Thus, after more than 20 years, three different molecules have proven to be important for HTLV-1 entry. It seems clear that HSPG, as for nearly all other viruses, is involved in binding but not in fusion. The properties displayed by both GLUT1 and NRP-1, i.e. binding to the HTLV-1 SU and modulation of infection, are consistent with those expected for a receptor but, as mentioned above, many questions on their exact function remain unsolved. In particular, it is not clear whether GLUT1 and NRP-1 cooperate to promote HTLV-1 entry. We will review very recent data that have started to clarify the respective roles of HSPG, NRP-1 and GLUT1 in HTLV-1 infection.
GLUT1 and NRP-1 bind distinct regions of the HTLV-1 SU
As was hypothesized previously, there is now definitive evidence that GLUT1 and NRP-1 bind to distinct regions of SU. Kim et al.
determined that the minimal RBD is located within the first 183 amino terminal residues of the HTLV-1 SU [52
]. Mutation of the tyrosine residue 114 completely abolished the binding of H1-RBD, suggesting it plays a critical role. However, the results of binding studies using RBD should be interpreted with caution, as they might not truly represent what occurs with the full length SU or the SU/TM complex in the native Env. Several lines of evidence suggest that this may be the case. For example, over expression of GLUT1 strongly increased binding of the H1-RBD [2
], but has little impact on the binding of full-length HTLV-1 SU [57
]. It is thus possible that shortening the SU modifies its quaternary structure and mimics a conformational change resulting in an "active" form capable of binding directly to GLUT1, possibly by exposing the tyrosine residue 114 [52
]. It is tempting to speculate that the conformational change allowing the association to GLUT1 might be dependent on the initial SU binding to NRP-1. This would be analogous to what occurs for HIV, where SU (gp120) binding to CD4 induces a conformational change that allows binding to the co receptors, which facilitates fusion.
The HSPG-interacting SU region is located outside the GLUT1 and NRP-1 binding sites, and thus should not decrease its affinity to the other two molecules; this interpretation is consistent with the data showing that increasing cell surface levels of HSPG increases SU binding to receptor-expressing cells [6
]. Blocking the ability of SU to interact with HSPG, either by enzymatic removal or incubation with a peptide mimicking the HSPG-binding domain of VEGFA165
(exon-7 peptide), decreased the DC-mediated transfer of HTLV-1 to CD4+ T cells by 50%. Moreover, blocking both HSPG and NRP-1 interactions decreased HTLV-1 infection by 80% [4
]. Thus, HSPG are important for the efficiency of HTLV-1 infection, likely by increasing and stabilizing the initial SU binding to NRP-1, a phenomenon already described for gp120 and CD4 [76
In 2009, Lambert et al.
found that the binding of the HTLV-1 SU to NRP-1 involved two distinct types of interaction: an indirect interaction mediated by HSPG and a direct interaction mediated by the amino-acids 90-94 of the SU [4
]. Moreover, they identified a region of SU (aa 90-94) that is homologous in both sequence and function to the exon 8 domain of the NRP-1 ligand VEGFA165
]. This demonstrates that the SU binds to NRP-1 using molecular mimicry of VEGFA165
. Interestingly, the aa 90-94 motif is conserved in the SU proteins of HTLV-2, HTLV-3 and related Simian T-Lymphotropic Viruses but not other retroviruses [4
] and is the target of neutralizing antibodies [78
]. Strikingly, Kim et al.
showed that mutating either aa 94 (R) or 114 (Y), which inhibits direct interaction with NRP-1 or GLUT1 respectively, is equally efficient in preventing interference due to overexpression of the H1-RBD in target cells [52
]. This reinforces the notion that the direct interaction of SU with both NRP-1 and GLUT1 is required for HTLV-1 entry.
Towards a hypothetical model of the HTLV-1 receptor
In light of these data, it is possible to draw a model for HTLV-1 entry based on an analogy with HIV CD4/CXCR4 or CCR5 receptor complexes, and what is known about interactions of NRP-1 with its ligand VEGFA165
, HSPG, and receptor (Figure ). During the initial binding phase, the SU would interact with HSPG via its C-terminal portion, allowing its attachment to the cell surface (Figure , step 1). This would enhance the local concentration of SU, and thus increase the probability of interactions with NRP-1. This notion is consistent with observations that removal of cell surface HSPG dramatically reduces the level of binding of the SU or virions to CD4+
T cells [6
], the titer of pseudotyped virus, and the infection of CD4+
T cells and DC. HSPG/SU complexes would then interact with the b domain of NRP-1, through both HSPG/NRP-1 interactions and direct SU/NRP-1 binding mediated by the 90-94 region of the SU (step 2). This binding to NRP-1 could trigger a conformational change within the SU, exposing the region necessary for GLUT1 binding including the critical Y114. Finally, the interaction of the SU with GLUT1 could trigger the fusion process necessary to HTLV-1 entry (step 3). In this model, NRP1, with the help of HSPG, is the primary binding receptor, while GLUT1 is the fusion receptor. This is consistent with the formation of NRP-1/Env/GLUT1 complexes and with the fact that GLUT1 over expression does not increase the level of HTLV-1 SU binding. This model applies to both primary CD4+
T cells and DC [4
]. Recent data showing that HTLV-1 entry into primary astrocytes requires both NRP-1 and GLUT1 [3
] suggest that this view is true for these cells as well.
Figure 1 Hypothetical model of the HTLV receptor complex and HTLV entry: 1- HSPG interaction with the SU allow the initial attachment and concentration of the virions at the cell surface. 2- HSPG facilitates the recruitment of the SU to NRP-1. 3- Using molecular (more ...)
Whether this model is relevant for HTLV-1 entry in other types of cells remains to be investigated. A recent paper from Alkhatib's group showed that GLUT1, but not NRP-1, was involved in HTLV-1 entry into Hela cells [3
]. Moreover, the same group was the first to show that HTLV-1 can efficiently enter into U87 cells, in which the surface level of GLUT1 is very low [56
]. Low level of GLUT1 at the cell surface could be sufficient to promote fusion if the density of surface NRP1 is high, which is the case of the U87 cell line [3
]. Indeed, it was shown that T cells expressing very low levels of CCR5 but high levels of CD4 were still competent for HIV envelope-mediated membrane fusion [79
]. Alternatively, NRP-1 could function on certain cell types as both the binding and fusion receptor, as suggested for CHO cells [3
]. Finally, one cannot exclude the possibility that, like HIV, HTLV-1 uses different fusion receptors on different cell types and other unidentified molecules could serve as fusion receptors for HTLV-1.
Other evidence supporting a NRP-1/GLUT1- complex comes from reports that a small intracytoplasmic protein called GLUT1CBP or PSD-95/Dlg/ZO-1 can bind both NRP-1 and GLUT1 via its PDZ domain [80
]. Moreover, syndecans, the main members of the HSPG family, also contain a PDZ-binding motif at their C-terminus of their cytoplasmic domain [82
]. Deletion of the PDZ-binding motif of GLUT1 was shown to preclude GLUT1 clustering at the plasma membrane [83
]. GLUT1CBP could therefore act as a "bridge" to stabilize the entire tripartite receptor complex. It could also provide cues for cytoskeletal polarisation, since it can associate with several cytoskeletal proteins [80
] (Figure , step 3). Altogether, this model not only reconciles nearly all of the recent data obtained on the HTLV-1 receptor but also appears to explain the in vivo
tropism of the virus better than the GLUT1-only hypothesis.