Research Article

Genetic Instability Caused by Loss of MutS Homologue 3 in Human

Astrid C. Haugen,1 Ajay Goel,2 Kanae Yamada,3 Giancarlo Marra,6 Thuy-Phuong Nguyen,2 Takeshi Nagasaka,2 Shinsaku Kanazawa,4 Junichi Koike,4 Yoshinori Kikuchi,3 Xiaoling Zhong,3 Michitsune Arita,3 Kazutoshi Shibuya,5 Mitsuo Oshimura,7 Hiromichi Hemmi,3 C. Richard Boland,2 and Minoru Koi2

1Laboratory of Molecular Genetics, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina; 2Gastrointestinal Cancer Research Laboratory, Baylor Research Institute and Sammons Cancer Center, Baylor University Medical Center, Dallas, Texas; Departments of 3Molecular Biology, 4Surgery, and 5Surgical Pathology, Faculty of Medicine, Toho University, Ohta-ku, Tokyo, Japan; 6Institute for Molecular Cancer Research, University of Zurich, Zurich, Switzerland; and 7Department of Biomedical Science, Institute of Regenerative Medicine and Biofunction, Graduate School of Medical Science, Tottori University, Yonago, Tottori, Japan

Abstract activity, usually following losses of hMSH2 or hMLH1 (1, 2). Most instability (MSI) is a hallmark of mismatch Lynch syndrome CRC tumor specimens with germline mutations in hMSH2 hMLH1 repair (MMR) deficiency. High levels of MSI at mononucleo- or exhibit MSI-H (3), as do all sporadic CRCs where hMLH1 tide and dinucleotide repeats in colorectal cancer (CRC) are the gene has been silenced by promoter hypermethylation attributed to inactivation of the MMR genes, hMLH1 and (4). Microsatellite loci containing mononucleotide, dinucleotide, hMSH2. CRC with low levels of MSI (MSI-L) exists; however,its trinucleotide, and tetranucleotide repeats may be affected in MSI-H molecular basis is unclear. There is another type of MSI— tumors (1). elevated microsatellite alterations at selected tetranucleotide MSI-L CRCs have mutations at <30% to 40% of microsatellite loci and are usually found in nonfamilial cases (1, 2). These cancers repeats (EMAST)—where loci containing [AAAG]n or [ATAG]n repeats are unstable. EMAST is frequent in non-CRCs; rarely exhibit MSI in loci containing mononucleotide repeats (1); MYCL however,the incidence of EMAST and its cause in CRC is not however, they often exhibit MSI at the 1 locus, which harbors known. Here,we report that MutS homologue 3 (MSH3) [AAAG]n tetranucleotide repeats (2, 5). In many noncolorectal knockdown or MSH3-deficient cells exhibit the EMAST human cancers, a subset of tumors shows high-frequency or low- phenotype and low levels of mutations at dinucleotide repeats. frequency MSI at loci containing mononucleotide or dinucleotide About 60% of 117 sporadic CRC cases exhibit EMAST. All of the repeats, similar to MSI-H and MSI-L colorectal tumors, respectively cases defined as MSI-H (16 cases) exhibited high levels of (2). However, there is another type of MSI where mutations rarely EMAST. Among 101 non–MSI-H cases,all 19 cases of MSI-L occur at loci with mononucleotide or dinucleotide repeats but and 35 of 82 cases of MSS exhibited EMAST. Although non– occur frequently at loci with trinucleotide or tetranucleotide MSI-H CRC tissues contained MSH3-negative tumor cells repeats (2, 6). Increase in instability at certain tetranucleotide ranging from 2% to 50% of the total tumor cell population, repeat markers containing [AAAG]n or [ATAG]n is frequently found the tissues exhibiting EMAST contained more MSH3-negative in non–small cell lung (7, 8), head and neck(8), bladder (8–10), cells (average,31.5%) than did the tissues not exhibiting kidney (8), skin (9), and ovarian (11) cancers. This mutational EMAST (8.4%). Taken together,our results support the concept signature has been termed elevated microsatellite alterations at that MSH3 deficiency causes EMAST or EMAST with low levels selected tetranucleotide repeats (EMAST; ref. 2). Both MSI-L and of MSI at loci with dinucleotide repeats in CRC. [Cancer Res EMAST rarely exhibit MSI in loci consisting of mononucleotide 2008;68(20):8465–72] repeats, but frequently exhibit MSI in loci containing [AAAG]n repeats, suggesting that MSI-L and EMAST may share a similar Introduction molecular basis. However, the mechanisms underlying MSI-L and EMAST are not known (11, 12). Defects in the mismatch repair (MMR) system result in a To date, six MMR genes have been identified in eukaryotic cells. mutator phenotype, manifesting as microsatellite instability (MSI) These include three homologues of the Escherichia coli MutS gene, in DNA of affected cells. Colorectal cancers (CRC) exhibiting MSI MSH2, MSH3, and MSH6, and the MutL homologues, MLH1, PMS2, can be classified as MSI-high (MSI-H) or MSI-low (MSI-L) based on and MLH3 (13). According to the current model, DNA mismatch frequency of mutations at microsatellite loci (1, 2). MSI-H CRCs recognition is mediated by MutSa, a heterodimer of MSH2 and have been defined as those exhibiting MSI at a threshold of >30% MSH6, or MutSh, a heterodimer of MSH2 and MSH3 (14–16). to 40% of loci examined. These tumors are defective in MMR MutSa recognizes the majority of base/base mismatches (14–16). Both MutSa and MutSh recognize small insertion/deletion loops (IDL) with up to 10 unpaired nucleotides (17, 18). In yeast, although a h Note: Supplementary data for this article are available at Cancer Research Online both MutS and MutS play a role in repairing IDLs containing (http://cancerres.aacrjournals.org/). one and two unpaired nucleotides (16, 17), MutSh plays a more Requests for reprints: C. Richard Boland and Minoru Koi, Gastrointestinal Cancer Research Laboratory (250 Hoblitzelle), Baylor Research Institute and Sammons Cancer dominant role in repairing IDLs with more than four unpaired Center, Baylor University Medical Center, 3500 Gaston Avenue, Hoblitzelle, Dallas, TX nucleotides (19). In humans, both MutSh and MutSa are 75246. Phone: 214-820-2692; Fax: 214-818-9292; E-mail: [email protected] or competent for repairing IDLs containing 1 to 10 unpaired [email protected]. h I2008 American Association for Cancer Research. nucleotides (18, 20–23), but MutS has a stronger affinity for doi:10.1158/0008-5472.CAN-08-0002 recognizing more than two unpaired nucleotides (18, 21). www.aacrjournals.org 8465 Cancer Res 2008; 68: (20). October 15, 2008

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. Cancer Research

Furthermore, genetic complementation of MSH3 deficiency in Storm imaging system. As a positive control, an extract from the human human cells increased stability at loci containing dinucleotide and CRC cell line, WiDr, was used. tetranucleotide repeats (22). Thus, it seems plausible that loss of Vector construction,DNA transfection,and small interfering RNA MutSh due to MSH3 inactivation in human cells may result in MSI expression. The tetracycline-regulated retroviral vector, TMP (Open Biosystems) was used to construct a plasmid expressing small interfering not only at loci containing dinucleotide repeats, but also at loci RNA (siRNA) against the hMSH3 message. The vector contains a with tetranucleotide repeats, such as EMAST loci. tetracycline regulated Pol II promoter that directs expression of down- In this study, we first addressed whether down-regulation or loss stream genes, including a puromycin resistance gene and GFP genes under of MSH3 results in instability at EMAST and other loci containing the regulation of the transcription factor tTA. Construction of the siRNA mononucleotide and dinucleotide repeats. To answer this question, expression plasmid was performed according to the manufacturer’s we generated MSH3-positive and MSH3-negative cell lines and protocols. Nineteen coded oligonucleotides (5¶-GCAAGGAGTTATGGAT- examined them for stability at mononucleotide and dinucleotide TAA-3¶) in exon 23 of the hMSH3 gene (26) were inserted into the TMP microsatellite and EMAST loci. We then examined whether MSI at vector. The resulting vector (pMSH3si) was clone purified, and the inserts EMAST loci is common in CRC and determined the relationships were verified by DNA sequencing. The HCT116+3+5 cells were cotransfected between EMAST and MSI-H, MSI-L, and MSS status in CRC as with a pMSH3si plus pTet-off vector (BD Biosciences) encoding tTA using defined by five consensus National Cancer Institute (NCI) micro- Effectene transfection reagent (QIAGEN). After transfection, the cells were selected using 0.6 Ag/mL of puromycin for 10 d. The resulting clones were satellite markers, including BAT25, BAT26 D2S123, D5S346, D17S250, D18S64 isolated and examined under a fluorescent microscope to detect GFP plus two additional dinucleotide markers ( and expression. To turn off the expression of the siRNA, 100 ng/mL of D18S69 ). Finally, we analyzed CRC tissues for expression of MSH3 doxycycline (Dox; BD Biosciences) was added to the culture medium. by immunohistochemical (IHC) staining using an anti-MSH3 Single-clone microsatellite assay. The method for single-cell MSI assay antibody. Our data show that MSH3 deficiency is associated with has been described (24). Genomic DNA from single clones was subjected to EMAST and low levels of instability at the loci with dinucleotide DNA amplification and microsatellite analysis, as described below. The repeat in both cell lines and CRC tissues. microsatellite markers used for this assay were BAT25, BAT26, D2S123, D5S346, D17S250, D17S791, MYCL1, D19S394, and D9S242. Tissue and DNA extraction. One hundred seventeen unselected Materials and Methods sporadic CRCs and corresponding normal tissues were collected at the Cell culture. The human CRC cell lines, HCT116 and HCT116+3-clone 6, University of California San Diego (UCSD) under Institutional Review were described previously (24). HCT116 cells were grown in DMEM with Board approval. For DNA extraction, tumor and normal tissues were 10% fetal bovine serum. HCT116+3-clone 6 was grown in growth medium microdissected separately from paraffin-embedded sections (5 Am) plus 400 Ag/mL G418 (24). Mouse A9 cells containing a blasticidin-resistant according to procedures previously described (27). Genomic DNA was gene-marked human 5 (A9+5; ref. 25) were maintained in isolated from microdissected tissues using GenReleaser (Bioventures, Inc.) growth medium containing 6 Ag/mL blasticidin S (Invitrogen). The human and treated with proteinase K. A second cohort of 88 unselected sporadic CRC cell line WiDr was maintained in growth medium containing 10% CRCs (bearing ‘‘CR’’ in their identifiers) and corresponding normal tissues serum. were collected from patients treated at Toho University Ohmori Hospital Microcell-mediated chromosome transfer. The method for microcell- (Tokyo). mediated chromosome transfer has been described (24). Microcells isolated DNA amplification and microsatellite analysis for CRC tissues. Two from A9+5 cells were fused to HCT116 and HCT116+3. After 24 h, the cells methods were used for microsatellite analysis. In the first method, genomic were trypsinized, pelleted, resuspended, and plated into growth medium DNA was amplified and labeled with radioisotope by 35 to 40 cycles of PCR, containing 6 Ag/mL of blasticidin S. After 10 d, growing colonies were as previously described (27). After denaturation in formamide at 95jC for isolated, expanded, and analyzed for hMSH3 expression. 5 min, PCR products were separated on an 8% polyacrylamide gel Single-stranded conformational polymorphism analysis. Genomic containing 7.5 mol/L urea and exposed to X-ray films. The resulting band DNA isolated from A9+5, HCT116, HCT116+3, HCT116+5, and HCT116+3+5 images between tumor and normal samples were visually compared. In the was subjected to PCR amplification of hMSH3 exon 7. Primers used were a second method, PCR amplifications were performed from genomic DNA forward primer (5¶-GAGATAATGACTGATACTTCTACC-3¶) and a reverse using fluorescently labeled primers. After heat denaturation, amplified PCR primer (5¶-CATTTGTTCCTCACCTGCAAAG-3¶). Samples were heated to products were electrophoresed on an ABI PRISM 3100 Avant Genetic 85jC for 5 min, loaded onto 20% Tris-borate EDTA (TBE) gel (Invitrogen), Analyzer (Applied Biosystems) and analyzed by GeneMapper fragment and run at 300 V and 9jC. The gel was silver-stained with a SilverXpress kit analysis software (Applied Biosystems). Five markers with dinucleotide (Invitrogen). repeats (D2S123, D5S346, D17S250, D18S64, and D18S69), two markers with cDNA isolation and sequencing. Total mRNA was isolated from normal mononucleotide repeats (BAT25 and BAT26), and seven EMAST markers human lymphocytes, HCT116+3, HCT116+5, and HCT116+3+5 using a (MYCL1, D20S82, D20S85, L17835, D8S321, D9S242, and D19S394) were used. QuickPrep total RNA extraction kit (GE Healthcare). Oligo dT-primed The primer sequences and PCR conditions for D2S123, D5S346, D17S250, cDNAs were synthesized from 2 Ag total RNA using a First-Strand cDNA D18S64, MYCL1, and L17835 were described previously (1, 7). Those of Synthesis kit (Invitrogen). hMSH3 cDNA was prepared by reverse D18S69, D20S82, D20S85, D8S321, D9S242, D18S64, and D19S394 were found transcriptase using a forward primer (5¶-GACTGTTTGTTCATGTACGC-3¶) in the genome database. Tumors were categorized as follows: (a) MSI-H, and a reverse primer (5¶-ACTATCAAACACAACCTCGC-3¶) and sequenced tumors exhibiting MSI at three or more of the seven mononucleotide or using an ABI PRISM dye terminator kit with an ABI 373 fluorescent dinucleotide; (b) MSI-L, tumors exhibiting MSI at one or two of the seven sequencer (Applied Biosystems). markers; (c) MSS, tumors which did not exhibit MSI at any of the seven Western blot analysis. Anti-hMSH3 monoclonal mouse antibody (Clone markers; (d) EMAST, tumors exhibiting MSI at more than one locus among 52) and anti-hMLH1 mouse monoclonal antibody (Clone G168-728) were the seven EMAST markers; and (e) non-EMAST, tumors which did not show obtained from BD Biosciences PharMingen. Heat-denatured cell lysates MSI at any of the EMAST markers. (50 Ag) were subjected to electrophoresis in a 10% polyacrylamide gel with MSH3 IHC. Nineteen cases (UCSD cohort) and 27 cases (Toho sodium dodecyl sulfate. Separated proteins were electroblotted to a cohort) were subjected to MSH3 IHC. Paraffin-embedded tissues were polyvinylidene difluoride membrane. The membrane was treated with deparaffinized and rehydrated. After antigen retrieval [121jC for 15 min in primary antibody followed by treatment with horseradish peroxidase– 0.01 mol/L citrate buffer (pH 6.0)], the tissues were treated with anti-MSH3 conjugated goat anti-mouse IgG. Horseradish peroxidase was detected by antibody (28) overnight at 4jC, followed by incubation with secondary exposure to film with ECL reagent (GE Healthcare) and visualized by the antibody. Staining was developed by diaminobenzide chromogen and

Cancer Res 2008; 68: (20). October 15, 2008 8466 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. MSI-L/EMAST Associated with MSH3 Deficiency in CRC counterstained with hematoxylin. The brown and blue signals indicated the conformational polymorphism (SSCP) analysis. As shown in presence (brown) or absence (blue) of MSH3 immunoreactivity. To Fig. 1A, HCT116+5 and HCT116+3+5 contained the wild-type (W) determine the percentage of MSH3-negative cells in a tissue, all of the hMSH3 allele observed in the chromosome donor (A9+5), in  nuclei present in 10 randomly chosen sites using the 40 objective lens addition to a mutated allele (M). In contrast, HCT116 and were scored for MSH3 expression. An average of f6,000 nuclei per case was HCT116+3 contained only the mutated allele. cDNA sequence data examined. As a negative control for MSH3 staining, CRC tissues from MSH2-negative patients were used. The examiner for the IHC was blinded for exon 7 from normal human lymphocyte cells, HCT116, to the MSI and EMAST data. HCT116+3, HCT116+5, and HCT116+3+5, revealed that HCT116+5 Statistical analysis. Groups were compared using Fisher’s exact test, m2 and HCT116+3+5 expressed wild-type and mutant MSH3 mRNA test, Student’s t test, and Steel-Dwass test. Analyses were performed using (Fig. 1B). The expression of MSH3 was detected in WiDr, JMP software, version 5.1 (SAS Institute) and Excel Statistics for Windows HCT116+5, and HCT116+3+5 but not in HCT116 or HCT116+3 by 2006 software (Social Survey Research Information Co., Ltd.). All P values Western blotting (Fig. 1C). These results indicate that in HCT116+5 are two-sided, and any P value of <0.05 was considered statistically and HCT116+3+5, the transferred hMSH3 is expressed at the mRNA significant. and protein level. We then determined whether expression of MSH3 correlates Results and Discussion with stabilization of microsatellite loci in HCT116+5 (MSH3+/ À MSI profiles caused by loss of MSH3 in cultured colon MLH1 ) and HCT116+3+5 (MSH3+/MLH1+) by performing single- cancer cell lines. The human colon cancer cell line HCT116 is clone MSI assays. HCT116+5 exhibited high levels of MSI at all loci À À defective in MLH1 (MLH1 ) and MSH3 (MSH3 ). The MLH1 examined, suggesting that lackof MLH1 expression results in MSI deficiency in HCT116 is corrected by stable transfer of a normal not only at loci with [A]n and [CA]n repeats but also at EMAST loci. copy of human chromosome 3 into HCT116+3 (24). To examine the In contrast, complete stabilization was observed at all micro- degree of instability at microsatellite loci containing mononucle- satellite loci in HCT116+3+5 (Table 1). When HCT116+3+5 was otide and dinucleotide repeats and EMAST loci in HCT116 compared with HCT116+3, the presence of MSH3 significantly À À À (MSH3 /MLH1 ) and HCT116+3 (MSH3 /MLH1+), we measured correlated with stabilization of the EMAST loci MYCL1(P < 0.0001) mutation frequencies at these loci using a single-clone micro- and D9S242 (P = 0.047) and strongly correlated with stabilization of satellite assay. One hundred sixty-five clones from HCT116+3 were D19S394 (P = 0.054; Table 1 and Supplementary Table S1). These analyzed. No mutations were detected in the BAT26 locus, results indicate that restoration of the expression of MSH3 by suggesting that transfer of wild-type hMLH1 corrects MSI at transfer into HCT116+3 is associated with mononucleotide repeats (Table 1 and Supplementary Table S1). stabilization of EMAST loci. D17S791, which contains [CA]n repeats, showed significantly lower To provide direct evidence for a role of MSH3 on instability of levels of instability in HCT116+3 than were observed in HCT116. In EMAST loci, we constructed an HCT116+3+5 cell line whose MSH3 contrast, a comparable mutation frequency was observed at was down-regulated by siRNA expression targeted to its message. EMAST loci in HCT116+3 and HCT116, including MYCL1, To generate this cell line, we constructed an MSH3-siRNA D19S394, and D9S242 (Table 1 and Supplementary Table S1). expression vector (pMSH3si) containing 19 coded nucleotides in These results suggest that complementation of an MLH1 defect via exon 23 of the hMSH3 locus (26). The cells were cotransfected with chromosome 3 transfer stabilizes loci containing [A]n and [CA]n pMSH3si and pTet-Off plasmids and selected with puromycin. repeats and fails to stabilize EMAST loci in HCT116. Drug-resistant clones were isolated and subjected to fluorescent Because HCT116 contains homozygous frameshift mutations of microscopy to detect expression of GFP. GFP-positive clones were the [A]8 repeat in exon 7 of the hMSH3 gene (22), we examined the further subjected to Western blot analysis for MSH3 expression. As effects of introduction of a normal copy of hMSH3 into HCT116 shown in Fig. 1C, clone G5 revealed down-regulated MSH3. and HCT116+3 via chromosome 5 transfer. Human chromosome 5 Reexpression of MSH3 was seen when G5 cells were cultivated in tagged with a blasticidin-resistant gene (25) was transferred from the presence of Dox (Fig. 1C). Comparing G5 cells with G5 cells mouse A9 hybrid cells to HCT116 and HCT116+3 using microcell exposed to Dox, there was no difference in the expression levels fusion. Several blasticidin-resistant clones were isolated and of any of the other MMR proteins, including MLH1 (Fig. 1C), analyzed for presence of a wild-type hMSH3 gene by single-strand MSH2, and MSH6 (not shown). These results indicate that

Table 1. The effect of MSH3 deficiency and/or MLH1 deficiency on stability of various microsatellite loci

Cell lines (MMR status) No.subclones with MSI/no.subclones examined at each locus (%)

BAT26 [A]n D17S791 [CA]n MYCL1 [AAAG]n D19S394 [AAAG]n D9S242 [AAAG]n

À À HCT116 (MLH1 , MSH3 ) 5/49 (10)* 22/56 (39)* 6/48 (13)* 6/33 (18)* 6/30 (20)* À c c HCT116+3 (MLH1+, MSH3 ) 0/165 (0) 4/178 (2) 16/186 (9)* 5/36 (14)* 4/32 (13)* À HCT116+5 (MLH1 , MSH3+) 4/54 (7)* 18/54 (33)* 6/55 (11)* 4/32 (13)* 2/32 (6)NS c c c c c HCT116+3+5 (MLH1+, MSH3+) 0/169 (0) 0/172 (0) 0/187 (0) 0/35 (0) 0/35 (0)

Abbreviation: NS, not significant. * Frequency of mutations was significantly different from HCT116+3+5 according to Fisher’s exact test (Supplementary Table S1). cFrequency of mutations was significantly different from HCT116 according to Fisher’s exact test (Supplementary Table S1).

www.aacrjournals.org 8467 Cancer Res 2008; 68: (20). October 15, 2008

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. Cancer Research

Figure 1. Expression of hMSH3 in HCT116 derivatives. A, hMSH3 SSCP analysis.DNA from A9+5, HCT116, HCT116+3, HCT116+5, and HCT116+3+5 was subjected to PCR using hMSH3-specific primers, and the PCR products were denatured and separated on an TBE gel.A9+5, HCT116+5, and HCT116+3+5 contain wild-type bands (W) in addition to mutated bands (M). B, sequencing of exon 7 containing [A]8 repeats of hMSH3 cDNA. cDNA from normal lymphocytes contains [A]8, whereas cDNA from HCT116+3 contains [A]7.A mixed peak signal (N)atthe3¶ end of the [A]7 tract was detected in cDNA from HCT116+5 and HCT116+3+5. C, Western blot analysis for MSH3 and MLH1 expression using mouse monoclonal anti-MSH3 and anti-MLH1 antibodies.WiDr was used as a positive control for MSH3 and MLH1 expression.No MSH3-specific band was detected in HCT116, HCT116+3, or HCT116+3+5 clone G5, whereas MSH3 was detected in HCT116+5, HCT116+3+5, and HCT116+3+5 clone G5 cultivated under the presence of Dox.MLH1 expression was detected in HCT116+3 and HCT116+3+5, but not in HCT116 and HCT116+5.

down-regulation of MSH3 can be specifically controlled by the mutation at EMAST loci including MYCL1, D19S394, and D9S242 expression of MSH3-specific siRNA. and at dinucleotide repeat loci including D2S123, D5S346, and We next isolated f70 independent clones each from G5 cells D17S250 in MSH3-negative cells were 10.3% and 1.9%, respectively, and G5 cells exposed to Dox during clonal growth and performed suggesting that EMAST loci are nearly five times as unstable as single-clone MSI analysis for five consensus NCI markers (BAT25, dinucleotide repeat loci when MSH3 is absent. Reexpression of BAT26, D2S123, D5S346, and D17S250) and three EMAST markers MSH3 in G5 cells by Dox greatly decreased frequency of mutations (MYCL1, D19S394, and D9S242). As shown in Table 2, the MSI at all loci with [CA]n repeats and at EMAST loci (Table 2). pattern obtained in clones from MSH3-negative G5 cells is similar We found no significant effect of MSH3 deficiency on the to that of HCT116+3 clones. Both exhibited complete stability at stability of dinucleotide repeats D2S123, D5S346, and D17S250 loci containing mononucleotide repeats, low levels of instability when each locus was subjected to Fisher’s exact test individually. at loci containing dinucleotide repeats, and high levels of instability However, a statistically insignificant but stronger effect was found at the EMAST loci (Tables 1 and 2). Average frequencies of when all three loci were combined for analysis (P = 0.056; Table 2).

Table 2. The effect of down-regulation of MSH3 on stability at various microsatellite loci

Cell lines No.subclones with MSI/no.subclones examined (%)

BAT25 BAT26 D2S123 D5S346 D17S250 D2S123 + MYCL1 [AAAG]n D19S394 D9S242 MYCL1 + [A]n [A]n [CA]n [CA]n [CA]n D5S346 + [AAAG]n [AAAG]n D19S394 + D17S250 D9S242

À G5 (MLH1+, MSH3 ) 0/68 0/68 2/70 (3) 1/67 (1) 1/69 (1) 4/206 (2) 9/64 (14) 7/70 (10) 5/70 (7) 21/204 (10) G5 plus Dox 0/68 0/68 0/72 0/72 0/72 0/216 0/72 0/71 1/69 (1) 1/212 (0.5) (MLH1+, MSH3+) P* 0.24 0.48 0.49 0.056 0.0008 0.006 0.21 <0.00001

*P values were determined according to Fisher’s exact test.

Cancer Res 2008; 68: (20). October 15, 2008 8468 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. MSI-L/EMAST Associated with MSH3 Deficiency in CRC

Table 3. High prevalence of EMAST in CRC MSI assay for HCT116+5 and HCT116+3+5 suggest that MLH1 deficiency results in a high frequency of mutations at EMAST No.EMAST No.non-EMAST Total markers (Table 1). Thus, our results suggest that the MSI-H/EMAST cases (%) cases (%) phenotype may be caused by MLH1 deficiency. Association of EMAST with down-regulation of MSH3 MSI-H 16 0 16 (13.7) expression in CRC tissues. To determine whether MSI-L/EMAST MSI-L 19 0 19 (16.2) and MSS/EMAST tumors exhibit loss or down-regulation of MSH3, MSS 35 47 82 (70.1) we performed IHC staining on 46 CRC tissues for the expression of Total 70 (59.8) 47 (40.2) 117 the MSH3 protein. Because only 19 cases from our UCSD cohort were available for IHC studies, an additional 27 samples were obtained from a second cohort of 88 sporadic CRC cases collected Abbreviations: EMAST, a tumor with more than one EMAST marker at Toho University. These 27 cases were analyzed for MSI using exhibiting mutation; non-EMAST, a tumor with no mutations in any of seven EMAST markers examined. the same panel of seven NCI markers and seven EMAST markers (Fig. 2; details of our analysis of these tumors will be published elsewhere). In total, 6 cases of MSI-L/EMAST, 27 cases of MSS/EMAST, and These results suggest that a larger sample may be needed to detect 13 cases of MSS/non-EMAST CRCs were stained with a rabbit anti- significant MSI in MSH3-negative cells at [CA]n repeat loci because MSH3 polyclonal antibody (28). As a positive control for MSH3 of the low level of instability at these loci. A significant effect of negativity, we stained CRCs known to be negative for MSH2 MSH3 deficiency was detected at MYCL1(P = 0.0008) and D19S394 expression, because MSH3 is unstable and not detected in these (P = 0.006) and the three EMAST loci combined (P < 0.00001). tumors. By immunostaining, we could readily distinguish between Taken together, the results indicate that loss of MSH3 results in a MSH3-positive and MSH3-negative cells. In the majority of high degree of instability at EMAST loci and in a lower degree of EMAST cases, we observed patchy staining for MSH3. Although instability at with [CA]n repeats. MSH3-negative cells tend to be seen in clusters, MSH3-positive High incidence of EMAST tumors in human CRC. The above cells are found adjacent to MSH3-negative cells (Fig. 3). The results suggest that some MSI-L in CRC may be due to MSH3 sporadic and heterogeneous MSH3 expression in EMAST cases deficiency. If this is the case, our results predict that these MSI-L is quite different from what was observed in MSH2-negative tumors will exhibit MSI at EMAST loci. To test this, we performed cases, where loss of MSH3 is clonal and homogeneous. Because experiments to determine whether MSI at EMAST loci is common there were obvious differences in the number of MSH3-negative in clinical CRC specimens and examined its relationship to MSI-H, cells among our cases, we determined the percentage of MSH3- MSI-L, and MSS status defined by consensus NCI microsatellite negative cells in a given sample by counting f6,000 tumor cell markers. nuclei. A significant difference in the presence of MSH3-negative Paired normal and tumor tissues from 117 unselected sporadic cells was found between EMAST and non-EMAST tumors CRC cases were collected at the UCSD. DNA from these samples when the average percentage of MSH3-negative cells for EMAST was analyzed for MSI using five consensus NCI microsatellite (31.5%; 95% confidence interval, 29.3–33.7) and the average markers (BAT25, BAT26, D2S123, D5S346, D17S250) plus D18S64 percentage of MSH3-negative cells for non-EMAST (8.4%; 95% and D18S69. Analysis identified 16 MSI-H cases (13.7%), 19 MSI-L confidence interval, 4.8–11.9) were compared using Student’s t cases (16.2%), and 82 MSS cases (70.1%; Table 3; Supplementary Fig. test (P < 0.0001). This shows that EMAST tumors contain sig- S1). We analyzed all cases for MSI at the seven EMAST loci, nificantly more MSH3-negative cells than do non-EMAST tumors. including MYCL1, D20S82, D20S85, L17835, D8S321, D9S242, and Based on the above, we defined tumors as MSH3-positive if MSH3- D19S394. Tumors showing MSI in at least one EMAST locus were negative cells occupy <15% of all tumor cells. If MSH3-negative cells categorized as MSI-H/EMAST, MSI-L/EMAST, or MSS/EMAST. MSS occupy >15% of the cellular population, the tumors were defined as tumors that did not show MSI at any EMAST markers were defined MSH3-negative. as MSS/non-EMAST. Seventy of the 117 CRC cases (60%) exhibited We next analyzed all 163 CRC cases for frameshift mutations at mutations in at least one EMAST locus (Table 3). All MSI-H cases exon 7 of the hMSH3 gene by fragment analysis and found no (16 cases) and MSI-L cases (19 cases) and 35 of 82 MSS cases were evidence of mutations, except in five MSI-H tumors (Fig. 2 and positive for EMAST, indicating that this mutational signature is Supplementary Fig. S1). These results were expected because common in CRC. MSI-L is always associated with EMAST in CRC. frameshift mutations in the [A]8 repeat are secondary to a This was found to be the case for the MSH3-deficient cell line previously occurring defect in the hMLH1 gene (28). described above (Tables 1 and 2). These results support the idea Despite a significant association between the quantity of MSH3- that MSI-L may be caused by MSH3 deficiency. It has been reported negative cells and the EMAST phenotype, it is not clear how that many, perhaps most, MSS CRC tumors are MSI-L (29, 30). MSH3 negativity relates to EMAST in CRC tissues. As mentioned Therefore, it is possible that the majority of the MSS/EMAST cases above, in most EMAST tumors, MSH3-positive and MSH3-negative found in this study could be identified as MSI-L/EMAST if cells are interspersed (Fig. 3). A possible explanation for this is additional dinucleotide markers were examined. If this is the case, that MSH3 negativity is a reversible trait. If this is the case, we 46% of CRC (19 cases of MSI-L/EMAST and 35 case of MSS/ may find some MSH3-positive cells exhibiting EMAST. To test this EMAST) could be affected by MSH3 dificiency. possibility, three cases of EMAST/MSH3-negative CRCs (samples In MSI-H/EMAST cases, the high frequency of mutations at 92, 113, and 118) and four cases of non–EMAST/MSH3-positive EMAST loci can be entirely explained by MLH1 deficiency. As all CRCs (samples CR058, CR337, CR067, and 115) were analyzed. are sporadic CRCs, it is likely that the MSI-H cases contain EMAST profiles were examined from two to three locations within hypermethylated hMLH1 alleles (27). Moreover, the results of the each tumor. Three EMAST/MSH3-negative tumors were found to www.aacrjournals.org 8469 Cancer Res 2008; 68: (20). October 15, 2008

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. Cancer Research contain subregions with a high percentage of MSH3-negative These results support the concept that down-regulation of cells (31–79%), as well as a subregion with a high percentage of MSH3 may be reversible. We also analyzed four cases of non– MSH3-positive cells (90–96%; Supplementary Fig. S2). DNA was EMAST/MSH3-positive CRC for the presence of local EMAST. In all isolated from each of these regions and subjected to EMAST cases except CR058, tissues were uniformly occupied by a high analysis. In all three cases, MSI was detected in the MSH3- level of MSH3-positive cells (90–99%). Three randomly chosen negative regions, as well as in the MSH3-positive regions regions of CR337, CR067, and 115 tumors were analyzed for (Supplementary Fig. S2). The same or similar loci were affected EMAST. MSI was not detected in any of the subregions of these in both the positive and negative regions, indicating that the three tumors (Supplementary Fig. S2). In contrast, tumor CR058 tumor cells residing in each region shared the same EMAST was found to contain an MSH3-negative subregion and MSH3- ancestor cells. positive regions. MSI was detected in MSH3-negative regions, but

Figure 2. Association of the EMAST phenotype with down-regulation of MSH3 in CRC.This figure provides detailed data from the 46 CRCs analyzed for expression of MSH3 by IHC. Columns, MSI data for two markers with mono-A repeats (B25 and B26), five markers with CA repeats (S123 through S64), seven EMAST markers (MYCL1 through S321), MSI status at NCI markers (NCI), EMAST status, IHC for MSH3 protein expression, and analysis for frameshift mutations in [A]8 in exon 7 of the hMSH3 locus.For MSI data, a solid box indicates the presence of a mutation.For MSI using the NCI panel, L indicates MSI-L and S indicates MSS.For EMAST status, E indicates EMAST positive and non-E indicates EMAST negative.For MSH3 IHC, the darker gray boxes represent MSH3-negative wit h >15% of cells down-regulating MSH3, and lighter gray boxes represent MSH3-positive with fewer than 15% of cells down-regulating MSH3.The numbers wi thin the boxes represent the percentage of MSH3-negative cells in each tissue.For MSH3 frameshift mutation, N represents no mutations detected.Abbrevi ations used for each marker are as follows: B25, BAT25; B26, BAT26; S123, D2S123; S346, D5S346; S250, D17S250; S69, D18S69; S64, D18S64; S82, D20S82; S242, D9S242; S85, D20S85; S394, D19S394; S321, D8S321.

Cancer Res 2008; 68: (20). October 15, 2008 8470 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. MSI-L/EMAST Associated with MSH3 Deficiency in CRC

Figure 3. MSH3 expression in non-EMAST and EMAST CRCs. A1 (Â100) and A2 (Â400), a non-EMAST CRC with homogenous MSH3 expression (brown). B1 (Â100) and B2 (Â400), an EMAST CRC with heterogeneous MSH3 expression. Arrows, MSH3-negative (blue) region.

not in the MSH3-positive regions of this tumor (Supplementary outcomes (34, 36–38), raising the possibility that MSH3 deficiency Fig. S2). Taken together, these results suggest the following: in might contribute to metastasis in CRC. Certain genes containing EMAST/MSH3-negative tumors, down-regulation of MSH3 expres- dinucleotide, trinucleotide, or tetranucleotide repeats in their 5¶ sion and its consequence, the appearance of EMAST, may take promoter, coding, and/or 3¶ untranslated regions could be place in the tumor progenitor cell before or during the latest clonal mutational targets of MSH3 deficiency, whose mutation might expansion. This is followed by reexpression of MSH3 in some contribute to initiation or progression of tumors (38, 39). progeny cells. Thus, MSH3-negative and MSH3-positive progeny MutSh is involved in repairing interstrand cross-links (40). Our have an MSI signature similar to that of their ancestor cells. In preliminary experiments using the MSH3-positive and MSH3- non–EMAST/MSH3-positive tumors, a tumor progenitor cell negative cells generated in this study indicate that MSH3-deficient would not have experienced down-regulation of MSH3 before the cells are more sensitive to cisplatin treatment. Information last clonal expansion. The presence of the local EMAST cells found obtained from this line of study may contribute to the rationale in CR058 could be explained by down-regulation of MSH3 in local for platinum-based adjuvant treatment for CRC with and without progeny cells. MSH3-negative tumor cells. What might be the mechanistic basis for reversible expression of The existence and significance of MSI-L CRCs have been the MSH3? It has been reported that hypoxia down-regulates MMR subject of debate, mainly due to our lackof knowledgeof the proteins in human tumor cell lines (31). Our preliminary experi- molecular features which distinguish them from MSI-H or MSS ments show that hypoxia down-regulates MSH3 expression and CRCs (2, 12) and the ambiguity of the definition of MSI-L (30). We reoxygenation induces reexpression of MSH3 in tissue cultured have found that MSI-L CRCs and a substantial number of MSS cells (data not shown). Furthermore, overexpression of glucose CRCs (f43%) were MSH3-negative EMAST tumors. Therefore, our transporter-1, an indicator of hypoxic response (32), correlates with results raise a possibility that sporadic non–MSI-H CRCs could be MSH3 negativity and EMAST in the CRC tissues examined in this recategorized as MSH3-negative EMAST tumors or MSH3-positive study. An intensive investigation to determine the exact roles of non-EMAST tumors. Further studies are needed to establish the hypoxia and reoxygenation for MSH3 expression and EMAST is validity of this classification. under way in our laboratory. Overall, our data show that loss of MSH3 is associated with Possible effect of MSH3 deficency on CRC. A mouse model of instability at EMAST loci and low levels of instability at micro- MSH3-deficiency showed that the MSH3 defect might cause late satellite loci with dinucleotide repeats in both cell lines and CRC onset MSI-positive gastrointestinal cancers (33), suggesting that tissues. Further studies are needed to determine the pathologic MSH3 deficiency could contribute to tumor initiation. Families of significance of MSH3 deficiency on CRC formation. late onset CRC with MSI-L or MSS exist (34). Therefore, it would be interesting to determine whether these families carry MSH3 germline mutations. MSH3 deficiency is also associated with Disclosure of Potential Conflicts of Interest progression of disease (28, 35). MSI-L CRCs have poor clinical No potential conflicts of interest were disclosed. www.aacrjournals.org 8471 Cancer Res 2008; 68: (20). October 15, 2008

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. Cancer Research

Acknowledgments Research of the Japan Society for the Promotion of Science 16591358 (H. Hemmi), NIH grants R01 CA72851, and funds from Baylor University Medical Center (C.R. Boland). Received 1/2/2008; revised 7/8/2008; accepted 7/30/2008. The costs of publication of this article were defrayed in part by the payment of page Grant support: NIH intramural grant 1Z01ES023016-05, Association for Interna- charges. This article must therefore be hereby marked advertisement in accordance tional Cancer Research project grant 03-103 (M. Koi), grant-in-aid for Scientific with 18 U.S.C. Section 1734 solely to indicate this fact.

References 14. Modrich P. Mechanisms in eukaryotic mismatch 28. Plaschke J, Kru¨ger S, Jeske B, et al. Loss of MSH3 repair. J Biol Chem 2006;281:30305–9. protein expression is frequent in MLH1-deficient 1. Dietmaier W, Wallinger S, Bocker T, Kullmann F, Fishel 15. Jiricny J. The multifaceted mismatch-repair system. colorectal cancer and is associated with disease R, Ru¨schoff J. Diagnostic microsatellite instability: Nat Rev Mol Cell Biol 2006;7:335–46. progression. Cancer Res 2004;64:864–70. definition and correlation with mismatch repair protein 16. Kolodner RD, Marsischky GT. Eukaryotic DNA 29. Halford S, Sasieni P, Rowan A, et al. Low-level expression. Cancer Res 1997;57:4749–56. mismatch repair. Curr Opin Genet Dev 1999;9:89–96. microsatellite instability occurs in most colorectal 2. Boland CR, Thibodeau SN, Hamilton SR, et al. A 17. Marsischky GT, Filosi N, Kane MF, Kolodner R. cancers and is a nonrandomly distributed quantitative National Cancer Institute Workshop on Microsatellite Redundancy of Saccharomyces cerevisiae MSH3 and trait. Cancer Res 2002;62:53–7. Instability for cancer detection and familial predisposi- MSH6 in MSH2-dependent mismatch repair. Genes Dev 30. Laiho P, Launonen V, Lahermo P, et al. Low-level tion: development of international criteria for the 1996;10:407–20. microsatellite instability in most colorectal carcinomas. determination of microsatellite instability in colorectal 18. Genschel J, Littman SJ, Drummond JT, Modrich P. Cancer Res 2002;62:1166–70. cancer. Cancer Res 1998;58:5248–57. Isolation of MutSh from human cells and comparison of 31. Bindra RS, Crosby ME, Glazer PM. Regulation of DNA 3. Lagerstedt Robinson L, Liu T, Vandrovcova J, et al. the mismatch repair specificities of MutSh and MutSa. repair in hypoxic cancer cells. Cancer Metastasis Rev Lynch syndrome (hereditary nonpolyposis colorectal J Biol Chem 1998;273:19895–901. 2007;26:249–60. cancer) diagnostics. J Natl Cancer Inst 2007;99:291–9. 19. Sia EAS, Kokoska J, Dominska M, Greenwell P, Petes 32. Koukourakis MI, Giatromanolaki A, Harris A, Sivridis 4. Herman JG, Umar A, PolyakK, et al. Incidence and TD. Microsatellite instability in yeast: dependence on E. Comparison of metabolic pathways between cancer functional consequences of hMLH1 promoter hyper- repeat unit size and DNA mismatch repair genes. Mol cells and stromal cells in colorectal carcinomas: a methylation in colorectal carcinoma. Proc Natl Acad Sci Cell Biol 1997;17:2851–8. metabolic survival role for tumor-associated stroma. U S A 1998;95:6870–5. 20. Acharya S, Wilson T, Gradia S, et al. hMSH2 forms Cancer Res 2006;66:632–7. 5. Iino H, Jass JR, Simms LA, et al. DNA microsatellite specific mispair-binding complexes with hMSH3 and 33. Edelmann W, Umar A, Yang K, et al. The DNA instability in hyperplastic polyps, serrated adenomas, hMSH6. Proc Natl Acad Sci U S A 1996;93:13629–34. mismatch repair genes Msh3 and Msh6 cooperate in and mixed polyps: a mild mutator pathway for 21. Palombo F, Iaccarino I, Nakajima E, Ikejima M, intestinal tumor suppression. Cancer Res 2000;60: colorectal cancer? J Clin Pathol 1999;52:5–9. Shimada T, Jiricny J. hMutSh, a heterodimer of hMSH2 803–7. 6. Mao L, Lee DJ, Tockman MS, Erozan YS, Askin F, and hMSH3, binds to insertion/deletion loops in DNA. 34. Kets CM, Hoogerbrugge N, Bodmer D, et al. Sidransky D. Microsatellite alterations as clonal markers Curr Biol 1996;6:1181–4. Unfavorable pathological characteristics in familial for the detection of human cancer. Proc Natl Acad Sci 22. Risinger JI, Umar A, Boyd J, BerchuckA, KunkelTA, colorectal cancer with low-level microsatellite instabil- U S A 1994;91:9871–5. Barrett JC. Mutation of MSH3 in endometrial cancer and ity. Mod Pathol 2006;19:1624–30. 7. Ahrendt SA, Chow JT, Xu L-H, et al. Molecular evidence for its functional role in heteroduplex repair. 35. Kawakami T, Shiina H, Igawa M, et al. Inactivation of detection of tumor cells in bronchoalveolar lavage fluid Nat Genet 1996;14:102–5. the hMSH3 mismatch repair gene in bladder cancer. from patients with early stage lung cancer. J Natl Cancer 23. Umar A, Risinger JI, Glaab WE, Tindall KR, Biochem Biophys Res Commun 2004;325:934–42. Inst 1999;91:332–9. Barrett JC, Kunkel TA. Functional overlap in mis- 36. Kohonen-Corish MR, Daniel JJ, Chan C, et al. Low 8. Xu L, Chow J, Bonacum J, et al. Microsatellite match repair by human MSH3 and MSH6. Genetics microsatellite instability is associated with poor prog- instability at AAAG repeat sequences in respiratory 1998;148:1637–46. nosis in stage C colon cancer. J Clin Oncol 2005;23: tract cancers. Int J Cancer 2001;91:200–4. 24. Koi M, Umar A, Chauhan DP, et al. Human 2318–24. 9. Danaee H, Nelson HH, Karagas MR, et al. Micro- chromosome 3 corrects mismatch repair deficiency 37. Wright CM, Dent OF, Newland RC, et al. Low level satellite instability at tetranucleotide repeats in skin and and microsatellite instability and reduces N-methyl-N¶- microsatellite instability may be associated with re- bladder cancer. Oncogene 2002;21:4894–9. nitro-N-nitrosoguanidine tolerance in colon tumor cells duced cancer specific survival in sporadic stage C 10. Catto JW, Azzouzi AR, Amira N, et al. Distinct with homozygous hMLH1 mutation. Cancer Res 1994;54: colorectal carcinoma. Gut 2005;54:103–8. patterns of microsatellite instability are seen in tumours 4308–12. 38. Iglesias D, Fernandez-Peralta AM, Nejda N, et al. of the urinary tract. Oncogene 2003;22:8699–706. 25. Kugoh H, Mitsuya K, Shigenami K, Schulz TC, RIS1, a gene with trinucleotide repeats, is a target in the 11. Singer G, Kallinowski T, Hartmann A, et al. Different Oshimura M. Mouse A9 cells containing single human mutator pathway of colorectal carcinogenesis. Cancer types of microsatellite instability in ovarian carcinoma. for analysis of genomic imprinting. DNA Genet Cytogenet 2006;167:138–44. Int J Cancer 2004;112:643–6. Res 1999;6:165–72. 39. Yamashita K, Arimura Y, Kurokawa S, et al. 12. Umar A, Boland CR, Terdiman JP, et al. Revised 26. Lin Y, Dion V, Wilson JH. Transcription promotes Microsatellite instability in patients with multiple pri- Bethesda Guidelines for hereditary nonpolyposis colo- contraction of CAG repeat tract in human cells. Nat mary cancers of the gastrointestinal tract. Gut 2000; rectal cancer (Lynch syndrome) and microsatellite Struct Mol Biol 2006;13:179–80. 46:790–4. instability. J Natl Cancer Inst 2004;96:261–8. 27. Goel A, Nagasaka T, Arnold CN, et al. The CpG island 40. Zhang N, Lu X, Zhang X, Peterson CA, Legerski RJ. 13. Lyer RR, PluciennikA, Burdett V, Modrich P. DNA methylator phenotype and chromosomal instability are hMutSh is required for the recognition and uncoupling mismatch repair: functions and mechanisms. Chem Rev inversely correlated in sporadic colorectal cancer. of psoralen interstrand cross-links in vitro. Mol Cell Biol 2006;106:302–23. Gastroenterology 2007;132:127–38. 2002;22:2388–97.

Cancer Res 2008; 68: (20). October 15, 2008 8472 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research. Genetic Instability Caused by Loss of MutS Homologue 3 in Human Colorectal Cancer

Astrid C. Haugen, Ajay Goel, Kanae Yamada, et al.

Cancer Res 2008;68:8465-8472.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/68/20/8465

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2008/10/07/68.20.8465.DC1

Cited articles This article cites 40 articles, 21 of which you can access for free at: http://cancerres.aacrjournals.org/content/68/20/8465.full#ref-list-1

Citing articles This article has been cited by 15 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/68/20/8465.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/68/20/8465. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2008 American Association for Cancer Research.