Purpose: The intratumoral heterogeneity (ITH) and the evolution of genomic architectures associated with the development of distant metastases are not well understood in colorectal cancers.Experimental Design: We performed multiregion biopsies of primary and liver metastatic regions from five colorectal cancers with whole-exome sequencing and copy number profiling.Results: In addition to a substantial level of genetic ITH, multiregion genetic profiling identifies the subclonal mutational architecture, leading to the region-based or spatial categorization of somatic mutations and the inference of intratumoral evolutionary history of cancers. The universal mutations (those observed in all the regional biopsies) are enriched in known cancer genes such as APC and TP53 with distinct mutational spectra compared with biopsy-or region-specific mutations, suggesting that major operative mutational mechanisms and their selective pressures are not constant across the metastatic progression. The phylogenies inferred from genomic data show branching evolutionary patterns where some primary biopsies are often segregated with metastastic lesions. Our analyses also revealed that copy number changes such as the chromosomal gains of c-MYC and chromothripsis can be region specific and the potential source of genetic ITH.Conclusions: Our data show that the genetic ITH is prevalent in colorectal cancer serving as a potential driving force to generate metastasis-initiating clones and also as a means to infer the intratumoral evolutionary history of cancers. The paucity of recurrent metastasis-clonal events suggests that colorectal cancer distant metastases may not follow a uniform course of genomic evolution, which should be considered in the genetic diagnosis and the selection of therapeutic targets for the advanced colorectal cancer.

Whole-genome sequencing data allow detection of copy number variation (CNV) at high resolution. However, estimation based on read coverage along the genome suffers from bias due to GC content and other factors. Here, we develop an algorithm called BIC-seq2 that combines normalization of the data at the nucleotide level and Bayesian information criterion-based segmentation to detect both somatic and germline CNVs accurately. Analysis of simulation data showed that this method outperforms existing methods. We apply this algorithm to low coverage whole-genome sequencing data from peripheral blood of nearly a thousand patients across eleven cancer types in The Cancer Genome Atlas (TCGA) to identify cancer-predisposing CNV regions. We confirm known regions and discover new ones including those covering KMT2C, GOLPH3, ERBB2 and PLAG1. Analysis of colorectal cancer genomes in particular reveals novel recurrent CNVs including deletions at two chromatin-remodeling genes RERE and NPM2. This method will be useful to many researchers interested in profiling CNVs from whole-genome sequencing data.

Although it has been suggested that kinesin family member 14 (KIF14) has oncogenic potential in various cancers, including hepatocellular carcinoma (HCC), the molecular mechanism of this potential remains unknown. We aimed to elucidate the role of KIF14 in hepatocarcinogenesis by knocking down KIF14 in HCC cells that overexpressed KIF14. After KIF14 knockdown, changes in tumor cell growth, cell cycle and cytokinesis were examined. We also examined cell cycle regulatory molecules and upstream Skp1/Cul1/F-box (SCF) complex molecules. Knockdown of KIF14 resulted in suppression of cell proliferation and failure of cytokinesis, whereas KIF14 overexpression increased cell proliferation. In KIF14-silenced cells, the levels of cyclins E1, D1 and B1 were profoundly decreased compared with control cells. Of the cyclin-dependent kinase inhibitors, the p27Kip1 protein level specifically increased after KIF14 knockdown. The increase in p27Kip1 was not due to elevation of its mRNA level, but was due to inhibition of the proteasome-dependent degradation pathway. To explore the pathway upstream of this event, we measured the levels of SCF complex molecules, including Skp1, Skp2, Cul1, Roc1 and Cks1. The levels of Skp2 and its cofactor Cks1 decreased in the KIF14 knockdown cells where p27Kip1 accumulated. Overexpression of Skp2 in the KIF14 knockdown cells attenuated the failure of cytokinesis. On the basis of these results, we postulate that KIF14 knockdown downregulates the expression of Skp2 and Cks1, which target p27Kip1 for degradation by the 26S proteasome, leading to accumulation of p27Kip1. The downregulation of Skp2 and Cks1 also resulted in cytokinesis failure, which may inhibit tumor growth. To the best of our knowledge, this is the first report that has identified the molecular target and oncogenic effect of KIF14 in HCC.

Intratumoural heterogeneity (ITH) leads to regional biases of the mutational landscape in a single tumour and may influence the single biopsy-based clinical diagnosis and treatment decision. To evaluate the extent of ITH in unifocal prostate cancers (PCAs), we analysed multiple regional biopsies from three PCAs, using whole-exome sequencing, DNA copy number and gene expression profiling analyses. A substantial level of ITH was identified, in that 0-61% and 18-71% of somatic variants were common or private, respectively, within a given cancer. The enhanced mutation detection rate in the combined sequencing dataset across intratumoural biopsies was demonstrated with respect to the total number of mutations identified in a given tumour. Allele frequencies of the mutations were positively correlated with the levels of intratumoural recurrence (private < shared < common), but some common mutations showed low allele frequency, suggesting that not all were clonally fixed. Regional biases in the presentation of a well-known TMPRSS2-ERG fusion was noted in one PCA and the somatic mutation- and copy number-based phylogenetic relationships between intratumoural biopsies were largely concordant. Genes showing intratumoural expression variability were commonly enriched in the molecular function of eicosanoid metabolism and PCA-relevant clinical markers. Taken together, our analyses identified a substantial level of genetic ITH in unifocal PCAs at the mutation, copy number and expression levels, which should be taken into account for the identification of biomarkers in the clinical setting.

Although the colorectal adenoma-to-carcinoma sequence represents a classical cancer progression model, the evolution of the mutational landscape underlying this model is not fully understood. In this study, we analyzed eight synchronous pairs of colorectal high-grade adenomas and carcinomas, four microsatellite-unstable (MSU) and four -stable (MSS) pairs, using whole-exome sequencing. In the MSU adenoma-carcinoma pairs, we observed no subclonal mutations in adenomas that became fixed in paired carcinomas, suggesting a ‘parallel’ evolution of synchronous adenoma-to-carcinoma, rather than a ‘stepwise’ evolution. The abundance of indel (in MSU and MSS pairs) and microsatellite instability (in MSU pairs) was noted in the later adenoma- or carcinoma-specific mutations, indicating that the mutational processes and functional constraints operative in early and late colorectal carcinogenesis are different. All MSU cases exhibited clonal, truncating mutations in ACVR2A, TGFBR2, and DNA mismatch repair genes, but none were present in APC or KRAS. In three MSS pairs, both APC and KRAS mutations were identified as both early and clonal events, often accompanying clonal copy number changes. An MSS case uniquely exhibited clonal ERBB2 amplification, followed by APC and TP53 mutations as carcinoma-specific events. Along with the previously unrecognized clonal origins of synchronous colorectal adenoma-carcinoma pairs, our study revealed that the preferred sequence of mutational events during colorectal carcinogenesis can be context-dependent.