Absence of somatic alterations of the EB1 gene adenomatous polyposis coli-associated protein in human sporadic colorectal cancers.

The human EB1 gene product was recently found, by a yeast two-hybrid screening, to be associated with the carboxy terminus of the APC (adenomatous polyposis coli) protein, the product of a tumour-suppressor gene thought to act as a gatekeeper in colorectal carcinogenesis. Because virtually all of the APC mutations result in the synthesis of carboxy-terminal truncated proteins, mutant APC proteins are expected to lose their ability to interact with EB1 gene product. Thus, the interaction between APC and EB1 proteins may be important for the tumour-suppressor activity of APC protein, and raises the hypothesis that EB1 is also involved in sporadic colorectal tumorigenesis. To investigate this hypothesis, somatic mutations in the entire coding sequence of EB1 cDNA were searched by reverse transcriptase single-strand conformational polymorphism (SSCP) analysis in 21 sporadic colorectal cancers and seven adenomas. None of these tumours contained somatic mutation, whereas a silent cDNA variant was identified in 14% of alleles. Furthermore, to investigate whether EB1 locus was included within a region subjected to losses of heterozygosity, four polymorphism markers surrounding EB1 locus were surveyed. Only one out of 28 colorectal tumours contained a loss of heterozygosity at the D20S107 marker. In conclusion, the present findings strongly suggest that EB1 gene is not involved in somatic colorectal carcinogenesis. ImagesFigure 2Figure 3

The adenomatous polposis coli gene (APC) on chromosome 5q2.1 encodes a tumour suppressor w-hich is assumed to act as a gatekeeper in colorectal carcinogenesis . The APC gene is mutated in about 75%c of colon cancer cell lines (Smith et al. 1993) and 60%c of sporadic colorectal cancers (Mlisoshi et al. 1992: Posell et al. 1992: Mis-aki et al. 1994. In these tumours. APC mutations are thouaht to be an earlv event as thev have been found in small benign adenomas (Powell et al. 1992). as well as in putative precursor of colorectal adenomas. i.e. aberrant crypt foci (Smith et al. 1994a). In addition. inherited mutations of APC gene are responsible for familial adenomatous poly-posis. an autosomal dominant disorder that predisposes to earlN deselopment of colorectal cancer (Groden et al. 1991: Kinzler et al. 1991. Both germline and somatic mutations are almost exclusiv elv nonsense or frameshift mutations encoding for truncated APC proteins lacking their carboxy -terminal half (Nagase et al. 1993: De Vries et al. 1996. The APC gene encodes a cvtoplasmic 2843-amino-acid protein which is beliesed to act as a tumour suppressor. blocking the cell cycle progression in G (Baeg, et al. 1995) and precipitating entry into apoptosis of the colorectal epithelial cells (Monn et al. 1996). In addition to these functional features. sesveral biochemical interactions betseen APC gene product and other proteins haxse been demonstrated. As a matter of fact. the APC protein can form stable homodimer swith its amino-terminal domains (Josly-n et al. 1993: Recerved 18 August 1997Revised 17Apnl 1998Accepted 22 April 1998 Correspondence to: B Bressac-de Paillerets Su et al. 1993). and can associate wsith seseral other proteins. includinc P-catenin (Rubinfeld et al. 1993: Su et al. 1993. plak-oalobin (y-catenin) (Shibata et al. 1994). tubulin (Munemitsu et al. 1994: Smith et al. 1994b). gl-cogen svnthase kinase-31 (a mammalian homologue of ZW-3 kinase) (Rubinfeld et al. 19961. hDLG (a homologue of the Drosophila disc large tumoursuppressor gene) (Matsumine et al. 1996) and a human protein named EB I (Su et al. 1995). EBI gene on chromosome 20q 1 1.2 encodes for a nos el 268amino-acid protein which has been found to associate w-ith the carboxyv-terminus of APC protein (codons 2167-2843) through a y east tu-o-hvbrid screening (Su et al. 1995). Recent immunoprecipitation assay have showsn that EB1 product can also associate with -catemnn (Monn et al. 1996). Ahich has been found to participate in the Wg/Wnt cell proliferation pathws ay. EB 1 protein shares little sequence similarity to other proteins except a calcium channel from carp. the bactenral RNA pol merase 6-chain. the Saccharomvces cerevisiae gene product YerO16p (Su et al. 1995) and the closely related gene RPI (Renner et al. 1997). Although its function remains unknown. the recent characterization of a putatise EBI homologue in urochordate marine insertebrates suggests that EB I protein has an important conservatisve cellular function (Pancer et al. 1996). Because virtuallv all of the APC mutations result in carboxv-terminal truncated products. APC mutant proteins are expected to lose their abilits to interact with EBI oene product. Thus. the interaction bets-een APC and EBI gene products may be important for the tumour-suppressor activity of APC protein. and raises the hypothesis that EBI max also be insolved in sporadic colorectal tumorigenesis. In order to insvesticate this hypothesis. A-e analy sed 21 sporadic colorectal cancers and sesen adenomas for EBI point mutations and losses of heterozvgositV.

MATERIALS AND METHODS
Tumour specimens Twenty-one sporadic colorectal adenocarcinomas. seven adenomas. and normal corresponding tissue specimens w ere obtained from 21 patients (11 males and ten females) with mean age 67.5 years (ranae 53-84 years) at the time of surgery at the Institut Gustaxe Roussy. Accordinga to Astler-Coller's staging.
adenocarcinomas w ere classified as A in one case. B 1 in tu o cases. B2 in five cases. C1 in one case. C2 in four cases and D in eiaht cases. Nine adenocarcinomas were located in the right colon. eight were located in the left colon and four were located in the rectum.
Sesen tubular colorectal adenomas containing, severe dysplasia in fise cases and moderate dysplasia in two other cases wvere also anals sed. Five adenomas were located in the right colon. and txo were located in the left colon. The sporadic nature of these tumours was supported by the following criteria: (1) absence of tumour microsatellite instabilitv searched Awith three A-mononucleotide repeat loci (BATRII. BAT26 and BAT40) essentially as described elsewhere (Markowitz et al. 1995: Liu et al. 1996: (2) presence of fewer than fi e colorectal adenomas: (3) and absence of Amsterdam's criteria in the patient's pedigree (Vasen et al. 1991). All the tissues were snap frozen and stored in liquid nitrogen until analysis. Tumour tissues x-ere dissected directly from the surrounding normal areas by light microscopically directed scraping of the specimens.  Figure 1 Schematic locabon of pnmers pairs used for RT-SSCP analysis of the EB1 cDNA coding sequence. The open box represents the coding region of EB1 cDNA, whereas the 5' and 3' untranslated regions are shown as shaded boxes. The location of the different pnmers are indicated in brackets. The amplificabon was successively performed by amplifying the entire coding sequence using the pnmers S1 and AS3, then re-amplifying the PCR product into three overapping segments using the primer pairs Sl-AS1, S2-AS2 or The following primers pairs were designed to amplify the entire EBI coding sequence in three overlapping segments ( Figure  1): S I (sense). 5'-CGAGACGAAGACGGAACC-3'. and AS I (antisense). 5'-AT-l-lGTCAACACCCATTCTCT-3': S2 (sense). 5'-CACGAGTACATCCAGAACT-T-3'. and AS2 (antisense). 5'-AGGGTTCTTTCGCACCACAC-3': S3 (sense). 5'-CCAGAG-GCCCATCTCAAC-3'. and AS3 (antisense). 5'-CCGATGT-TGCTCTGCTGGT-3'. To improve the specificity of the polymerase chain reaction (PCR). the reaction was successively performed by amplifying the entire coding sequence using the SI and AS3 primers. then reamplifying the PCR products with the primer pairs Sl-ASl. S2-AS2 or S3-AS3. The first PCR was performed in 20-gil reaction mixtures containing, 1 gl of cDNA mix. 150 nsi of SI and AS3 primers. 0.2 units of Taq polymerase (Perkins-Cetus). 200 sm each of deoxyribonucleoside triphosphate. 2 m-m of MgCl,. in Taq buffer. After an initial denaturation step of 94°C for 5 min. the PCR was carried out for 35 cycles of 94°C for 30 s. 54°C for 30 s and 72°C for 1 min. and terminated by a final extension at 72°C for 10 min. One microlitre of the 1:1000 diluted PCR products was reamplified as previously in Single-strand conformational polymorphism (SSCP) analysis Conformnational changes in PCR products were analysed by SSCP analysis as previously described (Lazar et al. 1994). Briefly. 5 gil of radiolabelled PCR products was diluted with 20 jl of 0.1% SDS. 10 m- ¶ EDTA pH 8.0 mixed 1:1 with formamide dye (deionized formamide. 0.05% bromophenol blue and 0.05% xylene cyanol). heat denatured at 950C for 2 min. and chilled on ice before loading. Five microlitres of the mixture was loaded to Hydrolink MDE gels (FMC. Rockland. ME. USA) with 8% glycerol for room temperature gels or without glycerol for 4°C gels. Electrophoresis was carried out for 14 and 18 h at 8 W constant power for 4°C and room temperature gels respectively. Gels were transferred to Whatman 3MM paper.
dried on a vacuum-slab dryer and autoradiographed for 12-24 h with an intensifying screen. Base pair changes were identified by shifts in the pattem of single-stranded DNA conformers.
Sequence analysis PCR products displaying abnormal pattern were purified using Microspin columns (Pharmacia. Uppsala. Sweden). and directly sequenced with the deoxytermination method using fluorescently tagged dideoxyribonucleoside (Applied Biosystems. Foster City. CA. USA) on the Applied Biosystems model 373A DNA sequencer. The sequences were compared with the published human EB1 cDNA sequence (Genbank accession number U24166) using the Sequence Navigator package (Applied Biosystems).

Analysis of allelic loss
To determine whether EBI was included within a region undergoing losses of heterozygosity in colorectal cancers. DNA from dissected frozen tumour specimens and normal corresponding tissues were extracted using standard methods (Maniatis et al. 1989) and subjected to PCR amplification of the polymorphic microsatellite markers surrounding EBI at loci D20S112 (AFM197xbl12 ). D20S195 (AFM321xcl). D20S107 (AFMI42xh4) and D20S178 (AFM240vd6) (Chumahov et al. 1995). After amplification for 40 cycles with annealing temperature at 50°C and incorporation of [ax-P]dATP. PCR products were separated on 6% denaturing polyacrylamide gels (acrylamide. N-N'-bisacrylamide: 28:2). and autoradiography was performed.
Allelic loss was scored if the autoradiographic signal was at least 50% reduced when compared with the corresponding normal allele.

RESULTS AND DISCUSSION
To improve the specificity of the reaction. EB I cDNA was amplified in two steps. In the first step. the entire coding, sequence was amplified using the primers SI and AS3 in 28 sporadic colorectal tumour specimens and normal corresponding tissues.
Amplification gave a unique product of 897 bp. indicating the lack of abnormal splicing of EB1 mRNA. In the second step. SSCP analysis of both tumour and normal specimens was performed by reamplifying the previous PCR products with the primer pairs S1-ASI, S12-AS2 or S3-AS3. No change in the electrophoretic mobility was found in colorectal cancer specimens only. indicating the lack of somatic mutation. Seven tumours and normal corresponding specimens contained sequence alterations of the S3-AS3 fragment as shifts in the electrophoretic mobility of singlestranded conformers (Figure 2A). Sequence analysis found a silent polymorphism changing a C to T in the third base of codon 191 (191-F to 191-P) (Figure 2B). No additional difference with the published sequence was found when sequencing the entire coding sequence of EBJ.
To determine whether EBI locus. which maps on chromosome 2Oq 1 1.2 (Su et al. 1995 normal tissues were heterozygoous for all four markers in the 28 analysed tumours. Allelic loss at the locus D2'O 107 on the 20q 1 1.2 region w-as found in only 1 out of the 28 tumours ( Figure   3). This finding is consistent with previous allelotype analysis of colorectal cancers showing the low frequency of 20q chromosome losses (Vogelstein et al. 1989: Thorstensen et al. 1996. Incidentally. microsatellite instability was found in two cancers wvith only one dinucleotide marker. In the present study. we have tested the hypothesis that EBI alterations may occur alternatively to APC mutations in human sporadic colorectal cancers. Although APC mutations were not searched for in the 28 tumours of this study. we probably analysed several tumours containing wild-type APC gene. since no APC mutations are found in about 40%7c of colorectal cancers (Mivoshi et al. 1992: Powell et al. 1992: Mivaki al. 1994. Reaardless of the APC gene status. the results described above provide evidence that EBI is not subjected to point mutations and is not included within a region undergoing losses of heterozvgosity in a significant number of human sporadic colorectal cancers. However. it remains possible that more subtle alterations. such as mutations in the promoter region or abnormal DNA methylation. could inactivate EBI alleles in human colorectal cancers. As a matter of fact. chances in the methvlation status have been show-n to affect various genes involved in sporadic colorectal carcinogenesis. such as APC (Hiltunen et al. 1997). and hMLHI (Kane et al. 1997).
In conclusion. the present data suggests that EBI is not involved in the development of human sporadic colorectal cancers.

Note added in proof
Since the submission of this article. EB 1 product has been shown to localize to microtubules in i-vo. in both fission (Beinhauer et al. 1997) and buddinga yeast (Schwartz et al. 1997). In addition. the function of EB 1 product has been recently clarified in yeast (Muhua et al. 1998). In those cells. mutation of EBI homologrue has been shown to abolish the cell-cycle delay induced by misalignment of the mitotic spindle. These findings suggest that EB 1 may be necessary to maintain neutral ploidy througrh this cellcycle checkpoint mechanism. ABBREVIATIONS APC. adenomatous polyposis coli: PCR. polymerase chain reaction: SSCP. single-strand conformational polymorphism: RT. reverse transcriptase: SDS. sodium dodecvl sulphate.