CD4 Receptor Localized to Non-raft Membrane Microdomains Supports HIV-1 Entry

Despite the preferential localization of CD4 to lipid rafts, the significance and role of these microdomains in HIV-1 entry is still controversial. The possibility that CD4, when localized to non-raft domains, might be able to support virus entry cannot be excluded. Because disintegration of rafts by extraction of cellular cholesterol with methyl-β-cyclodextrin suffers from various adverse effects, we investigated molecular determinants controlling raft localization of the CD4 receptor. Extensive mutagenesis of the receptor showed that a raft-localizing marker, consisting of a short sequence of positively charged amino acid residues, RHRRR, was present in the membrane-proximal cytoplasmic domain of CD4. Substitution of the RHRRR sequence with alanine residues abolished raft localization of the CD4 mutant, RA5, as determined biochemically using solubilization in nonionic detergents and by confocal microscopy. The possible inhibitory effect of the introduced mutations on the adjacent CVRC palmitoylation site was ruled out because wild type (wt) CD4 and RA5, but not a palmitoylation-deficient mutant, were efficiently palmitoylated. Nonetheless, the RA5 mutant supported productive virus entry to levels equivalent to that of wild type (wt) CD4. Sucrose gradient analysis of Triton X-100 virus lysates showed that Gag and envelope gp120 proteins accumulated in low buoyant, high-density fractions. This pattern was changed after virus incubation with cells. Whereas Gag proteins localized to lipid rafts in cells expressing wt CD4 and RA5, gp120 accumulated in rafts in cells expressing wt CD4 but not RA5. We propose that raft localization of CD4 is not required for virus entry, however, post-binding fusion/entry steps may require lipid raft assembly.

The interaction of viral glycoprotein envelope gp120 with CD4 (1, 2) induces conformational changes in gp120 that enable it to interact with coreceptors (3)(4)(5). In addition, gp120-CD4 interaction stimulates receptor phosphorylation on serine residues (6) that may lead to separation of CD4 from the noncovalently interacting tyrosine kinase p56 Lck and subsequent receptor endocytosis (7). Activation of multiple signaling pathways stimulated by the gp120-CD4 interaction (8 -12) may affect various stages of virus replication.
Membrane microdomain distribution of HIV-1 coreceptors CXCR4 and CCR5 remains a subject of debate. Although conformation and function of these coreceptors depend on membrane cholesterol (13)(14)(15) and the interaction with glycosphingolipids (16) during virus entry (17), HIV-1 coreceptors may localize to microdomains different from that occupied by CD4 (18,19). Thus, activation of the entry process through engagement of CD4 and one of the coreceptors, would require a mechanism allowing for the apposition of the entry receptors in the same membrane environment. Consequently, it has been suggested that coreceptors may migrate into rafts after HIV-1 gp120-induced clustering (16,20). This may result in destabilization of raft membranes required for fusion and entry of virus (21). Alternatively, the CD4-HIV-1 complexes may exit lipid rafts for the interaction with coreceptors localized in a non-raft, fusion-competent environment. Thus, in both models the role of lipid rafts would be limited to initial interaction with CD4 localized in these domains. However, the subsequent steps leading to virus fusion and entry would differ.
Because CD4 almost totally localizes to lipid rafts, we could not exclude the possibility that CD4, if present outside these microdomains, could also support virus fusion and entry. This possibility was emphasized by the observation that fusion of alphaviruses depended on the presence of cholesterol and sphingolipids rather than intact lipid rafts in the target membrane (22). Methods known to disintegrate rafts rely on the extraction of cholesterol with methyl-␤-cyclodextrin and suffer from numerous problems including toxicity, indiscriminate extraction of cholesterol from cell membranes, and various unexpected side effects. Thus, in this study, the significance of raft localization of CD4 in virus entry was addressed by using an approach based on the analysis of molecular determinant(s) controlling localization of this receptor to lipid microdomains without perturbation of cellular membranes.
Extensive mutagenesis of CD4 revealed that neither extracellular nor transmembrane domains of the receptor play a significant role in raft association. We have, however, identified a novel raft-localizing marker in the membrane-proximal cytoplasmic domain of CD4. When this positively charged RHRRR motif was substituted with alanine residues, CD4 was redirected to non-raft membranes despite its intact palmitoylation and the presence of the Lck binding motif. However, similar to wild type receptor, this CD4 mutant efficiently supported HIV-1 entry, suggesting that raft localization of CD4 is not required for productive virus entry. Based on our findings, we propose a new model of CD4-mediated virus fusion in nonraft membrane regions.
Cell Cultures-Human PM1 T cells and BJAB B cells were maintained in RPMI 1640 medium with L-glutamine (Invitrogen) supplemented with 10% fetal calf serum (Gemini Bio-Products, CA) and gentamicin (50 g/ml). Human 293T cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal calf serum, L-glutamine, and gentamicin.
Treatment with 3Ј Sialyllactose-PM1 cells (5 ϫ 10 6 ) were washed twice with cold PBS and resuspended in 1 ml of 1% BSA in PBS (BSA/PBS). 100 g of 3Ј-sialyllactose was added and the cells were incubated for 1 h at 37°C. The cells were collected by centrifugation, washed twice with ice-cold BSA/PBS, and were either processed for confocal microscopy or were infected with NL env(Ϫ)luc(ϩ) virus pseudotyped with HXB2 envelope and analyzed for virus entry.
Construction of CD4-GFP and CD4-GFP Mutants-The full-length cDNA encoding CD4 in pCG.CD4 vector was obtained from J. Skowronski (Cold Spring Harbor Laboratory). The CD4 coding region was amplified by PCR with the 5Ј primer designed to contain a EcoRI restriction site (underlined) followed by Kozak sequence and the first 24 nucleotides of CD4. The sense orientation of the primer is presented: GCTTCGAATTCTCGCCACCATGAACCGGGGAGTCCCTTTTAGG. The 3Ј primer (sense orientation) contained the coding sequence from the last 21 nucleotides of CD4 without a stop codon and an XmaI site (underlined): CAGAAGACATGTAGCCCCATTGCCCGGGATCCA. The PCR product containing full-length CD4 cDNA was digested with EcoRI and XmaI and cloned into respective sites of pEGFP-N1 (Clontech), creating the CD4-GFP construct. The coding region of the construct was confirmed by sequence analysis; functional expression and membrane localization were confirmed by transfection into 293T or BJAB cells and confocal microscopy. The resulting CD4-GFP fusion protein contains an 8-amino acid linker, ARDPPVAT, which connects the carboxyl terminus of CD4 with the first methionine residue of the GFP protein. The CD4-GFP mutants were constructed using the QuikChange site-directed mutagenesis kit (Stratagene) according to a protocol provided by the manufacturer. Primers used in site-directed mutagenesis reactions are available upon request. CD4 amino acid numbering is according to Swiss Prot P01730 and includes the 25-amino acid signaling peptide.
Metabolic Labeling with [1-14 C]Palmitic Acid-293T cells were grown in 60-mm dishes and transfected with wt CD4-GFP, RA5, or non-palmitoylated mutant CD4 C419, 422A. The next day, the cells were washed and incubated for 3 h in Dulbecco's modified Eagle's medium with 2% BSA supplemented with [1-14 C]palmitic acid (80 Ci/ ml; Amersham Biosciences). Radiolabeled cells were then washed three times in ice-cold PBS and lysed in 1% Triton X-100 TNE lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA) supplemented with protease inhibitor mixture (Complete; Roche Diagnostics). CD4 was immunoprecipitated using anti-GFP 3E6 antibody and protein A/G plus-agarose (Santa Cruz Biotechnology). Washed immunoprecipitates were resolved by SDS-PAGE, transferred to nitrocellulose filter, and exposed to x-ray film (Kodak).
Virus Preparation and Single-cycle Infectivity Assay-A plasmid DNA encoding envelope from a T-tropic HIV-1 clone HXB2 (15 g) was cotransfected with pNL4-3 env(Ϫ)luc(ϩ) (10 g) into 293T cells using SuperFect (Qiagen). Forty-eight hours after transfection, the culture supernatants were collected, filtered through 0.45-m filters, and concentrated by ultracentrifugation through a cushion of 20% sucrose in PBS. The pelleted virus was resuspended in PBS with 0.1% BSA, aliquoted, and stored frozen at Ϫ80°C. The virus titer was determined by the reverse transcriptase (RT) activity assay (23).
293T cells were plated in triplicate in 6-or 12-well Biocoat plates coated with fibronectin (BD Biosciences) 1 day prior to transfection with different CD4-GFP constructs. Twenty-four hours post-transfection, the cells were incubated for 1 h with a standardized amount of pseudotyped viruses (5 cpm of RT/cell), washed, and the cultures were propagated for 48 h in Dulbecco's modified Eagle's medium with 10% fetal calf serum. After incubation, the cells were washed and lysed in 100 l of lysis buffer (Promega). Luciferase activity was determined after adding 100 l of luciferase assay reagent (Promega) to 20 l of precleared cell lysate and counting the resultant scintillation for 10 s using a Monolight 2010 luminometer (Analytical Luminescence Laboratory). Background luciferase activity, determined by infection of cells with NL4-3 env(Ϫ)luc(ϩ) virions produced in the absence of envelope protein was negligible (150 relative light units for 10 s counting).
Biochemical Isolation of Lipid Rafts-Forty hours after transfection of BJAB cells (3-5 ϫ 10 7 ) with different CD4-GFP constructs, the cells were washed three times with ice-cold PBS and lysed on ice for 20 min in 1 ml of either 0.5% Triton X-100 or 0.5% W-1 TNE lysis buffer supplemented with protease inhibitor mixture. The cell lysates were homogenized with 10 strokes of a Dounce homogenizer and centrifuged for 5 min at 1,000 ϫ g at 4°C in a microcentrifuge to remove insoluble material and nuclei. The supernatant was mixed with 1 ml of 80% sucrose in lysis buffer, placed at the bottom of ultracentrifuge tubes, and overlaid with 6 ml of 30% and 3 ml of 5% sucrose in TNE lysis buffer. Lysates were ultracentrifuged at 4°C in a SW41 rotor (Beckman) for 16 -18 h at 38,000 rpm. Eleven 1-ml fractions were collected from the top with Auto Densi-Flow (Labconco) and analyzed by Western blotting or stored at Ϫ80°C.
Western Blot Analysis-Aliquots of 40 l of individual sucrose gradient fractions were analyzed by SDS-PAGE on 10% polyacrylamide gels. Proteins were transferred to nitrocellulose membranes and probed with specific antibodies. Bound antibodies were detected with secondary antibodies conjugated with horseradish peroxidase and the Super-Signal West Dura substrate (Pierce). Protein concentration was determined with the bicinchoninic acid protein assay reagent (Pierce).
Raft Aggregation and Confocal Microscopy-A total of 1 ϫ 10 6 CD4positive PM1 T cells or 5 ϫ 10 6 transfected BJAB cells were washed in serum-free RPMI and incubated on ice for 10 min with PBS supplemented with 1% BSA (ELISA grade; BSA/PBS). After centrifugation, the cell pellet was resuspended in 100 l of BSA/PBS and incubated on ice for 30 min with fluorescently labeled monoclonal antibodies at concentrations suggested by the manufacturer (typically 10 g/ml) followed by extensive washing with ice-cold BSA/PBS. To investigate the presence of untagged CD4 in lipid rafts, the cells were first incubated for 30 min on ice with biotinylated CT-B and washed extensively with PBS/BSA. Lipid raft aggregation was initiated by incubation with streptavidin-Alexa Fluor 594 conjugate for 30 min at 37°C. After washing, the cells were incubated on ice for 30 min with anti-CD4-FITC (clone 13B8.2), washed again, and fixed on ice for 15 min with fresh 4% paraformaldehyde in PBS. To analyze localization of CD4-GFP receptor in transiently transfected BJAB cells, the ganglioside GM1-resident microdomains were aggregated with biotin-CT-B/streptavidin-Alexa Fluor 594, as described above or for detection of other surface antigens like CD71 or CD48, rafts were aggregated using biotin-CT-B/anti-biotin BN-34 antibody. After washing and incubation of the cells on ice for 30 min with anti-CD71 or anti-CD48 mouse antibodies labeled with the Zenon One Alexa Fluor 594 mouse IgG1 labeling kit, the cells were washed, fixed, and mounted using ProLong antifade kit. Fluorescently labeled cells were analyzed using Nikon PCM 2000 laser scanning confocal microscope with ϫ60 objective lenses. The images were processed using Photoshop software (Adobe).

CD4 Receptor Glycosylation Is Not Necessary for Association with Lipid
Rafts-It has been reported that glycolipids can function as cofactors for HIV-1 fusion. The interaction between ganglioside GM3 and the CD4 receptor (24,25) and the observation that GM3 is enriched in lipid microdomains (26), suggest that CD4 may localize to lipid rafts as a result of the interaction with GM3. However, treatment of CD4-positive PM1 T cells with GM3-derived 3Ј-sialyllactose, the oligosaccharide moiety of GM3 that interacts with CD4 (24), did not significantly affect the CD4 receptor expression and raft localization (Fig. 1A). Accordingly, productive entry of NL env(Ϫ)luc(ϩ) pseudotyped with the HXB2 envelope was not considerably affected by pretreatment of PM1 T cells with 3Ј-sialyllactose (Fig. 1B). These results suggest that CD4-GM3 interaction is dispensable for raft localization and virus entry.
Further investigation of the putative raft-localizing signals in CD4 was performed using CD4-negative B cells, BJAB, transiently transfected with the CD4-GFP fusion receptor in which GFP was fused in-frame to the carboxyl terminus of the CD4 receptor (Fig. 1C). The resulting GFP-tagged CD4 was expressed on the cell surface (see Figs. 2-5) and supported productive virus entry as efficiently as untagged wt CD4 (Fig. 1D).
N-Glycans present on membrane proteins of non-lymphoid cells have been shown to mediate raft association, possibly through binding to lectin-like molecules residing in rafts (27).
To determine the role of N-glycosylation in CD4 sorting to rafts, N-glycosylation sites in CD4 were eliminated by substituting either Asn-296 or Asn-325 with glutamine residues (N296Q or N325Q, respectively). These CD4-GFP mutants were expressed on the cell membrane and colocalized with lipid raft-resident ganglioside GM1 (Fig. 1E) and were resistant to Triton X-100 extraction to the same degree as wt CD4 (Fig. 1F). In conclu-FIG. 1. Glycosylation of CD4 is not necessary for association with lipid rafts. A, PM1 T cells were treated with 3Ј-sialyllactose (3ЈSL; 100 g/ml) for 1 h at 37°C or left untreated (control). Lipid rafts were aggregated by incubation with biotinylated CT-B and streptavidin-Alexa Fluor 594, stained with anti-CD4-FITC, fixed, and visualized by confocal microscopy. The colocalization of CD4 with the raft marker GM1 is presented (overlay). B, 3Ј-sialyllactose-treated cells and untreated controls were infected for 1 h with NL env(Ϫ)luc(ϩ) virus pseudotyped with the HXB2 envelope (5 cpm of RT/cell). Luciferase activity was measured in cell lysates prepared 48 h post-infection as described under "Experimental Procedures." Uninfected cells (cells only) or untreated cells infected with envelope-defective NL env(Ϫ)luc(ϩ) virus (Env Ϫ Lucϩ) were included. Luciferase activity, presented as relative light units (RLU), is expressed relative to luciferase activity detected in infected untreated control cells (taken as 100%). Standard deviations of three independent experiments are shown. C, cartoon of CD4 outlining the location of GFP and sites of mutations used in this study. Extracellular immunoglobulin-like domains are labeled D1-D4. Mutations, represented by stars, at the sites of N-linked glycosylation are in D3 and D4; mutation of the Lck binding domain in the cytoplasmic tail of CD4 is indicated. The sequence of membrane proximal amino acids (single letter code) with palmitoylated cysteines (underlined) is also shown. D, CD4-GFP fusion receptor supports productive entry of NL env(Ϫ)luc(ϩ) virus pseudotyped with HXB2 envelope as efficiently as untagged wt CD4 (control). Transfected BJAB cells were infected with pseudotyped virus as described in B. E, BJAB cells were transiently transfected with CD4-GFP mutant receptors in which N-glycosylation sites, Asn-296 or Asn-325, were eliminated by substitution with Gln residues (N296Q or N325Q, respectively). Expression of CD4 N296Q or CD4 N325Q mutants and their colocalization with lipid raft GM1 (overlay) was analyzed by confocal microscopy after aggregation of lipid rafts with biotin-CT-B and streptavidin-Alexa Fluor 594 as described under "Experimental Procedures." F, wt CD4-GFP and N-glycosylation mutants, CD4 N296Q and CD4 N325Q, localize to lipid rafts. Postnuclear 0.5% Triton X-100 lysates prepared from transfected BJAB cells were fractionated on sucrose density gradients. Equal volumes (40 l) of individual fractions were resolved by SDS-PAGE, transferred to nitrocellulose, and probed with anti-CD4 (H-370) antibody. Localization of non-raft marker CD71 and raft-resident CD48 is shown. sion, neither association with ganglioside GM3 nor deletion of each N-glycosylation site was sufficient to change raft distribution of CD4.
The Extracellular Domains of CD4 and Raft Association-To test the possibility that CD4 may localize to microdomains through the interaction of its extracellular domain with other raft-anchored proteins, we have created mutants with deletions in different extracellular domains of the receptor. The constructs were transiently transfected into BJAB cells and the localization of CD4-GFP mutants was investigated by confocal microscopy. Both wild type CD4-GFP as well as two-domain mutants, CD4 D1-D2-GFP and CD4 D3-D4-GFP, colocalized extensively with GM1 ( Fig. 2A). In contrast, one-domain, CD4 D4-GFP, or three-domain D1-D2-D4 mutants (d.D3-GFP) were not detectably expressed on the cell surface and did not overlap with GM1, CD48, or CD71 (Fig. 2B). Intracellular expression of these mutants is most likely because of their misfolding and retention by quality control mechanisms (28). Interestingly, removing all four extracellular domains restored surface expression of the truncated CD4 TM-GFP (TM-GFP, Fig. 2A), supporting the idea that raft-localizing determinants may be present in transmembrane or cytoplasmic domains of the receptor.
Because the primary binding site for HIV-1 gp120 is localized in domain D1 of CD4 (29) and the glycosylphosphatidylinositol (GPI)-anchored CD4 D1-D2/Thy-1 chimeric receptor was shown to support virus entry and replication (30), we analyzed whether the CD4 D1-D2-GFP receptor was able to support productive entry of NL env(Ϫ)luc(ϩ) pseudotyped with the HXB2 envelope. Our results show that despite raft localization, the CD4 D1-D2-GFP did not support productive virus entry and replication (Fig. 2, A and C). This may suggest the importance of the remaining domains in virus entry or indicate that two-domain CD4 is not able to execute conformational changes in gp120, which would allow the envelope to interact with coreceptors.
CD4 Transmembrane Domain Is Not Essential for Raft Localization-To analyze the role of the transmembrane (TM) region of CD4 in raft association, we created chimeric molecules in which the TM domain of CD4 was replaced with the corresponding region of non-raft resident transferrin receptor CD71 (CD4/TM CD71) or tyrosine phosphatase CD45 (CD4/TM CD45). The chimeric molecules contained intact extracellular and cytoplasmic domains of CD4. However, replacement of the TM domain of CD4 did not affect partitioning of the chimeric CD4 receptor to lipid rafts when expressed in BJAB cells (Fig.  3A). Furthermore, the CD4 TM mutants colocalized with raftresident GM1 (Fig. 3B). This is in agreement with a recent observation that a CD4 mutant in which the TM domain was replaced with the respective domain from the non-raft partitioning low density lipoprotein receptor colocalized with lipid rafts in HEK-293 cells (31).
CD4 Receptor Palmitoylation or Association with Lck Does Not Affect Raft Partitioning-The above results indicate that the raft localization signals may be present in the cytoplasmic domain of the receptor. There are several features in the cytoplasmic tail of CD4 that may serve as potential raft-localization signals. Raft partitioning of various signaling molecules depends on their modification with saturated fatty acyl chains, which could preferably recruit these molecules into the liquidordered phase of raft microdomains (32). A number of palmitoylated signaling molecules including G proteins (33), Src kinases (34), as well as transmembrane LAT (35) and CD8␤ (36) localize to raft microdomains. Interestingly, palmitoylation of CD4 on Cys-419 and Cys-422 and its association with p56 Lck was shown to target CD4 to lipid rafts (37, 38) or was without effect on raft localization (31). Thus, to further investigate the role of CD4 palmitoylation in raft localization, the cells were transfected with wild type or mutated CD4 in which the palmitoylated, membrane-proximal cysteine residues (39) were mutated to alanines (CD4 C419A,C422A). Sucrose gradient analysis shows that raft distribution of palmitoylated wt CD4 and the non-palmitoylated mutant CD4 C419A,C422A was similar (Fig. 4A). These results were also confirmed by confocal microscopy analysis by showing that CD4 C419A,C422A or CD4 C419S,C422S mutants colocalized with lipid raft GM1 (Fig. 4B). Metabolic labeling with [ 14 C]palmitic acid confirmed that wt CD4 but not the CD4 C419S,C422S mutant (C/S) was palmitoylated (Fig. 5C). These results strongly suggest that palmitoylation of cysteine residues is not required for the association of CD4 with lipid rafts. Because BJAB cells do not express Lck, these results also suggest that interaction with Lck may be dispensable for raft association. However, we further analyzed the role of the amino acid sequence of the Lck binding site domain of CD4 as a potential raft determinant. Sucrose gradient analysis (Fig. 4A) and confocal microscopy (Fig. 4B) showed a similar pattern of distribution of wt CD4 and the CD4 C445A,C447A mutant in which the Lck binding motif 445CQC447 was eliminated by mutation into AQA (40). These results were also confirmed by using a CD4 deletion mutant CD4 d.437-442 unable to interact with Lck (40), which localized to lipid rafts (Fig. 4A). In conclusion, neither palmitoylation nor Lck interaction motifs play a critical role as raft localizing determinants of the CD4 receptor in the cells.

A Novel Dominant Raft Localizing Determinant Is Present in the Membrane-proximal Cytoplasmic Domain of CD4 -We
have determined that a CD4 mutant with most of its cytoplasmic domain deleted (amino acid residues 428 -458) colocalized with GM1 in lipid rafts (data not shown). We thus investigated the role of the remaining short sequence of positively charged amino acid residues, RHRRR, which is localized adjacent to the membrane-proximal palmitoylation sequence CVRC in the cytoplasmic domain of CD4 (Fig. 1C). Replacement of two amino acid residues, RH, with an alanine doublet did not significantly affect raft localization of the CD4 RH/AA mutant (Fig.5A). Replacement of the positively charged arginine triplet with negatively charged glutamic acid residues reduced raft localization of the RRR/EEE mutant by about 90% as compared with wt CD4. However, substitution of the entire RHRRR sequence with alanine residues (RHRRR/A5; RA5 mutant) completely abolished association with lipid rafts (Fig. 5, A and B).
Localization of RA5 to Triton X-100-soluble membranes does not rule out entirely the possibility that RA5 may localize to Triton X-100-soluble rafts or that its association with lipid rafts is very week. Indeed, the pentaspan membrane protein, prominin, present on microvilli localizes to Triton X-100-soluble but Lubrol WX-insoluble lipid microdomains (41). Therefore, we analyzed the distribution of RA5 in sucrose gradient fractions prepared from cells lysed with a non-ionic detergent similar to Lubrol WX, detergent W-1. Our results show that whereas wt CD4 localizes to W-1 rafts, the RA5 mutant is excluded from W-1-resistant rafts (Fig. 5A, lower panel). These results strongly support the notion that RA5 localizes to non-raft membrane domains independently of the detergent used for extraction.
Because the RHRRR sequence is positioned next to palmitoylated cysteine residues, mutations of RHRRR could interfere with cysteine palmitoylation and affect membrane domain localization of the mutant. To investigate this possibility, the cells were transfected with wt CD4, RA5, or CD4 C419S,C422S (C/S) mutant followed by metabolic labeling of the cells with [1-14 C]palmitic acid. After immunoprecipitation of the CD4 with anti-GFP antibodies, incorporation of [1-14 C]palmitate was analyzed by autoradiography. Results show that both wt and RA5 CD4 but not the C/S mutant incorporated palmitate (Fig. 5C). Thus, non-raft localization of the RA5 mutant is not because of inhibition of palmitoylation of the receptor. These results together with our previous observation that non-palmitoylated CD4 C419A,C422A mutant localized to rafts, strongly suggest that palmitoylation is not a major raft localization determinant for the receptor.
CD4 Localized in Non-raft Membranes Supports Productive Entry of HIV-1-To investigate the consequences of non-raft localization of the CD4 RA5 receptor for virus entry, we applied a single-cycle infectivity assay. In this assay, we used HIV-1 virions produced by cotransfection of an envelope-negative and luciferase gene expressing the reporter construct NL env(Ϫ)luc(ϩ) and HXB2 envelope expression vector (42). Thus, pseudotyped virions are able to establish only a single round of infection. Consequently, the amount of luciferase activity measured in cells infected with pseudotyped viruses may serve as an indirect estimation of productive virus entry. Our results show that the RA5 mutant supported virus entry and replication to levels only slightly lower (0 -30%) than wt CD4 in transfected 293T cells (Fig. 6A). Moreover, the CD4 D1-D2 mutant with intact HIV-1 gp120 binding site was unable to support productive virus entry and replication despite its raft localization ( Figs. 2A and 6B). Altogether, these results demonstrate that the CD4 receptor redirected to non-raft microdomains was still able to support productive virus entry and replication. However, we cannot rule out the possibility that rafts are required for fusion after the attachment of HIV-1 to the RA5 receptor. This possibility is suggested by the distribution of viral proteins in sucrose gradients prepared from 293T cells transfected with wt CD4 or RA5 and exposed to HIV-1 (Fig. 6, C and D). The pattern of gp120 expression in membrane microdomains was strikingly similar to that of CD4; gp120 was present in lipid rafts in cells transfected with wt CD4 and localized to non-rafts in cells expressing RA5. However, mature Gag p24 as well as p55 or p41 localized to lipid rafts in cells transfected with either wt CD4 or RA5. The presence of p55 and p41 may reflect contamination of virus preparations with microvesicles enriched in raft membranes containing assembling HIV-1 Gag precursor and envelope proteins, or with immature virus particles. Although the presence of Gag p24 in lipid rafts may represent a post-binding association of viral proteins with rafts during or after the fusion process, we cannot reject a possibility that the presence of Gag in rafts reflects association of Gag with viral membrane rafts. The existence of virion-associated lipid rafts has recently been suggested (43). However, sucrose gradient analysis of Triton X-100 virus lysates showed that both Gag and envelope gp120 proteins accumulated in low buoyant, high-density "non-raft" fractions (Fig.  6E). Thus, we conclude that although localization of CD4 to lipid rafts is not necessary for virus fusion and entry, postbinding stages in virus entry may require lipid raft assembly.

DISCUSSION
Preferential localization of CD4 to lipid rafts suggests that these cholesterol and glycosphingolipid-enriched lipid microdomains may represent HIV-1 entry sites. Experimental results suggest that integrity of lipid rafts is important for virus entry, and destabilizing rafts by either extraction of cholesterol or inhibition of glycosphingolipid synthesis inhibits virus entry (13,17,42). However, extraction of cholesterol with ␤-cyclodextrin suffers from various drawbacks, thereby raising questions about specificity of the observed inhibition of virus entry. Instead, mutation of the raft-localizing determinants, which would result in exclusion of the CD4 receptor from raft microdomains, could directly demonstrate the importance of raft localization of the receptor in virus entry. However, different approaches used to identify and eliminate markers controlling raft partitioning of the receptor produced conflicting results (31,37,38,44). Therefore, in an effort to resolve these discrepancies, we have investigated the localization of raft-sorting molecular determinants in the CD4 receptor. Because glycosylation was reported to mediate raft association of some glycoproteins (27,45), we first evaluated the role of CD4 glycosylation in association with lipid rafts. We have found that neither the interaction of CD4 with the raft-localized ganglioside GM3 (24,25) nor N-glycosylation of CD4 were required for raft localization of the receptor.
These observations did not, however, exclude the possibility that molecular determinants for raft association may be present elsewhere in the extracellular domains of the receptor. To explore this possibility, we created mutants with deletions in different extracellular domains of CD4 and analyzed their surface expression and raft localization. From different CD4 deletion mutants tested, the two-domain mutants, D1-D2 and D3-D4, were expressed on the cell surface and significantly colocalized with the raft resident marker GM1. However, in contrast to the D1-D2/Thy-1 chimera (30), the D1-D2 mutant only marginally supported virus entry. Interestingly, deletion all four extracellular domains of CD4 restored surface expression and raft association of the mutant, suggesting that the raft determinant may be present in transmembrane or cytoplasmic domains of CD4. However, mutagenesis of the transmembrane domain of CD4 was without effect on raft association, indicating that raft localization marker(s) are likely localized in the cytoplasmic domain of the receptor.
There are several features in the cytoplasmic domain of CD4 that may serve as potential raft-localizing determinants. One of them is receptor palmitoylation. However, the role of CD4 palmitoylation in raft association is controversial (31, 37, 38). We did not observe any significant changes in the raft localization of non-palmitoylated CD4. Moreover, the presence of acylated Lck, which interacts with CD4, was proposed to play an important role in anchoring CD4 with raft microdomains present on microvilli (44). Accordingly, mutations that disrupted binding of Lck to the CQC motif in CD4 also reduced but did not abolish raft localization of the receptor (37,38). Interestingly, these studies showed that only a synergistic effect of combined mutations of palmitoylation sites and Lck binding motif was able to remove CD4 from rafts.
We have shown that despite the absence of Lck in BJAB cells, wt CD4 and palmitoylation mutant CD4 C419A,C422A localized to lipid rafts. Also, mutations disrupting Lck binding sites in CD4 (40) were without effect on raft localization. These results lead us to conclude that neither palmitoylation nor Lck binding play a dominant role as raft localizing signals in CD4.
We have found, however, that a short sequence of positively charged amino acid residues, RHRRR, may control CD4 partitioning to lipid rafts. Although substitution of the first 2 amino acid residues of this sequence with alanine residues was without significant effect on raft localization, replacement of the arginine triplet with glutamic acid residues resulted in over 90% reduction in raft association, as compared with wt CD4. Importantly, substitution of the entire RHRRR sequence with alanine residues completely abolished raft localization of the RA5 mutant despite intact palmitoylation and Lck binding motifs.
We have also investigated the possibility that substitution of RHRRR with alanine residues could inhibit palmitoylation of the adjacent, membrane-proximal cysteine residues, and affect membrane microdomain distribution of the RA5 mutant. However, our results showed that the RA5 mutant was excluded from rafts despite palmitoylation of the receptor and thus suggest that the RHRRR sequence represents a dominant raft localization determinant.
Next, we investigated whether non-raft localization of RA5 may affect virus entry. We have found that productive entry of the HXB2 pseudotyped virus into cells expressing RA5 was similar to the entry levels supported by wt CD4. These results are in agreement with the recent observation that the R5 Ba-L envelope-pseudotyped virus enters cells in a raft-independent fashion (38), although in this study the CD4 receptor was redirected to non-rafts by simultaneous mutation of the palmitoylation sites and Lck binding motifs. However, rafts could be required for fusion after the attachment of HIV-1 to the RA5 receptor. This possibility was suggested by the observation that, in contrast to the membrane distribution pattern of viral gp120 resembling that of wt CD4 or RA5, viral Gag proteins were localized to lipid rafts both in cells transfected with wt CD4 and RA5.
Our results contrast with a report showing that a CD4-low density lipoprotein receptor chimera was unable to support virus entry (31). One explanation is that CD4-low density lipoprotein receptor chimera and RA5 mutants may localize to non-overlapping microdomains with differing permissiveness for virus fusion. In support of this possibility, it has been suggested that both lipid rafts and non-raft membranes may represent a complex mosaic of partially overlapping or nonoverlapping microdomains (41,46). Furthermore, low density lipoprotein receptors reside in a novel lipid microdomain that is distinct from that containing GPI-linked proteins (47).
We propose that this novel raft localization determinant may suggest a new mechanism for virus entry. It has been shown that CD4 and coreceptors are localized on microvilli of T cells (44,48). This could indicate that microvilli might be a place for virus fusion and entry. However, the diameter of viral particles (145 Ϯ 25 nm) (49) or viral core (on average 120 nm in length) usually exceeds the average diameter of a microvillus (less than 100 nm in diameter) and would present a significant barrier for virus fusion and entry. Thus, microvilli may simply anchor incoming viruses but fusion could take place only after disassembly of microvilli (50). Alternatively, the receptor-virus complexes would migrate from microvilli-localized rafts to nonraft membrane areas where fusion could take place. This, however, does not exclude the possibility that at some stages of virus entry lipid raft reassembly may be required for successful completion of the process. It has been shown that a similar cluster of positively charged amino acids in the juxta-membrane cytoplasmic domain of CD44, CD43, and ICAM-2 binds ERM (ezrin/radixin/moesin) proteins (51). Therefore, we speculate that the RHRRR domain in CD4 may represent a potential site for the interaction with ERM proteins that may anchor the receptor to cortical actin filaments. In support of this hypothesis, ezrin has been shown to colocalize with lipid rafts (52) FIG. 6. CD4 RA5 localized in nonraft membranes supports productive entry of HIV-1. A, BJAB cells were transfected with wt CD4 (WT control), CD4 RA5 (RA5), and two-domain CD4 mutants, D1-D2 and D3-D4, as controls. Twenty-four hours after transfection, the cells were incubated for 1 h with NL env(Ϫ)luc(ϩ) virus pseudotyped with HXB2 envelope (5 cpm of RT/cell). Luciferase activity was measured in cell lysates prepared 48 h post-infection as described in the legend to Fig. 1. Cells transfected with wt CD4 and infected with envelope-defective NL env(Ϫ)luc(ϩ) virus (Env Ϫ Lucϩ) were included as negative controls. Luciferase activity is expressed relative to the activity detected in cells transfected with wt CD4 (100%). Standard deviations of five independent experiments are shown. B, Triton X-100 postnuclear lysates prepared from transfected BJAB cells (before infection with NL env(Ϫ)luc(ϩ) virus pseudotyped with HXB2 envelope and used in A) were fractionated on sucrose density gradients and analyzed for the raft/non-raft distribution of wt CD4, RA5, and truncation mutants. Localization of non-raft CD71 and raftresident CD48 is shown. C, distribution of viral proteins in sucrose gradients prepared from 293T cells transfected with wt CD4 (left panel) or RA5 (right panel) and exposed to HIV-1. Forty hours after transfection, the cells were infected with NL4-3 (5 cpm of RT/cell) for 45 min at 37°C, washed, and lysed in Triton X-100 and fractionated on sucrose density gradients. Localization of viral envelope gp120 and Gag proteins was detected by Western blotting using human anti-HIV-1 serum. D, distribution of wt CD4 and RA5 in transiently transfected 293T cells and infected with NL4-3 as described in C. Localization of raft-resident marker GM1 and non-raft CD71 is shown. E, sucrose gradient distribution of viral proteins prepared from 0.5% Triton X-100 lysates of virus preparations. NL4-3 was partially purified by ultracentrifugation through a 20% sucrose cushion, the pellet was resuspended in 500 l of TNE buffer and mixed with an equal volume of TNE with 1% Triton X-100. HIV-1 Gag and gp120 proteins accumulated in low buoyant, high-density "non-raft" fractions. Similar results were obtained for virus lysates prepared in 1% Triton X-100 (not shown). and CD4 was found to colocalize with actin and ezrin in microvilli and membrane ruffles (53). Thus, binding of HIV-1 gp120 to CD4, resulting in phosphorylation of serine residues (7) and dissociation of activated Lck from the receptor (54), would decrease the positive charge of the cytoplasmic domain and reduce its binding to ERM proteins, promoting the CD4-HIV-1 aggregates to exit rafts in microvilli and enter a fusioncompetent environment. As a result of its reduced charge in the cytoplasmic domain, RA5 might interact weakly with ERM proteins and already exist in this fusion-competent domain. Experiments to test this hypothesis are currently in progress in our laboratory.
Our observations do not exclude the possibility that virus may also enter the cells through lipid rafts. A four-domain CD4-GPI (55), or two-domain D1-D2/Thy-1 chimeras (30), both anchored to the cell membrane through a GPI tail, support virus entry and replication, suggesting lipid rafts as viral entry sites. Although GPI-linked proteins are in general confined to lipid rafts, it was not shown that these particular CD4 chimeras localized to lipid rafts occupied by wt CD4. To the contrary, there is evidence suggesting that diverse and heterologous GPI-anchored proteins may be segregated to distinct lipid rafts (56).
In addition, our results may also suggest a mechanism regulating the transient recruitment of CD4 to the center of the immunological synapse (57,58). We propose that engagement of CD4 at the immunological synapse followed by CD4 tail phosphorylation and Lck dissociation will reduce CD4 binding to ERM proteins. Consequently, phosphorylated CD4 after delivery of Lck to the center of immunological synapse will localize to its periphery. Further studies will be needed to verify this hypothesis and to better understand the role of lipid rafts in virus entry and immune cell activation.