CREB Drives Acinar to Ductal Cells Reprogramming and Promotes Pancreatic Cancer Progression in Preclinical Models of Alcoholic Pancreatitis

Abstract

Background & Aims

Chronic alcoholism often leads to pancreatitis, which exacerbates pancreatic damage through acinar cell injury and fibrotic inflammation activating AKT/mTOR/cyclic adenosine monophosphate response element binding protein 1 (CREB) signaling axis. However, the molecular interplay between oncogenic Kras^G12D/+(Kras∗) and CREB in promoting pancreatic cancer progression under chronic inflammation remains poorly understood.

Methods

Experimental alcoholic chronic pancreatitis (ACP) induction was established in multiple mouse models, with euthanasia during the recovery stage to evaluate tumor latency. CREB was selectively deleted (Creb^fl/fl) in Ptf1a^CreERTM/+;LSL-Kras^G12D/+(KC) genetic mouse models (KCC^-/-). Pancreata from Ptf1a^CreERTM/+, KC, and KCC^-/- mice were analyzed using histological profiling, Western blotting, phosphokinase array, and quantitative polymerase chain reaction. Single-cell RNA sequencing was performed in ACP-induced KC mice. Lineage tracing analysis using YFP reporter mice and acinar cell explant cultures analysis were also conducted.

Results

ACP induction in KC mice significantly impaired pancreas’ repair mechanism. Acinar cell-derived ductal lesions demonstrated sustained CREB hyperactivation in acinar-to-ductal metaplasia/pancreatic intraepithelial neoplasia lesions associated with pancreatitis and pancreatic cancer. Persistent CREB activity reprogrammed acinar cells, and increased profibrotic inflammation. Notably, acinar-specific Creb deletion in ACP-induced models suppressed high-grade pancreatic intraepithelial neoplasia development, restrained tumor progression, and improved acinar cell function.

Conclusions

Our findings demonstrate that CREB and Kras∗ promote irreversible acinar-to-ductal metaplasia, accelerating pancreatic cancer progression with ACP. Targeting CREB may present a promising strategy to mitigate inflammation-driven pancreatic tumorigenesis.

Graphical abstract

Summary.

Exogenous induction of alcoholic chronic pancreatitis accelerates pancreatic cancer progression by activating cAMP response element binding protein 1 (CREB), which drives irreversible acinar-to-ductal reprogramming in the pancreas of Ptf1a^CreERTM/+;LSL-Kras^G12D/+ (KC) genetically engineered mouse model. Pancreas-specific deletion of Creb in this model limits advanced pancreatic intraepithelial neoplasia formation, reduces disease severity, and preserves pancreas tissue architecture- highlighting CREB as a key molecular driver of pancreatic tumorigenesis.

Alcoholic pancreatitis, both acute and chronic, is a risk factor for pancreatic cancer.^1,2 Prolonged alcohol abuse accounts for 60% to 90% of chronic pancreatitis (CP) cases.^3,4 Continuous inflammation due to alcoholic CP (ACP) causes pancreatic atrophy and fibrosis within and around the pancreas.⁵

Chronic pancreatic inflammation frequently involves loss of acinar cell homeostasis and ductal phenotype acquisition, termed as acinar-to-ductal metaplasia (ADM). This adaptive physiological response to inflammation protects the pancreas from further damage.⁶ Nevertheless, scientific experimental models consistently demonstrate that in pancreatic tissue with oncogenic Kras^G12D/+ (Kras∗), inflammation conspired by nongenetic environmental factors including excess alcohol consumption, obesity, and smoking impedes pancreatic regeneration and expedites7, 8, 9 the development of neoplastic precursor lesions, including pancreatic intraepithelial neoplasia (PanIN), and may progress rapidly toward invasive carcinoma.^10,11

Previous studies using caerulein and ACP mouse models have linked epithelial cell-intrinsic signaling (mitogen-activated protein kinases [MAPKs], phosphoinositide 3-kinase [PI3K]/AKT, and mechanistic target of rapamycin [mTOR]) to inflammation-driven pancreatic disease pathogenesis.12, 13, 14 Building on our lab’s long standing efforts in understanding the multifaceted role of transcription factor signaling in pancreatitis and pancreatic cancer, we have previously identified hyperactivation of cyclic adenosine monophosphate response element-binding protein 1 (CREB) via granulocyte-macrophage colony stimulation factor (GM-CSF) as a key mediator of chronic inflammation that is induced by smoking and regulates aggressiveness of pancreatic cancer.¹⁵ In our recent study using a C57BL/6 mouse model of alcohol-induced chronic inflammation, screening of specific kinase signaling pathways identified CREB phosphorylation at Ser133 as one of the highest upregulated target along side PI3K/AKT.¹⁶ Therapeutic targeting of PI3K/AKT signaling node mitigates the severity and progression of CP, primarily by modulating acinar cell death¹⁷; however, the role of transcription factors including CREB in acinar-to-ductal reprogramming in the context of ACP has not been explored. Despite the well-established relationship between inflammation and cancer, the molecular mechanisms linking chronic inflammation with oncogenic drivers remain poorly understood within the context of pancreatic cancer initiation and progression. Because a majority of patients with CP suffer from alcoholism, this study aimed to develop pancreas-specific genetically engineered mouse models (GEMMs) to better investigate these mechanisms.

Herein, we demonstrate persistent CREB activation in acinar cells transitioning toward a ductal phenotype in Kras∗ (KC) using GEMMs post-ACP induction. Creb deletion in KC mice with ACP reduced ADM reprogramming, hindered tumor growth, and prolonged tumor latency. These findings establish CREB as a key oncogenic node in Kras-mediated pancreatic cancer progression with chronic inflammation and also underscores its potential as a promising therapeutic target for pancreatic cancer.

Results

Establishing ACP in Ptf1a^CreERTM/+ Mice

An ACP induction experimental mouse model was established in Ptf1a^CreERTM/+ mice, as described previously.¹⁶ The mice were sacrificed after 3 or 21 days (ACP recovery period) (Figure 1A). All experimental cohort mice exhibited weight gain post ACP induction (Figure 1B); however, stopping exogenous ACP induction restored the mouse’s body weight (identical to the control) within 21 days during the ACP recovery period. Serum levels revealed an increased blood alcohol concentration with alcohol (A) or ACP induction (Figure 1C). Ptf1a^CreERTM/+ mice with ACP induction demonstrated markedly reduced pancreatic weight (implying pancreatic injury) (Figure 1D). ACP withdrawal completely restored pancreatic weight to its normal level after 21 days (Figure 1E). Alcohol-treated mice showed elevated serum amylase levels, whereas the experimental mice cohort with ACP induction exhibited the most substantial decline in serum amylase levels compared with the control, A, and caerulein (CP)-exposed mice (Figure 1F), suggesting maximum acinar injury. Histological assessment of mouse pancreata harvested 3 days after ACP induction revealed profound injuries to the pancreatic tissue architecture, including acinar cell loss, duct-like structures (cytokeratin-19 [CK-19⁺] positivity), activated pancreatic stellate cells (PSCs) displaying α-smooth muscle actin (α-SMA) positivity, increased collagen deposition (Sirius Red) (Figure 1G), and leukocyte infiltration (CD45⁺ immune cells) compared with the control, A-, or CP-exposed mice (Figure 1G). Pancreatic injury in Ptf1a^CreERTM/+ mice completely reversed after discontinuing the exogenous ACP stimulus (21-day recovery period) (Figure 2A). No histologically detectable PanINs occurred (Figure 2A and Figure 1G) in any mouse cohorts. Quantitative PCR (qPCR)-based transcriptional profiling of ACP-treated mouse pancreata demonstrated marked downregulation of genes implicated in acinar cell regulation and function (Figure 2B). Overall, these results communicate the transient and reversible nature of pancreatic acinar injury after ACP induction in Ptf1a^CreERTM/+ mice.

Figure 1.

Comparative analysis of severe damage mediated by alcoholic chronic pancreatitis (ACP) as compared to the impact of alcohol (A) or caerulein (CP) alone in Ptf1a^CreERTM/+ mice. (A) ACP induction in Ptf1a^CreERTM/+ mice exposed to ethanol (alcohol)-enriched liquid diet (A) and caerulein administration (CP). Mice were euthanized after 3 and 21 days of ACP recovery periods. (B) Dot plot depicting initial and final body weight measurements in mice treated with vehicle, (A), (CP), or ACP with 3- and 21-day recovery (n = 7 mice per group). (C) Blood alcohol concentration levels (mg/dL) shown in control (ctrl), A, and ACP-induced Ptf1a^CreERTM/+ mice (n = 7–9 mice per group). (D and E) Relative pancreas weight measurements (pancreas weight [gm]/body weight [gm] × 1000) of Ptf1a^CreERTM/+ mice in ctrl, A, CP, and ACP-induced groups with 3- and 21-day recovery (n = 5–7 mice per group). (F) Measurements of serum amylase activity in ctrl, A, CP, and ACP-induced mice (3-day recovery) (n = 6 mice per group). (G) Representative images depicting H&E with histological quantification of CK19⁺ ducts, PanINs (Alcian Blue), αSMA, collagen (Sirius red), and immune cells (CD45⁺) in pancreata of ctrl, A, CP and ACP-induced Ptf1a^CreERTM/+ mice (3-day recovery) (n = 4 mice per group). Scale bar, 50 μm. ns, nonsignificant; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001 by ANOVA.

Figure 2.

CREB activation in alcoholic chronic pancreatitis (ACP) in Ptf1a^CreERTM/+ mice. (A) Ptf1a^CreERTM/+ mice pancreas with H&E and quantification depicting ducts (CK19⁺), PanINs (Alcian Blue), collagen (Sirius red), and immune cells (CD45⁺) with ctrl and ACP induction (n = 5 mice per group). (B) qPCR analysis in pancreas tissue harvested from ctrl and ACP-induced Ptf1a^CreERTM/+ mice (n = 3 mice per group). (C) H&E-based quantification of mice pancreata highlighting acinar and ADM regions (n = 3 mice per group). (D) Representative images of pancreas depicting amylase (green)/CK19 (red) co-IF labeling and CK19⁺ amylase⁺ cell corresponding quantification (n = 3 mice per group). (E and F) Mouse kinase array in control (ctrl) and ACP-induced pancreata performed by using pooled tissue lysates from (n = 3) biological replicates for each group (Ctrl and ACP), with 2 membranes used for quantitative estimation of FC differences. (G) Western blotting showing pCREB expression with relative fold change expression values (normalized to total CREB) in pancreatic tissue lysates in ctrl, 3-, and 21-day ACP recovery period (n = 2 mice per group). (H) H&E and pCREB (green)/CK19 (red) co-I.F labeling with pCREB⁺ duct quantification in pancreata harvested from ctrl and ACP with recovery (n = 6 mice per group). Scale bar, 50 μm. ns, nonsignificant; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001 by ANOVA or unpaired t-test for 2-groups comparison.

Activation of CREB Is a Hallmark of ADM in Response to ACP

Hematoxylin and eosin (H&E) (Figure 2C) and co-immunofluorescence (IF) (Figure 2D) staining for CK19 and amylase in ACP-treated Ptf1a^CreERTM/+ mice revealed transient duct-like cells within 3 days of ACP induction. By the end of the 21-day recovery period, these cells redifferentiated and repopulated the acinar compartment, demonstrating the reversibility of ADMs.

We conducted phosphokinase array profiling on Ptf1a^CreERTM/+ mice pancreata post-ACP induction (Figure 2E). Our analysis revealed mTOR and CREB as the 2 highest upregulated signaling targets, compared with the control pancreatic lysates (Figure 2F). This study explores CREB’s role in acinar to ductal reprogramming with Kras∗. Further validation using immunoblotting (Figure 2G) and co-IF staining (Figure 2H) analysis uncovered elevated phosphorylated CREB (pCREB) expression in duct-like structures 3 days post-ACP recovery, compared with the controls. By day 21 of ACP recovery, pCREB expression returned to baseline levels. These findings suggest that ACP induces pancreatic injury, fosters a fibroinflammatory milieu, and enhances CREB activation. Notably, these changes reversed during the recovery period, reducing CREB activation to baseline.

ACP Induction Accelerates ADM Reprogramming to PanIN and Pancreatic Cancer in Kras^G12D Mutant Mice

To enhance the evidence of ACP-mediated acinar cell reprogramming toward the ductal phenotype, we used KC GEMM harboring Kras∗ and established ACP induction (described in Figure 1A). Mice receiving an alcohol-based diet showed higher blood alcohol concentrations than control mice (Figure 3A). Inducing ACP in KC mice considerably increased relative pancreatic weight (Figure 3B), without significant sex-specific differences (Figure 3C).

Figure 3.

ACP induction accelerates ADM reprogramming toward PanIN and pancreatic cancer in Kras^G12D/+ mutant mice. (A) Blood alcohol concentration measurement (mg/dL) in ctrl, A, and ACP-induced KC mice (n = 6–7 mice per group). (B) Relative pancreas weight measurements in respective KC mice cohorts (n = 5–8 mice per group). (C) Dot plot depicting comparing relative pancreatic weights between male and female KC mice with ACP (n = 3–5 mice per group). (D) Representative pancreatic images depicting H&E with quantification of CK19⁺ ducts, PanINs, collagen, and CD45⁺ in ctrl and ACP-induced KC mice with 3- and 21-day recovery periods (n = 5 mice per group). (E) H&E staining within the pancreas depicting acinar, ADMs, PanINs, and cancer regions (n = 3 mice per group). (F) Representative pancreas images displaying H&E, images, and quantification of CK19⁺ ducts, PanINs (Alcian Blue), collagen (Sirius red), αSMA, and immune cell positivity (CD45⁺) in the pancreata of ctrl, A, CP, and ACP-induced KC experimental cohorts (3-day recovery) (n = 4–7 mice per group). (G) Comparative histological evaluation of mouse pancreas using H&E-based analysis (n = 3 mice per group). (H) Representative images illustrating pCREB expression in the pancreata of KC mice, accompanied by corresponding quantification across all experimental mice cohort (n = 3–4 mice per group). Scale bar, 50 μm. ns, nonsignificant; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001 by ANOVA and unpaired t-test for 2-group comparison

The pancreas in control KC mice exhibited well-structured acinar compartments and low-grade PanIN lesions (Figure 3D and E). In contrast, ACP-induced KC mice demonstrated a substantially higher frequency of mucinous ductal lesions (low-grade PanINs) with fewer acinar cells (after a 3-day recovery period) than A- or CP-induced KC mouse pancreata (Figure 3F and G). After ACP induction and 21-day recovery, pancreatic tissue histological analysis revealed that these mucinous PanIN progressed towards high-grade lesions (Figure 3D and E), and a significant increase in CK-19⁺ ductal positivity, mucin content, increased collagen deposition, and infiltration of CD45⁺ immune cells was noted (Figure 3D). This increase along with activated PSCs were higher in the ACP experimental cohort of KC mice pancreata than in A- or CP exposed mice (Figure 3F). A substantial increase in the deposition of extracellular matrix and an increased in immune cells infiltration accompanied this progression (Figure 3F). Chronic inflammatory stimuli induced by ACP resulted in significantly elevated pCREB expression compared with control or alcohol alone, suggesting a potential association with increased pancreatic damage and heightened CREB (Figure 3H). Interestingly, chronic inflammatory insult due to CP alone also led to increased CREB activation compared with control KC mice (Figure 3H), although pCREB expression appeared to be higher in ACP induced mice compared with CP. Overall, these findings suggest that the addition of alcohol to cerulein-induced CP amplifies the fibroinflammatory milieu and enhanced CREB activation to accelerate tumor progression in KC mice pancreata.

Single-cell RNA-seq (scRNA-seq) was performed on live single-cell suspensions of KC mice pancreata, either control or ACP. Overall, 7903 and 7797 cells from the control and ACP groups were analyzed, respectively. Uniform manifold approximation and projection (UMAP) cell clusters were annotated by examining classic cell type markers.^18,19 We distinguished 9 distinct clusters representing different cell types (Figure 4A and B). ACP-induced KC mice exhibited an increased abundance of cell types, including ductal cells, macrophages, neutrophils, regulatory T cells, and fibroblasts, compared with the control mice (Figure 4C). Transcriptional analysis of the acinar cell compartment of KC pancreata showed upregulation of mRNA transcripts related to inflammation-induced stress, growth, and proliferation with ACP induction compared with the control mice (Figure 4D). Gene Ontology Biological Process (GOBP) analysis revealed that genes in the acinar subcluster exhibited differential expression related to cellular stress, apoptosis, alcohol metabolism response, and epithelial cell differentiation, consistent with ACP induction in KC GEMM mice (Figure 4E). The uninjured control KC pancreata showed more amylase⁺ acinar cells with CK19⁺ expression associated with the ductal structures, as revealed using co-IF analysis (Figure 4F). In ACP-treated KC mice, the acinar compartment progressively transformed and was replaced by a substantial proportion of ductal structures of acinar origin (dual CK19⁺amylase⁺ cells) after 3 days of recovery. Pancreatic tissue analysis at the end of the 21-day recovery period revealed a ductal phenotype with the highest proportion of CK19⁺ cells (Figure 4F). Notably, in ACP-induced KC mice, pancreata amylase distribution was predominantly localized to the basolateral membrane, contrary to the usual apical distribution, which is in line with previous reports suggesting pancreatic enzymes are released into interstitial space via this pathway.²⁰

Figure 4.

ACP induction accelerates pancreatic tumor progression and reduces tumor latency period in Kras^G12D/+ mutant mice. (A and B) UMAP projection displaying cell clusters along with bubble plot outlining the expression of canonical lineage cell cluster annotations in KC scRNA seq dataset. (C) Cell count table depicting changes in the number of cells across different cellular clusters in the pancreata of control and ACP-KC mice. (D) Volcano plot depicting differentially expressed genes (ACP-KC vs control) within the acinar cell compartment, based on scRNA-seq analysis of pancreata. (E) Bubble plot showing differentially enriched pathways (ACP-KC vs control) within the acinar cell compartment in scRNA seq dataset. (F) Representative pancreas images depicting amylase (green)/CK19 (red) co-IF (top) CK19⁺ amylase⁺ cell (bottom) quantification (within the epithelial cell compartment only) in ctrl and ACP-induced KC mice with 3- and 21-day recovery periods (n = 3 mice per group). (G) qPCR analysis in the pancreas of ctrl and ACP-induced KC mice (n = 3 mice per group). (H) Measurement of tumor latency period in ctrl, A, CP, and ACP-induced KC mice (n = 12–13 mice per group). Scale bar, 50 μm. ns, nonsignificant; ∗P < .05; ∗∗P < .01; ∗∗∗∗P < .0001 by ANOVA and unpaired t-test for 2-group comparison.

qPCR validation of transcriptional alterations in ACP-induced KC mice pancreata confirmed the reprogramming of acinar cells toward ductal metaplasia (Figure 4G). Advanced-grade histological lesions progressing to pathologically evident cancer in KC mice pancreata with ACP induction correlated with a significantly shortened tumor latency period (mean, 5.80 months) (Figure 4H) compared with control (mean, 9.76 months), A (mean, 8.67 months), or CP (mean, 7.56 months) -exposed KC mice, confirming a higher tumor burden induced by ACP than damage from A or CP induction alone.

CREB Is Persistently Activated During PanIN Progression in Kras^G12D Mutant Mice

Histological assessment using co-IF analysis of pancreatic tissues harvested after ACP induction in KC mice substantially increased pCREB-positive staining within CK19⁺ ducts during the recovery periods (3 and 21 days) compared with the control pancreata (Figure 5A). Western blotting (Figure 5B) confirmed heightened and sustained CREB activation in ACP-induced KC mice at 3, 21, and 42 days post-ACP recovery, suggesting a positive feed-forward loop in Kras∗ in driving CREB expression.

Figure 5.

Sustained hyperactivation of CREB in ACP Kras^G12D/+ driven ADM and PanIN progression. (A) H&E and co-IF labeling of pCREB (green) and CK-19 (red) expression (top) with pCREB⁺ duct quantification within the ctrl and KC mice pancreata at 3- and 21-day ACP recovery period (bottom) (n = 4 mice per group). (B) Western blotting of pCREB in pancreatic tissue lysates of control and KC mice at 3-, 21- and 42-day ACP recovery period (n = 3 mice per group). (C) H&E images of the pancreas along with co-IF labeling of pCREB (red)/CK19 (blue)/YFP (green), and YFP⁺ pCREB⁺ duct quantification in ctrl and Ptf1a^CreERTM/+; LSL-Kras^G12D/+; R26R-EYFP (KCY) mice at 3- and 21-day recovery periods (n = 4 mice per group). (D) Schematic depicting sustained hyperactivation of CREB driving acinar cell neoplastic reprogramming in KC mice upon ACP induction. Scale bar, 50 μm.∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001 by ANOVA for 3-groups comparison and unpaired t-test for 2 groups.

We combined the acinar-targeted KC GEMM with R26R-EYFP to generate the Ptf1a-^CreERTM/+;LSL-Kras^G12D/+;R26R-EYFP (KCY) mouse model for lineage tracing analysis (Figure 5C). The KCY mice underwent ACP induction (described in Figure 1A). Tamoxifen administration to KCY mice achieved approximately 95% recombination efficiency in acinar cells for 24 hours.²¹ In control KCY mice, YFP expression was mainly localized to acinar cells and excluded from CK19⁺ ducts. ACP-induced pancreatic injury in KCY mice accelerated PanIN progression (similar to KC-ACP), with many ductal structures originating from acinar cells (YFP⁺CK19⁺ducts) (Figure 5C). We observed sustained CREB activation within acinar-derived YFP/CK-19 dual-positive ductal lesions at 3- and 21-day recovery periods post-ACP induction. These results underscore the role of CREB in reprogramming acinar cells into preneoplastic precursors with ACP and Kras∗ (Figure 5D).

Acinar-specific Ablation of CREB Attenuates Spontaneous Kras^G12D-mediated PanIN Progression

We generated acinar-specific conditional Creb knockout mice with Kras∗ (Figure 6A) (Ptf1a^CreERTM/+; LSL-Kras^G12D/+; Creb^fl/fl or KCC^-/-), whereas KC mice with wild-type Creb were called as KC. The pancreata were collected from these 2 mouse cohorts at 5 and 10 months of age. The relative pancreatic weight of KC mice with Creb deletion (KCC^-/-) was markedly reduced compared with that of wild-type KC mice (Figure 6B and C). Histological examination showed that the pancreata of 5-month-old KC mice displayed a higher prevalence of ADM and low-grade PanIN lesions (Figure 6D and E), whereas that of 5-month-old KCC^-/- mice showed no precursor lesions and demonstrated a normal architecture. Pancreata harvested from 10-month-old KCC^-/- mice displayed infrequent ADM or PanIN1 lesions. Conversely, 10-month-old KC mice pancreata exhibited marked high-grade PanIN lesions and cancer (Figure 6D and E).

Figure 6.

Acinar-specific ablation of Creb attenuates spontaneous Kras^G12D/+ mediated PanIN progression. (A) Breeding strategy for generating acinar-specific Creb deficient Kras∗ mutant mice (Ptf1a^CreERTM/+; LSL-Kras^G12D/+; Creb^fl/fl or KCC^-/-). (B) Comparative measurement of relative pancreas weight in 10-month-old Creb wild type (KC) and KCC^-/- mice (n = 4–5 mice per group). (C) Representative photomicrographs of whole pancreas depicting significantly less tumor burden in KCC^-/- mice than KC mice at 10 months of age. (D) Representative H&E images of the pancreata harvested from 5- and 10-month-old KC and KCC^-/- mice. Scale bar, 10 and 50 μm. (E) Comparative assessment using H&E-based histology to examine the entire pancreas, illustrating acinar cells, ADMs, PanINs, and cancerous regions in 5- and 10-month-old KC and KCC^-/- mice (n = 3 mice per group). (F) Co-IF analysis of pCREB (green), CK-19 (red), and DAPI (blue) with quantification in pancreatic tissue sections harvested from 10-month-old KC and KCC^-/- mice (n = 4 mice per group). Scale bar, 50 μm. (G) Representative bright-field photomicrographs of primary 3D acinar cell cultures established from 10-month-old KC and KCC^-/- mice (scale bar, 20 μm) (n = 5 mice per group). (H) qPCR of pancreatic tissue harvested from KC and KCC^-/- mice (n = 3 mice per group). Scale bar, 10 and 50 μm. ns, nonsignificant; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001; 1-way ANOVA or unpaired t-test for 2-groups comparison.

Co-IF analysis in KC and KCC^-/- mice (10 months old) confirmed a significant reduction in pCREB expression in the pancreas ductal regions in KCC^-/- mice (Figure 6F). Pancreatic acinar cells were isolated from KC and KCC^-/- mouse pancreata for explant culture. Acinar-specific Creb ablation significantly blocked ductal-like structure formation compared with acinar cells with intact CREB (Figure 6G). qPCR analysis of 10-month-old KCC^-/- mice pancreata revealed reduced ductal cell phenotype genes, with a concomitant upregulation in genes linked to acinar cell regulation, function, proliferation, and differentiation compared with wild-type KC (Figure 6H). Taken together, these results suggest that CREB ablation reduces the ability of acinar cells to undergo ADM reprogramming.

Acinar-specific Ablation of Creb Attenuates Kras∗-mediated PanINs to Pancreatic Cancer Progression With ACP

In Ptf1a^CreERTM/+ mice with Creb deletion (CC^-/-) without Kras∗ (Figure 7A), histological assessment confirmed the loss of CREB expression within Ptf1a^CreERTM/+;Creb^fl/fl (CC^-/-) mice pancreata compared with wild-type Ptf1a^CreERTM/+ (C) mice (Figure 7B). CC^-/- mice exhibited enhanced acinar cell regenerative potential post-ACP induction (Figure 7C). The improved recovery in Creb-deleted mice was associated with reduced CK19⁺ ducts and an attenuated fibroinflammatory milieu.

Figure 7.

Histological profiling of Ptf1a^CreERTM/+;Creb^fl/fl (CC^-/-) mice pancreata with ACP induction. (A) Mouse breeding strategy to generate a genetic knockout of acinar cell-specific Creb, under a Ptf1a^CreERTM/+ promoter. (B) Representative images of the mouse pancreas with H&E staining, along with co-IF of total CREB expression (tCREB, red) and DAPI (blue), in control Ptf1a^CreERTM/+ (C) and CC^-/- mice. IF-based histological assessment, confirmed loss of CREB expression within the pancreata of Ptf1a^CreERTM/+;Creb^fl/fl (CC^-/-) as compared with wild type Ptf1a^CreERTM/+. (C) Comparative histological evaluation of mouse pancreas, accompanied by representative photomicrographs showcasing H&E, CK19⁺ ducts, Sirius Red, αSMA, and CD45⁺ staining within the pancreata of control (C) and CC^-/- mice, in ctrl or with ACP induction (3-day recovery) (n = 4 mice per group). Scale bar, 50μm. ns, nonsignificant; ∗∗∗∗P < .0001 by ANOVA.

To assess acinar-specific Creb ablation’s capacity to counteract Kras∗-driven ADM/PanIN originating from the acini, we next compared KC and KCC^-/- mouse pancreata (Figure 8A). KC and KCC^-/- mice maintained on a standard liquid diet were used as controls. ACP induction in KC mice considerably increased the relative pancreatic weight (21-day recovery period), confirming a higher tumor burden than that in the control KC mice (Figure 8A). This was less prominent in KCC^-/- mice with ACP induction, where CREB was deleted (Figure 8A and B), suggesting an association between attenuated tumor burden and CREB loss.

Figure 8.

Acinar-specific ablation of Creb attenuates Kras∗ induced progression of ADM/PanINs toward pancreatic cancer with ACP. (A) Relative pancreas weight of KC and Ptf1a^CreERTM/+;LSL-Kras^G12D/+; Creb^fl/fl (KCC^-/-) mice with ACP induction. (B) Representative photomicrographs of whole pancreas depicting significantly less tumor burden in KCC^-/- as compared with KC mice with ACP induction. (C) H&E-based histological examination illustrating acinar cells, ADMs, PanINs, and cancerous regions in ctrl or within KC and KCC^-/- mice pancreata harvested after 21 days of ACP recovery period (n = 3 mice per group). (D) Comparative histological evaluation of mouse pancreas, accompanied by images showcasing H&E, CK-19, Alcian Blue, Sirius Red, and CD45⁺ staining within control (KC and KCC^-/-) mice and at 3- and 21-day ACP recovery period (n = 5 mice per group). (E) Pancreas images depicting amylase (green)/CK19 (red) and DAPI (blue) immunofluorescent labeling (left) and CK19⁺ amylase⁺ cell corresponding quantification (right) (within the epithelial cell compartment only) in ctrl (KC and KCC^-/-) or with ACP induction (n=3 mice per group). Scale bar, 50μm. ns, nonsignificant; ∗∗∗ P < .001; ∗∗∗∗ P < .0001 by ANOVA.

Creb deletion in KCC^-/- mice pancreata remarkably protected against ACP-triggered ADM and PanIN progression as evidenced by H&E (Figure 8C), CK19, and Alcian blue staining (Figure 8D). Post-ACP induction, KC mice pancreata displayed a higher prevalence of high-grade PanIN lesions along with pancreatic cancer. Conversely, pancreatic tissue architecture in KCC^-/- mice displayed mostly normal organization of acinar cells, with fewer ADMs and low-grade PanIN lesions (Figure 8C and D). A notable reduction in the fibroinflammatory milieu (Sirius red and CD45⁺ cells) was observed (3- and 21-day ACP recovery periods) compared with that in KC GEMM (Figure 8D). Similar beneficial histological trends occurred in control KCC^-/- mice. Co-I.F staining within KC and KCC^-/- mice pancreatic tissues showed a notably heightened proportion of amylase-positive acinar cells in Creb-depleted KCC^-/- mice (Figure 8E). Conversely, KC mice pancreata exhibited substantial damage resulting from ACP induction, with high CK-19⁺ cell abundance and loss of amylase⁺ acini (Figure 8E).

Acinar-specific Ablation of CREB Attenuates ACP-induced Acinar Cells to Ductal Reprogramming and Increases Tumor Latency Period

We studied ACP-induced pancreatic injury in a lineage tracing mouse model of KC (KCY; Ptf1a-^CreERTM/+;LSL-Kras^G12D/+;R26R-EYFP) and KCYC^-/- (Creb deletion in KCY) mice (Figure 9A). Co-IF staining displayed more YFP⁺/CK19⁺ ductal lesions concurrent with the loss of the acinar cell compartment upon ACP induction in KCY mice. We found a substantial reduction in ductal structures, with minimal cells exhibiting dual positivity of YFP⁺/CK19⁺ during ACP-induced KCYC^-/-. qPCR analysis confirmed a notable decrease in the mRNA expression levels of genes associated with ductal cell identity and function (Sox9, Krt19, Aldh1a3, and Tspan8) in ACP-treated KCC^-/- mice pancreata compared with KC mice (Figure 9B).

Figure 9.

Acinar-specific ablation of Creb attenuates ADM reprogramming and increases tumor latency period with ACP induction in KC mice. (A) Representative H&E images of the pancreas along with co-IF labeling and YFP quantification (green)/CK19 (blue) in Ptf1a-^CreERTM/+;LSL-Kras^G12D/+;R26R-EYFP (KCY) and Creb^fl/fl (KCYC^-/-) in control or with ACP induction (n = 4 mice per group). (B) qPCR-based analysis of KC and KCC^-/- mice pancreatic tissue with ACP induction (n = 3 mice per group). (C) Tumor latency period assessment in ctrl or with ACP induction in KC and KCC^-/- mice (n = 12 mice per group, ctrl mice cohort represented here is similar to that shown in Figure 4H). (D) Schematic representation illustrating that acinar-specific Creb ablation diminishes pancreatic cancer progression in ACP induction in KC mice. Scale bar, 50 μm. ns, nonsignificant; ∗P < .05; ∗∗∗∗P < .0001 by ANOVA or unpaired t-test for 2-group comparison.

Creb-deficient GEMMs (KCC^-/- and KCYC^-/-) displayed low-grade PanIN lesions and a diminished pancreatic ductal phenotype, further reflected in tumor latency (Figure 9C). ACP-induced KCC^-/- mice pancreata had a mean pathological malignancy duration of 14.48 months compared with 5.80 months for KC-ACP mice. Conversely, control KC harboring wild-type Creb and KCC^-/- exhibited greater tumor latency (9.76 vs 19 months) (Figure 9C). Overall, these results suggest that acinar-specific Creb ablation reduces ACP-induced acinar-to-ductal reprogramming and Kras∗-induced neoplastic progression, delaying tumor burden (Figure 9D).

Discussion

A consensus exists that alcohol-dependent mouse models replicate the clinical presentation of human CP.^22,23 Chronic alcohol exposure induces membrane instability within the zymogen and lysosomal compartments of acinar cells, increasing the risk of premature enzyme activation and autodigestive injury.^24,25 When alcohol-exposed mice are subjected to additional stressors such as supraphysiological doses of cerulein, this priming effect leads to an additive and exacerbated pancreatic injury, profoundly damaging the pancreas and affecting other cellular constituents as well (including PSCs and ductal cells).²⁶ Based on these prior findings in animal models, irrespective of mouse sex,²⁷ we demonstrate that GEMMs of pancreatic disease (with or without Kras∗) recapitulate the clinical progression of chronic pancreatic injury through a combination of ethanol-based diet and cerulein administration, mimicking the pathological features of human ACP.^16,28, 29, 30, 31, 32, 33

Pancreatic cancer or pancreatic ductal adenocarcinoma (PDAC), is characterized by widespread genomic alterations and extensive tumor cell heterogeneity.34, 35, 36 Previous studies in mouse models have highlighted that both acinar and ductal cells can result in PDAC, with acinar cells more amenable to cellular plasticity and transform into a progenitor-like cell type with ductal characteristics (ADMs) with Kras∗.^6,8,11,37 In our current study, utilizing multiple GEMMs and lineage tracing studies, we demonstrated that under chronic inflammatory stimulation by ACP leads to sustained activation of CREB in ADMs undergoing ductal transition, which cooperates with Kras∗ to accelerate pancreatic cancer progression (Figure 10).

Figure 10.

(A) Schematic representation of ACP induction with Kras∗ driving irreversible ADM reprogramming toward accelerated PDAC progression along with sustained CREB activation and increased fibroinflammation. (B) Acinar cell-specific deletion of Creb generates anti-tumor responses and delays pancreatic cancer progression. The schematic was created with BioRender with License agreement numbers OW27SH91EN.

Crosstalk between epithelial cells and inflammation-driven stress response mediators promotes the aggressive biology of pancreatic cancer. We have previously discovered that GM-CSF mediated activation of CREB contributes to PDAC growth and therapeutic resistance within the context of smoking-associated chronic inflammation.¹⁵ Simultaneously, other studies have also reported that PDAC epithelial cells sustain their proliferation through obesity-associated stress responses, activating CREB at Ser133 via protein kinase-dependent signaling.^15,38 Thus, it is plausible that the additive stimulus of alcohol and cerulein exacerbates epithelial cell intrinsic and extrinsic stress response mediators, converging on CREB, as a central hub of signal integration to accelerate pancreatic cancer progression in GEMM models. This is supported by our scRNA-seq analysis of KC-ACP GEMM, which revealed acinar cell populations enriched in pathways prominently related to cellular responses to stress and reactive oxygen species (ROS). Simultaneously, we observed a profound accumulation of innate immunosuppressive myeloid subsets—macrophages and granulocytes—within the pancreata of ACP-exposed KC mice. Our ongoing research is focused on elucidating these complexities of epithelial-myeloid crosstalk, and specifically how it drives CREB activation not only in ductal cells but also across diverse cellular compartments through autocrine and paracrine signaling within the tumor microenvironment.

Recent studies from our group and others have firmly established CREB as a transcription activator and a key cooperative node between KRAS∗-mediated effector pathways and other genomic drivers, including TP53, in pancreatic cancer.^39,40 Upregulation of CREB expression has been associated with increased proliferation and migration of pancreatic cancer cells, evasion of apoptosis, and induction of epithelial to mesenchymal transition (EMT).41, 42, 43 Pharmacological targeting of CREB activation using small molecule inhibitor has been shown to dampen PDAC progression and metastatic spread.^15,40,44 However, pancreas-specific CREB deleted GEMMs were lacking, which limited our understanding of its oncogenic potential. Therefore, building on a growing body of evidence implicating CREB and its downstream signaling in pancreatic cancer progression, for the first time, our study extensively characterized CREB-deleted GEMMs with Kras∗. Acinar-specific CREB loss significantly attenuated both spontaneous and ACP-mediated ADM/PanIN formation while simultaneously promoting acinar cells regeneration in KC GEMM. Supporting this notion, CREB deletion further enhanced the transcriptional rewiring of gene hubs that promote acinar cell regenerative potential and function.

Pancreatic cancer cell heterogeneity is mainly governed by exocrine compartment plasticity. Acinar cells transition to a duct-like identity in response to exogenous stress, including ACP, making them susceptible to Kras∗-induced precancerous lesions. Given that multiple upstream signaling pathways can activate CREB, the current study focuses only on defining CREB’s role in the ductal epithelial compartment, both in the presence and absence of ACP. However, comparing the differential regulation of CREB expression in alcohol or CP settings alone with Kras∗ remains an important area for future investigation.

Several upstream kinases phosphorylate CREB at Ser133, which triggers the recruitment of the coactivator CBP.⁴⁵ This complex then binds to target gene promoters’ cAMP-response element (CRE) sequences and facilitates gene regulation via intrinsic and associated acetylase activities and/or interacting with core transcriptional machinery.⁴⁶ Previous studies have outlined the cancer cell-autonomous and extrinsic effect of CREB activation promoting PDAC aggressiveness and neutrophil-mediated immunosuppression by engaging with downstream effectors, including FOXA1, β-catenin, and CXCL1.^39,40 Our ongoing studies are focused on uncovering novel downstream targets by which CREB regulates ADM-PanIN progression with chronic inflammation and Kras∗. In future studies, we aim to evaluate whether genetic or pharmacological inhibition of these targets can phenocopy CREB loss and suppress pancreatic cancer progression.

In conclusion, ACP induction promotes pancreatic cancer progression via irreversible ADM reprogramming with Kras∗ cooperativity. These ductal lesions originating from acinar cells display continuous CREB hyperactivation. Attenuating high-grade PanIN formation and tumor progression through acinar-specific loss of Creb provide foundational insights for future studies on CREB’s role in pancreatic cancer progression in alcoholic pancreatitis with Kras∗.

Materials and Methods

Creb Deletion in a Genetically Engineered Ptf1a^CreERTM/+;LSL-Kras^G12D/+ (KC) Mice Model

Ptf1a^CreERTM/+ knock-in allele mice were obtained from Jackson Laboratory (stock number: 019378)²² and crossed with LSL-Kras^G12D/+ to generate acinar-specific Kras^G12D/+ mutant mice, referred to as Ptf1a^CreERTM/+;LSL-Kras^G12D/+ (KC) mice. To generate acinar-specific Creb1 (Creb) knockout mice, Ptf1a^CreERTM/+; Creb^fl/fl mice were crossed with LSL-Kras^G12D/+; Creb^fl/fl mice to generate Ptf1a^CreERTM/+;LSL-Kras^G12D/+; Creb^fl/fl (KCC^-/-) mice. The Creb^fl/fl mice were obtained from Professor Eric Nestler (Cold Spring Harbor Laboratory).⁴⁷ Creb was selectively excised from acinar cells in LSL-Kras^G12D/+ mice using inducible Cre recombinase, expressed from the pancreas-specific Ptf1a^CreERTM/+ promoter. Cre recombinase activity was induced in 6-week-old mice through daily tamoxifen administration (Sigma-Aldrich, cat. # T5648) for 6 consecutive days at a dose of 0.15 mg/g body weight, followed by a 1-week rest before starting the ACP induction phase.

Lineage Tracing in R26R^EYFP Reporter Mice

R26R^EYFP mice (obtained from Jackson Laboratory; stock number: 006148)⁴⁸ were used for lineage tracing. To precisely trace the lineage of recombined acinar cells, Ptf1a^CreERTM/+ mice containing the R26R^EYFP reporter were crossed with LSL-Kras^G12D/+ mice to generate Ptf1a^CreERTM/+;LSL-Kras^G12D/+; R26R^EYFP. These mice were subsequently bred with Creb^fl/fl mice to produce Ptf1a^CreERTM/+;LSL-Kras^G12D/+; R26R^EYFP; Creb^fl/fl (hereafter referred to as KCYC^-/-) mice.

Mouse Genotyping

Genotyping was conducted using the automated genotyping service provider, Transnetyx. Supplementary Table 1 presents the probe sequences used for genotyping analysis.

Animal Studies

Here, mice of both sexes weighing 20 to 25 g were used; they were housed in pathogen-free conditions under a 12-hour light-dark diurnal cycle with a controlled temperature of 21°C to 23°C and maintained on a standard rodent chow diet (Harlan Laboratories) before the experimental induction of the ACP protocol. The mice were euthanized upon the manifestation of signs of compromised health, including weight loss, accelerated respiration, hunched posture, piloerection, and reduced activity. All animal experiments were approved and performed in compliance with the regulations and ethical guidelines for experimental and animal studies of the Institutional Animal Care and Use Committee and the University of Miami guidelines (Miami, FL; Protocol Nos. 15-057, 15-099, 18-081, and 21-093). All authors had access to the study data and had reviewed and approved the final manuscript.

Establishing ACP In Vivo

The experimental mice were pair-fed with alcohol for 14 weeks using a Lieber-DeCarli alcohol-based liquid diet (A) (BioServ Inc, cat. #F1259SP) containing 5% v/v ethanol, whereas the control mice received a standard control liquid diet (C) (BioServ Inc, cat. #F1259SP) using 28% carbohydrates instead of ethanol. During the final 4 weeks of alcohol exposure, CP was induced by administering cerulein solubilized in phosphate-buffered saline to achieve a concentration of 10 mg/mL (ACP induction period). This feeding regimen mimics the pancreatic damage from chronic alcohol use in humans.^16,22,49 Cerulein was intraperitoneally delivered at a dose of 50 μg/kg through hourly injections (6 times daily, 3 days weekly) over 4 weeks. This combined effect of alcohol and cerulein was observed in the ACP group. The animals were euthanized humanely to harvest the blood and pancreas after ACP induction on days 3 and 21 (ACP recovery period). Supplementary Table 2 presents a comprehensive overview of the treatment groups spanning both the induction and recovery phases.

Tumor Latency Period Estimation

In the context of GEMMs, the tumor latency period is the duration between tumor initiation or formation and the point at which these pancreatic tumors become grossly discernible during biweekly abdominal palpation examinations. Mice exhibiting noticeable signs of illness attributed to an increased tumor burden were euthanized. Subsequently, the pancreatic tissues were collected and subjected to H&E staining for pathological analysis.

Pancreatic Digestion for scRNA-seq and Library Generation

The mouse pancreatic tissue harvested from control KC and KC with ACP induction were mechanically dissociated to generate single-cell suspensions, as previously described.¹⁸ Pancreatic fragments (1–2 mm) were immersed in 0.02% trypsin C-ethylenediamine tetra acetic acid 0.05% (Biological Industries) for 10 minutes at 37°C with agitation, then washed with 10% fetal calf serum (FCS)/Dulbecco’s Modified Eagle Medium (DMEM). For the next dissociation step, the cells were washed with Hanks’ balanced salt solution (HBSS) × 1 containing 1 mg/mL collagenase P, 0.2 mg/mL bovine serum albumin (BSA), and 0.1 mg/mL trypsin inhibitor. After 20 to 30 minutes incubation at 37°C with agitation, the samples were pipetted up and down, returned to 37°C, passed through a 70-mm nylon mesh (Corning #431751), and washed 2 times with 1 × HBSS containing 4% BSA and 0.1 mg/mL DNase I. The samples were divided into 2 equal volumes, with one sample subjected to centrifugation and washed 3 times at 60 × g to isolate large cells containing acinar cells. Simultaneously, the second sample was centrifuged and washed 3 times at 300 × g to collect all cells. If red blood cells were abundant in the second sample, they were treated with red blood cell lysis buffer (Sigma-Aldrich). Live cells were isolated using a MACS Dead Cell Removal Kit (Miltenyi Biotech #130-090-101). Finally, the 2 samples were combined at a cell ratio of 30% and 70% from the 60 × g and 300 × g samples, respectively, and processed for scRNA-seq library preparation by the Oncogenomics-Shared Resource Facility (University of Miami). Briefly, the cells were counted, and 10,000 cells were loaded per lane on 10× chromium microfluidic chips. Single-cell capture, barcoding, and library preparation were performed using the Chromium 85 Controller, Chromium Next GEM Single Cell 3’ GEM, Library & Gel Bead Kit v3.1, and Chromium Next GEM Chip G kit (10× Genomics), with a cell recovery target of 100,000 cells as per manufacturer’s guidelines. The cDNA and libraries were sequenced using an Illumina NovaSeq 6000. The Cell Ranger pipeline (version 7.0, 10× Genomics) transformed Illumina base call files into FASTQ files, aligned them to the GRCm38 reference genome, and generated a digital gene-cell count matrix through the Biostatistics and Bioinformatics Shared Resource facility (University of Miami). Subsequently, the count matrices were imported into R version 3.5.0 and analyzed using the R package Seurat version 4.0.13-15.

Serum Alcohol Analysis

The alcohol concentration in the serum was estimated using an ethanol assay kit (Abcam, cat. # ab65343) as per the manufacturer’s protocol and has been detailed previously.¹⁶

Serum Amylase Analysis

The serum amylase levels in blood samples collected from various study groups of mice were quantified through a colorimetric assay (Abcam, cat. # ab102523). This analysis was conducted in accordance with the manufacturer’s provided protocol and was described in detail previously.¹⁶

Immunohistochemistry and Immunofluorescence Tissue Staining

Harvested pancreas tissues from mice necropsies were fixed in 10% neutral buffered formalin and embedded in paraffin to carry out histological staining procedures including H&E, Sirius Red, Masson's trichrome, and Alcian Blue. For immunohistochemistry (IHC)-based detection, antigen retrieval was performed using citrate (pH 6.0) or Tris-EDTA buffer prior to incubation with BlockAid Blocking Solution (Thermo Fisher), and endogenous peroxidase activity was blocked by incubating with 3% H₂O₂. Tissue sections were then stained with primary antibodies at specified concentrations (Supplementary Table 3) overnight at 4⁰C. IHC slides were developed using 3,3’ diaminobenzidine (DAB) substrate (Vector) followed by counterstain using Meyer’s hematoxylin and imaged using DM750 microscope (Leica Microsystems). For IF-based staining, primary antibody was detected using species-specific Alexa Fluor 594 and/or Alexa Fluor 488 (Thermo Fisher) secondary antibodies incubated on sections for 1 hour at room temperature. Nuclear staining was performed using Hoechst 33342 dye (Thermo Fisher). All IF-stained slides were scanned using the Olympus Fluoview1000 confocal microscope. In the present study, we performed digital image conversion to 16-bit grayscale followed by threshold adjustment in ImageJ to distinguish true fluorescent signals from background autofluorescence. This thresholding strategy was validated, and we consistently excluded areas with nonspecific signal or histological artifacts from quantification. Secondly, fluorescence imaging was performed using an Olympus Fluoview1000 confocal microscope. The use of a confocal system, particularly the adjustable pinhole aperture, allowed for optical sectioning and rejection of out-of-focus background signal, thereby improving signal-to-noise ratio. Although paraffin-embedded pancreatic tissues are known to exhibit green-channel autofluorescence, we minimized its impact by optimizing laser power, detector gain, and imaging settings. High resolution multichannel images were exported for quantitative analysis, quantification of double positive cells was performed using ImageJ (NIH), for each mouse pancreas tissue sections, multiple nonoverlapping fields of view per section were selected ensuring representative sampling across acinar and ductal compartments. The CK19⁺ ducts, Alcian blue⁺ PanINs, and CD45⁺ immune cells were counted using Image J by converting the scanned image to 16-bit grayscale and adjusting the thresholds to differentiate the stained cells from the background. After that, the “analyze particles” function was used to automatically count the cells reported as discrete numbers.

For special staining including Sirius Red, the image scale was first calibrated to micrometers, followed by conversion to grayscale. A threshold was then applied to segment the Sirius Red-stained collagen areas, and the stained area was measured. Collagen deposition was reported as relative extracellular matrix (ECM) content normalized to the corresponding group in comparison for the analysis. IHC staining quantification of pCREB expression was done using ImageJ. The color deconvolution tool was applied to separate the true DAB positive signal from hematoxylin-stained purple counterstain. Measurement of percentage (%) area occupied was calculated across 3 to 4 representative fields of view within the pancreas for each biological replicate across all 4 experimental mice cohorts using consistent upper and lower threshold limits to minimize background noise and ensure specificity of DAB accumulation within the tissues.

H&E-based Assessment of Pancreatic Lesions

Pancreatic tissue sections from age-matched mice in each group were stained with H&E and subsequently examined for ADMs, pancreatic lesions, and carcinoma in situ. The percentage of acinar area and number of ducts that contained any grade of PanIN lesions were measured by examining 10 H&E-stained high-power fields (40× magnification) per slide. PanINs were graded according to established criteria.^50,51 In PanIN1 ducts, the normal cuboidal pancreatic epithelial cells transition to columnar architecture and can gain polyploid morphology. PanIN2 lesions are associated with a loss of polarity. PanIN3 lesions (or in situ carcinoma) show cribriform morphology, the budding off cells and luminal necrosis with marked cytological abnormalities, without invasion beyond the basement membrane.⁵⁰ The data was expressed as a percentage of total lesions in the whole pancreas.

Western Blot Analysis

The freshly harvested pancreas tissue was promptly flash-frozen and preserved at −80°C for long-term storage. To prepare protein lysates for Western blot analysis, the frozen tissues were thawed and homogenized in RIPA buffer (0.1% SDS, 50 mM Tris·HCl, 150 mM NaCl, 1% NP-40, and 0.5% Na deoxycholate) with protease inhibitor cocktail (Sigma) and PhosSTOP phosphatase inhibitor (Roche). Lysates were sonicated and centrifuged at 10,000 g for 15 minutes at 4°C to collect supernatant. The protein concentration of the cell and tissue lysate was determined by Bio-Rad protein assay kit (Bio-Rad). Next, 35 μg of whole-cell lysate or whole-tissue lysate was separated on NuPAGE Novex 4-12% Bis-Tris Gels and transferred on iBlot transfer stack using iBlot dry blotting transfer system (Life Technologies). For immune-detection, membranes were incubated with antibodies listed in Supplementary Table 3. The membranes were subsequently incubated with secondary anti-mouse or anti-rabbit secondary antibodies conjugated with horseradish peroxidase (Jackson ImmunoResearch). Finally, the immunoreactive bands were developed with Pierce ECL Western Blotting Substrate (Thermo Scientific) and recorded on blue basic autoradiography film (Bioexpress). Uncropped raw images of the blots are shown in Figure 11.

Figure 11.

Raw uncropped images of Western blot membranes forFigure 2G and5B.

Phosphokinase Array Analysis

The phosphorylation profiles of kinases were screened using a human profiler array as a screening tool (R&D System, cat. #ARY003B). Key targets including pCREB was subsequently validated by Western blotting and IF-based anlysis using species specific monoclonal antibodies. In brief, 200 mg of equal protein from pancreatic tissue lysates were loaded onto a nitrocellulose membrane with duplicate capture antibody spots. Phosphorylated protein levels were determined with phospho-specific antibodies and detected through chemiluminescent-based reaction chemistry. Spot density on the membrane was quantified using HL⁺⁺ image analysis software.

Cluster Identification and Annotation of Single-cell RNA Sequencing Dataset

Gene-cell matrices were analyzed with Seurat package (v4.3.0 RStudio). Cells fewer than 200 transcripts and ≤7.5% mitochondrial counts were removed. Feature measurements were normalized using the Normalize Data function with a scale factor of 10,000 and the LogNormalize normalization method. Variable genes were identified using the FindVariableFeatures function. Data were scaled and centered using linear regression on the counts and the cell cycle score difference. Principal component analysis (PCA) was performed with the RunPCA function using the previously defined variable genes to identify necessary dimensions for >90% variance within the data. Violin plots were then used to filter the data according to user-defined criteria. Cell clusters were obtained via the FindNeighbors and FindClusters functions, using a resolution of 0.7 for all samples, and nonlinear dimensional reduction was then performed using UMAP clustering. A FindAllMarkers table was created, and clusters were defined by user-defined criteria. Clusters that coexpressed known distinct marker genes were merged for subsequent analysis.

Differential Gene Expression Analysis of Single-cell RNA Sequencing Dataset

Prior to differential gene expression analysis, each cluster of interest was subjected to normalization, scaling, and PCA. Next, the function “FindMarkers” from Seurat v4.0 R package was utilized to finds the differentially expressed genes for identity classes. Genes were considered differentially expressed if detected in at least 25% of clusters, with default log fold change of 0.25. Wilcoxon Rank Sum test was used and listed in Supplementary Table 4. Volcano plots were generated based on these output files using the EnhancedVolcano package. Gene set enrichment analysis (GSEA) was performed using the “fgsea” R package on differentially expressed genes (log(FC) >0.5 and adjusted P-value < .050). “MSigDB” R package was utilized to access the following databases: C2 (KEGG, REACTOME, PID, BIOCARTA), C5 (GO:BP) and H (Hallmarks). Among different databases, the gene sets that met the statistical requirements were then curated, visualized via “ggplot2” R package, and ordered by normalized enrichment score (NES) listed in Supplementary Table 4.

RNA Isolation and qPCR Analysis

RNA was isolated from flash-frozen pancreas tissues using the RNeasy Kit (Qiagen) according to the manufacturer’s protocol. cDNA generated after performing reverse transcription of RNA product was subjected to qPCR analysis using gene-specific predesigned primers (RT² qPCR Primer Assay, Qiagen), listed in Supplementary Table 5. Gene expression was normalized to the housekeeping gene GAPDH using the comparative CT (^ΔΔCT) method and reported as fold change (FC) relative to control.

Isolation of Primary Pancreatic Acinar Cells and 3D Explant Culture

Isolation of primary pancreatic acinar cells and establishment of 3D explant cultures was performed as described previously.⁵² Briefly, the acinar cells were incorporated into a mixture of collagen and Waymouth medium. On days 1, 3, and 5, duct-like structures were counted and quantified in a blinded fashion. The area of the ducts was determined using ImageJ software. Quantification of ductal structures involved the manual counting of at least 4 distinct fields under 10× magnification in triplicates.

Statistical Analysis

Descriptive statistics were calculated using Prism software (GraphPad Software Inc). Results are shown as values of means ± standard deviations (SDs) unless otherwise indicated. Prior to applying parametric tests, data distributions were assessed for normality using Shapiro-Wilk test. To assess multiple comparisons, 1-way analysis of variance (ANOVA) was applied followed by Tukey’s or Dunnett’s post hoc tests when deemed appropriate. A 2-tailed Student’s t-test was used for 2-group comparisons. Statistical significance was defined using a cutoff of .050 unless otherwise specified in the figure legends.