Identification of host trafficking genes required for HIV-1 virological synapse formation in dendritic cells.

The lentivirus human immunodeficiency virus (HIV) targets and destroys CD4+ T cells, leaving the host vulnerable to life-threatening opportunistic infections associated with AIDS. Dendritic cells (DCs) form a virological synapse (VS) with CD4+ T cells, enabling the efficient transfer of virus between the two cells. We have identified cellular factors that are critical in the induction of the VS. We show that ADP-ribosylation factor 1 (ARF1), bridging integrator 1 (BIN1), and Rab GTPases RAB7L1 and RAB8A are important regulators of HIV-1 trafficking to the VS and therefore the infection of CD4+ T cells. We found these cellular factors were essential for endosomal protein trafficking and formation of the VS and that depletion of target proteins prevented virus trafficking to the plasma membrane by retaining virus in intracellular vesicles. Identification of key regulators in HIV-1 trans-infection between DC and CD4+ T cells has the potential for the development of targeted therapy to reduce trans-infection of HIV-1 in vivo.

D endritic cells (DCs) are key antigen-presenting cells that provide an important link between innate and adaptive immune systems, activating T cells (reviewed in references 1 and 2). Although HIV-1 is able to replicate in DCs, the process is inefficient and produces low levels of infectious virus (3)(4)(5)(6)(7)(8). However, DCs are able to transfer intact viral particles to target T cells via virological synapses (VS) by a process termed "trans-infection" (9), contributing to the spread of infection in vivo (10,11).
HIV-1 trans-infection has been shown to depend on the ability of the virus to "surf" along the surface of the DC via actin-rich dendrites to promote trans-infection (12)(13)(14). Several studies conducted in macrophages and DCs located viruses sequestered into plasma membrane invaginated compartments from which viral particles are released at the VS (15)(16)(17)(18). These compartments are thought to be surface accessible (15); however, there is evidence of a population becoming isolated from the cell surface (16). It is established in macrophages and DCs that these surface-accessible compartments may have complex morphologies that require membrane trafficking regulation, such as the virus-containing compartments found in macrophages (17).
It has been previously reported that HIV-1 enters the cell through the endolysosomal pathway, with evidence supporting roles for clathrin-mediated endocytosis (4,28,29), receptor-mediated endocytosis (30,31), and macropinocytosis (32). However, at later time points virus accumulates in virus-containing compartments rich in tetraspanins, such as CD81, that are continuous with the plasma membrane (4,15,17). More recent studies identified the importance of tetraspanin 7 (TSPAN7) and dynamin 2 (DNM2) in maintaining viral particles on dendrites and promoting efficient viral transfer. Disruption of these targets led to sequestration of virus in intracellular vesicles and a reduction in viral transfer (13).
To elucidate the role of membrane trafficking in the capture and trafficking of the virus through DCs to the VS, we performed a high-throughput small interfering RNA (siRNA) screen targeting membrane trafficking proteins. Our results identified proteins involved in vesicle trafficking between early endosomes, the trans-Golgi network (TGN), and the plasma membrane that reduce transfer of HIV-1 from DC to T cells. We show that HIV-1 is dependent on a functioning endocytic pathway, the disruption of which results in an accumulation of virus in intracellular vesicles, blocking trafficking of the virus to the virological synapse.

RESULTS
The siRNA membrane trafficking library identified genes involved in HIV-1 trans-infection between DCs and T cells. To identify the cellular trafficking pathways involved in the transfer of HIV-1 in trans-infection from DC to CD4 ϩ T cells, an siRNA library targeting 140 membrane trafficking genes was utilized. SMARTpool siRNA was transfected into monocyte-derived DCs (MDDCs) 48 h before infection with full-length CXCR4-tropic HIV-1 (R9) and cocultured with SUPT1 cells at 1:1 ratio. HIV-1-infected SUP-T1 cells were analyzed by flow cytometry 48 h later. No infection was detected in SUP-T1 cells inoculated with an HIV-1 fusion-mutant control (Fig. 1a). Nontarget siRNA was used to compare infection levels and showed Ͻ20% variation between replicates (Z score ϭ 1.5 standard deviation); therefore, the lower assay cutoff point was set at 20%.
Efficient HIV-1 trans-infection requires vesicle trafficking at the plasma membrane. To understand potential relationships between the genes selected in the siRNA screen, gene annotation enrichment analysis was used to identify common interactions between the candidate genes that may be involved in the trans-infection of HIV-1 between DCs and T cells. Analysis of the siRNA candidates was carried out for cellular compartments and biological processes (Table 1). Our results show that the genes required for optimal viral transfer are primarily involved in endocytic compartment regulation, whereas genes that restrict viral transfer are largely involved in clathrincoat-mediated endocytosis and actin-dependent processes at the plasma membrane. Taken together, the data suggest that preventing viral uptake via clathrin-coated vesicles enhances viral transfer, which is likely due to increased retention of virus on the cell surface. This finding is in agreement with studies that show HIV-1 transmitted in trans between DCs and T cells from the surfaces of DCs (12,13). In contrast, genes required for efficient trans-infection are strongly associated with cytoplasmic membranebound vesicles and vesicle-mediated transport, supporting the view that HIV-1 is sequestered into intracellular virus-containing compartments (VCC) (15,16,33). ARF1, BIN1, RAB7L1, and RAB8A are required for HIV-1 trans-infection. The siRNA library used to identify target genes is comprised of a set of four separate siRNA sequences that target different regions of the same gene; these are pooled to reduce the potential off-target effects of siRNA transfection. The knockdown of the pooled siRNA typically reflects the most functional siRNA within the pool. Therefore, the four individual siRNA can be analyzed for their ability to reduce viral transfer to validate whether the observed phenotype is a genuine on-target effect.
The main aim of our study was to identify cellular pathways involved in the delivery of HIV-1 to the VS to aid trans-infection; therefore, we focused our investigation on the genes that facilitate the transfer of HIV-1 between DC and T cells. Each of the four siRNA were transfected individually into MDDCs, infected with HIV-1 (R9) and cocultured with autologous T cells for 48 h. Of the final nine candidates, three showed a reduction in transfer (Ն20%) in at least two of the four individual siRNA, across four independent  donors: BIN1 (siRNAs B and C), RAB7L1 (siRNAs A, C, and D) and RAB8A (siRNAs A and B) (Fig. 2a). An average reduction of 50% in transfer was evident for ARF1 siRNA A, whereas ARF1 siRNA B produced a 17 to 20% knockdown in viral transmission in three of the four donors analyzed. Thus, in conjunction with the targeted reduction of ARF1 at the protein level, this result indicates that ARF1 siRNA A was the most functional siRNA in the pool, and it was therefore decided to pursue this candidate further.
To determine the level of protein depletion in MDDCs, cell lysates transfected with 200 nM pooled siRNA (Fig. 2b) targeting the entire length of the gene were analyzed by Western blotting. Knockdown was quantified by densitometry relative to protein expression levels in nontarget siRNA-transfected lysates. An efficient knockdown was achieved using pooled siRNA; a 35% Ϯ 17% reduction in protein expression was observed for ARF1, 52% Ϯ 6.6% for BIN1, 53% Ϯ 23.4% for RAB7L1, and 54% Ϯ 8.1% for RAB8A compared to nontarget siRNA (Fig. 2c).
To confirm whether siRNA is capable of reducing viral trans-infection independent of viral strain, MDDCs were transfected with the selected target siRNA and infected with either R8BAL (CCR5-tropic) and R9 (CXCR4-tropic) HIV-1. A significant reduction in viral transfer, ranging between 26 and 40% in R9-infected cells and between 35 and 45% in R8BAL-infected cells, was observed for all candidates, demonstrating that host factors involved in trafficking to the VS are shared for both CXCR4-and CCR5-tropic strains of HIV-1 ( Fig. 2d and e). All MDDC transfected with pooled siRNA remained Ͼ80% viable compared to control cells, ensuring that the reduction in transfer is not due to the cellular toxicity of the siRNA transfection (Fig. 2f). Further, siRNA transfection of MDDCs resulted in a marginal (Ͻ5%) increase in the DC maturation marker CD83. Viral binding of p24 Gag also saw a marginal increase compared to untreated and mock-transfected cells; however, HIV-1 internalization was not affected, confirming that the observed reduction in trans-infection is not due to decreased binding or internalization of the virus (data not shown).
Additional experiments were conducted on selected candidate siRNAs showing evidence of protein knockdown and a reduction in viral trans-infection in at least two of the four individual siRNAs tested. siRNA candidate genes that failed to meet these criteria (WASF1, EPN3, PAK1, and AMPH1) showed no evidence of a reduction in viral trans-infection when MDDCs were transfected with individual siRNAs, nor were we able to detect a specific knockdown in protein expression, suggesting that the previously observed reduction in viral trans-infection maybe due to off-target effects of those specific siRNAs. Therefore, these genes were eliminated from further analysis along with AP1M1 which showed high variability in the reduction of viral transfer between donors. The final candidates included ARF1, associated with retrograde transport at the Golgi compartment and protein transport to endosomes (34,35); BIN-1, known to form a complex with dynamin to control vesicle transport and scission (36); RAB7L1, a GTPase required for retromer recycling between the TGN and endosomes (37); and GTPase RAB8A, which is involved in polarized vesicular trafficking to the plasma membrane from the TGN (38).
Depletion of target proteins reduces virological synapse formation between MDDC and CD4 ؉ T cells. The efficient trans-infection of HIV-1 from DCs to T cells is dependent on the formation of VS, an adhesive structure that promotes viral transmission (39,40). To assess whether the observed reduction in trans-infection was due to a reduction in VS formation, siRNA-transfected MDDCs were infected with HIV-1 R9 or R8BAL and cocultured with autologous CD4 ϩ T-cells. Imaging of the transfected MDDCs revealed that in the case of BIN1 and RAB7L1 siRNA-transfected cells, HIV-1 R9 appeared to accumulate in large cellular vesicles at the plasma membrane and did not form VS with the T cells in spite of apparent interactions between the two cell types. In addition, ARF1-and RAB8A-depleted cells also appear to inhibit VS formation; however, the accumulation of virus can be seen in smaller vesicles at the cell periphery (Fig. 3a). Quantification of VS was similar in nontarget siRNA, untreated, and mocktransfected cells. All candidate siRNAs had a 40 to 60% reduction in VS formation between DCs and T cells compared to nontarget siRNA-transfected cells (Fig. 3b).
Similar results were seen for R8BAL-infected MDDCs, where a reduction in VS number with T cells was observed; however, BIN1-and RAB7L1-transfected cells did not accumulate virus in intracellular vesicles to the extent seen in R9-infected MDDCs ( Fig. 3c and d). In addition, we observed that LFA-1, a stabilizing component of the VS, did not become enriched at the interface between the MDDCs and T cells in the absence of virus (data not shown). These data suggest that virus targets cytoplasmic vesicles after entry into MDDCs; however, onward trafficking of virus to the plasma membrane is inhibited by depletion of the target genes, preventing VS formation and reducing efficient trans-infection between the DCs and CD4 ϩ T cells.
The integrity of virus-containing vesicles is compromised in BIN1-and RAB7L1depleted MDDC cells. CD81, a type II transmembrane protein, is one of the main tetraspanins recruited to the host cell membrane during HIV-1 trans-infection and is known to colocalize with HIV-1-containing compartments in macrophages and DCs (4,18,41). To determine whether target siRNAs altered endogenous CD81 localization in MDDCs, transfected cells were labeled for CD81 (Fig. 4a). In control cells (nontarget siRNA) CD81 is found at the cell periphery with a faint perinuclear staining. In contrast, ARF1 siRNA underwent a reduction in CD81-positive vesicles that was evident within both the cytoplasm and at the cell periphery. BIN-1 and RAB7L1 depletion reduced the  CD81 vesicle number and size, whereas no significant difference was observed in cells depleted of RAB8A ( Fig. 4b and c). In all three cases, an accumulation of CD81 vesicles was observed within the cytoplasm not at the cell periphery (Fig. 4a).
CD81 plays an important role in regulating viral trans-infection at the VS, and depletion of the tetraspanin can reduce viral trans-infection (42). In light of previous findings, we assessed CD81 localization during HIV-1 infection. As expected, we observed p24 Gag colocalization with CD81 at the cell periphery in CD81 tetraspaninenriched microdomains (TEMs) at 4 h postinfection in control cells. We observed that the numbers of CD81 p24 Gag TEMs in transfected MDDCs are reduced in ARF1depleted cells. In contrast, in cells transfected with siRNA targeting BIN1, RAB7L1, and RAB8A there were both virus and CD81 at the cell periphery; however, the staining of the TEM was diffuse and lacked the structure of the TEM ( Fig. 4d and e). This was confirmed by colocalization data, indicating that CD81 association with p24 was reduced in siRNA-transfected cells (Fig. 4f). Taken together, these data suggest that trafficking of CD81 and p24 Gag to the cell periphery to form the TEM is compromised by knockdown of ARF1, BIN-1, RAB7L1, and RAB8A, potentially preventing the efficient trans-infection of virus via the VS.

Retention of virus in endocytic compartments reduces HIV-1 transfer.
We hypothesized that the presence of virus and CD81 in cytoplasmic vesicles and the disrupted trafficking of CD81 and p24 Gag to the plasma membrane by target siRNA was due to retention in endocytic compartments. Therefore, we aimed to trap virus in endosomal derived vesicles to establish whether this directly affects viral trans-infection to T cells. MDDCs were treated with endocytic inhibitors prior to infection with R9 virus, and the level of trans-infection was measured. We utilized LY294002, a phosphatidylinositol 3-kinase (PI3K) inhibitor known to block macropinocytosis and the formation of early endosomes, and bafilomycin A1, a vacuolar type H ϩ -ATPase (V-ATPase) that prevents endosomal acidification.
Inhibition with LY294002 resulted in a mild increase of HIV-1 trans-infection (ϩ20%) compared to dimethyl sulfoxide (DMSO)-treated control cells. However, a 2-fold decrease in trans-infection was observed in cells treated with bafilomycin A1, indicating that efficient viral transfer requires a functioning endocytic pathway (Fig. 5a). This reduction was not due to inhibitor toxicity, with MDDCs remaining Ͼ80% viable during treatment and subsequent infection (Fig. 5b). To visualize any differences between MDDCs treated with LY294002 and bafilomycin, infected MDDCs were analyzed by confocal microscopy. HIV-1 was concentrated at the cell surface in cells pretreated with LY294002, which is in agreement with previous findings (13). In contrast, virus accumulates inside intracellular vesicles in cells treated with bafilomycin A1 (Fig. 5c), indicating that viral uptake into MDDCs was not inhibited, and that retention within endocytic vesicles reduced trans-infection. Controls confirmed that both horseradish peroxidase (HRP) taken into the MDDCs via the fluid phase and the lysosomal marker low-density lipoprotein (LDL) were lost in cells treated with the PI3K inhibitor LY294002, as predicted. In contrast, LDL-DIL labeling was diffuse and cytoplasmic in cells treated with bafilomycin A1, suggesting a block in LDL-DIL uptake by endosomes in the MDDCs. On the contrary, HRP taken up via the fluid phase was less affected, suggesting that unlike LDL, HRP is retained in endocyte-like compartments (Fig. 5d).
HIV-1 did not colocalize with the organelle markers EEA1, Rab5, Rab7, Rab11, LAMP2, or CHMP2B in either siRNA-transfected or bafilomycin A1-treated MDDCs in our exper- Taken together, these data indicate that HIV-1 transfer is reliant on a functioning endocytic pathway. Blocking virus in endosomal derived compartments results in the accumulation of virus in cytoplasmic vesicles, which in turn reduces viral transfer between MDDCs and T cells, as seen in siRNA-transfected MDDCs. In addition, bafilomycin A1 appears to block LDL-DIL but not HRP or HIV-1 uptake into MDDC, suggesting that HIV-1 is predominantly trafficked to cellular compartments that differ from those utilized by LDL.
Downstream trafficking from early endosomes is compromised in MDDCs transfected with target siRNA. Endosomal cargo has one of two fates; it is either recycled back to the cell surface (i.e., transferrin), or it is directed to lysosomes for degradation (i.e., LDL). To confirm that target siRNA is blocking endosomal trafficking in MDDCs, cells were transfected with pooled ARF1, BIN-1, RAB7L1, and RAB8A siRNAs and either stained for early endosomes with early endosome antigen 1 (EEA1), incubated with Alexa Fluor-labeled transferrin, or LDL-DIL, to assess the recycling and lysosomal trafficking pathways, respectively.
In nontarget siRNA-transfected MDDCs, EEA1 is seen in numerous vesicles of various sizes throughout the cell. MDDCs transfected with siRNA against ARF1, BIN1, and RAB8A resulted in the formation of abnormal endosomes marked by a decrease in both number and size. RAB8A siRNA resulted in more numerous, enlarged vesicles evident at the cell periphery (Fig. 6a to c).
In all instances, labeled transferrin was found localized at the cell periphery, with no discernible differences between control and siRNA-transfected cells (Fig. 6a, second  panel). However, a reduction in vesicle number and Alexa Fluor labeling was observed in BIN1-, RAB7L1-, and RAB8A-transfected cells (Fig. 6d and e).
In nontarget siRNA or mock-transfected control cells, LDL-DIL predominantly accumulates in lysosomes in the perinuclear region. In cells transfected with siRNA targeting ARF1, LDL has accumulated in various-sized vesicles in the cytoplasm (Fig. 6a, third  panel). Knockdown of BIN-1, RAB7L1 resulted in a reduction of LDL containing vesicles within the cells, indicating that the delivery of LDL-DIL to lysosomes is significantly reduced (Fig. 6a and f to g). RAB8A-silenced cells were also found to have a reduced number of LDL vesicles; however, diffuse staining is evident within the cytoplasm, suggesting that LDL is taken into the cell but not trafficked within the endolysosomal pathway.
These observations suggest ARF1 regulates endosomal morphology and vesicle formation and slows LDL trafficking to the perinuclear region but is dispensable for the recycling of transferrin to the plasma membrane. BIN1 and RAB7L1 also affect endosomal vesicle formation, resulting in the retention of vesicles at the cell periphery and reducing downstream trafficking from endosomes, as evidenced by a reduction in both transferrin-and LDL-containing vesicles, suggesting that BIN1 and RAB7L1 play a role in early endosomal protein trafficking. In contrast, RAB8A depletion appears to increase early endosome size, although trafficking of both LDL and transferrin is also reduced, suggesting that RAB8A action is targeted more downstream, regulating protein trafficking after the cargo has left the endocytic compartment. The disruption of endosomal vesicle trafficking at or after the early endosomal compartment by target siRNA creates a knock-on effect, altering endocytic trafficking to lysosomes and recycling of cargo to the plasma membrane. Taken together, the disruption of TGN-endosomalplasma membrane trafficking suggests that HIV-1 trafficking from internalized compartments relies on endosomal sorting pathways to traffic to and accumulate at the VS and potentially within VCC at the cell surface.
HIV-1 trans-infection requires retromer complex recycling of cargo. Based on the findings that trafficking between key endosomal compartments is compromised in siRNA-targeted cells, which in turn reduces HIV-1 trafficking to the VS, we wanted to confirm whether trans-infection of HIV-1 was reduced when trafficking to the plasma membrane from the TGN or early endosomes is compromised. Several of our selected genes play a key role in endosomal sorting to the TGN and plasma membrane, with RAB7L1 specifically involved in retromer activity. In addition, a proportion of transferrin and its receptor are recycled in a retromer-dependent manner to the plasma membrane (43). The retromer has also been found to play a key role in HIV-1 Env trafficking and viral assembly (44). Thus, we decided to investigate the role of the retromer complex in trans-infection using siRNA targeting key components, VPS26A and VPS35, of the retromer complex. HIV-1 trans-infection was significantly reduced in MDDCs transfected with each of the retromer siRNAs ( Fig. 7a and b) from 25 to 50%. A more marked reduction was evident in cells infected with CXCR4-tropic strain of the virus. A protein knockdown of approximately 60% was confirmed for both VPS26A and VPS35 (Fig. 7c to e), and no reduction in cell viability was evident resulting from siRNA transfection of the VPS genes (Fig. 7f). Therefore, we were able to confirm endosomal sorting between the TGN and the plasma membrane is required for HIV-1 transinfection.

DISCUSSION
DCs perform an essential role in the transmission of HIV-1 to target CD4 ϩ T cells, promoting the spread of infection. Although there have been numerous investigations  into the role of DCs in trans-infection, the cellular trafficking pathways exploited by HIV-1 remain unclear. The identification of host cell factors and intracellular pathways exploited by HIV-1 to aid trans-infection of T cells will facilitate the development of novel therapies and may reduce initial transmission of HIV-1.
In this study, we identified a number of host factors involved in trans-infection of HIV-1 from DCs to T cells. By conducting an siRNA screen targeting membrane trafficking proteins, we identified four genes involved in efficient trans-infection from DCs to T cells. Although one similar shRNA/siRNA screen has been conducted investigating the role of membrane trafficking in HIV-1 trans-infection, none to date has focused on the genes we identified in the present study. In a recent shRNA screen, Menager and Littman (13) also identified ARF1, ARF6, and ARPC1B as reducing viral transfer and CLTC, CLTB, and AP2M1 as enhancing viral transfer; however, the ability to draw direct comparisons between the two studies is complicated by the fact that these authors used shRNA technology in a screen that targets a different gene library, several of them not included in the siRNA screen we utilized. The study then proceeds to concentrate on TSPAN7 and DNM2 and their role in trans-infection at the cell surface, whereas we focused on the trafficking of internalized virus. In a study using an identical siRNA screen, Wen et al. identify a number of common genes such as RAB7L1, AP1M1, BIN-1, ARPC1B, DIAPH1, ARF6, WASF1, CLTC, and VAV2 required for HIV-1 and M-PMV virus release from HeLa cells (45). Overall, there is a high consistency of hits between previous screens conducted in DCs and our own membrane trafficking screen, verifying our findings.
Our initial siRNA screen shows that knocking down genes associated with clathrincoated vesicle formation enhanced trans-infection; this suggests that restricting viral uptake into MDDCs and retention of virus on the cell surface promotes HIV-1 transinfection. It has been previously demonstrated that soluble CD4 protein is able to inhibit infection, suggesting that virus particles bound to the surface of the MDDCs were the main source of trans-infection (12). In support of this model, Menager et al. demonstrated that DNM2 and TSPAN7, which coordinate actin nucleation and stabilization, had roles in restricting endocytosis and maintaining virus on cellular dendrites enabling transfer (13,46). On the other hand, there is compelling evidence for the model that HIV-1 is sequestered in plasma membrane-derived invaginated compartments induced upon HIV-1 uptake (33). From this compartment, viral particles can be released to the VS to initiate trans-infection (15)(16)(17). We initially identified nine genes from the siRNA screen that reduced trans-infection. These genes were predominantly associated with cytoplasmic, membrane-bound vesicles with direct involvement in vesicle-mediated transport and membrane organization, thus supporting a requirement for membrane-bound vesicles in HIV-1 trans-infection. These results provide evidence for both viral transmission via the cell surface and trafficking via intracellular compartments to promote trans-infection in MDDCs.
Although the use of primary MDDCs and CD4 ϩ T cells is a representative model of HIV-1 trans-infection, employing methods such as siRNA transfection within established MDDCs has its limitations. Generally, 50% transfection efficiency is achieved, which in turn does not completely block reduction of trans-infection within these cells. However, partial knockdown is still capable of producing a strong phenotype, and the study of these pathways in primary cells is essential to uncovering underlying mechanisms of trans-infection and is critical for investigating and identifying such cellular processes.
Here, we concentrate on studying genes required for efficient viral trans-infection and therefore aim to investigate how internalized virus is trafficked to the VS. We demonstrate that the reduction in viral trans-infection observed from the depletion of four genes-ARF1, BIN1, RAB7L1, and RAB8A-is due to the apparent retention of virus in intracellular vesicles and a reduction in virus accumulation at the VS between DC and T cells. MDDCs can capture and store HIV-1 virions in invaginations at the plasma membrane (9,15). Live imaging shows that viral puncta are trafficked into enclosed intracellular compartments (47); whether these compartments are enclosed or remain accessible to the cell surface is still a matter of debate (15,16). The integrity and formation of intracellular compartments are believed to be regulated by membrane trafficking processes (17). Based on these data, we propose that the reduction in VS formation observed in siRNA-treated MDDCs disrupts the regulation and trafficking of intracellular compartments, resulting in the retention of viral particles within intracellular vesicles, preventing onward trafficking to the VS and therefore inhibiting viral trans-infection.
VS formation and HIV-1 spread relies on the interaction of MDDCs and recipient T cells, triggering the active polarization of organelles and cell surface proteins. One such component, LFA-1, has been shown to induce T-cell polarization toward the VS to induce efficient viral T-cell-to-T-cell spread (48). In the context of VS formation between DCs and T cells, it has been reported that cell-to-cell contacts are not increased by the presence of HIV-1 and that the formation of the VS was decreased by 60% when the interaction between ICAM-1 and LFA-1 was blocked (49). Our findings agree with these data. We also observed several T cells interacting with HIV-1-infected siRNA-transfected MDDCs; however, accumulation of LFA-1 at the VS was only evident in the presence of HIV-1 p24 Gag. These data suggest that by blocking trafficking of HIV-1 to the cell periphery, enrichment of LFA-1 at the MDDC-T-cell interface is also prevented, restricting VS formation. It may be the case that virus alone is not the only trigger for VS formation, and it is plausible that by blocking trafficking of HIV-1 to the cell surface in MDDCs we may also be preventing the recruitment of other key components to form efficient VS.
We also observed the retention of endogenous CD81 in cytoplasmic vesicles and a reduction in localization at the cell periphery. In addition, at 4 h postinfection, TEMs are reduced or disrupted, potentially affecting the recruitment and budding of HIV-1 at the VS. The tetraspanin CD81 colocalizes with HIV-1 within VCC (4,18,41) and accumulates at the VS, promoting viral trans-infection, preventing cell-to-cell fusion, and providing a platform for viral budding (50,51). Our results are consistent with these findings, suggesting that trafficking of CD81 within MDDCs to the plasma membrane and recruitment to TEMs, along with HIV-1, are required for trans-infection. This is supported further by a study showing that blocking CD81 with specific antibodies reduces VS formation (52). Conversely, Krementsov et al. showed that direct depletion of CD81 actually enhances viral transmission between HeLa and Jurkat cells (53). The different outcomes observed in these studies may reflect the different methods and cell types employed to target CD81 and reduce its presence at the VS. Our data support the former approach, where CD81 is still present within the cell but is prevented from forming functioning TEMs at the cell periphery, whereas actual depletion of CD81 from cells may have a number of downstream effects, altering normal cell function.
Overall, we demonstrate that targeting host factors that regulate endocytic compartments and vesicle trafficking to the plasma membrane within MDDC results in the disruption of trafficking of CD81 and virus to the VS reducing trans-infection.
Our results show that upon disruption of target genes, protein trafficking to lysosomes and recycling of transferrin to the plasma membrane is reduced: this suggests that endosomal sorting and recycling to the plasma membrane are closely linked to trans-infection in MDDCs.
In conjunction with other ARF proteins, ARF1 plays an important role in the regulation of recycling endosome morphology and recycling pathways; however, depletion of the gene was not found to directly affect the recycling of transferrin receptor (54,55). Depletion of ARF1 in our study is consistent with this role in protein recycling in MDDCs, altering endosomal morphology but not affecting recycling of transferrin to the plasma membrane. In the context of infection, HIV-1 ability to mediate the downregulation of MHC-1 is achieved by targeting AP-1 and ARF1 activity (56), resulting in the accumulation of MHC-1 in the TGN or endosomes (57). HIV-1 Vpu also targets the same pathway (58,59) to counteract tetherin, which is known to block the release of progeny virus from the cell (60). These data, in conjunction with the fact that ARF1 binding partner AP1M1 was originally identified as a potential gene required for trans-infection in our screen, support the idea that the same recycling pathway could be utilized for the successful trans-infection of HIV-1 in DCs. Depletion of ARF1 is likely to impact the morphology of virus-containing compartments and the recycling of internalized virus to the cell surface, which in turn reduces the accumulation of virus at the VS and therefore trans-infection.
BIN-1 is a key player in membrane remodeling during endocytosis and endosomal sorting and is essential for the formation of plasma invaginations in muscle tissue (61). BIN-1 mutants were found to both impair membrane tabulation and cause compact membrane curvature (62). Our findings support these data: depletion of BIN-1 in MDDCs reduces endosomal size, producing small round vesicles and preventing downstream trafficking. A role for BIN-1 in HIV-1 infection is supported further by a study that identifies the upregulation of BIN-1 in CD4 ϩ and CD8 ϩ T cells from ex vivo patients (63). Based on this, we propose that BIN-1 is required for the efficient formation and function of plasma membrane invaginations and endosomal sorting that assist the trafficking of HIV-1 to the VS.
RAB7L1 also plays a role in intracellular trafficking and the endosomal sorting of lysosome-bound membrane proteins (64). Again, our results support a similar role for RAB7L1 in MDDCs; the transport of both LDL and transferrin was impaired in RAB7L1depleted MDDCs, suggesting that trafficking from endosomal compartments is compromised. The finding that RAB7L1, along with AP1M1, is involved in HIV-1 Gag trafficking and virion budding in the activated macrophage cell line MM6 and in CD4 ϩ Jurkat cells (65) supports a role for RAB7L1 in the recruitment of HIV-1 particles in MDDCs to the VS to assist viral budding at the cell surface.
RAB8A is known to control vesicular transport and promote membrane protrusions, which can be inhibited by blocking membrane recycling (66), in agreement with our findings. Knockdown of RAB8A by siRNA in previous studies was found to inhibit HIV-1 replication in HeLa P4/R5 cells and directly interact with nef, env, and gag-pol (67). Therefore, it seems plausible that the depletion of RAB8A in MDDCs inhibits membrane recycling and therefore membrane protrusions, reducing HIV-1 trans-infection. The data also support the idea that HIV-1 taken up by MDDCs could rely on the same recycling pathways to traffic to the cell membrane to accumulate virus at the VS.
The data presented here clearly point to a role for endocytic recycling pathways in HIV-1 trans-infection; therefore, we investigated the retromer complex implication in the trafficking of HIV-1 to the VS. Retromer-dependent protein sorting pathways provide an opportune target for a variety of viral and bacterial pathogens (68,69). For instance, HIV-1 envelope protein and herpesvirus saimiri, a T-lymphotrophic tumor virus, bind the retromer to aid infection and viral release (44,70), whereas influenza A M2 protein escapes degradation via transportation from early endosomes to the TGN (71). Our data confirm a role for the retromer in DC-mediated HIV-1 trans-infection and exploitation of recycling pathways by the virus to achieve efficient transfer between cells.
We hypothesize that VCC and VS formation is dependent on the retromerdependent endocytic-TGN-plasma recycling pathway. By exploiting the retromer pathway, internalized viral particles can be subverted to the plasma membrane, where virus becomes sequestered to promote VS formation and enable trans-infection between MDDCs and T cells (Fig. 8).
DCs are among the most important cellular targets in early HIV-1 transmission. HIV-1 is thought to accumulate in "viral endosomes" where the virus is able to exploit a pathway essential for the delivery of components to the immunological synapse and activation of T cells (4). Uptake into DCs using this method not only allows efficient trans-infection to target CD4 ϩ T cells but also evades detection by the immune system (27), the importance of which was shown in vivo using a humanized mouse model (10,11).
By using high-throughput siRNA screening, we were able to identify ARF1, BIN1, RAB7L1, and RAB8A as essential for endosomal trafficking between the TGN and early endosomes and coordinated transport to the plasma membrane in a retromerdependent manner. Thus, we identify key cellular trafficking proteins exploited by HIV-1 in DCs to efficiently disseminate virus to target T cells promoting trans-infection. A better understanding of the role of these proteins in viral transfer to T cells may serve as potential candidates for targeted therapy to control the transfer of HIV-1 between DCs and T cells in vivo.

MATERIALS AND METHODS
Ethics statement. Peripheral blood mononuclear cells (PBMCs) were derived from buffy coats obtained from healthy blood donors, anonymously provided by the Welsh Blood Service, UK. Written informed consent for the use of buffy coats for research purposes was obtained from blood donors, and the use of patient samples and procedures were approved by the local research ethics committee at Cardiff University.
CD4 ϩ T cells were isolated using CD4 ϩ magnetic beads (Miltenyi Biotech), maintained in the presence of IL-2, and activated 4 days before use with 2 g/ml PHA. SUP-T1 T lymphoblasts and 293T human embryonic kidney (HEK) cells (obtained from NIH AIDS Research and Reference Reagent Program) were maintained in supplemented RPMI 1640 or Dulbecco modified Eagle medium, respectively. Viral stock production. Viral stocks were produced by transfection of HEK293T cells with calcium phosphate DNA precipitation of proviral plasmids encoding full-length HIV-1ϫ4 and R5 provirus, pR9 and pR8BAL, respectively (the plasmids were provided by D. Trono D [EPFL, Lausanne, Switzerland]). Infectious titers were determined by titration onto SUP-T1 cells and quantification of HIV-1 p24 Gag by enzyme-linked immunosorbent assay using a Lenti-X p24 rapid titer kit (Clontech).
RNAi screen in MDDCs. An siRNA screen was performed using a commercially available SMARTpool ON-TARGET library containing 140 membrane trafficking genes (Dharmacon-GE Healthcare). MDDCs (1 ϫ 10 5 cells/well) seeded in 96-well plates were reverse transfected twice, 24 h apart, with 200 nM pooled siRNA or with control siRNA (SMARTpool ON-TARGET nontarget siRNA; Dharmacon-GE Healthcare) using HiPerFect transfection reagent (Qiagen) in serum-free media. After 48 h, the MDDCs were infected with 20 to 30 ng of p24 Gag HIV-1 R9 by spinoculation and cocultured with SUP-T1 or CD4 ϩ T cells prestained with Celltrace Far Red (Invitrogen) in the presence of indinivir at 2 g/ml (Sigma) for a further 48 h, as previously described (72).
Transfer assay. MDDCs (1 ϫ 10 5 cells/well) were reverse transfected twice, 24 h apart, with 200 nM pooled or individual siRNA using HiPerFect transfection reagent (Qiagen) in serum-free media. After 48 h, the MDDCs were infected with 5 to 10 ng of p24 Gag HIV-1 R9 or 2 to 5 ng of p24 Gag HIV-1 R8BAL by spinoculation for 2 h and then cocultured with CD4 ϩ T cells prestained with Celltrace Far Red (Invitrogen) at 37°C for a further 48 h.
Western blot analysis. At 72 h posttransfection, the cells were lysed with 1ϫ cell lysis buffer (Cell Signaling), and supernatants were harvested and reduced. Cell lysates were separated on a 4 to 12% SDS-PAGE gel and run next to a PAGEruler (Thermo Fisher) and then analyzed by Western blotting, followed by enhanced chemiluminescence detection and densitometry analysis (MyImage analysis; Thermo Scientific).
Uptake assays. Transfected MDDCs were incubated with HRP (Sigma) at 10 mg/ml for 1 h at 4°C, fixed on coverslips using 2% PFA, and labeled using the indicated antibodies.
Transfected MDDCs (1 ϫ 10 5 ) were seeded onto poly-L-lysine-coated coverslips and placed at 4°C for 10 min prior to the addition of either 12 g/ml LDL-DIL (Life Technologies) for 4 h or 25 g/ml transferrin Alexa Fluor 488 (Life Technologies) for 30 min, both at 37°C. Cells were fixed in 1% PFA and nuclei labeled with TOPRO-3 (Life Technologies).
Inhibition assays. Inhibitors LY294002 (50 M) and bafilomycin A1 (0.5 M) were added to MDDCs prior to and during infection with R9 HIV-1. DMSO was used as a control at equal concentrations. Cells were either seeded on coverslips and fixed in 2% PFA for confocal imaging, or they were washed and cocultured with CD4 ϩ T cells for 48 h at 37°C for analysis via flow cytometry.
Virological synapse assay. MDDCs transfected with siRNA were infected with HIV-1 for 2 h prior to incubation with CD4 ϩ T cells on poly-L-lysine-coated coverslips at 1:1 ratio for 40 min at 37°C. Fixed cells (2% PFA) were labeled for actin and p24 Gag-FITC and viewed on the confocal microscope. Virological synapse formation was counted if an accumulation (ϳ50% or greater) of p24 Gag was evident at or adjacent to the junction between T cells and MDDCs. T cells were identified by their smaller size and less cytoplasmic content in comparison to larger MDDCs.
Confocal immunofluorescence. Cells were adhered to poly-L-lysine coverslips (Corning), fixed in 2% PFA, permeabilized with 0.05% saponin, and stained with indicated primary antibodies in PBS-0.2% BSA-0.05% saponin, followed by Alexa Fluor-labeled secondary antibodies (1:400) when necessary. TOPRO3 in PBS (1:1,000) was used to stain nuclei (Life Technologies). Confocal microscopy analysis was carried out using a Zeiss LSM710 with a 100ϫ oil objective with 488, 546, and 633 nm acquired sequentially using ZENlite software (Zeiss). All confocal images represent a single plane. Colocalization analysis was performed using Zenlite software (Zen Blue) with the colocalization function.
Bioinformatics-protein interrelationship mapping. RNAi screen candidates were enriched using DAVID to identify significant gene ontology (GO) terms, and a protein-protein interaction network was visualized using the EnrichmentMap (Bader Lab) plug-in for Cytoscape 3.3.3. The top five significant values are reported. The minimum confidence score was set at 0.005 (74)(75)(76)(77).
Image analysis. Image analysis was performed using ImageJ software (National Institutes of Health) and analyzed with Excel software (Microsoft). A macro was designed to apply a set scale to all images, followed by the color threshold, to eliminate any background staining, and the particle analysis function was applied to quantify vesicles. Pixels were converted to m using the set scale.
Statistics. Data were analyzed using a two-sample t test, comparing nontarget to targeted siRNA samples. A one sample t test was used to compare siRNA transfer assays across donors. P values of Ͻ0.05, Ͻ0.005, and Ͻ0.0005 (indicated by asterisks [*, **, and ***, respectively] in the figures) were considered significant. Data were analyzed using Prism (GraphPad) software.