Cell differentiation and cell-cycle alterations by tyrosine kinase inhibitors in human melanoma cells.

Differentiation therapy is an attractive option for malignant melanoma, as traditional forms of chemotherapy seem to have little effect on this type of tumour. Among the several pathways for the experimental induction of differentiation of melanoma, we have focused on signal transduction mediated by protein kinases. We have examined the effects of calphostin C (a protein kinase C inhibitor), genistein and methyl 2,5-dihydroxycinnamate (tyrosine kinase inhibitors), and exogenous phosphotyrosine (an activator of protein tyrosine phosphatases) on the growth, morphology and differentiation of malignant melanomas in vitro. All four compounds tested were able to inhibit cell proliferation, but only genistein and methyl 2,5-dihydroxycinnamate were able to induce morphological changes, yielding a more dendritic or a rounder phenotype, respectively. The latter two drugs were also able to induce specific cell-cycle alterations, in contrast to calphostin C and phosphotyrosine. Melanin content was increased greatly in phophotyrosine treated cells and, to a smaller extent, in cells treated with genistein. RNA expression of specific genes encoding cytolytic T-cell antigens was not altered by the two tyrosine kinase inhibitors, in spite of the phenotypic changes observed. Together, these results suggest that tyrosine kinases are involved in cell cycle, growth, and differentiation pathways in malignant melanomas; however, these pathways may not be co-dependent. The results also suggest that these pathways may be sensitive to specific tyrosine kinase inhibitors or activators of protein tyrosine phosphatases.

Melanoma-reactive human cytotoxic T lymphocytes derived from skin biopsies of delayed-type hypersensitivity reactions induced by injection of an autologous melanoma cell line.

The expression by melanomas of multiple antigens that are recognized by specific MHC class I-restricted CTLs has been clearly demonstrated. The goal of many immunotherapy protocols being developed is, therefore, the induction and/or augmentation of CTLs specific for such antigens. One approach has been to immunize using irradiated autologous melanoma cells. Responses to this type of immunization and others are often subsequently measured by delayed-type hypersensitivity (DTH) reactions. The aim of this work was to characterize whether specific CTL responses occur at such DTH sites. Cutaneous DTH reactions were observed following injection of irradiated autologous melanoma cells expressing known tumor antigens. We isolated lymphocytes from biopsies of DTH reaction sites and could measure melanoma-specific CTL activity after 2-3 weeks of culture. The T-cell receptor-Vbeta repertoire of the cultured lymphocytes, assessed by flow cytometry, was highly skewed in both the CD4(+) and CD8(+) T-cell subsets. The repertoires were different among cultures derived from independent biopsies of simultaneous or subsequent DTH reaction sites and very different to that of fresh peripheral blood lymphocytes (PBLs) or PBLs cultured under the same conditions. No particular T-cell expansions dominated several DTH reaction sites, nor could they be detected in PBLs. It appears that T-cell responses to this type of immunization may be limited to the local microenvironment. Establishing the value of DTH reactions in determining levels of systemic antitumor immunity requires further investigation; however, such reactions may indicate a patient's competence to mount an antitumor immune response and enable the isolation of tumor-specific CTLs for use in tumor antigen identification.

Monoclonal antibodies against recombinant-MAGE-1 protein identify a cross-reacting 72-kDa antigen which is co-expressed with MAGE-1 protein in melanoma cells.

The MAGE-1 gene codes for tumor-associated peptides recognized by cytolytic T lymphocytes in association with MHC-class-1 molecules such as HLA-A1 and HLA-Cw16. In the course of a study aiming at the immunohistochemical detection of the MAGE-1 gene product in tumor samples, 2 mouse monoclonal antibodies (MAbs) directed against a full-length recombinant MAGE-1 fusion protein were found to react strongly not only with the 46-kDa MAGE-1 protein, but also with a 72-kDa product in immunoblots of lysates obtained from several MAGE-1-mRNA-positive melanoma cell lines. Pre-incubation of the antibodies with the recombinant MAGE-1 fusion protein abolished their reactivity both with MAGE-1 protein and with the 72-kDa product, thus confirming the occurrence of antigenic determinant(s) shared by the 2 proteins. The 72-kDa protein is not an alternative product of MAGE-1, since it was still detected in lysates of a MAGE-1 loss variant derived from a MAGE-1-positive melanoma cell line. Moreover, the 72-kDa protein does not appear to be a product of the other members of the MAGE gene family known to be expressed in tumors (such as MAGE-2, -3, -4 and -12). Interestingly, expression of the 72-kDa protein was found to be correlated with that of MAGE-1 protein. Thus, in 30 tumor cell lines analyzed by immunoblotting and RT-PCR, the 72-kDa protein was never detected in MAGE-1-mRNA-negative cell lines, while it was co-expressed with MAGE-1 protein in 12 out of 15 cell lines expressing MAGE-1. Furthermore, the 72-kDa protein was detected in lysates of human testis, the only normal tissue known to express MAGE-1. Finally, treatment of MAGE-1-mRNA-negative cell lines with 5-Aza-2'-deoxycytidine, a hypomethylating agent known to induce MAGE-1 expression, resulted in the expression of the 72-kDa protein. Taken collectively, these findings suggest that expression of the gene encoding the 72-kDa protein identified in this study through antigenic determinant(s) shared with MAGE-1 protein is regulated in a way similar to that of MAGE-1.

Expression of MAGE genes in primary and metastatic cutaneous melanoma.

Human genes MAGE-1 and MAGE-3 code for antigens that are recognized on melanoma cells by autologous cytolytic T lymphocytes. These antigens may constitute useful targets for specific anti-tumor immunization of cancer patients, since genes MAGE-1 and MAGE-3 are expressed in a number of tumors of different histological types, but are not expressed in normal adult tissues other than testis. This also applies to genes MAGE-2 and MAGE-4, which are closely related to MAGE-1 and MAGE-3. We have analyzed the expression of these 4 MAGE genes in cutaneous melanoma. Sixteen of 100 primary tumors vs. 69 (48%) of 145 metastases from individual patients expressed MAGE-1. Similar differences in the frequency of gene expression between primary and metastatic tumor samples were observed for MAGE-2, MAGE-3, and MAGE-4. MAGE expression in primary tumors was correlated with tumor thickness: there was a significantly increased frequency in the expression of MAGE-1, -2 and -3 in tumors of greater thickness. Benign and dysplastic nevi, as well as in situ melanomas, did not express any of the 4 MAGE genes.

Expression of type A and B tumor necrosis factor (TNF) receptors on melanoma cells can be regulated by dbc-AMP and IFN gamma.

Local administration of high-dose r-TNF alpha with IFN gamma in the limbs of melanoma patients has proved to be a very promising treatment. To understand the role played by the effect of TNF on melanoma cells in tumor destruction, we have investigated the expression of TNF-receptors in melanoma cells using monoclonal antibodies specific for the type-A (75-kDa) and the type-B (55-kDa) TNF receptors. Flow cytometric analysis of cultured melanoma cells indicated the presence of both types of receptor. Quantificative differences in the relative levels of receptors were observed for different cells lines, although the type-B receptor was generally more strongly expressed. Similar results were obtained by immunohistochemistry on cryosections from tumor samples. Positive staining of variable intensity was observed for the type-B TNF-receptor in a high percentage of tumor cells. The type-A TNF-receptor was also detected, but with a weaker staining. The total TNF-binding activity of cultured melanoma cells, as measured by binding of 125I-labeled TNF alpha, was up-regulated between 2- and 4-fold by incubation of cells with activators of protein kinase A or IFN gamma. Treatment of cultured melanoma cells with dbc-AMP resulted in a selective induction of type-A TNF-receptors, without affecting the type-B receptor level. In contrast, IFN gamma was able to induce either type of receptor in a cell-line-dependent fashion. Addition of TNF alpha to melanoma cells induced the activation of the nuclear transcription factor kappa B, as measured in an electrophoretic mobility shift assay, thus indicating the biological significance of the TNF-receptors on these cells.

Recognition of HLA-A1 by murine monoclonal antibodies.

Balb/c mice were immunized with cells from the mouse mastocytoma line P815 transfected with an HLA-A1 gene. The splenocytes of the immunized mice were fused with cells from the murine myeloma NS-1. In an initial screening, supernatants of growing cultures were tested for their binding capacity to the immunizing P815/A1+ cells as well as to P815/A2+ cells. Three out of 756 hybrids produced antibodies which bound to P815/A1+ cells only. They were cloned and further analyzed for their binding reactivity to reference B-lymphoblastoid cells from the Tenth International Histocompatibility Workshop. One monoclonal antibody, designated 6B11, reacted only with HLA-A1+ cells, while the two other antibodies, 3G3 and 7F10, appeared to detect antigenic determinants shared by HLA-A1, A3, A11, A26, and A30 (3G3) and by HLA-A1, A3, A11, A26, A28 and A30 (7F10). Flow cytometric studies on B-lymphoblastoid cell lines as well as on a series of tumor cell lines, including melanoma and colon carcinoma lines, confirmed the specificity of these antibodies. Monoclonal antibodies 7F10 and 6B11 were found to be of the IgM class and 3G3 of the IgG1 class. By complement-dependent 51Cr release experiments it was further shown that the two IgM antibodies 7F10 and 6B11 were able to lyse all cell lines of the HLA-A1 haplotype tested.

Expression of CALLA/CD10 on human melanoma cells.

The reactivity spectrum of an anti-CALLA/CD10 monoclonal antibody for cutaneous melanoma was analysed by immunohistochemistry in a series of lesions of different Breslow thickness. Similar proportions of small primary tumours, advanced primary tumours and metastatic lesions were found to express CALLA/CD10 (31-47%). However the proportion of stained cells within a given lesion increased with tumour progression. Up to 23% of the advanced primary lesions (> 3.0 mm) showed 26-50% cells stained with the anti-CALLA/CD10 antibody and up to 14% of the metastatic lesions showed 76-100% stained cells. The expression of CALLA/CD10 was further analysed in 15 ocular melanoma lesions of different histiotype. All five spindle type lesions, three of six epitheloid and two of five mixed type lesions stained positively with the anti-CALLA/CD10 antibody. The percentage of stained cells within a given lesion varied from 30% to 100%. A total of 63% of the ocular melanomas and 38% of the cutaneous melanomas tested expressed CALLA/CD10. Experiments with cultured melanoma cell lines showed that the surface expression of CALLA/CD10 can be modulated in vitro by treatment with interleukin 2 (IL-2) and an adenosine 3',5'-cyclic monophosphate (analogue).

Melanocyte tumor progression is associated with changes in angiogenesis and expression of the 67-kilodalton laminin receptor.

BACKGROUND: A number of experimental studies have substantiated changes in angiogenesis and in laminin/laminin-receptor interactions during tumorigenesis and tumor progression. However, these observations have never been verified objectively in tissues from a well-defined model of tumor progression. METHODS: Tissues from 97 proliferative lesions of the melanocyte lineage defining distinct steps in tumor progression were investigated immunohistochemically for changes in angiogenesis and expression of the laminin receptor (67-kilodalton molecule). RESULTS: Although the microvessel number was low in common nevi, it increased significantly in nevi with architectural disorder with varying degrees of melanocytic atypia (termed "nevi with ADMA"), and these changes persisted during transformation. Progression to primary melanomas was accompanied by a high microvessel number and progression to metastases by another significant increase. The number and diameter of microvessels were significantly higher at the lesion base than at the adjacent dermis of primary melanomas and higher in the lesion than in the adjacent tissue of metastatic foci. Expression of the laminin receptor, evaluated as percentages of positive lesions and positive cells per lesion, underwent upregulation in the course of progression. Changes in expression were associated mostly with nevi with ADMA, transformation, and deepening of the tumors into the dermis. CONCLUSIONS: These in situ data suggest that more frequent interactions between melanocytic cells and their microvasculature via adhesion protein laminin occur during tumor progression.

Effects of systemic interferon-alpha (IFN-alpha) on the antigenic phenotype of melanoma metastases. EORTC melanoma group cooperative study No. 18852.

In order to investigate the effects of in vivo treatment with interferon-alpha (IFN-alpha) on melanoma antigens, a clinical EORTC trial (No. 18852) was accompanied by an immunohistological study. Twenty patients with melanoma metastases of skin and soft tissues, eventually also of the lung, who were treated with systemic IFN-alpha, were evaluated for a comparison of metastases before (40) and during (42) treatment. Representative cryostat sections were studied immunohistologically with a panel of monoclonal antibodies against differentiation antigens (HMW-MAA, K-1-2, NKI-beteb, M-2-10-15), progression markers (transferrin receptor, ICAM-1, VLA-2), histocompatibility antigens (HLA-A, B, C, HLA-DR) and the proliferation-associated nuclear antigen Ki67. We found an overall reduction of the proliferation-associated antigen Ki-67 (p < 0.01), and an increase in expression of HLA-DR (p < 0.05) and ICAM-1 (trend) during treatment. The intensity of expression of HLA-A, B and C antigens as well as pigmentation (p < 0.01) was found to be increased. Early progression (< or = 8 weeks after onset of treatment) was associated with a lack of phenotypic changes. The data suggest an independent modulation of proliferation, pigmentation, and antigen expression by systemic treatment of metastatic melanoma with IFN-alpha.

Expression of IL-2 receptors in human melanoma cells.

Melanoma cells can secrete several cytokines and express various cell surface molecules, such as the intercellular adhesion molecule ICAM-1, class II histocompatibility antigens, and the CALLA antigen, typically found in cells of the immune system. We have investigated the possible expression of interleukin-2 (IL-2) receptors in melanoma using monoclonal antibodies specific for the p55/alpha chain (TAC antigen) and the p75/beta subunit. Flow cytometric analysis of cultured melanoma cells showed the presence of low levels of the TAC antigen and of the beta chain on the surface of several cell lines. Similar results were obtained in vivo by immunohistochemistry on cryosections prepared from cutaneous and ocular melanoma explants. Positive staining was observed for the alpha chain of the IL-2 receptor in a high percentage of tumour cells. The beta chain could also be detected, although in a limited number of specimens. Analysis of RNA from melanoma cell lines by Northern blot showed the presence of typical 4 Kb transcripts for the p75 subunit, while low-abundance message for the p55 chain could be detected using combined reverse transcription/polymerase chain reaction analysis. Together, these results suggest that melanoma cells may express high affinity receptors for IL-2.

Stimulation of human T cells by streptococcal "superantigen" erythrogenic toxins (scarlet fever toxins).

The pyrogenic (erythrogenic) exotoxins A and C (SPEA and SPEC) of Streptococcus pyogenes belong to the family of mitogenic toxins of which the staphylococcal enterotoxins are the prototypes. The erythrogenic toxin B (SPEB) is a proteinase precursor. All SPE have been reported to be superantigens. Here we have analyzed the human T cell response to these toxins. We used highly purified preparations of SPEA, SPEB, and SPEC from different S. pyogenes strains. These toxins were apparently homogenous in SDS-PAGE, IEF, and HPLC. In addition, recombinant SPEA and SPEC were produced in Escherichia coli. In cultures of PBMC, all three toxins expanded preferentially a fraction of T cells. Using mAb against V beta 2, -5, -6, -8, and -12, we investigated the phenotype of the stimulated cells. Natural SPEA, SPEB, and SPEC strongly stimulated V beta 8+ T cells, whereas recombinant SPEA and SPEC did not. Both natural and recombinant SPEA stimulated V beta 12+ cells and both natural and recombinant SPEC stimulated V beta 2+ cells. In accordance with these findings, a human V beta 8+ line responded to all three toxins derived from S. pyogenes but not to the recombinant proteins. An antiserum against natural SPEC neutralized specifically the V beta 2-stimulating activity of SPEC and the V beta 8-stimulating activity of all three toxins, but had no effect on the response to other superantigens. This shows that trace amounts of a potent novel V beta 8-stimulating activity not identical to SPEA and SPEC are responsible for the stimulation of V beta 8+ T cells by natural SPEA and SPEC reported previously. In a preliminary screening of S. pyogenes strains from patients, we found that this novel superantigen appears to be more widely distributed than SPEA and SPEC. Furthermore, we present evidence that also the superantigenic properties of SPEB are due to contaminations with this V beta 8 stimulator. The response to SPEB usually required 1000 times higher concentrations than to SPEA or SPEC. Antisera to SPEC but not to SPEB inhibited the response of PBMC and V beta 8+ Jurkat cells to SPEB. Furthermore, more stringent purification of SPEB yielded SPEB preparations devoid of mitogenic activity. These results indicate that the mitogenicity that is commonly attributed to SPEB is due to minute contaminations of the V beta 8 stimulator. These results raise two important caveats for the work with these highly potent T cell mitogens.(ABSTRACT TRUNCATED AT 400 WORDS)

Immunologic recognition of malignant melanoma by autologous T lymphocytes.

T lymphocytes specifically recognizing autologous tumor cells in vitro can be generated from melanoma patients. Recognition of tumor cells by both CD4 and CD8 lymphocytes is mediated through the T-cell receptor and is restricted by HLA antigens. Although HLA-A2 has been identified as a restricting allele for many melanoma-specific cytotoxic T lymphocytes, T cells directed against antigens unique to each patient's tumor as well as antigens common to melanomas from unrelated individuals can be restricted by several different HLA alleles. A common melanoma antigen recognized in association with HLA-A1 has now been identified. The antigen is a nonapeptide derived from the gene MAGE1, a normal cellular gene preferentially expressed in a variety of solid tumors. Melanoma cells have been found to produce a soluble form of the intracellular adhesion molecule-1. Soluble intercellular adhesion molecule-1 effectively inhibits cell-mediated cytotoxicity in vitro, raising the possibility that its expression in vivo could promote escape of the tumor cells from immune effectors.

Purification and characterization of murine lipopolysaccharide-binding protein.

The serum protein lipopolysaccharide (LPS)-binding protein (LBP) seems to play an important role in regulating host responses to LPS. Complexes of LPS and LBP form in serum and stimulate monocytes, macrophages, or polymorphonuclear leukocytes after binding to CD14. Previous reports have described the structure and properties of LBP from human and rabbit sera. Since mice are used in some experimental models of endotoxemia or gram-negative bacterial infections, information is needed about the properties of murine LBP. Murine LBP was purified by ion-exchange chromatography and high-pressure liquid chromatography; its NH2-terminal sequence (TNPGLVTRIT) was very similar to those of human and rabbit LBPs (80 to 90% amino acid identity). Murine LBP resembled LBPs from other species in that it promoted the binding of LPS to monocytes and enhanced the sensitivity of monocytes to LPS at least 100-fold. Mouse LBP, like rabbit and human LBPs, was found to be an acute-phase protein. Further in vivo studies with mice and anti-CD14 or anti-LBP reagents should help determine the role of LBP in response to LPS challenges.

Lectins and anti-T monoclonal antibodies-induced changes of second messengers generating enzymes in human peripheral blood mononuclear cells.

A five min. incubation of peripheral blood mononuclear cells (PBMN) with either phytohaemagglutinin (PHA) or concanavalin A (ConA) resulted in distinct subcellular redistribution patterns of phosphatidylinositol 4,5-bisphosphate phospholipase C [PtdIns(4,5)P2-PLC] and myo-inositol 1,4,5-trisphosphate monophosphatase [Ins(1,4,5)P3-monophosphatase] activities. When compared to control cells, PHA-treated PBMN cells displayed a significant increase of PtdIns(4,5)P2-PLC and Ins(1,4,5)P3-monophosphatase relative specific activities in the nuclear fraction along with an increment (D) in enzyme amount of 6.5% and 7.3%, respectively. Incubation with B66.6, an anti-CD4 monoclonal antibody (Mab) which specifically activates CD4(+)-T cells in the absence of any other stimuli, also induced changes of these activities in the nuclear fraction, thus mimicking the effect of PHA observed in helper T cell subpopulation. No changes were detected after incubation of PBMN cells with the non mitogenic anti-CD4 MAb 101-69, or with an anti-CD3 MAb which activates T cells only in the presence of a second stimulus. On the other hand, after incubation with ConA, PtdIns(4,5)P2-PLC relative specific activity was enhanced in the microsomal fraction while the Ins(1,4,5)P3-monophosphatase activity increased in both nuclear and microsomal fractions and decreased in cytosol. An increment D of 4.6% and 10.9% for PtdIns(4,5)P2-PLC and Ins(1,4,5)P3-monophosphatase, respectively, was measured in the microsomal fraction. Only after three days of incubation with a mitogenic anti-CD2 MAb Lau-2.1.2, the PtdIns(4,5)P2-PLC activity increased in the particulate fraction of PBMN similar to ConA treatment.

Expression and release of interleukin-1 by human glioblastoma cells in vitro and in vivo.

The present study demonstrates interleukin-1 (IL-1) production by human glioblastoma cells both in vitro and in vivo. The presence of IL-1 alpha and IL-1 beta transcripts was analyzed in 4 cell lines. IL-1 alpha mRNA was expressed constitutively in one cell line whereas constitutive IL-1 beta mRNA could not be detected in any of the cell lines. IL-1 alpha transcripts could be induced with phorbol myristate acetate (PMA) or PMA plus lipopolysaccharide (LPS) in 2 of 4 cell lines and IL-1 beta mRNA in 2 of 4 cell lines. Culture fluid from these cell lines was tested for the presence of IL-1 using a specific radio-immuno-assay for either IL-1 alpha or IL-1 beta. In agreement with the results on RNA, one of 4 cell lines was found to constitutively produce IL-1 alpha but not IL-1 beta. After treatment with PMA and LPS, IL-1 alpha was detected in the culture fluid from two other lines and IL-1 beta in the medium from three lines. That the IL-1 produced by these cell lines was biologically active was confirmed in a two step thymocyte proliferation assay. IL-1 like activity was detected in all samples that were positive in the radio-immuno-assay. Finally, immunohistological analysis on fresh frozen tumour sections provided evidence for IL-1 production by glioblastoma cells in vivo. Fourteen out of 28 glioblastomas were stained with an anti-IL-1 alpha monoclonal antibody while none of them was stained with an anti-IL-1 beta antibody.

Inhibition of phosphatidylserine synthesis induced by a CD4 mAb, B66.6 in Jurkat T cells.

A monoclonal antibody, B66.6, previously classified in the cluster of differentiation 4 (CD4), has been studied and compared with another CD4 monoclonal antibody, IOT4. It was found that B66.6 but not IOT4 was able to mobilize Ca2+ from intracellular stores in the Jurkat T cell line. Ca2+ mobilization was followed by a decrease in the extent of phosphatidylserine synthesis. In the presence of the phorbol ester, phorbol 12,13-dibutyrate, B66.6 induced interleukin-2 synthesis. Altogether, the results indicate that the CD4 monoclonal antibody, B66.6, mimics other T cell activators such as CD3 and confirm that the inhibition of phosphatidylserine synthesis in activated T cells follows the mobilization of Ca2+ from intracellular stores and is independent of the activation of the Ca(2+)-and phospholipid-dependent protein kinase C.

The differential reactivity of cells of the melanocytic lineage with four monoclonal antibodies against IFN-gamma inducible molecules.

The reactivity of four monoclonal antibodies (MAbs) directed against IFN-gamma inducible antigens with melanocytic cells was investigated in the course of local and systemic tumor progression of human malignant melanoma. Frozen sections of histologically defined melanocytic tissues at different stages of progression were stained with these MAbs using an indirect immunoperoxidase technique. The reactivity of MAbs Me15/B3 and Me15/F9, directed against two different epitopes of a 90-kDa molecule, was found to correlate with melanoma progression. Indeed, a significantly lower percentage of small than of advanced primary melanomas or metastases stained positively. A differential staining of nevocytic and dysplastic nevi was further observed for these two MAbs, which were also non-reactive with normal skin melanocytes. The reactivity of MAb Me14/D12, which identifies the intercellular adhesion molecule ICAM-1 and MAb Mel14/F12, directed against a 40-kDa molecule, was found to be independent of the Breslow thickness of primary melanomas. Both the latter MAbs stained a high proportion of nevocytic and dysplastic nevi. The co-expression of the surface molecules defined by MAbs Me14/D12, Me15/B3 and Me15/F9 in the course of melanoma progression was also analyzed. The frequency of this co-expression increased according to the Breslow thickness of primary melanomas. In addition, up to 100% of metastases, as opposed to 20% of dysplastic nevi, were found to be simultaneously stained by these three MAbs. It is therefore conceivable that high-risk melanocytic lesions might be identified by the use of a combination of MAbs directed against IFN-gamma regulated antigens.

Humoral immune response in disease-free advanced melanoma patients after vaccination with melanoma-associated gangliosides. EORTC Cooperative Melanoma Group.

Several studies have shown that melanoma-associated gangliosides are immunogenic in melanoma patients and that antibodies against them have a favorable prognostic effect. Our study aims at characterizing the humoral immune response in disease-free, advanced melanoma patients vaccinated with a total ganglioside fraction extracted from pooled metastases of human melanoma, containing as major gangliosides GM3 and GD3, and as minor ones GM2 and GD2. Prior to vaccination, all patients were made disease-free by surgical removal of skin, lymph-node or other distant metastases. Repeated vaccinations were carried out intradermally with gangliosides either in the native form in buffered solution, or in the form of liposomes. Serum samples were collected at regular intervals and assayed by ELISA for the presence of specific IgG and IgM antiganglioside antibodies. Selected samples were tested by immunostaining on thin-layer plates to specify the ganglioside species involved in the reactivity. Out of 32 evaluable patients, 17 presented a significant increase in antibody titer, mostly of the IgG isotype, which was maximal between 2 and 4 months after starting injections of gangliosides, and gradually disappeared within 1 year. No significant difference could be seen between the group of 20 patients treated with native gangliosides and the group of 12 vaccinated with the gangliosides in liposomes. All gangliosides seemed to be immunogenic, but GM2 and GD2 were somewhat more reactive. The disease-free intervals for the patients who showed an antibody response to the treatment were significantly higher (p less than 0001) than those of the non-responding group, as compared by the Kaplan-Meier method.