Restriction of V86M CA HIV-1 by R332G-R335G TRIM5α_hu is independent of cyclophilin A

Although V86 is located in the CypA-binding loop of HIV-1 CA, this mutant was previously shown to retain CA binding to CypA [35]. Accordingly, transduction of lymphocytes and macrophages by V86M HIV-1 was significantly affected by treatment by 5 μM of cyclosporine A (CsA), a drug that competes with CA for binding to CypA (Additional file 1: Figure S1). We nonetheless hypothesized that the mutation might affect functional interactions between CypA, TRIM5α and CA. To directly analyze the role of CypA in V86M-mediated resistance, we knocked down its expression in TE671 and in Sup-T1 cells that had been transduced with WT or R332G-R335G TRIM5α_hu or with the empty vector. The knockdown was efficient for all the TRIM5 variants and in both cell contexts (Additional file 1: Figure S2). Non-transduced cells were eliminated by antibiotic treatment, and cells were then challenged with either WT or V86M NL4-3 vectors in the presence or absence of 2 μM CsA (Figure 2). In control permissive cells, replication of both WT and V86M HIV-1_NL-GFP was decreased (25 to 35%) either by expression of the CypA shRNA or by CsA treatment [25,43] (Figures 2A and ​and2B,2B, left panels). As expected, combining CsA treatment and CypA knockdown had no additional inhibitory effect on HIV-1 replication in the two cell lines, confirming that CsA inhibits incoming HIV-1 in human cells by interfering with its interactions with CypA [25,27,43]. In TE671 and Sup-T1 cells expressing R332G-R335G TRIM5α_hu, replication of WT HIV-1 was strongly reduced compared with the control “vector” cells (~50-fold and 20-fold, respectively) and like before, V86M partly rescued infection of HIV-1 in these cells (5-fold in TE671 cells, 2.5-fold in Sup-T1 cells). CypA knockdown enhanced infectivity of WT HIV-1_NL-GFP by 5-fold in TE671 cells and by 3-fold in Sup-T1 cells, but had no significant effect on the replication of the V86M mutant. Likewise, CsA treatment strongly increased replication of WT HIV-1_NL-GFP in R332G-R335G TRIM5α_hu cells (about 10-times in TE671, 5-times in Sup-T1 cells) but had a much smaller effect on V86M HIV-1_NL-GFP (1.5- to 2-fold in both cell lines). Therefore, replication of V86M CA HIV-1 in cells expressing R332G-R335G TRIM5α_hu was not efficiently rescued by depletion or inhibition of CypA.

Open in a separate window

Figure 2

Restriction of CA-V86M HIV-1 by R332G-R335G TRIM5α_hu is independent of cyclophilin A. TE671 cells (A) and Sup-T1 cells (B) transduced with either WT TRIM5α_hu, R332G-R335G TRIM5α_hu or with the “empty” vector were also transduced with shRNAs targeting CypA or the non-relevant luciferase mRNAs. Following antibiotic selection, cells were challenged with a single dose of WT or V86M HIV-1_NL-GFP, in the presence or absence of 2 μM CsA. A multiplicity of infection of ~0.2 was used, as determined for the WT virus in parental cells and in the absence of drug. WT and V86M virus preparations were adjusted by RT assay. Cells were analyzed by FACS two days after infection, and the bars show the average % infected cells from 3 infections with standard deviations. Note the difference in scale between the y-axis of different graphs.

CsA concentration-dependent assays

In this experimental setting, instead of using a single volume of HIV-1_NL-GFP for all infections as in Figure 2, we adjusted the amount of virus used for each virus/cell combination so that approximately 1% of cells were infected in the absence of drug, and we then repeated the infection in presence of increasing CsA concentrations (Figure 3). Like before, CsA decreased replication of both WT and V86M HIV-1_NL-GFP in the permissive control “vector” cells by 30% to 50% in both TE671 cells (Figure 3A) and Sup-T1 cells (Figure 3B). CsA rescued replication of WT HIV-1_NL-GFP in cells expressing R332G-R335G TRIM5α_hu, by up to 6-fold in TE671 cells and 8-fold in Sup-T1 cells. In contrast, the impact of the drug on the restriction of the V86M mutant by this variant of TRIM5α_hu was much smaller: up to 1.5-fold and 2-fold in TE671 and Sup-T1 cells, respectively.

Open in a separate window

Figure 3

CsA dose-dependent analysis. TE671 cells (A) and Sup-T1 cells (B) expressing R332G-R335G TRIM5α_hu or transduced with the “empty” vector were challenged with WT or V86M HIV-1_NL-GFP in the presence of increasing concentrations of CsA. Virus doses were adjusted so that approximately 1% of cells were infected in the absence of CsA. The percentages of infected cells were determined two days later as before and results are shown as –fold increases relative to the relevant no-drug controls.

Altogether, data in Figures 2 and ​and33 show that restriction of V86M HIV-1 replication by R332G-R335G TRIM5α_hu is poorly sensitive to the presence of CypA compared to restriction of the WT control. However, replication of V86M HIV-1 in permissive human cells is as much sensitive as WT HIV-1 to the presence of CypA.

The sensitivity of HIV-1 restriction by TRIM5α_hu mutants to the V86M mutation correlates with its sensitivity to CsA treatment

Data shown in Figures 1 and ​and22 revealed that restriction of HIV-1 by R332G-R335G TRIM5α_hu was counteracted either by the V86M mutation or by CsA treatment, and suggested that the two interventions had redundant effects. In order to better characterize the role of CypA in the TRIM5α resistance conferred by V86M, we analyzed the effect of this mutation on other mutants of TRIM5α_hu. We also analyzed the effect of CsA treatment on restriction by these same mutants. G330E, R332G, R335G, and G330E-R332G-R335G TRIM5α_hu all restricted HIV-1 when expressed in TE671 cells, albeit with various efficacies (Additional file 1: Figure S3A), and as shown before [38,42]. Replication of HIV-1_NL-GFP in these cells was enhanced either by introducing the V86M mutation (Additional file 1: Figure S3A) or by CsA treatment (Additional file 1: Figure S3B). When we plotted the increase in HIV-1_NL-GFP infectivity resulting from V86M against the increase in infectivity resulting from CsA treatment in TE671 expressing these various TRIM5α_hu, we observed a highly significant correlation (Figure 4). Also, the level of rescue of HIV-1_NL-GFP infectivity by either V86M mutation or CsA treatment was grossly proportional to the level of restriction. This analysis supports our findings that V86M and CsA treatment rescue HIV-1 replication in restrictive conditions by redundant mechanisms. However, the slope of the linear curve (1.89) also shows that CsA is more efficient than V86M at rescuing HIV-1 replication.

Open in a separate window

Figure 4

Direct correlation between the effect of mutation V86M and the effect of CsA treatment. The increase in HIV-1 infectivity in TE671 cells expressing the indicated TRIM5α_hu in presence of 2 μM CsA was plotted against the increase in infectivity resulting from the V86M mutation. All values are –fold increases relative to the untreated WT control virus. Data were gathered at a multiplicity of infection yielding close to 1% infected cells for the untreated WT HIV-1 in the cell lines expressing each the various TRIM5α. The correlation is significant, as shown by the R² obtained and the Spearman’s P-value. The linear regression equation is shown.

Restriction of V86M CA HIV-1 by R332G-R335G TRIM5α_hu in cat cells is insensitive to CsA

In human cells expressing an HIV-1-restrictive TRIM5α, CsA treatment might cause dual, opposite effects on HIV-1: enhancement by counteracting TRIM5α, and inhibition by a yet unclear, CypA-dependent mechanism. In addition, although endogenous TRIM5α_hu is thought to be recessive against the over-expressed TRIM5α in our experimental system, there was still a slight possibility for interference effects. To address these issues, we analyzed the rescue of HIV-1 replication by V86M mutation and/or by CsA treatment under restriction by R332G-R335G TRIM5α_hu in CRFK cat fibroblasts. These cells do not express any TRIM5α, and CsA treatment does not inhibit WT HIV-1 infection in cat cells [26,44]. We found that R332G-R335G TRIM5α_hu strongly restricted HIV-1_NL-GFP, and the mutation V86M in CA decreased restriction by about 3-fold (Figure 5). WT HIV-1_NL-GFP replication was enhanced 10-times by CsA treatment while the effect of the drug on the replication of V86M HIV-1_NL-GFP was much smaller (1.5-fold), thus mirroring the data obtained in human cells.

Open in a separate window

Figure 5

V86M HIV-1 restriction by R332G-R335G TRIM5α_hu in cat cells is not counteracted by CsA treatment. CRFK cells were transduced with R332G-R335G TRIM5α_hu or with the empty vector as control and subsequently challenged with a single dose of WT or V86M HIV-1_NL-GFP in presence or absence of 2 μM CsA. Infections were done at a multiplicity of infection of ~0.2 as determined in parental cells infected with the WT virus, and the amounts of viruses used were normalized by RT assay. The percentages of infected cells were determined two days later by FACS.

V86M CA interactions with CypA studied by NMR

Data in Figures 1, ​,2,2, ​,3,3, ​,44 and ​and55 showed that V86M made HIV-1 resistant to CypA-dependent, TRIM5α-mediated restriction mechanisms. Thus, we hypothesized that V86M could modify the interactions between HIV-1 CA and CypA, or the isomerisation of the former by the latter [33,45]. The interaction of CA with CypA was studied by 2D ¹H,¹⁵N ZZ-exchange NMR spectroscopy (Figure 6). In the absence of CypA, we observed two G89 N-H correlation peaks in the V86M capsid spectra, indicating that, as in WT, the GP bond exists in both cis and trans states. However, the peak positions for both states in V86M were chemically shifted compared to WT, suggesting that the mutation influences the conformations adopted by the cyclophilin-binding loop (Figure 6A). Furthermore, there is evidence of a second minor G89_cis peak in the V86M spectra that is indicative of discrete CypA-binding loop isomers.

Open in a separate window

Figure 6

V86M CA interactions with CypA studied by NMR. (A) Expansions of 2D ¹H-¹⁵N ZZ-exchange spectra illustrating CypA catalysed cis-trans proline isomerisation of P90 in HIV1caN. The four panels represent backbone amide 1H,15N correlations for the preceding G89 residue in HIV1caN and in HIV1caN V86M in the presence and absence of CypA. In the absence of CypA, the isomerisation reaction is slow and only distinct G89 cis- and trans- correlation peaks are observed. The presence of CypA leads to an increase in the isomerisation rate and the accumulation of cis and trans exchange peaks. The dotted box denotes the location of the exchange peaks and indicates their symmetrical distribution with regard to G89_cis and G89_trans. Note that in the case of HIV1caN V86M a broad G89cis peak and an additional cis-exchange peak indicates a second conformer. All spectra were acquired at the same mixing time (69 ms) and processed with the same contour levels. (B) Evaluation of exchange rates for CypA catalyzed isomerisation of G89-P90 in HIV1caN and HIV1caN V86M. Normalised peak intensities of trans auto peaks and the corresponding exchange peaks for HIV1caN/CypA and HIV1caN V86M/CypA were plotted as a function of mixing time (s). Exchange rates (k_ex) in the order of 5 s^-1 for HIV1caN/CypA and 20 s^-1 for HIV1caN V86M/CypA were extracted by fitting auto peaks and exchange peaks as described by Bosco et al. (2002).

Upon addition of CypA, symmetrical exchange peaks were observed for WT and V86M indicating that both capsids are a substrate for CypA and that increased cis-trans exchange occurs during the ZZ-exchange experiment. However, in the V86M mutant, at least two cis exchange peaks were observed compared to a single peak in the WT. Furthermore, the intensities and line shapes of these additional peaks were directly affected by addition of CypA and dependent upon the mixing time of the experiment (Figure 6B). This supports the hypothesis that the V86M mutation results in distinct populations of conformers for the G89-P90 bond.

Next, we determined the catalysed isomerisation rate of the G89-P90 bond in both WT and V86M capsids in the presence of CypA (Figure 6B). For WT capsid we observed a similar rate as previously published, of ~5 s^-1. For the V86M capsid, the exchange rate for the major G89 peak was ~20 s^-1. This suggests that the V86M mutation has both altered the conformations adopted by the CypA-binding loop and increased the rate at which it is isomerised. The presence of an additional G89_cis peak in V86M in the absence of CypA suggests that the mutant capsid may also undergo faster uncatalysed isomerisation, however we were not able to quantify this under the conditions tested.

Effects of V86M on early stages of viral replication

qPCR assays were performed on DNA extracted from TE671 cells expressing either WT TRIM5α_hu, R332G-R335G TRIM5α_hu or, as a positive control, TRIM5α_rh, and then infected with WT or V86M HIV-1_NL-GFP (Figure 7). We quantified GFP DNA as a marker for late reverse transcription products and 2-LTR circles as a marker for nuclear viral DNA. As expected, accumulation of HIV-1 DNA was reduced in cells expressing TRIM5α_rh and R332G-R335G TRIM5α_hu, and levels of nuclear HIV-1 DNA were even more strongly decreased. This is consistent with HIV-1 replication being impeded by TRIM5α at both the reverse transcription and nuclear transport stages [17,22]. The amounts of viral DNAs were more decreased in TRIM5α_rh cells than in R332G-R335G TRIM5α_hu cells, consistent with the greater magnitude of restriction conferred by TRIM5α_rh. V86M had no effect on total DNA levels but significantly increased relative amounts of 2-LTR nuclear DNA species. However, V86M 2-LTR DNA levels in restrictive cells were still reduced 5- to 10-times compared with levels of WT 2-LTR DNA in non-restrictive conditions, as expected from the observation that V86M only partially rescues HIV-1 in restrictive conditions. In summary, escape from restriction by this mutant is associated with increased HIV-1 transport to the nucleus.

Open in a separate window

Figure 7

V86M enhances nuclear transport of HIV-1 DNA in restrictive conditions. TE671 cells expressing the indicated TRIM5α were challenged with WT or V86M HIV-1_NL-GFP. An RT inhibitor (nevirapine) control was included to show the absence of contaminating DNA. The levels of HIV-1 late reverse transcription products and 2-LTR circles were determined by real-time quantitative PCR using primers to detect GFP or 2-LTR junctions, respectively. DNA amounts were calculated using serially diluted, linearized DNAs as standards. Data are presented as mean ratio of late RT or 2-LTR DNA copies over the number of GAPDH DNA copies, normalized to the permissive control TRIM5α_hu. Shown are values with standard deviations of 3 or 5 independent infections (late RT products and 2-LTR, respectively). Student’s t-test was used to test for significance of data, specifically WT vs V86M viruses in cells expressing R332G-R335G or TRIM5α (p=0.0020 and p=0.0015, respectively).

Cells

Human rhabdomyosarcoma TE671 cells, human embryonic kidney 293 T cells and feline renal CRFK cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum and antibiotics at 37°C. Human T lymphoblast Sup-T1 cells and monocytic cells THP-1 were maintained in RPMI-1640 supplemented with 10% fetal bovine serum and antibiotics at 37°C. All cell culture reagents were from Hyclone (Thermo Scientific, Logan, UT, USA). Human PBMCs were isolated from diluted whole blood of a healthy donor by Ficoll density gradient centrifugation as previously described [65]. CD4⁺ T cells were then isolated using a negative isolation kit (Dynabeads Untouched Human CD4 T Cells, Invitrogen) according to manufacturer specifications. The isolated cells were cultivated in RPMI-1640 supplemented with 10% fetal bovine serum and antibiotics. Activation and proliferation of CD4⁺ T cells were performed by supplementing the culture medium with 60 U/ml IL-2 (Peprotech, Canada) and incubating the cells with anti-CD3/CD28 coated beads (Dynabeads Human T-Activator CD3/CD28, Invitrogen) according to the manufacturer instructions. Infections of these cells were performed in the absence of IL-2 and activating beads. THP-1 monocytic cells (7.5 x 10⁵) were differentiated into M0 macrophages by incubation in serum-free medium supplemented with 50 ng/ml PMA and 10 ng/ml GM-CSF for 48 h. Polarization into pro-inflammatory M1 macrophages was induced by addition of 100 ng/ml LPS and 25 U/ml IFN-γ. Anti-inflammatory M2 macrophages were induced by addition of 20 ng/ml IL-4 and 20 ng/ml IL-10 (all reagents were from Peprotech). In all cases, non-adherent cells were washed away before infection.

Retroviral vectors production

HIV-1 and MLV-based vectors were produced through transient transfection of 293 T cells and collected as previously described [38,44]. To produce HIV-1_NL-GFP, cells were co-transfected with pNL-GFP and pMD-G. To produce the MLV-based vectors SRBl and MIP, cells were transfected with the relevant pMIP or pSRBl plasmid and co-transfected with pCL-Eco and pMD-G. To produce HIV-1 VLPs, cells were co-transfected with pΔR8.9 and pMD-G. When necessary, amounts of viruses were normalized using the nonradioactive EnzChek RT Assay Kit (Invitrogen, Burlington, ON), or based on their titer in permissive human cells.

RNA interference

pSRBl-based retroviral vectors encoding an shRNA targeting either CypA or Luc mRNAs were produced in 293 T cells as described [28,38]. TE671 and Sup-T1 cells were plated at 250,000 cells per well in 12-well plates and exposed to 1 mL of either SRBL-CypA or -Luc vector preparation. Two days post-transduction, cells were placed in medium containing 10 μg/ml of blasticidin (Sigma-Aldrich). Selection was allowed to proceed for 1 week. Efficiency of the knockdown was verified by assaying CypA expression levels in the transduced cells by western blotting, using antibodies directed against CypA (rabit polyclonal IgG; Sigma-Aldrich, StLouis, MI) and actin (mouse monoclonal; Chemicon International).

Viral challenges

Cells were plated in 24-well plates at 50,000 cells/well (TE671 and CRFK), 100,000 cells/well (Sup-T1), 200,000 cells/well (primary CD4 T cells) or 750,000 cells/well (differentiated THP-1) and infected with HIV-1 vectors (either HIV-1_TRIP-CMV-GFP or HIV-1_NL-GFP) with or without CsA treatment (2 or 5 μM). Drugs were added 15 min prior to infections. In CsA dose-dependent experiments, virus doses were adjusted for each virus-cell combination so that approximately 1% of the cells would be infected in the absence of CsA. Two days post-infection, cells were trypsinised when necessary, fixed in 1 to 2% formaldehyde in a PBS solution. The % of GFP-positive cells were then determined by analyzing 10,000 to 20,000 cells with a FC500 MPL cytometer (Beckman Coulter) using the CXP software (Beckman Coulter). For spreading infections, quantification of RT activity in the supernatants of infected cells was done exactly as before [38].

Quantitative real-time PCR of HIV-1 DNA

Cells were plated in 12-well plates at 3x10⁵ cells/well and infected with HIV-1_NL-GFP WT or V86M vectors. Viruses were passed through 0.45 μm filters and pretreated for 1 hour at 37°C with 20U/ml DNAse I (New England Biolabs) to prevent contamination by carry-over plasmid DNA. In addition, control infections were performed in presence of 80 μM nevirapine to ascertain the absence of such contaminating plasmid DNA. Total cellular DNA was prepared after 6 hours of infection (late RT products) or 6 hours of infection followed by 18 hours of incubation in virus-free medium (2LTR-circles) and DNA preparation was done using the DNeasy Blood and Tissue Kit (Qiagen, California).

The primer sets to detect each DNA species were as follows: GFP forward, GACGACGGCAACTACAAGAC; GFP reverse, TCGTCCATGCCGAGAGTGAT; 2LTR-circles forward (MH535), AACTAGGGAACCCACTGCTTAAG [66]; 2LTR-circles reverse (MH536), TCCACAGATCAAGGATATCTTGTC; GAPDH forward, GTCAGTGGTGGACCTGACCT; GAPDH reverse, TGAGCTTGACAAAGTGGTCG. In each experiment, a standard curve of the amplicon being measured was run in duplicate ranging from 30 to 3×10⁵ copies plus a no-template control. Reactions contained 1× SensiFAST SYBR Lo-Rox kit (Bioline, UK), 300nM forward and reverse primers, and 5 μl template DNA (100-300 ng) in 20 μl-sized reactions. After initial incubation step of 3 min at 95°C, 40 cycles of amplification were carried out at 10s at 95°C, 10s at 62°C (GFP) or 65°C (2LTR, GAPDH) and 10s (GFP) or 15 s (2LTR, GAPDH) at 72°C on a MX3000P qPCR system (Agilent Technologies, California). Results were analyzed with the MxPro software (Agilent Technologies).

Statistical analyses

Statistical data were calculated using GraphPad Prism version 5.03. Student’s unpaired t-tests were used for tests of difference between means.