We analyzed the influence of Vpu on HIV cell-to-cell transmission. We cocultured for 2 h WT- or ΔVpu-infected cells with primary CD4+ lymphocytes, and then harvested the target lymphocytes. We then followed productive viral spread to lymphocytes by measuring by flow-cytometry the appearance of Gag+ cells, as outlined . We first used as donors HeLa cells, that constitutively express tetherin, or Hela-THN- cells, in which the protein was silenced (Figure S1a
). Productive entry of viruses in HeLa or HeLa-THN- cells (which lack the CD4 receptor) was ensured by pseudotyping WT or ΔVpu virions with the VSV-G envelope. HeLa cells, with similar levels of infection (15–20% of Gag-expressing cells, as assessed by flow cytometry), were then cocultivated with CD4+ lymphocytes. WT HIV was efficiently transmitted to targets, with about 20% of lymphocytes expressing Gag after 18 h (). Nevirapine, a reverse-transcriptase inhibitor, significantly decreased the appearance of Gag in targets, confirming that the signal mostly originates from newly synthesized viral proteins, and not from capture of incoming virions (not shown). Transmission of WT HIV was slightly affected by tetherin. This confirmed that tetherin inhibition by Vpu is not absolute 
and likely depends on the relative levels of the two proteins. ΔVpu was transmitted from HeLa-THN- cells, but much less potently from HeLa cells (). A compilation of independent experiments, with lymphocytes from different donors, indicated that tetherin significantly decreased Gag expression in ΔVpu recipient cells (two-fold reduction) (). Similar results were obtained when Jurkat lymphoid cells were used as targets (). Thus, tetherin decreases HIV cell-to-cell transmission. Vpu counteracts this phenomenon. Noteworthy, the inhibitory effect of tetherin on ΔVpu was counteracted by transient transfection of a Vpu expression plasmid in donor HeLa cells, excluding the possibility that the expression of Vpu in targets may have biased the results (not shown).
Tetherin reduces HIV cell-to-cell transmission.
We then used 293T cells as donors, since they do not naturally express tetherin. We examined if transient expression of tetherin inhibited viral cell-to-cell spread. To this end, 293T cells were cotransfected with WT or ΔVpu HIV proviruses, along with a control or a tetherin expression plasmid. An amount of 100 ng of tetherin plasmid was selected, since it potently inhibited release of ΔVpu, without affecting that of WT HIV (Figure S1b
). Upon coculture of transfected cells with Jurkat cells, ΔVpu transmission to target lymphocytes was decreased by tetherin, whereas WT HIV was minimally impaired (). A compilation of 6 independent experiments confirmed a significant reduction (two-fold) of ΔVpu transmission from tetherin positive cells, when compared to negative cells (). We next evaluated the contribution of cell-free viral particles to the productive infection of target cells. We previously reported that a gentle agitation of cocultures inhibits HIV spread through direct cell contacts without impairing infection by free virions 
. Shaking cocultures of 293T donor cells and Jurkat target cells strongly inhibited the appearance of Gag+ cells in targets, irrespectively of the presence of tetherin or Vpu in donors (). Therefore, under these experimental conditions, most of productive viral transmission occurs through intercellular contacts. The contribution of cell-free virions is minimal.
Tetherin reduces HIV cell-to-cell transmission from 293T and primary T cells.
To describe further the impact of tetherin on HIV cell-to-cell spread, we transfected different amounts of tetherin expression plasmids (Figure S1b
). At low amounts (20 ng of transfected DNA), ΔVpu release in supernatants was inhibited, without obvious effects on cell-to-cell transmission. With high levels of tetherin (200 ng), ΔVpu release and transmission were both restricted. This was also the case for WT HIV. Therefore, as previously reported for viral release 
, the effect of tetherin on cell-to-cell spread is dose-dependent. The anti-tetherin activity of Vpu is not absolute, and tetherin inhibits more easily viral release than cell-to-cell transmission.
We then asked whether tetherin, when induced by type-I interferon (IFN), restricts intercellular viral spread. We generated 293T cells that carry an shRNA against tetherin (293T-shTHN) or a control shRNA (293T-shCtr). Flow-cytometry indicated that IFN induced tetherin in 293T control cells, but not in 293T-shTHN cells (). Upon IFN treatment, cell-to-cell spread of ΔVpu was significantly impaired in control cells (two-fold decrease), and occurred normally in tetherin-silenced 293T cells (). Therefore tetherin is the major interferon-induced protein impairing HIV cell-to-cell transfer in this experimental setting.
It was important to determine whether tetherin also inhibits lymphocyte-to-lymphocyte viral transfer. Primary lymphocytes, naturally expressing tetherin (not shown), were infected with WT or ΔVpu HIV, and were then cocultivated with uninfected target lymphocytes, where the appearance of Gag+ cells was measured over time. With WT HIV, about 30% of Gag+ targets were detected at 48 h. This signal was only partly inhibited in the presence of Nevirapine, suggesting that it corresponds to a mix of newly synthesized Gag proteins and incoming viral materials bound to targets (). ΔVpu was significantly less transmitted to recipient lymphocytes. T. Interestingly, in the presence of Nevirapine, the % of Gag+ target cells was not significantly different for WT and ΔVpu (), suggesting that uptake of incoming viral materials by targets is not inhibited by tetherin. This dissociation between viral uptake and subsequent infection of target cells is studied further below. A decrease of ΔVpu productive transmission was observed when various T cell lines (MT4C5, Jurkat, or CEM) all expressing tetherin, were used as donors or targets (not shown and for an example of viral transfer from MT4C5 to Jurkat cells). To directly assess the role of tetherin in lymphocytes, we generated CEM cells in which expression of the protein was silenced (CEM-THN-, with about 90% of Tetherin down-regulation, Figure S2a
). CEM-THN- transmitted more efficiently ΔVpu to target lymphocytes than parental CEM cells (Figure S2b
). Altogether, these results show that tetherin significantly reduces HIV cell-to-cell transmission from various primary and permanent cell types (HeLa, 293T, and lymphocytes).
Which step of viral spread does tetherin alter? We examined whether the protein affected VS formation in lymphocytes. We measured the recruitment of Gag proteins at the contact zone between donors and recipients, a hallmark of VS formation 
. Lymphocytes (MT4C5 T cell line) were infected with Vpu positive or negative viruses. With ΔVpu, the cell surface Gag signal was generally more intense than with WT, and appeared mostly as large patches of fluorescence, reflecting the trapping of virions (Figure S3a
). In non-infected MT4C5 cells, tetherin was found at the cell periphery and in intracellular compartments (Figure S3a
), likely corresponding to the Golgi or endosomal network 
. As expected, tetherin colocalized with Gag at the surface of ΔVpu-infected cells, and was down-regulated in WT-infected cells (Figure S3a
). Infected cells were then incubated for 1 h with recipient lymphocytes (Jurkat cells stained with Far Red dye). The percentage of Gag+ donor cells forming conjugates with Far Red+ cells was similar with WT and ΔVpu viruses (not shown). About 30% of conjugates displayed a polarization of Gag at the junction, irrespectively of the presence of Vpu (). Interestingly, with ΔVpu, the large Gag-containing patches accumulated at the contact zone (). We then investigated the distribution of tetherin in conjugates of infected cells and targets. The protein colocalized with Gag at the synapse in about 80% of conjugates with ΔVpu, whereas it was much less present in WT-induced synapses, probably as a direct consequence of Vpu-mediated removal of tetherin from the cell surface (). Tetherin also accumulated at the intercellular zone when tetherin-negative Jurkat cells were used as targets (Figure S3b
), strongly suggesting that molecules found at the VS originated from donors, without requiring the presence of the antiviral protein in targets. Tetherin enrichment was not detected with control non-infected donors mixed with targets (not shown). Altogether, these results show that tetherin does not prevent Gag polarization and VS formation, and accumulates with Gag proteins at the junction zone.
Tetherin accumulates with Gag at the virological synapse.
We next visualized the spatio-temporal events leading to viral transfer in living cells. We used an infectious HIV-GagGFP virus 
and its Vpu-deleted counterpart. Jurkat cells producing WT and ΔVpu HIV-GagGFP were cocultivated with targets that expressed a red fluorescent protein (RFP)-actin, and images were acquired every 20 s for 2 h. As previously reported 
, virological synapses or polysynapses readily formed with WT HIV, illustrated by Gag polarization at the junction and subsequent transfer (Video S1
and ). In donor cells, ΔVpu HIV-GagGFP often appeared as patches which were larger than those observed with its Vpu-positive counterpart (Video S2
and ), likely reflecting the activity of tetherin 
. Time-lapse analysis showed that the large patches of Gag proteins originated from all regions of the plasma membrane and gained access to intercellular contact zones (Video S2
and ). Both WT and ΔVpu viral materials from donor cells were then in part transferred to recipient cells (Videos S1
What is the behavior of WT and ΔVpu viruses after their transfer to target cells? Infected HeLa cells were cocultivated with Jurkat cells, targets were harvested after 2 h, and Gag distribution was examined. We readily distinguished two types of Gag staining after transfer, the first corresponding to small and discrete puncta, and the second associated with large aggregates of Gag proteins (). These two viral species mirrored those observed in donor cells. WT HIV was mostly transferred as small clusters, whereas ΔVpu appeared as large aggregates in 70% of the targets (). The number of ΔVpu-infected cells displaying large clusters was strongly reduced when HeLa-THN- were used as donors (). To document further these large Gag-positive bundles in target cells, we followed the localisation of ΔVpu HIV-GagGFP on targets by correlative microscopy analysis. This technique combines fluorescence and scanning electron microscopy of the same samples over a wide range of magnification. The Gag signal corresponded to an agglomeration of viral-like particles (VLPs), each with a size of about 100 nm, assembled as large clusters (). Immunogold staining revealed that these VLPs were decorated with HIV Env+ dots, and likely corresponded to HIV-1 virions ( and S4
). These VLPs were not visible in non-infected cells (not shown and 
). Additional immunofluorescence stainings on target cells confirmed a colocalization of Gag and Env (Figure S5a
). Moreover, target cell membrane labelling with cholera toxin, a raft marker, suggested that these Gag clusters accumulated at the surface (Figure S5b
). These large patches were still observed 15 or 24 h after harvesting the targets, and are thus relatively long-lived (). The conglomeration of Gag in recipient Jurkats similarly occurred after coculture with ΔVpu-infected lymphocytes (), and is thus not due to the use of HeLa as donors. Furthermore, these large aggregates were positive for tetherin ( and S5c
). The tetherin signal originated from donors, since it was detected when using tetherin-negative Jurkat recipients (). Therefore, tetherin is transferred along with HIV particles to recipient cells. These observations are consistent with the incorporation of tetherin into virions 
Distribution of transferred WT or ΔVpu viruses on target Jurkat cells.
Tetherin promotes transfer of large viral patches and inhibits productive infection.
Productive cell-to-cell transfer of the R5 tropic AD8ΔVpu strain to MT4-CCR5+ cells was also inhibited by tetherin (Figure S6a
). With AD8ΔVpu, the large characteristic patches of Gag-positive material were also readily detected in target CCR5-negative Jurkat cells and CCR5+ primary CD4+ lymphocytes (Figure S6b
), and not in a CD4-negative Jurkat subclone (not shown). The results suggest that transfer of these viral patches requires CD4 binding but not coreceptor expression in recipient cells. This event, however, did not lead to productive infection in the absence of CCR5 (not shown). These results also demonstrate that tetherin can restrict intercellular spread of X4 and R5 viruses.
We further documented by real-time imaging the fate of viruses after their transfer to targets. After 2 h of coculture with WT and ΔVpu HIV-GagGFP producer cells, Jurkat cells were harvested and monitored for up to 6–8 h ( and Videos S3
). At time zero post coculture, ΔVpu viral aggregates were apparently larger, and more numerous than WT virions. A standardized quantification demonstrated a 7-fold increase in the fluorescent signal per cell with ΔVpu (). This confirmed that the impaired productive infection of targets did not result from a reduced transfer of viral material. The fate of incoming viral particles was apparently different with WT and ΔVpu. In the presence of Vpu, the punctate fluorescent signals decreased in number and intensity overtime. In addition to signal quenching, this decrease may reflect viral detachment, endocytosis, degradation or fusion events ( and Video S3
). In the absence of Vpu, the large patches were apparently stable, some of them remaining visible at the plasma membrane after 6–8 h ( and Video S4
We then assessed the early events of viral replication in targets by quantifying viral DNA synthesis. HeLa and HeLa-THN- cells, infected with WT and ΔVpu HIV, were cocultivated with Jurkat cells for 1 h. Target cells were then harvested, further incubated at 37°C and analyzed as a function of time by qPCR for the presence of reverse transcription (RT) products. Nevirapine was included as a control to ensure that the detected PCR products were the result of proviral DNA neosynthesis ( d, e). With WT, we observed an increase of RT products overtime, reaching about 10 copies per cell at 24 h after infection ( d,e). Viral DNA synthesis was similar when tetherin positive and negative cells were used as donors ( d,e). The situation was different with ΔVpu. With tetherin-negative donors, viral DNA synthesis occurred efficiently, reaching 10–15 copies at 24 h, which is even slightly higher than levels observed with the WT virus ( d,e). Tetherin drastically reduced the appearance of RT products, which barely exceeded background levels observed in Nevirapine-treated cells. Therefore, tetherin, when expressed in donor cells, imprints the virus, resulting in a strong decrease (5 fold reduction) of viral DNA synthesis after viral transfer to targets.
Our results indicate that tetherin impairs an early step of the viral cycle. We hypothesized that the fusion ability of the viral aggregates could be reduced. We adapted a cell-free virion-based assay 
to analyze viral fusion after cell-to-cell transfer. This assay involves the use of viruses containing a beta-lactamase-Vpr (BlaM-Vpr) protein chimera (see experimental outline ). After 2 h of coculture with infected cells, target cells are harvested. The successful cytoplasmic access of Blam-Vpr as a result of fusion is monitored by the enzymatic cleavage of CCF2-AM, a fluorogenic substrate of beta-lactamase loaded in targets. We used as donor HeLa cells endogenously expressing tetherin, and producing WT and ΔVpu HIV. Fusion of the wild-type virus was readily detected, with more or less 5% of target Jurkat harbouring cleaved CCF2-AM (). There was a significant (2.7 fold) decrease of fluorescent targets with ΔVpu (). We used as control donors cells producing an Env-deleted (ΔEnv) or a fusion-defective HIV (F522Y mutant, that carries a point mutation in Env abrogating fusion but retaining CD4 binding 
). None of these two mutants scored positive (), confirming that the assay detects viral fusion, and not virion endocytosis 
. Furthermore, a gentle shaking of the cocultures inhibited the appearance of cleaved CCF2-AM+ target cells (Figure S7
) strongly suggesting that cell-free viral particles play a minor role in this viral fusion assay. Altogether, our results demonstrate that tetherin impairs viral fusion and subsequent productive infection of target cells.
Tetherin reduces fusion after viral transfer to target cells.