Inhibition of hedgehog signaling ameliorates severity of chronic pancreatitis in experimental mouse models.

All in vivo mice experiments were carried out according to the guidelines set up by the Institutional Animal Care and Use Committee’s (IACUC) and Animal Resource Program (ARP), University of Alabama at Birmingham, USA. All the animal experiments were conducted using 6- to 8-week-old male mice of C57BL/6J (Jackson Laboratories, Catalog#000664, RRID:IMSR_JAX:000664) background weighing 22–27g. The mice were either in-bred in the mice facility of University of Alabama at Birmingham or ordered from Jackson Laboratories. Mice were housed in environment controlled micro-isolators under specific pathogen–free conditions with a 12-hour light/dark cycle and an ambient temperature of 23°C ± 2°C. The animals were given standard laboratory chow (LabDiet) as feed and provided with water ad libitum . Mice purchased from the vendors were acclimatized to this environment for at least 1 week prior to starting experiments. Two separate models involving caerulein and L-arginine were used for induction of CP in mice. In caerulein model, mice were administered with 7 hourly episodes of 50μg/kg caerulein (Bachem, Catalog #4030451) dissolved in phosphate buffered saline (Gibco, Catalog #10010) twice a week for 10 consecutive weeks (schematic, fig. 1a ) as previously described by us ( 79 ). For the L-arginine model of CP, a 9% solution of L-arginine (Sigma Aldrich, Catalog #11039) was made in normal saline (Cytiva, Catalog #Z1377) and adjusted to pH 7.0 with 1M Sodium hydroxide, followed by filtration using a 0.22-micron filter. This solution was stored at 4°C until use for a maximum period of four weeks, beyond which the left-over solution was discarded. The solution was pre-warmed to 37°C before it was injected into the mice. L-arginine-induced CP involved four episodes of intraperitoneal injections of L-arginine, with one episode occurring each consecutive week (schematic, fig. 2a ). Each episode comprised of two injections of L-arginine administered at one-hour intervals, followed by two hourly injections of normal saline (100 μl/g/dose, up to a maximum of 2.5 ml given subcutaneously), occurring one hour after the last L-arginine injection. The L-arginine dosage for the first episode was 4.5 g/kg ×2, for the second episode was 4.6 g/kg ×2, and for the third and fourth episodes, it was 4.75 g/kg ×2 ( 79 ). In the L-arginine model, treatment with Smo antagonist, vismodegib (50mg/kg; ip; once daily – MedChemExpress, Catalog #HY-10440) was started either after two weeks (i.e. after three episodes of L-arginine), referred to as the recurrent acute pancreatitis (RAP) model (schematic in fig. 3a ) or after four weeks (i.e. one week after the fourth L-arginine pancreatitis episode), referred to as the well-established model of CP (schematic in fig. 4a ). Treatment was continued for four weeks, and the mice were euthanized by CO 2 asphyxiation, one week after the last drug injection. In separate experiments using the well-established model of L-arginine CP, mice were euthanized after day 3, day 7 and day 10 of treatment to understand the earliest changes occurring in the pancreas after treatment (schematic in fig. 5a ). In the caerulein model, vismodegib treatment was started either after 6 weeks in the RAP model (schematic in Appendix fig. 4a ) or after 11 weeks (i.e. 1 week after the last set of caerulein injections), referred to as the well-established model of CP (schematic in Appendix fig. 6a ). After sacrifice, the mice body weight was measured prior to harvesting the pancreas. The pancreas was then harvested, weighed and a small portion was fixed in 10% neutral buffered formalin for histology. The remaining pancreas was flash frozen in liquid nitrogen and then transferred to −80°C until further use. For histologic evaluation, pancreatic tissues were fixed in 10% neutral phosphate buffered formalin for 24–48 hours and embedded in paraffin for further sectioning at histology core facility of UAB. Pancreatic tissue sections (4 μm) were stained with Hematoxylin and Eosin (H&E) and images were acquired on a light microscope using a camera attachment. The microscopic fields (10X magnification) were randomly selected per mouse by investigators blinded to groups, and pancreatic architecture, glandular atrophy, formation of pseudotubular complexes and fibrosis scores were calculated by histology scoring in a blinded fashion as previously described by us( 79 ). For immunofluorescence staining, pancreatic sections were de-paraffinized in xylene, re-hydrated in a series of alcohol solutions – (100%, 90%, 80%, 70% and finally PBS, each for 10 minutes) and then subjected to antigen retrieval process (30 minutes in a steamer at 95°C) using a pH 6 citrate buffer (Abcam, Catalog #ab93678). The sections were then permeabilized with 0.1% Triton-x-100 and blocked with Protein block (Abcam, Catalog #ab64226) for 15 minutes. The blocked sections were then probed with antibodies directed against Amylase (Santa Cruz Biotechnology, Catalog #sc-12821), Vimentin (Cell Signaling Technologies, Catalog #D21H3), Cytokeratin 19 (Abcam, Catalog #ab52625) and CD45 (Abcam, Catalog #ab10558). All the antibodies were diluted (1:100) in Background Sniper (Biocare Medical, Catalog #BS966L). Fluorochrome tagged secondary antibodies, Alexa Fluor ™ 488 anti-Goat IgG (Invitrogen, Catalog #A-11055) and Alexa Fluor ™ 594 anti-rabbit IgG (Invitrogen, Catalog #A-21207) were used at a dilution of 1:400. The sections were overlaid with Prolong ™ Gold antifade reagent containing DAPI (Invitrogen, Catalog # P36935 ). Visualization was performed under Nikon Ti fluorescent microscope. Representative images were captured using a camera attached to the microscope. Immunofluorescence quantification was performed using ImageJ software (Version 1.54g, RRID:SCR_003070) ( 80 ) and the percentage area of staining per microscopic field was calculated. RNA-FISH was performed on the pancreatic sections using the RNA Scope Multiplex Fluorescent Detection Kit (Advanced Cell Diagnostics, Catalog #323100) following the manufacturer’s instructions. Briefly, 4 μm sections were obtained from formalin fixed paraffin embedded tissue blocks and baked at 60°C for 1 hour followed by a series of de-paraffinization steps and treatment with hydrogen peroxide for 10 minutes. Target retrieval was performed in a steamer for 15 minutes followed by protease treatment of the slides for 30 minutes and probe hybridization for 2 hours at 40°C. Mouse specific RNA probes against Gli1 (Advanced Cell Diagnostics, Catalog #311001) and Collagen-1 (Advanced Cell Diagnostics, Catalog # 319371-C2) were used. After amplification of the signal using the reagents provided in the kit, the development was done with HRP linked to the appropriate probe channel. The fluorescent dyes used were Opal 570 (Akoya Biosciences, Catalog #FP1488001KT) for the Collagen probe and Opal 690 (Akoya Biosciences, Catalog #FP1497001KT) for the Gli1 probe, both used at a dilution of 1:750. The slides were counterstained with DAPI, mounted and visualized / imaged using a fluorescent microscope. For calculating the extent of co-localization between green (Gli1) and red (Col1a1) staining, we used an ImageJ plugin called “Just Another Co-Localization Plugin” (JACoP)( 81 ). This Plugin separately calculates the Mander’s overlap co-efficient (MOC)( 82 ) for both conditions – extent of green (Gli1) overlapping with red (Col1a1) and extent of red overlapping with green. The score varies from 0 (No overlap) to 1 (High overlap). Pancreatic stellate cells (PSCs) were isolated from C57Bl/6J mice according to a previously published protocol ( 83 ). Briefly, pancreas was quickly harvested after euthanizing mice by CO 2 asphyxiation, and the harvested pancreas from 3 mice were pooled together, washed once in ice cold RPMI medium (Gibco, Catalog #11875), followed by chopping into small chunks (~1 mm 3 ) and digestion in an enzyme mix (composition provided in table 1 ) for 20 minutes in a shaker incubator maintained at 37°C and 200 rpm. The digested slurry was dislodged using a 5ml pipette and passed through sterile gauge, collected in 50ml tube and then centrifuged for 10 minutes at 1500 rpm. The pellet was washed once with 25ml of 0.3% BSA solution (composition provided in table 1 ) and then re-suspended in 9ml of GBSS-A (composition provided in table 1 ) and mixed thoroughly by pipetting. This 9 ml of cell suspension was distributed into two 15ml tubes and each tube was overlaid carefully with 2ml of 0.3% BSA solution, without disturbing the bottom layer. The tubes were spun at 1400g for 20 minutes at 4°C without brake. The cells at the interphase were carefully collected from both tubes and pooled, centrifuged at 1700 rpm for 10 minutes. The pellet consisting of the PSCs were washed once with 0.3% BSA solution and resuspended in IMDM medium (Gibco, Catalog #12440) supplemented with 10% fetal bovine serum and transferred to a T25 flask for culture. After attaining confluence, the cells were sub-cultured at a ratio of 1:3 and the cells in P1 passage were used for in vitro experiments after attaining 80% confluency. PSCs were stimulated with 20ng/ml recombinant mouse TGFβ (R&D Systems, Catalog #7666-MB) and the experimental groups were treated with either vehicle (0.1% DMSO, Sigma Aldrich, Catalog #D8418) or vismodegib (100 μM) for 24h. Cells were then harvested for qPCR analysis and western blotting. For RNA isolation, harvested pancreatic tissue (either fresh or previously stored at −80°C) was homogenized using MagNa Lyser tubes (Roche, Catalog #03358941001) in 1ml of TRIzol reagent (Invitrogen, Catalog #15596018) and RNA was isolated using RNAEasy Plus Mini kit (Qiagen, Catalog #74134) according to manufacturer’s protocol. Briefly, 200ul of chloroform was added in the TRIzol lyzed tissue sample, mixed well by inversion and allowed to stand for 15 minutes. The tubes were spun at 10,000g for 15 minutes at 4°C and the clear aqueous phase containing the RNA fraction was collected and passed through the spin columns. The flow through was discarded. The columns were washed multiple times in a series of buffers provided in the kit (RW1 and RPE) and then RNA was finally eluted in appropriate volumes of Nuclease-free water (Corning, Catalog #46–000-CI). In case of PSCs, the cultured cells were harvested (after drug treatment) and lyzed in TRIzol by vortexing and RNA was isolated using RNAEasy Plus Micro kit (Qiagen, Catalog #74004) according to manufacturer’s instructions and as mentioned above. The harvested RNA was evaluated for quality and concentration using a nanodrop and cDNA was prepared from 2μg of RNA using high-capacity cDNA Reverse Transcription Kit (Applied Biosystems, Catalog #4368814) following kit instructions. Quantitative real-time analysis for target genes were performed using the Lightcycler 480 II Thermal cycle instrument (Roche) after adding the LightCycler 480 SYBR Green I Master (Roche, Catalog #04887352001) with the primer mixes (listed in table 2 ) in triplicates according to the manufacturer recommendations. The harvested PSCs were lyzed in RIPA buffer (Thermo Scientific, Catalog #J63306.AK) containing protease inhibitor (Roche, Catalog #04693132001). Protein quantification was done by the BCA method using Pierce ™ BCA Protein Assay Reagents A and B (Thermo Scientific, Catalog# 23228 and #23224). In total, 30μg of protein was loaded in each well, resolved by polyacrylamide gel electrophoresis using pre-cast Mini-PROTEAN TGX Stain-free gel (Biorad, Catalog #4568094), and then transferred to nitrocellulose blotting membrane (Cytiva, Catalog# 10600001). Membrane blocking was done by with 5% bovine serum albumin (BSA) (Sigma Aldrich, Catalog #A3294), followed by incubation with primary antibody against Collagen I (Invitrogen, Catalog #PA5–95137) and alpha-SMA (Abcam, Catalog#ab5694) overnight. Horseradish peroxidase–conjugated secondary antibody (Anti-Rabbit IgG, Cell Signaling Technology, Catalog #7074); was used. HRP conjugated antibody targeted against GAPDH (Cell Signaling Technology, Catalog #3683) was used as the loading control. The blots were developed using the SuperSignal ™ West Pico Plus Chemiluminescent substrate kit (Thermo Scientific, Catalog #345770) and visualized and imaged using GE ImageQuant LAS 200 Imager. Densitometry analysis of the blot was performed using ImageJ software using the analyze plot function. De-identified FFPE specimens of human patients suffering from chronic pancreatitis were obtained from the tissue biorepository of the University of Alabama at Birmingham. The patient details are listed in table 3 . H&E and Sirius red staining was performed in these specimens to understand the disease severity and extent of collagen deposition. Immunofluorescence was also performed in the FFPE sections to understand the expression and localization of hedgehog pathway components using the protocol described previously in the microscopy section. The antibodies used were human specific and directed against Gli1 (Invitrogen, Catalog #MA5–32553), Gli2 (Proteintech, Catalog # 18989–1-AP), Shh (Invitrogen, Catalog #701403) and Vimentin (Abcam, Catalog # ab20346). All were used at a dilution of 1:100. We used a publicly available dataset of single-nuclear RNA sequencing (snRNAseq) of human chronic pancreatitis specimen to evaluate the cellular localization of hedgehog pathway activation in CP( 84 ). We loaded the Seurat object file (chronic_pancreatitis.rds) containing pre-annotated clusters, downloaded from http://singlecell.charite.de/cellbrowser/pancreas/ into R (version# 4.3.1) and used the DotPlot parameter to create dotplots for the genes of interest (GLI1, GLI2 and SMO). Total RNA was isolated from the pancreatic tissue samples harvested from vehicle or vismodegib treated mice (n = 4 per group) as described above using the RNAEasy Plus Mini kit (Qiagen, Catalog #74134). Library preparation was performed with 200ng of total RNA using the NEB Next Ultra II Directional RNA Seq kit with poly A selection (NEB, Catalog# E7760), following manufacturer’s instructions. The libraries were evaluated on the Agilent Bioanalzyer 2100 for quality and size distribution and sequenced pair ended 100 bp on the Illumina NovaSeq 6000 using standard protocols. Quality control of RNA-Seq raw read data (FASTQ) involved evaluating GC bias, total number of reads, overall base quality score, rRNA content, and duplication rates. Assessment of sequencing performance was evaluated using FastQC prior to bioinformatics analysis. The raw sequencing reads were processed using the Illumina CASAVA package (V.1.8.2). Short reads FASTQ format were processed and aligned to the Mouse reference genome mm39. The output SAM file was used for transcript assembly and abundance analysis using Cufflinks, and for quality control. Fragments per kilobase of exon per million mapped reads (FPKM) values were calculated for each gene in each sample. Subsequent analyses were carried out using the R software, while pathway analysis was performed using the Ingenuity Pathway Analysis software (Ingenuity Systems/Qiagen, USA). The Kruskal-Wallis Test (Dunn’s Multiple Comparisons Test) was employed to assess differences among multiple groups, while Mann Whitney Test (non-parametric test) was utilized for comparing two groups. p value of less than 0.05 was considered statistically significant. The statistical significance between groups were calculated and graphs were generated using Graphpad Prism 10 software (San Jose, CA, RRID:SCR_002798). Results are presented as mean ± standard error of mean (SEM).

In the caerulein induced model of CP (CerCP), C57BL/6 mice were repeatedly administered caerulein injections for a period of 10 weeks (schematic, fig. 1a ). CP induced mice had significantly low pancreas to body weight ratio (pancreatic atrophy) in comparison to control healthy mice ( fig. 1b ), due to reduced acinar cell mass. Consistent with human CP ( 43 ), the loss of pancreatic acini, occurrence of pseudo-tubular complexes, alternately called Acinar-to-Ductal Metaplasia structures (ADMs) and collagen deposition were evident in the histology imaging and scoring after hematoxylin and eosin (H&E) and Sirius red staining in the CerCP group ( fig. 1c , Appendix fig. 1a , b ). There was an increase in the prevalence of ADMs as revealed by Cytokeratin 19 (CK19) staining ( Appendix fig. 1c ). As previously reported in this model ( 44 ), quantitative PCR (qPCR) analysis of the injured pancreatic tissue indicated that the markers of fibrosis such as Collagen-1 and alpha-smooth muscle actin (αSMA) were highly upregulated in the CerCP group ( Appendix fig. 1d ). In response to an injury, the pancreatic acinar cells undergo de-differentiation and starts exhibiting certain embryonic features. At this stage, some characteristics of ductal cell-like phenotype are acquired by acinar cells, in an event described as Acinar-to-Ductal Metaplasia ( 45 – 47 ). Among the most common features of ADMs are enhanced expression of ductal markers like CK19 and Sox9 along with decreased expression of amylase, which is characteristic of healthy acini ( 48 , 49 ). We observed a significant upregulation of the ADM marker, CK19 in the CerCP group ( Appendix fig. 1e ). Conversely, the acinar cell markers such as Protease serine (Prss1), Carboxypeptidase (Cpa1) and Amylase (Amy1) were significantly downregulated ( Appendix fig. 1f ). qPCR analysis also revealed a significantly upregulated expression of hedgehog pathway components such as the transcription factors, Gli1 and Gli2; the hedgehog pathway ligand, Indian hedgehog (Ihh) and the hedgehog pathway downstream gene, Zeb1 ( 50 ) and Snai1 ( 51 ) in the CerCP group ( fig. 1d , Appendix fig. 1g ). However, we did not observe any changes in the transcriptional levels of other known hedgehog pathway ligands such as Desert hedgehog (Dhh) and Sonic hedgehog (Shh) ( Appendix fig. 1g ). Other characteristic features of CP such as increased pancreatic infiltration of macrophages (F480), a bulk of which are alternatively polarized with increased expression of M2 markers such as CD206 and Arginase (Arg1)( 52 ) were evident in this model ( Appendix fig. 1h ). Further assessment by qPCR indicated elevated expression of pro-fibrotic cytokines like TGFβ along with pro-inflammatory cytokines such as TNFα and IL6 ( Appendix fig. 1i ), revealing the heterogeneity of the immune micro-environment upon CP induction. RNA-fluorescence in situ hybridization (RNA-FISH) studies reinforced the qPCR results suggesting that while the expression of Gli1 transcripts, which is widely considered as a surrogate marker of hedgehog pathway activation, is negligible in the control healthy pancreas, it is drastically overexpressed in the CerCP group ( fig. 1e ). Interestingly, this over-expression of Gli1 pre-dominantly co-localized with the Collagen-1 expressing cells in the pancreas, which are primarily the activated pancreatic stellate cells (PSCs). The Mander’s Overlap Co-efficient (MOC) index confirmed that green staining (Gli1) significantly overlapped with red staining (Col1a1). Conversely, the low MOC index for red (Col1a1) overlapping with green (Gli1) suggests that while Gli1 is expressed mainly in the PSCs, not all PSCs express Gli1. It is important to note that for calculating the co-localization index, ImageJ quantifies the area of green stain overlapping with red stain and vice versa . As seen from the figure ( fig. 1E ), Gli1 staining appears as dots, whereas Col1a1 staining appears diffuse. This may explain the relatively low MOC observed for Col1a1 overlapping with Gli1. Overall, this indicates that upregulation of hedgehog pathway in PSCs might be responsible for increased fibrosis and formation of scar tissue in the injured pancreas, thereby inhibiting acinar regeneration. To rule out the possibility that the activation of hedgehog pathway in CerCP model could be model specific, we employed the more severe L-arginine induced model of CP (L-argCP) in mice according to the schematic shown in fig. 2a and assessed the expression of hedgehog pathway components. As previously reported by us ( 53 ), the L-argCP model is characterized by severe atrophy ( fig. 2b ), with near complete loss of acinar cells, formation of pseudo-tubular complexes, fatty replacement and extensive collagen deposition as revealed by H&E and Sirius red staining ( fig. 2c ). Histological and Sirius red scoring corroborates with the severe injury and fibrosis observed in this model ( Appendix fig. 2a , b ). ADM prevalence as measured by CK19 staining is higher compared to the CerCP model ( Appendix fig. 2c ). The increase in the expression levels of pro-fibrotic gene transcripts (Collagen-1, αSMA) and ADM markers (CK19 and Sox9) were of a much higher magnitude compared to that of CerCP model ( Appendix fig. 2d , e ). Correspondingly, the decrease in the acinar cell markers (Prss1, Amy1 and Cpa1) was also evident ( Appendix fig. 2f ). Quelling concerns of any model specific effects at play, qPCR analysis confirmed an increased expression of hedgehog pathway components and the downstream genes (Ihh, Gli1, Gli2 and Zeb1) in the L-argCP group compared to the control healthy pancreas ( fig. 2d ). Interestingly the expression of hedgehog pathway ligands, (Dhh and Shh) that were unchanged in the CerCP model showed a significant increase in the L-argCP model ( Appendix fig. 2g ). Consistent with our findings of CerCP model, herein too we found increased macrophage infiltration (F480) with polarity shifting towards the M2 lineage (increase in CD206 and Arg1 expression) ( Appendix fig. 2h ). Other pro-fibrotic and inflammatory cytokines such as TGFβ, TNFα, IL6 and IL10 demonstrated a significant increase in the L-argCP group ( Appendix fig. 2i ). RNA-FISH and Mander’s overlap co-efficient (MOC) index re-confirmed predominant expression of Gli1 transcripts in the Collagen-1 expressing cells in the L-argCP model ( fig. 2e ). Clinical studies suggest that ~35% of patients with recurrent acute pancreatitis (RAP) will eventually progress to CP ( 54 ). Hence it is crucial to develop strategies to inhibit this progression from RAP to CP. Since the progression of CP was associated with an increased activation of hedgehog signaling, we questioned whether pharmacological inhibition of hedgehog signaling with vismodegib, when administered along with a concurrent ongoing injury, will prevent the development of CP. For this, experimental mice were randomized after 2 weeks of L-arginine treatment (i.e. after 3 episodes of L-arginine injections) into vismodegib (50mg/kg) or vehicle treatment groups. Mice were euthanized 1 week after the completion of the treatment and pancreas was harvested (schematic, fig. 3a ). We observed a significantly higher pancreas to body weight ratio in the vismodegib treatment (L-argCP + vismodegib) group at the end of treatment compared to the vehicle group (L-argCP 8wks), suggesting a reduced level of injury in the absence of hedgehog signaling ( fig. 3b ). H&E and Sirius red staining along with the corresponding blinded scoring also revealed an increased acinar cell mass and reduced collagen deposition post vismodegib treatment ( fig. 3c ). Reduced expression of Gli1 transcripts in the drug treatment group validated the treatment efficacy ( fig. 3d ). We also observed either a significant downregulation or at least a trend towards downregulation in the expression of other hedgehog pathway components (Gli2 and Ihh) and the hedgehog pathway downstream gene, Zeb1 in the treatment group (Appendix fig. 3b) . Furthermore, we observed a significant decrease in the expression of the pro-fibrotic marker, Collagen-1 in the vismodegib treated group ( fig. 3d ), while αSMA expression demonstrated a decreasing trend compared to the vehicle group (Appendix fig. 3c) . Immunofluorescence analysis of the pancreatic sections indicated an overall decrease in the pancreatic infiltration of CD45 immune cells (Appendix fig. 3a) . There was a drastic reduction in macrophage infiltration, especially the M2 polarized macrophages as evident by transcript levels of F480 and CD206 in the vismodegib treated pancreas compared to the vehicle group ( fig. 3d ). The vismodegib treated group exhibited reduced expression of CK19 and Sox9 genes signifying reduced ADM associated features ( fig. 3d , Appendix fig. 3d ). Conversely, we observed an increased expression of the markers of healthy acinar cells (Cpa1, Prss1 and Amy1) in the drug treatment group ( fig. 3d , Appendix fig. 3e ). Immunofluorescence imaging revealed an increase in acinar cell mass (amylase staining, green) with a concomitant reduction in vimentin (fibrosis marker, red) and CK19 (ADM marker, red) expression upon vismodegib treatment, which can be attributed to increased acinar cell mass replacing the fibrotic areas in the pancreas in the absence of active hedgehog signaling ( fig. 3e ). Additionally, the expression of all the pro-fibrotic and pro-inflammatory markers also showed a significant decrease in the treatment group (Appendix fig. 3f) . We re-confirmed these findings in the CerCP ongoing injury model, to eliminate any model specific effects. In this model, vismodegib treatment was started after 6 weeks of caerulein injections (schematic, Appendix fig. 4a ). Here again, we observed a significant improvement in pancreatic atrophy in the drug treatment group (CerCP + vismodegib) compared to the vehicle (CerCP 12wks) (Appendix fig. 4b) . The parameters associated with the histopathology of the pancreas also improved, as evident by H&E imaging and Sirius red staining (Appendix fig. 4c) . Immunofluorescence imaging confirmed similar findings as noted with L-argCP ongoing injury model, where vismodegib treatment reduced the expression of vimentin and CK19 and increased amylase expression. Further, hedgehog signaling inhibition also reduced CD45 immune cell infiltration in the pancreas (Appendix fig. 4d) . Often patients present to the healthcare facilities at advanced stage of CP, where the acinar cells are already replaced with a scar fibrous tissue leading to severe exocrine insufficiency. To simulate this clinical scenario, we investigated the therapeutic efficacy of vismodegib in well-established model of L-argCP, where the drug treatment was started one week after the last episode of injury ( fig. 4a ). Interestingly, even when the treatment was started at an advanced stage of the disease, H&E and Sirius red staining and the corresponding blinded scoring confirmed a profound improvement in the parameters of CP including pancreatic atrophy ( fig. 4b ) and histology, in the vismodegib treated group (L-argCP + vismodegib) compared to the vehicle group (L-argCP 16wks). Histological images demonstrated amelioration in fibrosis and collagen deposition as well as reduced persistence of ADM structures, while the acinar cell mass and the overall pancreatic architecture showed a considerable improvement ( fig. 4c ). Attenuated expression of hedgehog pathway transcription factors, Gli1 and Gli2 as well as the downstream gene, Snai1 validated the efficacy of drug treatment ( fig. 4d and Appendix fig. 5a ). Additionally, there was a reduced expression of the hedgehog pathway ligand, Ihh ( Appendix fig. 5a ). Similar to our observations in the recurrent acute pancreatitis model (ongoing injury model), we found lower infiltration of macrophages, as revealed by decreased pancreatic expression of F480, especially the alternatively polarized macrophages (CD206) ( fig. 4d ). Remarkably, vismodegib treatment not only arrested progression of fibrosis, as evident in the ongoing injury model, it was also able to resolve already established fibrosis in an established CP model as indicated by the qPCR results for pro-fibrotic markers, Collagen-1 and αSMA ( fig. 4d and Appendix fig. 5b ). The resolution in fibrosis co-related with a corresponding decrease in ADM markers such as CK19 and Sox9, and a concomitant increase in the expression of acinar markers like Amy1 and Cpa1 ( fig. 4d and Appendix fig. 5c , d ). The expression level of IL10, the cytokine most often associated with the polarity of M2 macrophages ( 55 , 56 ) also decreased on vismodegib treatment and so were the expression of other pro-fibrotic and pro-inflammatory cytokines such as TGFβ and TNFα ( Appendix fig. 5e ). Finally, immunofluorescence revealed increased acinar cell mass as shown by amylase staining (green), with a parallel decrease in the fibrosis marker, vimentin (red) and reduced persistence of the ADM marker, CK19 (red) ( fig. 4e ). Similar results were found post vismodegib treatment in the well-established model of CerCP, where the treatment with vehicle (CerCP 16wks) or vismodegib (CerCP + vismodegib) was commenced one week after the last set of caerulein injections (schematic, Appendix fig. 6a ). We observed an improvement in pancreatic atrophy as well as pancreatic histology, with significantly reduced fibrosis, collagen deposition and persistence of ADMs ( Appendix fig. 6b and 6c ). qPCR results further confirmed the decrease in expression levels of hedgehog pathway transcription factor (Gli1), fibrotic marker (Collagen-1, αSMA) and reduced infiltration by M2 polarized macrophages (CD206 and Arg1 expression) ( Appendix fig. 6d ). Immunofluorescence staining re-affirmed our findings that vismodegib treatment led to decrease in fibrosis (vimentin, red staining) and reduction in ADM marker (CK19, red staining) concomitant with an increase in amylase expression (green staining), indicative of acinar cell regeneration ( Appendix fig. 6e ). The clinical relevance of our murine based study can be appreciated, only if similar pattern of hedgehog pathway upregulation is evident in human patients suffering from chronic pancreatitis. We examined formalin fixed paraffin embedded (FFPE) sections of pancreatic tissues obtained from patients suffering from chronic pancreatitis ( Table 3 ). H&E imaging of the pancreas from two CP patients confirmed the associated fibro-inflammation typical to the disease with intense collagen deposition as indicated by Sirius red staining ( fig. 6a ). Additionally, the snRNAseq analysis of human pancreatic specimens of chronic pancreatitis revealed that the hedgehog pathway components - GLI1, GLI2 and SMO are pre-dominantly expressed in the activated pancreatic stellate cell compartment during CP ( fig. 6b ). Immunofluorescence staining of the human pancreatic sections revealed extensive fibrosis as seen by vimentin expression, with a concomitant upregulation of the hedgehog pathway components such as Gli1, Gli2 and Shh, especially in the fibrotic areas ( Appendix fig. 7 ). Taken together, the data clearly establishes the involvement of hedgehog pathway in the pathogenesis of chronic pancreatitis in humans. While treatment with vismodegib for four weeks was able to improve the parameters of pancreatic injury in multiple experimental mouse models of CP, it is important to understand the kinetics of recovery and evaluate early changes induced by the inhibition of the hedgehog signaling. In the L-argCP model, we euthanized mice after 3, 7 and 10 days of treatment with vismodegib or vehicle (schematic, fig. 5a ). Unsurprisingly, there was no improvement in the pancreatic atrophy and histology after 3 days of treatment. The day 7 pancreas showed a non-significant trend towards increased pancreatic weight, whereas the difference acquired significance by day 10 in the vismodegib treatment group compared to the vehicle ( fig. 5b and 5c ). There was a trend towards an attenuated hedgehog signaling as indicated by the expression of Gli1 and Gli2 by qPCR, as early as day 3 of treatment and continuing till day 7. By day 10, we observed a significant decrease in the expression levels of Gli1 and Gli2, in the vismodegib treatment group ( fig. 5d ). Surprisingly, although there was only a minor improvement in the gross histology of the pancreas as visualized by H&E imaging, the markers of fibrosis (Collagen-1 and αSMA) were significantly downregulated within 10 days of vismodegib treatment ( fig. 5d ). These results indicate that amelioration of fibrosis is a critical pre-requisite for rendering a pancreatic microenvironment conducive for acinar regeneration at later stages. To further understand the earliest gene expression changes and signaling pathways modulated by hedgehog pathway inhibition, we performed Bulk RNA-seq of the pancreas from the vehicle and the vismodegib treated groups; after 3 days of treatment. Principal component analysis (PCA) of the RNA-seq dataset revealed that samples from the two groups clustered distinct from each other ( fig. 7a ). The heatmap revealed the relative expression of significant DEGs in the dataset ( fig. 7b ). The volcano plot ( fig. 7c ) revealed significantly up-regulated and down-regulated DEGs in the dataset. The dataset revealed a total of 1978 significant differentially expressed genes (DEGs) with false discovery rate (FDR) =2. Of these, 1030 were up-regulated and 948 were down-regulated (Supp. Table 1). This included 1377 protein coding genes (632 upregulated and 745 downregulated). The top upregulated and downregulated protein-coding genes (with foldchange >= 5 and FDR < 0.05) between the vehicle and the vismodegib treated group are listed in Table 4 . Among the top upregulated genes were found the markers of acinar cells such as Amy2a1, Prss1, Prss3 and Amy2a5 indicating the commencement of regenerative process. The top downregulated DEGs included multiple histone transcripts, genes involved in metabolic pathways as well as other pro-fibrotic markers such as Fabp3( 57 ). Ingenuity Pathway Analysis (IPA) of DEGs (with an FDR = +/− 2) revealed 63 significant canonical pathways. Considering z-score > 2 for significant activation and z-score < −2 for significant inhibition, we found 21 pathways that were significantly inhibited ( Table 5 ) in the vismodegib treated group, while none of the pathways were significantly activated. The genes that contributed to the prediction of these pathways are listed in (Supp. Table 2). Among the top inhibited pathways include those pertinent to eukaryotic transcription and translation initiation and elongation, rRNA processing as well as other amino acid metabolic pathways. The Upstream Regulator Analysis (URA) function of IPA was applied to the DEGs identified in our dataset. The URA predicts the upstream transcriptional regulators from the dataset based on the literature and compiled data in the Ingenuity Knowledge Base. IPA identified 58 upstream regulators that were predicted to be most significantly regulated. Of these, 26 were predicted to be inhibited while 32 were predicted to be activated after vismodegib treatment ( Table 6 ). Among the inhibited upstream regulators were transcriptional regulators such as MLXIPL (chREBP), MYC and TEAD1 and the translation regulator EIF4E, with confirmed roles in tissue fibrosis( 58 – 61 ). Supp. Table 3 provides information on the target genes in the dataset which contributed to the prediction as well as the molecular type of the regulator. Myo-fibroblast like pancreatic stellate cells (PSCs) scantily populate the exocrine region of the pancreas. Recurrent pancreatic injury and subsequent wound healing response results in local upregulation of anti-inflammatory cytokines, which eventually results in transformation of quiescent PSCs to activated fibroblasts( 62 – 64 ). Based on our observation (RNA-FISH studies) that hedgehog pathway is pre-dominantly activated in the PSCs and owing to the critical role played by PSCs during CP, we sought to understand the role of PSC-specific hedgehog pathway signaling in the development of pancreatic fibrosis. For this, we isolated PSCs from the pancreas of C57BL/6 mice and stimulated them with TGFβ for 24 hours. TGFβ is a pro-fibrotic cytokine relevant in the context of CP associated pancreatic fibrosis( 65 ). We found an enhanced expression of the hedgehog pathway activation marker, Gli1 and Gli2 upon stimulation with TGFβ and this increase was expectedly countered by vismodegib treatment ( fig. 8a ). Concurrently, TGFβ treatment also led to increased expression of ECM markers such as Collagen-1 and αSMA, thus confirming activation of PSCs. In resonance to our in vivo findings, treatment of TGFβ activated PSCs with vismodegib resulted in reduced fibrosis, as evident by decreased expression of the pro-fibrotic markers by qPCR ( fig. 8b ). Additionally, western blot analysis confirmed the attenuated levels of Collagen and alpha-SMA protein in PSCs, after hedgehog pathway inhibition ( fig. 8c ).

Chronic pancreatitis (CP) is a fibro-inflammatory disease of the pancreas with unknown pathophysiology and no specific cure. CP typically involves recurring episodes of acute pancreatitis (AP) often resulting in pancreatic insufficiency due to replacement of the enzyme producing parenchymal cells with scar tissue. In this paper, we demonstrated that while hedgehog signaling is almost absent in healthy pancreas, it re-activates during CP. CP is caused by varied etiologies and pathogenesis and hence it is necessary to validate the findings across multiple experimental models to avoid any model specific effects. In this manuscript, we confirmed our findings using two different mouse models of experimental CP. The caerulein model is a widely used model of CP, where repeated administration of supramaximal doses of the cholecystokinin analogue caerulein, provokes sustained secretion of digestive enzymes by the pancreatic exocrine cells, leading to chronic injury, atrophy and development of CP. This is a highly reproducible model of mild CP where the histopathological evaluations closely resemble clinical scenarios. The L-arginine model on the other hand is a highly severe model of CP with almost complete ablation of the pancreatic acinar cells and extensive fibrosis. While it is not a preferred model of CP owing to its complexity, it is nevertheless important to validate the experimental findings in a severe model to account for severe cases of clinical CP. As expected, while there was an activation of hedgehog pathway in both the mild and severe models of CP, the magnitude was much higher in the L-arginine model as estimated by the expression of Gli1, which is a reliable indicator of hedgehog pathway activation. Furthermore, we also observed that the activation of hedgehog signaling was pre-dominantly localized to the collagen-1 expressing pancreatic stellate cells (PSCs) of the pancreas in both experimental mice models as well as human CP. This is in agreement to other studies in the context of pancreatic cancer, where activation of hedgehog signaling was restricted to cancer associated fibroblasts (CAFs) ( 26 ) as well as in other fibrotic diseases such as NASH, where hedgehog signaling is prominently activated in the hepatic stellate cells( 66 ). PSCs are myofibroblast-like cells in the pancreas that are maintained in a quiescent state in the healthy pancreas( 67 ). Injury to the pancreas leads to activation, increased proliferation and migration of these PSCs to the site of injury( 67 ). Activated PSCs help in re-modelling of the injured tissue by secreting extracellular matrix proteins such as laminin, fibronectin and collagen ( 68 , 69 ). Sustained injury stimulus leads to replacement of the pancreatic parenchyma with the fibrotic tissue commonly observed in CP( 70 , 71 ). Hedgehog signaling is known to influence activation of PSCs( 72 ). Inhibition of hedgehog signaling has been shown to exert potent anti-fibrotic effects in several fibrotic diseases as well as resulting in resolution of cancer. In this study, we used a Smo antagonist, vismodegib which is a pharmacological small molecule inhibitor to inhibit hedgehog signaling. To simulate clinical scenarios, we selected two treatment regimens – one where the treatment with vismodegib was started in the presence of an ongoing injury, which reflects the clinical scenario of recurrent acute pancreatitis (RAP). In the other regimen, we started with vismodegib treatment after the CP condition was well-established with irreversible pancreatic atrophy and fibrosis. Treatment with vismodegib significantly attenuated the severity of injury resulting out of recurrent episodes of acute pancreatitis. Strikingly, vismodegib was also able to facilitate resolution of pancreatic fibrosis from a well-established and irreversible CP condition. These findings were validated in both the caerulein as well as the highly severe L-arginine model of CP. Next, we investigated the underlying mechanistic aspects by which inhibition of hedgehog signaling results in improved CP parameters. Although a complete treatment regimen (four weeks in this study) offers insight into the effectiveness and feasibility of the approach, it fails to provide a mechanistic understanding of the pathways through which hedgehog signaling inhibition leads to disease improvement. Hence, we analyzed the pancreas at early timepoints of treatment, monitoring the kinetics of recovery and assessing early molecular changes. Our results indicate that inhibition of hedgehog signaling primarily attenuates fibrosis, thereby creating a microenvironment conducive for acinar cell regeneration. Our RNA-seq results provide comprehensive mechanistic details on the effect of vismodegib treatment on the recovering pancreas including the downstream signaling pathways that are directly influenced by inhibition of hedgehog signaling. We identified several pathways that have reported role in the fibrotic process. Interestingly, the pancreatic transcriptomics revealed that although there is no evident improvement in histology after 3 days of treatment, the regenerative process has already commenced as revealed by up-regulation of several acinar cell transcripts. Hedgehog signaling is a critical embryonic pathway involved in a multitude of developmental processes such as patterning of neural tube, brain and limb development as well as proper development of the foregut and gastrointestinal system ( 73 – 76 ). In contrast, the absence of hedgehog signaling in the embryonic pancreatic epithelium for the entirety of development process is surprising and it is generally believed that hedgehog signaling has an inhibitory function during the development of early pancreas( 76 ). While sonic hedgehog expression impairs pancreas formation, the inhibition of hedgehog signaling using a Smo antagonist, cyclopamine leads to development of ectopic pancreas buds in the developing chick embryos in the Pdx-1 expressing areas originally destined for differentiation into posterior stomach and proximal duodenum( 77 ). Additionally, there are contradicting reports on the role of hedgehog signaling during pancreatitis. For instance, Fendrich et al( 78 ) reported that hedgehog signaling is important for the regeneration of exocrine pancreas following AP, while our results in multiple experimental models of CP reveal a deleterious role of hedgehog signaling during CP. However, our study differs from their study in multiple aspects. While AP is mainly an inflammatory condition, involving local and systemic inflammation, the hallmark feature of CP is fibrosis leading to loss of pancreatic function. It is possible that during the progression of the disease from the acute to the chronic phase, the role of hedgehog signaling changes from a pro-regenerative and anti-inflammatory role in AP to a pro-fibrotic role in case of CP. In fact, this differential role of hedgehog signaling is well documented in literature( 15 ). We hypothesize that while it plays a critical protective role during AP by quelling inflammation, its activation in CP accentuates fibrosis and thus acquires a disease-promoting role. Our in vitro experiments also reveal a reversal in the activation of PSCs, from a pro-fibrotic to a quiescent phenotype after hedgehog pathway inhibition, further supporting this hypothesis. This study has immense translational benefits. CP and RAP have no known cure and the resulting pancreatic fibrosis is irreversible without therapeutic intervention. The therapeutic agent, vismodegib that we used in this study is an FDA approved drug with proven clinical safety profile and manageable adverse effects. Administration of this drug had no obvious adverse effects on any of the organs in multiple mice models of CP that we studied. These reasons make it a lucrative target for commencing a clinical trial using this drug for the treatment of CP and RAP. This study, however, has few limitations. Detailed mechanistic experiments are needed to understand the kinetics and mechanism of pancreatic recovery after hedgehog signaling inhibition. The pathways identified in the RNA-seq studies needs further exploration and validation. Identification of other targets within the differentially regulated pathways can pave way for exploring new therapeutic strategies with minimal off-target effects. A flow cytometric evaluation of the pancreas at early stage of treatment will offer insights on the immune modulation induced by inhibition of hedgehog signaling, either directly or mediated through other cells (such as the PSCs). It is also important to understand the impact of PSC-specific hedgehog signaling on the pancreatic microenvironment including the immune cells, acinar cells, and the PSC itself and how hedgehog signaling modulates the crosstalk that happens between them during CP. Using a PSC- specific conditional hedgehog signaling knockout mice and overexpression models will validate the hypothesis that PSC-specific hedgehog signaling is the main driver of CP. Last but not the least, incorporating clinically relevant experimental models in the study such as alcohol-smoking and smoking-Lipopolysaccharide (LPS) models of CP can increase human relevance of these findings. In summary, we demonstrated that hedgehog signaling is activated in CP and plays an important role in the pathogenesis of the disease. Inhibition of hedgehog signaling can be used as a novel therapeutic strategy against CP, thus improving the lives and quality of life of the patients suffering from this debilitating disorder.

Chronic Pancreatitis (CP) is an irreversible fibro-inflammatory disease of the pancreas with an incomplete understanding of its pathophysiology ( 1 ). There is no specific cure and currently, treatment is mainly symptomatic involving pancreatic enzyme replacement therapy (PERT)( 2 ) and pain management with opioids, often leading to opioid dependence( 3 ). Consequently, there is an unmet need to understand the pathogenesis as well as develop novel therapeutic strategies against this disease. Pancreatic atrophy and fibrosis are the hallmark characteristics of CP, where the pancreatic parenchyma, including the acinar cell mass is progressively replaced by fibrous tissue leading to pancreatic insufficiency( 4 , 5 ). Understanding signaling pathways that drive fibrosis is pivotal in addressing this disease. There is accumulating evidence that chronic dysregulated activation of several key developmental signaling pathways may play a role in development of tissue fibrosis( 6 – 9 ). In this regard, hedgehog signaling is an evolutionary conserved developmental signaling pathway that is highly active during embryonic and neonatal stages( 10 ), but with undefined roles in healthy adults and is often re-activated in pathological conditions such as cancer( 11 ). The binding of hedgehog pathway ligands (Sonic, Indian and Desert hedgehog)( 12 ) to the transmembrane receptor, Ptch (Patched) present in hedgehog pathway responsive cells leads to subsequent release of its repression on another transmembrane protein, Smo (smoothened, frizzled class receptor). The activation of Smo leads to a series of downstream signaling cascade finally culminating in the nuclear translocation of the transcription factors belonging to the Gli (glioma-associated oncogene) family - Gli1, Gli2 and Gli3. This leads to subsequent expression of hedgehog pathway dependent downstream genes( 10 ). In general, while hedgehog pathway is quiescent in healthy adult tissues( 13 – 15 ), reactivation of this pathway is observed as a reparative wound healing mechanism in response to an injury( 16 , 17 ). However, aberrant activation of this pathway is known to be implicated in several fibrotic diseases such as renal( 18 , 19 ), bile duct( 20 ), hepatic( 21 , 22 ), pulmonary( 23 ), skin fibrosis( 24 ) as well as in carcinogenesis including pancreatic ductal adenocarcinoma (PDAC) ( 25 – 27 ). Isolated reports also indicate activation of this pathway in human( 28 ) and experimental model of chronic pancreatitis( 29 ). However, these studies lack mechanistic details and to the best of our knowledge, none of the studies have explored the pharmacological inhibition of hedgehog signaling as a therapeutic strategy against CP. Inhibition of hedgehog signaling has been explored as a therapeutic strategy against various fibrotic conditions( 30 – 32 ). Hedgehog pathway inhibition using Smo antagonists such as Vismodegib and Cyclopamine have been widely reported in experimental animal models ( 33 , 34 ). Vismodegib is an FDA approved small molecule Smo antagonist that been indicated for use against advanced basal cell carcinoma( 35 ) with manageable adverse effects ( 36 ). In addition to its anti-tumorigenic effects( 37 , 38 ), vismodegib also has demonstrated anti-fibrotic properties in experimental mice models of advanced hepatic fibrosis( 31 , 39 ). On the other hand, Gli inhibitors, such as GANT-61( 40 ), target both the canonical as well as non-canonical (Smo independent TGFβ mediated activation of Gli1( 41 , 42 )) arms of hedgehog signaling. In this study, we have systematically explored the role of hedgehog signaling in the progression of chronic pancreatitis using multiple experimental mouse models of chronic pancreatitis as well as in human clinical specimens. Our findings revealed that hedgehog signaling is overexpressed during CP and it is predominantly activated in the pancreatic stellate cells of the pancreas. Finally, we explored the pharmacological inhibition of hedgehog signaling as novel therapeutic strategy against chronic pancreatitis.

Supplementary Table 1 - Vismodegib vs Vehicle DESeq2 annotated results with normalized counts Supplementary Table 2 - List of canonical pathways Supplementary Table 3 - List of upstream regulators URL: https://figshare.com/s/84ce2402475a3c6cc06f DOI: 10.6084/m9.figshare.27280917