Loss of heterozygosity on chromosome 22 in ovarian carcinoma is distal to and is not accompanied by mutations in NF2 at 22q12.

Frequent loss of heterozygosity (LOH) has been reported on 22q in ovarian carcinoma, implying the presence of a tumour-suppressor gene. The neurofibromatosis type 2 gene (NF2) at 22q12 is a plausible candidate. Analysis of 9 of the 17 exons of NF2 by single-strand conformational polymorphism (SSCP) in 67 ovarian carcinomas did not detect any somatic mutations, suggesting that NF2 is not involved in the pathogenesis of ovarian carcinoma. LOH data support this conclusion and that the putative tumour-suppressor gene lies distal to NF2, beyond D22S283.

Cytogenetic and allele loss studies point to the involvement of a chromosome 22 tumour-suppressor gene in a variety of malignancies. Loss of 22q markers has been observed in 11-38% of breast cancers (Sato et al., 1990;Chen et al., 1992;Lindblom et al., 1993), 21-50% of colon carcinomas (Miyaki et al., 1990;Okamoto et al., 1988) and 33% of hepatocellular carcinomas (Takahashi et al., 1993). In ovarian carcinoma, loss of 22q markers has been reported at frequencies of up to 71% (Cliby et al., 1993;Dodson et al., 1993). although others have reported lower frequencies in the range of 25% (Sato et al., 1991;Yang-Feng et al., 1993;Osborne & Leech. 1994). These studies have not defined a minimum region of LOH. However, with the exception of hepatocellular carcinomas, the 22q losses are consistent with deletion of the region containing NF2 on 22ql2 (Rouleau et al., 1993;Trofatter et al., 1993) and this therefore may be the target for the LOH.
Neurofibromatosis type 2 (NF2) is an autosomal dominant syndrome which predisposes to schwannomas, meningiomas, ependymomas and juvenile cataracts. LOH for DNA markers on chromosome 22 has been found in both hereditary and sporadic tumours associated with the NF2 phenotype (Sanson et al., 1993), suggesting that NF2 is acting as a tumour-suppressor gene. NF2 is expressed in a variety of non-central nervous system tissues and therefore could be implicated in tumours arising from these tissues. Although ovarian carcinoma is not part of the NF2 syndrome, somatic mutations in the tumour-suppressor gene NFI, for example, have been reported in tumour types not usually associated with the hereditary syndrome (Li et al., 1992). The strong genetic evidence for the presence of a tumour-suppressor gene on 22q prompted us to undertake a mutation analysis of NF2 in ovarian carcinomas and, if appropriate, a deletion analysis to determine the smallest region of loss.

Materials and methods
Tumour specimens and DNA extraction Tumour and blood samples were obtained from 67 patients undergoing primary surgery for ovarian carcinoma. Thirtyeight of the samples have been descnrbed previously and have been the subject of extensive molecular studies (Foulkes et al., 1993a(Foulkes et al., -c. 1994Allan et al., 1994). The remaining 29 samples were collected from hospitals in and around Southampton. The 67 tumours studied included 46 serous tumours. 11 mucinous. two mixed Muillerian. seven endomet-rioid and one granulosa cell tumour. DNA was extracted from the tumours and blood as described by Foulkes et al. (I993a).

Polymerase chain reactions (PCRs)
PCRs were performed on 10-200 ng of genomic DNA in a reaction volume of 20 ;LI with the inclusion of 1 iLCi of [a-32PJdCTP (Foulkes et al., 1993c) using the published PCR cycle conditions. SSCP anatvsis NF2 exons were amplified with primers within surrounding intronic sequence. Details of these primers, including the regions amplified by each set and the PCR conditions used, have been reported by Twist et al. (1994). Samples were prepared for single-strand conformational polymorphism analysis as described previously (Campbell et al., 1994) and electrophoresis carried out at room temperature in 6% nondenaturing acrylamide gels in the presence of 5% and 10% glycerol.

Results
Nine of the 17 coding exons of NF2 (exons 2, 5, 7-12 and 15) were examined for the presence of somatic mutations. The SSCP analysis was biased toward more 5' exons since the N-terminal region of the merlin protein is more frequently associated with mutations (Deprez et al., 1994), however no mutations were detected in any of the 67 ovarian carcinomas. The sensitivity of SSCP analysis has been shown to be related to the size of the PCR product (Sheffield et al., 1993). In our study, this was 188 bp, and we therefore estimate the SSCP analysis sensitivity to be 70-75%.
In view of the absence of detectable NF2 mutations we undertook a preliminary allelotype analysis of 37 ovarian carcinomas (from the collection described by Foulkes et al., 1993a-c) in order to identify tumours with partial loss of 22q, since previous studies have not enabled an assignment of the gene to any particular region on chromosome 22. All four 22q markers used map distal to NF2, with the most proximal marker, D22S430, located in a region approximately 300kb distal to NF2. The relative genetic and "6 P. ENGLEFIELD et al.  (Matise et al., 1994;Buetow et al., 1994). The inter-locus distances are in centimorgans. The position of the NF2 gene is indicated just proximal to D22S430. cytogenetic positions of these markers are shown in Figure 1. Twenty-three of 32 informative tumours (72%) showed loss of one or more 22q markers. In three of these tumours partial loss of 22q was observed (Figure 1). In tumour 28, complete LOH was observed with D22S274 and D22S282, while only partial LOH was seen with D22S283. The finding was consistently found on repeated testing and therefore might be accounted for by clonal variation. In tumour 72, heterozygosity is clearly maintained with D22S283 and completely lost with the more distal markers. In tumour 91, there was maintenance of heterozygosity with the most proximal marker D22S430 but clear loss with the distal markers D22S283 and D22S274.
LOH data suggest that a tumour-suppressor gene relevant in ovarian, breast and colon carcinomas resides on chromosome 22, but it is unclear at this stage if the same gene is involved in these and other malignancies. NF2, located at 22q12, is a plausible candidate for this gene, but to date somatic mutation analysis has been restricted largely to tumours of the central nervous system (CNS). However, in two reports, mutations were detected in the coding region of NF2 in a total of I of 69 breast carcinomas, 6 of 17 melanomas and 2 of 64 colon carcinomas (Arakawa et al., 1994;Biachi et al., 1994). The significance of the low frequencies of NF2 mutation in the breast and colon carcinomas is unclear. However, in this study of 67 ovarian carcinomas, the absence of NF2 mutations supports the conclusion that NF2 is not involved in the pathogenesis of ovarian carcinoma. Although we have not examined all the NF2 coding exons, studies in NF2related tumours (Rouleau et al., 1993;Deprez et al., 1994;Rubio et al., 1994;Ruttledge et al., 1994;Twist et al., 1994) and non-CNS tumours (Arakawa et al., 1994;Biachi et al., 1994) have shown that mutations are distributed throughout the gene with a bias towards more 5' exons. The distribution of NF2 mutations is unlikely to differ significantly in ovarian carcinomas, and given our large sample size we consider it improbable that NF2 is the 22q ovarian carcinoma tumoursuppressor gene.
In our preliminary allelotype study we demonstrated 72% LOH of 22q markers, consistent with the frequencies reported by Cliby et al. (1993) and Dodson et al. (1993). Three tumours were identified which had retained heterozygosity for markers proximal to NF2 but showed LOH for more distal markers. Although some somatic deletions might be the result of generalised genomic instability unrelated to tumour pathogenesis, the finding of three tumours with consistent losses distal to NF2, coupled with the absence of mutations in the NF2 gene, support the conclusion that the 22q ovarian carcinoma tumour-suppressor gene lies distal to NF2, beyond D22S283. Detailed deletion mapping of 22q in an expanded set of ovarian carcinomas is currently under way to refine the location of this gene. Abbreviado LOH, loss of heterozygosity; NF2, neurofibromatosis type 2; PCR, polymerase chain reaction; CNS, central nervous system. This work was fumded by a grant from the Medical Research Council. Additional support was provided by the Wessex Medical Trust. We thank-Mrs M. Baron and Mrs N. Thomas and the clinicians from hospitals in the Wessex region who provided tumour samples and Dr G. Rouleau for providing the NF2 primer sequences.