Over-expression of the c-myc proto-oncogene in colorectal carcinoma.

Alterations in the c-myc proto-oncogene in colorectal cancer were studied at the level of RNA expression, gene amplification and rearrangements. One hundred cases of colorectal cancer, stratified by Dukes' stage were examined. The level of messenger RNA expression was measured in tumours and matched normal mucosa from the same patient. Between 5 and 400 fold over-expression was found in 66% of tumours. Neither the presence nor the level of over-expression correlated with tumour staging. A significant correlation (P < 0.01) was found between over-expression of c-myc in tumours and the presence of synchronous adenomas elsewhere in the colon. In contrast to other tumours, no rearrangements of the gene were found on Southern analysis of colorectal cancers. Similarly, amplification of the gene was not found in the cancers examined. ImagesFigure 1Figure 2Figure 3Figure 4

Since the establishment of the National Cancer Registry in 1968, Singapore has seen a steady increase in the incidence of colorectal cancer, with a standardised rate of 19.9 and 15.7 per 100,000 for males and females for the period [1968][1969][1970][1971][1972] compared with rates of 31 and 26.3 for males and females for 1983-1987, giving an increase in incidence of 55% and 67% respectively. In terms of number of cases, there has been an average annual increase of 3.5% since 1968. Colorectal cancer was the sixth most common cancer at the start of the Cancer Registry; it ranks second today, and will be the most common cancer in Singapore by the end of this decade.
Being a small country with a dense population of 3 million, Singapore provides an ideal opportunity for studying the carcinogenic process of colorectal cancer, particularly, because it has three different races which manifest different risks. Among Singaporeans the incidence of colorectal cancer is highest in the Chinese population (which comprises 75% of the total). The Chinese also have the highest rate of increase. As most of the Chinese population originated from the Southern coastal provinces of China, any study here will provide an opportunity for comparing the aetiology of colorectal cancer for Chinese populations in China, Hawaii and California.
The role of c-myc in colorectal carcinomas is not well understood. Immunohistochemistry has shown that c-myc gene product in normal colonic tissue is located in the mid zone of the colonic crypts, which corresponds to the zone of maturation and differentiation of colonic epithelial cells (Stewart et al., 1986;Melhem et al., 1992). In  localisation extends into the proliferative zone while in colorectal carcinoma c-myc staining can be found in the mature zone as well as the maturation and proliferative zones of colonic crypts (Stewart et al., 1986;Melhem et al., 1992). Over-expression of the c-myc mRNA has been reported to occur in between 60%-80% (Finley et al., 1989;Sikora et al., 1987;Rothberg et al., 1985;Calabretta et al., 1985;Erisman et al., 1985;Tsuboi et al., 1987;Imaseki et al., 1989) of colon carcinomas, although the number of samples in these studies is rather small, from a minimum of six tumours (Calabretta et al., 1985) to a maximum of 38 Rothberg et al., 1985). One study by Rothberg et al. (Rothberg et al., 1985) reports a correlation between overexpression of c-myc and the location of the tumour, and although statistically significant, is based on a relatively small sample size of 38 tumours. However this result has not been supported by other workers (Imaseki et al., 1989), but again this is based on a small sample size (11 tumours). As yet no correlation has been found between over-expression of the c-myc proto-oncogene and either patient survival or disease recurrence (Erisman et al., 1988), or metastatic potential (Tsuboi et al., 1987).
Amplification of the c-myc oncogene has been reported in fresh colonic tumours, although the incidence is low varying from 6%, (2/32; Yokota et al., 1986 and 3/45;Meltzer et al., 1987), to 22% (2/9; Alexander et al., 1986). A better correlation is found when only aggressive subtypes of colorectal tumours (such as mucinous or poorly differentiated tumours) were examined. In these cases slight amplification of the c-myc gene is found in approximately 50% of cases (Heerdt et al., 1991). None of these papers report any rearrangement of the c-myc gene in colorectal carcinomas.
The present study was undertaken with two main points in mind. Firstly we wished to examine the type of oncogenic changes occurring in an Asian population which is showing a rapid increase in incidence of colorectal cancer. Secondly we wished to determine, using a larger sample base, accurate correlations between c-myc and various clinical correlates such as Dukes' stage, age, sex, and tumour site.

Tumour specimens
Samples used in this study were from patients admitted to the Department of Colorectal Surgery at Singapore General Hospital. No initial chemotherapy, radiotherapy or hormonal therapy was given prior to tumour excision. A portion of the surgically removed tumour was snap frozen in liquid nitrogen and stored at -80°C until required. The remainder of the tumour sample was sent for histopathological diagnosis. Control mucosa (sited at least 10 cm proximal to the site of the tumour) was also removed and similarly treated.
Isolation of RNA and Northern blotting RNA was extracted from tumour and mucosa samples by the method of Chomcznski and Sacchi (Chomcznski & Sacchi, 1987), followed by caesium chloride centrifugation (Sambrook et al., 1989) and quantitated by UV spectrophotometry. Total RNA was fractionated through formaldehyde-agarose gels, transferred to a solid matrix (Hybond-N, Amersham, Arlington Heights, IL), and hybridised to 32P random primed labelled (Feinberg & Vogelstein, 1983;1984) cDNA probes. The following double-stranded probes were employed: c-myc cDNA (pGl-5'-c-myc; American Type Culture Collection, Rockville, MD); P-actin cDNA clone (Clontech Laboratories, Palo Alto, CA). After hybridisation filters were exposed to Fuji-RX medical X-ray film (Japan) for between 2-5 days. Signal was quantitated on a CS-9000 scanning densitometer (Shimadzu, Japan). Corrected myc signal in the tumour was compared to the corrected myc signal in the mucosa by comparison with the 13-actin control signal to obtain a number representing the level of overexpression in the tumour. Reproducibility was assessed by 10% of the samples being analysed on separate Northern filters. Reproducibility for all samples was found to be ± 30%, with the majority being ± 20%. The greatest variation was found in one sample from which RNA was extracted from two separate portions of the tumour and probably reflects differing amounts of stromal cell contamination.
Isolation of DNA, Southern hybridisation and DNA dot blot analysis DNA was isolated by standard methods (Davis et al., 1986). For dot blot analysis approximately 10 gg of genomic DNA was transferred to a Nylon filter (Hybond-N, Amersham), denatured by soaking in 1.5 M NaCl, 0.5 M NaOH and neutralised by soaking in 2 M NaCl, 0.5 M Tris-HCl pH 6.0. After drying, DNA was cross-linked by UV radiation and filter hybridised overnight to random prime (Feinberg & Vogelstein, 1983;1984) labelled cDNA probes. Probes used were c-myc exon 1 cDNA (pGl-5'-c-myc; American Type Culture Collection, Rockville, MD); carboxypeptidase H (Manser et al., 1990); c-Ki-ras exon 1 (Barbacid, 1987) PCR (Saiki et al., 1988;Mullis & Faloona, 1987;Saiki et al., 1985) product (primers from Clontech Laboratories, Palo Alto, CA) and P-actin (Clontech Laboratories, Palo Alto, CA). After each exposure the filter was stripped by boiling in 0.1% SDS. For Southern analysis 10 tig of normal mucosa and tumour DNA was digested with EcoRl restriction endonuclease (New England Biolabs) and subjected to electrophoresis on a 0.8% agarose gel. After electrophoresis, DNA was denatured and neutralised as above and transferred to solid matrix (Hybond-N, Amersham) by overnight capillary action. The filter was hybridised overnight with c-myc cDNA probe (pGl-5'-c-myc; American type Culture Collection, Rockville, MD), labelled by the random prime method (Feinberg & Vogelstein, 1983;1984) with 32P, and autoradiography performed.

Results
Over-expression of c-myc RNA The level of c-myc messenger RNA in 100 colorectal tumours, equally divided by Dukes' stage (25 Dukes' A, 25 Dukes' B, 25 Dukes' C and 25 Dukes' D) was measured and compared to levels found in normal mucosa from the same patient. The tumours came from 55 male and 45 female patients, average age 62 years (range 24-89 years). There were 92 Chinese patients and eight others. In each case, 10 fig of total RNA from both the tumour and from matched normal mucosa was fractionated through formaldehyde-agarose gels and hybridised initially with c-myc cDNA. To compensate for variations in the amount of RNA loaded in each lane a second hybridisation with ,-actin was undertaken. Levels of c-myc were then quantitated against levels of P-actin. In all 66 of the tumours were shown to over-express c-myc (see Figure 1). Only tumours showing a greater than 3-fold increase in c-myc levels were considered to be over-expressing. Thirty-four per cent of the tumours showed no overexpression. Low levels of over-expression (3-10-fold increase in levels of c-myc RNA) was found in 20%, moderate levels of c-myc over-expression (11-30-fold increase) was found in 29%, and high levels of RNA over-expression (>30-fold increase) was found in 17% (see Table I) of tumours. No correlation was found between the presence of c-myc overexpression and the stage of the tumour (see Table II); the level of c-myc over-expression and Dukes' staging (see Table   III); the presence of c-myc over-expression and age (Table  IV) (Table V).
Furthermore in contrast to other workers , no correlation was found between the site of the tumour, i.e. left or right side tumours (where left side tumours are those of the rectum, sigmoid colon, descending colon and splenic flexure and right side tumours are those of the caecum, hepatic flexure, ascending colon and transverse colon) and c-myc over-expression (see Table VI). c-myc overexpression did however correlate with the presence of synchronous adenomas (Table VII). Of the 66 patients where over-expression was found in the tumour, 22 possessed synchronous polyps, while only two patients out of 34 not having c-myc over-expression also possessed synchronous polyps. Hence there is a significant correlation (P> 0.01, analysed by x2 test) between tumours over-expressing c-myc and the presence of synchronous polyps. Five patients had synchronous cancers and tumours from four out of the five      expressing the tumour and 10 fig of DNA from matched normal mucosa of the same patient was digested with the restriction endonuclease Eco RI, the digest products were separated on a 0.8% agarose gel and then transferred to solid nylon matrix support prior to hybridisation with a c-myc cDNA probe. Results shown in Figure 3 show only the expected single band at 12.5 kb (Taub et al., 1982;Dalla Favera et al., 1983) which is of equal intensity between the tumour sample and the normal mucosa. No amplification is therefore present. Furthermore no rearrangement of the c-myc gene was found at this level of resolution in any of the ten samples.

Gene rearrangements
The Southern analysis in Figure 3 shows that of the ten samples analysed no detectable rearrangements were found. For this reason a larger number of samples was examined. A further 32 samples were digested with the restriction endonuclease EcoRl (in this case only DNA from the tumour was examined) and the DNA separated by agarose gel electrophoresis. The final sample composition was 11 Dukes' A tumours, 13 Dukes' B tumours, five Dukes' C tumours and 13 Dukes' D tumours. After Southern transfer, rearrangements of the c-myc proto-oncogene were analysed by hybridisation with a c-myc cDNA probe. Representative results are shown in Figure 4. As can be seen at this level of resolution no gross rearrangement was found in any of the samples analysed.
Discussion patients had increased levels of c-myc in the main tumour analysed here. One of these patients with synchronous cancers also had synchronous polyps and expressed c-myc at moderate levels in the main tumour (9-fold over-expression).

Gene amplification
A total of 50 colon carcinomas were examined for amplification of the c-myc cellular proto-oncogene by dot blot hybridisation. Of the 50 samples, 12 were Dukes' A, 17 were Dukes' stage B, eight were Dukes' stage C and 13 were Dukes' stage D. No tumour was found to contain amplified c-myc, as compared against hybridisation of P-actin, c-Ki-ras exon 1 and carboxypeptidase H (Manser et al., 1990), a gene known to be present as a single copy in the human genome (DRS, unpublished data), see Figure 2 for representative dot blots. In confirmation of the dot blot analysis, no amplification could be discerned in ten samples analysed by genomic Southern blotting. In this case, 10 ytg of DNA from The c-myc proto-oncogene is the cellular homologue of the v-myc oncogene of avian myelocytomatosis virus (Venstrom et al., 1982;Dalla Favera, 1982b;Watt et al., 1983a,b) and is a member of the myc family of oncogenes, which contains five other members besides c-myc; namely N-myc, Lmyc, Rmyc, Pmyc and Bmyc (De Pinho et al., 1987;. The c-myc proto-oncogene is present as a single copy gene in the normal human genome and has been localised to chromosome 8 (specifically at 8q24) (Taub et al., 1982;Dalla Favera et al., 1982a;Neel et al., 1982), and consists of one non-coding exon and two coding exons separated by two introns (Hamelyn & Rabbitts, 1983;Watt et al., 1983a,b;De Pinho et al., 1987). We have examined the expression of c-myc in colorectal tumours. We have found that 66% of colorectal tumours show some degree of over-expression of c-myc RNA. The samples analysed were representative of the stages of colorectal tumour as determined by histopathologic analysis. The tumour samples analysed consisted of 25 Dukes' A tumours,  Table II (although in agreement with other workers (Finley et al., 1989) there is perhaps a slight reduction of the percentage of late state tumours over-expressing c-myc i.e. Dukes' stage C and D tumours, although as with Finley et al., this reduction is not statistically significant), or with the age or sex of the donor (Table IV and V respectively). The degree of overexpression ranges from 5-fold to in excess of 400-fold overexpression, but again this does not correlate with the stage of the tumour (Table III). Rothberg et al. (1985), found a significant correlation between the site of the tumour and the over-expression of c-myc in a study on 38 colorectal tumours. In their study they find that 81% of left side tumours (those of the rectum, splenic fixture, sigmoid colon and descending colon) overexpress c-myc, whereas only 36% of right side tumours (those of the caecum, hepatic flexure, ascending colon and transverse colon) show elevated levels of c-myc expression (or alternatively 85% of elevated c-myc expression is found in left side tumours, whereas only 15% of elevated expression is found in right side tumours). In our study, however, we find that an almost identical proportion of left and right side tumours over-express c-myc, i.e. 65% and 60% respectively, although it should be noted that 80% of the tumours in our sample cohort are left side tumours, whereas only 15% of the tumours are right side tumours (five tumours are not included in this analysis as the donors had multiple colorectal tumours). Nevertheless our sample size is more than 2.5 times the size of that used in the study by Rothberg et al.
A significant correlation was found in our study between c-myc being over-expressed in the tumour and the presence of synchronous polyps elsewhere in the colon (Table VII). Synchronous polyps were found in 24 of the patients. Twenty-two of these cases occurred in patients with c-myc over-expression in the tumour, whereas only two of the cases of synchronous polyps were found in those patients whose tumours did not over-express c-myc. This correlation is statistically significant (P<0.01). Furthermore the fact that 5% of the patients in this study showed synchronous cancers and that 80% of these patients showed c-myc over-expression in the main tumour, might indicate that c-myc changes are more widespread throughout the colon than has been found with other oncogenes and tumour suppressor genes such as c-Ki-ras and p53, whose changes tend to be localised to the site of the tumour. It is possible to speculate therefore that a wide spread colonic alteration could be genetically predetermined. This supposition can be partially supported by the wide range of basal levels of c-myc messenger RNA noted by ourselves (data not shown) and others (Finley et al., 1989). Amplification and rearrangement of the c-myc cellular proto-oncogene has been shown to be associated with several different malignancies (Yokota et al., 1986;Little et al., 1983;Ocadiz et al., 1987;Asker et al., 1988;Alitalo et al., 1983;Kozbar & Croce, 1984;Collins & Groudine, 1982;McCarthy et al., 1984;Dalla Favera et al., 1982c;Lu et al., 1988;Rothberg et al., 1984;Nakasato et al., 1984;Heerdt et al., 1991). However, we have examined 50 colorectal tumours of different stages and find no evidence of gene amplification. Furthermore, examination of some 42 tumours by genomic Southern analysis shows no evidence of gene rearrangements. The lack of amplification or rearrangement of the c-myc gene in colorectal carcinomas clearly indicate that a different mechanism of activation is occurring in these tumours as opposed to tumours of the lymphatic system such as Burkitt's lymphoma, and tumours derived from uterine cervix (Ocadiz et al., 1987), esophageal cancers (Lu et al., 1988), hematopoietic malignancies (Rothberg et al., 1984) and stomach cancers (Nakasato et al., 1984), where amplification and/or rearrangement have been shown to be correlated with tumorigenesis in primary biopsy samples.
Hence, in colorectal cancers as activation of c-myc is not a result of either amplification or rearrangement then activation could result from either point mutations in the c-myc gene, either in the promoter region or within the first exon as has been shown for some other malignancies, or possibly by the activation or deactivation of a trans-activating factor. This latter possibility is supported by studies that show that c-myc over-expression is correlated with loss of chromosome 5 alleles (Erisman et al., 1989). Furthermore the introduction of chromosome 5 by microcell fusion into colon carcinoma cell lines leads to the suppression of c-myc deregulation (Rodriguez-Alfageme et al., 1992), possibly by the reintroduction of a functional APC gene, a gene that is implicated in the genesis of spontaneous colorectal cancers and highly implicated in the familial adenomatous polyposis syndrome, an inherited susceptibility to colon cancer and which is known to reside on chromosome 5 Joslyn et al., 1991).