PVNOXT regulates acinar cell-mediated inflammatory response via DMVACh projections to the acinar cell

The protocols for the use and care of laboratory animals were followed in every operation. The experimental mice ranged from 4 to 25 weeks of age. All C57BL/6 mice were purchased from Jiangsu Gempharmatech Co., Ltd. (Nanjing, China), and Cela1-P2A-iCre mice (Stock No: I001074, MGI No: 8235537) and Rosa-LSL-tdTomato mice (Stock No: C001181) were generated by Cyagen Biosciences (Suzhou, China). Given the well-established sexual dimorphism in oxytocinergic signaling, where neural regulation in females can be confounded by fluctuating estrogen levels, we exclusively employed male mice in this study. Mice were housed in a specific pathogen-free facility at the Anhui Medical University Experimental Animal Center under controlled conditions: 22 °C ambient temperature, 40–60% humidity, and a 12-hour light/dark cycle (06:00–18:00 light phase), with ad libitum access to standard rodent chow. The general health of the mice was good, with all surgical procedures following approved guidelines for postoperative monitoring. For retrograde tracing experiments with intraductal injection of PRV or FG, 10–12-week-old male mice were used, using the C57BL/6 or Cela1-P2A-iCre line. For Fos experiments, 21–25-week-old male mice were used, using the Cela1-P2A-iCre line. For all electrophysiology, optical fiber, and DREADD experiments, 21–25-week-old male mice were used, using the C57BL/6 line. Mice were maintained in group housing for the study duration. To establish the model, mice were fed a high-fat diet (HFD) for 16 weeks. The HFD consisted of 58.6% standard chow, 15% lard, 20% sucrose, 5% casein, 1.2% cholesterol, and 0.2% sodium cholate [ 11 ]. To induce the HTGP model, HFD-fed mice were intraperitoneally (i.p.) injected with ceruletide (100 µg/kg/h, A0190, MCE, USA) six times, and lipopolysaccharide (10 mg/kg, L2880, Sangon Biotech, Shanghai, China) was added to the final injection of ceruletide [ 12 , 13 ]. Mice were treated at 18 h after model induction, and venous blood and pancreatic tissues were collected. Mice were anesthetized with 2–2.5% isoflurane and positioned for sterile surgery. A 1.5–2.0-cm midline laparotomy was performed via an incision through the linea alba. The stomach and duodenum were exteriorized and carefully manipulated to expose the biliary-pancreatic duct and sphincter of Oddi. A proximal microclamp on the common bile duct and dual duodenal microclamps flanking the sphincter of Oddi were used to block infusion toward the liver and prevent backflow of the virus to the duodenum. The glass microelectrode was advanced through the duodenal Oddi sphincter into the biliary-pancreatic duct until the tip was positioned at the pancreatic duct orifice. Thereafter, a 3 µL solution (PRV-EGFP, titer: 2.0 × 10 9 PFU/mL; Ba2001, titer: 2.0 × 10 9 PFU/mL; or FG) was infused into the pancreatic duct at 6 µL/min. Following a 5-minute dwell period, all instruments were removed, and the puncture site was sealed with tissue adhesive (Vetbond, 3 M, St. Paul, MN, USA). The abdominal contents were repositioned, and the cavity closed in layers using sutures (Chenghe, Ningbo, China). Mice were allowed to recover on heating pads. Mice under isoflurane anesthesia were secured in a stereotaxic apparatus, and a scalp incision was made to expose the skull. Viral delivery into the target region was performed at a rate of 35 nL/min using a glass microelectrode connected to an infusion pump (micro 4, World Precision Instruments, Sarasota, FL, USA). Animals with inaccurate injection sites were excluded from the data analysis. Mice were deeply anesthetized using isoflurane and then perfused with ice-cold oxygenated N -methyl-D-glucamine artificial cerebrospinal fluid (NMDG ACSF). Coronal slices (300 μm) were sectioned with a vibrating microtome (VT1200s, Leica, Wetzlar, Germany) and were initially incubated in oxygenated HEPES ACSF solution (28 °C) for at least 1 h. Then, the brain slices were transferred to a slice chamber (RC-26, Warner Instruments, Hamden, CT, USA) for whole-cell recording and were continuously perfused with oxygenated standard ACSF solution (32 °C). Neurons in regions of interest were visualized using a water immersion objective (×40) on an upright microscope (BX51WI, Olympus, Tokyo, Japan). The signals were low-pass filtered at 2.8 kHz and digitized at 10 kHz using a Multiclamp 700B amplifier (Molecular Devices, San Jose, CA, USA). The data were collected from the neurons with the appropriate input resistance (more than 100 MΩ) and series resistance (less than 30 MΩ) using Clampex 10 (Molecular Devices, San Jose, CA, USA). The current-evoked firing was recorded in current-clamp mode (I = 0 pA) using pipettes (5–7 MΩ) filled with potassium gluconate-based internal solution. Mice were deeply anesthetized with 2–2.5% isoflurane and transcardially perfused sequentially with ice-cold 0.9% saline and 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). For peripheral tissues (pancreas, duodenum, stomach, liver, lung, and heart), tissues were post-fixed overnight in 4% PFA at 4 °C. They were then dehydrated through a graded ethanol series, cleared in xylene, and embedded in paraffin. Serial Sects.  (4–5 μm) were cut using a rotary microtome (HM340E, Thermo Scientific, Waltham, MA, USA), mounted onto poly-L-lysine-coated slides, and baked overnight at 60 °C. For brain tissue, brains were post-fixed overnight in 4% PFA at 4 °C, cryoprotected in 30% sucrose in PBS until the brains sank (≥ 48 h), and sectioned coronally at 30 μm using a sliding microtome (NX70, CryoStar, Thermo Scientific, Waltham, MA, USA). Sections were collected in 0.1 M PBS for short-term analysis or in PBS containing sodium azide for long-term storage. For immunofluorescence staining, both paraffin-embedded and frozen sections were washed in 0.1 M PBS for 10 min, blocked with 5% normal donkey serum and 0.5% Triton X-100 in PBS for 2 h, and then incubated with primary antibodies at 4 °C overnight. After three washes in PBS (10 min each), sections were incubated with secondary antibodies, washed again (3 × 10 min in PBS), incubated with DAPI solution, washed briefly, and then mounted onto slides. Images were captured using a confocal microscope (LSM980 NLO, Zeiss, Oberkochen, Germany) and a fluorescence microscope (DMIL LED, Leica, Wetzlar, Germany). Fluorescent signal quantification was performed by manual neuron counting. The pancreas tissues were cut into 1 mm 3 sections and fixed in 2.5% glutaraldehyde for 24 h, followed by post-fixation in 1% osmium tetroxide for 3 hours. After dehydration in a graded ethanol series (50%, 70%, 90%, and 100%) for 10 min per step, the samples were embedded in resin and cut to a thickness of 50–60 nm. The sections were stained with uranyl acetate and lead citrate, and then examined under a transmission electron microscope (HT7700, Hitachi, Tokyo, Japan). To prevent bias, no specific areas were selected. Whole pancreas protein was extracted using radioimmunoprecipitation assay (RIPA) lysis buffer containing 1 mM phenylmethanesulfonyl fluoride (PMSF). Protein concentration was determined using a bicinchoninic acid (BCA) protein assay kit (Beyotime Biotechnology, Shanghai, China). After separation by 10% SDS-PAGE, the samples were electrotransferred onto PVDF membranes. The membranes were blocked for 1.5 h with Tris-buffered saline containing 0.05% Tween-20 and 5% non-fat milk. The blots were then incubated with primary antibodies, followed by horseradish peroxidase-conjugated secondary antibodies. Protein bands were visualized using an enhanced chemiluminescence (ECL) kit (JS-M6P, Peiqing, Shanghai, China). β-actin was utilized as a loading control for normalization. ImageJ software (version 1.6.0, NIH, Bethesda, MA, USA) was used to perform densitometric analysis of the bands. Total RNA was extracted using TRIzol. cDNA was synthesized using the Transcriptor First-Strand cDNA Synthesis Kit (Roche Diagnostics, Basel, Switzerland). Quantitative real-time PCR (qRT-PCR) was performed using FastStart Universal SYBR Green Master (Roche Diagnostics, Basel, Switzerland) according to the manufacturer’s instructions. Genomic DNA was extracted using a mouse tail genotyping kit (D7283M, Beyotime, Shanghai, China) following the manufacturer’s instructions. Agarose gel (1.0 g in 50 mL 1× TAE buffer containing 40 mM Tris-acetate and 1 mM EDTA) was dissolved by heating in a microwave, mixed with 5 µL nucleic acid dye, poured into a casting tray with a comb, and solidified at room temperature for 30 min. After solidification, the gel was placed in an electrophoresis tank with wells facing the cathode, submerged in 1× TAE buffer, loaded with DNA samples and a molecular weight marker, and run at 120 V for 40 min. The gel was then imaged, and fragment sizes were estimated based on the marker. Pancreatic tissues were fixed in 4% formalin, embedded in paraffin, and sectioned at a thickness of 4 μm. Sections were stained with hematoxylin and eosin (H&E), Masson’s trichrome, and Sirius Red. Histological changes were examined by light microscopy (BX53, Olympus, Tokyo, Japan). Histopathologic damage was evaluated using the Schmidt pathological score [ 14 ]. To record calcium signals from acetylcholinergic neurons in the dorsal motor nucleus of the vagus (DMV ACh ), we injected rAAV-ChAT-GCaMP6s-WPRE-pA (rAAV-ChAT-GCaMP6s, AAV2/9, titer: 6.22 × 10 12 vg/mL, 180 nL) into the DMV of C57 mice and implanted an optical fiber (200 μm outer diameter, 0.37 numerical aperture, Inper Technology, Hangzhou, China) directly above the DMV. After 3 weeks of viral expression, neural activity was recorded using fiber photometry with the Inper imaging system. The implanted optical fiber was connected to a single-mode fiber patch cord (MFO-1 × 2-F-W1.25-200-0.37-100, Inper Technology, Hangzhou, China) using a ceramic ferrule and black heat-shrink tubing to ensure stable optical coupling. The optical pathway was configured with a 470 nm LED light source (3 ms pulse width, 40 Hz), filtered through a 470/10 nm bandpass filter (FB470-10, Thorlabs, Newton, NJ, USA), reflected by a dichroic mirror (MD498, Thorlabs, Newton, NJ, USA), and focused through a 20× objective lens (PLN20X, NA = 0.4, Olympus, Tokyo, Japan). This optical assembly was then coupled to the fiber via a light adapter (Doric Lenses, Québec, Canada), delivering 25–40 mW of excitation power at the fiber tip. Fluorescence signals from GCaMP6s were collected through a 525/40 nm emission filter (MF525-39, Thorlabs, Newton, NJ, USA) and acquired by a CMOS camera (Orca-Flash 4.0, Hamamatsu Photonics, Hamamatsu, Japan) at 60 Hz, with a 405 nm laser (L405P100, Thorlabs, Newton, NJ, USA) serving as the isosbestic control. All optical signals were processed in real-time using Inper Plot software (Inper Technology, Hangzhou, China), which calculated the average fluorescence intensity within the region of interest at the fiber tip. The relative fluorescence change (ΔF/F) was quantified using the standard formula: ΔF/F (%) = [(Fduration – Fbaseline)/Fbaseline] × 100%, where Fbaseline represented the pre-stimulation fluorescence intensity. During the experiment, we maintained noise, light, and temperature within standard ranges. Mice in which the optical fiber missed the target or viral injection failed were excluded from the analysis. The levels of serum TG (A110-1-1), TC (A111-1-1), AMY (C016-1-1), LPS (H255-1-2), IL-1β (H002), IL-6 (H007), and TNF-α (H052-1-2) were measured using commercially available ELISA kits (Jiancheng Corp, Nanjing, China), following manufacturer’s instructions. Mice were deeply anesthetized with 2–2.5% isoflurane and then perfused with ice-cold 1× PBS to remove blood from the vasculature. Brain tissues were immediately collected and sectioned coronally (50 μm). The PVN regions were isolated via laser-capture microdissection (CellCut System, MMI, Glattbrugg, Switzerland) and stored at − 80 °C until analysis. To generate single-cell suspensions, tissue samples were dissociated using the MobiSolution Tissue Dissociation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Single-cell libraries were prepared using the MobiNova-100 system with the MobiCube single-cell nucleic acid capture and library construction kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Finally, the constructed libraries were sequenced [ 15 , 16 ]. After quality assessment of the raw data, gene quantification and quality control were performed. The normalized gene expression data were subjected to customized analysis, including dimensionality reduction clustering, marker gene identification, cell type identification, differential gene expression analysis, gene enrichment analysis, and cell-cell communication analysis. Single-cell suspension, library preparation, sequencing, and bioinformatics analyses were performed by OE Biotech (Shanghai, China). For immunofluorescence, the following primary antibodies were used: rabbit anti-c-Fos (1:1000, 226008, Synaptic Systems, Göttingen, Germany), goat anti-ChAT (1:500, AB144P, Sigma, St. Louis, MO, USA), rabbit anti-AChE (1:200, AF5274, Affinity, Cincinnati, OH, USA), mouse anti-c-Fos (1:250, AB20894, Abcam, Cambridge, UK), rabbit anti-CRH (1:100, 10944-1-AP, Proteintech, Rosemont, IL, USA), rabbit anti-PGP9.5 (1:100, 14730-1-AP, Proteintech, Rosemont, IL, USA), mouse anti-F4/80 (1:300, AF6139, Santa Cruz, Dallas, TX, USA), rabbit anti-IL-6 (1:300, 26404-1-AP, Proteintech, Rosemont, IL, USA), rabbit anti-IL-1β (1:500, ab283818, Abcam, Cambridge, UK), mouse anti-TNF-α (1:200, sc-52746, Santa Cruz, Dallas, TX, USA), mouse anti-AVP (1:50, sc-390723, Santa Cruz, Dallas, TX, USA), mouse anti-OXT (1:50, sc-515809, Santa Cruz, Dallas, TX, USA), rabbit anti-glutamate (1:500, G6642, Sigma, St. Louis, MO, USA), rabbit anti-GABA (1:500, A2052, Sigma, St. Louis, MO, USA), mouse anti-amylase (1:50, sc-46657, Santa Cruz, Dallas, TX, USA). The following secondary antibodies were used: Alexa Fluor 594-conjugated anti-goat IgG (1:500, A32758, Invitrogen, Waltham, MA, USA), Alexa Fluor 594-conjugated anti-rabbit IgG (1:500, A21207, Invitrogen, Waltham, MA, USA), Alexa Fluor 488-conjugated anti-rabbit IgG (1:500, A21206, Invitrogen, Waltham, MA, USA), and Alexa Fluor 488-conjugated anti-mouse IgG (1:500, A32766, Invitrogen, Waltham, MA, USA). We used the following reagents and materials: BSA (A600332-0025, BBI, Shanghai, China), antifade mounting medium for fluorescence (BL739A, Biosharp, Hefei, China), clozapine N-oxide (CNO, 17366, MCE, Monmouth Junction, NJ, USA), PBS (BL302A, Biosharp, Hefei, China), sucrose (BS085, Biosharp, Hefei, China), DNA marker (Y1318, Tiangen, Beijing, China), DAPI (C1005, Beyotime, Shanghai, China), 4% PFA (BL539A, Biosharp, Hefei, China), SuperRed (BS354B, Biosharp, Hefei, China), atropine (HY-B1205, MCE, Monmouth Junction, NJ, USA), mecamylamine (M9020, Sigma, St. Louis, MO, USA), darifenacin (HY-A0033, MCE, Monmouth Junction, NJ, USA), MLA (HY-N2332A, MCE, Monmouth Junction, NJ, USA), and 4-Color 5-Plex Multicolor Immunofluorescence Kit Plus (D1005plus-05RM, Zhongkang Bio, Wuhan, China). We conducted stereotaxic delivery of the following viruses: PRV-CAG-EGFP (PRV-EGFP, 2.0 × 10 9 PFU/mL, P03001 , Brain VTA, Wuhan, China), PRV-CMV-△TK-LSL-EGFP-2 A-TK (Ba2001, 2.0 × 10 9 PFU/mL, P04002 , Brain VTA, Wuhan, China), FG (FC10000, Fluorochrome, LLC, Denver, CO, USA), rAAV-ChAT-hM3D(Gq)-mCherry-WPRE-pA (AAV2/9, 5.92 × 10 12 vg/mL, PT2213, Brain VTA, Wuhan, China), rAAV-ChAT-hM4D(Gi)-mCherry-WPRE-pA (AAV2/9, 5.67 × 10 12 vg/mL, PT3108, Brain VTA, Wuhan, China), rAAV-ChAT-mCherry-WPRE-pA (AAV2/9, 5.25 × 10 12 vg/mL, PT11947, Brain VTA, Wuhan, China), rAAV-mOXT-hM3D(Gq)-mCherry-WPRE-pA (AAV2/9, 5.59 × 10 12 vg/mL, PT1058, Brain VTA, Wuhan, China), rAAV-mOXT-hM3D(Gq)-EGFP-WPRE-pA (AAV2/9, 5.59 × 10 12 vg/mL, PT1059, Brain VTA, Wuhan, China), rAAV-mOXT-hM4D(Gi)-mCherry-WPRE-pA (AAV2/9, 5.30 × 10 12 vg/mL, PT5607, Brain VTA, Wuhan, China), rAAV2-retro-mOXT-Cre-WPRE-pA (AAV2/R, 5.30 × 10 12 vg/mL, PT6086, Brain VTA, Wuhan, China), rAAV-EF1α-DIO-hM3D(Gq)-mCherry-WPRE-pA (AAV2/9, 5.25 × 10 12 vg/mL, PT0042, Brain VTA, Wuhan, China), rAAV-EF1α-DIO-hM4D(Gi)-mCherry-WPRE-pA (AAV2/9, 5.25 × 10 12 vg/mL, PT0043, Brain VTA, Wuhan, China), rAAV2-retro-hSyn-EGFP-WPRE-hGH pA (AAV2/R, 5.36 × 10 12 vg/mL, PT1990, Brain VTA, Wuhan, China), rAAV-ChAT-GCaMP6s-WPRE-pA (AAV2/9, 6.22 × 10 12 vg/mL, PT1973, Brain VTA, Wuhan, China), rAAV-mOXT-EGFP-5’miR-30a-shwdfy1-3’-miR30a-WPRE-pA (AAV2/9, 5.25 × 10 12 vg/mL, PT14190, Brain VTA, Wuhan, China) and rAAV-mOXT-EGFP-5’miR-30a-scramble-3’-miR30a-WPRE-pA (AAV2/9, 5.46 × 10 12 vg/mL, PT14177, Brain VTA, Wuhan, China). The data are presented as mean ± SD. A one-way analysis of variance (ANOVA) was used to evaluate statistical differences, and the post hoc Bonferroni test was used for multiple comparisons. All statistical analyses were performed using SPSS version 21.0. Two treatment groups were compared using two-tailed unpaired t tests. A p -value of less than 0.05 indicated statistical significance.

Mice were fed with a normal or HFD for 16 weeks (from 3 to 19 weeks of age). By week 9, the mice exhibited significantly greater body weight gain than normal (Fig.  1 A). This indicated that the HFD could elicit a significant increase in the weight of the mice. The feed intake of the mice in the normal group increased sharply in the first 7 weeks, and then gradually decreased, whereas the feed intake of the HFD group increased sharply in the first 5 weeks and then plateaued (Fig.  1 B). Normal-diet consumption consistently exceeded HFD intake throughout the study period, indicating that the HFD could result in a decrease in the feed intake of the mice, possibly due to the increased calories compared to normal feed. Fig. 1 Model Establishment. A Changes in body weight in HFD and normal groups ( n  = 10). B The changes of feed intake in HFD and normal groups ( n  = 10). Data are represented as mean ± SD. Significance was assessed using two-way analysis of variance (ANOVA) ( *p  < 0.05; **p  < 0.01; ***p  < 0.001). C Paraffin-embedded pancreas sections were stained with H&E. Areas of edema (solid arrowhead), inflammatory infiltrates (hollow arrow), hemorrhage (solid arrow), and necrosis (hollow arrowhead). The inset is a higher magnification of the boxed area (scale bar, 100 μm, 20 μm higher magnification inset) ( n  = 3 per group). D Representative columns show the pathological scores of the different groups ( n  = 3 per group). E-F Serum levels of TG, TC, AMY, and LPS ( n  > 3 per group). G Serum levels of IL-1β, IL-6, and TNF-α ( n  > 3 per group). H The protein expression of IL-1β, IL-6, and TNF-α in pancreas extracts prepared from control, HFD, and HTGP mice pancreas tissues ( n  > 3 per group). I Densitometry analysis of the intensity of the protein bands. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( **p  < 0.01 vs. control; ***p  < 0.001 vs. control; ## p  < 0.01 vs. control; ### p  < 0.001 vs. control; ns vs. control). Dots represent individual mice Model Establishment. A Changes in body weight in HFD and normal groups ( n  = 10). B The changes of feed intake in HFD and normal groups ( n  = 10). Data are represented as mean ± SD. Significance was assessed using two-way analysis of variance (ANOVA) ( *p  < 0.05; **p  < 0.01; ***p  < 0.001). C Paraffin-embedded pancreas sections were stained with H&E. Areas of edema (solid arrowhead), inflammatory infiltrates (hollow arrow), hemorrhage (solid arrow), and necrosis (hollow arrowhead). The inset is a higher magnification of the boxed area (scale bar, 100 μm, 20 μm higher magnification inset) ( n  = 3 per group). D Representative columns show the pathological scores of the different groups ( n  = 3 per group). E-F Serum levels of TG, TC, AMY, and LPS ( n  > 3 per group). G Serum levels of IL-1β, IL-6, and TNF-α ( n  > 3 per group). H The protein expression of IL-1β, IL-6, and TNF-α in pancreas extracts prepared from control, HFD, and HTGP mice pancreas tissues ( n  > 3 per group). I Densitometry analysis of the intensity of the protein bands. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( **p  < 0.01 vs. control; ***p  < 0.001 vs. control; ## p  < 0.01 vs. control; ### p  < 0.001 vs. control; ns vs. control). Dots represent individual mice To establish the HTGP model, the HFD mice were i.p. injected with ceruletide (100 µg/kg/h, six times) and a single dose of lipopolysaccharide (10 mg/kg). Ceruletide injections were administered hourly to induce pancreatic acinar cell injury, while lipopolysaccharide was used to exacerbate systemic inflammation. Pancreatic tissues were collected 18 h after the final injection and subjected to H&E staining to evaluate histopathological changes (Fig. 1 C). In the control group, pancreatic tissue sections exhibited normal histological architecture, characterized by well-organized acinar structures, intact islets of Langerhans, and clear ductal epithelium. In contrast, pancreatic tissue sections from the HTGP group displayed pronounced histopathological changes indicative of severe inflammation and tissue damage, including pancreatic edema, inflammatory infiltrates, hemorrhage, and necrosis [ 17 ] (Fig. 1 C and D). The HFD group did not exhibit significant pathological changes, except for some hypertrophy of the islet tissue. Extended HFD intake periods often result in lipid metabolism abnormalities. In our study, HFD feeding led to a significant increase in serum triglyceride (TG) and cholesterol (TC) levels (Fig. 1 E). Elevated amylase and lipase are key indicators of acute pancreatitis [ 18 ]. As expected, ceruletide and lipopolysaccharide significantly increased serum AMY and LPS activities compared with the control groups (Fig. 1 F). To further determine inflammation severity, we examined the serum levels and protein expression levels of IL-1β, IL-6, and tumor necrosis factor (TNF)-α (Fig. 1 G and H, and 1 I). These results showed that the mouse model of HTGP was successfully established. To enable specific retrograde tracing from target acinar cells and identify central neuronal populations synaptically connected to these cells, we developed an in vivo tracing approach that integrates a fluorescence-based retrograde viral tracer, acinar cell-specific transgenic mice, and a reproducible viral delivery technique [ 19 ]. Specifically, we generated Cela1-P2A-iCre transgenic mice (Figure S2A), which selectively express Cre recombinase in pancreatic acinar cells, and injected the Cre-dependent recombinant PRV strain Ba2001 into their pancreatic acinar ducts [ 20 ] (Figure S1 A and S1B). This virus does not inherently express Cre, and EGFP expression is activated only in the presence of Cre. Following uptake by acinar cells, intracellular Cre mediates excision of a STOP cassette, thereby initiating EGFP expression. The virus is subsequently transported retrogradely from the initial acinar cells, enabling the labeling of central and peripheral neurons that innervate the pancreas. The Cela1-P2A-iCre mouse line was generated by replacing the endogenous stop codon (TGA) of the mouse Cela1 gene with a P2A-iCre expression cassette via gene editing. Polymerase chain reaction (PCR) analysis confirmed recombinase gene expression (Figure S2B). To validate tissue-specific recombinase activity, this line was further bred with Rosa-LSL-tdTomato mice to generate ROSA26_tdTomato [CKI/+]; Cela1-P2A-iCre [KI/+] dual-genotype mice. Fluorescence staining of corresponding tissues (Figure S2C) confirmed the pancreatic specificity of Cela1-P2A-iCre mice. Within the observation window of 12–48 h after Ba2001 injection, EGFP+ fluorescence was detected in acinar cells at the 48-hour time point (Fig. 2 A and B, and 2 C). Meanwhile, through fluorescence staining with PGP9.5 (a pan-neuronal marker) and choline acetyltransferase (ChAT, another critical marker of acetylcholine), we identified the presence of intrapancreatic cholinergic ganglia, which co-localized with neurons retrogradely labeled by Ba2001 (Figure S8A). This provides direct morphological evidence that intrapancreatic cholinergic ganglia serve as a critical relay center in the neural circuit from the central nervous system to pancreatic acinar cells. In contrast, no EGFP+ fluorescence was observed in the pancreatic acinar cells of C57BL/6 mice injected with Ba2001, likely because the virus failed to undergo effective recombination in these cells (Fig. 2 D). Fig. 2 EGFP signal in the pancreatic acinar cells and CNS after PRV administration. A-C Representative images of acinar cell-specific (amylase antibody staining, red) localization of Ba2001 (EGFP fluorescence, green) in Cela1-P2A-iCre mice at 12, 24, and 48 h after administration (scale bar, 50 μm) ( n  = 3 mice per condition, three images per mice). D-E Representative images of acinar cell-specific (amylase antibody staining, red) localization of Ba2001 or PRV-EGFP (EGFP fluorescence, green) in C57BL6 mice at 48 h after administration (scale bar, 50 μm) ( n  = 3 mice per condition, three images per mice). F-G Representative images of EGFP fluorescent labeling (green) shows the localization of Ba2001. All sections were counterstained with nuclear marker DAPI (blue); the inset is a higher magnification of the boxed area (scale bar, 250 μm, 25 μm higher magnification inset) ( n  = 3 per mice, three images per region) EGFP signal in the pancreatic acinar cells and CNS after PRV administration. A-C Representative images of acinar cell-specific (amylase antibody staining, red) localization of Ba2001 (EGFP fluorescence, green) in Cela1-P2A-iCre mice at 12, 24, and 48 h after administration (scale bar, 50 μm) ( n  = 3 mice per condition, three images per mice). D-E Representative images of acinar cell-specific (amylase antibody staining, red) localization of Ba2001 or PRV-EGFP (EGFP fluorescence, green) in C57BL6 mice at 48 h after administration (scale bar, 50 μm) ( n  = 3 mice per condition, three images per mice). F-G Representative images of EGFP fluorescent labeling (green) shows the localization of Ba2001. All sections were counterstained with nuclear marker DAPI (blue); the inset is a higher magnification of the boxed area (scale bar, 250 μm, 25 μm higher magnification inset) ( n  = 3 per mice, three images per region) In addition, we constructed a recombinant PRV-EGFP strain, in which EGFP expression is constitutively driven by a ubiquitous CAG promoter, independent of the Cre-loxP system [ 21 ] (Figure S1 A). This virus was administered via the same intraductal injection into the pancreatic ducts of C57BL/6 mice. As expected, PRV-EGFP injection successfully induced EGFP+ expression in the acinar cells of C57BL/6 mice (Fig. 2 E). Notably, no EGFP+ signals were detected in the duodenum, stomach, or liver during this period (Figure S1 C), indicating the absence of viral leakage into adjacent organs. The central circuit mechanism of regulating acinar cell secretion is still unclear, which is why we utilized whole-brain slices to observe EGFP signals. In the CNS, the first EGFP+ cells appeared in the pons, where they were located almost exclusively in the DMV (Fig. 2 F). Consistent with previous reports, the DMV is a crucial component of the parasympathetic nervous system, and a key nucleus in the CNS that innervates the pancreas [ 3 , 22 ]. Meanwhile, EGFP+ cells were also evident, though to a lesser extent, in the nucleus of the solitary tract (Sol) 5 days after injection (Figure S3D). Thereafter, many EGFP+ cells were also observed in the hypothalamus, specifically in the PVN (Fig. 2 G). A small number of EGFP+ cells were also observed in the Barrington’s nucleus (Bar) and gigantocellular reticular nucleus (Gi) (Figure S3B and S3C). According to previous reports, corticotropin-releasing hormone (CRH) neurons in the Bar project to the DMV [ 23 ]. Therefore, the EGFP+ cells observed in our PRV tracing experiment may reflect the existence of a Bar-DMV pathway. Meanwhile, we hypothesize that Gi neurons in the pons may interconnect with the DMV through neural networks. No EGFP+ cells were observed in the striohypothalamic nucleus (stHy), the locus coeruleus (LC), and the nucleus ambiguus (Amb) up to 5 days after viral injection (Figure S3A, S3C, and S3D). EGFP+ cells were also not detected in these brain sections (Figure S3E). Together, these findings demonstrate an interaction that connects the brain with pancreatic acinar cells (Figure S3F). The absence of Ba2001 EGFP+ signals in the brain regions suggests that these regions are not the principal neural pathways to acinar cells, though their potential functional involvement cannot be entirely ruled out. To screen specific brain regions involved in acinar cell-mediated HTGP processes, we introduced c-Fos as a proxy to examine recent neural activity [ 24 ]. The PVN is an autonomic control center in the brain, with neurons playing essential roles in various processes such as stress regulation, metabolism, growth, and immunity. Notably, the PVN showed much stronger c-Fos signals in HTGP mice than in the naive controls [ 25 ] (Fig. 3 A and B). In accordance with the ‘rest and digest’ function of the parasympathetic nervous system, vagal motor neurons originating from the DMV provide parasympathetic motor input to the stomach, small intestine, large intestine, gallbladder, and pancreas. These DMV neurons engage enteric neurons to regulate multiple digestive processes, including gastric contraction, relaxation, and acid secretion, pancreatic endocrine and exocrine secretions, gallbladder contraction, and intestinal motility [ 26 ]. The DMV also showed much stronger c-Fos signals in the HTGP mice than in the naive controls (Fig. 3 A and B). Furthermore, the DMV showed dense c-Fos signals co-labeled with acetylcholinesterase (AChE), a critical marker of acetylcholine (ACh) (Fig. 3 C and D). This suggests that cholinergic neurons may be activated. Moreover, prominent labeling was observed in several other nuclei related to sleep control and the cardiovascular system, such as the nucleus of the Sol, Amb [ 27 , 28 ] (Fig. 3 A and B). These findings suggest that, among the acinar cell-brain axis regions identified by PRV tracing, the PVN and DMV are the ones also associated with HTGP as revealed by c-Fos mapping, indicating a potential neural pathway through which the central nervous system could communicate with the pancreas. Fig. 3 C-Fos activation of PVN and DMV in acinar cell-mediated HTGP. A Representative images show c-Fos expression in the PVN, DMV, Bar, LC, Gi, Amb, StHy, and Sol of the control and HTGP groups of mice (scale bar, 1 mm) ( n  = 3 per group). B Quantification of c-Fos+ cell numbers in nuclei of the control and HTGP mice. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( *p  < 0.05; ***p  < 0.001; ns, not significant). C Representative images of DMV sections showing c-Fos (green) and AChE immunostaining in HTGP mice (scale bar, 800 μm) ( n  = 3 per mice). D Quantification of the percentage of AChE+ & c-Fos+ neurons among total AChE+ neurons in the DMV. E Schematic representation of recording configuration in DMV slices of HTGP mice. F Representative samples of action potentials in the DMV from control and HTGP mice (control, n  = 24 cells; HTGP, n  = 19 cells). G Summary data for firing rate recorded in control and HTGP mice. The action potential frequency is increased in HTGP mice. The statistical analysis involved a two-way ANOVA followed by t-test ( **p  < 0.01, F (1, 656) = 115.5) ( n  = 5 per group). H Two-sample t-test was used to compare action potential thresholds ( *p  < 0.05) ( n  = 5 per group) C-Fos activation of PVN and DMV in acinar cell-mediated HTGP. A Representative images show c-Fos expression in the PVN, DMV, Bar, LC, Gi, Amb, StHy, and Sol of the control and HTGP groups of mice (scale bar, 1 mm) ( n  = 3 per group). B Quantification of c-Fos+ cell numbers in nuclei of the control and HTGP mice. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( *p  < 0.05; ***p  < 0.001; ns, not significant). C Representative images of DMV sections showing c-Fos (green) and AChE immunostaining in HTGP mice (scale bar, 800 μm) ( n  = 3 per mice). D Quantification of the percentage of AChE+ & c-Fos+ neurons among total AChE+ neurons in the DMV. E Schematic representation of recording configuration in DMV slices of HTGP mice. F Representative samples of action potentials in the DMV from control and HTGP mice (control, n  = 24 cells; HTGP, n  = 19 cells). G Summary data for firing rate recorded in control and HTGP mice. The action potential frequency is increased in HTGP mice. The statistical analysis involved a two-way ANOVA followed by t-test ( **p  < 0.01, F (1, 656) = 115.5) ( n  = 5 per group). H Two-sample t-test was used to compare action potential thresholds ( *p  < 0.05) ( n  = 5 per group) Using whole-cell recordings in DMV brain slices, we characterized the electrophysiological properties of pancreas-projecting DMV ACh neurons identified by FG pancreatic infusion (Fig.  3 E). Representative action potential (APs) traces from both control and HTGP mice are shown in Fig.  3 F. The firing rates were increased in the HTGP mice (Fig.  3 G). The minimal current required to elicit APs was significantly lower in the HTGP mice compared to the control mice, indicating enhanced intrinsic excitability (Fig.  3 H). These results suggested that HTGP caused hyperactivity of DMV ACh neurons. To identify the brain origins of pancreatic innervation, we performed retrograde tracing of the upstream neurons through pancreatic FG injection (Fig. 4 A). Vagal efferents displayed FG+ labeling 1 week post-injection (Figure S4A). Within the DMV, FG+ neurons were observed and showed colocalization with ChAT immunoreactivity, while sparse labeling appeared in the Sol (Fig. 4 B, 7 A, 8A, and 8B). Other brain regions remained essentially unlabeled (Figure S4A and 4B). These results indicate that DMV ACh neurons directly project to the pancreas. The DMV is the primary source of efferent projections to the celiac ganglion, which subsequently provides major vagal innervation to the pancreas, including immunomodulatory pathways [ 29 , 30 ]. These findings demonstrate the pivotal role of the DMV in pancreatic inflammatory regulation, as it exhibited the strongest projection signals among all brain regions innervating the pancreas, thus constituting our primary focus. To confirm the direct projection from the DMV to the pancreas, we transected the vagus before injecting FG into the pancreas. In these mice, no signal was found in the DMV (Figure S4C). Fig. 4 DMV ACh neurons directly project to the acinar cells. A Schematic diagram for FG injection into the pancreas of HTGP mice. B FG-labeled neurons colocalized with ChAT immunofluorescence, the magnified image shows the left DMV and right DMVs (right, scale bar, 100 μm) ( n  = 5). C-E Experimental design and representative image of virus injection site (scale bar, 100 μm) ( n  = 8). F-G Serum levels of AMY, LPS, IL-1β, IL-6, and TNF-α ( n  > 3 per group). Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( ***p  < 0.001 control vs. HTGP; ## p  < 0.01 HTGP vs. HTGP + Gi; ### p  < 0.001 HTGP vs. HTGP + Gi; $ p  < 0.05 HTGP vs. HTGP + Gq; $$ p  < 0.01 HTGP vs. HTGP + Gq; $$$ p  < 0.001 HTGP vs. HTGP + Gq; ns, not significant control vs. control + Gi; control vs. control + Gq). Dots represent individual mice DMV ACh neurons directly project to the acinar cells. A Schematic diagram for FG injection into the pancreas of HTGP mice. B FG-labeled neurons colocalized with ChAT immunofluorescence, the magnified image shows the left DMV and right DMVs (right, scale bar, 100 μm) ( n  = 5). C-E Experimental design and representative image of virus injection site (scale bar, 100 μm) ( n  = 8). F-G Serum levels of AMY, LPS, IL-1β, IL-6, and TNF-α ( n  > 3 per group). Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( ***p  < 0.001 control vs. HTGP; ## p  < 0.01 HTGP vs. HTGP + Gi; ### p  < 0.001 HTGP vs. HTGP + Gi; $ p  < 0.05 HTGP vs. HTGP + Gq; $$ p  < 0.01 HTGP vs. HTGP + Gq; $$$ p  < 0.001 HTGP vs. HTGP + Gq; ns, not significant control vs. control + Gi; control vs. control + Gq). Dots represent individual mice To further confirm whether the DMV promotes acinar cell secretion by selectively targeting ACh neurons, we investigated this connection by chemogenetic stimulation of DMV ACh neurons. Recombinant adeno-associated viruses (rAAV) carrying the ChAT gene promoter were stereotaxically injected into the DMV of C57BL/6 mice: rAAV-ChAT-hM3D(Gq)-mCherry for activation or rAAV-ChAT-hM4D(Gi)-mCherry for inhibition [ 31 ]. This strategy enabled the specific expression of hM3D(Gq) or hM4D(Gi) reporter in DMV ACh neurons, thereby establishing the DMV ACh : hM3D(Gq) and DMV ACh : hM4D(Gi), respectively (Fig. 4 D and E). Subsequently, neuronal activity was modulated by i.p. injection of the hM3Dq receptor agonist CNO (2 mg/kg, Fig. 4 C). Enzyme activity assays were performed 2 h after CNO administration. Serum results showed that under normal physiological conditions, activation of Gi-coupled receptors did not alter the levels of pancreatic enzyme and inflammatory mediator secretion, nor did activation of Gq-coupled receptors. However, under pathological conditions, compared with the HTGP group, activation of Gi-coupled receptors reduced the levels of pancreatic enzyme secretion and inflammatory mediator secretion, whereas activation of Gq-coupled receptors increased these levels, although they did not fully return to the levels of the control group (Fig. 4 F and G). Western blotting analysis validated this conclusion (Figure S9A). To further evaluate histopathological changes, pancreatic tissues were collected and subjected to H&E staining (Figure S9B). The results showed that under normal physiological conditions, pancreatic tissue sections from the control group exhibited typical histological architecture, and activation of either Gi- or Gq-coupled receptors did not induce significant histological alterations. In contrast, the HTGP group presented marked pathological damage. Compared with the HTGP group, activation of Gi-coupled receptors partially reversed the pathological changes, as evidenced by reduced infiltration of inflammatory cells such as neutrophils and attenuated tissue necrosis. Conversely, activation of Gq-coupled receptors exacerbated tissue injury, manifested by increased erythrocyte extravasation and aggravated inflammatory infiltration. This indicates that under HTGP conditions, DMV ACh neurons are specifically activated and drive disease progression, but their modulation only partially reverses the detected parameters, a finding entirely consistent with the complex in vivo pathological process. Possible contributing factors include: (1) The progression of HTGP encompasses acinar cell damage, local necroinflammation, and systemic inflammation. Our intervention targeted one specific central cholinergic efferent pathway, while other pathological processes have their own progression trajectories that may not be entirely dependent on neural regulation; (2) Once the pathological cascade is initiated, even if a key driving factor is removed, the established pathological damage and inflammatory environment can persist for a period; (3) Other neuronal populations may be involved in the regulation; (4) Other nuclei within the central nervous system might constitute parallel regulatory pathways. These findings suggest that DMV ACh represents a key neuronal population that is activated under HTGP conditions and contributes to disease progression, and combined intervention strategies may represent a more effective therapeutic approach in the future, with its physiological function requiring further investigation. Collectively, our findings support the existence of a DMV ACh -acinar cell axis that coordinately promotes pancreatic enzyme and inflammatory mediator secretion in vivo. To investigate the potential neural circuit providing input to the DMV in the HTGP mice, the recombinant retrograde virus rAAV2-retro-hSyn-EGFP was injected into the DMV [ 32 ] (Fig. 5 A and B). We detected abundant EGFP signals in the PVN (Fig. 5 C), as indicated by co-staining, which identified EGFP+ oxytocin neurons (82%) (Fig. 5 D and G). However, we did not observe ample EGFP+ glutamatergic neurons (14%) and EGFP+ GABAergic neurons (8%) (Fig. 5 E and F, and 5 G). These results suggested that the DMV receives direct innervation from the PVN. However, we cannot confirm the role of OXT neurons in these innervations. Therefore, we conducted retrograde tracing by pancreatic acinar duct injection with Ba2001. Five days after virus expression, numerous EGFP+ neurons could be observed in the PVN that co-stained with an oxytocin antibody (83%), a glutamatergic antibody (12%), or a GABAergic antibody (9%) (Figure S5C and S5D). Together, these results suggested that DMV ACh neurons receive direct innervation from PVN OXT . Fig. 5 DMV ACh receives direct inputs from PVN OXT neurons. A Experimental design of the retrograde tracing from the DMV to the PVN. B-C Representative images showing the DMV injection site and EGFP+ neurons in the PVN (scale bar, DMV, 400 μm and PVN, 1 mm) ( n  = 5). D-G Representative images and summarized data showing that EGFP+ neurons within the PVN are colocalized with oxytocinergic (OXT), glutamatergic (Glu), and GABAergic (GABA) neurons (scale bar, 50 μm) ( n  = 3) DMV ACh receives direct inputs from PVN OXT neurons. A Experimental design of the retrograde tracing from the DMV to the PVN. B-C Representative images showing the DMV injection site and EGFP+ neurons in the PVN (scale bar, DMV, 400 μm and PVN, 1 mm) ( n  = 5). D-G Representative images and summarized data showing that EGFP+ neurons within the PVN are colocalized with oxytocinergic (OXT), glutamatergic (Glu), and GABAergic (GABA) neurons (scale bar, 50 μm) ( n  = 3) To explore the influence of the PVN OXT -DMV ACh axis on inflammation, we first tested the effects of activating or inhibiting PVN OXT neurons. We delivered the chemogenetics viruses rAAV-mOXT-hM3D(Gq)-mCherry or rAAV-mOXT-hM4D(Gi)-mCherry into the PVN (Fig. 6 B and C). These viruses contain the mice oxytocin gene promoter fragment to specifically drive OXT expression in OXT neurons, enabling targeted manipulation of PVN OXT neurons [ 33 ]. PVN OXT neurons were activated by i.p. injection of CNO (Fig. 6 A). Serum analysis revealed that under normal physiological conditions, activation of either Gi- or Gq-coupled receptors did not significantly affect the secretion levels of pancreatic enzymes or inflammatory mediators. Under pathological conditions, however, activation of Gi-coupled receptors partially reduced the secretion of pancreatic enzymes and inflammatory mediators compared with the HTGP group, whereas activation of Gq-coupled receptors elevated these parameters (Fig. 6 D and E). Consistent with our previous findings regarding the modulation of DMV ACh activity, these changes did not fully return to the levels observed in the control group. Western blotting further validated these results (Figure S10A). Fig. 6 Activation of the PVN OXT -DMV ACh axis produces inflammatory response. A-C Experimental design and representative image of the virus injection site (scale bar, 100 μm) ( n  = 6). D-E Serum levels of AMY, LPS, IL-1β, IL-6, and TNF-α ( n  = 6). Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( ***p  < 0.001 control vs. HTGP; ## p  < 0.01 HTGP vs. HTGP + Gi; ### p  < 0.001 HTGP vs. HTGP + Gi; $$ p  < 0.01 HTGP vs. HTGP + Gq; $$$ p  < 0.001 HTGP vs. HTGP + Gq; ns, not significant control vs. control + Gi; control vs. control + Gq). Dots represent individual mice. F-H Experimental design and representative image of virus injection site (scale bar, 100 μm) ( n  = 3). I Representative images of pancreatic resin slice observed by TEM. Zymogen granules (solid arrowhead), vacuoles (solid arrow), and ER (hollow arrowhead) (scale bar, 2 μm) ( n  = 5 per group). J The protein expression of IL-1β, IL-6, and TNF-α in pancreas extracts prepared from control and HTGP mice pancreas tissues ( n  > 3 per group). K Densitometry analysis of the intensity of the protein bands. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( ***p  < 0.001 control vs. HTGP; ### p  < 0.001 HTGP + Gi + NS vs. HTGP + Gi + CNO; $$$ p  < 0.001 HTGP + Gq + NS vs. HTGP + Gq + CNO). Dots represent individual mice Activation of the PVN OXT -DMV ACh axis produces inflammatory response. A-C Experimental design and representative image of the virus injection site (scale bar, 100 μm) ( n  = 6). D-E Serum levels of AMY, LPS, IL-1β, IL-6, and TNF-α ( n  = 6). Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( ***p  < 0.001 control vs. HTGP; ## p  < 0.01 HTGP vs. HTGP + Gi; ### p  < 0.001 HTGP vs. HTGP + Gi; $$ p  < 0.01 HTGP vs. HTGP + Gq; $$$ p  < 0.001 HTGP vs. HTGP + Gq; ns, not significant control vs. control + Gi; control vs. control + Gq). Dots represent individual mice. F-H Experimental design and representative image of virus injection site (scale bar, 100 μm) ( n  = 3). I Representative images of pancreatic resin slice observed by TEM. Zymogen granules (solid arrowhead), vacuoles (solid arrow), and ER (hollow arrowhead) (scale bar, 2 μm) ( n  = 5 per group). J The protein expression of IL-1β, IL-6, and TNF-α in pancreas extracts prepared from control and HTGP mice pancreas tissues ( n  > 3 per group). K Densitometry analysis of the intensity of the protein bands. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA ( ***p  < 0.001 control vs. HTGP; ### p  < 0.001 HTGP + Gi + NS vs. HTGP + Gi + CNO; $$$ p  < 0.001 HTGP + Gq + NS vs. HTGP + Gq + CNO). Dots represent individual mice We next examined the changes in inflammatory mediator secretion by acinar cells and pancreatic macrophages at different stages of inflammation. PVN OXT neurons were activated at 2 h and 18 h after HTGP induction, and the cellular sources of inflammatory mediator were assessed by immunofluorescence staining. At the 2-hour time point, acinar cells exhibited extensive colocalization with IL-1, IL-6, and TNF-α, with the number of colocalized cells being significantly higher than that observed for macrophages (Figure S10B). At the 18-hour time point, the number of macrophages colocalized with IL-1, IL-6, and TNF-α was markedly increased and significantly higher than that of acinar cells colocalized with these mediators (Figure S10C). These findings indicate that the elevation of serum inflammatory mediators is not attributable to a single cell type. In the early phase of inflammation, acinar cells undergo injurious stress and activation, followed by macrophage activation, which subsequently becomes the predominant source of circulating inflammatory mediators. We also assessed pancreatic histopathological changes using H&E staining (Figure S10D). In line with the serological findings, activation of Gi- or Gq-coupled receptors did not induce notable structural alterations in pancreatic tissue under normal physiological conditions. Under HTGP pathological conditions, however, activation of Gi-coupled receptors attenuated histopathological damage, as evidenced by reduced infiltration of inflammatory cells such as neutrophils. In contrast, activation of Gq-coupled receptors exacerbated tissue injury, characterized by increased inflammatory infiltration. Next, we tested the effects of activating PVN OXT -DMV ACh neurons on inflammatory mediator secretion. We delivered the retrograde transport virus rAAV2-retro-mOXT-Cre into the bilateral DMV ACh and rAAV-EF1α-DIO-hM3D(Gq)-mCherry into the bilateral PVN OXT of C57BL/6 mice (Fig. 6 G). Similarly, we delivered the retrograde transport virus rAAV2-retro-mOXT-Cre into the bilateral DMV ACh and rAAV-EF1α-DIO-hM4D(Gi)-mCherry into the bilateral PVN OXT of C57BL/6 mice (Fig. 6 H). PVN OXT -DMV ACh neurons were activated by i.p. injection of CNO (Fig. 6 F). Transmission electron microscopy (TEM) was employed to evaluate ultrastructural alterations in pancreatic tissues (Fig. 6 I). Pancreatic acinar cells exhibited intact cellular architecture, the tight junctions between adjacent cells remained structurally intact, and no signs of cytoplasmic vacuolization or organelle damage were observed. In contrast to the control, TEM revealed significant ultrastructural abnormalities in the HTGP group, characterized by marked dilation of the rough endoplasmic reticulum (rER), with evident vesiculation, reduced numbers of zymogen granules, and irregular granule morphology. Meanwhile, we noted the appearance of endoplasmic reticulum (ER) whorls in the HTGP group, multi-laminar concentric membrane structures recognized as markers of pronounced ER stress [ 34 , 35 ]. Conversely, the Gi-coupled receptor activation group showed partial amelioration of cellular damage, and decreased vacuolization, suggesting attenuated cellular stress. However, activation of the Gq-coupled receptor group exacerbated acinar cellular damage, with near-complete loss of zymogen granules. Severe rER fragmentation and prominent autophagic vacuoles were observed, indicating serious cellular degeneration. Western blotting analysis revealed that under normal physiological conditions, activation of either Gi- or Gq-coupled receptors did not significantly alter the protein expression levels of IL-1β, IL-6, and TNF-α. Under pathological conditions, however, the Gi group exhibited varying degrees of reduction in IL-1β, IL-6, and TNF-α expression levels compared with the HTGP group, whereas the Gq group showed corresponding increases in these inflammatory factors (Fig. 6 J and K). Serum analysis further corroborated these findings (Figure S11A). Consistent with earlier findings, H&E staining further confirmed pancreatic histopathological changes induced by modulating the PVN OXT -DMV ACh neurons (Figure S11B). The results demonstrated that under normal physiological conditions, pancreatic tissue sections from the control group exhibited typical histological architecture, and activation of either Gi- or Gq-coupled receptors did not induce significant histological alterations. In contrast, the HTGP group presented marked pathological damage. Compared with the HTGP group, activation of Gi-coupled receptors partially reversed the pathological changes, as evidenced by reduced erythrocyte extravasation and alleviated inflammatory infiltration. Conversely, activation of Gq-coupled receptors exacerbated tissue injury, manifested by increased tissue necrosis. However, the regulation of pancreatic inflammation by the nervous system exhibits a high degree of context dependence and bidirectionality, depending on the receptor subtypes involved, namely pro-inflammatory muscarinic acetylcholine receptors (mAChR) versus anti-inflammatory nicotinic acetylcholine receptors (nAChR) [ 36 ]. Based on this, we further investigated the downstream receptor types mediating the pro-inflammatory response induced by activation of the PVN OXT -DMV ACh axis. To this end, we inhibited downstream receptor function using atropine (40 µg/kg, 1 i.p. administration 30 min prior the first and third ceruletide injections, a non-selective mAChR antagonist) or mecamylamine (1 mg/kg, i.p., a non-selective nAChR antagonist) [ 37 , 38 ]. Western blotting analysis revealed that, compared with the HTGP group, activation of the PVN OXT -DMV ACh axis led to increased expression levels of inflammatory factors. Notably, this pro-inflammatory effect was attenuated upon inhibition of mAChR expression, whereas no significant change was observed following inhibition of nAChR expression (Figure S11C). To further delineate the receptor subtypes involved, we employed darifenacin (20 mg/kg, i.p., a selective M3 receptor antagonist) and MLA (1 mg/kg, i.p., once daily for one week, a selective α7nAChR antagonist) to block the respective downstream receptors [ 39 , 40 ]. Consistently, inhibition of M3 receptor expression attenuated the pro-inflammatory effect, whereas inhibition of α7nAChR expression resulted in no significant alteration (Figure S11D). Collectively, these findings suggest that in the HTGP model, activation of the PVN OXT -DMV ACh axis likely exerts its pro-inflammatory effects primarily through M3 muscarinic receptor-mediated signaling. To elucidate the molecular mechanisms by which the PVN regulates pancreatic enzyme secretion and inflammatory response, we established the HTGP group and subsequently isolated the PVN brain region for single-cell RNA sequencing analysis (Fig. 7 A). After filtering low-quality cells, approximately 20,995 high-quality cells were obtained in each group. Subsequently, we grouped the cells into 18 clusters based on similar gene expression patterns (Fig. 7 B). By cross-referencing the gene signature unique to each cluster with known mouse brain cell-type signatures from previous studies and other databases, we divided the cells into five cell types, including neurons, astrocytes, OPC, oligodendrocytes, and endothelial cells [ 41 ] (Fig. 7 C and D). We categorized neurons into four primary neuronal subtypes, including glutamatergic, GABAergic, histaminergic, and oxytocinergic neurons, based on the expression of key neurotransmitters or neuropeptides [ 42 ] (Fig. 7 E and F). Fig. 7 Single-cell transcriptomic atlas of the PVN in HTGP. A Schematic diagram of the single-cell sequencing workflow. B Uniform Manifold Approximation and Projection (UMAP) embedding of single-cell RNA sequencing (scRNA-seq) data for all 20,995 cells. C The UMAP plots of the five cell types in HTGP and NS. D Bubble plot showing the expression of marker genes used to define the five major cell types. E Four principal neuronal subclusters were delineated by UMAP analysis. F Bubble plot for marker genes of neuronal subclusters. G-H Dot plot and bar graph displaying the KEGG and GO enrichment analysis results of differentially expressed genes in the HTGP group, respectively. I Representative differentially expressed Wdfy1 genes in the PVN region of the brain sections. The vertical axes of the violin plots indicate normalized gene expression values Single-cell transcriptomic atlas of the PVN in HTGP. A Schematic diagram of the single-cell sequencing workflow. B Uniform Manifold Approximation and Projection (UMAP) embedding of single-cell RNA sequencing (scRNA-seq) data for all 20,995 cells. C The UMAP plots of the five cell types in HTGP and NS. D Bubble plot showing the expression of marker genes used to define the five major cell types. E Four principal neuronal subclusters were delineated by UMAP analysis. F Bubble plot for marker genes of neuronal subclusters. G-H Dot plot and bar graph displaying the KEGG and GO enrichment analysis results of differentially expressed genes in the HTGP group, respectively. I Representative differentially expressed Wdfy1 genes in the PVN region of the brain sections. The vertical axes of the violin plots indicate normalized gene expression values Oxytocin neurons, which are well known to be predominantly restricted to the PVN, have been shown to regulate the expression of pyroptotic and pro-inflammatory factors, including NOD-like receptor protein 3 (NLRP3), apoptosis-associated speck-like protein containing a CARD (ASC), gasdermin D (GSDMD), caspase-1, IL-1β, and IL-18 [ 42 , 43 ]. These findings attracted our interest. We then performed an enrichment analysis of the differentially expressed genes from oxytocin neurons, comparing the HTGP and NS groups. The results showed that the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis mainly included the TNF, mitogen-activated protein kinase (MAPK), atherosclerosis, and toll-like receptor signaling pathways (Fig. 7 G). The GO function enrichment was mainly positive regulation of the toll-like receptor signaling pathway, including TLR3, TLR4, and TLR9 (Fig. 7 H). Therefore, we focused on the toll-like receptor (TLR) signaling pathway, as it was not only significantly enriched in both GO and KEGG analyses but had also been explicitly demonstrated by multiple studies to play a critical role in the pathogenesis of pancreatitis [ 44 ]. Further, we analyzed all members of the TLR family in oxytocin neurons, including Wdfy1 , Tnip2 , Rac1 , Rtn4 , Pik3r1 , Pdpk1 , Nfkb1 , Mapk1 , Map2k7 , Map2k3 , Fos , Chuk , and App (Figure S6A and 7I). Based on the P-values ( P < 0.05), we focused our analysis on Wdfy1 , App , Rtn4 , Fos , Map2k7 , Map2k3 , Tnip2 , and Mapk1 . Among these, Wdfy1 has been identified in microarray studies as a definite gene influencing the degree of chronic pancreatitis in mice [ 45 ]. Functionally, Wdfy1 can positively regulate signaling pathways mediated by TLR3 and TLR4, thus promoting the activation of the TLR3/4 ligand-induced transcription factors NF-κB [ 46 ]. Regarding App , aspartic proteinases (Saps) hydrolyze App into amyloid β (Aβ)-like peptides that bind to TLR4 and promote fungal killing in vitro, whereas the fungal pathogen Candida albicans is linked to chronic brain diseases such as Alzheimer’s disease [ 47 , 48 ]. Downregulation of Rtn4 ameliorates cerebral ischemia/reperfusion injury in mice through regulating microglia polarization via the TLR4/NF-κB pathway, and inhibits the expression of inflammatory mediators, such as CHOP, IL-6, and TNF-α, in a muscle cell line [ 49 , 50 ]. Fos and other transcription factors are involved in the immunosuppressive activity of artemisinin-type compounds toward inflammatory and autoimmune diseases [ 51 ]. Moreover, Tnip2 mediates GRβ-promoted inflammation and is associated with the severity of major depressive disorder [ 52 , 53 ]. Finally, MAP2K7 , MAP2K3 , and MAPK1 are all members of the MAPK family. MAP2K7 primarily functions in the JNK signaling pathway, regulating proliferation, apoptosis, and cytodifferentiation [ 54 ]. MAP2K3 specifically activates the p38 pathway, regulating cancer and heart failure [ 55 , 56 ], while MAPK1 (also known as ERK2) serves as the terminal kinase in the ERK cascade, regulating cell apoptosis and the cell cycle [ 57 ]. Consequently, among the TLR family members analyzed, Wdfy1 was the only gene with a reported association with pancreatitis in the literature, prompting us to conduct further investigations. The rAAV-mOXT-shWdfy1-EGFP virus was injected to downregulate the expression of Wdfy1 in PVN OXT neurons in the HTGP model group. Pancreatic tissues were collected and subjected to H&E staining to evaluate histopathological changes (Figure S7A). The levels of Wdfy1 protein and mRNA were analyzed by Western blotting and qRT-PCR. The results of qRT-PCR analysis indicated that compared with the control group, the mRNA expression level was significantly increased in the HTGP group, while it was decreased in the virus-injected group (Figure S7B). Western blotting further validated the findings (Figure S7C). To investigate the role of Wdfy1 in the activity of the PVN OXT -DMV ACh axis, we performed fiber photometry to monitor ChAT neural calcium signals in the DMV (Fig.  8 A). We first injected rAAV-ChAT-GCaMP6s into the DMV of C57 mice. After a 21-day viral expression period, DMV ACh neural activity was recorded to establish baseline (BL) calcium activity in DMV neurons (Fig.  8 C). Then, we injected rAAV-mOXT-shWdfy1-EGFP into the PVN of C57 mice to inhibit Wdfy1 expression (day 1, D1) (Fig.  8 B) and injected rAAV-mOXT-shScramble-EGFP into the PVN to serve as a control. Finally, we established the HTGP model in rAAV-mOXT-shWdfy1-EGFP mice and rAAV-mOXT-shScramble-EGFP mice at day 21 (D21). The results of fiber photometry recordings revealed that the spontaneous calcium transients and average calcium responses in GCaMP6s-expressing DMV ACh neurons innervated by the PVN were not significantly changed in rAAV-mOXT-shScramble-EGFP mice. However, the spontaneous calcium transients and average calcium responses in PVN-innervated GCaMP6s-expressing DMV ACh neurons were decreased in rAAV-mOXT-shWdfy1-EGFP mice (Fig.  8 D and E), suggesting that Wdfy1 is required to maintain BL calcium activity in DMV neurons. Fig. 8 Wdfy1 is involved in the PVN OXT -DMV ACh axis-mediated inflammatory response of acinar cells. A Schematic diagram of the real-time optical fiber photometry assay. B Representative image of the virus injection site in the PVN (scale bar, 100 μm) ( n  = 6 per group). C A typical image of GCaMP6s expression and track of the optical fiber in the DMV (scale bar, 50 μm) ( n  = 6 per group). D Example traces (left) and heatmaps (right) showing DMV ACh neuronal calcium activity in rAAV-mOXT-shScramble-EGFP or rAAV-mOXT-shWdfy1-EGFP virus injection mice. The color scale on the right indicates ΔF/F (%). E Summary data for average ΔF/F (%) (left) and frequency (right) of spontaneous calcium events in DMV ACh neurons in rAAV-mOXT-shScramble-EGFP or rAAV-mOXT-shWdfy1-EGFP virus injection mice. The statistical analysis involved a two-way ANOVA followed by t-test ( **p  < 0.01 D21 vs. D1; ****p  < 0.0001 D21 vs. D1; ns, not significant) ( n  = 5 per group). F Example traces (left) and heatmaps (right) showing DMV ACh neuronal calcium activity in rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shScramble-EGFP virus injection mice or rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shWdfy1-EGFP virus injection mice. The color scale on the right indicates ΔF/F (%). G Summary data for average ΔF/F (%) (left) and frequency (right) of spontaneous calcium events in DMV ACh neurons in rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shScramble-EGFP virus injection mice or rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shWdfy1-EGFP virus injection mice. The statistical analysis involved a two-way ANOVA followed by t-test ( ***p  < 0.001 BL vs. POST; ****p  < 0.0001 BL vs. POST; ns, not significant) ( n  = 5 per group). H The protein expression of IL-1β, IL-6, and TNF-α in pancreas extracts prepared from each group. Representative blots are shown (left, n  = 3 per group). Densitometry analysis of the intensity of the protein bands (right). I qRT-PCR analysis of IL-1β, IL-6, and TNF-α expression in pancreas extracts prepared from each group. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA (*** p  < 0.001 control vs. HTGP; ## p  < 0.01 HTGP + hM3Dq vs. HTGP + shWdfy1; ### p  < 0.001 HTGP + hM3Dq vs. HTGP + shWdfy1; $$ p  < 0.01 HTGP + hM3Dq vs. HTGP + shWdfy1 + hM3Dq; $$$ p  < 0.001 HTGP + hM3Dq vs. HTGP + shWdfy1 + hM3Dq). Dots represent individual mice Wdfy1 is involved in the PVN OXT -DMV ACh axis-mediated inflammatory response of acinar cells. A Schematic diagram of the real-time optical fiber photometry assay. B Representative image of the virus injection site in the PVN (scale bar, 100 μm) ( n  = 6 per group). C A typical image of GCaMP6s expression and track of the optical fiber in the DMV (scale bar, 50 μm) ( n  = 6 per group). D Example traces (left) and heatmaps (right) showing DMV ACh neuronal calcium activity in rAAV-mOXT-shScramble-EGFP or rAAV-mOXT-shWdfy1-EGFP virus injection mice. The color scale on the right indicates ΔF/F (%). E Summary data for average ΔF/F (%) (left) and frequency (right) of spontaneous calcium events in DMV ACh neurons in rAAV-mOXT-shScramble-EGFP or rAAV-mOXT-shWdfy1-EGFP virus injection mice. The statistical analysis involved a two-way ANOVA followed by t-test ( **p  < 0.01 D21 vs. D1; ****p  < 0.0001 D21 vs. D1; ns, not significant) ( n  = 5 per group). F Example traces (left) and heatmaps (right) showing DMV ACh neuronal calcium activity in rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shScramble-EGFP virus injection mice or rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shWdfy1-EGFP virus injection mice. The color scale on the right indicates ΔF/F (%). G Summary data for average ΔF/F (%) (left) and frequency (right) of spontaneous calcium events in DMV ACh neurons in rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shScramble-EGFP virus injection mice or rAAV-mOXT-hM3D(Gq)-EGFP + rAAV-mOXT-shWdfy1-EGFP virus injection mice. The statistical analysis involved a two-way ANOVA followed by t-test ( ***p  < 0.001 BL vs. POST; ****p  < 0.0001 BL vs. POST; ns, not significant) ( n  = 5 per group). H The protein expression of IL-1β, IL-6, and TNF-α in pancreas extracts prepared from each group. Representative blots are shown (left, n  = 3 per group). Densitometry analysis of the intensity of the protein bands (right). I qRT-PCR analysis of IL-1β, IL-6, and TNF-α expression in pancreas extracts prepared from each group. Data are represented as mean ± SD. Significance was assessed using two-way ANOVA (*** p  < 0.001 control vs. HTGP; ## p  < 0.01 HTGP + hM3Dq vs. HTGP + shWdfy1; ### p  < 0.001 HTGP + hM3Dq vs. HTGP + shWdfy1; $$ p  < 0.01 HTGP + hM3Dq vs. HTGP + shWdfy1 + hM3Dq; $$$ p  < 0.001 HTGP + hM3Dq vs. HTGP + shWdfy1 + hM3Dq). Dots represent individual mice Next, to assess whether Wdfy1 modulates DMV calcium responses to inputs from the PVN, we injected rAAV-mOXT-shScramble-EGFP and rAAV-mOXT-hM3D(Gq)-EGFP into the PVN. Chemogenetic activation of PVN OXT neurons by CNO administration elicited robust calcium elevations in the DMV, similar to the spontaneous calcium transients and average calcium responses. However, this effect was eliminated after injecting rAAV-mOXT-shWdfy1-EGFP and rAAV-mOXT-hM3D(Gq)-EGFP simultaneously, with no significant difference in calcium signals compared to BL levels (Fig.  8 F and G), suggesting that Wdfy1 knockdown attenuates the calcium signals in the DMV induced by activating PVN OXT neurons. These results demonstrate that the Wdfy1 gene is essential for the activity of the PVN OXT -DMV ACh axis. These findings were confirmed by PCR and Western blotting. Western blotting and qRT-PCR analyses were performed to evaluate alterations in inflammatory factors (IL-1β, IL-6, and TNF-α) at both protein and mRNA levels in the HTGP model. Significant downregulation of inflammatory factors was observed following rAAV-mOXT-shWdfy1-EGFP administration. Combined delivery of rAAV-mOXT-shWdfy1-EGFP and rAAV-mOXT-hM3D(Gq)-EGFP resulted in more pronounced suppression of inflammatory factors expression compared to rAAV-mOXT-hM3D(Gq)-EGFP alone (Fig.  8 H and I). In summary, our results demonstrate that the Wdfy1 gene is involved in the PVN OXT -DMV ACh axis-mediated inflammatory response of acinar cells.

Pancreatic innervation was first described over a century ago, in the time of Paul Langerhans, and the role of neural stimuli in regulating secretory function was later confirmed by Ivan Pavlov and subsequent investigators [ 1 , 2 ]. Anatomical and functional studies have recently revealed that the pancreas is connected to the central nervous system (CNS) and enteric nervous system, and contains intrapancreatic ganglia. The nerve bundles entering the pancreas include sensory afferent and autonomic fibers of both sympathetic and parasympathetic branches [ 3 ]. Furthermore, recent advances in volume tissue imaging have illuminated the organ-wide neuronal architecture and spatial organization of the pancreas, highlighting new opportunities to explore cellular interrelationships, particularly neuronal activity in health and diseases [ 4 ]. The pancreas is structurally organized into distinct compartments for endocrine and exocrine functions [ 5 ]. Although the pancreas is functionally similar across species, it exhibits distinct anatomical differences. In humans, it is a single organ composed of the head, body, and tail regions, whereas in mice, it is dispersed along the mesentery of the small intestine in a branched structure with duodenal, gastric, and splenic lobes. The exocrine portion, making up most of the organ’s mass, is embedded in the parenchyma and comprises acinar and ductal cells. The former synthesizes and releases digestive enzymes, including amylase (AMY) and lipase (LPS), which serve as the most widely used laboratory diagnostic biomarkers for pancreatitis [ 6 , 7 ]. Studies employing multi-synaptic viral tracers have explored CNS-pancreas connectivity [ 8 ]. However, the precise neuroanatomical pathways and functional interactions between distinct brain neuronal populations and pancreatic acinar cells have not been fully elucidated. The efferent vagus nerve originates in the dorsal motor nucleus of the vagus (DMV) in the brainstem medulla oblongata and projects to visceral organs in the thoracic and abdominal cavities, including the lungs, heart, liver, gastrointestinal tract, kidneys, and pancreas [ 9 , 10 ]. We therefore aimed to establish a hypertriglyceridemia-associated acute pancreatitis (HTGP) model that is closely related to acinar cell pathology. Using pseudorabies virus (PRV) and Fluoro-Gold (FG) tracers in HTGP model mice, we detected enhanced green fluorescent protein (EGFP)-positive signals in the DMV. Subsequent monosynaptic viral tracing from the DMV revealed positive labeling in the paraventricular hypothalamic nucleus (PVN), a key regulatory center for vagal activity. These findings delineate a functional neural circuit connecting oxytocin neurons in the paraventricular hypothalamic nucleus (PVN OXT ) with digestive enzyme-secreting acinar cells. Furthermore, we demonstrate that WD repeat and FYVE domain-containing 1 ( Wdfy1 ) in PVN OXT neurons promotes digestive enzyme secretion and is essential for pancreatitis development.

Acute pancreatitis (AP) is an inflammatory disorder of the pancreas triggered by multiple etiological factors. The condition features premature activation and systemic release of pancreatic enzymes, leading to autodigestion and multi-organ complications. The leading causes of AP globally are gallstones and alcohol abuse [ 58 – 60 ]. Hypertriglyceridemia is another common cause, accounting for 5.9% of cases in North America, 2–4% of all cases in the general population, and approximately 50% of cases during pregnancy [ 61 , 62 ]. Hypertriglyceridemia is often under-recognized as an etiology in patients with AP if other etiological risk factors are present [ 63 ]. Triglycerides can cause pancreatic injury through the generation of free fatty acids (FFAs) from their hydrolysates [ 58 , 64 ]. Hypertriglyceridemia (HTG) can not only cause AP but can also modify its severity [ 65 , 66 ]. HTGP is typically triggered by severe HTG, although the precise mechanisms underlying HTG-induced HTGP remain incompletely understood. Elevated serum FFAs may play a key role, primarily through the following pathways: (1) Microcirculatory disturbance around acinar cells. High serum FFAs can damage vascular endothelial cells surrounding pancreatic acinar cells, leading to microthrombus formation in periacinar capillaries. This impairs pancreatic microcirculation, induces local acidosis, and results in premature activation of pancreatic enzymes. (2) Acinar cell metabolic dysfunction. Elevated FFAs disrupt mitochondrial function in acinar cells, reducing ATP synthesis and causing abrupt calcium overload, which further precipitates premature enzyme activation. (3) Intra-acinar inflammatory response. Excessive FFAs generate pro-inflammatory mediators, triggering localized inflammation [ 64 , 67 , 68 ]. These cascades culminate in the hallmark of AP: acinar cell death and localized or systemic inflammatory response [ 64 , 69 , 70 ]. The pancreas serves dual endocrine and exocrine functions [ 71 ]. Its endocrine component comprises islets that regulate glucose homeostasis through hormone secretion (insulin, glucagon, and somatostatin). At the same time, the exocrine part assists in the breakdown of carbohydrates, proteins, and fats by secreting a variety of digestive enzymes [ 72 ]. However, emerging evidence highlights the crucial role of neural signaling in pancreatic hormone secretion and metabolic control [ 73 ]. Deciphering the physiological functions of specific pancreatic nerves and their dysregulation in disease requires methodologies for comprehensive ganglionic mapping and accurate neural activity modulation. Jimenez-Gonzalez used tissue clearing combined with retrograde tracing and AAV-based targeting to successfully map pancreatic ganglionic inputs and reveal glucose-regulatory functions of both parasympathetic and sympathetic pancreatic innervation [ 30 ]. For more than a century, the central regulation of blood glucose and pancreatic endocrine function has been recognized, with insulin secretion considered to be under central neural control [ 19 , 74 – 77 ]. Hypothalamic glucose-sensing neurons can directly modulate insulin secretion, with this neural regulation originating from a specific subpopulation of oxytocin neurons in the PVN of the hypothalamus, via sympathetic autonomic nerve branches that innervate pancreatic islets to regulate β-cells [ 19 , 78 , 79 ]. Evidence demonstrates well-characterized glucose-sensing properties in GLP-1R-expressing neurons within the dorsomedial hypothalamic nucleus (DMH). These neurons project to the pancreas via brainstem vagal pathways, and activation of this neural circuit enhances insulin secretion, leading to blood glucose reduction [ 75 , 80 , 81 ]. These findings demonstrate that the pancreas–brain axis plays a pivotal role in regulating pancreatic endocrine function, providing crucial evidence for understanding the central neural mechanisms governing glycemic control. However, the regulation of the pancreas–brain axis in pancreatic exocrine function remains poorly understood, with both the anatomical connections and functional relationships between the brain and acinar cells still unclear. Unlike parenchymal injections, we delivered the tracing virus into the pancreatic ducts of Cela1-P2A-iCre transgenic mice, ensuring both consistent administration and acinar cell-specific targeting across all subjects in our tracing experiments [ 19 ]. Studies have used viral vectors to regulate acinar cells in pancreatitis, with ductal delivery as the primary viral vector administration route [ 82 , 83 ]. Using the Ba2001 in combination with the Cela1-P2A-iCre mouse model, we found that the regions projecting most closely and robustly to acinar cells were the PVN, DMV, and Bar, and to a lesser extent, EGFP+ cells in the Gi and Sol. By combining c-Fos mapping, we systematically screened and identified key brain regions that are associated with HTGP in the acinar-brain axis. Retrograde tracing revealed the PVN as the predominant forebrain region with direct synaptic connections to acinar cells, while the DMV (and to a lesser degree, the Sol) appeared to function as intermediate relays for PVN signals. Other nuclei, including the stHy, Amb, LC, Bar, and Gi, may also contribute to acinar secretory regulation. Additional neuroanatomical tracing is required to delineate the hierarchical connectivity between these nuclei and acinar cells. Notably, a discrepancy was observed between the sparse neural connectivity from acinar cells to the Sol revealed by PRV retrograde tracing and the robust c-Fos activation in this nucleus in the HTGP model. This suggests that the activation of the Sol may arise from mechanisms other than direct synaptic input from acinar cells. Given that the Sol is a primary brainstem target of vagal afferents and plays a critical role in integrating peripheral inflammatory signals, it is plausible that the pronounced c-Fos expression in the Sol reflects systemic inflammation-induced vagal activation rather than acinar cell-derived signals [ 84 ]. Our retrograde tracing results yielded different innervation time courses of pancreatic cells when compared with a previously published study [ 19 ]. The two studies differ substantially in several key aspects that could account for their divergent results, including the distinct viral tracers and target cell populations employed. A key methodological difference lies in the analysis time points. Papazoglou et al. examined tissues at 3 days post-viral administration, compared with our 5 days endpoint. The 3-day time point takes into account the highly infectious nature and rapid propagation of such viruses in the brain. However, extended incubation to 5 days enhances EGFP labeling intensity and distribution throughout the brain and hypothalamic regions [ 8 ]. FG retrograde tracing following pancreatic duct injection revealed distinct labeling of ChAT+ neurons in the DMV, demonstrating direct acinar cell-to-DMV ACh neuroanatomical connectivity. Recombinant retrograde viral tracing with rAAV2-retro-hSyn-EGFP injected into the DMV revealed the largest fraction of labeling of OXT+ neurons in the PVN, demonstrating direct PVN OXT -DMV ACh neuroanatomical connectivity. To validate the functional circuitry, we chemogenetically stimulated PVN OXT neurons, which induced secretion of AMY and LPS and ultimately increased the inflammatory response in HTGP, consistent with our findings from DMV ACh neurons activation. When we chemogenetically activated the PVN OXT -DMV ACh axis, the inflammatory response in HTGP was exacerbated. Our findings revealed a functional brain-to-acinar cell transneuronal circuit that ultimately induced the inflammatory response in HTGP. We performed unbiased transcriptomic analyses, which revealed neuronal circuit targets specific to acinar cells. Our analysis revealed Wdfy1 , a differentially expressed gene in OXT neurons, as a focal point. KEGG and GO enrichment analyses revealed that Wdfy1 is involved in the TLR3 and TLR4 signaling pathways. Wang et al. suggested that ω-3 fatty acids mitigate early-stage severe AP (SAP) inflammation via TLR4/NF-κB p56 pathway suppression, while HFD exacerbates intestinal barrier injury in SAP rat models through TLR4-RIP3 activation [ 85 , 86 ]. Building on these findings, future investigations could elucidate the functional contributions within this neural pathway. We demonstrate that the PVN OXT -DMV ACh -acinar cell axis, in which Wdfy1 plays a key role, underlies the exacerbation of pancreatitis severity in the mouse model, providing a mechanistic explanation for the observed pancreatic inflammation. Evidence suggests that the pancreas receives sympathetic innervation originating from the intermediolateral column of the spinal cord at the T6-L2 segments, with postganglionic fibers primarily distributed to the islets and vasculature, while innervation of the exocrine pancreas is relatively sparse [ 73 ]. Parasympathetic innervation, originating from the dorsal motor nucleus of the vagus, relays at intrapancreatic ganglia via the vagus nerve, with postganglionic fibers widely distributed to both the islets and the exocrine pancreas, thereby facilitating exocrine pancreatic secretion [ 2 , 30 , 87 ]. In addition, the pancreas is also innervated by sensory nerves arising from the dorsal root ganglia and nodose ganglia, as well as by enteropancreatic nerves extending from the gastrointestinal tract [ 88 ]. Our data confirm that the final relay of this pathway involves intrapancreatic cholinergic ganglia. Of note, the signaling of ACh exhibits dual characteristics. On the one hand, postganglionic parasympathetic fibers innervating the pancreas act on mAChR, predominantly the M3 subtype, on acinar cells via acetylcholine, thereby promoting pancreatic exocrine secretion. On the other hand, vagal cholinergic signaling can activate the cholinergic anti-inflammatory pathway by stimulating α7 nicotinic acetylcholine receptors (α7nAChR) on immune cells, thereby exerting an anti-inflammatory effect and conferring tissue protection. Our findings indicate that in HTGP models, the PVN OXT -DMV ACh axis preferentially drives pro-inflammatory effects through M3 signaling. This observation underscores the complexity of neural regulation in pancreatic inflammation. Furthermore, the pro-inflammatory role of PVN OXT neurons observed in our study appears to contradict the well-documented anti-inflammatory response of oxytocin reported in a previous study [ 43 , 89 ]. Those established effects are primarily mediated through systemic hormonal actions: OXT released into the circulation acts on OXT receptors in various organs to suppress pro-inflammatory pathways such as NF-κB in immune cells. In our study, the PVN OXT neurons exert their function through a specific synaptic neural circuit. The ultimate effect of this pathway is achieved by the direct action of ACh released by vagal cholinergic neurons on pancreatic acinar cells. This might explain the different effects of OXT. In addition, given the well-established sexual dimorphism in oxytocinergic signaling, where neural regulation in females can be confounded by fluctuating estrogen levels, we exclusively employed male mice in this study. This strategic approach was critical to mitigate the potential confounding effects of cyclical hormonal variations, thereby allowing us to unequivocally attribute our functional findings to the identified neural circuitry rather than to endocrine interference. By establishing this foundational framework in males, we performed the first mapping of a previously undefined hypothalamic-acinar cell pathway, which provides an essential baseline for future investigations into sex-specific differences in this neural axis. To further elucidate the cellular sources of elevated serum inflammatory mediators following PVN OXT neuron activation, we investigated the temporal dynamics of mediator secretion by macrophages and acinar cells at different stages of inflammation. Previous studies have demonstrated that acinar cell activation originates from cellular stress and serves as the initiating event in the inflammatory response of pancreatitis, with acinar cells entering an activated state immediately after ceruletide injection and recruiting immune cells through the release of signaling molecules. At this stage, early activated tissue-resident macrophages can be observed. However, their presence largely reflects the cumulative effect following ceruletide injection. Macrophages, as a highly heterogeneous cell population, are capable of rapidly altering their activation status and function in response to local microenvironmental signals [ 90 ]. Activated by recruitment signals from acinar cells, macrophages reach peak activation at 24 h, becoming the principal source of elevated circulating inflammatory mediators and thereby driving the propagation of systemic inflammatory response [ 91 ]. In the present study, pancreatic tissues were collected at 18 h after ceruletide injection to assess inflammatory mediators secretion, while PVN OXT neurons were activated at 2 h after CNO administration. Accordingly, we selected two time points, the 2-hour (early phase) and the 18-hour (late phase), to evaluate the colocalization of inflammatory mediators with acinar cells and macrophages, respectively. Consistent with previous reports, at the 2-hour time point, acinar cells exhibited extensive colocalization with IL-1β, IL-6, and TNF-α, with the number of colocalized cells being significantly higher than that observed for macrophages. At the 18-hour time point, the number of macrophages colocalized with IL-1β, IL-6, and TNF-α was markedly increased and significantly higher than that of acinar cells colocalized with these mediators. These findings indicate that the elevation of serum inflammatory mediators is not attributable to a single cell type. This temporal cascade provides a cellular basis for understanding how PVN OXT neuron activation ultimately leads to exacerbated inflammation in HTGP. Our study provides anatomical and functional evidence of a well-defined neuronal circuit, and to our knowledge, no studies have characterized the central control of pancreatic exocrine cells in this model system. However, given species-specific differences in pancreatic neuroanatomy, the translational relevance of our findings to humans requires further investigation. Current constraints, particularly the inability to employ viral vector-based circuit manipulation in humans, represent a fundamental limitation in neural circuit research. To overcome this barrier, our future work will focus on establishing a human-relevant experimental platform using co-cultures of human pancreatic organoids and OXT neurons. Within this system, viral vectors will be applied to modulate the PVN OXT -DMV ACh neural pathway. Furthermore, we will validate the efficacy and safety of targeting Wdfy1 in higher animal models, such as non-human primates. Although selectively identifying neurons projecting to acinar cells through retrograde tracing techniques would be the ideal approach, current tracing tools have inherent limitations, such as uneven infection efficiency leading to incomplete neuronal labeling. Furthermore, while our study focuses on the role of PVN OXT neurons in regulating acinar cell secretory function and their impact on inflammatory response, we cannot exclude potential contributions from other PVN neuronal populations or neurons in additional brain regions. Currently, treatment for HTGP is primarily supportive, lacking targeted therapies for precise intervention. Wdfy1 , as a potential target in OXT neurons within the PVN, offers a promising avenue for the development of targeted inhibitory drugs. By specifically reducing the hyperexcitability of these neurons, such an approach could attenuate abnormal stimulation of the pancreas while preserving normal functions of other neurons, thereby minimizing potential side effects. Future studies could employ more sophisticated methodologies, including cell-type-specific labeling, advanced circuit tracing, and functional manipulation techniques, to comprehensively characterize this hypothesized PVN-acinar cell axis. We propose that such transneuronal circuits represent critical hubs within integrated multi-organ networks. Elucidating analogous pathways will advance understanding of central neural regulation in both pancreatic physiology and pathology.

Using the HTGP mouse model, we describe a functional transneuronal circuit connecting the hypothalamus to acinar cells in mice. This circuit originates from a subpopulation of oxytocin neurons in the PVN OXT , and reaches the exocrine pancreas via the acetylcholinergic neurons in the DMV ACh to innervate acinar cells. Silencing PVN OXT neurons suppresses digestive enzyme secretion and inhibits inflammatory response in HTGP. Conversely, stimulation of these neurons induces inflammatory response by dysregulating secretory pathways in acinar cells. Single-cell RNA sequencing revealed that Wdfy1 plays a critical role in acinar cell-mediated inflammatory response, and its function is necessary for the PVN OXT -DMV ACh axis.

Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2.