Activation of Pancreatic Acinar FXR Protects against Pancreatitis via Osgin1-Mediated Restoration of Efficient Autophagy

Pancreatitis is the leading cause of hospitalization in gastroenterology, and no medications are available for treating this disease in current clinical practice. FXR plays an anti-inflammatory role in diverse inflammatory diseases, while its function in pancreatitis remains unknown. In this study, we initially observed a marked increase of nuclear FXR in pancreatic tissues of human patients with pancreatitis. Deleting the FXR in pancreatic acinar cells (FXRacinarΔ/Δ) led to more severe pancreatitis in mouse models of caerulein-induced acute and chronic pancreatitis, while the FXR agonist GW4064 significantly attenuated pancreatitis in caerulein or arginine-induced acute pancreatitis and caerulein-induced chronic pancreatitis. FXR deletion impaired the viability and stress responses of pancreatic exocrine organoids (PEOs) in vitro. Utilizing RNA-seq and ChIP-seq of PEOs, we identified Osgin1 as a direct target of FXR in the exocrine pancreas, which was also increasingly expressed in human pancreatitis tissues compared to normal pancreatic tissues. Pancreatic knockdown of Osgin1 by AAV-pan abolished the therapeutic effects of FXR activation on pancreatitis, whereas pancreatic overexpression of Osgin1 effectively alleviated caerulein-induced pancreatitis. Mechanistically, we found that the FXR-OSGIN1 axis stimulated autophagic flux in the pancreatic tissues and cell lines, which was considered as the intrinsic mechanisms through which FXR-OSGIN1 protecting against pancreatitis. Our results highlight the protective role of the FXR-OSGIN1 axis in pancreatitis and provided a new target for the treatment of this disease.


Introduction
Pancreatitis is an inflammatory disorder of the pancreas with considerable morbidity, mortality, and socioeconomic burden [1]. Acute pancreatitis (AP), with a global incidence of approximately 0.034% [2], usually starts from acinar cell death and quickly progresses to systemic inflammation that requires emergency management. AP patients usually com-plain of severe acute pain in left upper abdomen and are at a high risk of developing systemic inflammation or multiple organ dysfunction (MODS). About 50 out of every 100,000 persons in the world suffer from chronic pancreatitis (CP) [3], manifesting slowly developed pancreatic inflammation. Some patients with AP will develop into CP due to recurrent pancreatitis [4]. In the pathogenesis of CP, repetitive episodes of pancreatic inflammation cause normal tissues being gradually replaced by fibrotic tissues, leading to chronic pain, pancreatic exocrine and endocrine dysfunction, and reduced life expectancy [5]. To date, there is no specific pharmacological therapeutics for both acute and chronic pancreatitis in the clinic.
The major cause of AP, accounting for up to 40% of the cases, is gallstone disease [6]. When the common bile duct is obstructed, biliary-pancreatic reflux causes pancreatic acinar cell death and tissue inflammation. Previous studies demonstrated that one of the major bile acid receptors, cell surface G protein-coupled bile acid receptor-1 (GPBAR1), is involved in the pathogenesis of AP [7,8]. Besides GPBAR1, there is another key bile acid receptor, namely, FXR (also called nuclear receptor subfamily 1 group H member 4, NR1H4), that is present in the pancreas [9][10][11]. Whether FXR plays a role in the pathogenesis of pancreatitis remains unclear.
FXR is a ligand activated transcription factor. Upon activation, FXR predominately binds inverted repeats spaced by one nucleotide (IR-1) on its target genes [12] and plays critical roles in the regulation of glucose, lipid, bile acid, and amino acid metabolism within liver and intestine [13,14]. Recent studies also showed that FXR was key for various inflammatory diseases. FXR promoted the transactivation of SHP to inhibit c-Jun-mediated osteopontin production in NKT cells and mitigated acute hepatitis [15]. By antagonizing the activator protein-1 (AP-1), stat3, and nuclear factor-kappa B signaling pathways, FXR alleviates inflammation and protects impaired intestinal epithelial barrier integrity in inflammatory bowel disease (IBD) [16]. In addition to transcriptional regulation, FXR was found to directly interact with NLRP3 and caspase 1, negatively regulating the NLRP3 inflammasome in cholestasis-associated sepsis [17]. These studies give us a clue that FXR may participate and play an anti-inflammatory role in the pathogenesis of pancreatitis.
In the present study, we observed that FXR was dramatically increased in the nuclei of pancreatic acinar cells in human patients with pancreatitis. This observation drove us to generate mice with pancreatic acinar-specific deletion of FXR (FXR acinarΔ/Δ ) to further investigate the functional role of FXR in pancreatitis. We found that activation of pancreatic acinar FXR protected against pancreatitis. Activation of FXR by the agonist GW4064 was able to alleviate the disease in different mouse pancreatitis models. We further identified that oxidative stress induced growth inhibitor 1 (Osgin1) was a major direct FXR target gene in pancreatic exocrine cells and mediated the protective effects of FXR activation on pancreatitis through restoration of efficient autophagy.

Nuclear FXR Level Was Elevated in Human and Murine
Pancreatitis Tissues. To see whether FXR is involved in pancreatitis, we assessed FXR expression in pancreatic tissue sections from patients with pancreatitis by immunohistochemical (IHC) staining, with normal pancreatic tissues from patients with pancreatic cancer as the control. FXR was mainly expressed in the cytoplasm of normal pancreatic tissues. There was no difference in cytoplasmic FXR levels between pancreatitis and normal pancreatic tissues (Figures 1(a) and 1(b)). However, nuclear FXR was markedly increased in pancreatitis tissues compared to that in normal pancreatic tissues (Figures 1(a) and 1(b)). Next, we assessed FXR expression in murine models of caeruleininduced pancreatitis. Dramatically increased nuclear FXR expression was also observed in pancreatitis tissues (Figure 1(c)). These data demonstrated that FXR was upregulated in human and murine pancreatitis tissues.

Pancreatic Acinar Deletion of FXR Aggravated
Pancreatitis in Murine AP and CP Models. We next generated FXR acinarΔ/Δ mice (Figure 1(d) and Figure S1) to further investigate the possible function of FXR on pancreatitis (Figure 1(e)). Pancreatic acinar deletion of FXR did not directly induce alterations in pancreatic histology and exocrine function. However, when challenged with caerulein, FXR acinarΔ/Δ mice presented more severe pancreatitis phenotypes compared with FXR f/f mice, including elevated serum amylase and lipase level ( Figure 1(f)), more severe tissue edema and necrosis (Figures 1(g) and 1(h)), and more neutrophil infiltrations and apoptotic acinar cells in pancreas (Figures 1(i)-1(k)). Moreover, we challenged FXR acinarΔ/Δ and FXR f/f mice with repeated caerulein injection (Figure 1(l)). The results showed that FXR deletion led to a significant increase in CP-associated tissue edema, fibrosis, and inflammation infiltration (Figures 1(m)-1(o)). These findings suggested a protective role of FXR in pancreatitis.

FXR Deletion Compromised the Viability and Stress
Tolerance of Pancreatic Exocrine Organoids. As pancreatic exocrine organoids (PEOs) are excellent in vitro models for studying pancreatic exocrine functions [18], we next generated FXR f/f PEOs from FXR f/f mice and FXR -/-PEOs from mice with FXR deletion in the whole pancreas (FXR pancreasΔ/Δ ) (Figure 3(a) and Figure S2, a and b). FXR -/-PEOs exhibited damaged organoid structures and a lower viability, indicating FXR is required for normal pancreatic organoids function (Figure 3(a)). We then treated the PEOs with different types of bile acids (BAs) as physiological FXR agonists and antagonists in vitro ( Figure 3(b)). Upon treatment with the FXR antagonists tauro-β-muricholic acid (T-βMCA), glycoursodeoxycholic acid (GUDCA), and Z-guggulsterone (ZGG), the viability of PEOs was slightly reduced (Figure 3(b)). To establish a model of pancreatic injury in PEOs, we tested the effects of several commonly used stress stimuli on wildtype (WT) PEOs ( Figure S2c). Among these candidates, caerulein, tumor necrotic factor (TNF), asymmetric dimethylarginine (ADMA), and cortisone did not damage the PEOs, while palmitic acid (PA) and linoleic acid (LA) markedly impaired the viability and growth of PEOs ( Figure S2, c and d). Moreover, we performed RNA-seq analysis on PA-treated PEOs and found that genes upregulated upon PA treatment in FXR f/f PEOs were primarily related to the NLRP3 inflammasome, mitophagy, and noncanonical NF-κB signaling ( Figure S3, a-f), which are well-recognized molecular cascades implicated in the pathogenesis of AP. Thus, we considered PA-induced PEOs damage model as an appropriate in vitro model of pancreatitis.
Next, we compared the responses of FXR f/f and FXR -/-PEOs to PA and found that FXR -/-PEOs exhibited lower tolerance to PA stress (Figures 3(c) and 3(d)). The PA-induced reduction in PEO viability was attenuated by the FXR agonists GW4064, TDCA, and TCDCA, while aggravated by the FXR antagonist ZGG (Figure 3(e)). Furthermore, we analyzed genes that exhibited differential expression only in FXR -/-PEOs but not in FXR f/f PEOs under PA stress. The upregulated genes were mainly enriched in inflammatory and defense responses (Figures 3(f) and 3(g)), while the downregulated genes were involved in DNA replication and cell cycle (Figures 3(h) and 3(i)). In addition, some genes related to inflammation, defense response, cell cycle, and DNA replication in FXR f/f PEOs under PA-induced stress exhibited more prominent expression changes in FXR -/-PEOs, which was also validated via quantitative real time PCR (RT-qPCR) (Figures 3(j)-3(l)). These results indicated that FXR ablation caused more severe inflammation and impaired renewal ability in PEOs.

Osgin1 Was Transcriptionally Upregulated by FXR Activation in PEOs.
To identify the downstream signaling of FXR that majorly functions in exocrine pancreatic cells, we subjected PEOs with or without GW4064 treatment to RNA-seq ( Figure S4) and ChIP-seq (Figure 4(a) and S5). Upon activation, FXR is bound to the promoter of Shp and Slc51b, enhancing their mRNA expression ( Figure S5, c-f), which is consistent with previous reports [19] in liver and intestine. Motif analysis revealed that the binding motif of FXR in PEOs was an inverted repeat spaced by one nucleotide (IR-1) (Figure 4(b)). Venn analysis identified 29 genes that were transcriptionally upregulated by FXR activation in PEOs (Figures 4(c) and 4(d)). Among these, FXR is bound to the transcription start site (TSS) and promoter regions of eight genes (Figure 4(e)). Considering that the FPKM value of Osgin1 was the highest (Figure 4(f)), we hypothesized that Osgin1 is a major target of FXR within cells of the exocrine pancreas. Indeed, as shown in Figure 4(g), the mRNA levels of Osgin1 increased~13-fold in GW4064-treated WT PEOs, while this effect was abolished in GW4064-treated FXR KO PEOs. In vivo, the mRNA levels of Osgin1 increased~2-fold after GW4064 treatment (Figure 4(h)). Next, we performed ChIP-qPCR and found that FXR is bound to 6592-6774, 6331-6518, and 63-243 upstream of the Osgin1 TSS (Figures 4(i) and 4(j)). By analyzing the public GEO dataset (GSE133700), we also found that Osgin1 could harbor the binding sites of four FXR isoforms in liver organoids ( Figure S5g) [20].

Osgin1 Mediated the Protective Effects of FXR on
Pancreatitis. We further assessed the clinical significance of OSGIN1 in pancreatitis by IHC and found that OSGIN1 level in human pancreatitis tissues was significantly elevated compared to that in normal pancreatic tissues (Figures 4(k) and 4(l)). Additionally, the OSGIN1 level was significantly correlated with nuclear FXR level (Figure 4(m)).
To investigate the functional role of Osgin1 in vivo, we manipulated Osgin1 in the pancreatic tissues utilizing AAV-pan, in which the AAV2 genomes were encapsulated in the Y447F and Y773F mutant AAV8 capsids to generate a viral vector with high transduction efficiency in pancreatic acinar cells [21]. We infected mice with AAV-pan with a cytomegalovirus (CMV) promoter driving green fluorescent protein (AAV-pan-CMV-GFP) expression and found a high infection efficiency of AAV-pan in pancreatic tissues using an imaging system for live animals and fluorescence microscopy of frozen slices ( Figure 5(a)). We then infected mice with AAV-shOsgin1 or AAV-shControl by intraperitoneal injection (i.p.). After 2 weeks, the mice were subjected to AP modeling and GW4064 treatment. RT-qPCR and immunoblots analyses showed that AAV-shOsgin1 successfully suppressed the mRNA and protein levels of Osgin1 ( Figure 5(b)). In mice with Osgin1 knockdown, no difference was observed in the serum amylase and lipase levels as well as tissue edema and necrosis between the vehicle and GW4064 group (Figures 5(c)-5(e)), suggesting that the therapeutic effects of GW4064 on pancreatitis were mediated by the promotion of Osgin1 expression.

Research
Chip-seq FXR binding RNA-seq FXR activating 168 29 261 Site1 Site2 Site3 Gclm AAV-pan with a CMV promoter driving green fluorescent protein (AAV-pan-CMV-GFP) expression served as a control. Overexpression of Osgin1 in the pancreas was confirmed through RT-qPCR and western blotting ( Figure 5(f)). Two weeks after the injection of AAV-Osgin1 or AAV-GFP, mice were challenged with caerulein for seven hours. Osgin1 overexpression in the pancreas mitigated the elevation of serum amylase and lipase levels, tissue edema and necrosis, inflammation infiltration, and acinar cell apoptosis in mice with caerulein challenge (Figures 5(g)-5(k)), which was consistent with the results of GW4064 administration.

FXR Activation Restores Efficient Autophagy in
Pancreatitis by Enhancing Osgin1 Expression. Previous studies showed that OSGIN1 is an important regulator of autophagy [22][23][24]. In smoking-induced stress in human airway epithelium, OSGIN1 induced the formation of autophagosomes to enhance autophagy [22]. Autophagy is critical for pancreatic injury repair, eliminating injured and useless cell components and recycling their constituents [25]. To look into the downstream signaling of OSGIN1, we next conducted RNA-seq on pancreatic tissues with OSGIN1 overexpression (OSGIN1-OE) and performed GO enrichment analyses. It revealed that genes altered by OSGIN1-OE were significantly enriched in autophagyrelated biological processes (Figures 6(a) and 6(b)). We treated pancreatic cells (SW1990) with GW4064 in presence of Torin1, which inhibits the mTORC1/2 complex to induce autophagy [26]. The results showed that in the presence of chloroquine, GW4064 treatment significantly increased levels of LC3-II ( Figure 6(c)). Furthermore, we constructed an adenovirus vector with an Ubi promoter driving the expression of a dual-fluorescence fusion protein, stubRFP-sensGFP-LC3B. This protein could reflect the level of autophagic flux in cells, autophagosomes that are not fused with lysosomes present yellow signals (overlapping of red and green fluorescence), whereas the red signal corresponds to autolysosomes. Here, SW1990 cells were transfected with the stubRFP-sensGFP-LC3B adenovirus (Ad-Ubi-stubRFP-sensGFP-LC3B) followed by Torin1 and GW4064 treatment.
The results showed that the yellow and red puncta were both increased upon GW4064 treatment ( Figure 6(d)), suggesting that autophagosomes and autolysosomes were significantly increased by FXR activation. The RT-qPCR analysis showed that the Osgin1 mRNA level was also elevated by GW4064 treatment in SW1990 cells ( Figure S6a). We then overexpressed OSGIN1 in SW1990 cells ( Figure S6a) and detected the alterations of autophagic flux. The results were similar to those with GW4064 treatment (Figures 6(e) and 6(f)). Furthermore, we overexpressed OSGIN1 in both 293T and SW1990 cells and conducted immunoprecipitation followed by mass spectrometry (IP-MS). The Venn analysis identified 48 proteins that interacted with OSGIN1 ( Figure S6b). The Reactome pathway enrichment revealed that OSGIN1 was interacted with OSGIN1 interacted with most of the critical components of the ATP-dependent chaperonin complex TRiC (also called CCT) ( Figure S6, b and c), which is reportedly involved in the autophagic flux [27,28], further supporting the regulation of autophagy by OSGIN1. Next, we investigated the effects of FXR-OSGIN1 signaling on autophagy in pancreatic tissues, and we assessed LC3-I, LC3-II, and p62 protein levels in vivo using immunoblot. In caerulein-induced mouse models of pancreatitis, OSGIN1 overexpression significantly reduced the accumulation of p62 and increased the ratio of LC3-II to LC3-I ( Figure 6(g)), which was consistent with the results in GW4064 treatment (Figure 6(h)). We observed the autophagic vesicles in pancreatic tissues utilizing transmission electron microscopy (TEM). It showed that autophagic vesicles were dramatically accumulated in pancreatitis tissues and were significantly reduced by OSGIN1-OE or GW4064 treatment ( Figure S6d). Moreover, the autophagy promoting effect of GW4064 was almost totally abolished by Osgin1knockdown ( Figure 6(i)), and FXR deletion in pancreatic acinar cells led to a more severe impairment of autophagy ( Figure 6(j) and S6d). These findings indicated that FXR activation restored efficient autophagy by promoting the expression of Osgin1.

Discussion
In this study, we find that the key bile acid receptor FXR in pancreatic acinar cells is activated in pancreatic tissues and required for the protective effects against pancreatitis. The FXR agonist GW4064 exhibited excellent therapeutic effects on pancreatitis; it could directly activate pancreatic acinar FXR to restore efficient autophagy mediated by enhanced Osgin1 expression, thus mitigate the severity of pancreatitis.
Previous reports showed that global FXR knockout did not have significant effects on experimental AP [29]. The impact of global FXR deletion on the whole organism, especially the liver and intestine, is considerably complex, and specific FXR function in pancreatitis may not be easily observed. Tissue-specific deletion of FXR is a better way to investigate its unique roles in pancreas. To this end, we generated mutant mice of pancreatic acinar cell-specific deletion of FXR (FXR acinarΔ/Δ ), serving as an appropriate in vivo model to investigate the functions of FXR in pancreatitis. Zhou et al. previously considered FXR activation in CP as a deleterious pathological alteration based on their finding that nuclear FXR expression was positively correlated with the severity of CP in human patients [30]. However, without detailed mechanistic study in vitro and in vivo, their conclusion ruled out the possibility that FXR upregulation was compensatory and beneficial in pancreatitis. Indeed, here, we found that acinar cell-specific deletion of FXR in mice caused more severe pancreatitis both in acute and chronic pancreatitis models, indicating that FXR played a protective role against pancreatitis.
Impaired autophagy has been shown to be an important pathological event in AP. Inhibition of autophagy through genetic or pharmacological approaches led to more severe pancreatitis in mice, while restoration of efficient autophagy via disaccharide trehalose mitigated the disease [31,32]. Therefore, restoration of efficient autophagy in the acinar cells could be beneficial to alleviate pancreatitis [33]. Indeed,  (h) Representative western blot images and summarized plot of p62 protein levels as well as the ratio of LC3-II to LC3-I in pancreatic tissues from mice treated with saline + vehicle (n = 3), caerulein + vehicle (n = 5), or caerulein + GW4064 (n = 5). (i) Representative western blot images and summarized plot of p62 protein levels as well as the ratio of LC3-II to LC3-I in pancreatic tissues from caerulein challenged mice treated with AAV − shControl + vehicle (n = 5), AAV − shControl + GW4064 (n = 5), AAV − shOsgin1 + vehicle (n = 5), and AAV − shOsgin1 + GW4064 (n = 5). (j) Representative western blot images and summarized plot of p62 protein levels as well as the ratio of LC3-II to LC3-I in pancreatic tissues from FXR f/f (n = 5) and FXR acinarΔ/Δ (n = 5) mice treated with caerulein. Data are presented as the mean ± SEM. ns: not significant. * P < 0:5, * * P < 0:01, * * * P < 0:001, and * * * * P < 0:0001 via one-way ANOVA. # P < 0:01 and ## P < 0:01 via unpaired Student's t test. 10 Research we provided sufficient evidences in this study showing that FXR actually stimulated autophagy in the pathogenesis of pancreatitis. This is very different from the situation in the liver, where FXR is an autophagy suppressor in the fasting condition [34,35]. As shown by our results, FXR agonist GW4064 and overexpression of its downstream OSGIN1 stimulated autophagic flux in pancreatic cell lines. In vivo, deletion of FXR resulted in a more severe impairment of autophagic flux, whereas the FXR agonists GW4064 significantly stimulated the autophagic flux and reduced the accumulation of autophagic vesicles in the pancreatic tissues of mice with pancreatitis. These results demonstrate that FXR is an autophagic regulator that promotes autophagic flux in pancreatic cells.
Of note, only 11% of FXR-binding sites in the liver and intestine are overlapped [36], suggesting that FXR may exert specific molecular functions in different organs. It is possible that FXR plays unique functions in the pancreas or at least via regulating a different set of downstream genes. In the liver, FXR inhibits the expression of autophagic genes, such as Map1lc3a, Map1lc3b, Atg2a, Atg2b, Atg7, and Atg10 through directly binding to their promotor or disrupting the CREB/CRTC2 complex [34,35]. In our ChIP-seq and RNA-seq data of PEOs, we did not observe binding sites or transcriptional regulation of FXR on autophagy genes, suggesting that, in the exocrine pancreas, FXR regulates autophagy not by directly transcriptional promote or inhibit autophagic gene expression but by a unique way.
It is already known that all the four FXR isoforms bind to the promoter of Osgin1 in the liver organoids [20], but no literatures report the biological or clinical significance of this transcriptional regulation. Here, our study gives FXR-OSGIN1 axis important biological and pathological implications for the first time. We find that OSGIN1 is elevated in human pancreatitis tissues and positively correlated with nuclear FXR expression. OSGIN1 is transcriptionally activated by direct binding of FXR to its promotor in pancreatic exocrine cells and mediates the protective effects of FXR on pancreatitis in vivo. OSGIN1 has been recognized as a regulator of autophagy in several studies. In human airway basal cells, overexpressed OSGIN1 upregulated and colocalized with LC3-II [22]. Overexpression of OSGIN1 also increased LC3-II and decreased p62 protein levels in MCF-7 cells, while knockdown impaired autophagy in HeLa cells [24]. lncRNAs UCA1 and METTL3 were found to regulate autophagy by targeting Osgin1 to affect cell death [23,37]. Consistent with the findings with these studies, overexpression of OSGIN1 stimulated autophagic flux in pancreatitis tissues and cell lines, which was similar to the effect of GW4064 treatment. As illustrated in Figure S6, b and c, OSGIN1 interacted with most of the critical components of CCT. CCT plays important roles in maintaining efficient autophagic flux. A loss of CCT function leads to the inability to process aggregated proteins and inhibition of autophagy [27]. Recently, it was reported that the chaperonin subunit CCT2 is an aggrephagy receptor, regulating the clearance of aggregation-prone proteins [28]. CCT also folds both mLST8 and Raptor to play a key role in mTORC assembly and signaling [38].
We consider the interaction between OSGIN1 and CCT is likely to be the intrinsic mechanism by which OSGIN1 regulates the autophagy. In our future study, we will focus on this interaction and investigate whether it is involved in the regulation of autophagy by OSGIN1.
Overall, our results showed that the FXR-OSGIN1 axis plays an important role in the regulation of autophagy and protects against pancreatitis in the exocrine pancreas. Activation of pancreatic acinar FXR restores efficient autophagy by promoting OSGIN1 expression to ameliorate pancreatitis. Our results highlight the FXR-OSGIN1 axis as a promising pharmacological target for treating pancreatitis.

Experimental Design.
We assessed FXR and OSGIN1 protein levels in human pancreatitis tissues through IHC to demonstrate the clinical significance of our study. We specifically depleted pancreatic acinar FXR in mice and treated different mouse models of pancreatitis with FXR agonists GW4064 to explore the effects of FXR activation in pancreatitis. PEOs with or without FXR deletion were used as in vitro models. RNA-seq and ChIP-seq analyses of PEOs were conducted to identify factors acting downstream of FXR in the exocrine pancreas. We utilized AAV-pan to knockdown or overexpress Osgin1 specifically in the pancreas of mice in order to investigate the role of Osgin1 in AP. Immunoblots and Ad-LC3B-dual reporter of autophagy flux were used to evaluate autophagy in vitro. Immunoblots and TEM were conducted to evaluate autophagy in vivo.

Human Samples.
All patients were from The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China. All patients and/or legal guardians signed the informed consent documentation prior to experiments. Normal pancreatic tissues (n = 10) were obtained from patients with pancreatic cancer in the pancreatectomy. Pancreatitis tissues (n = 15) were obtained from patients diagnosed with pancreatitis. Six of them were obtained through biopsy and nine were from pancreatectomy. Detailed information of these patients was collected in Table S1. All research procedures on human samples were conducted with approval of the Ethics Committee of the First Affiliated Hospital of Zhengzhou University and in line with the Declaration of Helsinki.

Mice.
Mice of the C57BL/6J strain (Department of Laboratory Animal Science in Fudan University) were maintained at room temperature (22-25°C) and with free access to food and water with 12 h light/dark cycle in. FXR f/f mice with a C57BL/6J background were provided by Prof. Changtao Jiang. Mist1 CreERT2 mice with a C57BL/6J background were purchased from The Jackson Laboratory. PDX1 cre mice with a C57BL/6J background were purchased from GemPharmatech. Pancreatic acinar deletion of FXR (FXR acinarΔ/Δ ) was induced in Mist1 CreERT2 ; FXR f/f mice via intraperitoneal injection of tamoxifen (100 mg/kg) for 5 days. The FXR in whole pancreas of PDX1 Cre ; FXR f/f mice was depleted from birth. The growth and development of 11 Research FXR acinarΔ/Δ and FXR pancreasΔ/Δ mice were normal, and the body weight was comparable among FXR acinarΔ/Δ FXR pancreasΔ/Δ , and FXR f/f mice of the same sex and age.
All animal studies were performed according to protocols approved by the Animal Ethics Committee of the Fudan University School of Basic Medical Sciences. All mice were used after 1-or 2-weeks acclimatization after importing into the facility. Before each experiment, mice of the same group were placed in the same appropriately sized cage and equilibrated for a week. During equilibration, mice in control group and treated group were kept in the same condition. The number of mice was decided with reference to Guideline for the Care and Use of Laboratory Animals in China. Humane endpoints were defined as any obvious disease-related symptoms in mice. We established humane endpoints, but no mice were humanely sacrificed in this study.

Pancreatitis Modeling.
Pancreatitis modeling was performed as described in the figures. Briefly, to establish caerulein-induced pancreatitis models, mice were intraperitoneally injected with 50 μg/kg caerulein 6 hourly and were sacrificed 1 hour after the last caerulein injection. To establish caerulein-induced CP models, the above procedure was repeated 6 times in two weeks, and mice were sacrificed 1 hour after the last injection. To establish L-arginineinduced AP models, mice were intraperitoneally injected with 4 g/kg L-arginine (10%) twice with 1 hour interval and were sacrificed 6 hours after the first injection.

Histological
Evaluation. IHC staining of FXR and OSGIN1 was analyzed by histological scoring. Scoring criteria was divided into two parts: the intensity of protein expression (score = 0, none; score = 1, mild; score = 2, moderate; score = 3, severe) and the quantity of protein expression (score = 0, none; score = 1, 1%~25%; score = 2, 26%~50%; score = 3, 51%~75%; and score = 4, 76%~100%). Pancreatic injury of mice was evaluated by pancreatic edema, necrosis. Scoring criteria for quantity (the estimated proportion of edema and necrosis) were as follows: score = 0, none; score = 1, 1%~25%; score = 2, 26%~50%; score = 3, 51%~75%; and score = 4, 76%~100%. Scoring criteria for intensity of pancreatic edema were as follows: score = 0, none; score = 1, acinar swelling; score = 2, acinar swelling and interlobular interstitial separation; and score = 3, severe interlobular interstitial separation. Scoring criteria for intensity of pancreatic necrosis were as follows: score = 0, none; score = 1, mild spotty necrosis; score = 2, severe spotty necrosis; and score = 3, diffuse spotty necrosis with lobular loss. Each sample was evaluated in a blinded manner by three senior pathologists. Final score of each sample was the average of three scores. The quantity and intensity scores were then multiplied to obtain a total score, which could range from 0 to 12. Inflammatory infiltration in AP was evaluated by the number of MPO positive cells per high power field. For each individual, we obtained one tissue section and performed MPO immunochemistry staining on it. We choose 5-10 high power fields to count the MPO positive cells in each tissue section. Cell apoptosis was evaluated by TdT-mediated dUTP nick end labeling (TUNEL) staining. We calculated the number of TUNEL positive cells per high power field. We also observed 5-10 high power fields of one section from one sample. Every point in column chart of inflammation and apoptosis evaluation performed the number of MPO or TUNEL positive cells in one high power field. Tissue fibrosis and inflammation infiltration in CP models were evaluated by Sirius Red staining and F4/80 immunostaining followed by analyses using Visiopharm (Hamamatsu). The area of total tissues and fibrotic tissues as well as the number of F4/80+ cells was calculated, respectively. We calculated the fibrotic percentages and number of F4/80+ cells/μm 2 to represent the fibrotic and inflammatory degree of each individual.

Pancreatic Exocrine Organoid (PEO) Initiation and
Culturing. WT PEOs were initiated from FXR f/f mice, and FXR KO PEOs were established from FXR f/f ; PDX1 cre mice. Mice were sacrificed after anesthesia with 3% isoflurane. After cutting up, pancreatic tissues from mice were digested five times using a tissue dissociation cocktail containing 3.75 ml collagenase IV (1 mg/ml), 3.75 ml dispase (1 U/ml), 300 μl DNase I solution (1 mg/ml), and 22.2 ml DMEM/F-12 with 5 mM HEPES. The digested liquid was filtered through a reversable strainer to preserve the ductal fragments and remove other cells. Remaining ductal fragments on the filter were collected and suspended in Matrigel. A 50 μl suspension was added to the wells of a 24-well plate to form a dome. After 10 min of incubation in a 37°C incubator with 5% CO 2 , PEOs were cultured in 750 μl commercial culture medium per well (PancreaCult™ Organoid Growth Medium). PEOs were passaged-5-6 days after establishment. Briefly, PEOs were broken into small pieces using a pipette in Advanced DMEM/F-12. The diameter of fragments was limited to less than 70 μm using a strainer. The fragments were collected, mixed with Matrigel, and plated on cell culture plates. PEOs were cultured in commercial culture medium and passaged every 3-6 days. 4.7. Viability Assays of PEOs and the Area Calculation. The grown PEOs were broken into small pieces in Advanced DMEM/F-12 using a pipette. The diameter of fragments was also limited to less than 70 μm using a strainer. Fragments were collected and suspended in Matrigel. To ensure that the organoids were evenly distributed, the suspension was pipetted up and down 5-8 times and immediately added into the wells of a 96-well plate (10 μl per well). PEOs were cultured in 100 μl culture medium per well for 2-3 days. To explore the effect of different substances on PEOs, the culture medium was removed carefully, and new culture medium with different content was added without touching the PEO dome. The viability of organoids was evaluated using the CellTiter-Glo® Luminescent Cell Viability Assay (Promega). Image J (Version 1.53) [39] was used to calculate the area of PEOs. Briefly, we set scales according to the ruler in images, drew the outline of PEOs along the edges, and measured the area.  [42,43]. Heatmap was generated through pheatmap package on R Studio.
4.12. Osgin1 Manipulation in Pancreas of Mice. The viral tools were all packaged by BrainVTA (Wuhan) Co., Ltd. AAV viral vectors were constructed according to Chen et al. [21]. AAV2 genomes were encapsulated in the Y447F and Y773F mutant AAV8 capsids to generate a viral vector with high transduction efficiency for the pancreatic acinar cells (AAV-pan). For Osgin1 knockdown, AAV-pan with a U6 promoter driving shOsgin1 (GATGCTATACCCTGAG TACCA) and cytomegalovirus (CMV) promoter driving mCherry (AAV-pan-U6-shOsign1-CMV-mCherry) was constructed, and AAV-pan with a U6 promoter driving scramble (CCTAAGGTTAAGTCGCCCTCG) and CMV promoter driving mCherry protein (AAV-pan-U6-shControl-CMV-mCherry) expression served as a control. For Osgin1 overexpression, AAV-pan with a cytomegalovirus (CMV) promoter driving Osgin1 expression (AAV-pan-CMV-Osign1) was conducted and AAV-pan with CMV promoter driving green fluorescent protein (AAV-pan-CMV-GFP) served as a control. AAV was injected (1 * 10 11 vg per mouse) intraperitoneally two weeks before modeling. Successful transduction for every mouse was confirmed by in vivo fluorescence imaging after sacrifice.
4.13. Immunoblot. Samples were snap-frozen in liquid nitrogen immediately upon harvesting. RIPA buffer with freshly added protease inhibitor cocktail, and phenylmethylsulfonyl fluoride was used to lyse frozen-samples and extract proteins. We loaded equal amounts of protein (30-50 μg) on an SDS-polyacrylamide gel to separate them. Proteins were then electrotransferred onto PVDF membranes (Millipore), which were then blocked in 5% skim milk in Tris-buffered saline containing Tween 20 for 1 h at room temperature. The membranes were probed with primary antibodies overnight at 4°C and incubated at room temperature for 1 h with an appropriate secondary antibody conjugated to horseradish peroxidase. The protein signals were visualized with an ECL Western Blotting kit (Tanon). ImageJ was used to calculate protein relative expression by plot every gel lane. All protein expression was normalized by Ponceau S or loading control (β-actin, HSP90, or GAPDH).
4.14. Statistical Analysis. All comparisons between two groups were analyzed by unpaired, two-tailed Student's