Several studies demonstrated that HIV-1 replication cycle is restricted at an early post-entry step in primary human quiescent CD4+ T cells in vitro (reviewed in 12). To better understand the restriction block observed in resting G0 cells in vitro, human primary quiescent CD4+ T cells were isolated from PBMCs by a two-step process. First, unwanted cell populations were labeled with biotin-conjugated antibodies (ab) to CD8, CD16, CD19, CD36, CD56, CD123, TCRγδ and glycophorin A, and removed with anti-biotin magnetic beads on an AutoMacs cell separator. Next, recovered cells were stained with anti-CD8-FITC (clone SK1, BD Bioscences), anti-CD25-PE (clone 4E3, Miltenyi Biotec), anti-CD14 (Clone TUK4, Miltenyi Biotec) and anti-HLA-DR (L243, BD Bioscences) ab and sorted on a FACSVantage cell sorter. Typically, 98% of the cells expressed CD4 and 99% were negative for activation markers (data not shown). Next, purified quiescent CD4+ T cells were infected with the NL4.3 strain of HIV-1 at a multiplicity of infection (moi) of 1 and the subcellular localization of incoming sub-viral complexes was studied by immunofluorescence and confocal microscopy. Infected and control cells were co-stained with antibodies against HIV-1 capsid (CA) protein and against γ-tubulin, a cellular marker for the centrosome 10. We observed that, at day 2 and day 9 post-infection, CA antigens co-localized with γ-tubulin in 58 to 75% of CA-positive cells, respectively (Fig 1A). These observations demonstrate that, in the absence of viral replication, incoming HIV-1 sub-viral complexes concentrate at the centrosome of quiescent T lymphocytes in vitro. Note that the quiescent phenotype of target CD4+ T cells did not significantly change upon infection, as determined by monitoring the surface expression of T cell activation markers (CD25 and HLA-DR) of infected and control cells by flow cytometry (Fig 1B).

To rule out the possibility that the pericentrosomal distribution of incoming CA at later time points was the result of a spreading infection which might occur in few cells, single-round viral vectors pseudotyped with the glycoprotein G of vesicular stomatitis virus (VSVg) were used for further studies. These vectors maintain the biological properties that govern early events in the replication cycle of their parental counterpart, but are unable to achieve late stages of the viral replication. Additionally, although VSVg-pseudotyped viral particles enter by fusion out of acidified endosomes, instead of receptor-mediated fusion at the plasma membrane, the post-fusion events are analogous to that of wild-type HIV-1. Therefore, human primary quiescent CD4+ T cells were transduced with a VSVg-pseudotyped HIV-1-based lentivector carrying the GFP transgene and the localization of incoming sub-viral complexes was analyzed. As in the case of the wild-type virus, incoming HIV-1 CA proteins from lentivectors were localized in the pericentriolar area from day 2 to day 9 post-transduction (Fig. 1C and data not shown) in 60 to 82% of CA-positive cells, respectively. These results indicate that the route of entry and the viral accessory proteins are not implicated in early HIV-1 intracellular trafficking. As expected, transduced quiescent cells did not support GFP expression and their activation status was not significantly altered when compared to that of control cells (data not shown). Altogether, these results indicate that in quiescent CD4+ T cells, incoming HIV-1 sub-viral complexes concentrate in close proximity to the centrosome.

We then asked whether the pericentrosomal localization of incoming HIV-1 was observed also in other resting cell systems. To this aim, cycling or resting human primary fibroblast MRC5 cells were transduced with a VSVg-pseudotyped HIV-1-based lentivector carrying the GFP transgene. Analysis of GFP expression at 48, 72 and 96 h post-transduction by flow cytometry showed that only cycling, but not resting, MRC5 cells supported HIV-1 viral gene expression (Fig 2A).

We next analyzed the subcellular distribution of incoming sub-viral complexes in resting MRC5 cells. Immunostaining of transduced resting MRC5 revealed that incoming HIV-1 CA targeted the centrosome as early as 4 hours post-transduction and persisted at this site up to 28 days post-transduction (Fig 2B). By staining these cells with an antibody against HIV-1 matrix (MA) protein, we visualized dots or patches on the cell surface, which disappeared within 24 hours (data not shown). Persistence of HIV-1 CA and loss of MA antigens in quiescent MRC5 cells were confirmed by Western blotting on total cell lysates. As shown in figure 2C, HIV-1 CA was still detectable at day 28 post-transduction, while MA was not detected in the extracts from transduced cells as soon as 24 h following transduction, confirming our immunofluorescence studies (Fig 2C). Indeed, upon entry, most of MA, which directly binds to the viral envelope, remains associated with the inner surface of the cellular membrane and is subsequently degraded 11. Partial disassembly and/or degradation of incoming HIV-1 cores in quiescent cells might account for the reduction of CA signal intensity over time (Fig 2C). Consistently, we never visualized structured and assembled incoming HIV-1 cores in quiescent cells by electron microscopy (data not shown). Once the inside the cytoplasm, a structural reorganization and/or partial disassembly of the capsid shell might occur, regardless of the activation status of the target cell (reviewed in 12). These observations demonstrate that incoming HIV-1 virions undergo a certain degree of uncoating soon after entry into quiescent cells.

Since HIV-1 CA has been found to be still associated with entering virions at the onset of reverse transcription 13, we wished to establish whether centrosomal-associated sub-viral complexes detected at the centrosome might represent reverse transcription complexes (RTCs). For that purpose, we investigated the localization of the reverse-transcribed viral DNA in transduced resting cells using fluorescent in situ hybridization (FISH). HIV-1 reverse transcription has been reported to be completed within 3 days in quiescent cells in vitro 89. Thus, resting MRC5 cells were transduced with the VSVg-pseudotyped NL4.3 virus and FISH was performed 4 days later, using the full-length proviral genome as a probe. Remarkably, we found that the reverse-transcribed viral genome localized at the centrosome in resting cells (Fig. 3A) and that the frequency of co-localization vDNA/γ-tubulin was similar to that of CA/γ-tubulin. Since both incoming CA antigens and the viral DNA genome reside at the MTOC of resting primary cells, we concluded that they likely represent RTCs.

To assess whether sub-viral complexes concentrated at the centrosome constitute stable pre-integration intermediates, which might be subsequently reactivated for productive infection, quiescent MRC5 cells were first transduced with a VSVg-pseudotyped HIV-1 vector and later stimulated to divide by splitting and serum addition. At different time points post-transduction, contaminant cycling cells supporting direct GFP expression were eliminated by cell sorting and the purity of the resulting cell population was typically 98% (Fig 3B). The percentage of cells expressing GFP was then monitored by flow cytometry 48, 72 and 96h following reactivation. As shown in figure 3B, GFP expression could be detected following reactivation of transduced cells up to day 21 post-transduction, demonstrating that part of viral DNA present at the MTOC reaches the nucleus to integrate into host chromosomes. These results demonstrated that the sub-viral complexes, which persist at the centrosome, in cells maintained quiescent for an extended period of time, are stable, functional and inducible upon cell stimulation.