Frequent alteration of DNA damage signalling and repair pathways in human colorectal cancers with microsatellite instability

Transcription

1 (2007) 26, & 2007 Nature Publishing Group All rights reserved /07 $ ORIGINAL ARTICLE Frequent alteration of DNA damage signalling and repair pathways in human colorectal cancers with microsatellite instability C Miquel 1, S Jacob 1, S Grandjouan 2, A Aime 3,4, J Viguier 1, J-C Sabourin 5, A Sarasin 1, A Duval 3,4 and F Praz 1,3,4 1 Centre National de la Recherche Scientifique Genetic Instability and Cancer, Paris, France; 2 Service d Oncogénétique Digestive, Hôpital Cochin, Paris, France; 3 INSERM, U762, Paris, France; 4 Université Pierre et Marie Curie-Paris6, Paris, France and 5 Département de Pathologie, Institut Gustave Roussy, Villejuif, France Accumulation of frameshift mutations at genes containing coding mononucleotide repeats is thought to be the major molecular mechanism by which mismatch repair-deficient cells accumulate functional alterations. These mutations resulting from microsatellite instability (MSI) can affect genes involved in pathways with a putative oncogenic role, but may also arise in genes without any expected role in MSI carcinogenesis because of the high mutation background of these tumours. We here screened 39 MSI colorectal tumours for the presence of mutations in 25 genes involved in DNA damage signalling and repair pathways. Using a maximum likelihood statistical method, these genes were divided into two different groups that differed significantly in their mutation frequencies, and likely represent mutations that do or do not provide selective pressure during MSI tumour progression. Interestingly, the so-called real-target mutational events were found to be distributed among genes involved in different functional pathways of the DNA metabolism, for example, DNA damage signalling (DNA-PKcs, ATR), double-strand break (DSB) repair (DNA-PKcs, RAD50), mismatch repair (MSH3, MSH6, MBD4) and replication (POLD3). In particular, mutations in MRE11 and/or RAD50 were observed in the vast majority of the tumours and resulted in the concomitant loss of immunohistochemical expression of both proteins. These data might explain why MSI colorectal cancers (CRC) behave differently in response to a wide variety of chemotherapeutic agents, notably those targeting DNA. More generally, they give further insights into how MSI leads to functional changes with synergistic effects in oncogenic pathways. (2007) 26, ; doi: /sj.onc ; published online 26 March 2007 Keywords: colorectal cancer; MSI phenotype; DNA repair; DNA damage signalling; chemotherapy Correspondence: Dr F Praz, Instabilité des Microsatellites et Cancers, INSERM, U762-Universite Pierre et Marie Curie-Paris6, CEPH, 27 rue Juliette Dodu, Paris, France. praz@cephb.fr Received 29 November 2006; revised 10 January 2007; accepted 6 February 2007; published online 26 March 2007 Introduction The instability of microsatellite repeated sequences in tumours underlies the hereditary non-polyposis colorectal cancer syndrome (HNPCC), which is frequently associated with a germline mutation in one of the mismatch repair (MMR) genes, predominantly MLH1 and MSH2 (Aaltonen et al., 1993; Ionov et al., 1993; Thibodeau et al., 1993). Apart from HNPCC, microsatellite instability (MSI) is also observed in 10 15% sporadic colorectal cancer (CRC), owing to the aberrant methylation of MLH1 promoter. MSI is not confined to only intergenic non-coding repeated sequences but also affects intronic and coding mononucleotide repeats (cmnr), leading to frameshift mutations in numerous genes (Mori et al., 2001; Duval and Hamelin, 2002; Suzuki et al., 2002; Woerner et al., 2003). Until now, accumulation of such mutational events at genes containing cmnr is thought to be the major molecular mechanism by which MMR-deficient cells experience functional changes with putative oncogenic effects, defining the so-called mutator pathway. In absence of functional studies, several attempts have been made to discriminate between the genes that are true targets for colorectal MSI-driven mutagenesis, and the others whose mutations occur randomly, because of the structural or yet undefined criteria in the context of the high level of instability characterizing these tumours. A high mutation prevalence, a bi-allelic inactivation and a role in a growth suppressor pathway are among the most widely accepted criteria, but no consensus has been achieved, especially no mutation incidence cut-off value could be defined (Boland et al., 1998; Duval and Hamelin, 2002). To overcome this problem, different statistical analyses independent of functional considerations have been proposed (Duval et al., 2001; Woerner et al., 2003). Until now, they have been applied to the analyses of mutational data sets concerning target genes that belong to different signalling pathways with no direct relationship. As alterations of these genes could be of functional significance if associated with other mutations in the same pathway, owing to their synergistic effects, it seems of interest to take into consideration all possible target genes involved in closely related biological networks, as proposed previously (Suzuki et al., 2002).

2 5920 In the present study, we have exhaustively taken into consideration the presence of mutations in almost all genes containing cmnr covering the major DNA damage signalling and repair pathways. Based on this method, an overview of how MSI leads to genomic changes with putative synergistic effects in these pathways could be obtained. Results Patients and tumours The clinicopathological characteristics of the patients and their tumours are reported in Table 1. The comparison between the familial and the sporadic cases did not show any significant differences in any of the clinicopathological characteristics. Among the 31 familial cases, defects in MLH1 and MSH2 were observed in 15 and 10 cases, respectively (Table 1). Characterization of the mutations in genes involved in DNA damage signalling and repair pathways We have performed a systematic in silico search for genes containing an exonic mononucleotide repeat of at least eight bases (Ensembl release 32). We have studied 24 genes involved in base excision repair (BER), nucleotide excision repair (NER), MMR, double-strand break (DSB) repair or DNA damage signalling. Among them, seven had not been previously identified as potential target genes; four of them are involved in DNA damage signalling or metabolism (RIF1, TOPORS, TOPBP1 and SMC1L1) and three in replication, either in translesion synthesis (REV1L and REV3L, the polymerase zeta catalytic subunit) or as a PCNA-interacting subunit of polymerase d (POLD3). We further looked for mutations in the intronic MRE11 T11 tract because they impair the splicing process, resulting in a strong decrease of the MRE11 protein level (Giannini et al., 2002). The presence of frameshift mutations is based on peak pattern alterations compared with the normal profile. Examples of representative chromatograms obtained for various genes and tumours are illustrated in Figure 1. The presence of mutations located within the repeated sequence of the 25 selected genes is shown for each tumour in Figure 2. MRE11 mutations have been found in the majority of colorectal MSI tumours (74%). Five genes (BRCA1, BRCA2, SMC1L1, TOPBP1 and TOPORS) were devoid of mutation and two (WRN and XRCC2) displayed a single mutation in our series of tumours. The number of mutations varied greatly among the tumours one tumour (FK15) lacked mutations in these genes but carried mutations in TGFb-RII and TCF4, two well-known target genes in MSI CRC (data not shown). The range of mutated genes within the tumours varied from 0 to 12, with a median of five mutations per tumour. Identification of the real-target genes according to the maximum likelihood method In order to identify the so-called real-target genes for instability, the mutation frequencies of the 24 cmnr were analysed using the statistical maximum likelihood method described previously (Duval et al., 2001). The MRE11 gene whose mutations occur in a T11 intronic tract has been excluded from the statistical analysis because the consequences of MRE11 alterations are less clear than those of frameshift mutations. Yet, the extent of aberrant splicing increases with the size of the deletions at the MRE11 T11 tract, resulting in variable wild-type MRE11 expression. The likelihood method allowed us to individualize two groups of genes according to their mutation frequency with an observed cutoff value at 33%. The first group consisted of seven genes that are considered as real-targets for instability, shown as black histograms (Figure 3). This group included three already known target genes (RAD50, MSH6 and MSH3), as well as four additional genes that had not been studied previously (POLD3) or had not been definitely assigned as real-target genes (ATR, MBD4, DNA-PKcs). All other 17 genes represented the second group of genes that are unlikely to play a major individual role during MSI colorectal tumour progression (Figure 3, grey histograms). Analysis of RAD50 and MRE11 protein expression in MSI CRC according to their mutational status In order to assess how the mutations detected in the MRE11 genes affected MRE11 and RAD50 protein expression, we performed an immunohistochemical study on 28 of the 39 MSI cases on serial sections. In the normal colonic epithelium, MRE11 and RAD50 expression exhibited a strong ubiquitous nuclear immunostaining, as in the case of lymphocytes, endothelial cells and smooth muscle cells (Figure 4). Among the 28 tumours analysed, six were devoid of mutations in both MRE11 and RAD50 and displayed a strong and homogeneous immunostaining of both proteins. In the nine cases that displayed mutations in both MRE11 and RAD50, their expression was decreased in nine and seven cases, respectively (illustrated in Figure 4; case FK14). Among the 13 cases displaying a mutation in MRE11 only, we observed a clear loss of expression, with no or very faint staining in all or part of the tumour in 11 cases (illustrated in Figure 4; cases SK2, SK7). In two remaining cases (FK18 and FK23), no decrease in MRE11 expression could be detected, whatever tumour area was considered; FK18 retained a wild-type MRE11 allele associated with a mutation consisting in a 1 bp deletion, which has previously been reported to moderately affect MRE11 expression (Ottini et al., 2004). As illustrated in Figure 4, the expression of RAD50 fully overlapped that of MRE11, being undetectable or very faint in the tumour areas where MRE11 was either absent or very weak, in agreement with previous reports showing that mutations in MRE11 are associated with a reduced expression of the three members of the MRE11/ RAD50/NBS1 complex. Interestingly, we could also evaluate the expression of MRE11 and RAD50 in an adenoma derived from a patient (FK14) whose tumour had mutations in both genes (Figure 4). Although the adenoma displayed a mutation of MRE11 identical to that of the adenocarcinoma, it lacked the RAD50

3 Table 1 Clinicopathological characteristics of the patients and their tumours Age (years) Sex Tumour location Grade T N M staging UICC stage IHC defect Germline mutation 5921 Familial CRC FK1 48 F Right G1 T1 N0 M0 I MLH1 FK2 38 M Right G1 T4 N0 M0 II MSH2 FK3 65 F Right G2 T1 N0 M0 I MSH2 MSH2 FK4 46 F Left Mucinous T4 N1 M0 III MSH2 FK5 65 F Right G1 T2 N0 M0 I MLH1 MLH1 FK6 67 F Right G2 T3 N0 Mx NDMLH1 MLH1 FK7 27 M Right G1 T3 N0 M0 II MLH1 MLH1 FK8 55 M Right G2 T4 N1 M1 IV MLH1 FK9 42 M Right G1 T3 N2 M0 III MLH1 FK10 52 F Right G2 T4 N2 M1 IV MLH1 FK11 48 M Left G2 T4 N2 M0 III MLH1 FK12 30 M Left G1 T4 N2 M1 IV MLH1 MLH1 FK13 43 M Right G3 T2 N0 M0 I MSH2 FK14 37 F Right G1 T2 N0 M0 I MSH2 MSH2 FK15 32 M Rectum Mucinous T3 N0 M0 II MLH1 FK16 37 M Right G3 T4 N2 M0 III MLH1 MLH1 FK17 51 M Left G1 T1 N0 M0 I MLH1 FK18 69 F Right G1 T1 N0 M0 I FK19 28 F Right G1 T3 N0 M0 II FK20 34 M Right G2 T4 N2 M1 IV MSH2 MSH2 FK21 35 M Rectum G1 T3 N0 Mx NDMSH2 FK22 37 F Left G2 T3 N1 M0 III MSH2 MSH2 FK23 55 F Left G1 T2 N0 Mx NDMLH1 MLH1 FK24 36 F Right G2 T2 N0 M0 I FK25 38 M Left G1 T2 N0 M1 IV FK26 37 M Right Mucinous T3 N1 M0 III FK27 59 F Right G1 T3 N0 M0 II MLH1 MLH1 FK28 40 M Right G3 T4 N2 M0 III MSH2 FK29 72 M Right Mucinous T4 N0 M0 II FK30 39 M Right G2 T3 N0 M0 II MLH1 FK31 36 M Right G3 T3 N0 Mx NDMSH2 MSH2 Sporadic CRC SK1 55 F Right G3 T3 N0 M0 II SK2 66 M Right Mucinous T3 N2 M0 III SK3 41 M Right G3 T3 N0 M0 II SK4 27 M Rectum G2 T3 N0 M0 II SK5 54 F Right G3 T3 N1 M0 III SK6 70 F Right G3 T4 N1 M0 III SK7 51 F Right Adenocarcinoid T4 Nx M1 IV SK8 48 M Right G2 T4 N2 M0 III Abbreviations: UICC, International Union Against Cancer; IHC, immunohistochemical; CRC, colorectal cancer; ND, not determined. The tumours were located either right (ascending and transverse colon) or left (descending and sigmoid colon) to the splenic flexure or in the rectum. The grading of differentiation was performed according to the WHO criteria; tumours were classified as mucinous when an area of extracellular mucin secretion was greater that 50%. Staging was based on tumour size (T), lymph node invasion (N) and distant metastasis (M), according to the classification defined by the UICC. Defects in MSH2 or MLH1 were defined for the familial cases either by immunohistochemical staining or through the determination of the germline mutation. mutation; the expression of both MRE11 and RAD50 was identical and lost in part of the adenoma, confirming further that a defect in MRE11 is sufficient to downregulate RAD50 and that it may be an early event in colorectal carcinogenesis (Figure 4). Analysis of cumulative mutation frequencies in the different genes involved in a given DNA damage signalling or repair pathway We further performed mutation analyses of the 24 selected genes with cmnr based on their involvement in a particular functional network to calculate the cumulative mutation frequencies, shown as dotted histograms (Figure 3). We observed that the great majority of the tumours carried a mutation in at least one of the genes involved in MMR (82%) or in DNA damage signalling (74%). In comparison, the highest mutation rates observed for the genes analysed individually did not exceed 51% for the MMR MSH3 gene or 44% for the DNA damage signalling gene ATR. Similarly, DSB repair was affected in 59% of the tumours, whereas RAD50 and DNA-PKcs, the genes presenting the highest mutation rate in their coding repeated tract, were mutated in no more than 34% of the cases. The two error-prone polymerases capable of bypassing DNA lesions, REV1L and REV3L, were mutated in, respectively, five and four of the 39 MSI tumours; again, as no tumour had a concomitant mutation in these genes, translesion polymerases were targeted in 23% of

4 5922 Figure 1 Examples of the detection of frameshift mutations in coding mononucleotide repeated sequences. The regions encompassing the repeated sequences were amplified by PCR using fluorescent primers and migrated on an ABI3100 genetic analyser (Applied Biosystems). The profiles were analysed using the GeneScan 3.2 software. The wild-type allele is shown in plain, and the mutated allele is indicated by an arrow. tumours. Thus, by using this approach, the presence of an abnormality in a given pathway was repeatedly twice as frequent than that of a particular gene participating in the pathway. Discussion Gene mutation frequency by itself may not be the only reliable criterion in the case of tumours with an exacerbated mutator phenotype, because the accumulation of mutations targeting several genes involved in a particular biological process may synergize and have significant functional consequences even when the mutation frequencies of genes taken independently are too low to be considered as real targets (Suzuki et al., 2002). Thus, we here develop for the first time an exhaustive mutational analysis focused on 25 target genes containing cmnr covering the major DNA damage signalling and repair in a large series of MSI CRC, owing to the role of these processes in the control of the genome integrity and tumour sensitivity to chemotherapeutic agents. Our results show that DNA damage signalling, DSB repair and MMR are frequently altered in MSI CRC, whereas NER and BER pathways are seldom targeted. Interestingly, the role of DNA damage response as an anticancer barrier in early human tumorigenesis has recently been proposed (Bartkova et al., 2005). Indeed, constitutive activation of the DNA damage response frequently occurs at preinvasive stages of major types of human tumours, including CRC, and this checkpoint response is believed to delay or prevent malignant transformation of early lesions (Bartkova et al., 2005). By impairing cell-cycle arrest, senescence and apoptosis, mutations compromising the ATR/ATM-regulated DNA damage response network, such as those occurring in MSI CRC, may allow cell proliferation, survival and tumour progression (Bartkova et al., 2005). Using the statistical maximum likelihood method designed to discriminate Real-Target genes from By- Stander genes described previously (Duval et al., 2001), we further show that genes are divided into two groups that differ significantly in their respective mutation frequencies. Of interest, we observed that real-target genes are scattered over several major DNA repair pathways, which also comprise a number of ByStander genes that behave as additional potential targets for MSI-driven mutations (Figure 3). For example, DSB repair may be hampered because of mutations in Real- Target genes, such as DNA-PKcs or RAD50/MRE11, or in ByStander genes, like BLM or XRCC2. It is noteworthy that defects in any of the above-mentioned genes have been further implicated in hypersensitivity to bleomycin, a DSB inducer, or camptothecin, a topoisomerase I inhibitor (Li et al., 2004; Pommier, 2006; Takemura et al., 2006). Interestingly, our results show that only a minority of MSI CRC is free from mutations in both MRE11 and RAD50 genes. We could further show by immunohistochemistry that mutations in either gene lead to concomitant lack of MRE11 and RAD50 expression in CRC, in keeping with the data obtained by Western blot showing that mutations in either gene result in a significantly decreased expression of the MRE11/RAD50/NBS1 complex (Giannini et al., 2004, 2002). Based on these observations, it is tempting to speculate that the increased chemosensitivity of MSI CRC to irinotecan, an analogue of camptothecin, may result from a defect in the MRE11/RAD50/NBS1 complex (Fallik et al., 2003). To our knowledge, our study is the first one to report that translesion polymerases, REV1L and REV3L, may be targeted in MSI CRC, although mutations occur at

5 5923 Figure 2 MSI-driven target gene mutations in DNA damage signalling and repair pathways in MSI colorectal tumours. Insertion/ deletion mutations in familial (FK) and sporadic (SK) CRC. For each gene, mutations are scored as present (black), absent (grey) or not determined (white). The mutation frequency for each gene represents the percentage of mutated tumours versus the total number of tumours. Tumours are ranked according to the number of mutations detected in the genes studied. rates below the cutoff value previously defined. Regarding the particular cases of genes believed to modulate the response to DNA damage-inducing drugs, their mutation rates need not be elevated to impair the clinical outcome, as mutations are likely to be selected upon treatment. Translesion polymerases play a unique role in maintaining the integrity of the genome particularly when damaged DNA needs to be replicated, their defect may thus be particularly deleterious in response to various chemotherapeutic agents. Yet, REV3L-deficient vertebrate cells are hypersensitive to methylating-agents, platinum-containing drugs and ionizing radiation (Sonoda et al., 2003). In addition to post-replicative repair, some MMR components play a direct role in DNA damage signalling, cell cycle checkpoint and apoptosis (Jacob and Praz, 2002; Yanamadala and Ljungman, 2003; Jiricny, 2006). Thus, a defect in MMR may contribute to the particular sensitivity of MSI CRC to chemotherapy. In the future, defining which DNA repair pathways are functionally relevant to predict the response of MSI CRC to the treatment with chemotherapeutic drugs should be made easier by such an approach and represents a major achievement in the management of patients. Additional exhaustive studies investigating the genomic consequences of MSI-driven mutations

6 5924 Figure 3 Cumulative gene mutation frequencies according to the various DNA damage signalling and repair pathways. Histograms in black represent the mutation frequencies of genes identified as Real-Target genes using the maximum likelihood statistical method; other genes are shown in grey. Dotted histograms represent the cumulative mutation frequency obtained by taking into consideration the genes involved in a given pathway, for example DNA damage signalling, DSB repair, MMR and TLS. Figure 4 Immunohistochemical analysis of MRE11 and RAD50 in normal colonic mucosa and MSI colorectal tumours. Normal mucosa displays a strong nuclear MRE11 and RAD50 immunostaining. Adenocarcinomas carrying MRE11 mutations, with (FK14) or without (SK2 and SK7) RAD50 mutations, display a severe reduction or complete loss of expression of both proteins. The adenoma from patient FK14 that carries an MRE11 mutation only also shows loss of expression of both MRE11 and RAD50. Original magnification 20. accumulating in other pathways with an expected important role in tumour progression would be of interest as well. Materials and methods Tumour samples Primary colon adenocarcinomas displaying MSI were obtained from 31 patients with familial history of colorectal cancer and eight sporadic cases. Specimens were collected following the Institutional Ethics Committee approval and French Medical Research Council guidelines. MSI was tested using the reference Bethesda panel and all were classified as MSI-high according to the National Cancer Institute recommendations (Boland et al., 1998). The presence of a germline mutation in MSH2 or MLH1 has been established for 19 patients. The histologic type and the grading of differentiation were assessed according to the World Health Organization criteria. Tumour staging was determined according to the criteria defined by the World Health Organization and the International Union Against Cancer. Immunohistochemistry Immunohistochemistry for MSH2 and MLH1 was performed as previously described (Fallik et al., 2003). Immunohistochemistry for MRE11 and RAD50 were performed on 4-mm sections from formalin-fixed, paraffin-embedded tissues

7 previously deparaffinized in xylene and rehydrated through a graded alcohol series to distilled water. Antigen retrieval was carried out by heating in a microwave oven for 20 min at 991C in 10 mm citrate buffer (ph 7.3). Then, endogenous peroxydase activity was blocked by immersion in 3% H 2 O 2 for 5 min. The sections were then incubated for 1 h at room temperature with a mouse monoclonal antibody against RAD50 (clone13b3/2c6, 1:50 dilution, GeneTex, San Antonio, TX, USA) or with a rabbit polyclonal antibodies against MRE11 (1:300, Novus Biologicals, Littleton, CO, USA). The immunoreactivity was revealed with the Dako EnVision system (Glostrup, Denmark) according to the manufacturer s instructions and using diaminobenzidine as chromogen. Sections were counterstained with hematoxylin and coverslipped for microscopic evaluation. Negative control staining was performed by omitting the primary antibody. In all cases analysed, the normal colon mucosa flanking the tumour tissue served as internal positive control. Only nuclear staining was taken into account. The MRE11 and RAD50 immunostaining patterns were assessed independently by two pathologists who were not aware of the mutation status. DNA extraction Genomic DNA was extracted from 7-mm-thick formol-fixed paraffin-embedded tissue sections as described (Fallik et al., 2003). The level of contamination of tumours by normal cells has been estimated from the relative intensities of BAT-26 and BAT-25 mutated and normal profiles upon electrophoresis, as described (Brennetot et al., 2005). Only two cases (FK23 and FK29) contained less than 40% tumour DNA that has been established to be the limit for an optimal detection of mutations in target genes (Brennetot et al., 2005). Detection of frameshift mutations The regions encompassing the repeated sequences contained in the 25 selected genes were amplified either in monoplex or in multiplex polymerase chain reaction (PCR). PCR was carried out on approximately 50 ng genomic DNA using fluorescent primers and HotStarTaq DNA Polymerase (Qiagen, Hilden, Germany), as described (Fallik et al., 2003). Oligonucleotide sequences and protocol details are available upon request. The electrophoresis profiles of the PCR products were analysed using the GeneScan 3.2 software (Applied Biosystems, Foster City, CA, USA). The presence of frameshift mutations is based on peak pattern alterations compared with the reference peak size and profile patterns established for each gene using numerous normal DNA samples. All PCR with abnormal profiles were repeated at least twice in independent monoplex PCR to confirm the presence of a mutation. Statistical analysis Differences in the mean mutation frequency according to a number of clinicopathological parameters, including gender, tumour location, tumour stage and differentiation have been performed using the Epi6.1 software (World Health Organization). In order to discriminate between the real-target and the ByStander genes, we used the maximum likelihood method described previously (Duval et al., 2001). Acknowledgements We thank Dr Brigitte Bressac-de-Paillerets and Dr Sylviane Olschwang for sharing data, and Vale rie Velasco, Nadine Leonard, Joe lle Lacombe and Olivier Buhard for their expert technical assistance. We also thank Dr Pierre Duvillard, Professor Pierre Netter and Dr Filippo Rosselli for constant support and stimulating discussions. Financial supports have been provided by the Fondation pour la Recherche Me dicale (to CM), the Ministe` re de l Education Nationale de la Recherche et de la Technologie, and by the Ligue Nationale contre le Cancer (to SJ) and the Association pour la Recherche sur le Cancer (to JV). This research was granted by the Association pour la Recherche sur le Cancer (#4683 and 3835 to FP) References Aaltonen LA, Peltomaki P, Leach FS, Sistonen P, Pylkkanen L, Mecklin JP et al. (1993). Clues to the pathogenesis of familial colorectal cancer. Science 260: Bartkova J, Horejsi Z, Koed K, Kramer A, Tort F, Zieger K et al. (2005). DNA damage response as a candidate anticancer barrier in early human tumorigenesis. Nature 434: Boland CR, Thibodeau SN, Hamilton SR, Sidransky D, Eshleman JR, Burt RW et al. (1998). A National Cancer Institute Workshop on Microsatellite Instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer. Cancer Res 58: Brennetot C, Buhard O, Jourdan F, Flejou JF, Duval A, Hamelin R. (2005). Mononucleotide repeats BAT-26 and BAT-25 accurately detect MSI-H tumors and predict tumor content: implications for population screening. Int J Cancer 113: Duval A, Hamelin R. (2002). Mutations at coding repeat sequences in mismatch repair-deficient human cancers: toward a new concept of target genes for instability. Cancer Res 62: Duval A, Rolland S, Compoint A, Tubacher E, Iacopetta B, Thomas G et al. (2001). Evolution of instability at coding and non-coding repeat sequences in human MSI-H colorectal cancers. Hum Mol Genet 10: Fallik D, Borrini F, Boige V, Viguier J, Jacob S, Miquel C et al. (2003). Microsatellite instability is a predictive factor of the tumor response to irinotecan in patients with advanced colorectal cancer. Cancer Res 63: Giannini G, Rinaldi C, Ristori E, Ambrosini MI, Cerignoli F, Viel A et al. (2004). Mutations of an intronic repeat induce impaired MRE11 expression in primary human cancer with microsatellite instability. 23: Giannini G, Ristori E, Cerignoli F, Rinaldi C, Zani M, Viel A et al. (2002). Human MRE11 is inactivated in mismatch repair-deficient cancers. EMBO Rep 3: Ionov Y, Peinado MA, Malkhosyan S, Shibata D, Perucho M. (1993). Ubiquitous somatic mutations in simple repeated sequences reveal a new mechanism for colonic carcinogenesis. Nature 363: Jacob S, Praz F. (2002). DNA mismatch repair defects: role in colorectal carcinogenesis. Biochimie 84: Jiricny J. (2006). The multifaceted mismatch-repair system. Nat Rev Mol Cell Biol 7: Li HR, Shagisultanova EI, Yamashita K, Piao Z, Perucho M, Malkhosyan SR. (2004). Hypersensitivity of tumor cell lines with microsatellite instability to DNA double strand break

8 5926 producing chemotherapeutic agent bleomycin. Cancer Res 64: Mori Y, Yin J, Rashid A, Leggett BA, Young J, Simms L et al. (2001). Instabilotyping: comprehensive identification of frameshift mutations caused by coding region microsatellite instability. Cancer Res 61: Ottini L, Falchetti M, Saieva C, De Marco M, Masala G, Zanna I et al. (2004). MRE11 expression is impaired in gastric cancer with microsatellite instability. Carcinogenesis 25: Pommier Y. (2006). Topoisomerase I inhibitors: camptothecins and beyond. Nat Rev Cancer 6: Sonoda E, Okada T, Zhao GY, Tateishi S, Araki K, Yamaizumi M et al. (2003). Multiple roles of Rev3, the catalytic subunit of polzeta in maintaining genome stability in vertebrates. EMBO J 22: Suzuki K, Dai T, Suzuki I, Dai Y, Yamashita K, Perucho M. (2002). Low mutation incidence in polymorphic noncoding short mononucleotide repeats in gastrointestinal cancer of the microsatellite mutator phenotype pathway. Cancer Res 62: Takemura H, Rao VA, Sordet O, Furuta T, Miao ZH, Meng L et al. (2006). Defective Mre11-dependent activation of Chk2 by ataxia telangiectasia mutated in colorectal carcinoma cells in response to replication-dependent DNA double strand breaks. J Biol Chem 281: Thibodeau SN, Bren G, Schaid D. (1993). Microsatellite instability in cancer of the proximal colon. Science 260: Woerner SM, Benner A, Sutter C, Schiller M, Yuan YP, Keller G et al. (2003). Pathogenesis of DNA repair-deficient cancers: a statistical meta-analysis of putative Real Common Target genes. 22: Yanamadala S, Ljungman M. (2003). Potential role of MLH1 in the induction of p53 and apoptosis by blocking transcription on damaged DNA templates. Mol Cancer Res 1: