Gastroenterology 2018 04 15;154(5):1494-1508.e13. Epub 2017 Dec 15.
Department of Medicine, Stony Brook University School of Medicine, Stony Brook, New York. Electronic address:
Background & Aims: Activating mutations in KRAS are detected in most pancreatic ductal adenocarcinomas (PDACs). Expression of an activated form of KRAS (KrasG12D) in pancreata of mice is sufficient to induce formation of pancreatic intraepithelial neoplasia (PanINs)-a precursor of PDAC. Pancreatitis increases formation of PanINs in mice that express KrasG12D by promoting acinar-to-ductal metaplasia (ADM). We investigated the role of the transcription factor Krüppel-like factor 5 (KLF5) in ADM and KRAS-mediated formation of PanINs.
Methods: We performed studies in adult mice with conditional disruption of Klf5 (Klf5) and/or expression of Kras (LSL-Kras) via Cre recombinase regulated by an acinar cell-specific promoter (Ptf1a). Activation of Kras and loss of KLF5 was achieved by administration of tamoxifen. Pancreatitis was induced in mice by administration of cerulein; pancreatic tissues were collected, analyzed by histology and immunohistochemistry, and transcriptomes were compared between mice that did or did not express KLF5. We performed immunohistochemical analyses of human tissue microarrays, comparing levels of KLF5 among 96 human samples of PDAC. UN-KC-6141 cells (pancreatic cancer cells derived from Pdx1-Cre;LSL-Kras mice) were incubated with inhibitors of different kinases and analyzed in proliferation assays and by immunoblots. Expression of KLF5 was knocked down with small hairpin RNAs or CRISPR/Cas9 strategies; cells were analyzed in proliferation and gene expression assays, and compared with cells expressing control vectors. Cells were subcutaneously injected into flanks of syngeneic mice and tumor growth was assessed.
Results: Of the 96 PDAC samples analyzed, 73% were positive for KLF5 (defined as nuclear staining in more than 5% of tumor cells). Pancreata from Ptf1a-Cre;LSL-Kras mice contained ADM and PanIN lesions, which contained high levels of nuclear KLF5 within these structures. In contrast, Ptf1a-Cre;LSL-Kras;Klf5 mice formed fewer PanINs. After cerulein administration, Ptf1a-Cre;LSL-Kras mice formed more extensive ADM than Ptf1a-Cre;LSL-Kras;Klf5 mice. Pancreata from Ptf1a-Cre;LSL-Kras;Klf5 mice had increased expression of the tumor suppressor NDRG2 and reduced phosphorylation (activation) of STAT3, compared with Ptf1a-Cre;LSL-Kras mice. In UN-KC-6141 cells, PI3K and MEK signaling increased expression of KLF5; a high level of KLF5 increased proliferation. Cells with knockdown of Klf5 had reduced proliferation, compared with control cells, had reduced expression of ductal markers, and formed smaller tumors (71.61 ± 30.79 mm vs 121.44 ± 34.90 mm from control cells) in flanks of mice.
Conclusion: Levels of KLF5 are increased in human PDAC samples and in PanINs of Ptf1a-Cre;LSL-Kras mice, compared with controls. KLF5 disruption increases expression of NDRG2 and reduces activation of STAT3 and reduces ADM and PanINs formation in mice. Strategies to reduce KLF5 activity might reduce progression of acinar cells from ADM to PanIN and pancreatic tumorigenesis.