|Home | About | Journals | Submit | Contact Us | Français|
Feline immunodeficiency virus (FIV) is a lentivirus that causes AIDS in domestic cats, similar to human immunodeficiency virus (HIV)/AIDS in humans. The FIV accessory protein Vif abrogates the inhibition of infection by cat APOBEC3 restriction factors. FIV also encodes a multifunctional OrfA accessory protein that has characteristics similar to HIV Tat, Vpu, Vpr, and Nef. To examine the role of vif and orfA accessory genes in FIV replication and pathogenicity, we generated chimeras between two FIV molecular clones with divergent disease potentials: a highly pathogenic isolate that replicates rapidly in vitro and is associated with significant immunopathology in vivo, FIV-C36 (referred to here as high-virulence FIV [HV-FIV]), and a less-pathogenic strain, FIV-PPR (referred to here as low-virulence FIV [LV-FIV]). Using PCR-driven overlap extension, we produced viruses in which vif, orfA, or both genes from virulent HV-FIV replaced equivalent genes in LV-FIV. The generation of these chimeras is more straightforward in FIV than in primate lentiviruses, since FIV accessory gene open reading frames have very little overlap with other genes. All three chimeric viruses exhibited increased replication kinetics in vitro compared to the replication kinetics of LV-FIV. Chimeras containing HV-Vif or Vif/OrfA had replication rates equivalent to those of the virulent HV-FIV parental virus. Furthermore, small interfering RNA knockdown of feline APOBEC3 genes resulted in equalization of replication rates between LV-FIV and LV-FIV encoding HV-FIV Vif. These findings demonstrate that Vif-APOBEC interactions play a key role in controlling the replication and pathogenicity of this immunodeficiency-inducing virus in its native host species and that accessory genes act as mediators of lentiviral strain-specific virulence.
Feline immunodeficiency virus (FIV) is a lentivirus closely related to human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV) that infects numerous feline species (1–3). FIV infection of the domestic cat is an important model for HIV, as it is the only nonprimate lentivirus that causes an AIDS-like syndrome in its natural host. Feline and human AIDS are both characterized by recrudescence of high levels of circulating virus, progressive loss of CD4+ T-lymphocytes, neutropenia, weight loss, gingivitis, neurological impairment, and opportunistic infections (4–8). FIV is also similar to HIV in its cellular tropism (9, 10), genomic organization (11), coding of regulatory accessory proteins (12), and utilization of a two-receptor cellular entry mechanism (13–16). These similarities and a well-developed experimental cat infection model (9, 12, 17, 18) facilitate the study of lentiviral pathogenesis in a native host species, as well as the development of preventative and therapeutic strategies.
HIV disease progression to AIDS varies dramatically between infected individuals, and understanding the basis for this variability is critical to combating the disease. While host genetics, immunity, and underlying medical conditions account for a portion of this variability (19–21), the pathogenicity of the infecting virus has also been associated with disease progression (22–29). Furthermore, HIV replication capacity and pathogenicity can vary between viral subtypes (30–35) or between strains within the same subtype (36–39). The basis for this variability in disease potential is still poorly understood.
Similar to HIV, FIV exhibits structural and functional diversity that dictates the nature and extent of its pathogenicity in vivo. There are multiple genetic subtypes of FIV (40–42), and strains within these subtypes exhibit various degrees of pathogenicity (5, 43–46). One subtype C isolate causes particularly high disease incidence and severity, with 60% of young cats developing severe acute immunodeficiency syndrome within 12 to 18 weeks following experimental infection (5). We previously characterized a molecular clone, FIV-C36 (referred to hereinafter as high-virulence FIV [HV-FIV]), that recapitulates the highly pathogenic phenotype of the parental virus (17). In contrast, the subtype A molecular clone FIV-PPR (referred to hereinafter as low-virulence FIV [LV-FIV]) exhibits relatively low pathogenicity in experimentally infected cats and replicates to significantly lower levels than HV-FIV in vitro and in vivo (46, 47). To understand the basis for this difference in pathogenicity, we previously produced chimeras between these HV- and LV-FIV strains and compared their replication in vitro and pathogenicity in vivo. FIV-PCenv, a chimera containing the 3′ half of HV-FIV, including vif, orfA, rev1, and env, on the LV-FIV background demonstrated intermediate, although delayed, viral loads and hematological pathology during primary infection of cats (47). Passage of FIV-PCenv from infected cats into naive cats resulted in higher viral loads, similar to those observed during parental HV-FIV infections, which are not delayed compared to the infection kinetics of parental constructs (48). Thus, the results from experiments with subtype A/C chimeric FIV-PCenv suggest that elements from the 3′ half of the genome, including the accessory genes vif and orfA, contribute to the heightened virulence observed during infections with the HV-FIV strain.
The FIV vif gene is necessary for virus replication in vitro (49–51) and in vivo (52, 53). Lentiviral Vif promotes viral replication by binding host APOBEC3 (A3) cytidine deaminases and inducing their degradation via the proteasome (54–57). In the absence of functional Vif, A3 proteins are incorporated into assembling virus particles (55, 58, 59). After virions carrying A3 protein infect a new cell, A3 can deaminate cytidines in lentiviral minus-strand DNA, resulting in G-to-A hypermutation in the provirus (58, 60–62). Additionally, A3 proteins also appear to reduce the accumulation of reverse transcription products and inhibit lentiviral replication by a deamination-independent mechanism (63–66). Feline species encode four A3 genes, including three very similar A3Z2 genes (a to c) and an A3Z3 gene (67–69). These genes were formerly known as A3C (c, a, and b) and A3H, respectively (70). In addition to these four one-domain proteins, a two-domain A3Z2-Z3 protein (formerly A3CH) is produced by alternative splicing (68, 71). The replication of FIV lacking Vif or wild-type HIV is strongly restricted by feline A3Z3 and A3Z2-Z3, while the A3Z2 proteins have no effect (68, 69, 71, 72). The expression of FIV Vif reverses feline A3Z3- and A3Z2-Z3-mediated lentiviral restriction (68, 71, 72).
FIV does not encode distinct regulatory proteins that are directly comparable to HIV Tat, Vpr, Vpu, or Nef. However, the FIV genome does include a multifunctional 77-amino-acid nonstructural viral protein expressed from the orfA gene, located between vif and env. OrfA is necessary for efficient FIV replication in lymphocytes in vitro and in vivo (52, 73–76). Multiple potential functions have been ascribed to OrfA, including transactivation of viral protein expression (77), effects on virion formation and infectivity (73), cell cycle arrest (78), effects on cellular gene expression (79), and downregulation of cell surface expression of the primary FIV receptor, CD134 (80).
Our previous studies implicating 3′ viral genomic elements in the rapid-growth kinetics and heightened virulence of HV-FIV relative to the growth kinetics and virulence of LV-FIV (47, 48) led us to hypothesize that the FIV regulatory accessory genes vif and/or orfA may be responsible for this difference in viral replication capacity. Here, we describe the construction of three chimeric viruses in which LV-FIV vif, orfA, or both were replaced with HV-FIV versions of these genes. Comparison of the growth kinetics of these constructs in multiple ex vivo contexts and in the presence or absence of feline A3 suggests that (i) accessory genes convey different capacities for in vitro replication and (ii) Vif-A3 interactions are largely responsible for the increased replication capacity of HV-FIV. Thus, this study suggests that lentiviral accessory proteins may play a critical role in determining the differences in viral replication capacity and pathogenicity between strains.
Blood donor cats were housed in a specific-pathogen-free AAALAC International-accredited animal facility at Colorado State University. All procedures were approved by the CSU Institutional Animal Care and Use Committee prior to initiation.
FIV chimeric clones were produced in which vif, orfA, or vif/orfA from high-virulence FIV (HV-FIV) (FIV-C36 ) replaced these genes in low-virulence FIV (LV-FIV) (FIV-PPR ). Chimeric virus construction was performed using a modified PCR-driven overlap extension technique (82). This technique allowed us to insert HV-FIV vif, orfA, or vif/orfA into the LV-FIV genome with highly specific gene junctions that were not dependent on restriction enzyme site locations. We used plasmids containing full-length LV-FIV and HV-FIV parental constructs as the PCR templates to generate three overlapping fragments: AB (LV), CD (HV), and EF (LV). The primers for these reactions were designed to ensure accurate junctions between parental constructs. All primer sequences are listed in Table S1 in the supplemental material. Primers A and F, specific for pLV-FIV, were used for the generation of all chimeras. Primers B.1 and E.1 were used in the construction of FIV-HVvif, while primers B.2 and E.2 were used for FIV-HVorfA. Based on the 5′-to-3′ arrangement of vif and orfA on the FIV genome, primers B.1, C.1, D.2, and E.2 were used for the construction of FIV-HVvif/orfA. PCRs were conducted with Platinum Pfx DNA high-fidelity polymerase (Life Technologies, Carlsbad, CA) using the following cycling conditions: 30 cycles of 94°C for 15 s, 55°C for 30 s, and 68°C for 1 min. Products were separated on an agarose gel, stained with crystal violet, and extracted using the QIAquick gel extraction and purification kit (Qiagen, Valencia, CA). After the self-primed overlapping reaction between fragments AB, CD, and EF, we amplified the full-length chimeric AF product with primers A and F to generate sufficient amounts for agarose gel extraction and ligation into pLV-HIV. The self-primed and AF amplification cycling conditions were as follows: 30 cycles of 94°C for 15 s, 58°C for 30 s, and 68°C for 2 min.
The chimeric AF fragments were digested with restriction endonucleases EcoRV, Bsu36I, and BclI and ligated into pLV-FIV using T4 DNA ligase according to the manufacturer's specifications (New England BioLabs, Ipswich, MA). All chimeric clones were transformed in MAX Efficiency Stbl2 competent Escherichia coli (Life Technologies) and grown at 30°C in order to avoid instability and recombination of lentiviral sequences. All constructs were validated by restriction enzyme analysis and PCR using primer sets AD, CD, and CF. Clones were directly sequenced by Laragen, Inc. (Los Angeles, CA), and the sequences were analyzed using Sequencher software (Gene Codes Corporation, Ann Arbor, MI) to verify the LV-/HV-FIV junctions and the reestablishment of restriction sites.
Crandell feline kidney cells (CrFK) (83), Mya-1 feline T-lymphoblastoid cells (84), and human 293T cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA). CrFK-CD134 cells, also known as GFox cells, are CrFK cells stably transfected with the FIV primary binding receptor CD134 (14). CrFK and CrFK-CD134 cells were grown at 37°C in 5% CO2 in Dulbecco's modified Eagle's medium (DMEM) with GlutaMAX-1, 1 g/liter d-glucose, and 110 mg/liter sodium pyruvate (Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT), 0.075% sodium bicarbonate, 0.1 mM minimal essential medium (MEM) nonessential amino acids, and 1× penicillin-streptomycin (10,000 U/liter penicillin and 10,000 μg/liter streptomycin; Life Technologies). We purified primary feline peripheral blood mononuclear cells (PBMC) from blood samples obtained from specific-pathogen-free adult cats in a breeding colony at Colorado State University. This procedure was approved by the CSU Institutional Animal Care and Use Committee. PBMC were isolated on a Histopaque-1077 (Sigma-Aldrich, St. Louis, MO) gradient and washed with phosphate-buffered saline (PBS). Primary feline PBMC, as well as Mya-1 cells, were grown at 37°C in 5% CO2 in RPMI 1640 medium with GlutaMAX-1 (Life Technologies) supplemented with the following: 20% FBS, 9 g/liter d-glucose (Sigma-Aldrich), 1× penicillin-streptomycin, 0.075% sodium bicarbonate, 0.1 mM MEM nonessential amino acids, 1 mM sodium pyruvate, 0.055 mM 2-mercaptoethanol (Life Technologies), and 10 ng/ml human recombinant interleukin-2 (Millipore, Temecula, CA). Human 293T cells were grown at 37°C in 5% CO2 in DMEM medium with 4.5 g/liter d-glucose and 110 mg/liter sodium pyruvate, supplemented with the following: 2× GlutaMAX-1 (Life Technologies), 10% FBS, and 1× penicillin-streptomycin.
Plasmids containing full-length FIV genomes were transfected into CrFK cells using Lipofectamine 2000 as described by the manufacturer (Life Technologies, Carlsbad, CA). Briefly, 1 × 106 CrFK cells were plated in 6-cm dishes in antibiotic-free medium and incubated for 24 h. Cells were transfected by adding 6 μg FIV plasmid and 15 μl Lipofectamine 2000 diluted in Opti-MEM I (Life Technologies). At 15 h posttransfection, the medium was removed and replaced with antibiotic-containing medium. Cell supernatants were collected at 48 h posttransfection, centrifuged at 900 × g for 10 min to remove cells and cellular debris, and tested for the presence of FIV p26 capsid by enzyme-linked immunosorbent assay (ELISA) (85). One milliliter of each CrFK supernatant was then used to infect 4.4 × 107 Mya-1 cells in 44 ml medium, and the level of FIV p26 in the supernatant was monitored daily. On day 8, all infected cultures had reached peak p26 levels, so supernatants were collected and centrifuged at 900 × g for 10 min to remove cells, and aliquots were frozen at −80°C. The 50% tissue culture infectious dose (TCID50) of viral stocks was then determined by titration on Mya-1 cells. Viral stocks were diluted in five replicate 10-fold dilution series and added to 2 × 105 Mya-1 cells per well in 96-well plates. FIV reverse transcriptase (RT) activity in day 14 supernatant was determined as previously described (86, 87), and the Spearman-Karber method was used to calculate the TCID50.
FIV replication kinetics were assessed in Mya-1 T cells and cat PBMC. Mya-1 T cells at a density of 1 × 106 cells/ml medium were infected with FIV at a multiplicity of infection (MOI) of 0.001 TCID50 per cell. Freshly isolated PBMC from four pathogen-free domestic cats were pooled and stimulated in culture at 2 × 106 cells/ml with 5 μg/ml concanavalin A (ConA; Sigma-Aldrich, St. Louis, MO). After 3 days of ConA treatment, PBMC were readjusted to 2 × 106 cells/ml in fresh medium without ConA and infected with FIV at an MOI of 0.001. FIV replication was assessed in cell culture supernatant by measuring FIV p26 capsid antigen using a previously described capture ELISA (85) or by determining FIV RT activity as previously described (86, 87). For quantitation of FIV DNA load, cells were pelleted by centrifugation at 580 × g for 5 min, and DNA was extracted using the DNeasy blood and tissue kit (Qiagen, Valencia, CA). The FIV DNA load in cellular DNA was then determined by quantitative PCR (qPCR) as previously described (48) using primer and probe sets specific for FIV clade A and FIV clade C (45).
FIV-infected and uninfected PBMC (5 × 105 cells) were collected by centrifugation at 580 × g for 5 min, resuspended in 150 μl cold flow buffer (phosphate-buffered saline with 5% fetal bovine serum), and blocked with 30 μl goat serum (MP Biomedicals, Solon, OH) for 30 min at 4°C. Cells were washed twice with flow buffer, resuspended in 100 μl flow buffer, and incubated with 0.5 μl CD4-fluorescein isothiocyanate (FITC) antibody (clone 3-4F4; Southern Biotech, Birmingham, AL) and 5 μl CD134-Alexa Fluor 647 antibody (clone 7D6; AbD Serotec, Raleigh, NC) for 30 min at 4°C. Cells were washed three times with flow buffer and resuspended in 170 μl flow buffer containing 0.3 μg/ml propidium iodide (PI; Sigma-Aldrich, St. Louis, MO). Flow cytometry was performed immediately on a CyAn ADP flow cytometer (Beckman Coulter, Brea, CA). Isotype controls produced by incubating PBMC with mouse IgG1 (clone 15H6; Southern Biotech) conjugated to FITC or Alexa Fluor 647 were used to set gates at less than 1% positive cells for each isotype control. The percentages of cells staining positive for PI (dead or dying cells) were compared between viral infections, and PI-positive cells were excluded from the CD134 analyses. The mean fluorescence intensities (MFI) of CD134 on CD4+ cells of FIV-infected cultures were compared to the MFI on uninfected PBMC.
FIV RNA was purified from 140 μl of FIV-infected Mya-1 supernatant using the QIAamp viral RNA minikit (Qiagen, Valencia, CA). Viral RNA was converted to cDNA using random primers and Superscript II RT (Life Technologies, Carlsbad, CA) according to the manufacturer's instructions. The FIV vif/orfA region was amplified by PCR with primers Vif/Orf-F1 and Vif/Orf-R1 (see Table S1 in the supplemental material) using Taq DNA Polymerase (Life Technologies) according to the manufacturer's instructions, using the following cycling conditions: 35 cycles of 94°C for 45 s, 55°C for 30 s, 72°C for 90 s. The integrase portion of FIV pol was amplified with primers Int-F2 and Int-R2 (see Table S1) using identical conditions except that a PCR annealing temperature of 52°C was used. DNA sequencing was performed by the Colorado State University Proteomics Facility. Chromatograms and sequences were analyzed using Sequencher software (Gene Codes Corporation, Ann Arbor, MI).
For specific depletion of feline A3Z3 and A3Z2-Z3 expression, we designed a small interfering RNA (siRNA) (Dharmacon, Lafayette, CO) which targets feline A3Z3 (see Table S1 in the supplemental material). To differentiate between the effects of A3Z3 knockdown and nonspecific effects of RNA transfection, we also utilized siGENOME control nontargeting siRNA number 3 from Dharmacon as a negative control. A3Z3 knockdown for virus production was conducted in CrFK cells by plating 1 × 106 cells in 6-cm dishes. After 6 h of incubation, cells were transfected with siRNA at a 50 nM concentration using 2 μl/ml Lipofectamine RNAiMAX (Life Technologies, Carlsbad, CA) according to the manufacturer's instructions. At 24 h after siRNA transfection, the medium was removed, cells were washed with PBS, and each FIV plasmid (2 μg) was transfected using 15 μl Lipofectamine 2000 as described by the manufacturer (Life Technologies, Carlsbad, CA). At 15 h after FIV plasmid transfection, the medium was removed, cells were washed with PBS, and fresh medium was added. Cell supernatants were collected at 48 h after FIV plasmid transfection, centrifuged at 900 × g for 10 min to remove cells and cellular debris, aliquoted, and frozen at −80°C. Additional virus stocks were produced in control siRNA-transfected cells, which express A3Z3 normally, by identical procedures.
To accurately compare the levels of FIV p26 capsid between virus stocks, we performed FIV p26 ELISA on triplicate 2-fold dilution series of each virus stock. Linear plots of FIV p26 (at an optical density of 450 nm [OD450]) versus dilutions of virus stock were used to determine the amount of each virus stock equivalent to 0.2 OD450, and subsequent infections were initiated with this amount of each virus, containing equal FIV p26. To compare FIV replication in the presence or absence of A3Z3, 1.8 × 105 CrFK-CD134 cells in 12-well plates were transfected with control siRNA or anti-A3Z3 siRNA as described above and, 24 h later, infected with FIV stocks produced in A3Z3-expressing cells or A3Z3-depleted cells, respectively. FIV replication was quantified by measuring FIV p26 levels in cell culture supernatants on days 2, 4, and 6 following infection.
RNA was extracted from cells using TRIzol reagent (Life Technologies, Carlsbad, CA) according to the manufacturer's instructions. Any remaining DNA was then eliminated by treatment with DNase I, amplification grade (Life Technologies), using 2 U in a 50-μl volume. RNA preparations were further purified using the RNeasy minikit (Qiagen, Valencia, CA) RNA cleanup protocol and quantified using an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE). Cellular RNA was converted to cDNA according to the manufacturer's instructions for Superscript II (Life Technologies), using random hexamer primers.
Feline APOBEC3 (A3) cDNA was quantified by qPCR using primers specific for A3Z2, A3Z3, and A3Z2-Z3 (see Table S1 in the supplemental material). The A3Z2 primers detect the expression of all three feline A3Z2 isoforms (A3Z2a, A3Z2b, and A3Z2c) collectively. The A3Z2-Z3 primers detect the expression of all A3Z2-Z3 variants (A3Z2b-Z3, A3Z2c-Z3, and splice variants). The reaction mixtures were prepared with SsoFast EvaGreen supermix (Bio-Rad, Hercules, CA) and 400 nM primers and were run on a CFX96 real-time PCR detection system (Bio-Rad) using the following conditions: 95°C for 30 s, 40 cycles of 95°C for 5 s, and 60°C for 10 s, followed by a 65-to-95°C melt curve analysis. An A3Z2b-Z3 plasmid standard curve was used to determine copy number for all three A3 qPCRs. Since the A3Z2 and A3Z3 primers also detect A3Z2-Z3 expression, we subtracted the A3Z2-Z3 copy number from the A3Z2 and A3Z3 copy numbers for each sample. The validity of this subtractive method is supported by the use of a single A3Z2b-Z3 plasmid standard curve for all A3 qPCRs and qPCR efficiencies that were consistently in the 95-to-100% range for these assays. A3 mRNA copy numbers were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression using a previously described qPCR assay for feline GAPDH (88) with a feline-GAPDH-encoding plasmid standard curve. All primer pairs span exon junctions, and melt curves run on all qPCRs showed single-amplification products, demonstrating target specificity for all assays.
For all comparisons of FIV replication over time, we determined whether the level of FIV differed significantly between the virus strains over time by performing repeated-measures analysis of variance (ANOVA) on each pairwise combination of viruses and examining the time by virus interaction. CD134 expression over time was similarly assessed with repeated-measures ANOVA on each pairwise combination of viruses. The replication of FIV in the siRNA knockdown experiments was compared by repeated-measures ANOVA with Bonferroni posttests. For all experiments, a P value of <0.05 was considered the threshold for significance.
Chimeric FIV constructs were produced in which vif, orfA, or both genes from high-virulence FIV (HV-FIV) (FIV-C36) replaced these genes in the low-virulence FIV (LV-FIV) (FIV-PPR) viral genome, generating FIV-HVvif, FIV-HVorfA, and FIV-HVvif/orfA (Fig. 1; for details of chimera junctions, see Fig. S1 in the supplemental material). Sequencing of each clone verified each chimeric construct (see Fig. S1) with intact insertion of HV-FIV vif and vif/orfA into the LV-FIV genome. Two conservative amino acid changes were identified near the carboxy terminus of integrase (IN). Infectious FIV was recovered for comparative studies by transfecting CrFK cells with FIV-HVvif, FIV-HVorfA, and FIV-HVvif/orfA chimeric plasmids, as well as HV-FIV and LV-FIV parental plasmids. Cell supernatants removed at 48 h posttransfection all had levels of FIV p26 capsid that were detectable by ELISA (data not shown); each supernatant was then used to generate amplified virus stocks by growth in Mya-1 T cells. Supernatants were harvested at day 8 when maximum p26 levels were detected. We determined the 50% tissue culture infectious dose (TCID50) by infecting Mya-1 cells with serial dilutions of viral stocks and measuring RT activity at 14 days postinfection.
We compared the replication of vif and orfA chimeric viruses to that of parental strains by infecting Mya-1 T cells with equal TCID50s of each virus and monitoring FIV p26 capsid in cell culture supernatants by ELISA for 15 days (Fig. 2). As expected, HV-FIV replicated with dramatically increased kinetics compared to the replication kinetics of LV-FIV. FIV-HVorfA demonstrated replication kinetics intermediate to those of parental strains, indicating that HV-FIV orfA provides a significant replication advantage over LV-FIV orfA. Both HV-FIV vif-containing chimeric viruses (FIV-HVvif and FIV-HVvif/orfA) had strongly increased replication kinetics compared to the replication kinetics of LV-FIV. In fact, HV-FIV vif chimeric virus replication approached that of the virulent FIV parental strain. This result indicates that both OrfA and Vif contribute to the inherent differences noted for the in vitro replication capacities of HV-FIV and LV-FIV.
Multicycle replication of chimeric viruses has the potential to result in the selection of adaptive mutations within or surrounding the chimeric region. We have previously observed this phenomenon in the integrase-encoding portion of pol in the FIV-PCenv chimera (48). To determine whether any mutations were selected for in our chimeric virus production and subsequent replication in Mya-1 cells, we sequenced viral RNA from the day 15 supernatants of Mya-1 cell infections, whose replication kinetics are shown in Figure 2. We evaluated vif, orfA, and flanking regions, including the entire integrase portion of pol, for all five viruses and found that the consensus sequences were identical to the original plasmid clone sequences (data not shown). While lentiviral infections inherently result in the sporadic occurrence of mutations, these data indicate that virus production during 15 days of replication in Mya-1 T cells was consistent with input virus and that outgrowth was not a result of in vitro selection.
In order to confirm these results using viral detection assays other than p26 ELISA, we repeated the Mya-1 T-cell infections and measured RT activity in the supernatants and FIV proviral loads by qPCR. FIV growth kinetics as detected by RT activity (Fig. 3A) were similar to those detected by FIV p26 ELISA (Fig. 2). RT activity recapitulated earlier observations; namely, FIV-HVorfA replicated with increased kinetics compared to the replication of parental LV-FIV, while the HV-FIV vif-containing viruses (FIV-HVvif and FIV-HVvif/orfA) had replication kinetics that approached the replication of parental HV-FIV (Fig. 3A). The proviral DNA loads also indicated that all three chimeric viruses (FIV-HVvif, FIV-HVvif/orfA, and FIV-HVorfA) had greater FIV DNA loads than LV-FIV (Fig. 3B). Thus, three different methodologies for assessing viral replication confirmed that HV-FIV vif- and orfA-expressing chimeric viruses had increased replication capacities compared to the replication of the parental LV-FIV.
As a more-relevant biological model, we next sought to determine whether HV-FIV vif and orfA could also confer a replication advantage to LV-FIV in cat PBMC ex vivo. We infected pathogen-free cat PBMC with equal TCID50s of each of the five viruses and monitored FIV p26 capsid in cell culture supernatants by ELISA for 15 days (Fig. 4A). The differences in the magnitude and kinetics of viral growth were similar to the observations in Mya-1 T cells. FIV-HVorfA had increased replication kinetics compared to the replication of LV-FIV; HV-FIV vif-containing chimeric viruses (FIV-HVvif and FIV-HVvif/orfA) had significantly increased growth kinetics compared to the growth of LV-FIV and FIV-HVorfA (P < 0.05). Furthermore, HV-FIV vif chimeric viruses actually trended toward increased replication relative to that of parental HV-FIV. Thus, FIV growth in feline PBMC supports the observations in Mya-1 T cells—HV-FIV orfA confers a moderate, yet significant replication advantage to LV-FIV, while HV-FIV vif confers a strong replication advantage in primary cells isolated from outbred animals.
FIV infection of cells results in downregulation of the FIV primary binding receptor CD134 (80, 89), which is analogous to the ability of HIV to downregulate the CD4 receptor (90). Recently, FIV OrfA was found to be sufficient to mediate CD134 downregulation (80). We measured CD134 expression on CD4+ T cells in infected PBMC cultures to investigate whether variations in orfA genes affected the extent of CD134 downregulation. The expression of CD134 in FIV-infected cultures relative to its expression in uninfected PBMC is shown in Figure 4B. Comparison of CD134 downregulation (Fig. 4B) and FIV replication kinetics (Fig. 4A) revealed an inverse relationship; cultures infected with the more-rapidly replicating FIV-HVvif and FIV-HVvif/orfA had decreased CD134 expression compared with the CD134 expression in cultures infected with more-slowly replicating viruses, such as LV-FIV. Propidium iodide staining did not reveal significant differences in viral cytopathicity at any time point (data not shown). The degree of CD134 downregulation is apparently independent of the identity of the orfA gene, i.e., LV-FIV orfA-containing viruses had both the lowest (FIV-HVvif) and the highest (LV-FIV) CD134 expression. To our knowledge, this is the first study to demonstrate that FIV can induce downregulation of CD134 in cat CD4+ T cells ex vivo.
Thus, CD134 downregulation correlates more strongly with the quantity or rapidity of productive virus than with the strain identity of the orfA gene. As noted from the results shown in Figures 2, ,3,3, and and4A,4A, orfA may influence replication capacity and, thus, result in differences in the degree of CD134 cell surface expression on PBMC.
The primary function of lentiviral Vif is to bind and induce the degradation of cognate host APOBEC3 (A3) proteins, thereby protecting the virus from the antiviral effects of these proteins (91, 92). Since Vif-based differences in replication may involve interactions with A3 proteins, we quantified the levels of A3 mRNA expression in the feline cell types used in this study to determine whether these cells express A3 mRNA and whether cell type-specific differences in expression exist. We determined the A3 mRNA expression of CrFK cells, Mya-1 cells, and cat PBMC by quantitative real-time PCR and found that all three cell types expressed A3Z2, A3Z3, and A3Z2-Z3 mRNAs (Fig. 5). Total A3Z2 (including all three A3Z2 homologs) mRNA expression was highest, followed by A3Z3 mRNA expression. The expression of A3Z2-Z3 double-domain mRNAs was very low relative to the expression of single-domain A3Z2 and A3Z3 mRNAs. For instance, PBMC had >300-fold higher levels of A3Z3 than of A3Z2-Z3 mRNA. Since feline A3Z3 and several A3Z2-Z3 variant proteins can inhibit FIV (in the absence of Vif), while the three A3Z2 homologs do not inhibit FIV (68, 71), the low abundance of A3Z2-Z3 mRNA compared to the abundance of A3Z3 suggests that A3Z3 is the major antilentiviral A3 protein present in domestic cats. Notably, the expression level of A3Z3 in cat PBMC and Mya-1 cells is very similar to the levels of APOBEC3G and APOBEC3F reported for macaque PBMC and CD4+ T cells (93), suggesting that important antilentiviral A3s in other species are expressed at comparable levels. Comparing the three feline cell types examined, CrFK cells had lower A3Z3 mRNA levels than PBMC and Mya-1, while Mya-1 cells had lower A3Z2-Z3 mRNA levels than PBMC and CrFK cells. However, all three cell types expressed appreciable levels of A3 mRNAs and, therefore, have the potential to exhibit A3-mediated viral restriction.
We hypothesized that the difference in the replication capacity of FIV-HVvif relative to that of the parental LV-FIV would be dependent on the action of feline APOBEC3 (A3) proteins. To assess this hypothesis, we sought to compare the replication of HV-FIV, LV-FIV, and FIV-HVvif in cells depleted of feline A3 proteins. Since all of the anti-FIV feline A3 proteins contain a Z3 domain, we designed an siRNA targeting A3Z3 to allow us to specifically deplete anti-FIV A3 from cells. Transfection of CrFK cells with anti-A3Z3 siRNA reduced A3Z3 and A3Z2-Z3 mRNA expression by >85% for 7 days, while the expression of A3Z2 was unaffected (Fig. 6).
To compare FIV replication under conditions of transient siRNA transfection, we designed a short-course infection experiment utilizing highly transfectable CrFK cells that have been engineered to express the FIV receptor, CD134, allowing multiple rounds of viral replication and production of progeny virions (GFox cells ). Because A3 exerts antiviral effects via incorporation into a lentiviral virion from a virus producer cell, thereby resulting in subsequent viral inhibition upon infection of a naive target cell, we depleted A3 from both producer and target cells using siRNA to assess FIV replication in the near-complete absence of A3 as follows: (i) two parental strains of FIV and FIV-HVvif were produced in A3-depleted CrFK, and (ii) viral inocula were normalized by gag equivalents and used in equal amounts (iii) to infect CrFK-CD134 cells depleted of A3 following siRNA transfection. Concurrently, to assess FIV replication under conditions of normal A3 expression in both producer and target cells, CrFK-CD134 cells expressing A3 were infected with FIV produced in CrFK cells expressing A3. Since LV-FIV and FIV-HVvif only differ in vif, any difference in replication between these viruses would be attributable to the action of HV-FIV Vif versus LV-FIV Vif. In CrFK-CD134 cells expressing normal levels of A3Z3 and A3Z2-Z3, we found that FIV-HVvif was detected in significantly higher levels in the supernatant than LV-FIV (Fig. 7A), consistent with the results of replication experiments in Mya-1 T cells and domestic cat PBMC (Fig. 2 and and4A).4A). However, when A3-depleted virus was used to infect A3-depleted cells, the difference in replication capacity was mitigated to insignificant levels (Fig. 7B). This implies that the replication advantage of HV-FIV Vif is exerted through enhanced anti-A3Z3 and anti-A3Z2-Z3 effects, or, conversely, that the relative growth disadvantage mapped to LV-FIV Vif can be attributed to a relative inability to counteract the anti-FIV effects of these cellular restriction enzymes—i.e., the difference in replication between HV-FIV vif- and LV-FIV vif-expressing viruses is dependent on feline A3Z3 and/or A3Z2-Z3.
Studies of FIV, SIV, and HIV have demonstrated that the pathogenicity of the infecting lentiviral strain plays an important role in determining the viral load and the severity of host immunodeficiency (22, 24–26, 37, 43, 45). However, the specific viral characteristics associated with pathogenicity remain relatively poorly understood. FIV infection of the domestic cat offers a model system for research on lentiviral pathogenesis in an authentic host species with a disease pathology highly similar to that of HIV/AIDS, since one of the best-characterized features of the FIV system is the exhaustive documentation of strain-specific pathologies associated with genetically divergent FIV strains (5, 17, 43–48). Therefore, studies on the viral genetic basis for FIV pathogenicity can inform our understanding of other lentiviruses, such as HIV, by examining important genetic elements which lentiviruses share, such as Vif. Likewise, elements that are unique to FIV, such as OrfA, can be informative at a comparative level in understanding lentiviral strategies for replication and evasion of host defenses.
In this study, we examined the basis for FIV pathogenicity using vif and orfA accessory gene chimeras between two FIV molecular clones with reproducibly and strikingly contrasting disease potentials: HV-FIV (FIV-C36), which is highly virulent in vivo and replicates rapidly in vitro; and LV-FIV (FIV-PPR), which is less virulent in vivo and replicates more slowly and/or to lower titers in vitro. We used overlapping PCR (82) to clone vif and orfA accessory gene chimeras. The generation of such chimeras is feasible in FIV but not in human or primate lentiviruses, which contain overlapping transcription reading frames in accessory and structural genes (Fig. 1) (94).
We generated 3 chimeric FIVs in which vif, orfA, or vif/orfA from HV-FIV was substituted for these genes on the LV-FIV backbone, producing FIV-HVvif, FIV-HVorfA, and FIV-HVvif/orfA, respectively. In all cell systems evaluated and by all measures of viral replication, FIV-HVorfA displayed moderately increased replication kinetics compared to the replication of the isogenic parental LV-FIV, and chimeric viruses expressing HV-FIV vif or vif/orfA had strongly increased replication kinetics compared to the replication of LV-FIV, nearly equaling that of the virulent parental FIV. This finding suggests that these nonstructural elements may play a significant role in the capacity for replication and, thus, may be a mechanism for enhancing virulence.
In Mya-1 cells, the degree of viral replication for chimeric viruses (versus the replication of the parental LV-FIV, as measured by p26 ELISA [Fig. 2] or RT activity [Fig. 3A]) appeared to be relatively more enhanced than the viral DNA load (as measured by qPCR [Fig. 3B]). While the assays used cannot be directly compared and the scales of comparison are different (linear for viral production and log for proviral integration), this observation suggests that high-virulence Vif and OrfA may have a greater enhancement of postintegration events (viral transcription, translation, assembly, and release) than of preintegration events. This observation is consistent with a number of possible mechanisms, including an enhanced ability to counteract deamination-dependent inhibition by feline A3.
We further demonstrated that the enhancement of FIV-HVvif replication is apparent only in the presence of feline A3 proteins, suggesting that naturally occurring differences in lentiviral Vif contribute to viral replication capacity and virulence via interaction with host A3. The interaction between lentiviral Vif and host A3 is a critical factor determining the species specificity of lentiviruses. For instance, the inability of HIV-1 to infect species other than humans and chimpanzees is based largely on the species-specific adaptation of Vif to human A3 proteins (72, 95–97). Likewise, we have found that challenge of domestic cats with the puma strain of FIV (FIV-Pco) results in an initial productive infection that diminishes to undetectable levels over time (98). The reductions in virus expression coincide with G-to-A hypermutation in the provirus, consistent with the interpretation that FIV-Pco Vif is not well adapted to target domestic cat A3 (99).
While pathogenicity is likely related to numerous complex viral traits, the ability to counteract A3 intracellular restriction via a more “robust” Vif may be an important virulence determinant. Vif from one particular lentiviral species typically protects against A3 proteins from its native host; i.e., FIV Vif mitigates feline A3 activity, HIV Vif mitigates human A3 activity, etc. Vif antagonism of a matched-host A3 protein is not necessarily complete, however. A3 proteins can cause sublethal levels of G-to-A mutation even in the presence of a functional Vif (100, 101), suggesting that sufficient A3 evades Vif-mediated degradation to directly affect virus evolution/divergence. Additionally, when transfected at high levels, A3 proteins are able to overwhelm matched-host Vif, resulting in a virus restriction phenotype compared to the phenotype seen with normal A3 expression (55, 59, 61, 62). In a clinical setting, polymorphisms in human A3 genes have been associated with the risk of HIV acquisition (102–104) and with HIV disease progression (102, 105). Likewise, several reports have found that higher levels of A3 expression are associated with lower HIV loads (106, 107) and resistance to infection (108, 109). These studies suggest that there is a dynamic balance between A3-mediated restriction and Vif-mediated protection that logically extends to a role for lentiviral Vif with significant impact on viral replication and pathogenicity. Indeed, it has been shown that the ability of HIV Vif to neutralize A3 proteins in vitro can differ between HIV subtypes (110, 111) or between strains of the same subtype (112).
While the studies cited above utilized in vitro single-replication-cycle assays to support these observations, differences in Vif activity between naturally occurring lentiviral strains have not been explored in depth, especially in the context of multicycle whole-virus replication. Here, we examined fully replication-competent vif gene chimeras constructed between two FIV strains with well-characterized and divergent pathological consequences that directly correlate with in vitro growth characteristics (47, 48). These results suggest a model in which lentivirus-host adaptation is a continuum that is highly dependent upon Vif and that naturally occurring differences in lentiviral Vif can control the virus replication that is associated with pathogenic potential. Explanations for these findings might include strain-specific differences in (i) vif transcription levels, (ii) Vif-induced degradation of A3, (iii) Vif protein stability, or (iv) sequestration of A3 proteins which does not result in degradation.
The finding that HV-FIV Vif confers greater replication capacity than LV-FIV Vif in a common viral backbone additionally raises the question of what regions of Vif are essential for conferring this phenotype. Comparison of the Vif amino acid sequences of these two viruses shows 84% identity, with 40 amino acid differences, 10 of which are considered amino acids with highly dissimilar properties (see Fig. S2 in the supplemental material). Amino acid changes occur in a putative Cullin5 zinc-coordinating motif and immediately adjacent to a putative BC box (proposed by Stern et al. ), both considered highly relevant functional domains. Alternatively, sequence differences which affect the Vif expression level or stability might be relevant.
Multiple potential functions have been ascribed to the small, 77-amino-acid FIV OrfA protein. Originally characterized as a transactivator of viral protein expression (77), similar to HIV Tat, the mechanism of action is distinct from those of other lentiviral transactivators and the net upregulation of transcription is relatively weak (73, 113). OrfA has also been shown to affect virion formation and infectivity (73), as well as to localize to the nucleus and induce G2 cell cycle arrest (78), similar to HIV Vpr. Most recently, it has been demonstrated that OrfA downregulates the surface expression but not the transcription or translation of the primary FIV receptor CD134 and that OrfA-negative FIVs are unable to productively infect CD134-expressing cells (80). OrfA inhibition of CD134 may facilitate more-efficient virus release due to decreased receptor interactions with progeny virus, similar to the effects attributed to HIV Vpu and Nef on CD4 (90). However, another recent study suggests that this effect may be related to cell type and, potentially, other virus-specific factors (89). Interestingly, the orfA gene is the most genetically heterogeneous portion of the FIV genome when compared across all known FIV isolates. HV-FIV and LV-FIV OrfA proteins share only 65% amino acid identity, with 27 total differences, of which 8 are highly dissimilar amino acids (see Fig. S2 in the supplemental material). This high level of diversity provides a potential basis for functional differences between strains.
We found that replacing the LV-FIV orfA with the HV-FIV orfA resulted in a moderate increase in viral replication capacity. The double accessory gene chimera FIV-HVvif/orfA displayed replication kinetics similar to those of FIV-HVvif (Fig. 2, ,3,3, and and4),4), suggesting that the gain of function resulting from the high-virulence orfA is weaker than that conferred by the high-virulence vif. FIV OrfA-mediated downregulation of the FIV primary binding receptor CD134 is associated with the ability of the virus to replicate in CD134-expressing cells (80). Thus, a possible mechanism to explain the replication differences between strains could be that increased CD134 downregulation might facilitate efficient virus release and promote replication. We observed that CD134 surface expression appeared to correlate inversely with the rate of virus replication; i.e., lower expression of CD134 was associated with more-rapid viral replication (Fig. 4). However, ascribing effects on cellular protein expression to different viruses in a low-MOI, multiple-replication-cycle experiment is likely to be confounded by viruses replicating and spreading to new cells at different rates due to potentially unrelated elements (such as Vif).
In this study, we chose to focus on a comparison of vif and orfA chimeric and parental strain replication in T cells and PBMC, since these are major targets of FIV infection and replication in these cell types has been shown to correlate with replication in vivo (47, 114). However, macrophages also play an important role in establishing productive lentiviral infection, and both vif and orfA are required for productive FIV infection of macrophages (49, 74, 76). Therefore, future work aimed at understanding the mechanism of the observed differences in vif and orfA chimeric virus replication may benefit from comparison of virus replication in feline macrophages.
The results of this study identify an important mechanism for lentiviral virulence in a model relevant for HIV/AIDS. While it is well established that A3 plays a key role in restricting lentiviral cross-species infection, these findings support the concept that host cellular A3 can also affect the replication of host-adapted lentiviral strains, resulting in a continuum of consequences from aborted infection to significant virulence. The dynamic balance between A3-driven host restriction and Vif-driven viral escape from restriction is ongoing and may influence the ability of lentiviruses to cause progressive disease. This system will provide an opportunity to study specific Vif-A3 interactions that are relevant in vivo and determine the mechanism of increased Vif-dependent lentiviral replication capacity, suggesting rational and novel modalities for additional lentiviral therapeutics. The chimeric system outlined here will also provide a useful basis for more clearly delineating the function of OrfA with respect to FIV replication and virulence and for potentially identifying additional host restriction factors relevant to limiting lentiviral pathogenicity.
We thank Scott Carver for statistical assistance, Xin Zheng for technical assistance, and Wendy Sprague for flow cytometry advice. We thank Carsten Münk and Martin Löchelt for providing pcDNA3.1-A3Z2b-Z3-HA.
This work was supported by grants R01 AI48411 (J.E. and S.V.) and R01 AI25825 (J.E.) from the National Institute of Allergy and Infectious Diseases of the National Institutes of Health.
Published ahead of print 8 May 2013
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JVI.00752-13.