- Email: [email protected]
Marked allelic imbalance on chromosome 5q31 does not explain α-catenin expression in epithelial ovarian cancer
Gynecologic Oncology 94 (2004) 416 – 421 www.elsevier.com/locate/ygyno
Marked allelic imbalance on chromosome 5q31 does not explain a-catenin expression in epithelial ovarian cancer Hanna Tuhkanen, a,b Maarit Anttila, a,c Veli-Matti Kosma, a,b,d Jukka Puolakka, e Matti Juhola, f Seppo Heinonen, c and Arto Mannermaa a,d,g,* a
Department of Pathology and Forensic Medicine, University of Kuopio and Kuopio University Hospital, 70210 Kuopio, Finland b Department of Oncology, Kuopio University Hospital, 70210 Kuopio, Finland c Department of Obstetrics and Gynecology, University of Kuopio and Kuopio University Hospital, 70210 Kuopio, Finland d Department of Pathology, Center for Laboratory Medicine, University of Tampere and Tampere University Hospital, 33521 Tampere, Finland e Department of Obstetrics and Gynecology, Jyva¨skyla¨ Central Hospital, 40620 Jyva¨skyla¨, Finland f Department of Pathology, Jyva¨skyla¨ Central Hospital, 40620 Jyva¨skyla¨, Finland g Department of Clinical Genetics, University of Oulu and Oulu University Hospital, 90029 Oulu, Finland Received 18 November 2003 Available online 19 June 2004
Abstract Objective. Human a-catenin gene (CTNNA1) on chromosome 5q31 is aberrantly expressed in various types of cancer including epithelial ovarian tumors. Allelic imbalance on this region has also been described in several malignant diseases. In the present work, the role of CTNNA1 as a candidate tumor suppressor gene was studied by comparing protein expression with allelic imbalance in human epithelial ovarian tumors. Methods. a-Catenin protein expression was determined from two areas of 41 tumors, and tissues from these areas were microdissected. After DNA extraction, AI analysis was carried out with eight microsatellite markers. Results. Altogether, 93% of the tumors (38 of 41) showed allelic imbalance at one or more loci. Two distinct common regions of allelic imbalance were identified, one comprising markers D5S2002 and D5S1995 and the other markers D5S393 and D5S476. Loss of the CTNNA1 gene did not appear to be involved in down-regulation of a-catenin in ovarian tumors, since allelic imbalance with a variety of markers, including CTNNA1 associated marker D5S476, was found in tumor samples independently of a-catenin expression. Furthermore, allelic imbalance analyses of two different samples from the same tumor revealed genetic heterogeneity. Conclusions. High allelic imbalance frequency indicates that chromosomal region 5q31 is functionally important in epithelial ovarian cancer. Allelic imbalance occurs at two distinct regions of which one includes the CTNNA1 gene. However, this gene is likely to be inactivated by mechanisms other than allelic imbalance. In addition, genetic heterogeneity observed in these tumors demonstrates the multiclonal nature of epithelial ovarian tumors. D 2004 Elsevier Inc. All rights reserved. Keywords: a-Catenin; Allelic imbalance; Chromosome 5q31; Epithelial ovarian cancer
Introduction Human a-catenin gene (CTNNA1) located to 5q31 is a member of the E-cadherin– catenin complex that plays an important role in cell – cell adhesion. Abnormal a-catenin expression has been detected in several cancers [1,2]. We * Corresponding author. Department of Pathology and Forensic Medicine, University of Kuopio, P.O. Box 1627, 70211 Kuopio, Finland. Fax: +358-17-162753. E-mail address: [email protected] (A. Mannermaa). 0090-8258/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2004.04.029
have previously reported that reduced a-catenin expression is a useful marker of those stage I FIGO (International Federation of Gynecology and Obstetrics) tumors that are likely to run an unfavorable course [3]. Frequent allelic imbalance (AI), an abnormal ratio between two alleles at one locus, is a useful tool to identify genes that could play a critical role in the development and progression of the malignant tumor. A number of AI regions have been found in ovarian cancer [4,5] and one of them is found on chromosomal region 5q [4,6]. AI on this specific region has been detected also in many other malignant
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
diseases, making this region potentially interesting in cancer research [7,8]. However, as yet, the target gene for AI on 5q31 has not been specified. The aim of this study was to compare a-catenin expression and AI frequency within a tumor on chromosomal area under consideration and to test whether the CTNNA1 gene acts as a tumor suppressor gene in epithelial ovarian cancer. AI region on 5q was characterized by analyzing 41 epithelial ovarian adenocarcinomas using seven specific microsatellite markers adjacent to CTNNA1 and one microsatellite marker from the short arm of the chromosome.
Materials and methods Samples The samples of the present study were collected from patients diagnosed and treated for epithelial ovarian cancer at Kuopio University Hospital and Jyva¨skyla¨ Central Hospital, Finland, between 1976 and 1992 [3]. Suitable paraffinembedded material was available from 41 tumors. The patients were followed up until September 1996. Tumor staging was re-evaluated from patients’ operative and histopatological files and based on FIGO standards and histological typing and grading was performed according to the World Health Organization classification [9] (Table 1). From Table 1 Clinicopathological data of the patients Characteristic
No.
Epithelial histologic subtype Serous Endometrioid Clear cell Miscellaneousa
13 11 9 8
32 27 22 19
Histologic grade 1 2 3
4 10 27
10 24 66
FIGO stage I II III IV
13 11 16 1
32 27 39 2
Chemotherapy responseb Yes No
15 12
56 44
End statec Dead Alive
22 14
61 39
Total
41
100
a
%
Includes 1 mucinous, 4 mixed epithelial and 3 unclassified epithelial cases. b No data in 14 cases. c Dead, because of other reason (not ovarian cancer): 5 cases.
417
early stage patients, 70% were clinically staged. Normal tissue from the same patient, mainly uterine tissue, was used as a control for AI analyses. Immunohistochemistry and evaluation of staining Tissue samples were fixed in 10% formalin and embedded in paraffin. a-Catenin expression was detected by indirect avidin – biotin peroxidase immunohistochemistry. An anti-mouse a-catenin antibody that reacts with human homologue [immunoglobulin G1 (IgG); Transduction Laboratories; Lexington, KY, USA] was used in a dilution of 1:150. As described previously in detail, immunostaining results were classified either to normal when all cells expressed a-catenin continuously on cell membranes or reduced when clearly less than 100% of the tumor cells expressed a-catenin on cell membranes or the positivity was fragmented (uncontinuous) [3,10] (Fig. 1). Tissue microdissection Five sequential sections (5 AM) from 41 ovarian tumors stained by immunohistochemistry were microdissected manually for AI analyses. The microdissection was made using a 24-G needle under a light microscope according to a-catenin protein expression (Figs. 1A, B). Altogether, 35 sample pairs could be formed from areas of normal and areas of reduced acatenin expression for comparison of CTNNA1 expression level and AI frequency. In addition, six tumors with normal a-catenin expression were microdissected. Microdissected areas were confirmed by an experienced pathologist to contain at least 70% tumor cells. For each specimen, DNA was extracted by the proteinase-K – phenol – chloroform method following standard protocols [11]. AI analysis Seven microsatellite markers D5S2057, D5S2002, D5S1995, D5S808, D5S396, D5S393 and D5S467 spanning 7.1 Mb were chosen from http://www.ncbi.nlm.nih.gov/. Of these markers, D5S467 locates close to CTNNA1. In addition, D5S807, which lies on the short arm of the chromosome 5, was applied to evaluate the changes at the level of larger chromosomal regions (Fig. 2). Primer sequences and PCR conditions were obtained from http://www.gdb.org/ and the forward primers were labeled with 6-FAM. The target sequences were amplified by PCR in a total volume of 10 Al with AmpliTaq Gold polymerase (Applied Biosystems, Foster City, CA, USA) following manufacturer’s recommendations. Fragment analysis was done utilizing ABI Prism 310 genetic analyzer (Applied Biosystems) from 41 tumors and their controls. AI was calculated from the informative cases, where heterozygosity could be seen in the normal tissue samples. Because PCR fragments of different sizes are amplified with different efficiencies, the ratio of allele peak heights
418
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
[12]. Values >1.67 or <0.60 were assessed for AI, meaning that one of the alleles has decreased more than 40% [13] (Fig. 3A). Values between 0.6 and 0.7 or between 1.55 and 1.67 were considered as grey zone values. Samples that had a grey zone AI value in the first analysis were reamplified and reanalyzed at this particular locus. If this second result was positive, the tumor was assessed to have AI. A locus with AI in more than 30% of the cases was considered significant. Statistical analysis Statistical analyses were carried out using SPSS 9.0 software. Significance levels for association between categorical variables were computed by performing Chi-squared and Fisher’s exact tests. Univariate survival analyses were based on Kaplan –Meier method. The differences between curves were analyzed using the log-rank test. The level of statistical significance was set to P < 0.01.
Results
Fig. 1. (A) Tumor area of a clear cell ovarian cancer with normal a-catenin expression. (B) An area of the same tumor with significantly reduced acatenin expression. Bar = 50 Am.
was calculated using matched tumor and normal samples. The AI value was calculated from formula AI = (T2 N1)/ (T1 N2), where T is a tumor allele and N is a normal allele
Altogether, 93% (38/41) of the tumors exhibited AI at one or more loci and all eight microsatellite markers showed significant AI (Fig. 2). Large-scale losses reaching from the markers D5S807 to D5S476 could be seen in several tumors such as in number 26 (Fig. 3B). Two separate common regions of AI were defined on 5q, one depicted by the markers D5S2002 and D5S1995 and the other by D5S393 and D5S476 at the CTNNA1 locus. Both these regions were clearly distinguished in tumors such as number 7, 11 and 34 and 15, 18 and 23, respectively (Fig. 3B). AI was not related to clinicopathological parameters such as histologic subtype, stage, histologic grade nor end state (Table 1). AI dependence on possible recurrence and month of survival were calculated, but did not show significant P values.
Fig. 2. Chromosome 5 idiogram, microsatellite markers, their physical distances and the results of AI analyses in the tumors as well as in the areas of normal and reduced a-catenin protein expression. aHeterozygosity in normal sample. bA case was assessed to have AI if either normal or reduced area or both had AI.
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
419
Fig. 3. (A) An example of AI assessment. Alleles are shaded, peak heights are given in boxes, and the AI value in this case is 0.42. (B – D) AI on chromosome 5. The order of genetic markers is shown on the left. Each vertical line represents one tumor and case numbers are shown under each vertical line. Only the tumors with AI are marked in the figure. The tumor areas of (B) normal and (C) reduced a-catenin protein expression. Percentages of areas expressing acatenin are marked in C. (D) Intratumor heterogeneity seen between the areas of normal and reduced a-catenin protein expression.
However, AI on the marker D5S808 was associated with poor chemotherapy response ( P = 0.001). There were no significant differences between AI frequencies with CTNNA1 adjacent marker D5S476 or any other marker in samples classified according to a-catenin protein expression (Figs. 1 and 3B, C). The level of reduced a-catenin expression was not related to AI either. At region D5S393 – D5S476, 39% of tumors (9/23) showed heterogeneous AI result between normal and reduced expression areas. Intratumor AI analyses revealed extensive genetic heterogeneity. Heterogeneity was detected in 79% of the tumors (23/29) (Fig. 3D) when only the tumors with AI results from both normal and reduced a-catenin expression areas were taken into account. Indeed, of the eight microsatellite marker loci examined in one tumor, up to five microsatellite markers presented discrepancy between the samples studied. At both AI regions D5S2002 – D5S1995 and D5S393 –D5S476, 39% of tumors exhibited heterogeneous results. On the other hand, in the middle of the identified AI-rich regions, markers D5S808 and D5S396 did not show significantly different AI frequencies in normal and reduced a-catenin expression areas ( P = 0.001 and P = 0.002). Intratumor heterogeneity was not associated with clinicopathological characteristics.
Discussion The main finding of this study is the more specific location of AI to two distinct areas on 5q31. Another
interesting finding was that AI seems not to be behind the inactivation of the CTNNA1 gene, since tumor samples with normal and reduced a-catenin expression both showed AI with all chromosome 5 specific markers studied. In the present work, we searched for evidence of ovarian cancer tumor suppressor gene(s) on chromosome 5q by means of AI analyses. For the AI analyses, immunohistochemically determined areas having normal or significantly reduced a-catenin protein expression in the same tumor were microdissected. These two distinct areas allowed us to study both the relationship between a-catenin expression and AI and heterogeneity of the tumors. We have shown previously that reduction of a-catenin expression is an important step in ovarian tumorigenesis [3]. However, the data presented here show no evidence for AI as a principal mechanism behind reduced a-catenin expression. It is therefore probable that the CTNNA1 gene is inactivated mainly by other mechanisms, such as methylation, mutations or splice variants [14] or through posttranscriptional machinery. Precise CTNNA1 inactivation mechanism is thus not yet resolved and should be elucidated in future studies. Abundant AI in the tumor areas of normal a-catenin expression could be targeted to other genes close to the CTNNA1 gene, such as EGR1, which has been shown to have decreased expression in several types of tumor cell lines and is suggested to be a tumor suppressor gene [15]. AI was exhibited abundantly with all seven informative microsatellite markers on 5q. This finding strongly suggests that some of the ovarian cancer tumor suppressor gene(s) are
420
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
located on chromosomal area 5q, as reported previously [4,6]. However, in those studies, the entire chromosome 5 has been screened, and long marker distances have made it difficult to define a specific common region of AI. We could classify two distinct common regions on 5q on the basis of AI frequency in tumors (Fig. 3B), the most frequent AI region being at the marker D5S1995. Another very frequent AI region is located at the markers D5S393 and at D5S476, the latter of which is a CTNNA1 associated marker. Large deletions reaching apparently both p- and q-arms of the chromosome 5 could be detected in several tumors, although 5p AI at the marker D5S807 does not necessarily confirm the whole chromosome loss. Our results suggest that AI on 5p could be independent from 5q AI since there are tumors that show AI on 5p but not at D5S1995 (Figs. 3B, C). Other AI studies on ovarian cancer report diverse results on the chromosomal region 5p, with AI percents ranging from 0 to 32 [4– 6]. This is partially due to different material and microsatellite markers used. Also methodological development may play a role herein. Furthermore, we tested the possible association between AI on 5p and various clinicopathological factors. We assessed the significance level to P < 0.01 instead of P < 0.05 to avoid the type 1 statistical error due to large number of parameters. Significant association was found between the marker D5S808 and poor chemotherapy response ( P = 0.001), whereas no other specific clinical features could be linked to any of these regions. These results imply that either AI in the detected two regions is not restricted to any specific clinical subtype of ovarian cancer or that our material was too narrow to be able to show this. Intratumor heterogeneity has been reported in a number of molecular pathologic studies recently and the concept of genetic heterogeneity in tumors is generally accepted [16]. In the present study, AI analyses revealed abundant intratumor genetic heterogeneity reflecting the multiclonal nature of epithelial ovarian tumors. This also supports the view of multistep carcinogenesis of epithelial ovarian tumors similar to mucinous ovarian tumors presented by Takeshima et al. [17]. At D5S808 and D5S396, AI data showed significant similarity, but these two are the loci with lowest AI in this study. Due to intratumoral heterogeneity, it is crucial to carefully select the material for AI and other analyses of cancer specimens. For example, analysis of specific cell types separately could help to better understand the role of both genetic alterations and changes at protein expression levels in tumor growth and progression. The detection of AI has been used to identify genomic regions that harbor tumor suppressor genes. In the present work, contamination by normal cells was minimized by microdissection and evaluation of samples by an experienced pathologist. In addition, the same material was used in our previous study [18], where we used both manual microdissection and laser capture microdissection, and the results showed that carefully done manual microdissection gives results that are comparable to laser capture microdissection.
Technical flaws of manual AI analysis were prevented by using a fluorescence-based fragment analyzer. In addition, to improve the AI assessment, we have analyzed ambiguous samples twice. Microsatellite markers on the 5q region were arranged according to a sequence based map on http:// www.ncbi.nlm.nih.gov/ instead of genetic maps. Taken together, we believe our methods allowed us to reliably further characterize AI region 5p31.1 and to determine the role of AI as an inactivation mechanism of CTNNA1 in ovarian cancer. In summary, this study strengthened the conclusion that AI on 5q31 is abundant in human epithelial ovarian cancer, and furthermore, two minimal deleted regions were characterized. Expression of a-catenin, however, was not principally reduced by means of AI, since there was also marked AI detected in areas showing normal a-catenin expression. AI on this chromosomal region is probably targeted for other tumor suppressor gene(s) that most probably locate in the two here described regions. Finally, frequent intratumor genetic heterogeneity was shown, indicating that the selection of the tumor area is critical and impacts to results of molecular analyses in epithelial ovarian cancer.
Acknowledgments This study was supported by the Special Government Funding (EVO) of Kuopio and Tampere University Hospitals and the Cancer Fund of North Savo.
References [1] Davies BR, Worsley SD, Ponder BAJ. Expression of E-cadherin, acatenin and h-catenin in normal ovarian surface epithelium and epithelial ovarian cancers. Histopathology 1998;32:69 – 80. [2] Bo¨hm J, Niskanen L, Kiraly K, Kellokoski J, Eskelinen M, Hollmen S, et al. Expression and prognostic value of a-, h-, and g-catenins in differentiated thyroid carcinoma. J Clin Endocrinol Metab 2000;85: 4806 – 11. [3] Anttila M, Kosma VM, Ji H, Wei-Ling X, Puolakka J, Juhola M, et al. Clinical significance of a-catenin, collagen IV, and Ki-67 expression in epithelial ovarian cancer. J Clin Oncol 1998;16:2591 – 600. [4] Cliby W, Ritland S, Hartmann L, Dodson M, Halling KC, Keeney G, et al. Human epithelial ovarian cancer allelotype. Cancer Res 1993;53(10 Suppl.):2393 – 8. [5] Dodson MK, Hartmann LC, Cliby WA, DeLacey KA, Keeney GL, Ritland SR, et al. Comparison of loss of heterozygosity patterns in invasive low-grade and high-grade epithelial ovarian carcinomas. Cancer Res 1993;53:4456 – 60. [6] Wang VW, Bell DA, Berkowitz RS, Mok SC. Whole genome amplification and high-throughput allelotyping identified five distinct deletion regions on chromosomes 5 and 6 in microdissected early-stage ovarian tumors. Cancer Res 2001;61:4169 – 74. [7] Ogasawara S, Tamura G, Maesawa C, Suzuki Y, Ishida Y, Satoh N, et al. Common deleted region on the long arm of chromosome 5 in esophageal carcinoma. Gastroenterology 1996;110:52 – 7. [8] Lai F, Godley L, Joslin J, Fernald A, Liu J, Espinosa R, et al. Transcript map and comparative analysis of the 1.5 Mb commonly deleted segment of human 5q31 in malignant myeloid diseases with a del(5q). Genomics 2001;71:235 – 45. [9] Scully RE. Histological typing of ovarian tumours. World Health
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
[10]
[11]
[12]
[13]
[14]
Organization, International Histological Classification of Tumours. Second ed. New York: Springer; 1999. Hashizume R, Koizumi H, Ihara A, Ohta T, Uchikoshi T. Expression of h-catenin in normal breast tissue and breast carcinoma: a comparative study with epithelial cadherin and a-catenin. Histopathology 1996;29:139 – 46. Isola J, De Vries S, Chu L, Ghazvini S, Waldman F. Analysis of changes in DNA sequence copy number by comparative genomic hybridization in archival paraffin-embedded tumor samples. Am J Pathol 1994;145:1301 – 8. Lassus H, Laitinen MPE, Anttonen M, Heikinheimo M, Aaltonen LA, Ritvos O, et al. Comparison of serous and mucinous ovarian carcinomas: distinct pattern of allelic imbalance at distal 8p and expression of transcription factor GATA-4. Lab Invest 2001; 81:517 – 26. Canzian F, Salovaara R, Hemminki A, Kristo P, Chadwick R, Aaltonen LA, et al. Semiautomated assessment of loss of heterozygosity and replication error in tumors. Cancer Res 1996;56:3331 – 7. Vanpoucke G, Nollet F, Tejpar S, Cassiman JJ, van Roy F. The human
[15]
[16]
[17]
[18]
421
aE-catenin gene CTNNA1: mutational analysis and rare occurrence of a truncated splice variant. Biochim Biophys Acta 2002;1574:262 – 8. Huang RP, Fan Y, de Belle I, Niemeyer C, Gottardis MM, Mercola D, et al. Decreased Egr-1 expression in human, mouse and rat mammary cells and tissues correlates with tumor formation. Int J Cancer 1997; 72:102 – 9. Zborovskaya I, Gasparian A, Karseladze A, Elcheva I, Trofimova E, Driouch K, et al. Somatic genetic alterations (LOH) in benign, borderline and invasive ovarian tumours: intratumoral molecular heterogeneity. Int J Cancer 1999;82:822 – 6. Takeshima Y, Amatya VJ, Daimaru Y, Nakayori F, Nakano T, Inai K. Heterogeneous genetic alterations in ovarian mucinous tumors: application and usefulness of laser capture microdissection. Hum Pathol 2001;32:1203 – 8. Tuhkanen H, Anttila M, Kosma VM, Yla-Herttuala S, Heinonen S, Kuronen A, et al. Genetic alterations in the peritumoral stromal cells of malignant and borderline epithelial ovarian tumors as indicated by allelic imbalance on chromosome 3p. Int J Cancer 2004;109:247 – 52.
Marked allelic imbalance on chromosome 5q31 does not explain a-catenin expression in epithelial ovarian cancer Hanna Tuhkanen, a,b Maarit Anttila, a,c Veli-Matti Kosma, a,b,d Jukka Puolakka, e Matti Juhola, f Seppo Heinonen, c and Arto Mannermaa a,d,g,* a
Department of Pathology and Forensic Medicine, University of Kuopio and Kuopio University Hospital, 70210 Kuopio, Finland b Department of Oncology, Kuopio University Hospital, 70210 Kuopio, Finland c Department of Obstetrics and Gynecology, University of Kuopio and Kuopio University Hospital, 70210 Kuopio, Finland d Department of Pathology, Center for Laboratory Medicine, University of Tampere and Tampere University Hospital, 33521 Tampere, Finland e Department of Obstetrics and Gynecology, Jyva¨skyla¨ Central Hospital, 40620 Jyva¨skyla¨, Finland f Department of Pathology, Jyva¨skyla¨ Central Hospital, 40620 Jyva¨skyla¨, Finland g Department of Clinical Genetics, University of Oulu and Oulu University Hospital, 90029 Oulu, Finland Received 18 November 2003 Available online 19 June 2004
Abstract Objective. Human a-catenin gene (CTNNA1) on chromosome 5q31 is aberrantly expressed in various types of cancer including epithelial ovarian tumors. Allelic imbalance on this region has also been described in several malignant diseases. In the present work, the role of CTNNA1 as a candidate tumor suppressor gene was studied by comparing protein expression with allelic imbalance in human epithelial ovarian tumors. Methods. a-Catenin protein expression was determined from two areas of 41 tumors, and tissues from these areas were microdissected. After DNA extraction, AI analysis was carried out with eight microsatellite markers. Results. Altogether, 93% of the tumors (38 of 41) showed allelic imbalance at one or more loci. Two distinct common regions of allelic imbalance were identified, one comprising markers D5S2002 and D5S1995 and the other markers D5S393 and D5S476. Loss of the CTNNA1 gene did not appear to be involved in down-regulation of a-catenin in ovarian tumors, since allelic imbalance with a variety of markers, including CTNNA1 associated marker D5S476, was found in tumor samples independently of a-catenin expression. Furthermore, allelic imbalance analyses of two different samples from the same tumor revealed genetic heterogeneity. Conclusions. High allelic imbalance frequency indicates that chromosomal region 5q31 is functionally important in epithelial ovarian cancer. Allelic imbalance occurs at two distinct regions of which one includes the CTNNA1 gene. However, this gene is likely to be inactivated by mechanisms other than allelic imbalance. In addition, genetic heterogeneity observed in these tumors demonstrates the multiclonal nature of epithelial ovarian tumors. D 2004 Elsevier Inc. All rights reserved. Keywords: a-Catenin; Allelic imbalance; Chromosome 5q31; Epithelial ovarian cancer
Introduction Human a-catenin gene (CTNNA1) located to 5q31 is a member of the E-cadherin– catenin complex that plays an important role in cell – cell adhesion. Abnormal a-catenin expression has been detected in several cancers [1,2]. We * Corresponding author. Department of Pathology and Forensic Medicine, University of Kuopio, P.O. Box 1627, 70211 Kuopio, Finland. Fax: +358-17-162753. E-mail address: [email protected] (A. Mannermaa). 0090-8258/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2004.04.029
have previously reported that reduced a-catenin expression is a useful marker of those stage I FIGO (International Federation of Gynecology and Obstetrics) tumors that are likely to run an unfavorable course [3]. Frequent allelic imbalance (AI), an abnormal ratio between two alleles at one locus, is a useful tool to identify genes that could play a critical role in the development and progression of the malignant tumor. A number of AI regions have been found in ovarian cancer [4,5] and one of them is found on chromosomal region 5q [4,6]. AI on this specific region has been detected also in many other malignant
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
diseases, making this region potentially interesting in cancer research [7,8]. However, as yet, the target gene for AI on 5q31 has not been specified. The aim of this study was to compare a-catenin expression and AI frequency within a tumor on chromosomal area under consideration and to test whether the CTNNA1 gene acts as a tumor suppressor gene in epithelial ovarian cancer. AI region on 5q was characterized by analyzing 41 epithelial ovarian adenocarcinomas using seven specific microsatellite markers adjacent to CTNNA1 and one microsatellite marker from the short arm of the chromosome.
Materials and methods Samples The samples of the present study were collected from patients diagnosed and treated for epithelial ovarian cancer at Kuopio University Hospital and Jyva¨skyla¨ Central Hospital, Finland, between 1976 and 1992 [3]. Suitable paraffinembedded material was available from 41 tumors. The patients were followed up until September 1996. Tumor staging was re-evaluated from patients’ operative and histopatological files and based on FIGO standards and histological typing and grading was performed according to the World Health Organization classification [9] (Table 1). From Table 1 Clinicopathological data of the patients Characteristic
No.
Epithelial histologic subtype Serous Endometrioid Clear cell Miscellaneousa
13 11 9 8
32 27 22 19
Histologic grade 1 2 3
4 10 27
10 24 66
FIGO stage I II III IV
13 11 16 1
32 27 39 2
Chemotherapy responseb Yes No
15 12
56 44
End statec Dead Alive
22 14
61 39
Total
41
100
a
%
Includes 1 mucinous, 4 mixed epithelial and 3 unclassified epithelial cases. b No data in 14 cases. c Dead, because of other reason (not ovarian cancer): 5 cases.
417
early stage patients, 70% were clinically staged. Normal tissue from the same patient, mainly uterine tissue, was used as a control for AI analyses. Immunohistochemistry and evaluation of staining Tissue samples were fixed in 10% formalin and embedded in paraffin. a-Catenin expression was detected by indirect avidin – biotin peroxidase immunohistochemistry. An anti-mouse a-catenin antibody that reacts with human homologue [immunoglobulin G1 (IgG); Transduction Laboratories; Lexington, KY, USA] was used in a dilution of 1:150. As described previously in detail, immunostaining results were classified either to normal when all cells expressed a-catenin continuously on cell membranes or reduced when clearly less than 100% of the tumor cells expressed a-catenin on cell membranes or the positivity was fragmented (uncontinuous) [3,10] (Fig. 1). Tissue microdissection Five sequential sections (5 AM) from 41 ovarian tumors stained by immunohistochemistry were microdissected manually for AI analyses. The microdissection was made using a 24-G needle under a light microscope according to a-catenin protein expression (Figs. 1A, B). Altogether, 35 sample pairs could be formed from areas of normal and areas of reduced acatenin expression for comparison of CTNNA1 expression level and AI frequency. In addition, six tumors with normal a-catenin expression were microdissected. Microdissected areas were confirmed by an experienced pathologist to contain at least 70% tumor cells. For each specimen, DNA was extracted by the proteinase-K – phenol – chloroform method following standard protocols [11]. AI analysis Seven microsatellite markers D5S2057, D5S2002, D5S1995, D5S808, D5S396, D5S393 and D5S467 spanning 7.1 Mb were chosen from http://www.ncbi.nlm.nih.gov/. Of these markers, D5S467 locates close to CTNNA1. In addition, D5S807, which lies on the short arm of the chromosome 5, was applied to evaluate the changes at the level of larger chromosomal regions (Fig. 2). Primer sequences and PCR conditions were obtained from http://www.gdb.org/ and the forward primers were labeled with 6-FAM. The target sequences were amplified by PCR in a total volume of 10 Al with AmpliTaq Gold polymerase (Applied Biosystems, Foster City, CA, USA) following manufacturer’s recommendations. Fragment analysis was done utilizing ABI Prism 310 genetic analyzer (Applied Biosystems) from 41 tumors and their controls. AI was calculated from the informative cases, where heterozygosity could be seen in the normal tissue samples. Because PCR fragments of different sizes are amplified with different efficiencies, the ratio of allele peak heights
418
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
[12]. Values >1.67 or <0.60 were assessed for AI, meaning that one of the alleles has decreased more than 40% [13] (Fig. 3A). Values between 0.6 and 0.7 or between 1.55 and 1.67 were considered as grey zone values. Samples that had a grey zone AI value in the first analysis were reamplified and reanalyzed at this particular locus. If this second result was positive, the tumor was assessed to have AI. A locus with AI in more than 30% of the cases was considered significant. Statistical analysis Statistical analyses were carried out using SPSS 9.0 software. Significance levels for association between categorical variables were computed by performing Chi-squared and Fisher’s exact tests. Univariate survival analyses were based on Kaplan –Meier method. The differences between curves were analyzed using the log-rank test. The level of statistical significance was set to P < 0.01.
Results
Fig. 1. (A) Tumor area of a clear cell ovarian cancer with normal a-catenin expression. (B) An area of the same tumor with significantly reduced acatenin expression. Bar = 50 Am.
was calculated using matched tumor and normal samples. The AI value was calculated from formula AI = (T2 N1)/ (T1 N2), where T is a tumor allele and N is a normal allele
Altogether, 93% (38/41) of the tumors exhibited AI at one or more loci and all eight microsatellite markers showed significant AI (Fig. 2). Large-scale losses reaching from the markers D5S807 to D5S476 could be seen in several tumors such as in number 26 (Fig. 3B). Two separate common regions of AI were defined on 5q, one depicted by the markers D5S2002 and D5S1995 and the other by D5S393 and D5S476 at the CTNNA1 locus. Both these regions were clearly distinguished in tumors such as number 7, 11 and 34 and 15, 18 and 23, respectively (Fig. 3B). AI was not related to clinicopathological parameters such as histologic subtype, stage, histologic grade nor end state (Table 1). AI dependence on possible recurrence and month of survival were calculated, but did not show significant P values.
Fig. 2. Chromosome 5 idiogram, microsatellite markers, their physical distances and the results of AI analyses in the tumors as well as in the areas of normal and reduced a-catenin protein expression. aHeterozygosity in normal sample. bA case was assessed to have AI if either normal or reduced area or both had AI.
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
419
Fig. 3. (A) An example of AI assessment. Alleles are shaded, peak heights are given in boxes, and the AI value in this case is 0.42. (B – D) AI on chromosome 5. The order of genetic markers is shown on the left. Each vertical line represents one tumor and case numbers are shown under each vertical line. Only the tumors with AI are marked in the figure. The tumor areas of (B) normal and (C) reduced a-catenin protein expression. Percentages of areas expressing acatenin are marked in C. (D) Intratumor heterogeneity seen between the areas of normal and reduced a-catenin protein expression.
However, AI on the marker D5S808 was associated with poor chemotherapy response ( P = 0.001). There were no significant differences between AI frequencies with CTNNA1 adjacent marker D5S476 or any other marker in samples classified according to a-catenin protein expression (Figs. 1 and 3B, C). The level of reduced a-catenin expression was not related to AI either. At region D5S393 – D5S476, 39% of tumors (9/23) showed heterogeneous AI result between normal and reduced expression areas. Intratumor AI analyses revealed extensive genetic heterogeneity. Heterogeneity was detected in 79% of the tumors (23/29) (Fig. 3D) when only the tumors with AI results from both normal and reduced a-catenin expression areas were taken into account. Indeed, of the eight microsatellite marker loci examined in one tumor, up to five microsatellite markers presented discrepancy between the samples studied. At both AI regions D5S2002 – D5S1995 and D5S393 –D5S476, 39% of tumors exhibited heterogeneous results. On the other hand, in the middle of the identified AI-rich regions, markers D5S808 and D5S396 did not show significantly different AI frequencies in normal and reduced a-catenin expression areas ( P = 0.001 and P = 0.002). Intratumor heterogeneity was not associated with clinicopathological characteristics.
Discussion The main finding of this study is the more specific location of AI to two distinct areas on 5q31. Another
interesting finding was that AI seems not to be behind the inactivation of the CTNNA1 gene, since tumor samples with normal and reduced a-catenin expression both showed AI with all chromosome 5 specific markers studied. In the present work, we searched for evidence of ovarian cancer tumor suppressor gene(s) on chromosome 5q by means of AI analyses. For the AI analyses, immunohistochemically determined areas having normal or significantly reduced a-catenin protein expression in the same tumor were microdissected. These two distinct areas allowed us to study both the relationship between a-catenin expression and AI and heterogeneity of the tumors. We have shown previously that reduction of a-catenin expression is an important step in ovarian tumorigenesis [3]. However, the data presented here show no evidence for AI as a principal mechanism behind reduced a-catenin expression. It is therefore probable that the CTNNA1 gene is inactivated mainly by other mechanisms, such as methylation, mutations or splice variants [14] or through posttranscriptional machinery. Precise CTNNA1 inactivation mechanism is thus not yet resolved and should be elucidated in future studies. Abundant AI in the tumor areas of normal a-catenin expression could be targeted to other genes close to the CTNNA1 gene, such as EGR1, which has been shown to have decreased expression in several types of tumor cell lines and is suggested to be a tumor suppressor gene [15]. AI was exhibited abundantly with all seven informative microsatellite markers on 5q. This finding strongly suggests that some of the ovarian cancer tumor suppressor gene(s) are
420
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
located on chromosomal area 5q, as reported previously [4,6]. However, in those studies, the entire chromosome 5 has been screened, and long marker distances have made it difficult to define a specific common region of AI. We could classify two distinct common regions on 5q on the basis of AI frequency in tumors (Fig. 3B), the most frequent AI region being at the marker D5S1995. Another very frequent AI region is located at the markers D5S393 and at D5S476, the latter of which is a CTNNA1 associated marker. Large deletions reaching apparently both p- and q-arms of the chromosome 5 could be detected in several tumors, although 5p AI at the marker D5S807 does not necessarily confirm the whole chromosome loss. Our results suggest that AI on 5p could be independent from 5q AI since there are tumors that show AI on 5p but not at D5S1995 (Figs. 3B, C). Other AI studies on ovarian cancer report diverse results on the chromosomal region 5p, with AI percents ranging from 0 to 32 [4– 6]. This is partially due to different material and microsatellite markers used. Also methodological development may play a role herein. Furthermore, we tested the possible association between AI on 5p and various clinicopathological factors. We assessed the significance level to P < 0.01 instead of P < 0.05 to avoid the type 1 statistical error due to large number of parameters. Significant association was found between the marker D5S808 and poor chemotherapy response ( P = 0.001), whereas no other specific clinical features could be linked to any of these regions. These results imply that either AI in the detected two regions is not restricted to any specific clinical subtype of ovarian cancer or that our material was too narrow to be able to show this. Intratumor heterogeneity has been reported in a number of molecular pathologic studies recently and the concept of genetic heterogeneity in tumors is generally accepted [16]. In the present study, AI analyses revealed abundant intratumor genetic heterogeneity reflecting the multiclonal nature of epithelial ovarian tumors. This also supports the view of multistep carcinogenesis of epithelial ovarian tumors similar to mucinous ovarian tumors presented by Takeshima et al. [17]. At D5S808 and D5S396, AI data showed significant similarity, but these two are the loci with lowest AI in this study. Due to intratumoral heterogeneity, it is crucial to carefully select the material for AI and other analyses of cancer specimens. For example, analysis of specific cell types separately could help to better understand the role of both genetic alterations and changes at protein expression levels in tumor growth and progression. The detection of AI has been used to identify genomic regions that harbor tumor suppressor genes. In the present work, contamination by normal cells was minimized by microdissection and evaluation of samples by an experienced pathologist. In addition, the same material was used in our previous study [18], where we used both manual microdissection and laser capture microdissection, and the results showed that carefully done manual microdissection gives results that are comparable to laser capture microdissection.
Technical flaws of manual AI analysis were prevented by using a fluorescence-based fragment analyzer. In addition, to improve the AI assessment, we have analyzed ambiguous samples twice. Microsatellite markers on the 5q region were arranged according to a sequence based map on http:// www.ncbi.nlm.nih.gov/ instead of genetic maps. Taken together, we believe our methods allowed us to reliably further characterize AI region 5p31.1 and to determine the role of AI as an inactivation mechanism of CTNNA1 in ovarian cancer. In summary, this study strengthened the conclusion that AI on 5q31 is abundant in human epithelial ovarian cancer, and furthermore, two minimal deleted regions were characterized. Expression of a-catenin, however, was not principally reduced by means of AI, since there was also marked AI detected in areas showing normal a-catenin expression. AI on this chromosomal region is probably targeted for other tumor suppressor gene(s) that most probably locate in the two here described regions. Finally, frequent intratumor genetic heterogeneity was shown, indicating that the selection of the tumor area is critical and impacts to results of molecular analyses in epithelial ovarian cancer.
Acknowledgments This study was supported by the Special Government Funding (EVO) of Kuopio and Tampere University Hospitals and the Cancer Fund of North Savo.
References [1] Davies BR, Worsley SD, Ponder BAJ. Expression of E-cadherin, acatenin and h-catenin in normal ovarian surface epithelium and epithelial ovarian cancers. Histopathology 1998;32:69 – 80. [2] Bo¨hm J, Niskanen L, Kiraly K, Kellokoski J, Eskelinen M, Hollmen S, et al. Expression and prognostic value of a-, h-, and g-catenins in differentiated thyroid carcinoma. J Clin Endocrinol Metab 2000;85: 4806 – 11. [3] Anttila M, Kosma VM, Ji H, Wei-Ling X, Puolakka J, Juhola M, et al. Clinical significance of a-catenin, collagen IV, and Ki-67 expression in epithelial ovarian cancer. J Clin Oncol 1998;16:2591 – 600. [4] Cliby W, Ritland S, Hartmann L, Dodson M, Halling KC, Keeney G, et al. Human epithelial ovarian cancer allelotype. Cancer Res 1993;53(10 Suppl.):2393 – 8. [5] Dodson MK, Hartmann LC, Cliby WA, DeLacey KA, Keeney GL, Ritland SR, et al. Comparison of loss of heterozygosity patterns in invasive low-grade and high-grade epithelial ovarian carcinomas. Cancer Res 1993;53:4456 – 60. [6] Wang VW, Bell DA, Berkowitz RS, Mok SC. Whole genome amplification and high-throughput allelotyping identified five distinct deletion regions on chromosomes 5 and 6 in microdissected early-stage ovarian tumors. Cancer Res 2001;61:4169 – 74. [7] Ogasawara S, Tamura G, Maesawa C, Suzuki Y, Ishida Y, Satoh N, et al. Common deleted region on the long arm of chromosome 5 in esophageal carcinoma. Gastroenterology 1996;110:52 – 7. [8] Lai F, Godley L, Joslin J, Fernald A, Liu J, Espinosa R, et al. Transcript map and comparative analysis of the 1.5 Mb commonly deleted segment of human 5q31 in malignant myeloid diseases with a del(5q). Genomics 2001;71:235 – 45. [9] Scully RE. Histological typing of ovarian tumours. World Health
H. Tuhkanen et al. / Gynecologic Oncology 94 (2004) 416–421
[10]
[11]
[12]
[13]
[14]
Organization, International Histological Classification of Tumours. Second ed. New York: Springer; 1999. Hashizume R, Koizumi H, Ihara A, Ohta T, Uchikoshi T. Expression of h-catenin in normal breast tissue and breast carcinoma: a comparative study with epithelial cadherin and a-catenin. Histopathology 1996;29:139 – 46. Isola J, De Vries S, Chu L, Ghazvini S, Waldman F. Analysis of changes in DNA sequence copy number by comparative genomic hybridization in archival paraffin-embedded tumor samples. Am J Pathol 1994;145:1301 – 8. Lassus H, Laitinen MPE, Anttonen M, Heikinheimo M, Aaltonen LA, Ritvos O, et al. Comparison of serous and mucinous ovarian carcinomas: distinct pattern of allelic imbalance at distal 8p and expression of transcription factor GATA-4. Lab Invest 2001; 81:517 – 26. Canzian F, Salovaara R, Hemminki A, Kristo P, Chadwick R, Aaltonen LA, et al. Semiautomated assessment of loss of heterozygosity and replication error in tumors. Cancer Res 1996;56:3331 – 7. Vanpoucke G, Nollet F, Tejpar S, Cassiman JJ, van Roy F. The human
[15]
[16]
[17]
[18]
421
aE-catenin gene CTNNA1: mutational analysis and rare occurrence of a truncated splice variant. Biochim Biophys Acta 2002;1574:262 – 8. Huang RP, Fan Y, de Belle I, Niemeyer C, Gottardis MM, Mercola D, et al. Decreased Egr-1 expression in human, mouse and rat mammary cells and tissues correlates with tumor formation. Int J Cancer 1997; 72:102 – 9. Zborovskaya I, Gasparian A, Karseladze A, Elcheva I, Trofimova E, Driouch K, et al. Somatic genetic alterations (LOH) in benign, borderline and invasive ovarian tumours: intratumoral molecular heterogeneity. Int J Cancer 1999;82:822 – 6. Takeshima Y, Amatya VJ, Daimaru Y, Nakayori F, Nakano T, Inai K. Heterogeneous genetic alterations in ovarian mucinous tumors: application and usefulness of laser capture microdissection. Hum Pathol 2001;32:1203 – 8. Tuhkanen H, Anttila M, Kosma VM, Yla-Herttuala S, Heinonen S, Kuronen A, et al. Genetic alterations in the peritumoral stromal cells of malignant and borderline epithelial ovarian tumors as indicated by allelic imbalance on chromosome 3p. Int J Cancer 2004;109:247 – 52.