Sulfatide Inhibits HIV-1 Entry into CD4−/CXCR4+Cells

VIROLOGY

246, 211±220 (1998) VY989216

ARTICLE NO.

Sulfatide Inhibits HIV-1 Entry into CD42/CXCR41 Cells Jacques Fantini,*,1 Djilali Hammache,* Olivier DeleÂzay,* GeÂrard PieÂroni,² Catherine Tamalet,² and Nouara Yahi³ *Laboratoire de Biochimie et Biologie de la Nutrition, CNRS ESA 6033, Faculte des Sciences St. JeÂroÃme, 13397 Marseille Cedex 20, France; ³Laboratoire de Virologie, UF SIDA, CHRU de la Timone, 13005 Marseille, France; and ²INSERM U130, 13009 Marseille, France Received January 27, 1998; returned to author for revision February 26, 1998; accepted April 28, 1998 Sulfatide (39sulfogalactosylceramide) is the natural sulfated derivative of galactosylceramide (GalCer), a glycosphingolipid receptor allowing HIV-1 infection of CD4-negative cells from neural and intestinal tissues. The incorporation of exogenous sulfatide into the plasma membrane of HT-29 (a CD42/GalCer1/CXCR41 human intestinal cell line) or RD (CD42/GalCer2/ CXCR41 human rhabdomyosarcoma) resulted in a dose-dependent inhibition of HIV-1 infection. Experiments with luciferase reporter viruses pseudotyped with HIV-1 or amphotropic murine leukemia virus envelopes demonstrated that sulfatide acts at the level of viral entry. Paradoxically, the transfer of sulfatide in the plasma membrane of various CD42 cells resulted in increased binding of HIV-1. Surface pressure measurements were conducted to study the interaction of gp120 with glycosphingolipid monolayers. The data showed that gp120 could penetrate into a monomolecular film of GalCer, confirming the role of this glycosphingolipid as a functional receptor for HIV-1. In contrast, the insertion of gp120 into a monolayer of sulfatide was very limited. Moreover, the incorporation of sulfatide in a monomolecular film of GalCer specifically inhibited the penetration of gp120. In conclusion, these data show that sulfatide mediates gp120 binding but, in marked contrast with GalCer, is not able to initiate the fusion event. © 1998 Academic Press Key Words: HIV-1 infection; coreceptor; CXCR4; GalCer; CD4; glycolipids.

can serve as an alternative receptor for HIV-1 (Berson et al., 1996b; Bhat et al., 1991; Fantini et al., 1997; Harouse et al., 1991; Long et al., 1994; Yahi et al., 1992). This glycolipid is the prototype of a new family of alternative HIV-1 receptors including also seminolipid on spermatozoa (Brogi et al., 1996) and lactosylsulfatide on vaginal epithelium (Furuta et al., 1994). Thus, studying the interactions between HIV-1 and glycolipids may help to improve the understanding of the sexual transmission of HIV-1, as well as of the pathogenesis of this virus in the brain (Albright et al., 1996; Kimura-Kuroda et al., 1996) and in the intestinal mucosa (DeleÂzay et al., 1997a). The domain of gp120 involved in GalCer recognition has been mapped to the V3 loop based on the inhibitory activity of anti-V3 mAbs and V3-derived synthetic peptides (Cook et al., 1994; Yahi et al., 1994a, 1995). Upstream regions in the C2 domain may also be involved in the folding of the V3 loop that is optimal for binding to GalCer (Bhat et al., 1993). In addition, infection assays with chimeric viruses have identified the V3 loop as the principal determinant of HIV-1 tropism for various GalCer1 cells (Harouse et al., 1995; Trujillo et al., 1996; Yahi et al., 1996). HIV-1 isolates able to infect the intestinal HT-29 cell line belong to a subclass of T-cell-lineadapted and primary HIV-1 isolates (Yahi et al., 1996) that use the coreceptor CXCR4 to fuse with CD41 cells, i.e., X4 viruses according to the new classification for HIV-1 (Berger et al., 1998). In a recent study, we reported that GalCer and CXCR4 are coexpressed on the surface of HT-29 cells and that HIV-1 entry into these cells can be

INTRODUCTION The interaction of HIV-1 surface envelope glycoprotein gp120 with the plasma membrane of CD41 cells involves at least two distinct binding sites: (i) CD4 and (ii) a member of the chemokine receptor family, among which the SDF-1 receptor CXCR4 and the RANTES receptor CCR5 have recenly been characterized as major HIV-1 coreceptors. Following a primary interaction with CD4, a conformational change in gp120 renders the V3 domain of the viral glycoprotein available for secondary interactions with either CXCR4 or CCR5 (Alkhatib et al., 1996; Berson et al., 1996a; Deng et al., 1996; Dragic et al., 1996; Feng et al., 1996; Lapham et al., 1996; Wu et al., 1996). In the absence of CD4, the V3 loop is not correctly exposed to allow direct binding of gp120 to chemokine receptors. Yet HIV-1 can infect in vitro several CD42 cell types, including fibroblasts (Tateno et al., 1989) and neural (Harouse et al., 1989) and epithelial cells (Fantini et al., 1993). The relevance of these observations is underscored by the increasing evidence that CD42 cells can be targets for human and simian immunodeficiency viruses in vivo (Basgara et al., 1996; Dean et al., 1996; Livingstone et al., 1996). Galactosylceramide (GalCer), a glycosphingolipid abundantly expressed in neural and intestinal tissues,

1 To whom reprint requests should be addressed. Fax: 133 491-288440. E-mail: [email protected].

211

0042-6822/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.

212

FANTINI ET AL.

blocked by either anti-GalCer or anti-CXCR4 mAbs (DeleÂzay et al., 1997b). These data suggested that CXCR4 can cooperate with GalCer during the fusion process. Moreover, HIV-1 entry into HT-29 cells is determined by dual recognition of GalCer and CXCR4 since isolates like HIV-1(89.6) that can use CXCR4 (Doranz et al., 1996), but do not bind to GalCer, do not enter the cells (DeleÂzay et al., 1997b). This isolate also does not infect the neural SKNMC cell line which expresses both GalCer and its sulfated derivative 39sulfo-GalCer (sulfatide) (Harouse et al., 1995). Based on the observation that gp120 recognized GalCer and sulfatide with equal affinity (Bhat et al., 1991), it has been assumed that sulfatide could, like GalCer, function as an alternative receptor for HIV-1. Curiously, studies on the binding of gp120 to various lipids gave contradictory results as to the ability of sulfatide to recognize gp120. Using a liposome binding assay, Long et al. (1994) could confirm the interaction of gp120 with GalCer, but not with sulfatide. By ELISA, McAlarney et al. (1994) could, in contrast, detect binding of gp120 to sulfatide, but not GalCer. Moreover, the transfer of sulfatide into the plasma membrane of Blymphocytes rendered the cells competent for gp120 binding (McAlarney et al., 1994). Finally, Harouse et al. (1995) demonstrated that intact viral particles of HIV1(IIIB) could bind to sulfatide immobilized on ELISA plates, whereas HIV-1(89.6) virions could not. These data provided a molecular basis for the mechanism of infection of neural SKNMC cells and suggested that sulfatide can behave as a potential HIV-1 receptor in these cells. However, the lack of gp120 binding to sulfatide in the liposome membrane remained unclear. In this report, we present evidence that sulfatide, like GalCer, promotes the binding of HIV-1 to the plasma membrane of CD42 cells through an interaction with the V3 loop of gp120. However, gp120 makes a clear distinction between these two glycosphingolipids, as demonstrated by surface pressure measurements of GalCer and sulfatide monolayers. These data give some insight into the role of GalCer as a receptor for HIV-1 and explain why membrane-associated sulfatide behaves like a molecular decoy preventing HIV-1 entry into CD42/CXCR41 cells. RESULTS Binding of HIV-1 gp120 to sulfatide In a first series of experiments, increasing amounts of sulfatide were adsorbed on ELISA plates and probed with gp120. As shown in Fig. 1, gp120 binding was easily detected, starting from less than 100 ng of sulfatide per well. The specificity of the binding reaction was demonstrated by the lack of interaction with GluCer used as a negative glycosphingolipid control, whereas GalCer was fully recognized by the viral glycoprotein. Moreover, gp120 binding to sulfatide immobilized on ELISA plates

FIG. 1. Binding of recombinant HIV-1 gp120 to GalCer and sulfatide. Sulfatide (h), GalCer (■), GluCer (E) or solvent alone (F) was coated on ELISA plates as indicated under Materials and Methods. Binding of recombinant gp120 (IIIB isolate) was revealed with a mouse antigp120 mAb.

was inhibited by suramin, a V3 loop-binding sulfonylurea (Yahi et al., 1994a), and by anti-V3 antibodies (data not shown). However, sulfatide did not bind to gp120 when the glycoprotein was first coated on ELISA plates. Correspondingly, preincubation of HIV-1 with a solution of sulfatide up to 1 mg/ml, followed by ultracentrifugation of the virus, did not decrease HIV-1 infectivity. Taken together, these data suggest that the V3 domain of gp120 can bind to sulfatide adsorbed on a solid substratum, but not in solution. Transfer of sulfatide in CD42 cells results in increased binding of HIV-1 Since exogenously added glycosphingolipids incorporate spontaneously into the plasma membrane of recipient cells (Callies et al., 1977), these data prompted us to study the binding of HIV-1 to cells following sulfatide transfer. In a previous study, we isolated clonal cell derivatives of the CD42 human colon epithelial cell line Caco-2, which are not sensitive to HIV-1 infection and express neither GalCer nor CXCR4 (DeleÂzay et al., 1997b). Incubation of Caco-2/Cl2 cells with various concentrations of sulfatide resulted in the incorporation of the glycolipid in the plasma membrane. The sulfatidetreated cells were recognized by the anti-GalCer/sulfatide R-mAb (Rantsch et al., 1982) and the binding of the antibody increased with the amount of sulfatide incorporated (Fig. 2). Moreover, expression of sulfatide in the plasma membrane of these cells was sufficient to allow the attachment of HIV-1 and the level of virus binding

SULFATIDE INHIBITS HIV-1 INFECTION

FIG. 2. Incorporation of sulfatide into GalCer2/CD42/CXCR42 cells promotes HIV-1 binding. Caco-2/Cl2 cells grown in 96-well plates were incubated with the indicated concentrations of sulfatide, rinsed, and fixed with paraformaldehyde. The presence of sulfatide was detected by CELLISA with the R-mAb (h). The data are expressed as the mean of 10 independent experiments 6 SD. For HIV-1 binding, the cells were treated with the indicated concentrations of sulfatide and subsequently incubated with HIV-1(LAI) for 2 h at 4°C. The bound virus was detected by CELLISA using anti-gp120 and anti-gp41 mAbs as described under Materials and Methods (■). The data are expressed as the mean of 5 independent experiments 6 SD.

increased in proportion with the amount of sulfatide to which the cells were exposed (Fig. 2). Taken together, these data show a good correlation between the detection of sulfatide with the R-mAb and the binding of HIV-1 to sulfatide-treated cells. Similar results were obtained with the CD42/GalCer1/CXCR41 cell line HT-29, which showed an increased binding of the R-mAb following sulfatide incorporation (Fig. 3). Interestingly, sulfatide did not affect the binding of an anti-HLA mAb, showing that the transfer procedure did not induce a general masking of cell surface epitopes. Cell surface accessibility of CXCR4 was even slightly increased in sulfatide-treated HT-29 cells (Fig. 3). Finally, the transfer of sulfatide in the plasma membrane of HT-29 cells resulted in an increased binding of HIV-1 (data not shown), confirming the data obtained with Caco-2/Cl2 cells.

213

inhibition of HIV-1 infection of HT-29 cells, with an IC50 of 90 mg/ml. This concentration is clearly overestimated, since less than 5% of the total glycolipid in solution can be incorporated into the plasma membrane of the target cells under our experimental conditions (Callies et al., 1977). The specificity of the antiviral activity of sulfatide was demonstrated by normal replication noted with the use of other lipids: cholesterol and ganglioside GM1. Moreover, a concentration of sulfatide of 100 mg/ml was sufficient to prevent the infection of HT-29 cells by another isolate, i.e., HIV-1(LAI). Since this virus replicates less efficiently in these cells, the level of infection was analyzed by measuring the p24 antigen in the supernatant of human PBMC cocultivated with the HT-29 cells. The results in Fig. 4B show that PBMC efficiently rescued the virus from HT-29 cells exposed to HIV-1(LAI). In contrast, the virus was not rescued from HT-29 cells treated with sulfatide before HIV-1(LAI) exposure, indicating that these cells were not infected. To test whether sulfatide blocks infection at the level of viral entry, we used HIV-1-based luciferase reporter viruses. The luciferase reporter viruses infect cells in a single round but are not competent for further replication (Deng et al., 1996). Thus measurement of luciferase activity with pseudotypes of these viruses allows comparison of the relative efficiency of entry mediated by different viral envelopes. As shown in Fig. 5, HT-29 cells were infected by pseudotypes bearing the HXB2 Env, but not the macrophage-tropic Env of JRFL, which does not recognize the GalCer receptor (J. Fantini, unpublished data). Sulfatide inhibited infection with virus pseudotyped by

Sulfatide blocks HIV-1 entry into CD42/CXCR41 cells Human epithelial intestinal HT-29 cells are CD42 but express both CXCR4 and GalCer (DeleÂzay et al., 1997b). The cells were pretreated with various concentrations of sulfatide and subsequently exposed to HIV-1(NDK), an isolate that infects these cells productively. The infectivity of HT-29 cells was then analyzed by p24 concentration in the culture supernatant at 7 days postinfection. As shown in Fig. 4A, sulfatide induced a dose-dependent

FIG. 3. Incorporation of sulfatide in HT-29 cells. HT-29 cells were treated with the indicated concentration of sulfatide and assayed for binding of the anti-CXCR4 mAb 12G5 (Œ), the anti-HLA mAb IOT2 (E), the anti-GalCer/sulfatide R-mAb (■), or an irrelevant mouse mAb (‚) by CELLISA. The data are expressed as the mean of 3 independent determinations (SD ,10%).

214

FANTINI ET AL.

FIG. 5. Sulfatide blocks infection at the level of viral entry. HT-29 cells were either not treated (2) or treated (1) with sulfatide (1000 mg/ml) and then infected with luciferase reporter viruses pseudotyped by Envs of HIV-1 (JRFL or HXB2) or with A-MLV. Luciferase activity was measured 5 days later and expressed in arbitrary units [counts per seconds (cps)]. B, background luminescence.

FIG. 4. Inhibition of HIV-1 infection of HT-29 cells by sulfatide. (A) HT-29 cells were treated with various concentrations of sulfatide (h), cholesterol (E), or GM1 (■) for 2 h at 37°C. After thorough washing, the cells were exposed to HIV-1(NDK) overnight at 37°C. The cells were then trypsinated to remove excess inoculum and the state of infection was determined at day 7 postinfection by measuring the amount of p24 in the cell-free supernatant. The results shown are representative of 4 separate experiments. (B) HT-29 cells were either not treated (h) or treated (■) with 100 mg/ml of sulfatide. The cells were then exposed to HIV-1(LAI) overnight at 37°C. The cells were trypsinated, cultured for 7 days, and then cocultured with normal human PBMC as described under Materials and Methods. The rescued virus produced in cell-free supernatants of cocultured PBMC was quantified by measuring the amount of p24.

HXB2 Env, whereas the glycolipid had no effect on infection with virus bearing the A-MLV Env. Taken together, these data strongly suggest that sulfatide inhibition of HT-29 infection by T-tropic HIV-1 is due to a block at the level of viral entry and not to postentry or nonspecific effects. A similar study was conducted with the human rhabdomyosarcoma cell line RD, which expresses CXCR4 but neither CD4 nor GalCer (data not shown). Exposure of these cells to undiluted stocks of HIV-1(NDK) (m.o.i. up to

1 TCID50 per cell) did not result in productive infection. Nevertheless, HIV-1(NDK) could infect the RD cell line, as demonstrated by PCR amplification of HIV-1 pol sequences (Fig. 6, lanes 1 and 2). Preincubation of these cells with sulfatide (500 mg/ml) resulted in a total blockade of infection (lanes 5 and 6), whereas at 100 mg/ml the glycosphingolipid inhibited HIV-1 entry in two of four experiments (PCR signal detected in lanes 3 and 10, but not in lanes 4 and 9). Taken together, these data showed that sulfatide could inhibit HIV-1 entry into RD cells, with an approximate IC50 of 100 mg/ml.

FIG. 6. Inhibition of HIV-1 infection of RD cells by sulfatide. RD cells were either not preincubated (lanes 1, 2) or preincubated with sulfatide at 100 mg/ml (lanes 3, 4, 9, 10) or 500 mg/ml (lanes 5, 6). Forty-eight hours after infection, DNA was extracted and PCR-amplified as described under Materials and Methods. The arrow shows the amplified fragment of 805 bp. Positive controls (cultures of PBMC infected with primary isolates of HIV-1) were analyzed in lanes 7 and 8. A negative control for PCR (lane 11) and molecular weight markers (12) are also shown.

SULFATIDE INHIBITS HIV-1 INFECTION

215

DISCUSSION

FIG. 7. Interaction of gp120 with glycosphingolipid monolayers. Monomolecular films of GalCer (F), GluCer (h), sulfatide (■), or mixed films [GalCer/GluCer (E); GalCer/sulfatide (‚)] were prepared at various initial surface pressures. The variation of surface pressure induced by recombinant gp120 (IIIB isolate) added in the aqueous phase was measured. Results with mixed films are illustrated with a dotted line.

Sulfatide and GalCer are distinguished by gp120 Surface pressure experiments were conducted to study the interaction of gp120 with glycosphingolipid monolayers. In these experiments, the increase in surface pressure (DP) caused by penetration of the monolayer by the viral glycoprotein was measured as a function of initial surface pressure of the monolayer. According to this model, lipid monolayers show decreased compressibility with increasing surface pressure, so that DP is expected to decrease as the initial surface pressure of the monolayer increases (Lear and Rafalski, 1993). Under these conditions, Fig. 7 shows that gp120 interacts specifically with a monomolecular film of GalCer and less efficiently with a monomolecular film of sulfatide. Indeed, sulfatide was at best as active as GluCer, a control glycosphingolipid that is not recognized by gp120: for an initial surface pressure of 10 mN/m, the viral glycoprotein induced a variation (DP) of 8.2, 3.9, and 3.4 mN/m for GalCer, sulfatide, and GluCer, respectively. To further study the interaction of gp120 with glycosphingolipids, a mixed monolayer was prepared with an equal amount of GalCer and sulfatide. As shown in Fig. 7, gp120 did not significantly penetrate into this monolayer. In contrast, the viral glycoprotein could efficiently interact with a monolayer containing equal amounts of GalCer and GluCer. These data suggest that sulfatide, which is specifically recognized by gp120, impairs the functional interactions between the viral glycoprotein and GalCer.

The main result of the present study is the discovery that sulfatide can promote the binding of HIV-1 gp120 but, unlike GalCer, is unable to initiate the subsequent steps leading to the fusion process. The binding of gp120 to both GalCer and sulfatide can be evidenced by ELISA, which contrasts with a previous study reporting gp120 binding to sulfatide alone (McAlarney et al., 1994). In our assay, the protocol for glycosphingolipid adsorption on ELISA plates is based on the solubilization of the lipid with a mixture of chloroform, ethanol, and n-hexane, which, following addition of methanol and evaporation of the solvent, optimizes the orientation of glycosphingolipids with the hydrophobic moiety bound to the ELISA plate and the polar head accessible to the ligand (D. Hammache and J. Fantini, manuscript in preparation). Interestingly, gp120 did not recognize sulfatide in solution, which may suggest that several sulfatide molecules associated by hydrogen bonds are necessary for an optimal interaction, as previously suggested for GalCer (Long et al., 1994). Indeed, a threshold level of GalCer was required to allow gp120 binding to liposomes (Long et al., 1994) and, consistently, to confer sensitivity to HIV infection (Fantini et al., 1993). Thus, assuming that a lattice of sulfatide molecules stabilized by hydrogen bonds is required for gp120 binding, this supramolecular organization may develop only when the glycosphingolipid is adsorbed on a solid sustratum, which can be either a plastic support (ELISA plate) or a plasma membrane (e.g., after sulfatide transfer). This may explain why the preincubation of HIV-1 with sulfatide in solution does not reduce viral infectivity. One major outcome of this study is the evidence that gp120 does make a difference between GalCer and sulfatide, as assessed by using monomolecular films of glycosphingolipids at the air±water interface as a model for glycosphingolipid patches of the plasma membrane. Addition of gp120 in the aqueous phase underneath a monolayer of GalCer induced a marked increase of the surface pressure, and the effect was gradually decreased as the initial pressure of the monolayer increased. The influence of the initial surface pressure on the compressibility of the monolayer induced by gp120 demonstrates the high specificity of the interaction, as previously established for several other lipids and ligands (Lear and Rafalski, 1993). From a physical point of view, these data are interpretated as an evidence of ligand insertion into the lipid monolayer (Maggio, 1996). In the case of HIV-1 gp120, this may represent a complex network of interactions between amino acid residues of the V3 loop and several domains of the GalCer molecule, including both the polar sugar head and the ceramide moiety. The initial event is probably a primary interaction with galactose, since gp120 does not bind to GluCer or ceramides and does not affect the surface pressure of

216

FANTINI ET AL.

monolayers formed by these lipids. Then, secondary interactions with the ceramide moiety may lead to partial insertion of gp120 inside the monomolecular film of GalCer. This secondary binding step would significantly strengthen viral adhesion and prepare the fusion event, as recently pointed out by Haywood (1994). In the case of sulfatide, the sulfatation of galactose in position 39 may not affect the first binding step, as evidenced by immunoenzymatic techniques. However, according to surface pressure measurements, gp120 does not penetrate into a monomolecular film of sulfatide. Thus, the interaction between sulfatide and gp120 is completed at the first step, suggesting that the sulfated group impairs accessibility to secondary binding sites in the hydrophobic part of the glycosphingolipid. Consequently, binding of gp120 to sulfatide is less efficient than that to GalCer and may not be tight enough to initiate the fusion process. An illustration of the relative weakness of the gp120/sulfatide association is given by Long et al. (1994), who reported that the binding of gp120 to sulfatide in liposomes does not resist high-speed centrifugation. Similar data have been reported by Haywood and Boyer (1982) for the attachment of Sendai virus to liposomes bearing ganglioside receptors. The transfer of sulfatide into the plasma membrane of Caco-2/Cl2, a CD42/GalCer2/CXCR42 human cell line, confers HIV-1 binding, but not entry (DeleÂzay et al., 1997b). For HT-29, a CD42/GalCer1/CXCR41 cell line, the incorporation of sulfatide leads to increased HIV-1 binding and decreased infection. The hypothesis that sulfatide could inhibit HIV-1 infection though a general antiviral activity is dismissed by the lack of effect of the glycosphingolipid on the infection of various CD41 cells, including PBMC and HT-29/CD41 cells (data not shown). Moreover, sulfatide inhibited infection of luciferase reporter viruses pseudotyped by HXB2, but not A-MLV Envs. These findings strongly suugest that the inhibitory activity of sulfatide is due to a block at the level of viral entry. Thus, one can reasonably hypothesize that sulfatide acts as a decoy which prevents gp120 from establishing functional interactions with GalCer. In this respect, it is worth noting that transfer of a synthetic ganglioside analogue on the surface of receptor-negative cells resulted in binding of influenza virus, but not entry (Brossmer et al., 1993). A tentative model is proposed in Fig. 8 to explain the anti-HIV-1 activity of sulfatide in CD42/CXCR41 cells. GalCer and CXCR4 are coexpressed on the surface of most HT-29 cells, and only viruses able to interact with both receptors can enter these cells (DeleÂzay et al., 1997b). The simpliest explanation to account for the inhibition of HIV-1 entry into HT-29 cells by either anti-GalCer or anti-CXCR4 mAbs is that the virus interacts sequentially with GalCer and CXCR4 (Fig. 8, top left). Since the binding of gp120 to GalCer and CXCR4 involves the same domain, i.e., the V3 loop (Cook et al., 1994; Speck et al., 1997), a given gp120

molecule cannot bind to both receptors. Thus, the attachment of HIV-1 to a GalCer patch on the surface of HT-29 cells may involve several gp120, which may increase the probability of a functional interaction between available gp120 units and CXCR4 molecules. The oligomeric association of gp120 allows the same viral spike to interact with both GalCer and CXCR4 through distinct gp120 molecules. In the absence of GalCer, e.g., for RD cells, the probability of a direct interaction between gp120 and CXCR4 is very low, and only selected isolates like HIV1(NDK) can infect these cells, yet at a high m.o.i. and with a poor efficiency (Fig. 8, bottom left). Indeed, RD cells are 1000-fold less sensitive to HIV-1(NDK) infection than HT-29 cells (data not shown). Since CXCR4 expression is higher in RD than in HT-29 cells (not shown), one can reasonably conclude that it is GalCer which renders HT-29 cells particularly sensitive to HIV-1 infection. Once incorporated in the plasma membrane of HT-29 cells, sulfatide competes with GalCer for gp120 binding (Fig. 8, top right). This is demonstrated by the inability of gp120 to penetrate into a mixed monolayer of GalCer and sulfatide. At high concentrations, sulfatide may occupy most of the available viral spikes, which therefore cannot bind to CXCR4. Alternatively, the virus attached to sulfatide cannot be delivered to CXCR4 due to the instability of the gp120±sulfatide complex. A similar mechanism may explain the inhibitory activity of sulfatide incorporated in the plasma membrane of RD cells (Fig. 8, bottom right). It should be emphasized that the inhibitory effect of exogenously added sulfatide on HIV-1 entry does not imply that cells expressing both GalCer and sulfatide are protected from HIV-1 infection (Albright et al., 1996). For instance, HIV-1 can infect SKNMC cells (Harouse et al., 1995) as well as primary oligodendrocytes (Albright et al., 1996) which are GalCer1/sulfatide1. However, it can be postulated that CD42/CXCR41 cells expressing high amounts of sulfatide and low amounts of GalCer may not be infected. The identification (or the genetic engineering) of such cells would help to validate this model. In conclusion, the data reported here show that membrane-associated sulfatide can efficiently prevent infection of CD42/CXCR41 cells, whether or not they express the GalCer receptor. These data shed some light on the respective roles of GalCer and CXCR4 in the mechanism of HIV-1 entry into intestinal epithelial HT-29 cells. Studies are now in progress to determine whether other galactolipids, such as lactosylsulfatide in vaginal cells (Furuta et al., 1994) and seminolipid in spermatozoa (Brogi et al., 1996), are involved in CD4-independent pathways of HIV-1 infection. MATERIALS AND METHODS Viruses HIV-1(LAI) (BarreÂ-Sinoussi et al., 1983) and HIV-1(NDK) (Ellrodt et al., 1984) were harvested from chronically

FIG. 8. HIV-1 interaction with CD42/CXCR41 cells: effect of sulfatide. (Top left): (1) HIV-1 binds to a patch of GalCer on the surface of HT-29 cells. The GalCer-mediated attachment of the virion increases the probability of a functional interaction between available gp120 units and CXCR4 molecules. Following gp120 binding to CXCR4, the fusion process (2) is initiated by a conformational change in the transmembrane glycoprotein gp41 which releases its N-terminal fusion peptide. (Top right): sulfatide incorporated in the plasma membrane of HT-29 cells competes with GalCer for HIV-1 binding. The virions bound to sulfatide are not delivered to CXCR4, which results in an inhibition of viral entry. (Bottom left): selected HIV-1 isolates can infect RD cells through a direct interaction with CXCR4. (Bottom right): the incorporation of sulfatide in the plasma membrane of RD cells prevents the binding of HIV-1 to CXCR4.

SULFATIDE INHIBITS HIV-1 INFECTION 217

218

FANTINI ET AL.

infected CEM cells. The macrophage-tropic HIV-1(89.6) isolate (Collman et al., 1992) was produced in peripheral blood mononuclear cells (PBMC) obtained from healthy donors.

160-A (anti-gp41), 110-B (anti-gp120), 25 (anti-gp120) obtained from the Agence Nationale de Recherche sur le SIDA (ANRS), and RL 16-76 (anti-gp120), purchased from Immunotech. The mAbs were detected as described above.

Cell culture The human colon epithelial cell lines HT-29 and Caco2/Cl2 (Fantini et al., 1994) and the human rhabdomyosarcoma cell line RD (DeleÂzay et al., 1997b) were routinely grown in 25-cm2 flasks (Costar) in DMEM/F12 supplemented with 10% fetal bovine serum, penicillin (100 units/ ml), and streptomycin (100 mg/ml) as described previously (Fantini et al., 1993). gp120 binding to glycosphingolipids A stock solution of the indicated lipid (1 mg/ml) was prepared in hexane:chloroform:ethanol: (11:5:4, v:v:v) and diluted in methanol immediately before use. The glycolipids (100 ml) were then coated on Greiner ELISA plates (Osi, Elancourt, France) by evaporation of the solvent under a chemical hood. The wells were saturated for 2 h at 37°C in phosphate-buffered saline (PBS) containing 2% bovine serum albumin. Recombinant gp120 (10 mg/ ml) was incubated 2 h at 37°C. The plates were washed with PBS containing 0.05% Tween 20, incubated with an anti-gp120 mAb (Immunotech, Marseille, France), and revealed with peroxidase-conjugated goat anti-mouse IgG (Sigma) at a dilution of 1:1000). Reaction products were developed with o-phenylenediamine and the absorbance was measured at 490 nm with a Biotek EL311 ELISA reader (Osi). ELISA on intact cells (CELLISA) The cells grown in 96-well plates (Costar) were treated as indicated, washed, fixed in 3.7% paraformaldehyde, and incubated with PBS containing 2% bovine serum albumin (PBS/BSA) to reduce nonspecific binding. The indicated primary antibody was added in PBS/BSA for 1 h at room temperature. The cells were then washed and incubated with peroxidase-conjugated goat antimouse for 45 min. The labeling was revealed as described above. The primary antibodies used in this study were the anti-HLA-ABC mAb IOT2 (Immunotech), the anti-CXCR4 mAb 12G5 (Endres et al., 1996), the antiGalCer/sulfatide R-mAb (Rantsch et al., 1982), or the anti-gp120 RL 16-76 (Immunotech) as an irrelevant mouse IgG. For HIV-1 binding, the cells were treated with the indicated amount of sulfatide in serum-free medium, washed with culture medium, and incubated with HIV1(LAI) (100 ng of p24/ml) for 2 h at 4°C. After washing in PBS, the cells were fixed with 3.7% paraformaldehyde and incubated with PBS/BSA for 1 h at room temperature. The virus was detected with a mixture of mouse nonneutralizing anti-gp120 and anti-gp41 mAbs: 41-A (anti-gp41),

HIV-1 infection of CD42 cells When indicated, the cells were treated with sulfatide (or other lipids) in serum-free medium for 2 h at 37°C (DeleÂzay et al., 1997b). HT-29 cells were exposed to various isolates of HIV-1 as previously described (Fantini et al., 1993; Yahi et al., 1994b). The infections were done at a multiplicity of infection (m.o.i.) of 0.1 tissue culture infectious dose (TCID50) per cell. After 2 h of incubation with the virus, the cells were trypsinized and subcultured at least twice before analysis of the production of the HIV-1 p24 antigen. The amount of p24 was measured in the cell-free culture supernatant with an antigen capture ELISA assay (DuPont). To detect low levels of HIV-1 infection in HT-29 cells, an aliquot of the cells was cocultivated with indicator cells (normal human PBMC obtained from healthy donors), and HIV-1 p24 was measured in the cell-free culture supernatant at the time indicated after coculture. RD cells were exposed to HIV1(NDK) at 1 m.o.i. and analyzed by PCR after 7 days of culture. Briefly, genomic DNA was prepared by using Qiagen columns following the maufacturer's protocol (Qiagen, France). The RT region of HIV-1 DNA was amplified by two rounds of nested PCR with Taq DNA polymerase and supplied buffer (Boehringer-Mannheim). The first round was performed with primers MJ3 (59AGTAGGACCTACACCTGTCAAC-39) and MJ4 (59-CTGTTAGTGCTTTGGTTCCTCT-39) at 1 mM MgCl2 for an initial 7-min denaturation at 94°C and then 35 cycles of denaturing for 40 s at 94°C, 1 min of annealing at 55°C, and a 2-min 30-s extension at 72¡C. The second round used primers A35 (59-TTGGTTGCACTTTAAATTTCCCCATTAGTCCTATT-39) and NE135 (59-CCTACTAACTTCTGTATGTCATTGACAGTCCAGCT-39) under the first round conditions. Amplified RT regions of 805 bp were then electrophoresed on 1% agarose gels and visualized by ethidium bromide staining. Normal human PBMC were infected and analyzed as previously reported (Fantini et al., 1997). Pseudotyped virus infection assays Luciferase reporter viruses pseudotyped by Envs of HIV-1 macrophage-tropic JRFL or T-cell-line-adapted HXB2 or with amphotropic murine leukemia virus (AMLV) were generously provided by D. Littman. The HIV reporter proviruses are single round defective viruses which bear deletions in env, vif, and vpr and thus require coexpression of envelope to produce an infectious particle. Target cells were plated in 24-well plates at 105 cells per well. Forty-eight hours later, the cells were either not treated or treated with 1000 mg/ml sulfatide for

SULFATIDE INHIBITS HIV-1 INFECTION

2 h at 37°C. After washing, the cells were infected with 0.5 ml of luciferase reporter virus per well as described by Hill et al. (1997). The culture medium was changed 48 h later. At day 5 postinfection, the cells were lysed and assayed for luciferase activity using the Packard LucLite luciferase reporter assay kit and a Packard Tri-Carb 2100TR counter. Surface pressure measurements The surface pressure was measured with a Langmuir film balance (A&D Instruments, Oxford, UK) using the Collect software (Labtronics Inc., Guelph, Ontario, Canada). After being dissolved in a mixture of hexane:chloroform:ethanol 11:5:4 (v:v:v), lipids were spread inside a Teflon tank. In all experiments, the subphases were pure water obtained by filtration through a milli-Q water purification system (Millipore, Saint-Quentin, France). To measure the interaction of gp120 with monomolecular films of glycolipids, the lipids were spread inside the Teflon tank at various initial surface pressures. ACKNOWLEDGMENTS This work was supported by the Fondation pour la Recherche MeÂdicale (SIDACTION). O.D. is recipient of a SIDACTION fellowship. D.H. is recipient of a ReÂgion PACA fellowship. We thank F. Gonzalez-Scarano for the design of sulfatide transfer experiments. We are very grateful to D. Littman and V. Kewalramani for the generous gift of luciferase reporter viruses and their help with the pseudotyped virus infection assay. We also thank the M.R.C. for providing the CHO cell line producing recombinant gp120 and J. Hoxie for the 12G5 mAb.

REFERENCES Albright, A. V., Lavi, E., O'Connor, M., and Gonzalez-Scarano, F. (1996). HIV-1 infection of a CD4-negative primary cell type: The oligodendrocyte. Perspect. Drug Discov. Des. 5, 43±50. Alkhatib, G., Combadiere, C., Broder, C. C., Feng, Y., Kennedy, P. E., Murphy, P. M., and Berger, E. A. (1996). CC CKR5: A RANTES, MIP1a, MIP1b receptor as a fusion cofactor for macrophage-tropic HIV-1. Science 272, 1955±1958. BarreÂ-Sinoussi, F., Chermann, J. C., Rey, F., Nugeyre, M. T., Chamaret, S., Gruest, J., Axler-Blin, C., Brun-Vezinet, F., Rouzioux, C., Rozenbaum, W., and Montagnier, L. (1983). Isolation of a T-lymphotropic retrovirus from a patient at risk for acquired immune deficiency syndrome. Science 220, 868±870. Basgara, O., Lavi, E., Bobroski, L., Khalili, K., Pestaner, J. P., and Pomerantz, R. (1996). Cellular reservoirs of HIV-1 in the central nervous system of infected-individuals: Identification by combination of in situ PCR and immunohistochemistry. AIDS 10, 573±585. Berger, E. A., Doms, R. W., FenyoÈ, E. M., Korber, B. T. M., Littman, D. R., Moore, J. P., Sattentau, Q. J., Schuitemaker, H., Sodroski, J., and Weiss, R. A. (1998). A new classification for HIV-1. Nature 391, 240. Berson, J. F., Long, D., Doranz, B. J., Rucker, J., Jirik, F. R., and Doms, R. W. (1996a). A seven-transmembrane domain receptor involved in fusion and entry of T-cell-tropic human immunodeficiency virus type 1. J. Virol. 70, 6288±6295. Berson, J. F., Doms, R. W., and Long, D. (1996b). Interaction of human immunodeficiency virus type 1 envelope protein with liposomes contaning galactosylceramide. Perspect. Drug Discov. Des. 5, 169±180. Bhat, S., Spitalnik, S. L., Gonzalez-Scarano, F., and Silberberg, D. H. (1991). Galactosyl ceramide or a derivative is an essential compo-

219

nent of the neural receptor for human immunodeficiency virus type 1 envelope glycoprotein gp120. Proc. Natl. Acad. Sci. USA 88, 7131± 7134. Bhat, S., Mettus, R. V., Reddy, E. P., Ugen, K. E., Srikanthan, V., Williams, W. V., and Weiner, D. B. (1993). The galactosyl ceramide/sulfatide receptor binding region of HIV-1 gp120 maps to amino acids 206-275. AIDS Res. Hum. Retroviruses 9, 175±181. Brogi, A., Presentini, R., Solazzo, D., Piomboni, B., and CostantinoCeccarini, E. (1996). Interaction of human immunodeficiency virus type 1 envelope glycoprotein gp120 with a galactoglycerolipid associated with human sperm. AIDS Res. Hum. Retroviruses 12, 483±489. Brossmer, R., Isecke, R., and Herrler, G. (1993). A sialic acid analogue acting as a receptor determinant for binding but not for infection by influenza C virus. FEBS Lett. 323, 96±98. Callies, R., Schwarzmann, G., Radsak, K., Siegert, R., and Wiegandt, H. (1977). Characterization of the cellular binding of exogenous gangliosides. Eur. J. Biochem. 80, 425±432. Collman, R., Balliet, J. W., Gregory, S. A., Friedman, H., Kolson, D. L., Nathanson, N., and Srinivasan, A. (1992). An infectious molecular clone of an unusual macrophage-tropic and highly cytopathic strain of the human immunodeficiency virus type 1. J. Virol. 66, 7517±7521. Cook, D. G., Fantini, J., Spitalnik, S. L., and Gonzalez-Scarano, F. (1994). Binding of human immunodeficiency virus type 1 (HIV-1) gp120 to galactosylceramide (GalCer): Relationship to the V3 loop. Virology 201, 206±214. Dean, G. A., Reubel, G. H., and Pedersen, N. C. (1996). Simian immunodeficiency virus infection of CD81 lymphocytes in vivo. J. Virol. 70, 5646±5650. DeleÂzay, O., Yahi, N., Tamalet, C., Baghdiguian, S., Boudier, J. A., and Fantini, J. (1997a). Direct effect of type 1 human immunodeficiency virus (HIV-1) on intestinal epithelial cell differentiation: Relationship to HIV-1 enteropathy. Virology 238, 231±242. DeleÂzay, O., Koch, N., Yahi, N., Hammache, D., Tourres, C., Tamalet, C., and Fantini, J. (1997b). Co-expression of CXCR4/fusin and galactosylceramide in the human intestinal epithelial cell line HT-29. AIDS 11, 1311±1318. Deng, H., Liu, R., Ellmeier, W., Choe, S., Unutmaz, D., Burkhart, M., Di Marzio, P., Marmon, S., Sutton, R. E., Hill, C. M., Davis, C. B., Peiper, S. C., Schall, T. J., Littman, D. R., and Landau, N. R. (1996). Identification of a major co-receptor for primary isolates of HIV-1. Nature 381, 661±666. Doranz, B. J., Rucker, J., Yi, Y., Smyth, R. J., Samson, M., Peiper, S. C., Parmentier, M., Collman, R. G., and Doms, R. W. (1996). A dual-tropic primary HIV-1 isolate that uses fusin and the b-chemokine receptors CKR-5, CKR-3, and CKR-2b as fusion cofactors. Cell 85, 1149±1158. Dragic, T., Litwin, V., Allaway, G. P., Martin, S. R., Huang, Y., Nagashima, K. A., Cayanan, C., Maddon, P., Koup, R. A., Moore, J. P., and Paxton, W. A. (1996). HIV-1 entry into CD41 cells is mediated by the chemokine receptor CC-CKR-5. Nature 381, 667±673. Ellrodt, A., BarreÂ-Sinoussi, F., Le Bras, F., Nugeyre, M. T., Palazzo, L., Rey, F., Brun-Vezinet, F., Rouzioux, C., Segond, P., Caquet, R., Montagnier, L., and Chermann, J. C. (1984). Isolation of a new human T-lymphotropic retrovirus (LAV) from a married couple of Zairians, one with AIDS, the other with prodromes. Lancet 1, 1383±1385. Endres, M. J., Clapham, P. R., Marsh, M., Ahuja, M., Davis Turner, J., McKnight, A., Thomas, J. F., Stoebenau-Haggarty, B., Choe, S., Vance, P. J., Wells, T. N. C., Power, C. A., Sutterwala, S. S., Doms, R. W., Landau, N. R., and Hoxie, J. A. (1996). CD4-independent infection by HIV-2 is mediated by fusin/CXCR4. Cell 87, 745±756. Fantini, J., Cook, D. G., Nathanson, N., Spitalnik, S. L., and GonzalezScarano, F. (1993). Infection of colonic epithelial cell lines by type 1 human immunodeficiency virus (HIV-1) is associated with cell surface expression of galactosyl ceramide, a potential alternative gp120 receptor. Proc. Natl. Acad. Sci. USA 90, 2700±2704. Fantini, J., Yahi, N., Delezay, O., and Gonzalez-SCarano, F. (1994). GalCer, CD26 and HIV infection of intestinal epithelial cells. AIDS 8, 1347±1348.

220

FANTINI ET AL.

Fantini, J., Hammache, D., DeleÂzay, O., Yahi, N., AndreÂ-BarreÁs, C., RicoLattes, I., and Lattes, A. (1997). Synthetic soluble analogs of galactosylceramide (Galcer) bind to the V3 domain of HIV-1 gp120 and inhibit HIV-1-induced fusion and entry. J. Biol. Chem. 272, 7245±7252. Feng, Y., Broder, C. C., Kenedy, P. E., and Berger, E. A. (1996). HIV-1 entry cofactor: functional cDNA cloning of a seven-transmembrane, G protein-coupled receptor. Science 272, 872±877. Furuta, Y., Eriksson, K., Svennerholm, B., Fredman, P., Horal, P., Jeansson, S., Vahlne, A., Holmgren, J., and Czerkinsky, C. (1994). infection of vaginal and colonic epithelial cells by the human immunodeficiency virus type 1 is neutralized by antibodies raised against conserved epitopes in the envelope glycoprotein gp120. Proc. Natl. Acad. Sci. USA 91, 12559±12563. Harouse, J. M., Kunsch, C., Hartle, H. T., Laughlin, M. A., Hoxie, J. A., Wigdahl, B., and Gonzalez-Scarano, F. (1989). CD4-independent infection of human neural cells by human immunodeficiency virus type 1. J. Virol. 63, 2527±2533. Harouse, J. M., Bhat, S., Spitalnik, S. L., Laughlin, M., Stefano, K., Silberberg, D. H., and Gonzalez-Scarano, F. (1991). Inhibition of entry of HIV-1 in neural cell lines by antibodies against galactosyl ceramide. Science 253, 320±323. Harouse, J. M., Collman, R., and Gonzalez-Scarano, F. (1995). Human immunodeficiency virus type 1 infection of SK-N-MC cells: Domains of gp120 involved in entry into a CD4-negative, galactosyl ceramide/39 sulfo-galactosyl ceramide-positive cell line. J. Virol. 69, 7383±7390. Haywood, A. M. (1994). Virus receptors: Binding, adhesion strenghtening, and changes in viral structure. J. Virol. 68, 1±5. Haywood, A. M., and Boyer, B. P. (1982). Sendai virus membrane fusion: Time course and effect of temperature, pH, calcium, and receptor concentration. Biochemistry 21, 6041±6046. Hill, C. M., Deng, H., Unumatz, D., Kewalramani, V. N., Bastiani, L., Gorny, M. K., Zolla-Pazner, S., and Littman, D. R. (1997). Envelope glycoproteins from human immunodeficiency virus types 1 and 2 and simian immunodeficiency virus can use human CCR5 as a coreceptor for viral entry and make direct CD4-dependent interactions with this chemokine receptor. J. Virol. 71, 6296±6304. Kimura-Kuroda, J., Nagashima, K., and Yasui, K. (1996). Inhibitory effects of HIV-1 gp120 on myelin formation. Perspect. Drug Discov. Des. 5, 17±29. Lapham, C. L., Ouyang, J., Chandrasekhar, B., Nguyen, N. Y., Dimitrov, D. S., and Golding, H. (1996). Evidence for cell surface association between fusin and the CD4-gp120 complex in human cell lines. Science 274, 602±605. Lear, J. D., and Rafalski, M. (1993). Peptide-bilayer interactions: physical measurements related to peptide structure. In ``Viral Fusion Mechanisms,'' (J. Bentz, Ed.), pp 55±66. CRC Press, Boca Raton, FL. Livingstone, W. J., Moore, M., Innes, D., Whitelaw, J., Wyld, R., Robertson, J. R., Brettle, R. P., Bell, J. E., and Simmonds, P. (1996). Frequent infection of CD8 lymphocytes by HIV-1 in peripheral blood of AIDS patients. Lancet 348, 649±654.

Long, D., Berson, J. F., Cook, D. G., and Doms, R. W. (1994). Characterization of human immunodeficiency virus type 1 gp120 binding to liposomes containing galactosylceramide. J. Virol. 68, 5890±5898. Maggio, B. (1996). The surface behavior of glycosphingolipids in biomembranes: A new frontier of molecular ecology. Prog. Biophys. Mol. Biol. 62, 55±117. McAlarney, T., Apostolski, S., Lederman, S., and Latov, N. (1994). Characteristics of HIV-1 gp120 glycoprotein binding to glycolipids. J. Neurosci. Res. 37, 453±460. Rantsch, B., Clapshaw, P. A., Price, J., Noble, M., and Seifert, W. (1982). Development of oligodendrocyte and Schwann cells studied with a monoclonal antibody against galactocerebroside. Proc. Natl. Acad. Sci. USA 79, 2936±2939. Speck, R. E., Wehrly, K., Platt, E. J., Atchison, R. E., Charo, I. F., Kabat, D., Chesebro, B., and Goldsmith, M. A. (1997). Selective employment of chemokine receptors as human immunodeficiency virus type 1 coreceptors determined by individual amino acids within the envelope V3 loop. J. Virol. 71, 7136±7139. Tateno, M., Gonzalez-Scarano, F., and Levy, J. A. (1989). Human immunodeficiency virus can infect CD4-negative human fibroblastoid cells. Proc. Natl. Acad. Sci. USA 86, 4287±4290. Trujillo, J. R., Wank, W. K., Lee, T. H., and Essex, M. (1996). Identification of the envelope V3 loop as a determinant of a CD-negative neuronal cell tropism for HIV-1. Virology 217, 613±617. Wu, L., Gerard, N. P., and Wyatt, R. (1996). CD4-induced interaction of primary HIV-1 gp120 glycoproteins with the chemokine receptor CCR-5. Nature 384, 179±183. Yahi, N., Baghdiguian, S., Moreau, H., and Fantini, J. (1992). Galactosyl ceramide (or a closely related molecule) is the receptor for human immunodeficiency virus type 1 on human colon epithelial HT-29 cells. J. Virol. 66, 4848±4854. Yahi, N., Sabatier, J. M., Nickel, P., Mabrouk, K., Gonzalez-Scarano, F., and Fantini, J. (1994a). Suramin inhibits binding of the V3 region of HIV-1 envelope glycoprotein gp120 to galactosylceramide, the receptor for HIV-1 gp120 on human colon epithelial cells. J. Biol. Chem. 269, 24349±24353. Yahi, N., Spitalnik, S. L., Stefano, K. A., De Micco, P., Gonzalez-Scarano, F., and Fantini, F. (1994b). Interferon-g decreases cell surface expression of galactosylceramide, the receptor for HIV-1 gp120 on human colonic epithelial cells. Virology 204, 550±557. Yahi, N., Sabatier, J. M., Baghdiguian, S., Gonzalez-Scarano, F., and Fantini, J. (1995). Synthetic multimeric peptides derived from the principal neutralization domain (V3 loop) of human immunodeficiency virus type 1 (HIV-1) gp120 bind to galactosylceramide and block HIV-1 infection in a human CD4-negative mucosal epithelial cell line. J. Virol. 69, 320±325. Yahi, N., Ratner, L., Harouse, J. M., Gonzalez-Scarano, F., and Fantini, J. (1996). Genetic determinants controlling HIV-1 tropism for CD42/ GalCer1 human intestinal epithelial cells. Perspect. Drug Discov. Des. 5, 161±168.