Volume 145, Issue 3, Pages 668-678.e3 (September 2013) Let-7b and miR-495 Stimulate Differentiation and Prevent Metaplasia of Pancreatic Acinar Cells by Repressing HNF6  Pierre–Paul Prévot, Cécile Augereau, Alexandru Simion, Géraldine Van den Steen, Nicolas Dauguet, Frédéric P. Lemaigre, Patrick Jacquemin  Gastroenterology  Volume 145, Issue 3, Pages 668-678.e3 (September 2013) DOI: 10.1053/j.gastro.2013.05.016 Copyright © 2013 AGA Institute Terms and Conditions

Figure 1 Deficient expression of Dicer is associated with HNF6-dependent morphogenic anomalies. (A) miRNA expression (real-time quantitative polymerase chain reaction) is decreased at late stages of development in Foxa3-Cre/DicerloxP/− pancreas as compared with wild-type (WT) pancreas. Values (means ± SEM; n = 4; ∗P < .05; ∗∗ P < .01; ∗∗∗ P < .001) are normalized to β-actin expression. (B) At E15.5, a subset of acini in Foxa3-Cre/DicerloxP/− pancreata show enlarged lumina (white arrow), and a number of acinar cells display HNF6 overexpression (yellow arrow). Expression of Mucin1a and HNF1β was restricted to the ducts, in WT and Foxa3-Cre/DicerloxP/− pancreata. (C) H&E staining of Foxa3-Cre/DicerloxP/− pancreata at E18.5 shows enlarged acinar lumina (red arrows). Apical polarity markers are diffusely located (ezrin and atypical protein kinase C [aPKC]) or redistributed along the basolateral pole (Mucin 1a) of the acinar cells in Foxa3-Cre/DicerloxP/− pancreata. The absence of HNF6 abrogates the effect of Dicer inactivation on ezrin and aPKC location, and partially normalizes Mucin 1a distribution. Bar = 20 μm. Gastroenterology 2013 145, 668-678.e3DOI: (10.1053/j.gastro.2013.05.016) Copyright © 2013 AGA Institute Terms and Conditions

Figure 2 Inactivation of Dicer represses differentiation of acinar cells and induces hepatic gene expression, in a HNF6-dependent way. (A) Real-time quantitative polymerase chain reaction of early and late acinar markers and liver genes at E18.5. The Ct values for Apo4, ApoH, α-fetoprotein (Afp), Alb, and transthyretin are respectively 26.7, 29.2, 27.8, 28.7, and 31.3 in Foxa3-Cre/DicerloxP/− pancreas. Values (means ± SEM; n = 6; ∗P < .05; ∗∗ P < .01; ∗∗∗ P < .001) are normalized for β-actin. Asterisks indicate statistically significant differences between WT and Foxa3-Cre/DicerloxP/− pancreata. (B) Expression of HNF6 expression is high in ducts and faint in acinar cells, and absent from islets (asterisk) in wild-type (WT) pancreas at E18.5. In Foxa3-Cre/DicerloxP/− embryos, expression of HNF6 is increased in a subset of acinar cells (arrows). (C) Increased expression of HNF6 in Foxa3-Cre/DicerloxP/− acini at E18.5 is associated with normal expression of Ptf1a (arrows), but with repression of Mist1 and Rbjl (arrows). In Foxa3-Cre/DicerloxP/−/Hnf6−/− or Hnf6−/− pancreata, Ptf1a, Mist1, and Rbpjl are normal. Bar = 20 μm. Gastroenterology 2013 145, 668-678.e3DOI: (10.1053/j.gastro.2013.05.016) Copyright © 2013 AGA Institute Terms and Conditions

Figure 3 HNF6 overexpression in acinar cell−enriched cultures represses late differentiation genes while inducing hepatic genes. E15.5 pancreata were dissociated and cultured ex vivo in the presence of adenoviruses expressing GFP (Ad-GFP) or HNF6 (Ad-HNF6). (A) Real-time quantitative polymerase chain reaction analysis shows repression of acinar genes (left) and induction of hepatic gene expression (right). Data are means ± SEM; n = 6; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. (B) HNF6 binding consensus sequence and putative HNF6 binding sites, and their position relative to the transcription start site in hepatic gene promoters. (C) Chromatin immunoprecipitation using anti-HNF6 antibodies and cultured embryonic pancreata infected with Ad-HNF6 followed by quantitative polymerase chain reaction amplification shows binding of HNF6 to the transthyretin, ApoH, and Alb promoter. Values obtained with AdHNF6 or AdGFP in the presence of IgG were arbitrarily set to 1. No binding was seen in a region of the Hnf1β gene promoter devoid of any HNF6 binding sequence. Data are means ± SEM (n = 4); ∗P < .05; ∗∗P < .01; ns, not significant. Gastroenterology 2013 145, 668-678.e3DOI: (10.1053/j.gastro.2013.05.016) Copyright © 2013 AGA Institute Terms and Conditions

Figure 4 Acinar cell differentiation is perturbed in Ela-Cre/DicerloxP/− mice. (A) Wild-type (WT) (white) and Ela-Cre/DicerloxP/− (agouti) mice at P15. (B) Acinar-enriched miR-802 is reduced in Ela-Cre/DicerloxP/− pancreata at P0 (real-time quantitative polymerase chain reaction [RT-qPCR]). Values (means ± SEM; n = 4; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001) are normalized to β-actin. (C) Amy expression and apoptosis (cleaved [cl.] caspase3) in Ela-Cre/DicerloxP/− pancreata at P1. (D) Apical polarity markers ezrin, atypical protein kinase C (aPKC), and Mucin 1a in Ela-Cre/DicerloxP/− are diffusely located in acinar cells at P1. Increased expression of HNF6 in Ela-Cre/DicerloxP/− acini at P1 correlates with repression of Mist1 and Rbpjl, but with normal expression of Ptf1a. Bar = 20 μm. (E) Normal expression of early acinar markers (Ptf1a, Rbpj, Amy, Cpa), decreased expression of late markers (Mist1, Rbpjl, Prss1, Cel, Try), and induction of hepatic markers in Ela-Cre/DicerloxP/− pancreata at P1 (RT-qPCR). Given the acinar hypoplasia in Ela-Cre/DicerloxP/− pancreata, marker/β-actin ratios were further normalized for the ratio of acinar vs total cells. Acinar cells-to-total cells ratio is 56.7% ± 1.8% in WT and 12.5% ± 1.2% in Ela-Cre/DicerloxP/−. Total number of cells counted: 9145 in WT and 6341 in Ela-Cre/DicerloxP/− in 2 individuals. The Ct values for Apo4, ApoH, Afp, Alb, and transthyretin are respectively 27.7, 26.2, 25.8, 25.7, and 24.3 in Ela-Cre/DicerloxP/− pancreas. Data are means ± SEM; n = 5; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. Gastroenterology 2013 145, 668-678.e3DOI: (10.1053/j.gastro.2013.05.016) Copyright © 2013 AGA Institute Terms and Conditions

Figure 5 Let-7a and miR-495 target the HNF6 3′UTR and are expressed during acinar differentiation. (A) Transient transfection of HEK293 cells with a HNF6 3′UTR reporter plasmid and miR-Mimics shows that most miRNAs, selected by bioinformatic analyses, repress reporter activity by 20%−70%. Data are means ± SEM; n = 4. (B) EYFP-expressing acinar cells were fluorescence-activated cell sorted from Ela-Cre/Rosa26R-EYFP mice and expression of the miRNAs that most potently repress HNF6-luciferase reporter activity was measured by real-time quantitative polymerase chain reaction. Gastroenterology 2013 145, 668-678.e3DOI: (10.1053/j.gastro.2013.05.016) Copyright © 2013 AGA Institute Terms and Conditions

Figure 6 Regulation of acinar differentiation by let-7b and mir-495 depends on HNF6. (A) real-time quantitative polymerase chain reaction (RT-qPCR) analysis of acinar cells from mice injected with phosphate-buffered saline (control) or cerulein shows down-regulation of let-7b, miR-495, and of late acinar markers, and up-regulation of Hnf6, transthyretin, and Alb. Expression of Afp, ApoA4, and ApoH was not detected. EYFP-expressing acinar cells were fluorescence-activated cell sorted from Ela-Cre/Rosa26R-EYFP pancreas. Gene expression data (means ± SEM; n = 4. ∗P < .05; ∗∗P < .01) were normalized to control pancreas. (B) Induction of HNF6 in acini after cerulein injection correlated with down-regulation of Mist1, Rbpjl, and Amy. Bar = 20 μm. (C) Transfection of anti−let-7b (upper panel) and anti−miR-495 (lower panel) in AR42J cells up-regulates expression of Hnf6 and of most hepatic and ductal markers, while down-regulating most late acinar markers. The effect of anti−let-7b and anti−miR-495 antagomirs is nearly completely abrogated when HNF6 expression is inhibited by anti-HNF6 siRNA (siHNF6). Gene expression data (RT-qPCR; means ± SEM; n = 4; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001) were normalized to control (anti-miR Ctl+ siCtl). Asterisks indicate significant differences between antagomir and antagomir + siHNF6 treatment. (D) Model of the gene network that controls differentiation of pancreatic acinar cells. Gastroenterology 2013 145, 668-678.e3DOI: (10.1053/j.gastro.2013.05.016) Copyright © 2013 AGA Institute Terms and Conditions