Loss of RE-1 silencing transcription factor accelerates exocrine damage from pancreatic injury

Animals

All animal work was done in compliance to a protocol approved by the University of Florida Institutional Animal Care and Use Committee. p48-Cre^+/− mice (also known as Ptf1a-Cre^+/− mice) from a C57BL/6J background were a generous gift from Dr. Paul Grippo. REST^floxP/floxP mice came from a C57BL/6J background and were a generous gift from Dr. Jenny Hsieh³⁰. Conditional REST knockout mice (p48-Cre^+/−/REST^fl/fl) mice were bred in-house by crossing p48-Cre^+/−/REST^fl/fl females with REST^fl/fl males. Pup tail snips were digested in DirectPCR Lysis Reagent (Qiagen) supplemented with 20 mg/ml Proteinase K (Invitrogen) overnight at 55 °C then heat-treated at 85 °C for 45 min. Crude DNA was genotyped for the Cre Recombinase transgene and floxed Rest by PCR using LongAmp Taq DNA Polymerase (BioLabs). PCR products were resolved on a 1.2% agarose gel for 45 min at 120 V, resulting in a band at 450 base pairs for Cre transgene positive pups, a band at 550 base pairs for WT Rest allele, and a band at 700 base pairs for floxed Rest allele. PCR primer sequences are as follows: Cre Sense Primer (P1): 5′-CCTGGA AAATGCTTCTGTCCG-3′; Cre antisense primer (P2): 5′-CAGGGTGTTATAAGCAATCCC-3′; REST sense primer (P3): 5′-ACAGGATCTCTAGGAGCTCAGACTGG-3′; REST antisense primer (P4): 5′-CCAGGGTTCAGTTCTCTACACCCAC-3′. Mice were randomized to experimental groups according to age and sex so that near equal representation were in both groups.

Genomic pancreas DNA isolation and sequencing

To confirm Cre-mediated excision of pancreatic DNA, a subset of 1 to 2-month-old mice were euthanized and their pancreases excised before flash freezing. Tissue was crushed using a chilled stainless steel mortar and pestle then lysed in digestion buffer supplemented with 20 mg/ml Proteinase K (Invitrogen) using a Genomic DNA Isolation Kit (Norgen Biotek Corp). PCR amplification was performed on 250 ng genomic DNA using the LongAmp Taq DNA polymerase kit (New England Biolabs) and primers flanking the targeted Rest exon two (sense primer (P5): 5′-GAGCCGTTTCCTGTGATGGCATTC-3′; antisense primer (P4): 5′-CCAGGGTTCAGTTCTCTACACCCAC-3′). DNA was amplified for 30 cycles (94 °C for 30 s; 64 °C for 1 min; 65 °C for 2.5 min), and PCR products were electrophoresed on a 0.8% agarose gel at 80 V for imaging. Sanger Sequencing of PCR products were performed by Eton Biosciences using primers P4 and P5.

Western blot

Pancreas from 1 to 3-month-old mice were flash frozen and crushed using a cooled stainless steel mortar and pestle. Tissue was lysed in Cell Lytic MT Cell Lysis buffer (Sigma), 5 mM EDTA (Thermo), and Halt Protease and Phosphatase Inhibitor Cocktail (Thermo) then homogenized for 15 s. Samples were centrifuged for 15 min at 14,000 × g at 4 °C, and supernatant protein concentration was quantified using the BCA Assay (Thermo). Protein was prepared using Bolt LDS Sample Buffer (Novex) and Bolt Sample Reducing Buffer (Novex) then heat denatured at 70 °C for 10 min. Protein samples (15–44 μg) were separated on a 4–12% Bis-Tris Plus Gel (Invitrogen) (80 V for 20 min; 120 V for 60 min) then transferred to a PDVF membrane using Trans-Blot Turbo Transfer System (BioRad). The membrane was blocked in 5% milk solution for 1 h at room temperature and incubated in primary antibody overnight at 4 °C. Membranes were incubated in goat anti-rabbit-HRP or goat anti-rabbit-HRP antibody for one hour at room temperature and imaged using densitometry methods (Amersham ECL Detection (GE Healthcare); Amersham Imager 680 (GE Healthcare)). The following primary antibodies were used: anti-amylase (1:1000, Cell Signaling, 3796), anti-Snap25 (1 μg/ml, Abcam, ab41455, anti-REST (1:1000, Abcam, ab202962), anti-REST (1:500, Millipore, 07-579), anti-REST (1:1000, Abcam, ab21635), anti-REST (1:500, Hsieh514³¹), anti-REST (1:500, Aviva, ARP32086_P050), anti-REST (1:1000, Proteintech 2242–1-AP), anti-β-actin (1:1000, Cell Signaling, 4970S), and anti-GAPDH (1:1000, Cell Signaling, 5174S).

Quantitative PCR

RNA was harvested and isolated from mouse pancreas according to a previously established protocol³² using the miRNeasy Mini Kit (Qiagen). RNA samples had a RIN value of 7.0 or greater (Agilent 2100 Bioanalyzer). Reverse Transcription PCR was done using M-MLV Reverse Transcriptase (Thermo Fisher). Quantitative PCR was performed using EXPRESS SYBR GreenER (Invitrogen) or TaqMan (Applied Biosystems), and fluorescence measurements were collected using the Applied Biosystems, QuantStudio 7 Flex. Primer sequences are listed in Supplemental Table 1. In-house primers were designed using Primer Express (Applied Biosystems). Data were analyzed as gene expression using the 2^−∆CT method relative to 18S rRNA or as fold change in gene expression using the 2^−∆∆CT method. Statistical analysis was performed on either ∆Ct or fold change data.

Fasted blood glucose and glucose tolerance test

For determining glucose tolerance, 2 to 4-month-old male and female mice were transferred to clean cages for an overnight fast before blood glucose levels were determined using a glucometer³³. Mice were injected intraperitoneally with 1 mg glucose/g body weight, and blood glucose levels measured at 0, 30, 60, 90, and 120 min. Blood glucose clearance was quantified as area under the curve using the trapezoidal method.

Organoid culture for ADM

Acinar-to-ductal metaplasia from mouse acini was induced in vitro using modified Matrigel organoid-based culture techniques described previously³⁴. Briefly, 6–8-week-old mouse pancreas was minced to 1–3 mm sections and digested in 0.2 mg/ml Collagenase P (Roche) for 25 min at 37 °C. Clusters were filtered through a 500 µm cell strainer (pluriSelect) and resuspended in Waymouth media supplemented with 10% FBS, 1% Pen Strep, 100 µg/ml trypsin inhibitor (Sigma), and 20 µg/ml dexamethasone (Sigma). For REST KO studies, cells were plated in 50% Matrigel culture (Corning) on a 96-well plate at 100 µl/well. For virus infection studies, primary mouse acinar cells were infected with mouse control expressing adenovirus (abm, Cat. No. m009) or mouse Rest adenovirus (abm, Cat. No. 213793A). For adenovirus transduction, equal volumes of cell suspension were added to 5 µl of either control or Rest stock mouse adenovirus (10⁶ pfu/ml), incubated at 37 °C and 5% CO₂ for 1.5 h with gentle rocking every 10 min. Two parts of Matrigel (Corning) was added to one part of virus transduced cell suspension then plated in a 24-well collagen coated plate. The collagen coat was made using collagen I, rat tail (Thermo Fisher Scientific) at 2.6 mg/ml. For all organoid experiments, gelated Matrigel cultures were supplemented with prepared Waymouth media and refreshed every 1–2 days. Cultures were maintained at 37 °C in 5% CO₂. Microscope slide-grids (Sigma) beneath the plate were used to manually count number of clusters at designated time points. Ductal-like clusters and acinar clusters were defined as 3–10 co-adherent cells, and % ADM was calculated as (# ducts/# total clusters)/well. RNA was isolated at day 1 and day 4 of plating by lysing the Matrigel culture in cold Trizol LS (Ambion).

Acute pancreatitis model

Two to three-month-old mice were administered eight hourly intraperitoneal injections per day of saline control or 100 µg/kg caerulein (Bachem) over 2 consecutive days for 16 injections total. Mice were euthanized at day 2, 4, or 7 after injections, and the pancreas and blood were collected for analysis.

Pancreatic histology

Mice were euthanized by isoflurane followed by cervical dislocation. The pancreas was excised and immediately fixed in 10% neutral buffered formalin or 4% formaldehyde. Four-micrometer pancreatic tissue sections were stained with H&E, and histopathology was evaluated by a pathologist blinded to treatment and genotype. Percentage ADM was estimated as the ratio of tissue area containing ADM lesions relative to whole tissue area. Total necrosis score was the product of intensity score (0 = none, 1 = mild, 2 = moderate, 3 = severe) and area of necrosis (0 = none, 1 < 25%, 2 = 25–50%, 3 = 50–75%, 4 > 75%). Total inflammation score was the product of intensity score (0 = none, 1 = mild, 2 = moderate, 3 = severe) and area of inflammation (0 = none, 1 < 25%, 2 = 25–50%, 3 = 50–75%, 4 > 75%). Atrophy score and edema score were defined according to percent area (0 = none, 1 < 25%, 2 = 25–50%, 3 = 50–75%, 4 > 75%). Total histology score was the sum of total necrosis, total inflammation, and edema score.

Serum amylase levels

Mice were euthanized following isoflurane administration, and blood was collected by cardiac drainage using a needle-less 1-ml syringe to minimize hemolysis. Blood was incubated at room temperature for 25 min to allow coagulation. Samples were centrifuged at 1600 × g for 10 min at 4 °C, and the resulting serum was aliquoted and frozen. Aliquots were assayed by the Ohio State University Comparative Pathology and Mouse Phenotyping Shared Resource for amylase activity using the chromogenic method at 408/647 nm wavelength (Vet Axcel Chemistry Analyzer, Wasserman).

Pancreatic edema

The entire pancreas was excised, and the wet weight was recorded. Tissue was incubated at 85 °C for 24 h, and the dry weight was measured. Percent water content was calculated as ((wet weight − dry weight)/wet weight) × 100.

Transmission electron microscopy

Pancreas was excised and immediately preserved in Tyrode’s buffer containing 4% paraformaldehyde and 1% glutaraldehyde (pH 7.4), minced into 1-mm sections, and stored overnight at 4 °C. Samples were processed by the University of Florida College of Medicine Electron Microscopy Core Facility according to previously used protocol³⁵.

Immunofluorescence

Mouse pancreas was fixed in 10% neutral buffered formalin and processed to paraffin sections. Sections were deparaffinization using xylene and rehydrated to water using previously reported methods³⁶. Antigen retrieval was performed by heating for 20 min in EDTA buffer using a steamer then cooled to room temperature for 20 min. Samples were blocked in 10% goat serum (Vector) for 20 min. Tissue sections were incubated in rabbit anti-E-cadherin (1:300, GeneTex, GTX100443) overnight at 4 °C followed by detection using Alexa Fluor 555-conjugated goat anti-rabbit antibody for 20 min. For amylase and cytokeratin19 staining, sections were deparaffinized and antigen retrieval performed by citrate buffer (AR6) using a microwave for 90 s at full power followed by 15 min at 20% of full power and cool down 30 min. Sections were incubated with TRIS buffer containing 10% normal goat serum for 10 min. Tissue sections were incubated in rabbit anti-amylase (1:2000, Sigma A8273) for 1 h followed by Opal Polymer HRP Ms + Rb for 10 min then tertiary Opal 520 (1:500) for 10 min. Slides were blocked in 10% normal goat serum for 20 min and incubated in rabbit anti-cytokeratin19 (1:100, Abcam ab52625) for 1 h followed by Alexa Fluor 555 for 1 h. Slides were counterstained with DAPI (1 μg/ml, Fisher Scientific, 62248) or Hoechst S769121 (0.5 μg/ml, Abcam, ab138903) stain and imaged using a Zeiss Axioskop 2 Plus Fluorescence microscope.

Gene set enrichment analysis

Gene set enrichment analysis (GSEA) was performed using the GSEA software at the Broad Institute³⁷. The gene expression of eight normal human pancreas and 14 pancreatic ductal adenocarcinoma tissues (GSE71989) was used as the expression dataset. For the gene set database, we used 971 REST target genes³⁸, as well as a much smaller set of 24 REST target genes that were indicative of an aggressive subset of breast cancers³⁹. Another GSEA was performed using the gene sets to 572 transcription factors and transcription factor binding sites. For all analyses, the permutation type was set to phenotype and 1000 permutations were run.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 7 (San Diego, CA) and Sigma Plot 13.0 (Systat Software, Chicago, IL). No animals were excluded from the analysis as outliers. Two-sided p-values for two-sample comparisons were calculated using the Student’s t-test or Mann–Whitney rank sum test. For multiple comparisons, one-way or two-way ANOVA was used, and Holm-Sidak method was applied for post hoc multiple comparison adjustments. Data were transformed as needed to obtain normality and equal variance for experiments with n > 5. Expression values of Gene Expression Omnibus datasets were calculated by the anti-log2 of the signal intensity.

REST gene expression is increased during pancreatic injury and acinar de-differentiation

To investigate the role of REST in the exocrine pancreas, GSEA was performed on the gene expression profiles of normal human pancreas and pancreatic ductal adenocarcinoma tissues as described in the “Materials and methods” section. The REST target genes were highly enriched (i.e., upregulated in the normal pancreas with low REST expression) in both the large (Fig. 1a) and small REST target gene sets (Suppl Fig. 1A). In the GSEA that used 572 transcription factors and transcription factor binding sites as the gene set database, 57 of the 572 gene sets were upregulated in normal pancreas. While none of these 57 genes sets were significantly enriched (p < 0.05), the REST gene set was the top ranked (Suppl Table 2). These data suggest that REST activity is enhanced in human pancreatic ductal adenocarcinoma. Gene expression profiling from independent Gene Expression Omnibus (GEO) datasets^40,41 confirms that REST mRNA levels are elevated in human pancreatic tumor samples (Fig. 1b,c) and correspond with lower levels of REST target genes such as SNAP25, CHGA, and PDGB5 (Fig. 1c). Gene expression profiles from two independent datasets⁴² do not show significantly differential REST expression among stages of human PDAC (Suppl Fig. 1B,C), although the low number of available early and late human tumor samples makes analysis limited. Furthermore, survival of PDAC patients does not correlate with levels of REST expression (Suppl Fig. 1D).

The role of REST during the initiation of PDAC was then investigated. Gene expression profiling⁴³ of a PDAC mouse model shows that Rest expression increases prior to cancer development and is elevated in ADM lesions relative to normal acinar tissue (Fig. 1d). Similarly, available profiling data from mice with pancreatic Mst1/2 knockout (DKO)⁴⁴ that experience a loss of acinar mass and gain of ductal population also show an increase in REST expression compared with control (Fig. 1e). Pancreatitis is a known inducer of ADM in mice, so Rest gene expression was investigated from GEO datasets^45,46 of caerulein-induced acute pancreatitis studies. Rest expression was increased in response to acute administration of caerulein (Fig. 1f, g) and decreased to baseline upon recovery (Fig. 1g). Of note, one GEO dataset⁴⁷ of chronic mouse pancreatitis samples did not show increase of any Rest coding exons, but did show increased levels of the 3′UTR region (Suppl Fig. 1E). Overall, most independent datasets investigated suggest that Rest levels increased in response to injury and during ADM.

To better understand differences of Rest expression at the cellular level, the USCS Genome Browser was used to evaluate single-cell RNA-seq results from the Tabula Muris RNA-seq dataset^48–50. Isolated pancreatic ductal cells showed greater levels of Rest expression compared with acinar and endocrine (pancreatic polypeptide, alpha, delta) pancreas cells (Fig. 1h). In agreement to this finding, independent GEO datasets^51,52 that performed gene profiling of isolated mouse exocrine pancreas cells show greater Rest levels in adult ductal cells compared with embryonic acinar cells (Fig. 1i) or adult acinar cells (Fig. 1j). Deeper investigation of REST expression during mouse acinar maturation shows that adult acinar cells have less Rest expression than embryonic acini (Fig. 1k), and embryonic acini have lower Rest mRNA levels than embryonic progenitor pancreas cells (Fig. 1l), based on independent GEO datasets^43,51. These data indicate that Rest expression decreases during acinar exocrine maturation and is maintained at lower levels than in duct cells.

Development of a mouse model with conditional loss of REST in the exocrine pancreas

To investigate the role of REST during acinar homeostasis and response to injury, a novel transgenic mouse model was developed with Rest conditionally knocked out in the exocrine pancreas. Mice with homozygous intronic LoxP sites flanking exon two of Rest (REST^fl/fl)³⁰ were crossed with mice expressing the Cre Recombinase transgene under the pancreas-specific p48 promotor (p48-Cre^+/−) (Fig. 2a) to produce pancreas-specific REST knockout mice (Cre/REST^fl/fl).

Mouse tail snips were genotyped to confirm the presence of the Cre Recombinase transgene (primers P1 & P2) and the homozygous Rest LoxP mutant allele (primers P3 & P4) (Fig. 2b). Genomic DNA from Cre/REST^fl/fl mouse pancreas showed a high Cre-mediated excision efficiency by PCR analysis of the targeted Rest exon two region (primers P5 & P4) (Fig. 2c). Sanger Sequencing of the PCR amplified exon two region from REST KO pancreatic DNA shows the loss of exon two between intron 1–2 and intron 2–3 (Fig. 2d). As expected, the presence of the targeting vector with the LoxP sequence was observed at the excision site, indicating a successful genomic KO event. Gene expression of Rest was quantified to confirm knockout. In accordance with a previous report using the REST^fl/fl model¹¹, gene expression analysis showed loss of the targeted Rest exon two but expression of a truncated exon three and four Rest transcript (Fig. 2e). To evaluate loss of full-length REST protein in Cre/REST^fl/fl mice, western blot was performed on pancreatic lysate using five commercially available anti-REST antibodies and one custom-made antibody (Hsieh514³¹). While one antibody shows decrease in full-length REST protein (Fig. 2f), results among the other antibodies tested were variable (Suppl Fig. 2). Nevertheless, KO of Rest exon two was adequate to increase gene expression of known REST target genes such as Chga, Chgb, Snap25, and Pgbd5^16,39 in the pancreas and increase protein of Snap25 (Fig. 2g, h), suggesting successful abrogation of REST function as a transcriptional repressor.

REST knockout increases production of zymogen granules in pancreatic acini

Mice devoid of pancreatic REST had the same pancreatic weight normalized to body weight as control mice (Fig. 3a). Interestingly, the gross appearance of the Cre/REST^fl/fl mouse pancreas appeared whiter in color, distended, and fewer separations between lobules compared with control mice (Fig. 3b). Histologic observation of pancreatic acini showed increased eosinophilic staining of the cytoplasm from Cre/REST^fl/fl mice (Fig. 3c, d), suggesting an increase in intracellular protein content. Pancreas from REST KO mice displayed atypia and acinar cell swelling, observable by e-cadherin staining of acinar cell boundaries (Fig. 3d). Although Cre/REST^fl/fl mice do not appear to have greater total amylase levels (Fig. 3e), immunofluorescence staining suggests loss of REST disrupts the cytosolic distribution of amylase within the acinar cell (Fig. 3d, f). Of note, pancreatic endocrine function was euglycemic in REST KO mice, as evaluated by fasted blood glucose levels and glucose tolerance testing in comparison to control mice (Suppl Fig. 3A-C).

At the cellular level, brightfield microscopic images of REST KO acinar clusters from digested mouse pancreas showed aberrant dark granule localization from the apical to the basal surface of the acinar cluster (Fig. 3g). Transmission electron microscopy of Cre/REST^fl/fl and Cre control tissue revealed that REST KO pancreata had a significant increase in intracellular zymogen granules with most cells essentially filled with granules versus the apical accumulation in the Cre mice (Fig. 3h).

Loss of REST in pancreatic acini inhibits ductal transdifferentiation in vitro

To investigate the capability of acinar cells to transdifferentiate into ductal-like cells in the absence of REST, acini from Cre/REST^fl/fl mice were cultured as organoids and induced to undergo ADM by plating in Matrigel. Mouse acini with loss of REST produced fewer and smaller ductal morphologies relative to control pancreas over 4 days of culture (Fig. 4a, b). As ADM is a stepwise succession of acinar de-differentiation then progenitor transdifferentiation into ductal-like identity, gene expression was investigated to determine molecular differences in pancreatic identity. Acinar markers Amy2a, Cpa1, and Cpa2 decreased in both groups over the course of 4 days, but REST KO cells show less decrease in acinar marker Cpa2 compared with control cells (Fig. 4c). Congruously, REST KO cells showed a greater increase in progenitor marker Sox9 at day 4 and a lower increase in ductal marker Krt19 at day 4 compared with control cells (Fig. 4c). Model validation is supported by increased REST target gene expression (Pgbd5, Snap25, Chgb) in Cre/REST^fl/fl cells compared with control (Fig. 4c). Together, these data imply that REST is important for pancreatic acini to properly transdifferentiate into ductal cells in response to induced transdifferentiation.

The inverse relationship was observed when REST is overexpressed in wild-type mouse acini and ADM is induced. Adenoviral-REST was transfected into wild-type pancreatic acini that were subjected to organoid culturing over the course of 4 days. Acini with elevated levels of Rest produced more ducts per well compared with adenoviral-control (Fig. 4d). Furthermore, the seven-fold increase in Rest consequently doubled the expression of ductal identity marker Krt19 and dramatically decreased expression of acinar identity markers (Amy2a, Cpa2) and acinar transcription factors (Ptf1a, Nr5a2, Rbpjl, Xbp1, Gata4, Atf6) relative to control (Fig. 4e). These data suggest that REST promotes ductal-like transformation of acinar cells.

Loss of Rest in pancreatic acini increases caerulein-induced acute pancreatitis

Aberrant upregulation of zymogen granules during homeostatic conditions suggests that REST knockout mice would be more sensitive to autodigestion-related injury. Furthermore, if REST is required for acinar transdifferentiation into a refractory duct-like state as a protective mechanism, then loss of REST in acinar cells would lead to greater injury severity. To test the effects of REST knockout on acinar cells during tissue damage and recovery, mice were induced with acute pancreatitis using elevated doses of caerulein. Histopathology revealed REST KO mice had greater tissue injury (Fig. 5a). Microscopic scoring showed trending but not statistical increases in four tissue parameters scored at day 2 in REST KO mice (Suppl Fig. 4A-E). Sensitive biochemical methods were used to quantify the differences in pancreatitis severity at the day 2 of peak injury. Serum amylase levels were higher in REST knockout mice compared with controls (Fig. 5b). Furthermore, REST knockout mice had greater pancreatic edema in response to caerulein-induced injuries (Fig. 5c). Cre/REST^fl/fl had greater gene expression of clusterin (Fig. 5d), a molecular chaperone that protects against acinar cell stress⁵³. Acute caerulein treatment was sufficient to induce an increase in the cytokeratin19⁺ ductal population and decrease in the amylase⁺ acinar population (Fig. 5e). These data show REST knockout mice are more sensitive to caerulein-induced acini overstimulation and transdifferentiation, suggesting REST plays a protective role against pancreatic tissue damage.