Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Cancer Research Molecular and Cellular Pathobiology Colorectal Cancer Cell Lines Are Representative Models of the Main Molecular Subtypes of Primary Cancer Dmitri Mouradov1,2,4, Clare Sloggett5, Robert N. Jorissen1,2,4, Christopher G. Love1,2,4, Shan Li1,2, Antony W. Burgess3,4, Diego Arango8,9, Robert L. Strausberg10,11, Daniel Buchanan7, Samuel Wormald2,4, Liam O'Connor2,4, Jennifer L. Wilding12, David Bicknell12, Ian P.M. Tomlinson13, Walter F. Bodmer12, John M. Mariadason6, and Oliver M. Sieber1,2,4 Abstract Human colorectal cancer cell lines are used widely to investigate tumor biology, experimental therapy, and biomarkers. However, to what extent these established cell lines represent and maintain the genetic diversity of primary cancers is uncertain. In this study, we profiled 70 colorectal cancer cell lines for mutations and DNA copy number by whole-exome sequencing and SNP microarray analyses, respectively. Gene expression was defined using RNA-Seq. Cell line data were compared with those published for primary colorectal cancers in The Cancer Genome Atlas. Notably, we found that exome mutation and DNA copy-number spectra in colorectal cancer cell lines closely resembled those seen in primary colorectal tumors. Similarities included the presence of two hypermutation phenotypes, as defined by signatures for defective DNA mismatch repair and DNA polymerase e proofreading deficiency, along with concordant mutation profiles in the broadly altered WNT, MAPK, PI3K, TGFb, and p53 pathways. Furthermore, we documented mutations enriched in genes involved in chromatin remodeling (ARID1A, CHD6, and SRCAP) and histone methylation or acetylation (ASH1L, EP300, EP400, MLL2, MLL3, PRDM2, and TRRAP). Chromosomal instability was prevalent in nonhypermutated cases, with similar patterns of chromosomal gains and losses. Although paired cell lines derived from the same tumor exhibited considerable mutation and DNA copy-number differences, in silico simulations suggest that these differences mainly reflected a preexisting heterogeneity in the tumor cells. In conclusion, our results establish that human colorectal cancer lines are representative of the main subtypes of primary tumors at the genomic level, further validating their utility as tools to investigate colorectal cancer biology and drug responses. Cancer Res; 74(12); 1–10. 2014 AACR. Introduction Colorectal cancer is a leading cause of cancer-related morbidity and mortality (1). Human colorectal cancer cell lines are an important, commonly used preclinical model system for studying this disease, and have provided essential insights into tumor molecular and cell biology. Cell lines are a fundamental tool used in the discovery of new antitumor compounds and for the discovery of drug sensitivity, resisAuthors' Affiliations: 1Ludwig Institute for Cancer Research; 2Systems Biology and Personalised Medicine Division; 3Structural Biology Division; Walter and Eliza Hall Institute of Medical Research; 4Faculty of Medicine, Dentistry and Health Sciences, Department of Medical Biology, University of Melbourne, Parkville; 5VLSCI Life Sciences Computation Centre, a collaboration between Melbourne, Monash and LaTrobe Universities, c/o The University of Melbourne, Carlton; 6Oncogenic Transcription Laboratory, Ludwig Institute for Cancer Research, Austin, VIC, Australia; 7 Cancer and Population Studies Group, Queensland Institute of Medical Research, Herston, QLD, Australia; 8Group of Molecular Oncology, CIBBIM-Nanomedicine, Vall d'Hebron University Hospital, Research Institute noma de Barcelona, Passeig Vall d'Hebron, 119(VHIR), Universitat Auto 129, 08035 Barcelona; 9CIBER de Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN); Spain; 10Ludwig Collaborative Laboratory for Cancer Biology and Therapy, Department of Neurosurgery, Johns Hopkins University School of Medicine, Baltimore, Maryland; 11Ludwig Institute for Cancer Research Ltd., New York, New York; 12Cancer and Immunogenetics Laboratory, Weatherall Institute of Molecular Medicine (WIMM), John Radcliffe Hospital, University of Oxford; and 13Molecular and Pop- tance, and toxicity biomarkers, with molecular markers of response to conventional chemotherapies and targeted agents showing clinical utility in patients (2–7). Examples include the relationship between tumor microsatellite instability-high (MSI-H) status and lack of 5-fluorouracil (5FU) response (2), and mutations in KRAS, BRAF, and PIK3CA exon 20 and resistance to anti-EGFR antibody therapy (3). However, to what extent colorectal cancer cell lines ulation Genetics Laboratory, Wellcome Trust Centre for Human Genetics, Oxford, United Kingdom Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). Corresponding Authors: John Mariadason, Oncogenic Transcription Laboratory, Ludwig Institute for Cancer Research Ltd., Austin Hospital, Level 5, Olivia Newton-John Cancer & Wellness Centre, Studley Road, Heidelberg, VIC 3084, Australia. Phone: 61-3-9496-3068; Fax: 61-3-94965334. E-mail: [email protected]; and Oliver Sieber, Colorectal Cancer Genetics Laboratory, Systems Biology and Personalised Medicine Division, Walter and Eliza Hall Institute of Medial Research, 1G Royal Parade, Parkville, VIC 3052, Australia. Phone: 61-3-9345-2885; Fax: 61-3-9347-0852; E-mail: [email protected] doi: 10.1158/0008-5472.CAN-14-0013 2014 American Association for Cancer Research. www.aacrjournals.org Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. OF1 Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Mouradov et al. represent and maintain the genetic diversity of primary cancers remains controversial. More than the past two decades, major cancer genes and pathways central to colorectal cancer development have been delineated, including the WNT, MAPK, PI3K, TGFb, and p53 pathways. Two broad molecular subtypes of colorectal cancer have emerged, characterized by MSI-H (15%) or chromosomal instability (CIN, 60%; refs. 8–10). MSI-H colorectal cancers occur predominantly in the proximal colon, and often show poor differentiation, mucinous histology, and increased peritumoral lymphocytic infiltration (11). These tumors exhibit hypermutation caused by defective DNA mismatch repair (MMR), tend to be near-diploid and to have a CpG island methylator phenotype (CIMP), and harbor mutations in a distinct set of driver genes, including BRAF, PTEN, and TGFBR2 (12–14). In contrast, chromosomally unstable tumors are more common in the distal colorectum and tend to develop along the classical genetic pathway of colorectal tumorigenesis, with mutations in APC, KRAS, SMAD4, and TP53 (15). Recent cancer genome sequencing studies have revealed additional details of the genetic profiles of human colorectal cancer, highlighting their diversity. An initial whole-exome sequencing study on 11 microsatellite stable (MSS) colorectal cancers demonstrated that such tumors harbor 80 coding sequence mutations with a small number of commonly mutated driver genes and a large number of "private" mutations (16). Subsequently, The Cancer Genome Atlas (TCGA) Network reported comprehensive data on 223 unselected sporadic colorectal cancers (17). Hypermutation was identified in 15% of carcinomas, with three quarters of these displaying the expected MSI-high (MSI-H) phenotype associated with epigenetic silencing and/or mutation of MMR genes. However, one-quarter of hypermutated tumors did not show MSI-H and seemed to be associated with DNA polymerase e (POLE) mutations, perhaps representing a distinct colorectal cancer subtype. Twenty-four genes were identified as significantly mutated in colorectal cancer, including several novel candidate genes such as ATM, ARID1A, TCF7L2, SOX9, and FAM123B. A number of recurrent DNA copy-number alterations were reported comprising potentially drug-targetable amplifications of ERBB2 and IGF2. A similar study on 74 pairs of primary human colon tumors reported highly concordant results, and in addition identified low frequency fusion transcripts involving R-spondin family members RSPO2 and RSPO3 (18). Here, we report the first comprehensive whole-exome and DNA copy-number analyses of 70 of the most widely used colorectal cancer cell lines, and undertake a detailed comparison of genetic alterations with those reported for TCGAanalyzed primary cancers. We demonstrate that the spectra of mutations and DNA copy-number aberrations in colorectal cancer cell lines are representative of primary tumors, including hypermutation phenotypes and targeting of major signaling pathways. Our data further highlight novel aspects of colorectal cancer biology, including significant enrichment of mutated genes involved in chromatin remodeling and histone methylation or acetylation. Our results verify established colorectal cancer cell lines as a useful preclinical model system, and provide a comprehensive genomic data resource enabling OF2 Cancer Res; 74(12) June 15, 2014 informed choices when selecting cell lines for functional and pharmacogenomics research. Materials and Methods Colorectal cancer cell lines and TCGA-analyzed primary cancers The 70 colorectal cancer cell lines studied were obtained from a range of sources, listed below, over a period spanning several years (Supplementary Table S1). C10, C106, C125, C135, C32, C70, C80, C84, and C99 were generated by the W.F. Bodmer laboratory, CACO2, COLO201, COLO205, COLO320-DM, DLD1, HCC2998, HCT116, HCT15, HCT8, HT29, LOVO, LS174T, LS180, LS411, LS513, NCI-H716, NCI-H747, RKO, SKCO-1, SNU-C1, SNU-C2B, SW1116, SW1417, SW1463, SW403, SW48, SW480, SW620, SW837, SW948, and T84 were obtained from the American Type Culture Collection, KM12 was obtained from DCTD, COLO678, and CX-1 were obtained from DSMZ, Gp2D, Gp5D, HT115, and HT55 were obtained from ECACC, CCK81, and CoCM-1 were obtained from HSRRB, VACO10, and VACO4S were provided by Dr. J.A. McBain (University of Wisconsin School of Medicine, Madison, WI), SNU-175, SNU283, and SNU-C4 were obtained from KCLB, LIM1215, LIM1899, LIM2099, LIM2405, and LIM2551 were obtained from the Ludwig Institute for Cancer Research, SW1222 were provided by Prof. M. Herlyn (The Wistar Institute, Philadelphia, PA), HDC101, HDC54, HDC82, HDC87, and HDC90 were provided by Prof. M, Schwab (DKFZ, Germany), HCA46, HCA7, and HRA19 were provided by Dr. S.C. Kirkland (Royal Postgraduate Medical School, United Kingdom), and RW2982 and RW7213 were provided by Dr. P. Calabresi (Roger Williams General Hospital, Providence, RI). Cells were cultured with Dulbecco's Modified Eagle Medium and 10% FBS at 37 C and 5% CO2. Cell line authentication was performed using the Promega StemElite ID System at the Queensland Institute of Medical Research (QMIR, Queensland, Australia) DNA Sequencing and Fragment Analysis Facility (January 2013). Cells were tested for mycoplasma by the MycoAlert Mycoplasma Detection Kit (Lonza). Published exome-capture sequencing data on 223 patients with colorectal cancer were retrieved from the TCGA (19). SNP array data were available for 213 of these patients. Mutation data were filtered for exons and at splice sites (3 bp). Exome-capture sequencing Exome-capture was performed using the TruSeq Exome Enrichment Kit (Illumina) and 100 bp paired-end read sequencing performed on an Illumina HiSeq 2000 at the Australian Genome Research Facility (AGRF). Variant detection was implemented using a modified GATK Best Practice protocol; variants were filtered against known germline variations and those detected in 114 in-house normal colorectal tissues (Supplementary Methods). Regions of known germline chromosomal segmental duplications were excluded (20). Transcriptome sequencing RNA-Seq analysis was performed at the AGRF on an Illumina HiSeq2000 to a depth of >100 million paired reads. Alignment to the human reference genome (hg19) was achieved using Cancer Research Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Genomic Landscape of Colorectal Cancer Cell Lines TopHat (21), and RPKM values calculated. Absence of gene expression was defined as a RPKM value of <1 (Supplementary Methods). SNP microarray analysis SNP assays were performed by the AGRF using Illumina Human610-Quad BeadChips (GSE55832) and processed using GenomeStudio (Illumina). SNPs showing germline alterations, based on 637 normal samples, were excluded. DNA copynumber segmentation was performed using OncoSNPv2.18 (22). Regions of significantly altered genome were identified using GISTIC2.0 (23). Microsatellite instability analysis MSI analysis was performed for the National Cancer Institute recommended microsatellite marker panel BAT25, BAT26, D2S123, D5S346, and D17S250 using fluorescently labeled primers on a 3130xl Genetic Analyzer (Applied Biosystems; ref. 24). MSI-H was diagnosed if instability was evident at 2 or more markers. CIMP marker and MLH1 promoter methylation analysis MethyLight real-time PCR was performed for the Weisenberger and colleagues CIMP marker panel (IGF2, SOCS1, RUNX3, CACNA1G, NGN1), MLH1, and the reference ALU (C-4; ref. 25). The percentage of methylated reference (PMR) was calculated for GENE:ALU ratio of template amount in a sample against GENE:ALU ratio of template amount in methylated reference DNA. Samples with a PMR greater than 10 for 3 CIMP markers were classified as CIMP-positive. Statistical analysis Statistical analyses were conducted using the R statistical computing software (http://www.R-project.org/). Differences between groups were assessed using Fisher exact test or x2 test for categorical variables and the Student t test for continuous variables. Correlation between RNA-Seq and microarray gene expression data were assessed using Spearman rank correlation coefficient. Overlap of gene lists was assessed using a hypergeometric test. GISTIC2.0 analysis for significantly altered focal regions of chromosomal deletion or gain/amplification were adjusted for multiple testing, implementing a BHFDR adjusted Q-value cut-off of 0.05. All statistical analyses were 2-sided and considered significant when P < 0.05. pipeline sensitivity and specificity for nonsilent variants were shown to be 79.2% and 98.6% in an analysis of 10 primary colorectal cancers with and without paired normal tissue, with the majority (93.4%) of false-negative calls resulting from somatic mutations mimicking annotated germline variants (Supplementary Data). Accuracy of mutation calling was verified by Sanger resequencing for 12 selected genes (APC, CTNNB1, KRAS, BRAF, NRAS, PIK3CA, PTEN, SMAD2, SMAD3, SMAD4, FBXW7, and TP53) on 43 cell lines, with validation of 97.6% (165/169) of mutations (Supplementary Table S3). For the 63 unique cell lines, the total number of putative somatic mutations ranged from 219 to 8657 (mean ¼ 1498.3). Similar to the primary colorectal cancers reported by The TCGA (17), the prevalence of mutations varied substantially, ranging from 6.6 to 260.0 per 106 bases, with evidence for nonhypermutated and hypermutated groups (Fig. 1). Cell lines with a mutation prevalence of >25 per 106 bases were designated as hypermutated. Hypermutation was observed in 34.9% (22/63) of cell lines as compared with 15.7% (35/223) of the TCGA cancers (P < 0.001). 86.3% (19/22) of hypermutated cell lines exhibited MSI-H, similar to the proportion found in cancers (80.0%, 28/35, P ¼ 0.725). MSI-H status in cell lines was associated with MLH1 hypermethylation, mutations in MMR genes and presence of CIMP (P < 0.009; Fig. 1). Consistent with the established association between MSI-H and proximal tumor location in colorectal cancer, all MSI-H cell lines with available site details originated from the proximal colon (Supplementary Table S1). There was no evidence of imbalance in site distribution between cell lines and cancers (P ¼ 0.937). Mutation spectra in hypermutated colorectal cancer cell lines Compared with nonhypermutated cell lines, MSI-H cell lines displayed a higher number of InDels (41.1-fold; P < 0.001) and SNVs (9.7-fold; P < 0.001) as observed in the TCGA cancers Results Exome mutation profiles in 70 colorectal cancer cell lines Seventy commonly used colorectal cancer cell lines, comprising 63 unique lines and 7 duplicate lines, were analyzed for mutations in protein-coding exons and exon–intron boundaries by whole-exome capture sequencing (Supplementary Table S2). In all cases, >20-fold coverage of >80% of targeted exons was achieved. As matched normal tissue for these cell lines was not available, putative somatic mutations were identified by annotation against databases of known human germline variants. Regions of known germline chromosomal segmental duplications were excluded to reduce the possibility of false-positive variants caused by read mismapping. Mean www.aacrjournals.org Figure 1. Mutation frequencies in 70 human colorectal cancer cell lines. Cell lines were separated into distinct hypermutated and nonhypermutated cases. MSI-H, microsatellite instability-high; CIMP, CpG island methylator phenotype; MLH1 meth., MLH1 promoter hypermethylation, MMR mut., DNA mismatch repair gene mutation; inset, mutations in MMR genes among the hypermutated samples, with cases in the same order as in the main graph. Cancer Res; 74(12) June 15, 2014 Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. OF3 Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Mouradov et al. Figure 2. Types of mutations for 70 human colorectal cancer cell lines and 223 TCGA-analyzed primary cancers. A–C, counts of SNVs and InDels (A and B), and proportions of nucleotide transitions and transversions (C). Paired cell lines originating from the same tumor are indicated by identical colored stars. MSIH, microsatellite instability-high; NSHP, nucleotide-substitution hypermutator phenotype; POLE EDM, missense mutation in the POLE exonuclease domain. (InDels: 18.3-fold, SNVs: 10.8-fold; P < 0.001; Fig. 2A and B). As expected, InDels in MSI-H cases tended to occur at nucleotiderepeat (N5 or NN3) sequences (MSI-H vs. nonhypermutated: cell lines 82.7% vs. 17.9%, cancers 73.4% vs. 38.4%; P < 0.001), a bias not observed for the SNVs (cell lines 2.9% vs. 3.7%, cancers 2.3% vs. 2.9%; Supplementary Table S2). The SNV spectrum differed significantly between MSI-H and nonhypermutated cases, with an increased proportion of A>G/T>C transitions (cell lines 1.5-fold, cancers 1.8-fold; P < 0.011) and decreased C>G/G>C transversions (cell lines 4.5-fold, cancers 5.2-fold; P < 0.001; Fig. 2C). A second type of hypermutated tumor was observed among cell lines and the TCGA cancers. These cases were MSS and compared with nonhypermutated cases exhibited a nucleotidesubstitution hypermutator phenotype (NSHP) characterized by a substantial increase of SNVs (cell lines 16.8-fold, cancers 56.8fold; P < 0.005; Fig. 2A and B). The SNV spectrum in 2 of 3 NSHP cell lines (HT115, HCC2998) and 7 of 7 NSHP cancers further displayed increased proportions of C>A/G>T transversions (MSI-H vs. nonhypermutated: cell lines 1.5-fold, cancers 2.4fold) and A>C/T>G transversions (cell line 3.8-fold, cancers 2.0fold), as well as decreased C>G/G>C transversions (cell lines 13.6-fold, cancers 22.3-fold; Fig. 2C). Combining cell lines and cancers, this mutator phenotype was significantly associated with the presence of POLE mutations (NSHP: 90.0%, 9/10 vs. MSIH: 29.8%, 14/47 vs. nonhypermutated: 1.3%, 3/229; P < 0.001). In addition, all nine NSHP cases with POLE mutation had at least 1 missense mutation in the POLE exonuclease domain (EDM), as compared with only 1 of 17 non-NSHP cases with POLE mutation (P < 0.001; Supplementary Table S4). 82.4% (14/17) of POLE mutated non-NSHP samples were MSI-H, and there was no evidence that the non-EDM POLE mutations in these cases OF4 Cancer Res; 74(12) June 15, 2014 modified the MSI-H phenotype with similar SNV frequencies and transition/transversion spectra compared with MSI-H cancers without POLE mutation (P > 0.05 for all comparisons). A single NSHP cell line, HT55, exhibited a distinct bias to A>T/T>A transversions (49.0% vs. nonhypermutated mean 4.6%) and was wild type for POLE, but a similar case was not present among the TCGA cancers (Fig. 2C). This cell line harbored a heterozygous truncating mutation in the MMR gene PMS1, which may be related to this mutator phenotype, although Pms1-deficient mice have been reported to show no evidence of tumor predisposition or hypermutation (26). Chromosomal and subchromosomal aberrations DNA copy-number alterations in cell lines were profiled using Illumina Human 610-Quad BeadChips. As anticipated, DNA copy-number profiles were similar between cell lines and primary cancers (Fig. 3A). Hypermutated MSI-H and NSHP cases showed stable profiles with a modal chromosome copynumber of 2n. In contrast, nonhypermutated cases tended to exhibit unstable profiles with modal chromosome copy numbers ranging from 2n to 4n in cell lines and 2n to 6n in cancers. For nonhypermutated groups, the most common deleted chromosome arms were 8p, 17p (including TP53), and 18q (including SMAD4), and the most common gained regions were chromosome 7, 8q (including MYC), 13, and 20q. Some differences were apparent between cell lines and primary cancers, including more frequent chromosome 4 deletions in nonhypermutated and chromosome 7 gains and 18q deletions in hypermutated cell lines. Focal regions significantly targeted by DNA copy-number alterations in both cell lines and cancers were deduced using GISTIC2.0 software (23). Eleven and 4 significant focal regions Cancer Research Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Genomic Landscape of Colorectal Cancer Cell Lines Figure 3. Genome-wide DNA copynumber aberrations in 63 unique colorectal cancer cell lines and 213 TCGA primary cancers stratified into nonhypermutated and hypermutated cases. A, absolute DNA copy-number gains and losses relative to diploid status (2n) deduced using oncoSNP. B and C, minimal significant regions of chromosomal loss (blue) and gain/ amplification (red) deduced using GISTIC2.0. Selected candidate genes in overlapping regions are indicated and total numbers of genes are shown in brackets. were found to overlap between nonhypermutated and hypermutated groups, respectively (Fig. 3B and C and Supplementary Table S5). For nonhypermutated cases, overlapping regions included gain/amplification of MYC, a known key mediator of colorectal tumorigenesis (27), and deletion of the candidate cancer genes PARK2 and WRN, detected in 29.5% and 36.3% of cell lines and 21.9% and 31.6% of primary cancers, respectively. The 4 recurrent regions identified in hypermutated cases—which were also present in the nonhypermutated group—were all known fragile sites (including FHIT, A2BP1, WWOX, or MACROD2), the functional relevance of which is uncertain. Cancer genes and pathways Gene mutation profiles, excluding silent mutations, were compared between colorectal cancer cell lines and primary www.aacrjournals.org cancers. Only genes with well-defined expression in cell lines [reads per kilobase per million reads mapped (RPKM) >1 in at least 3/13 colorectal cancer cell lines analyzed by RNA-Seq] were considered (n ¼ 20,702 genes; Supplementary Table S6). NSHP cases were excluded from this comparison because of the small sample size. Mutation landscapes of cell lines markedly resembled those of primary cancers. In both cohorts, nonhypermutated cases displayed a small number of distinct mutation peaks, with APC, TP53, and KRAS being the most common targets, whereas hypermutated MSI-H cases showed frequent mutations in multiple genes, with the anticipated bias to genes comprising nucleotide repeats (Supplementary Tables S7). Significant overlap was observed for the top 5% of mutated genes (based on the proportion of affected samples), with 54 and 62 genes intersecting for nonhypermutated and MSI-H groups, respectively Cancer Res; 74(12) June 15, 2014 Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. OF5 Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Mouradov et al. Integrating mutation and DNA copy-number data across cell lines and primary cancers showed the anticipated tumor suppressor signatures for APC, TP53, SMAD4, and SOX9 with a significant overrepresentation of 2 mutational hits (2þ mutations or 1 mutation and loss of heterozygosity; P < 0.023). There was further an association of mutation in BRAF, ERBB3, PIK3CA, and KRAS with copy-number gain of 5 at the respective chromosomal regions (P < 0.018). Trends to mutual exclusivity between pathway member mutations were observed for KRAS and BRAF, TP53, and ATM (P < 0.005). Figure 4. Mutation frequencies for WNT, MAPK, PI3K, TGFb, and p53 pathway members as well as chromatin regulators identified among the top 5% of mutated genes overlapping between colorectal cancer cell lines and TCGA-analyzed primary cancers for nonhypermutated or hypermutated MSI-H samples. (P < 0.001; Supplementary Tables S8). Overlapping genes included the expected members of the WNT, MAPK, PI3K, TGFb, and p53 pathways. For nonhypermutated cases, these comprised APC, CTNNB1, FBXW7, KRAS, BRAF, PIK3CA, SMAD4, and TP53, as well as the candidate cancer genes, AXIN2, BCL9L, FAT1, SOX9, ERBB3, PIK3C2B, TIAM1, and ATM. For MSI-H cases these encompassed APC, FBXW7, PIK3CA, and TGFBR2, along with the candidates CREBBP, TCF7L2, RNF43, and ACVR2A. Mutation distributions across colorectal cancerassociated signaling pathways were also similar between nonhypermutated cell lines and cancers, with the same pathway members tending to show the highest mutation frequencies (Fig. 4). Greater variability in mutation frequencies between pathway members was observed for MSI-H cases, at least partly related to the higher mutation background and smaller sample size. A notable difference for MSI-H cases was a differential prevalence of CTNNB1 mutations, which were frequent in cell lines (47%) but not reported in primary cancers (P < 0.001). In addition to the established colorectal cancer-associated pathways, significant enrichment was observed for mutations in chromatin-state regulators (P < 0.001; gene ontology enrichment analysis). These comprised ASH1L, CHD6, PRDM2, and MLL3 in nonhypermutated and ARID1A, EP300, EP400, MLL2, CHD6, PRDM2, TRRAP, and SRCAP in MSI-H cases (Supplementary Table S8). Approximately 49% and 19% of nonhypermutated and 100% and 93% of MSI-H cell lines and primary cancers carried mutations across these candidates, respectively. OF6 Cancer Res; 74(12) June 15, 2014 Mutation and DNA copy-number differences in paired colorectal cancer cell lines Included in our colorectal cancer cell line panel were 5 pairs/ triplets originally derived from the same tumor (COLO201/ COLO205, CX-1/HT29, Gp2D/Gp5D, LS174T/LS180, and DLD1/HCT8/HCT15), and 1 pair derived from a primary tumor and subsequent lymph node metastasis (SW480/SW620). LS174T and LS180 were established by trypsin treatment and scraping of primary cultures from the same tumor, respectively (28), and have been shown to differ with respect to E-cadherin expression and cell adhesion, with LS174T displaying complete loss of E-cadherin protein (2, 29). Assessment of the overlap between mutations detected in paired cell lines identified substantial discrepancies, with 63 and 356 mutational differences in the nonhypermutated pairs COLO201/COLO205 and CX-1/HT29, and 2,763, 480, 6,503, 5,377, and 6,369 mutational differences in the MSI-H pairs Gp2D/Gp5D, LS174T/LS180, DLD1/HCT8, DLD1/HCT15, and HCT8/HCT15 (Fig. 5). Eight hundred and forty-nine mutations differed in the nonhypermutated primary/metastasis pair SW480/SW620. Nonsilent and silent mutations contributed in similar proportions to these discrepancies (54.4% vs. 56.6%), suggesting no selection for these differential alterations. DNA copy-number profiles showed multiple differences for nonhypermutated pairs, with 41.1% and 53.8% of the genome varying for COLO201/COLO205 and CX-1/HT29 (Fig. 5). In contrast, discrepancies were limited in MSI-H pairs with 0.2%, 0.4%, 6.6%, 2.9%, and 4.3% of the genome differing between Gp2D/Gp5D, LS174T/LS180, DLD1/HCT8, DLD1/HCT15, and HCT8/HCT15, respectively. 47.5% of the genome differed in the nonhypermutated primary/metastasis pair SW480/SW620. Notably, mutation differences between paired cell lines did not obscure known driver genes. For the established colorectal cancer genes APC, TP53, SMAD4, PIK3CA, KRAS, and BRAF, 29 of 33 (87.9%) nonsilent mutations were concordant between cell line pairs. Discussion In this study we show that the mutation and DNA copynumber landscapes determined for 70 colorectal cancer cell lines closely resemble those of primary tumors, underscoring the utility of cell lines as an appropriate model system for this malignancy. The 3 molecular subtypes of colorectal cancer recently defined by the TCGA, nonhypermutated, hypermutated with microsatellite instability, and hypermutated without microsatellite instability were faithfully captured in the cell line panel. As expected, MSI-H cell lines exhibited Cancer Research Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Genomic Landscape of Colorectal Cancer Cell Lines Figure 5. Overlap of mutations and DNA copy-number states between paired colorectal cancer cell lines. COLO201/COLO205, CX-1/HT29, Gp2D/Gp5D, LS174T/LS180, and DLD1/HCT8/HCT15 were derived from the same tumor, SW480/SW620 from a primary tumor and subsequent lymph node metastasis. hypermethylation of the MLH1 promoter and/or mutations in MMR genes, whereas hypermutated lines without microsatellite instability were instead characterized by mutations in the exonuclease domain of POLE. Consistent with defective DNA POLE proofreading function, the latter tumors showed a nucleotide substitution hypermutator phenotype with a bias to C>A/G>T and A>C/T>G transversions. Germline missense mutations in the POLE exonuclease domain have recently been associated with familial predisposition to colorectal cancer (30), and elevated tumor predisposition and base-substitution mutations observed in Pole exonuclease-mutant mice (31). Similarly, somatic missense mutations in the POLE exonuclease domain have been reported in 7% of endometrial cancers, with good evidence of associated hypermutation (32). Although differences in prognosis and chemotherapy response have been extensively documented for nonhypermutated versus MSI-H tumors (2–4), clinical characteristics of POLEmutant NSHP tumors are unknown. Our identification of 2 cell lines representative of this colorectal cancer subtype will facilitate an investigation of the specific aspects of the biology of these tumors, particularly in relation to identifying therapeutics that may exploit their unique genomic instability. Consistent with previous lower-resolution data (33), DNA copy-number profiles were similar between colorectal cancer cell lines and primary tumors, with nonhypermutated cases tending to exhibit chromosomal instability, whereas both MSI-H and NSHP cases had overall stable copy-number profiles. Patterns of whole and partial chromosome gains and losses were largely concordant. Besides recurrent gain/ www.aacrjournals.org amplification of the established colorectal cancer gene MYC, recurrent regions of deletion in nonhypermutated cases included the candidate cancer genes PARK2 and WRN. WRN has important roles in homologous recombination repair, MUTYH-mediated repair of oxidative DNA damage, and telomeric recombination (34–36), and WRN germline mutations are associated with chromosomal instability and cancer predisposition (37). PARK2 deletion has been previously reported in sporadic colorectal cancer, and Park2 deficiency shown to accelerate intestinal adenoma development in Apc mutant mice (38). The mutation landscapes in nonhypermutated and hypermutated MSI-H cases showed close resemblance in cell lines and primary tumors, with the expected alterations in the WNT, MAPK, PI3K, p53, and TGFb pathways. Besides these main colorectal cancer–associated pathways, multiple chromatinstate regulators were enriched among the top 5% of mutated genes, including proteins involved in chromatin remodeling (ARID1A, CHD6, and SRCAP) and histone methylation or acetylation (ASH1L, EP300, EP400, MLL2, MLL3, PRDM2, and TRRAP). Overall, 24% of nonhypermutated and 96% of MSIH cases harbored mutations in these genes. Trends for mutations to cluster in chromatin regulators have been reported in multiple other cancer types, including liver cancers (39), gastric adenocarcinoma (40), and transitional cell carcinoma of the bladder (41), suggesting potential pathogenicity of these alterations. Paired cell lines originating from the same tumor showed considerable differences for both mutations and DNA copy- Cancer Res; 74(12) June 15, 2014 Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. OF7 Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Mouradov et al. Figure 6. Simulation of the acquisition of mutations in cell culture. The process of serial passage was modeled with cell numbers repeatedly increasing 5 6 from 1 10 to 2 10 cells. Proportions of cells containing the most frequent mutation by passage number for five simulations, using fixed mutation 8 6 probabilities of 10 per base per cell replication for nonhypermutated (A) and 10 for hypermutated (B) MSI-H tumors. The black diagonal trendline is the least squares fit on the log–log scale and the small vertical lines are the 99% prediction intervals. The horizontal red line corresponds to a sequencing mutation detection threshold of 10%. number profiles. Nonhypermutated pairs differed for up to 356 mutations and 54% of genome copy number, whereas MSI-H pairs differed for up to 6,503 mutations and 6.6% of genome copy number. The two main possible reasons for these differences are preexisting heterogeneity between the cells from the original tumor grown out to establish these paired lines, or acquisition of alterations as a result of longterm culture. Assuming a normal mutation rate of 108 (per bp per cell generation) for nonhypermutated and 100-fold elevated rate for hypermutated pairs and absence of selection (42), cell lines established from tumor cells separated by an average of 328 and 66 replications would be expected to exhibit the observed mutational differences with a >99% probability, respectively (Supplementary Data and Table S9). In contrast, mutations acquired during serial passaging in culture are not anticipated to reach a detectable 10% level before 9,034 and 4,266 replications, respectively (Supplementary Data and Fig. 6). Although serial cell passaging may have been common-place at the time when these paired cell lines were established >20 years ago, stock-keeping practices to limit the number of replications were soon introduced. Preexisting mutation heterogeneity in the original tumor therefore seems highly likely to account for the majority of the detected differences. Numbers of replications may be expected to be larger between cells from a primary tumor and subsequent metastasis, and accordingly our corresponding nonhypermutated cell line pair showed a 4-fold higher number of mutational differences than the other nonhypermutated pairs. Importantly, mutational differences across paired cell lines did not obscure known driver genes, with 88% of nonsilent mutations in established colorectal cancer genes concordant between pairs. Despite the high level of similarity between colorectal cancer cell lines and primary tumors, there were a number of differences. These included overall higher mutation and DNA copynumber frequencies, as well as a greater proportion of detected OF8 Cancer Res; 74(12) June 15, 2014 InDels in cell lines. These discrepancies may be related to differences in exome-capture and sequencing platforms, bioinformatics pipelines, the presence of contaminating normal tissue in primary cancers, and accuracy of assigning somatic alterations in cell lines. Cell lines further showed a higher proportion of hypermutated cases, and exhibited differences in the prevalence of aberrations for certain genes and genomic regions such as CTNNB1 mutations in MSI-H cases. These latter findings may be a reflection of preferential growing out of cell lines from primary tumors (or their respective subclones) that contain these aberrations, a contention supported by the observation that only 10% to 15% of colorectal cancers give rise to cell lines (43). Another possibility is that some of the mutations have been acquired and selected for in tissue culture. A caveat to our analysis of protein-coding genes is that we could not report on untranslated exonic regions (UTR), as the latter were inconsistently covered by our study and the TCGA. UTR mutations can impact on RNA splicing, stability, or translation as previously highlighted for MSI-H colorectal cancer (44). In the comparison of gene mutation profiles, we chose to exclude silent mutations (other than those affecting splice sites), although some of these may similarly have functional consequences (45). In conclusion, our comparative analysis of the genomic landscapes of human colorectal cancer cell lines and TCGAanalyzed primary cancers identified cell lines representative of the three main mutational colorectal cancer subtypes. Within these molecular subtypes, although some differences were evident, cell lines showed globally similar genetic alterations to primary cancers, including genome-wide mutation, DNA copy number, and driver gene mutation profiles. Accordingly, gene expression profiles of colorectal cancer cell lines have previously been shown to broadly represent those of primary tumors (46). Our genomic data significantly expand on cancer cell line characterization efforts by the major cancer genome Cancer Research Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Genomic Landscape of Colorectal Cancer Cell Lines centers, such as the Cancer Cell Line Encyclopedia project, which currently reports mutation data for 1,500 selected genes on 62 colorectal cancer cell lines (5). Our data will help to inform investigations of the molecular basis of colorectal cancer pathogenesis, inherent and acquired drug resistance, and exploration of novel treatment modalities for this malignancy. Writing, review, and/or revision of the manuscript: D. Mouradov, C.G. Love, A.W. Burgess, D. Buchanan, J.L. Wilding, I.P.M. Tomlinson, W.F. Bodmer, J.M. Mariadason, O.M. Sieber Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Mouradov, S. Li, S. Wormald Study supervision: J.M. Mariadason, O.M. Sieber Acknowledgments The authors thank the Victorian Cancer BioBank for patient specimens and Prof. M. Schwab at the DKFZ for providing cell lines. Disclosure of Potential Conflicts of Interest No potential conflicts of interest were disclosed. Grant Support Authors' Contributions Conception and design: A.W. Burgess, R.L. Strausberg, J.M. Mariadason, O.M. Sieber Development of methodology: D. Mouradov, R.N. Jorissen, S. Li, D. Bicknell, O.M. Sieber Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Arango, D. Buchanan, S. Wormald, D. Bicknell, J.M. Mariadason, O.M. Sieber Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Mouradov, C. Sloggett, R.N. Jorissen, C.G. Love, A.W. Burgess, D. Arango, D. Buchanan, L. O'Connor, J.L. Wilding, W.F. Bodmer, J.M. Mariadason, O.M. Sieber This work was supported by Cancer Australia through a Project Grant (APP1030098; O.M. Sieber), Ludwig Institute for Cancer Research (J.M. Mariadason, A.W. Burgess, O.M. Sieber), NHMRC Overseas Postdoctoral Fellowship (519795; S. Wormald), and a Victorian State Government Operational Infrastructure Support grant. J.M. Mariadason holds an Australian Research Council Future Fellowship. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received January 7, 2014; revised March 17, 2014; accepted April 8, 2014; published OnlineFirst April 22, 2014. References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J Clin 2011;61:69–90. 2. Bracht K, Nicholls AM, Liu Y, Bodmer WF. 5-Fluorouracil response in a large panel of colorectal cancer cell lines is associated with mismatch repair deficiency. Br J Cancer 2010;103:340–6. 3. Ashraf SQ, Nicholls AM, Wilding JL, Ntouroupi TG, Mortensen NJ, Bodmer WF. Direct and immune mediated antibody targeting of ERBB receptors in a colorectal cancer cell-line panel. Proc Natl Acad Sci U S A 2012;109:21046–51. 4. Weickhardt AJ, Price TJ, Chong G, Gebski V, Pavlakis N, Johns TG, et al. Dual targeting of the epidermal growth factor receptor using the combination of cetuximab and erlotinib: preclinical evaluation and results of the phase II DUX study in chemotherapyrefractory, advanced colorectal cancer. J Clin Oncol 2012;30: 1505–12. 5. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S, et al. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 2012;483:603–7. 6. Basu A, Bodycombe NE, Cheah JH, Price EV, Liu K, Schaefer GI, et al. An interactive resource to identify cancer genetic and lineage dependencies targeted by small molecules. Cell 2013;154: 1151–61. 7. Fogli S, Caraglia M. Genotype-based therapeutic approach for colorectal cancer: state of the art and future perspectives. Expert Opin Pharmacother 2009;10:1095–108. 8. Aaltonen LA, Salovaara R, Kristo P, Canzian F, Hemminki A, Peltomaki P, et al. Incidence of hereditary nonpolyposis colorectal cancer and the feasibility of molecular screening for the disease. N Engl J Med 1998;338:1481–7. 9. Lengauer C, Kinzler KW, Vogelstein B. Genetic instability in colorectal cancers. Nature 1997;386:623–7. 10. Miyazaki M, Furuya T, Shiraki A, Sato T, Oga A, Sasaki K. The relationship of DNA ploidy to chromosomal instability in primary human colorectal cancers. Cancer Res 1999;59:5283–5. 11. Alexander J, Watanabe T, Wu TT, Rashid A, Li S, Hamilton SR. Histopathological identification of colon cancer with microsatellite instability. Am J Pathol 2001;158:527–35. 12. Duval A, Reperant M, Hamelin R. Comparative analysis of mutation frequency of coding and non coding short mononucleotide repeats in mismatch repair deficient colorectal cancers. Oncogene 2002;21: 8062–6. www.aacrjournals.org 13. Kambara T, Simms LA, Whitehall VL, Spring KJ, Wynter CV, Walsh MD, et al. BRAF mutation is associated with DNA methylation in serrated polyps and cancers of the colorectum. Gut 2004;53: 1137–44. 14. Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, Issa JP. CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A 1999;96:8681–6. 15. Rowan A, Halford S, Gaasenbeek M, Kemp Z, Sieber O, Volikos E, et al. Refining molecular analysis in the pathways of colorectal carcinogenesis. Clin Gastroenterol Hepatol 2005;3:1115–23. 16. Sjoblom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD, et al. The consensus coding sequences of human breast and colorectal cancers. Science 2006;314:268–74. 17. The Cancer Genome Atlas Network. Comprehensive molecular characterization of human colon and rectal cancer. Nature 2012;487: 330–7. 18. Seshagiri S, Stawiski EW, Durinck S, Modrusan Z, Storm EE, Conboy CB, et al. Recurrent R-spondin fusions in colon cancer. Nature 2012;488:660–4. 19. Koo BK, Spit M, Jordens I, Low TY, Stange DE, van de Wetering M, et al. Tumour suppressor RNF43 is a stem-cell E3 ligase that induces endocytosis of Wnt receptors. Nature 2012;488:665–9. 20. Bailey JA, Gu Z, Clark RA, Reinert K, Samonte RV, Schwartz S, et al. Recent segmental duplications in the human genome. Science 2002;297:1003–7. 21. Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 2009;25:1105–11. 22. Yau C, Mouradov D, Jorissen RN, Colella S, Mirza G, Steers G, et al. A statistical approach for detecting genomic aberrations in heterogeneous tumor samples from single nucleotide polymorphism genotyping data. Genome Biol 2010;11:R92. 23. Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biol 2011;12:R41. 24. Boland CR, Thibodeau SN, Hamilton SR, Sidransky D, Eshleman JR, Burt RW, et al. 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 1998;58: 5248–57. Cancer Res; 74(12) June 15, 2014 Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. OF9 Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Mouradov et al. 25. Weisenberger DJ, Siegmund KD, Campan M, Young J, Long TI, Faasse MA, et al. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet 2006;38:787–93. 26. Prolla TA, Baker SM, Harris AC, Tsao JL, Yao X, Bronner CE, et al. Tumour susceptibility and spontaneous mutation in mice deficient in Mlh1, Pms1 and Pms2 DNA mismatch repair. Nat Genet 1998;18:276–9. 27. Sansom OJ, Meniel VS, Muncan V, Phesse TJ, Wilkins JA, Reed KR, et al. Myc deletion rescues Apc deficiency in the small intestine. Nature 2007;446:676–9. 28. Rutzky LP, Kaye CI, Siciliano MJ, Chao M, Kahan BD. Longitudinal karyotype and genetic signature analysis of cultured human colon adenocarcinoma cell lines LS180 and LS174T. Cancer Res 1980;40: 1443–8. 29. Efstathiou JA, Liu D, Wheeler JM, Kim HC, Beck NE, Ilyas M, et al. Mutated epithelial cadherin is associated with increased tumorigenicity and loss of adhesion and of responsiveness to the motogenic trefoil factor 2 in colon carcinoma cells. Proc Natl Acad Sci U S A 1999;96:2316–21. 30. Palles C, Cazier JB, Howarth KM, Domingo E, Jones AM, Broderick P, et al. Germline mutations affecting the proofreading domains of POLE and POLD1 predispose to colorectal adenomas and carcinomas. Nat Genet 2012;45:136–44. 31. Albertson TM, Ogawa M, Bugni JM, Hays LE, Chen Y, Wang Y, et al. DNA polymerase epsilon and delta proofreading suppress discrete mutator and cancer phenotypes in mice. Proc Natl Acad Sci U S A 2009;106:17101–4. 32. Church DN, Briggs SE, Palles C, Domingo E, Kearsey SJ, Grimes JM, et al. DNA polymerase e and d exonuclease domain mutations in endometrial cancer. Hum Mol Genet 2013;22:2820–8. 33. Douglas EJ, Fiegler H, Rowan A, Halford S, Bicknell DC, Bodmer W, et al. Array comparative genomic hybridization analysis of colorectal cancer cell lines and primary carcinomas. Cancer Res 2004;64: 4817–25. 34. Prince PR, Emond MJ, Monnat RJ Jr. Loss of Werner syndrome protein function promotes aberrant mitotic recombination. Genes Dev 2001; 15:933–8. 35. Kanagaraj R, Parasuraman P, Mihaljevic B, van Loon B, Burdova K, Konig C, et al. Involvement of Werner syndrome protein in MUTYH- OF10 Cancer Res; 74(12) June 15, 2014 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. mediated repair of oxidative DNA damage. Nucleic Acids Res 2012;40:8449–59. Mendez-Bermudez A, Hidalgo-Bravo A, Cotton VE, Gravani A, Jeyapalan JN, Royle NJ. The roles of WRN and BLM RecQ helicases in the alternative lengthening of telomeres. Nucleic Acids Res 2012;40: 10809–20. Lauper JM, Krause A, Vaughan TL, Monnat RJ Jr. Spectrum and risk of neoplasia in Werner syndrome: a systematic review. PLoS ONE 2013;8:e59709. Poulogiannis G, McIntyre RE, Dimitriadi M, Apps JR, Wilson CH, Ichimura K, et al. PARK2 deletions occur frequently in sporadic colorectal cancer and accelerate adenoma development in Apc mutant mice. Proc Natl Acad Sci U S A 2010;107:15145–50. Fujimoto A, Totoki Y, Abe T, Boroevich KA, Hosoda F, Nguyen HH, et al. Whole-genome sequencing of liver cancers identifies etiological influences on mutation patterns and recurrent mutations in chromatin regulators. Nat Genet 2012;44:760–4. Zang ZJ, Cutcutache I, Poon SL, Zhang SL, McPherson JR, Tao J, et al. Exome sequencing of gastric adenocarcinoma identifies recurrent somatic mutations in cell adhesion and chromatin remodeling genes. Nat Genet 2012;44:570–4. Gui Y, Guo G, Huang Y, Hu X, Tang A, Gao S, et al. Frequent mutations of chromatin remodeling genes in transitional cell carcinoma of the bladder. Nat Genet 2011;43:875–8. Tomlinson IP, Novelli MR, Bodmer WF. The mutation rate and cancer. Proc Natl Acad Sci U S A 1996;93:14800–3. Liu Y, Bodmer WF. Analysis of P53 mutations and their expression in 56 colorectal cancer cell lines. Proc Natl Acad Sci U S A 2006;103:976–81. Wilding JL, McGowan S, Liu Y, Bodmer WF. Replication error deficient and proficient colorectal cancer gene expression differences caused by 30 UTR polyT sequence deletions. Proc Natl Acad Sci U S A 2010;107:21058–63. Kudla G, Murray AW, Tollervey D, Plotkin JB. Coding-sequence determinants of gene expression in Escherichia coli. Science 2009; 324:255–8. Schlicker A, Beran G, Chresta CM, McWalter G, Pritchard A, Weston S, et al. Subtypes of primary colorectal tumors correlate with response to targeted treatment in colorectal cell lines. BMC Med Genomics 2012;5:66. Cancer Research Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research. Published OnlineFirst April 22, 2014; DOI: 10.1158/0008-5472.CAN-14-0013 Colorectal Cancer Cell Lines Are Representative Models of the Main Molecular Subtypes of Primary Cancer Dmitri Mouradov, Clare Sloggett, Robert N. Jorissen, et al. Cancer Res Published OnlineFirst April 22, 2014. Updated version Supplementary Material E-mail alerts Reprints and Subscriptions Permissions Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-14-0013 Access the most recent supplemental material at: http://cancerres.aacrjournals.org/content/suppl/2014/04/22/0008-5472.CAN-14-0013.DC1 Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at [email protected]. To request permission to re-use all or part of this article, contact the AACR Publications Department at [email protected]. Downloaded from cancerres.aacrjournals.org on April 29, 2017. © 2014 American Association for Cancer Research.