Proc. Natl. Acad. Sci. USA Vol. 90, pp. 2700-2704, April 1993 Medical Sciences Infection of colonic epithelial cell lines by type 1 human immunodeficiency virus is associated with cell surface expression of , a potential alternative gpl20 receptor JACQUES FANTINI*tt, DAVID G. COOK*t, NEAL NATHANSONt, STEVEN L. SPITALNIK§, AND FRANCISCO GONZALEZ-SCARANO*t¶ Departments of *Neurology, tMicrobiology, and §Pathology and Laboratory Medicine, University of Pennsylvania Medical Center, Philadelphia, PA 19104-6146 Communicated by Stanley Prusiner, December 22, 1992

ABSTRACT The gastrointestinal tract plays a major role For these reasons, understanding the mechanisms by in the pathogenesis and pathophysiology ofinfection by the type which gastrointestinal cells are infected by HIV-1 is of 1 human immunodeficiency virus (HIV-1). It is a potential fundamental importance. In vitro infection studies demon- route for viral entry and it is the site of a number of compli- strated a marked heterogeneity in the sensitivity of intestinal cations, including both opportunistic infections and a primary cell lines to HIV infection. For example, HT-29, a human HIV-induced enteropathy. Correspondingly, both in vivo and colonic adenocarcinoma cell line, was productively infected in vitro studies have demonstrated HIV infection of gastroin- with several HIV-1 and HIV-2 isolates (5, 10-13). In contrast, testinal cells ofIymphoid and epithelial origin. HT-29, a human Caco-2, a cell line obtained from a different human adeno- colonic epithelial cell line that is infectable with many HIV-1 carcinoma, appeared much less sensitive to HIV infection strains, does not express CD4 protein or mRNA. Recent studies (5). Infection of HT-29 cells did not involve CD4, the main showed that antibodies recognizing a neutral related HIV receptor, since it was not blocked by anti-CD4 mono- to galactosylceramide (GalCer) in HT-29 cells inhibited HIV-1 clonal antibodies OKT4A and Leu-3a (10, 12, 13), which are infection of this cell line, extending previous findings in neural known to block infection in T lymphocytes and lymphoid cell cells. In the current studies, we further analyzed the neutral lines (14). Furthermore, CD4 protein or mRNA has not been consistently detected in this cell line (10). These results of HT-29 cells and showed that they contained suggested that, like several fibroblastic and neural lines (15, authentic GalCer and that recombinant gpl20 bound to this 16), HT-29 cells are infected by an alternative pathway. It glycolipid. Moreover, by analyzing GalCer expression in clones was recently reported that a glycolipid related to galactosyl- derived from HT-29 and Caco-2 (another human colonic cell (GalCer), an essential component of the neural line), we observed that the level of expression of this glycolipid receptor for HIV surface glycoprotein gpl20 (17, 18), was was associated with the sensitivity to HIV-1 infection. Sub- expressed by HT-29 cells and that anti-GalCer antibodies clones ofCaco-2 did not express GalCer and were not infectable inhibited HIV-1 infection in these cells (19). These results with any of three HIV-1 strains. These results strengthen the suggested that GalCer (or a close derivative) was involved in possibility that GalCer is an alternative receptor in CD4- cell HIV-1 entry into the HT-29 cells. lines. Furthermore, since GalCer is a major glycolipid in In this report, we extend the original findings by conclu- epithelial cells of the small intestine and colon, these results sively proving that the glycolipid originally detected in HT-29 provide a structural basis for the binding of HIV-1 by gastro- cells was indeed GalCer and not a related ceramide mono- intestinal epithelial cells and the entry of the virus into those hexoside. Furthermore, by analyzing GalCer expression in cells. two human colon epithelial cell lines and clones derived from them, we demonstrate that the presence of this glycolipid is The gastrointestinal tract plays a central role in the patho- associated with sensitivity to HIV-1 infection. These results genesis ofthe acquired immunodeficiency syndrome (AIDS). suggest that GalCer, one of the major neutral glycolipids The rectal mucosa may represent a major route of horizontal expressed by normal epithelial intestinal cells in humans (20, transmission of type 1 human immunodeficiency virus 21), may behave as the HIV-1 receptor of gastrointestinal (HIV-1) during anal intercourse. Furthermore, a number of epithelium. gastrointestinal disorders, including diarrhea, malabsorption, and weight loss, are commonly reported during the course of MATERIALS AND METHODS HIV-1 infection (1). Although the origin of these gastroin- testinal symptoms is still under considerable debate (2), in Cell Culture. The human colonic adenocarcinoma cell lines many HIV-infected patients the only pathogenic organism HT-29 [American Type Culture Collection (ATCC) HTB38] identified in the intestine is HIV itself, and this has led to the and Caco-2 (ATCC HTB37) were cloned by limiting dilution, hypothesis that some of these syndromes may be due to a as described (22, 23). Parental and clonal cell lines were primary HIV-induced enteropathy (3, 4). This concept has routinely grown in Dulbecco's modified Eagle's medium been supported by two independent findings: (i) both primary and transformed intestinal cells derived from intestine have Abbreviations: GalCer, galactosylceramide; HIV, human immuno- deficiency virus; HPTLC, high-performance thin-layer chromatog- been infected by HIV in vitro (5, 6) and (ii) HIV viral RNA raphy; rgpl20, recombinant HIV-1(IIIB) gpl20; TCIDso, 50% tissue and core proteins have been detected by in situ hybridization culture infectious dose. and immunohistochemistry in jejunal and colonic biopsy lTo whom reprint requests should be addressed at: Department of specimens obtained from HIV-infected individuals (7-9). Neurology, University ofPennsylvania, Clinical Research Building, 422 Curie Boulevard, Philadelphia, PA 19104-6146. tCurrent Address: Universite d'Aix-Marseille I and Institut National The publication costs of this article were defrayed in part by page charge de la Sant6 et de la Recherche Medicale U270, Facult6 de Medecine payment. This article must therefore be hereby marked "advertisement" Nord, Boulevard Pierre Dramard 13916, Marseille, Cedex 20, in accordance with 18 U.S.C. §1734 solely to indicate this fact. France. 2700 Downloaded by guest on September 23, 2021 Medical Sciences: Fantini et al. Proc. Natl. Acad. Sci. USA 90 (1993) 2701 (DMEM)/Ham's F12 medium, 1:1 (vol/vol), supplemented silica-gel-60 plates in pyridine/ethyl acetate/acetic acid/ with 10% heat-inactivated fetal bovine serum and 15 mM water, 5:5:1:3, and the plates were scanned with an automatic Hepes (pH 7.4). Human T-lymphoblastoid SUP-Ti cells and TLC linear analyzer (Berthold, Pittsburgh, PA). Glucose and the T-B hybrid cell line CEMx174 were grown in RPMI 1640 galactose standards were chromatographed in adjacent lanes supplemented with 10% fetal bovine serum. and the migration positions were measured after orcinol Virus Stocks. Viruses used were the prototype HIV-1 strain spraying. HIV-1(IIIB) (24), the Zairian strain HIV-1(NDK) (25), and the macrophage-tropic strain HIV-1(89-6) (26). To prepare virus stocks, culture supernatants from chronically infected RESULTS CEMx174 cells were passed through a 0.22-,um Millipore Cell Surface Expression of GalCer in Human Colonic Epi- filter and stored in aliquots at -80°C until use. The stocks had thelial Cell Lines. The presence of GalCer and related gly- a titer in CEMx174 cells of 105 1 50% tissue culture infective colipids on the surface of human colonic epithelial cell lines doses (TCID50)/ml for HIV-1(IIIB), 104-9 TCID50/ml for was analyzed by indirect immunofluorescence microscopy of HIV-1(NDK), and 105.8 TCID50/ml for HIV-1(89.6). live cells with an anti-GalCer monoclonal antibody (R-Mab). HIV-1 Infection of Colonic Cells. Exponentially growing This antibody revealed a marked heterogeneity in antigen human colonic cells were exposed to HIV-1 at a multiplicity expression among clonal subpopulations derived from a of infection of 0.001-1 TCID50 per cell for 16 hr at 37°C. After single adenocarcinoma cell line. More than 50% of parental extensive washing in serum-free culture medium, the cells HT-29 cells were strongly labeled by the antibody. A similar were treated with trypsin (2.5 mg/ml) to remove excess proportion of positive cells was found in the HT-29-A7 inoculum, harvested, and subcultured three times before subpopulation (Fig. 1A), while 20% of HT-29-D4 cells were analysis. Infection was characterized from day 3 postinfec- weakly labeled by the antibody (data not shown). Finally, tion (i) by measuring HIV-1 p248a8 antigen in the culture only rare clusters ofHT-29-D9 cells (representing <5% ofthe supematant with an antigen-capture assay (Coulter), (ii) by overall population) demonstrated immunoreactivity with the indirect immunofluorescence detection of p24wn in acid/ anti-GalCer antibody (Fig. 1B). alcohol-fixed cells (27), and (iii) by cocultivation with the Caco-2 cells were also subcloned into several populations, CD4+ indicator cell lines SUP-Ti and CEMx174. and using R-Mab we have found some variation in the surface Inhibition of Infection by Anti-GalCer Monoclonal Anti- immunoreactivity (data not shown). One clone (Caco-2/14) bodies. Cells were treated at 4°C for 2 hr with -20 ,ug ofeither initially demonstrated a high level of labeling, comparable to anti-GalCer (R-Mab, a mouse IgG3) (28) or anti-CD4 that of HT-29. However, further analysis (see below) dem- (OKT4A) monoclonal antibody and subsequently exposed to onstrated that the antibody was reacting with a related HIV-1 (0.1 TCID50 per cell) for 1 hr at 37°C (shorter incuba- antigen on the Caco-2/14 cells, but not with GalCer. This tion period than in experiments described above) in the related antigen does not bind gpl20. After more extensive continuous presence of monoclonal antibodies. passaging of Caco-2/14, immunoreactivity with R-Mab be- Immunofluorescence Labeling. Indirect immunofluores- came undetectable by immunofluorescence. cence detection of GalCer was performed on live cells as The parental and clonal populations expressed carcinoem- described (19). A mouse monoclonal antibody against carci- bryonic antigen, a surface glycoprotein of human colonic noembryonic antigen (Boehringer Mannheim) was used as a epithelial cells (30), ruling out contamination as an explana- marker for colonic epithelial cells. The specificity of R-Mab tion for the heterogeneity of GalCer expression. (anti-GalCer) binding was demonstrated by using mouse IgG3 Characterization of GalCer-Related Glycolipids in Human (Sigma) as a negative control. Colonic Epithelial Cells. The GalCer-related glycolipids rec- Glycolipid Analysis and gpl20 Binding. Alkali-stable Folch ognized by the anti-GalCer monoclonal antibody were ana- lower-phase (29) extracted from the cell lines were lyzed by HPTLC. The alkali-stable Folch lower-phase gly- chromatographed on high-performance thin-layer chroma- colipids were chromatographed on silica-gel plates and then tography (HPTLC) plates (silica gel 60; Merck) with chloro- visualized with orcinol. The main glycolipid isolated from form/methanol/water, 60:35:8 (vol/vol), as the mobile HT-29 cells (Fig. 2A, lane 2) was a ceramide monohexoside phase. Glycolipids were visualized with orcinol reagent (29). with the same chromatographic mobility as the lower band of For gpl20 binding, the chromatography plates were first bovine brain GalCer, which was used as a standard (lane 1). treated for 2.5 min in 0.1% poly(isobutyl methacrylate) in This lower band is the form of bovine GalCer containing n-hexane. The dried plates were then immersed in 50 mM a-hydroxylated fatty acids, the subspecies that binds gp120 Tris HCl, pH 7.4/1% gelatin for 2 hr at room temperature to (ref. 18 and unpublished results). A comigating ceramide block nonspecific binding. The plates were further incubated with recombinant HIV-1(IIIB)BHlo clone gpl20 (rgpl20, 1 ,ug/ml) for 1 hr at room temperature. Binding of rgpl2O was detected by incubating the plates with a 1:200 dilution of rabbit anti-rgpl20 (a gift from Ray Sweet, SmithKline Beecham) followed by 125I-labeled goat anti-rabbit polyclonal antibodies (2 x 106 cpm/ml). The plates were thoroughly washed in phosphate-buffered saline, dried, and exposed to Kodak x-ray film. Characterization of Carbohydrates. The glycolipids in HT-29 cells were metabolically labeled by exposing expo- nentially growing cells in T-75 flasks to either [3H]- or [14C]galactose (New England Nuclear) at 10 p.Ci (370 kBq)/ml for 16 hr at 37°C. Metabolically labeled alkali-stable FIG. 1. Immunofluorescence of human colonic cell lines with Folch lower-phase glycolipids were then chromatographed as anti-GalCer antibodies. Live cells were incubated with undiluted anti-GalCer (R-Mab) supernatant, and binding was detected with above, and the band corresponding to the ceramide mono- fluorescein-conjugated rabbit anti-mouse IgG. Control antibody hexoside was scraped; extracted in chloroform/methanol, (mouse IgG3) was negzative on all cell lines. Like the parental HT-29 1:1; desalted by chromatography on Sephadex G-25; and cells, >50% of HT-29-A7 cells (A) were strongly labeled. by the hydrolyzed in 4 M trifluoroacetic acid at 100°C for 4 hr. The antibody. Fewer than 5% of HT-29-D9 cells (B) were strongly released monosaccharides were then chromatographed on labeled; many cells had low levels offluorescence, as shown. (x130.) Downloaded by guest on September 23, 2021 2702 Medical Sciences: Fantini et al. Proc. Natl. Acad. Sci. USA 90 (1993) A B moiety was released by acid hydrolysis. The monosaccharide was then identified as galactose by its comigration with a 1 2 3 4 5 6 7 8 1 2 3 galactose standard on HPTLC (Fig. 3C). Thus, these results GalCerC demonstrated that the rgpl20-binding glycolipid extracted from HT-29 cells was GalCer. Correlation Between GalCer Expression and HIV-1 Infec- tion. To study the sensitivity of HT-29- and Caco-2-derived cell lines to HIV-1 infection, cells were exposed to either Gml- ; IIIB, NDK, or 89.6 isolates of HIV-1 at a multiplicity of infection of 1 TCID50 per cell. Infection was monitored by (i) measurement ofp24549 antigen in the culture supematant, (ii) FIG. 2. (A) extracts from colonic cell lines yield GalCer-like indirect immunofluorescence with a monoclonal antibody bands. Glycolipids (200 dig) extracted from human colonic cell lines against p24wn, and (iii) cocultivation with CD4+ SUP-Ti or (corresponding to 107 cells) were separated on HPTLC plates with CEMx174 cells. At this multiplicity of infection, the mac- chloroform/methanol/water, 60:35:8, and sprayed with orcinol. Lanes: 1, GalCer, galactosyl (Sulf), and GM1 standards rophage-tropic strain HIV-1(89.6) (26) did not appear to infect obtained from bovine brain; 2, HT-29; 3, HT-29-A7; 4, HT-29-D4; 5, any ofthe colonic cell lines used and, thus, served as a control HT-29-D9; 6, Caco-2; 7, Caco-2/14; 8, Caco-2/2. A prominent band for potential residual inoculum in some of the experiments. comigrated with the a-hydroxylated GalCer standard in lanes 2-5 HT-29, HT-29-A7, and HT-29-D4 were infected by both IIIB (HT-29 and clones) but not in lanes 6-8 (Caco-2 and clones). The and NDK isolates, and in all cases, p245ag was easily mea- GalCer band was quantified with a scanning densitometer. With surable in the supematant (Table 1). However, the level of HT-29 as standard (100%), the intensity of orcinol staining was as follows: HT-29-A7, 96%; HT-29-D4, 128%; HT-29-D9, 38%; Caco-2 p24w production was higher in HT-29 and HT-29-A7 cells, and Caco-2 clones, s5%. As determined by immunofluorescence, which expressed high levels of GalCer (Fig. 1). HT-29 and surface expression for HT-29-D4 was lower than for HT-29 or HT-29-A7 cells were also infectable by lower-titer inocula HT-29-A7. (B) rgpl2O binds to a GalCer-like band from HT-29 cells. (down to 0.001 TCID50 per cell). In addition to p24gag, the Glycolipids extracted from HT-29 (lane 2) and Caco-2/2 (lane 3) cells supematants contained virus that was fully infectious for were separated on HPTLC plates and incubated with rgpl20. Lane SUP-Ti cells (data not shown). The HT-29-D9 which 1 contained the standards (GalCer, Sulf, and GM1). A single band line, corresponding in mobility to GalCer (arrow) bound rgpl2O in HT-29 expressed lower levels of GalCer by immunofluorescence extracts (lane 2) but not in Caco-2/2 extracts (lane 3). Note that (Fig. 1), was infectable, but virus was detected only after rgp120 also bound Sulf in lane 1 (18) but not in colonic cell extracts. A B monohexoside was also detected in lipid extracts from HT- 29-A7 (lane 3), HT-29-D4 (lane 4), and, to a lesser extent, 1 2 1 2 HT-29-D9 cells (lane 5), and this glycolipid was recognized by .. the anti-GalCer antibody (data not shown). In contrast, it was GaICer[ 3 '* virtually absent from Caco-2 cells (lane 6) and its two subclones (lanes 7 and 8), although a slower migrating species P Suif- ._W was recognized by the R-Mab antibody; this species did not ,, bind gp120 (see below). Thus, we concluded that a glycolipid similar in chromatographic mobility to GalCer was present in iw ..... the HT-29 cells and to various extent in clones derived from this cell line, but not in Caco-2 cells. We next determined whether HIV-1 gpl20 could bind this ceramide. Binding of gp120 to a Ceramide Monohexoside Extracted C from HT-29 Cells. Alkali-stable Folch lower-phase lipids from HT-29 and Caco-2/2 cells were separated on HPTLC plates 1 I and incubated with rgpl20. The viral surface glycoprotein specifically bound to the ceramide monohexoside from a lipid extract of HT-29 cells (Fig. 2B, lane 2); this compound had the same mobility as a-hydroxylated bovine brain GalCer, a0 which, as previously reported (18), was also recognized by ~0 rgpl20 (lane 1). In contrast, no specific binding was observed m:a with lipids extracted from Caco-2/2 cells (lane 3) or from Caco-2/14 and parental Caco-2 cells (data not shown). These data emphasized the specificity of gpl20 binding to neutral glycolipids and demonstrated that the expression of a glyco- Migration lipid binding site for HIV-1 gpl20 may vary from one colon FIG. 3. Characterization ofthe gpl2O-binding glycolipid in HT-29 cell line to the other. extracts. HT-29 cells were metabolically labeled with [14C]- or The gpl2O-Binding Glycolipid Extracted from HT-29 Is [3H]galactose and alkali-stable Folch lower-phase lipids were sepa- GalCer. The above results demonstrated the presence of a rated on HPTLC plates. (A) Orcinol stain of standards (lane 1) and gpl20-binding ceramide monohexoside in colonic extracts. HT-29 extract (lane 2). Band corresponding to R-Mab and rgpl20 To identify the nature of its sugar moiety, HT-29 cells were binding is indicated by an arrow. (B) Autoradiogram of radioactive metabolically labeled with [14C]- or [3H]galactose, and their extracts from HT-29 cells labeled with [14C]galactose (lane 1) or glycolipid content was analyzed. The results (Fig. 3 A and B) [3H]galactose (lane 2). (C) The band indicated by the arrow in the 14C extract was purified and acid-hydrolyzed, and the sugar moiety was showed that these cells were able to use galactose as a identified by chromatography followed by quantitative scanning of precursor for the biosynthesis of various glycolipids, includ- the HPTLC plate. A single peak was seen and is shown. Abcissa, ing the ceramide monohexoside (arrow) recognized by the migration from origin; ordinate, arbitrary radioactive units. The anti-GalCer antibody and rgpl20 (Fig. 2). This radioactive migration of galactose (Gal) and glucose (Glc) standards is indicated glycolipid was purified by preparative TLC and its sugar by the arrows. Downloaded by guest on September 23, 2021 Medical Sciences: Fantini et al. Proc. Natl. Acad. Sci. USA 90 (1993) 2703

Table 1. Infection of human colonic cell lines by HIV-1 p24949, pg/ml Cell line HIV-1(IIIB) HIV-1(NDK) HT-29 1450 7312 HT-29-A7 2051 11380 HT-29-D4 380 654 HT-29-D9* 13 '10 Caco-2t '10 '10 Caco-2/2t '10 '10 Caco-2/14t '10 '10 Cells were inoculated with HIV-1 isolates at a multiplicity of infection of 1 TCID50 per cell. The concentration of supernatant p24Sag was determined by antigen-capture assay 9 days after infec- tion. The sensitivity of the assay was 10 pg/ml. *HIV-1(IIIB) and HIV-1(NDK) were reproducibly rescued by co- cultivation with SUP-Ti cells. FIG. 4. Indirect immunofluorescence of human colonic cells for tCocultivation with SUP-Ti and CEMx174 did not rescue virus. p24m. Approximately 20%o of HT-29-A7 cells infected with HIV- 1(NDK) at 1 TCID50 per cell demonstrated intracellular expression cocultivation with CD4+ cell lines. In contrast, Caco-2 cells of p249ag on day 9 after infection. The number of infected cells and their clonal derivatives were not infected. increased with time to reach -50% of the HT-29-A7 cells infected Indirect immunofluorescence of HT-29-A7 cells exposed with HIV1(NDK) 1 month after infection (data not shown). (x90.) to either HIV-1(IIIB) or HIV-1(NDK) demonstrated the presence of intracellular p245ag [Fig. 4 shows HIV-1(NDK)]. antibodies, similar experiments were performed in the HT-29 Mock-infected cells, as well as nonproductively infected cells cell line. Yahi et al. (19) were able to inhibit infection by two HIV-1 strains (LAV and NDK) with a monoclonal antibody (HT-29-D9), were negative. As expected, the infection of with high activity against GalCer, and they demonstrated the HT-29, HT-29-A7, and HT-29-D4 cells was not inhibited by presence of a ceramide monohexoside in lipid extracts of the T anti-CD4 antibodies that block CD4-dependent infection of HT-29 line. In the present study we have further character- lymphocytes (Table 2). In contrast, anti-GalCer antibodies ized this glycolipid and shown that, like GalCer from brain, efficiently inhibited p24gag production (data not shown), it is specifically recognized by rgpl20. Previous experiments giving further evidence that GalCer is involved in the binding have indicated that the presence of galactose on a ceramide and/or fusion process of HIV-1 during infection of human is not by itself sufficient to allow binding of rgpl2O (18). This colonic epithelial cells. is supported by the finding that a number of HT-29 neutral glycolipids incorporated radioactive galactose, yet only Gal- DISCUSSION Cer bound rgpi20. Having conclusively shown the presence of a gpl20- Several groups have reported that HIV-1 infection of colonic binding glycolipid on these intestinal cells, we correlated its epithelial cells is not mediated by CD4 (10, 13, 31), suggesting expression with the efficiency of HIV-1 infection. Cells that the existence of an alternative HIV-1 receptor. Following expressed either low or undetectable levels of GalCer reports of inhibition of neural cell infection with anti-GalCer (Caco-2 and its subclones) were not infected by HIV-1. Cells Table 2. HIV infection correlates with GalCer expression in human colonic epithelial cell lines GalCer HIV-1 Inhibition of infection§ Cell line expression* Isolatet Infectiont Anti-CD4 Anti-GalCer HT-29 + + IIIB ++ No Yes NDK ++ No Yes HT-29-A7 + + IIIB + + No Yes NDK ++ No Yes HT-29-D4 + IIIB + No ND NDK + No ND HT-29-D9 + IIIB + ND ND NDK + ND ND Caco-2 - IIIB - ND ND NDK - ND ND Caco-2/2 - IIIB - ND ND NDK - ND ND Caco-2/14 - IIIB - ND ND NDK - ND ND *+ +, >50% positive cells; +, >20o positive cells; +, <5% positive cells; -, GalCer undetectable. Results are based on cell surface detection with anti-Gal Cer Mab (Fig. 1) and HPTLC analysis of alkali-stable Folch lower phase glycolipids extracted from the cells (Fig. 2 and 3). tAll colon cell lines were treated with IIIB, NDK, and 89.6 isolates of HIV-1 at an equivalent multiplicity of infection, 1 TCID50 per cell. In all cases, HIV-1 89.6 produced no infection (see text). t+ +, p24gag >5000 pg/ml of cultured supernatant of colon cells 3 weeks postinfection; +, 10 pg/ml c p24gag < 1000 pg/ml and virus was rescued by cocultivation with SUP-Ti or CEMx174 cells; ±, p24gag < 10 pg/ml but virus was rescued by cocultivation; -, p24 < 10 pg/ml and virus was not rescued by cocultivation. §Antibodies were used at a concentration of 20 ,ug/ml as described in Materials and Methods. ND, not done. Downloaded by guest on September 23, 2021 2704 Medical Sciences: Fantini et al. Proc. NatL Acad Sci. USA 90 (1993) that expressed high levels of GalCer (HT-29, HT-29-A7, and, 8. Heise, C., Dandekar, S., Kumar, P., Duplantier, R., Donovan, to a lesser extent, HT-29-D4) were productively infected by R. M. & Halsted, C. H. (1991) Gastroenterology 100, 1521- HIV-1, and the infection was blocked by anti-GalCer anti- 1527. 9. Kotler, D. P., Reka, S., Borcich, A. & Cronin, W. J. (1991) bodies. The last clonal cell line, HT-29-D9, appeared very Am. J. Pathol. 139, 823-830. heterogeneous for GalCer expression (=5% positive cells 10. Omary, M. B., Brenner, D. A., De Grandpre, L. Y., Roebuck, stained by immunofluorescence) and was less sensitive to K. A., Richman, D. D. & Kagnoff, M. F. (1991) AIDS 5, HIV-1 infection. This cell line did not produce p24ww in the 275-281. supernatant ofinfected cells, and virus was rescued only after 11. Barnett, S. W., Barboza, A., Wilcox, C. M., Forsmark, C. E. cocultivation with CD4+ cell lines, much like other infectable & Levy, J. A. (1991) Virology 182, 802-809. CD4- lines (15, 16). Since GalCer was detected in HT-29-D9 12. Fantini, J., Yahi, N., Baghdiguian, S. & Chermann, J. C. (1992) J. Virol. 66, 580-585. extracts (Fig. 2), it is possible that the relative inefficiency of 13. Fantini, J., Yahi, N. & Chermann, J. C. (1991) Proc. Natl. HIV-1 infection in these cells was the consequence of a Acad. Sci. USA 88, 9297-9301. reduced accessibility of the glycolipid in the plasma mem- 14. Sattentau, Q. J., Arthos, J., Deen, K., Hanna, N., Healey, D., brane (32, 33). Beverley, P. C. L., Sweet, R. & Truneh, A. (1989) J. Exp. The association between the level of GalCer expression Med. 170, 1319-1334. and the infectability of CD4- colonic cell lines supports the 15. Tateno, M., Gonzalez-Scarano, F. & Levy, J. A. (1989) Proc. hypothesis that this glycolipid behaves as an alternative Natl. Acad. Sci. USA 86, 4287-4290. 16. Harouse, J. M., Kunsch, C., Hartle, H. T., Laughlin, M. A., receptor for HIV-1. GalCer is one of the major neutral Hoxie, J. A., Wigdahl, B. W. & Gonzalez-Scarano, F. (1989)J. glycolipids found in epithelial cells from human small intes- Virol. 63, 2527-2533. tine (20) and colon (21), as well as transformed cell lines, and 17. Harouse, J. M., Bhat, S., Spitalnik, S. L., Laughlin, M., we recently found that rgpl20 specifically binds GalCer from Stefano, K., Silberberg, D. H. & Gonzalez-Scarano, F. (1991) normal human ileal and colonic mucosal epithelium with a Science 253, 320-323. specificity similar to that found with HT-29 cells (unpublished 18. Bhat, S., Spitalnik, S. L., Gonzalez-Scarano, F. & Silberberg, work). The presence of a putative HIV-1 receptor on epithe- D. H. (1991) Proc. Natl. Acad. Sci. USA 88, 7131-7134. lial cells ofthe intestinal mucosa may clarify the pathogenesis 19. Yahi, N., Baghdiguian, S., Moreau, H. & Fantini, J. (1992) J. Virol. 66, 4848-4854. ofgastrointestinal disorders in patients with AIDS (4, 12) and 20. Bjork, S., Breimer, M. E., Hansson, G. C., Karlsson, K. A. & provide a mechanism for anorectal transmission of the virus Leffler, H. (1987) J.; Biol. Chem. 262, 6758-6765. (34). 21. Holgersson, J., Jovall, P. A. & Breimer, M. E. (1991) J. Biochem. 110, 120-131. We thank Dr. Patrice Spitalnik and Dr. Peter Pahlsson for their 22. Fantini, J., Abadie, B., Tirard, A., Remy, L., Ripert, J. P., El help with the TLC binding assays, Dr. Brian Wolffor use ofthe TLC Battari, A. & Marvaldi, J. (1986) J. Cell Sci. 83, 235-249. linear analyzer, Dr. Ray Sweet (SmithKline Beecham) for providing 23. Lehmann, M., Rabenandrasana, C., Rognoni, J. B., Verrier, rgpl20, Andrew Albright for his technical assistance, and all of the B., Marvaldi, J. & Fantini, J. (1991) Cytotechnology 5, 117-127. members of the Gonzalez and Nathanson laboratories for their 24. Popovic, M., Sarngadharan, M. G., Read, E. & Gallo, R. C. helpful advice and comments. This work was supported in part by (1984) Science 224, 497-500. Public Health Service Grants NS27405, NS31067, and NS30606 and 25. Ellrodt, A., Barre-Sinoussi, Le Bras, F., Nugeyre, M. T., by grants from the North Atlantic Treaty Organization (NATO) and Palazzo, L., Rey, F., Brun-Vezinet, F., Rouzioux, C., Segond, the Fondation pour la Recherche Medicale to J.F. P., Caquet, R., Montagnier, L. & Chermann, J. C. (1984) Lancet 1, 1383-1385. 1. Kotler, D. P., Gaetz, H. P., Klein, E. B., Lange, M. & Holt, 26. Collman, R., Balliet, J. W., Gregory, S. A., Friedman, H., P. (1984) Ann. Int. Med. 101, 421-428. Kolson, D. L., Nathanson, N. & Srinivasan, A. (1992) J. Virol. 2. Bartler, J. G., Belitsos, P. C. & Sears, L. C. (1992) Clin. Infect. 66, 7517-7521. Dis. 15, 726-735. 27. Coilman, R., Hassan, N. F., Walker, R., Godfrey, B., Cutillik, 3. Ullrich, R., Zeitz, M., Heise, M., L'age, M., Hoffken, G. & J., Hastings, J., Friedman, H., Douglas, S. D. & Nathanson, N. Riecken, E. 0. (1989) Ann. Int. Med. 111, 15-21. (1989) J. Exp. Med. 170, 1149-1163. 4. Ullrich, R., Heise, W., Bergs, C., L'age, M., Riecken, E. 0. & 28. Ranscht, B., Clapshaw, P. A., Price, J., Noble, M. & Seifert, Zeitz, M. (1992) Gut 33, 1080-1084. W. (1982) Proc. Natl. Acad. Sci. USA 79, 2709-2713. 5. Adachi, A., Koenig, S., Gendelman, H. E., Daugherty, D., 29. Leden, R. W. & Yu, R. (1982) Methods Enzymol. 83, 139-191. Gattoni-Celli, D., Fauci, A. S. & Martin, M. A. (1987) J. Virol. 30. Fantini, J. (1992) Cancer J. 5, 78-83. 61, 209-213. 31. Phillips, D. M. & Bourinbaiar, A. S. (1992) Virology 186, 6. Moyer, M. P., Huot, R. I., Ramirez, A., Jr., Joe, S., Meltzer, 261-273. M. S. & Gendelman, H. E. (1990) AIDS Res. Hum. Retrovi- 32. Karlsson, K. A. (1989) Annu. Rev. Biochem. 58, 309-350. ruses 6, 1409-1415. 33. Lloyd, K. O., Gordon, C. M., Thampoe, I. J. & DiBenedetto, 7. Nelson, J. A., Wiley, C. A., Reynolds-Kohler, C., Reese, C. (1992) Cancer Res. 52, 4948-4953. C. E., Margaretten, W. & Levy, J. A. (1988) Lancet 2, 259- 34. Yahi, N., Baghdiguian, S., Bolmont, C. & Fantini, J. (1992) J. 262. Acquired Immune Defic. Syndr. 5, 993-1000. Downloaded by guest on September 23, 2021