A number of genes of the MAGE family have been shown to code for antigens that are recognized on many human tumors by autologous CTLs. These antigens should be strictly tumor specific because the encoding MAGE genes are not expressed in normal adult cells, except for male germ-line cells, which lack HLA expression. Here, we report that a distant relative of the previously identified MAGE genes is expressed in many, if not all, normal tissues. This gene, which was named MAGE-D, is located in Xp11. Its exon-intron structure is completely different from that of the other MAGE genes. None of the 20 MAGE antigenic peptides presently known to be recognized by T lymphocytes is encoded by the new MAGE gene. It appears, therefore, that this new finding leaves intact the tumor specificity of the antigens encoded by the MAGE genes that are expressed only in tumor and germ-line cells.
The human gene MAGE-A1 was identified by a gene transfection approach involving the stimulation of CTLs directed against a melanoma cell line (1). By hybridization of cosmids with a MAGE-A1 probe, 11 closely related genes were identified (2). Together, these 12 genes form the MAGE-A cluster, located in the q28 region of the X chromosome (3). A sequencing effort directed at the Xp21 region led to the identification of the MAGE-B cluster, which comprises four genes (4, 5, 6). MAGE-C, a third group, comprises two genes located in Xq26-27 (7).3 They were identified following the analysis of cDNA libraries enriched for tumor- and testis-specific sequences by representational difference analysis. These MAGE-A, -B and -C genes are not expressed in normal tissues, except in male germ-line cells, and, for some of these genes, in placenta. Seven MAGE-A genes, two MAGE-B genes, and two MAGE-C genes are expressed in a significant proportion of tumors of various histological types. The activation of these genes appears to result from the demethylation of their promoter region (8). In an attempt to identify new MAGE genes that may encode tumor-specific antigens, we performed a search of protein databanks for MAGE-related protein sequences. We report here that one MAGE homologue recorded in databanks is expressed ubiquitously.
Materials and Methods
A search for proteins that are homologous to protein MAGE-A10 was performed with the blastp program on the Internet server of the National Center for Biotechnology Information.4 The numerous blast hits that were obtained comprised mostly known MAGE protein sequences as well as a protein previously unknown to us, which we named MAGE-D. The MAGE-D protein sequence was a translation of a genomic nucleotide sequence and of a cDNA nucleotide sequence deposited in databanks. The genomic nucleotide sequence data were produced by the Human Chromosome X Sequencing Group at the Sanger Center.5 This sequence is from clone dJ1409 of the P1 artificial chromosome library RPCI1 constructed at the Roswell Park Cancer Institute by the group of Pieter de Jong.6 The cDNA nucleotide sequence has accession number U92544.
Protein Sequence Alignments.
The multiple alignments of protein sequences and the dendrogram representation were generated with the MacVector Version 6.5 software (Oxford Molecular Ltd., Oxford, England), using the ClustalW algorithm.
RNA purification and cDNA synthesis were performed as described (9). The cDNAs produced from 50 ng of total RNA or 200 ng of genomic DNA were PCR-amplified in a TRIO-Thermoblock (Biometra, Göttingen, Germany) for 30 cycles of 1 min at 94°C and 4 min at 72°C with primers SL162 (in exon 8; 5′-GCTGGGTCTGCTCATGGTGCT-3′) and SL163 (in exon 12; 5′-CCCAATGCCCATTCGAGCTCT-3′).
Results and Discussion
A New Distant Relative of the MAGE-A, -B, and -C Genes.
A blastp homology search in protein databanks was performed with MAGE-A protein sequences. It revealed a putative protein, hitherto unknown to us, that was distantly related but clearly homologous to the known MAGE proteins. A genomic clone and also a cDNA sequence coding for this protein were found in databanks. The gene coding for the MAGE-related protein had been identified in the course of systemic sequencing of recombinant P1 artificial chromosome clones containing sequences of the human X chromosome (Human Chromosome X Sequencing Group at the Sanger Center). The artificial chromosome containing the new MAGE gene had been mapped by fluorescence in situ hybridization to the Xp11 region, a region clearly distinct from those of the MAGE-A, -B, and -C clusters (Fig. 1 A). It seemed, therefore, appropriate to name the new gene MAGE-D.
The MAGE-D gene codes for a putative protein of 606 amino acids. This compares with a size of ∼320 amino acids for the MAGE-A proteins. The region of homology is located in the central part of the MAGE-D protein, at positions 276–478 (Fig. 2,A). It is homologous to the two hundred amino acids located at the COOH-terminal end of the other MAGE proteins. Pairwise alignments were carried out with a set of protein sequences comprising MAGE-D, other known MAGE proteins, and also human necdin, a protein that was previously identified as a distant MAGE relative (10). The results summarized in the dendrogram of Fig. 1 B, indicate that MAGE-D is about equally distant from the MAGE-A, -B, and -C groups and slightly less distant from necdin.
MAGE-D Is Expressed in Normal Tissues.
The expression pattern of gene MAGE-D was determined by RT-PCR7 analysis. Specific primers were selected in different exons to enable us to distinguish the amplification product of cDNA templates (414 bp) from that of contaminating genomic DNA (2.4 kb). The specificity of the primers was verified by sequencing the PCR amplification product of testis cDNA: the sequence was unambiguous and identical to that found in the MAGE-D cDNA clone recorded in databanks.
We tested samples from 13 different normal tissues. Remarkably, all were found to express MAGE-D (Fig. 3). This result could be due to the ubiquitous expression of gene MAGE-D or to its expression in a single cell type present in all normal tissues tested. To distinguish between these possibilities, we tested the expression of MAGE-D in 38 cell lines of various origins. The cell lines were derived from four different normal cell types as well as from tumors of 22 different histological types (Fig. 3). All were positive, indicating that gene MAGE-D is expressed in most, if not all, cell-types, in complete contrast to the MAGE-A, -B, and -C genes, which are expressed only in male germ-line and tumor cells.
The MAGE proteins were reported previously to be homologous to mouse necdin (2). The mouse necdin gene was reported to be expressed only in fetal and adult brain, in postmitotic neurons (11). However, the NDN gene, which codes for human necdin and is located on chromosome 15, has been reported recently to be expressed in many cell types (10). Thus, there are at least two distant MAGE homologues with ubiquitous expression.
MAGE-D Has a Unique Exon-Intron Structure.
The MAGE-D gene is composed of 14 exons (Fig. 2 B). Two of these are alternative first exons. Alternative exon 1 is found in the complete MAGE-D cDNA clone found in databanks. Alternative exon 1′ is found in a human fetal brain expressed sequence tag (accession no. T05278). The open reading frame that encodes the putative MAGE-D protein starts with an ATG located in exon 2 and ends with a stop codon in exon 12. The region of protein MAGE-D that shows clear homology to the other MAGE proteins is encoded in exons 4–12.
This exon-intron structure of the MAGE-D gene contrasts with the structure of the MAGE-A, -B, and -C genes. These genes are composed of three or four exons, with the entire MAGE coding sequence comprised in the last exon. A possible interpretation of the difference in structure between MAGE-D and the other MAGE genes would be that MAGE-D kept the structure of the ancestor of all the MAGE genes, whereas the other MAGE genes would have originated by retrotransposition of mRNA of this ancestral gene. In this regard, the NDN gene resembles the majority of MAGE genes, having been reported to be intronless (10).
MAGE-D Does Not Encode Any of the Known MAGE Antigenic Peptides.
The MAGE-A and -B genes code for antigens that are recognized by autologous T cells. These antigens ought to be strictly tumor specific because these genes are not expressed in normal cells, except for male germ-line cells, which cannot present antigens to T cells because of a lack of HLA molecules. However, the tumor specificity of these antigens could vanish if related genes with ubiquitous expression were found to code for the same antigenic peptides. So far, ∼20 antigenic peptides encoded by the MAGE-A and -B genes have been shown to be recognized by T lymphocyte clones on various HLA molecules (Fig. 4; Refs. 12, 13, 14, 15, 16, 17, 18, 19, 20).8,9 We examined the sequences of the MAGE-D putative protein as well as that of human necdin to establish whether these proteins contain any of the known MAGE antigenic peptides. None of these peptides was present in the MAGE-D or necdin sequence. It could not be excluded, however, that some minor changes in amino acids would preserve the ability of an antigen to bind to the same HLA molecule and to be recognized by the same T cell. But for all the known MAGE antigenic peptides, there were at least four differences with the corresponding MAGE-D or necdin sequence, making this extremely unlikely. Moreover, there is not a single nonameric peptide sequence present in the MAGE-D or necdin sequences, which is identical to any sequence present in the MAGE proteins displayed in Fig. 4. We found only one instance of nonameric peptides with eight identical amino acids shared between MAGE-D and another MAGE protein: four overlapping peptides were found in a 13-amino acid stretch of MAGE-B2.
Arguably, if there was a large number of presently unknown MAGE genes with ubiquitous expression, they might code for some or many of the antigenic peptides encoded by MAGE-A, -B, and -C, resulting in tolerance for the antigens encoded by the latter genes. Even though this possibility appears to be unlikely, it cannot be rigorously excluded at this time. What can be said, however, is that mice do not appear to be tolerant for the antigens encoded by the Smage genes, mouse homologues of MAGE genes (21). We have obtained strong and specific CTL responses in mice by vaccinating mice with a Smage antigenic peptide.10
The existence of a MAGE-related gene with ubiquitous expression offers new perspectives for the analysis of the function of the MAGE proteins. So far, it does not appear to reduce the potential usefulness of MAGE-encoded antigens for cancer immunotherapy.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Belgian Programme on Interuniversity Poles of Attraction initiated by the Belgian State, Prime Minister’s Office, Science Policy Programming, and by grants from the Association contre le Cancer (Brussels, Belgium), the BIOMED2 program of the European Community, the Fonds J. Maisin (Brussels, Belgium), CGER-Assurances and VIVA (Brussels, Belgium), and the Fonds National de la Recherche Scientifique (TELEVIE grants; Brussels, Belgium). S. L. is supported by the Fonds National de la Recherche Scientifique (Collaborateur Scientifique; Brussels, Belgium).
S. Lucas, unpublished results.
http://www.ncbi.nlm.nih.gov/Entrez/nucleotide.htm/. Accession no. Z98046.
The abbreviation used is: RT-PCR, reverse transcriptase-PCR.
P. Chaux, R. Luiten, N. Demotte, V. Vantomme, V. Stroobant, C. Traversari, V. Russo, E. Schultz, R. Guy, G. R. Cornelis, T. Boon, and P. van der Bruggen. Identification of five MAGE-A1 epitopes recognized by cytolytic T lymphocytes obtained by in vitro stimulation with dendritic cells transduced with MAGE-A1, submitted for publication.
M. T. Duffour, P. Chaux, C. Lurquin, G. Cornelis, T. Boon, and P. van der Bruggen. A MAGE-A4 peptide presented by HLA-A2 is recognized by cytolytic T lymphocytes, submitted for publication.
A. Van Pel, E. De Plaen, M. T. Duffour, G. Warnier, M. Perricaudet, and T. Boon. Induction of cytolytic T lymphocytes by immunization of mice with adenovirus containing a mouse homologue of the human MAGE-A genes, manuscript in preparation.
We thank M. Panagiotakopoulos for expert technical assistance, E. De Plaen for critically reviewing the manuscript, and J-P. Szikora for his help in computer analysis of protein sequences. The editorial assistance of S. Khaoulali and S. Mapp is gratefully acknowledged.