2646 Diabetes Volume 66, October 2017

SMAD3/Stat3 Signaling Mediates b-Cell Epithelial- Mesenchymal Transition in Chronic Pancreatitis–Related Diabetes

Xiangwei Xiao,1 Shane Fischbach,1 Tina Zhang,1,2 Congde Chen,1 Qingfeng Sheng,1 Ray Zimmerman,1 Sneha Patnaik,1 Joseph Fusco,1 Yungching Ming,1 Ping Guo,1 Chiyo Shiota,1 Krishna Prasadan,1 Nupur Gangopadhyay,1 Sohail Z. Husain,3 Henry Dong,2 and George K. Gittes1

Diabetes 2017;66:2646–2658 | https://doi.org/10.2337/db17-0537

Many patients with chronic pancreatitis develop diabetes Many patients with chronic pancreatitis develop insulinopenia, (chronic pancreatitis–related diabetes [CPRD]) through an intolerance, resistance, and eventually diabetes undetermined mechanism. Here we used long-term par- (2), largely as a result of the intimate proximity of the tial pancreatic duct ligation (PDL) as a model to study CPRD. endocrine pancreas to the exocrine pancreas (3). Moreover, We found that long-term PDL induced significant b-cell de- patients with chronic pancreatitis often develop a fibrotic differentiation, followed by a time-dependent decrease in pancreas with a reduced b-cell mass and have a 15- to b — fi functional -cell mass all speci cally in the ligated tail 16-fold increased risk for pancreatic (4). To date, the portion of the pancreas (PDL-tail). High levels of transform- understanding of the development and pathogenesis of b b ing growth factor 1(TGF 1) were detected in the PDL-tail chronic pancreatitis–relateddiabetes(CPRD)isverylimited. and were mainly produced by M2 macrophages at the In addition, the mechanisms of b-cell loss in CPRD may in early stage and by activated myofibroblasts at the later part be similar to those in (T2D) (5,6) and in stage. Loss of b-cell mass was then found to result from ISLET STUDIES cystic fibrosis (7). Thus, elucidation of the underlying mech- TGFb1-triggered epithelial-mesenchymal transition (EMT) by b-cells, rather than resulting directly from b-cell apo- anisms common to chronic pancreatitis, CPRD, and pancre- ptosis. Mechanistically, TGFb1-treated b-cells activated atic cancer is critical. expression of the EMT regulator Snail in a SMAD3/ Among animal models for acute and chronic pancreatitis Stat3-dependent manner. Moreover, forced expression of (8), partial pancreatic duct ligation (PDL) has been used to forkhead box O1 (FoxO1), an antagonist for acti- generate chronic pancreatitis in mammals (9,10). Ligation vated Stat3, specifically in b-cells ameliorated b-cell EMT of the pancreatic duct immediately at the beginning of the and b-cell loss and prevented the onset of diabetes in splenic or tail part of the pancreas blocks the drainage of duc- mice undergoing PDL. Together, our data suggest that tal fluid from the distal pancreas, resulting in autodigestion chronic pancreatitis may trigger TGFb1-mediated b-cell of acinar cells and severe inflammation specifically in the EMT to lead to CPRD, which could substantially be pre- ligated tail of the pancreas, although initially the islets are vented by sustained expression of FoxO1 in b-cells. largely unaffected. Acinar cell death in the tail of the pancreas leads to the complete loss of acinar cells, without significant acinarcellregeneration(11).Onthecontrary,thenonligated The prevalence of chronic pancreatitis is roughly 50 per part, or head, of the pancreas appears to be normal, thus of- 100,000 people worldwide (1). Chronic pancreatitis in the fering an excellent internal control (12,13). We recently re- United States results in more than 122,000 outpatient vis- ported an inflammatory molecular signature in PDL, which its and more than 56,000 hospitalizations each year (2). induced b-cell proliferation in a transforming growth factor

1Division of Pediatric Surgery, Department of Surgery, Children’s Hospital of Received 7 May 2017 and accepted 27 July 2017. Pittsburgh, University of Pittsburgh School of Medicine, Pittsburgh, PA This article contains Supplementary Data online at http://diabetes 2 ’ Division of Immunogenetics, Department of Pediatrics, Children s Hospital of .diabetesjournals.org/lookup/suppl/doi:10.2337/db17-0537/-/DC1. Pittsburgh, University of Pittsburgh School of Medicine, Pittsburgh, PA © 2017 by the American Diabetes Association. Readers may use this article as 3Department of Pediatrics, Children’s Hospital of Pittsburgh, University of Pitts- long as the work is properly cited, the use is educational and not for profit, and the burgh School of Medicine, Pittsburgh, PA work is not altered. More information is available at http://www.diabetesjournals Corresponding author: Xiangwei Xiao, [email protected], or George K. .org/content/license. Gittes, [email protected]. diabetes.diabetesjournals.org Xiao and Associates 2647 b (TGFb) signaling–dependent manner (12–15). In Vitro Culture and Treatment of Mouse Islets As a strong stimulant of epithelial-mesenchymal transition Mouse islets were cultured in Ham’s F10 medium (Life Tech- (EMT) in epithelial cells (16–18), TGFb1 is highly upregulated nologies, St. Louis, MO) supplemented with 0.5% BSA (Sigma- in the ligated tail of the pancreas after PDL (PDL-tail) (14). Aldrich, St. Louis, MO), 2 mmol/L glutamine, 2 mmol/L Thus we were prompted to evaluate the effects of PDL- calcium, and 5 mmol/L glucose at 37°C in 95% air and 5% induced TGFb1ontheEMTofb-cells. CO2. After overnight culture, islets were either treated or (FoxO1) is a pivotal factor in not treated with 20 ng/mL TGFb1 (with or without SMAD3 orchestrating the response of b-cell mass and function to and Stat3 inhibitors) and harvested after 0, 0.5, 1, 3, 12, overnutrition and obesity (19) and to and48h.TheSMAD3inhibitorSIS3(2mmol/L) or Stat3 (20–22). FoxO1 is predominantly expressed by b-cells in inhibitor cryptotanshinone (CTSN; 5 mmol/L) or control the adult pancreas. We and others have shown that FoxO1 DMSO solution was added with the TGFb1. nuclear translocation increases NeuroD1, MafA, and Nkx6.1 Western Blotting expression in b-cells, contributing to the maintenance of a Western blotting was performed as previously described functional differentiated phenotype to resist stress-induced (14,30). Primary antibodies for Western blotting are rabbit dedifferentiation, dysfunction, and failure (23–25). Never- polyclonal anti-MafA (Abcam), anti-GAPDH, anti-Snail, anti- theless, a role for FoxO1 during the pathogenesis of CPRD Slug, anti-Twist, anti-ZEB1, anti-ZEB2, anti-SMAD3, anti- is unknown. pSMAD3, anti-Stat3, anti-pStat3, anti-NeuroD1, anti-Nkx6.1, Here we studied long-term PDL (12–24 weeks) as a anti-Pdx1, and anti–epithelial cadherin (E-cadherin), all from model of chronic pancreatitis in humans. We further ana- Cell Signaling Technologies, San Jose, CA). The secondary lyzed the molecular mechanisms underlying the gradual antibody was horseradish peroxidase–conjugated antirabbit b-cell loss in this model, which mimics CPRD in humans. (Dako, Carpinteria, CA).

RESEARCH DESIGN AND METHODS Virus Production AAV serotype 8 vectors were used to generate AAV-RIP-null Mouse Manipulation and AAV-RIP-FoxO1 viruses, as described before (25). All mouse experiments were approved by the Animal Re- search and Care Committee at the Children’s Hospital of Isolation of RNA and Quantitative RT-PCR Pittsburgh and the University of Pittsburgh Institutional An- RNA extraction and quantitative RT-PCR have been de- imal Care and Use Committee. BAC transgenic rat insulin scribed previously (12,13,27). The following primers all were (RIP) Cre reporter (RIP-Cre) mice, MIP-GFP mice purchased from Qiagen (Valencia, CA): CycloA (QT00247709), (green fluorescent protein reporter under the control of a FoxO1 (QT00116186), Pdx1 (QT00102235), NeuroD1 mouse insulin promoter), and Rosa26CAG-mTmG (mTmG) (QT00251265), MafA (QT01037638), Nkx6.1 (QT00143318), mice have been described before (13). These mice and TGFb1 (QT00145250), vimentin (QT00159670), a-SMA C57BL/6 mice were all purchased from The Jackson Labo- (QT00140119), F4/80 (QT00099617), collagen I (QT00 ratory (Bar Harbor, ME). TGFb receptor II (TBR2) fx/fx 162204), E-cadherin (QT00121163), Snail (QT00240940), mice were generous gifts from Professor Stefan Karlsson Slug (QT00098273), ZEB1 (QT00105385), ZEB2 (QT00148 (University of Lund, Sweden) and have been described pre- 995), and fibronection (QT00135758). Quantitative RT-PCR viously (12). All mice were 10-week-old males and had a values were normalized against CycloA, which proved to be C57BL/6 background. PDL was performed and validated as stable across the samples. Fold changes from the control are described elsewhere (12–15). Intraductal viral infusion was showninthefigures. performed as described previously (26). Adeno-associated Histology and Immunohistochemistry virus (AAV) 8 viruses (titration of 1012 genome copy All pancreas samples were fixed and cryoprotected in 30% particles/mL in a 150-mL volume) were delivered via a cath- sucrose overnight before freezing, as described before (12). eter at a rate of 6 mL/min. BrdU-supplemented drinking Masson trichrome staining was performed using a Trichrome water was given to mice 1 week before the analysis, as Stain (Masson) Kit (Sigma-Aldrich). The fluorescent membrane- previously described (12). targeted Tomato (mT) and membrane-targeted EGFP (mG) Pancreatic Digestion, Islet Isolation, and FACS were detected by direct fluorescence. Primary antibodies for for b-Cells immunostaining were guinea pig polyclonal anti-insulin Pancreas digestion, islet isolation, and b-cell purification (Dako), rabbit polyclonal anti-F4/80 (Invitrogen, CA, Carls- from MIP-GFP mice have been described previously, taking bad), goat polyclonal antiamylase (Santa Cruz Biotechnol- advantage of the specific expression of green fluorescent ogy, Dallas, TX), rabbit polyclonal anti-FoxO1 (made by protein on b-cells (12–14,27–29). Dissociated pancreatic H.D.), goat polyclonal anti-E-cadherin (R&D Systems, Los cells were sorted based on a-smooth muscle actin (SMA) pos- Angeles, CA), rabbit polyclonal anti-a-SMA (Abcam), and itivity after the cells were fixed in 4% formalin and then rat polyclonal anti-CD45 (BD Biosciences). BrdU staining sequentially incubated with anti-a-SMA (Abcam, Cambridge, has been described before (12). Secondary antibodies for MA) and phycoerythrin-conjugated antirabbit antibody (BD indirect fluorescent staining were Cy2, Cy3, or Cy5 conju- Biosciences, San Jose, CA). gated rat-, rabbit- goat-, and guinea pig–specific antibodies 2648 b-Cell EMT Contributes to CPRD Diabetes Volume 66, October 2017

(Jackson ImmunoResearch Labs, West Grove, PA). Nu- their loss correlates with b-cell dysfunction and b-cell loss clear staining was performed with Hoechst solution (BD (23–25). As such, we performed PDL on MIP-GFP mice, and Biosciences). b-Cell mass was quantified as described pre- sorted b-cells via FACS at serial time points after PDL to viously (12). asses Pdx1, NeuroD1, Nkx6.1, and MafA mRNA levels in b Data Analysis -cells by quantitative RT-PCR and Western blotting. We de- tected a gradual decrease in both mRNA and protein levels All values are depicted as means 6 standard errors. Five for these in b-cells (Fig. 2L and Supplementary Fig. repeats were analyzed in each condition. All data were sta- 3), suggesting that b-cell dedifferentiation may occur in the tistically analyzed using one-way ANOVA with a Bonferroni correction, followed by the Fisher exact test. Data were tail of the pancreas over the long-term after PDL. considered significant when P , 0.05. b-Cell Loss Over the Long-term After PDL Initially Results From b-Cell EMT Rather Than b-Cell RESULTS We then analyzed whether dedifferentiation of b-cells fol- Long-term PDL Is a Model of Chronic Pancreatitis lowed by b-cell apoptosis may be responsible for the pro- Recent studies of the effects of PDL on b-cell mass have gressive loss of b-cell mass. We performed both cleaved primarily focused on the short-term (0–8weeks).PDLcan caspase 3 (Fig. 3A) and terminal deoxynucleotidyl TUNEL induce robust pancreatitis, as shown here in immunostain- (Fig. 3B) staining to evaluate b-cell apoptosis. To our sur- ing for the pan-leukocyte marker CD45 (Fig. 1A)andthe prise, no significant increase in b-cell apoptosis was de- macrophage-specific marker F4/80 (Fig. 1B). Long-term tected by either method at different time points after PDL is a model of chronic pancreatitis. However, the effect PDL (Fig. 3A and B), suggesting that b-cell loss over the of long-term PDL on b-cell mass has not been studied. long-term after PDL is not primarily attributable to b-cell Fasting blood glucose remained normal (Supplementary apoptosis. Fig. 1), presumably as a result of the head part of the pan- We previously showed that large amounts of TGFb1were creas remaining intact after PDL (PDL-head; Supplementary secreted by M2 macrophages in the PDL-tail, which led to Fig. 2). Here, with gross imaging, we detected increasing SMAD7 expression and TGFb receptor signaling in b-cells exocrine tissue atrophy and fibrosis in the PDL-tail over to coordinate a modest increase in b-cell proliferation. These the12to24weeksafterPDL(Fig.1C–F), which was con- actions occurred in conjunction with simultaneously acti- firmed by fibrosis quantified with Masson trichrome stain- vated epidermal growth factor receptor signaling (14). ing (Fig. 1G) and by quantification of transcript levels of TGFb1 is also a potent and well-defined trigger for EMT, a fibrotic markers—collagen I, fibronectin, and vimentin—in process by which epithelial cells lose their cell polarity and the PDL-tail compared with the PDL-head (Fig. 1H). These cell-cell adhesion and gain migratory and invasive proper- findings are consistent with what has been seen in patients ties to become mesenchyme-like cells (16–18). Because we with chronic pancreatitis (2) and suggest that long-term detected high TGFb1 levels (14) and progressive fibrosis in PDL is a reasonable model of human chronic pancreatitis. the PDL-tail (Fig. 1A–G), we hypothesized that TGFb1may also trigger EMT of b-cells. Loss of E-cadherin is well estab- Gradual b-Cell Loss Is Detected Over the Long-term lished as a fundamental event in EMT, during which the After PDL factors ZEB1, ZEB2, Snail, and Slug bind di- When we examined the morphology of the pancreas over rectly to E-box consensus sequences on the promoter region time after PDL, we found that islets were easily visualized of E-cadherin and repress its expression (16–18). Thus we with the naked eye early after PDL as a result of elimination analyzed expression levels of these EMT-related genes in of acini, as shown grossly in Fig. 1C and D and in previous purified b-cells from the tails of pancreases of MIP-GFP publications (12,13,31). Surprisingly, however, although ac- mice after PDL. We found that ZEB1 and Snail protein inar cells do not regenerate after PDL, islets were barely vis- levels in b-cells increased with time, whereas E-cadherin ible by 12 weeks after PDL (Fig. 1E) and were not grossly levels decreased with time. In addition, ZEB2 and Slug mRNA visible at all by 24 weeks after PDL (Fig. 1F). Further, in- (but not protein) increased with time (Fig. 3C and Supple- sulin staining and b-cell mass quantification revealed a grad- mentary Fig. 3). These results suggest that an EMT process ual b-cell loss, with a time-dependent decrease in b-cell may indeed occur in b-cells over time after PDL. mass, all specifically in the PDL-tail (Fig. 2A–K). Thus, To further confirm this possibility, we performed b-cell loss occurs over the long-term after PDL. a lineage tracing study using our previously published b-Cell Dedifferentiation Occurs Over the Long-term RIP-Cre;mTmG mouse model (13). These mice should ex- After PDL press membrane-bound red fluorescence in all cells except Recent studies strongly suggest that protection of the dif- insulin-positive b-cells and their progeny, whose floxed mT ferentiated phenotype of existing b-cells is critical to main- cassette is deleted, leading to constitutive expression of the tain a functional b-cell mass and thus to prevent T2D mG cassette located just downstream. We noticed that fi- (23,24). In line with this concept, the b-cell–specifictran- broblasts tend to not express the mT well, and therefore scription factors Pdx1, NeuroD1, Nkx6.1, and MafA seem many of the cells in a fibrotic pancreas do not fluoresce. to be required for b-cells to be fully functional, whereas In these mice, if any b-cells undergo dedifferentiation to diabetes.diabetesjournals.org Xiao and Associates 2649

Figure 1—PDL is a model for both acute and chronic pancreatitis in humans. A and B: Immunostaining with CD45 (A)andF4/80(B)inthePDL- tail 1 week after PDL (PDL1w). C–F: Gross images of the pancreas 1, 4 (PDL4w), 12 (PDL12w), and 24 (PDL24w) weeks after PDL. G: Quantification of pancreatic fibrosis in the PDL-tail and PDL-head based on the percentage of the area positively stained by Masson trichrome staining. H: Quantification of transcript levels of the fibrotic markers collagen I, fibronectin, and vimentin in the PDL-tail compared with the PDL- head through quantitative RT-PCR. Scale bars are 50 mminA and B and 200 mminC–F. Values were normalized to the sham-treated tail. *P < 0.05, PDL1w vs. sham; &P < 0.05, PDL4w vs. PDL1w; #P < 0.05, PDL12w vs. PDL4w (n = 5). HO, Hoechst nuclear staining; INS, insulin; NS, not significant.

become insulin-negative, they still keep mG expression and enrichment in the PDL-tail (Fig. 4A). Because tissue fibrosis appear green. We detected quite a few mG-positive/vimentin- and b-cell EMT seem to occur over the long-term after PDL, positive/insulin-negative cells 12 weeks after PDL; these cells we hypothesized that either TGFb1 levels remain high over likely represent b-cells that have undergone EMT (Fig. 3D). the long-term after PDL because the TGFb1comesfrom These data strongly suggest the occurrence of a b-cell EMT, other sources, or non-TGFb1 EMT triggers may be in- which may be a critical contributor to b-cell loss over the volved. We analyzed the transcript levels of TGFb1inthe long-term after PDL. PDL-tail over time and found that they remained high, TGFb1 Is a Necessary Trigger of b-Cell EMT After PDL peaking ;4 weeks after PDL, when F4/80 transcript levels We previously reported that PDL entails recruitment of M2 were already significantly decreased (Fig. 4A). These data macrophages that secrete large amounts of TGFb1(14). suggest that TGFb1 may be the trigger of tissue fibrosis and However, macrophage infiltration in the PDL-tail peaked b-cell EMT but that macrophages may not be the major 1 week after PDL and then subsequently decreased, as shown source of TGFb1 at these later time points after PDL previously (12–14), and as shown here by F4/80 transcript (14). Indeed, previous studies have highlighted a specific 2650 b-Cell EMT Contributes to CPRD Diabetes Volume 66, October 2017

Figure 2—Gradual b-cell loss is detected over the long-term after PDL. A–J: Representative images of immunohistochemistry for insulin (INS) by 3-39-diaminobenzidine (A–E) and for INS (green) and amylase (AMY; red) by fluorescence (F–J). K:Quantification of b-cell mass in the PDL-tail at serial time points after PDL. L: PDL was performed in MIP-GFP mice and b-cells were isolated by flow cytometry based on green fluorescent protein at different time points after PDL in order to analyze the transcript levels of NeuroD1, Nkx6.1, Pdx1, and MafA using quantitative RT-PCR. Scale bars are 50 mm. Values were normalized to the sham-treated tail (n =5).*P < 0.05, 1 week after PDL (PDL1w)/4 weeks after PDL (PDL4w) vs. sham; #P < 0.05, 12 weeks after PDL (PDL12w) vs. PDL1w/PDL4w; &P < 0.05, 24 weeks after PDL (PDL24w) vs. PDL12w. HO, Hoechst nuclear staining; NS, not significant.

cell population called myofibroblasts, also called pancreatic macrophages in pancreatic tissue remodeling, fibrosis, and stellate cells (32), which, when activated, produce high lev- b-cell EMT. els of TGFb1 to promote EMT and fibrosis (16–18). Be- To confirm TGFb1 as a necessary trigger for b-cell EMT cause a-SMA is a specificmarkerformyofibroblasts, we in this model, we performed PDL on b-cell–specificTBR2- examined the transcript levels of a-SMA in the PDL- null mutant mice (RIP-Cre; TBR2fx/fx) (12,13) (Fig. 5A). tail and found a significant and sustained increase in While b-cell–specific suppression of TGFb1-activated a-SMA by 4 weeks after PDL (Fig. 4A); this was further TGFb receptor signaling did not alter pancreas fibrosis in confirmed by immunostaining (Fig. 4B). Moreover, when PDL-tails (Fig. 5B), the gradual loss of b-cell mass over the we sorted a-SMA–negative and a-SMA–positive cells from long-term after PDL was significantly reduced (Fig. 5C and the PDL-tail, we detected significantly higher levels of D). These data strongly suggest that TGFb1 is a necessary TGFb1inthea-SMA–negative population than in the trigger for b-cell EMT and b-cell loss over the long-term a-SMA–positive population 1 week after PDL, but signifi- after PDL. cantly higher levels of TGFb1inthea-SMA–positive pop- ulationthaninthea-SMA–negative population 4 weeks TGFb1 Triggers EMT of b-Cells Through a SMAD3/Stat3 after PDL (Fig. 4C). This indicates that activated myofibro- Signaling Cascade blasts may be the major source of TGFb1 over the long- To understand the molecular basis underlying TGFb1- term after PDL and may play a more important role than activated b-cell EMT, we treated cultured primary mouse diabetes.diabetesjournals.org Xiao and Associates 2651

Figure 3—b-Cell loss over the long-term after PDL results from b-cell EMT rather than b-cell apoptosis. A and B: Quantification of the percentage of caspase3-positive (A) and TUNEL-positive (B) b-cells at different time points after PDL. C: PDL was performed in MIP-GFP mice and b-cells were sorted to analyze transcript levels for E-cadherin, ZEB1, ZEB2, Snail, and Slug using quantitative RT-PCR. D: PDL was performed in RIP- Cre;mTmG mice; a representative confocal image at 12 weeks after PDL is shown. Yellow arrows point to mG-positive/vimentin-positive/insulin (INS)-negative cells, also shown in high magnification. mT and insulin appear red; mG, green; and vimentin, blue. Values were normalized to the sham-treated tail (n = 5). Scale bar is 50 mm. #P < 0.05, 12 weeks after PDL (PDL12w) vs. 4 weeks after PDL (PDL4w). NS, nonsignificant; PDL1w, 1 week after PDL; PDL24w, 24 weeks after PDL.

islets with 20 ng/mL TGFb1 and harvested the cells at 0, 0.5, treatment, and Snail expression seemed to be sustained 1, 3, 12, 24, and 48 h after TGFb1 administration. Western (Fig. 6A). In addition, Snail mRNA was activated in a pat- blotting was performed to examine the EMT regulatory pro- tern similar to that of the protein (Fig. 6B), suggesting that teins Twist, Snail, Slug, ZEB1, and ZEB2; among these, Snail Snail may be activated primarily at the transcription level. was the only one found to be strongly activated (Fig. 6A). Thus we focused on the regulation of Snail by TGFb signal- Moreover, Snail was activated as early as 1 h after TGFb1 ing. SMAD3 is a key factor in mediating signal transduction 2652 b-Cell EMT Contributes to CPRD Diabetes Volume 66, October 2017

Figure 4—TGFb1 is secreted by M2 macrophages and activated myofibroblasts in the PDL-tail. A: Quantification of transcript levels of F4/80, TGFb1, and a-SMA in the PDL-tail using quantitative RT-PCR. Values were normalized to the sham-treated tail (n =5).B:Representativeimages of immunostaining for a-SMA in the PDL-tail. C: Quantitative RT-PCR for TGFb1insorteda-SMA–negative and a-SMA–positive cells from the PDL-tail. Values were normalized to a-SMA–negative cells from the tail of the pancreas 1 week after PDL (PDL1w). Scale bar is 50 mm. *P < 0.05, PDL1w vs. sham; &P < 0.05, 4 weeks after PDL (PDL4w) vs. PDL1w; #P < 0.05, 12 weeks after PDL (PDL12w) vs. PDL4w. INS, insulin; PDL24w, 24 weeks after PDL.

of activated TGFb receptors. Activated TGFb/SMAD3 has of Stat3 , consistent with a previous been shown to induce phosphorylation and activation of report (33). Stat3 to promote the transcription of Snail (33). TGFb1- treated mouse islets were cotreated with either a specific FoxO1 Prevents EMT of b-Cells Over the Long-term SMAD3 phosphorylation inhibitor (SIS3) or a specific Stat3 After PDL phosphorylation (Tyr705) inhibitor (CTSN) in order to sup- FoxO1 is a that plays a key pro- press phosphorylation-associated activation of SMAD3 and tectiveroleagainstb-cell failure during stress (23–25). Stat3, respectively. Islets were analyzed 12 h after treat- Previous reports have shown that FoxO1 and Stat3 antag- ment. We found that TGFb1 induced phosphorylation of onize each other in the regulation of T-cell function (34,35) SMAD3 and Stat3 and activation of Snail (Fig. 6C–F). Sup- and leptin signaling transduction (36,37). We were thus pression of SMAD3 phosphorylation by SIS3 prevented prompted to evaluate the role of FoxO1 during the process both TGFb1-induced phosphorylation of Stat3 and TGFb1- of b-cell EMT in this CPRD model, because the inflamma- induced activation of Snail, but suppression of Stat3 phos- tory environment in the PDL-tail presumably induces sig- phorylation by CTSN prevented only TGFb1-induced nificant local cell stress. We performed immunostaining for activation of Snail, without affecting TGFb1-induced phos- FoxO1 and, compared with the PDL-head and the sham- phorylation of SMAD3 (Fig. 6C–F). Thus TGFb1maytrig- operated control, found enhanced expression and nuclear ger EMT of b-cells through a SMAD3/Stat3 signaling localization of FoxO1 in b-cells in the PDL-tail 4 weeks after cascade, and SMAD3 phosphorylation occurs upstream PDL, suggesting that FoxO1 may be activated during b-cell diabetes.diabetesjournals.org Xiao and Associates 2653

Figure 5—TGFb1 is a necessary trigger of b-cell EMT in the PDL-tail. A: Schematic of a b-cell–specific TBR2-null mutant mouse. B: Quan- tification of fibrosis in the PDL-tail based on the area positively stained with Masson trichrome staining; RIP-Cre;TBR2fx/fx and control TBR2fx/fx mice were used. C and D:Quantification (C) and representative images (D)ofb-cell mass in the PDL-tail using RIP-Cre;TBR2fx/fx and control TBR2fx/fx mice. Values were normalized to the sham-treated tail (n = 5). Scale bars are 50 mm. *P < 0.05. HO, Hoechst nuclear staining; INS, insulin; NS, nonsignificant; PDL1w, 1 week after PDL; PDL4w, 4 weeks after PDL; PDL12w, 12 weeks after PDL; PDL24w, 24 weeks after PDL.

stress after PDL (Fig. 7A). However, FoxO1 was absent forcibly expressed FoxO1 specifically in b-cells using an from b-cells 12 weeks after PDL (Fig. 7A). To document a intraductal viral infusion technique recently developed in protective role for FoxO1 against EMT of b-cells, we our laboratory (26) (Fig. 7B). Here, forced expression of 2654 b-Cell EMT Contributes to CPRD Diabetes Volume 66, October 2017

Figure 6—TGFb1 triggers EMT of b-cells through a SMAD3/Stat3 signaling cascade. A and B: Cultured primary mouse islets were treated with 20 ng/mL TGFb1andharvestedat0,0.5,1,3,12,24,and48hafterTGFb1 administration. A: Representative Western blotting for the EMT regulatory Twist, Snail, Slug, ZEB1, and ZEB2. B: Results of quantitative RT-PCR for Snail. C–F: Cultured primary mouse islets were treated with 20 ng/mL TGFb1aloneorwith2mmol/LSIS3and5mmol/L CTSN and harvested 12 h after TGFb1 administration. C: Representative Western blotting for SMAD3, pSMAD3, Stat3, pStat3, and Snail. D–F:Quantification of the pSMAD3-to-SMAD3 ratio (D), the pStat3-to-Stat3 ratio (E), and the Snail-to-GAPDH ratio (F). *P < 0.05. N = 5 mice for each experimental group. NS, nonsignificant; und, undetected.

FoxO1 in b-cells by intraductal infusion of AAV carrying and increased E-cadherin (Fig. 7E). Together, these data sug- recombinant FoxO1 under RIP in mice undergoing PDL gest that FoxO1 may prevent b-cell EMT over the long- prevented FoxO1 loss in islets after PDL (Fig. 7C)and term after PDL (Fig. 7F). significantly ameliorated b-cell loss over time (Fig. 7D). Moreover, forced expression of FoxO1 by AAV-mediated DISCUSSION gene transfer in purified b-cells from tails of pancreases Chronic pancreatitis is highly prevalent (1,2) and predis- 12 weeks after PDL inhibited activation of EMT genes poses patients to a high risk of developing diabetes diabetes.diabetesjournals.org Xiao and Associates 2655

Figure 7—FoxO1 seems to play a pivotal role in preventing b-cell EMT after PDL. A: Representative images of immunostaining for FoxO1 in the PDL-tail 4 weeks (PDL4w) and 12 weeks (PDL12w) after PDL. B: Schematic of the experiment. AAV carrying either RIP-FoxO1 or RIP-null as a control was intraductally infused immediately before PDL, and b-cell mass was then analyzed at 12 and 24 weeks (PDL24w) after PDL. C:FoxO1 mRNA in mouse islets. Values were normalized to the sham-treated tail at week 0. D: Analysis of b-cell mass showed that expression of FoxO1 specifically in b-cells, forced by intraductal viral infusion, significantly reduced b-cell loss over the long-term after PDL. E: b-Cells were isolated 2656 b-Cell EMT Contributes to CPRD Diabetes Volume 66, October 2017

(CPRD) (5). The current treatment of chronic pancreatitis pancreatitis in many humans. Third, acute pancreatitis by includes surgical removal of part or all of the pancreas, with PDL progresses to chronic pancreatitis relatively quickly, the potential to isolate and transplant the islets back into which not only significantly shortens the experimental pe- the patient. Many patients already have diabetes, however, riod but also allows easier detection of changes in the pa- and therefore are not candidates for this islet salvage. Also, rameters of interest (9–11,41–43). islet isolation from a pancreas with chronic pancreatitis char- In this study we show that long-term PDL becomes a acterized by severe tissue fibrosis is difficult and inefficient. chronic pancreatitis model, and the gradual loss of b-cell Moreover, transplanted islets are situated in the liver, mass after PDL mimics the similar phenomenon in patients where they may gradually lose function (1,2). Therefore a with CPRD. We found that b-cell dedifferentiation followed better understanding of the molecular mechanisms under- by b-EMT occurs in the PDL-tail, resulting in a progressive lying the development and pathogenesis of CPRD may lead loss of functional b-cell mass. In the lineage tracing exper- to novel, more effective therapeutic strategies. iment, no vimentin/insulin double-positive cells were ever The development of CPRD may largely result from the detected. However, vimentin-positive b-cell progeny cells were proximity of the endocrine pancreas to the exocrine pancreas. broadly detected by 12 weeks after PDL, suggesting wide- When the exocrine pancreas is severely inflamed and fibrotic, spread b-cell EMT. Of note, EMT b-cells may then undergo the inflammatory milieu results in the presence of many cell death, since the fold-increase in EMT-associated genes cytokines and factors that negatively affect the neighboring was not dramatic in purified b-cells at serial time points islet cells in a paracrine way. In addition, the vasculature is after PDL. typically severely disrupted. These alterations may signifi- Using a loss-of-function experiment in which TGFb re- cantly affect the biology and homeostasis of islet b-cells. ceptor signaling was blocked (TBR2-null mutant mice), we Complete PDL and partial PDL have been used for confirmed that TGFb receptor signaling was necessary for decadesasmodelsforacuteandchronicpancreatitisin b-cell EMT in the PDL-tail. We also showed evidence that humans (9,10). Compared with complete PDL, the partial TGFb1 ligand is initially derived predominantly from M2 PDL model has several advantages. The nonligated head of macrophages soon after PDL, but it comes from activated the pancreas (PDL-head) is minimally affected and thus can myofibroblasts in the later stages after PDL. Myofibroblasts be used as an internal control for the PDL-tail. Also, the can be activated by TGFb1, but as a result they begin to islets and acinar cells in the head of the pancreas are ade- produce and secrete TGFb1 themselves, as a positive feed- quate to meet physiological needs for glucose homeostasis back loop for TGFb1 production (16–18). Early after and secretion of pancreatic digestive enzymes, respectively. PDL, the recruited macrophages may act as activators of The disadvantage of the partial PDL model is that the myofibroblasts through their production and secretion of intact head of the pancreas prevents changes in blood TGFb1 (14). Furthermore, high TGFb1 has been detected glucose at the late stage of CPRD. Since 2008, PDL over in the pancreas in patients with acute (44) and chronic the short-term—traditionally regarded as a model of acute pancreatitis (45). Thus a critical role for TGFb1inb-cell pancreatitis—has been extensively investigated as a model EMT over the long-term after PDL seems a likely mecha- of b-cell regeneration. Although controversy still exists, nism of CPRD pathogenesis. Mechanistically, we showed that most of these short-term studies have shown that little TGFb1 activated TGFb receptor signaling in b-cells to in- or no b-cell neogenesis occurs, but b-cell proliferation in- duce phosphorylation and activation of SMAD3, which creases modestly, which contributes minimally to b-cell subsequently induced activation of Stat3, possibly through mass (13,31,38–40). competitive binding to protein inhibitor of activated Stat3 (33). Compared with other chronic pancreatitis models (e.g., serial FoxO1 and Stat3 are known to antagonize one another cerulein injections, serial L-arginine injections, alcohol feeding, in several cell types, but this has not been studied in b-cells. and genetic modifications), PDL confers several advantages. However, recent studies have demonstrated a critical role First, PDL requires only one treatment, which substantially for Stat3 in pancreatic cell homeostasis (46,47). Here, early reduces stress to the animals and reduces experimental activation of Stat3 overcomes FoxO1 suppression, leading variability. Second, PDL induces robust pancreatitis through to b-cell dysfunction. Strengthening FoxO1 expression in ductal occlusion, which may mimic the pathogenesis of b-cells suppresses Stat3 signaling and thereby suppresses

from the PDL-tail of MIP-GFP mice at PDL12w and placed in culture. These b-cells were transduced with either AAV-RIP-FoxO1 or AAV-RIP- null as a control. After infection (48 h), the b-cells were analyzed for expression of E-cadherin, Snail, Slug, ZEB1, and ZEB2. Values were normalized to the AAV-RIP-null treated control. F: Proposed signaling cascade in CPRD. High levels of TGFb1intheinflamed pancreas are produced by M2 macrophages early and by activated myofibroblasts at later stages. TGFb1-activated TGFb receptor (TGFbR) signaling in b-cells, in which phosphorylation and activation of SMAD3 induce phosphorylation and activation of Stat3 to activate transcription of Snail, leads to EMT in b-cells, resulting in b-cell dysfunction and loss (CPRD). FoxO1 and Stat3 antagonize one another, so early activation of Stat3 overcomes FoxO1 suppression, leading to b-cell dysfunction. However, strengthening FoxO1 expression in b-cells suppresses Stat3 signaling and thereby suppresses b-cell EMT. Scale bars are 50 mm. *P < 0.05. HO, Hoechst nuclear staining; INS, insulin; NS, nonsignificant. N =5mice for each experimental group. diabetes.diabetesjournals.org Xiao and Associates 2657 b-cell EMT. In our mouse model, we further showed that 12. Xiao X, Wiersch J, El-Gohary Y, et al. TGFb receptor signaling is essential for FoxO1 overexpression is sufficient to prevent b-cell EMT, inflammation-induced but not b-cell workload-induced b-cell proliferation. Diabetes and the late loss of FoxO1 seemed to portend the comple- 2013;62:1217–1226 b tion of b-cell EMT. 13. XiaoX,ChenZ,ShiotaC,etal.Noevidencefor cell neogenesis in murine adult pancreas. J Clin Invest 2013;123:2207–2217 The relatively delayed onset of b-cell EMT in vivo, com- 14. Xiao X, Gaffar I, Guo P, et al. M2 macrophages promote beta-cell proliferation pared with in vitro, possibly results from the in vivo environ- by up-regulation of SMAD7. Proc Natl Acad Sci U S A 2014;111:E1211–E1220 ment of the islets that might protect against EMT through 15. Xiao X, Guo P, Shiota C, et al. Neurogenin3 activation is not sufficient to direct cell-cell communication. In summary, we present here a duct-to-beta cell transdifferentiation in the adult pancreas. J Biol Chem 2013;288: mouse model for CPRD. To the best of our knowledge, this 25297–25308 study is the first to show b-cell EMT in an experimental 16. Chaudhury A, Hussey GS, Ray PS, Jin G, Fox PL, Howe PH. TGF-beta-mediated setting of CPRD. Because b-cell EMT seems to underlie the phosphorylation of hnRNP E1 induces EMT via transcript-selective translational in- process of b-cell loss, prevention of b-cell EMT may be an duction of Dab2 and ILEI. Nat Cell Biol 2010;12:286–293 effective treatment for CPRD. 17. Park SH, Jung EH, Kim GY, Kim BC, Lim JH, Woo CH. Itch E3 ligase positively regulates TGF-b signaling to EMT via Smad7 ubiquitination. Mol Cells 2015;38:20–25 Funding. This work was supported by the startup from the Department of 18. Song J. EMT or apoptosis: a decision for TGF-beta. Cell Res 2007;17:289–290 Surgery of Children’s Hospital of Pittsburgh (to X.X.), and by the National Institutes of 19. Zhang T, Kim DH, Xiao X, et al. FoxO1 plays an important role in regulating Health/National Institute of Diabetes and Digestive and Kidney Diseases (grant nos. b-cell compensation for insulin resistance in male mice. Endocrinology 2016;157: R01DK111460 and R01DK112836 to G.K.G.). 1055–1070 Duality of Interest. No conflicts of interest relevant to this article were 20. Kitamura T, Ido Kitamura Y. Role of FoxO proteins in pancreatic beta cells. reported. Endocr J 2007;54:507–515 Author Contributions. X.X. conceived and designed the study. X.X., S.F., 21. Mezza T, Shirakawa J, Martinez R, Hu J, Giaccari A, Kulkarni RN. Nuclear T.Z., C.C., Q.S., R.Z., S.P., J.F., Y.M., P.G., C.S., K.P., and N.G. acquired data. X.X., export of FoxO1 is associated with ERK signaling in b-cells lacking insulin receptors. S.Z.H., H.D., and G.K.G. analyzed and interpreted the data. X.X. wrote the J Biol Chem 2016;291:21485–21495 manuscript. All authors revised the article and approved the final version to be 22. Gannon M. BuMP-ing up insulin secretion by pancreatic beta cells. Cell Metab published. X.X. and G.K.G. are the guarantors of this work and, as such, had full 2007;5:157–159 access to all the data in the study and take responsibility for the integrity of the data 23. Guo S, Dai C, Guo M, et al. Inactivation of specific transcription factors in and the accuracy of the data analysis. type 2 diabetes. J Clin Invest 2013;123:3305–3316 Prior Presentation. Part of this study was presented orally (no. 2015-A- 24. Talchai C, Xuan S, Lin HV, Sussel L, Accili D. Pancreatic b cell dedifferentiation 1495-Diabetes) at the 75th Scientific Sessions of the American Diabetes Association, as a mechanism of diabetic b cell failure. Cell 2012;150:1223–1234 Boston, MA, 5–9 June 2015. 25. Xiao X, Chen C, Guo P, et al. Forkhead box protein 1 (FoxO1) inhibits accelerated b cell aging in pancreas-specific SMAD7 mutant mice. J Biol Chem References 2017;292:3456–3465 1. Yadav D, Timmons L, Benson JT, DierkhisingRA,ChariST.Incidence,prev- 26. Xiao X, Guo P, Prasadan K, et al. Pancreatic cell tracing, lineage tagging and alence, and survival of chronic pancreatitis: a population-based study. Am J Gas- targeted genetic manipulations in multiple cell types using pancreatic ductal infusion troenterol 2011;106:2192–2199 of adeno-associated viral vectors and/or cell-tagging dyes. Nat Protoc 2014;9:2719– 2. DiMagno MJ, DiMagno EP. Chronic pancreatitis. Curr Opin Gastroenterol 2013; 2724 29:531–536 27. Xiao X, Guo P, Chen Z, et al. Hypoglycemia reduces vascular endothelial growth 3. Pasca di Magliano M, Forsmark C, Freedman S, et al. Advances in acute and factor A production by pancreatic beta cells as a regulator of beta cell mass. J Biol chronic pancreatitis: from development to inflammation and repair. Gastroenterology Chem 2013;288:8636–8646 2013;144:e1–e4 28. Xiao X, Prasadan K, Guo P, et al. Pancreatic duct cells as a source of VEGF in 4. Lowenfels AB, Maisonneuve P, Cavallini G, et al.; International Pancreatitis mice. Diabetologia 2014;57:991–1000 Study Group. Pancreatitis and the risk of pancreatic cancer. N Engl J Med 1993;328: 29. Song Z, Fusco J, Zimmerman R, et al. Epidermal growth factor receptor signaling 1433–1437 regulates b cell proliferation in adult mice. J Biol Chem 2016;291:22630–22637 5. Pan J, Xin L, Wang D, et al. Risk factors for diabetes mellitus in chronic 30. Xiao X, Fischbach S, Song Z, et al. Transient suppression of TGFb receptor signaling pancreatitis: a cohort of 2,011 patients. Medicine (Baltimore) 2016;95:e3251 facilitates human islet transplantation. Endocrinology 2016;157:1348–1356 6. Brodovicz KG, Kou TD, Alexander CM, et al. Impact of diabetes duration and 31. Pan FC, Bankaitis ED, Boyer D, et al. Spatiotemporal patterns of multipotentiality chronic pancreatitis on the association between type 2 diabetes and pancreatic in Ptf1a-expressing cells during pancreas organogenesis and injury-induced facul- cancer risk. Diabetes Obes Metab 2012;14:1123–1128 tative restoration. Development 2013;140:751–764 7. Ode KL, Moran A. New insights into cystic fibrosis-related diabetes in children. 32. Bhatia V, Cao Y, Ko TC, Falzon M. Parathyroid hormone-related protein interacts Lancet Diabetes Endocrinol 2013;1:52–58 with the transforming growth factor-b/bone morphogenetic protein-2/gremlin sig- 8. Lerch MM, Gorelick FS. Models of acute and chronic pancreatitis. Gastroen- naling pathway to regulate proinflammatory and profibrotic mediators in pancreatic terology 2013;144:1180–1193 acinar and stellate cells. Pancreas 2016;45:659–670 9. Tanaka T, Ichiba Y, Fujii Y, Itoh H, Kodama O, Dohi K. New canine model of 33. Saitoh M, Endo K, Furuya S, et al. STAT3 integrates cooperative Ras and TGF-b chronic pancreatitis due to chronic ischemia with incomplete pancreatic duct ob- signals that induce Snail expression. Oncogene 2016;35:1049–1057 struction. Digestion 1988;41:149–155 34. Oh HM, Yu CR, Dambuza I, Marrero B, Egwuagu CE. STAT3 protein interacts 10. Boerma D, Straatsburg IH, Offerhaus GJ, Gouma DJ, van Gulik TM. Experi- with class O Forkhead transcription factors in the and regulates nucle- mental model of obstructive, chronic pancreatitis in pigs. Dig Surg 2003;20:520–526 ar/cytoplasmic localization of FoxO1 and FoxO3a proteins in CD4(+) T cells. J Biol 11. Yamaguchi Y, Matsuno K, Goto M, Ogawa M. In situ kinetics of acinar, duct, Chem 2012;287:30436–30443 and inflammatory cells in duct ligation-induced pancreatitis in rats. Gastroenterology 35. Hsu P, Santner-Nanan B, Hu M, et al. IL-10 potentiates differentiation of human 1993;104:1498–1506 induced regulatory T cells via STAT3 and Foxo1. J Immunol 2015;195:3665–3674 2658 b-Cell EMT Contributes to CPRD Diabetes Volume 66, October 2017

36. Ma W, Fuentes G, Shi X, Verma C, Radda GK, Han W. FoxO1 negatively reg- 42. Meyerholz DK, Williard DE, Grittmann AM, Samuel I. Murine pancreatic duct ulates leptin-induced POMC transcription through its direct interaction with STAT3. ligation induces stress kinase activation, acute pancreatitis, and acute lung injury. Biochem J 2015;466:291–298 Am J Surg 2008;196:675–682 37. Yang G, Lim CY, Li C, et al. FoxO1 inhibits leptin regulation of pro-opiomela- 43. Kishi S, Takeyama Y, Ueda T, et al. Pancreatic duct obstruction itself induces fi nocortin promoter activity by blocking STAT3 interaction with speci city protein 1. expression of alpha smooth muscle actin in pancreatic stellate cells. J Surg Res – J Biol Chem 2009;284:3719 3727 2003;114:6–14 38. Rankin MM, Wilbur CJ, Rak K, Shields EJ, Granger A, Kushner JA. b-Cells are 44. Friess H, Lu Z, Riesle E, et al. Enhanced expression of TGF-betas and their not generated in pancreatic duct ligation-induced injury in adult mice. Diabetes 2013; receptors in human acute pancreatitis. Ann Surg 1998;227:95–104 62:1634–1645 45. van Laethem JL, Deviere J, Resibois A, et al. Localization of transforming 39. Kopp JL, Dubois CL, Schaffer AE, et al. Sox9+ ductal cells are multipotent growth factor beta 1 and its latent binding protein in human chronic pancreatitis. progenitors throughout development but do not produce new endocrine cells in the – normal or injured adult pancreas. Development 2011;138:653–665 Gastroenterology 1995;108:1873 1881 40. Cavelti-Weder C, Shtessel M, Reuss JE, et al. Pancreatic duct ligation after 46. Valdez IA, Dirice E, Gupta MK, Shirakawa J, Teo AKK, Kulkarni RN. Proin- fl almost complete b-cell loss: exocrine regeneration but no evidence of b-cell re- ammatory cytokines induce endocrine differentiation in pancreatic ductal cells via generation. Endocrinology 2013;154:4493–4502 STAT3-dependent NGN3 activation. Cell Rep 2016;15:460–470 41. Nepomnyashchikh LM, Lushnikova EL, Viskunov VG, Protsenko SI, Molodykh 47. Gorogawa S, Fujitani Y, Kaneto H, et al. Insulin secretory defects and impaired OP, Koldysheva EV. Ultrastructure of acinar cell injuries in experimental acute pan- islet architecture in pancreatic beta-cell-specific STAT3 knockout mice. Biochem creatitis created by common bile duct ligation. Bull Exp Biol Med 2011;150:747–752 Biophys Res Commun 2004;319:1159–1170