Dendritic cells (DCs) have essential roles in early detection of pathogens and activation of both innate and adaptive immune responses. Whereas human DCs are resistant to productive HIV-1 replication, they have a unique ability to take up virus and transmit it efficiently to T lymphocytes. By doing that, HIV-1 may evade, at least in part, the first line of defense of the immune system, exploiting DCs instead to facilitate rapid infection of a large pool of immune cells. While performing an shRNA screen in human primary monocyte-derived DCs, to gain insights into this cell biological process, we discovered the role played by tetraspanin-7 (TSPAN7). This member of the tetraspanin family appears to be a positive regulator of actin nucleation and stabilization, through the ARP2/3 complex. By doing so, TSPAN7 limits HIV-1 endocytosis and maintains viral particles on actin-rich dendrites for an efficient transfer toward T lymphocytes. While studying the function of TSPAN7 in the control of actin nucleation, we also discovered the existence in DCs of two opposing forces at the plasma membrane: actin nucleation, a protrusive force which seems to counterbalance actomyosin contraction.
Dendritic cells (DCs) are part of the innate immune system and constitute a first layer of defense against pathogens. One of their roles is to capture viruses and bacteria in order to present antigenic peptides to T lymphocytes to stimulate an optimal and specific adaptive immune response [1,2]. Although human DCs are able to capture HIV-1 through multiple different receptors, they are relatively resistant to productive viral infection in comparison with T lymphocytes, which are the main targets of the virus in vivo. This appears to be due to a block in reverse transcription mediated by the host dNTP hydrolase SAMHD1 [3–7]. Interestingly, a cell-intrinsic sensor for HIV-1, cGAS, has the potential to activate a type I interferon response to reverse-transcribed viral DNA in DCs, but is not typically engaged, because of this block [8–9]. However, DCs, at least in vitro, have been shown to be able to transfer HIV to T lymphocytes without first being infected [10,11]. This process, called trans-infection or trans-enhancement, has been proposed to happen through several mechanisms such as the exosome pathway following classical endocytosis , membrane protrusion before internalization , or more recently, via a tetraspanin-rich compartment (containing CD81), called the ‘invaginated pocket’, which is still accessible to solvent but is topologically distinct from the cell membrane . Thus, DCs have the potential to detect HIV-1 infection and to strongly activate virus-specific T cells, but we think this is barely happening in vivo, due to the poor replication of the virus in those cells. It is tempting to speculate that HIV-1 may have evolved to avoid infecting DCs, evading at least, in part, the first line of defense of the immune system in mucosal tissues, exploiting DCs instead to facilitate rapid infection of a large pool of T cells and perhaps to reside in non-replicating reservoirs within DCs.
A better understanding of the molecular and cell biological aspects of DC-mediated enhancement of T-cell infection in trans is needed to evaluate the importance of this process in infected individuals and potentially to identify new therapeutic targets for limiting HIV replication and eradicating viral reservoirs.
To gain more molecular insights into the transfer of HIV-1 from DCs to T lymphocytes, we performed an shRNA screen of 455 genes involved in organelle and membrane trafficking, in primary human monocyte-derived dendritic cells (MDDCs). We found 81 genes leading to a >30% decrease in HIV-1 transfer upon knockdown and another 53 genes corresponding to a >30% increase in HIV-1 transfer. In particular, we identified the protein tetraspanin-7 (TSPAN7) as a positive regulator of actin nucleation and stabilization. We discovered the important roles played by actin nucleation and cortical actin stabilization in maintaining viral particles on dendrites and limiting HIV-1 endocytosis, to allow an efficient transfer of HIV-1 to T lymphocytes .
The protein Tspan-7, encoded by the gene TSPAN7 localized on the X chromosome, belongs to the transmembrane 4 superfamily also known as tetraspanin family. There are 33 tetraspanins in mammals, mostly cell membrane proteins, which interact with each other and dynamically with numerous other protein partners to organize a hierarchical network of protein interactions in tetraspanin-enriched microdomains. By acting somewhat like scaffolding proteins and creating a dynamic web of proteins, tetraspanins have been involved in multiple different aspects of cell biology and physiology such as cell motility, signaling, morphology, neurite outgrowth and tumors [16,17]. Tspan-7 is a 27.5 kDa cell surface glycoprotein composed of 249 amino acids, highly expressed in the brain. This member of the tetraspanin family is still poorly characterized compared with other tetraspanins and not yet reported to be involved in cell-to-cell transfer of HIV, in contrast with CD9, CD63, CD81 and CD82 . Mutations of its gene, localized on the X chromosome, have been associated with non-syndromic X-linked mental retardation [19–21]. Interestingly, in cultured hippocampal neurons, TSPAN7 is required for filopodia and dendritic spine formation and stabilization, a role that is consistent with the subcellular phenotypes associated with the characterized X-linked mental retardation pathologies.
TSPAN7 appears to be essential for the formation of membrane protrusions at the surface of dendritic cells and for HIV-1 transfer from DC to T cells
Among the hits of our shRNA screen leading to a decrease in HIV-1 transfer to T cells, TSPAN7 was a top candidate with a decrease in its expression in dendritic cells, leading to a 50% decrease in HIV-1 transfer to T cells . After confocal and electron microscopy experiments, we observed a loss of long and thin membrane protrusions at the surface of dendritic cells (‘dendrites’) (Figure 1A,B). Upon TSPAN7 knockdown, those dendrites, normally enriched in actin, appeared to be replaced by short blebs devoid of actin. Linked with this switch from dendrites to blebs, we also observed an increase in HIV-1 internalization, in gigantic vesicles (≥1 µm in diameter) resembling macropinosomes (Figure 1A). This increase in macropinocytosis was not specific to HIV-1, as an increased internalization of molecules of dextran (mostly internalized by macropinocytosis) was identically observed.
Effect of inhibition of TSPAN7/actin nucleation on HIV-1 transfer.
TSPAN7 as a positive regulator of actin nucleation
Four of seven subunits of the ARP2/3 complex, a master regulator of actin nucleation, have been identified in our shRNA screen. This complex through the branching of actin filaments is at the center of multiple different cellular biological processes such as formation of membrane protrusions, filopodia, dendrites, ruffles, formation and stabilization of subcortical actin and endocytosis . By using drugs to specifically inhibit the ARP2/3 complex, we were able to mimic all the phenotypes due to TSPAN7 knockdown: switch from actin-rich dendrites to blebs, increase in HIV-1 and dextran internalization and decrease in HIV-1 transfer (Figure 1B).
No synergistic effect was observed on actin-rich dendrites, HIV-1 internalization and transfer, when inhibiting actin nucleation after decreasing TSPAN7 expression. Based on those experiments, we then concluded that TSPAN7 was a positive regulator of actin nucleation.
Actin nucleation leads to the formation of a dense and branched network of actin filaments, which are required for the development of actin-rich dendrites that can capture HIV-1 and transfer it directly to T lymphocytes (Figure 2). Interestingly, a destabilization of cortical actin, defined by a lack of continuous actin presence underlying the plasma membrane, is also observed following the loss of actin nucleation and TSPAN7 expression .
Schematic representations of the requirement for TSPAN7/actin nucleation in HIV-1 transfer from MDDCs to T cells.
Inhibition of actomyosin contraction can compensate the defect of TSPAN7 expression and actin nucleation function at the plasma membrane
Blebs, which can be defined as rounded outgrowth on the surface of a cell, are usually formed following interaction between actin filaments and non-muscle myosin 2, which generates a contractile force called actomyosin . Using drugs known to inhibit bleb formation, we were able to rescue the excess of HIV-1 endocytosis following the loss of TSPAN7 function, suggesting an uptake through macropinocytic vesicles following bleb retractions (Figure 2) . In contrast with other forms of membrane protrusion, blebs are largely devoid of branched and polymerized actin . Thus, an ARP2/3-independent, bleb-dependent macropinocytosis exists in the absence of actin nucleation that leads to HIV-1 internalization. This alternative form of macropinocytosis has already been described for other viruses like vaccinia .
In a more unexpected way, we discovered that inhibition of actomyosin contraction could also rescue dendrite formation, maintenance of HIV-1 on their surface and transfer to T lymphocytes, in the absence of TSPAN7 expression/actin nucleation (Figures 1A,B and 2). In a context where actin nucleation is still intact, inhibiting the function of non-muscle myosin 2 leads to a drastic increase in actin-rich dendrites with the maintenance of HIV-1 at their surface, resulting in an overall increase in HIV-1 transfer to T cells (Figures 1A,B and 2).
By screening for protein expressed in dendritic cells and involved in HIV-1 transfer to T cells, we discovered the essential role played by actin nucleation and its effectors. We identified TSPAN7 as a positive regulator of actin nucleation. Its function, through regulation of actin nucleation, is required to stabilize subcortical actin, which limits HIV-1 endocytosis through bleb-driven macropinocytosis. It also promotes the formation of actin-rich dendrites able to capture HIV-1 and efficiently transfer it to CD4+ T lymphocytes (Figure 2).
Tetraspanins are already known to regulate cytoskeletal rearrangements as demonstrated by their influence on integrin signaling and cell spreading after adhesion [26–29]. Complexes of tetraspanin–integrins have been proposed to regulate the actin cytoskeleton by bringing together important regulators of the ARP2/3 complex in close proximity to phosphoinositides .
Tspan-7 has been shown to interact, in other cell types and in particular in neuronal cells, with the phosphatidylinositol (PI) 4-kinase (PI4K)  and β1-integrin [21,30]. We could then imagine an indirect but classical role on the actin cytoskeleton through β1-integrin signaling and/or biosynthesis of phosphatidylinositol 4,5-bisphosphate, which could positively influence a large number of actin cytoskeleton regulators .
Owing to the indistinguishable phenotypes in MDDCs, following TSPAN7 knockdown and inhibition of ARP2/3 complex, we suspect a more direct link between TSPAN7 and actin nucleation. No direct links have been established between TSPAN7 and the ARP2/3 complex in neurons, but since dendritic spine formation is dependent on actin nucleation, a direct role of TSPAN7 on the actin cytoskeleton was proposed . A direct interaction between the C-terminal tail of TSPAN7 and the PDZ domain of PICK1 was reported from those studies . PICK1 is a PDZ and BAR domain containing protein which has been shown to interact with ARP2/3 and F-actin filaments and therefore inhibit actin nucleation [33,34]. The function of TSPAN7 described in the neurons strongly supports its newly identified role in dendritic cells and confirms some clear connection with actin nucleation processes. It would be interesting to see if TSPAN7 can also interact with PICK1 in human dendritic cells. We could then imagine a model where TSPAN7 by interacting with PICK1 at the plasma membrane, would sequester it and prevent it from interacting and inhibiting the ARP2/3 complex.
Although not expressed as highly as in the brain, TSPAN7 expression seemed to be induced during monocyte to dendritic cell differentiation in vitro and correlates well with the appearance of actin-rich dendrites . As actin nucleation is a fast, dynamic and reversible process, we could imagine that TSPAN7, due to its low expression in MDDCs, would be a good candidate to modulate in order to rapidly affect the status of actin nucleation.
Finally, while studying the function of TSPAN7 in the control of actin nucleation, we discovered the existence in DCs of two opposing forces at the plasma membrane. Actin nucleation seems to be a protrusive force which functions to counterbalance an actomyosin-driven contractile force (Figure 2). The balance between those two forces appears to be key for the distribution of actin-rich membrane protrusions versus actin-devoid blebs and to control the maintenance of HIV-1 at the surface of the cell versus being internalized through macropinocytic events (Figure 2).
In conclusion, a better understanding of the molecular mechanisms linking TSPAN7 and actin nucleation is key and could also provide us with new ways to control and modulate actin nucleation function in a more cell-specific and fine-tuned way. This would be important not only for controlling HIV-1 transfer from DCs to T lymphocytes but also for more fundamental cell biology processes such as membrane protrusions and a certain type of endocytic pathways.
The Author declares that there are no competing interests associated with this manuscript.
Present affiliation: Laboratory of inflammatory Responses and Transcriptomic Networks in Diseases, Institut Imagine, INSERM UMR 1163, Paris Descartes-Sorbonne Paris Cité University, Paris, France.