Frequent Amplification of Chromosomal Region 20q12-q13 in Ovarian Cancer¹

Ovarian cancer is thought to arise because of somatic alterations in genes involved in regulating cell proliferation,apoptosis, and genomic integrity. Alterations in several proto-oncogenes and tumor suppressor genes have been described, many of which are the same as in breast cancer. The ERBB2 oncogene is amplified and subsequently overexpressed in 9–30% of ovarian cancers (1, reviewed in 2). In most studies, overexpression of ERBB2 correlates with poor survival of patients (reviewed in ref. 2). Amplification of the MYC oncogene has been found in approximately 10–30% of ovarian cancers, more frequently in serous than in mucinous cancers (2). Other oncogenes amplified in ovarian cancer include KRAS, INT2, FMS, MDM2, and AKT2, but these amplifications appear to be relatively uncommon (3–5%; 1, reviewed in Ref. 2). However, cytogenetic studies indicate that homogeneously staining regions and double minute chromosomes are much more common, suggesting that ovarian cancers contain currently unknown gene amplifications (2).

CGH has revealed that chromosome arm 20q shows frequently copy number gains in ovarian cancer, similar to what has been observed in breast cancer (3, 4). The frequency of the copy number gain at 20q ranges from 20 to 48% (5, 6, 7, 8). Studies in breast cancer have revealed that the amplification consists of at least three smaller subamplicons, at 20q11, 20q12–20q13.1, and 20q13.2 (9). These regions are frequently coamplified and separated with large regions of nonamplified DNA (9). The region 20q13.2 is the most commonly amplified region in breast cancer (9). Recent studies have identified several putative candidate oncogenes from the amplified regions. Chromosome microdissection and hybrid selection studies have recovered three overexpressed and amplified candidate target genes originating from 20q: AIB3 and AIB4 mapping to 20q11, and AIB1 gene at 20q12. The latter is a member of SRC-1 family of nuclear receptor coactivators (10, 11). Amplification of the AIB1 gene also has been found in an ovarian cancer cell line (11). The MYBL2 gene, a member of v-MYB–related oncogenes, and the phosphotyrosine-phosphatase 1 gene (PTPN1; both mapping to 20q13.1) are independently amplified in a small fraction of breast cancers (9, 12). Two candidate genes for the 20q13.2 amplicon include ZNF217 and BTAK (13, 14). The former encodes a putative C2H2 Kruppel-like transcription factor, and the latter a centrosome-associated kinase involved in chromosome replication (13, 14).

The frequency of 20q amplification and the involvement of the candidate genes are not known in ovarian carcinoma. Here, we studied the amplification of the chromosomal region 20q in sporadic and hereditary ovarian cancers and cell lines with CGH³ and FISH using probes detecting the candidate genes found to be amplified in breast cancer.

Eight ovarian cancer cell lines (UT-OV-1, UT-OV2, UT-OV3, UTOV4,UT-OV5, UT-OV7, UT-OV8, UT-OV10) were derived from primary or metastatic epithelial ovarian carcinomas. Before the experiments the cells were cultured in complete Eagle’s minimal essential medium supplemented with 2 mm l-glutamine, 1%nonessential amino acids, 100 IU/ml penicillin, 100 IU/ml streptomycin,and 10% fetal bovine serum. The cells were kept in logarithmic growth with regular passages. The cells were harvested at confluency to obtain interphase nuclei from cells that were predominantly in the G₁-phase of the cell cycle (9). Trypsinized cells were cytocentrifuged and air-dried at room temperature until fixed and processed for FISH. Thirty-one primary untreated sporadic ovarian carcinomas were obtained from the University Lund and the University Tampere. These included 24 patients without a family history of breast or ovarian cancer, and 7 patients with a family history of breast or ovarian cancer. Five of the 7 familial ovarian cancer cases have been analyzed for BRCA1 and BRCA2 mutations. Two were found to be carriers of BRCA1 germline mutations 3172ins5 (termination at codon 1025) and 4744C→G (Ser1542Cys), respectively. Two patients were identified as carriers of BRCA2 germline mutations 6174delT (ter2003) and 7829C→T (affecting the splice donor site of exon 15), whereas no mutation was identified in the fifth patient. The two remaining familial cases had not been analyzed for BRCA1and BRCA2 mutations. Thus, 4 of 31 patients were considered as having a hereditary disease, 3 of 31 as familial ovarian cancer, and 24 of 31 as having sporadic ovarian carcinoma. Estrogen and progesterone status was determined immunohistochemically (sporadic tumors) or with EIA (hereditary tumors; Refs. 15, 16). The clinical and pathobiological characteristics of 31 primary tumors are shown in Table 1. The primary tumors were freshly frozen and stored at −70°C. Imprint touch preparations were made for FISH analysis by lightly pressing a semithawed frozen tumor onto Superfrost Plus microscope slides (Menzel, Germany) microscope slides and air-dried. Before hybridization, slides were stained with May–Grunwald–Giemsa staining to ensure that imprint touch preparations were cytologically representative of tumor tissue.

CGH was performed according to a published protocol (17). Briefly, genomic DNA from ovarian cancer cell lines and normal female reference DNA were extracted using a standard protocol and labeled with FITC–dUTP and Texas Red–dUTP (DuPont, Boston, MA) using nick translation. Labeled DNAs (400 ng each) and 10 μg of unlabeled Cot-1 DNA (Life Technologies, Gaithersburg, MD) were hybridized onto commercially available normal metaphase chromosomes (Vysis Inc.,Downers Grove, IL). The hybridizations were evaluated using the QUIPS digital image analysis system (Vysis Inc.).

Nine P1, PAC, or BAC probes defining genes and loci along 20q (10, 11, 12, 13, 18) were used in the study. The probes were physically mapped by fractional length from p-telomere (FLpter)measurements as described (18). In case of closely spaced probes, pairwise interphase mapping and data from contiguous assembly of the 20q13 region was used (13). The probes and their physical chromosomal localizations are listed in Table 2. The probes were used in our previous studies, except for BTAK, which was screened from the human PAC library for this study (18). A probe recognizing the pericentromeric region of chromosome 20 (RMC20L116) was used as reference probe to determine the overall copy number of chromosome 20. The probe cocktail containing the ERBB2 oncogene and the chromosome 17 centromere probe was obtained from Vysis Inc. (Chicago,IL).

Samples were fixed with 50 and 70%, and twice in 100% Carnoy’s solution (3:1 methanol:acetic acid, 10 min each). Two-color FISH was carried out as previously described (9) with minor modifications. In brief, the slides were denatured in 70%formamide–2× SSC at 72°C for 3 min. Five nanograms of differentially labeled pericentromeric probe, 20 ng locus- or gene-specific probe (when appropriate), and 10 μg of human placental DNA were hybridized overnight at +37°C as described (9). FISH of ERBB2 was done according to the manufacturer’s instructions (Vysis Inc., Chicago, IL). After posthybridization washes,bound probes were detected immunohistochemically with avidin–FITC(Vector Laboratories, Burlingame, CA) and anti-digoxigenin rhodamine(Boehringer; Ref. 9). Slides were counterstained with 0.2μ m 4,6 diamino-2-phenylindole in an antifade solution (Vectashield; Vector Laboratories). Hybridization signals were scored using Zeiss Axioplan 2 epifluorescence microscope equipped with dual band-pass fluorescence filter (Chromatechnology, Brattleboro,NV), which enables simultaneous detection of both fluorescein and Texas Red fluorescence. Hybridization signals from at least 300 nuclei were scored to assess the average copy number status per cell. The scoring results were expressed as the number of signals relative to the centromere signals (level of amplification). Digital images were taken with a Hamamatsu 9585 camera (Hamamatsu Inc, Hamamatsu City, Japan)operated via ISIS image analysis software (MetaSystems GmbH,Altslussheim, Germany). Control hybridizations to normal fibroblasts were done to ascertain that the hybridization efficiencies of the test and reference probes were similar.

Three of eight ovarian cancer cell lines showed a copy number gain at chromosome 20q (Fig. 1). In the cell line UT-OV1 the entire 20q was gained, without evidence of highly amplified regions within 20q. In UT-OV5, a copy number gain was found at 20q11-q13.2. A regional copy number gain was found at 20q12-qter in UT-OV-8. (Fig. 1). The copy number profiles were in the normal range in the remaining five cell lines.

Analysis of gene amplification by FISH confirmed the findings obtained by CGH (Figs. 1 and 2). As expected, CGH failed to show difference between copy number gain and amplification(Fig. 1). In UT-OV1 all studied genes and loci between AIB1and BTAK showed either a copy number gain (ratio to 20 centromere copy number, 1.5–3) or a high-level amplification (ratio to 20 centromere copy number, >3). In UT-OV5 the low-level amplification extended from 20q11 (AIB3/AIB4) to 20q13.2(BTAK). High-level amplification of PTPN1 and ZNF217 was found both in UT-OV1 and UT-OV8. The amplification pattern was discontinuous in UT-OV8, with a nonamplified region between PTPN1 and ZNF217 (defined by RMC20P4001).

Gene amplification for at least one region along 20q arm was found in 13 of 24 (54%) primary sporadic ovarian carcinomas, in 1 of the 3 familial tumors, and in all 4 hereditary tumors (Fig. 2, examples of FISH in Fig. 3). The mean level of amplification ranged from 2- to 12-fold, and individual tumor cells showed up to 15-fold amplification (compared with 20 centromere count). All hereditary, 1 familial (33%), and 12 sporadic tumors (50%) showed high-level amplification (>3-fold) with one or more of the following genes/loci tested: AIB1 at 20q12, MYBL2 and PTPN1 at 20q13.1, ZNF217 and BTAK at 20q13.2 (Fig. 2). Other genes/loci did not show high-level amplification (Fig. 2).

When the prevalence of high-level amplification was evaluated (Table 3), amplicons defined by PTPN1and AIB1 were the most common in sporadic ovarian cancer,but because of coamplifications, regions defined by BTAK, MYBL2, and ZNF217 also were amplified almost as often (Table 3). No constant patterns of coamplification were found in these tumors (Fig. 2). Two or more separate amplicons were found together in various combinations. Amplified regions were in most tumors interspersed with nonamplified regions (Fig. 2), indicating discontinuous amplicons along the 20q. The PTPN1 gene was amplified alone at 20q in three tumors, AIB1 in two tumors,whereas MYBL2, ZNF217, and BTAK were never amplified alone at high level (Fig. 2). In hereditary tumors, BTAK was amplified in all samples and was the only highly amplified gene in one case (Fig. 2, Table 3).

Five of six sporadic tumors with high-level AIB1amplification were estrogen receptor positive, compared with 4 of 18 tumors without AIB1 amplification (P =0.0147, Fischer’s exact test, two-sided P value). No significant correlation was found between progesterone receptor and AIB1 amplification. Other highly amplified genes showed no correlation with estrogen or progesterone receptor status. In sporadic tumors, all three endometrioid tumors, four of six undifferentiated and 6 of 12 serous adenocarcinomas showed 20q amplification. These differences were not statistically significant. No 20q amplifications were found in mucinous tumors or borderline tumors. None of the 20q amplifications correlated with tumor staging. When cumulative survival of the patients was studied, only tumors with AIB1amplification showed a tendency to correlate with poor patient survival(P = 0.066, Mantel-Cox test, Fig. 4). Tumors in Fig. 2 and Table 1 are presented with same numbers to allow cross-reference.

To further characterize the tumors with 20q amplification, we analyzed amplification of ERBB2 in the same set of tumors. Three cell lines (UT-OV-1, UT-OV4, and UT-OV-5), two sporadic tumors(8.3%), and none of the familial or hereditary tumors showed amplification for ERBB2. ERBB2 and 20q were not amplified in the same primary tumors (Table 1).

On the basis of the high frequency of amplification, genes located at chromosome 20q are, together with ERBB2 and MYC, the most common targets of gene amplification in ovarian cancer. In the present tumor material, ERBB2 and genes at 20q were not amplified in the same primary tumors. Thus, gene amplification at 20q may identify a new subgroup of sporadic ovarian cancers. Amplification at 20q was found also in four of four hereditary ovarian tumors, suggesting that carcinogenesis involving inactivation of susceptibility genes BRCA1 and BRCA2 may specifically lead to a tumor type that includes amplification at 20q. ERBB2 amplification was not found in the hereditary tumors,which is in agreement with recent observations of hereditary breast and ovarian tumors (19, 20).

The main finding in sporadic ovarian carcinoma was that the amplification at 20q involves at least five separate regions. Similarly, as in breast cancer (9), these regions are distinct (i.e., separated by nonamplified DNA), but consistently coamplified with each other in various combinations. Despite similar discontinuous amplicon structure, several differences between breast and ovarian cancer also were found. The overall frequency of amplification of any region was at least two times more common in ovarian than in breast cancer. In breast cancer, the ZNF217 gene has been defined as the core of the 20q amplification (9), whereas in ovarian carcinomas it was amplified less frequently than the more proximally located genes AIB1 and PTPN1. The steroid-receptor coactivator gene AIB1 was amplified in a higher proportion of ovarian than in breast cancers. More importantly, AIB1 also was amplified alone, suggesting that it also may be selected independently in ovarian cancer. Another striking difference was a lack of high-level amplification at 20q11, suggesting that AIB3 and AIB4 are unlikely to be the targets for amplification in ovarian cancer.

The high frequency of amplification at 20q by FISH is in accordance with studies by CGH (5, 6, 7, 8). However, as also indicated in the present study, the resolution of CGH is too low to detect small amplification peaks and to distinguish high-level amplification from a low-level copy number gain of a larger region. Amplification at 20q has been studied in ovarian cancer in two previous studies (1, 21). The clone RMC20C001, which defines the minimal amplicon at 20q13.2 (including the ZN217 gene), has previously shown amplification in ∼3% of ovarian tumors by Southern analysis (1). This differs from our results (ZN217 amplified in 12.5% in our study), but can be explained by the lower sensitivity of Southern blot analysis to detect low-level amplification. Furthermore,the clone RMC20C001 may not detect all amplicons, which include ZNF217 (13). The ZNF217 gene, a candidate gene for 20q13.2 amplification, was recently cloned by positional cloning (13). It encodes a putative C2H2 Kruppel-like transcription factor, and it was the only gene constantly amplified and overexpressed in the ∼250-kb minimal common region of amplification in breast cancer (13). In the current study ZNF217 was not amplified alone, suggesting a less central role in ovarian cancer. The same holds for MYBL2, which is sometimes amplified without coamplification of the 20q13.2 region in breast cancer (12).

The AIB1 gene, mapping to 20q12, has previously been shown amplified and overexpressed in one ovarian carcinoma cell line (11) and in 7.4% of primary tumors by Southern analysis (21). We found AIB1 amplified in 25% of sporadic ovarian tumors, again indicating a higher sensitivity of FISH. Amplification was significantly associated with positive estrogen receptor status of the tumor (P = 0.01). Interestingly, AIB1 has been shown to interact with the estrogen receptor and to enhance estrogen receptor–dependent transcription (11). The ovarian surface epithelium, from which ovarian carcinomas arise, has been described to express estrogen receptor protein (22). The ovarian surface epithelium undergoes rapid proliferation after ovulation (reviewed in Ref. 23), similar to that in mammary epithelium during the menstrual cycle. This may explain why this gene is selected for amplification in ovarian cancer, because AIB1 amplification may activate an estrogen-dependent,growth-promoting pathway in transformed epithelial cells forming the malignant tumors. AIB1 amplification also has been associated with estrogen and progesterone receptor status in breast tumors (21). Despite the small number of patients, AIB1 amplification also showed nearly statistically significant correlation with poor patient survival, underlining the importance of this amplification in the tumor progression of ovarian cancer.

The PTPN1 gene, a phosphotyrosine phosphatase gene mapping to 20q13.1, was highly amplified in 29% of ovarian tumors and 25% of cell lines. PTPN1 is overexpressed in breast and ovarian (24, 25) cancer, but according to our previous results in breast cancer (9),⁴PTPN1 is almost always constantly expressed without any clear correlation between amplification and expression. Therefore, it is possible that another gene in this region is the target for this amplification. The BTAK gene was amplified in 21% of sporadic tumors, and in four of four hereditary ovarian tumors with the BRCA1 or BRCA2 germline mutation. BTAKencodes a centrosome-associated kinase, which has recently been shown to be the target gene of the 20q13 amplification in breast and colon cancer (14). Up-regulation of BTAK leads to centrosome amplification, chromosomal instability, and aneuploidy (14). Interestingly, all sporadic and hereditary ovarian tumors analyzed here were aneuploid for chromosome 20 in this material(data not shown), suggesting that BTAK amplification may be specifically associated with aneuploidization in breast and ovarian cancers.

According to the present results and previous data from breast cancer,the coamplification of separate regions at 20q is likely to reflect the molecular mechanisms underlying gene amplification. Similar coamplifications of separate regions along the same chromosome arm have been described in 17q (ERBB2 at 17q12 and in unidentified regions at 17q23–24) and 8q24.1 (MYC) and 8q22 regions (26, 27). The exact molecular mechanisms are unknown,although multiple chromosomal breaks and unbalanced translocations may be predisposing cytogenetic aberrations. Coamplifications also suggest synergism between genes up-regulated by gene amplification. The amplified genes may, for example, share a common pathway in growth regulation.

We thank Dr. Anne Kallioniemi for providing a BAC probe for the MYBL2-gene.