13Tumor metastasis is the cause of death for 90% of cancer patients, and no 14 currently-available therapies target this multi-step process in which cancer cells 15 spread from the local tissue of a primary tumor to distant organs where they 16 establish secondary tumors 1 . Although epithelial-to-mesenchymal transition 2 , 17 tumor-secreted exosomes 3 , epigenetic regulators as well as other genes 4-8 have 18 been implicated in metastasis, little is known about how cells adapt to and 19 colonize new tissue environments. Here, we show that the epigenetics-mediated 20reprogramming of tissue-specific gene transcription in cancer cells promotes 21 metastasis. Using colorectal cancer (CRC) as a model, we found in both clinical 22 3 reprogramming in several cohorts of clinical CRC tumor samples and in multiple 33 other forms of metastatic cancers, indicate that this reprogramming may be a 34 common feature of metastasis in multiple cancers and suggest the targeted 35 disruption of this epigenetic reprogramming as a strategy for the development of 36 therapies to treat metastasis, the leading cause of cancer-related mortality. 37
MAIN 39Tumor metastasis refers to the movement of tumor cells from a primary site 40to distant organs that they progressively colonize 9 . More than 100 years ago, 41Paget suggested the idea of metastasis as the interaction of "seeds" and "soil" 10 , 42 but subsequent research has yielded only a limited understanding of the 43 mechanism(s) through which metastatic cancer cells ("seeds") adapt to and 44 colonize a new tissue environment ("soil"), the crucial steps of the metastasis 45 process 11 . 46It has been reported that the expression of tissue-specific or cell-lineage 47 7 the genome regions around loci encoding up-regulated liver-specific genes had 113 enriched H3K27ac (P value = 2.2e-16, by t-test) and H3K4me2 (P value = 2.2e-16, 114 by t-test) deposition (Fig. 2c). 115So-called super-enhancers are a small fraction of total enhancers and 116 encompass broad chromatin domains with H3K27ac deposition near genes 117 essential for defining cell identity 20,21 . By identifying super-enhancers in SW480 118 and SW620 cells based on our H3K27ac ChIP-seq datasets, we found that in 119 addition to the 264 super-enhancers common to both cell lines, there are 280 and 120 215 unique super-enhancers in SW620 and SW480 cells, respectively (Extended 121 Data Fig. 4d). Comparison between our ChIP-seq and RNA-seq data revealed a 122 high Pearson correlation coefficient (R = 0.807, P value < 2.2e-16) between the 123 genome-wide distribution of super-enhancers and the expression levels of the 124 nearest genes in these two cell lines ( Fig. 2d and Extended Data Fig. 4e). Notably, 125 many of the liver-specific genes in SW620 cells were found near SW620-unique 126 super-enhancers; while the colon-specific genes in SW480 cells were found close 127 to SW480-unique super-enhancers ( Fig. 2e and Extended Data Fig. 4f, g). Our 128 epigenetic and transcriptomics experiments demonstrate that reshaped enhancer 129 ...