Control of Viral Infectivity by Tripartite Motif Proteins

Abstract

It is of great interest to understand the molecular details of the pathways that constitute species barriers to viral infection. The tripartite motif protein TRIM5α has emerged as an important mediator of species-specific retroviral replication and innate immunity. This review considers the role of TRIM5α as an antiviral protein in mammals. The methods used to identify species-specific restriction to retroviral infection, and the identification of TRIM5α itself, are outlined. TRIM5α mediates an early postentry block to sensitive retroviral infection, usually before viral DNA synthesis. Results from mutational analysis of TRIM5α and their contribution to a mechanistic model for TRIM5α antiviral activity are discussed. The antiviral role of other TRIM proteins is considered, as is the role of TRIM5α cytoplasmic bodies.

INTRODUCTION

Retroviral infections are responsible for significant disease in mammals, causing a variety of pathologies including immune deficiency, malignancies, and neurological and immunological symptoms. They are not strictly species specific and on rare occasions are able to jump from one species to another, a process known as zoonosis. If a zoonotic virus is able to replicate in its new host and infect further individuals it can cause widespread disease. Human immunodeficiency viruses HIV-1 and HIV-2 have transferred into humans from chimpanzees and sooty mangabeys respectively (Gao et al., 1992, 1999). Neither virus causes disease in its simian host but both cause acquired immunodeficiency syndrome (AIDS) in humans. Further examples of zoonotic viruses are the coronavirus from Chinese civet cats, which causes severe acute respiratory syndrome (Li et al., 2005), as well as avian influenzaviruses that periodically cause influenza pandemics (Li et al., 2004). It is of great interest to understand how mammalian species barriers ensure that zoonotic viral transfer is rare and to identify the molecular mechanisms involved. It is also important to understand how viruses bypass species barriers as this information might aid the design of novel antiviral therapeutics as well as help predict which viruses pose the greatest future threat to human health.

Phylogenetic analysis of HIV-1 sequences from infected individuals has revealed three HIV-1 groups termed M (main/major), O (outlier/other), and N (non-M, non-O/new), each resulting from an independent zoonosis into humans (Sharp et al., 2001). Similarly, HIV-2 sequences reveal eight independent zoonotic transfers of SIVsm (simian immunodeficiency virus from sooty mangabeys) from sooty mangabeys into humans (Apetrei et al., 2005). Remarkably, of these 11 transfers, only that which led to the HIV-1 M group of sequences is responsible for the AIDS pandemic, currently numbering about 40 million infected individuals and more then 20 million casualties (Marais, 2004). HIV-1 groups O and N are epidemiologically unsuccessful and have led to a relatively small number of cases, mostly in West Africa. HIV-2 sequence types C to H each comprise a single sequence and almost certainly represent dead-end infections that have not been able to spread in humans. It is likely that many more viruses have transferred into humans, been unable to spread, and remain undetected. These observations illustrate the power of species barriers that protect us from viral infections from our simian cousins as well as from more distantly related animals.

Work on host factors that control viral permissivity has focused attention on intracellular antiviral proteins that have been referred to as restriction factors (Goff, 2004a,b). These proteins are part of an innate or intrinsic immune system that actively blocks viral infection in a species-specific way (Bieniasz, 2004). Certain members of a family of deaminase enzymes, exemplified by human APOBEC3G (apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3G), are able to deaminate viral nucleic acids, leading to hypermutation and abortive infection (Sheehy et al., 2002). The sole function of primate lentiviral Vif proteins appears to be to recruit APOBECs to the proteasome and allow successful viral replication (Harris et al., 2003; Mangeat et al., 2003; Sheehy et al., 2003). A second family of antiviral proteins is defined by a tripartite motif and exemplified by the tripartite motif protein TRIM5α (Stremlau et al., 2004). This review focuses on identification of TRIM5α as an important antiviral protein in primates, on its mechanism of action, and on whether TRIM5α is unique within the TRIM family as having antiviral activity.

ASSAY FOR VIRAL INFECTIVITY

The science of gene therapy has driven the design of increasingly sophisticated viral vector systems for the delivery of therapeutic genes to patient cells (Sinn et al., 2005). Retroviral vectors, derived from a range of retroviruses, have been useful for the study of the control of viral permissivity (Fig. 1). The vectors can be coated or “pseudotyped” with the envelope protein from vesicular stomatitis virus, thereby obviating specific envelope/receptor requirements and allowing infection of almost all cell lines tested. They can be engineered to encode green fluorescent protein and retroviral infection can be measured by fluorescence-activated cell sorting (FACS). Cell lines from a range of species have been tested for permissivity and the results are summarized in Table 1 (Hoffman et al., 1999; Towers et al., 2000; Besnier et al., 2002; Cowan et al., 2002; Ikeda et al., 2002; Munk et al., 2002; Hatziioannou et al., 2003; Ylinen et al., 2005). In each case BALB/c (B)-tropic murine leukemia virus tagged with green fluorescent protein (MLV-B GFP) is infectious, allowing the demonstration that the cells are specifically able to block infection of the restriction-sensitive viruses.

FIG. 1.

An assay for retroviral infectivity. Plasmids encoding a gag-pol expression cassette, a VSV-G (envelope protein from vesicular stomatitis virus) expression cassette, and a retroviral vector encoding GFP are transfected into 293T cells. Forty-eight hours later the supernatant is collected, filtered, and serially diluted. These dilutions are used to infect cells plated in six-well plates in the presence of Polybrene (5 μg/ml). Forty-eight hours later the green cells are enumerated by FACS and the percentage infection is plotted against the viral input dose, measured in reverse transcriptase units or infectious units per milliliter as determined with an indicator cell line.

Table 1.

Retroviral Restrictions

Species	Restriction-sensitive viruses	Restriction saturable	TRIM5α required
Human (Ref1)	MLV-N, E1AV	Yes	Yes
African green monkey (Lv1)	MLV-N, HIV-1, HIV-2, SIVmac, EIAV	Yes	Yes
Owl monkey	HIV-1	Yes	Yes (TRIM-Cyp)
Rhesus macaque	HIV-1, HIV-2, MLV-N, SIVagm	Yes	Yes
Squirrel monkey	SIVmac	No	Yes
Cynomolgus monkey	HIV-1	Untested	Yes
Rabbit	HIV-1, HIV-2	No	Untested
Bats, cows	MLV-N	Untested	Untested
Pigs	MLV-N	Untested	Untested
Chimpanzeea	MLV-N	Untested	Yes
Spider monkeya	HIV-1, SIVagm	Untested	Yes
Orangutana	MLV-N	Untested	Yes
Tamarina	SIVmac	Untested	Yes

Abbreviations: EIAV, equine infectious anemia virus; HIV-1 and HIV-2, human immunodeficiency virus types 1 and 2; N-MLV, N-tropic murine leukemia virus; SIVagm, simian immunodeficiency virus from African green monkeys; SIVmac, simian immunodeficiency virus from rhesus macaques; TRIM5α, tripartite motif protein 5α; TRIM-Cyp, TRIM with cyclophilin A insertion.

^aRestriction specificities were determined in permissive cells overexpressing the TRIM5α alleles specified. In these cases experiments have not been performed in cell lines from the species indicated.

SATURATION OR ABROGATION ASSAYS REVEAL DOMINANT RESTRICTION FACTORS IN A RANGE OF SPECIES

Cells might be nonpermissive to retroviral infection for one of two reasons. It might be that the cells lack a factor that the restricted virus requires for the completion of its life cycle, or it might be that the cells express a factor that specifically blocks the incoming restricted virus. To differentiate between these two possibilities, saturation or “abrogation” assays have been used (Boone et al., 1990; Besnier et al., 2002; Cowan et al., 2002; Towers et al., 2002; Hatziioannou et al., 2003). In these experiments virus-like particles (VLPs; in general, retroviral vector encoding puromycin resistance, rather than GFP) were titrated onto nonpermissive cell lines in the presence of a fixed low dose of virus encoding GFP. If the GFP-encoding virus became more infectious in the presence of the VLPs, in a VLP concentration-dependent way, then a dominant antiviral factor is present that the VLPs titrate, making the cells more permissive to the GFP-encoding virus. These experiments revealed that in most cases, low permissivity to retroviral infection could indeed be titrated by VLPs, indicating that antiviral factors are common in mammalian cell lines. Nonsaturable examples include HIV-1 in rabbit cells and SIVmac (simian immunodeficiency virus from rhesus macaques) in squirrel monkey cells. In squirrel monkey cells, squirrel monkey TRIM5α is responsible for SIVmac restriction but the rabbit restriction factor has not yet been identified (Besnier et al., 2002; Ylinen et al., 2005). Where saturation occurs, only VLPs that are sensitive to the particular restriction factor are able to saturate restriction, suggesting that a sensitive virus can interact with the antiviral factor, whereas an insensitive virus cannot. The antiviral factors have been termed restriction factors; the human factor is named Ref1 and the simian factor is named Lv1 (Towers et al., 2000; Cowan et al., 2002). Saturation assays have also been used effectively to demonstrate that in species in which infection by a number of unrelated viruses is blocked, such as humans and African green monkeys, saturation with VLPs from any of the sensitive viruses can render the cells permissive to restricted virus (Hatziioannou et al., 2003). This observation strongly suggested that there was a single factor, now known to be TRIM5α, that was able to recognize broadly unrelated viruses. The blocks to infection are realized early after infection, usually resulting in a strong inhibition of viral DNA synthesis. However, squirrel monkey TRIM5α can block SIVmac infection without blocking DNA synthesis (Ylinen et al., 2005), and this is discussed further below.

TRIM5α IS A RESTRICTION FACTOR IN PRIMATES

The identification of TRIM5α as a potent restriction factor, active against HIV-1 in rhesus macaques (Stremlau et al., 2004), revolutionized the study of the control of retroviral permissivity. On the basis that rhesus macaque cells were poorly permissive to infection by HIV-1, Stremlau screened a rhesus macaque cDNA library for cDNAs that could block HIV-1 GFP infectivity. Cells expressing the cDNA library were repeatedly infected with HIV-1 GFP and the infected cells discarded. The resultant HIV-1 resistant clones were found to encode rhesus TRIM5α. When TRIM5α was expressed in human HeLa cells they became nonpermissive to HIV-1 GFP infection but permissivity to Moloney MLV was unaffected. On publication of this work a number of groups immediately set out to test whether human and simian TRIM5α alleles could explain the examples of poor species-specific retroviral infectivity summarized in Table 1. It was quickly shown that the human TRIM5α gene was indeed responsible for the low titer of MLV-N in human cells and for the low titer of HIV-1, SIVmac, equine infectious anemia virus (EIAV), and MLV-N in cells from African green monkeys (Hatziioannou et al., 2004a; Keckesova et al., 2004; Perron et al., 2004; Yap et al., 2004). Restricted virus infectivity was rescued by specifically reducing TRIM5α expression, using small interfering RNA (siRNA). Furthermore, expression of TRIM5α in permissive cells such as feline CRFK cells or murine MDTF cells rendered them able to restrict viral infection with the expected specificities. More recently, TRIM5α has been shown to explain restriction of SIVmac in cells from squirrel monkeys and of HIV-2 in cells from rhesus macaques (Song et al., 2005; Ylinen et al., 2005).

Tripartite motif proteins, also known as RBCC proteins, are named for a series of three motifs with a characteristic order and spacing (Reymond et al., 2001). They comprise a RING (really interesting new gene) domain encoding two atypical zinc fingers, either one or two B-box 2 domains, and a coiled coil domain. Some TRIM proteins, including TRIM5α, additionally contain a SPRY domain, also found in members of the immunoglobulin superfamily. The B-box, coiled coil, and SPRY domains are likely to function in protein–protein interactions. Only the TRIM5α splice variant, the longest and only one containing a SPRY domain, is able to block retroviral infection, indicating the importance of the SPRY domain for restriction (Stremlau et al., 2004).

MUTATIONAL ANALYSIS OF TRIM5α

Sequence comparison between TRIM5α alleles from different primates suggested that variable regions in the SPRY domain might be responsible for their differential species-specific antiviral properties. A number of groups performed mutational analyses on TRIM5α, swapping sections of the SPRY domains between human and simian TRIM5α genes and testing the resultant fusions for antiviral activity (Nakayama et al., 2005; Perez-Caballero et al., 2005; Sawyer et al., 2005; Stremlau et al., 2005; Yap et al., 2005). Together these studies concluded that a variable section of the SPRY domain, between approximately residues 320 and 345, was responsible for much of the species-specific restriction activity of TRIM5, although residues in the coiled coil were also important, particularly for restriction of MLV (Yap et al., 2005). Notably, changing residue 332 in the SPRY domain of human TRIM5 to the corresponding residue in rhesus and African green monkey TRIM5, R332P, enabled the human protein to restrict HIV-1, although not as strongly as the wild-type rhesus protein (Stremlau et al., 2005; Yap et al., 2005). Interestingly, this mutant was also able to strongly restrict SIVmac, a property not shared by either of the wild-type TRIM5α proteins (Stremlau et al., 2005). This result is reminiscent of the observation that combining the antiviral determinants of the murine restriction factor Fv1 can result in an Fv1 mutant with novel anti-MLV specificities (Bock et al., 2000). The corresponding TRIM5α proline-to-arginine mutation in the rhesus protein also strengthened its ability to restrict HIV-1, further confirming the importance of this position in antiviral specificity (Yap et al., 2005). In certain cases it is also possible to fuse the SPRY domain from one TRIM to the RBCC domain of a second TRIM and confer the antiviral specificity of the first to the fusion protein (Yap et al., 2005). Together these observations indicate that the SPRY domain encodes the determinants of antiviral specificity.

TRIM5 is multiply spliced, resulting in at least three isoforms, each increasingly shorter from the C terminus. The TRIM5γ isoform, which lacks the C-terminal SPRY domain, has dominant negative activity against the antiviral activity of TRIM5α (Stremlau et al., 2004). This activity is mediated through the coiled coil domain, which is necessary and sufficient for dominant negative activity (Perez-Caballero et al., 2005). Furthermore, dominant negative activity has been shown to correlate with the ability to multimerize, in a yeast two-hybrid assay, and it therefore seems likely that molecules with a coiled coil, but without a SPRY domain, can heterodimerize with TRIM5α to create a complex with a single SPRY domain that is unable to restrict infection (Perez-Caballero et al., 2005). This hypothesis holds that multivalent interactions are required between the TRIM5α complex and the virus for effective restriction. Indeed, one study has shown that capsid protein from restriction-sensitive MLV-N, but not from restriction-insensitive MLV-B, can be copurified with TRIM5α in a SPRY domain-dependent way (Sebastian and Luban, 2005). This interaction is apparent only when using detergent-stripped virions, suggesting that the interaction occurs between TRIM5α multimers and capsid multimers in the context of the intact virion core. Both lentiviral (HIV-1) and gammaretroviral (MLV) cores are made up of similar arrays of hexameric rings of capsid monomers (Li et al., 2000; Mortuza et al., 2004), supporting a similar mechanism for restriction of viruses unrelated at the sequence level. Whether multiple interactions are required because the TRIM5α is required to bind and cross-link multiple hexamers or whether this requirement simply reflects the increased affinity of multiple binding sites remains unclear.

The TRIM5 RING domain has been shown to act as a ligase for autoubiquitination in vitro in the context of the TRIM5δ splice variant (Xu et al., 2003). This observation suggests that restriction might involve ubiquitination of the incoming capsid. Ubiquitination can have complex consequences and may not simply lead to recruitment to the proteasome (Weissman, 2001). Ubiquitination of incoming core could lead to mislocalization, alteration of conformation, or simply perturbation of capsid rearrangement. However, whether the TRIM5α RING domain plays a role in restriction remains unclear. TRIM5α with the critical RING domain zinc-coordinating cysteine residues mutated clearly has reduced antiviral activity, although this mutant retains significant antiviral activity when overexpressed (Stremlau et al., 2004; Javanbakht et al., 2005). The role of the RING domain therefore remains uncertain. However, it is required for maximal restriction and we must be careful in assuming that it is redundant, because TRIM5α RING mutants retain some anti-viral activity when overexpressed. It is also possible, however, that the TRIM5α RING domain is required for a function unrelated to TRIM5α antiviral activity.

VIRAL DETERMINANT FOR TRIM5α RESTRICTION IS IN THE CAPSID

That the viral determinant for TRIM5α restriction is in the capsid was originally shown by differential restriction of MLV-N, which is restricted in a range of species, and the closely related MLV-B, which is not (Towers et al., 2000; Besnier et al., 2003). This difference in tropism depends on a single amino acid in the MLV capsid: arginine in MLV-N promotes sensitivity to TRIM5α restriction, whereas glutamate renders MLV-B insensitive to restriction. TRIM5α sensitivity determinants are also found in the HIV-1 capsid and mutations to residues on the capsid surface alter HIV-1 titer in a species-specific way (Kootstra et al., 2003; Hatziioannou et al., 2004b; Ikeda et al., 2004). More recent work considering HIV-2 and SIVmac has shown that changes in the external loop of the SIVmac capsid, homologous to the HIV-1 cyclophilin-binding site, can render SIVmac insensitive to TRIM5α-mediated restriction in squirrel monkey cells, while retaining its insensitivity to human and rhesus TRIM5αs (Ylinen et al., 2005).

MODEL FOR RESTRICTION OF RETROVIRAL INFECTIVITY

The experimental data can be used to propose a mechanism for TRIM5α-mediated restriction. Mapping the viral and TRIM5α sensitivity determinants to the capsid (CA) and SPRY domains, respectively, suggests that the SPRY domain interacts directly with the incoming viral CA. What happens after TRIM5α–CA interactions is unclear. It is reasonable to assume that, simply by interacting with multiple CA molecules in the incoming core, TRIM5α can perturb the continuation of the infectious cycle. MLV and HIV-1 are known to have different uncoating requirements, with HIV-1 shedding most of its CA early after infection and MLV holding on to its CA during DNA synthesis (Fassati and Goff, 1999, 2001). However, it is likely that both viruses need to rearrange their viral core, possibly after specific interaction with host factors, before reverse transcription can efficiently proceed. In support of a role for CA in reverse transcription there are a number of viral CA mutants that form apparently normal particles but are unable to reverse transcribe (Craven et al., 1995; Alin and Goff, 1996; Cairns and Craven, 2001). One explanation for this phenotype is that the mutant capsids cannot efficiently perform the rearrangements required for reverse transcription.

Interaction with TRIM5α, either directly or indirectly, leads to a block to reverse transcription in most cases. However, squirrel monkey TRIM5α, active against SIVmac, does not block SIVmac DNA synthesis (Ylinen et al., 2005). Although this observation might suggest a later interaction with the virus, after DNA synthesis, it seems likely that squirrel monkey TRIM5α interacts with virus early after infection but that the perturbation of capsid activity is subtle enough to disrupt viral infectivity, without significantly affecting reverse transcription.

ROLE OF CYTOPLASMIC BODIES REMAINS UNCLEAR

Certain TRIM proteins including TRIM5α have been reported to exist in discrete areas of the cytoplasm, known as cytoplasmic bodies, when exogenously expressed (Reymond et al., 2001). Whether endogenous TRIM5α forms cytoplasmic bodies is unclear, but it appears that bodies are not required for TRIM5α antiviral activity (Perez-Caballero et al., 2005). The issue is complicated by the fact that many proteins, including TRIM5α, form bodies called aggresomes when they are over-expressed (Kopito, 2000). Differentiating between true cytoplasmic bodies and aggresomes is challenging as they may be functionally related. Aggresomes are not simply spontaneously forming protein aggregates, as illustrated by their involvement in the regulation of the cellular enzyme inducible nitric oxide synthetase (Kolodziejska et al., 2005). It will be interesting to look for endogenous TRIM5α cytoplasmic bodies, using TRIM5α-specific antibodies, and to determine the effects of manipulations that affect restriction such as viral infection and drugs such as cyclosporin A.

ROLE FOR OTHER TRIM PROTEINS IN VIRAL PERMISSIVITY

The human genome encodes about 60 TRIM genes and it is obviously of great interest to consider whether there are more TRIM proteins with antiviral activity. There is much circumstantial evidence supporting an antiviral role for PML, the most well-studied TRIM protein, otherwise known as TRIM19. PML is found fused to retinoic acid receptor (RAR) by chromosome translocations in a series of malignancies including promyelocytic leukemia, after which it is named. The PML–RAR fusion deregulates both PML and retinoic acid receptor function and causes the malignant phenotype (Zelent et al., 2001). In the nucleus PML exists in bodies called PML or ND10 bodies (Negorev and Maul, 2001). The cellular functions of the bodies and of PML are not completely solved but PML clearly has roles in the control of apoptosis, transcription, DNA repair, chromatin metabolism, and cell signaling (Borden, 2002; Lin et al., 2004). Many studies have shown that replicating viral DNA tends to associate with PML bodies and that viral immediate-early gene transcription takes place there (Everett, 2001). Furthermore, the bodies are often disrupted by viral infection; usually the association between virus and the bodies is best seen with viral mutants that do not affect them. The degradation of PML by herpes simplex virus type 1 (HSV-1) infection has been well characterized. HSV-1-encoded E3 ubiquitin ligase ICP0 catalyzes PML ubiquitinylation and its recruitment to the proteasome (Chelbi-Alix and de Thé, 1999; Maul et al., 2003). These observations imply that PML has antiviral activity and that certain viruses need to degrade or move PML in order to complete their life cycle. Further circumstantial evidence of an antiviral role for PML is provided in a number of studies including reduced permissivity to vesicular stomatitis virus and influenza A of hamster cells overexpressing PML. Moreover, cells derived from PML knockout mice have an impaired antiviral interferon response to HSV-1 infection, suggesting that PML is an essential component of the interferon response in these cells (Chee et al., 2003). However, absolute proof of an antiviral role for PML remains elusive and the issue is controversial. An elegant study has shown that PML bodies form in association with transcriptionally active HSV-1 nucleoprotein complexes shortly after their arrival in the nucleus of an infected cell (Everett and Murray, 2005). The HSV-1 protein ICP0 then disrupts PML bodies associated with HSV-1 complexes as well as unassociated PML bodies. This work shows that the cell can deposit PML body proteins at specific sites in response to viral replication. Whether this is a defensive act remains unclear. PML has also been suggested to have anti-HIV-1 activity in human cells, although this observation is controversial (Turelli et al., 2001).

As yet there is no strong evidence of further TRIM genes having antiviral roles. TRIM1 genes from human, African green monkey, and owl monkey cells have been reported to weakly restrict MLV, but only by a few fold, and it remains to be seen whether such restriction might be strong enough to be protective (Yap et al., 2004).

ROLE OF CYCLOPHILIN A IN RESTRICTION

An interesting twist in the TRIM5 tale was provided by identification of the factor responsible for the low titer of HIV-1 in owl monkey cells (Nisole et al., 2004; Sayah et al., 2004). Owl monkeys are New World monkeys and an exception to the general rule that New World monkey cells are permissive to HIV-1 infection (Hofmann et al., 1999). In cells from owl monkeys HIV-1 can be rendered insensitive to restriction by preventing interactions between the HIV-1 capsid protein and the immunophilin molecule cyclophilin A (CypA) (Towers et al., 2003). Immunophilins comprise three families of abundant enzymes able to isomerize proline residues. Examples of peptidyl prolyl isomerase function suggest a role in the control of protein function. Isomerization of prolines by Pin1 has been associated with modulation of phosphorylation status, protein–protein interactions, subcellular localization, and protein stability (Ryo et al., 2003). HIV-1 and CypA have a long history. CypA was identified in a yeast two-hybrid assay as a binding partner for HIV-1 Gag (Luban et al., 1993). It was later shown that this interaction led to CypA being incorporated into the HIV-1 virions and mutations that blocked the interaction, as well as drugs that competed the CypA away from HIV-1, lowered HIV-1 infectivity in human cells (Franke et al., 1994a,b; Thali et al., 1994). It was assumed that HIV-1 capsids required CypA in order to fold properly or to properly uncoat and efficiently reverse transcribe. Prior to the identification of TRIM5α as encoding Ref1/Lv1 a role for restriction was hypothesized for the anti-HIV-1 effects of cyclosporin A (CSA) in human cells (Towers et al., 2003). However, the role of human TRIM5α, if any, in the effects of CSA remains unclear.

The most important CA–CypA interactions take place after the virus has entered the target cells, compared with those that incorporate CypA into HIV-1 particles (Kootstra et al., 2003; Towers et al., 2003; Sokolskaja et al., 2004; Hatziioannou et al., 2005). Experiments in owl monkeys showed that the effects of CypA were also important in monkeys, but in this case CypA–CA interactions were inhibitory. The same mutations or drugs that decreased titer in human cells completely rescued restricted infectivity in owl monkey cells to high titer (Towers et al., 2003). These observations were followed by those of Sayah et al., who were intrigued by the observation that whereas knocking down CypA expression with siRNA rescued HIV-1 infectivity in owl monkey cells, replacing the CypA with an siRNA-insensitive CypA did not return the cells to their nonpermissive state (Sayah et al., 2004). This observation implied that there was a second CypA-related protein in owl monkey cells that was essential for restriction of HIV-1. Sayah and coworkers identified a fusion protein between TRIM5 and CypA, expressed uniquely in owl monkeys, that explained the sensitivity of HIV-1 to manipulation of CA–CypA interactions. It appears that a CypA pseudogene had inserted into exon 7 of the TRIM5 gene by retrotransposition to produce the TRIM–Cyp allele. This finding was quickly followed by Nisole et al., who found the same TRIM–Cyp fusion in owl monkey cells and demonstrated that TRIM–Cyp binds the HIV-1 capsid through its CypA domain in vitro (Nisole et al., 2004). Owl monkeys appear to be unique in encoding a TRIM–Cyp fusion. The CypA pseudogene replaces the SPRY domain and presumably has the same function in bringing the RBCC domain into contact with the incoming HIV-1 core. The presence of the CypA domain therefore explains the absolute sensitivity of HIV-1 restriction in owl monkey cells to CSA and mutations such as HIV-1 CA G89V, which prevent CypA, and therefore TRIM–Cyp, from interacting with the HIV-1 capsid. It is intriguing, and as yet unexplained, that restricted HIV-1 infectivity is also rescued in Old World monkey cells such as African green monkey CV1 cells and rhesus FRhK4 cells by CSA treatment (Ikeda et al., 2004). The reason for this TRIM–Cyp-independent CSA sensitivity is likely to be mechanistically important and warrants further investigation.

FUTURE FOR POSTENTRY RESTRICTION RESEARCH

There remain a number of examples of poor permissivity to retroviral infection for which the factor(s) responsible have not yet been identified. The Lv2 phenotype (Schmitz et al., 2004) and the restriction of certain lentiviruses in HeLa cells (Sokolskja et al., 2004; Hatziioannou et al., 2005) are clearly TRIM5 independent. Also, rodent cells (Hatziioannou et al., 2004c), rabbit cells (Hofmann et al., 1999; Besnier et al., 2002), and bat, cow, and pig cells (Towers et al., 2000) restrict retroviral infection, possibly through TRIM5. MLV-N restriction in hamster and dog cells (Towers et al., 2000) has not survived closer examination (G.J. Towers, unpublished observations). The factors responsible for these restriction phenotypes are likely to be identified soon and they may swell the ranks of antiviral TRIM proteins.

The fact that HIV-2 can be restricted by human TRIM5 when overexpressed suggests that it may play a role in controlling HIV-2 infection in humans (Ylinen et al., 2005). Further analysis of these observations with primary cells and virus is likely to be informative. It will be interesting to characterize TRIM5 polymorphisms in the human population and to determine whether these could have any impact on HIV-1 or HIV-2 replication. The role of cytoplasmic bodies, if any, in restriction is an interesting question, as is whether TRIM5 has functions unrelated to its antiviral properties.

As well as being a cause of disease, retroviruses have shown great promise in the delivery of therapeutic genes in gene therapy protocols. A significant hurdle in successful gene therapy is the efficient delivery of the therapeutic gene to the correct target cells. Often the most important target cells, for example, stem cells, have low permissivity to retroviral infection. A better understanding of the host–virus interactions that determine permissivity will enable the design of retroviral vectors with higher infectivities in relevant cell types. Furthermore, understanding the determinants of species-specific retroviral replication will enable us to improve animal models for AIDS. Current models comprise rhesus macaques infected with simian immunodeficiency virus (SIVmac) and its derivatives as HIV-1 is strongly restricted by both rhesus TRIM5α and APOBEC3G. The current models have been successful in understanding the immunology of AIDS (Mattapallil et al., 2005), but their use in testing vaccines and anti-HIV therapeutics is limited. Altering HIV-1 so that it can replicate and cause disease in rhesus macaques would be expected to significantly improve the AIDS animal model. The study of host factors controlling viral permissivity is therefore likely to provide information that is both interesting and useful to a broad audience.