Loss of heterozygosity in sporadic oesophageal tumors in the ... - Nature

Oncogene (1998) 17, 2101 ± 2105 ã 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc

Loss of heterozygosity in sporadic oesophageal tumors in the tylosis oesophageal cancer (TOC) gene region of chromosome 17q M von Brevern1, MC Hollstein1, JM Risk3, J Garde3, WP Bennett2, CC Harris2, K-R Muehlbauer1 and JK Field3,4 1

German Cancer Research Centre (Deutsches Krebsforschungszentrum), Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany; Laboratory of Human Carcinogenesis, National Cancer Institute, Bethesda, Maryland 20892, USA; 3Molecular Genetics and Oncology Group, Department of Clinical Dental Sciences, University of Liverpool, Liverpool L69 3BX, UK; and 4Roy Castle International Centre for Lung Cancer Research, 200 London Road, Liverpool L3 4TA, UK

2

From the genotyping of UK and US tylotic families with a high risk of oesophageal cancer we have previously localized the tylosis-associated cancer susceptibility gene (TOC gene, tylosis oesophageal cancer gene) to a 1 cM region on the long arm of chromosome 17 (Kelsell et al., 1996). In the present study we investigated loss of heterozygosity (LOH) patterns of 35 sporadic squamous cell carcinomas of the oesophagus using six polymorphic microsatellite markers encompassing this locus. Twentyfour of the 35 cases (69%) revealed LOH at one or more loci. Deletion was most frequently observed with the marker D17S801 (64% LOH, informative cases), which shows signi®cant linkage to the TOC locus. The LOH analysis in sporadic oesophageal cancer we report here is thus consistent with the hypothesis that the tylosis oesophageal cancer susceptibility gene is also involved in the pathogenesis of a proportion of sporadic squamous cell carcinomas of the oesophagus. Keywords: tylosis; loss of heterozygosity; oesophageal cancer

Introduction Tylosis (hereditary hyperkeratosis palmaris et plantaris) is an autosomal dominant trait characterized by thickening of the skin on the palms and soles that can be associated with a very high risk of squamous cell oesophageal cancer. In a UK pedigree with the cancer-prone syndrome variant, almost all (90 ± 95%) aected members will develop oesophageal cancer by the time they reach the age of 70 (Ellis et al., 1994). We (JMR, JKF) have previously reported our results from haplotyping and recombination analysis of individuals from the `S' tylotic family in Liverpool, which permitted location of the genetic defect to a 6 cM region at 17q23-qter, telomeric to the type II keratin gene cluster (Risk et al., 1994). Further work with this family and with material from one other family from the US has allowed a more exact localization to within a map distance of 1 cM (Kelsell et al., 1996), and identi®cation of the tylosis oesophageal cancer gene (TOC) by positional cloning is underway. Correspondence: JK Field, Molecular Genetics and Oncology Group, Department of Clinical Dental Sciences, University of Liverpool Liverpool L69 3BX, UK Received 11 September 1997; revised 18 May 1998; accepted 18 May 1998

Mapping of an inherited genetic defect in a familial cancer syndrome can lead to discovery of the same genetic alteration in acquired cases. While hereditary oesophageal carcinoma is uncommon, non-familial oesophageal cancer is one of the ten leading causes of cancer mortality world-wide, and accounts for more cancer deaths than any other neoplasm in some populations (Munoz and Castellague 1994; Tomatis 1990). The greater than 50-fold geographical variation in incidence of this cancer (for example, 2 ± 3 cases per year /100 000 males, age-adjusted average incidence in Europe, compared to 80 ± 100 cases per year /100 000 reported in some areas of northern China), has been attributed essentially to environmental risk factor exposure, and not to the in¯uence of genetic background on risk. Allelotyping of sporadic oesophageal cancer has shown that LOH occurs frequently on 17q, as well as on 9p, 17p, 13q, 5q and 18q (Aoki et al., 1994; Mori et al., 1994; Shibagaki et al., 1994). The breast/ovarian cancer gene on 17q could be considered a candidate tumor suppressor gene in oesophageal tumors and other cancer types showing a high LOH on 17q. Nevertheless, mapping of the TOC in tylotic families has shown the locus to be clearly distinct from BRCA1. In this study we wished to make use of our re®nement of the TOC gene localization and the recent development and ordering of new microsatellite markers on 17q in order to learn whether there is support for the proposal that the tumor susceptibility gene causing hereditary oesophageal cancer in tylosis families is involved in the sporadic form of the disease. Normal and tumor DNA of 35 patients with nonfamilial squamous cell carcinoma of the oesophagus were tested for allelic loss using six polymorphic microsatellite markers spanning the TOC region on chromosome 17q. (Risk et al., 1994, Kelsell et al., 1996). Frequent loss of heterozygosity (LOH) at or near this locus in tumors would be regarded as an indication that the TOC locus is a tumor suppressor gene with a role in the development of familial and sporadic squamous cell cancers of the oesophagus. Results Tumor cells and histologically normal mucosa of biopsies from 35 patients with sporadic squamous cell carcinoma of the oesophagus were examined for LOH at six microsatellite loci within a 5 cM region of chromosome 17q25.1-17q25.3 containing the TOC

LOH in sporadic oesophageal tumors in the TOC region M von Brevern et al

2102

gene (Figure 1). Mapping of the gene has been further re®ned to an interstitial 1 cM region on 17q25.2-q25.3 where ®ve of the six markers we used are located; the remaining marker in this study lies proximal to this minimal region. Linkage of this chromosomal location to hereditary oesophageal cancer has been demonstrated in three tylosis syndrome families (Risk et al., 1994, Hennies et al., 1995, Stevens et al., 1996). Initally we analysed 18 cases from Europe (series A) at all six loci by ``classical'' non-automated methodology involving PAGE, and gel staining to visualize DNA fragments and estimate signal intensity. Using a newly available automatic DNA sequencer with DNA fragment sizing applications, we then re-analysed series A samples at three markers in the minimal 1 cM TOC region: D17S1603; D17S1839, and D17S801. There was 85% concordance for all marker data informative by the PAGE gel procedure; however, there were several instances where PAGE gel analysis yielded a `noninformative' (homozygous) case (only one DNA fragment signal was apparent following PCR amplification of DNA from normal cells), whereas, upon microcapillary electrophoresis and laser detection the sample was shown to be informative (heterozygous). The resolution of the latter procedure allows clear distinction of two alleles diering by only a few base pairs. On the basis of this methodology comparison, the second set of 17 cases (series B, China) was analysed exclusively by the automatic procedure, and the results for series A patients generated by the automatic capillary procedure were considered the de®nitive data for markers tested by both procedures. Frequency of LOH A summary of LOH analysis at chromosome 17q25.1-q25.3 in oesophageal cancer are shown Figure 2. Twenty-four of the

the six markers on 35 patients with in Table 1 and 35 tumors (69%)

revealed allelic loss at one or more microsatellite loci within the 5 cM region spanned by the six markers. In 13 of these 24 cases (54%) with LOH, heterozygosity at one or more of the other markers in the set of six we tested was retained. LOH was most frequent (64%) at locus D17S801 (Table 1) within the 1 cM internal minimal TOC region. The order of markers in this area is not yet certain, and they are shown in Figure 1 according to our most current data from physical mapping and haplotyping (Risk et al., unpublished). The LOH percentages given for the 35 oesophageal tumors analysed here and shown in Table 1 may be conservative ®gures because we adhered strictly to our 40%-loss-of-signal criterion for judging a tumor of an informative case to have undergone LOH. Current LOH evaluation by automatic ¯uorescent-tagged DNA fragment analysis de®nes cut-o points re¯ecting allele reduction of 25% to 40% (allelic imbalance factor between 1.3 to 1.7). See Materials and methods. We have chosen the more stringent criterion of 1.7 (approximately 40% loss). Eleven cases with LOH at 17q25.1-q25.3 demonstrated loss with each informative marker we examined; in three of these tumors LOH was observed at all six loci (case 2060, 2062 and 2064, Figure 2), indicating deletions spanning at least 5 cM. There was sucient DNA from ten of the eleven cases (i.e. all except case 155) to test for complete loss of chromosome 17 with four outlying microsatellite markers D17S799, D17S1872, D17S1818 and D17S934. Six of the ten cases showed heterozygosity and retention of both alleles for one or more of these chromosome 17 markers, thus indicating that LOH at the TOC region is not due to loss of the entire chromosome. Microsatellite instability (MI) was observed in only one case (150), determined by the presence of a DNA fragment unique to the tumor sample of the matched normal/ tumor tissue pair (data not shown).

Table 1 LOH in the TOC region in sporadic squamous cell carcinomas of the oesophagus Chromosome order of marker ®ter

MS marker on chromosome 17q

No. of homozygous cases Patient series (uninformative)

No. of heterozygous cases (informative)

Loss of heterozygosity No. LOH/no. informative cases No. of cases (in per cent)

1

D17S1864

A B A+B

10 4 14

8 12 20

4 7 11

50% 58% 55%

2

D17S1603

A B A+B

7 2 9

11 14 25

4 9 13

36% 64% 52%

3

D17S1839

A B A+B

8 2 10

10 15 25

4 11 15

40% 73% 60%

4

D17S801

A B A+B

3 2 5

14 14 28

8 10 18

57% 71% 64%

5

D17S1817

A B A+B

4 4 8

14 13 27

4 7 11

29% 54% 41%

6

D17S785

A B A+B

9 2 11

9 15 24

3 7 10

33% 47% 42%

Series A: Europe; Series B: China; Italics: microsatellite analyses were done by conventional polyacrylamide gels (see Materials and methods)


2103

Figure 1 Location of microsatellite markers used in this study with respect to one another. The composite map of the microsatellite order used in this paper has been based on our physical mapping and haplotype data. In addition the published data on marker order from GeÂneÂthon has been included for comparison. Finally the data from our radiation hybrid mapping has been shown (Risk et al., unpublished). The cytogenetic locations of these microsatellites were obtained from the latest GDB composite map. (a) Composite map. (b) Haplotype data (Risk et al unpublished). (c) Physical mapping data (Risk et al unpublished). (d) GeÂneÂthon map. (e) Radiation hybrid mapping (Risk et al unpublished)

Minimal region A minimal region of loss was observed between D17S1839 and D17S801 in the majority of cases showing LOH, however, only one tumor (2098) de®ned the proximal limit and two tumors de®ned distal limit (2112, 2078). The positioning of the minimal region of loss between two adjacent markers indicates that the 12 (31%) tumors showing no LOH anywhere within this region may still contain a mutation in the same gene (between D17S1839 and D17S801) as those demonstrating LOH. Two tumors did not concur with this minimal region and may show sporadic LOH (tumor 210 at D17S785; tumor 2108 at D17S1817). Thus LOH at D17S1817 to D17S785 most likely represents sporadic loss, however, there is a possibility that there is a second minimal region of loss between D17S1817 ± D17S785, but this is unlikely as there are only two tumors supporting this hypothesis (patient number 210 and 2108) and because the proposed minimal region is adjacent (D17S1839 ± D17S801). Discussion

MI

Figure 2 Pattern of allelic loss in squamous cell oesophageal tumors in the TOC region of chromosome 17q25.1-q25.3. Closed rectangle, loss of heterozygosity with more than 40% loss of one allele; white rectangle, heterozygosity retained; MI, microsatellite instability; ni, non-informative; n.d., no data obtainable. Samples numbers 141 ± 210=Series A (Europe), 2060 ± 2120=Series B (China)

A high frequency of loss of heterozygosity (LOH) in tumors at a speci®c locus is regarded as a potential indicator that a tumor suppressor gene maps to the region. A putative tumor suppressor gene (TOC) important in hereditary oesophageal cancer in tylotic families is located at chromosome 17q25.2-q25.3. The frequent loss of heterozygosity at this chromosomal location in the 35 oesophageal tumors of non-familial cases we observed is consistent with the hypothesis that the same gene may be important in the etiopathology of a proportion of sporadic carcinomas of the oesophagus. While it would seem obvious that such is the case, cancer genotyping studies of the last several years have illustrated that this assumption cannot be made a priori: paradoxically the type of cancer induced by an inherited genetic defect in the germline of aected families does not always predict the type of malignacy that arises when the defect occurs in somatic tissues, nor does the


2104

identi®cation of a high-penetrance, inherited defect leading to a given cancer type imply that the same defect will be found frequently in the etiopathogenesis of sporadic forms of the same disease. Characterization of the genetic alterations typical in acquired human oesophageal cancers and their implications in pathogenesis has progressed considerably in the last decade (Montesano et al., 1996). In squamous cell carcinomas at this anatomical site, ampli®cation/overexpression of the cyclin D1 and EGFR oncogenes is often present, as well as loss of p16, Rb and p53 tumor suppressor function by mutation, gene silencing, and/or allelic loss. Mutation of one p53 gene copy, accompanied by loss of the remaining allele is the most frequently found speci®c genetic change in this cancer (Hollstein et al., 1990, 1996; Montesano et al., 1996). P53 tumor suppressor loss of function can occur in carcinomas with or without a functional Rb tumor suppressor gene. Similarly, we found no correlation between p53 gene lesions (previously examined in these tumors; see von Brevern et al., 1996; Bennett et al., 1991) and LOH at the TOC locus in the 35 cases tested in the present study. This would argue that the p53 gene and the TOC gene have distinct roles in tumor suppression. Archived tumor samples identi®ed by this study to have LOH at or near D17S801 will be cases of choice for examining the retained allele for mutation when the TOC gene has been cloned.

Materials and methods Patient samples and PCR ampli®cation at polymorphic loci on chromosome 17q25.1-q25.3 Thirty-®ve cases of squamous cell oesophageal cancer were investigated in this study: the country of origin of 18

patients (cases 141 ± 210, series A) is in western Europe; the remaining 17 cases (2060 ± 2120) are from a high-incidence area of China. Samples were either frozen fresh biopsies or ethanol-®xed and paran-embedded tissue. Fixed Tissues: normal-appearing mucosa and tissue areas in which cancerous cells predominated were isolated separately. Embedded tissues were serially sectioned and mounted on slides. Every tenth section was stained with haematoxylineosin and tumor areas containing 435% neoplastic cells were marked by the histopathologist for subsequent microdissection of cells from unstained slides under a light dissecting microscope. Tissues from normal uninvolved mucosa were similarly examined to verify normal histology and mucosal cells then scraped from the slides for DNA extraction. Cases for which fresh biopsies could be obtained: tumor and uninvolved mucosa from each patient were snap frozen separately in liquid nitrogen at the time of removal from the patient, and DNA was extracted directly. Tumor diagnosis was veri®ed by removal of approximately a quarter of the tumor biopsy, ®xation, paran embedding, mounting, and staining with haematoxylin and eosin, and examination of histopathology. DNA was prepared by digestion with proteinase K and phenol/chloroform extraction according to standard procedures. After ethanol precipitation, DNA was resuspended in sterile distilled water and used as template for polymerase chain reaction (PCR) ampli®cation of six polymorphic microsatellite loci on chromosome 17q25.1q25.3 spanning a map distance of 5 cM: D17S1864/ AFMc008we1, D17S1603/AFMa135xd5, D17S1839/ AFMb054zf9, D17S801/AFM203xg5, D17S1817/ AFMa312ya5, D17S785 (Dib et al., 1996). The oligonucleotide pairs used as primers for ampli®cation of DNA for gel LOH analysis were from Isogen, Amsterdam, The Netherlands, or from Perkin Elmer, Applied Biosystems International, (Foster City, CA, USA) for PCR material analysed with the ABI Gene Scan System. Reaction volumes were 20 or 25 ml, and contained 50 ng of each primer; MgCl2 (1.5 mM); BSA (200 mg/ml); dNTP (160 mM each); Formamide, 2.5% (Merck, Germany); and 0.5 to 0.6 units Taq DNA polymerase (Boehringer Mannheim,

Figure 3 LOH analysis by microcapillary electrophoresis and data analysis with the ABI310 Genetic Analyzer showing loss of heterozygosity at both loci in this patient. Peaks falling within the allele fragment size range for a given marker are selected. The area under each selected peak is given in the table below the electropherogram. Peak area ratios are calculated from the electropherograms of tumor and normal DNA. A peak area from the tumor electropherogram for a given allele that re¯ects a reduction of 40% or more is de®ned as LOH. (See Materials and methods). This ®gure shows that in tumor tissue from patient B2062 there is retention of the smaller allele (183 bp) and almost complete loss of the second allele (187 bp) at marker D17S785. The same patient has also lost an allele (189 bp) at marker D17S1817 in the tumor sample


Mannheim, Germany) in 16Taq Reaction Buer (Boehringer Mannheim, Germany). Reactions were performed in an MJ Research Minicycler (Biozym Diagnostik, Oldendorf, Germany), programmed as follows: 948C 2 ± 4 min (initial denaturation step), 25 ± 35 cycles of denaturation at 948C/30 s; annealing of primer and polymerization at 54 ± 588C/30 s depending on the marker, with an additional polymerization step at 728C for 1 min for samples to be analysed with the ABI Gene Scan system. Five microliters of completed PCR volume were electrophoresed in a 3.5% Wide-Range agarose gel (BRL, Bethesda, USA) to con®rm that DNA ampli®cation was successful. LOH determination by microsatellite marker analysis Polyacrylamide gel analysis Initally, allelic loss at microsatellites was determined by electrophoresis of 10 ± 14 ml PCR product on 10% nondenaturing polyacrylamide gels (acrylamide/Bis-acrylamide 29 : 1). Electrophoresis was for 7 ± 16 h (depending on marker locus) at 250 Volts; visualization and intensity were recorded by densitometer readings following ethidium bromide staining of gels. Tumors of heterozygous individuals were considered to have sustained allelic loss when a 40% or greater reduction of intensity could be observed for one of the two PCR signals. Automatic DNA fragment capillary electrophoresis, laser detection of ¯uorescence and ¯ourescent signal peak analysis LOH at microsatellite loci was also evaluated by PCR of tumor and normal sample pairs with ¯uorescent dye primers and microcapillary electrophoresis of PCR products in an ABI Prism 310 Genetic Analyzer and data analysis with Genescan software from Perkin Elmer Applied Biosystems International (ABI), Foster City, CA. One microliter of PCR reaction was added to the Template Suppression Reaction Buer (ABI), and three markers,

each tagged with a dierent ¯uorescent dye, were examined stimultaneously per sample injection run. Automatic capillary electrophoresis and data recording proceeded according to manufacturer's instructions. For a heterozygous patient, a 40% or greater reduction in peak area (i.e. 60% of expected area, or less) of one allele in the tumor DNA PCR product, normalized against the retained allele, was considered evidence for allelic loss at that marker, (Figure 3). This corresponds to an allele imbalance factor (IF) of 1.7 IF

=

PAR (T) PAR (N)

PAR (T) =

Peak area of retained allele Peak area of lost allele

PAR (N) =

Peak area corresponding to retained T allele Peak area corresponding to lost T

Allele reduction is calculated by determining PAR(N) and calculating expected peak area in tumor assuming PAR (T)=PAR (N). An observed peak area of the putative lost allele that is 60% or lower than the calculated expected value (had there been no allele loss) is considered evidence of LOH in this study.

Abbreviations LOH, loss of heterozygosity; SCC, squamous cell carcinoma; MS, microsatellite; TOC, tylosis oesophageal cancer gene; PCR, polymerase chain reaction; PAGE, polyacrylamide gel electrophoresis; MSI, microsatellite instability; Rb, retinoblastoma; EGFR, epidermal growth factor receptor.

References Aoki T, Mori T, XiQun D, Nisihira T, Matsubara T and Nakamura Y. (1994). Genes, Chromosomes & Cancer, 10, 177 ± 182. Bennett WP, Hollstein MC, He A, Zhu SM, Reseau JH, Trump BF, Metcalf RA, Welsh JA, Midgley C, Lane DP and Harris C. (1991). Oncogene, 6, 1779 ± 1784. Dib C, Faure S, Fizames C et al. (1996). Nature, 380, 152 ± 154. Ellis A, Field JK, Field EA, Friedman PS, Fryer A, Howard P, Leigh IM, Risk J, Shaw JM and Whittaker J. (1994). Oral Oncol., Eur. J. Cancer, 30B, 102 ± 112. GDB(TM) Human Genetics Database version 6.3 (database online). Baltimore (Maryland, USA): John Hopkins University 1990-. Updated daily. Available from Internet: URL http://www.gdb.org/gdb/gdbtop.html. Hennies H-C, Hagedorn M and Reis A.(1995). Genomics, 29, 537 ± 540. Hollstein MC, Metcalf RA, Welsh JA, Montesano R, and Harris CC. (1990). Proc. Natl. Acad. Sci. USA, 87, 9958 ± 9961. Hollstein MC, Shomer B, Greenblatt M, Soussi T, Hovig E, Montesano R and Harris CC. (1996). Nucleic Acids Res, 24, 141 ± 146. Kelsell DP, Risk JM, Leigh IM, Stevens HP, Ellis A, Hennies HC, Reis A, Weissenbach J, Bishop DT, Spurr NK and Field, JK.(1996). Hum. Mol. Genet., 5, 857 ± 860.

Montesano R, Hollstein M and Hainaut P.(1996). Int. J. Cancer, 69, 225 ± 235. Mori T, Aoki T, Matsubara T, Iida F, XiQun D, Nishihira T, Mori S and Nakamura Y.(1994). Cancer Res., 54, 1638 ± 1640. Munoz N and Castellsague X.(1994). Eur. J. Gastroenterol. Hepatol., 6, 649 ± 655. Risk JM, Field EA, Whittaker J, Fryer A, Ellis A, Shaw JM, Friedmann PS, Bishop DT, Bodmer J, Leigh IM and Field JK. (1994). Nature Genetics, 8, 319 ± 340. Shibagaki I, Shimada Y, Wagata T, Ikenaga M, Imamura M and Ishizaki K.(1994). Cancer Res., 54, 2996 ± 3000. Stevens HP, Kelsell DP, Bryant SP et al. (1996). Arch. Dermatol., 132, 640 ± 651. Tomatis L (editor-in-chief). (1990). Cancer: Causes, Occurrence and Control, IARC Scienti®c Publications No. 100, IARC Press, Lyon. von Brevern M-C, Hollstein MC, Cawley HM, De Benedetti VMG, Bennett WP, Liang L, He A-G, Zhu SM, Tursz T, Janin N and Trivers GE. (1996). Cancer Res., 56, 4917 ± 4921.

2105