Restriction of Foamy Viruses by Primate Trim5{alpha}

Foamy viruses (FVs) are unconventional retroviruses with a replicationstrategy that is significantly different from orthoretrovirusesand bears some homology to that of hepadnaviruses. Althoughsome cellular proteins, such as APOBEC3, have been reportedto block FVs, no restriction by Trim5 ${alpha}$ has been described todate. The sensitivity of three FV isolates of human-chimpanzeeor prototypic (PFV), macaque (SFVmac), and feline (FFV) originto a variety of primate Trim5 ${alpha}$ s was therefore tested. PFV andSFVmac were restricted by Trim5 ${alpha}$ s from most New World monkeys,but not from other primates, whereas FFV-based vectors wererestricted by Trim5 ${alpha}$ s from the great apes gorilla and orangutan.Trim5 ${alpha}$ s from Old World monkeys did not restrict any FV isolatetested. Capuchin Trim5 ${alpha}$ was unique, as it restricted SFVmac andFFV but not PFV. Trim5 ${alpha}$ specificity for FVs was determined bythe B30.2 domain, interestingly involving, in some instances,the same residues of the variable regions previously implicatedas major determinants for human immunodeficiency virus type1 restriction. FVs with chimeric Gags were made to map the viraldeterminants of sensitivity to restriction. The N-terminal halfof the Gag molecule was found to contain the regions that controlsusceptibility. This region most likely corresponds to the capsidof conventional retroviruses. Due to their unique replicationstrategy, FVs should provide a valuable new system to examinethe mechanism of retroviral restriction by Trim5 ${alpha}$ .

INTRODUCTION

Top

INTRODUCTION

Foamy viruses (FVs), also known as spumaviruses, are the solemembers of the Spumaretrovirinae subfamily of retroviruses.They have been isolated from a wide range of mammalian hosts,including primates, cats, cows, and horses, and are characterizedin vitro by a very broad tropism, allowing the infection ofa great variety of different cell types and species (reviewedin reference 32). In cell culture, FV replication often resultsin the formation of vacuolated, multinucleated syncytia witha characteristic foamy appearance (53). Nevertheless, they haveyet to be strongly associated with any disease in vivo.

FVs display a number of remarkable differences in their lifecycles compared to conventional retroviruses. Interestingly,some of the unique features of the FV replication strategy showstrong homology to those of hepatitis B virus. For example,unlike other members of the retrovirus family, reverse transcriptionof FVs occurs predominantly in producer cells, not targets.Although FV particles contain both viral RNA and DNA at a ratioof about 5:1, studies on the kinetics of reverse transcription,together with those on the rescue of infectious virus aftercell transfection, strongly suggest that the functionally relevantnucleic acid within virions is DNA (36, 51, 70, 71). Furthermore,the FV Pol protein is expressed as a separate protein insteadof the Gag-Pol fusion protein found in other retroviruses (6,15, 24, 29, 70). The FV Gag protein is unusual because it isnot clearly divided and processed into matrix (MA), capsid (CA),and nucleocapsid (NC) subunits, as is observed for conventionalretroviruses. During particle assembly and release, proteolyticprocessing of FV Gag at a single C-terminal site seems to takeplace, resulting in FV virions composed of the Gag precursor(p71Gag) and its processing product (p68Gag) (reviewed in reference17). Further, FV particles resemble the immature forms of conventionalretroviruses, suggesting the absence of Gag cleavage and consequentCA rearrangement in the extracellular phase of the life cycle(37). However, additional secondary FV Gag processing sitesby the FV protease have been identified with recombinant proteinsand peptides in vitro, and mutagenesis of these sites in theprovirus suggests that they play an important role during theearly steps in FV replication, involving a cellular and FV protease-dependentdisassembly pathway during entry into the target cell (27, 42).The major homology region and zinc finger(s), which are foundin CA and NC, respectively, in conventional retroviruses, arealso absent in FV Gag (31). Finally, the requirement of FV Envcoexpression for Gag membrane association and budding suggeststhe absence of a membrane-targeting domain in FV Gag that isnormally located in the retroviral MA subunit (2, 16, 45).

Retroviral infection can be restricted by a number of host cellfactors, some of which may play a role in the innate immunesystem. Two major types of restriction factors that act earlyin the viral life cycle have previously been described (40).The APOBEC family of deaminases acts predominantly on viralnucleic acids, while Fv1 (Friend virus susceptibility factor1)/Trim5 ${alpha}$ targets the retroviral CA. One of the first restrictionfactors to be described was the Fv1 gene, which confers resistancein mice against the murine leukemia virus (MLV) (28). Fv1 restrictionis dependent on the MLV CA, where a single amino acid changeat residue 110 can alter the susceptibility of the virus torestriction (26). It blocked virus replication at a stage thatwas post-reverse transcription but before integration (46).Fv1 appears to be derived from the Gag protein of a class ofendogenous retrovirus known as MuERV-L (4). A second retroviralCA-dependent restriction activity that arrests virus replicationprior to reverse transcription and affects a number of retroviruses,including human immunodeficiency virus type 1 (HIV-1), simianimmunodeficiency virus (SIV), MLV, and equine infectious anemiavirus, was subsequently reported (3, 7, 19). Trim5 ${alpha}$ , a memberof the tripartite motif (Trim) family of genes, was shown tobe responsible for this activity (20, 25, 41, 58, 67). The tripartitemotif comprises the RING, B-BOX, and coiled-coil (RBCC) domains(50). In addition, Trim5 ${alpha}$ also possesses a B30.2 domain at theC terminus. It is this domain that is mainly responsible forretroviral CA recognition (38, 54, 58, 68) and binding (55,59). There is no similarity between Fv1 and Trim5 at the sequencelevel, but their functional organization is remarkably similar,with both having a multimerization domain preceding a retroviralCA-binding domain (66). Artificial restriction factors thatrestrict HIV-1 can be constructed by fusing a multimerizationdomain to cyclophilin A, which binds HIV-1 CA (23, 66).

The APOBEC family members have been shown to act on a wide rangeof elements, including lentiviruses, MLVs, FVs, and retrotransposons,as well as the hepadnavirus hepatitis B virus and the parvovirusadeno-associated virus, suggesting that they have evolved ageneral function of protecting against invading viral nucleicacids (22). By contrast, the target range so far reported forretroviral CA-dependent restriction factors appears much lessextensive. Fv1 acts only on MLV (5), and restriction activityof Trim5 ${alpha}$ has been shown only for lentiviruses and gammaretroviruses(20, 25, 41, 58, 67). Restriction of other classes of retroviruseswould provide support for the notion that Trim5 ${alpha}$ had also evolvedto protect against retroelements. Since the FVs have propertiesthat are most distinct from other members of the retrovirusfamily (49), it seemed of particular interest to ask if theycould be restricted by Trim5 ${alpha}$ . Any such restriction might providesignificant insight into the life cycle of FVs; for example,it might enable the assignment of FV Gag into functional regionscorresponding to those found in conventional retroviruses. Conversely,the unique life cycle of the FVs might enhance our understandingof the mechanism of action of retroviral CA-dependent restrictionfactors. With these considerations in mind, we set out to investigatethe possibility that Trim5 ${alpha}$ might restrict FVs.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Cells and viruses. Human HT1080 (47), 293T (12), and common marmoset CM0203F cellswere maintained in Dulbecco modified Eagle medium, while cotton-toptamarin B95-8 cells (35) were maintained in RPMI media. Allmedia were supplemented with 10% fetal calf serum and 1% penicillinand streptomycin. CM0203F cells were generated by simian virus40 large T antigen-mediated immortalization of common marmosetbone marrow stroma cells. Restriction factors were deliveredinto cells using Moloney MLV (MoMLV)-based vectors. Viral vectorswere produced by transfection of 293T cells as previously described(67). MoMLV-based delivery vectors were made by cotransfectingVSVG, pHIT60, and either pLgatewaySN or pLgatewayIRESEYFP containingthe restriction gene. FV vectors (Fig. 1) were generated bycotransfection of pciSFV-1env (provides Env) and either pczDWP001,pcDWS001, or pcDWF003 (contributing Gag, Pol, and the vectorgenome with the enhanced green fluorescent protein [EGFP] markergene) to produce prototypic FV (PFV), simian FV macaque (SFVmac),and feline FV (FFV) isolates, respectively (two-plasmid system).FVs with chimeric gag genes were produced by either a three-plasmidSFVmac vector or four-plasmid PFV vector cotransfection system.In the three-plasmid SFVmac vector system, pciSFV-1env, pcDWS003(providing Pol and the vector genome with EGFP, but not Gag),and a Gag-expressing plasmid were used, while in the four-plasmidPFV vector system, pciSFV-1env, pcziPol PFV vector (providingPol), pMD9 (a minimal vector genome with an EGFP marker gene),and a Gag-expressing construct were cotransfected. FV vectorsupernatants were harvested 48 h posttransfection, aliquoted,and stored at –80°C until further use. Subsequently,individual vector supernatant aliquots were pretitrated on HT1080cells using the EGFP marker gene and flow cytometric analysis.For the two-color restriction assay described below, FV vectorsupernatants were then used at dilutions that resulted in 3to 40% EGFP-positive HT1080 cells.

View larger version (18K):
[in this window]
[in a new window]

FIG. 1. Schematic illustration of the FV vector systems employed. Schematic outline of the two-, three-, and four-plasmid FV vector systems consisting of a gene transfer vector and up to three packaging constructs. CMV, cytomegalovirus virus promoter; R, long terminal repeat (LTR) region; U5, LTR unique 5' region; U3, LTR unique 3' region; ${Delta}$ U3, enhancer-deleted U3 region; ${Delta}$ gag, gag gene with inactivated translation start (ATG $->$ GTG); CAS I or II, cis-acting sequence I or II; SFFV U3, spleen focus-forming virus U3 promoter; EG, EGFP; SD, splice donor; SA, splice acceptor; pA, polyadenylation signal sequence.

Restriction assays. Restriction was determined either by virus titration on HT1080cells expressing the restriction factors or by the two-colorfluorescence-activated cell sorter (FACS) assay. In the firstmethod, HT1080 cells were transduced with the MLV-based pLgatewaySNretroviral vector carrying the restriction factor and the neomycinresistance gene. The transduced cells were selected on 1 mg/mlof G418 for 2 weeks before seeding into 12-well plates at adensity of 5 x 10⁴ cells per well. B95-8 cells were plated ata density of 1 x 10⁵ cells per well. The cells were then challengedwith increasing amounts of FV carrying the EGFP marker, andthe percentage of infected cells was determined by FACS 3 dayspostinfection. The two-color FACS assay has previously beendescribed (5, 67). Briefly, HT1080 cells were transduced withthe MLV-based pLgatewayIRESEYFP retroviral vector carrying therestriction gene and an EYFP marker gene 2 days prior to challengingwith FVs carrying the EGFP marker. The cells were then analyzedby FACS to determine the percentage of YFP/restriction factor-positivecells that were FV infected and EGFP positive. This was comparedto the percentage of FV-infected cells (EGFP positive) thatdid not express the restriction factor (EYFP negative). A ratiothat was less than 0.3 was taken to represent restriction, whilea ratio greater than 0.7 indicated the absence of restriction.Ratios between 0.3 and 0.7 were indicative of partial restriction.

DNA constructs. (i) Restriction factors. Chimeric human RBCC-primate B30.2 Trim5 ${alpha}$ constructs have previouslybeen described (39). Total cotton-top tamarin RNA from the B95-8cell line was a kind gift from Peter Horn. mRNA from the CM0203F(common marmoset) cell line and an orangutan B-cell line (akind gift from Nel Otting, Biomedical Primate Research Center,The Netherlands) was isolated using the Oligotex direct mRNAkit from Qiagen. Full-length Trim5 ${alpha}$ s from cotton-top tamarin,common marmoset, and orangutan were derived from the RNA byreverse transcription (RT)-PCR using the first strand cDNA synthesiskit from Roche and then amplified with PfuUltra (Stratagene)using primer pairs TopoTrim5F (5'-CACCATGGCTTCTGGAATCCTGG-3')and Trim5Rev (5'-TCAAGAGCTTGGTGAGCACAG-3').

Trim5 ${alpha}$ constructs containing chimeric B30.2 domains were generatedby overlapping PCR. The fragments amplified from human Trim5 ${alpha}$ with primer pairs TopoTrim5F and HO2Rev (5'-GTAATGTTTCCCTGATGTGATAC-3')or HO3Rev (5'-TAACCCTATAACCCAGTAGCC-3') were joined to thoseamplified from human RBCC-orangutan B30.2 Trim5 ${alpha}$ with primerpairs Trim5Rev and HO2F (5'-GTATCACATCAGGGAAACATTAC-3') or HO3F(5'-GGCTACTGGGTTATAGGGTTA-3') to form HuOU2 and HuOU3, respectively,in a second PCR using TopoTrim5F and Trim5Rev. Similarly, thefragments amplified from human RBCC-orangutan B30.2 Trim5 ${alpha}$ withprimer pairs TopoTrim5F and HO2Rev or HO3Rev were joined tothose amplified from human Trim5 ${alpha}$ with primer pairs Trim5Revand HO2F or HO3F to form OUHu2 and OUHu3, respectively. ThePCR fragments were cloned into pENTRDTopo (Invitrogen) beforethey were transferred to pLGatewayIRESEYFP (67) or pLGatewaySN(66) by LR clonase II (Invitrogen).

In order to convert the glutamine at position 389 of the orangutanTrim5 ${alpha}$ to lysine, site-directed mutagenesis was performed using150 ng each of primer pair Q389K-F (5'-GCAATGTATAATATTGAAAAAAATGAAAATTATCAAC-3')and Q389K-R (5'-GTTGATAATTTTCATTTTTTTCAATATTATACATTGC-3') toamplify 10 ng of the template. This was followed by digestionwith DpnI to remove the template before transforming XL10 Goldultracompetent cells (Stratagene) with the mutagenized plasmid.

(ii) FVs. The PFV transfer vector pczDWP001 with the env and bel openreading frames (ORF) deleted but that expresses Gag and Poland contains an internal spleen focus-forming virus (SFFV) U3promoter-driven EGFP-neomycin (EGN) marker gene cassette, theminimal PFV transfer vector pMD9 expressing EGFP from an internalSFFV U3 promoter, the PFV Pol packaging construct pcziPol, andthe SFVmac Env packaging expression construct pciSFV-1env haspreviously been described (13, 21, 44, 57). Based on the SFVmac(34) and FFV (51) proviral genomes, the transfer vector constructspcDWS001 and pcDWF003 were generated with structure analogousto pczDWP001, except that they contained an EGFP instead ofan EGN marker gene (Fig. 1). In order to make pcDWF003, pcDWF001was first generated by digesting the FFV proviral expressionvector pChatul-3 (51) with PshAI and BamHI to delete the Envand Bel-1 ORFs and subsequently inserting a recombinant overlapPCR fragment comprising the 3' end of the FFV Pol ORF that hasthe overlapping translation start codons of the Env ORF inactivatedwithout changing the Pol protein sequence, which was generatedusing primers 2013 (5'-CGCGGATCCGCGGCCGCCTGCAGTTAATTAATCAGGATGAGTCAACTGAAGTTTCTG-3'),2014 (5'-AGAACATGTGACGACCTTGAAGGAATGGACGGAATG-3'), 2015 (5'-CCTTCAAGGTCGTCACATGTTCTTGTTCCGTTATGC-3'),and 2016 (5'-GAGTTACAAAATGATAAATGTATGG-3'). A 1,206-bp PacI/NotIfragment containing an SFFVU3 promoter-EGFP expression cassettewas then inserted into the unique PacI/NotI restriction sitesof pcDWF001 downstream of the Pol ORF, resulting in pcDWF003.In order to construct pcDWS001, pMR2 was first generated bycombining a 5,906-bp SacI/PmlI fragment of the SFVmac proviralclone pSFV-1 (34), comprising R-U5-Gag-Pol, a 1,218-bp SwaI/NotI(blunted) fragment containing an SFFVU3 promoter-EGFP expressioncassette, and a 1,972-bp EheI/PstI fragment of pSFV-1 comprisingpart of ORF-2 and a complete 3' long terminal repeat (LTR) inpcDNA3 (Invitrogen). The EcoRI/PacI fragment of pMR2 was thenreplaced with a recombinant overlap PCR fragment comprisingthe 3' end of the SFVmac Pol ORF having the overlapping translationstart codons of the Env ORF inactivated without changing thePol protein sequence, which was generated using primers 2003(5'-CTTAAGTTAACAGCTTATCAGGATAACGGCACCTCCAACGACTCTGG-3'), 2004(5'-GTCACTGCAGTTAATTAATCAAGTGCTTGATGTGCTTGACTCGTCTTCTTC-3'),2005 (5'-CCAAGGGGCAGCATTCACCTC-3'), and 2006 (5'-TATCCTGATAAGCTGTTAACTTAAG-3').The SFVmac vector pcDWS003 is a variant of pcDWS001 not expressingSFVmac Gag due to inactivation of the gag ORF translation startsite (Fig. 1), which was generated by exchanging a 1,076-bpMluI/HindIII fragment of pcDWS001 with a corresponding PCR-derivedfragment using primer 594 (5'-GTCGCTGAGTAGTGCGCGAGC-3') andmutagenesis primer 2620 (5'-CTAAAGCTTGGACATCTAAATCACCTTaTATTGCTGCCACttctattaag-3'),thus inactivating the Gag translation start by an ATG-to-GTGchange and introducing an additional in-frame stop codon afterthe fourth amino acid. To complement its Gag deficiency, anSFVmac Gag packaging construct (pcziSG) was generated by amplifyingthe complete SFVmac gag ORF from pcDWS001 using primer 2623(5'-ttaggtaccATGGCAGCAATAGAAGGTGAT-3') and 2624 (5'-tttgaatTCAGTTCCCTTGATTTCCGCTTC-3')and inserting the amplicon after KpnI/EcoRI digestion into thepcDNA3.1+zeo-based cytomegalovirus intron A-driven eukaryoticexpression vector pczi (21). The chimeric PFV/SFV Gag packagingconstructs were generated by recombinant overlap PCR and arebased on the SFVmac pcziSG and PFV pcziGag4 (PG) packaging plasmids.The natural FV primary and secondary cleavage sites were usedas chimera fusion sites for both ORFs, resulting in the followingconstructs: pcziPSG-1 (PFV Gag amino acids [aa] 1 to 621, SFVmacGag aa 617 to 647), pcziPSG-2 (PFV Gag aa 1 to 352, SFVmac Gagaa 343 to 647), pcziPSG-3 (PFV Gag aa 1 to 339, SFVmac Gag aa330 to 647), pcziPSG-4 (PFV Gag aa 1 to 311, SFVmac Gag aa 302to 647), pcziSPG-1 (SFVmac Gag aa 1 to 616, PFV Gag aa 622 to648), pcziSPG-2 (SFVmac Gag aa 1 to 342, PFV Gag aa 353 to 648),pcziSPG-3 (SFVmac Gag aa 1 to 329, PFV Gag aa 340 to 648), andpcziSPG-4 (SFVmac Gag aa 1 to 301, PFV Gag aa 312 to 648).

Western blot analyses. HT1080 cells were seeded in 12-well plates at a density of 5x 10⁴ cells per well 24 h prior to transduction with MoMLV-basedvectors carrying the restriction gene. Transduced cells weretransferred to 6-cm-diameter dishes 2 days posttransductionand grown to confluence before lysing with 500 µl of lysisbuffer containing 1% NP-40, 150 mM NaCl, and 50 mM Tris-HCl,pH 8. The lysate was cleared by centrifugation at maximum speedin a microcentrifuge, and the protein concentration determinedby Bradford assay. Western blot analysis was carried out using25 µg of total protein per sample. Goat anti-Trim5 ${alpha}$ (Abcam)was used at a 1:1,000 dilution, and horseradish peroxidase-conjugatedanti-goat (Vector Laboratories) was used at 1:1,000. Detectionwas performed using the chemiluminescent horseradish peroxidasesubstrate from Amersham.

RESULTS

Top

RESULTS

Restriction of FVs by Trim5 ${alpha}$ . Trim5 ${alpha}$ s from primates have been shown to restrict a diverse rangeof retroviruses with restriction specificity mapping primarilyto the B30.2 domain (38, 54, 60, 68). We had previously clonedthe Trim5 ${alpha}$ B30.2 domain from a panel of primates representingthe great apes, Old World monkeys (OWM), and New World monkeys(NWM) (39). These were fused to the RBCC domain of human Trim5 ${alpha}$ and tested for activity against HIV-1, SIVmac, and N-tropicMLV. All constructs had shown restriction activity against atleast one virus. We now set out to investigate whether FVs aresensitive to restriction by Trim5 ${alpha}$ using the same panel of chimericmolecules. HT1080 cells, which are permissive to PFV, SFVmac,and FFV, were transduced with the human RBCC-primate B30.2 chimerasand subsequently challenged with FV vectors encoding the EGFPmarker, which were produced using the two-plasmid system. Restrictionwas evaluated after 3 days as described previously using a two-colorFACS assay (5, 67).

Other than the squirrel monkey, all of the Trim5 ${alpha}$ chimeras withthe NWM B30.2 domains were seen to restrict SFVmac (Fig. 2).PFV and SFVmac were sensitive to restriction by the Trim5 ${alpha}$ chimerascontaining the B30.2 domains from the tamarin and marmoset Trim5 ${alpha}$ s.Interestingly, the tamarins and marmosets, which possessed identicalFV restriction properties, belong to the same Callitrichinaesubfamily in the Cebidae family of the NWM. The squirrel monkeyTrim5 ${alpha}$ chimera had the only NWM B30.2 domain in the collectionthat did not restrict any of the FVs tested, while the browncapuchin B30.2 domain was unique in that it showed restrictionof both SFVmac and FFV but not PFV. No restriction of PFV, SFVmac,or FFV was observed for the chimeras containing the OWM B30.2domains. Among the great apes, only the orangutan and gorillaTrim5 ${alpha}$ chimeras showed restriction, clearly inhibiting FFV, whereasthe human and chimpanzee chimeras lacked activity against allthree viruses.

View larger version (25K):
[in this window]
[in a new window]

FIG. 2. Restriction of FVs by human-RBCC-primate-B30.2 Trim5 ${alpha}$ chimeras. HT1080 cells were transduced with the different Trim5 ${alpha}$ chimeras and challenged with either PFV, SFVmac, or FFV vectors that encode the EGFP marker, produced using the two-plasmid system, 2 days posttransduction. The cells were analyzed by FACS 3 days later, and the ratio of infected Trim5 ${alpha}$ -positive cells to Trim5 ${alpha}$ -negative cells was scored. A ratio of less than 0.3 was taken as restriction, while that which was greater than 0.7 was scored as no restriction. Numbers in between were taken to represent partial restriction. The origins of the B30.2 domain from the various primates are indicated on the left of the figure, along with a phylogenetic tree that shows the relationship between the different primates (39). Restriction of the three different FVs is shown on the right. Numbers indicating restriction are boxed. Results are the averages and errors of two experiments.

Restriction of primate FVs by the cotton-top tamarin Trim5 ${alpha}$ . The restriction of FVs described in the previous section wasby chimeras that contain the human Trim5 ${alpha}$ RBCC. It was possiblethat this could contribute to the restriction and that the blockmight be an artifact of the combined human RBCC and primateB30.2 domains. To test this possibility, full-length cotton-toptamarin Trim5 ${alpha}$ was isolated from the total RNA of a cotton-toptamarin lymphocytic cell line, B95-8, by RT-PCR. This was usedboth in a two-color restriction FACS assay, yielding identicalresults with the chimeric construct described in the previoussection (data not shown), and to prepare a stable cell lineexpressing the full-length Trim5 ${alpha}$ by transducing HT1080 cellswith a vector that also carried the G418 resistance marker.Single G418-resistant colonies were isolated after 2 weeks ofselection. One of these cotton-top tamarin Trim5 ${alpha}$ -expressingclones was challenged with increasing amounts of PFV, SFVmac,or FFV encoding the EGFP marker, produced using the two-plasmidsystem. The percentage of cells transduced was determined byFACS; FV titers in the cotton-top tamarin Trim5 ${alpha}$ -expressing HT1080cells were compared to those in untransduced cells.

Restriction of PFV and SFVmac was observed for the cotton-toptamarin Trim5 ${alpha}$ -expressing HT1080 cell line (Fig. 3), though restrictionof PFV was somewhat less pronounced than that of SFVmac. Incontrast, there was no difference in titers of FFV between thetransduced cells and the negative control, indicating the absenceof restriction. Hence, the ability to restrict PFV and SFVmacwas a true property of the cotton-top tamarin Trim5 ${alpha}$ and notan artifact of the human RBCC primate B30.2 fusion. A similarconclusion could be reached from studies with full-length marmoset(data not shown) and orangutan Trim5 ${alpha}$ s (see below).

View larger version (18K):
[in this window]
[in a new window]

FIG. 3. Restriction of FVs by the full-length cotton-top tamarin Trim5 ${alpha}$ in transduced HT1080 cells. Full-length Trim5 ${alpha}$ from a cotton-top tamarin cell line, B95-8, was cloned into a retroviral vector that also carried the G418 resistance marker. HT1080 cells were transduced with the vector and treated with 1 mg/ml G418 for 2 weeks until single colonies were observed. A G418-resistant colony (CT Tam) was picked, expanded, and seeded into 12-well plates before infecting with increasing amounts of FVs carrying the EGFP marker, produced using the two-plasmid system. Untransduced HT1080 cells were included as a control. The cells were analyzed by FACS 3 days postinfection, and the percentage of infected cells was plotted against infectious units of FVs, which was determined by titrating on nonrestricting HT1080 cells. The data shown represent the averages and errors of two experiments.

Taken together, these results indicated that, like MLV and lentiviruses,FVs are susceptible to restriction by primate Trim5 ${alpha}$ s. As before,there are considerable variations in the levels of susceptibilityto the panel of Trim5 ${alpha}$ s between the different viruses. The specificitydeterminants for this variation in FV restriction reside inthe B30.2 domain, since the chimeras all contain the human RBCCdomain.

Restriction of FVs in New World monkey cell lines. The restriction of FVs by the primate Trim5 ${alpha}$ described abovewas observed for transduced cell lines, where stable expressionlevels are likely to be substantially higher than those seenfrom endogenous promoters. In an effort to demonstrate restrictionwith "normal" levels of Trim5 ${alpha}$ expression, the infectivity ofFVs on two New World monkey cell lines was investigated withB95-8 cells, which are a suspension cell line derived from cotton-toptamarin peripheral lymphocytes, and CM0203F cells, which area fibroblastic line from a common marmoset. In general, bothcell lines were less susceptible to FVs compared to HT1080s.However, the FFV titers were significantly higher on both celllines compared to those of SFVmac and PFV (data not shown),consistent with the data showing that SFVmac and PFV are restrictedby the endogenous cotton-top tamarin and common marmoset Trim5 ${alpha}$ s(Fig. 2).

However, it should be noted that neither line is ideal for thesestudies, since both are very weakly infected by retrovirusesand, at least in the case of B95-8, poorly transfected by nucleicacids, but our choices were constrained both by the paucityof such lines and by import/export regulations. Attempts toconfirm the specificity of these effects by abrogation and smallinterfering RNA have therefore proved fruitless (data not shown).Hence, further evidence will be required to fully establishthe in vivo relevance of the restriction of FV by Trim5.

Determinants of FFV restriction by ape Trim5 ${alpha}$ . The B30.2 domain contains three variable regions (V1, V2, andV3) (54, 56). Residues in all three regions can contribute tothe specificity of lentivirus and MLV recognition. To map thedeterminants of FFV restriction in the great ape Trim5 ${alpha}$ B30.2domain, chimeras were made between the human (does not restrict)and orangutan (restricts) B30.2 domains. These V-region chimerasall contain the human RBCC domain. The structures of these chimerasand their restriction properties against FVs produced usingthe two-plasmid system are shown in Fig. 4A; expression dataare shown in Fig. 4B. The different chimeric molecules are expressedat similar levels (Fig. 4B). Replacing V1 from the orangutanB30.2 with that from the human in HuOU2 resulted in a loss ofFFV restriction. By contrast, replacing only the V1 from thehuman B30.2 with that from the orangutan in OUHu2 resulted inrestriction of FFV. This indicated that the primary determinantsof FFV restriction lay in the V1 region of the ape Trim5 ${alpha}$ . Interestingly,HuOU3, which contained the V1 and V2 from human B30.2 and V3from the orangutan, restricted SFVmac, while neither the humannor orangutan Trim5 ${alpha}$ had any activity against SFVmac, a findingconsistent with our previous observations of the genetic reshufflingof the B30.2 variable domains (39).

View larger version (45K):
[in this window]
[in a new window]

FIG. 4. Determination of the regions of the B30.2 domain involved in FFV restriction. (A) Human-RBCC-primate-B30.2 Trim5 ${alpha}$ chimeras containing chimeric B30.2 domains between the human and orangutan Trim5 ${alpha}$ were constructed and tested for FV restriction as described in the legend for Fig. 2 using FV produced with the two-plasmid system. The compositions of different variable regions (V1, V2, and V3) of the different chimeras are shown on the left, while the restriction data are presented on the right. Numbers that indicate restriction are boxed. (B) Western blot analysis of HT1080 cells transduced with the primate Trim5 ${alpha}$ s described for panel A using anti-Trim5 ${alpha}$ . Lanes: 1, human; 2, HuOU2; 3, HuOU3; 4, orangutan; 5, OUHu2; 6, OUHu3; 7, untransduced control. (C) Point mutations were created by site-directed mutagenesis on residue 332 in the human and gorilla V1 region. The sequences of the V1 region in different apes are shown on the left, while the restriction data are presented on the right. Numbers that indicate restriction are boxed. Results are the averages and errors of two experiments. (D) Western blot analysis of HT1080 cells transduced with the primate Trim5 ${alpha}$ s described for panel C using anti-Trim5 ${alpha}$ . Lanes: 1, human; 2, chimpanzee; 3, gorilla; 4, orangutan; 5, human R332Q; 6, gorilla Q332R; 7, untransduced control.

The V1 regions of the hominidae Trim5 ${alpha}$ are very similar betweenthe different family members (Fig. 4C). In fact, there are onlytwo residues that differ between the human (does not restrict)and gorilla (restricts) sequences. These are arginine and glutamineat residue 332 and valine and methionine at residue 340. Wereasoned that residue 340 was not likely to be involved in FFVrestriction, as the chimpanzee (does not restrict) V1 also hadmethionine at this position. To investigate the involvementof residue 332 in FFV restriction, two Trim5 ${alpha}$ point mutants weretested for their restriction to FFV produced using the two-plasmidsystem. The restriction properties of these mutants are shownin Fig. 4C; expression data are shown in Fig. 4D. A change fromarginine to glutamine at position 332 in the human V1 (humanR332Q) resulted in a gain in activity against FFV, while substitutingthe glutamine for an arginine in the gorilla V1 (gorilla Q332R)resulted in a loss of FFV restriction. This indicated that thisresidue was the primary determinant for FFV recognition by theV1 region. Interestingly, this residue has also been shown tobe the main determinant of HIV-1 restriction by Trim5 ${alpha}$ (60, 68).

We have previously observed an effect of the V2 region of theorangutan Trim5 ${alpha}$ on HIV-1 restriction (39). In particular, theresidue at position 389 seemed to exert an influence in combinationwith the V1 region. Since FFV restriction by the orangutan Trim5 ${alpha}$ appeared to be governed by the same residue in V1 that determinedHIV-1 restriction, we wondered if residue 389 in V2 could alsoexert an effect on FFV restriction. In order to test this possibility,site-directed mutagenesis of full-length orangutan Trim5 ${alpha}$ aswell as the human RBCC-orangutan B30.2 chimera was carried outto convert the glutamine at position 389 to lysine. Restrictionassays using the two-color FACS analysis revealed a decreasein the restriction of FFV, while PFV and SFVmac remained unrestricted(Fig. 5A). The decrease in restriction was confirmed by titrationof FFV in HT1080 cells that had been stably transduced with,and expressed equal levels of, either full-length orangutanTrim5 ${alpha}$ or the Q389K mutant (Fig. 5B and C). Hence, the restrictionof FFV by orangutan Trim5 ${alpha}$ seems to be governed by the same residuesthat influence the restriction HIV and SIVmac, in particular,residues 332 and 389 in V1 and V2 of the B30.2 domain, respectively(39, 60, 68).

View larger version (30K):
[in this window]
[in a new window]

FIG. 5. Mapping of variation in FFV restriction in orangutan Trim5 ${alpha}$ . (A) Cells expressing human-RBCC-orangutan-B30.2 chimeric or full-length orangutan Trim5 ${alpha}$ with either glutamine (wild type [WT]) or lysine (Q389K mutant) at residue 389 were challenged with various FVs carrying the EGFP marker, produced using the two-plasmid system, and restriction was scored as described in the legend for Fig. 2. Numbers that indicate restriction are boxed. Results are the averages and errors of two experiments. (B) Titration curves of FFV, produced with the two-plasmid system, on HT1080 cells that were stably transduced with either full-length WT orangutan Trim5 ${alpha}$ or the Q389K mutant. The graphs were plotted using results from the averaging of two experiments. (C) Western blot analysis of the cells described for panel B using anti-Trim5 ${alpha}$ . Lanes: 1, nontransduced control; 2, Q389K mutant; 3, WT orangutan Trim5 ${alpha}$ .

Determination of the regions involved in the restriction of SFVmac by Trim5 ${alpha}$ . To further map the determinants of SFVmac restriction, chimericB30.2 domains were constructed and tested for activity againstthe three FVs produced using the two-plasmid system (Fig. 6A).The variable regions in the B30.2 domain of the capuchin, whichrestricts SFVmac and FFV but not PFV, were swapped with thosein the squirrel monkey, which does not restrict FVs. In SCC,where the V2 and V3 of the squirrel monkey were replaced withthose from the capuchin, SFVmac restriction was acquired, butthe chimera had no activity against FFV. Conversely, in CSS,where the V2 and V3 of the capuchin B30.2 were replaced withthat from the squirrel monkey, restriction of SFVmac and FFVwas lost, suggesting that these regions played a role in restriction.However, since the inactive chimera had no activity againstother viruses tested (SIVmac, HIV-1, or N-tropic MLV; resultsnot shown), we cannot exclude the fact that it could be nonfunctional,although its expression was comparable to those of both parentalTrim5 ${alpha}$ s in Western blot analyses (Fig. 6B). In addition, therewere no major differences, like insertions or deletions, betweenthe B30.2 domains of capuchin and the squirrel monkey, suggestingthat the lack of activity of CSS was most likely due to thefailure to recognize the CA of the retroviruses tested. Dueto the difficulty in interpreting the results of nonactive chimeras,further mapping analysis was not undertaken.

View larger version (20K):
[in this window]
[in a new window]

FIG. 6. Determination of the regions of the B30.2 domain involved in SFVmac restriction. (A) Human-RBCC-primate-B30.2 Trim5 ${alpha}$ chimeras containing chimeric B30.2 domains between the capuchin and squirrel monkey Trim5 ${alpha}$ were constructed and tested for FV restriction as described in the legend for Fig. 2, with FVs produced using the two-plasmid system. The compositions of the different variable regions (V1, V2, and V3) of the different chimeras are shown on the left, while the restriction data are presented on the right. Numbers that indicate restriction are boxed. Results are the averages and errors of two experiments. (B) Western blot analysis of HT1080 cells transduced with the primate Trim5 ${alpha}$ s as described for panel A, using anti-Trim5 ${alpha}$ . Lanes: 1, capuchin; 2, CSS; 3, squirrel monkey; 4, SCC; 5, untransduced control.

Target of Trim5 ${alpha}$ restriction in FVs. The Gag proteins of various orthoretroviruses have been identifiedas the viral targets for restriction by different Trim5 ${alpha}$ s (7,19, 58, 63, 66). However, the FV replication strategy differsin many aspects from that of orthoretroviruses, with the structuralproteins displaying many unique functions not found for thecorresponding orthoretroviral proteins (reviewed in reference9). We therefore wondered what the restriction determinant ofFVs might be. The results presented above demonstrated differentialrestriction behaviors of PFV, SFVmac, and FFV vectors, dependingon the Trim5 ${alpha}$ s used. However, these experiments did not allowus to identify unambiguously the viral component responsiblefor the restriction phenotype, since both the Gag and Pol proteinsof the respective FV species are coexpressed from the transfervectors in the packaging cell employing the two-plasmid vectorsystems (Fig. 1). On the other hand, an influence of the FVglycoprotein could be excluded because the SFVmac Env proteinwas used for all vector preparations. To further characterizethe determinant of FVs for restriction by Trim5 ${alpha}$ s, we thereforeemployed a three-plasmid vector system based on SFVmac (unpublishedresults) and a four-plasmid PFV vector system (13, 21, 44, 57),again using in all cases the SFVmac Env protein for pseudotyping.First we observed that the packaging of PFV or SFVmac vectorgenomes and their corresponding Pol protein as well as particleinfectivity was compatible with both PFV and SFVmac Gag (D.Lindemann, unpublished results). Titers of corresponding vectorpreparations differing only in their Gag proteins were withina twofold range of each other and reached absolute titers of2 x 10⁵ to 1 x 10⁶ EGFP FFU/ml (results not shown). This allowedus to clearly identify the FV Gag protein as the target forrestriction since particles containing PFV Gag and either SFVmacvector and Pol or PFV vector and Pol were not restricted byTrim5 ${alpha}$ with capuchin or chimeric squirrel monkey SCC B30.2 domains(Fig. 7A). In contrast, otherwise identical vector particleswith SFVmac Gag protein substituting for PFV Gag were restrictedby both Trim5 ${alpha}$ s. Thus, PFV and SFVmac particles generated eitherby the two-plasmid or by the three- and four-plasmid vectorsystems displayed the same restriction phenotype. We thereforeconclude that, as with orthoretroviruses, it is the Gag protein,rather than packaged vector genome or Pol protein, which representsthe critical determinant of viral restriction.

View larger version (18K):
[in this window]
[in a new window]

FIG. 7. FV Gag determinants of Trim5 ${alpha}$ restriction. Two-color FACS restriction assay using HT1080 cell populations expressing human-RBCC-primate-B30.2 Trim5 ${alpha}$ chimeras with capuchin or chimeric squirrel monkey SCC B30.2 domains and FV vector supernatants containing the different FV Gag proteins or chimeras as indicated. FV vector particles containing the SFVmac transfer vector were generated with the three-plasmid system by transient cotransfection of 293T cells with pcDWS003, pciSFV-1env (SFVmac), and the different Gag protein expression constructs, while particles containing the PFV transfer vector were made with the four-plasmid system by cotransfection of pMD9, pcziPol, pciSFV-1env (PFV), and the different Gag protein expression plasmids as indicated. The domain compositions of the different Gag chimeras are shown on the left, while restriction data are presented on the right. Numbers that indicate restriction are boxed. Results are the averages and errors of two experiments. (A) Parental Gags. (B) Chimeric Gags. Chimeras PSG-2, PSG-3, and SPG-2 yielded FV vector supernatants with very low titers ( $≤$ 1 x 10³ EGFP FFU/ml) that were insufficient for restriction analysis and were therefore not tested.

For orthoretroviruses, it has been demonstrated that the processedCA subunit and not the Gag precursor protein is the target forrestriction by Trim5 ${alpha}$ s (11, 18). FV Gag proteins, however, arenot cleaved into the MA, CA, and NC subdomain structure, andprocessing of the Gag precursor at only a single site is observedupon particle assembly and release (reviewed in reference 17).However, additional well-conserved secondary cleavage sites,first characterized by in vitro assays, also appear essentialduring early steps of the FV replication cycle at virus entry,presumably due to an FV and cellular protease-dependent CA disassemblypathway in the target cells (27, 42). To map further the determinantsof restriction on FV Gag, chimeras were generated between PFVand SFVmac Gag proteins (Fig. 7B). The ORFs were fused at thewell-conserved natural primary and secondary FV protease cleavagesites to increase the chances of obtaining functional chimericCA proteins. Out of eight chimeras made, five gave rise to infectiousvector particles with titers sufficient for use in two-colorrestriction assays. As with the wild-type PFV or SFVmac Gagproteins, chimeras PSG-1, PSG-4, SPG-1, and SPG-4 gave riseto similar titers (within a threefold range) to both PFV (four-plasmid)and SFVmac (three-plasmid) vector systems (D. Lindemann, unpublisheddata). SPG-3 produced particles of lower titers, but this stillresulted in at least 2% EGFP-positive cells in nonrestrictingHT1080 cell populations. The data on the restriction propertiesof the PFV-SFVmac Gag chimeras are shown in Fig. 7B. Similarto full-length PFV Gag, use of the PSG-4 chimera with the N-terminalp33Gag domain derived from PFV Gag and the C-terminal regionsfrom SFVmac did not lead to restriction of FV vector particlesby Trim5 ${alpha}$ molecules with capuchin or squirrel monkey SCC chimericB30.2 domains. By contrast, the presence of the N-terminal domainof SFVmac Gag (e.g., SPG-4) was sufficient to result in restrictionby both Trim5 ${alpha}$ constructs, in the same manner as FV vector particlescontaining the full-length SFVmac Gag. Thus, the restriction-determiningdomains of PFV and SFVmac reside within the N-terminal halfof the FV Gag protein.

DISCUSSION

Top

DISCUSSION

FVs have been reported to be restricted by several host factors.The promyelocytic leukemia protein, which is another memberof the Trim family, was shown to inhibit PFV by reducing transcriptionon binding to the PFV transactivator, Tas (48), although thisconclusion has been questioned by others (33). More recently,APOBEC3G was demonstrated to restrict FFV, PFV, and SFV (8,30, 52). This restriction could be alleviated by the Bet protein.In this study, we have identified for the first time Trim5 ${alpha}$ sthat restrict PFV, SFV, or FFV. The spumaviruses therefore nowrepresent another genus of Retroviridae that are restrictedby Trim5 ${alpha}$ , in addition to gammaretroviruses and lentiviruses.

Similar to the other retroviruses, Trim5 ${alpha}$ appears to target theGag of the FVs. Unlike the orthoretroviruses, the organizationof Gag into MA, CA, and NC with distinct cleavage sites is absentin FVs (17). The only processing of FV Gag observed in the courseof FV particle morphogenesis in vivo occurs at the C terminusof the molecule, removing a three-kilodalton peptide (p3) andproducing a shortened protein of 68 kDa (14, 43). However, threeinternal secondary cleavage sites have been characterized invitro that seem to be important during early steps of the FVreplication cycle upon entry into target cells (27, 42). Byanalogy with MLV and HIV-1, it may be that further processingof the FV Gag is important for recognition by restriction factors,as was reported for MLV and HIV-1 (11, 18). This suggests thatrearrangement of the FV Gag occurs as a consequence of eitherp3 or internal cleavage. Indeed, mutation of the internal secondarycleavage sites leads to reduced or abortive infection, presumablydue to the negative effect on cleavage of Gag in the targetcells upon entry (27). We have mapped the determinant of FVrestriction to the N-terminal half of FV Gag, in particular,the region before the first predicted internal secondary cleavagesite. In PFV, cleavage at this site would result in a 33-kDaprotein, while a much smaller product would be formed in FFV.As this protein is recognized by Trim5 ${alpha}$ , it is likely to correspondto the part of FV Gag that is exposed to the exterior of thevirion, similar to the CA subunit from conventional retroviruses.It will be interesting to further map the target of Trim5 ${alpha}$ inthe N-terminal fragment of the FV Gag, in comparison with theorthoretroviral CA.

We have also characterized the regions in the B30.2 domain thatdetermine the specificity of the restricted virus. As describedpreviously for the lentiviruses (39), all three variable regionsof the B30.2 can contribute to the specificity determinationof FV restriction: V2 and V3 are involved in SFV recognition,while V1 and V2 influence FFV restriction. In addition, specificresidues in the orangutan V1 (residue 332) and V2 (residue 389)that affected HIV-1 restriction (39) were also observed to determineFFV restriction. These data support the notion that all threevariable regions could form contacts with the FV Gags in a waysimilar to those of the gammaretroviral and lentiviral CAs.It also suggests that despite the lack of sequence and organizationalhomology between the FV Gag and orthoretroviral CA, they couldbe structurally similar. We might therefore predict that theGag moiety of FV forms hexamers within viral particles.

The biology of FVs is distinct from those of the other retrovirusesthat have been reported to be inhibited by Trim5 ${alpha}$ . Reverse transcriptionof FVs would have taken place predominantly in the producercells before entry into the Trim5-containing target cells (10,36, 51, 70, 71). Hence, binding to incoming viral particlesas well as inhibition of FV infection must be mainly post-reversetranscription. This would be similar to the Fv1 block of MLV(46) and the squirrel monkey Trim5 block of SIVmac (69) butcontrasts with the effect of most Trim5 ${alpha}$ s on virus replication,where both binding and inhibition occur before DNA synthesis(25, 58). It has been argued that proteasome-mediated degradationplays an important role in pre-reverse transcription restriction(1, 64). Many proteins that are targeted for the proteasomeare ubiquitinated by a series of enzymes, which include theE3 ligase. The RING motif of Trim5 ${alpha}$ has been demonstrated topossess E3 ligase activity (65), and it would be a good candidatefor targeting retroviral CAs for proteasomal degradation bytagging them with ubiquitin. Nevertheless, recent reports showedthat inhibition of the proteasomal machinery with MG132 resultedin a rescue of reverse transcription, although the virus wasstill blocked at a later stage, suggesting that proteasomaldegradation was involved in the early block and that more thanone mechanism might be at work during restriction (1, 64). FVsmay offer particular opportunities for studying this secondphase of Trim5 ${alpha}$ restriction. SFVmac, which is restricted by someNWM Trim5 ${alpha}$ s, has no lysine residues in its Gag (72). Hence, itis unlikely that ubiquitination of SFVmac Gag takes place duringTrim5 ${alpha}$ restriction. Since SFVmac is potently restricted in apost-reverse transcription fashion, this is consistent withthe notion that ubiquitin-mediated proteasomal degradation isnot involved in the late block in retroviral restriction. Thismeans that we can study restriction in the absence of drugsthat inhibit proteasomal function with viruses that are notsensitive to the proteasome-dependent mechanism of restriction.

Trim5 ${alpha}$ has clearly evolved under significant positive pressure(54). In light of the widespread distribution of FVs, it istempting to suggest that such viruses might have played a rolein driving Trim5 ${alpha}$ evolution. However, due to our failure to dateto demonstrate a clear protective role for Trim5 ${alpha}$ or indeed theabsence of clear pathology induced by FVs, this study, whichis the first description of FV restriction by Trim5 ${alpha}$ , can formonly the basis for further investigations to establish the roleof Trim5 in the evolution between FVs and their hosts. We notethat primate Trim5 ${alpha}$ s that restrict PFV very often also restrictSFVmac but not FFV. Conversely, with the exception of that fromthe capuchin monkey, FFV is restricted by Trim5 ${alpha}$ s that do notrestrict PFV or SFVmac. This suggests that FVs, which are evolutionarilyclosely related, have similar restriction phenotypes. In addition,Trim5 ${alpha}$ s from primates which are closely related to one anotherrestrict similar FVs. In particular, the Trim5 ${alpha}$ s from the subfamilyCallitrichinae of the NWM seem to confer resistance againstPFV and SFV, while those from apes block FFV. The restrictionphenotypes do not necessarily reflect the relationship betweenthe host and FV, as the biological relevance of FFV restrictionby ape Trim5 ${alpha}$ is unclear. However, they are an indication thatGags from related FVs have affinities for Trim5 ${alpha}$ s that have similarB30.2 domains. This is consistent with the notion that Trim5 ${alpha}$ evolved to protect against invasion of the genome by retroelements.Since there appear to be very few endogenous spumaretroviruses,perhaps Trim5 ${alpha}$ prevented invasion by endogenous FV-like elements.Evidence of the cospeciation of simian FVs and primates hasbeen shown, although this is apparently rare in RNA viruses,where the dominant mode of evolution is through cross-speciestransfer (61). The idea that restriction factors such as Trim5 ${alpha}$ and APOBECs act to drive cospeciation by preventing cross-speciestransmission of FVs is therefore most attractive. However, thephysiological role of Trim5 ${alpha}$ with respect to the simian FVs remainsto be determined, since the restricted viruses are geographicallyseparated from the restricting NWM. PFV, which was isolatedfrom an ape, and SFVmac, which comes from an OWM, are restrictedby NWM Trim5 ${alpha}$ but not by ape or OWM Trim5 ${alpha}$ . It would be interestingto find out if a recently sequenced NWM FV from spider monkeys(62) would have a reciprocal restriction phenotype.

ACKNOWLEDGMENTS

We thank Peter Horn and Nel Otting for providing cell linesand RNA samples.

This work was supported by the United Kingdom Medical ResearchCouncil and by grants from the DFG (LI621/3-1 and LI621/4-1)and the BMBF (01ZZ0102) to D.L.

FOOTNOTES

* Corresponding author. Mailing address: Medical Research Council, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, United Kingdom. Phone: 44 20 88162140. Fax: 44 20 89064477. E-mail: jstoye{at}nimr.mrc.ac.uk

Published ahead of print on 26 March 2008.

M.W.Y. and D.L. contributed equally to this study.

REFERENCES

Top