups M and O and HIV-2, but not simian lentiviruses including SIVMAC1A11, SIVAGMTan-1 or SIVAGMSAB-1. TRIM5αhu did not show notable late restriction activities against these lentiviruses. TRIM5αag and TRIM5αcy showed intermediate restriction phenotypes against HIV-1 and HIV-2, but showed no restriction activity against SIV production. A series of chimeric TRIM5α constructs indicated that the N-terminal region of TRIM5αag and TRIM5αcy are essential for the late restriction activity, while the C-terminal region of TRIM5αcy negatively regulates the late restriction activity against HIV-1. When select human TRIM family proteins were examined, TRIM21 and 22 were efficiently incorporated into HIV-1 VLPs, while only TRIM22 reduced HIV-1 titers up to 5-fold. The antiviral activities and encapsidation efficiencies did not correlate with their relative expression levels in the producer cells. Conclusions/Significance Our results demonstrated the variations in the late restriction activities among closely related TRIM5α orthologues and a subset of human TRIM family proteins, providing further insights into the late restriction activities of TRIM proteins.


Introduction
Approximately 8% of the human genome is comprised of retroviral elements, implicating an extensive history of competition between hosts and retroviruses [1,2]. To counteract these viruses, primates have developed defensive measures which target various aspects of the retroviral life cycle. Cellular restriction factor TRIM5a is one such contributing element in this antiviral defense against retroviruses [3,4,5,6]. TRIM5a belongs to the TRIM family of proteins, which are characterized by sequential domains in the N-terminal half of the protein, RING, with one or two bboxes followed by a coiled-coil motif and its a isoform includes a C-terminal B30.2(PRYSPRY) domain. The rhesus monkey TRIM5a (TRIM5arh) recognizes the incoming HIV-1 core through its C-terminal B30.2(PRYSPRY) domain and promotes its premature disassembly or degradation before reverse transcription [7,8,9,10]. Primate TRIM5a orthologues have distinct post-entry restriction activities against a range of retro-and lentiviruses; however, they generally lack strong restriction activity against their own host-specific viruses. For instance, human TRIM5a (TRI-M5ahu) restricts N-tropic murine leukemia virus (N-MLV) as well as equine infectious anemia virus (EIAV), but not human immunodeficiency virus type-1 (HIV-1) or simian immunodeficiency virus (SIV) [3,4,6,11]. In contrast, TRIM5arh expression in HIV-1-permissive cells confers strong antiviral activity against HIV-1, EIAV, N-MLV and SIV from African green monkeys (SIV AGM ), but not against SIV from rhesus macaques (SIV MAC ) [3,4,6,9,11,12]. The African green monkey TRIM5a orthologue (TRIM5aag) restricts HIV-1, SIV MAC , EIAV and N-MLV, but not SIV AGM [3,11], while the cynomolgus monkey orthologue (TRIM5acy) restricts HIV-1 and HIV-2, but not SIV MAC infection [13]. These post-entry restriction patterns of TRIM5a orthologues suggest that lentiviruses have evolved to evade TRIM5a-mediated post-entry restriction when colonizing respec-tive species. In response, host species also appear to have evolved their TRIM5a proteins, especially the coiled-coil and B30.2(PRYSPRY) domains, against retro-and lentiviruses [14,15].
TRIM5arh also exhibits an additional antiviral activity against HIV-1 production, independently of the well-characterized postentry restriction, to block the late phase of HIV-1 replication [16,17]. High levels of TRIM5arh expression blocks HIV-1 production predominantly by reducing the number of HIV-1 virions, while modest TRIM5arh expression blocks the late phase of HIV-1 replication by reducing virion infectivity as well as virion numbers [16,18]. When HIV-1 virus-like-particles (VLPs) are produced in the presence of TRIM5arh, TRIM5arh is efficiently incorporated into VLPs, implicating the interaction between cellular and viral components during viral assembly [16]. This TRIM5arh-mediated restriction of HIV-1 production is mediated by the N-terminal RBCC domain, but not the C-terminal B30.2(PRYSPRY) domain [16]. Further studies have identified several determinants for this late restriction. A RING structure is essential for the efficient interaction with HIV-1 Gag, while two amino acid residues in TRIM5arh coiled-coil domain (M133 and T146) are critical for the late restriction activity [19]. Our data suggest that the TRIM5arh-mediated late restriction involves at least two distinct activities: (i) interaction with HIV-1 Gag polyprotein through the N-terminal, RING and b-box 2 regions of a TRIM5arh monomer, and (ii) an effector function(s) that depends upon the coiled-coil and linker 2 domains of TRIM5arh [19]. Although TRIM5ahu does not show strong late restriction activities against HIV-1 group M viruses [16], it remains to be determined if the human orthologue has any late restriction activity against other human and non-human primate lentiviruses.
Previous studies have shown that TRIM5a is interferonresponsive [20,21]. Recent study identified TRIM6, 21, 22 and 34, which are located on chromosome 11p together with TRIM5, are also interferon responsive [22]. Given that the late restriction of TRIM5a is dependent on the RBCC domain and these TRIM family proteins possess an N-terminal RBCC, it is possible that TRIM5ahu paralogues may have similar late-restriction activities.
In the present study, we determined the antiviral spectra and Figure 1. Amino acid sequence alignment of TRIM5a orthologues examined in this study. Consensus amino acid sequences of TRIM5a. as well as TRIM5ahu (Hu), TRIM5arh (Rh), TRIM5aag (Ag) and TRIM5acy (Cy) amino acid sequences are shown. Identical residues are indicated by dots. Gaps are indicated by dashes. HindIII site in the coiled-coil domain was used as the junction for the chimeras used in this study. doi:10.1371/journal.pone.0016121.g001 encapsidation efficiency of TRIM5a orthologues and paralogues. We found that TRIM5a orthologues from African green (TRIM5aag) and cynomolgus (TRIM5acy) monkeys have similar, but weaker late restriction activities against HIV-1 production. Similar to TRIM5arh, TRIM5aag and TRIM5acy proteins were efficiently incorporated in HIV-1 VLPs, and their RBCC domains were essential for the late restriction activities. Intriguingly, the Cterminal regions of TRIM5aag and TRIM5acy proteins negatively regulated the late restriction activities. Studies using human TRIM5 paralogues with conserved RBCC domains demonstrated that TRIM21 and TRIM22 specifically incorporated into HIV-1 VLPs, while only TRIM22 mildly restricted HIV-1 production. Our results therefore demonstrate the variable late restriction activities of TRIM5a orthologues and paralogues. The involvement of the C-terminal sequences of TRIM5a proteins in determining the potency of late restriction activities suggests more complex mechanisms underlying TRIM protein-mediated late restriction activities than previously reported.
Virus-like-particle incorporation assay 1.0 mg of TRIM5a-or TRIM-expressing plasmid and 0.2 mg of a codon-optimized HIV-1 Gag-Pol expression plasmid, pH-GP [36], were co-transfected into 293T cells using FuGene 6. Two days post-transfection, transfected cells were harvested in RIPA buffer to assess TRIM5a and HIV-1 GagPol expression. Culture supernatants were also harvested, and passed through a 0.45 mmpore syringe filter for VLP purification. The filtered supernatants were then centrifuged at 18,0006g for 90 min through a 20% sucrose cushion, resuspended in PBS and centrifuged at 18,0006g for 90 min. Pelleted VLPs were lysed in 5 ml of RIPA buffer and to which 5 ml of sample buffer was added, heat-denatured and subjected to immunoblot analysis.

Viral production assay
For the assessment of TRIM5a orthologue antiviral activities, increasing amounts of pHuT5a, pRhT5a, pAgmT5a and pCynT5a (0.1, 0.3 and 1.0 mg) were co-transfected with 0.1 mg of infectious lentiviral proviral plasmid into 293T cells (1.0610 6 cells) using FuGene 6 (Roche). A control plasmid (pBlueScrip-tIIKS(+), Stratagene) was added to each transfection reaction to bring the final plasmid concentrations to 1.2 mg per transfection. Human TRIM family protein antiviral activities were examined similarly, except 1.0 mg of TRIM-expression plasmid was used in each reaction. Two days post-transfection, cellular supernatants were passed through a 0.45 mm-pore syringe filter and the viral titers in the supernatants were determined as infectious units/ml (IU/ml) in GHOST(3)R3/X4/R5 indicator cells.

Late restriction activities of TRIM5arh, TRIM5aag and TRIM5acy against primate lentiviruses
In the well-characterized TRIM5a post-entry restriction, the TRIM5a B30.2(PRYSPRY) determines the potency and specificity of restriction. In contrast, the late restriction of HIV-1 depends on the TRIM5a RBCC domain [16,19]. The highly conserved RBCC sequences among simian TRIM5a orthologues suggest the possible late restriction activities in other primate TRIM5a proteins (Fig. 1). We therefore examined the late-restriction activities of TRIM5aag and TRIM5acy against a series of human and non-human primate lentiviruses. Human lentiviruses, HIV-1 group M viruses, HIV-1 group O and HIV-2 viruses were examined for their sensitivity to TRIM5a-mediated late restriction. HIV-1 group O viruses are divergent from group M viruses [37] and show different sensitivities to TRIM5arh-mediated Lv1 restriction [38], partly due to their cyclophilin A-independence [38,39]. Since group O and HIV-2 viruses have been less successful in colonizing human population than the moreprevalent HIV-1 group M viruses, we hypothesized that TRIM5ahu might be able to block the late phase of group O HIV-1 or HIV-2 replication. SIV MAC and SIV AGM were also included to test the late restriction activities of TRIM5a proteins against non-human primate lentiviruses.
Immunoblot analysis was performed to verify the proper expression of TRIM5aag and TRIM5acy proteins following transfection into 293T cells ( Fig. 2A). TRIM5arh reduced the titers of two HIV-1 group M clones (NL4-3, 89.6) and HIV-1 group O clones (CMO2.41 and CMO2.5) by up to 60-fold in a dose-dependent manner (Fig. 2B). TRIM5arh also reduced the titers of HIV-2 up to 40-fold. Although SIV MAC and SIV AGM SAB-1 titers were largely unaffected by TRIM5arh, we observed a 10-fold decrease in the titers of SIV AGM Tan-1 in the presence of TRIM5arh (Fig. 2B). TRIM5aag and TRIM5acy showed similar patterns of late restriction activity to its rhesus monkey orthologue, although the reductions in viral titers were modest. NL4-3 titers were reduced by 7-fold in cells expressing TRIM5aag or TRIM5acy, while 89.6 titers were only reduced by 2-fold. Although group O isolate CMO2.41 virus titers were reduced by up to 7-fold when producer cells expressed TRIM5aag or TRIM5acy, the titers of another Group O clone CMO2.5, which is based on the CMO2.41 isolate but containing the 59LTR to vpr sequences from the MVP2171 O-type isolate [40], were strongly affected by TRIM5aag but not TRIM5acy. TRIM5aag and TRIM5acy showed little late restriction activity on SIV MAC 1A11, SIV AGM Tan-1, SIV AGM SAB-1 and HIV-2 production. Intriguingly, TRIM5aag reduced SIV AGM Tan-1 titers by 6-fold (Fig. 2B), which may be explained by the difference in the host species of SIV AGM and TRIM5aag: SIV AGM Tan-1 is isolated from the Chlorocebus tantalus [31] but the TRIM5aag protein used in these experiments were derived from Chlorocebus aethiops-derived CV1 cells [41]. TRIM5ahu marginally reduced HIV titers, while it did not affect SIV titers (Fig. 2B).

Efficient encapsidation of rhesus monkey, African green monkey and cynomolgus monkey TRIM5a into HIV-1 virus-like particles
Previously, we have shown that when HIV-1 proteins were produced in the presence of TRIM5arh, HIV-1 Gag polyproteins were rapidly degraded [16]. In contrast, over-expression of codonoptimized HIV-1 GagPol was able to saturate the late restriction activity, leading to production of sufficient amounts of VLPs in the presence of TRIM5arh and efficient incorporation of TRIM5arh in the VLPs [16,17]. Efficient encapsidation of TRIM5arh into VLPs generated without HIV-1 protease suggests specific interaction between TRIM5arh and HIV-1 Gag before or during HIV-1 assembly [16]. HIV-1 protease appears to cleave TRIM5arh in the B30.2(PRYSPRY) domain to produce the truncated, 20 kDa form of TRIM5arh in the VLPs, because formation of the 20 kDa form was not seen in the VLPs made without HIV-1 Pol, or in the VLPs treated with HIV-1 protease inhibitors [16]. TRIM5arh and TRIM5ahu chimeric constructs demonstrated that processing of TRIM5a proteins was not necessary for their late restriction activities [16,17]. Since three simian TRIM5a proteins showed different late restriction activities (Fig. 2B), we examined the differences in their HIV-1 Gagassociation efficiencies by the VLP encapsidation assay. When high levels of HIV-1 VLPs were generated in the presence of TRIM5a proteins, efficient incorporation of truncated forms of TRIM5arh, TRIM5aag and TRIM5acy, but not TRIM5ahu, were observed (Fig. 3A). The truncated forms of TRIM5a likely resulted from the cleavage of TRIM5a proteins in the B30.2(PRYSPRY) domain by the HIV-1 protease. The efficient incorporation of the simian TRIM5a proteins into VLPs suggests a specific interaction between simian TRIM5a proteins and HIV-1 Gag proteins. Since this experiment was performed under conditions where the late restriction activities were saturated by the over-expressed HIV-1 Gag, the slight reduction of p24 in the TRIM5aag lane (Fig 3A) did not reflect the level of restriction.

Simian TRIM5a orthologues form prominent cytoplasmic bodies
TRIM5a proteins self-associate to form cytoplasmic bodies, while dimerization is also required for efficiently binding to retroviral capsid [42,43]. Immunohistochemistry studies in 293T cells suggest that TRIM5arh efficiently form cytoplasmic bodies, while TRIM5ahu primarily displayed more diffused cytoplasmic localizations [19]. In order to address whether the varying late restriction activities of three simian TRIM5a proteins were due to their different subcellular localizations, we determine the localizations of simian TRIM proteins by immunostaining. 293T cells were transfected with 1.0 mg of TRIM5a-expressing plasmid. TRIM5a proteins were then detected with anti-HA antibodies and analyzed by confocal microscope. TRIM5ahu, which does not strongly affect HIV-1 production, showed predominantly diffuse cytoplasmic distribution with little discernible cytoplasmic bodies (Fig. 3B). In contrast, 59.8%, 52.8% and 52.7% of 293T cells expressing TRIM5arh, TRIM5aag and TRIM5acy, respectively, showed discrete cytoplasmic bodies (.3 discrete cytoplasmic body formations per cell) throughout the cytoplasm (Fig. 3 C-E). Our data therefore demonstrated similar subcellular localizations of TRIM5arh, TRIM5aag and TRIM5acy, despite their varying late restriction activities.
The RBCC sequences of TRIM5aag and TRIM5acy are essential for the late restriction activity TRIM5ahu amino acid sequences are the most divergent and showed the weakest late restriction activity among the tested TRIM5a orthologues (Fig. 2B). We therefore assessed whether the substitution with TRIM5ahu N-or C-terminal sequences can relieve the late restriction activities of the simian TRIM5a orthologues. Chimeric TRIM5a constructs were generated as depicted in Figure 4A. Simian TRIM5a proteins with N-terminal TRIM5ahu sequences showed no effect on HIV-1 (Fig. 4B, upper panel) or SIV MAC production (data not shown), underscoring the importance of the RBCC sequences in the late restriction activity against HIV-1. Although the TRIM5arh-TRIM5ahu chimera, R/H, showed a late restriction activity as potent as wild-type TRIM5arh, the two simian TRIM5a chimeras with human Nterminal sequences, A/H and C/H, showed stronger late restriction activities than wild-type TRIM5aag and TRIM5acy (Fig. 4B, upper panel). No prominent differences in protein expression levels were observed between the chimeric TRIM5a proteins (Fig. 4B, lower panel). These data indicate that the RBCC sequences of TRIM5aag and TRIM5acy are essential for the late restriction activity and suggest the possibility that the C-terminal regions of simian TRIM5aag and TRIM5acy proteins negatively regulate TRIM5a late restriction activities against HIV-1.

The influence of variations in the RBCC sequences of simian TRIM5a proteins on the late restriction activities
Previously, we demonstrated that introduction of two TRI-M5ahu-specific amino acid residues into TRIM5arh (M133T and T146A in the coiled-coil region) abrogates the late restriction activity of TRIM5arh [19]. In the RING, b-box 2 and partial coiled-coil domains, there are three amino acid differences between TRIM5arh and TRIM5aag, while only one residue separates TRIM5arh and TRIM5acy. In order to address the possible contributions of these residues on the modest late restriction activities of TRIM5aag and TRIM5acy, we introduced single amino-acid substitutions into the N-terminal half of TRIM5arh using corresponding TRIM5aag or TRIM5acy sequences (Fig. 5A). After verification of similar chimeric TRIM5a expression levels by immunoblot (Fig. 5B, lower panel), we examined respective late restriction activities against HIV-1. When HIV-1 NL4-3 was produced in the presence of TRIM5arh, TRIM5aag and TRIM5acy, HIV-1 titers were reduced approximately 60-, 7-and 6-folds, respectively, while the TRIM5arh mutants with single amino acid substitutions, Rh(H29P)Ag, Rh(K45E)Ag, Rh(G52E)Ag and Rh(M150L)Cy, showed potent late restriction activities, comparable to that of wild-type TRIM5arh (Fig. 5B, upper panel). These data indicate that variations in the single amino acid residues alone (29P, 45E, 52E or 150L) cannot explain the modest late restriction activities of TRIM5acy and TRIM5aag.

C-terminal sequences of TRIM5aag and TRIM5acy negatively regulates the potency of late restriction activity
To test whether TRIM5aag or TRIM5acy C-terminal sequences can impair the late restriction activity against HIV-1, we generated chimeric TRIM5a constructs as depicted in Figure 6A. Similar levels of TRIM5a expression were confirmed via immunoblot (Fig. 6B). Although the restriction activity of the TRIM5arh protein with N-terminal TRIM5acy sequence (RhM150LCy) did not notably differ from wild-type TRIM5arh restriction activities (Fig. 4B), the restriction against HIV-1 was relieved when C-terminal TRIM5acy sequences were fused with the N-terminal region of TRIM5arh (Fig. 6C). TRIM5aag Cterminal sequences in TRIM5arh (R/A) impaired the late restriction against HIV-1 by 3-fold when compared to wild-type TRIM5arh (Fig. 6C). These data suggest that C-terminal sequences of TRIM5a proteins can negatively regulate the late restriction activities, offering partial explanation to the modest late restriction activities of TRIM5aag and TRIM5acy against HIV-1.

Encapsidation and late restriction activities of human TRIM proteins
The 5a isoform of human TRIM protein had very little effect on HIV-1 production; however TRIM5ahu is one of over 80 members of the RBCC family of TRIM proteins. We therefore sought to examine whether other human TRIM family proteins may (1) be incorporated into HIV-1 VLPs and (2) restrict HIV-1 production. From the vast numbers of TRIM family proteins, we examined TRIM6, TRIM34 and TRIM22 since they are located in a paralogous cluster which includes TRIM5a [44]. We also examined the influence of TRIM1, TRIM18 and TRIM21 expression on HIV-1 VLP incorporation. TRIM1 has antiviral activity against N-tropic murine leukemia virus infection [6] and TRIM18 is its paralogue. TRIM21 can modulate TRIM5a ubiquitination [45] as well as the interferon-mediated antiviral response [46].
To obtain sufficient amounts of VLPs in the presence of TRIM proteins, we used pH-GP, which generates high levels of HIV-1 GagPol and abrogates the late restriction activities of TRIM5arh [16]. 293T cells were co-transfected with 1.0 mg of TRIMexpressing plasmids and 0.25 mg of codon-optimized HIV-1 GagPol-expression plasmid, pH-GP. Cell lysates and HIV-1 VLPs were harvested as described in materials and methods. Immunoblot analysis was performed to detect HA-tagged TRIM proteins as well as HIV-1 Gag proteins. All TRIM proteins were detected in the producer cells (Fig. 7A). Efficient VLP incorporation was evident with TRIM5arh, human TRIM21 and TRIM22 (Fig. 7A). Previously, we have demonstrated that HIV-1 Gag maturation delays in the presence of TRIM5a, resulting in accumulation of premature Gag proteins in producer cells and VLPs [16,17]. This was also true with human TRIM21 and TRIM22, where VLPs made in the presence of TRIM21 and TRIM22 showed notable accumulation of premature Gag proteins, particularly in VLPs (Fig. 7A). No HA signal was detected in the VLPs made in the presence of TRIM1, 6 and 34 (Fig. 7A). Intracellular expression levels of the TRIM proteins were not necessarily correlated with their incorporation into VLPs, indicating that the incorporation of TRIM proteins is not due to non-specific packaging of RBCC proteins into VLPs. These observations suggest the direct or indirect interaction of TRIM21 and TRIM22 with HIV-1 Gag proteins in producer cells. No effects on HIV-1 Gag levels in the presence of TRIM5a or TRIM22 are likely due to the saturation of late restriction activities by over-expression of codon optimized HIV-1 Gag.
Next, we examined the correlation between VLP incorporation status of human TRIM proteins and late restriction activities. To assess the antiviral activities of these TRIM proteins, 1.0 mg of TRIM-expressing plasmids were co-transfected with 0.1 mg of pNL4-3, p89.6, pROD10 or pSIVmac1A11 into 293T cells, and viral titers were determined in GHOST indicator cells. Of the human TRIM proteins that were assessed, only TRIM22 showed a slight reduction in NL4-3 and 89.6 titers, while the other TRIM proteins had no effect (Fig. 7B). None of the human TRIM proteins showed notable effects on HIV-2 or SIV MAC 1A11 production (Fig. 7B).

Discussion
In this report, we examined the late restriction activities and VLP encapsidation efficiencies of simian TRIM5a orthologues and related human TRIM proteins, and their late restriction activities against a panel of lentiviruses. Our results revealed the antiviral spectra and varying restriction activities of these TRIM proteins against lentiviral production. The relative expression levels or subcellular localizations of TRIM5a could not explain the encapsidation efficiency or the potency of late restriction activity. The RBCC domains of TRIM5aag and TRIM5acy were essential for the late restriction, while the C-terminal regions of TRIM5aag and TRIM5acy negatively regulated the restriction activities. Similar antiviral spectra between simian TRIM5a orthologues may suggest a conserved restriction mechanism among these proteins. Of the examined human TRIM5 paralogues, TRIM21 and TRIM22 were efficiently incorporated into HIV-1 VLP, while only TRIM22 showed marginal late restriction activity.
Among primate lentiviruses, HIV-2 and SIV MAC are closely related, as sooty mangabey SIV has transferred to humans and rhesus monkeys as HIV-2 and SIV MAC [47]. However, TRI-M5arh blocks the infection of HIV-2, but not SIV MAC , mainly due to the difference in the structure in the capsid protein that recruits cyclophilin A into HIV-1 virions [48]. Examination of HIV-2 and SIV MAC production in the presence of simian TRIM5a proteins demonstrated the most remarkable differences between the two closely related lentiviruses. Although SIV MAC was resistant to all four TRIM5a orthologues in the late-restriction, up to a 40-fold reduction in HIV-2 production was observed in the presence of TRIM5arh. It is possible that the variations in the CA loop regions of SIV MAC and HIV-2, which correspond to the HIV-1 CA cyclophilin-binding loop, may in part contribute to the differential late restriction phenotypes. However, the CA loop region cannot solely explain the resistance of SIV MAC to the TRIM5arh-mediated late restriction, because an HIV-1 cyclophilin-binding loop mutant with corresponding SIV MAC sequence was still sensitive to the late restriction [36]. Intriguingly, a recent study has shown that a single amino acid change in the HIV-1 CA cyclophilin-binding loop allowed the virus to escape from TRIM5arh-mediated post-entry restriction, suggesting the importance of the CA loop structure for restriction factor recognition [49]. Our previous study has also suggested that SIV MAC resists TRIM5arh-mediated late restriction by counteracting or saturating the TRIM5a late restriction machinery, rather than escaping TRIM5arh recognition altogether [36]. We therefore speculate that the sensitivity of HIV-2 to TRIM5arh-mediated late restriction is partly due to the relatively inefficient HIV-2 production from pROD10, where production of progeny virions may not be sufficient to overcome the TRIM5arh-mediated late restriction. The different replication kinetics of CMO2.41 and CMO2.5 in PBMCs can be attributed to the gag-pol region of CMO2.5, which is derived from a separate primary type-O isolate [50]. The differences in viral gag-pol sequences offers partial explanation as to why CMO2.41 and CMO2.5 responded differently to the late restriction activities of TRIM5aag (Fig. 2B).
The incorporation of TRIM5arh into the VLPs made with HIV-1 Gag suggests a specific interaction between TRIM5arh and HIV-1 Gag polyproteins [16]. Determinants for this interaction lie in the RING and coiled-coil domains of TRIM5arh, and the B30.2(PRYSPRY) motif is not required for the interaction or the late restriction activity of TRIM5arh [19]. TRIM5arh mutants with the M133T and/or T146A amino acid substitutions in TRIM5arh coiled-coil domain showed efficient encapsidation but impaired late restriction activity [19]. Although the RBCC sequences, including the M133 and T146, of three simian TRIM5a proteins are highly conserved among the three simian TRIM5a proteins, we found that TRIM5aag and TRIM5acy showed weaker late restriction activities than TRI-M5arh. Similar to our previous study, replacement of the Nterminal RBCC sequences of TRIM5aag and TRIM5acy with the corresponding TRIM5ahu sequences resulted in the loss of prominent late restriction effects of these two simian TRIM5a proteins, indicating the essential roles of the RBCC domains in their late restriction activities (Fig. 4B). Unexpectedly, C-terminal TRIM5ahu sequences mildly strengthened the late restriction activities of TRIM5aag and TRIM5acy; while impaired late restriction activities were observed when the N-terminal region of TRIM5arh was fused with C-terminal TRIM5aag or TRIM5acy sequences (Fig. 6C). These results indicate that TRIM5a RBCC sequences are required for the late restriction, and C-terminal amino acid sequences can modulate the potency of the late restriction against HIV-1, adding further complexity to the mechanisms of TRIM5a-mediated late restriction. Further studies will determine the extent of inter-domain communication within the TRIM5a protein and the amino acid residues which are responsible for the negative regulation.
Proteins that have a conserved RING, b-box 1 and/or b-box 2 and coiled-coil domains are included in the superfamily of TRIM genes, and many TRIM proteins have been implicated to be interferon-responsive and a contributing factor in the defense against infectious agents [42,51]. Located in the same genetic locus on chromosome 11p15, TRIM5, TRIM22, TRIM6 and TRIM34 were classified in the same clade [22]. Of the human TRIM proteins assessed in this study, TRIM21 and TRIM22 were efficiently incorporated into VLPs. A faint TRIM18 signal was also detected in the purified VLPs, suggesting weak interaction between human TRIM18 and HIV-1 Gag (Fig. 7A). In contrast to the efficient encapsidation of these proteins, only TRIM22 showed a modest late restriction activity against HIV-1 ( Fig. 7B and C). These data suggest that encapsidation efficiency alone could not fully explain the differences in the late restriction activities of TRIM proteins. It is likely that additional determinant(s) controls the potency of the late restriction activity following initial binding of a TRIM protein with HIV-1 Gag.
In summary, we demonstrated the examples of TRIM proteinmediated late restriction activities and their potential to interact with viral proteins. Our data provide further insights into the complex host-pathogen interplay in TRIM protein-mediated late restriction.