[CANCERRESEARCH56, 2633-2640. June 1, 1996)

ABSTRACT

p120ca3 (CAS) is a protein tyrosine kinase substrate that associates

directly with the cytoplasmic tail of the cell-cell adhesion molecule Ecadherin. CAS Is thus part of a multimolecular complex that, along withother cadherin-bbding proteins (catenins), mediates interactions betweenE-cadherin and the actin cytoskeleton. Down-regulation of E-cadherinexpression and defects in catenin function have been implicated in tumormetastasis, but the role of CAS in these processes has not been addressed.Recently, the study of CAS was complicated when new anti-CAS antibodlea revealed the presence of at least four putative CAS isoforms thatappeared to vary in abundance between cell types. Here, we identify thefour major isoforms expressed In murine fibroblasts, and we show thatthey are products of alternative splicing. Analysis of CAS isoforms In avariety of murine cell lines Indicates that motile cells like fibroblasts and

macrophages preferentially express CASt (i.e., CAS1A and CAS1B isoforms), and epithelial cells prefrrentiaHy express CAS2 (Le., CAS2A andCAS2B Isoforms), whereas nonadherent cells (e.g., B cells, T cells, andmyelold cells) do not express detectable levels of CAS. Interestingly, CAStexpression is dramatically up-regulated in a Src-transformed MadinDarby canine kidney cell line, Indicating that the pattern of isoformexpression can be altered by cell transformation. Analysis of a variety ofdifferentiated and metastatic human tumor cell lines reveals that CASisoform expression in these cells Is quite heterogeneous. Furthermore,several poorly differentiated cell lines fail to express particular Isoformsthat are typically observed In well-differentiated cell lines. These dataraise the possibifity that unbalanced expression ofCAS isoforms in humancarcinomas may influence cadherin function and contribute to malignantor metastatic cell phenotypes.

INTRODUCTION

CAS3 is a tyrosine kinase substrate implicated previously in ligandinduced receptor signaling through the epidermal growth factor, platelet-derived growth factor, and colony-stimulating factor 1 receptors(1, 2), and in cell transformation by Src (3). Recently, we and othersidentified CAS as a component of the multiprotein cell-cell adhesioncomplex containing E-cadherin, a-catenin, @3-catenin,and y-catenin(plakoglobin; Refs. 4—6).Like @-cateninand plakoglobin (7), CASassociates direcfly with E-cadherin via its armadillo repeats (8), aninteraction that is lost when the cytoplasmic catenin-binding domainof E-cadherin is deleted (5). However, unlike f3-catenin and plakoglobin, the presence of which in cadherin complexes is mutuallyexclusive, CAS coexists in E-cadherin complexes with either @3-catenm or plakoglobin. Therefore, E-cadherin has at least two distinctprotein-binding epitopes in its cytoplasmic tail, one that associateswith either (3-catenin or plakoglobin, and another that simultaneously

Received 1/3/96; accepted 3/28/96.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 with18U.S.C.Section1734solelyto indicatethisfact.

I Supported in part by NIH Grant CA55724 (A. B. R.), NIH Cancer Center CORE

Grant P30 CA21756, and the American Lebanese Syrian Associated Charities of St. JudeChildren's Research Hospital.

2 To whom requests for reprints should be addressed, at Department of Tumor Cell

Biology. St. Jude Children's Research Hospital, 332 North Lauderdale, Memphis, TN38105.Phone:(901)495-3542;Fax:(901)495-2381.

3 The abbreviations used are: CAS, pl20@, a tyrosine kinase substrate; RT-PCR,

reverse transcription-PCR; MDCK, Madin-Darby canine kidney; Mab, monoclonalantibody.

associates with CAS. Recently, new anti-CAS antibodies revealedseveral putative isoforms of CAS. As with @3-catethnand plakoglobin,the associations of different CAS isoforms with a given cadherincomplex appear to be mutually exclusive in their binding to the CASepitope on cadherins (6). These data suggest that cadherin function ismodulated in part by a variety of mutually exclusive interactions ofArm domain-containing catenins with cadherins.

Several lines of evidence suggest that defects in various components of the E-cadherin complex result in decreased cell-cell adhesiveness and increased tumor metastasis. In highly metastatic carcinomas, expression of E-cadherin is frequently lost, reduced, orheterogeneous (9—15),and human tumors containing genetic defectsin the E-cadherin gene have been identified (1 1). Furthermore, perturbing E-cadherin function promotes invasive behavior in otherwisenormal cells (9, 16—18),whereas its forced expression often suppresses the invasive potential of malignant cells (10, 18, 19). Metastasis has also been linked to genetic defects in catenins. Tumor celllines with homozygous deletions in a-catenin (20—22)or @3-catenin(23—25) have been reported, and in some cases, the invasive pheno

types can be suppressed by forced expression of the normal genes (23,25, 26). Other links to carcinogenesis include the recent demonstration that f3-catenin associates directly with the tumor suppressoradenomatous polyposis coli (27, 28) and genetic evidence in Drosophila and Xenopus that links the (3-catenin homologue armadillo tothe winglessfWnt-J oncogene signaling pathways (29—31).Both adenomatous polyposis coli (32) and Wnt-1 (33, 34) modulate the levelsof j3-catenin in cells, and this appears to contribute to their roles in thegenesis of colon and breast cancer, respectively.

The association of CAS with E-cadherin complexes thus raises thepossibility that defects in CAS might also be involved in tumorigenesis. Antibodies to CAS recognize several isoforms, none of whichhave been extensively characterized in normal or tumor cell lines. Theisoforms are thought to arise by alternative splicing, but other possibilities (e.g., family members) have not been ruled out (4, 6). Here, wereport the identification of the major CAS isoforms expressed inmurine cell lines and their derivation by alternative splicing. Inaddition, cell-type-specific patterns of isoform expression were observed, suggesting that regulation of CAS alternative splicing in cellsmight modulate cadherin function. Furthermore, a striking inductionof CAS I expression was observed in Src-transformed MDCK cells,indicating that the pattern of isoform expression can be altered by celltransformation. Interestingly, CAS isoform expression was remarkably heterogeneous in human carcinomas, and in several carcinomacell lines, particular isoforms that are generally expressed in epithelialcell types were absent. These data raise the possibility that defects inthe alternative splicing of CAS might contribute to the morphologicalor invasive changes observed in some malignant cells.

MATERIALS AND METhODS

RT-PCR and Plasmids. Alternativelyspliced sequences were located byRT-PCR, using total RNA derived from NIH 3T3 fibroblasts as a template.

Oligonucleotide primers used in this study are listed in Table 1. Total RNAwas isolated from confluent NIH 3T3 cells after cell lysis in 2 ml Triazolsolution (Bethesda Research Laboratories, Bethesda, MD)/l00-mm dish, ac

cording to the supplier's instructions. Contaminating DNA was removed by

2633

Identification of Murine pl2ocas Isoforms and Heterogeneous Expression ofpl2O―@@Isoforms in Human Tumor Cell Lines'

Yin-Yuan Mo and Albert B. Reynolds2

Depart,nent of Tumor Cell Biology, Si. Jude Children ‘sResearch Hospital. Memphis, Tennessee 38105

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Table I Oligonucleotide primers usedforRT-PCRPrimerSequenceNucleotides―Used

forYM-l

YM-2YM-3GAAGGATGGAAGAAATCAGACT

TATCACCCAGATCCCCAGATCGGGG1TrGmC'VFAAAGGAcTGGC2848-2869(+)

3016-3037(-)205-228(+)RT-PCR

RT-PCRRT-PCR,

subcloningYM-4

YM-5pl2O—44CACAAC

Fill! 1GACTGTGGTCFCCATGTCTFCATAGCTCCFGAGGAGTCATCAG1TrGTCAG754-777(-)

I293-1 3 l4(—)3478-3494(-)RT-PCR,

subcloningFirst strand

cDNAsynthesisp120-27TCCGAGTrGTCATrGTC78O-796(-)First

strandcDNAsynthesisp1

20-3p120-10pl2O—3lTGCCGTAGTrGTCACTGC

GTATCTGCCATATGTCCATI'CGAGTAGQTGGAAGC101

3-lO3O( —)24l0-2427(—)I 130-I l46(+)RT-PCR

RT-PCRRT-PCRa

Nucleotide position refers to the published sequence (37); (—),antisense strand; (+), sense strand.

p12o――ISOFORMS

treatment with RNase-free DNase followed by extraction with phenol:chloroform and ethanol precipitation. The RNA was then heated for 10 mm at 68°Cand transferred to ice. RT-PCR was carried out as described previously (35),using 1@ total RNA as a template. The PCR products were cloned into pCR

II (Invitrogen), and at least two clones of each RT-PCR product were Sequenced using the Sequenase version II sequencing kit (Amersham).

To construct expression plasmids, the EcoRI-MstI fragments from thesePCR products were then subcloned into pRcCMV/CAS1A and pRcCMV/CAS1B (4) to generate pRcCMV/CAS2A and pRcCMV/CAS2B, respectively.

Cell Culture and DNA Transfection. All cell lines were culturedat 37°Cwith 5% CO2 in media supplemented with 10% FCS. The murine fibroblastcell lines NIH 3T3, Swiss 3T3, and C3H lOTl/2 and the canine kidneyepithelial MDCK cell line were cultured in DMEM. The colon adenocarcinoma cell lines Cob 205, Lovo, DLD-l, and SW620; the choriocarcinoma cellline Bewo; and the pancreas adenocarcinoma cell line Capan2 were grown inRPMI. Two other colon adenocarcinoma cell lines, SW480 and HCT1I6, werecultured in Leibovitz's L-l5 medium and McCoy's medium, respectively. Thehuman breast cancer cell lines MDA-MB23I , MCF7, MDA-MB453, MDAMB468, BT474, ZR75B, SKBr3, and T47D were cultured in Ham's Fl2medium.

Expression plasmids containing the CAS isoform cDNAs were transiently

transfected (5 p@g/60-mmdish) into NIH 3T3 cells, using lipofectamine (Bethesda Research Laboratories) as described previously (4). Expression wasanalyzed after 18h by immunoprecipitation and Western blotting, as describedpreviously (4). For analysis of CAS isoforms by immunoprecipitation, cells

were lysed in radioimmunoprecipitation assay buffer [50 mr@iTris (pH 7.2);150 mM NaCI; 1% NP-40; 1% deoxycholate; 0. 1% SDS; I mist EDTA; 0. 1 1jiM

sodium vanadate; and proteinase inhibitors: I m@tphenylmethylsufonyl fluoride, 10 @ag/mlaprotinin, and 5 @.tWmlleupeptin].

Most of the antibodies used in this study have been described previously (4,8). Briefly, Pab Fl is an affinity-purified rabbit antibody generated to theNH2-terminal amino acids 9—349ofCAS1B. It reacts with all isoforms of CASbut is somewhat less reactive with CAS1 isoforms than is Mab ppl2O (Transduction Laboratories, Lexington, KY), which also reacts with all CAS iso

forms. Mab 2B12 (36) is CASI specific and is known to recognize an epitopein the NH2-terminal region of the CAS1 isoforms (4). The E- and P-cadherin

monoclonal antibodies were obtained from Transduction Laboratories. The

N-cadherin monoclonal antibody l3A9 was generously provided by Dr.Margaret Wheelock.

Anti-S2 (Fig. 4, antipeptide) is a polyclonal antibody to the 21-amino acidpeptide (KKPDREEIPMSNIKSNTKSLD) predicted by the sequence of theCOOH-terminal, alternatively spliced sequence inserted between codons 874and 875 of the original CAS cDNA encoding CaslB (37; see Fig. 2A, 0designated S2). This peptide was glutaraldehydeconjugated to keyhole limpethemacyanin and used to generate antisera in rabbits. To affinity purify peptidespecific antibodies, peptides were conjugated to Sepharose 4B (Pharmacia)

according to the manufacturer's instructions and used as an affinity matrix foradsorption of antibodies from serum. Antibodies were eluted from the beadsusing Actisep (Sterogene), desalted on PD1O columns (Pharmacia), and quan

titated by spectrophotometry.

RESULTS

Cell-type-specific Expression of Murine CAS Isoforms. A vanety of putative CAS isoforms has been observed in murine (NIH 3T3),canine (MDCK), and human cell lines (4, 6), and we have previouslysuggested a correlation between the preferential expression of CASIisoforms in fibroblasts and CAS2 isoforms in epithelial cells (4). Toexamine this trend in more detail and eliminate species differences,we analyzed the isoforms in a wide variety of murine cell lines. Foreach cell line, direct comparison of the isoforms was facilitated byloading adjacent lanes with immunoprecipitates generated with Mabppl2O (recognizes all isoforms) or Mab 2B 12 (CAS 1 specific), followed by Western blotting with Mab ppl2O. Consistent with previousobservations in NIH3T3 fibroblasts, the fibroblast cell lines Swiss3T3and C3H1OT1/2 preferentially expressed CAS1 (Fig. 1, Lanes 1 and3), whereas epithelial cell lines such as TM4 and Comma D expressedmostly CAS2 isoforms (Fig. 1, Lanes 9 and 11, respectively). Although the WEG-l cell line was derived from mouse uterine epithehum, it has been immortalized by transfection with SV4O T-antigen.It does not express E-cadherin, and its morphology strongly resemblesthat of fibroblasts (38). Thus, the expression of primarily CAS1

Fig. I. Analysis of CAS isoforms in murine cell lines. The CAS isoformsin cell lysates from a variety of different murine cell types (top) wereimmunoprecipitated with Mab ppl2O (odd-numbered lanes), which recognizes all known CAS isoforms, or with Mab 2B12 (even-numbered lanes),which recognizes only CASI isoforms. The amount of Mab 2Bl2 immunoprecipitates loaded onto the gel was doubled to better detect the CASIisoforms. which are of low abundance in many epithelial cell lines. Immunoprecipitates were Western blotted with Mab ppl2O. Right, positions of themurine isoforms and their tentative designations.

pp120+ + + + + + 4. + + + ÷2B12@@ ÷ ÷ ÷ ÷ + ÷ ÷ ÷ ÷

@ •O— CASIB—CAS2A@ ...• -@ ,.

1 2 3 4 5 6 7 8 9 1011 12 13 141516 17 18 19@21@

2634

;@8a@)O @i<0)U-0.

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

pl2O'°'ISOFORMS

ATG1 ATG2 Mat IIVM@'1 i I I

YM-1

4636

S2 YM-2V

I- 4-YM.4 ‘Ovl-5

Si

Si . GAGTGTGAAGTGAGG. ACCTrCAI@GACGAC. ACTCGTCGGCATCAGAAC.

.94 195

S2 KKP D A E E I P MS@ S NT KSLDTMGWCCTGACCGGGAAGWUCCMTGAGCMTAT@Q@TCAMCACAWTCA1TAGATAI I

2628 2629

Fig. 2. Identification of CAS isoforms by PCR. NIH3T3 cell mRNA was scanned by RT-PCR using pairs of cas-specific oligonucleotides spanning the CAS cDNA sequence. Thelocations of the alternatively sliced sequences Sl and S2 are illustrated schematically. Below, the splice sites are marked by vertical arrows; splice insert sequences are in lightface,flanking regions are in boldface. The Sl splice removes the 5' ATG (ATG1) used by CAS1A and CAS1B. The putative internal ATG (ATG2), which is flanked by a good Kozacsequence. is the second ATG following the 51 splice and is the likely start codon for CAS2A and CAS2B. The guanine residues underlined in S2 correct a previous sequencing error.The correct sequence encodes methionine and glycine at insert amino acid positions 13 and 14 (underlined). All numbering is relative to the originally cloned CAS gene, which encodesCAS1B (see Ref. 37).

isoforms in WEG-l cells (Fig. 1, Lane 13) is not necessarily inconsistent with the above-noted trend. Like fibroblasts, the macrophagecell line Bad .2F5 (Fig. 1, Lane 5) expressed high levels of CAS 1. Inaddition, Bacl.2F5 cells reproducibly expressed low levels of a highermolecular weight putative isoform (Fig. 1, Lanes 5 and 6, tentativelydesignated CAS1*), which, before these experiments, had not beenobserved in murine cells. In contrast, in two nonadherent murine celllines, 32D myeloid cells (Fig. 1, Lanes 15 and 16) and Ag8 Blymphocytes (Fig. 1, Lanes 17 and 18), CAS expression was undetectable. One very weakly adherent cell line, a putative trophectodermal carcinoma cell line Fekete (39), did not express detectable levelsof any CAS isoform. The significance of this observation is currentlyunder study. These data indicate a pattern of cell-type-specific cxpression of particular isoforms. Specifically, highly motile cells likefibroblasts and macrophages preferentially express CAS1 isoforms,whereas more static cells like epithelial cells express mostly CAS2isoforms. In addition, the absence of CAS in myeloid cells and Blymphocytes, in which cadherins are not expressed, is consistent withthe notion that CAS is important for cadherin-mediated functions.

Identification and Characterization of Murine Fibroblast CASIsoforms. As a first step toward defining the functional significanceof the isoforms, we have cloned the four predominant isoformsexpressed in murine cells. To identify alternatively spliced sequences,CASmRNAwasscreenedbyRT-PCR,usinga seriesofoligonucleotide primers spanning the full-length CAS cDNA. Most of the primerpairs resulted in a single band, as expected (data not shown). Mab2B12 is CAS1 specific and recognizes an epitope in the NH2-terminalend of CAS. Consistent with these data, two species of mRNA weredetected by RT-PCR amplification using the primers YM-3 and YM-4(see Table I for primer sequences) located in the 5' end of cas (Fig.2, SI). DNA sequencing revealed that the shorter sequence resultsfrom the loss of nucleotides —94to + 195 (all numbering is relativeto the originally cloned CAS gene, which encodes CAS1B; see Ref.37), suggesting that the ATG encoding the first methionine of CAS1is spliced out of the mRNA encoding CAS2. The first ATG followingthe splice (at nucleotide 215) has an unfavorable Kozak sequence (40)and is out of frame with the rest of the protein. The second ATG atnucleotide 304 (Fig. 2, ATG2) is in frame, has an excellent Kozacsequence, and is therefore likely to act as the translation start codonfor CAS2. Initiation of translation at ATG2 results in the loss of theNH2-terminal 101 amino acids from CAS2 isoforms. Two bands werealso detected by using the primers YM-1 and YM-2 located in the 3'end of the CAS coding sequence. Sequencing of these bands revealedthe alternatively spliced 63-bp fragment (identified previously bycDNA cloning), which inserts between nucleotides 2628 and 2629 ofthe gene encoding CAS1B (4). However, the sequencing of PCR

derived clones from several different experiments indicated that residues 39—42of this 63-bp insert are GGGG instead of the previouslyreported AAAA (Fig. 2, S2). Amino acids 13 and 14 of this 21-aminoacid insertion are therefore methionine and glycine, respectively,instead of isoleucine and lysine. Both of the alternatively splicedsequences are flanked by 5' GT and 3' AG nucleotides, which areconserved nucleotides for RNA splicing (41).

To further confirm these assignments, we cloned cDNAs encodingeach of the isoforms into the Rc/CMV expression vector and cxpressed them individually by transient transfection in NIH3T3 fibroblasts (Fig. 3). Because the exogenous proteins were expressed at veryhigh levels, they were easily distinguished on Western blots fromendogenous CAS. Lanes 2—6in Fig. 3 were exposed for the sameamount of time to allow direct comparison of the expression levels ofendogenous and exogenous CAS. Lanes 1 and 7 in Fig. 3 are longerexposures of Lane 2; these were necessary to visualize the lessabundant CAS2 isoforms. All four exogenously expressed isoformscomigrated with their endogenous counterparts. In general, the exog

0wU,

LI)

ECL oa@ Cl)Exposure Cl) 0Time@ c%J- (!) __________

0@@

Txn:cDG@<<<< a@>>0000

IA

a.. S@IB

@ —@ 0 —2A

1234567Fig. 3. Comigration of cloned and endogenous CAS isoforms. NIH3T3 cells were

transiently transfected with vector alone (Lanes I. 2. and 7), CAS1A (Lane 3), CAS1B(Lane 4), CAS2A (Lane 5), or CAS2B (Lane 6). After 18 h, cells were lysed, and CASisoforms were immunoprecipitated and Western blotted with Mab ppl2O. The center lanes(Lanes 2—6)were exposed for 5 s to visualize the overexpressed exogenous CASisoforms. To better visualize the less abundant endogenous isoforms (exposed for 5 s inLane 2), 30-s (Lane 1) and 120-s (Lane 7) exposures of Lane 2 are shown.

2635

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

p120°'ISOFORMS

A@p@@@

IA.1B-.@@ —@ —1A

2A@@ —2A

Fig. 4. The C-terminal insertion 52 is present inboth CASIA and CAS2A. A, the CAS isoformswere immunoprecipitated by Mab 2Bl2 (Lanes 1. 3.and5)orMabppl20(Lanes2.4.and6),and I 2 3 4 5 6Westem blotted with Mab 2Bl2 (Lanes 1 and 2),Mab ppl2O (Lanes 3 and 4), or an antipeptide an- WB 2B12 ppl2O anti -tibody generated against the 21-amino acid insertionencoded by S2 in CAS1A. B, schematic representa- P@ etion of the major isoforms expressed in NIH3T3cells. The A forms of CAS1 and CAS2 contain theS2 insertion, whereas the B forms have spliced out ATG1 A102this sequence (see Figs. 3 and 4A). @,protein@coding sequence; 0, alternatively spliced se-@@ I— CA5 1Aquences encoded by Sl and 52.

:@@:@i:; @1tJ—cAS lB

. l@ I—C@S2A

@—(‘@j-—--CAB2B

enous CAS isoforms were expressed at approximately equal levels,although in the particular experiment shown in Fig. 3, this was not thecase for CAS1B.

To definitively demonstrate the presence of the S2 splice inCAS2A, we generated antibodies against the 21-amino acid peptideencoded by the 63-bp S2 sequence and probed CAS immunoprecipitates by Western blotting. The peptide-specific antibodies bound toboth CAS1A and CAS2A, indicating that the A and B forms of bothCAS 1 and CAS2 differ from one another by the presence or absenceof this sequence (Fig. 4A, Lane 6). As expected, immunoprecipitationof NIH 3T3 cell lysates with Mab 2B 12, which is CAS1 specific,revealed only CAS1A and CAS1B (Fig. 4A, Lane 3), whereas Mabppl2O precipitated all four CAS isoforms (Fig. 4A, Lane 4). Together,the data in Figs. 2, 3, and 4A strongly suggest that the isoformsexpressed in vivo result from the four possible combinations derivedby the presence or absence of these two alternatively spliced sequences, as illustrated schematically in Fig. 4B.

CAS Isoform Switching in Src-transformed MDCK Cells. Tyrosine phosphorylation of CAS is highly correlated with Src-inducedcell transformation (3). However, little is known about how celltransformation affects the pattern of isoform expression. Hence, weexamined the expression of CAS isoforms in Src-transformedNIH3T3 cells and in MDCK cells. In both cell types, all of theisoforms were heavily phosphorylated on tyrosine in Src-transformedcells but not in normal cells (Fig. 5B, Lanes 1—4).Interestingly, insome of the MDCK cell lines, but not in NIH 3T3 cells, Src transformation induced a large increase in the levels of CASI isoforms,which are normally minor isoforms in this cell type (Fig. 5C, compareLanes 5 and 6; Fig. 5A, compare Lanes 3 and 4). The isoform

A B C D3T3 MOCK 3T3 MDCK MOCK MDCK

Txn: .+ .÷ -÷ .÷ -+ -+

@ A.@. -Casi

12 34

Anti-GAS Anti-pTyr(PabFl) 56 56

Anti-CAS Anti-pTyr(Mab2Bl2)

Fig. 5. Up-regulation of CAS1 isoforms in Src-transfonned MDCK cells. The CASisoforms in lysates from the indicated normal (Lanes 1. 3. and 5), and Src-transformed(Lanes 2, 4, and 6) cells were compared by immunoprecipitation with a CAS-specificpolyclonal antibody Pab Fl (Lanes 1—4)and Western blotting with Pab Fl (A) oranti-pTyr antibodies (B). The experiment with the normal (Lane 5) and Sic-transformed(Lane 6) MDCK cells was then repeated using the CASI-specific Mab 2Bl2 for immunoprecipitation (Lanes 5 and 6) and Western blotting with Mab 2Bl2 (C) or antibodies tophosphotyrosine (D).

2636

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

pl2O@ ISOFORMS

CD C%J‘-,- U,

@ r-° a.Fig. 6. Heterogeneous CAS isoform expression in human carcinoma cell

lines. CAS isoforms from a wide variety of human carcinomas and twononadherent hematopoietic cell lines (Molt 4 and Raji) were compared byimmunoprecipitation using Mab ppl2O (odd-numbered lanes) or Mab 2Bl2(even-numbered lanes) and Western blotting with Mab ppl2O. The amount ofMab 2B 12 immunoprecipitates loaded onto the gel was doubled to better detecttheCAS1isoforms,whichareof lowabundancein manyepithelialcelllines.Va-2, foreskin fibroblast; HeLa and A 431, cervical carcinoma; Hep G2, livercarcinoma; Bewo, choriocarcinoma; PC3, prostate carcinoma; Molt 4, T lymphocyte; Raji, B lymphocyte; SW620 and HCTJI6, colon carcinoma; ZR75B,breast carcinoma; Capan 2, pancreatic carcinoma.

switching did not strictly correlate with transformation, because notall Src-transformed MDCK cell lines induced this effect (data notshown). Instead, the effect appears to correlate with high-level Srcexpression. A similar result has been reported previously in an MDCKcell line expressing temperature-sensitive Src (42).

Analysis of CAS Isoforms in Human Tumor Cell Lines. TheSrc-induced up-regulation of CAS1 isoforms in MDCK cells, coupledwith the high-level expression of CAS1 isoforms in fibroblasts andmacrophages, suggests that CAS1 might play a key role in invasive ormotile cells and hence be up-regulated in tumors that have becomehighly metastatic. To address this possibility, we analyzed the isoforms in a variety of human tumor cell lines derived from differentcell types (Fig. 6). Mab ppl2O was used to immunoprecipitate anddetect all four isoforms, and Mab 2B 12 immunoprecipitates were runin adjacent lanes to facilitate comparison and identification of CAS1isoforms. Because CAS1 isoforms are frequently expressed at lowlevels in epithelial cells, the relative amount of Mab 2B12 immunoprecipitates loaded in the Mab 2Bl2 lanes was doubled. Althoughcertain epithelial cell lines that have a marked fibroblast-like appearance (e.g. , HeLa, Fig. 6, Lanes 3 and 4) expressed increased levels ofCAS1 isoforms(ascomparedto well-differentiatedepithelialcells),the most striking observation was the extreme diversity in patterns ofCAS isoform expression in the tumor cell lines. For example, approximately equal amounts of CAS1 and CAS2 isoforms were detected inHeLa (cervical carcinoma) and Hep-G2 (liver carcinoma; Fig. 6,Lanes 3 and 5, respectively), whereas mainly CAS2 was detected inHCT116 (colon carcinoma) and ZR75B (breast carcinoma; Fig. 6,Lanes 19 and 21, repectively). Frequently, similar cell types displayeddifferent patterns of isoform expression (e.g. , compare cervical carcinomas A-431 and HeLa, Fig. 6, Lanes 3 and 7).

To simplify the interpretation of the observed isoform diversity, weobtained and compared CAS expression in multiple cell lines from thesame types of tumors (Fig. 7). Interestingly, several cell lines failed to

express one or more isoforms normally expressed in well-differentiated model epithelial cell lines like MDCK cells. For example,among the colon tumors (Fig. 7B, Lanes 3—14),the isoform cxpression pattern of the well-differentiated cell line DLD-l wasquite similar to that displayed by MDCK cells, whereas some ofthe poorly differentiated cell lines like HCT1 16 (Fig. 7B, Lane 11)and Colo2O5 (Fig. 7B, Lane 13) expressed only a subset of theseCAS isoforms. The isoform patterns expressed by the breast carcinomas were also quite heterogeneous (Fig. 7A, Lanes 5—18).Forexample, the cell lines BT474 and T470 expressed easily detectable levels of the CAS1@ isoforms (Fig. 7A, Lanes 15 and 17),which were absent from the other cell lines. Many of the cell lines(e.g., AR 75B, SKB-3, MDA 468, BT 474, and T470) expressedCAS2*, which was absent from MDA 231.

During the course of these analyses, it became clear that there aremore than four CAS isoforms. There is at least one high molecularweight isoform that contains the NH2-terminal epitope found in CAS1isoforms but that must contain additional sequence(s) that have not yetbeen cloned. Although this isoform was frequently encountered inhuman cells, a putative counterpart in murine cells was observed onlyin the macrophage cell line Bac 1.2F5. For purposes of discussion, wehave tentatively designated this isoform CAS1* (Fig. 1, Lanes 5 and6; Fig. 6, Lanes 18 and 23; Fig. 7A, Lanes 1, 2, 9, 10, 15—18;Fig. 7B,Lanes 4, 9, 10, and 14). Another putative isoform that comigrated inhuman and canine MDCK cells lacked the Mab 2B12 epitope butmigrated more slowly on electrophoretic gels than the two murineCAS2 isoforms (Fig. 6, Lanes 7, 17, and 23; Fig. 7, Lanes 1, 5, 9, 13,15, and 17). This isoform has been tentatively designated CAS24.Thus, it appears that there are at least six isoforms expressed in humancells. Although the respective isoforms observed in murine, canine,and human cell lines seem to comigrate on polyacrylamide gels, it isnot yet clear whether these bands are directly analogous to oneanother.

B ColonA Breast

,@J.1'..

I-@m I-

pp12O@ + + + + + +2B12@@ + ÷ + + +

._CAS1*_. —@ @ø 4@CAS2*.ssI.@ a

2637

@.c@ 0> I

0,- ‘e

@ 0)

<m@

ppl2O ÷ ÷ ÷ + ÷ ÷ + + + + + ÷2B12@@@ + ÷ + ÷ + ÷ + ÷ +

....a_ •... . a

..mI.s•-CAS1°a-. m

1 2 3 4 5 6 7 8 9 10 11 12 13 1415161718 1920 21@ 2324

>@@ C,)@

ppI2O + + + + + + + + +2B12@@ + + + + + + +

I 2 3 4 5 6 7 8 9 10 11 12 13 1415 16 1718 1 2 3 4 5 6 7 8 9 10 11 1213 14

Fig. 7. Heterogeneous CAS isoform expression in breast and colon carcinoma cell lines. CAS isoforms from panels of breast (A. Lanes 5—18)and colon (B. Lanes 3—14)carcinomacell lines were examined by immunoprecipitation with Mab ppl2O (odd-numbered lanes) or Mab 2Bl2 (even-numbered lanes) and Western blotting with Mab ppl2O. The amount ofMab 2B12 immunoprecipitates loaded onto the gel was doubled to better detect the CAS1 isoforms, which are of low abundance in many epithelial cell lines. Isoforms from MDCK(A, Lanes I and 2) and Va-2 (A, Lanes 3 and 4; B, Lanes I and 2) were includedfor comparison.Two new putative CAS isoforms,which have not yet been identified by cDNAsequencing, are designated CAS15 and CAS25.

111@@ .--w@.,•0@ø$@.

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

pl2O―°ISOFORMS

Fig. 8. E-. P-, and N-cadherin expression in thebreast and colon carcinoma cell lines. The cell linesanalyzed in Fig. 7 for CAS isoforms were examinedby Western blotting to determine their expressionlevels of E-, P-, and N-cadherins. Cell lysates wereprepared as described for immunoprecipitation(e.g., Fig. 7) but were then denatured by the addition of lOX Laemmli sample buffer. Fifty-sag aliquota of each cell lysate were separated on 8%SDS-polyacrylamide gels, transferred to nitrocellulose, and Western blotted with monoclonal antibodies specific for E-cadherin, P-cadherin, and N-cadherin (Mab l3A9). The N-cadherin panels (data notshown) were negative, with the exception of minoramounts of N-cadherin, which were detectable inMDCK, BT474, and T470 cells after very longenhanced chemiluminescence exposures.

1 234 5 6 7 8 91011121314

Analysis of E-, P-, and N-Cadherin in Breast and Colon TumorCell Lines. To determine whether particular patterns of CAS isoformexpression in breast and colon tumor cell lines might be linked toexpression levels of common cadherins, we analyzed our breast andcolon cell lines for expression of E-, P-, and N-cadherin (Fig. 8). Notsurprisingly, E-cadherin expression was absent in two (SKB-3 andMDA231) of the seven breast cell lines, and reduced in two (SW 620and SW480) of the six colon cell lines. The Lovo cell line may expressan aberrantly large form of P-cadhenn (Fig. 8, Lane 12). With theexception of Cob 205, which made only E-cadherin, most of the celllines expressed both E-and P-cadherins, but not N-cadherin. On verylong exposures, minor amounts of N-cadherin were detected in ourE-cadherin control cell line, MDCK, and in three of the breast celllines (ZR758, BT474, and T470; data not shown). Interestingly,SW620 and SW480, which are different isolates from the samepatient, were similar with respect to E- and P-cadherin levels (Fig. 8,Lanes 10 and 11) but expressed different CAS isoforms (Fig. 7, Lanes5 and 7). Although there was no obvious correlation between cadherinand CAS isoform expression patterns, there were many examples ofheterogeneous CAS isoform patterns in cell lines that were otherwisetypical with respect to E- and P-cadherin expression (compare Figs. 7and 8, breast cell lines ZR75B, MCF-7, MDA468, BT474, and T470,and colon cell lines DLD-l and HCT1 16).

DISCUSSION

The down-regulation or loss of E-cadhenn expression in highlymetastatic tumors is well documented (for a review, see Ref. 43), anddefects in both a- and j3—cateninhave been demonstrated to promoteinvasive cellular behavior (20—25).The direct physical associationbetween E-cadherin and CAS (8) implies that CAS is a new catenin,the role of which is physiologically relevant to cadherin function.Hence, the characterization of CAS in normal and malignant cells isan important step toward elucidating its physiological roles withrespect to cadherin function and tumorigenesis.

In previous work, we defined CAS 1 (CAS lA and CAS 1B) andCAS2 (CAS2A and CAS2B) isoform types according to their ability

(CAS1) or inability (CAS2) to bind to Mab 2B12 (4). Here, we havedirectly identified and characterized the four CAS isoforms expressedin murine cells. The CAS2 isoforms lack the Mab 2B12 epitope as aresult of an alternative splicing event that removes the CAS1 startcodon and the NH2-terminal 101 amino acids. The B forms of eachtype (CAS1B and CAS2B) differ from their respective A forms by theabsence of a alternatively spliced 21-amino acid sequence in theCOOH terminus. The data suggest that the major murine isoformsresult from the four possible combinations generated by the presenceor absence of these two alternatively spliced sequences. Interestingly,isoform expression was modulated in different cell types; high levelsof CAS1 expression were observed in cells that are highly motile (e.g.,fibroblasts and macrophages), whereas CAS2 isoforms were moreabundant in epithelial cells. The absence of CAS expression in nonadherent cells like T and B lymphocytes suggests that CAS functionis not required in cells that do not express cadherins. These datasuggest that modulation of CAS isoform expression by alternativesplicing is an important means of modifying the role of cadherins indifferent types of cells.

Our analysis also indicates the existence of additional isoforms ofCAS that have yet to be identified. Whereas most of the murine celllines examined expressed only the four isoforms described here, anadditional high molecular weight isoform was reproducibly immunoprecipitated from Bacl.2F5 macrophage cell lysates (Fig. 1, CASJ*)by both Mab ppl2O and Mab 2B 12, indicating that this new isoformis of the CAS1 type. Similar high molecular weight bands were alsodetected by both antibodies in several of the human cell lines and inthe canine cell line MDCK (Figs. 6 and 7, CASJ*). In addition, apreviously unrecognized CAS2 isoform was abundant in several human cell lines and in MDCK cells (Figs. 6 and 7, CAS2*), but not inany of the murine cell lines. The expression of this isoform in severaldifferent human cell lines indicates that it is most likely the result ofalternative splicing rather than mutation of the CAS gene. It is unlikelythat CAS2* could be a result of proteolysis, because a) it wasreproducibly present or absent from specific cell lines prepared underidentical conditions, and b) in Co1o205 cells, it was the only isoform

2638

t@ Colon

,@, LI)

00@ ,— C@J

0 A Cc@@oF-O@ I_i@ •@ > —‘-I -J@@o0O@@ ØCOJ10

Breast

N. 0

@ I—

E-cadherin

P-cadhenn ••@

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

pl2tf°°ISOFORM5

2. Kanner, S. B., Reynolds, A. B., and Parsons, J. T. Tyrosine phosphorylation of a120-kilodalton pp60@r@substrate upon epidermal growth factor and platelet-derivedgrowth factor receptor stimulation and in polyomavirus middle-T-antigen-transformed cells. Mol. Cell. Biol., 11: 713—720,1991.

3. Reynolds,A. B., Roesel,D. J., Kanner,S. B., and Parsons,J. T. Transformation-specffictyrosine phosphotylation ofa novelcellular protein in chicken cells expressing oncogenicvariants of the avian cellular sic gene. Mol Cell. BioL, 9: 629—638, 1989.

4. Reynolds, A. B., Daniel, J., McCrea, P., Wheelock, M. M., Wu, J., and Zhang, Z.Identification of a new catenin: the tyrosine kinase substrate pl2O'@°@'associates withE-cadherin complexes. Mol. Cell. Biol., 14: 8333—8342,1994.

5. Shibamoto, S., Hayakawa, M., Takeuchi, K., Hon. T., Miyazawa, K., Kitamura, N.,Johnson, K. R., Wheelock, M. J., Matsuyoshi, N., Takeichi, M., and Ito, F. Association of p120. a tyrosine kinase substrate, with E-cadherin/catenin complexes. J. CellBiol., 128: 949—957,1995.

6. Staddon, J. M., Smales, C., Schulze, C., Each, F., and Rubin, L. p120, a p120-relatedprotein (p100), and the cadherin/catenin complex. J. Cell Biol., 130: 369—381,1995.

7. Hulsken, J., Birchmeier, W., and Behrens, J. E-cadherin and APC compete for theinteraction with @-cateninand the cytoskeleton. J. Cell Biol., 127: 2061—2069,1994.

8. Daniel, J. M., and Reynolds, A. B. The tyrosine kinase substrate pl20@aSbindsdirectly to E-cadherin but not to the adenomatous polyposis coli protein or a-catenin.Mol.Cell.Biol.,15:4819—4824,1995.

9. Behrens, J., Marcel, M. M., Roy, F. V., and Birchmeier, W. Dissecting tumor cellinvasion: epithelial cells acquire invasive properties after the loss of uvomorulinmediated cell-cell adhesion. J. Cell. Biol., 108: 2435—2447,1989.

10. Frixen, U., Behrens, J., Sachs, M., Eberle, G., Voss, B., Warda, A., Lochner, D., andBirchmeier, W. E-cadherin mediated cell-cell adhesion prevents invasiveness ofhuman carcinoma cell lines. J. Cell. Biol., 117: 173-185, 1991.

11. 0th, T., Kanai, Y., Oyama, T., Yoshiura, K., Shimoyama, Y., Birchmeier, W.,Sugimura, T., and Hirohashi, S. E-cadherin gene mutations in human gastric carcinoma cell lines. Proc. Natl. Acad. Sci. USA, 91: 1858—1862, 1994.

12. Schipper, J. H., Frixen, U. H., Behrens, J., Unger, A., Jahnke, K., and Birchmeier, W.E-cadherin expression in squamous cell carcinomas of head and neck: inversecorrelation with tumor dedifferentiation and lymph node metastasis. Cancer Res., 51:6328—6337,1991.

13. Shimoyama, Y., and Hirohashi, S. Expression of E- and P-cadherin in gastriccarcinomas. Cancer Res., 51: 2185—2192,1991.

14. Shimoyama, Y., and Hirohashi, S. Cadherin intercellular adhesion molecule inhepatocellular carcinomas: loss of E-cadhenn expression in an undifferentiated carcinoma. Cancer LeU., 57: 131—135,1991.

15. Shiozaki, H., Tahara, H., Oka, H., Miyata, M., Kobayashi, K., Tamura, S., lihara, K.,Koki, Y., Hems, S., Takeichi, M., and Mon. T. Expression of E-cadherin adhesionmolecules in human cancers. Am. J. Pathol., 139: 17-23, 1991.

16. Behrens, J., Birchmeier, W., Goodman, S. L., and Imhof, B. A. Dissociation ofMadin-Darby canine kidney epithelial cells by the monoclonal antibody anti-Arc-l:mechanistic aspects and identification of the antigen as a component related touvomorulin.J.Cell.Biol.,101:1307-1315,1985.

17. Gumbiner, B., and Simons, K. A functional assay for proteins involved in establishingan epithelial occluding bamier-@identification ofa uvomorulin-like polypeptide. J. CellBiol., 102: 457—468,1986.

18. Vleminckx,K.,Vakaet,L., Marcel,M.,Fiers,W.,andRoy,F. V. Geneticmanipulation of E-cadherin expression by epithelial tumor cells reveals an invasion suppressor role. Cell, 66: 107—119, 1991.

19. Chen, W., and Obrink, B. Cell contacts mediated by E-cadhenn (uvomorulin) restrictinvasive behavior of L-cells. J. Cell. Biol., 114: 319-327, 1991.

20. Morton, R. A., Ewing, C. M., Nagafuchi, A., Tsukita, S. H., and Isaacs, W. B.Reduction of E-cadherin levels and deletion of the a-catenin gene in human prostatecancer cells. Cancer Res., 53: 3585—3590,1993.

21. Ode, T., Kanai, Y., Shimoyama, Y., Nagafuchi, A., Tsukita, S., and Hirohashi, S.Cloning of the human a-catenin cDNA and its aberrant mRNA in a human cancer cellline. Biochem. Biophys. Res. Commun., 193: 897—904,1993.

22. Shimoyama, Y., Nagafuchi, A., Fujita, S., Gotoh, M., Takeichi, M., Twukita, S., andHirohashi, S. Cadherin dysfunction in a human cancer cell line: possible involvementof loss of a-catenin expression in reduced cell-cell adhesiveness. Cancer Res., 52:1—5,1992.

23. Kawanishi, J., Kato, J., Sasaki, K., Fujii, S., Watanabe, N., and Niitsu, Y. Loss ofE-cadherin-dependent cell-cell adhesion due to mutation of the @-cateningene in ahuman cancer cell line, HSC-39. Mol. Cell. Biol., 15: 1175—1181, 1995.

24. Oyama, T., Kanai, Y., Ochial, A., Akimoto, S., Ode, T., Yanagihara, K., Nagafuchi,A.,Tsukita,S., Shibamoto,S., Ito,F.,Takeichi,M.,Matsuda,H.,andHirohashi,S.A truncated @.3-catenindisrupts the interaction between E-cadherin and a-catenin: acause of loss of intercellular adhesiveness in human cancer cell lines. Cancer Res., 54:6282—6287,1994.

25. Watabe, M., Nagafuchi, A., Tsukita, S., and Takeichi, M. Induction of polarizedcell-cell association and retardation of growth by activation of the E-cadherin-cateninadhesion system in a dispersed carcinoma line. J. Cell Biol., 127: 247—256,1994.

26. Hirano,S.,Kimoto,N.,Shimoyama,Y.,Hirohashi,S.,andTakeichi,M.Identificationof a neural a-catenin as a key regulator of cadherin function and multicellularorganization. Cell, 70: 293-301 , 1992.

27. Su, L.-K., Vogelstein, B., and Kinzler, K. W. Association of the APC tumor suppressor protein with catenins. Science (Washington DC), 262: 1734—1737,1993.

28. Rubinfeld, B., Sousa, B., Albert, I., Muller, 0., Chamberlain, S. H., Masiarz, F. R.,Munemitsu, S., and Polakis, P. Association of the APC gene product with @-catenin.Science (Washington DC), 262: 1731—1734,1993.

29. Funayama, N., Fagono, F., McCrea, P., and Gumbiner, B. M. Embryonic axisinduction by the Armadillo repeat domain of @-catenin:evidence for intracellularsignaling. J. Cell Biol., 128: 959—968,1995.4 A. B. Reynolds, unpublished observations.

2639

expressed. In addition, the bands are unlikely to derive from phosphorylation induced gel shifts, because Staddon et a!. (6) have alsoobserved six electrophoretic species of CAS, the elecrophoretic mobilities of which were unaffected by phosphatase treatment. Althoughwe cannot yet rule out the possibility that these new isoforms arefamily members rather than splicing derivatives, their recognition bya variety of CAS-specific antibodies (data not shown) argues for thelatter alternative.

The simultaneous presence of CAS isoforms in complexes witheither @3-cateninor plakoglobin (4, 5) complicates models for cadherinregulation. Several lines of evidence indicate that the various CASisoforms are likely to compete with one another for a single bindingsite on cadherins. First, CAS2 isoforms cannot be coprecipitated withCAS1-specific antibodies (6), indicating that CAS1 and CAS2 isoforms do not coexist in cadherin complexes. Second, different cadherins, such as E-, N-, and P-cadherin, do not appear to have apreference for particular CAS isoforms. Instead, CAS isoforms varyfrom one cell type to another, as do cadherins, and their associationwith a particular cadherin parallels their relative abundance in thecell.4 Interestingly, the splicing of CAS removes regions at the NH2-and COOH-terminal ends, leaving the central ARM domain intact.The data imply that the armadillo repeats are crucial for targeting CASto a conserved epitope on cadherins, whereas sequences unique tospecific CAS isoforms cater to the specialized requirements of specific cell types, possibly through the recruitment of novel bindingpartners. Indeed, CAS1 but not CAS2 isoforms have been implicatedin interactions with the FER tyrosine kinase (44), although the relevence of this interaction to cadherin function is not yet clear.

Interestingly, the patterns of isoform expression in human carcinomas were extremely heterogeneous, raising the possibility that alteredCAS isoform expression might contribute to cadherin malfunction inmalignant cells. The critical question with regard to the role of CASin malignancy is whether the failure to express a particular CASisoform is analogous to the complete or partial loss of cadherinfunction that occurs as a result of genetic damage to a- or j3-catenin.Given the above data and the direct interaction of CAS with Ecadherin, it seems likely that severely unbalanced expression of CASisoforms will alter cadherin function. In some of the carcinoma celllines tested, there was clear evidence for down-regulation or loss of Eand/or P-cadherin, which might account for their invasive behavior.However, unusual CAS isoform expression may be sufficient topromote invasiveness in the other cell lines, many of which displayedtypical levels of E- and P-cadherins. Transfection experiments shouldreveal whether restoring expression of specific CAS isoforms willalter or reverse the invasive phenotype of cell lines like Co1o205 andHCT116. In addition to mechanistic studies, it will be important todetermine whether CAS isoform expression patterns in malignantcells can be used as prognostic indicators of metastatic potential.Studies are under way to identify the human CAS isoforms and todetermine their roles in normal and malignant cells.

ACKNOWLEDGMENTS

We thank Drs. Neal Rosen, Sally Parsons, and Michael Bratton for thehuman colon and breast tumor cell lines used in this work. We are also

indebted to Drs. Pierre McCrea, Dan Medina, and Dan Carson for providingmurine cell lines.

REFERENCES

I. Downing,J. R., and Reynolds,A. B. PDGF,CSF-l, and EGF inducetyrosinephosphorylation of p120, a pp6O―°transformation-associated substrate. Oncogene, 6:607—613,1991.

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

pl2O'°'ISOFORMS

30. McCrea, P. D., Brieher, W. M., and Gumbiner, B. M. Induction of a secondary bodyaxis in Xenopus by antibodies to @-catenin.J. Cell Biol., 123: 477-484, 1993.

31. Peifer, M. Cell adhesion and signal transduction: the Armadillo connection. TrendsCell Biol., 5: 224—229,1995.

32. Munemitsu, S., Albert, I., Souza, B., Rubinfeld, B., and Polakis, P. Regulation ofintracellualr @3-cateninlevels by the adenomatous polyposis coli (APC) tumorsuppressor protein. Proc. Nail. Acad. Sci. USA, 92: 3046—3050,1995.

33. Hinck, L., Nelson, W. J., and Papkoff, J. Wnt-l modulates cell-cell adhesion inmammalian cells by stabilising f3-catenin binding to the cell adhesion protein cadherin. J. Cell Biol., 124: 729—741,1994.

34. Bradley, R. S., Cowin, P., and Brown, A. M. C. Expression of Wnt-1 in PC12 cellsresults in modulation of plakoglobin and E-cadherin and increased cellular adhesion.J. Cell Biol., 123: 1857—1865,1993.

35. Shen, J., and Hush, J. cis-Regulatory sequences responsible for alternativesplicing of the Drosophila dopa decarboxylase gene. Mol. Cell. Biol., 14:7385—7393,1994.

36. Kanner, S. B., Reynolds, A. B., Vines, R. R., and Parsons, J. T. Monoclonalantibodies to individual tyrosine-phosphorylated protein substrates of oncogeneencoded tyrosine kinases. Proc. Natl. Acad. Sci. USA, 87: 3328—3332,1990.

37. Reynolds, A. B., Herbert, L., Cleveland, J. L., Berg, S. T., and Gaut, J. R. p120, anovel substrate of protein tyrosine kinase receptors and of p6O@°@,is related to

cadherin-binding factors (3-catenin,plakoglobin and armadillo. Oncogene, 7: 2439—2445, 1992.

38. Wegner,C. C., Cherington,V.,Clemens,J. W.,Jacobs,A. L., Julian,J., Surveyor,0. A., Bell, E. C., and Carson, D. D. Production and characterization of WEG-l, an

epidermal growth factor/transforming growth factor-a-responsive mouse uterine epithelial cell line. Endocrinology, 137: 175—184,1996.

39. Damjanov, I., and Damjanov, A. Trophectodermal carcinoma: mouse teratocarcinoma-derived tumour stem cells differentiating into trophoblastic and yolk sac elements.J. Embryol.Exp.Morphol.,86: 125—141,1985.

40. Kozak, M. Point mutations define a sequence flanking the AUG initiator codon thatmodulates translation by eukaryotic ribosomes. Cell, 44: 283—292,1986.

41. Padgett,R. A., Grabowski,P. J., Konarska,M. M., Seller,S., and Sharp,P. A.Splicing ofmessenger RNA precursors. Annu. Rev. Biochem., 55: 1119—1150,1986.

42. Aghib, D. F., and McCrea, P. D. The E-cadherin complex contains the Sic-substratep120. Exp. Cell Res., 218: 359—369,1995.

43. Birchmeier, W., and Behrens, J. Cadherin expression in carcinomas: role in theformation of cell junctions and the prevention of invasiveness. Biochim. Biophys.Acta, 1198: 11—26,1994.

44. Kim, L., and Wong, T. W. The cytoplasmic tyrosine kinase FER is associated with thecatenin-like substrate ppl2O and is activated by growth factors. Mol. Cell Biol., 15:4553—4561,1995.

2640

Association for Cancer Research. by guest on August 23, 2020. Copyright 1996 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from