A new gene coding for a differentiation antigen recognized by autologous cytolytic T lymphocytes on HLA-A2 melanomas.

It has been reported previously that antitumor cytolytic T lymphocyte (CTL) clones can be isolated from blood lymphocytes of HLA-A2 melanoma patients, after stimulation in vitro with autologous tumor cells, and that some of these CTL clones lyse most HLA-A2 melanomas. A first antigen recognized by such CTL clones was previously shown to be encoded by the tyrosinase gene. We report here the identification of another gene that also directs the expression of an antigen recognized on most melanomas by CTL clones that are restricted by HLA-A2. The gene, designated Melan-A, is unrelated to any known gene. It is 18 kb long and comprises five exons. Like the tyrosinase gene, it is expressed in most melanoma tumor samples and, among normal cells, only in melanocytes.

A nonapeptide encoded by human gene MAGE-1 is recognized on HLA-A1 by cytolytic T lymphocytes directed against tumor antigen MZ2-E.

We have reported the identification of human gene MAGE-1, which directs the expression of an antigen recognized on a melanoma by autologous cytolytic T lymphocytes (CTL). We show here that CTL directed against this antigen, which was named MZ2-E, recognize a nonapeptide encoded by the third exon of gene MAGE-1. The CTL also recognize this peptide when it is presented by mouse cells transfected with an HLA-A1 gene, confirming the association of antigen MZ2-E with the HLA-A1 molecule. Other members of the MAGE gene family do not code for the same peptide, suggesting that only MAGE-1 produces the antigen recognized by the anti-MZ2-E CTL. Our results open the possibility of immunizing HLA-A1 patients whose tumor expresses MAGE-1 either with the antigenic peptide or with autologous antigen-presenting cells pulsed with the peptide.

A gene encoding an antigen recognized by cytolytic T lymphocytes on a human melanoma.

Many human melanoma tumors express antigens that are recognized in vitro by cytolytic T lymphocytes (CTLs) derived from the tumor-bearing patient. A gene was identified that directed the expression of antigen MZ2-E on a human melanoma cell line. This gene shows no similarity to known sequences and belongs to a family of at least three genes. It is expressed by the original melanoma cells, other melanoma cell lines, and by some tumor cells of other histological types. No expression was observed in a panel of normal tissues. Antigen MZ2-E appears to be presented by HLA-A1; anti-MZ2-E CTLs of the original patient recognized two melanoma cell lines of other HLA-A1 patients that expressed the gene. Thus, precisely targeted immunotherapy directed against antigen MZ2-E could be provided to individuals identified by HLA typing and analysis of the RNA of a small tumor sample.

DNA methylation is the primary silencing mechanism for a set of germ line- and tumor-specific genes with a CpG-rich promoter.

A subset of male germ line-specific genes, the MAGE-type genes, are activated in many human tumors, where they produce tumor-specific antigens recognized by cytolytic T lymphocytes. Previous studies on gene MAGE-A1 indicated that transcription factors regulating its expression are present in all tumor cell lines whether or not they express the gene. The analysis of two CpG sites located in the promoter showed a strong correlation between expression and demethylation. It was also shown that MAGE-A1 transcription was induced in cell cultures treated with demethylating agent 5'-aza-2'-deoxycytidine. We have now analyzed all of the CpG sites within the 5' region of MAGE-A1 and show that for all of them, demethylation correlates with the transcription of the gene. We also show that the induction of MAGE-A1 with 5'-aza-2'-deoxycytidine is stable and that in all the cell clones it correlates with demethylation, indicating that demethylation is necessary and sufficient to produce expression. Conversely, transfection experiments with in vitro-methylated MAGE-A1 sequences indicated that heavy methylation suffices to stably repress the gene in cells containing the transcription factors required for expression. Most MAGE-type genes were found to have promoters with a high CpG content. Remarkably, although CpG-rich promoters are classically unmethylated in all normal tissues, those of MAGE-A1 and LAGE-1 were highly methylated in somatic tissues. In contrast, they were largely unmethylated in male germ cells. We conclude that MAGE-type genes belong to a unique subset of germ line-specific genes that use DNA methylation as a primary silencing mechanism.

Identification on a human sarcoma of two new genes with tumor-specific expression.

Genes MAGE, BAGE, GAGE, and LAGE-1/NY-ESO-1 code for antigens that are recognized on melanoma cells by autologous CTLs. Because the pattern of expression of these genes results in the presence of antigens on many tumors of various histological types and not on normal tissues, these antigens qualify for cancer immunotherapy. To identify new genes with tumor-specific expression, we applied a cDNA subtraction approach, ie., representational difference analysis, to a human sarcoma cell line. We obtained two cDNA clones that appeared to be tumor specific. The corresponding genes were named SAGE and HAGE because they have the same pattern of expression as genes of the MAGE family. SAGE encodes a putative protein of 904 amino acids and shows no homology to any recorded gene. Like the MAGE-A genes, it is located in the q28 region of chromosome X. Expression of gene SAGE was observed mainly in bladder carcinoma, lung carcinoma, and head and neck carcinoma but not in normal tissues, with the exception of testis. Gene HAGE, which is located on chromosome 6, encodes a putative protein of 648 amino acids. This protein is a new member of the DEAD-box family of ATP-dependent RNA helicases. Gene HAGE is expressed in many tumors of various histological types at a level that is 100-fold higher than the level observed in normal tissues except testis. Because of this tumor-specific expression, genes SAGE and HAGE ought to encode antigens that could be useful for antitumoral therapeutic vaccination.

Identification of a new MAGE gene with tumor-specific expression by representational difference analysis.

Human genes expressed exclusively in tumors and male germ line cells, such as those of the MAGE, BAGE, and GAGE families, encode antigens recognized by T lymphocytes, which are potentially useful for antitumor immunotherapy. To identify new genes of this type, we generated cDNA populations enriched in sequences expressed only in testis and melanoma, using the representational difference analysis approach. A testis cDNA library enriched by subtraction with cDNA from four other normal tissues was hybridized with radiolabeled melanoma cDNA enriched by subtraction with normal skin cDNA. A cDNA fragment sharing significant homology with MAGE genes was identified, and a cosmid containing this new gene, named MAGE-C1, was isolated. MAGE-C1 is composed of four exons and encodes a putative protein of 1142 amino acids. It is about 800 residues longer than the other MAGE proteins due to the insertion of a large number of short repetitive sequences in front of the MAGE-homologous sequence. The MAGE-C1 gene appears to be located on band Xq26, whereas the MAGE-A and MAGE-B genes are located on Xq28 and Xp21, respectively. Like other MAGE genes, MAGE-C1 is expressed in a significant proportion of tumors of various histological types, whereas it is silent in normal tissues except testis. It is probable, therefore, that like other MAGE genes, MAGE-C1 encodes antigens that may constitute useful targets for cancer immunotherapy because of their strict tumoral specificity.

Differential requirement for CD4 help in the development of an antigen-specific CD8+ T cell response depending on the route of immunization.

Previous studies in our laboratory have shown that DBA/2 mice injected i.p. with syngeneic P815 tumor cells transfected with the HLA-CW3 gene (P815-CW3) showed a dramatic expansion of activated CD8+CD62L- T cells expressing exclusively the Vbeta10 segment. We have used this model to study the regulatory mechanisms involved in the development of the CW3-specific CD8+ response, with respect to different routes of immunization. Whereas both intradermal (i.d.) and i.p. immunization of DBA/2 mice with P815-CW3 cells led to a strong expansion of CD8+CD62L-Vbeta10+ cells, only the i.d. route allowed this expansion after immunization with P815 cells transfected with a minigene coding for the antigenic epitope CW3 170-179 (P815 miniCW3). Furthermore, depletion of CD4+ T cells in vivo completely abolished the specific response of CD8+CD62L-Vbeta10+ cells and prevented the rejection of P815-CW3 tumor cells injected i.p., whereas it did not affect CD8S+CD62L-Vbeta10+ cell expansion after i.d. immunization with either P815-CW3 or P815 miniCW3. Finally, the CW3-specific CD8+ memory response was identical whether or not CD4+ T cells were depleted during the primary response. Collectively, these results suggest that the CD8+ T cell response to P815-CW3 tumor cells injected i.p. is strictly dependent upon recognition of a helper epitope by CD4+ T cells, whereas no such requirement is observed for i.d. injection.

The activation of human gene MAGE-1 in tumor cells is correlated with genome-wide demethylation.

Human gene MAGE-1 encodes tumor-specific antigens that are recognized on melanoma cells by autologous cytolytic T lymphocytes. This gene is expressed in a significant proportion of tumors of various histological types, but not in normal tissues except male germ-line cells. We reported previously that reporter genes driven by the MAGE-1 promoter are active not only in the tumor cell lines that express MAGE-1 but also in those that do not. This suggests that the critical factor causing the activation of MAGE-1 in certain tumors is not the presence of the appropriate transcription factors. The two major MAGE-1 promoter elements have an Ets binding site, which contains a CpG dinucleotide. We report here that these CpG are demethylated in the tumor cell lines that express MAGE-1, and are methylated in those that do not express the gene. Methylation of these CpG inhibits the binding of transcription factors, as seen by mobility shift assay. Treatment with the demethylating agent 5-aza-2'-deoxycytidine activated gene MAGE-1 not only in tumor cell lines but also in primary fibroblasts. Finally, the overall level of CpG methylation was evaluated in 20 different tumor cell lines. It was inversely correlated with the expression of MAGE-1. We conclude that the activation of MAGE-1 in cancer cells is due to the demethylation of the promoter. This appears to be a consequence of a genome-wide demethylation process that occurs in many cancers and is correlated with tumor progression.

Identification of human testis-specific transcripts and analysis of their expression in tumor cells.

Tumor-specific antigens recognized by autologous T lymphocytes are encoded by genes, including those of the MAGE, BAGE, and GAGE gene families, that are expressed in a significant fraction of tumors of various types, but not in normal adult tissues, except for testis where they appear to be expressed in germ cells. Because male germ cells are known to express many genes that are not expressed in other normal adult tissues, we wished to determine whether most of these genes are occasionally activated in tumor cells. Representational difference analysis was used to obtain testis-specific transcripts. The expression of 15 testis-specific cDNA sequences was tested by RT-PCR in a series of tumor cell lines. Only one cDNA sequence showed a significant level of expression in some tumor cell lines. Remarkably, this cDNA clone proved to be a new gene of the MAGE family. These results suggest that MAGE, BAGE, and GAGE genes belong to a minor subset of testis-specific genes that is often activated in tumors of various types, whereas most testis-specific genes are either never or very rarely activated in tumors.

Involvement of two Ets binding sites in the transcriptional activation of the MAGE1 gene.

The MAGE1 gene codes for an antigen recognized on melanoma cell line MZ2-MEL by autologous cytolytic T lymphocytes. It is expressed in a number of tumors of different histological origins, but not in normal tissues except in testis. The MAGE1 promoter region was analyzed by performing transient transfections in MZ2-MEL cells with luciferase reporter plasmids. A fragment extending from nucleotide -792 to +118 exhibited high transcriptional activity. By deletional analysis of this fragment, we identified five activating regions designated C, A, B', B, and D. The activity of region A depends on the presence of region B' and vice versa. Two inverted Ets motifs contained in regions B' and B were found to drive 90% of the activity of the MAGE1 promoter in MZ2-MEL cells. Electrophoretic mobility shift assays performed with a nuclear extract from MZ2-MEL cells and with competitor oligonucleotides containing an Ets consensus site showed that nuclear proteins bind to the Ets motif of regions B' and B. Similar experiments suggested that region A binds transcription factors of the Sp1 family. The MAGE1 promoter was found to exert transcriptional activity in tumor cells where the MAGE1 gene is not expressed, suggesting that other mechanisms, such as demethylation, may contribute to the tumor-specific expression of the gene.

A mutated intron sequence codes for an antigenic peptide recognized by cytolytic T lymphocytes on a human melanoma.

We have identified an antigen recognized on a human melanoma by autologous cytolytic T lymphocytes. It is encoded by a gene that is expressed in many normal tissues. Remarkably, the sequence coding for the antigenic peptide is located across an exon-intron junction. A point mutation is present in the intron that generates an amino acid change that is essential for the recognition of the peptide by the anti-tumor cytotoxic T lymphocytes. This observation suggests that the T-cell-mediated surveillance of the integrity of the genome may extend to some intronic regions.

Immunogenic (tum-) variants obtained by mutagenesis of mouse mastocytoma P815. VIII. Detection of stable transfectants expressing a tum- antigen with a cytolytic T cell stimulation assay.

Mutagen treatment of mouse mastocytoma P815 produces highly immunogenic "tum-" variants. Most of these variants express potent transplantation antigens which are not present on the original P815 tumor cells. These tum- antigens, which appear to be specific for each variant, elicit a strong cytolytic T lymphocyte (CTL) response, but do not seem to induce a specific antibody response. As a first step in the isolation of the gene of a tum- antigen, we attempted DNA-mediated gene transfer. As a DNA recipient cell we used P1.HTR, a highly transfectable P815 cell line, whose selection has been previously described. For the detection of antigen-expressing cells in transfected populations we developed a procedure that relies on the ability of these cells to stimulate the proliferation of the relevant CTL. Using DNA from tum- variant P91 mixed with a plasmid carrying an antibiotic resistance gene, we obtained several independent transfectants expressing a tum- antigen, at a frequency of approximately 1 in 13,000 antibiotic-resistant transfectants. These transfectants express only one of the two tum- antigens that were identified on P91, suggesting that these tum- antigens correspond to different genes. We expect that the detection procedure described here will be suitable for the identification of transfectants for any gene that determines the expression of an antigen recognized by CTL.

Induction of a cytolytic T-cell response in mice with a recombinant adenovirus coding for tumor antigen P815A.

We investigated the efficacy of a recombinant adenovirus in inducing a cytolytic T-lymphocyte (CTL) response in mice against tumor antigen P815A, which is present on mouse mastocytoma P815. The recombinant adenoviral vector (Adeno.PIA) contained the sequence coding for the antigenic nonapeptide which binds to the H-2.Ld molecule to form antigen P815A. We verified that murine cells infected in vitro with Adeno. PIA were lysed by an anti-P815A CTL clone. Mice then received a single intradermal injection of Adeno. PIA, and after a few weeks their spleen cells were stimulated in vitro with tumor cells expressing antigen P815A. An anti-P815A CTL response was observed with the spleen lymphocytes of nearly all the mice, providing the lymphocytes were re-stimulated in vitro with cells expressing both P815A and co-stimulatory molecule B7.1. When the stimulatory cells did not express B7.1, a specific CTL response was observed in only 45% of the mice, and it was less intense. The Adeno. P1A viral vector was unable to raise an anti-P815A response in mice that had been previously infected with a recombinant adenovirus carrying the beta-galactosidase gene or with a defective adenovirus. We conclude that adenoviral vectors may be very useful for the priming of cytolytic T-cell responses directed against human tumor antigens. Other modes of immunization may be necessary to boost the responses induced with adenoviral vectors.

Quantitative evaluation of the expression of MAGE genes in tumors by limiting dilution of cDNA libraries.

The MAGE-A genes are expressed in tumor cells but not in healthy tissues, except in male germ line cells and in placenta. They encode tumor-specific antigens recognized by autologous cytolytic T lymphocytes (CTLs). On the basis of semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR) assays, 6 of the 12 members of the MAGE-A family, including MAGE-A1, were previously reported to have a high level of expression in tumors, whereas 5 other members, including MAGE-A10, were expressed at a much lower level, deemed to be insufficient for CTL recognition. However, analysis with antibodies has shown that some melanoma cell lines contain equivalent amounts of MAGE-A1 and MAGE-A10 proteins. This discrepancy appeared to be due to the low efficacy of the primers that had been used for the previous MAGE-A10 RT-PCR assays. This led us to develop a method that is independent of the efficacy of the PCR primers to evaluate MAGE-A gene expression. cDNA libraries from tumor cell lines were introduced into bacteria, of which 200 pools of about 500 bacteria were maintained in microcultures. The frequencies of the MAGE-A cDNA clones in each library were evaluated by performing PCR assays on each of these pools. The abundance of MAGE-A10 cDNAs was found to be similar to that of MAGE-A1 in 3 of the libraries that were analyzed, including 2 with high expression (1/6,400), confirming that MAGE-A10 is expressed at a high level. MAGE-A2, A3, A4, A6 and A12 cDNAs were also confirmed often to be present at a frequency of more than 1/10,000, a level of expression that should suffice for recognition of antigenic peptides encoded by these genes by cytolytic T cells. The remaining MAGE genes are either not expressed in tumors or are expressed at a very low level, with the exception of MAGE-A8 and 11, which show high expression in a very small number of tumors. This method also allowed us to isolate 5 MAGE-A cDNAs that we had not obtained previously, enabling us to delineate the exons in the sequences of genes MAGE-A5, A8, A9, A10 and A11.

A MAGE-A4 peptide presented by HLA-A2 is recognized by cytolytic T lymphocytes.

The MAGE-encoded antigens that are recognized by cytolytic T lymphocytes (CTL) are shared by many tumors and are strictly tumor specific. Clinical trials involving therapeutic vaccination of cancer patients with MAGE antigenic peptides or proteins are in progress. To increase the range of patients eligible for therapy with peptides, it is important to identify additional MAGE epitopes. We have used a method to identify CTL epitopes, which selects naturally processed peptides. CD8(+) T cells, obtained from individuals without cancer, were stimulated with autologous dendritic cells infected with a recombinant adenovirus containing the MAGE-A4 coding sequence. Responder cell microcultures that specifically lysed autologous EBV-transformed B cells infected with vaccinia-MAGE-A4 were cloned using autologous stimulator cells infected with a Yersinia enterocolitica carrying the MAGE-A4 sequence. An anti-MAGE-A4 CTL clone was obtained and the epitope was found to be decapeptide GVYDGREHTV (amino acids 230-239) presented by HLA-A2 molecules. The CTL clone lysed HLA-A2 tumor cells expressing MAGE-A4. This is the first reported antigenic peptide encoded by MAGE-A4. It may be valuable for cancer immunotherapy because MAGE-A4 is expressed in 51% of lung carcinomas and 63% of esophageal carcinomas, whereas about 50% of Caucasians and Asians express HLA-A2.

Sequence and expression pattern of the human MAGE2 gene.

We reported previously identification of the human MAGE1 gene, which encodes an antigen recognized on human melanoma MZ2-MEL by autologous cytolytic T lymphocytes. In addition to MAGE1, melanoma MZ2-MEL expresses several closely related genes, one of which has been named MAGE2. The complete MAGE2 sequence was obtained and it comprises 3 exons homologous to those of MAGE1 and an additional exon homologous to a region of the first MAGE1 intron. Like the open reading frame of MAGE1, that of MAGE2 is entirely encoded by the last exon. The MAGE1 and MAGE2 sequences of this exon show 82% identity and the putative proteins show 67% identity. The MAGE2 gene is expressed in a higher proportion of melanoma tumors than MAGE1. It is also expressed in many small-cell lung carcinomas and other lung tumors, laryngeal tumors, and sarcomas. No MAGE1 and MAGE2 gene expression was found in a large panel of healthy adult tissues, with the exception of testis.

Efficient expression of tum- antigen P91A by transfected subgenic fragments.

Mutagen treatment of mouse P815 tumor cells produces immunogenic mutants that express new transplantation antigens (tum- antigens) recognized by cytolytic T cells. The gene encoding tum- antigen P91A comprises 12 exons and a mutation located in exon 4 is responsible for the production of a new antigenic peptide. Transfection experiments showed that the expression of the antigen could be transferred not only by the entire gene but also by gene segments comprising only the mutated exon and parts of the surrounding introns. This was observed with subgenic regions that were not cloned in expression vectors. Antigen expression did not require the integration of the transfected gene segment into a resident P91A gene by homologous recombination. It also occurred when the subgenic segment was transfected without the usual selective gene, which comprises an eucaryotic promoter, and also without plasmid sequences, which are known to contain weak promoters. When a stop codon was introduced at the beginning of exon 4, the expression of the antigen was maintained and evidence was obtained that an ATG codon located in this region served as initiation site for the translation of the antigenic peptide. But we have not obtained evidence indicating that antigenic peptides are direct translation products rather than degradation products of entire proteins.

Structure of the gene of tum- transplantation antigen P91A: the mutated exon encodes a peptide recognized with Ld by cytolytic T cells.

Mutagen treatment of mouse P815 tumor cells produces immunogenic mutants that express new transplantation antigens (tum- antigens) recognized by cytolytic T cells. We found that the gene conferring expression of tum- antigen P91A contains 12 exons, encoding a 60 kd protein lacking a typical N-terminal signal sequence. The sequence shows no significant similarity with sequences in current data bases. A mutation that causes expression of the antigen is located in exon 4; it is the only apparent difference between the normal and the antigenic alleles. A short synthetic peptide corresponding to a region of exon 4 located around this mutation makes P815 cells sensitive to lysis by anti-P91A cytolytic T cells. The mutation creates a strong aggretope enabling the peptide to bind the H-2 Ld molecule. Several secondary tumor cell variants that no longer express tum- antigen P91A were found to carry deletions in the gene.

Identification of tumour rejection antigens recognized by T lymphocytes.

On the basis of the results reviewed here, there are two major mechanisms whereby tumour rejection antigens may arise. The first mechanism is mutational. Point mutations occurring in a large variety of genes may produce new antigenic peptides, either by providing them with the ability to bind to MHC class I molecules or by providing them with a new epitope (Fig. 2). The second mechanism is the activation of a gene that is silent in normal tissues and for which no strong natural tolerance has been established. Plausible candidates for the mutational mechanism are the "tumour specific transplantation antigens" observed on methylcholanthrene induced tumours and tumours induced by ultraviolet light. The diversity of these antigens appears to be very large, like that of the tum- antigens. Moreover, these tumours have been obtained with high doses of carcinogens, which are proven mutagens. On the other hand, a P815 tumour rejection antigen appears to arise through the activation of a silent gene, and it may turn out that this is the rule for most tumour rejection antigens. It is our hope that other genes coding for mouse and human tumour rejection antigens will soon be identified, so that it will become clear whether the activational mechanism is the rule or the exception. In our view, this is a crucial issue. Insofar as tumour rejection antigens result from mutations, they may be highly specific for every individual tumour. The tumour specific nature of these antigens would then be easily ascertained. However, active immunization of cancer patients would require that a tumour cell line be obtained from each patient, a most unpractical prospect. If, on the other hand, production of tumour rejection antigens results from the activation of a normal gene, then there is a good probability that the same gene may be activated in many different tumours, being perhaps preferentially shared by tumours of the same histological type. This would probably not result in the expression of the same antigen in all these tumours, because the patients would differ in their presenting molecules, which are determined by their HLA haplotype. However, a subset of the tumours expressing the same "tumour rejection" gene should share the same class I restricting element, so that all of these patients could be immunized with a cell that would express the gene and carry the appropriate HLA molecule.(ABSTRACT TRUNCATED AT 400 WORDS)