Fusion of Cyclophilin A to Fv1 Enables Cyclosporine-Sensitive Restriction of Human and Feline Immunodeficiency Viruses

Abstract

TRIM5α is a potent intracellular antiviral restriction factor governing species-specific retroviral replication. In the New World species owl monkey the coding region for the viral binding B30.2 domain of TRIM5α has been replaced by a cyclophilin A (CypA) pseudogene by retrotransposition. The resultant TRIM5-CypA fusion protein restricts human immunodeficiency virus type 1 (HIV-1), as well as feline immunodeficiency virus (FIV), by recruitment of the CypA domain to the incoming viral capsids. Infectivity is rescued by agents such as cyclosporine that disrupt CypA binding to its substrates. Mice encode an antiviral restriction factor called Fv1 (for Friend virus susceptibility gene 1), which is active against murine leukemia virus and related to endogenous gag sequences. Here we show that fusing CypA to Fv1 generates a restriction factor with the antiviral specificity of TRIMCyp but the antiviral properties of Fv1. Like TRIMCyp, Fv1-Cyp restricts HIV-1 and FIV and is sensitive to inhibition by cyclosporine. TRIM5α is known to have a short half-life and block infectivity before viral reverse transcription. We show that Fv1-Cyp has a long half-life and blocks after reverse transcription, suggesting that its longer half-life gives the restricted virus the opportunity to synthesize DNA, leading to a later block to infection. This notion is supported by the observation that infectivity of Fv1-Cyp restricted virus can be rescued by cyclosporine for several hours after infection, whereas virus restricted by TRIMCyp is terminally restricted after around 40 min. Intriguingly, the Fv1-Cyp-restricted HIV-1 generates closed circular viral DNA, suggesting that the restricted virus complex enters the nucleus.

Consideration of host factors influencing murine leukemia virus (MLV) infection in mice led to the discovery of the Fv1 (for Friend virus susceptibility gene 1) antiviral phenotype (28, 39). The Fv1 gene was identified as an almost full-length endogenous retroviral gag protein and is unique to mice (5). Fv1 protects mice from infection by MLV, allowing the division of MLV isolates according to their Fv1 sensitivity. N-tropic MLV (MLV-N) infects NIH mice, which encode the Fv1 N allele (Fv1^n/n) but not BALB/c mice, which are Fv1^b/b. Conversely, B-tropic MLV (MLV-B) infects BALB/c mice but not NIH mice. Cell lines derived from these mice have similar MLV sensitivities and NIH/BALB/c Fv1 heterozygotes (Fv1^n/b) expressing both proteins restrict both MLV-N and MLV-B (36). A third group of MLV, which includes Moloney MLV, are NB-tropic in that they are insensitive to both Fv1 N and Fv1 B. The viral determinants for sensitivity to Fv1 lie in the MLV capsid. Notably, an N-tropic virus can be made B-tropic by switching the amino acid at position capsid (CA) 110 from arginine to glutamate and vice versa (24). Making N- or B-tropic MLV NB-tropic is more complex and requires a number of changes (26, 40). The details of the antiviral mechanism remain unclear, but recent data suggest that incoming retroviral capsids interact with Fv1 early after entry and are rendered uninfectious (32). Fv1-restricted MLV completes viral DNA synthesis by reverse transcription (RT) but does not form a provirus. The observation that viral DNA circles are reduced implies that Fv1 blocks infectivity before viral nuclear entry (21, 51). An important feature of Fv1 restriction is that it is saturable. This means that restrictive cells can be rendered permissive by titrating the Fv1 protein by coinfecting with virus-like particles (8). The virus-like particles must be restriction sensitive and encode protease, demonstrating that gag cleavage is essential for recognition by Fv1 (16).

The identification of TRIM5α as a potent antiretroviral restriction factor, active against a variety of divergent retroviruses, has awoken interest in Fv1 (19, 23, 35, 41, 53). Restriction by TRIM5α bears a striking resemblance to restriction by Fv1. Both factors target the incoming viral CA protein, and the restriction of MLV-N by both Fv1 N and human or simian TRIM5α molecules is dependent on an arginine at CA 110 (4, 25, 35, 45). Like Fv1, TRIM5α is saturable and in some cases restricted virions synthesize viral DNA, although in most cases TRIM5α inhibits viral DNA synthesis (41, 54, 55). TRIM5α from Old World monkeys (OWM), but not humans, strongly restricts human immunodeficiency virus type 1 (HIV-1) and contributes to the inability of HIV-1 to replicate in OWM (19, 23, 35, 41, 53). Strong restriction of HIV-1 in OWM cells depends on the activity of the peptidyl prolyl isomerase cyclophilin A (CypA) (2, 20, 22, 43). Since CypA is known to bind the HIV-1 CA molecule (30, 44) and change its shape by catalyzing cis/trans isomerization of the peptide bond at CA G89-P90 (9, 56), it has been proposed that this isomerization makes the HIV-1 CA a better target for the OWM TRIM5α molecule (22). In other words, maximal restriction of HIV-1 by OWM TRIM5α depends on CypA activity. Remarkably, in cells from the New World owl monkey CypA has been fused to TRIM5 by insertion of an in-frame CypA pseudogene that completely replaces the coding region of the TRIM5α viral binding B30.2 domain (31, 38). The resultant molecule, named TRIMCyp, restricts HIV-1 and feline immunodeficiency virus (FIV) via recruitment of the CypA domain to the viral CA. Inhibition of the interaction between CypA using the competitive inhibitor of CypA binding, cyclosporine (CSA), rescues HIV-1 and FIV infectivity in owl monkey cells (15, 29, 31, 38, 46).

Here we have sought to further our understanding of the relationship between CypA and restriction by TRIM5α and Fv1 by fusing CypA from TRIMCyp to Fv1 and characterizing the ability of the fusion proteins to restrict retroviral infection.

MATERIALS AND METHODS

Generation of HAFv1 N and Fv1-Cyp encoding plasmids.

HAFv1 N was made by PCR using oligonucleotides fwd TS33 (5′-CAGTGAATTCATGAATTTCCCACGTGCGCTTGC-3′) and rev TS34 (5′-CGATGCGGCCGCTCAGAGTTTTGTAGCTGCTGTTGG-3′). Fv1 N encoded within the CXCR plasmid was used as a template. The Fv1 PCR product was inserted into the retroviral vector pEXN (57), a derivative of pLNCX, by using the EcoRI and NotI sites (underlined). The resulting vector expresses the N-terminal hemagglutinin (HA)-tagged Fv1 protein from the cytomegalovirus promoter and the neomycin resistance gene from the MLV long terminal repeat (LTR) in the transduced cells. Single cell clones of CRFK expressing HAFv1 N were generated by limiting dilution in media containing 0.8 mg of G418/ml.

Fv1-Cyp fusions were made by generating MluI sites in the Fv1 coding region at amino acid position 358 or 432 by site-directed mutagenesis as described previously (58) using the position 358 oligonucleotides fwd TS30 (5′-GGATTAAAACGCGTAGGGAACTTGGAATG-3′) and rev TS29 (5′-CCCTACGCGTTTTAATCCCATCCTC-3′) and the position 432 oligonucleotides fwd TS31 (5′-GTTTAACGCGTACAGCAGCTACAAAACTCTG-3′) and rev TS32 (5′-GCTGTACGCGTTAAACTTAATTCTATTAG-3′). The MluI sites are underlined. Fv1 N encoded within the CXCR plasmid between the EcoRI and Csp45I sites was used as a template. An MluI site was generated at the beginning of the CypA coding region from owl monkey TRIMCyp by PCR using fwd TS27 (5′-TAGCACGCGTATGGTCAATCCTACTGTGTTC-3′) and rev TS28 (5′-CGATTTCGAATTAAAGTTGTCCACAGTCAG-3′). The CypA PCR product was then joined to Fv1 between the MluI site engineered into Fv1 and the Csp45I site in CXCR by using the Csp45I site in the reverse primer (underlined). The resulting fusion proteins were therefore encoded within the MLV-based expression vector CXCR in the X position between the EcoRI site and the Csp45I site as described previously (23). This vector expresses the fusion proteins from the MLV LTR and DsRed-Express red fluorescent protein (Clontech) from a cytomegalovirus promoter in the transduced cells. All PCR products were sequenced (Lark Technologies, Takely, United Kingdom). Oligonucleotides were designed and sequence analyzed by using DNADynamo analysis software (Bluetractor Software UK). Single cell clones of the CRFK line expressing Fv1-Cyp fusions were generated by dilution. CRFK cells expressing Fv1 N encoded within CXCR have been described previously (32).

Viruses.

Vesicular stomatitis virus protein G (VSV-G)-pseudotyped, green fluorescent protein (GFP)-encoding retroviral vectors derived from HIV-1, HIV-2, FIV and MLV-N and B were prepared by transfection of 293T cells as previously described (3, 17, 18, 37). Viruses were titrated onto cells plated at 10⁵ cells per well in six-well plates, and infection was measured by enumerating GFP-expressing cells by fluorescence-activated cell sorting (FACS) 48 h after infection as described previously (3). Infections were performed in the presence or absence of 5 μM CSA (Sandoz, Frimley, United Kingdom) or vehicle (dimethyl sulfoxide [DMSO] at 5 μl/ml) where specified.

TaqMan qPCR.

In order to measure viral DNA synthesis, we infected 2 × 10⁵ cells per well in six-well plates in triplicate with DNase-treated virus (120 min, 70 U of DNase/ml; Promega UK) and incubated them for 6, 12, or 18 h. Total DNA was extracted (QiaAmp; QIAGEN, Crawley, United Kingdom) from two samples, and the third was subjected to FACS to enumerate infected cells 48 h after infection. Ten percent of the sample (ca. 100 ng of DNA) was subjected to TaqMan quantitative PCR (qPCR) for linear DNA encoding GFP as described previously (3), LTR gag as described previously (11, 50), or viral 2-LTR circles as described previously (11, 50). The copy number of PCR product was expressed per 100 ng of total DNA, which was equivalent to around 20,000 cells.

Saturation assays.

We plated CRFK cells stably expressing Fv1 N, or unmodified CRFK cells as a control, in six-well plates at 10⁵ cells per well. They were infected the following day with neat MLV encoding the Fv1-Cyp fusions. After 48 h the cells were split 1/6 and then infected the day after that with MLV-N or MLV-B encoding GFP. Cells expressing GFP were enumerated by FACS 48 h after GFP-encoding virus infection, and titers were calculated in infectious units per milliliter.

Determination of half-life.

Cycloheximide (Merck) was prepared in ethanol at 100 mg/ml and diluted in tissue culture media for each experiment to 100 μg/ml. CRFK cells and CRFK cells expressing Fv1-Cyp358, HA-Fv1 N, or human TRIM5α-HA were plated in six-well plates at 10⁵ cells per well. The next day, cells were treated with cycloheximide for different durations and then lysed in 2× sodium dodecyl sulfate loading buffer. Equal volumes of supernatants from sonicated and lysed cells were subjected to Western blotting, using antibodies raised against CypA (rabbit polyclonal; SA296; Biomol), actin (Sigma), or horseradish peroxidase-linked anti-HA monoclonal clone 3F10 (Roche).

Timed inhibition of restriction with CS.

CRFK cells expressing Fv1-Cyp358, Fv1-Cyp432, or TRIMCyp were incubated with VSV-G-pseudotyped, GFP-encoding HIV-1 vector at 4°C for 2 h to allow the virus to bind to the cells. Infection was initiated by warming up the cells to 37°C for 10 min. A total of 40 mM ammonium chloride was then added to prevent acidification of endosomes and thus further VSV-G-mediated viral entry. At this time 5 μM CSA was added for the zero time point or added at subsequent time points as shown in Fig. 5. In this way we chased a 10-min pulse of virus for 10 h with doses of 5 μM CSA. A parallel experiment was performed with a similar dose of unrestricted MLV-B. At 48 h after infection infected cells were enumerated by FACS.

FIG. 5.

HIV-1 restricted by Fv1-Cyp can be rescued for significantly longer after infection than HIV-1 restricted by TRIMCyp. CRFK cells expressing Fv1-Cyp358 (A), Fv1-Cyp432 (B), or TRIMCyp (C) were infected with a 10-min pulse of restricted HIV-1 (▪) or unrestricted MLV-B (□). Infection was stopped by addition of 40 mM ammonium chloride, and 5 μM CSA was added at time points after ammonium chloride addition as shown. At 48 h after infection infected cells were enumerated by FACS. Virus restricted by Fv1-Cyp is long-lived and can be rescued several hours after infection. Conversely, virus restricted by TRIMCyp is irreversibly restricted after around 1 h.

RESULTS

We made two Fv1-Cyp fusion proteins by attaching a full-length CypA coding region from the owl monkey TRIMCyp cDNA (31, 38) onto the Fv1 N 3′ end, joining Fv1 N amino acid residues 1 to 358 or 1 to 432 to CypA from owl monkey TRIMCyp. The last 82 or 8 residues of Fv1 N were therefore replaced with full-length CypA. These fusions were called Fv1-Cyp358 and Fv1-Cyp432, respectively, and were stably expressed in feline CRFK cells using the MLV vector CXCR as previously described (23).

(i) Restriction of retroviral infection by Fv1-Cyp fusion proteins.

We prepared VSV-G-pseudotyped, GFP-encoding retroviral vectors derived from HIV-1, HIV-2, FIV, and MLV-N and -B by transfection of 293T cells as previously described (3, 17, 18, 37). We then titrated these vectors into CRFK cells expressing Fv1 N or the Fv1-Cyp fusions in the presence or absence of 5 μM CSA as described previously (3). Unmodified CRFK cells were infected as a control. Infected cells were enumerated by FACS 48 h after infection, and infectious titers were calculated (Fig. 1). The data show that Fv1 N does not restrict the lentiviruses HIV-1, HIV-2, and FIV. However, both Fv1-Cyp fusion proteins restrict HIV-1 and FIV infectivity by 1 to 2 orders of magnitude, and restriction is relieved by treatment with CSA (Fig. 1A and C). Importantly, neither Fv1-Cyp fusion protein restricts the HIV-1 mutant, CA G89V, which cannot recruit CypA due to mutation of the CypA binding site (56) (Fig. 1B). HIV-2 is also not restricted by Fv1-Cyp, a finding consistent with its reported inability to recruit CypA (10) (Fig. 1D). The restriction specificity of Fv1-Cyp is therefore identical to that of the natural TRIMCyp fusion protein expressed in owl monkeys, which restricts both HIV-1 and FIV, in a CSA-sensitive way, via recruitment of CypA to the incoming viral capsids (15, 29, 31, 38). Importantly TRIMCyp also cannot restrict either HIV-1 G89V or HIV-2. We conclude, therefore, that by fusing CypA to Fv1 we have enabled Fv1 to restrict HIV-1 and FIV by recruitment to the incoming HIV-1/FIV capsids via the CypA domain. Titration curves of HIV-1 on unmodified CRFK cells and cells expressing TRIMCyp or the Fv1-Cyp fusions are shown in Fig. 1E to H. Note that restriction by Fv1-Cyp does not impact on the shape of the curve.

FIG. 1.

Restriction of retroviral infection by Fv1-Cyp fusion proteins (A to J). Unmodified CRFK cells (C) and CRFK cells expressing Fv1 N, Fv1-Cyp358, or Fv1-Cyp432 were infected with wild-type or mutant viruses as shown, and infectious titers per milliliter were calculated. Cells were treated with 5 μM CSA or vehicle control (DMSO). Errors are standard deviations of titers derived at three doses of virus. The results are representative of titers of two independent virus preparations on two independently derived cell clones. Titrations of HIV-1 on unmodified CRFK (E) cells or CRFK cells expressing TRIMCyp (F), Fv1-Cyp358 (G), Fv1-Cyp432 (H) are also shown. HIV-1 dose was standardized by RT ELISA (Caviditech, Sweden). MLV-N (I) or MLV-B (J) was also titrated onto CRFK cells expressing Fv1 N or Fv1-Cyp fusions as shown.

Next we tested whether the Fv1-Cyp fusions were able to restrict MLV. They are derived from Fv1 N and might be expected to restrict MLV-B but not MLV-N. We compared the infectivity of MLV-N (Fig. 1I) and MLV-B (Fig. 1J) on CRFK cells expressing wild-type Fv1 N or the Fv1-Cyp fusions or unmodified CRFK cells as a control. Although Fv1 N restricted MLV-B but not MLV-N as expected, neither of the Fv1-Cyp fusions restricted MLV. It appears that Fv1's antiviral activity against MLV has been abrogated by its fusion to CypA. The infectivity of MLV was not affected by 5 μM CSA in the presence of the Fv1-Cyp fusions. Knowing that the antiviral specificity determinants for Fv1 are at residues 358 and 399 and C-terminal to 438 (5, 6), we imagine that in Fv1-Cyp432 we have covered and/or replaced the MLV binding site of Fv1, and in Fv1-Cyp358 we have replaced it with CypA.

(ii) Fv1-Cyp restricted HIV-1 is competent for RT and forms DNA circles.

An important difference in the restriction of MLV by TRIM5α or Fv1 is that TRIM5α, but not Fv1, activity leads to a severe reduction of MLV DNA synthesis (21, 22, 34). We therefore examined the ability of Fv1-Cyp restricted HIV-1 to reverse transcribe by measuring the products of RT by TaqMan qPCR. We infected CRFK cells expressing the Fv1-Cyp fusions in the presence or absence of 5 μM CSA, incubated the cells for 6 h, and then extracted the total DNA. We quantified viral DNA corresponding to early RT product (GFP) and expressed the data as the number of RT products per 100 ng (ca. 10% of the sample) of total DNA (Fig. 2A). A second set of parallel infections was subjected to FACS 48 h after infection to determine infectivity (Fig. 2B). The results show that while HIV-1 infectivity is reduced by an order of magnitude by both Fv1-Cyp fusions, formation of early RT products is not inhibited. FIV DNA synthesis was also not inhibited in this assay using a GFP amplicon as described previously (3) (data not shown). As a control to ensure that DNA detected by qPCR was due to viral RT and not plasmid from the virus prep, we performed identical experiments with virus that had been heat treated at 95°C for 3 min. In these samples, there was no early RT product present 6 h after infection and no infected cells 48 h after infection (data not shown). To test whether late RT products were inhibited by Fv1-Cyp, we measured HIV-1 LTR-gag products, by TaqMan qPCR, as previously described (11, 50) at 6, 12, and 18 h after infection. Late HIV-1 RT products were also not inhibited by Fv1-Cyp expression (Fig. 2C to E) and are maximal at 12 h postinfection as previously observed (11). Viral DNA 2 LTR circles, formed from linear double-stranded RT products, are inhibited in Fv1-restricted infection (21, 45, 51). The enzymes required for circle formation are thought to reside uniquely in the cell nucleus, so circle formation is used as a surrogate for nuclear entry. The lack of circles after Fv1-restricted infection is interpreted as Fv1 preventing restricted MLV from entering the nucleus (21, 51). We measured HIV-1 DNA circles at 18 h postinfection (Fig. 2F) and found that Fv1-Cyp does not block HIV-1 2 LTR circle formation. We used samples infected at higher dose (multiplicity of infection [MOI] = 2) than for the measurement of linear DNA (Fig. 2A to E, MOI = 0.1) since the lower-dose infection did not produce a measurable circle signal (data not shown). Infections performed in parallel were subjected to FACS 48 h after infection to enumerate infected cells (Fig. 2G) and to demonstrate that even with a high-dose infection Fv1-Cyp restricts HIV-1 infectivity. Relieving the block by treatment with CSA reduced the number of circles detected, suggesting that in the absence of restriction some of the linear RT products integrate into the genome, reducing the amount of substrate for circle formation.

FIG. 2.

Fv1-Cyp restricted virus is competent for RT and forms DNA circles. Unmodified CRFK cells (□), CRFK cells expressing Fv1-Cyp358 (▪), or Fv1-Cyp432 (□) were infected with equal doses of DNase-treated HIV-1 encoding GFP in the presence of 5 μM CSA or vehicle control (DMSO). TaqMan PCR was performed on 100 ng (10% of total) samples. The TaqMan amplicon encoded GFP (A); LTR-gag, a late RT product (C to E); or 2 LTR circles (F). DNA samples were prepared 6 h (A and C), 12 h (D), or 18 h (E and F) after infection. As a control CRFK cells (□) or CRFK cells expressing TRIMCyp (□) were infected with HIV-1 as described above, and DNA synthesis was measured as described above using the GFP amplicon (H). The percentage of infected cells was determined in parallel samples by FACS 48 h after infection (B, G, and I). Errors are standard deviations of qPCR values derived from duplicate infections.

As a positive control we analyzed HIV-1 RT in CRFK cells and CRFK cells expressing owl monkey TRIMCyp (22). We found that TRIMCyp strongly blocked HIV-1 DNA synthesis (Fig. 2H), as well as HIV-1 infectivity (Fig. 2I), as has been previously described (1, 38, 46). qPCR using DNA from cells inoculated with heat-treated virus (95°C, 3 min) yielded levels of GFP DNA that were below the limit of reliable detection, indicating that the signals shown are derived from viral RT and not plasmid DNA. Note that the magnitude of the block to DNA synthesis is smaller than the block to infectivity, as is usually the case for restriction by TRIM5α and TRIMCyp (1, 48).

(iii) Fv1-Cyp432 but not Fv1-Cyp358 acts as a dominant-negative inhibitor against wild-type Fv1.

Both Fv1 and TRIM molecules are able to act as dominant-negative inhibitors against related proteins of different antiviral specificity. For example, if Fv1 B is expressed in NIH 3T3 cells, then the cells lose their ability to restrict MLV-B, through endogenous expression of Fv1 N, but become able to restrict MLV-N due to the exogenously expressed Fv1 B (7). The simplest explanation for this observation is that Fv1 molecules multimerize and that the overexpressed protein titrates the endogenous protein. This notion is consistent with Fv1's relatedness to Gag, a protein that readily multimerizes. Similarly, TRIM molecules can homo- and hetero-multimerize, and overexpression of TRIM 34 in human cells oblates TRIM5α's antiviral activity against MLV N (57). The shorter TRIM5 splice variants TRIM5δ and TRIM5γ are also dominant negative to TRIM5α (32, 41). We therefore examined whether Fv1-Cyp fusions were able to act as dominant-negative inhibitors of wild-type Fv1 activity. We expressed the Fv1-Cyp fusions in CRFK cells expressing Fv1 N and tested their ability to restrict MLV (Fig. 3). The data show that the expression of Fv1 N reduces the MLV-B titer by around 2 orders of magnitude. Strikingly, the expression of Fv1-Cyp432 but not Fv1-Cyp358 rescued the infectivity of MLV-B, indicating that the long Fv1-Cyp432 protein but not the short Fv1-Cyp358 is able to act as a dominant-negative agent against Fv1 N. The dose of the virus delivering the Fv1-Cyp proteins in these experiments resulted in an MOI of ca. 1 as tested by measurement of red fluorescent protein expression by FACS in CRFK cells infected in parallel (data not shown).

FIG. 3.

Fv1-Cyp432 but not Fv1-Cyp358 acts as a dominant-negative inhibitor against wild-type Fv1. CRFK cells expressing Fv1 N were infected with MLV vector encoding Fv1-Cyp358 or Fv1-Cyp432. After 48 h the cells were infected with MLV-N or MLV-B encoding GFP. At 48 h postinfection the GFP-expressing cells were enumerated by FACS. Infectious titers were calculated and are expressed as infectious units per milliliter. Errors are standard deviations of titers determined at three doses. The results are representative of two experiments performed with two independent preparations of virus.

(iv) Fv1-Cyp and Fv1 have long half-lives and are not dependent on the proteasome for antiviral activity.

Recent work has demonstrated that TRIM5α has a short half-life in cells and is rapidly turned over by the proteasome (14, 27). We therefore examined the turnover rate of HA-tagged Fv1 and Fv1-Cyp358 by inhibiting protein synthesis with cycloheximide and measuring specific protein levels at various time points after cycloheximide addition, by Western blotting (Fig. 4). The data show that Fv1 (Fig. 4A) and Fv1-Cyp (Fig. 4B) half-lives are long and that there is no significant reduction of protein levels 7 h after the addition of cycloheximide. As a positive control we performed the same experiment with TRIM5α. TRIM5α has a shorter half-life and has almost disappeared 2 h after cycloheximide addition (Fig. 4C), as has previously been described for both TRIM5α and TRIMCyp (14, 27). As a control for protein loading we stripped the Western blots and reprobed using anti-CypA (Fig. 4A,B) or actin (Fig. 4C) antibodies. Fv1 and TRIM5α were detected via an N-terminal or C-terminal HA tag, respectively, and Fv1-Cyp was detected via its CypA domain.

FIG. 4.

Fv1-Cyp and Fv1 have long half-lives and are not dependent on the proteasome for antiviral activity. (A) Western blot showing Fv1-Cyp358 protein levels in CRFK cells stably expressing Fv1-Cyp358 from an MLV vector. Cells were either untreated (UT) or treated with cycloheximide for various durations. CypA protein levels are measured by Western blot as a loading control. (B) Western blot showing HA-Fv1 N protein levels in CRFK cells stably expressing HA-Fv1 N from an MLV vector. Cells were either untreated (UT) or treated with cycloheximide for various durations. Unmodified CRFK cells are shown as a negative control. CypA protein levels are measured by Western blot as a loading control. (C) Western blot showing TRIM5α-HA protein levels in CRFK cells stably expressing TRIM5α-HA from an MLV vector. Cells were either untreated (UT) or treated with cycloheximide for various durations. Unmodified CRFK cells are shown as a negative control. Actin protein levels are measured by Western blot as a loading control. (D) HIV-1 or FIV titers were determined on CRFK cells or CRFK cells expressing Fv1-Cyp358 or Fv1-Cyp432 in the presence (□) or absence (▪) of 1 μg/ml MG132. (E) MLV-N or MLV-B titers were determined on CRFK cells expressing Fv1 N in the presence (□) or absence (▪) of 1 μg/ml MG132. Errors are the standard deviation of titers determined at different doses. As a control for MG132, human TE671 cells were infected in triplicate with MLV-N (□) or MLV-B (▒). (F) Two samples had total DNA extracted 6 h after infection and were subjected to TaqMan QPCR using a GFP amplicon. (G) One sample was subjected to FACS 48 h after infection to determine the percentage of infected cells. Errors are standard deviations of two qPCR values determined from parallel infections. The data are representative of two independently performed experiments.

Inhibition of the proteasome has recently been shown to allow TRIM5α-restricted virus to synthesize viral DNA by RT (1, 48). The virus remains restricted however, suggesting that the antiviral mechanism is not simply recruitment of the virus to the proteasome by TRIM5α. In order to examine the role of the proteasome in restriction by Fv1 and Fv1-Cyp, we measured infectious titer of HIV-1 and FIV on CRFK cells expressing the Fv1-Cyp fusions or MLV, on cells expressing Fv1 N. Cells were treated with 1 μg of the proteasome inhibitor MG132/ml at the time of infection or left untreated as a control (Fig. 4D and E). The data show that treatment with MG132 cannot completely rescue restricted infection. Interestingly, restriction of FIV by Fv1-Cyp was reliably rescued by a few fold, whereas FIV infectivity on unrestricting cells was not. This effect was small, however, and FIV was still strongly restricted by Fv1-Cyp in cells treated with MG132. As a control for MG132 activity we measured the infectivity of, and DNA synthesis by, MLV-N and MLV-B in the presence or absence of MG132 on TE671 cells, a human cell line that restricts MLV-N. MLV-N DNA synthesis (Fig. 4F), but not infectivity (Fig. 4G), was rescued by treatment with MG132, as described previously (1). Neither infectivity nor DNA synthesis of unrestricted MLV-B is significantly affected by MG132 treatment on TE671 cells. These data demonstrate that restriction by Fv1 or Fv1-Cyp does not rely on proteasome activity.

HIV-1 restricted by Fv1-Cyp can be rescued for significantly longer after infection than HIV-1 restricted by TRIMCyp.

The observation that Fv1 and Fv1-Cyp block infectivity after viral DNA synthesis, combined with the observation that both proteins have long half-lives and restrict independently of the proteasome suggests that the restricted HIV-1/Fv1-Cyp complex might be long-lived within cells. HIV-1 restricted by Fv1-Cyp may therefore be sensitive to rescue by CSA for longer after infection than HIV-1 restricted by TRIMCyp, which has been shown to be rendered insensitive to CSA rapidly after infection (33). To test this possibility, we exposed CRFK cells expressing either Fv1-Cyp or TRIMCyp to a 10-min pulse of HIV-1 and then chased this pulse with addition of CSA at intervals. The percentages of infected cells were determined 48 h later by FACS. The data presented in Fig. 5 clearly show that HIV-1 restricted by Fv1-Cyp358 is long-lived and can be rescued by the addition of CSA even when CSA is added 10 h after infection. Virus restricted by Fv1-Cyp432 is also significantly rescued, even if CSA is added 5 h after infection. Conversely, and as previously described (33), HIV-1 restricted by TRIMCyp can only be rescued for a short time after infection and cannot be rescued at all after 2 h. In a parallel experiment MLV-B remained unaffected by Fv1-Cyp or TRIMCyp, and its infectivity was not impacted by treatment with CSA.

DISCUSSION

This study demonstrates that fusing CypA to the murine restriction factor Fv1 allows it to restrict HIV-1 and FIV (Fig. 1). CSA treatment relieves restriction and restores infectivity, indicating that interaction between CypA and viral gag is essential for restriction. We conclude that these viruses recruit incoming CypA and therefore Fv1-Cyp to their capsid and become restricted by the Fv1 moiety. Intriguingly, HIV-1 restricted by Fv1-Cyp synthesizes DNA, as does Fv1-restricted MLV (21, 51) (Fig. 2). Neither late LTR gag RT products nor the earlier GFP encoding RT products were inhibited by Fv1-Cyp fusions. TRIM5α tends to significantly reduce viral DNA synthesis, although both Fv1 and TRIM5α interact with the virus early after its entry and can compete for the incoming virus (32). We also measured HIV-1 2 LTR circles, which account for ca. 10% of the circles formed (11). Fv1 blocks viral DNA circle formation (21, 51) but, strikingly, circles are not blocked during restriction of HIV-1 by Fv1-Cyp (Fig. 2F). Importantly, relieving the block to infection with CSA reduced the amount of 2 LTR circles. This suggests that a proportion of the unintegrated DNA from restricted virions is circularized in the presence of Fv1-Cyp. This is reminiscent of the observed increase in 2 LTR circle formation in the presence of integrase inhibitors (11). Relieving restriction with CSA reduces circles presumably because linear DNA integrates to form proviruses, before it gets circularized. These data also indicate that Fv1-Cyp blocks HIV-1 after RT and suggest that the restricted virus enters the nucleus. This result is perhaps surprising since Fv1 blocks MLV circle formation. This might suggest that Fv1 and Fv1-Cyp have different mechanisms of restriction, but it is also possible that it reflects the different nuclear entry properties of HIV-1 compared to MLV. Our data are reminiscent of a study by Yap et al. (52) in which CypA was fused to a variety of human tripartite motif proteins, some of which restricted HIV-1 infection but allowed normal levels of circle formation.

Expression of the Fv1-Cyp fusions in the presence of Fv1 indicates that the longer Fv1-Cyp432 is dominant negative against the anti-MLV activity of Fv1 but the shorter Fv1-Cyp358 is not. The most likely explanation for this is that replacing the last 82 residues of Fv1 with CypA generates a molecule that cannot interact with full-length Fv1. Presumably, fusing CypA close to the C-terminal end of Fv1 does not prevent the fusion protein from multimerizing with Fv1. Formally, it is possible that the shorter Fv1-Cyp fusion does not multimerize at all, but we believe that this is unlikely since the identical antiviral specificities and similar stage of the block to infection mediated by the two fusion proteins strongly suggests that they restrict infection by the same mechanism.

The rescue of TRIM5α restricted DNA synthesis, by inhibition of the proteasome (Fig. 4) (1, 48), suggests that, normally, the TRIM5-virus complex is degraded by the proteasome too rapidly for the virus to make a significant amount of DNA. Thus, inhibition of the proteasome protects the TRIM5/virus complex and allows efficient RT. The block to RT might therefore be a consequence of rapid TRIM5α turnover rather than a direct consequence of its antiviral activity (1, 48). We propose that the later block to Fv1-restricted virus is due to its longer half-life, as demonstrated in Fig. 4, as well as by the ability of CSA to rescue the infectivity of Fv1-Cyp-restricted virus for at least 10 h after infection (Fig. 5). We assume that the longer half-life allows the restricted virus to reverse transcribe. Such a model is consistent with evidence that TRIM5 is an E3 ligase for ubiquitin, capable of auto-ubiquitinylation (49), and has a short half-life due to recruitment to the proteasome (14). Conversely, Fv1 is gag-like and unlikely to have such properties.

The molecular details of the mechanism of action of both TRIM5 and Fv1 remain incompletely solved. The loss of a particulate capsid fraction in the presence of TRIM5 has been interpreted as TRIM5 rapidly uncoating sensitive, incoming cores (34, 42). Intriguingly, Fv1 has been reported to block infectivity without having an impact on particulate capsid levels (34). The loss of particulate capsid may therefore be associated with proteasome activity. It will be interesting to follow the fate of the capsid in the presence or absence of TRIM5 and MG132, where the TRIM5-virus complex is protected from the proteasome but TRIM5 retains its antiviral properties.

The involvement of cyclophilins in innate immunity in plants (13), as well as hepatitis C (47) and vaccinia replication (12), suggests that the role of cyclophilins in immunity and viral replication might be more general than previously thought. Certainly, the continued search for antiviral molecules active against HIV-1 and the study of their antiviral mechanism will provide insight essential for the development of novel antiviral therapeutics and the improvement of animal models for infection.