Involvement of the prostate and testis expression (PATE)-like proteins in sperm-oocyte interaction.

BACKGROUND: The prostate and testis expression (PATE)-like family of proteins are expressed mainly in the male genital tract. They are localized in the sperm head and are homologous to SP-10, the acrosomal vesicle protein also named ACRV1. Our aim was to characterize the expression and functional role of three PATE-like proteins in the testis and ejaculated sperm. METHODS: The expression and localization of PATE-like proteins in human testis biopsies (n= 95) and sperm cells were assessed by RT-PCR, immunohistochemistry and immunofluorescence staining (at least 600 sperm cells per specimen). The function of the PATE protein was tested by the hemizona assay and hamster egg penetration test (HEPT). RESULTS: PATE and PATE-M genes and proteins were present almost exclusively in germ cells in the testis: immunoflourescence showed that the percentage of germ cells positive for PATE, PATE-M and PATE-B was 85, 50 and 2%, respectively. PATE and PATE-M proteins were localized in the equatorial segment of the sperm head, while PATE-B protein was localized in the post-acrosomal region. A polyclonal antibody (Ab, at 1:50 and 1:200 dilutions) against the PATE protein did not inhibit sperm-zona binding in the hemizona assay (hemizona index of 89.6 ± 10 and 87 ± 36%, respectively). However, there was inhibition of sperm-oolemma fusion and penetration in the HEPT (penetration index: without Ab 7 ± 3.9; Ab dilution of 1:100, 4 ± 3.5; Ab dilution of 1:20, 0.6 ± 1.2, P < 0.001). CONCLUSIONS: Our data suggest that PATE protein is involved in sperm-oolemma fusion and penetration but not sperm-zona binding.

Muc13, a novel human cell surface mucin expressed by epithelial and hemopoietic cells.

Transmembrane mucins are glycoproteins involved in barrier function in epithelial tissues. To identify novel transmembrane mucin genes, we performed a tblastn search of the GenBanktrade mark EST data bases with a serine/threonine-rich search string, and a rodent gene expressed in bone marrow was identified. We determined the cDNA sequence of the human orthologue of this gene, MUC13, which localizes to chromosome band 3q13.3 and generates 3.2-kilobase pair transcripts encoding a 512-amino acid protein comprised of an N-terminal mucin repeat domain, three epidermal growth factor-like sequences, a SEA module, a transmembrane domain, and a cytoplasmic tail (GenBanktrade mark accession no. ). MUC13 mRNA is expressed most highly in the large intestine and trachea, and at moderate levels in the kidney, small intestine, appendix, and stomach. In situ hybridization in murine tissues revealed expression in intestinal epithelial and lymphoid cells. Immunohistochemistry demonstrated the human MUC13 protein on the apical membrane of both columnar and goblet cells in the gastrointestinal tract, as well as within goblet cell thecae, indicative of secretion in addition to presence on the cell surface. MUC13 is cleaved, and the beta-subunit containing the cytoplasmic tail undergoes homodimerization. Including MUC13, there are at least five cell surface mucins expressed in the gastrointestinal tract.

Analysis of the promoter of the MUC1 gene overexpressed in breast cancer.

The MUC1 gene encodes a mucin glycoprotein and is overexpressed in breast cancer. Knowledge of the mechanisms leading to MUC1 overexpression may help in the development of molecular approaches for breast cancer therapy. In order to study the regulation of the MUC1 gene transcription, we analyzed functional activities of various deletion mutants of the MUC1 promoter. We established that transcriptional cis-elements present in the SacI/XmnI fragment of the promoter are competent and sufficient for expression of, at least, tandem repeats containing isoform(s) of the MUC1 protein. CAT transfection analysis showed that both the 3' and 5' regions of the SacI/XmnI fragment possess transcription activities. Promoter activities associated with the SacI/XmnI fragment were confirmed by a RNase protection assay, which demonstrated multiple transcription start sites (TSSs) in the MUC1 gene transcribed in epithelial T47D cells. We show that treatment of the T47D cells with TGFbeta1 leads to activation of additional TSSs in the MUC1 gene. The roles of the structural and functional properties of the MUC1 promoter in MUC1 gene transcription are discussed.

MUC1 isoform specific monoclonal antibody 6E6/2 detects preferential expression of the novel MUC1/Y protein in breast and ovarian cancer.

The products of the MUC1 gene are known to be highly expressed in human breast cancer cells. The best characterized MUC1 protein is a polymorphic, type 1 transmembrane molecule containing a large extracellular domain composed primarily of a variable number of 20 amino acid tandem repeats. We have recently identified a novel protein product of the MUC1 gene, the MUC1/Y protein, that is also a transmembrane protein but is devoid of the tandem repeat array and its immediate flanking sequences. To analyze its expression in tumor cells we generated monoclonal antibodies directed against the MUC1/Y extracellular domain (anti-MUC1/Yex MAbs). Epitope mapping identified the MAb, 6E6, which recognized the MUC1/Y isoform with exquisite specificity- the repeat-array-containing MUC1 isoform could not compete out this immunoreactivity. A 30mer peptide which is unique for MUC1/Y and corresponds to the "join" region generated by the MUC1/Y specific splice, abrogated all 6E6 MAb immunoreactivity towards MUC1/Y. Immunoprecipitation of the MUC1/Y protein with 6E6 MAbs revealed that, in contrast with the proteolytic cleavage of the tandem-repeat-array-containing MUC1 isoform, MUC1/Y is not cleaved. Flow cytometry analyses using the 6E6 MAbs demonstrated that the MUC1/Y isoform is expressed on the cell surface of both MCF-7 breast cancer cells and malignant epithelial cells present in effusions obtained from breast and ovarian cancer patients. Our results unequivocally establish that the MUC1/Y protein is expressed on the surface of breast cancer cells and cells of other epithelial malignancies. The anti-MUC1/Y MAbs described here can target MUC1/Y expressing tumor cells in vivo and are likely to be important reagents both for epithelial tumor diagnosis and immunotherapy.

The breast cancer-associated MUC1 gene generates both a receptor and its cognate binding protein.

MUC1 proteins, some of which contain a mucin-like domain and others lacking this region, can be generated from the human breast cancer-associated MUC1 gene by alternative splicing. The MUC1/Y isoform is devoid of the mucin domain and is a cell membrane protein that undergoes transphosphorylation on both serine and tyrosine residues. We have identified cognate binding proteins that specifically interact with the extracellular domain of MUC1/Y. Coimmunoprecipitation analyses clearly revealed the presence of complexes composed of MUC1/Y and its cognate binding proteins in primary breast tumor tissue. MUC1/Y-expressing mammary tumor cells can be specifically targeted, in vivo, with the labeled cognate binding protein. The k(D) of MUC1/Y for its binding proteins was estimated as 1.2 nM. The MUC1/Y binding proteins are also derived from the MUC1 gene and represent the secreted mucin-like polymorphic MUC1 proteins MUC1/SEC and MUC1/REP, which contain a tandem repeat array. Whereas nonposttranslationally modified MUC1/Y bound efficiently to MUC1/SEC, the latter mucin-like protein had to be posttranslationally modified in a cell-type specific manner to bind MUC1/Y. The interaction of MUC1/Y with MUC1/SEC has important biological functional correlates: (a) it induces MUC1/Y phosphorylation; and (b) it has a pronounced effect on cell morphology. These findings suggest that MUC1/Y and MUC1/SEC form an active receptor/ cognate binding protein complex that can elicit cellular responses. The proteins comprising this complex are, thus, generated by alternative splicing from one and the same gene, namely the MUC1 gene.

Preferential expression of novel MUC1 tumor antigen isoforms in human epithelial tumors and their tumor-potentiating function.

The human MUC1 gene expresses at least 2 type 1 membrane proteins: MUC1/REP, a polymorphic high m.w. MUC1 glycoprotein often highly expressed in breast cancer tissues and containing a variable number of tandem 20 amino acid repeat units, and the MUC1/Y protein, which lacks this repeat array and, therefore, is not polymorphic. Despite their documented importance in signal transduction processes, the relative expression of the 2 isoforms in epithelial tumors is unknown. Using antibody reagents which recognize different MUC1 domains, the expression of these isoforms in malignant epithelial cells has been evaluated. A comparison of the amounts of the 2 isoforms revealed preferential expression of the novel MUC1/Y protein in breast cancer tissue samples. Furthermore, although the MUC1/REP protein is almost undetectable in HeLa cervical adenocarcinoma epithelial cells, the MUC1/Y isoform is extensively expressed in these cells. The presence of the MUC1/Y sequence as well as that of an additional tandem-repeat-array-lacking isoform, designated MUC1/X, were demonstrated by reverse transcriptase PCR amplification of RNA extracted from HeLa and ovarian carcinoma cells. It has been shown previously that the MUC1 cytoplasmic domain interacts with the SH2 domain containing GRB2 protein, which transduces signals to ras, a protein which in its activated form can lead to cell transformation. We present here data demonstrating that MUC1/Y isoform expression increases the tumorigenic potential of DA3 mouse mammary epithelial cells; in contrast, potentiation of tumorigenicity is not observed with MUC1/REP expression. Our studies thus demonstrate that expression of the MUC1 gene in epithelial tumors can give rise to substantial levels of MUC1 proteins devoid of the tandem repeat array, which are generated by alternative splicing mechanisms.

Detection of a secreted MUC1/SEC protein by MUC1 isoform specific monoclonal antibodies.

Although MUC1 proteins are known to be secreted by breast cancer cells, the mechanism of their release from the cell is still obscure. Our previously reported MUC1 cDNA sequences suggested the existence of a secreted MUC1 isoform, MUC1/SEC, that includes a sequence of intron 2, and terminates prematurely at a stop codon within this intron. It is thus devoid of a transmembrane domain. As no formal evidence for MUC1/SEC expression at the protein level had been provided, we generated monoclonal antibodies (mAbs) against a peptide sequence (sec peptide) that is unique for the MUC1/SEC protein. Two anti-sec peptide mAbs were obtained which reacted strongly with (a) the immunizing peptide, (b) recombinant MUC1/SEC protein, and (c) MUC1 proteins secreted from breast cancer cells. The immunoreactivity of the anti-sec peptide mAbs with MUC1 proteins secreted by breast cancer cells was specifically inhibited by the sec peptide-it was completely unaffected by a peptide sequence that represents a MUC1 repeat motif. Significantly, the anti-sec peptide mAbs also detected MUC1/SEC protein in sera of breast cancer patients. We have established here that these mAbs recognize the MUC1/SEC isoform via a peptide sequence which is unique for the MUC1/SEC protein. Our studies thus demonstrate that the MUC1/SEC protein is a bona-fide MUC1 isoform and that its expression may contribute to the secretion of MUC1 proteins by secretory epithelial cells in general and breast cancer cells in particular.

Preoperative diagnosis of thyroid papillary carcinoma by reverse transcriptase polymerase chain reaction of the MUC1 gene.

As thyroid nodules are common, it is imperative to recommend operation only to those with a high risk of malignancy. Fine-needle aspiration (FNA) biopsy, which is widely used for this purpose, is limited by the considerable rate of non-diagnostic or non-interpretable conclusions. Therefore, it is highly desirable to acquire a new diagnostic means for thyroid cancer. We have recently described a marked over-expression of the MUC1 gene in thyroid papillary-carcinoma tissue, as compared with various benign thyroid pathologies. As the amount of mRNA obtainable by FNA is not amenable to hybridization analysis, we amplified the mRNA sequence of the MUC1-gene upstream of the variable number tandem repeat array using the reverse-transcription polymerase chain reaction (RT-PCR). Seven out of 8 FNA samples obtained from thyroid papillary carcinoma resulted in 336 and 309 base-pair products. In contrast, in all 13 FNA samples obtained from various benign pathologies, only the smaller RT-PCR product was observed. Sequence analysis of the RT-PCR products indicates that alternative splicing of the exon 2 acceptor site accounts for the difference between the 2 amplification products. It is suggested that RT-PCR of the MUC1-gene transcript may add a biomolecular diagnostic dimension to the routine cytological preoperative FNA diagnosis.

DNA methylation status of the MUC1 gene coding for a breast-cancer-associated protein.

The MUC1 gene codes for protein products that are highly expressed in human breast-cancer tissue and that serve as tumor markers for disease progression. The factors contributing to the disease-specific over-expression of the MUC1 gene are under intensive investigation and are yet to be determined. A large transcribed region of the human MUC1 gene is a CpG island that consists of 60-bp tandemly repeating units, each of which contains one SmaI restriction site. The methylation status of regulatory regions, upstream to the transcriptional start site, is essential for the regulation of gene expression. We therefore evaluated whether the methylation status of the various regions of the MUC1 gene may affect its expression. Using SmaI, and its isoschizomer XmaI endonucleases, we demonstrated that in peripheral-blood leukocytes (PBL-DNA) that do not express the MUC1 gene, the repeat array is completely methylated, whereas the same sequences are entirely non-methylated in breast-tumor-tissue DNA (BT-DNA). In contrast, sequences upstream and downstream to the repeat array showed no difference in the methylation pattern in PBL-DNA and BT-DNA. Hypomethylation within the repeat array was also observed in other epithelial tissues that express the MUC1 gene at much lower levels to those seen in breast-cancer tissue. These studies demonstrate that hypomethylation of the tandem repeat array is an absolute requirement for MUC1 gene expression in epithelial tissues, although in breast-cancer tissue additional regulatory mechanisms must pertain for its over-expression.

Tyrosine phosphorylation of the MUC1 breast cancer membrane proteins. Cytokine receptor-like molecules.

Phosphorylation on tyrosine residues is a key step in signal transduction pathways mediated by membrane proteins. Although it is known that human breast cancer tissue expresses at least 2 MUC1 type 1 membrane proteins (a polymorphic high molecular weight MUC1 glycoprotein that contains a variable number of tandem 20 amino acid repeat units, and the MUC1/Y protein that is not polymorphic and is lacking this repeat array) their function in the development of human breast cancer has remained elusive. Here it is shown that these MUC1 proteins are extensively phosphorylated, that phosphorylation occurs primarily on tyrosine residues and that following phosphorylation the MUC1 proteins may potentially interact with SH2 domain-containing proteins and thereby initiate a signal transduction cascade. As with cytokine receptors, the MUC1 proteins do not harbor intrinsic tyrosine kinase activity yet are tyrosine phosphorylated and the MUC1/Y protein participates in a cell surface heteromeric complex whose formation is mediated by two cytoplasmically located MUC1 cysteine residues. Furthermore, the MUC1/Y protein demonstrates sequence similarity with sequences present in cytokine receptors that are known to be involved in ligand binding. Our results demonstrate that the two MUC1 isoforms are both likely to function in signal transduction pathways and to be intimately linked to the oncogenetic process and suggest that the MUC1/Y protein may act in a similar fashion to cytokine receptors.

Characterization and molecular cloning of a novel MUC1 protein, devoid of tandem repeats, expressed in human breast cancer tissue.

The human breast cancer marker protein, MUC1, is a polymorphic transmembrane molecule containing a large extracellular domain that is primarily composed of a variable number of highly conserved 20-amino-acid tandem repeats. We report here the detection of a novel invariantly sized 1.2-kb MUC1 mRNA, in addition to the large polymorphic mRNAs, by probing Northern blots with MUC1-cDNA-unique-sequence probes. The nucleotide sequence of this novel MUC1 mRNA demonstrates that it is identical to the MUC1 cDNA sequences downstream and upstream to the tandem-repeat array of the transmembrane form of MUC1. However, it contains neither the central tandem repeat array itself nor its directly flanking sequences that are deleted by a differential splicing event utilizing splice acceptor and donor sequences 5' and 3' to the tandem-repeat array. The splice event retains, downstream to the splice acceptor site, an open reading frame identical to that of the repeat-array-containing MUC1 thereby generating the novel MUC1/Y protein. Cells transiently transfected with the novel MUC1/Y cDNA express the MUC1/Y protein that is modified by glycosylation. The MUC1/Y protein is also readily detected in human breast cancer cells grown in vitro. Furthermore, primary breast cancer tissue samples demonstrate significant levels of the MUC1/Y protein whereas expression in tissue adjacent to the tumor is undetectable. Molecular characterization presented here, of the novel MUC1/Y molecule lacking the repeat array, suggests that it is likely to play a role distinct to that of the polymorphic repeat-array-positive MUC1 protein and that it may act as a new marker protein for human breast cancer.

Does a novel form of the breast cancer marker protein, MUC1, act as a receptor molecule that modulates signal transduction?

Molecular analysis of a protein highly expressed in human breast cancer, indicates the presence of a polymorphic tandem repeat domain that encodes a conserved 20 amino acid repeat motif rich in serine and threonine residues that in the mature protein, designated MUC1, are linked via O-glycosidic linkages to sugar residues. Recent studies performed in our laboratory have led to the molecular characterization of a novel MUC1 repeat array minus mRNA, generated by an alternative splicing event that deletes the central tandem repeat array and its flanking sequences. The conceptually derived amino acid sequence of the novel MUC1 protein shows that it is identical with the previously reported transmembrane MUC1 amino acid sequence except for the deletion of the central 20 amino acid tandem repeat array and sequences immediately flanking the repeat array. This indicates that the novel MUC1 protein, which is devoid of the "hallmark" feature of mucins, the tandem repeat array, may be functionally different to the much larger, heavily glycosylated polymorphic repeat array containing MUC1 proteins, that affect cell-cell interactions. Based on an analysis of its peptide sequence, we propose the hypothesis that the novel MUC1 protein may act as a receptor molecule that modulates signal transduction. Preliminary experimental data supports this hypothesis. It appears, therefore, that the MUC1 gene is multifunctional with regard to its protein products- the repeat array containing MUC1 proteins may alter cellular adhesion processes whereas the novel MUC1 protein could be acting as a receptor-like molecule participating in signal transmission.

Recognition of peptidyl epitopes by polymorphic epithelial mucin (PEM)-specific monoclonal antibodies.

Peptidyl epitope recognition by several murine monoclonal antibodies (MAbs E29, H23, HMFG-1, HMFG-2, MA5, MA6 and MA9) which react with the polymorphic epithelial mucins [PEM; epithelial membrane antigen (EMA)] was studied by using ten synthetic peptides representative of the 20 residue tandem repeat as test antigens. Antibody binding to 6-10 residue overlaps and to peptides having a common carboxy-terminus and staggered amino-termini (8-31 residues) was assessed by solid phase and competition ELISA techniques. From these analyses, all MAbs except MA9 were found to react predominantly with the carboxy-terminal half of the repeat motif. Polyclonal antibody responses in mice immunized with intact EMA/PEM-containing preparations also displayed significant reactivities against synthetic repeat peptide antigens and, conversely, synthetic peptides as carrier-conjugated immunogens induced antibodies recognizing intact antigens. These results are discussed vis-à-vis peptide conformation, the potential effects of O-glycosylation on secondary structure, and the possible effects of these parameters on immunogenicity and antigenicity.

Vaccinia recombinants expressing secreted and transmembrane forms of breast cancer-associated epithelial tumour antigen (ETA).

Monoclonal antibody H23 identifies a polymorphic epithelial tumour antigen (ETA) that is aberrantly expressed in breast cancer and which may afford a target for active immunotherapy. We recently reported the cloning of H23-ETA genomic and cDNA clones. H23-ETA contains a multiple internal tandem repetition of a 20 amino acid motif and sequence analysis predicted two mRNA species encoding different ETA proteins, one harbouring a C-terminal potentially transmembrane hydrophobic zone (T) and a second form (S) that lacks this zone. We report that both RNA species can be detected in breast cancer cells. To further characterize the encoded proteins we have constructed vaccinia virus recombinants, VV-ETA-S and VV-ETA-T, separately expressing the alternative forms. Despite selective loss of internal tandem repeat elements during propagation of recombinant vaccinia, the encoded polypeptides were efficiently recognized by H23 monoclonal antibody. Immunoprecipitation revealed that ETA encoded by the S recombinant was secreted into the culture medium whereas the T form remained tethered at the cell surface. Both forms were readily detected in infected cells by immunofluorescence. Abnormal mobility of the T polypeptide indicated post-translational cleavage that may permit the extracellular domain of the T-polypeptide to be shed from the cell surface. Further, fluorescence-activated cell sorting analysis shows that the S form of the polypeptide is also partly present at the cell surface. Vaccinia recombinants expressing ETA may be of utility in the active immunotherapy of breast cancer.

Expression of a gene coding for breast tumor-associated antigen in thyroid papillary carcinoma.

The expression of the H23 gene, previously shown to be overexpressed in breast cancer tissue, was examined in various thyroid pathologies. Thyroid papillary carcinomas demonstrate significant H23 mRNA levels, whereas benign thyroid pathologies have very low levels of expression. H23 gene expression in thyroid cancer inversely correlated with that of thyroperoxidase and thyroglobulin genes. Immunoblot assays of thyroid cancer tissues revealed overexpression of H23 gene at the protein level as well The data presented indicate that dedifferentiation of thyroid tissue to the malignant state is associated with increased H23 gene expression and suppression of some thyroidal differentiation marker genes.

Isolation and characterization of an expressed hypervariable gene coding for a breast-cancer-associated antigen.

A human gene and cDNA coding for a breast-cancer-associated antigen (H23Ag) were isolated and characterized. The gene contains two exons and one intron. Part of the second exon is a tandem repeat array (TRA) consisting of multiple 60-bp G + C-rich units. We report here the characterization of unique sequences that are found in the H23Ag gene and cDNA, in addition to the 60-bp repeats. Analysis of the cDNA sequences revealed a putative ATG start codon preceded by two overlapping initiation consensus sequences (CCACC). The open reading frame determines an amino acid (aa) sequence consisting of three regions. The first region contains an initiating methionine and a highly hydrophobic putative signal peptide. This is followed by a variable number of highly conserved 20-aa repeat units (TRA). The last region, C-terminal to TRA, contains four potential N-linked glycosylation sites. The genomic nucleotide sequences demonstrate a putative promoter region that includes a 'TATA' box. A putative estrogen regulatory element is located 5' to the promoter region. The characterization of the gene and cDNA coding for the H23Ag presented here, may help to elucidate its possible function in human breast cancer.

Multiple protein forms of the human breast tumor-associated epithelial membrane antigen (EMA) are generated by alternative splicing and induced by hormonal stimulation.

Cloning and sequencing of epithelial membrane antigen (EMA) has demonstrated the existence of a variable number of tandem repeats (VNTR) flanked by unique sequences, and alternative splicing has been proposed to result in secreted and membrane-bound antigenic forms. Antisense oligonucleotides, specific for the VNTR region and various alternative splice forms, were used as probes to define EMA transcripts in poly A+ RNA from a mucinous breast tumor cell line. The BT549 line has been shown to exhibit enhanced expression and secretion of EMA when the cells are cultivated in a medium supplemented with hydrocortisone and insulin, and Northern blot analysis demonstrated that EMA-related RNA transcripts are commensurately enhanced. As a result of the large increase in EMA RNA levels, two major transcripts in BT549 have been identified as coding for either the secreted or transmembrane EMA forms and two antigenic forms have been immunoprecipitated from BT549 cell layer and medium translation products.

Expression of genes coding for pS2, c-erbB2, estrogen receptor and the H23 breast tumor-associated antigen. A comparative analysis in breast cancer.

Expression of the gene coding for a new breast tumor-associated antigen, H23, was compared to expression of genes coding for pS2, c-erbB2 and estrogen receptor (ER). Comparison involved mRNA expression in normal and malignant breast tissues as well as in non-breast tumors. Results obtained by RNA dot blot and Northern hybridizations showed that expression of the H23 antigen coding gene is a discriminatory marker in human breast cancer. It is expressed in 92% of breast tumors whereas 69%, 62% and 56% of breast tumors demonstrate significant mRNA levels of c-erbB2, ER and pS2, respectively. Non-malignant or normal breast tissue expresses much lower levels of the H23 antigen mRNA. From the comparative analysis presented here it is concluded that the gene coding for H23 antigen furnishes a most useful marker for human breast cancer.

Human epithelial tumor antigen cDNA sequences. Differential splicing may generate multiple protein forms.

The isolation and characterization of complementary DNAs (cDNAs) which code for an epithelial antigen aberrantly expressed in human breast tumor tissue are described here. The only information regarding the primary structure of this potentially important antigen has been a 20-amino-acid repeat motif. We now report the complete amino acid sequences of different forms of the human epithelial tumor antigen as deduced from the nucleotide sequence of isolated non-repeat cDNAs. The diversity of protein forms is generated by a series of alternative splicing events that occur in the regions located upstream and downstream to a central tandem repeat array. Isolated cDNAs coding for the upstream region show that differential usage of alternative splice acceptor sites may generate two protein forms containing putative signal peptides of varying hydrophobicities. The complexity of possible antigen forms is further compounded by alternative splicing events occurring in the region 3' to the repeat array. The isolated cDNAs 3' to the tandem repeats indicate that whereas one mRNA transcript is colinear with the gene, and defines an open reading frame (ORF) containing 160 amino acids downstream to the repeat array, a second cDNA correlates with a mRNA that is generated by a series of splicing events. The deduced amino acid sequence of the spliced cDNA contains an ORF that is identical for 149 amino acids downstream to the repeat array with the amino acid sequence of the unspliced cDNA. At this point it diverges and continues for an additional 179 amino acids. The sequence contains a highly hydrophobic 28-amino-acid peptide, located towards the carboxyl terminus, that may correspond to a transmembrane region. The cDNAs and deduced amino acid sequences, presented here, define the complete amino acid sequences of the epithelial tumor antigen and demonstrate the existence of multiple protein forms that probably localize to different cellular and extracellular compartments.