Hypoxia-dependent inhibition of tumor cell susceptibility to CTL-mediated lysis involves NANOG induction in target cells.

Hypoxia is a major feature of the solid tumor microenvironment and is known to be associated with tumor progression and poor clinical outcome. Recently, we reported that hypoxia protects human non-small cell lung tumor cells from specific lysis by stabilizing hypoxia-inducible factor-1α and inducing STAT3 phosphorylation. In this study, we show that NANOG, a transcription factor associated with stem cell self renewal, is a new mediator of hypoxia-induced resistance to specific lysis. Our data indicate that under hypoxic conditions, NANOG is induced at both transcriptional and translational levels. Knockdown of the NANOG gene in hypoxic tumor cells is able to significantly attenuate hypoxia-induced tumor resistance to CTL-dependent killing. Such knockdown correlates with an increase of target cell death and an inhibition of hypoxia-induced delay of DNA replication in these cells. Interestingly, NANOG depletion results in inhibition of STAT3 phosphorylation and nuclear translocation. To our knowledge, this study is the first to show that hypoxia-induced NANOG plays a critical role in tumor cell response to hypoxia and promotes tumor cell resistance to Ag-specific lysis.

LFA-1 and ICAM-1 expression induced during melanoma-endothelial cell co-culture favors the transendothelial migration of melanoma cell lines in vitro.

BACKGROUND: Patients with metastatic melanoma have a poor median rate of survival. It is therefore necessary to increase our knowledge about melanoma cell dissemination which includes extravasation, where cancer cells cross the endothelial barrier. Extravasation is well understood during travelling of white blood cells, and involves integrins such as LFA-1 (composed of two chains, CD11a and CD18) expressed by T cells, while ICAM-1 is induced during inflammation by endothelial cells. Although melanoma cell lines cross endothelial cell barriers, they do not express LFA-1. We therefore hypothesized that melanoma-endothelial cell co-culture might induce the LFA-1/ICAM ligand/receptor couple during melanoma transmigration. METHODS: A transwell approach has been used as well as blocking antibodies against CD11a, CD18 and ICAM-1. Data were analyzed with an epifluorescence microscope. Fluorescence intensity was quantified with the ImageJ software. RESULTS: We show here that HUVEC-conditioned medium induce cell-surface expression of LFA-1 on melanoma cell lines. Similarly melanoma-conditioned medium activates ICAM-1 expression in endothelial cells. Accordingly blocking antibodies of ICAM-1, CD11a or CD18 strongly decrease melanoma transmigration. We therefore demonstrate that melanoma cells can cross endothelial monolayers in vitro due to the induction of ICAM-1 and LFA-1 occurring during the co-culture of melanoma and endothelial cells. Our data further suggest a role of LFA-1 and ICAM-1 in the formation of melanoma cell clumps enhancing tumor cell transmigration. CONCLUSION: Melanoma-endothelial cell co-culture induces LFA-1 and ICAM-1 expression, thereby favoring in vitro melanoma trans-migration.

Down-regulation of mcl-1 by small interfering RNA sensitizes resistant melanoma cells to fas-mediated apoptosis.

Resistance of malignant melanoma cells to Fas-mediated apoptosis is among the mechanisms by which they escape immune surveillance. However, the mechanisms contributing to their resistance are not completely understood, and it is still unclear whether antiapoptotic Bcl-2-related family proteins play a role in this resistance. In this study, we report that treatment of Fas-resistant melanoma cell lines with cycloheximide, a general inhibitor of de novo protein synthesis, sensitizes them to anti-Fas monoclonal antibody (mAb)-induced apoptosis. The cycloheximide-induced sensitization to Fas-induced apoptosis is associated with a rapid down-regulation of Mcl-1 protein levels, but not that of Bcl-2 or Bcl-xL. Targeting Mcl-1 in these melanoma cell lines with specific small interfering RNA was sufficient to sensitize them to both anti-Fas mAb-induced apoptosis and activation of caspase-9. Furthermore, ectopic expression of Mcl-1 in a Fas-sensitive melanoma cell line rescues the cells from Fas-mediated apoptosis. Our results further show that the expression of Mcl-1 in melanoma cells is regulated by the mitogen-activated protein kinase/extracellular signal-regulated kinase (ERK) and not by phosphatidylinositol 3-kinase/AKT signaling pathway. Inhibition of ERK signaling with the mitogen-activated protein/ERK kinase-1 inhibitor or by expressing a dominant negative form of mitogen-activated protein/ERK kinase-1 also sensitizes resistant melanoma cells to anti-Fas mAb-induced apoptosis. Thus, our study identifies mitogen-activated protein kinase/ERK/Mcl-1 as an important survival signaling pathway in the resistance of melanoma cells to Fas-mediated apoptosis and suggests that its targeting may contribute to the elimination of melanoma tumors by the immune system.

The context of HLA-DR/CD18 complex in the plasma membrane governs HLA-DR-derived signals in activated monocytes.

HLA-DR-derived signals in activated monocytes mediate both pro-inflammatory cytokine production and caspase-independent death, and have been postulated to play a role in inflammation and in its resolution, respectively. Herein, using the monocytic/macrophagic human cell line THP-1 primed with IFNgamma (IFNgamma-primed THP-1), we investigated how HLA-DR may integrate both signals. Our inhibition studies demonstrated that if cell death is dependent on PKCbeta activation, the induction of TNFalpha gene expression relies on PTK activation, in particular the Src family of kinases, but both cell responses implicate the beta2-integrin CD18. Accordingly, sequential immunoprecipitation experiments demonstrated that following engagement of HLA-DR on IFNgamma-primed THP-1 cells, the HLA-DR/CD18 complex physically associates with PKCbeta and with PTK. Pharmacological disruption of lipid rafts microdomains abolished the assembly of HLA-DR/CD18/PTK signaling complex, HLA-DR-mediated tyrosine activation, and the PTK-dependent TNFalpha expression in IFNgamma-primed THP-1 cells. In contrast, HLA-DR/CD18/PKCbeta complex was still formed and able to mediate cell death after cholesterol depletion of these cells. These results indicate that while the integrity of lipid rafts is necessary for the transduction of cytokine gene expression through the HLA-DR/CD18 complex, it is not necessary for the induction of the HLA-DR/CD18-dependent cell death. Thus, our study provides experimental evidence indicating the compartmentalization of HLA-DR/CD18 complex within or outside lipid rafts as a mechanism through which HLA-DR can integrate both PTK and PKCbeta signals leading to activation and death, respectively, of activated monocytes. This might provide new insights into how MHC class II signaling may regulate inflammatory response.

Identification and Molecular Characterization of the Chloroplast Targeting Domain of Turnip yellow mosaic virus Replication Proteins.

Turnip yellow mosaic virus (TYMV) is a positive-strand RNA virus infecting plants. The TYMV 140K replication protein is a key organizer of viral replication complex (VRC) assembly, being responsible for recruitment of the viral polymerase and for targeting the VRCs to the chloroplast envelope where viral replication takes place. However, the structural requirements determining the subcellular localization and membrane association of this essential viral protein have not yet been defined. In this study, we investigated determinants for the in vivo chloroplast targeting of the TYMV 140K replication protein. Subcellular localization studies of deletion mutants identified a 41-residue internal sequence as the chloroplast targeting domain (CTD) of TYMV 140K; this sequence is sufficient to target GFP to the chloroplast envelope. The CTD appears to be located in the C-terminal extension of the methyltransferase domain-a region shared by 140K and its mature cleavage product 98K, which behaves as an integral membrane protein during infection. We predicted the CTD to fold into two amphipathic α-helices-a folding that was confirmed in vitro by circular dichroism spectroscopy analyses of a synthetic peptide. The importance for subcellular localization of the integrity of these amphipathic helices, and the function of 140K/98K, was demonstrated by performing amino acid substitutions that affected chloroplast targeting, membrane association and viral replication. These results establish a short internal α-helical peptide as an unusual signal for targeting proteins to the chloroplast envelope membrane, and provide new insights into membrane targeting of viral replication proteins-a universal feature of positive-strand RNA viruses.

Class II transactivator (CIITA) isoform expression and activity in melanoma.

In contrast with melanocytes, melanomas display constitutive expression of HLA-DR (HLA-DR+). This abnormal expression has been associated with tumour progression and metastatic dissemination. We have previously reported that this deregulation of HLA-D genes is due to the abnormal constitutive expression of the lymphocyte-specific isoform of class II transactivator (B-CIITA), in addition to its fibroblast form (F-CIITA), which is usually expressed in major histocompatibility complex (MHC) class II-negative interferon-gamma-induced cell types, such as melanocytes. In this study, we investigated the abnormal expression of B-CIITA in a panel of melanoma cell lines displaying differential HLA-DR expression profiles, and analysed whether such a molecular event can participate in tumour progression. Our results showed that the abnormal expression of B-CIITA did not have any particular effect, in comparison with F-CIITA, on the classical activity of CIITA HLA-D gene regulation. As CIITA has also been shown to regulate genes other than HLA-D, we evaluated the modulation of those encoding cyclin D1, YARS (tyrosyl-tRNA synthetase) and TRIP1 (transforming growth factor (TGF)-beta receptor-interacting protein), proteins involved in cell cycle/apoptosis balance, angiogenesis and resistance to TGF-beta, respectively. In contrast with other cell types, neither B-CIITA nor F-CIITA was able to modulate these genes in melanoma cell lines. Thus, the activity of CIITA, whether lymphocyte-specific or fibroblast-specific, is restricted to HLA-D gene expression in these tumours. Accordingly, our data suggest that CIITA is not involved per se in tumour progression; rather, it is the MHC class II molecules themselves, through tumour antigen presentation and the induction of tumour antigen-specific CD4 lymphocyte anergy, that may participate in immune escape and melanoma progression.

Nanog and Oct4 overexpression increases motility and transmigration of melanoma cells.

PURPOSE: Melanoma tumors are highly heterogeneous and can undergo phenotypic modifications depending on their plasticity and the microenvironment, with shifts between proliferative and invasive states. We have shown that melanoma cells, grown as spheroids in a neural crest cell medium, polarize toward an invasive and motile phenotype, in agreement with transcriptomic modulations, including the up-regulation of Nanog and Oct4. Overexpression of these genes was shown to be associated with poor prognosis and metastatic forms of some cancers. We thus investigated implication of Nanog and Oct4, two embryonic transcription factors, in melanoma motility. METHODS: Our team used stable transfection of Nanog or Oct4 in A375 melanoma cell line to investigate motility in a wound healing assay and a transendothelial migration assay. Using semiquantitative RT-PCR, expression of two gene panels involved either in mesenchymal motility or in amoeboid migration was studied. RESULTS: Strongly enhanced capacities of motility and extravasation were observed with cells overexpressing Oct4 and Nanog. The A375 cell line has been described as having a mesenchymal migration type. However, in the Oct4 and Nanog transfectants, several amoeboid migration markers are strongly induced. Accordingly, amoeboid migration inhibitors decrease significantly the transmigration of Oct4- and Nanog-expressing cells through endothelial cells. CONCLUSIONS: We propose here that Nanog and Oct4 pluripotency marker expression in melanoma cells increases the transmigration capacity of these cells through the gain of amoeboid motility, leading to higher invasiveness and aggressiveness.

PHF19 and Akt control the switch between proliferative and invasive states in melanoma.

Melanoma tumor cells shift between proliferative and invasive states based on their plasticity and microenvironmental conditions. Our team has shown that melanoma cells, grown as spheroids in a neural cell crest medium, polarize toward an invasive phenotype, characterized by a higher motility, a poor proliferation rate and a gain of pluripotency gene expression (Nanog and Oct4) when compared with cells grown in two dimensions in a serum-contaning medium. In agreement with the phenotypic switching hypothesis, most of these features are reversible. Microarray studies comparing two- vs. three-dimensional cultures revealed the downregulation of a polycomb-like protein, PHF19 (PHD finger protein 19), in the spheroids. As Polycomb proteins are involved in the epigenetic control of gene expression, we hypothesized that PHF19 might play a role in the switch between proliferative and invasive phenotypes. In this report, we show that PHF19 silencing reduces the cell proliferation rate and increases the transendothelial migration capacities of melanoma cell lines. However, PHF19 does not modulate the transcription level of Oct4 and Nanog. In the search of an upstream transcriptional regulator of the above genes, we identified the Akt signaling cascade as an inhibitor of Oct4 and Nanog expression and an activator for PHF19 expression. Through chromatin immunoprecipitation, we further provide evidence that phospho-Akt is part of the transcriptional complex associated to the promoters of all three genes. Our data therefore indicate the role of PHF19 and its upstream regulator, Akt, in the phenotype switch of melanoma cells from proliferative to invasive states.

Plasticity of melanoma cells induced by neural cell crest conditions and three-dimensional growth.

Melanoma tumors have been shown to comprise both invasive and proliferative cell subpopulations. These populations are highly plastic, thus hampering full characterization and therapeutic targeting of dormant and partially dedifferentiated invasive cells. We have reported, previously, that melanoma cells grown in a serum-free neural crest medium, in which they propagate as spheroids, show higher invasiveness and increased immune escape. In addition, in spheroids, we showed the increased expression of several genes which are involved in pluripotency, differentiation, and invasion. We therefore proposed that these culture conditions favor the polarization of proliferative melanoma cells toward an invasive state. As plasticity may suggest a reversible polarization, the aim of this report is to assess the transient phenotype of invasive cells generated through this procedure. We provide evidence that spheroid cells mimic dormant populations, and that this phenotype is fully reversible when cells are reintroduced into culture media that contain serum in which they grow as a monolayer. We also show that most transcriptional deregulations can be reversed. To further explain this plasticity in melanoma cells, we explored the epigenetic status of four gene promoters, assuming changes in acetylation or dimethylation on histone 3. We show reversible modifications on lysine 9 and lysine 4. We propose that spheroids allow the transient polarization of melanoma cells toward enhanced dormancy, loss of differentiation, and invasiveness, thereby reproducing the properties and plasticity of invasive subpopulations in melanoma tumors. This in-vitro model will allow further characterization and targeting of melanoma invasive cell populations.

Junctional adhesion molecules are required for melanoma cell lines transendothelial migration in vitro.

One of the main steps of metastasis is extravasation, a phenomenon well described in lymphocytes but remaining to be fully uncovered for melanoma. Junctional adhesion molecules (JAMs) control the transendothelial migration of leukocytes. To date, the role of the JAM proteins, notably JAM-A and JAM-C, has not been examined in melanoma. Here, we compared two melanoma tumor cell lines, A375 and SLM8 cells, the A375 cell line being four times more efficient than the SLM8 cells in the crossing of the endothelial monolayer. We show evidence of the differential expression of JAM-A and JAM-C in these cell lines with JAM-C mainly expressed in the A375 cell line, and JAM-A detected preferentially in the SLM8 cells. To further dissect the respective roles of these proteins, we used both siRNA and blocking antibodies to decrease JAM-A and JAM-C expression.

Melanoma spheroids grown under neural crest cell conditions are highly plastic migratory/invasive tumor cells endowed with immunomodulator function.

BACKGROUND: The aggressiveness of melanoma tumors is likely to rely on their well-recognized heterogeneity and plasticity. Melanoma comprises multi-subpopulations of cancer cells some of which may possess stem cell-like properties. Although useful, the sphere-formation assay to identify stem cell-like or tumor initiating cell subpopulations in melanoma has been challenged, and it is unclear if this model can predict a functional phenotype associated with aggressive tumor cells. METHODOLOGY/PRINCIPAL FINDINGS: We analyzed the molecular and functional phenotypes of melanoma spheroids formed in neural crest cell medium. Whether from metastatic or advanced primary tumors, spheroid cells expressed melanoma-associated markers. They displayed higher capacity to differentiate along mesenchymal lineages and enhanced expression of SOX2, NANOG, KLF4, and/or OCT4 transcription factors, but not enhanced self-renewal or tumorigenicity when compared to their adherent counterparts. Gene expression profiling attributed a neural crest cell signature to these spheroids and indicated that a migratory/invasive and immune-function modulating program could be associated with these cells. In vitro assays confirmed that spheroids display enhanced migratory/invasive capacities. In immune activation assays, spheroid cells elicited a poorer allogenic response from immune cells and inhibited mitogen-dependent T cells activation and proliferation more efficiently than their adherent counterparts. Our findings reveal a novel immune-modulator function of melanoma spheroids and suggest specific roles for spheroids in invasion and in evasion of antitumor immunity. CONCLUSION/SIGNIFICANCE: The association of a more plastic, invasive and evasive, thus a more aggressive tumor phenotype with melanoma spheroids reveals a previously unrecognized aspect of tumor cells expanded as spheroid cultures. While of limited efficiency for melanoma initiating cell identification, our melanoma spheroid model predicted aggressive phenotype and suggested that aggressiveness and heterogeneity of melanoma tumors can be supported by subpopulations other than cancer stem cells. Therefore, it could be constructive to investigate melanoma aggressiveness, relevant to patients and clinical transferability.

Coexpression of major histocompatibility complex class II with chemokines and nuclear NFkappaB p50 in melanoma: a rational for their association with poor prognosis.

The constitutive expression of major histocompatibility complex class II (MHC II) molecules in melanoma is highly unusual and has been associated with unfavorable clinical outcome and higher metastatic dissemination. This association remains poorly understood and therefore, in this study we looked to whether it is caused by intracellular events that promote tumor progression. We previously reported that MHC II expression in melanoma cells requires active mitogen-activated protein kinase/extracellular signal-related kinase. However, our comparative and molecular analyses of a panel of melanoma cell lines herein provide clear evidence that mitogen-activated protein kinase/extracellular signal-related kinase is not sufficient for HLA-DR expression. We found that the expression of HLA-DR in these tumors rather coincides with the expression of CXCL-1 and CXCL-8 chemokines, both known to be expressed in tumors that invade early and are related to invasive stages of melanoma. The expression of HLA-DR also nicely paralleled that of the nuclear NFkappaB p50 subunit, regulating the expression of these chemokines in melanoma and previously correlated with poor prognosis of melanoma patients, although we provide evidence that NFkappaB is not directly regulating MHC II expression level. The molecular basis for class II transactivator and HLA-DR expression in melanoma therefore remains unsolved, but our findings linking together the expression of HLA-DR, of chemokines involved in invasiveness, and of nuclear NFkappaB p50 strongly support the content that MHC II may be a marker of invasive primary melanoma, and could explain the long-standing association of MHC II expression with overall poor prognosis and unfavorable clinical outcome.

Pathologic expression of MHC class II is driven by mitogen-activated protein kinases.

The class II transactivator (CIITA) is the master regulator of MHC class II molecules (MHC II). In melanoma, the MHC II are constitutively expressed due to an abnormal transcription of CIITA from its promoter III (pIII), and requires the presence of a 1-kb enhancer located upstream from this latter. Since mitogen-activated protein kinases (MAPK) have been shown to be activated in most melanomas, we sought to analyze their possible involvement in CIITA expression. Using chemical inhibitors and dominant-negative constructs of MAPK-ERK kinase (Mek1) and MAPK-JNK, we evidenced the inhibition of MHC II and CIITA expression in melanoma cell lines displaying activated MAPK. Transcriptional regulation by MAPK is known to involve the AP-1 transcription factor family. Sequence analysis revealed an AP-1-responsive motif in the enhancer of CIITA pIII at -5954/-5947 from the site of transcription initiation. Its mutagenesis reduced CIITA expression four- to fivefold in melanoma cell lines and alleviated the effect of dominant-negative constructs of the MAPK pathway. Together, our findings demonstrate that MAPK-ERK and MAPK-JNK are regulators of CIITA transcription in melanoma, and pinpoint an AP-1-responsive site in the CIITA gene pIII. This should have considerable impact on our understanding of the physio-pathologic expression of MHC II.

Splicing defect in RFXANK results in a moderate combined immunodeficiency and long-duration clinical course.

MHC class II deficiency provokes a severe immunodeficiency characterized by a lack of antigen-specific immune response. In the absence of bone marrow transplantation (the only curative treatment), patients affected by this genetic recessive disease die in early childhood. However, others and we have recently described cases of mild or asymptomatic immunodeficiencies with defects in either CIITA (class II transactivator) or RFX5, both proteins required for the transcription of HLA-D genes. We describe in this report the first case of moderate immunodeficiency resulting from a defect in RFXANK, another transcription factor essential for HLA-D expression. The patient did not display any detectable expression of MHC class II molecules on B lymphocytes, monocytes or activated T lymphocytes. Accordingly HLA-D transcription was altered in the corresponding B-lymphoblastoid cell line. The defect in RFXANK was observed both at the transcript and protein level. Indeed a homozygous IVS4+5G>A mutation was evidenced in RFXANK, and shown to hamper the splicing of intron 4. However, we had shown previously that a defect in intron 4 can lead to the skipping of exon 4, and that the resulting truncated protein retains the capacity to activate HLA-DR expression. Therefore, like the two cases of moderate immunodeficiencies described previously, we demonstrate that the RFXANK defect presented here is coherent with a residual activity of the mutant protein. We thus propose that the common feature displayed by mildly immunodeficient patients is the leakiness of the mutations, which might allow a local or temporal expression of MHC class II molecules.

HLA-DR signaling inhibits Fas-mediated apoptosis in A375 melanoma cells.

Although melanocytes are devoid of the human major histocompatibility complex class II (HLA II) molecules, melanomas often display constitutive expression of these molecules, particularly HLA-DR. This constitutive expression of HLA-DR molecules is associated with tumor progression and poor prognosis but the molecular basis for this association remains poorly understood. Within the hypothesis of a role in immune escape, we analyzed the regulation of Fas-mediated apoptosis by HLA-DR signaling in the HLA-DR-positive malignant melanoma cell line A375. Our study demonstrates that engagement of HLA-DR molecules with anti-HLA-DR-specific monoclonal antibody L243 significantly reduces Fas-mediated apoptosis; DNA fragmentation and cell death were decreased by 50% and 40%, respectively. We found that while HLA-DR signaling does not affect Fas receptor expression, it significantly reduces Fas-induced activation of caspase-8 and Bid. Furthermore, inhibition studies and expression of dominant negative form of Mek-1 demonstrated that HLA-DR-mediated inhibition of caspase-8/Bid activation and apoptosis are dependent on the activation of the MAPK/Erk pathway. Together, our results provide evidence that HLA-DR signaling activates the MAPK/Erk pathway in A375 melanoma cells, which has a functional role in the resistance of these cells to Fas-mediated apoptosis. These observations underline the potential importance that HLA-DR signaling might have in melanoma immune escape and tumor progression.

MHC class II-deficient tumor cell lines with a defective expression of the class II transactivator.

MHC class II expression defects have been evidenced in several human tumor cell lines originating from lung cancers or retinoblastoma. Accordingly, the mouse adenocarcinoma and fibrosarcoma cell lines, RAG and L(tk-), do not express I-A and I-E molecules even when treated with IFN-gamma. Here we show that fusion of both cell lines restores the inducible expression of MHC class II, thereby demonstrating that they present different and recessive alterations outside the MHC class II locus. CIITA, the MHC class II transactivator, controls the tissue-specific expression of MHC class II genes and creates the architecture of the transcriptional complex that binds to the MHC class II gene promoters. In L(tk-) cells, C2ta transcripts, expressed from the gene encoding CIITA, were indeed detected in severely limited amounts, with a defect in C2ta transcription initiation. In agreement we show here that the L(tk-) cell line does not express the CIITA protein. In contrast, in the RAG cell line, C2ta transcripts were expressed at normal levels, from the proper initiation site. The nucleotide sequencing of the CIITA cDNA from RAG did not reveal any mutation. However, the CIITA protein was not detected. These data evidence a new type of defect in a MHC class II-defective tumor cell line, as we show here that the alteration in the RAG cells occurs downstream of C2ta transcription. The RAG mutation might therefore reside in the C2ta transcript nuclear export or translation, or in the stability of the CIITA protein.

Isoforms of the class II transactivator protein.

The class II transactivator (CIITA) controls both the constitutive and IFN-gamma inducible expression of HLA-D genes. In addition, through the squelching of another transactivator CREB-binding protein, CIITA was more recently shown to have a wider cellular function, including cell cycle control or cellular response to IFN-gamma and IL-4. However, due to its low expression level, its analysis mainly relies on the study of recombinant overexpressed forms of the protein. We report here the analysis of native CIITA in various cell types. We first show the precise timing of CIITA protein expression in a fibroblast cell line in response to IFN-gamma. This expression is observed 2 h after the cytokine addition with a peak of expression ranging from 16 to 24 h. We next show the existence of two major isoforms of the CIITA protein differentially expressed in fibroblast, B lymphocyte or melanoma cell lines. We present the first demonstration that these isoforms originate from alternative translation initiation codons. We finally show that CIITA isoforms translocate to the nucleus with an apparently similar efficiency. Our data therefore demonstrate the existence of CIITA isoforms whose respective ratio depends on the cell type examined. However, we present evidence for a modulation of this ratio in a melanoma cell line with an abnormal constitutive expression of MHC class II molecules.