Diverse Mechanisms of β-Catenin Deregulation in Ovarian Endometrioid Adenocarcinomas

INTRODUCTION

Ovarian cancer causes more deaths of women in the United States than any other gynecological malignancy (1) . Most malignant ovarian tumors are epithelial (carcinomas) and are thought to arise from the ovarian surface epithelium or the secondary Müllerian system, which includes endometriosis (2) . Although the existing classification scheme for ovarian carcinomas is based entirely on morphological criteria, clinical and molecular genetic analyses suggest that the major histological subtypes (e.g., serous, clear cell, mucinous, and endometrioid) likely represent distinct disease entities (3, 4, 5) . Moreover, ovarian carcinoma precursor lesions may well be subtype specific (6) . Hence, a clearer understanding of ovarian cancer pathogenesis might be more readily attained from focused molecular genetic studies of important cell signaling pathways in distinct ovarian carcinoma subsets, rather than through studies of a more heterogeneous collection of carcinomas.

OEAs 3 share molecular genetic features with uterine endometrioid adenocarcinomas, including frequent mutations of the PTEN tumor suppressor gene (7) , microsatellite instability (8 , 9) , and mutations of the CTNNB1 gene (10, 11, 12, 13, 14) . CTNNB1 encodes β-catenin, a vertebrate homologue of Drosophila armadillo and a critical component of the highly conserved Wnt signaling pathway (15) . The Wnt pathway has a critical role in regulating cell fate specification, proliferation, and differentiation in various tissues (15) . β-Catenin levels are regulated by a protein complex containing the APC protein, GSK3β, and AXIN (16 , 17) . This protein complex promotes degradation of free cytosolic β-catenin via GSK3β-mediated phosphorylation of NH₂-terminal β-catenin sequences and subsequent ubiquitination and proteasome degradation of β-catenin. When the Wnt pathway is activated, GSK3β activity and β-catenin ubiquitination and degradation are inhibited, and β-catenin enters the nucleus where it complexes with TCF/LEF transcription regulator proteins (17) . The TCF/LEF proteins cooperate with nuclear β-catenin to activate transcription of target genes, including perhaps c-MYC (18) , cyclin D1 (19 , 20) , and MMP-7 (21) , although many of the TCF/LEF target genes remain unknown. Oncogenic mechanisms leading to β-catenin stabilization include inactivation of the APC tumor suppressor protein and mutational activation of β-catenin itself via localized mutations of its NH₂-terminal GSK3β regulatory motif (22, 23, 24) . More recently, inactivating mutations in AXIN (AXIN1) in hepatocellular carcinomas and colon cancer cell lines (25 , 26) , and inactivating mutations in its homologue, AXIN2 (AXIL/Conductin), have been detected in microsatellite unstable colorectal cancers (27) . Rare mutations of CTNNG1, which encodes γ-catenin, have also been identified in human cancer (28) . γ-Catenin is another vertebrate homologue of Drosophila armadillo that, similar to β-catenin, functions in cell adhesion and Wnt signaling (29 , 30) . Although some prior studies have provided evidence that a sizable percentage of OEAs harbor β-catenin mutations (10, 11, 12, 13, 14) , studies of APC in OEAs have been extremely limited thus far, and no APC mutations have been reported (31) . Whether CTNNG1, AXIN1, or AXIN2 mutations play a role in the pathogenesis of these tumors remains unknown.

In an effort to better understand the molecular pathogenesis of OEAs, we collected 45 primary OEAs and 2 OEA-derived cell lines and carried out a comprehensive molecular analysis of several genes encoding proteins known to function in the Wnt signaling pathway. We found Wnt pathway defects in both OEA cell lines and in nearly half of the primary OEAs analyzed. β-Catenin deregulation was most often attributable to mutation of β-catenin itself and less frequently to inactivating mutations in APC, AXIN1, or AXIN2. Collectively, our findings provide evidence for diverse mechanisms of β-catenin deregulation in this particular subtype of ovarian cancer.

MATERIALS AND METHODS

Tumor Samples.

A total of 45 snap-frozen primary OEAs were analyzed. Six specimens were obtained from the Johns Hopkins Hospital; 2 from the University of Michigan Hospital; and 37 from the Cooperative Human Tissue Network/Gynecological Oncology Group Tissue Bank. A small portion of formalin-fixed, paraffin-embedded tissue from each specimen was histologically verified as OEA by a gynecological pathologist (K. R. C.) and classified as well, moderately, or poorly differentiated, based on the tumor’s histological features. Tumor stage (I–IV) was assigned according to the International Federation of Gynecology and Obstetrics (FIGO) system. Analysis of tissues from human subjects was approved by the University of Michigan’s Institutional Review Board (IRB-MED 2001-0022).

Cell Lines.

Two OEA-derived cell lines (TOV-112D and MDAH-2774), two ovarian clear cell carcinoma-derived cell lines (TOV-21G and ES-2), two ovarian adenocarcinoma (histological type unknown)-derived cell lines (NIH:OVCAR-3 and SKOV-3), and colon cancer cell line SW480 were obtained from the American Type Culture Collection (Manassas, VA). Ovarian serous carcinoma-derived cell lines HOC-1, HOC-7, HOC-8, and ovarian serous cystadenoma cell lines ML3 and ML10 were a gift from L. Dubeau (USC School of Medicine, Los Angeles, CA). Ovarian carcinoma cell lines (histological type unknown) OVCAR-4, OVCAR-5, OVCAR-8, PEO1, and PEO4 were a gift of T. Hamilton (Fox Chase Cancer Center, Philadelphia, PA). Ovarian serous carcinoma cell line DOV13 was a gift of D. Fishman (Northwestern University, Chicago, IL). IOSE-80 cells (human ovarian surface epithelial cells expressing SV40 large T Antigen) were a gift of N. Auersperg (University of British Columbia, Vancouver, British Columbia, Canada). TOV-112D and TOV-21G were cultured in 1:1 mixture of MCDB 105 medium (Sigma Chemical Co., St. Louis, MO) and Medium 199 (Life Technologies, Inc., Gaithersburg, MD) with 15% FBS (Life Technologies, Inc.). MDAH-2774 cells were cultured in RPMI 1640/10% FBS. ES-2 cells were cultured in McCoy’s 5A/10% FBS. All other cell lines were maintained in DMEM with 10% FBS.

DNA, RNA, and cDNA Preparation.

Primary tumor tissues were manually microdissected prior to nucleic acid extraction to ensure that each tumor sample contained at least 70% tumor cells. H&E-stained sections of frozen tumor tissues were used as dissection guides. Genomic DNA was isolated from pooled frozen tissue sections, using standard SDS/proteinase K digestion followed by phenol/chloroform extraction. Total RNA was extracted with Trizol (Life Technologies, Inc.) according to the manufacturer’s protocol. First-strand cDNA was synthesized from DNaseI-treated mRNA samples using random hexamer primers (Pharmacia Biotech, Piscataway, NJ) and Superscript II (Life Technologies, Inc.).

PCR.

The PCR primers used for each reaction are specified in Table 1 ⇓ . GenBank accession numbers for the nucleotide sequences used to determine suitable primer sequences are as follows: CTNNB1, X87838; CTNNG1, AF233882; APC, M74088; AXIN1, AF009674; and AXIN2, AF078165. The entirety of CTNNB1 exon 3 was amplified from cDNA using a forward primer in exon 2 and reverse primer in exon 4. A portion of CTNNG1 (γ-catenin) exon 1 was amplified from genomic DNA using exon-based primers flanking the region encoding the GSK3β regulatory domain near the γ-catenin NH₂ terminus. The coding region of AXIN1 was amplified from genomic DNA exon-by-exon using 10 pairs of intron-based primers. The coding region of AXIN2 was amplified from cDNA using primers to generate six overlapping cDNA segments. PCR for APC was performed using either cDNA (exons 1–14) or genomic DNA (exon 15) as templates. All PCR reactions were carried out in a final volume of 50 μl containing 50 mm KCl, 1.5 mm MgCl₂, 10 mm Tris-HCl (pH 8.3), 200 μm of each deoxynucleotide triphosphate, 0.2 μm each primer, and 2.5 units Taq polymerase (Life Technologies, Inc.). The annealing temperature used for each primer pair is specified in Table 1 ⇓ . After an initial denaturation at 95°C for 3 min, templates were amplified for 35 cycles (95°C for 30 s, 54–66°C for 30 s, and 72°C for 45 to 90 s), followed by a 5-min extension at 72°C. PCR products were visualized on 1- 2% agarose gels and purified with Qiaquick gel extraction kits (Qiagen, Valencia, CA).

Nucleotide Sequence Analysis.

Sequencing of PCR products was carried out using Thermo-Sequenase and ³³P-labeled dideoxynucleoside triphosphates (United States Biochemical, Cleveland, OH) or by automated sequencing conducted by the University of Michigan’s Sequencing Core. All sequence alterations were confirmed by bidirectional manual sequencing of PCR products generated by at least two independent reactions.

PCR Amplification of Microsatellite Sequences.

Two microsatellite loci on chromosome 11q23 (D11S1647 and D11S1987; Research Genetics, Huntsville, AL) were amplified from genomic DNA extracted from tumor and matched normal tissue. Each PCR mixture contained 37.5 mm Tris, 2.2 mm MgCl₂, 200 μm each dATP, dGTP, dTTP, 25 μm dCTP, 2 μCi (3000 Ci/mmol) [³²P]dCTP, 1 μm each primer, and 1 unit of Taq polymerase. Target sequences were amplified for 30 cycles as described above. PCR products were resolved by electrophoresis on 6% polyacrylamide gels, which were dried and subjected to autoradiography.

PTT.

A PTT was used to screen samples for protein truncating mutations in the APC gene. A large portion of the APC coding region (codons 1–2247; exons 1–14 and a portion of exon 15) was amplified from cDNA or genomic DNA (exon 15) in four overlapping segments encompassed by the T7-modified primers (T7-trans) specified in Table 1 ⇓ . The great majority of the APC mutations identified to date are within this region of the gene. 4 After verifying the presence of PCR-amplified products for each segment on 1% agarose gels, radioactive in vitro transcription/translation of the PCR-generated fragments was performed using the TNT Quick Coupled Transcription/Translation System (Promega Corp., Madison, WI) according to the manufacturer’s instructions. Products were separated on NuPAGE 10% Bis-Tris precast gels (Invitrogen, San Diego, CA) at 40 V for 60 min. The gels were fixed in methanol and acetic acid for 30 min and then soaked in Amplifying Solution (Amersham, Arlington Heights, IL) for 30 min for fluorographic enhancement of signal. After fixation in 7% acetic acid/7% methanol/1% glycerol, gels were dried and exposed to X-OMAT AR scientific imaging film (Kodak, Rochester NY). For samples with bandshifts in the PTT, the appropriate APC gene regions were amplified by PCR and bidirectionally sequenced.

Immunohistochemical Analysis of β-Catenin.

Five-μm sections of formalin-fixed, paraffin-embedded tissues were mounted on Probe-On slides (Fisher Scientific, Itasca, IL), deparaffinized in xylene, and then rehydrated into distilled H₂O through graded alcohols. Antigen retrieval was enhanced by microwaving the slides in citrate buffer (pH 6.0; Biogenex, San Ramon, CA) for 10 min. Endogenous peroxidase activity was quenched by incubation with 6% hydrogen peroxide in methanol, and then the sections were postfixed in 10% buffered formalin, washed, and blocked with 1.5% normal horse serum for 1 h. Sections were then incubated with a mouse monoclonal anti-β-catenin antibody (C19220; Transduction Laboratories, Lexington, KY) at a dilution of 1:500 overnight at 4°C. Slides were washed in PBS and then incubated with a biotinylated horse antimouse secondary antibody for 30 min at room temperature. Antigen-antibody complexes were detected with the avidin-biotin peroxidase method using 3,3′-diaminobenzidine as a chromogenic substrate (Vectastain ABC kit; Vector Laboratories, Burlingame, CA). Immunostained sections were lightly counterstained with hematoxylin and then examined by light microscopy. Immunostaining was scored on a three-tiered scale for intensity (+, absent/weak; ++, moderate; and +++, strong) in the tumor cell nuclei, cytoplasm, and cell membranes.

Immunofluorescence Studies of β-catenin.

Cells were grown on coverslips for 1–2 days, then fixed with 3% paraformaldehyde for 15 min at room temperature, and permeabilized with 1% goat serum/0.5% Triton X-100/PBS for 15 min at room temperature. After washing with PBS, the slides were blocked with 20% goat serum/0.2% Triton X-100/PBS for 30 min at room temperature. Cells were incubated with anti-β-catenin antibody (C19220; Transduction Laboratories) diluted 1:300 at room temperature for 2 h. After washing with 2% goat serum/0.2% Triton-X-100/PBS, cells were incubated with fluorescein (FITC)-conjugated AffiniPure donkey antimouse IgG (Jackson Immuno Research Laboratories, Inc., West Grove, PA) at a dilution of 1:200 for 60 min at room temperature. Slides were washed with PBS and then counterstained with propidium iodide (Sigma Chemical Co., St. Louis, MO) at a concentration of 1 μg/ml for 5 min. After mounting in aqueous medium, fluorescent microscopy was used to observe β-catenin localization.

β-Catenin/TCF Transcription Reporter Assay.

β-Catenin/TCF transcriptional reporter assays were performed essentially as described previously (24 , 32) . To measure TCF reporter activity in the ovarian carcinoma and other cell lines, cultured cells were plated in six-well plates 16 h prior to transient transfection with the reporter constructs pTopflash or pFopflash, which were gifts from B. Vogelstein (Johns Hopkins University, Baltimore, MD). All transfections were performed with FuGene 6 reagent (Roche, Indianapolis, IN) and 0.5 μg of pTopflash or pFopflash. To normalize transfection efficiency in the reporter assays, cells were cotransfected with 0.5 μg of pCH110 (Amersham), which contains a functional LacZ gene cloned downstream of a cytomegalovirus early region promoter-enhancer element. Forty-eight h after transfection, the cells were washed with PBS and then lysed with reporter lysis buffer (Promega). Luciferase activities were measured with a luminometer (model TD-20E; Turner Corp., Mountain View, CA) after adding luciferase assay reagent (Promega) to the cell lysates. β-galactosidase activities were measured using the β-galactosidase enzyme assay system (Promega) and a microplate reader (model 3550-UV; Bio-Rad, Hercules, CA). Each experiment was performed in triplicate.

RESULTS

TCF-dependent Transcription Is Activated in OEA-derived Cell Lines.

Most available ovarian cancer cell lines were derived from serous carcinomas, not endometrioid, clear cell, or mucinous primary tumors. In some cases, the histological subtype of primary carcinoma from which a cell line was derived was unknown, but the origin was most likely serous, because serous carcinomas are more common than all other histological subtypes combined (33) . We assayed TCF transcriptional activity as a strategy for assessing alterations in the Wnt signaling pathway in ovarian carcinoma cell lines, two of which are known to be derived from OEAs (TOV-112D and MDAH-2774). Cells were transfected with either a wild-type TCF-dependent luciferase reporter plasmid (pTopflash) or a mutated control version (pFopflash), and the ratio of the luciferase activity of pTopflash versus pFopflash in a given cell line was considered a measure of relative TCF activity. Representative data are shown in Fig. 1 ⇓ . Both TOV-112D and MDAH-2774 demonstrated constitutively increased TCF-dependent transcriptional activity (about 30- and 7-fold, respectively) compared with ovarian surface epithelial cells (IOSE-80), serous cystadenoma-derived cells (ML3 and ML10), or various cell lines derived from nonendometrioid ovarian carcinomas (TOV-21G, ES-2, DOV13, HOC-1, HOC-7, HOC-8, A1847, NIH:OVCAR-3, OVCAR4, OVCAR5, OVCAR8, OVCAR10, PEO1, PEO4, and SKOV3). Increased (5.3-fold) TCF-dependent transcription in MDAH-2774 cells has been noted previously by Furlong and Morin (34) .

Nuclear Localization of β-Catenin in OEA Lines and a Subset of Primary Tumors.

Immunofluorescent staining of the two OEA-derived cell lines, TOV-112D and MDAH-2774, showed nuclear accumulation of β-catenin in a subset of tumor cells, similar to the pattern of β-catenin staining observed in cell lines with known Wnt pathway defects (Fig. 2, A–D) ⇓ . Nuclear immunoreactivity for β-catenin was more diffuse in TOV-112D than in MDAH-2774 cells. The basis for nonuniformity of nuclear β-catenin accumulation in cell lines with pathway defects is unclear, but differences may relate to nonuniformity of β-catenin expression in individual cells and/or to the levels of other proteins, such as ICAT, that may regulate β-catenin or its interaction with TCF/LEF factors (35) . Nuclear immunoreactivity for β-catenin was also observed in 14 of the 45 primary OEAs (Table 2 ⇓ and Fig. 2, E–J ⇓ ). Primary tumors with nuclear accumulation of β-catenin frequently showed reduced membrane staining. Similar to the situation in cell lines, nuclear localization was typically noted in many, but not all, neoplastic cells within a given tumor.

CTNNB1 and CTNNG1 Mutational Analyses.

Somatic missense mutations in CTNNB1 sequences encoding the NH₂-terminal portion of β-catenin were identified in 14 of 45 (31%) primary OEAs. All tumors with mutations were well or moderately differentiated, and many showed squamous differentiation. CTNNB1 mutations affected serine (or immediately adjacent) residues in the presumptive GSK3β regulatory motif at the β-catenin NH₂ terminus (Table 2 ⇓ and Fig. 3 ⇓ ; Ref. 16 ). A missense CTNNB1 mutation at codon 37 (Ser→Ala) was also found in the OEA-derived cell line TOV-112D (Fig. 3) ⇓ . Twelve of the 14 primary OEAs with CTNNB1 mutations showed nuclear accumulation of β-catenin protein (Table 2) ⇓ . Exon 1 of CTNNG1, which encodes γ-catenin sequences that are also presumptive targets for phosphorylation by GSK3β, was sequenced in all primary OEAs lacking CTNNB1 mutations and in MDAH-2774 cells. No CTNNG1 mutations were identified.

APC Mutational Analysis.

PTTs for APC mutations were performed on 3 primary tumors; 2 with nuclear β-catenin accumulation and no demonstrable CTNNB1 mutation (OE-13T and OE-32T), and one with moderate cytoplasmic but no detectable membrane staining for β-catenin (OE-15T). The OEA-derived MDAH-2774 cell line was also analyzed for truncating mutations in APC. Four overlapping gene segments spanning APC codons 1–2247 were evaluated. One tumor (OE-32T) showed aberrantly migrating translation products from the first (codons 1–796) and second (codons 686-1211) segments in addition to the wild-type protein product (Fig. 4A ⇓ and data not shown). Sequence analysis of OE-32T DNA revealed nonsense mutations at codons 554 and 1114; both mutations resulted in the conversion of an arginine residue to a stop codon (Fig. 4B) ⇓ .

AXIN1 Mutational Analysis.

Interactions between AXIN and APC and AXIN and β-catenin are essential for promoting β-catenin phosphorylation and degradation (36) . All primary OEAs with wild-type CTNNB1 and APC genes were evaluated for mutations in exons 1–5 of AXIN1. Although studies to date are somewhat limited, all previously identified AXIN1 mutations in human cancers have been shown to occur within these exons, which encode the domains implicated in binding APC, GSK3β, and β-catenin (Fig. 5A ⇓ ; Ref. 26 ). The entirety of the AXIN1 coding region was examined in MDAH-2774 cells and in the 2 primary tumors with wild-type CTNNB1 sequences and nuclear β-catenin accumulation (OE-13T and OE-32T) as well as the tumor with significantly increased cytoplasmic β-catenin staining but no membrane staining (OE-15T). A protein truncating mutation in AXIN1 (exon 4, codon 416, Lys→stop) was identified in tumor OE-29T (Fig. 5B) ⇓ , and a missense sequence alteration in AXIN1 (exon 5, codon 555, Val→Ile) was identified in MDAH-2774 cells (Fig. 5C) ⇓ . Although this is a conservative amino acid change, the mutation affects a residue predicted to lie within the region of AXIN1 thought to bind β-catenin (26) . Three silent single-nucleotide polymorphisms within AXIN1 were also found, affecting nucleotide 93 (codon 31) in exon 1, nucleotide 873 (codon 291) in exon 1, and nucleotide 2517 (codon 839) in exon 9. The codon 31 and 291 single-nucleotide polymorphisms and several other polymorphisms in AXIN1 have been reported previously (37) .

AXIN2 Mutational Analysis.

Previous studies of colorectal carcinomas have shown AXIN2 mutations in four mononucleotide repeats within exon 7 (27) . These mutations were found only in colorectal cancers with defective mismatch repair and were not seen in mismatch repair-proficient tumors. Hence, the region encompassing these mononucleotide tracts (codons 640–718) was evaluated in all primary OEAs with wild-type CTNNB1 alleles. In the 3 primary OEAs with abnormal β-catenin levels and/or localization and no detectable CTNNB1 mutations (OE-13T, OE-32T, and OE-15T), the entire AXIN2 coding region was sequenced. A frameshift mutation at codon 665 (single G deletion) was found in one primary tumor (OE-5T; Fig. 6A ⇓ ). In previous work (38) , this tumor was found to exhibit instability at two of two microsatellite sequences (Fig. 6B) ⇓ .

DISCUSSION

Mutations of NH₂-terminal β-catenin sequences have been demonstrated in various human tumor types, including colorectal carcinomas (24 , 39) , melanomas (40) , hepatoblastomas (41) , medulloblastomas (42) , prostatic carcinomas (43 , 44) , and uterine endometrial carcinomas (45, 46, 47, 48) , particularly endometrioid adenocarcinomas with squamous differentiation (10) . Prior studies of ovarian carcinomas have indicated that β-catenin mutations are common only in OEAs, with estimated mutation frequencies of 16–54% (10, 11, 12, 13, 14) .

We have pursued our studies of the prevalence and mechanisms underlying β-catenin deregulation in 45 histologically verified frozen primary OEA tissues. This is, in fact, a substantial number of primary tumors of this particular histological subtype, because only 10–20% of ovarian carcinomas are endometrioid (49) . We identified CTNNB1 exon 3 mutations in 31% of our primary ovarian endometrioid carcinomas. Mutations inactivating the APC, AXIN1, and AXIN2 genes were each observed in 1 primary OEA and were found in tumors that showed wild-type CTNNB1 alleles. One of the 45 tumors we studied exhibited nuclear β-catenin accumulation not accounted for by mutations in the CTNNB1, APC, AXIN1, or AXIN2 genes. To our knowledge, our study is the first to identify mutations of APC, AXIN1, or AXIN2 in ovarian carcinomas. Collectively, our findings provide evidence for diverse mechanisms of β-catenin deregulation in OEAs, with demonstrable genetic mechanisms leading to β-catenin deregulation present in roughly 40% of our primary OEA specimens and in both OEA-derived cell lines. Interestingly, we observed that 11 of the 14 OEAs with mutant β-catenin were well differentiated, and none were poorly differentiated. Our findings are consistent with prior reports that OEAs with mutant β-catenin are more frequently low grade, exhibit squamous differentiation, and are associated with a favorable prognosis (10 , 12 , 13) .

Furlong et al. (50) identified a frameshift mutation of APC exon 15 in SW626, a cell line originally thought to be of ovarian origin. On the basis of this finding and immunohistochemical analyses showing expression of cytokeratin 20 and lack of expression of cytokeratin 7 and the estrogen and progesterone receptors in SW626 cells, it is now thought that SW626 was more likely derived from a colorectal carcinoma that metastasized to the ovary rather than a primary ovarian carcinoma. Nevertheless, we believe that confusion about tumor type is an unlikely explanation for the APC mutations identified in our primary OEA (OE-32T), because we have verified that OE-32T is a well-differentiated endometrioid adenocarcinoma, arising in the setting of endometriosis. Furthermore, the tumor was confined to one ovary. Metastatic colon carcinomas typically involve both ovaries, and nearly half of all OEAs are associated with endometriosis (51) .

A number of studies have shown that AXIN is critical for mediating the down-regulation of β-catenin [reviewed by Polakis (16)] . AXIN has binding sites for a number of proteins including GSK3β, β-catenin, APC, PP2Ac (the catalytic subunit of protein phosphatase 2A), and itself (26 , 52) . Although much remains to be learned about how AXIN functions, it appears to play a crucial role in facilitating the phosphorylation of β-catenin and APC by GSK3β (53) . On the basis of its ability to down-regulate β-catenin and Wnt signaling, AXIN can be regarded as a tumor suppressor. Moreover, the somatic inactivating mutations found in AXIN1 and AXIN2 in some tumors are consistent with the tumor suppressor gene designation. Although biallelic inactivation of AXIN1 and AXIN2 has been identified in some human cancers, in others, one allele is mutant and the other is wild type (26 , 27) . Our findings, particularly when considered in light of data in the literature, offer support for the view that the AXIN mutations we report alter AXIN function and deregulate β-catenin. Akin to the mutations we describe in CTNNB1 and APC, the mutations detected in AXIN1 and AXIN2 were clonal in nature. Two of the three mutant AXIN alleles described encode significantly truncated mutant proteins, removing defined functional domains in the AXIN1 and AXIN2 proteins. In the case of the missense mutant AXIN1 allele identified in MDAH-2774 cells, we have presented clear evidence of nuclear accumulation of β-catenin (Fig. 2D) ⇓ and increased TCF transcriptional activity (Fig. 1) ⇓ in the tumor cells harboring the mutation. Although the TCF-dependent transcriptional activity in MDAH-2774 is modest compared with TOV-112D, our results are comparable with those of the Morin laboratory, which identified MDAH-2774 as one of four ovarian cancer cell lines with increased β-catenin/TCF transcriptional activity (34) . The AXIN1 sequence alteration in MDAH-2774, which affects a residue in the region of AXIN1 implicated in β-catenin binding, has not been identified in any of over 100 AXIN1 alleles we and others have characterized thus far in other cancers (Refs. 25 , 26 and data not shown). Collectively, these findings support the likely functional significance of the AXIN1 defect in MDAH-2774, because these cells do not have mutations in CTNNG1, CTNNB1, APC, or AXIN2. Finally, we note that the primary tumor with the nonsense AXIN1 mutation (OE-29T) did not show nuclear accumulation of β-catenin by immunohistochemistry. Given the limited number of tumors with AXIN1 mutations identified to date, little is known about how well AXIN1 mutation correlates with aberrant β-catenin localization. Even tumors with documented CTNNB1 mutations do not invariably show readily recognizable nuclear or cytoplasmic accumulation of β-catenin protein (see for example, OE-14T and OE-19T; Table 2 ⇓ ).

Previous functional studies of human tumor-derived mutants of AXIN2 provide support for a dominant-negative effect of the mutant protein (27) . Specifically, transfection of a cDNA encoding the same AXIN2 mutant as the one we identified in tumor OE-5T (2083, del G) into cells expressing wild-type AXIN2 resulted in activation of TCF-dependent transcription. Moreover, this mutant protein was found to be more stable than the wild-type AXIN2 protein in colon cancer cells. These findings are consistent with a dominant-negative role for the mutant AXIN2 protein.

In summary, we have shown here that β-catenin deregulation is a common feature of OEAs, and mechanisms for deregulating β-catenin include mutation of β-catenin itself and mutational inactivation of APC, AXIN1, and AXIN2. The defects in β-catenin regulation presumably lead to activated expression of TCF/LEF-regulated target genes. Although many of these remain unknown, a number of downstream targets have been proposed and include c-MYC, CCND1/cyclin D1, PPARδ, gastrin, Cx43 (connexin 43), WISP-1, WISP-2, and MMP-7 (matrilysin; Refs. 16 , 18 , 19 , 21 , 54, 55, 56, 57 ). Downstream target genes of the β-catenin/TCF signaling pathway crucial for ovarian epithelial cell transformation will likely be up-regulated in all, or at least most, OEAs with β-catenin/TCF pathway defects. Further work evaluating gene expression patterns in OEAs with and without Wnt pathway defects will likely enrich our view of critical downstream TCF/LEF-regulated target genes.