Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer Int. J. Cancer: 71, 741–749 (1997) r 1997 Wiley-Liss, Inc. PREFERENTIAL EXPRESSION OF NOVEL MUC1 TUMOR ANTIGEN ISOFORMS IN HUMAN EPITHELIAL TUMORS AND THEIR TUMOR-POTENTIATING FUNCTION Amos BARUCH1,2, Mor-li HARTMANN1,2, Sheila ZRIHAN-LICHT1, Shulamit GREENSTEIN1, Matti BURSTEIN1, Iafa KEYDAR1, Mordechai WEISS1, Nechama SMORODINSKY1 and Daniel H. WRESCHNER1* 1Department of Cell Research and Immunology, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Ramat Aviv 69978, Israel 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. Int. J. Cancer 71:741–749, 1997. r 1997 Wiley-Liss, Inc. The MUC1 gene is frequently expressed with high intensity in human carcinomas, especially in breast cancer tissue (Ceriani et al., 1977; Burchell et al., 1987). In these patients, augmented concentrations of MUC1 proteins (previously referred to as episialin, H23Ag; ETA, epithelial tumor antigen; PEM, polymorphic epithelial mucin; EMA, epithelial membrane antigen; CA15-3; MCA, mammary carcinoma antigen) may be one of the earliest signs of disease progression, and, as such, the MUC1 gene has aroused much interest. A major MUC1 gene product (designated MUC1/ REP) is a polymorphic type 1 transmembrane molecule which consists of a large, heavily glycosylated extracellular domain, a transmembrane domain and a 72 amino acid cytoplasmic tail (Wreschner et al., 1990; Hareuveni et al., 1990; Ligtenberg et al., 1990; Gendler et al., 1990). This isoform is proteolytically cleaved early after translation (Ligtenberg et al., 1992a) and is shed from the cell by an unknown mechanism. The MUC1 isoform, MUC1/Y, generated by alternative splicing (Zrihan-Licht et al., 1994b), has a molecular mass of 42–45 kDa and, in contrast to MUC1/REP, does not undergo proteolytic cleavage (Zrihan-Licht et al., 1994b). Furthermore, this novel MUC1 isoform retains the N-terminal as well as the transmembrane and cytoplasmic MUC1 domains but lacks the tandem repeat array that has been considered the hallmark of MUC1. To date, the relative extent of expression in epithelial tumors of the MUC1 protein containing the tandem repeat array vs. that of the novel isoform devoid of the tandem repeats has not been studied. To address this issue, we determined the expression of MUC1 gene products in epithelial tumors using not only a panel of antibodies directed against the 20 amino acid tandem repeat motif but also antibodies which specifically recognize the MUC1/Y cytoplasmic and extracellular domains. Reverse transcriptase PCR (RT-PCR) was used to assess the expression of MUC1 isoforms devoid of the tandem repeat array. Our studies demonstrate that the MUC1 isoforms lacking the tandem repeat array not only are extensively expressed in epithelial tumors but in certain cases they are almost the only MUC1 protein products observed. Significantly, the tyrosine and, to a lesser extent, the serine residues of the cytoplasmic domains of the MUC1 proteins undergo phosphorylation (Zrihan-Licht et al., 1994a). One study confirmed the phosphorylation of the MUC1 tyrosine residues (Pandey et al., 1995) and demonstrated in vivo interaction between the MUC1 cytoplasmic domain and the GRB-2 protein. As membrane receptor proteins participating in signal transduction processes that involve GRB-2, and thereby ras, are often important for tumorigenesis, we analyzed whether MUC1 expression can affect the tumorigenic potential of MUC1 isoform transfectants. We show that the novel MUC1/Y protein, in contrast to MUC1/REP, can increase the transforming potential of DA3 mouse mammary tumor cell transfectants. Our findings suggest that MUC1 isoforms lacking the tandem repeats may act as novel marker proteins of malignant epithelial cells and play an active role in the oncogenetic process. It is thus likely that the novel tandem repeat array lacking MUC1 isoforms described here perform functions distinct from those of the MUC1/REP protein. MATERIAL AND METHODS Cells Stable transfectants expressing MUC1 isoforms were generated by co-transfecting an expression plasmid harboring either MUC1/ REP or MUC1/Y cDNA with the neomycin plasmid (pSV2 neo) selection marker into 3T3 ras-transformed fibroblasts (Wreschner et al., 1990; Hareuveni et al., 1990; Zrihan-Licht et al., 1994a, b) or DA3 mouse mammary tumor cells (Fu et al., 1990). The cell line T47D is derived from a human mammary carcinoma and is a high MUC1 expressor. HeLa cells are derived from a uterine cervical adenocarcinoma and express low levels of the MUC1/REP isoform. All cell lines were grown in DMEM supplemented with 10% FCS and 2 mM L-glutamine. cDNA constructs Generation of expression vectors harboring either the full-length transmembrane MUC1/REP or the novel MUC1/Y cDNA and driven by the HMG co-enzyme A reductase promoter (expression Contract grant sponsors: Israel Cancer Association, Israel Cancer Research Fund, Israel Academy of Sciences, Federico Fund of Tel Aviv University, Barbara Friedman Fund, Simko Chair for Breast Cancer Research. *Correspondence to: Department of Cell Research and Immunology, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Ramat Aviv 69978, Israel. Fax: 972-3-6422046. 2The first two authors contributed equally to these studies. Received: 10 September 1996; accepted 6 January 1997 742 BARUCH ET AL. vector pCL642) has been described previously (Wreschner et al., 1990; Hareuveni et al., 1990; Zrihan-Licht et al., 1994a,b). Western blot analyses Cell lysates were prepared by adding lysis buffer (50 mM NaCl, 20 mM Tris-HCl [pH 7.4], 100 µg/ml leupeptin and 0.5% Nonidet P-40) either to cell pellets or to surgically resected breast tissues, pulverized under liquid nitrogen, followed by vortex mixing and sonication (3 3 10 sec bursts using a Branson sonicator). Cell debris were removed by centrifugation at 12,000g for 10 min. All procedures were performed at 4°C or on ice. Protein samples were denatured by boiling in SDS-buffer containing mercaptoethanol and analyzed on SDS/acrylamide gels. Gels were electrotransferred for 3 hr at 1 A to nitrocellulose filters, which were then blocked in PBS containing 5% skimmed milk, followed by incubation with the primary antibody. Filters were washed in PBS and then incubated with a secondary anti-rabbit (or anti-mouse) antibody conjugated to horseradish peroxidase followed by ECL (Amersham, Aylesbury, UK) detection. Antibodies Western blot analyses were performed with (i) monoclonal antibodies (H23 MAbs) that recognize an epitope contained within the tandem repeat domain of the MUC1/REP protein (Keydar et al., 1989), (ii) a polyclonal antibody (a gift from Dr. S. Gendler, London, UK) directed against the 17 C-terminal amino acids of the MUC1 cytoplasmic domain, (iii) an affinity-purified polyclonal antibody (designated anti-MUC1/Yex) prepared by injecting rabbits with the MUC1/Y extracellular domain synthesized as a recombinant protein using the PET expression system according to the protocol described below, and (iv) affinity-purified antibodies recognizing MUC1/Y extracellular domain sequences that are N-terminal to the site of proteolytic cleavage in the MUC1/REP protein (described below). Generation of bacterial recombinant MUC1/Yex protein and truncated MUC1/Yex protein DNA coding for the MUC1/Y extracellular domain was generated by PCR using an up-stream oligonucleotide (ACAAGTACTGGTCATGCA AGCTCTACCCCAGGTGGAG) that introduces a Sca1 restriction enzyme site (AGTACT) at the 58 terminus. The first nucleotide of the sequence CTGGTCA corresponds to the MUC1/Y nucleotide at position 128 (Zrihan-Licht et al., 1994b) and the trinucleotide ACT codes for an N-terminal threonine in place of the serine located at amino acid 24 in the MUC1/Y sequence (Zrihan-Licht et al., 1994b) and represents the first N-terminal amino acid following cleavage of the MUC1/Y signal peptide sequence. The down-stream anti-sense oligonucleotide used introduces a termination codon after the nucleotide sequences coding for the amino acid sequence SAQSGAGV (amino acids 149–156; Zrihan-Licht et al., 1994b) located just up-stream to the MUC1/Y transmembrane domain; the termination codon is then followed by a BamH1 site. For PCR amplification, the MUC1/Y cDNA was used as a template with the above up-stream and down-stream primers. The PCR product obtained was cleaved with Sca1 and BamH1 and ligated to Stu1- and BamH1-cleaved pET11d-DHFR vector DNA (this pET 11d vector [Novagen, Madison, WI] codes for a non-relevant bacterial dihydrofolate reductase protein that at its carboxyl terminus contains the factor Xa cleavage site [amino acids IEGR] followed by Stu1 and BamH1 restriction sites; the vector was obtained from Dr. M. Mevarech [Tel Aviv University]). The recombinant plasmid pET11d-DHFRMUC1/Yex was introduced into BL21(DE3)pLysS bacteria, and recombinant protein was induced by the addition of IPTG. Affınity purification of anti-MUC1/Yex antibodies The recombinant DHFR-MUC1/Yex fusion protein (see above) was purified from inclusion bodies by preparative SDS-PAGE and injected with Freund’s adjuvant into New Zealand white rabbits. The anti-MUC1/Yex serum obtained was purified by passage through an Affigel-10 (Bio-Rad, Hercules, CA) affinity column to which N-terminal histidine tagged recombinant MUC1/Yex protein (see below) had been covalently bound. The use of this affinity resin ensures that only antibodies recognizing the MUC1/Y extracellular domain will be purified. Anti-MUC1/Yex antibodies were eluted from the affinity column with 0.1 M glycine-HCl, pH 2.7, buffer. Eluted fractions were immediately neutralized by the addition of 1 M Tris base. Generation of N-terminal histidine tagged recombinant MUC1/Yex protein The expression vector coding for N-terminal histidine tagged recombinant MUC1/Yex protein was generated by utilizing the pET16b (Novagen) expression system. The MUC1/Y extracellular domain (nucleotides 127–525; Zrihan-Licht et al., 1994b) was amplified by PCR using the full-length MUC1/Y cDNA as template and up-stream and down-stream primers that introduced Nde1 and BamH1 sites at the 58 and 38 termini, respectively. The purified PCR fragment was cleaved with Nde1 and BamH1 and ligated to Nde1/BamH1 cleaved pET16b vector DNA. Recombinant DNA plasmid was introduced into BL21(DE3)pLysS bacteria and recombinant protein prepared as follows. An overnight culture of recombinant bacteria was diluted 1:100 into fresh LB medium containing 100 µg/ml ampicillin. After the culture reached an optical density (at 600 nm) of 0.3, IPTG was added to a final concentration of 1 mM and cells were grown for an additional 5 hr. Cells were centrifuged at 5,000 g for 10 min, and the bacterial pellet was washed with 0.9% NaCl, resuspended in lysis buffer (0.1 M phosphate buffer [pH 7.0], 10 mM EDTA, 0.1% Triton X-100; 30 ml/l culture) followed by vortexing and sonication (5 3 10 sec bursts, using a Branson sonicator). Inclusion bodies were pelleted by centrifugation at 12,000g for 10 min. The pellet was resuspended in buffer A (10 mM Tris-HCl [pH 8.0], 50 mM Na2HPO4, 6 M guanidine HCl; 5 ml/g of cell pellet wet weight) and incubated for 30 min in the same buffer at 25°C. The suspension was centrifuged at 10,000 g and the supernatant incubated with a 50% Ni-agarose slurry for 30 min at 25°C. The slurry was loaded onto a 2 cm diameter column and washed with 10 column volumes of buffer A, 20 column volumes of buffer B (10 mM Tris-HCl [pH 8.0], 50 mM Na2HPO4, 8 M urea) and buffer C (10 mM Tris-HCl [pH 6.3], 50 mM Na2HPO4, 8 M urea) until an optical density at 280 nm of less than 0.01 was obtained. Elution was performed with 150 mM imidazole in buffer C. For protein renaturation, fractions containing recombinant protein were loaded onto a 50 ml Sephadex G-50 column pre-equilibrated with 0.1 M NaHCO3 (pH 8.4), and 0.5 ml elution fractions were collected. Generation of truncated MUC1/Y extracellular domain protein Generation of bacterial recombinant MUC1/Y truncated extracellular domain cDNA was performed utilizing the pET11d expression system. The truncated MUC1/Yex construct (designated MUC1/Y/Ssp1) was generated by replacing the entire MUC1/Yex coding region of the pET11d-DHFR-MUC1/Yex vector with a PCR fragment (nucleotides 128–328) that extends to the Ssp1 site at nucleotide 328 and introduces a stop codon immediately down-stream to the Ssp1 site, using full-length MUC1/Y cDNA as template. BL21 bacteria were transformed with the pET11d-DHFRMUC1/Y/Ssp1 vector coding for the truncated protein. The MUC1/ Y/Ssp1 truncated protein was purified from inclusion bodies as described above and the purified protein covalently bound to an Affigel-10 resin. To isolate antibodies recognizing MUC1/Y/Ssp1, the affinity-purified anti-MUC1/Yex antibodies (see above) were passed through the MUC1/Y/Ssp1 affinity column and the antibodies eluted with 0.1 M glycine-HCl buffer (pH 2.7) as described above. RT-PCR RNA isolated from HeLa or Lucy ovarian carcinoma cells was reverse-transcribed using the anti-sense down-stream oligonucleotide primer ‘‘c’’ (complementary to nucleotides 1050–1018, nucleotide numbering for location of the primers as in Wreschner et al., 1990), and 10% (2 µl) of the reverse transcription reaction NOVEL MUC1 ISOFORMS IN EPITHELIAL TUMORS underwent PCR amplification in a final volume of 40 µl with the same down-stream primer and up-stream sense oligonucleotide primer ‘‘a’’ (nucleotides 18–42), using 35 cycles of 95°C, 30 sec; 55°C, 60 sec; and 72°C, 60 sec. PCR products were purified by agarose gel electrophoresis and, in some cases, subjected to nested PCR amplification using the same down-stream anti-sense primer (c) and the up-stream primer (b) corresponding to nucleotides 60–88. All reagents used for RT-PCR were obtained from a Gene Amp RNA PCR kit (Perkin Elmer Cetus, Norwalk, CT), and the manufacturer’s protocols were followed. DNA sequencing MUC1-specific RT-PCR products were subcloned into M13 and sequenced using the dideoxynucleotide chain-termination method. The M13 universal primer or synthetic oligonucleotides prepared according to known sequences were used to prime ssDNA. RESULTS To examine the levels of the various MUC1 isoforms in epithelial tissues, we assessed MUC1 expression with antibodies directed against various domains of the MUC1 proteins. Characterization of MUC1/REP expression with anti-repeat array antibodies Expression of MUC1 proteins in epithelial tumors was initially analyzed with H23 MAbs that recognize an epitope present within the 20 amino acid tandem repeat motif. As previously reported (Wreschner et al., 1990; Hareuveni et al., 1990; Ligtenberg et al., 1990; Gendler et al., 1990), these antibodies detect polymorphic high m.w. MUC1 proteins of different sizes and with molecular masses greater than 200 kDa (Fig. 1a). The polymorphism is due to varying lengths of the tandem repeat array included in the MUC1 alleles. Immunoreactive polymorphic MUC1 proteins migrating FIGURE 1 – MUC1/REP and MUC1/Y expression in tumor tissue. Identical protein lysates prepared from primary breast tumor tissue samples (BT), normal breast tissue (BN), paired tumor and normal breast tissues from the same patients (BT*, BN* and BT9, BN9), T47D breast cancer cells or HeLa adenocarcinoma cells were analyzed on (a) 6% or (b) 15% SDS/acrylamide gels. This was followed by Western blotting and probing with H23 MAbs that recognize an epitope within the tandem repeat array (a) or with polyclonal antibodies generated against the cytoplasmic domain (b). Detection was with horseradish peroxidase-conjugated secondary antibodies followed by enhanced chemiluminescence. Numbers to the left indicate the molecular mass in kDa. (a) Expression of repeat-array-containing MUC1 proteins detected with anti-repeat antibodies. The immunoreactive bands represent the high m.w. polymorphic repeat-array-containing MUC1 proteins. The mature (fully glycosylated) repeat-array-containing proteins present in the T47D breast cancer cells are indicated by the top 3 arrows to the left; the bottom 3 arrows indicate the core or partially glycosylated repeat-array-containing proteins in the T47D sample. (b) Expression of MUC1/Y proteins detected with anti-cytoplasmic domain antibodies. The mature 42–45 kDa MUC1/Y protein is indicated at the left and right of the figure by the full arrow; the arrowhead indicates the non-glycosylated MUC1/Y (see also c and d). Brackets to the left and right indicate the cytoplasmic domain-containing proteolytic cleavage products migrating with a molecular mass of 25–30 kDa. (d) The MUC1/Y protein is N-glycosylated. The MUC1/Y protein was expressed in 35S-methionine-labeled BSC-1 monkey epithelial cells that had been infected with recombinant vaccinia virus coding for T7 RNA polymerase and transfected with the pTM1 vaccinia expression vector harboring cDNA coding for the MUC1/Y protein, as described in Zrihan-Licht et al. (1994b). The labeled MUC1/Y protein was immunoprecipitated from cell lysates and the immunoprecipitate collected with protein A-Sepharose beads. The MUC1/Y protein was then incubated without or with N-glycanase (control and N-glyc, lanes 1 and 2, respectively), analyzed on SDS-polyacrylamide gel and visualized by autoradiography. The mature N-glycosylated 42–45 kDa MUC1/Y protein is indicated to the right by the full arrow; the arrowhead indicates the nonglycosylated MUC1/Y. 743 with molecular masses ranging from 100 to slightly more than 200 kDa were also observed; pulse-chase experiments previously showed that these represent the precursor MUC1/REP proteins (Hilkens and Buijs, 1988). Despite the high amount of MUC1/REP expression in the T47D breast cancer cell line as well as in several breast tumor samples, other breast tumor samples displayed only very low MUC1/REP protein amounts, as did normal breast tissue samples. Furthermore, exceptionally low to undetectable concentrations of MUC1/REP proteins were observed in HeLa cells (Fig. 1a). Ponceau red staining of the blotted proteins revealed that this lack of immunoreactivity was not due to variations in protein loading. Characterization of MUC1 isoform expression with anti-cytoplasmic domain antibodies shows preferential MUC1 isoform expression To test for differential expression of the MUC1/REP and MUC1/Y isoforms in the various epithelial tumor tissues, the 744 BARUCH ET AL. samples analyzed above were resubjected to Western blot analysis using antibodies that specifically bind to the C-terminal 17 amino acids of the MUC1 cytoplasmic domain and that do not recognize epitopes contained within the tandem repeat array. It has been shown that early after translation, the transmembrane form of the MUC1/REP protein undergoes proteolytic cleavage in a region that is 45–60 amino acids N-terminal to the transmembrane domain (Ligtenberg et al., 1992a), thereby generating a heterodimer MUC1/REP protein. This consists of the high molecular mass tandem-repeat-array-containing extracellular domain that is bound non-covalently in an SDS-sensitive linkage to the remaining C-terminal cleavage products containing the 45–60 extracellular MUC1 amino acids, the transmembrane domain and the MUC1 cytoplasmic tail, which migrate as diffuse immunoreactive species in the region of 25–30 kDa. Accordingly, Western blot analysis with anti-cytoplasmic domain antibodies of proteins resolved by SDS/15% acrylamide gels clearly demonstrated that, in the high MUC1/REP-expressing samples, immunoreactive species migrated as a diffuse band in the region of 25–30 kDa (Fig. 1b, lanes 2, 3, 6 and 7). In line with the MUC1/REP expression analyses described above, these immunoreactive proteins were absent from the HeLa epithelial tumor cell sample (Fig. 1b, c, lanes 1) as well as from those breast tissue samples which expressed very low amounts of the MUC1/REP protein (compare Fig. 1a with b). In contrast, prominent anti-cytoplasmic domain-reactive species migrating in the region of 42–45 kDa were observed not only in the high MUC1/REP-expressing samples but also conspicuously in the MUC1/REP-negative HeLa epithelial cell sample and in the low MUC1/REP-expressing breast tumor tissue samples (Fig. 1b). In several samples, there were 2 closely spaced immunoreactive bands in the 42–45 kDa region (e.g., Fig. 1b, lane 9), which may represent variations in the glycosylation status of the MUC1 proteins (see also ‘‘Discussion’’). The 42–45 kDa MUC1 isoforms were not observed in any of the normal breast tissue samples (Fig. 1b). By analyzing MUC1/Y-expressing transfectants with both anti-cytoplasmic domain antibodies and anti-MUC1/Y extracellu- lar domain antibodies (Fig. 2a), the 42–45 kDa immunoreactive species was identified as the MUC1/Y protein. A comparison of the expression of repeat array-containing MUC1 proteins (Fig. 1a) and MUC1/Y proteins (Fig. 1b) demonstrates that some breast tumor samples preferentially express the MUC1/Y protein and have only low expression levels of the repeat array containing MUC1 proteins. This is particularly striking in the breast tumor samples analyzed in lanes 5 and 9 of Figure 1. A different epithelial adenocarcinoma, represented by the HeLa cervical uterine adenocarcinoma cells, also exhibits this pattern of MUC1 expression (compare Fig. 1a with b, lanes 1, HeLa; this is more evident in Fig. 1c). A faint 33 kDa band that was immunoreactive with the anticytoplasmic domain antibodies was also detected in several samples expressing the MUC1/Y protein (Fig. 1b, c). This band represents a precursor MUC1/Y protein, as demonstrated by digesting immunoprecipitated radioactively labeled MUC1/Y protein with N-glycanase (Fig. 1d). The treatment reduced the mass of the 42–45 kDa proteins to about 33 kDa, thus indicating that the mature MUC1/Y protein is N-glycosylated. Whereas the 33 kDa precursor MUC1/Y protein was easily observed in stable MUC1/Yexpressing transfectants (Fig. 2a, lane 3), it was present in only very small amounts in HeLa and T47D cells (Fig. 1c) and barely detectable in the primary breast tumor tissue samples (Fig. 1b). MUC1/Y expression detected with antibodies directed against the MUC1/Y extracellular domain To obtain a reagent directed against the MUC1/Y protein, polyclonal antibodies (designated anti-MUC1/Yex) were generated against the bacterial recombinant MUC1/Y extracellular domain protein. Western blotting of the MUC1/Y-expressing transfectant with these antibodies demonstrated the presence of mature 42–45 kDa and 33 kDa precursor MUC1/Y proteins (Fig. 2a, lane 3). When anti-MUC1/Yex antibody was used to probe lysates prepared from either T47D breast tumor cells (Fig. 2a, lane 1) or primary breast tumor tissues (Fig. 2a, b, lanes indicated by BT), the FIGURE 2 – MUC1/Y expression in tumor tissue detected by anti-MUC1/Y antibodies. (a,b) Protein lysates prepared from primary breast tumor tissue samples (BT), normal breast tissue (BN), paired tumor and normal breast tissues from the same patient (BT*, BN*), T47D breast cancer cells (T47D), HeLa adenocarcinoma cells (HeLa), control 3T3 neomycin-resistant transfectants (control), 3T3 MUC1/REP-expressing transfectants (M.1/REP) and 3T3 MUC1/Y-expressing transfectants (M.1/Y) were resolved on SDS-polyacrylamide (12%) gels, Western blotted and probed with affinity-purified antibodies that had been generated against the recombinant MUC1/Y extracellular domain. The mature MUC1/Y protein and the cytoplasmic domain containing proteolytic cleavage products are indicated by the full arrow and bracket to the right. Panel (b) contains samples different from those presented in (a) and was analyzed with anti-MUC1/Yex antibodies. Note the expression of the MUC1/Y protein in the breast tumor samples (BT), breast cancer cells (T47D) and the MUC1/Y transfectant (M.1/Y). Expression was not detected in the normal breast tissue samples (BN), the MUC1/REP transfectant or the control neomycin 3T3 transfectant (control). (c) Competition of antibody binding. Protein lysate from a breast tumor tissue sample was analyzed by SDS-acrylamide gel, and the Western blot was probed with affinity-purified antibodies that had been generated against the recombinant MUC1/Y extracellular domain, as described above for (a) and (b). Competing recombinant MUC1/Y extracellular domain at a final concentration of 100 µg/ml was either added to (1 comp., lane 2) or deleted from (2comp., lane 1) the antibody probing solution. NOVEL MUC1 ISOFORMS IN EPITHELIAL TUMORS MUC1/Y protein was clearly expressed. The MUC1/Y extracellular domain contains sequences common to the cytoplasmic domaincontaining proteolytic cleavage products (Fig. 3d). In some samples, these 25–30 kDa proteins also were detected by the anti-MUC1/ Yex antibodies (Fig. 2a, lanes 1 and 7). MUC1/Y expression was not detected in the normal breast tissue samples (Fig. 2a, b, lanes indicated by BN). Western blotting of the HeLa cell lysates with the anti-MUC1/Yex antibodies revealed the presence of only the 42–45 kDa MUC1/Y protein (Fig. 2b, lane 5), commensurate with the results obtained with the anti-cytoplasmic domain antibodies (Fig. 1b). Western blot analyses thus demonstrated that tumor tissue samples differentially express the MUC1 isoforms. To analyze whether specific cells in the same tumor tissue are exclusively expressing one or the other of the MUC1 isoforms, adjacent breast tumor tissue sections were individually immunohistochemically stained with MUC1 isoform-specific antibodies (anti-MUC1/Yex antibodies and anti-repeat domain antibodies). These analyses revealed that the tumor cells can simultaneously express both 745 MUC1 isoforms, no exclusivity in MUC1 isoform expression in specific tumor cells being observed (data not shown). Repeated use of the anti-cytoplasmic domain antibodies markedly reduced their immunoreactivity against the MUC1/Y isoform, indicating lability of the antibody population able to recognize the MUC1/Y isoform. Conversely, the affinity-purified anti-MUC1/ Yex antibodies consistently detected, even following their repeated usage, the 42–45 kDa MUC1/Y isoform (Fig. 3). A comparison of the immunoreactivity of the 2 polyclonal antibodies against the MUC1/Y isoform and the cytoplasmic domain-containing proteolytic cleavage products present in primary tumor tissue (Fig. 3a, b) clearly shows that the anti-MUC1/Yex antibodies are far more sensitive than the anti-cytoplasmic domain antibodies in detecting MUC1/Y expression. Indeed, in some samples, the MUC1/Y protein was detectable only with the MUC1/Yex antibodies (compare Fig. 3a with b, lanes 3). The MUC1/Y protein is extensively phosphorylated within its cytoplasmic domain (Zrihan-Licht et al., 1994a). It may well be that the MUC1/Y isoform immunoreactivity of the anti-cytoplasmic domain antibodies, generated against an FIGURE 3 – Detection of MUC1/Y expression with antibodies recognizing different regions of the MUC1/Y extracellular domain. The scheme presented in (d) illustrates the repeat-array-containing MUC1/REP (MUC1, upper molecule) and MUC1/Y proteins. The sites in MUC1/REP corresponding to the splice donor and splice acceptor sites that generate the MUC1/Y protein by deleting the tandem repeat array (block of closely spaced vertical lines) are indicated by black arrows facing downward. The region comprising the proteolytic cleavage site of the repeat array containing MUC1 protein is indicated by the 2 white arrows facing upward. Proteins present in tumor (BT) or normal (BN) breast tissue lysates were resolved by SDS/PAGE, transferred and immunoblotted with (a) affinity-purified anti-MUC1/Yex polyclonal antibody, (b) anti-cytoplasmic domain polyclonal antibody and (c) antibodies directed against the extracellular domain of MUC1/Yex and affinity-purified by passage through an Affigel-10 affinity column to which recombinant truncated MUC1/Yex protein (designated MUC1/Y/Ssp1) terminating at the Ssp1 site just up-stream to the proteolytic cleavage site had been bound. The regions recognized by the antibodies used in (a), (b) and (c) are indicated in panel (d) by (a), (b) and (c), respectively. Detection was performed with horseradish peroxidase-conjugated secondary antibodies followed by enhanced chemiluminescence. Molecular mass markers are indicated in kDa on the left. The MUC1/Y protein is indicated in (a), (b) and (c) by the horizontal arrow and the cytoplasmic domain containing proteolytic cleavage products is indicated in (a) and (b) by the bracket. 746 BARUCH ET AL. unmodified peptide sequence, is compromised by post-translational modifications (such as phosphorylation) occurring on amino acid residues located within the cytoplasmic domain. Characterization of MUC1/Y isoform expression with antibodies directed toward the N-terminal sequences of MUC1/Y To obtain antibodies which exclusively recognize MUC1/Y sequences that are N-terminal to the site at which proteolytic cleavage occurs in the MUC1/REP protein, we adopted the following approach. As cleavage occurs in the MUC1/REP protein within a region immediately down-stream to an Ssp1 site (at nucleotide number 328 in the MUC1/Y sequence [Zrihan-Licht et al., 1994b]), we generated a truncated bacterial recombinant MUC1/Yex protein (designated MUC1/Y/Ssp1) terminating at the Ssp1 site. The purified MUC1/Y/Ssp1 protein was covalently bound to an Affigel-10 (Bio-Rad) resin, which was then used to affinity-purify anti-MUC1/Yex antibodies (see ‘‘Material and Methods’’). The affinity-purified antibodies thus obtained will recognize only MUC1/Yex sequences up-stream to the proteolytic cleavage site. Western blot analysis of proteins derived from breast cancer cells grown in vitro (Fig. 3c, lane 1) or of breast tumor tissue (Fig. 3c, lane 2) with these anti-MUC1/Y/Ssp1 antibodies revealed the MUC1/Y protein. These antibodies were not immunoreactive against proteins extracted from normal breast tissue (Fig. 3c, lane 3). As expected, the cytoplasmic domain-containing MUC1/REP proteolytic cleavage products with molecular masses of 25–30 kDa were not detected by this antibody preparation. The immunoreactivity of the MUC1/Y isoform against antibodies which recognize epitopes N-terminal to the proteolytic cleavage site present in MUC1/REP (designated ‘‘(c)’’ in Fig. 3) provides additional evidence that MUC1/Y, in contrast to MUC1/REP, does not undergo proteolytic cleavage. These results not only confirm MUC1/Y expression in breast tumor tissue but also rule out the possibility that the MUC1/Y protein derives from proteolytic breakdown of the MUC1/REP isoform. RT-PCR characterization of MUC1 isoforms lacking the tandem repeat array To ascertain the exact nucleotide sequence of the MUC1 isoforms expressed in malignant epithelial cells, HeLa RNA as well as RNA isolated from the ovarian carcinoma cell line Lucy were subjected to RT-PCR analysis, using primers located down-stream and up-stream to the tandem repeat array. Ethidium bromide staining of the amplified products revealed the presence of several prominent bands (Fig. 4a) which hybridized extensively to an internal MUC1 probe (Fig. 4b), suggesting that they were all derived from the MUC1 gene. PCR products deriving from the MUC1/REP mRNA were not detected. This is most likely due to the low MUC1/REP expression levels in these samples, the large size of MUC1/REP and/or its high G/C content. Nucleotide sequencing using both sense and anti-sense strands of the RT-PCR DNA products subcloned in M13 revealed, as expected, a sequence corresponding to the MUC1/Y isoform. MUC1/Y sequences also were observed which contained a 27 bp insert (designated MUC1/ Yalt) located down-stream to the 58 terminus, which is generated by a previously described alternative splicing event (Wreschner et al., 1990; Ligtenberg et al., 1990) occurring in the first intron (Fig. 4c). Several additional M13 clones revealed the existence in both the HeLa and Lucy samples of another novel MUC1 isoform that is also devoid of the tandem repeat array. Nucleotide sequencing showed that this isoform (designated MUC1/X) is generated by an alternative splicing event which has the same splice donor site as MUC1/Y ( gtgag) but utilizes an alternative splice acceptor (SA) site located 54 nucleotides up-stream to that used by MUC1/Y. The flanking sequence at this divergent SA site (58 ttctccccag-TTG 38) is composed exclusively of pyrimidine residues immediately up-stream to the invariant ag SA site and thus corresponds to consensus SA sequences (Fig. 4c). RT-PCRs utilizing different down-stream primers that were more distal to those used here consistently demonstrated the MUC1/X sequence (data not shown). As a consequence of this splice event, the expected MUC1/X protein, except for an additional 18 amino acid insert, will be identical to the MUC1/Y protein (Fig. 4e). The major DNA products appearing in the HeLa and Lucy RT-PCRs are thus derived from MUC1/Y, MUC1/Yalt and MUC1/X sequences. Expression of the MUC1/Y isoform in mouse mammary epithelial cells promotes in vivo tumorigenicity Having demonstrated MUC1/Y isoform expression in epithelial tumors, the next issue of obvious concern was its possible function within the epithelial cell. Previous studies have demonstrated that the cytoplasmic domains of the MUC1 proteins are phosphorylated on serine as well as on tyrosine residues (Zrihan-Licht et al., 1994b), and co-immunoprecipitation analyses revealed that the tyrosine-phosphorylated MUC1 cytoplasmic domain interacts with the GRB-2 protein (Pandey et al., 1995). These results indicate that the MUC1 proteins may participate in cell growth and proliferation and that they are thus likely to be functionally important in the oncogenetic process. To investigate whether expression of one or more of the MUC1 isoforms can transform cells or promote cell tumorigenicity, we generated stable MUC1/Y- and MUC1/REP-expressing transfectants in several different cell lines. Expression of either isoform in NIH/3T3 fibroblasts was not transforming (data not shown). To answer the question whether MUC1 expression can promote epithelial cell tumorigenicity, stable MUC1 isoform-expressing transfectants were generated using DA3 mouse mammary epithelial cells transformed by the carcinogen DMBA (Fu et al., 1990). Expression of the various MUC1 isoforms was confirmed by probing cell lysates with the appropriate anti-MUC1 antibodies. The tumorigenic potential of 5 independent MUC1/Y and 3 independent MUC1/REP transfectants was investigated by inject- FIGURE 4 – Detection by RT-PCR of MUC1 isoforms devoid of the tandem repeat array in HeLa and Lucy ovarian carcinoma cells. (a, b) RT-PCR products of MUC1 mRNA isolated from epithelial carcinoma cells. Poly A1 RNA isolated from Lucy ovarian carcinoma cells and HeLa cervical carcinoma cells was reverse-transcribed with the down-stream oligonucleotide ‘‘c’’, followed by addition of the upstream oligonucleotide ‘‘a’’ and PCR amplification as described in ‘‘Material and Methods’’ (see c for location of primers). PCR products were purified on agarose gel and subjected to nested PCR amplification, using the up-stream and down-stream oligonucleotides ‘‘b’’ and ‘‘c’’, respectively. The Lucy and HeLa RT-PCR products (lanes 1 and 2) were analyzed by agarose gel electrophoresis and ethidium bromide staining (a) followed by Southern blotting and hybridization with an internal MUC1 DNA probe (b). Sizes (in bp) of the DNA markers are indicated to the right of (a). (c, d) Location of splice donor (S.D.) and acceptor (S.A.) sites that generate the MUC1/Y and MUC1/X isoforms. The MUC1-specific RT-PCR products (a) were subcloned in M13 followed by dideoxynucleotide sequencing of the single-stranded DNA. The DNA sequence obtained was compared with that of the repeat-array-containing MUC1/REP (Wreschner et al., 1990), allowing identification of the splice donor (GTgag) and splice acceptor sites (cacttctccccAG and tttcaaacctccAG) generating the MUC1/X and MUC1/Y isoforms, respectively, shown in (c). As a consequence of these splice events, the tandem repeat array and its flanking sequences are deleted in the MUC1/X and MUC1/Y isoforms. Down-stream to the splice site, both isoforms retain the same reading frame (RF) as the repeat-array-containing MUC1 protein. The previously described (Wreschner et al., 1990; Ligtenberg et al., 1990) alternative splicing event occurring in the first intron is designated in (c) by ‘‘alt’’. The abbreviations SP, TM and CYT in (d) indicate the signal peptide, transmembrane and cytoplasmic domains, respectively. (e) The deduced amino-terminal amino acid sequences of the MUC1/Y and MUC1/X isoforms. The signal peptide is boxed and followed by the N-terminal amino acid sequence of the mature protein. The location of the splice deleting the tandem repeat array is indicated by the first black dot. This is followed by the smaller letters indicating the 18 amino acids present in the MUC1/X isoform. The MUC1/Y amino acid sequence continues from the second black dot; the full sequence appears in Zrihan-Licht et al. (1994b). NOVEL MUC1 ISOFORMS IN EPITHELIAL TUMORS ing transfected cells into the hind-leg muscle of syngeneic BALB/c mice. Controls were parental, non-transfected DA3 cells as well as DA3 cells transfected only with the neomycin-resistance gene. Inoculated mice were then monitored for both tumor latency and tumor size. Tumor latency in all 5 MUC1/Y-expressing transfectants was markedly shorter than that observed either in the MUC1/REP-expressing transfectants or in the control neomycinresistant transfectant (Fig. 5). The difference in tumorigenic potential of the various transfectants was subjected to a statistical analysis using the Mantel-Cox and Breslow tests. Significant 747 differences ( p , 0.05) were noted between the MUC1/Y and the MUC1/REP transfectants; no statistical differences were noted between the MUC1/REP transfectants and the control cells (not expressing MUC1). Furthermore, the average diameter of tumors at 100 days post-tumor cell inoculation, developing from the MUC1/Y transfectants, was 35 mm compared to 13 mm for tumors developing from either the control neomycin transfectant or the MUC1/REPexpressing transfectants. In a separate experiment, by 30 days following tumor cell inoculation, 100% of mice injected with the MUC1/Y transfectant had developed visible tumors, whereas only FIGURE 4 BARUCH ET AL. 748 FIGURE 5 – Tumorigenicity of MUC1 DA3 transfectants. DA3 mouse mammary carcinoma cells were co-transfected with both G418 (neomycin)-resistant cDNA and MUC1/REP cDNA (MUC1/REP#), G418-resistant cDNA and MUC1/Y cDNA (MUC1/Y#) or with the G418-resistant cDNA alone (neo#). Cells from stable transfectants (indicated by number) were injected into BALB/c mice (5 3 105 cells/animal). Mice were examined at several time points post-injection for the appearance of tumor. 50% of mice injected with the MUC1/REP form showed tumor formation. The increase in tumorigenic potential of the MUC1/Y transfectants was observed at 2 different cell concentrations (data not shown). In summary, a comparison of the tumorigenicity of 5 independent MUC1/Y-expressing transfectants revealed that all were significantly more tumorigenic than both the control neomycinresistant DA3 cells and the 3 MUC1/REP-expressing transfectants. Notably, no significant differences in tumorigenic potential were observed within the MUC1/Y-expressing transfectant group. DISCUSSION Using novel antibody reagents which recognize different regions of the MUC1 isoforms, we have shown that MUC1 protein isoforms are differentially expressed in malignant epithelial cells. Indeed, in certain breast tumor tissues, as well as in HeLa cervical adenocarcinoma cells, almost exclusive expression of the MUC1/Y isoform was observed. These findings indicate that decisions taken at the level of RNA splicing are likely to determine the outcome of MUC1 protein isoform expression and that epithelial cells at a specific stage of differentiation or ‘‘dedifferentiation’’ (i.e., malignant cells) preferentially express a specific MUC1 isoform. Malignancy-specific splicing events have been demonstrated (Matsumara and Tarin, 1992; Dall et al., 1994; Itoh et al., 1994). Alternative CD44 mRNA transcripts in breast, colon and uterine cervical cancers are both quantitatively and qualitatively specific to the malignant state (Matsumara and Tarin, 1992; Dall et al., 1994). Another striking example is the K-sam gene, which encodes a receptor for keratinocyte growth factor. In malignant tissues, an alternatively spliced mRNA transcript codes for a carboxylterminal truncated receptor with high affinity for its ligand, whereas in normal tissues the non-truncated receptor is the major isoform and has a lower affinity for its ligand (Itoh et al., 1994). Although mechanisms operative at the level of RNA splicing are likely to determine which specific MUC1 isoform is expressed, the decision whether to express the MUC1 gene is probably transcriptionally regulated (Abe and Kufe, 1993). In this context, it is notable that the CpG dinucleotide, present within the CCCGGG hexanucleotide found in every 60 bp repeat unit comprising the tandem repeat array, is fully methylated in cells that cannot express MUC1 but is completely non-methylated in tissues that can express the MUC1 gene (Zrihan-Licht et al., 1995). The specific MUC1 isoforms to be expressed will be determined, however, at the level of RNA splicing. By Western blotting we have shown that different tumor tissue samples express the MUC1 isoforms to varying extents. The relative expression of the various MUC1 isoforms may thus correlate with the stage of the breast cancer or with the specific type of epithelial adenocarcinoma in question. Research is presently in progress to resolve this issue. As the MUC1/REP and MUC1/Y proteins are both tyrosinephosphorylated (Zrihan-Licht et al., 1994a), they are likely to be involved in signal transduction processes. Our findings are also consistent with data showing that the MUC1 cytoplasmic domain can interact with the GRB-2 protein (Pandey et al., 1995). The GRB-2 protein is known to serve as the link between growth factor receptors and the ras signaling pathway. These data indicate that, as with other cell surface receptors (e.g., receptor tyrosine kinases, cytokine receptors), MUC1 may form a signaling complex which is intimately related to the oncogenetic process. Furthermore, tumor development is delayed in MUC1 null mice (Spicer et al., 1995), thereby implicating MUC1 protein expression in tumor progression. Consistent with these findings, we present data showing that expression of the MUC1/Y protein in DA3 mouse mammary epithelial cells increases the tumorigenic potential of these cells. MUC1/Y expression did not, however, lead to the transformation of NIH/3T3 fibroblasts. This is likely due to the absence in the fibroblast cell of accessory proteins that are likely required for MUC1 signal-transducing functions. As reported previously, such proteins could be cell surface molecules that generate heteromeric complexes with MUC1/Y (Zrihan-Licht et al., 1994a) or secreted proteins (ligands) that bind to and activate MUC1/Y function. The marked difference in the tumorigenic potential of the MUC1/Y and MUC1/REP transfectants may be due to different immunological responses elicited by these molecules in the host animal. Since the 2 molecules have a distinct structure, it could also well be that the MUC1/Y and MUC1/REP proteins are involved in NOVEL MUC1 ISOFORMS IN EPITHELIAL TUMORS different cellular processes. As previously shown, the MUC1/REP isoform, in contrast to MUC1/Y, is involved in cell to cell and cell to ECM repulsion via its large repeat-array-containing extracellular domain (Ligtenberg et al., 1992b; Hartman et al., 1992) and can inhibit integrin-mediated and E-cadherin-mediated cell adhesion. Significantly, MUC1/Y is devoid of this tandem repeat array and is apparently not involved in anti-cell-adhesive processes. It is thus probable that, although the MUC1/REP and MUC1/Y proteins are likely involved in signal transduction pathways, each isoform has both a unique activating mechanism and function in malignant epithelial cells. It is noteworthy that RT-PCR revealed an additional MUC1 isoform, MUC1/X, that is also devoid of the tandem repeat array and, except for an 18 amino acid insert in the extracellular domain, is identical to the MUC1/Y isoform. As previously noted, by probing immunoblots of breast cancer cell lysates with anticytoplasmic domain antibodies (Fig. 1b), a doublet band migrating in the region of 42–45 kDa was detected in certain samples. This heterogeneity could result from post-translational modifications, such as glycosylation and/or phosphorylation of the MUC1/Y protein. Another possibility is that the slower migrating band represents the MUC1/X isoform, which we deduce as being about 2 kDa larger than the MUC1/Y isoform. To resolve this issue and to demonstrate MUC1/X expression in malignant epithelial cells, we are now developing antibody reagents specific for the MUC1/X isoform. It should be noted that hundreds of isoforms for the neurexin neuronal cell surface proteins are generated by alternative splicing events and that a splice site-specific ligand for one 749 particular neurexin isoform has been documented (Ichtchenko et al., 1995). Indeed, using a receptor gel overlay assay, we have identified proteins that specifically recognize and bind to the MUC1/Y extracellular domain. Variations in the extracellular domain sequences of the MUC1/Y and MUC1/X isoforms could thus result in different affinities for putative interacting ligands. We are now investigating whether the ligand proteins that specifically interact with the MUC1/Y isoform are recognized in a similar manner by the MUC1/X protein. Our results indicate that MUC1 isoforms lacking the tandem repeat array may be useful tumor marker proteins and that in the cancer cell they are likely to perform functions different from those of the MUC1/REP protein. In particular, our results indicate that MUC1/Y expression may be intimately related to the uncontrolled growth and proliferation of breast cancer cells. Furthermore, the fact that the MUC1/Y isoform is expressed in some malignant epithelial cells displaying very low levels of the MUC1/REP protein suggests that the novel MUC1 isoforms, rather than the MUC1/REP proteins, are a hallmark of epithelial malignancy. ACKNOWLEDGMENTS This work was supported in part by the Israel Cancer Association, the Israel Cancer Research Fund and the Israel Academy of Sciences (D.H.W.) and by the Simko Chair for Breast Cancer Research, the Federico Fund for Tel Aviv University and the Barbara Friedman Fund (I.K.). REFERENCES ABE, M. and KUFE, D., Characterization of cis-acting elements regulating transcription of the human DF3 breast carcinoma-associated antigen (MUC1) gene. Proc. nat. Acad. Sci. (Wash.), 90, 282–286 (1993). BURCHELL, J., GENDLER, S., TAYLOR-PAPADIMITRIOU, J., GIRLING, A., LEWIS, A., MILLIS, R. and LAMPORT, D., Development and characterization of breast cancer reactive monoclonal antibodies directed to the core protein of the human milk mucin. Cancer Res., 47, 5476–5482 (1987). CERIANI, R.L., THOMPSON, K.E., PETERSON, J.A. and ABRAHAMS, S., Surface differentiation antigens of human mammary epithelial cells carried on the human milk fat globule. Proc. nat. Acad. Sci. (Wash.), 74, 582–587 (1977). DALL, P., HEIDER, K.H., HAKELE, A., VON MINCKWITZ, G., KAUFMANN, M., PONTA, P. and HERRLICH, P., Surface protein expression and messenger RNA-splicing analysis of CD44 in uterine cervical cancer and normal cervical epithelium. Cancer Res., 54, 3337–3341 (1994). FU, Y., WATSON, G., JIMENEZ, J.J., WANG, Y. and LOPEZ, D.M., Expansion of immunoregulatory macrophages by granulocyte-macrophage colonystimulating factor derived from a murine mammary tumor. Cancer Res., 50, 227–234 (1990). GENDLER, S.J., LANCASTER, C.A., TAYLOR-PAPADIMITRIOU, J., DUHIG, T., PEAT, N., BURCHELL, J., PEMBERTON, L., LALANI, E.-N. and WILSON, D., Molecular cloning and expression of the human tumor associated polymorphic epithelial mucin, PEM. J. biol. Chem., 265, 15286–15293 (1990). HAREUVENI, M., TSARFATY, I., ZARETSKY, J., KOTKES, P., HOREV, J., ZRIHAN, S., WEISS, M., GREEN, S., LATHE, R., KEYDAR, I. and WRESCHNER, D.H., A transcribed VNTR gene codes for a human epithelial tumor antigen. cDNA cloning, expression of the transfected gene and overexpression in breast cancer tissue. Europ. J. Biochem., 189, 475–486 (1990). HARTMAN, M., TSARFATI, I., HAREUVENI, M., KEYDAR, I. and WRESCHNER, D.H., Expression of breast cancer mucin, MUC1, alters cell adhesion and a MUC1 variant may act as a receptor molecule on breast cancer cells. In: L. Ioanndou-Mouzaka, N.J. Agnantis and I. Karydas (eds.), Senology, proceedings of the 7th international congress on senology, pp. 379–383, Excerpta Medica, Amsterdam (1992). HILKENS, J. and BUIJS, F., Biosynthesis of MAM-6, an epithelial sialomucin. Evidence for involvement of a rare proteolytic cleavage step in the endoplasmic reticulum. J. biol. Chem., 263, 4215–4222 (1988). ICHTCHENKO, K., HATA, Y., NGUYEN, T.M., ULLRICH, B., MISSLER, M., MOOMAW, C. and SUDHOF, T.C., Neuroligin 1: a splice site specific ligand for beta-neurexins. Cell, 81, 435–443 (1995). ITOH, H., HATTORI, Y., SAKAMOTO, H., ISHII, H., KISHI, T., SASAKI, H., YOSHIDA, T., KOONO, M., SUGIMURA, T. and TERADA, M., Preferential alternative splicing in cancer generates a K-sam messenger RNA with higher transforming activity. Cancer Res., 54, 3237–3241 (1994). KEYDAR, I., CHOU, C.S., HAREUVENI, M., TSARFATY, I., SAHAR, E., SELZER, G., CHAITCHIK, S. and HIZI, A., Production and characterization of monoclonal antibodies identifying breast tumor associated antigens. Proc. nat. Acad. Sci. (Wash.), 86, 1362–1366 (1989). LIGTENBERG, M.J., BUIJS, F., VOS, H.L. and HILKENS, J., Suppression of cellular aggregation by high levels of episialin. Cancer Res., 52, 2318–2324 (1992). LIGTENBERG, M.J.L., KRUIJSHAAR, L., BUIJS, F., VAN MEIJER, M., LITVINOV, S.V. and HILKENS, J., Cell associated episialin is a complex containing two proteins derived from a common precursor. J. biol. Chem., 267, 6171–6177 (1992). LIGTENBERG, M.J.L., VOS, H.L., GENISSEN, A.M.C. and HILKENS, J., Episialin, a carcinoma associated mucin, is generated by a polymorphic gene encoding splice variants with alternative amino termini. J. biol. Chem., 265, 15573–15578 (1990). MATSUMARA, Y. and TARIN, D., Significance of CD44 gene products for cancer diagnosis and disease evaluation. Lancet, 340, 1053–1058 (1992). PANDEY, P., KHARBANDA, S. and KUFE, D., Association of the DF3/MUC1 breast cancer antigen with the Grb2 and the sos/ras exchange protein. Cancer Res., 55, 4000–4003 (1995). SPICER, A.P., ROWSE, G.J., LIDNER, T.-K. and GENDLER, S.J., Delayed mammary tumor progression in MUC-1 null mice. J. biol. Chem., 270, 30093–30101 (1995). WRESCHNER, D.H., HAREUVENI, M., TSARFATY, I., SMORODINSKY, N., HOREV, J., ZARETSKY, J., KOTKES, P., WEISS, M., LATHE, R., DION, A.S. and KEYDAR, I., Human epithelial tumor antigen cDNA sequences. Differential splicing may generate multiple protein forms. Europ. J. Biochem., 189, 463–473 (1990). ZRIHAN-LICHT, S., BARUCH, A., ELROY-STEIN, O., KEYDAR, I. and WRESCHNER, D.H., Tyrosine phosphorylation of the MUC1 breast cancer membrane proteins. Cytokine receptor-like molecules. FEBS Lett., 356, 130–136 (1994a). ZRIHAN-LICHT, S., VOS, H.L., BARUCH, A., ELROY-STEIN, O., SAGIV, D., KEYDAR, I., HILKENS, J. and WRESCHNER, D.H., Characterization and molecular cloning of a novel MUC1 protein, devoid of tandem repeats, expressed in human breast cancer tissues. Europ. J. Biochem., 224, 787–795 (1994b). ZRIHAN-LICHT, S., WEISS, M., KEYDAR, I. and WRESCHNER, D.H., DNA methylation status of the MUC1 gene coding for a breast cancer associated protein. Int. J. Cancer, 62, 245–251 (1995).