Gangliosides shed by tumor cells enhance tumor formation in mice.

The role of tumor cell membrane gangliosides in tumor formation was probed using a series of cloned murine AKR lymphoma cell lines. Tumor formation was directly related to high expression and shedding of membrane gangliosides. In vivo, as little as 1 pmol of purified total gangliosides of highly tumorigenic cells, injected intradermally with poorly tumorigenic cells (which lacked and did not shed gangliosides), markedly increased the tumorigenicity of these cells in syngeneic normal mice. Thus, gangliosides shed by tumor cells are a previously unrecognized, extremely potent enhancer of tumor formation in vivo.

Modulation of the immune response by gangliosides. Inhibition of adherent monocyte accessory function in vitro.

Gangliosides are potent inhibitors of lymphoproliferative responses. Selectively greater inhibitory effects of gangliosides on antigen-induced (vs. mitogen-induced) proliferation have been documented; e.g., 50 nmol of highly purified bovine brain gangliosides (BBG)/ml caused greater than or equal to 87% inhibition of proliferative responses of human peripheral blood mononuclear cells (PBMC) to three soluble specific antigens (Candida, streptokinase-streptodornase, and tetanus toxoid) vs. less than or equal to 37% inhibition of responses to three nonspecific mitogens (phytohemagglutinin, concanavalin A, and pokeweed mitogen). The possibility that BBG interfere with adherent monocyte accessory function, upon which responses to soluble specific antigens are strictly dependent, was therefore considered. PBMC were separated into the adherent and nonadherent subpopulations, exposed to BBG, recombined, and their proliferative responses were measured. Unseparated PBMC preincubated for 48-72 h with 100 nmol BBG/ml and then washed to remove unbound BBG exhibited 73-76% inhibition of subsequent antigen-induced lymphoproliferation. Separate pretreatment of both adherent and nonadherent cell subpopulations in BBG under the same conditions resulted in similar (72-82%) inhibition, which was reproduced by preincubation of only the adherent cells in BBG. Preincubation of only the nonadherent cells in BBG was not inhibitory. Inhibition (a) was independent of whether gangliosides were added in solution or incorporated into liposomes, (b) was abrogated by adding untreated monocytes to cultures containing adherent cells that were preexposed to BBG (excluding the possibility that BBG was inducing suppression mediated by adherent cells), and (c) was reversible by further incubation of BBG-pretreated adherent cells in control medium. Together, these results delineate a mechanism by which gangliosides modulate lymphoproliferative responses--direct, noncytotoxic, and ultimately reversible inhibition of the accessory function of adherent monocytes.

Influence of cellular ganglioside depletion on tumor formation.

BACKGROUND: Gangliosides are immunosuppressive cell surface molecules that are often present in high concentrations in and shed actively by tumor cells. These molecules inhibit the antitumor immune response that is implicated in tumor rejection. We therefore determined the ability of tumor cells pharmacologically depleted of gangliosides to form tumors in mice. METHODS: We tested a ganglioside-rich subline of B16 murine melanoma, MEB4, and MEB4 cells that had been depleted of endogenous gangliosides by incubation with 0.5 microM 1-phenyl-2-hexadecanoylamino-3-pyrrolidino-1-propanol, a specific inhibitor of the enzyme glucosylceramide synthase. Tumor formation was assessed twice a week for 10 weeks after the intradermal injection of tumor cells, and metastatic potential was assessed 4 weeks after tail vein injection of tumor cells. All P values are from two-sided tests. RESULTS: Reduction of the ganglioside content of MEB4 cells, which was not cytotoxic to cells and did not inhibit cell proliferation in vitro, markedly reduced their ability to form tumors. Only 40% of the mice given an intradermal injection of 10(5) ganglioside-depleted MEB4 cells developed tumors compared with 100% of the mice given an injection of 10(5) control MEB4 cells (P<.001). Ganglioside depletion also reduced metastasis: A mean of five pulmonary metastases was detected per mouse given an injection of 2 x 10(5) ganglioside-depleted MEB4 cells compared with a mean of 25 per mouse given an injection of 2 x 10(5) control MEB4 cells. CONCLUSION: Tumor cells with a pharmacologically decreased concentration of gangliosides produce fewer tumors in mice than do untreated cells, suggesting that pharmacologic depletion of gangliosides should be explored further as a therapeutic approach to cancer.

Abrogation of shedding of immunosuppressive neuroblastoma gangliosides.

Shedding of tumor cell gangliosides may contribute to tumor cell escape from host immune destruction. Thus, it would be of interest to block the shedding of these immunosuppressive molecules. To this end, we studied a ceramide analogue, D-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (D-PDMP). D-PDMP is a potent inhibitor of glucosylceramide synthase and thereby the synthesis of cellular glycosphingolipids. Exposure of LAN-5 human neuroblastoma cells to 10 microM D-PDMP for 5 days almost completely abolished the shedding of gangliosides (from 240 to 8 pmol/10(8) cells/h), whereas cellular ganglioside synthesis was reduced by 90%. A shorter (3-day) treatment of LAN-5 cells with 10 microM D-PDMP was already effective in inhibiting shedding (by 86%) even while the cellular ganglioside content was still high. Specificity was evidenced by the only minimal effect of D-PDMP on the synthesis of sphingomyelin and phosphatidylcholine. Therefore, certain pharmacological agents, such as D-PDMP, may be useful in abrogating tumor ganglioside shedding and its consequent biological effects in vivo.

Human tumor gangliosides inhibit murine immune responses in vivo.

Gangliosides which are shed by tumor cells clearly inhibit cellular immune responses in vitro. However, the immunosuppressive activity of these molecules have been more difficult to ascertain in vivo. Here we have adapted a murine model to determine the effects of tumor gangliosides in an in vivo microenvironment, the lymph node draining the site of stimulation by allogeneic cells. In this model, allogeneic splenocytes (BALB/c) are s.c. injected into C3H mice. The cellular immune response in the draining popliteal lymph nodes 4 days later is evidenced as an increase in lymph node mass (2-fold), lymphocyte number (6-fold), and lymphocyte DNA synthesis (6-fold). Purified human neuroblastoma gangliosides (10 nmol) coinjected with the stimulating allogeneic cells significantly suppressed this in vivo immune response. The increase in the lymph node mass was reduced by 65% (0.66 versus 1.89 mg), the increase in lymphocyte number (4.0 x 10(6) cells/node) was almost completely inhibited (1.1 x 10(6) cells/node), and in vitro [3H]thymidine uptake by the lymphocytes recovered in vivo was reduced by 80%. In contrast to the inhibition by tumor gangliosides, liposomes of cholesterol:lecithin were not inhibitory. Thus, tumor gangliosides, specifically, modulate cellular immune responses in vivo, which may contribute to the observed enhancement of tumor formation by these molecules.

Localization of shed human tumor gangliosides: association with serum lipoproteins.

Tumor gangliosides are biologically active (immunosuppressive and tumor-enhancing) cell surface molecules which are shed into the circulation in vivo. However, the mechanism of transport of these molecules (i.e., in solution or bound to proteins or other lipids) is not known. To resolve this question we traced, by direct chemical detection, the serum localization of a specific human tumor ganglioside, GD2, shed by neuroblastoma cells. Sera from patients with tumors were separated into the lipoprotein fractions [very low-density lipoprotein, low-density lipoprotein (LDL), and high-density lipoprotein] and lipoprotein-depleted serum. All three lipoprotein fractions contained GD2. 73% of the total GD2 was present in the LDL fraction, while very low-density lipoprotein and high-density lipoprotein contained 21 and 6%, respectively. Significantly, lipoprotein-depleted serum, which would contain both albumin-bound and free ganglioside, was devoid of GD2. Thus, shed neuroblastoma tumor gangliosides are exclusively associated with the serum lipoprotein (and predominantly LDL) fractions in vivo. These findings have implications for the immunological detection of these molecules and the development of approaches to their removal.

Expression of GD2 ganglioside by untreated primary human neuroblastomas.

Primary neuroblastomas obtained before therapy from 36 patients were studied to determine the frequency of tumors expressing a specific glycosphingolipid, GD2 ganglioside. Total tissue gangliosides were purified by a new partition method, quantitated, and analyzed by high-performance thin-layer chromatography. All 36 neuroblastoma tumors, representing all clinical stages, contained GD2 ganglioside. The mean relative and absolute concentrations of GD2 were substantial (12% of the total tissue gangliosides and 50 nmol/g of tissue) and were independent of the clinical stage of the tumor. In contrast, 6 samples of related but more differentiated tumors (ganglioneuroblastoma and ganglioneuroma) had little or no detectable GD2 (less than or equal to 1.5% of total gangliosides and less than or equal to 4 nmol/g of tissue). These results suggest that GD2 is a sensitive marker for neuroblastoma tissue and may be an excellent target antigen for immunotherapy of this tumor.

Shedding and immunoregulatory activity of YAC-1 lymphoma cell gangliosides.

YAC-1 lymphoma cells, both when cultured in vitro and when passaged in ascites form in vivo, synthesize gangliosides (means of 22.1 and 14.7 nmol lipid-bound sialic acid isolated per 10(8) cells, respectively) with potent inhibitory effects on mitogen- and antigen-induced lymphoproliferation: 10 to 30 nmol highly purified YAC-1 gangliosides/ml caused greater than 90% inhibition of proliferative responses of murine lymphocytes to concanavalin A, lysozyme (a soluble specific antigen), and allogeneic cells (mixed-lymphocyte response). Measureable quantities of these gangliosides were shed by the tumor cells in vitro and also were recovered from the ascites fluid in vivo. Furthermore, the gangliosides isolated from ascites fluid (mean of 15.3 nmol/ml) had inhibitory activity of a magnitude similar to that of the gangliosides isolated from the tumor cells. Therefore, significant inhibition of normal lymphoproliferative responses by tumor-derived gangliosides occurred at ganglioside concentrations which are actually present in the fluid surrounding the tumor cells in vivo. These results support the hypothesis that shedding of gangliosides may serve to protect tumor cells from host immune destruction.

Acceleration of myeloid recovery from cyclophosphamide-induced leukopenia by pretreatment with Bacillus Calmette-Guérin.

Treatment of C57BL/6 mice with Bacillus Calmette-Guérin i.p. 8 days prior to the induction of leukopenia by cyclophosphamide (300 mg/kg i.p.) significantly (p less than 0.002) increased peripheral granulocyte counts on each day during the recovery from leukopenia. The recovery of lymphocyte counts was unaffected by Bacillus Calmette-Guérin treatment. Further experiments indicated that the accelerated granulocyte recovery was the result of an earlier initiation of the recovery process rather than of the release of stored granulocytes. Bacillus Calmette-Guérin may have clinical value as a stimulator of myelopoiesis in patients rendered leukopenic by antineoplastic chemotherapy.

Bacillus Calmette-Guérin enhancement of colony-stimulating activity and myeloid colony formation following administration of cyclophosphamide.

Pretreatment of mice with Bacillus Calmette-Guérin (BCG) enhances recovery of the peripheral granulocyte count from cyclophosphamide (CTX)-induced leukopenia. In the present study, in vitro bone marrow culture was used to assess the effect of BCG pretreatment on serum levels of myeloid colony-stimulating activity and on regeneration of bone marrow myeloid colony-forming units (CFU-c) following CTX. C57BL/6 mice received either BCG or 0.9% NaCl solution i.p. 8 days prior to the administration of CTX (300 mg/kg i.p.). Following CTX, peak serum levels of myeloid colony-stimulating activity were strikingly higher in the BCG-pretreated mice [1320 +/- 30 (S.E.) units] than in the 0.9% NaCl solution-pretreated mice (764 +/- 37 units). BCG pretreatment also resulted in higher numbers of marrow CFU-c during recovery from CTX (e.g., 43.9 +/- 1.8 versus 20.0 +/- 0.9 myeloid colonies/10(5) marrow cells, Day 3 post-CTX). However, neither the rate of decline nor the absolute nadir of CFU-c, nor CFU-c cell cycle characteristics were affected by the pretreatment. The initial effect of i.p. BCG pretreatment on recovery from CTX-induced granulocytopenia is an augmentation of serum myeloid colony-stimulating activity which precedes the enhanced regeneration of CFU-c and the accelerated recovery of the peripheral granulocyte count.

Gangliosides block antigen presentation by human monocytes.

Gangliosides, immunosuppressive molecules shed by tumor cells, are potent inhibitors of monocyte accessory cell function. However, the specific monocyte cellular defect caused by gangliosides is unknown. Here we sought to delineate whether this abnormality is in the induction of suppressor cells, in intracellular antigen processing, or in intercellular antigen presentation. Three sets of studies of the tetanus toxoid (TT)-induced lymphoproliferative response, which is dependent upon monocyte accessory function, address this issue: (1) Antigen (TT)-primed human monocytes incubated with 50-100 microM human brain gangliosides for 24-48 h, washed, and then combined with T-cells, were inhibited in triggering T-cell proliferation, showing that the effect was occurring after antigen processing was complete. (2) T-cell responses to immobilized anti-CD3 or to antigen-primed control monocytes in the presence of ganglioside-exposed monocytes were unaffected, showing that ganglioside-exposed monocytes did not act as suppressor cells. (3) Stimulation by TT peptide fragment 830-843, which does not require processing, was completely inhibited by exposure of monocytes to gangliosides. These findings identify ganglioside interference with monocyte accessory cell function at the level of antigen presentation. We conclude that tumor gangliosides may inhibit host anti-tumor cellular immune responses by preventing the effective cellular interactions of the antigen-primed monocyte with the responding T-lymphocyte.

Shedding of human neuroblastoma gangliosides.

Gangliosides shed by tumor cells are implicated in modulating tumor formation. For example, rapid progression of human neuroblastoma tumors is associated with high circulating levels of shed GD2 ganglioside. To elucidate the kinetic and qualitative characteristics of tumor cell ganglioside shedding, which is difficult to accomplish in vivo, we examined this process in LAN-5 human neuroblastoma cells in vitro. Three major gangliosides, GD2, GM2 and GT1b, comprise 82% of the mean total of 38 nmol LBSA/10(8) cells. These molecules are shed very rapidly (1-3 pmol/10(6) cells per h, or approx. 0.5% of the total cell gangliosides per h). Quantitative and qualitative characteristics of ganglioside shedding are remarkably constant over a 40-fold range of cell density. Not only GD2, but every major carbohydrate species is shed, in proportion to its concentration in the cell, with slightly greater shedding of ceramide subspecies containing shorter chain fatty acids. These findings were confirmed in three other neuroblastoma cell lines, LAN-1, IMR and KCNR. We suggest that the high expression and rapid generalized shedding of human neuroblastoma tumor gangliosides results in significant in vivo accumulation of these biologically active molecules during tumor growth.

Immunosuppression by human gangliosides: I. Relationship of carbohydrate structure to the inhibition of T cell responses.

Causes of cellular immunodeficiency frequently associated with cancer remain poorly understood. One possible mechanism is tumor cell membrane shedding of immunosuppressive molecules, such as the sialic acid-containing glycosphingolipids, gangliosides. To explore this interesting hypothesis and establish structure-activity relationships, we examined the effects of a series of highly purified human gangliosides on T cell function. In all, ten individual molecular species of two major biosynthetic pathways were compared for their ability to inhibit human T cell proliferative responses. They include GM1, GD1a, GD1b, and GT1b (the predominant normal brain species), and GM4, GM3, GM2, GD3, GD2 and GQ1b. Strikingly, each HPLC-purified molecule, from the simplest monosialoganglioside to the most complex polysialoganglioside, had potent inhibitory activity; even the ganglioside with the most elemental carbohydrate structure (GM4, one sialic acid linked to a monosaccharide) strongly inhibits T cell proliferative responses to tetanus toxoid (ID90 = 1.5 microM). The data also reveal a complex interplay between elements of oligosaccharide structure in determining immunosuppressive activity. Sialic acid is critical to maximal activity, and (i) immunosuppression is most potent in gangliosides containing a terminal sialic acid. (ii) Total desialylation almost abolishes activity and (iii) partial alteration (lactone formation) reduces activity. (iv) Activity is generally but not always higher with higher numbers of sialic acid residues/molecule, and (v) some larger neutral glycosphingolipids retain measurable immunosuppressive activity. Overall, the potent inhibition by gangliosides supports the hypothesis that shedding of these molecules by tumors creates a highly immunosuppressive microenvironment around the tumor, thereby inhibiting the function of infiltrating host leukocytes and contributing to diminished T cell responses in cancer.

Biosynthesis and shedding of murine lymphoma gangliosides.

Ganglioside biosynthesis and subsequent shedding are a potential mechanism contributing to tumor cell escape from the host immune response. As a first step in identifying active molecular species, structural characterization and quantification of the purified individual cellular and shed gangliosides of YAC-1 murine lymphoma cells were undertaken. These studies uncovered three striking changes in ganglioside metabolism in cells passaged in vivo, compared with cells cultured in vitro. (i) Marked inhibition of GalNAcGM1b synthesis: GM1b was present in an equal proportion to its biosynthetic product GalNAcGM1b in vitro, but was present in a 6-fold higher concentration in vivo. (ii) Marked inhibition of NeuGc synthesis: NeuGc, present in vitro in an up to 7-fold higher concentration than its biosynthetic precursor NeuAc, was decreased in relative concentration in vivo (1:1). (iii) Selectivity of shedding: ganglioside shedding in vitro was generalized with respect to both carbohydrate structure and ceramide structure (mainly d18:1-C24:1 and d18:1-C16:0), while in vivo, there was selective shedding of gangliosides containing NeuGc and the shorter chain fatty acid. The reduced synthesis of NeuGc and of GalNAcGM1b in vivo, and the selective shedding of more polar ganglioside species, also in vivo, show that the extracellular environment can markedly affect cellular ganglioside metabolism.

Natural forms of shed tumor gangliosides.

Gangliosides shed by tumor cells are immunosuppressive molecules, but the mechanisms of shedding are poorly understood. We therefore conducted a comprehensive study of shedding to identify the natural forms of shed gangliosides. By chemical detection and mass spectrometric analysis of the gangliosides of YAC-1 murine lymphoma cells, we first confirmed that all major ganglioside species are released. Then, by the combination of metabolic and cell surface radiolabeling, we further demonstrated that gangliosides are released directly from the cell plasma membrane, i.e. by shedding. Ultracentrifugation separated the conditioned medium of metabolically radiolabeled cells cultured in either serum-free or serum-containing medium into: (1) a pellet of 100-200 nm membrane vesicles (visualized by electron microscopy) containing nearly one-third of total shed gangliosides; and (2) the supernatant, which contained soluble gangliosides (two-thirds of the total shed gangliosides). Although the ganglioside concentration in the conditioned medium (6-14x10-8 M) was above the critical micelle concentration of purified YAC-1 gangliosides (<1x10-8 M), by gel filtration >90% of the soluble gangliosides were found in monomeric form (MW <2 kDa) and only <10% in micelles (130 kDa). Ultrafiltration of fresh conditioned medium likewise showed the existence of monomers, and the findings were confirmed in human Daoy medulloblastoma and mouse MEB4 melanoma cells. Thus, in their natural states, shed tumor cell gangliosides exist in three forms: membrane vesicles, micelles, and monomers.

Tumor gangliosides enhance alpha2 beta1 integrin-dependent platelet activation.

Gangliosides, sialic acid-containing glycosphingolipids, enhance platelet adhesion to collagen and consequent platelet activation. For example, gangliosides shed by neuroblastoma tumor cells (NBTG) added to a subthreshold (non-activating) concentration (1 microgram/ml) of collagen, cause platelet aggregation (59 +/- 10%) and ATP release (2.3 +/- 0.2 nmol) equivalent to that caused by 10 micrograms/ml collagen alone. Here we report further studies to characterize this effect. Platelet aggregation and ATP release were not induced by NBTG in combination with subthreshold concentrations of adenosine diphosphate, epinephrine, thrombin or arachidonic acid, suggesting that NBTG specifically influences collagen-mediated platelet activation. Maximal platelet aggregation and ATP release required extracellular magnesium and only a short (1 min) preincubation with NBTG, suggesting a collagen receptor-mediated mechanism of this ganglioside activity. Since gangliosides interact with several integrin receptors, we determined whether NBTG influences alpha 2 beta 1, a major integrin collagen receptor on platelets. Incubation of platelets with a monoclonal antibody directed against the alpha 2 chain (5E8) blocked the increase in platelet aggregation (9 +/- 3% vs. 80 +/- 2%) and ATP release ( < 0.2 vs 2.5 +/- 0.1 nmol) induced by NBTG and 1 microgram/ml collagen. Incubation with an antibody to the non-integrin collagen receptor, CD36, or with an isotype control antibody did not abrogate the effect of NBTG. Finally, NBTG and its major component, GD2, enhanced alpha 2 beta 1-mediated platelet adhesion to immobilized collagen in an antibody 5E8-inhibitable manner. These findings implicate the alpha 2 beta 1-collagen interaction as a target of the effect of tumor-derived gangliosides.

Enrichment and localization of ganglioside G(D3) and caveolin-1 in shed tumor cell membrane vesicles.

Tumor cell ganglioside shedding has been implicated in the process of tumor formation. Previously, we identified three forms of tumor ganglioside shedding: micelles, monomers and membrane vesicles. Here, we have explored the membrane vesicle form of ganglioside shedding, using a newly identified human ovarian carcinoma cell line, CABA I. These cells synthesize and express a spectrum of gangliosides, including the disialoganglioside, G(D3). Immunostaining using the monoclonal antibody R24 confirmed G(D3) expression and its presence in the plasma membrane of these cells. Cellular gangliosides were detected in the culture supernatant by HPTLC autoradiography, confirming an active shedding rate of 3% of cellular gangliosides/24 h. CABA I cell membranes also express caveolin-1, a characteristic protein marker for caveolae, which was detected by flow cytometric analysis and by Western blotting in both the cell membranes and the isolated membrane vesicles. To further define the expression of G(D3) and caveolin-1, we used immunogold electron microscopy. This revealed localization of G(D3) in small clusters in the plasma membrane as well as enrichment and localization of ganglioside G(D3) and caveolin-1 in shed membrane vesicles, with 58-78% of vesicles carrying both G(D3) and caveolin-1. Together, these results suggest that membrane vesicle shedding originates in plasma membrane domains enriched in gangliosides and caveolin-1.

Modulation of human myelopoiesis by human gangliosides.

Cell surface gangliosides are potent modulators of cellular proliferation. We hypothesize that gangliosides shed by tumor cells modulate hematopoiesis and contribute to human tumor-associated suppression of hematopoiesis. To test this hypothesis, we determined the effects on myeloid colony formation by human bone marrow mononuclear cells of total gangliosides isolated from human brain and of seven highly purified individual ganglioside species (GM1, GM2, GD1a, GD1b, GD2, GD3, and GT1b). Total human brain gangliosides and certain individual species, GD1a, GD1b, and GT1b, significantly inhibited myeloid colony formation (number as well as size). The most complex molecules, GD1a, GD1b, and GT1b, were the most inhibitory, suggesting that the degree of inhibition is related to ganglioside structural complexity. To extend these findings, we also investigated certain tumor-derived (neuroblastoma) gangliosides, which we found inhibited both myeloid colony formation and 3H-thymidine incorporation by human bone marrow mononuclear cells. These data suggest a role for gangliosides, which are shed by proliferating cells, in the regulation of human hematopoiesis and may explain the bone marrow hypoplasia observed in association with many human malignancies.

Intercellular transfer of shed tumor cell gangliosides.

Three distinct steps underlie immumosuppression by tumor gangliosides: (i) their shedding by the tumor cell, (ii) binding to target leukocytes in the tumor microenvironment, and (iii) action upon the target cell. While shedding is well documented, cell to cell transfer of shed gangliosides is not. To address this, we employed a dual chamber culture system. In this system, metabolically radiolabeled lymphoma cells shed gangliosides into the conditioned medium of the contralateral chamber, which contained normal fibroblasts as the target cell. The shed lymphoma cell gangliosides bound avidly to the target fibroblasts in a trypsin-resistant manner (1-2 x 10(6) and 7 x 10(6) molecules/fibroblast in 24 and 48 h). Significantly higher than binding rates of purified lymphoma gangliosides added exogenously, these binding rates in a system which models the in vivo microenvironment suggest that cell to cell ganglioside transfer is a highly efficient process.

Detection of complex gangliosides in human amniotic fluid.

Gangliosides are possibly very potent immunosuppressive molecules. Here we show that human amniotic fluid contains high concentrations of a number of previously unnoted, structurally complex and highly polar gangliosides. These unusual molecules are present early in pregnancy (first trimester), increase in concentration with gestational age, and reach maximum levels (0.8 microM) at term. Since similar gangliosides have been detected in human placenta, trophoblast, and amnion, we suggest that these molecules are shed into the amniotic fluid bathing these tissues.