Independent evolution of an antiviral TRIMCypin rhesus macaquesSam J. Wilson*, Benjamin L. J. Webb*, Laura M. J. Ylinen*, Ernst Verschoor†, Jonathan L. Heeney†‡,and Greg J. Towers*§

*Medical Research Council Centre for Medical Molecular Virology, Department of Infection, Royal Free and University College Medical School, UniversityCollege London, London W1T 4JF, United Kingdom; †Department of Virology, Biomedical Primate Research Centre, 288 GJ, Rijswijk, The Netherlands;and ‡Department of Veterinary Medicine, University of Cambridge, Madingley Road, Cambridge CB3 0ES, United Kingdom

Edited by John M. Coffin, Tufts University School of Medicine, Boston, MA, and approved December 27, 2007 (received for review September 21, 2007)

The antiretroviral restriction factor TRIM5 has recently emerged as animportant mediator of innate immunity and species-specific inhibi-tion of retroviral replication in mammals. Selection pressure frompathogenic infection has driven rapid evolution of TRIM5 genes,leading to the antiviral specificities we see today. Remarkably, theNew World owl monkey (Aotus trivirgatus) encodes a TRIM5 proteinin which the antiviral determinants in the B30.2 domain have beenreplaced by cyclophilin A (CypA) encoded by a retrotransposed cDNA.The owl monkey TRIMCyp protein restricts infection by a subset oflentiviruses that recruit CypA to their capsids, including HIV-1 andfeline immunodeficiency virus. Here, we show that the Old Worldmonkey, rhesus macaque (Macaca mulatta), also encodes a TRIMCypprotein that has arisen independently from that in owl monkeys. Therhesus TRIMCyp is encoded by a single, but common, allele (Mamu7)of the rhesus TRIM5 gene, among at least six further alleles thatencode full-length TRIM5 proteins with no homology to CypA. Theantiviral specificity of the rhesus TRIMCyp is distinct, restrictinginfection of HIV-2 and feline immunodeficiency virus but not HIV-1.Restriction by rhesus TRIMCyp is before reverse transcription andinhibited by blocking CypA binding, with cyclosporine A, or bymutation of the capsid CypA binding site. These observations suggesta mechanism of restriction that is conserved between TRIMCypproteins. The lack of activity against HIV-1 suggests that Mamu7homozygous animals will be null for TRIM5-mediated restriction ofHIV-1 and could contribute to improved animal models for HIV/AIDS.

cyclophilin � lentivirus � restriction � TRIM5 � zoonosis

TRIM5 has recently been identified as a powerful restrictionfactor responsible for species-specific restriction of retroviral

infectivity as part of the innate immune system (1–6). TRIM5 hasa tripartite, or RBCC, motif consisting of RING, B Box 2, andcoiled coil domains with the characteristic ordering and spacing thatdefines the TRIM family. Together with many other members ofthe TRIM family, TRIM5 has a C-terminal PRY/SPRY, or B30.2,domain. In the case of TRIM5, this domain defines antiviralspecificity, probably by interacting directly with the incoming viralcapsid (7–13). Antiretroviral TRIM5 variants have been describedin primates, cattle, and rabbits (1–3, 14–16). These TRIM5 se-quences form a monophyletic group, indicating that they arederived from a common ancestor that probably had antiviralproperties (16). TRIM5 blocks retroviral infectivity by an incom-pletely characterized mechanism that involves the proteasome (6,17, 18) and may involve capsid uncoating (13, 19). In the New Worldowl monkey, the TRIM5 locus has been modified by insertion of acyclophilin A (CypA) cDNA by retrotransposition into the seventhintron (20, 21), leading to the expression of a TRIM5 variant(TRIMCyp), in which CypA replaces the exon eight-encoded B30.2domain. This change effectively replaces the antiviral specificitydeterminant, with CypA, leading to restriction of retroviruses thatrecruit CypA to their incoming capsid, including HIV-1, felineimmunodeficiency virus (FIV), and simian immunodeficiency virusfrom Tantalus monkey (SIVtan) (20–23).

Here, we demonstrate that rhesus macaques also encode anantiviral TRIMCyp molecule that has evolved independently fromthat in owl monkeys. Convergent evolution is indicated by the twoCypA sequences being in different positions in the owl monkey andrhesus TRIM5 loci. Furthermore, unlike the situation in owlmonkeys, the rhesus TRIMCyp fusion is encoded by a singleTRIM5 allele, among at least six further alleles that encode anintact B30.2 domain.

ResultsIdentification of a TRIMCyp in Rhesus Macaques. The rhesus macaqueTRIM5 gene has been demonstrated to be polymorphic, particu-larly within exon eight, which encodes the B30.2 domain anddetermines antiviral specificity (24). This finding suggests balancingselection at the TRIM5 locus, under selective pressure frompathogenic retroviruses. We sought to further characterize thedegree of polymorphism in rhesus macaques by sequencing exoneight from a cohort of rhesus macaques (Macaca mulatta), whichrevealed the existence of a TRIM5 allele that we named Macacamulatta TRIM5 allele seven (Mamu7) (Fig. 1A). Mamu7 has aframe shift in exon eight leading to a B30.2 domain that is truncatedby 59 aa and a G to T mutation in the splice acceptor site at the endof the sixth intron that is likely to cause exon skipping (Fig. 1B).Inspection of the rhesus genome sequence revealed CypA se-quences downstream of TRIM5 (data not shown). Knowing thatowl monkeys encode a TRIM5 protein with a CypA domain,encoded by a retrotransposed cDNA, in place of its B30.2 domain(20, 21), we sought an RNA transcript in rhesus cells in which theTRIM5 RBCC domain is fused to CypA by PCR. Remarkably, wewere able to amplify a rhesus macaque TRIMCyp (rhTRIMCyp)cDNA from a rhesus macaque cell line heterozygous for Mamu7and Mamu1 (LLC-MK2) by using primers specific to the RING andto CypA (Fig. 1C). The sequence of the Mamu7 coding sequencefrom a homozygous rhesus macaque and LLC-MK2 cells wereidentical (data not shown). RhTRIMCyp and owl monkey TRIM-Cyp (omTRIMCyp) protein sequences are shown aligned (Fig. 1C).The rhTRIMCyp protein is distinct from the owl monkey proteinin that it is encoded by exons 2 to 6 of TRIM5 and a CypA sequence,which replaces exons seven and eight. The owl monkey protein isencoded by exons two to seven of TRIM5 and a CypA cDNAinserted into the owl monkey genome by retrotransposition (20, 21).We then used genomic primers to PCR-amplify the CypA cDNA,and we sequenced it, thus revealing that it is in a different position

Author contributions: S.J.W., L.M.J.Y., J.L.H., and G.J.T. designed research; S.J.W. andB.L.J.W. performed research; L.M.J.Y., E.V., J.L.H., and G.J.T. contributed new reagents/analytic tools; S.J.W., B.L.J.W., and G.J.T. analyzed data; and G.J.T. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequence reported in this paper has been deposited in the GenBankdatabase (accession no. EU157763).

See Commentary on page 3177.

§To whom correspondence should be addressed. E-mail: g.towers@ucl.ac.uk.

© 2008 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0709003105 PNAS � March 4, 2008 � vol. 105 � no. 9 � 3557–3562

MIC

ROBI

OLO

GY

SEE

COM

MEN

TARY

than that in owl monkeys, indicating independent evolution of thetwo TRIMCyps. In owl monkeys, the CypA cDNA lies betweenexons seven and eight, whereas in the rhesus genome, the CypAcDNA is �900 nt downstream of the end of exon eight (Fig. 1D).The Mamu7 CypA sequence has the features of a cDNA that hasbeen reinserted into the genome by retrotransposition by an L1 lineelement, including target site duplication and a polyadenylationsequence (Fig. 1E).

To examine the frequency of Mamu7 (TRIMCyp) in rhesus

macaques, we PCR-amplified TRIM5 exon eight from DNApurified from 31 Indian and 38 Chinese Macaca mulatta.Fifteen of 31 Indian animals were heterozygous for Mamu7,and 1 was homozygous (Table 1). Strikingly, we did not detectMamu7 in any of the 38 Chinese macaques analyzed. A PCRscreen using oligos directed against the genomic sequenceeither side of the CypA cDNA sequence was consistent withthe exon eight sequencing data and revealed that the CypAinsertion is unique to the Mamu7 allele (Fig. 1F). It also

Fig. 1. Identification of a rhesus TRIMCyp. (A) Diagram of the TRIM5� protein showing polymorphisms between proteins encoded by Mamu1 and Mamu7.The position of nonsynonymous (above) and synonymous differences (below) are shown. (B) The sixth intron of Mamu7 bears a mutation at the spliceacceptor site (star). Exon7 sequence is shown (boxed). Conserved nucleotides are indicated (asterisk). (C) Alignment of the protein sequences of rhesusTRIMCyp (Mamu7) and owl monkey TRIMCyp (TCyp). Asterisk, identical residue; colon, conserved substitution; period, semiconserved substitution; gap,no conservation. RING, Bbox2 (BB), coiled coil (CC), and CypA domains are shown. (D) Diagram indicating splicing of TRIM5� encoded by Mamu1, rhesusTRIMCyp (Mamu7), and owl monkey TRIMCyp. Noncoding (gray) and coding (black) exons and CypA (striped) sequences are shown. In owl monkey,TRIMCyp CypA is encoded by a CypA cDNA in the seventh intron (20). (E) Genomic sequence of the Mamu7 CypA sequence. Target site duplication (TSD),splice acceptor (asterisk), start codon (circle), and polyadenylation signal and poly(A) sequences (underlined), which are typical of L1-mediatedretrotransposition, are shown. (F) PCR using primers on either side of the CypA insertion indicates that it is unique to the Mamu7 allele. Mamu genotypes,as determined by sequencing the exon eight PCR product, are shown. H2O denotes water control.

Table 1. Frequency of TRIM5 alleles in rhesus macaque cohort

Macaca mulatta origin

Allelic frequency

Mamu1 (%) Mamu3 (%) Mamu4 (%) Mamu5 (%) Mamu7 (%)

Indian 12/62 (19.35) 19/62 (30.65) 7/62 (11.29) 8/62 (12.9) 16/62 (25.81)Chinese 4/76 (5.26) 37/76 (48.68) 25/76 (32.89) 10/76 (13.16) 0/76 (0)

3558 � www.pnas.org�cgi�doi�10.1073�pnas.0709003105 Wilson et al.

confirms the fact that FRhK4 cells do not encode a TRIMCypand that LLC-MK2 cells are heterozygous for Mamu7 andMamu1.

Rhesus TRIMCyp Restricts HIV-2 and FIV but Not HIV-1, SIV fromMacaques (SIVmac), Equine Infectious Anemia Virus (EIAV), or MurineLeukemia Virus (MLV). To examine the restriction specificity ofrhTRIMCyp, we expressed it in permissive feline CRFK cells, asdescribed in ref. 23. As a negative control we expressed emptyvector, and as positive controls we expressed rhesus TRIM5�encoded by Mamu1 or omTRIMCyp. We then determined infec-tious titers of vesicular stomatitis virus (VSV)-G pseudotypedGFP-encoding retroviral vectors derived from HIV-1, HIV-2, FIV,SIVmac, MLV, and EIAV on pools of transduced cells, as well ason unmodified CRFK cells, as described in ref. 23 (Fig. 2).Remarkably, rhTRIMCyp was able to strongly restrict retroviralinfectivity but had a significantly different antiviral specificity fromomTRIMCyp. RhTRIMCyp strongly restricted both HIV-2 andFIV but not HIV-1. OmTRIMCyp restricted HIV-1 and FIV, ashas been described (20–22). SIVmac, EIAV, and MLV were notrestricted by either of the TRIMCyp proteins. Rhesus TRIM5�restricted HIV-1, HIV-2, FIV, and EIAV as has been described (1,3, 24–26). Expression levels of TRIM5�, owl monkey TRIMCyp,and rhTRIMCyp were shown to be similar by Western blotting forthe N-terminal HA tag (Fig. 2H). �-Actin served as a loadingcontrol.

Overexpression of Rhesus TRIMCyp Inhibits Restriction by RhesusTRIM5�. The fact that rhesus TRIM5� and TRIMCyp have differ-ent antiviral specificities suggests that each factor might havedominant negative activity against the other. To test this possibility,we transiently overexpressed rhTRIMCyp in rhesus FRhK4 cells,

which are homozygous for Mamu1 (Fig. 1 and data not shown). Asa positive control, we expressed human TRIM34. We then testedthe permissivity of the modified FRhK4 cells to HIV-1, HIV-2, orMLV-B vectors encoding GFP (Fig. 3). Expression of RhTRIM-Cyp was strongly dominant negative against the antiviral activity ofrhesus TRIM5� against HIV-1 (Fig. 3A). HIV-2 infectivity wasslightly reduced by TRIMCyp expression, presumably because it issensitive to both factors (Fig. 3B). Expression of human TRIM34also was strongly dominant negative against rhesus TRIM5�,rescuing infectivity of HIV-1 and HIV-2. Infectivity of unrestrictedMLV-B remained unaffected by expression of rhTRIMCyp,TRIM5�, or TRIM34 (Fig. 3C). These data suggest that, whenoverexpressed, TRIMCyp forms heteromultimers with endogenousTRIM5� that cannot restrict HIV-1. Exogenous expression ofTRIM5� (Mamu1) reduced infectivity of both HIV-1 and HIV-2,as expected. Viral stocks encoding rhTRIMCyp, TRIM5�, andTRIM34 were equalized by measuring provirus copy number ininfected cells by quantitative PCR (Q-PCR) as described in ref. 27(data not shown). Expression levels of expressed TRIM34,TRIM5�, and rhTRIMCyp proteins were similar, as shown byWestern blot analysis (Fig. 3D). The phenomenon of dominantnegative activity of TRIM5-related proteins, on overexpression, hasbeen described for related TRIMs (23) and deleted TRIM5 (7) andnatural shorter splice variants of TRIM5 itself, TRIM5� or -� (1,28). We presume that the dominant negative activity is caused byoverexpression and that at endogenous levels both factors are likelyto restrict in a codominant way. This finding is true for the murinerestriction factor Fv1, which is codominant in the mouse (29), butdominant negative when overexpressed in homozygous cell lines inexperiments similar to those described here (30).

Drugs or Mutations That Inhibit CypA Recruitment to Capsid PreventrhTRIMCyp Antiviral Activity. It is likely that, like omTRIMCyp, therhesus TRIMCyp protein restricts infection via recruitment of theTRIM5 RBCC domain to incoming capsids via CypA–capsidinteractions. To test this, we used the drug cyclosporine A (CSA),which binds and blocks the CypA active site and rescues omTRIM-Cyp-restricted infection (20, 21, 31). We infected CRFK cellsexpressing rhTRIMCyp with HIV-2- or FIV-derived vectors in thepresence and absence of 5 �M CSA. Blocking the CypA binding site

Fig. 2. Rhesus TRIMCyp restricts HIV-2 and FIV but not HIV-1, SIVmac, EIAV,or MLV. (A–G) Infectious titers of the viruses indicated were determined onunmodified feline cells (CRFK), CRFK cells bearing empty vector (EXN), CRFKcells expressing TRIM5� (Mamu1), rhesus (Mamu7), or owl monkey TRIMCyp(OMTC). Errors are standard deviations of titers determined at three multi-plicities of infection and are representative of experiments performed withindependent viral stocks. (H) Expression levels of rhesus TRIM5� (Mamu1),rhTRIMCyp (Mamu7), or owl monkey TRIMCyp (OMTC) were compared inCRFK cells or CRFK cells bearing empty vector (EXN) by Western blot analysisfor the HA tag. �-actin was detected on parallel blots as a loading control.

Fig. 3. Overexpression of rhesus TRIMCyp inhibits restriction by rhesusTRIM5�. FRhK4 cells were transduced with equal doses of MLV vector encod-ing TRIM5� or rhesus TRIMCyp. A human TRIM34 expression vector was usedas a positive control. GFP-encoding VSV-G pseudotypes of HIV-1 (A), HIV-2 (B),or MLV-B (C) were titrated on the modified cells 48 h later. Titers are expressedas infectious units/ml (IU/ml) determined as described (Fig. 2). Errors arestandard deviations derived from triplicate experiments and are representa-tive of at least two experiments using independent viral stocks. (D) Parallelinfections were harvested at 48 h, and TRIM protein expression was deter-mined by Western blotting analysis for the N-terminal HA tag.

Wilson et al. PNAS � March 4, 2008 � vol. 105 � no. 9 � 3559

MIC

ROBI

OLO

GY

SEE

COM

MEN

TARY

with CSA specifically rescued HIV-2 or FIV infectivity whenrestricted by rhTRIMCyp but not when restricted by rhesusTRIM5� (Fig. 4 A and B). HIV-2 infectivity is completely rescuedby CSA and FIV infectivity is rescued specifically, although notcompletely. Titration of CSA up to 15 �M indicated that even highconcentrations of CSA did not completely rescue FIV infectivity(data not shown). We interpret these observations as showing that,like owl monkey TRIMCyp, rhTRIMCyp restricts virus via recruit-ment of the CypA domain to the viral capsid.

Previous work has shown that, unlike HIV-1, HIV-2 does notrecruit CypA into viral particles (32). However, HIV-2 does havea conserved glycine-proline motif homologous to the CypA bindingmotif of HIV-1. To test whether this motif is responsible forsensitivity to rhTRIMCyp, we used an HIV-2 in which this motif hasbeen mutated to a sequence that does not recruit CypA in thecontext of HIV-1 (HIV-2 capsid G87A). In fact, HIV-2 capsidG87A was insensitive to rhTRIMCyp (Fig. 4C), suggesting that thismotif recruits rhTRIMCyp. Importantly, HIV-2 G87A remainsrestricted by rhesus TRIM5� (Fig. 4B). An FIV capsid mutantcapsid P90A has been described as being insensitive to restrictionby omTRIMCyp, suggesting that the CypA domain is recruited tothis proline (33). However, in our hands, this mutant is three to fourorders of magnitude less infectious than the wild-type virus, makingit difficult to test whether it is sensitive to rhTRIMCyp (data notshown).

Rhesus TRIMCyp Blocks Restricted Viral Reverse Transcription in aProteasome-Dependent Way. Both TRIM5� and omTRIMCypstrongly block retroviral DNA synthesis in most cases of restrictedinfectivity. Moreover, restricted viral DNA synthesis, but not in-fectivity, is rescued by inhibition of the proteasome (17, 18), whichsuggests that restricted virions are rapidly degraded by the protea-some, before they can reverse-transcribe, but that the proteasomeis not required for the block to infectivity. We examined whetherrestriction by rhTRIMCyp conserves these features. We infectedCRFK cells expressing rhTRIMCyp with HIV-2 or FIV in thepresence and absence of proteasome inhibitor and measured in-fection, or DNA synthesis, in parallel samples (Fig. 5). The resultsshow that rhTRIMCyp causes a strong block to HIV-2 and FIVinfectivity and DNA synthesis. We note that, as is often the case, theblock to infectivity is greater than the block to DNA synthesis,suggesting that some restricted virions are able to reverse-transcribe. Importantly, DNA synthesis, but not infectivity, ofrhTRIMCyp-restricted virus is rescued by inhibition of the protea-some. These observations are similar to those reported for restric-tion of HIV-1 by rhesus TRIM5� and omTRIMCyp (17, 18). Ittherefore is likely that the mechanism of restriction is conserved

between TRIM5� and TRIMCyp. Indeed, the only significantdifference between TRIM5� and the two TRIMCyps is likely to bethe mechanism of recruitment to the viral capsid, by a B30.2 domainor a CypA domain, respectively. Boiling the virus before infectionabrogates the production of viral DNA (Fig. 5 B and D), indicatingthat the Q-PCR signal is caused by viral reverse transcription andnot plasmid contamination.

DiscussionThis study describes a TRIMCyp protein in Indian rhesus ma-caques. The protein is expressed from a single allele of the TRIM5gene (Mamu7) by exon skipping from exon six to a CypA sequencederived from a reverse-transcribed, integrated CypA cDNA, prob-ably generated by L1-mediated retrotransposition (Fig. 1). Remark-ably, this is the second case of fusion of a simian TRIM5 RBCCdomain to CypA. In the New World owl monkey, a similarL1-mediated retrotransposition event has inserted a CypA cDNAinto the seventh intron of the TRIM5 gene, leading to the expres-sion of the owl monkey TRIMCyp. This protein strongly restrictsinfection by a variety of unrelated retroviruses (20–23). Impor-tantly, the different locations of the CypA cDNA in the owl monkeyand rhesus TRIM5 loci indicate that they have evolved indepen-dently (Fig. 1E).

Despite their similarity, the two TRIMCyp proteins have distinctantiviral specificities, differentiating between viruses that havearisen from different SIV lineages, i.e., HIV-1 and HIV-2 (Fig. 2).The Mamu7 CypA sequence differs from the rhesus CypA se-quence, from which it was derived, at two positions close to theactive site (D369N and R372H), which suggests that the rhesus andowl monkey proteins have been under selection pressures fromdifferent pathogenic viruses, consistent with their Old and NewWorld origins, respectively. Such evidence for change at the site ofvirus recruitment is reminiscent of positive selection in the TRIM5B30.2 domain, which also leads to differences in antiviral specificitybetween species (24, 34). The observation that Mamu7 is commonin Indian but not Chinese rhesus macaques is consistent withgeographic variation in selection pressure on the TRIM5 locus. Thefact that rhTRIMCyp does not restrict HIV-1 (Fig. 2) suggests that

Fig. 4. Drugs or mutations that inhibit CypA recruitment to capsid preventrhTRIMCypantiviralactivity. Titersofwild-typeHIV-2 (A), FIV (B), orHIV-2bearingcapsidmutationG87A(C)weremeasuredonCRFKcells (CRFK),CRFKcellsbearingemptyvector (EXN),CRFKcellsexpressingTRIM5� (Mamu1),CRFKcellsexpressingrhesus TRIMCyp (Mamu7), or CRFK cells expressing owl monkey TRIMCyp (OMTC)in the presence (black bars) or absence (gray bars) of 5 �M CSA. Errors arestandard deviations of titers determined at three multiplicities of infection andare representative of experiments performed with independent viral stocks.

Fig. 5. Rhesus TRIMCyp blocks restricted viral reverse transcription in a protea-some-dependent way. Cells expressing rhTRIMCyp (Mamu7), or unmodified cells(CRFK), were infected with HIV-2 (A and B) or FIV (C and D) in triplicate in thepresence (black bars) or absence (gray bars) of proteasome inhibitor. One samplewas subjected to FACS to determine infectious titer at 48 h (A and C), and totalDNA was purified from the second and third sample 6 h after infection formeasurement of viral DNA synthesis (B and D). Viral DNA synthesis is expressed asmoleculesofGFPtemplateper100ngoftotalDNA.Errorsarestandarddeviationsof values derived from parallel infections, and data are representative of inde-pendent experiments performed by using independent stocks of virus.

3560 � www.pnas.org�cgi�doi�10.1073�pnas.0709003105 Wilson et al.

Mamu7 homozygous animals will be more permissive to HIV-1replication and therefore useful in the development of a macaqueanimal model for HIV-1 infection.

RhTRIMCyp-restricted infectivity is rescued by the com-petitive inhibitor of CypA binding, CSA, indicating that thefactor is recruited to virions via CypA–capsid interactions. Thereason for the incomplete rescue of FIV infectivity by CSA isunclear, but high concentrations of the factor in the transducedcells, or high affinity binding between rhTRIMCyp and theFIV capsid, are possible explanations. Importantly, HIV-2 isrendered insensitive to restriction by mutation of the glycine-proline motif homologous to the site in HIV-1 that recruitsCypA. It is surprising that rhTRIMCyp restricts HIV-2, a virusthat has hitherto been thought not to recruit CypA to itscapsid. However, we speculate that lentiviruses in general mayrecruit CypA after entering the target cell cytoplasm asevidenced by the conservation of proline-rich structures inprimate lentiviral capsid sequences (6, 22). We propose thatthese motifs lead to the restriction of HIV-1, SIVtan, and FIVby omTRIMCyp and FIV and HIV-2 by rhTRIMCyp (Fig. 2)(20–23, 33). The role of CypA in lentiviral infection remainsunclear. It has been suggested to be involved in uncoating, butmutation of the CypA recruiting motif does not significantlyreduce infectivity of, for example, HIV-1 capsid G89V in felinecells (35) or HIV-2 capsid G87A in human cells (25). Morerecently, it has been suggested that CypA acts to inf luencesensitivity of retroviruses to restriction factors (6, 31, 35, 36).

It is remarkable that the same two genes have been fusedtogether by a process of exon shuff ling by L1-mediatedretrotransposition on more than one occasion. It is unclearwhich property of TRIM5 makes it an attractive candidate forfusion to CypA, especially given that artificial fusion of CypAto divergent TRIMs (37), and unrelated molecules (38–40) canmake effective restriction factors. It is also unclear whichviruses might have been responsible for providing selection,but it is notable that a growing number of unrelated viruseshave been shown to be inf luenced by cyclophilins, includingvaccinia virus (41), hepatitis C (42), and murine cytomegalo-virus (43). If cyclophilins are commonly recruited to virionproteins, they may be particularly effective as virus bindingdomains for restriction factors. The attraction of the TRIM5RBCC domain remains unclear but is presumably mechanistic.The fact that restriction by TRIM5� and TRIMCyp leads to astrong block to reverse transcription that is sensitive to inhi-bition of the proteasome (Fig. 5) (17, 18) suggests a commonmechanism. We presume that restricted complexes are rapidlyrecruited to the proteasome, but that virus infectivity is lost,even if the restricted virus is protected from degradation byinhibition of the proteasome (17, 18). We hypothesize thatdisruption of capsid rearrangement, uncoating, or traffickingcauses this proteasome-independent block to infectivity.

Transposition of Alu elements and processed pseudogenes isthought to have played an important role in mammalian evolution.It is assumed that in most cases, retrotransposition and integrationof cDNA sequences into the genome leads to the production of

inactive pseudogenes. These sequences decay in the absence ofpurifying selection and usually are mutated when compared withthe original cDNA sequence. Our data suggest that mammaliangenomes may have sequences that appear to be pseudogenes but arein fact part of a functional gene modified by retrotransposition, asis the case with TRIMCyps. Characterization of open, and relativelyintact, ‘‘pseudogenes’’ may well lead to the discovery of furtherexamples of exon shuffling mediated by retrotransposition inmammalian genomes.

MethodsCloning Rhesus TRIMCyp. Rhesus TRIMCyp was cloned from cDNA prepared fromthe LLC-MK2 Macaca mulatta rhesus kidney cell line RNA by using SuperScript(Invitrogen) according to manufacturer’s instructions. PCR was performed byusing primers (forward) 5�-GATCGAATTCATGGCTTCTGAAATCCTGCTTAATG-3�and (reverse) 5�-CAAGTTCGAATTATTTGAGTTGCCCACAGTCAGC-3� and pfuturbo (Stratagene). Sequences of three independent clones were determined toensure accuracy. CDNAs encoding TRIM5 Mamu1, omTRIMCyp, rhTRIMCyp, orhuman TRIM34 were cloned into the LNCX2-based MLV retroviral vector pEXN (agift fromPaulBieniasz,TheRockefellerUniversity,NewYork)betweentheEcoR1and Csp54I/ClaI sites (underlined) in frame with an N-terminal HA tag (23). TheMLV vectors were packaged into VSV-G pseudotyped Moloney MLV cores andexpressed in permissive feline CRFK cells as described (23, 44). Transduced CRFKcells were selected in G418 (1 mg/ml; Invitrogen), and pools of cells were used forsubsequent titration experiments. The rhesus TRIMCyp sequence has GenBankaccession no. EU157763.

Sequencing TRIM5 Alleles from Rhesus Monkeys (Macaca mulatta). Exon eightwas amplified by PCR from rhesus macaque genomic DNA samples (BiomedicalPrimate Research Centre, Rijswijk, The Netherlands) by using primers (forward)5�-CTTCTGAACAAGTTTCCTCCCAG-3� and (reverse) 5�-ATGAGATGCACATGGA-CAAGAGG-3�. PCR products were sequenced directly. Exon eight heterozygositywas resolved by cloning PCR products and sequencing multiple clones. Sequenceswere analyzed by using DNADynamo (Bluetractor Software). PCR of the CypAinsertion used primers (forward) 5�-TGACTCTGTGCTCACCAAGCTCTTG-3� and(reverse) 5�-ACCCTACTATGCAATAAAACATTAG-3� and GoTaq polymerase (Pro-mega) according to manufacturer’s instructions. The PCR product for the CypAinsertion was sequenced directly.

Viral Vectors. VSV-G protein pseudotyped retroviral vectors encoding GFPand derived from HIV-1, HIV-2, SIVmac, FIV, EIAV, or MLV were producedby triple transfection of 293T cells as described (44, 45). Viral stocks weretitrated onto cells and infected cells enumerated by fluorescence-activatedcell sorting (FACS), as described in ref. 46. Titers were determined atmultiplicities of infection between 0.01 and 0.3. The HIV-2 capsid mutantG87A has been described (25). Cyclosporine (Sandoz) was diluted in DMSOand added to cells at the time of infection, where stated, at 5 �M.

Q-PCR. Retroviral DNA synthesis by reverse transcription was measured byTaqman Q-PCR as described (27, 44). Samples were infected with DNase-treated virus (70 units/ml; Promega) in the presence or absence of MG132(1 �g/ml; Sigma) for 6 h. Total DNA was purified (QiaAmp; Qiagen) andsubjected to Taqman PCR (GFP amplicon) (27, 44). A parallel sample wassubjected to FACS to determine an infectious titer.

ACKNOWLEDGMENTS. We thank Welkin Johnson and Ruchi Newman for shar-ing unpublished observations, Salvatore Butera for helpful discussion and advice,and Ivonne Nieuwenhuis, Claire Pardieu, Imogen Lai, Paul Clapham, Paul Bien-iasz, Eric Poeschla, Andrew Lever, Jonathan Stoye, Didier Trono, Francois LoicCosset, Kyriacos Mitrophanous, and Adrian Thrasher for reagents. We thank theWellcome Trust for funding this work through a senior fellowship (to G.J.T.).

1. Stremlau M, et al. (2004) The cytoplasmic body component TRIM5� restricts HIV-1infection in Old World monkeys. Nature 427:848–853.

2. Keckesova Z, Ylinen LM, Towers GJ (2004) The human and African green monkeyTRIM5� genes encode Ref1 and Lv1 retroviral restriction factor activities. Proc NatlAcad Sci USA 101:10780–10785.

3. Hatziioannou T, Perez-Caballero D, Yang A, Cowan S, Bieniasz PD (2004) Retrovirusresistance factors Ref1 and Lv1 are species-specific variants of TRIM5�. Proc Natl AcadSci USA 101:10774–10779.

4. Yap MW, Nisole S, Lynch C, Stoye JP (2004) Trim5� protein restricts both HIV-1 andmurine leukemia virus. Proc Natl Acad Sci USA 101:10786–10791.

5. Perron MJ, et al. (2004) TRIM5� mediates the postentry block to N-tropicmurine leukemia viruses in human cells. Proc Natl Acad Sci USA 101:11827–11832.

6. Towers GJ (2007) The control of viral infection by tripartite motif proteins and cyclo-philin A. Retrovirology 4:40.

7. Perez-Caballero D, Hatziioannou T, Yang A, Cowan S, Bieniasz PD (2005) Humantripartite motif 5{�} domains responsible for retrovirus restriction activity and speci-ficity. J Virol 79:8969–8978.

8. Yap MW, Nisole S, Stoye JP (2005) A single amino acid change in the SPRY domain ofhuman Trim5� leads to HIV-1. Restriction Curr Biol 15:73–78.

9. Stremlau M, Perron MJ, Welikala S, Sodroski J (2005) Species-specific variation in theB30.2(SPRY) domain of TRIM5� determines the potency of human immunodeficiencyvirus restriction. J Virol 79:3139–3145.

10. Perron MJ, Stremlau M, Sodroski J (2006) Two surface-exposed elements of theB30.2/SPRY domain as potency determinants of N-tropic murine leukemia virus restric-tion by human TRIM5�. J Virol 80:5631–5636.

Wilson et al. PNAS � March 4, 2008 � vol. 105 � no. 9 � 3561

MIC

ROBI

OLO

GY

SEE

COM

MEN

TARY

11. Ohkura S, Yap MW, Sheldon T, Stoye JP (2006) All three variable regions of the TRIM5� B30.2domain can contribute to the specificity of retrovirus restriction. J Virol 80:8554–8565.

12. Sebastian S, Luban J (2005) TRIM5� selectively binds a restriction-sensitive retroviralcapsid. Retrovirology 2:40.

13. Stremlau M, et al. (2006) Specific recognition and accelerated uncoating of retroviralcapsids by the TRIM5� restriction factor. Proc Natl Acad Sci USA 103:5514–5519.

14. Ylinen LM, et al. (2006) Isolation of an active Lv1 gene from cattle indicates thattripartite motif protein-mediated innate immunity to retroviral infection is wide-spread among mammals. J Virol 80:7332–7338.

15. Si Z, et al. (2006) Evolution of a cytoplasmic tripartite motif (TRIM) protein in cows thatrestricts retroviral infection. Proc Natl Acad Sci USA 103:7454–7459.

16. Schaller T, Hue S, Towers GJ (2007) An active TRIM5 in rabbits indicates an antiviralcommon ancestor for mammalian TRIM5 proteins J Virol 81:11713–11721.

17. Wu X, Anderson JL, Campbell EM, Joseph AM, Hope TJ (2006) Proteasome inhibitorsuncouple rhesus TRIM5� restriction of HIV-1 reverse transcription and infection. ProcNatl Acad Sci USA 103:7465–7470.

18. Anderson JL, et al. (2006) Proteasome inhibition reveals that a functional preintegra-tion complex intermediate can be generated during restriction by diverse TRIM5proteins. J Virol 80:9754–9760.

19. Perron MJ, et al. (2007) The human TRIM5{�} restriction factor mediates accelerateduncoating of the N-tropic murine leukemia virus capsid. J Virol 81:2138–2148.

20. Sayah DM, Sokolskaja E, Berthoux L, Luban J (2004) Cyclophilin A retrotranspositioninto TRIM5 explains owl monkey resistance to HIV-1. Nature 430:569–573.

21. Nisole S, Lynch C, Stoye JP, Yap MW (2004) A Trim5–cyclophilin A fusion protein foundin owl monkey kidney cells can restrict HIV-1. Proc Natl Acad Sci USA 101:13324–13328.

22. Diaz-Griffero F, et al. (2006) Requirements for capsid-binding and an effector functionin TRIMCyp-mediated restriction of HIV-1. Virology 351:404–417.

23. Zhang F, Hatziioannou T, Perez-Caballero D, Derse D, Bieniasz PD (2006) Antiretroviralpotential of human tripartite motif-5 and related proteins. Virology 353:396–409.

24. Newman RM, et al. (2006) Balancing selection and the evolution of functionalpolymorphism in Old World monkey TRIM5�. Proc Natl Acad Sci USA 103:19134 –19139.

25. Ylinen L, Keckesova Z, Wilson SJ, Ranasinghe S, Towers GJ (2005) Differential restrictionof HIV-2 and SIVmac by TRIM5� alleles. J Virol 79:11580–11587.

26. Saenz DT, Teo W, Olsen JC, Poeschla EM (2005) Restriction of feline immunodeficiencyvirus by Ref1, Lv1, and primate TRIM5� proteins. J Virol 79:15175–15188.

27. Towers GJ, et al. (1999) One step screening of retroviral producer clones by real timequantitative PCR. J Gene Med 1:352–359.

28. Passerini LD, Keckesova Z, Towers GJ (2006) Retroviral restriction factors Fv1 and TRIM5{�}act independently and can compete for incoming virus before reverse transcription. J Virol80:2100–2105.

29. Pincus T, Rowe WP, Lilly F (1971) A major genetic locus affecting resistance to infectionwith murine leukemia viruses II: Apparent identity to a major locus described forresistance to friend murine leukemia virus. J Exp Med 133:1234–1241.

30. Bock M, Bishop K, Towers G, Stoye JP (2000) Use of a transient assay for studying thegenetic determinants of Fv1 restriction. J Virol 74:7422–7430.

31. Towers GJ, et al. (2003) Cyclophilin A modulates the sensitivity of HIV-1 to hostrestriction factors. Nat Med 9:1138 –1143.

32. Braaten D, Franke EK, Luban J (1996) Cyclophilin A is required for the replication ofgroup M human immunodeficiency virus type 1 (HIV-1) and simian immunodeficiencyvirus SIV (CPZ)GAB but not group O HIV-1 or other primate immunodeficiency viruses.J Virol 70:4220–4227.

33. Lin TY, Emerman M (2006) Cyclophilin A interacts with diverse lentiviral capsids.Retrovirology 3:70.

34. Sawyer SL, Wu LI, Emerman M, Malik HS (2005) Positive selection of primate TRIM5�

identifies a critical species-specific retroviral restriction domain. Proc Natl Acad Sci USA102:2832–2837.

35. Keckesova Z, Ylinen L, Towers GJ (2006) Cyclophilin A renders HIV-1 sensitive to oldworld monkey but not human TRIM5a antiviral activity. J Virol 80:4683–4690.

36. Berthoux L, Sebastian S, Sokolskaja E, Luban J (2005) Cyclophilin A is required forTRIM5{�}-mediated resistance to HIV-1 in Old World monkey cells. Proc Natl Acad SciUSA 102:14849–14853.

37. Yap MW, Dodding MP, Stoye JP (2006) Trim-cyclophilin A fusion proteins can restricthuman immunodeficiency virus type 1 infection at two distinct phases in the viral lifecycle. J Virol. 80:4061–4067.

38. Schaller T, Ylinen LM, Webb BL, Singh S, Towers GJ (2007) Fusion of cyclophilin A to fv1enables cyclosporine-sensitive restriction of human and feline immunodeficiency vi-ruses. J Virol 81:10055–10063.

39. Yap MW, Mortuza GB, Taylor IA, Stoye JP (2007) The design of artificial retroviralrestriction factors. Virology 365:302–314.

40. Javanbakht H, et al. (2007) The ability of multimerized cyclophilin A to restrictretrovirus infection. Virology 367:19–29.

41. Castro AP, Carvalho TM, Moussatche N, Damaso CR (2003) Redistribution of cyclophilinA to viral factories during vaccinia virus infection and its incorporation into matureparticles. J Virol 77:9052–9068.

42. Watashi K, et al. (2005) Cyclophilin B is a functional regulator of hepatitis C virus RNApolymerase. Mol Cell 19:111–122.

43. Kawasaki H, Mocarski ES, Kosugi I, Tsutsui Y (2007) Cyclosporine inhibits mousecytomegalovirus infection via a cyclophilin-dependent pathway specifically in neuralstem/progenitor cells. J Virol 81:9013–9023.

44. Besnier C, Takeuchi Y, Towers G (2002) Restriction of lentivirus in monkeys. Proc NatlAcad Sci USA 99:11920–11925.

45. Hatziioannou T, Cowan S, Goff SP, Bieniasz PD, Towers GJ (2003) Restriction of multipledivergent retroviruses by Lv1 and Ref1. EMBO J 22:385–394.

46. Towers G, et al. (2000) A conserved mechanism of retrovirus restriction in mammals.Proc Natl Acad Sci USA 97:12295–12299.

3562 � www.pnas.org�cgi�doi�10.1073�pnas.0709003105 Wilson et al.