Gangliosides from human melanoma immunomodulate response of T cells to interleukin-2

ELLULARIMMUNOLOGY

III,410-419(1988)

Gangliosides from Human Melanoma lmmunomodulate Response of T Cells to Interleukin-2 DAVE S. B. HOON,’ REIKO F. IRIE, AND ALISTAIR J. COCHRAN* Division of Surgical Oncology, John Wayne Clinic, Armand Hammer Laboratories, Jonsson Comprehensive Cancer Center, and *Division of Surgical PathoIogy, U3.A School of Medicine, Los Angeles, California 90024 Received July 10, 1987; accepted September 20, 1987 The gangliosides expressedby normal melanocytes are predominantly GM3 (>90%) and CD3 (~5%). Malignant melanoma can express several other types of gangliosides in significant quantities, including GM2 and GD2. Melanoma patients can develop an immune response against some of these ganglioside antigens on autologous melanoma cells. The four major gangliosides expressed by human melanoma cells (GM3, GD3, GM2, and GD2) were examined for their immunomodulatory effect on lymph node lymphocytes from melanoma patients. Gangliosides were added exogenously to lymphocytes grown in the presence of K-2. Preferential interactions of specific melanoma gangliosides on IL-2 stimulation were found. While GM2 and CD2 enhanced the lymphocyte response to IL-2, GM3 and CD3 significantly inhibited this response. GM2 and CD2 differ from GM3 and CD3 by the presence ofa terminal N-acetylgalactosamine. Since different gangliosides can upregulate and down-regulate lymphocyte responsesto IL-2, the ganglioside phenotype of melanoma cells may play a major role in determining whether an individual tumor causesimmune stimulation or suppression. @1988 Academic Pm, 1~.

INTRODUCTION Human melanoma cells synthesize and express high concentrations of membrane gangliosides (sialic acid-bearing glycolipids) in comparison with cells of other types of tumors ( l-3). The four commonly detected gangliosides of human melanoma cells are GM3, GD3, GM2, and GD2. Of these, GM3 and GD3 are expressed at high density, whereas GM2 and GD2 are present at relatively low density but have been shown to be autoimmunogenic in man (3-5). The exact ratio of individual gangliosides varies from one melanoma to another (3) and there is a correlation between ganglioside composition and prognosis in melanoma (6, 7). The immune response to cutaneous melanoma is biphasic; during early tumor progression there is a tumordirected immune response; however, as the tumor progresses,the immune response is suppressed(8). Carbohydrate surface molecules including gangliosides (9- 12) are shed from melanoma cells and may affect the functional capacity of lymphocytes adjacent to a tumor or, in the caseof lymph node lymphocytes, at a distance from a tumor. In this study ’ To whom correspondence should be addressed at the Division of Surgical Oncology, 9th flr., Louis Factor Bldg., UCLA School of Medicine, Los Angeles, CA 90024. 410 0008-8749/88 $3.00 Ck~pyrisht 0 1988 byAcademic Press,Inc. AUrightsofreproduction in any form reserved.

GANGLIOSIDES

MODULATE

T-CELL RESPONSE TO IL2

411

GM3

GM1 GD3 GD~A GD2 GDlg GTlg

STDS

MEL

FIG. 1. A representative thin-layer chromatogram of the ganglioside content of a human melanoma (autopsy specimen). Left column: purified ganglioside standards. Right column: gangliosides purified from melanoma. Isolation of gangliosides was as described by Tsuchida et al. (3).

we examined the effect of exogenous melanoma-derived gangliosides on the response of lymphocytes to interleukin-2 (IL-2).* The interaction of IL-2 to lymphocytes is essential to lymphocyte proliferation and the generation of immune responsesin vivo and in vitro ( 13- 16). MATERIALS AND METHODS Ganglioside isolation. Gangliosides were extracted from human melanomas and isolated from human melanomas as described previously (3). Purified gangliosides were assessedby applying them on TLC plates (high-pressure TLC plate of Silica Gel 60; E. Merck A. G., Darmstadt, Federal Republic of Germany). Ganglioside patterns were visualized by spraying with resorcinol-HCl reagent. As standard gangliosides for TLC analyses we used both purified human brain and melanoma gangliosides. Figure 1 is a representative example of a typical TLC chromatogram of a human melanoma biopsy specimen. The gangliosides GM3 and GD3 are expressed at high density, whereas GM2 and GD2 are expressedat low density. Lymphocytes from melanoma patients. Lymphocytes were obtained from tumorfree (as determined by histologic examination) lymph nodes regional to a primary melanoma ( 17) and from blood. Lymphocytes were isolated by standard Ficoll-Hypaque gradient centrifugation. IL-Zdependent human T-cell lines were established from lymphocytes obtained from draining lymph nodes of melanoma patients. Basically, lymphocytes were grown in the presence of natural IL-2 (lo-20 units) (Collaborative Research, Lexington, MA) and PHA (10 pg/ml, Burroughs Welcome, NC) for 4-5 days and then reseededat lower concentrations in 24-well plates in the presence of 10 units IL-2. Cultures which responded to IL-2 were expanded into cell lines for 2 Abbreviations used: cpm, counts per minute; IL2, interleukin-2; PHA, phytohemmaglutinin; rIL-2, recombinant IL2; TDLNs, tumor-draining lymph nodes; [‘H]TdR, [‘Hlthymidine reagent; TLC, thinlayer chromatography.

412

HOON,

IRIE, AND

COCHRAN

study of ganglioside effect. These cell lines consisted predominantly of T cells and could be maintained 3 to 4 months in culture. Establishment of IL-2-dependent Tcell lines from peripheral blood lymphocytes was much more difficult. Proliferation assa)ls. A standard [3H]thymidine proliferation assay was used to evaluate growth of lymphocytes ( 17). Briefly, cells were seededin quadruplicate wells at 105/well for individual tests in 96-well microplates (Flow Laboratories). Culture medium consisted of RPM1 1640 medium containing 5% heat-inactivated fetal calf serum (GIBCO, Grand Island, NY). Cells were grown at 37°C for 72 hr and then treated for 18 hr with 1 PCi of [3H]thymidine, harvested with a PHD harvester (Cambridge, MA), and counted with liquid scintillation fluid. Results were analyzed in counts per minute. Statistical analysis. Statistical analysis was performed using the Student two-tailed t test. RESULTS Eflect of Purified Whole Melanoma Gangliosides When stimulated with natural IL2 lymphocytes proliferate 2 to 10 times more than unstimulated lymphocytes. Purified mixtures of all the gangliosides derived from individual melanomas significantly inhibited the response of fresh tumor-draining lymph node (TDLN) lymphocytes to IL-2 in a dose-dependent fashion (Fig. 2A). The actual extent of inhibition varied with ganglioside samples prepared from different melanomas. Some ganglioside preparations significantly enhanced the response of lymphocytes to IL-2 at low concentrations (
GANGLIOSIDES

MODULATE

0 I I 62.50 31.25 15.63

I 7.61

413

T-CELL RESPONSE TO IL-2

1 3.91

, 1.95

I 0.93

I 0.49

, 0.24

J 0.12

Ganglioside concentration (pg/mll

01 15.60

3.90

0.98 Gonglioside

0.24 concentration

0.06

0.02

3 0

(pg /ml)

FIG. 2. The immunomodulating effect of purified whole gangliosides derived from a melanoma on the IL-2 and PHA responseof lymphocytes. (A) Whole purified gangliosides were added at different concentrations to lymphocytes from TDLNs ( IO5cells per well in 96well microtiter plates) with rIL-2 or PHA. In the absence of rIL-2 or PHA, TDLN lymphocyte proliferation was less than 5000 cpm. Results were expressedas percentages of control cultures (rIL-2 or PHA stimulation in the absenceof gangliosides). Open circles represent PHA stimulation at 0.5 &ml (Burrough Welcome) and solid circles represent rIL-2, 20 units (Amgen). (B) Effects of gangliosides on an IL-2dependent human T-cell line. The standard error of mean for each point was ~5%. At ganglioside concentrations of 15.60 and 0.06 &ml, the inhibition and stimulation, respectively, were significantly (P i 0.05) greater than those in the absence of gangliosides. For the IL-2-dependent T-cell lines grown in absenceof rIL-2 counts were lessthan 1500 cpm.

most enhancement. GM3 and GD3, the major gangliosides of human melanoma, strongly inhibited the response to IL-2 (Fig. 3). GM 1, a component of many normal tissues (20), did not significantly enhance or inhibit IL-2 responses.To further investigate the specificity of the enhancing response of IL-2 on incubation with GM2 and GD2 we established IL-Zdependent T-cell lines from breast cancer-draining lymph nodes. The gangliosides GM2 and GD2 were mostly inhibitory on rIL-2 stimulation to these breast lymph node T-cell lines (Table 1).

Toxicity of Individual Gangliosideson T Cells While investigating mechanisms whereby ganglioside molecules mediate stimulation and inhibition, we sought evidence of a toxic effect of gangliosides on lymphocytes at concentrations at which inhibitory activity was seen. Lymphocyte viability

414

HGGN, IRIE, AND COCHRAN

39. I GANGLIOSIDES

2.4 @g/ml

~1~100)

FIG. 3. Effect of purified individual gangliosides on the lymphocyte responseto rIL-2. Individual gangliosideswere added exogenously at different concentrations to IL-2dependent human T-cell lines established from melanoma patients. In the absence of IL-2 proliferation of cells was 5000 cpm. The graphs depict representative examples of several experiments with different T-cell lines. Gangliosides and 20 units of rIL-2 were added to the cells together. Melanoma gangliosides used: GM1 n ; GM2 0, GM3 t& GD2 0, GD3 q I. Percentage stimulation and inhibition was calculated as follows: (1 - (cpm of cells + rIL-2 + gangliosides/cpm of cells + r&2)) X 100. The standard error of mean for each point was lessthan 5%.

was >85% after 12 or 24 hr of incubation, with each of the gangliosides (similar to incubation without gangliosides) ruling out toxicity. Ganglioside Binding to IL-2 We also investigated the possibilities that the inhibitory effect of gangliosides was due to their binding to rIL-2 and that this interaction by gangliosides neutralized the ability of IL-2 to react with TAC on cells. To examine the binding of GM3 and GD3

TABLE 1 Effect of GM2 and GD2 on Human Breast T Cells Ganglioside concentration (&ml 156

2.4

(% stimulation of control)

Ganglioside GM2 GD2

39.1

X l/ 100)

-16.3 -13.2

-10.6 -8.4

1.2 -11.1

Note. Individual gangliosides were added exogenously at different concentrations to IL-2-dependent human T-cell lines established from breast cancer-draining lymph nodes. In the absence of IL-2 proliferation of cells was a mean 1739 cpm, whereas in the presence of IL-2 alone proliferation was a mean 19,190 cpm. This is a representative experiment. Gangliosides and 20 units of rIL-2 were added to the cells together. Percentage stimulation was calculated as described in Fig. 3. The negative sign refers to inhibition of stimulation compared to controls.

GANGLIOSIDES

MODULATE

T-CELL RESPONSE TO IL-2

415

TABLE 2 The Binding of Gangliosides to IL-2

cm Sample rIL-2

58,422 + 4594

Dialyzed sample rIL-2 rIL-2 + GM3 rIL-2 + GD3

52,254 r 3939 5,953 -t 1439 7,990 z!z1670

Note. Solutions of rIL-2 (20 units/ml) with or without gangliosides (25 r&ml) in RPM1 1640were mixed and dialyzed for 24 hr using a dialysis tubing with a 14-12,000 MW cutoff. The dialyzed samples were removed and tested for activity on IL2dependent T cells in a standard proliferation assay. The cpm in the proliferation assay are the means of quadruplicate samples. The stimulation of cells with IL2 alone, 20 units/ml (nondialyzed), is shown as a control. In comparison of rIL-2 dialyzed to nondialyzed there was no statistical significance. The comparison of rIG2 dialyzed vs rIL-2 + GD3 or GM3 dialyzed was highly significant (P < 0.001). These are representative experiments.

(the most inhibitory gangliosides) to IL-2, we incubated them with IL2 and then removed unbound gangliosides by a 24-hr period of dialysis. The gangliosides used are lessthan 4000 MW, and therefore were easily dialyzed out from the dialysis tube. IL-2 is approximately 15,000 MW and is not dialyzable from this dialysis tube. The retained solution was examined for IL-2 activity on IL-2-dependent T cells. Relevant controls included dialyzed IL-2 alone. IL-2 treated with GM3 and GD3 lost greater than 80% of its stimulatory activity, indicating that these gangliosides do bind to IL2 (Table 2).

Preincubation of T Cells with GangliosidesbeforeIL-2 Stimulation To determine whether ganglioside-induced alterations in cell surface membrane activity change IL-2 responsiveness,we treated lymphocytes with inhibitory concentrations of gangliosides before and after addition of rIL-2. Incubation with individual gangliosides for 12 to 24 hr before addition of rIL-2 significantly reduced the lymphocyte response to rIL-2 (Fig. 4). The simultaneous addition of individual gangliosides and rIL-2 also inhibited proliferation but to a lesserextent than preincubation of cells with gangliosides before rIL-2. Control glycolipid molecules (N-acetylneuraminic acid, ceramide trihexoside, and sphingosine) had no significant inhibitory effects. When gangliosides were added 12 to 24 hr after exposure to rIL-2 they did not significantly inhibit the lymphocyte response to IL-2 (Fig. 4).

GangliosidesBlocking IL-2 Receptor We then examined the effect of individual gangliosides on the IL-2 receptor Tat. Lymphocytes were stimulated with PHA ( 10 pg/ml) for 5 days to increase Tat expression, incubated with and without gangliosides (GM3 and GD3), stained with fluorescein-labeled antibody to Tat (Becton-Dickinson, Mountain View, CA), and analyzed by flow cytometry. There was no evidence that gangliosides bound to or blocked Tat.

416

HOON, IRIE, AND CGCHRAN GANGL IOSI DES

A;obB NANA CTH GD3 GO2

E 00 60

40

GM3 GM2

20

20

40

60

00

FIG. 4. Effect of preincubation of T cells with gangliosides on IL-2 responsiveness.(A) Individual purified gangliosides derived from melanoma were added to IL-2dependent T cells (10’ cells per well in 96-well microplates) and incubated for 24 hr at 37’C, after which 20 units rIL-2 was added (A). Cells were in culture for 72 hr and then pulsed during the last 18hr with [‘H]TdR. (B) The reverse of this experiment was carried out by incubating IL-2-dependent T cells with rIL-2 (20 units) for 24 hr, after which gangliosides were added. Culture time and pulse times were as described above(B). Purified gangliosides (GM2, GD2, GM3, and GD3) derived from melanoma were used at 25 &ml. For controls GM 1, NANA (N-acetylneuraminic acid), and CTH (ceramide trihexoside neutral glycolipid) were used at 25 &ml. The lymphocyte response to IL-2 in the absence of glycolipids was used as a control. Results are expressed in [‘H]TdR uptake in cpm. In (A) GM2, GM3, GD2 and GD3 only gave significant inhibition to controls (P < 0.05). In (B) no significant inhibition was found with any of the molecules. Preincubation of cells with rIL-2 or ganglioside for 12 hr gave results identical to those for 24 hr.

DISCUSSION Melanoma gangliosides were used in these studies since they are the sphingoglycolipids to which patients’ lymphocytes are exposed, and it has been suggestedthat tumor-derived gangliosides differ chemically from normal gangliosides. Purified whole melanoma gangliosides gave both stimulatory and inhibitory activities against unseparated lymphocytes and IL-Zdependent T-cell lines. These “whole” gangliosides were usually more stimulatory to IL-2-dependent T-cell lines than fresh peripheral blood lymphocytes. The latter population of cells contains macrophages and other T and non-T cells not responsive to IL-2 that may have inhibitory effects on IL-2 stimulation of responsive T cells. In examining purified individual gangliosides we found a different effect among the gangliosides, some stimulating and others inhibiting. The addition of individual gangliosides to K-2-dependent cell lines in the absence of IL-2 did not significantly alter lymphocyte proliferation. However, the addition of GM2 and GD2 to freshly isolated melanoma TDLN lymphocytes did stimulate proliferation in the absenceof rIL-2, with the degree of proliferation varying from patient to patient. Lymphocytes from human breast cancer-draining lymph nodes did not respond to exogenous GM2

GANGLIOSIDES

MODULATE

T-CELL RESPONSE TO IL-2

417

and GD2 in the way that melanoma nodal lymphocytes did. Hence this effect may be due to a preexisting immune response to GM2 and GD2 in melanoma patients. Breast tumors express little GM2 and GD2 compared to melanoma (unpublished data). The inhibitory activity of certain gangliosides such as GM3 and GD3 may be one mechanism of immune suppression in melanoma-bearing hosts. Gangliosides GM3 and GD3 are shed by melanoma and their serum levels were elevated in melanoma patients (10). Also in melanoma patients there is a significant increase in binding of these gangliosides to red blood cells (10). Similarly, these gangliosides can bind to lymphocytes, a potential mechanism for immunosuppression of lymphocyte activation. We investigated whether the inhibitory action of GM3 and GD3 on lymphocyte activation may be via neutralization of IL-2 or acting on TAC. Other investigators have reported preliminary evidence that gangliosides can bind to IL-2 (22). The inhibitory activities of the GM3 and GD3 may be responsible for neutralizing free IL2 in vivo, especially in melanoma patients with advanced disease,thereby inhibiting IL-2-associated immune reactivities. This may be of importance since IL-2 given systemically for the treatment of melanoma could be neutralized to a degree,dependent on the amount of GM3 and GD3 in the patients’ blood. Also GM3 is a normal component of many types of human tissues and may act as a natural inhibitor of free IL2 in tissue fluid and blood. Presensitization of gangliosides before and after IL-2 treatment gave differential responses.Lymphocytes exposed to gangliosides prior to IL-2 responded significantly less to IL-2. These studies strongly support the hypothesis that gangliosides can act at the level of the cell surface membrane. However, in our studies and those of others, gangliosides do not appear to inhibit lymphocyte responsesto IL-2 by directly blocking TAC (23). However, T lymphocytes have high-, intermediate, and low-affinity IL-2 receptors, the first apparently being more important physiologically (16). On the basis of our studies we cannot exclude the possibility of interactions between highaffinity IL-2 receptors and individual gangliosides. This study shows that different molecules of gangliosides from human melanoma can respectively inhibit and enhance the response of human T cells to IL-2. Although others have demonstrated various gangliosides found in brain and other tissues have inhibitory effects (24) we have examined the tumor-related gangliosides to which patients’ lymphocytes are exposed. This distinction is very important since tumor-derived gangliosides have been shown to differ from normal gangliosides. Gangliosides are certainly active at the cell surface membrane as well as against the IL-2 molecule itself, the effect varying with the concentration and type of ganglioside examined. Our studies and those of others suggestthat individual gangliosides have different effects on cell growth and that they may modulate specific growth factor receptors (25-27). The modulation of growth factor receptor responses appears to be related to the glycosphingolipid environment of the cell membrane (26, 27). Gangliosidereceptor interactions may shift receptor affinity or accessibility, causing alterations in membrane cytoplasmic events. The inhibitory effect of GM3 and GD3 appears to be due to the presence of the sialic acid group and its position on the molecule, since asialyated GM3 and GD3, which are ceramide trihexoside, had no inhibitory effect. We have shown previously that the sialic acids on the sugar molecules (galactose) of gangliosides contribute to their immunogenicity and reactivity with specific antibodies. The major difference between GM2 and GD2 relative to GM3 and GD3 is the

418

HOON, IRIE, AND COCHRAN

presence of the terminal N-acetylgalactosamine group. This terminal sugar presented on the ganglioside oligosaccharide chain may contribute to the stimulatory activity of the molecules. We have shown that GM2 and GD2 are immunogenic in humans (4,6). The enhanced lymphocyte response to IL-2 induced by GM2 and GD2 may reflect previous in vivo immune stimulation, Recently, we have demonstrated that some human natural killer cells can recognize GM2 as a target structure (28). The presence of GM2 on neoplastic cells may be immunogenic and able to activate cell-mediated immunity. Human melanoma early in its progression induces T-cell responses but later suppressesthese responses (8). It is known that primary melanomas have associated abundant lymphocytic infiltrates whereas melanoma metastasesseldom show prominent lymphocyte infiltrates (29). In normal human melanocytes GM3 is a major ganglioside component, GD3 a minor component, and GM2 and GD2 are minimum (2). By contrast, in melanoma cells, lesser quantities of GM3 and greater quantities of GD3 as well as low but significant quantities of GM2 and GD2 are expressed (l-3). However, the composition of these four gangliosides on melanoma is widely heterogenous (6). The particular combination and amount of gangliosides expressed on melanoma cell surfaces may be critical in determining whether immune stimulation or suppression occurs. The shedding of immunosuppressive gangliosides in vivo, particularly in the presence of tumor lymphocyte infiltrates, could incapacitate host immunity. The type of immunomodulation induced may be of great significance in determining the rate and extent of tumor progression. Recently, human monoclonal antibodies to the immunogenic gangliosides GD2 and GM2 were developed in our laboratory (4, 30) and anti-GD2 antibody has successfully been used to treat cutaneous metastasesof melanoma in man (31). These studies indicate that ganglioside expression in human melanoma has a major influence on the patient’s immune response and that the development of a specific individualized therapy that takes into account the ganglioside profile ofthe patient’s tumor may be therapeutically effective. ACKNOWLEDGMENTS We thank the testing core facility ofthe Jonsson Comprehensive Cancer Center, UCLA, for FACs analysis, and Drs. D. L. Morton and S. H. Golub for critical reading of the manuscript and helpful discussions. This research was supported by Grants CA12582, CA36034, CA42396, and CA30647 from NIH.

REFERENCES 1. Portokoulian, J., Zwingelstein, G., and Dore, J. F., Eur. .I. Biochem. 94, 19, 1979. 2. Carubia, J. M., Yu, R. K., Macala, L. J., Kirkwood, J. M., and Varga, J. M., Biochem. Biophys. Rex Commun. 120,500,1984. 3. Tsuchida, T., Saxton, R. E., Morton, D. L., and Irie, R. F., J. Natl. Cancer Inst. 78,45, 1987. 4. Irie, R. F., Sze, L. L., and Saxton, R. E., Proc. Natl. Acad. Sci. USA 79,5666, 1982. 5. Watanabe, T., Pukel, C. S., Takeyama, Lloyd, K. O., Shiku, H., Li, L. T. C., Travassos, L. R., Oettgen, H. F., and Old, L. J., J. Exp. Med. 156, 1884, 1982. 6. Tai, T., Cahan, L., Tsuchida, T., Saxton, R. E., and Morton, D. L., ht. J. Cancer 35,607, 1985. 7. Tsuchida, T., Saxton, R. E., and Irie, R. F., Med. Immunol. 13,479, 1987. 8. Hoon, D. B. S., Bowker, R., and Cochran, A. J., CancerRex 47, 1529, 1987. 9. Morgan, A. C., Jr., Gallaway, D. R., Imai, K., and Reisfeld, R. A., J. Immunol. 126,365, 1981. 10. Portoukalian, J., Zwingelstein, G., Abdul-Malak, N., and Dore, J. F., Biochem. Riophys. Res. Commun. 85,916, 1978. 11. Herlyn, M., Guerry, D., and Koprowski, H., J. Immunol. 134,4226,1985. 12. Tai, T., Paulson, J. C., Cahan, L. D., and Irie, R. F., Proc. Natl. Acad. Sci. USA 80,5392, 1983.

GANGLIOSIDES

MODULATE

T-CELL RESPONSE TO IL2

419

13. Brooks, C. G., and Henny, C. S., Contemp. Top. Mol. Immunol. 10,63, 1985. 14. Morgan, D. A., Ruscetti, F. W., and Gallo, R., Science 193,1007, 1976. 15. Robb, R., J. Immunol. Today 5,203,1984. 16. Robb, R. J., Greene, W. C., and Rush, M. R., J. Exp. Med. 160, 1126, 1984. 17. Hoon, D. S. B., Kom, E. L., and Cochran, A. J., Cancer Res. 47, 1740, 1987. 18. Hoon, D. S. B., McBride, B., Irie, R. F., Jung, T., Naungayan, J., and Cochran, A. J., J. Leukocyte Biol. 40,255, 1986. 19. Ladisch, S., Ulsh, L., Gillard, B., and Wong, C., J. Clin. Invest. 74,2074, 1984. 20. Hakomori, S. I., Cancer Rex 45,2405, 1985. 2 1. Kannagi, R., Stroup, N. A., Cochran, N. A., Urdalm, D. L., Young, W. W., and Hakomori, S., Cancer Res. 43,4997, 1983. 22. Parker, J. G., Caldini, G., Krishnamurti, C., Ahrens, P. B., and Ankel, H., FEBS Lett. 170,39 1,1984. 23. Robb, R. J., J. Immune!. 136,971,1986. 24. Bremer, E. G., Hakomori, S. I., Bowen-Pope, D. F., Raines, E., and Ross, R., J. Biol. Chem. 259, 6818, 1984. 25. Spiegal, S., Fishman, P. H., and Weber, R. J., Science230, 1285, 1986. 26. Hakomori, S. I., Sci. Amer. 254,44, 1984. 27. Symington, F. W., and Hakomori, S. I., Lymphokines 12,201, 1985. 28. Ando, I., Hoon, D. B., Suzuki, Y., Saxton, R. E., Golub, S. H., and Irie, R. F., Znt. J. Cancer40, 12, 1987. 29. Elder, D. E., Ainsworth, A. M., and Clark, W. H., In “Human Malignant Melanoma” (W. H. Clark, L. I. Goldman, and M. J. Mastranglo, Eds.), p. 55. Grune & Stratton, New York, 1979. 30. Irie, R. F., Tai, T., and Morton, D. L., Zn “Basic Mechanisms and Clinical Treatment of Tumor Metastasis” (M. Torisu and T. Yoshida, Eds.), p. 37 1. Academic Press,New York/Tokyo, 1985. 3 1. Irie, R. F., and Morton, D. L., Proc. Natl. Acad. Sci. USA 83,8694, 1986.