Early metastasis is a hallmark of pancreatic ductal adenocarcinoma and responsible for >90% of pancreatic cancer death. Because little is known about the biology and genetics of the metastatic process, we desired to elucidate molecular pathways mediating pancreatic cancer metastasis in vivo by an unbiased forward genetic approach.

Highly metastatic pancreatic cancer cell populations were selected by serial in vivo passaging of parental cells with low metastatic potential and characterized by global gene expression profiling, chromatin immunoprecipitation, and in vivo metastatic assay.

In vivo selection of highly metastatic pancreatic cancer cells induced epithelial-mesenchymal transition (EMT), loss of E-cadherin expression, and up-regulation of mesenchymal genes such as Snail. Genetic inactivation of E-cadherin in parental cells induced EMT and increased metastasis in vivo. Silencing of E-cadherin in highly metastatic cells is mediated by a transcriptional repressor complex containing Snail and histone deacetylase 1 (HDAC1) and HDAC2. In line, mesenchymal pancreatic cancer specimens and primary cell lines from genetically engineered Kras^G12D mice showed HDAC-dependent down-regulation of E-cadherin and high metastatic potential. Finally, transforming growth factor β-driven E-cadherin silencing and EMT of human pancreatic cancer cells depends on HDAC activity.

We provide the first in vivo evidence that HDACs and Snail play an essential role in silencing E-cadherin during the metastatic process of pancreatic cancer cells. These data link the epigenetic HDAC machinery to EMT and metastasis and provide preclinical evidence that HDACs are promising targets for antimetastatic therapy.