Ras-related TC21 is activated by mutation in a breast cancer cell line, but infrequently in breast carcinomas in vivo.

Activating ras mutations are found in many types of human tumour. Mutations in Harvey (H-), Kirsten (K-) and neuronal (N-) ras are, however, rarely found in breast carcinomas. TC21 is a ras family member that shares close homology to H-, K- and N-ras, and activating mutations have been found in ovarian carcinoma and leiomyosarcoma cell lines. We have examined panels of cDNAs from breast, ovarian and cervical cell lines, and primary and metastatic breast tumours for mutations in TC21 using a single-strand conformational polymorphism (SSCP)-based assay. One breast cancer cell line, CAL51, exhibited an altered SSCP pattern, compared with normal tissue, which was due to an A-T base change in codon 72, causing a predicted Gln-Leu activating mutation. Of nine primary and 15 metastatic breast tumour cDNAs analysed, none exhibited an altered pattern by SSCP. The apparently wild-type pattern by SSCP analysis was confirmed by sequence analysis of some of the cDNAs assayed. Thus, we conclude that mutations in TC21 are uncommon in breast carcinomas.

Accepted 28 January 1998 Correspondence to: KT Barker oxarian cancers (Ichikaxxa et al. 1994: Cuatrecasas et al. 1997. but generally appear to be a rare ex-ent in ox-arian carcinomas xvan t Veer et al. 1988: Berchuck andCarnev. 1997). The incidence in cerv ical cancer is not as clear. some reports describe a high frequency of ras mutations (Riou et al. 1988: Wong et al. 1995. others haxe found a verx lowx frequencv (Bos. 1988: Willis. 1993. One possible explanation for the absence of ras mutations in breast cancer could be that deregulation of RAS function does not lead to uncontrolled growth of breast epithelial cells. Human phaeochromocytomas do not haxe mutations in ras. and the introduction of oncogenic RAS into PC 12 cells. w-hich are rat phaeochromocytoma cells. leads to differentiation and groxxth arrest (Bar-Sagi andFeramisco. 1985: Noda et al. 1985). This does not appear to be the case for breast epithelial cells. For example. expression of oncogenic H-ras leads to the transformation of the breast epithelial cell line MCF 1OA (Basolo et al. 1991). Elex ated lexvels of H-ras product have been detected in human breast tumours (Watson et al. 1990). Another possibility is that deregulation of RAS pathways xia other mechanisms not inxolxing mutations in ras genes may be important. Alternatixely. other members of the ras family may play more major roles in growth control in breast cancer. and mutations may be found in these genes. TC2 1 is a ras family member cloned from a human teratocarcinoma cell line. It shoxws overall 60%' nucleotide and 55%1c amino acid identity w-ith H-. Kand N-ras. xith a strictly conserxed effector domain . This is the closest homoloax to H-. K-and N-ras genes of any of the ras superfamilx cloned to date. There are sexeral pieces of exidence to suggest that TC2 1 may play a role in human cancer. Codons 22. 23 and 72 in TC21 are analogous to those A-hose in xvixo mutation leads to oncogenic actixation in H-. K-and N-ras (Graham et al. 1994). Ahen codons 22 and 72 wxere experimentally mutated in TC2 1. 22 Glu to Val. and 72 Gln to Leu. the result w as a transforming actixvitx equivalent to that of oncogen c ras (Graham et al. 1994. A form of the 72 Gln to Leu mutation in TC2 1 has been cloned from an ox arian carcinoma cell line (Chan et al. 1994). and the transforming capacity and tumongenicity of this mutation in NIH/3T3 cells was confirmed. An insertional mutation of 9 bp at codon 24 in TC21 has been identified in a human leiomyosarcoma cell line. and this displays high transforming activity in NlH/3T3 cells (Huang et al, 1995).
Whereas most experiments on transformation by mutant TC21 have been performed in murine fibroblast cells, mutants 22 Glu to Val and 72 Gln to Leu have been shown to transform MCF lOA.
spontaneously immortalized human mammary epithelial cells (Soule et al. 1990). causing altered cell morphology and allowing colony formation in soft agar (Clark et al, 1996). It is therefore possible that aberrant TC21 signalling may be involved in breast tumour progression. To examine the possibility that TC2 1 may be mutated in breast cancer, we have developed a single-strand conformational polymorphism (SSCP)-based assay to look for equivalent mutations in breast cancer cell lines and primary and metastatic breast carcinomas. In view of the low incidence of Kand H-ras mutations reported in ovarian tumours, and the identification of a TC2 1-activating mutation in an ovarian cancer cell line.
A2780. we surveyed a panel of ovarian cancer cell lines. ras mutations may play an important role in the progression of cervical cancer. but the reported frequency of mutation varies between studies. As altemative ras gene family members may be mutated in those not harbouring Kor H-ras mutations. we examined a panel of cervical carcinomas and cell lines for activating TC21 mutations.

MATERIALS AND METHODS SSCP
RNA was isolated using the RNAgents® Total RNA Isolation System (Promega). by a modification of the method of Wilkinson ( 1988) or using guanidinium isothiocyanate (Chirgwin et al. 1979). cDNA was prepared as described previously (Barker et al. 1995).
Oligonucleotides primers 5'TACCGGCTCGTGGTGGTCGG3' (sense) and 5'TATCTGTGACTGAAAAGACC3' (antisense) were designed to amplify a 249-bp region of TC21 from nucleotides 40 to 289, spanning all three potential activating mutation sites Amplifications were carried out using cDNA derived from 0.2 jg of orginal RNA in 60 mM KCI 15 mm Tris-HCI pH 8.8. 1.75 mM MgCl,. 200 of each dNTlP 3 jCi [a-'2PJdCIT. 20 pmol of each pnmer and 1 unit of Taq polymerase in 25 j. Each reaction was cycled 30 times at 94°C for 1 min, 550C for 2 min and 720C for 1 min. An aliquot of 4 of each reaction was mixed with 4 of SSCP gel loading buffer (95% deionized fonnamide, 0.025 M EDTA pH 8.0.0.005% bromophenol blue, 0.005% xylene cyanol). heated at 95°C for 5 min. and chilled on ice for 1 min. Undenatuned samples were kept on ice. An aliquot of 4 of eidter denatured or undenatured samples was loaded onto 6% acrylamnide, x TBE, 10% glycerol gels. Fragments were resolved by electophoresis at 5 W for 15 h; gels were dried and exposed to X-ray film at -70°C ovemnight Subcloning and sequencing PCR reactions were carried out as for SSCP but without [3n-P] dCTP. Amplified products were resolved on 1.5% agarose TAE gels, excised and purified using a QLAEX H Gel Extraction Kit (Qiagen). Purified products were subcloned into the pGEMI-T Easy vector (Promega) according to the manufacturer's instructions. Sequences were determined by the dideoxy method of Direct PCR sequencing PCR amplifications were carried out either as above for subcloning. or with cloned Pfu DNA polymerase (Stratagene).
using the manufacturer's buffer and 200 gM of each dNTP.
Amplified products were purified using a QIAEX II Gel Extraction Kit. Sequence reactions using a Sequenase Version 2.0 Kit (USB) were performed on approximately 30 ng of purified product and an excess of primer (1 ig 10 jl ').

SSCP analysis of breast cancer cell lines
The genomic structure of TC21 has not yet been reported. therefore mutation analysis was performed on cDNA. Primers that bracket the region containing all three potential activating mutation sites were used to amplify a single band of the predicted size by PCR. SSCP gel running conditions that detect an altered mobility caused by a single base pair change were assessed using cDNA from normal mammary tissue. and from the ovarian cell line A2780. which has an A-T transversion in codon 72 leading to British Journal of Cancer (1998) 78(3) -468 (not shown), SKBr3. BT474, MCF 7. ZR75.1 and T-47D all exhibited a wild-type pattern on SSCP analysis. SSCP analysis of cDNA from MCF 1OA cells revealed a wild-type pattern of bands. consistent with the observations that these cells when transfected with mutant TC21 exhibit altered morphology and enhanced growth in soft agar. One cell line. CAL51 (Gioanni et al. 1990). revealed a shift in mobility of the upper band characteristic of the pattern seen with A2780 cells. The bands marked with arrows in Figure lB represent the two complementary strands. Extra bands are sometimes seen (marked with an asterisk): these are irreproducible and are not indicative of sequence changes.

TC21 is mutated in CAL51 cells
To investigate the cause of the shift in mobility seen in the SSCP analysis of CAL51 cell line cDNA. the sequence of the derived PCR product was determined. An A to T base change in codon 72. causing a predicted Gln-Leu mutation identical to that seen in the cell line A2780. was seen. Figure 2 shows the sequence of CAL5 1 in this region compared with the wild-type TC21 sequence from normal tissue. SSCP analysis of A2780 failed to detect a wild-type allele; this is consistent with the observations of Huang et al (1995). A wild-type allele was not observed in the SSCP analysis of CAL51 cells also. and sequencing revealed only the mutant sequence at this codon. Sequencing of the PCR products of other breast cancer cell lines confirmed the SSCP results. in that only wild-type sequences were present (data not shown). SSCP fails to detect TC21 mutations in ovarian or cervical cancer cell lines Twelve ovarian and eight cervical cancer cell lines were examined for TC21 mutation by SSCP. Amplified products were detected using cDNA from eight of the ovarian and seven of the cervical cell lines. and in all cases the SSCP analysis gave a wild-type pattern. When two of the cell line PCR products were sequenced. only wild-type sequences were observed. Five normal mammary tissue-derived cDNAs were subjected to SSCP analysis and all gave wild-type TC21 pattems. Each of nine primary breast cancers that gave an amplified TC2 1 product gave a wild-type pattem. Of 15 metastatic breast tunmour firom lymph nodes. 13 produced a wild-type TC21 pattem. Figure 3 shows a representative sample. whereas one product produced an apparently altered pattern. This cDNA and one obviously normal cDNA were amplified by PCR and sequenced. Sequencing of the product from one node revealed some silent changes with respect to the published cDNA sequence. but none resulting in amino acid changes.

DISCUSSION
The low frequency of H-. K-and N-ras mutations in breast tumours has led us to undertake this study to examine the possibility that the ras superfamily member TC2 1 is mutated in breast cancers. None of the nine primary or 15 metastatic breast carcinoma samples that we analysed were mutant for TC21. We conclude that TC2 1 mutations are an infrequent event in the development of breast tumours. Of the panel of breast. ovarian and cervical cancer cell lines analysed, only one. the breast carcinoma cell line CAL5 1, had a mutation in TC2 1. CALS 1 cells were isolated from a malignant pleural effusion of a woman with a metastatic breast adenocarcinoma. The cells exhibit morphological. structural and immunohistochemical characteristics of mammary epithelial cells. They will form colonies in soft agar and are tumorigenic in nude mice (Gioanni et al. 1990). Unlike the majority of established breast carcinoma cell lines they have an apparently normal karyotype (Gioanni et al. 1990;S Birdsall. personal communication). The genetic alterations responsible for the neoplastic phenotype of CAL5 1 are not known (they lack mutations in the p53 gene. Theile et al. 1994). To our knowledge. this British Joumal of Cancer (1998) 78(3), 296-300 CAL-51 A C G T A C G T c A A G I mutation in TC2 1 is the onl-genetic aberration reported for CAL5 1 cells. Ahen this mutant form (72. Gln-Leu) was transfected into MICF1OA cells. it conferred the abilitv to form colonies in soft a2ar. but not tumorigenicitv in nude mice (Clar-k et al. 1996). It is possible that the ability of CAL5 1 cells to form colonies in soft agar is at least partially due to this mutation in TC2 1.
Mutations in ras genes do not appear to be a common event in breast. cervical or ovarian tumours (with the possible exception of mucinous ovarian tumours. see Introduction). Although it could be hypothesized that cell type-specific differences in the likelihood of acquiring co-operating mutations account for the apparent tumour type bias of ras mutations. the deregulation of RAS pathways in the development of tumours such as those of the breast should not be discounted. Activated RAS is able to transform breast epithelial cells (Basolo et al. 1991). and chemicallv induced rat mammary tumours harbour ras mutations (Sukumar. 1990). Human breast epithelial cells transformed by carcinogens acquire activatingy ras mutations (Zhang et al. 1994). In addition. several studies in transgenic mice have found that activated and/or overexpressed ras alleles can predispose to the development of mammary tumours (Andres et al. 1987: Sinn et al. 1987: Mangyues et al. 1992. In view of the above findingys. there is a possibility that other components of RAS signalling pathways are altered in breast. cervical and ovarian tumours. thus having the equi-alent effect without an activating ras mutation. Oncogyenic RAS is less sensitive to GAPs leading to an accumulation of active GTPbound RAS: a reduction in GAP activity could have the same effect as mutated RAS. The two best characterized mammalian RAS GAPs are p'2O GAP and NFl GAP. Neurofibromatosis type I (NF ) is an autosomal dominant condition that predisposes to certain types of carcinoma. There is evidence to suggest that NFl may function as a tumour suppressor (Legius et al. 1993): this may be due to decreased RAS-GAP activitv. Tumours and cell lines derived from tumours from NFl patients have decreased NE1 GAP activitv and increased levels of RAS-GTP (Basu et al. 1992: DeClue et al. 1992: Bollag et al. 1996: Guha et al. 1996. NFl mutations have also been found to occur in spontaneously arising tumours (Li et al. 1992: Andersen et al. 1993). However. some cell lines established from tumours occumn2 in non-NFl patients have reduced levels of NFl but not elevated RAS-GTP levels (Johnson et al. 1993). It may be that loss of NT1 function(s) independent from GAP activitv play a role in the phenotype of these cells. It is hypothesized that GAPs may function as downstream effectors of RAS in addition to their regulatory activity (Clark and Der. 1995).
There have been cases of breast cancers of an aggressive nature reported in NFl patients (Teh et al. 1997). It may therefore prove informative to make a more detailed analyNsis of NF l or p12O GAP expression and sequence in breast and other tumours.