(2003) 22, 3506–3510 & 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc

MBD1, MBD2 and CGBP at 18q21 are infrequently mutated in human colon and lung

Scott Bader*,1, Marion Walker1, Heather A McQueen2, Robert Sellar1, Elizabeth Oei1, Suzan Wopereis1,3, Yanhua Zhu1, Audrey Peter1, Adrian P Bird2 and David J Harrison1

1Sir Alastair Currie Research UK Laboratories, Division of Pathology Molecular Medicine Centre, University of Edinburgh, Edinburgh EH42XU, UK ; 2Wellcome Trust Centre for Cell Biology, University of Edinburgh, Edinburgh EH9 3JR, UK; 3Instituut Life Science Technology, 7500KB Enschede, The Netherlands

The genes MBD1 and MBD2 encode methyl-CpG binding previously from this region, DCC (Fearon et al., 1990), that suppress from methylated DPC4/SMAD4 (Hahn et al., 1996) and JV18-1/SMAD2 promoters. In contrast, CGBP encodes a that (Riggins et al., 1996); however, frequencies of binds promoters containing unmethylated CpG and these genes in colorectal and lung carcinomas reported stimulates transcription. All three are located on human by many groups are lower than the observed LOH and chromosome 18q21, a region of frequent loss of hetero- their roles in colorectal tumorigenesis remain contro- zygosity in several cancers. These genes therefore versial (Fazeli et al., 1997; Fabre et al., 1999). MBD1 represent candidate tumour suppressor genes, whose loss and MBD2 were found to lie on human chromosome of function could affect the normal regulation of 18q21 by PCR and fluorescent in situ hybridization expression, whether by lack of complete suppression of (FISH) when originally cloned (Hendrich et al., 1999a). genes normally silenced (via loss of MBD1 and MBD2)or CGBP was colocalized following the discovery of its 30 by some loss of activation of genes normally expressed exon sequences within a PAC containing the (via loss of CGBP), either way contributing to the region and the CpG island of MBD1 (Cross et al., 1999). tumorigenic . We have confirmed by fluorescent of CpG dinucleotides is an important in situ hybridization that MBD1 and MBD2 bracket the epigenetic mechanism used by vertebrate cells to repress DCC giving a gene order of MBD1/CGBP–DCC 50- transcription of many tissue-specific genes (Bird, 2002). DCC 30-MBD2. Mutation analyses by single-stranded MBD1 and MBD2 are members of the MBD family that conformation polymorphism in colon and lung cancer cell act as intermediaries in this process, binding to methyl- lines and primary tumours revealed a small number of CpG and actively recruiting other protein complexes , suggesting only a limited role of these genes in that alter chromatin conformation (via chromatin human tumorigenesis. remodelling and histone deacetylation) to interfere with Oncogene (2003) 22, 3506–3510. doi:10.1038/sj.onc.1206574 the transcription process (Jones et al., 1998; Nan et al., 1998; Ng et al., 1999). One group has also shown that Keywords: 18q21 gene mutations; colon cancer; lung MBD2 is capable of actively demethylating DNA cancer (Bhattacharya et al., 1999), and its activity as transcrip- tional repressor or demethylase may depend on the context of the promoter with which it interacts (Detich et al., 2002). The demethylase activity of MBD2 remains The genes MBD1, MBD2 and CGBP are closely linked controversial (Ng et al., 1999; Wade et al., 1999; Boeke on chromosome 18q21. This is a region of the genome et al., 2000), but if true would activate gene transcription known to suffer loss of heterozygosity (LOH) in a by removing the repressive methyl-C signals. significant proportion of cancers, including about 75% Localized concentrations of CpG dinucleotides, colorectal carcinomas (CRC) (Vogelstein et al., 1988, termed CpG islands, are mainly kept unmethylated 1989; Thiagalingam et al., 1996) for which there are also and are commonly associated with promoters of house- reports of prognostic significance of 18q21 LOH (Jen keeping and constitutively expressed genes. How CpG et al., 1994; Lanza et al., 1998; Martinez-Lopez et al., islands are kept unmethylated is not certain, but one 1998), and 32% small cell (SCLC) and 56–65% mechanism may involve binding of a protein to nonsmall cell lung carcinomas (NSCLC) (Shiseki et al., unmethylated CpG that prevents DNA methylation 1994; Takai et al., 1998; Virmani et al., 1998). Three and its biological consequences. CGBP (CpG-binding candidate tumour suppressor genes had been cloned protein) may act in this way. CGBP has homology to MBD1 by having one CXXC domain, a domain that is highly conserved among several proteins including also *Correspondence: S Bader; E-mail: [email protected] Received 8 January 2003; revised 19 February 2003; accepted 7 March DNA methyltransferase 1 (DNMT1) (Bestor and 2003 Verdine, 1994), human trithorax (HRX) (also known Mutations of MBD1, MBD2 and CGBP genes S Bader et al 3507

Table 1 Samples used in mutation analyses of genes MBD1 Six CRC cell lines, 80 lung cancer cell lines, 64 MSS primary CRC tumours, five MSI primary CRC tumours, 11 NSCLC primary tumours MBD2 Seven CRC cell lines, 80 lung cancer cell lines, 12 MSS primary CRC tumours, 18 MSI primary CRC tumours CGBP Seven CRC cell lines, 20 lung cancer cell lines, 51 MSS primary CRC tumours as MLL and ALL-1) (Gu et al., 1992; Tkachuk et al., 1992; Ma et al., 1993; Zeleznik-Le et al., 1994; Prasad et al., 1995) and MLL-2 (FitzGerald and Diaz, 1999). CGBP has been shown to bind to DNA elements containing unmethylated CpG, and it has further been shown to have transcription activating properties from such unmethylated promoters (Voo et al., 2000). Whether involved in repression or activation of transcription, these genes together have a role in maintenance of the profile in cells and might therefore be considered to be candidate tumour suppressor genes: loss of normal function of MBD1, MBD2 and/or CGBP could lead to abnormal re- expression or silencing of a host of other genes, allowing tumorigenic progression via epigenetic instability. It has been shown that the level of expression of MBD2 (and the MBD prototype MeCP2) is decreased in a variety of Figure 1 Cytogenetic ordering of loci on 18q21: FISH of human primary human tumours including those of the colon interphase nuclei showing probe order for exons 1 (green) and 26 (Kanai et al., 1999; Muller-Tidow et al., 2001), (red) of the DCC gene with respect to (a) the MBD1 gene (yellow ¨ arrow) (b) the MBD2 gene (yellow arrow); (c) cartoon of suggested consistent with an association of MBD2 abnormality gene order with tumorigenicity. Although there is no evidence for increased tumorigenesis in mouse models totally lacking the Mbd2 gene (Guy et al., 2001; Hendrich et al., 2001), it was important to conduct a formal study looking for mutations in the human MBD2 gene and its MBD1 and MBD1 signal was clearly more distant to the other three, CGBP neighbours. a probe order of MBD1–DCC1–DCC26–MBD2 was We first used FISH to ascertain the order of MBD1 seen in approximately two-thirds of informative nuclei. and MBD2 genes with respect to DCC. Owing to the This places MBD1 probably near to and on the proximity of these three genes the signals were not centromeric side of SMAD4, which is itself about separable on metaphase , so we chose to 0.7 Mb centromeric to DCC1 (Thiagalingam et al., examine interphase nuclei. We used probes from the first 1996). CGBP location was earlier discovered by analysis and the 26th exons of DCC, which lie approximately of a PAC containing the CpG island of MBD1 that 1.3Mb apart (Cho et al., 1994), to orient the group with showed 30 exon sequences of CGBP (Cross et al., 1999). respect to centromere and telomere. PAC probes (Hu- The integrity of this genomic proximity of CGBP to man Genome Mapping Project Resource Centre gridded MBD1 in vivo was confirmed by performing PCR PAC library screened by PCR) for MBD1, MBD2 and between the gene sequences from total genomic DNA both ends of DCC were labelled with either biotin or (data not shown). However, CGBP and MBD1 orienta- digoxygenin for detection with Texas red and FITC- tion relative to the centromere could not be determined conjugated antibodies, respectively. Probes were hybri- by our analyses. Figure 1 shows our inferred order of dized in groups of three in all four possible combina- genes in the region. This arrangement is in total tions. For each set of three probes, one was labelled and agreement with mapping data for the genes as presented detected simultaneously with both systems to give an on the NCBI Genome website for . The overlapping red/green signal. In all, 57 informative project sequence clarifies the orientation interphase nuclei were scored, which included examples of MBD1–CGBP, hence the relative placement of these from each of the four groups of probes. Examples of two genes in our figure. informative nuclei are shown in Figure 1. The MBD2, For the mutation screen by single-stranded conforma- DCC1 and DCC26 probes gave fairly evenly spaced tion polymorphism (SSCP), were extracted by signals with a consistent probe order of DCC1–DCC26– standard methods from pellets of cell lines, from frozen MBD2. Since the distance between DCC1 and DCC26 is archival primary tumours and from normal blood known to be about 1.3Mb, and since the MBD2 probe samples. Cell lines were obtained from ECACC (the appears to be at least as close to the DCC26 probe as is European Collection of Cell Cultures/ATCC) or Drs the DCC1 probe, we inferred that the MBD2 gene lies Minna and Gazdar (University of Texas Southwestern within about 1 Mb of the 30 end of DCC. Although the Medical Centre). Seven colon cancer, 47 SCLC, 33

Oncogene Mutations of MBD1, MBD2 and CGBP genes S Bader et al 3508 NSCLC, one carcinoid and one mesothelioma cell lines MBD1, two mutations were found in cell lines, both of were used (see Table 1 the full list is available upon which affected the third CXXC domain. The cell line request). Primary tumour samples were part of an H1092 mutation is especially interesting as it creates a unselected, anonymized collection from patients at the premature stop codon just proximal to the CXXXXXC Royal Infirmary Edinburgh. Normal blood samples motif within the third CXXC domain. This mutation from independent healthy donors were obtained from only affects some of the five isoforms of the protein since the Scottish National Blood Transfusion Service. exon 11 is not included in all splice forms of the gene Primers were designed for SSCP that spanned each (Fujita et al., 1999), but we know that the relevant translated exon (overlapping where necessary) and transcripts are indeed expressed (cDNA PCR followed included splice acceptor and donor sites. The total by sequencing; data not shown). It is not known exactly numbers of primer pairs were 17 for MBD1, five for what role each isoform plays, but it has been shown that MBD2 and 16 for CGBP (list of sequences available the isoforms containing all three CXXC domains repress upon request). Standard SSCP PCR and gels, and cycle gene expression from both unmethylated and methy- sequencing were performed as described before (Bader lated promoters, while those that contain just the first et al., 1999). two CXXC domains only repress transcription when We found a large number of differences in our promoters are methylated (Fujita et al., 1999, 2000). The samples with respect to the published sequences for mutation in H1092 is predicted to produce truncations the three genes. Most were either silent, or confirmed as in isoforms v1, v2 (Fujita et al., 1999) and exon 7 (Cross polymorphisms by being seen frequently in other et al., 1997) that would have an incomplete third CXXC tumour samples and when checked in 100 normal blood domain but would also lack the C-terminal transcription samples. Others were predicted to be rare single- repression domain (TRD) and so would not mimic the nucleotide polymorphisms (SNP) by being seen in normal v3and v4 isoforms. We are currently determin- normal tissue from the same patient at least but absent ing whether the truncated protein isoforms are stable in from the other individuals studied, for example, in cells, since, if stable, they may have dominant-negative MBD1 of colon primary I9A (Table 2). It is theoretically effects in addition to the partial decline of normal possible that the heterozygous sequence differences seen MBD1 levels. Similar truncated proteins made experi- in the cell lines for which no normal tissue DNA was mentally of the related MeCP2 protein, comprising the available are also rare SNPs. Allowing for this caveat, prototypical MBD (Nan et al., 1993), have been shown however, some mis-sense and nonsense mutations were to be capable of binding to methylated DNA in vitro,as found in the samples screened (Table 2). Of interest for was a similar form of MBD4 (Hendrich et al., 1999b).

Table 2 Changes detected in MBD1, MBD2 and CGBP genes Sample DNA change Amino-acid change Comments

MBD1 Colon I9A Exon 4, C to A H77N Between MBD and NLS, constitut. het. Misc. Exon 4, G to A A80 Silent polymorphism Misc. Intron 7, C to T SA site, exon À9 nt Polymorphism Colon, HCT15 Exon 11, A to G K359E Within third CXXC domain, het. SCLC, H1092 Exon 11, C to T Q370ter Truncation just proximal to third CXXXXXC; het. Misc. Exon 12, C to G P451A Polymorphism SCLC, H1573Exon 12, A to T T429 Silent, het. Colon I17 Exon 12, A to G T427A Het. SCLC, H1514 Exon 12, C to T L459 Silent, het. Colon, HCT15 Intron 12, G to A SA site, exon À4 nt Het. Misc. Exon 15b, C to T C581 Silent polymorphism

MBD2a Colon, HCT15 Exon 2, T to G S183R C-terminal part of MBD, het. Colon, LOVO Exon 2, À1T F206L +frameshift Premature stop 8 codons from mutation, het. Colon, I10 Exon 2, G to T A194S C-terminal part of MBD, het. SCLC, H889 Exon 5, T to C A361V Conservative change, het. Colon, LS180 Intron 6+14 nt, +AG – Unstable microsatellite B10 nt from SD site

CGBP Colon, LOVO Exon 4, C to T R97W Het. Colon, HCT15 Exon 12, G to A C505Y Within PHD2, het. Colon 4.6 Intron 4 G to A SA site, exon À1 nt Conserved G of SA site, het. Misc. Exon 4, G to A K88 Silent polymorphism Misc. Exon 6 ND Polymorphism Misc. Exon 7, A to G S263Silent polymorphism Misc. Exon 10, T to C T442 Silent polymorphism Misc. Intron 12, C to A +14 nt Polymorphism

aAmino-acid numbering refers to the full-length MBD2a form. het, heterozygous; Constitut, constitutive; misc: miscellaneous; ND, not determined.

Oncogene Mutations of MBD1, MBD2 and CGBP genes S Bader et al 3509 The mis-sense mutation in cell line HCT15 was at a have significance in tumorigenesis. Most of the muta- residue with no known significance, but if it significantly tions we found were in MSI cases (colon cancer cell lines altered the structure of the third CXXC domain, this HCT15, LOVO, LS180 and primary tumour I17), which could lead to an important loss of MBD1’s ability to have an increased level of mutations but no chromoso- repress transcription from unmethylated promoters. The mal abnormality, while few mutations were found in the mutation is expressed (cDNA PCR followed by SSCP MSS colon or lung cancers, even though these tend to and/or sequencing; data not shown). lose chromosome 18q21 as seen cytogenetically or by Most of the mutations in MBD2 were located within LOH studies. If haploinsufficiency is important here, the overlap of the MBD and Sin3A-interacting (Boeke either mechanism of loss of normal MBD1/MBD2/ et al., 2000) regions and one, found in cell line LOVO, CGBP may be effective to reduce the amount of normal was a frameshift mutation that was expressed as seen by protein present. Consistent with this hypothesis, mRNA cDNA PCR þ sequencing (data not shown), and when levels of MBD2 were reduced by 50–66% in colorectal translated would produce a truncated protein leaving cancers in one study (Kanai et al., 1999), 20–80% lower only the MBD. If stable in living cells, the truncated in a variety of solid tumours (colon, lung, ovary) protein would be predicted to bind to methyl-CpG and (Mu¨ ller-Tidow et al., 2001), significantly reduced in exert a dominant-negative effect similar to that pre- SCLC and NSCLC when normalized to PCNA (Sato dicted for the MBD1 mutation mentioned above. Exon et al., 2002) and 25% reduced in hepatocellular 1, which encodes the N-terminal part of the MBD, was carcinomas (Saito et al., 2001). However, none of these not screened at the time because the region is extremely studies looked at the methylation status of the gene. As C/G rich (B75%) and thus awkward to PCR amplify, a common mechanism for abnormal silencing of genes especially in small segments for SSCP. Although we in tumours, demonstration of aberrant methylation were later able to PCR amplify the entire 650 bp would strengthen the significance of partial loss of covering the open reading frame using a high-fidelity expression of MBD2 in tumorigenesis. Biallelic loss of polymerase with C/G-rich enhancing additive, this these transcription regulatory genes may be rare because fragment has not yet been found to be amenable to it could be lethal. Homozygous loss of CGBP is available assays for mutation detection. There is thus embryonic lethal (Carlone and Skalnik, 2001). However, scope for further important mutations to be found in there is a lack of morbid phenotype in mouse knockout this gene since DNA encoding the N-terminal half of the models where Mbd2 is lacking (Guy et al., 2001; MBD is included in this missing section. Hendrich et al., 2001), although this may be because We found few sequence changes in CGBP. The most of species differences. interesting of these was that found in a primary colon In summary, we looked for mutations in the genes tumour where the conserved exon À1 nt G of the splice MBD1, MBD2 and CGBP involved in the regulation of acceptor site was altered to an A, predicted to disrupt gene expression that are located on chromosome 18q21 normal splicing such that exon 5 is simply omitted in a region of LOH in tumours. The genes lie intermixed joining exon 4 to 6. Alternatively, a cryptic splice with other candidate tumour suppressor genes DCC, acceptor site within exon 5 may be used, or some other SMAD4, and like them we have found that there is a low splicing occurs involving exon skipping around exon 5. frequency of mutation. We conclude that the abnorm- The new transcript formed by splicing exons 4 and 6 alities may be merely passenger mutations with no together would remain in frame, but the protein would relevance to tumorigenicity, or they reflect further lose the CXXC domain with potential changes in examples of gene abnormalities in that chromosomal protein function. In the more complicated cases region that are important for only a subset of colorectal depending on how the transcript is spliced, the reading and lung cancer cases. Analysis of the methylation frame may alter, creating a new protein sequence and status of at least MBD2 remains to be carried out to probably causing a premature truncation and so loss of implicate that mechanism as the alternative route by normal protein function. Attempts continue to PCR which the gene could play a major role in tumorigenesis. amplify between various combinations of primers from cDNA made from RNA extracted from archival frozen tissue of this tumour to look for an aberrant smaller Acknowledgements We thank Drs John Minna and Adi Gazdar (UTSMC, Dallas) transcript size. for DNAs of 60 of the lung cancer cell lines. Fixed The fact that most of the mutations we found were preparations of human metaphase and interphase cells were present only in one allele could have alternative a gift from Shelagh Boyle (MRC Human Genetics Unit, implications: either the mutations are simply random Edinburgh). This study was supported by the events consequent to the genomic instability occurring UK (SB, AP, DJH), Chief Scientist Office of the Scottish in tumour cells and with no selective advantage, or Executive (SB, MW), the BBSRC (HAMcQ) and the Well- haploinsufficiency of this group of genes is enough to come Trust (APB).

References

Bader S, Walker M, Hendrich B, Bird A, Bird C, Hooper M Bhattacharya SK, Ramchandani S, Cervoni N and Szyf M. and Wyllie A. (1999). Oncogene, 18, 8044–8047. (1999). Nature, 397, 579–583.

Oncogene Mutations of MBD1, MBD2 and CGBP genes S Bader et al 3510 Bestor TH and Verdine GL. (1994). Curr. Opin. Cell Biol., 6, Lanza G, Matteuzzi M, Gafa R, Orvieto E, Maestri I, Santini 380–389. A and del Senno L. (1998). Int. J. Cancer, 79, 390–395. Bird A. (2002). Genes Dev., 16, 6–21. Ma Q, Alder H, Nelson KK, Chatterjee D, Gu Y, Nakamura Boeke J, Ammerpohl O, Kegel S, Moehren U and Renkawitz T, Canaani E, Croce CM, Siracusa LD and Buchberg AM. R. (2000). J. Biol. Chem., 275, 34963–34967. (1993). Proc. Natl Acad. Sci. USA, 90, 6350–6354. Carlone DL and Skalnik DG. (2001). Mol. Cell. Biol., 21, Martinez-Lopez E, Abad A, Font A, Monzo M, Ojanguren I, 7601–7606. Pifarre A, Sanchez JJ, Martin C and Rosell R. (1998). Cho KR, Oliner JD, Simons JW, Hedrick L, Fearon ER, Gastroenterology, 114, 1330–1338. Preisinger AC, Hedge P, Silverman GA and Vogelstein B. Mu¨ ller-Tidow C, Ku¨ gler K, Diederichs S, Klu¨ mpen S, Mo¨ ller (1994). , 19, 525–531. M, Vogt U, Metzger R, Schneider PM, Berdel WE and Serve Cross SH, Clark CH and Bird AP. (1999). Nucleic Acids Res., H. (2001). Br. J. Cancer, 85, 1168–1174. 27, 2099–2107. Nan X, Meehan RR and Bird A. (1993). Nucleic Acids Res., Cross SH, Meehan RR, Nan X and Bird A. (1997). Nat. 21, 4886–4892. Genet., 16, 256–259. Nan X, Ng H-H, Johnson CA, Laherty CD, Turner BM, Detich N, Theberge J and Szyf M. (2002). J. Biol. Chem., 277, Eisenman RN and Bird A. (1998). Nature, 393, 386–389. 35791–35794. Ng H-H, Zhang Y, Hendrich B, Johnson CA, Turner BM, Fabre M, Martin M, Ulloa F and Real FX. (1999). Int J. Erdjument-Bromage H, Tempst P, Reinberg D and Bird A. Cancer, 81, 799–807. (1999). Nat. Genet., 23, 58–61. Fazeli A, Dickinson SL, Hermiston ML, Tighe V, Steen RG, Prasad R, Yano T, Sorio C, Nakamura T, Rallapalli R, Gu Y, Small CC, Stoeckli ET, Keino-Masu K, Masu M, Rayburn Leshkowitz D, Croce CM and Canaani E. (1995). Proc. Natl H, Simons J, Bronson RT, Gordon JI, Tessier-Lavigne M Acad. Sci. USA, 92, 12160–12164. and Weinberg RA. (1997). Nature, 386, 796–803. Riggins GJ, Thiagalingam S, Rozenblum E, Weinstein CL, Kern Fearon ER, Cho KR, Nigro JM, Kern SE, Simmons JW, SE, Hamilton SR, Willson KJKV, Markowitz SD, Kinzler Ruppert JM, Hamilton SR, Preisinger AC, Thomus G, KW and Vogelstein B. (1996). Nat. Genet., 13, 347–349. Kinzler KW and Vogelstein B. (1990). Science, 247, Saito Y, Kanai Y, Sakamoto M, Saito H, Ishii H and 49–56. Hirohashi S. (2001). Hepatology, 33, 561–568. FitzGerald KT and Diaz MO. (1999). Genomics, 59, 187–192. Sato M, Horio Y, Sekido Y, Minna JD, Shimokata K and Fujita N, Shimotake N, Ohki I, Chiba T, Saya H, Shirakawa Hasegawa Y. (2002). Oncogene, 21, 4822–4829. M and Nakao M. (2000). Mol. Cell. Biol., 20, 5107–5118. Shiseki M, Kohno T, Nishikawa R, Sameshima Y, Mizoguchi Fujita N, Takebayashi S-I, Okumura K, Kudo S, Chiba T, H and Yokota J. (1994). Cancer Res., 54, 5643–5648. Saya H and Nakao M. (1999). Mol. Cell. Biol., 19, 6415– Takai K, Kohno T, Hamada K, Takita J, Noguchi M, 6426. Matsuno Y, Hirohashi S, Uezato H and Yokota J. (1998). Gu Y, Nakamura T, Alder H, Prasad R, Canaani O, Cimino Cancer Res., 58, 3700–3705. G, Croce CM and Canaani E. (1992). Cell, 71, 701–708. Thiagalingam S, Lengauer C, Leach FS, Schutte M, Hahn SA, Guy J, Hendrich B, Holmes M, Martin JE and Bird A. (2001). Overhauser J, Willson JKV, Markowitz S, Hamilton SR, Nat. Genet., 27, 322–326. Kern SE, Kinzler KW and Vogelstein B. (1996). Nat. Genet., Hahn SA, Schutte M, Shamsul Hoque ATM, Moskaluk CA, 13, 343–345. da Costa LT, Rozenblum E, Weinstein CL, Fischer A, Yeo Tkachuk DC, Kohler S and Cleary ML. (1992). Cell, 71, 691– CJ, Hruban RH and Kern SE. (1996). Science, 271, 700. 350–353. Virmani AK, Fong KM, Kodagoda D, McIntire D, Hung J, Hendrich B, Abbott C, McQueen H, Chambers D, Cross S and Tonk V, Minna JD and Gazdar AF. (1998). Genes, Bird A. (1999a). Mammal. Genome, 10, 906–912. Chromosomes Cancer, 21, 308–319. Hendrich B, Guy J, Ramsahoye B, Wilson VA and Bird A. Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger (2001). Genes Dev., 15, 710–723. AC, Leppert M, Nakamura Y, White R, Smits AMM and Hendrich B, Hardeland U, Ng H-H, Jiricny J and Bird A. Bos JL. (1988). N Engl. J. Med., 319, 525–532. (1999b). Nature, 401, 301–304. Vogelstein B, Fearon ER, Kern SE, Hamilton SR, Preisinger Jen J, Kim H, Piantadosi S, Liu ZF, Levitt RC, Sistonen P, AC, Nakamura Y and White R. (1989). Science, 244, 207– Kinzler KW, Vogelstein B and Hamilton SR. (1994). N Engl. 211. J. Med., 28, 213–221. Voo KS, Carlone DL, Jacobsen BM, Flodin A and Skalnik Jones PL, Veenstra GJC, Wade PA, Vermaak D, Kass SU, DG. (2000). Mol. Cell. Biol., 20, 2108–2121. Landsberger N, Strouboulis J and Wolffe AP. (1998). Nat. Wade PA, GegonneA, Jones PL, Ballestar E, Aubry F and Genet., 19, 187–191. Wolffe AP. (1999). Nature Genet., 23, 62–66. Kanai Y, Ushijima S, Nakanishi Y and Hirohashi S. (1999). Zeleznik-Le NJ, Harden AM and Rowley J. (1994). Proc. Natl Biochem. Biophys. Res. Commun., 264, 962–966. Acad. Sci. USA, 91, 10610–10614.

Oncogene