HIV-1 ENTRY INTO T-CELLS IS NOT DEPENDENT ON CD4 AND CCR5 LOCALIZATION TO SPHINGOLIPID-ENRICHED, DETERGENT-RESISTANT, RAFT MEMBRANE DOMAINS.

The contribution of raft domains to human immunodeficiency virus (HIV) 1 entry was assessed. In particular, we asked whether the CD4 and CCR5 HIV-1 receptors need to associate with sphingolipid-enriched, detergent-resistant membrane domains (rafts) to allow viral entry into primary and T-cell lines. Based on Triton X-100 solubilization and confocal microscopy, CD4 was shown to distribute partially to rafts. In contrast, CCR5 did not associate with rafts and localized in nonraft plasma membrane domains. HIV-1-receptor partitioning remained unchanged upon viral adsorption, suggesting that viral entry probably takes place outside rafts. To directly investigate this possibility, we targeted CD4 to nonraft domains of the membrane by preventing CD4 palmitoylation and interaction with p56(lck). Directed mutagenesis of both targeting signals significantly prevented association of CD4 with rafts, but did not suppress the HIV-1 receptor function of CD4. Collectively, these results strongly suggest that the presence of HIV-1 receptors in rafts is not required for viral infection. We show, however, that depleting plasma membrane cholesterol inhibits HIV-1 entry. We therefore propose that cholesterol modulates the HIV-1 entry process independently of its ability to promote raft formation.


INTRODUCTION
Numerous studies dealing with biological membrane organization and composition have emphasized the non-random distribution of lipids and proteins into distinct membrane domains (1). Domains composed of cholesterol and saturated lipids, e.g. sphingolipids, or rafts have recently been shown to support a wide range of cellular events, including signal transduction, sorting and cellular trafficking of proteins and lipids, as well as pathogen entry into cells (2). Non-ionic detergent insolubility of these domains at 4°C was found to result from tight packing of cholesterol and sphingolipids in a liquid-ordered state (3). This property allows recovery of rafts as low-density, floating membranes by gradient centrifugation and makes it possible to characterize raft lipids and proteins (4).
Entry of human immunodeficiency virus type 1 (HIV-1) into host cells relies primarily upon interaction of the viral glycoprotein envelope (Env) gp120 subunit with cell surface CD4 (5). Conformational changes of gp120 upon CD4 binding trigger interactions of the Env with HIV-1 coreceptors CCR5 or CXCR4 (6). Subsequently, this binding to coreceptors exposes the Env gp41 trans-membrane subunit and promotes fusion of viral and cellular membranes (7). It has been proposed that oligomeric assembly of Env proteins (8,9) facilitates the recruitment of viral receptor (CD4 and coreceptors) molecules and ultimately HIV-1 entry (10). Since a correlation between the cell surface density of HIV-1 receptors and efficiency of infection has been emphasized (10), the clustering of CD4 and the coreceptor in delimited plasma membrane domains would be expected to favor HIV-1 entry.
The CD4 antigen is among the few membrane-spanning proteins found to partition into raft domains enriched in sphingolipids (e.g. the GM1 ganglioside, a prototypic marker of these domains) (11)(12)(13). To establish the contribution of rafts to HIV-1 entry, inhibition of glycosphingolipid synthesis (14,15) and depletion of cell plasma membrane cholesterol have been investigated (16)(17)(18). The inhibition of HIV-1 infection was shown in both instances and believed to result from disruption of raft integrity. However, considering that inhibition of glycosphingolipids synthesis is not detrimental to raft domain formation (19), and that cholesterol is distributed throughout plasma membranes (20)(21)(22), inhibition of HIV-1 entry following these two means of lipid perturbation (14)(15)(16)(17) cannot be attributed solely to disruption of rafts.
In the present work, a requirement for an association of CCR5 and CD4 with rafts to support HIV-1 entry was specifically addressed. A large body of evidence points to fatty acylation (specially, post-translational palmitoylation of cysteine residues) as a critical signal for targeting several inner leaflet signaling proteins to rafts (23) and for a few raft-seeking transmembrane proteins (24)(25)(26). It is conceivable that C-terminal palmitoylation of CD4 (27) might account for localization of this protein to rafts. This prompted us to investigate whether interference with CD4 palmitoylation would alter the distribution of the CD4 receptor to rafts.
Such a non-invasive approach led us to further investigate if CD4 maintains its HIV-1 receptor function when localized outside rafts.
Here we show that palmitoylation of CD4 and its interaction with the tyrosine kinase p56 Lck are important for the distribution of CD4 to rafts. In contrast to CD4, we present evidence that the CCR5 coreceptor preferentially partitions to non-raft domains, despite its palmitoylation on three C-terminal cysteine residues (28,29), both in T-cell lines and primary T cells. Importantly, when predominantly redistributed to non-raft domains, CD4 still displays full receptor function for monocytotropic (R5)-HIV-1 strains. Together, our results indicate that CCR5-dependent HIV-1 infection does not depend upon the presence of CD4 and CCR5 receptors in rafts. Furthermore, we show that depleting plasma membrane cholesterol in target cells inhibits viral entry, suggesting that cholesterol-dependent membrane properties other than rafts formation come into play to promote efficient HIV-1 infection. However, we confirmed others' data (31) that in 10% FCS growth medium, the sterol contained in plasma membranes of these cells is mainly cholesterol, as assessed by thin layer chromatography (TLC) (our data not shown). Human peripheral blood mononuclear cells (PBMC) were isolated from healthy donors using Ficoll-Hypaque (Amersham Pharmacia Biotech, Uppsala, Sweden) density gradient centrifugation. For lymphocyte experiments, freshly prepared PBMC at 2-3 X 10 6 cells per ml were cultured in growth medium containing phytohemagglutinin (PHA (Sigma)) at 1 mg/ml for three days and afterwards growth medium was supplemented with recombinant interleukin 2 (IL-2) at 150 units/ml (Chiron) for two weeks. Cells were mostly CD14 negative and more than 70% CD4 positive, as assessed by FACscan analysis (results not shown).  (31,32). Protein concentration was determined with the bicinchoninic acid protein assay reagent (Pierce) with bovine serum albumin as a standard. Total lipids were extracted according to Bligh and Dyer (33). The amount of phospholipid was estimated by a phosphate assay after total digestion in the presence of perchloric acid (34). Cholesterol was quantified using a colorimetric method based on the oxidation of the hydroxyl group at the carbon atom 3 in the β-position (Roche, Molecular Biochemicals).

HIV-1 infection, Luciferase assays and real-time quantitative PCR (Taq-Man). HIV-1
particles were produced and their concentration estimated by measuring both the HIV-1 Gagp24 antigen (ELISA detection kit (NEN TM Life Science Products)) and the βgalactosidase activity in infected Hela CD4 LTRLacZ indicator cells, as previously described (29). Briefly, HEK-293T cells were transiently transfected either with the WT R5-HIV-1 YU2 proviral DNA or the R5-HIV-1 JR-CSF-luc , which carries the firefly luciferase (luc) reporter gene instead of nef (a gift from Dr.V. Planelles, University of Utah School of Medicine, UT), or were co-transfected with an HIV-1 pNL4-3-luc envelope-deficient (env(-)), proviral DNA containing luc instead of nef and the R5-HIV-1-Ba-L Env-expressing vector. Luciferase activity was measured at the times indicated in the figure legends using a luminometer as described (29). Infection of T-cells with the indicated HIV-1 Gagp24 quantities is detailed in labeling were performed as previously described (29). Radiolabeled cells were lysed in TKM buffer containing 1% Brij 96 (w/v) and CD4 was immunoprecipitated using OKT4 antibody.
Where indicated, cells were incubated in the presence of 2-bromopalmitate (2BP) (100 µM) for 16 hours throughout the labeling time.
Immunofluorescence Microscopy. A3.01 cells (1 X 10 5 ) were treated to aggregate GM1 as described (26). Briefly, phycoerythrin (PE)-conjugated CTx (Sigma Chemical Co.) was applied for 30 min at 4°C, followed after washing by an anti-CTx monoclonal antibody overlap extension with T7, Sp6 and two internal primers containing the mutation, as described previously (29). The forward and reverse primers were, respectively : Virus stock production and infection were as described in (37)  Lck-or Palm-Lck-). Viral titers were estimated as described above.
Fluorescence-based assay of syncitia induced by HIV-1 envelope. The fluorescence-based assay of syncytia formation induced by HIV-1 Env (FLASH method) is described elsewhere (38). Briefly, BHK cells were infected for eight hours using a Semliki forest virus-based strategy permitting surface expression of the R5-HIV-1-BX08 Env. Infected BHK cells were labeled with SNARF®-1 and A2.01R5 target cells expressing various CD4 variants (WT,

Role of plasma membrane cholesterol in HIV-1 entry
The

CD4 but not CCR5 localizes in rafts of T lymphocytes
The insolubility of some lipids and proteins in cold non-ionic detergent Triton X-100 (TX-100) correlates well with their partitioning into ordered phases of biological membranes (3,41). This makes it possible to recover detergent-resistant membrane components as lowdensity material after centrifugation to equilibrium on a sucrose density gradient.
Accordingly, the partitioning of CD4 and CCR5 was investigated in primary T (Figure 2A) and A3.01R5 cells ( Figure 2B). Low-density fractions of primary T cells (Figure 2A (42)). We found that both the tyrosine kinase p56 Lck , which is known to associate with the cytoplasmic leaflet (Figure 2A, Lck), and the exoplasmic CD55 leaflet-glycophosphatidylinositol-anchored protein float with insoluble material at the top of the gradient (Figure 2A, fractions 3-5), as reported by others (43,44).
In A3.01R5 cells ( Figure 2B, upper panel), CCR5 partitioning was similar to that of the endogenous CCR5 protein in primary T cells (Figure 2A), i.e. CCR5 remained fully solubilized in 1% TX-100 in both cell types. Moreover, in both A3.01R5 ( Figure 2B) and A2.01R5 cells transduced with the wild-type (WT) CD4 molecule (Figures 4 and 5), the relative distribution of CD4 in both low-and high-density fractions was similar to that observed in primary T cells (Figure 2A). One may thus conclude that the behavior of CCR5 and CD4 after stable ectopic expression in T-cell lines is similar to that of their endogenous counterparts in primary T lymphocytes.
The above results agree with previously reported partitioning of CD4 to sphingolipidenriched, detergent-resistant domains (11)(12)(13). This contrasts strongly with the distribution pattern of CCR5, which was recovered in 1% TX-100 soluble domains where GM1 remains only marginally localized. Sphingolipids (i.e., glycosphingolipids and sphingomyelin) differ from the other class of membrane lipids, glycerolipids, in that they contain long, highly saturated fatty acyl chains. In appropriate admixture with cholesterol, sphingolipids form detergent-resistant domains in the plasma membrane (i. e. rafts) (46) that are sensitive to cholesterol depletion (22).
The nature of CD4-and GM1-enriched TX-100-resistant domains was further characterized following extraction of cholesterol-depleted A3.01R5 cells. As shown in Figure   2C,  Figure 2D). Moreover, quantification of CCR5 in low-density fractions expressed as a function of TX-100 concentration, revealed a distribution ( Figure 2D) identical to that of CD46 ( Figures 2B and D), which is representative of TX-100-soluble membrane proteins (45). These observations further support our previous conclusion that CCR5 is excluded from rafts and highlight that only a very small amount of the receptor is found in membrane domains of lesser detergent resistance.
To rule out that CCR5 might transiently distribute into CD4-containing rafts, we forced stabilization of these domains into visible patches by specific clustering of GM1 by cross-linking with CTx. The coalescence of Influenza Hemagglutinin (HA)-and GM1containing domains following GM1-CTx cross-linking was taken to reflect the propensity of HA to localize to GM1-enriched rafts (26). Prior to CTx cross-linking, both CD4 and CCR5 receptors were found evenly distributed over the plasma membrane ( Figures 3A and B, left panels). Cross-linking of GM1-CTx complexes with a rhodamine-labeled CTx antibody induced redistribution of CD4 into patches ( Figure 3A, second-panel) that overlapped GM1 patches ( Figure 3A, GM1 and overlay panels). In marked contrast, CCR5 distribution remained unchanged and did not significantly overlap with GM1-patches following CTxdependent cross-linking ( Figure 3B, GM1 and overlay panels). This clearly shows that CCR5 is mainly excluded from GM1-enriched domains and behaves differently from CD4 in this respect.
As concerns the requirement of HIV-1 entry on rafts, several possibilities remain. It was proposed that HIV-1 binding to CD4 might occur within rafts and that, subsequently, CD4-Env complexes redistribute to coreceptor-containing non-raft domains (52 (16)(17)(18). To test these possibilities, we investigated whether receptor cross-linking induced by R5-HIV-1 Env caused membrane redistribution of either CD4 or CCR5. A2.01R5 CD4 WT cells, exposed to R5-HIV-1-Ba-L virions for 45 or 90 min, were lysed in 1% TX-100, and fractioned on a sucrose gradient. As shown in Figure 4 adsorption of virions onto these cells did not modify CCR5 or CD4 distribution compared to uninfected control cells This non-raft distribution of CD4 in cells treated with 2BP indicated that palmitoylation is a key factor in regulating the distribution of this receptor within the cell membrane. Interestingly, 2BP also inhibited p56 Lck -palmitoylation and prevented its localization to rafts (59). CD4 interacts non-covalently with p56 Lck through its C-terminal cysteine residues (Cys 445 and Cys 447) (60) and it was proposed that this association occurs early in the secretory pathway, allowing both proteins to reach the plasma membrane together (61). Aside from blocking CD4 palmitoylation, 2BP treatment may indirectly affect CD4 distribution by preventing p56 Lck localization to rafts.
To explore these possibilities, CD4 cDNAs carrying mutations in cysteine residues involved in CD4-palmitoylation (CD4 Palm -), in its interaction with p56 Lck (CD4 Lck -) or both (CD4 Palm -Lck -), were expressed in CD4-negative A2.01R5 T-cells. Membrane distributions of these molecules were compared to that in CD4 WT cells following 1% TX-100 solubilization ( Figure 6A). Similar to the effect of 2BP ( Figure 5), direct prevention of CD4 palmitoylation by mutation (CD4 Palm -, Figure 5B, left panel, lane 2) shifted CD4 to high-density fractions ( Figure 6A). This strongly suggests that palmitoylation per se directs membrane localization of CD4 without affecting its biosynthesis ( Figure 5B, right panel Met labeling) and cell surface expression (data not shown). Similar results obtained for the CD4 Lckmutant show that, in addition, CD4/p56 Lck interaction also directs CD4 to rafts ( Figure 6A, (CD4 Lck -)). In keeping with these observations, the double mutation (CD4 Palm -Lck -) almost abolished CD4 association with low-density fractions ( Figure 6A and B, 6±2 % remaining in the low-density fractions). The distribution of the WT and mutant CD4 proteins in 1% TX-100-soluble and resistant fractions are compared in Figure 6B. The lower recovery of the CD4 Palm -Lckmutant protein in TX-100-resistant low-density fractions (3)(4)(5) is consistent with an additive effect of palmitoylation and lack of interaction with p56 Lck in the targeting of CD4 to rafts. In agreement with previous reports (61), the distribution of p56 Lck at the plasma membrane in raft and non-raft domains ( Figure 6C, D and S fractions, respectively) is independent of its interaction with CD4 ( Figure 6D). The reduced cell surface expression of the CD4 protein in the Lckand Palm --Lckcells, compared to its expression in CD4 WT or Palmcells (data not shown), is in keeping with CD4 down-regulation observed in cells lacking p56 Lck (62).

CD4 redistribution into non-raft domains does not impair CCR5-dependent HIV-1 entry
The consequences of expression of the CD4 double mutant (Palm -Lck -) on R5-HIV-1 entry were explored. We first used an HIV-Env glycoprotein-mediated cell fusion assay that is independent of the post-entry steps of the viral cycle and allows accurate quantification of fusion events (38). This method relies on syncytia formation between cells expressing the HIV-1 R5 BX08 Env glycoprotein and CD4 -stabilized A2.01R5 target cells ( Figure 7A).
CD4-WT and CD4-Palm -Lckcells showed the same capacity to fuse with R5-Env expressing cells. Cell fusion was prevented by blocking with either monoclonal antibody Q4120 that recognizes the gp120-binding domain of CD4 (63), or TAK779 that binds to CCR5 (64).
Another alternative assay was used to verify the apparent independence of the HIV-1- Although CCR5 appears to be present in detergent-soluble membrane domains, we envisaged that HIV-1 receptor engagement by viral Env might recruit CCR5 to rafts, to which CD4 and GM1 localize. This explanation was recently advanced to account for redistribution of CXCR4 by soluble X4-HIV-1 Env (16,18). However, the finding by Kozak et al. that CXCR4 coreceptor distribution into detergent-soluble domains is not modified by virion adsorption (52) challenges previous observations (16,18). HIV-1 entry is expected to be a cooperative process requiring the assembly of HIV-1 virions with CD4 and coreceptors. We therefore investigated whether R5-HIV-1 virion binding, which facilitates initial CD4 crosslinking unlike soluble monomeric viral Env, would induce CCR5 redistribution and colocalization with CD4 in T-cell rafts. In flotation studies, bridging of viral receptors by R5-HIV-1 virions adsorbed on T-cells did not shift CCR5 to raft fractions. Taken together, our data strongly suggest that recruitment of CCR5, to rafts is not a prerequisite for virus entry.
One could speculate that CCR5-enriched detergent-soluble membrane domains may form a ring around a raft area where virus could interact with the "raft "fraction of CD4. This hypothetical situation would be similar to the membrane organization described in T lymphocytes interacting with antigen-presenting cells (67), where the T-cell receptor interacts with an MHC II-bound cognate peptide not included in rafts. The role played by CD4 residing in rafts on HIV-1 infection was explored expressly to test this hypothesis.

Redistribution of CD4 outside raft domains does not affect HIV-1 infection
In contrast to CCR5, CD4 was equally distributed in both TX-100-resistant and TX-100soluble membranes. The non-raft redistribution of the non-palmitoylated CD4 protein strongly suggests that palmitoylation determines the targeting of this receptor to rafts. The role of fatty acylation in targeting membrane proteins to rafts (68,69) has been established for a number of integral palmitoylated proteins, such as LAT and CD8αβ (24,25 (73). intermediates (73,76). Indeed, fusion depends very much on lipids of spontaneous negative curvature (i.e. unsaturated phosphatidylethanolamine, cholesterol) to promote membrane bending (77,78). Likewise, synthetic peptides mimicking the HIV-1 gp41 fusion domain were found to promote negative curvature that facilitates formation of stalk intermediates (79). The fusogenic activity of these peptides was stimulated by insertion of cholesterol into a non-raft environment (80). Therefore, the observed interference of the lack of cholesterol with HIV-1 entry might reflect the altered capacity of plasma membranes to undergo viral-induced fusion.

MBCD
Furthermore, the role played by cellular cholesterol in HIV-1 entry may be related to its capacity to modulate the activity of membrane proteins. This has been shown for several G-protein coupled receptors (81,82) and also applies to the HIV-1 coreceptors, CCR5 and CXCR4. Effectively, depletion of cell membrane cholesterol has been shown to modify binding of CCR5 and CXCR4 to their natural chemokine ligands, MIP1-β and SDF-1α, respectively (65,74). Nonetheless, whether diminished HIV-1 infection in cholesteroldepleted cells relates to modulation of CCR5 or CXCR4 still remains speculative.
In conclusion, our findings exclude a significant participation of membrane rafts in HIV-1 binding to host cells and point to an important role for cholesterol in the mechanisms of viral entry into primary CD4 T lymphocytes. Further study, using real time and quantitative approaches, are needed to address the particularly intriguing modulator effect of plasma membrane cholesterol on the HIV-1 entry process.