Stabilization of HIF-2α impacts pancreas growth

Hypoxia inducible factors (HIFs) are critical regulators of the response to oxygen deficiency by activating target genes involved in a variety of biological functions. HIFs have been implicated in the pathophysiology of numerous pathologies including cancer. Patients with mutations in the von Hippel-Lindau (VHL) gene, an essential regulator of HIF activity, develop tumors in several organs including the pancreas. Previous functional studies of HIF activation in the pancreas have used Vhlh (the murine homolog of VHL) deficient mice. However, the role of each specific HIF transcription factors in the pancreas has not been thoroughly examined. We derived mice that constitutively express a normoxia-stable form of HIF2α in the pancreas. Activation of HIF2α in the pancreas severely impairs postnatal exocrine pancreas. Mice with pancreas-specific activation of HIF2α develop histological features reminiscent of pancreatitis including loss of acinar cells, ductal dilation and fibrosis. Moreover, we provide evidence that signaling pathways important for acinar cell homeostasis are altered in HIF2α-overexpressing pancreata.

the activation of HIF pathway on pancreas formation. To this end, we derived mice that constitutively express HIF2α in the pancreas. We focused on HIF2α given the prominent role of this HIF factor on proliferation, apoptosis and tumor formation 23 . Also, HIF2α appears to play a critical role in pancreas embryonic development 24 . We show that activation of HIF2α in the pancreas severely impairs postnatal exocrine homeostasis. Mice with pancreas-specific activation of HIF2α display histological features reminiscent of pancreatitis including loss of acinar cells, ductal dilation and fibrosis. Moreover, we provide evidence that signaling pathways important for acinar cell homeostasis are altered in HIF2α-overexpressing pancreas.

Results
Specific activation of HIF2α in the pancreas. To investigate the role of HIF2α in pancreas formation, we specifically activated HIF2α in the pancreas by crossing mice expressing Cre recombinase under control of the pancreatic and duodenal homeobox 1 (Pdx1) promoter (Pdx1-Cre mice) 25 with HIF2dPA mice (Pdx1-Cre;HIF2dPA mice). In HIF2dPA mice 26 , a form of human HIF2α that escapes recognition by the von Hippel-Lindau tumor suppressor protein (i.e. it is not degraded under normal oxygen conditions) is expressed upon Cre recombination. Pdx1-Cre;HIF2dPA mice were born at the expected Mendelian ratio and reached adulthood without any sign of compromised health. No differences in total body weight were found between Pdx1-Cre;HIF2dPA mice and littermate control mice at both 2 and 8 weeks of age (Fig. 1A). However, a substantial reduction in total pancreatic mass of Pdx1-Cre;HIF2dPA mice was found as early as 2 weeks of age, even when pancreatic weight was normalized to body weight (Fig. 1A). Western Blot (Fig. 1B) and immunofluorescence analysis (Fig. 1D) confirmed the efficient accumulation of HIF2α in Pdx1-Cre;HIF2dPA pancreas. In control pancreas, endogenous HIF2α expression was found in islets but not in the exocrine compartment (Fig. 1C), as previously reported 20 . Exocrine cell atrophy and ductal dilation in postnatal Pdx1-Cre;HIF2dPA pancreata. Gross morphology inspection and analysis of Haematoxylin and Eosin-stained tissue of newborn (P0) Pdx1-Cre;HIF2dPA pancreata did not reveal any apparent defects (Fig. 1E,F). However, loss of acinar cells and increased dilation of duct-like structures was observed at 2 weeks of age (Fig. 1I,J). These lesions rapidly progressed, and by 8 weeks of age most of the Pdx1-Cre;HIF2dPA acinar tissue had been replaced by duct-like tubular structures, stroma and adipose tissue (Fig. 1M,N). To further characterize the pancreatic exocrine defects of Pdx1-Cre;HIF2dPA mice we performed double immunohistochemistry for a ductal marker, duct-specific cytokeratin 19 (KRT19), and an acinar cell marker, amylase. The Pdx1-Cre;HIF2dPA pancreata appeared normal at P0 (Fig. 1G,H) but beginning at 2 weeks of age amylase expression was substantially reduced concomitant with expansion of duct-like tubular structures (Fig. 1K,L). The acinar architecture appeared disorganized in two-week-old Pdx1-Cre;HIF2dPA mice, marked by the loss of the typical round morphology of the acini as well as the dilation of the intracinar lumen (Fig. 1L) and concomitant increase of KRT19-positive epithelial cells. This acinar-ductal metaplasia, broadly defined as replacement of acinar cells by ductal-like cells, did not seem to involve transdifferentiation of acinar cells into ductal-like cells since intermediate cells co-expressing acinar and ductal markers were not observed ( Fig. 1L and Supplementary Fig. 1). By 8 weeks of age, the Pdx1-Cre;HIF2dPA pancreata appeared severely atrophic with extensive loss of acinar cells and dilated ducts; only residual clusters of acinar cells could be observed (Fig. 1O,P). Islets of Pdx1-Cre;HIF2dPA mice displayed a morphology similar to those observed in control mice ( Supplementary Fig. 1). Indeed, blood glucose levels in Pdx1-Cre;HIF2dPA mice were normal compared to control mice ( Supplementary Fig. 1). These results indicate that ectopic HIF2α stabilization in the pancreas leads to extensive loss of acinar cells and expansion of duct-like cells.
HIF2α activation leads to pancreatic abnormalities reminiscent of pancreatitis. Loss of acinar cells and ductal expansion are histological hallmarks of pancreatitis. We examined whether other classical features of pancreatitis were also present in Pdx1-Cre;HIF2dPA postnatal pancreata. Haematoxylin and Eosin staining showed increased stroma in 8-week-old Pdx1-Cre;HIF2dPA pancreata ( Fig. 1M,N). Gomori trichome staining revealed an marked increase in connective tissue in Pdx1-Cre;HIF2dPA mice ( Fig. 2A,B). Pdx1-Cre;HIF2dPA pancreatic tissue exhibited a substantial increase in cells expressing smooth muscle actin (SMA), a fibroblast activation marker (Fig. 2C,D) and vimentin, a mesenchymal marker (Fig. 2E-G), thus confirming the formation of pancreatic fibrosis. An increase in inflammatory cells was also apparent in 8-week-old Pdx1-Cre;HIF2dPA mice, as evidenced by CD11b immunohistochemistry, a marker of leukocytes (macrophages, neutrophils and granulocytes) (Fig. 2H-J) and F4/80 immunohistochemistry, a specific marker of macrophages ( Fig. 2K-M). Thus, HIF2α activation results in fibrosis and increased inflammatory cells in the pancreas, histological features of pancreatitis.
Increased acinar cell apoptosis and proliferation in Pdx1-Cre;HIF2dPA mice. To elucidate the underlying cause of the massive acinar loss in Pdx1-Cre;HIF2dPA mice, we assessed acinar cell proliferation, apoptosis and differentiation in two-week-old mice, an age in which Pdx1-Cre;HIF2dPA pancreata exhibit areas with preserved acinar tissue. Apoptosis was measured using TUNEL assay. Pdx1-Cre;HIF2dPA pancreata showed a marked increase in apoptotic cells compared to control pancreata ( Fig. 3A-C). Immunohistochemistry for cleaved caspase-3 confirmed the increase in apoptotic cells of Pdx1-Cre;HIF2dPA pancreata ( Fig. 3D,E). Cell proliferation was assessed by Ki-67 immunohistochemistry. Control mice displayed substantial pancreatic cell proliferation, as expected at this early postnatal age. However, both acinar (Fig. 3F,G,J) and ductal cell (Fig. 3H,I,J) proliferation was significantly higher in Pdx1-Cre;HIF2dPA pancreata. Thus, despite extensive acinar cell loss, pancreatic cell proliferation was prominent in Pdx1-Cre;HIF2dPA mice.
Remodeling of pancreatic acini upon HIF2α activation. Although the Pdx1-Cre;HIF2dPA pancreata appeared largely unaffected at birth, a substantial replacement of acinar cells by cells expressing ductal markers was observed as early as two weeks of age. We analyzed whether acinar cell dedifferentiation could also play a role in this postnatal exocrine pancreatic degeneration. Close examination of acini in two-week-old Pdx1-Cre;HIF2dPA mice revealed that dilation of intracinar ducts was associated with reduction of amylase immunoreactivity ( Fig. 1L and Supplementary Fig. 1). A key characteristic of the mature exocrine pancreas is the architecture of acini consisting of individual acinar cells that exhibit apical-basal cellular polarity and secrete digestive enzymes into a central lumen. Disruption of acinar cell organization and apical-basal polarity has been associated with postnatal pancreatic exocrine degeneration in various mutant mouse strains [27][28][29] . Thus, we decided to analyze the accumulation of several markers of acinar cell organization in Pdx1-Cre;HIF2dPA mice at two weeks of age when substantial acinar tissue was still preserved. The localization of the basal marker laminin (Fig. 4A,B) as well as basolateral markers β-catenin (Fig. 4C,D) and E-cadherin ( Fig. 4E-L) was not largely affected in Pdx1-Cre;HIF2dPA acini. In control acinar cells, carboxypeptidase A1 (CPA1) is localized close to the lumen of the acini (Fig. 4E). However, CPA1 did not show a clear apical localization in Pdx1-Cre;HIF2dPA acinar cells (Fig. 4F). Immunofluorescence for apical markers mucin-1 (Fig. G,H) and PKCε (Fig. 4I,J) confirmed the mislocalization of apical markers in Pdx1-Cre;HIF2dPA mice. Mucin-1 and PKCε were localized exclusively at the apical surface of control acinar cells (Fig. 4G,I). While these proteins still displayed an apical localization in Pdx1-Cre;HIF2dPA acinar cells, a marked cytoplasmic accumulation was observed (Fig. 4H,J). Immunofluorescence for mucin-1 also revealed the dilation of acini lumens in Pdx1-Cre;HIF2dPA mice ( Fig. 4H) compared to the small lumens found in control acini (Fig. 4G). The increase in lumen size of Pdx1-Cre;HIF2dPA acini was confirmed by immunofluorescence analysis of the tight junction protein ZO-1 (Fig. 4K,L). Therefore, HIF2α stabilization leads to extensive acini disorganization including intracinar lumen dilation and mislocalization of apical markers while basal markers are appropriately localized.  VHL patients commonly exhibit pancreatic lesions that include cysts and microcystic adenomas 10,11 . These lesions display prominent fibrous stroma and endothelial cells 11 . Also, epithelial cells lack the characteristic papillary structures and mucin found in pancreatic mucinous cystic neoplasms 11 . Similar to the pancreatic lesions of VHL patients, the duct-like tubular structures of Pdx1-Cre;HIF2dPA pancreata did not exhibit mucin-producing papillary structures, as revealed by alcian blue (Fig. 5A,B) and PAS staining (Fig. 5C,D). Indeed, Pdx1-Cre;HIF2dPA duct-like epithelial cells expressed markers of ductal cells including KRT19 (Fig. 1L,P) and duct binding lectin Dolichos biflorus agglutinin (Fig. 2C,D). Also, cilia, which are normally present in control ductal cells 30,31 but not in pancreatic intraepithelial neoplasia (PanIN) lesions and pancreatic cancer cells 32 , were found in pancreatic duct-like epithelial cells of Pdx1-Cre;HIF2dPA mice ( Fig. 5E,F). Pdx1-Cre;HIF2dPA pancreata displayed increased vascularization (Fig. 5G-K). MECA32-positive endothelial cells were closely intermixed with mucin-1-positive epithelial cells (Fig. 5I,J), as described in pancreatic cysts of VHL patients 11 . In agreement with the increased vascularization, an upregulation of the expression of the HIF-target gene VEGF was found in Pdx1-Cre;HIF2dPA pancreata ( Fig. 5K). Thus, HIF2α stabilization in the pancreas leads to the formation of pancreatic lesions similar to those found in the pancreas of VHL patients.

Ductal metaplasia and increased vascularization in
Altered signaling pathways in exocrine pancreas upon HIF2α activation. The expression of ductal transcription factors such as Sox9 is increased during acinar-ductal metaplasia in models of pancreatic injury [33][34][35] . In agreement with this, we observed prominent Sox9 expression in both disorganized acini and well-defined duct-like tubular structures of Pdx1-Cre;HIF2dPA pancreata (Fig. 5L,M). The MAPK signaling pathway plays an important role in acinar cell homeostasis in response to injury. During pancreatitis the MAPK pathway is up-regulated 36,37 and it has been shown to be necessary for acinar-ductal metaplasia 38,39 . We evaluated activation of the MAPK pathway, as assessed by immunofluorescence of phosphorylated ERK (pERK). In control pancreata, only a few scattered cells were positive for pERK (Fig. 6A, top panels). In stark contrast, strong nuclear pERK signal was observed in Pdx1-Cre;HIF2dPA acinar cells. This activation of MAPK pathway was largely limited to disorganized acini (Fig. 6A, bottom panels), the preserved acini in Pdx1-Cre;HIF2dPA mice did not shown substantial pERK signal (Fig. 6A, bottom panels). Of note, prominent MAPK activity was also observed in the stromal cells of Pdx1-Cre;HIF2dPA pancreata (Fig. 6A, bottom panels). The AKT/mTOR pathway has also been implicated in acinar-ductal metaplasia [40][41][42] . Thus, we evaluated the phosphorylation status of the ribosomal protein S6 (RPS6), a downstream target of the AKT/mTOR signaling pathway. As previously shown 42,43 , RPS6 was phosphorylated in acinar (Fig. 6B, top panels), but not ductal cells of control pancreata (Fig. 6C, top panels). In the pancreas of Pdx1-Cre;HIF2dPA mice, phosphorylated RPS6 (pRPS6) was easily detected in well-preserved acini (Fig. 6B, bottom panels). However, a dramatic decrease in pRPS6 was observed in areas with disorganized acini (Fig. 6B, bottom panels). Interestingly, we observed an induction of the stress protein clusterin in disorganized acini (Fig. 6D, bottom panels). No expression of clusterin was detected in normal acini of Pdx1-Cre;HIF2dPA mice (Fig. 6D, bottom panels) as well as in acini of control mice (Fig. 6D, top panels). Thus, HIF2α activation in the pancreas affects signaling pathways important for exocrine pancreas homeostasis.

Discussion
Here, we show that specific activation of HIF2α in the pancreas severely disrupts homeostasis of the exocrine pancreas. We derived transgenic mice that specifically expressed a normoxia-stable form of HIF2α in the pancreas. Continuous activation of HIF2α in pancreatic cells resulted in severe abnormalities in the postnatal pancreas including loss of acinar cells, acinar-ductal metaplasia, increased vascularization, and fibrosis.
To specifically activate HIF2α in the pancreas, we used the Pdx1-Cre transgenic line. Pdx1 is expressed during early stages of the developing pancreatic epithelium and HIF2α is expressed in pancreatic progenitor cells 24 . Indeed, HIF2α inactivation causes pancreatic hypoplasia indicating an important role for HIF2α in pancreas development 24 . However, newborn Pdx1-Cre;HIF2dPA pancreata did not display apparent pancreatic abnormalities thus indicating the HIF2α overactivation do not overtly affect embryonic development of the pancreas. After birth, a gradual loss of acinar cells was observed in Pdx1-Cre;HIF2dPA pancreata. Indeed, Pdx1-Cre;HIF2dPA mice develop features reminiscent of chronic pancreatitis including acinar cell loss, acinar-ductal metaplasia, inflammatory cell infiltration, fibrosis and lipomatosis. Despite the extensive acinar cell loss, Pdx1-Cre;HIF2dPA mice did not display loss of body weight after birth. However, it should be noted that the loss of pancreatic enzyme synthesis needs to be almost complete to cause overt physiological defects 44,45 .
Concomitant with the acinar cell loss, a marked increase in duct-like tubular structures was observed in Pdx1-Cre;HIF2dPA pancreata. Acinar-ductal metaplasia is commonly observed in chemical models of pancreatic injury as well as in several mutant mice deficient in genes critical for exocrine cell homeostasis and it is considered an adaptive response to acinar cell injury 46,47 . Although increased acinar cell proliferation was prominent in Pdx1-Cre;HIF2dPA mice, this did not result in tissue regeneration. Indeed, two-month-old Pdx1-Cre;HIF2dPA mice displayed a massive loss of acinar cells indicating that sustained HIF2α was detrimental to acinar cell regeneration. Lineage tracing experiments have demonstrated transdifferentiation of acinar cells into ductal-like cells in several animal models of acinar cell injury (reviewed in 47,48 ). In our study, lineage tracing could not be utilized to determine the acinar origin of ductal-like cells in Pdx1-Cre;HIF2dPA pancreata because the Pdx1-Cre transgene is expressed in common pancreatic progenitors. We did not detect intermediate cells co-expressing acinar and ductal markers in Pdx1-Cre;HIF2dPA pancreata that could suggest acinar-duct transdifferentiation but this is just circumstantial evidence. We did observe increased cell apoptosis in postnatal Pdx1-Cre;HIF2dPA pancreata. Thus, apoptosis could account for the severe loss of acinar cells although we cannot rule out an additional contribution from transdifferentiation of acinar cells into ductal-like cells. Although HIF2α usually promote cell survival and proliferation, a role of HIF2α in inducing cell apoptosis has also been reported. For example, HIF2α overexpression in hepatocellular carcinoma 49 and glioblastoma cells 50 increases tumor cell apoptosis. The proapoptotic effect of HIF2α overexpression on acinar cells could also be indirect as a response to acinar cell injury, as observed in pancreatitis. In agreement with this notion, we observed expression of the injury marker clusterin in Pdx1-Cre;HIF2dPA pancreata.
The duct-like tubular structures of Pdx1-Cre;HIF2dPA mice are reminiscent of the cysts and microcystic adenomas commonly found in VHL patients 10,11 , which are characterized by increased vascularization and prominent fibrous stroma. However, most pancreatic lesions in VHL patients are asymptomatic and no extensive acinar cell loss is usually found. Only rare cases of pancreatitis have been reported and these appear to be secondary to obstruction of the main pancreatic duct of cysts. A potential explanation for the difference in pancreatic phenotypes between VHL patients and Pdx1-Cre;HIF2dPA mice is that HIFα activation (by VHL inactivation) occurs locally in VHL patients (second hit hypothesis of hereditary cancer) while broad HIFα activation in the pancreas is achieved in Pdx1-Cre;HIF2dPA mice. In agreement with this notion, pancreas-specific Vhlh (the murine homolog of VHL) inactivation in mice results in the formation pancreatic cysts, loss of exocrine pancreas and fat replacement 21 . Interestingly, these pancreatic lesions were observed only in older Vhlh mutant mice, not in young mice 21 , as we observed in Pdx1-Cre;HIF2dPA mice. This apparent discrepancy might be related to the specificity and/or levels of HIF2α activation obtained in our mouse genetic model. Nonetheless, it is important to note that Vhlh mutant mice display severe perinatal lethality 15,21 and thus, the analysis of pancreata in older Vhlh mutant mice was limited to few surviving mice 21 , which could have not achieved full HIF activation on exocrine pancreas. In this regard, it would be interesting to determine which specific HIF transcription factors are activated in the pancreatic lesions of VHL patients. Vhlh mutant mice exhibit impaired glucose homeostasis including severe hypoglycemia and defective insulin and glucagon secretion. Although we did not perform a complete characterization of glucose homeostasis, Pdx1-Cre;HIF2dPA mice showed normal fed glucose levels. This apparent discrepancy might be due to differences in the specific HIFα overactivated between Pdx1-Cre;HIF2dPA and Vhlh mutant mice. In support of this hypothesis, the abnormal glucose homeostasis of ß-cell-specific Vhlh mutant mice is HIF1α-dependent since deletion of HIF1α restores impaired glucose homeostasis 12,16 . Thus, our results suggest that HIF2α overactivation in the pancreas do not markedly alter glucose homeostasis.
The pancreatic defects of Pdx1-Cre;HIF2dPA mice are similar to those observed in mice with defects in primary cilia formation and it has been shown that pVHL contributes to cilia maintenance and stability in kidney cells [51][52][53] . Indeed, cilia are lost in kidney cysts from VHL patients 53 . To the best of our knowledge, it has not been reported whether pancreatic cysts in VHL patients lack primary cilia 54 . However, primary cilia were clearly present on the ductal cells of Pdx1-Cre;HIF2dPA pancreata indicating that HIF2α overactivation does not overtly compromise primary cilia integrity. Interestingly, it has been reported that only HIF1α (not HIF2α) activity is important for cilia formation 53 .
The loss of acini organization in Pdx1-Cre;HIF2dPA mice was associated with alteration of the MAPK and AKT/mTOR signaling pathways. These pathways play a pivotal role in acinar cell homeostasis in response to injury. Thus, MAPK signaling is required for the dedifferentiation of acinar cells into duct-like cells 39 . In concordance with this notion, upregulation of pERK signal was observed in acinar cells located in disorganized acini of Pdx1-Cre;HIF2dPA mice. Interestingly, we also observed prominent MAPK activity in the stromal cells of Pdx1-Cre;HIF2dPA pancreata, a phenomenon also reported in fibrosis associated to pancreas cancer formation 55 . Disorganization of acini structure was also accompanied by decreased levels of phosphorylated RPS6, a downstream target of the AKT/mTOR signaling pathway that has been implicated in acinar-ductal metaplasia [40][41][42] . Crosstalk between these signaling pathways and HIFs have been reported 23,56 , thus providing a potential causal link between HIF2α overactivation and impairment of acinar cell homeostasis.
Depletion of Ptf1a, a master regulator of acinar differentiation and function in adult mice also results in decreased levels of phosphorylated RPS6 in acinar cells concomitant with loss of acinar cell identity 43 . Indeed, the postnatal pancreatic exocrine degeneration observed in Pdx1-Cre;HIF2dPA mice is similar to defects observed in mutant mice deficient in transcription factors critical for acinar cell formation and identity such as MIST1, NR5A2, GATA6 and PTF1A 28,29,43,57 . Pdx1-Cre;HIF2dPA pancreata displayed features compatible with loss of acinar cell identity including decreased expression of amylase, disruption of acini organization, mislocalization of acinar cell apical markers and increased expression of ductal markers. However, these phenotypes could be secondary effects of acinar cell injury, as reported in animal models of pancreatitis. Temporal, synchronous activation of HIF2α in adult acinar cells would be necessary to elucidate the direct role of HIF2α in acinar cell homeostasis.
SCIENtIFIC REPORTS | (2018) 8:13713 | DOI:10.1038/s41598-018-32054-5 A potential role for HIF2α in pancreatic cancer formation has been recently described. HIF2α is highly expressed in preneoplastic lesions but is downregulated during malignant progression of pancreatic cancer 20 . HIF2α inactivation impairs the progression of cancerous lesions in the Kras-mediated mouse model of pancreatic cancer 20 . We did not observe tumor formation in Pdx1-Cre;HIF2dPA mice but our analysis was limited to relatively young mice (two-month-old mice). Future studies in genetic models of pancreatic cancer will help to determine the effect of HIF2α activation in pancreatic cancer.

Methods
Mice. Mice were housed in a pathogen-free facility at the Institute of Biomedicine of Seville (IBiS). All procedures involving experimental animals were performed in accordance with European and local animal welfare laws, guidelines and policies. All animal studies were approved by the IBiS-Virgen del Rocio Ethics Committee. Pdx1-Cre and Hif2dPA (Rosa26Sor Tm4(Hif2A*)Kael ) mice carrying a floxed allele of human HIF2a have been previously described 25,26 . Hif2dPA mice were purchased from Jackson Labs.
Histological and immunohistochemical analyses. Histological (including alcian blue, periodic acid-Schiff and Gomori's one-step trichrome staining) as well as immunohistochemical analyses were performed as previously described 58,59 . Fluorescent and bright-field images were captured using a BX-61 microscope (Olympus) and LTC SP2 confocal microscope (Leica). All photomicrographs shown are representative of at least 3 independent samples of the indicated genotype. Primary antibodies were used at the To quantify cell proliferation, following double immunofluorescence for the proliferation marker Ki-67 (RM-9061; Thermo Scientific) and amylase or KRT19, positive cells were then counted from 10 non-overlapping fields at 200x magnification from 3 control and 3 Pdx1-Cre;Hif2dPA mice. Cell apoptosis was detected using In Situ Cell Death Detection Kit (Roche) according to the manufacturer's protocol in 10 non-overlapping fields at 200x magnification from 4 control and 5 Pdx1-Cre;Hif2dPA mice. Pancreatic sections treated with DNAse I (Sigma-Aldrich) were used as positive control. Acinar cell apoptosis was quantified by averaging the number of TUNEL-positive cells per total number of amylase-positive cells. Morphometric analysis of vimentin, CD11b and F4/80 were performed by using ImageJ software. For CD11b and F4/80, the number of positive cells per field was calculated. The extent of vimentin accumulation was measured as the percentage of vimentin-positive area per total pancreatic area. For each group, 10 randomly selected low-power fields from pancreatic sections at least 60 µm apart in three mice were analyzed. Western Blot analysis. Tissue samples were lysed and homogenized in lysis solution (50 mM Tris-HCl [pH 7.4], 100 mM NaCl, 5 mM MgCl 2 , 0,1% SDS, 1% Triton X-100) and Complete Ultra protease inhibitors (Roche). Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). After SDS-PAGE separation, samples were transferred to PVDF membranes (BioRad), processed for immunoblotting against anti-HA (1:500, 1167423001; Roche) and anti ß-actin (1:1000, ab8226; Abcam) antibodies, developed using HRP-conjugated secondary antibodies (Jackson) and Western ECL Substrate (Bio-Rad). ImageQuant LAS 4000 Mini Gold (GE Healthcare) was used for imaging the blots.

RNA isolation and Quantitative Real Time PCR. RNA isolation was performed using RNeasy Kit
(Qiagen) with deoxyribonuclease treatment, following manufacturer's instruction. Total RNA purity and concentration was assessed using Nanodrop 2000 spectrophotometer (Thermo Scientific, Wilmington, NC, USA), and subsequently reverse-transcribed using random hexamer primers and the cDNA Omniscript reverse transcriptase (Qiagen). Quantitative PCR was performed using with SYBR Green PCR Master Mix (Applied Biosystems) using a 7900HT real-time PCR system (Applied Biosystems). Relative quantification of RNA levels was calculated using the ΔΔCt method. Cyclophilin A (peptidylprolyl isomerase A-Mouse Genome Informatics) was used as housekeeping gene 59 . The results were expressed as fold relative to levels in control pancreata (value of 1). Primer sequences for cyclophylin A and VEGF were as follows: CycloA-forward 5′-TCACAGAATTATTCCAGGATTCATG-3′; CycloA-reverse 5′-TGCCGCCAGTGCCATT-3′; VEGF-forward 5′-CCAGCACATAGGAGAGATGAGCTT-3′; VEGF-reverse 5′-TCTGTCTTTCTTTGGTCTGCATTC-3′.

Statistical analysis.
Significance was determined using two-tailed Student's t-test, and one-way ANOVA (post-hoc Tukey HSD test). P < 0.05 was considered significant. Data are presented as mean ± SD.