Artesunate ameliorates acute pancreatitis by enhancing p62-mediated zymophagy through non-canonical NRF2 activation

Abstract

Objective

Zymophagy is a selective autophagy that eliminates activated trypsinogen during acute pancreatitis (AP). However, the upstream regulatory mechanisms and therapeutic potential of zymophagy remain poorly understood. This study investigated the mechanisms of zymophagy activation and its protective role in AP, using artesunate (AS) as a pharmacological tool.

Methods

AP models were induced by cerulein combined with lipopolysaccharide and evaluated by histopathological examination and enzymatic assays. Gene expression was knocked down with specific shRNA or siRNA, and expression levels were detected by real-time polymerase chain reaction (PCR). Zymophagy and its regulatory mechanisms were characterized through multiple experiments, and molecular interactions were validated by co-localization, immunoprecipitation (IP), chromatin immunoprecipitation (ChIP), thermal shift assay, and molecular docking.

Results

AS mitigated AP in both in vivo and in vitro models by inhibiting trypsinogen hyperactivation in acinar cells. This protective effect was mediated by enhanced Atg5-mediated zymophagy activation via a p62-dependent and vacuole membrane protein 1 (VMP1)–ubiquitin-involved pathway, which promotes the targeted engulfment and degradation of trypsinogen. Mechanistically, AS binds the PB1 domain of p62, strengthens its interaction with Kelch-like ECH-associated protein 1 (Keap1), triggers ubiquitin–p62-mediated autophagic degradation of Keap1, facilitates the nuclear translocation of nuclear factor erythroid 2-related factor 2 (NRF2) and its selective binding to the antioxidant response element in the p62 promoter, and ultimately upregulates p62 transcription.

Conclusion

Zymophagy is activated via a VMP1–ubiquitin–p62-dependent and Atg5-mediated canonical autophagy pathway. AS ameliorates AP by enhancing p62-mediated zymophagy via non-canonical NRF2 activation, which provides mechanistic evidence for therapeutic targeting of acinar zymophagy in AP management.

Graphical Abstract

Graphical Abstract.

AS potentiates the p62–Keap1 interaction by binding to the PB1 domain of p62, thereby promoting the p62-dependent autophagic degradation of Keap1, triggering non-canonical NRF2 activation to upregulate p62 mRNA expression, and ultimately enhancing p62-mediated pancreatic zymophagy. The figure was created in BioRender.com.

Introduction

Acute pancreatitis (AP) is an inflammatory disorder of the exocrine pancreas caused by the abnormal activation of pancreatic enzymes due to pancreatic trypsinogen activation and subsequent autodigestion [1]. AP is typically a self-limiting disease, yet ~15%–20% of patients will progress to severe acute pancreatitis (SAP), which is associated with high mortality, especially when triggered by pathogenic factors such as gallstones, alcohol, and infection [2–4]. As of now, no targeted therapies have been approved for the treatment of AP [2]. Existing pharmacological agents, including protease inhibitors, growth inhibitors, and anti-inflammatory drugs, have proven ineffective in preventing the progression of AP to SAP [2]. It is therefore imperative to identify efficacious early intervention drugs.

Autophagy plays a pivotal role in maintaining cell homeostasis by sequestering damaged organelles, misfolded proteins, and other redundant components into autophagosomes, ultimately forming autolysosomes for degrading cargoes via the lysosomal pathway [5]. Accumulating evidence indicates that the imbalance of pancreatic autophagy exacerbates AP [6]. Excessive activation of trypsinogen to trypsin is a critical early event in AP, leading to pancreatic cell death (both necrosis and apoptosis) and disease exacerbation [7]. Recent studies report that zymophagy, a selective autophagy process targeting overactivated trypsinogen for degradation, is activated during early stages of AP and exhibits protective effects against cell death and AP progression [8, 9]. Therefore, zymophagy has emerged as a novel target for future AP therapy, as it facilitates the isolation and degradation of overactivated trypsinogen in the endocrine pancreas.

Selective autophagy relies on adaptor proteins such as p62/SQSTM1 (hereafter p62), nuclear dot protein 52 kDa (NDP52), and optineurin to recognize ubiquitinated cargoes and initiate autophagosome formation [10]. In 2011, a study identified p62 as the autophagy adapter of zymophagy in both experimental and human pancreatitis, with ubiquitin (Ub) and vacuole membrane protein 1 (VMP1) participating in the formation of autophagosome of zymophagy (hereafter zymophagosome) [9]. Recently, another indicated that Trim33, an E3 ligase of Ub system, upregulated p62 and VMP1, thereby enhancing zymophagy to reduce trypsinogen activation and promote acinar cell recovery [8]. Although the repertoire of functional proteins involved in zymophagy requires further investigations, current studies have revealed the critical role of p62 in zymophagosome formation.

In selective autophagy, p62 is primarily recognized for its role in directing ubiquitinated proteins (i.e. aggrephagy and zymophagy) or organelles (i.e. mitophagy and lysophagy) to the growing autophagosome [10, 11]. Specifically, p62 interacts with ubiquitinated cargoes via its ubiquitin-associated domain (UBA) and recruits microtubule-associated protein 1 light chain 3 (LC3) through its LC3-interacting region (LIR) to initiate autophagosome formation [12]. Additionally, p62 interacts with other proteins with diverse functions, such as binding to Kelch-like ECH-associated protein 1 (Keap1) with a Keap1-interacting region (KIR) for its further degradation via autophagy [12]. Notably, Keap1 is a negative regulator of nuclear factor erythroid 2-related factor 2 (NRF2); thus, p62-Keap1 binding leads to non-canonical NRF2 activation, which in turn upregulates a series of NRF2-primed genes, such as autophagy-related genes including p62, Ndp52, and Gabarapl1 [12, 13]. In addition, this interaction establishes a positive feedback loop regulating p62 transcription. However, whether modulation of this triplet loop can enhance p62-zymophagy and protect against AP remains unknown.

Previously, we reported that artesunate (AS), a common antimalarial drug, exerted a protective effect in SAP rats by suppressing AP-associated digestive enzyme activity and inflammation [14]. In preliminary experiments, we found that AS ameliorated AP in mice, potentially linked to the marked increase of zymophagosome-like vesicles induced in the AP pancreas. In this study, we delve deeper into the protective role of AS in AP mice by examining zymophagy regulation, with a focus on the modulation of the non-canonical NRF2 activation mechanism.

Materials and methods

In vivo AP model and treatment

Specific pathogen-free KM mice (4–6 weeks old) and C57BL/6N mice (8–12 weeks old) were obtained from the Experimental Animal Center of our university. Mice were housed at 24°C on a 12-h light/dark cycle with free access to water and standard laboratory chow. Male KM or C57BL/6N mice were fasted for 8 h (water allowed), and AP was induced by six intraperitoneal injections (i.p.) of cerulein (CR; 50 μg/kg/h; Bidepharm, Shanghai, China) at 1-h intervals, followed by a final i.p. injection of lipopolysaccharide (LPS; 10 mg/kg; Sigma, St. Louis, MO). To evaluate the protective effect of AS, KM mice were randomly divided into five groups (n ≥ 6; Fig. 1A). At every recurrence cycle, AP mice received two intramuscular injections (i.m.) of AS (0, 2.5, 5.0, and 7.5 mg/kg; Nanyao Ltd, Guangxi, China) at 0 and 4 h post-AP induction, or two intravenous injections (i.v.) of Ulinastatin (UTI) (200 U/kg; Tianpu Pharmaceutical, Guangdong, China), a commonly used protease inhibitor in AP. Sham group mice received normal saline. The dosages of AS and UTI were matched to the clinical dosages (AS, 0.8–2.4 mg/kg/day; UTI, 1500 U/kg/day) using the body surface area conversion method.

Figure 1.

The protective role of AS in experimental AP. (A–D) Experimental AP in male mice was induced by CR (50 μg/kg/h; 6 injections; i.p.) followed by LPS (10 mg/kg; i.p.), after which mice were treated twice with AS (2.5, 5.0, 7.5 mg/kg; i.m.) or UTI (200 U/kg; i.v.) as shown in Panel A. Photographs and HE staining of pancreas are shown in B, bar = 50 μm. Histograms depict the pancreatic index and histology score (n = 6). (E–G) Levels of serum amylase, pancreatic trypsin, and lipase in AP mice (n = 6). (H) AR42J cells seeded in a 96-well plate (2 × 10⁴ cells/well) were treated with CR (0.1, 1, 10 nM) and LPS (0.05, 0.1, 0.5, 1 μg/ml) for 24 h; supernatants were collected to measure amylase activity (n = 5). (I, J) AR42J cells cultured as in Panel H were treated with 10 nM CR and 1 μg/ml LPS (hereafter referred to as AP) with or without AS (2 μg/ml) for up to 48 h, or with AP plus AS (0.5, 1, 2 μg/ml) for 24 h; supernatants were used to measure amylase activity (n = 5). (K) AR42J cells cultured as in Panel H were treated with AP ± AS (2 μg/ml) for 24 h; supernatants were analyzed for trypsinogen-activating peptide (n = 5). (L) AR42J cells treated as in Panel K were incubated with a BZiPAR fluorescent substrate for 0.5 h and then used to visualize trypsin activity. **P < 0.01, *P < 0.05, ^#P > 0.05.

At the end of the in vivo experiment, mice were euthanized under deep anesthesia (by isoflurane), followed by exsanguination via the orbital vein 18 h post-AP induction. Pancreas and serum samples were collected. Pancreas were (1) homogenized in PBS for immunoblotting (IB) or enzymatic assays; (2) fixed in 4% neutral paraformaldehyde (m/v) for histological observations; (3) fixed in 2.5% glutaraldehyde (m/v) for transmission electron microscopy (TEM) observation; and (4) frozen for immunofluorescence (IF) staining. Serum amylase, pancreatic lipase, and trypsinogen activity peptide (TAP) were measured using the corresponding commercial assay kits (Jiancheng, Jiangsu, China).

Histological observations

Morphological changes in the pancreas were assessed by calculating the pancreatic index (pancreas weight/tibia length; mg/cm). Paraffin sections (5 μm thickness) of mouse pancreas treated as indicated were used for subsequent hematoxylin–eosin (HE) staining, or immunohistochemistry (IHC) using an anti-NRF2 antibody (Sangon, Shanghai, China; RRID:D221053; 1:200) using an IHC kit (Boster, Hubei, China). Histological images were captured using a light microscope and analyzed using ImageJ (http://rsb.info.nih.gov/ij/). The histological score was calculated by observing edema, vacuolization, inflammatory cell infiltration, and necrosis according to a previous report [15].

In vitro AP model and treatment

AR42J cells (ATCC, Manassas, VA, RRID:CVCL_0143) were cultured in F12K medium (Boster, Nanjing, China) supplemented with 20% fetal bovine serum (Gibco, Grand Island, NY) and antibiotics (100 U/ml penicillin and 0.1 mg/ml streptomycin). Cells were treated with 10 nM of CR plus 1 μg/ml of LPS (hereafter termed AP) for 1–48 h as indicated to induce the in vitro AP model in the presence of AS (2 μg/ml), bafilomycin A1 (Baf; 10 nM; MedChem Express, Monmouth Junction, NJ), rapamycin (Rapa; 10 nM; MedChem Express), and/or 3-methyladenine (3-MA; 4 mM; MedChem Express). Supernatants were collected for detecting amylase activity as described above. To observe trypsinogen activation, AR42J cells were treated as indicated and the levels of trypsinogen-activating peptide in supernatants were detected according to the user’s manuals, or cells were then incubated for 20 min in HEPES buffer (pH 7.5) containing 100 μM of rhodamine 110 coupled bis-(CBZ-L-isoleucyl-L-prolyl-L-arginine amide) dihydrogen chloride (BZiPAR; Invitrogen, Carlsbad, CA), and the catalytic activity was determined using a fluorescence microscope or a microtitre plate reader.

Transmission electron microscopy

In vivo samples treated as indicated were fixed in 2.5% (m/v) glutaraldehyde, and then prepared for observation of subcellular changes using an HT7700 TEM (HITACHI, Tokyo, Japan). Specifically, the double-layered cargo-containing vesicle was defined as the autophagosome, within which the trypsinogen granule-positive vesicle was termed the zymophagosome.

RNA interference

Recombinant adeno-associated virus type 9 (AAV9) can efficiently infect a variety of cells and tissues, including brain and pancreas. For p62 knockdown in vivo, C57BL/6N mice (5 weeks old) were infected with AAV9 containing p62 shRNA or negative control AAV9 (2.5 × 10¹¹ vg/kg; i.p.; Genemine, Chongqing, China). After 63 days post-injection, p62 knockdown was confirmed by IB, and then mice were randomly divided into sham, AP, and AS groups for further treatment.

In vitro knockdown was performed using specific siRNA. The sequence of siRNA for rat Vmp1, p62, Atg5, and Nrf2 was designed and synthesized by Sangon (Supplementary Table 1). siRNAs were transfected into AR42J cells using a liposome method. Briefly, AR42J cells were seeded in 6-well plates (2.0 × 10⁵ cells/well) or 96-well plates (2.0 × 10⁴ cells/well) and cultured to 70% confluence, and then transfected with siRNA using Lipofectamine 3000 (Invitrogen) according to the user’s manual. After overnight renewal, cells were treated as indicated, then cells and supernatants were harvested for IB assays and amylase detection, respectively.

Real-time PCR

AR42J cells were treated as indicated and harvested for total RNA extraction using RNAiso Plus (Takara, Liaoning, China) according to the manufacturer’s instructions. Subsequently, 1 μg of RNA was reverse-transcribed into cDNA using a PrimeScript™ RT Reagent Kit (Takara). Finally, cDNA samples were used as templates for real-time polymerase chain reaction (PCR) with specific primers (Supplementary Table 1; Sangon) using SsoAdvanced SYBR Green Super Mix (Bio-Rad, CA, USA) under a CFX PCR system (Bio-Rad). The mRNA expression was analyzed and calculated using the 2^−△△Ct method by normalizing to β-actin.

Immunofluorescence

To assess zymophagy in vivo, frozen or paraffin sections of the pancreas were rehydrated in ethanol solutions, washed with 10 mM PBS (pH 7.2), incubated in 0.5% BSA (m/v) for 30 min, probed overnight with a rabbit anti-Trypsinogen antibody (Abcam, Cambridge, MA; Cat#ab166898; RRID:AB_3712100; 1:100) plus with a mouse anti-LC3B antibody (Cell Signaling, Beverly, MA; 83 506; RRID:AB_2800018, 1:200), and then stained with the corresponding Alexa Fluor 488 (AF488) and AF555 conjugated secondary antibodies (Invitrogen; 1:200). Cell nuclei were stained with 5 μg/ml DAPI. Slides were observed using an LSM900 confocal laser microscope (Zeiss, Jena, Germany). Images were analyzed, and the colocalization (Pearson’s correlation coefficient) was calculated using ImageJ.

For in vitro zymophagy detection, AR42J cells were seeded on glass slides in a 24-well plate (1 × 10⁵ cells/well), treated as indicated, fixed in 4% neutral paraformaldehyde (m/v), blocked in 1% BSA (m/v) containing 0.1% Triton X-100 (v/v), and probed as described for in vivo zymophagy observation. To detect p62-related zymophagy in vitro, AR42J cells were seeded on glass slides, transfected with an Ad-GFP-LC3B kit (Beyotime, Shanghai, China), treated as described, and probed with a rabbit anti-Trypsinogen antibody (Abcam; Cat#ab166898; RRID:AB_3712100; 1:100) and a mouse anti-p62 antibody (Abcam; Cat#ab56416; RRID:AB_945626; 1:200), followed by staining with the corresponding AF555 and AF647-conjugated secondary antibodies (Invitrogen; 1:200). To detect autophagic flux, AR42J cells were seeded on glass slides, transfected with a pH-sensitive RFP-GFP-LC3 kit (Thermo Fisher, Waltham, MA), and treated as indicated. Autophagosomes (yellow dots) and autolysosomes (red dots) were observed by confocal imaging and analyzed using ImageJ.

To detect colocalization of trypsinogen with VMP1, Ub, or p62, AR42J cells seeded on glass slides were treated as indicated and probed in a similar manner for observing in vitro zymophagy. Specifically, trypsinogen was probed with a mouse anti-Trypsinogen (Sigma–Aldrich; Cat#SAB1402883; RRID:AB_10738967; 1:200) antibody followed by an AF555-conjugated secondary antibody (Invitrogen; 1:200). Other proteins were probed by rabbit antibodies against VMP1 (Sangon, Cat#D262380; RRID:AB_3712104; 1:200), Ub (Cell Signaling; Cat#3936; RRID:AB_331292; 1:200), or p62 (Cell Signaling; Cat# 5114; RRID:AB_10624872; 1:200) followed by an AF488-conjugated secondary antibody (Invitrogen; 1:200). To detect colocalization of Keap1 with Ub, p62, or LC3B, AR42J cells seeded on glass slides were treated as indicated and probed as described above. Specifically, Keap1 was probed with a rabbit anti-Keap1 antibody (Cell Signaling; Cat#4678; RRID:AB_10548196; 1:200), followed by an AF555-conjugated secondary antibody (Invitrogen; 1:200), and Ub (Beyotime; Cat#AG3157; RRID:AB_3712105; 1:200), p62 (Abcam; Cat#ab56416; RRID:AB_945626; 1:200), or LC3B (Zen-bio; 206 019, RRID:AB_3675210; 1:200) was probed by corresponding mouse antibodies followed by an AF488-conjugated secondary antibody (Invitrogen, Cat#A-11029; RRID:AB_2534088; 1:200). The colocalization was observed by confocal imaging and analyzed using ImageJ.

To detect the intracellular distribution of NRF2, AR42J cells were seeded on glass slides and treated as indicated. The cells were then probed with a rabbit NRF2 antibody (Sangon; Cat#D121053; RRID:AB_3712104; 1:200) and processed similarly to in vitro zymophagy. Specifically, mean fluorescence intensity of NRF2 in the cytoplasm and nucleus was measured in Zen packages (Zeiss).

Immunoblotting and immunoprecipitation

Total protein lysates were extracted from AR42J cells or pancreas using IP Lysis Buffer (Beyotime) for IB or immunoprecipitation (IP). Cytoplasmic and nuclear lysates of AR42J cells were extracted using a Nuclei Extraction kit (KeyGEN; Shanghai, China). Equal amounts of protein samples were separated by SDS-PAGE, and the protein bands were then transferred to polyvinylidene fluoride (PVDF; 0.2 μm) membrane, blocked with 5% skim milk (m/v), and probed with the designated primary antibodies and horseradish peroxidase-conjugated secondary antibodies (Thermo Fisher). Chemiluminescence imaging was performed under a ChemiDoc™ Touch imaging system using the SuperSignal substrate and analyzed by the ImageLab packages (Bio-Rad). Rabbit antibodies against p62 (Cat# 5114; RRID:AB_10624872), LC3B (Cat#83 506; RRID:AB_2800018), Keap1 (Cat#4678; RRID:AB_10548196), and Atg5 (Cat#12 994; RRID:AB_2630393) were purchased from Cell Signaling Technology (1:1000), rabbit antibodies against Vmp1 (Cat#D262380; RRID:AB_3712104; 1:500), NRF2 (Cat#D121053; RRID:AB_3712104; 1:500), and Prss1 (Cat#D154078; RRID:AB_3712108; 1:500) were purchased from Sangon (1:500), and mouse antibodies against Histone H3 (Cat#AF0009; RRID:AB_3712106), Hsp90 (Cat#AH732; RRID:AB_3712106), and Ub (Cat#AG3157; RRID:AB_3712105) were purchased from Beyotime (1:1000). IP experiments were performed with Protein A/G magnetic beads (Thermo Fisher). Briefly, total, cytoplasmic, or nuclear lysates were incubated with the indicated primary antibodies overnight at 4°C. Immunocomplexes were then collected using Protein A/G beads and subjected to IB.

Chromatin IP

Chromatin IP (ChIP) was performed according to the protocol of an Agarose ChIP kit (Pierce, Rockford, IL). Protein-DNA complexes were cross-linked with 1% formaldehyde (v/v) supplemented with 10% glycine solution (v/v). Chromatin was fragmented using micrococcal nuclease (25 U/ml). Fragmented chromatin was isolated using a ChIP-grade rabbit anti-NRF2 antibody (Cell Signaling; Cat#20 733; RRID:AB_2820200). The ChIP products were amplified by real-time PCR (ChIP-qPCR) using primers specific for the antioxidant response element (ARE) site within the promoter region of p62, Ndp52, or Gabarapl1 (Sangon; Supplementary Table 1). Input samples were analyzed using control primers specific for the Gapdh promoter provided by the kit. Data were calculated using the 2^−△△Ct method and normalized to Gapdh.

Thermal shift assay

AR42J cells in logarithmic growth phase were lysed by NP-40, the supernatant was incubated with 100 μM AS for 30 min, followed by heating at 37–55°C (2°C increments) for 10 min. Protein thermal stability was assessed by IB, and thermal melting curves were generated. T_m values (indicating AS binding to p62, Keap1, or NRF2) were calculated using SPSS Statistics 24. △T_m (T_{m AS} − T_{m control}) value ≥2°C with statistical significance (P < 0.05) defined binding establishment. Three independent control replicates were included.

Molecular docking

The crystal structure of p62 (NO. 2KKC) was obtained from the PDB database. The homologous protein model of Keap1 (NO. AF-P57790-F1) was obtained from AlphaFold. AS was used as the ligand, and p62 or Keap1 was used as the receptor in docking using SWISSDOCK (www.expasy.org/). The results were analyzed and represented in PyMOL.

Statistical analysis

Data are shown as the mean ± standard deviation (SD). The experimental data were analyzed using ImageJ and SPSS Statistics 24. One-way analysis of variance was used for multiple group comparison. A two-tailed unpaired Student’s t-test was used to determine statistical significance between groups. Statistical significance is indicated as follows: *P < 0.05; **P < 0.01, ^#P > 0.05.

Results

The protective role of AS in experimental AP

Previously, we found that AS exerted a protective effect on SAP rats subjected to retrograde injection of 3.5% sodium taurocholate (ST) into the biliopancreatic duct [14]. In this study, we further investigated the pharmacological activities of AS in the mouse model of experimental AP treated with CR plus LPS as indicated (Fig. 1A). No mortality or toxicity of AS was detected in in vivo experiments. In the AP group, the pancreatic changes were mainly characterized by interstitial edema, inflammation, and microhemorrhages (Fig. 1B). As expected, these morphological changes and pathological changes were mitigated by AS treatments in a dose-dependent manner (5.0–7.5 mg/kg; Fig. 1B–D). In addition, the pancreas index and pancreas histology score were significantly reduced (Fig. 1C–D). Moreover, serum amylase activity, pancreatic trypsin level, and lipase activity were also increased in the AP group; but these decreased significantly with AS treatment (Fig. 1E–G). Meanwhile, the therapeutic effect of UTI (200 U/kg) used as a positive control on AP mice is consistent with that of AS (5.0–7.5 mg/kg). Compared with UTI, although AS did not demonstrate superior efficacy in reducing amylase activity or trypsin levels in pancreatitis models, it is noteworthy that the reductions in pancreatic index and pancreatic histology score achieved with high-dose AS (7.5 mg/kg) were significantly greater than those achieved with UTI. Taken together, these data suggest that AS exerts a significant protective effect in AP mice and may even provide superior protection compared with UTI in certain parameters.

The protective effect of AS on the AP model in vitro was also examined in the acinar cell line AR42J treated with CR plus LPS. We found that the amylase activity induced by CR and LPS increased in a dose-dependent manner, and that the appropriate concentrations to induce AP were 0.1–10 nM for CR and 0.05–1 μg/ml for LPS (Fig. 1H). Furthermore, AR42J cells were treated with 10 nM of CR plus 1 μg/ml of LPS (hereafter termed AP), amylase activity peaked at 24 h in the AP group, whereas AS treatment caused a sustained decline (Fig. 1I). Similar to the in vivo results, we also found the AP-induced increase in amylase activity in AR42J was inhibited by AS in a dose-dependent manner and was completely reduced by 2 μg/ml AS (Fig. 1J). Therefore, 24 h was selected as the key time point for evaluating AP, and 2 μg/ml was used as the dosage of AS for subsequent experiments.

In addition to amylase activity, we also examined whether AS directly influenced trypsinogen activation. As expected, AS strongly inhibited TAP, which indicated trypsinogen activation (Fig. 1K). Moreover, BZiPAR, a fluorescent and cell-permeable protease substrate indicating trypsinogen activation, was used here. Consistently, AS treatment showed complete inhibition of BZiPAR signal in live AR42J cells compared to the AP group (Fig. 1L). Overall, these data suggest that AS inhibits trypsinogen activation in acinar cells, which is associated with its protective effect on AP mice.

AS enhances acinar zymophagy in AP

Autophagy maintains cellular homeostasis during AP [1]. We used multiple methods to observe autophagy changes in AP pancreas with or without AS treatment. The experiments started with TEM scanning to observe changes in subcellular morphology of acinar cells, and the results showed that AP slightly increased the number of total autophagosomes, of which only a few were zymophagosomes (the maximum percentage was 22%) characterized by a double-layer surrounded and trypsinogen granule-containing vesicle, and the number of either autophagosome or zymophagosome peaked at 18 h (Fig. 2A–B). Notably, AS treatment continuously induced a marked increase of both autophagosome and zymophagosome (Fig. 2A–B). Importantly, AS-induced increase of autophagosome predominantly reflected zymophagosome accumulation, peaking at 64% among whole autophagosomes and representing a five-fold increase over the AP group at the same time. Moreover, the number of zymophagosome in AS treatment was peaking at 6 h, which was earlier than that of autophagosome (Fig. 2A–B). Consistently, the IB results from AP pancreatic lysate showed that AS treatment indeed upregulated the protein level of LC3B-II while reducing activated trypsinogen (Tryp-A) compared to the AP group (Fig. 2C). Taken together, these data indicate that AS treatment remarkably increases acinar zymophagy in AP.

Figure 2.

AS enhances acinar zymophagy in AP. (A, B) Experimental AP was induced in mice for 6, 18, and 24 h with or without AS treatment (7.5 mg/kg). Fresh pancreas was processed for TEM to observe subcellular changes (bar = 2 μm). Quantitative analysis is shown in B (n = 10 views). **P < 0.01, *P < 0.05 vs. AP. (C) Immunoblotting (IB) analysis of the protein levels of LC3B-I/II and total/activated trypsin (Tryp-T/A) in AP pancreas (0 and 18 h) obtained from Section A (n = 3). (D) AR42J cells were treated with AP for up to 24 h. Total cell lysates were used to detect protein levels of Atg5 and LC3B-II by IB (n = 2). (E) AR42J cells were treated with AP with or without the presence of AS (2 μg/ml) for 4 h. Protein levels were detected by IB as described in Panel D. (F, G) AR42J cells grown on glass slides were treated as shown in Panel E, and then cells were fixed and used for IF staining as shown to observe the expression of trypsinogen (Tryp) and LC3 by confocal imaging. Tryp-LC3 colocalization and residual LC3 dots were calculated by ImageJ (n = 5; bar = 10 μm). Protein ratios shown under the bands are representative of three or two independent experiments. **P < 0.01, *P < 0.05, and ^#P > 0.05.

To further characterize autophagic responses to AP and AS in vitro, studies were performed in AR42J treated with CR plus LPS. Similar to the in vivo studies, we first observed temporal changes in the response of the autophagy machinery to AP. The IB results showed that AP induced a weak increase in Atg5 levels, but a significant increase in LC3B-II levels, and the peak occurred at 4 h for both (Fig. 2D). Thus, 4 h was chosen as the key time point for studying autophagy changes in vitro. IB then re-examined whether AS affected the expression of these two. Although AS treatment showed no effect on Atg5 levels, it was able to remarkably increase LC3B-II expression (Fig. 2E), which was very similar to its activity in enhancing zymophagy in AP mice. More intuitively, IF staining showed that AP induced both zymophagy (LC3 merged with Tryp) and other types of autophagy (residual LC3; Fig. 2F–G). However, AS treatment only increased zymophagy but had no significant effect on the number of remaining LC3 puncta, suggesting that the upregulation of autophagy by AS was due to activated zymophagy rather than other types of autophagy. Collectively, both in vivo and in vitro results suggest that AS treatment further induces massive zymophagy in acinar cells, which is probably related to its protective role in AP.

AS ameliorates AP by enhancing Atg5-dependent zymophagy activation

To further characterize the zymophagy enhancement effect by AS in the AP pancreas, acinar cells transfected with a pH-sensitive GFP-RFP-LC3 construct were used here. In IF experiments, we found that the number of autophagosomes (yellow dots resulting from GFP-LC3 and RFP-LC3), but not autolysosomes (red dots indicating RFP-LC3), was increased by AS treatment compared to the AP group (Fig. 3A–B). In contrast, AS treatment increased the number of autolysosomes in the presence of bafilomycin A1 (Baf), an autophagy flux inhibitor that blocks autolysosome degradation. These data suggest that AS enhances zymophagy by facilitating its activation rather than its degradation.

Figure 3.

AS ameliorates AP by enhancing Atg5-dependent zymophagy activation. (A, B) AR42J cells grown on glass slides were transfected with a tandem RFP-GFP-LC3B construct and then treated with AP plus AS (2 μg/ml) and bafilomycin A1 (Baf; 10 nM) for 4 h. Autophagosomes and autolysosomes were observed by confocal imaging and counted in ImageJ (n = 5; bar = 10 μm). (C) AR42J cells grown in 96-well plates were treated with AP with or without AS (2 μg/ml) and Baf (10 nM), rapamycin (Rapa; 10 nM), or 3-methyladenine (3-MA; 4 mM) for 24 h, then supernatants were harvested for measuring amylase activity (n = 3). (D, E) AR42J cells were transfected with control and Atg5 siRNA, and then were treated with AP and AS described in Fig. 2G. IF staining and confocal imaging were performed as shown in Fig. 2F (n = 3). (F) AR42J cells grown in 6-well plates were treated as shown in Fig. 3D, then total cell lysates were extracted for detecting Atg5 and LC3B-II expression. Protein ratios shown under the bands are representative of three independent experiments (n = 3). (G) AR42J cells grown in 96-well plates were treated as shown in Panel D, then the supernatants were used to measure amylase activity (n = 3). **P < 0.01, *P < 0.05, and ^#P > 0.05.

A range of chemical modulators of autophagy were utilized to explore whether the protective role of AS in AP is mediated through enhanced zymophagy. In the in vitro AP model, the AP-induced increase in amylase activity was attenuated by the autophagy activator Rapa and the autophagy degradation inhibitor Baf, whereas it was further exacerbated by the autophagy initiation inhibitor 3-MA (Fig. 3C). Notably, the suppression of AP-induced amylase activity by AS was sustained only in the presence of Rapa but was abolished in the presence of either 3-MA or Baf (Fig. 3C). These findings strongly suggest that autophagy exhibits a protective role in AP, as even when autophagy flux is inhibited by Baf, the formation of autophagosomes still mitigates AP severity. More importantly, these data collectively indicate that AS alleviates AP by enhancing zymophagy.

Canonical selective autophagy, including zymophagy, is known to be activated via the Atg5-Atg12-Atg16L1 pathway critical for LC3 conjugation [5]. Using specific siRNA targeting Atg5 mRNA, we found that zymophagy was impaired in the AP group, and the enhancement of zymophagy (Tryp-LC3 colocalization) by AS was abolished in Atg5 knockdown acinar cells (Fig. 3D–E). The IB results consistently showed that AS treatment could not further increase the LC3B-II levels (Fig. 3F). As expected, amylase activity assays showed Atg5 knockdown exacerbated AP and abolished the protective effect of AS (Fig. 3G). Together, these results suggest that zymophagy has a positive effect on alleviating AP and that AS ameliorates AP by enhancing Atg5-dependent zymophagy activation.

AS upregulates p62-mediated zymophagy in AP

Selective autophagy employs various autophagy receptors and adapters such as the Pink1/Parkin axis, p62, optineurin, and NDP52 to facilitate autophagy activation [10]. A previous study reported that zymophagy was triggered by p62, a well-studied autophagy adaptor, with the participation of Ub and VMP1 [9]. Therefore, we first detected the protein levels of p62, Ub, and VMP1 in total cell lysates from the in vitro AP model for up to 24 h. The IB results showed that the protein levels of p62 and Ub were upregulated rapidly and then downregulated gradually, and the peak time of p62 was 1 h (Fig. 4A). In contrast, VMP1 showed no significant upregulation. Therefore, 1 h was chosen as the time point for determining p62-mediated zymophagy changes in vitro.

Figure 4.

AS upregulates p62-mediated zymophagy in AP. (A) AR42J cells were treated as shown in Fig. 2F. The levels of p62, ubiquitin (Ub), VMP1 and Tryp were detected by IB. (B) AR42J cells grown in 10-cm dishes were treated as described in Fig. 2G. Total cell lysates were then subjected to immunoprecipitation (IP) using an antibody against Tryp, and the associated p62, Ub, and VMP1 were detected by IB. (C–F) AR42J cells were treated as described in Fig. 2H. The colocalization of Tryp with VMP1, Ub, or p62 was detected as described in Fig. 2G (n = 3). (G, H) AR42J cells transfected with Ad-GFP-LC3B were treated as shown in Fig. 2G. The colocalization of Tryp with p62 and LC3 was detected as shown in Fig. 2G (n = 3). Protein ratios shown under the bands are representative of two independent experiments. **P < 0.01, ^#P > 0.05.

Subsequently, a co-IP experiment was performed to determine whether p62, Ub and VMP1 were recruited by trypsinogen and responded to AS treatment. In the whole cell lysates used as input, IB results showed that AS treatment could further increase the levels of p62 and Ub except that of VMP1 compared to AP (Fig. 4B). Whereas, only Tryp-associated p62 was further increased, while Tryp-associated Ub and VMP1 were weakly decreased by AS treatment in Tryp pulldown products. Similarly, confocal images depicting the colocalization of Tryp with VMP1, Ub or p62 showed that although these recruitments by Tryp were weakly increased in the AP group, only Tryp-p62 colocalization was further increased by AS treatment (Fig. 4C–F). To further confirm the involvement of p62 in zymophagy, immunofluorescent imaging was performed again to observe the co-localization of Tryp-p62-LC3. We found that colocalization was slightly increased by AP and was further augmented by AS treatment (Fig. 4G–H). Overall, these results suggest that although Ub, p62, and VMP1 are involved in zymophagy, only upregulated p62 is responsible for AS’s activity.

p62 knockdown impairs zymophagy and the protective role of AS in acinar cells

To further verify that p62 is involved in zymophagy and contributes to the protective role of AS in AP, p62 knockdown acinar cells were used to assess AS effects on AP and thereby-induced zymophagy. Since VMP1 was crucial for the activation of zymophagy, although it was not responsible for the activity of AS, Vmp1 knockdown acinar cells were used here as a control. First, IF experiments assessing colocalization of Tryp-LC3 showed that the enhancement of AP-induced zymophagy (yellow faculae) by AS persisted in cells treated with control siRNA, but was lost in cells treated with either Vmp1 siRNA or p62 siRNA (Fig. 5A–B). Consistently, the IB results also showed that AS treatment could further upregulate the protein levels of LC3B-II and p62 in cells treated with control siRNA, but this activity was lost in cells treated with Vmp1 siRNA and p62 siRNA (Fig. 5C–D). Finally, we examined amylase activity in acinar cells treated with these siRNAs. Similar to the results of cells treated with control siRNA, amylase activity was markedly increased by AP, while it was completely restored by AS treatment in cells transfected with Vmp1 siRNA (Fig. 5E). However, AP could induce much higher amylase activity and AS treatment lost the inhibitory effect in p62 knockdown cells. These data suggest that although both VMP1 and p62 contribute to zymophagy and the enhancing effect of AS, only p62 is directly responsible for AP and the protective role of AS. Our data also suggest the different roles of VMP1 and p62 in zymophagy, such that VMP1 may only support physiological zymophagy, while p62 is related to massively activated zymophagy, which plays a protective role in AP.

Figure 5.

p62 knockdown impairs zymophagy and the protective role of AS in AP. (A, B) AR42J cells were transfected with control siRNA, Vmp1 siRNA, or p62 siRNA, and then cells were treated as shown in Fig. 2H. The colocalization of Tryp with LC3 was detected as described in Fig. 2G (n = 3). (C–E) AR42J cells were treated as described in Panel A, then total cell lysates were used for detecting protein levels of VMP1, p62, and LC3B-II by IB, and supernatants were used to measure amylase activity (n = 3). Protein ratios shown under the bands are representative of three independent experiments. “n.d.” indicates not detected, *P < 0.05, **P < 0.01, and ^#P > 0.05.

p62 is required for zymophagy and mediates the protective effect of AS in AP

The above findings strongly suggest the pivotal role of p62 in initiating zymophagy and the protective role of AS in the in vitro AP model. Here, we further verified the role of p62 in AP using the in vivo AP model with a genotype of pancreatic p62 knockdown through a p-AAV9 harboring p62 shRNA and a tandem GFP gene (Fig. 6A–B). Pancreases transfected with p-AAV9 exhibited uniform green fluorescence due to tandem GFP expression. Morphological and histological observations showed that pancreatic changes, including edema and microhemorrhages, increased the pancreatic index and histology score in the AP group (Fig. 6C–F). As expected, these changes were significantly attenuated by AS in negative control mice. Notably, pancreatic index and histology scores were further increased in the AP group, and the protective effect of AS was abolished in mice transfected with p62 shRNA (hereafter referred to as p62 knockdown). Consistently, serum amylase activity assays produced similar data (Fig. 6G). These results suggest that p62 plays a protective role in AP and is involved in the therapeutic effect of AS.

Figure 6.

p62 is required for zymophagy and mediates the protective effect of AS in AP. (A, B) Schematic of the animal experiment. Male C57BL/6N mice were infected by p-AAV9 containing p62 shRNA or empty p-AAV9 used as a negative control (n = 4), and then treated as described in Fig. 1A. The efficacy of pancreatic p62 knockdown was verified by IB, and protein ratios are indicated under the bands. (C, D) Photographs and pancreatic index of pancreas. (E, F) HE staining and pancreatic histology score. (G) Serum amylase activity. (H, I) TEM observations of pancreas. (J) Expression of p62 in pancreatitis human specimens. RNA-sequencing data were obtained from the GEO database. *P < 0.05, **P < 0.01, and ^#P > 0.05.

We further investigated the impact of p62 knockdown on zymophagy in AP using TEM. In the negative control group, the number of zymophagosome in acinar cells increased slightly in the AP group, whereas AS treatment induced a substantial increase (Fig. 6H–I). In the p62 shRNA group, AP could still induce a weak increase in zymophagosomes in acinar cells; however, AS treatment failed to augment this increase in AP pancreas (Fig. 6H–I). These results indicate that p62 plays a pivotal role in acinar zymophagy and the zymophagy-enhancement effect of AS. Moreover, we obtained RNA sequencing data of human pancreatitis specimens from the GEO database to analyze the p62 mRNA expression pattern. The results showed that pancreatic p62 mRNA expression in patients with pancreatitis did not change significantly compared with that in normal controls (Fig. 6J; GSE326582, GSE30134, GSE143754). In one dataset (GSE101462), p62 expression was even downregulated. These findings suggest that p62-mediated zymophagy may not be adequately activated in patients with pancreatitis, consistent with our observation in this section that knockdown of p62 in experimental pancreatitis impaired zymophagy and aggravated pancreatitis.

AS upregulates p62 and zymophagy via selectively activating the NRF2-ARE-p62 pathway

Given that p62 serves as a pivotal activator in initiating acinar zymophagy and was responsible for the anti-AP activity of AS based on the above findings, we further explored the mechanism by which AS upregulated p62 in AP. Previous reports indicate that the transcription factor NRF2 regulates the transcription process of macroautophagy genes (including p62, Ndp52, and Gabarapl1) through direct binding to the ARE in their promoter regions [16]. We therefore used Nrf2 knockdown AR42J cells (via Nrf2 siRNA) to assess the role of NRF2 in p62-zymophagy and the enhancement activity of AS (Fig. 7A).

Figure 7.

AS upregulates p62 and zymophagy via selectively activating the NRF2-ARE-p62 pathway. (A, B) AR42J cells grown in 12-well plates were transfected with control and Nrf2 siRNA. Cells were treated as described in Fig. 2G for 1 h, then whole cell lysate was harvested for detecting NRF expression by IB (n = 3), and total RNA was extracted for real-time PCR analysis of p62, Ndp52, and Gabarapl mRNA expression by normalizing to β-actin (n = 4). (C, D) AR42J cells grown on glass slides were transfected with control and Nrf2 siRNA. The treatment, IF staining, and colocalization analysis were performed as shown in Fig. 3D (n = 3). (E) AR42J cells grown in 96-well plates were treated with AP with or without AS (2 μg/ml) for 24 h, then supernatants were used for measuring amylase activity (n = 3). (F) The binding motif for NRF2 in the promoters of target genes. Schematic diagrams indicate the loci of primer pairs designed for ChIP-PCR targeting the promoter region of p62, Ndp52, or Gabarapl. (G) ChIP-real-time PCR assays of the binding of NRF2 to the promoter of p62, Ndp52, or Gabarapl normalized to Gapdh in the input (n = 3). A–F, *P < 0.05, **P < 0.01, ^#P > 0.05. (H) AR42J cells grown in 12-well plates were treated with AP with or without 2 μg/ml AS for up to 1440 min as indicated, then cells were harvested for detecting the mRNA levels of p62, Ndp52, and Gabarapl normalized to β-actin (n = 4). **P < 0.01 vs. 0 h.

For mRNA expression, we found that AP downregulated Gabarapl1, and AS treatment only selectively upregulated p62 in AR42J cells treated with control siRNA. Markedly, the results showed that the upregulation effect of AS on p62 was completely abolished in Nrf2 knockdown cells (Fig. 7B). We also found that Ndp52 showed consistent downregulation across all treatments, and Gabarapl1 was upregulated by AP while downregulated by AS treatment in Nrf2 knockdown cells. Consistent with the results at the mRNA level, we found that the zymophagy potentiation activity of AS was maintained in control cells but lost in Nrf2 knockdown cells (Fig. 7C–D). Importantly, amylase activity was further increased in the AP group and could not be inhibited by AS treatment in Nrf2 knockdown cells (Fig. 7E). Together, these results suggest that AS enhances p62-zymophagy and protects AP via NRF2.

We therefore performed ChIP assays to determine whether AS affected the binding of NRF2 to promoters of these genes, using primer pairs targeting predicted ARE motifs identified via the JASPAR database (Fig. 7F). The results showed that although the binding of NRF2 to the ARE of Ndp52 and Gabarapl1 was increased by AP, which could not be further increased by AS treatment (Fig. 7G). By contrast, the binding of NRF2 to p62-ARE showed no significant response to AP, but was significantly increased by AS treatment. Furthermore, the temporal changes in mRNA expression of these genes were determined, and the results were consistent with the ChIP results. As expected, the mRNA level of Ndp52 and Gabarapl1 was rapidly upregulated by AP (for up to 24 h), but this upregulation could not be further augmented by AS (Fig. 7H). In contrast, the mRNA level of p62 showed no significant change in response to AP over 24 h, but was gradually upregulated by AS (Fig. 7H). These findings demonstrate that AS upregulates p62 mRNA expression by specifically activating the NRF2-ARE-p62 pathway.

AS promotes non-canonical NRF2 activation by binding to p62

Here, we further investigated the detailed mechanisms by which AS primed NRF2-dependent transcription of p62 in AP. To assess the nuclear translocation required for NRF2 activation, the AP pancreata from in vivo experiments (Fig. 1A) were further analyzed by IHC. The results showed that the number of NRF2-positive nuclei (brown staining) was slightly increased in AP, while massively increased following AS treatment (Fig. 8A–B). Subsequently, IB results indicated that AP induced an increase in both the cytoplasmic and nuclear levels of NRF2 in acinar cells; however, AS treatment showed the opposite impact, in that it remarkably reduced the cytoplasmic level while increasing the nuclear level of NRF2 (Fig. 8C). Together, these in vivo and in vitro data consistently indicate that AS induces NRF2 activation, which is associated with the upregulation of p62 mRNA.

Figure 8.

AS promotes non-canonical NRF2 activation by binding to p62. (A, B) AP pancreas obtained from experiments described in Fig. 1A were used for detecting NRF2 expression by IHC, and the positive staining of nuclear NRF2 (indicated by red arrows) was counted using ImageJ (n = 3; **P < 0.01). (C) AR42J cells grown in 10-cm dishes were treated as shown in Fig. 2E for 1 h. The cytoplasmic and nuclear levels of NRF2 were detected by IB (n = 2). (D–F) AR42J cells grown in 10-cm dishes were treated as shown in Fig. 2E. Whole cell lysates were subjected to IP using antibodies against p62, Keap1, or NRF2, and associated proteins as indicated were detected by IB, respectively. Protein ratios shown under the bands are representative of three or two independent experiments. (G–J) AR42J cells were treated as described in Fig. 2F. The colocalization of Keap1 with Ub, p62, or LC3B was observed and calculated as described in Fig. 3D (n = 3). (K) The thermal shift assay was applied to observe the protein thermal stability of p62, Keap1, and NRF2 treated with or without 2 μg/ml AS (n = 3). (L) The computed model for the p62 PB1 domain–AS interactions was shown in cartoon (p62) and stick (AS) mode. Hydrogen bonds between AS and p62 are depicted by green dashed lines; hydrophobic contacts are indicated with yellow dashed lines. *P < 0.05, **P < 0.01, and ^#P > 0.05.

Besides the canonical NRF2 activation driven by redox stress, p62 can also promote the non-canonical NRF2 activation through Keap1 sequestration [13, ¹⁷]. Here, co-IP experiments were performed to verify the interactions between p62, Keap1, and NRF2 in AP. In IP products of p62, AS treatment markedly increased Keap1 level compared to AP, suggesting that AS facilitates the p62-Keap1 interaction, thereby restricting the binding of Keap1 to NRF2 (Fig. 8D). In IP products of NRF2, we observed that AP induced an increase in Keap1 level, whereas AS treatment led to a decrease, which was thought to be associated with elevated nuclear level of NRF2 (Fig. 8E). In IP products of Keap1, AS treatment indeed reduced NRF2 level compared to AP, corroborating the prior Co-IP findings (Fig. 8F). Interestingly, AS treatment markedly elevated levels of Ub, p62, and LC3B-II, and reduced the level of Keap1 in Keap1 immunoprecipitates compared to AP treatment. This suggests an autophagy-dependent degradation pathway of Keap1 that is related to non-canonical NRF2 activation (Fig. 8F). Consequently, IF staining was conducted to observe the co-localization of Keap1 with Ub, p62, or LC3B in the in vitro AP model. The results demonstrated that AP could weakly induce co-localization of Keap1 with these proteins (Fig. 8G–J). Notably, AS treatment markedly increased the co-localization of Keap1 with Ub, p62, or LC3, corroborating the Co-IP findings. In addition, AS treatment induced p62 oligomerization, which was related to cargo interactions [18]. Collectively, these findings indicate that AS enhances p62-associated autophagic degradation of Keap1, leading to non-canonical NRF2 activation.

To further investigate, we employed the CETSA method to evaluate whether AS could directly bind to p62, Keap1, or NRF2. Compared to controls, AS markedly increased p62 tolerance from 45°C to 49°C (ΔT_m = 3.17°C), while Keap1 and NRF2 showed minimal changes (ΔT_m = 1.41°C and 0.91°C, respectively; Fig. 8K), indicating that AS selectively enhances the thermal stability of p62. Docking studies suggested that AS selectively bound to p62 (score = −7.26) but not to Keap1 or NRF2. The predicted binding pose revealed that AS interacted with the PB1 domain of p62, and key contacts included hydrogen bonds with residues MET2 and SER28, as well as hydrophobic packing against PHE27 (Fig. 8L). The AS-PB1 binding interface lied distant from the KIR and UBA domains, providing a mechanistic explanation for AS-mediated promotion of p62 oligomerization (Fig. 8H). These findings demonstrate that AS directly binds to p62, thereby strengthening the p62-Keap1 interaction and facilitating the non-canonical activation of NRF2.

Discussion

In this study, we demonstrate that augmenting p62-mediated pancreatic zymophagy with AS, a common anti-malarial agent, represents a promising therapeutic strategy for AP. Mechanistically, AS enhances the p62–Keap1 interaction by binding to the PB1 domain of p62, thereby promoting the autophagic degradation of Keap1 in a p62-dependent manner. This leads to the subsequent non-canonical activation of NRF2 and ultimately upregulates p62 mRNA expression.

AP is an acute abdominal disease caused by multiple etiological factors, including biliary disorders, ethanol, lipid disorders, and infection [19]. Clinical and experimental findings indicate pathologic intrapancreatic activation of trypsinogen is the main cause of AP [9, 20]. Consequently, the selection of experimental models is critical for precise observation of trypsinogen hyperactivation and related early changes during AP. ST, L-arginine (L-Arg), and CR are the most frequently used stimuli for establishing experimental AP models [19]. AP models induced by ST and L-Arg show severe pathophysiology with rapid progression and pancreatic necrosis, while cholecystokinin peptide homologue CR-induced AP models present mild reversible pathology characterized mainly by acinar cell apoptosis and interstitial edema [21]. The CR model preserves acinar cell physiological characteristics during experimentation, recapitulates clinical AP initiation mechanisms, and provides an optimal platform for investigating therapeutic targets in cell biology [21, 22]. This methodology has been widely adopted in the past decades due to its strong reproducibility, technical simplicity, and adaptability for generating other pancreatitis forms through combination with various stimuli [22]. In this study, we employed a murine AP model combining CR with LPS. This model exhibited pathophysiological features of rapid AP progression, pancreatic edema and microhemorrhage, and remarkable trypsinogen hyperactivation within acinar cells, which was in line with CR-induced AP but induced more pronounced early intracellular changes, particularly in massive autophagosome-like vesicle accumulation. Our findings align with previous reports that CR-based models are suitable for studying early pathophysiological alterations in acinar cells and AP therapeutic intervention strategies. Moreover, amylase activity was detected in this study to reflect trypsinogen activation and AP severity in both in vitro and in vivo models.

Currently, the core clinical principles for the management of AP include inhibition of pancreatic enzyme secretion, fluid resuscitation, organ support, analgesia, elimination of underlying causes, and prevention and management of complications. UTI, a broad-spectrum protease inhibitor, is used to mitigate inflammatory responses in patients with moderate-to-severe AP. Over the past decade, no specific drug has been available that can directly inactivate already activated pancreatic enzymes [3, 4, 23]. In this study, AS was observed to reduce pancreatic enzyme activation and to demonstrate superior efficacy to UTI in ameliorating pathological damage in pancreatitis. Beyond its previously reported anti-inflammatory pharmacological effects, AS may exhibit additional properties that further support its potential application in the treatment of AP [14].

Numerous studies have indicated that autophagy is a critical early event in AP [20]. However, the mechanistic role of autophagy in AP remains incompletely characterized, with existing literature demonstrating conflicting conclusions. An earlier report suggested that conditional knockout of Atg5 in pancreatic acinar cells reduced severity of AP murine model and reduced trypsinogen activation, indicating an initiating role of autophagy in AP [24]. Another study indicated experimental AP (CR and L-Arg model) in transgenic GFP-LC3 mice was more severe than in wildtype mice, suggesting uncontrolled autophagy exacerbates AP [25]. By contrast, two independent studies respectively reported that pancreas-specific depletion of either Atg5 or Atg7 resulted in the development of pancreatitis in transgenic animals, providing some considerable evidence on the protective role of autophagy in the exocrine pancreas during AP [26, 27]. Although the pathophysiological involvement of autophagy in AP remains contentious, emerging consensus confirms that basic autophagy protects acinar cells from stimuli while impaired autophagy potentiates pancreatitis progression [28, 29]. Here, our findings indicate inhibition of autophagy activation (using 3-MA or Atg5 siRNA) aggravates AP, whereas enhancement of autophagy activation (using rapamycin) mitigates AP. Interestingly, we found that blocking autophagy degradation (using bafilomycin A1) also alleviated AP, suggesting that the sequestration of activated trypsinogen within autophagosomes still exerts a protective effect despite impaired autophagy flux. Overall, our findings support the emerging consensus that autophagy plays a protective role in AP, and further suggest that sequestering harmful cargo in autophagosomes still confers a protective effect even under conditions of autophagic flux blockade.

Several forms of autophagy, including nonselective autophagy, zymophagy (targeting activated trypsinogen), mitophagy (targeting damaged mitochondria), and non-canonical autophagy, have been reported to occur in acinar cells during AP [6, 30]. Although these forms utilize similar LC3 conjugation systems and lysosomal degradation processes, they exhibit distinct cargo recognition and triggering mechanisms. Among these autophagic processes, zymophagy is unique in that it selectively eliminates hyperactivated trypsinogen and has been observed in both CR-related AP models and clinical patients [9]. Moreover, existing studies consistently suggest that zymophagy is also the only non-controversial autophagy form that exerts a protective role in AP models induced by CR or ST [8, 9]. Nevertheless, existing studies have not established the quantitative distribution of autophagic subtypes or resolved the temporal dynamics of zymophagy during AP progression. In this study, multimodal morphological and molecular monitoring of autophagy revealed that zymophagy was modestly activated in AP pancreas, peaking at 22% of total autophagic events (18 h post-induction). However, we also found a marked increase of nonselective autophagy in AP pancreas, which was temporally correlated with zymophagy. This indicates that zymophagy constitutes a minor part of AP-related autophagy, suggesting zymophagy-targeting therapy may have significant clinical potential in the future. However, to date, the therapeutic targeting of zymophagy in AP remains unexplored. Our study demonstrates that pharmacological enhancement of zymophagy by AS exerts a potent anti-AP effect, as evidenced by complete therapeutic abolishment following either zymophagy or pan-autophagy inhibition.

The mechanistic studies integrated zymophagy observation and pharmacological validation to delineate the AS-enhanced zymophagy pathway in both in vivo and in vitro AP models. Our findings demonstrate that AS upregulates zymophagy activation through the Atg5-Atg12-Atg16L1 pathway involved in the LC3-conjugation process, which closely resembles the phagophore elongation (autophagosome formation) stage observed in most canonical selective autophagy [30]. Our data further reveal that AS enhances zymophagy through the VMP1-Ub-p62 axis, consistent with the original characterization of zymophagy by Grasso et al. [9]. The molecular details of zymophagosome assembly are not fully understood. Our data suggest a model in which (1) p62, via its UBA domain, recognizes ubiquitinated trypsinogen and links the cargo to nascent zymophagosomes through LC3 binding; and (2) VMP1 promotes phagophore expansion and closure, thereby contributing to zymophagosome formation [31, 32]. Our findings reveal that AS exerts therapeutic effects through p62-dependent potentiation of zymophagy, and p62 knockdown in cellular and animal AP models completely abrogates AS-mediated protection. A previous study showed that AS enhanced nonselective autophagy in the retinas of diabetic rats by upregulating Atg6 expression and LC3-II/I ratio while down-regulating p62 [33], and our prior findings indicated that AS downregulated p62 expression in LPS-induced nonselective autophagy of macrophages [34]. Moreover, another study revealed that AS attenuated liver fibrosis by triggering ferritinophagy (a selective autophagy targeting ferritin), which correlated with p62 down-regulation in hepatic stellate cells [35]. Our findings unveil a specific AS-mediated mechanism where p62 upregulation enhances zymophagy, explaining its selective potentiation of zymophagy over other AP-related autophagy forms. Moreover, RNA sequencing data from clinical specimens revealed that pancreatic p62 mRNA expression was not significantly changed, or was even downregulated, in patients with pancreatitis, suggesting that p62-mediated zymophagy may be insufficiently activated in these patients. Based on this finding, we further propose that AS may be a potential precision therapy for patients with pancreatitis who possess low p62 expression.

Given that the pancreas employs NRF2 and nuclear factor κB (NF-κB), two key transcriptional factors that sense oxidative stress, to maintain redox homeostasis under physiological conditions as well as in AP and chronic pancreatitis [36]. A previous report revealed that p62 was strongly upregulated by NRF2 via an ARE element in the promoter of p62 gene, and could also be primed by NF-κB [37]. Our previous study has revealed that AS inhibited NF-κB activation in a rat SAP model [14]. Therefore, we investigated NRF2-mediated p62 upregulation by AS using a multi-method approach centered on Nrf2 knockdown and ChIP. Intriguingly, AS promoted NRF2 nuclear translocation to selectively upregulate p62 mRNA expression without affecting other NRF2-regulated autophagy genes (e.g. Ndp52 and Gabarapl1), demonstrating a unique transcriptional profile distinct from its regulation of other autophagy forms reported earlier. Similarly, a previous work showed that AS-activated p62/NRF2 signaling ameliorated arthritic bone erosion by inhibiting osteoclastogenesis, although this study focused on arthritis pathology rather than AP [38]. To our knowledge, NRF2 can be activated through canonical and non-canonical pathways. The canonical NRF2 activation involves electrophilic modification of Keap1 cysteines, which inhibits its ability to target NRF2, leading to NRF2 stabilization and nuclear translocation under redox stress [39, 40]. Furthermore, Keap1 undergoes Ub- and p62-associated autophagic degradation, thereby driving non-canonical NRF2 activation [40]. Here, our findings elucidate a selective autophagic degradation pathway of Keap1 via the Ub-p62 pathway, demonstrating both the involvement of non-canonical NRF2 activation in AP pathogenesis and the protective role of AS. In summary, both the direct upregulation of p62 through AS binding to p62 and activation of the p62–NRF2 signaling, and the indirect activation of p62 via p62-related autophagic degradation of Keap1, jointly contribute to AS-induced non-canonical NRF2 activation, with AS–p62 binding being the primary mechanism.

AS, a succinyl ester derivative of artemisinin, is widely used as an anti-malarial drug with repurposed therapeutic potential. Over the past two decades, numerous studies have validated its pleiotropic anti-inflammatory activity in multiple inflammatory diseases, including autoimmune disorders, neuroinflammation, and infectious diseases [41]. Our previous study demonstrates that AS ameliorates SAP in rats by inhibiting Toll-like receptor 4 (TLR4)-mediated pro-inflammatory cytokine production [14]. Complementary evidence from the murine AP model suggests that AS protects via dual suppression of TLR4-driven inflammation and nonselective autophagy [42]. Overall, existing studies suggest that AS protects inflammatory diseases, including AP via inflammation-related pathways, yet its molecular mechanism remains incompletely understood. Although some of them have mentioned nonselective autophagy, there is no study focused on the regulatory role of zymophagy by AS to date. Specifically, AS exhibited potent anti-AP activity by enhancing protective zymophagy, directly acting on p62 to potentiate its binding to Keap1 and disrupting the Nrf2-Keap1 complex. It is well-established that small molecules such as ML334, Cpd16, and CPUY192018 can directly disrupt the Nrf2-Keap1 complex [17]. Besides these NRF2 agonists, p62 ZZ-domain agonists have been explored as anti-microbial agents [43], and oroxylin A was shown to act as a p62 agonist by binding to the PB1 domain, inhibiting NF-κB activation in macrophages [44]. Here, AS is suggested as a p62 agonist acting on the PB1 domain to promote non-canonical NRF2 activation. Given the well-established safety profile of AS after decades of use in the treatment of malaria, and given that the doses of AS used in experimental models of AP do not exceed those used clinically in antimalarial therapy, AS may represent a candidate drug for the treatment of AP.

This study has several limitations. First, although the relationship between p62, zymophagy, and AP has been investigated in p62 knockdown models, we did not perform experiments using acinar cell-specific p62 knockout mouse models, which would provide more direct evidence for the role of p62-zymophagy in anti-AP effects of AS. In addition, we did not use human AP models, which limits the potential clinical relevance of our findings. Second, the mechanisms by which AS interacts with p62 remain to be clarified, and future studies combining site-directed mutagenesis with structural biology would help to address this question. Finally, the mechanism underlying the selective upregulation of p62 by AS among autophagy-related genes requires further investigation. A better understanding of this NRF2-mediated selective transcriptional activation may be valuable for both fundamental autophagy research and the potential clinical application of AS in AP.

Ethics statement

All animal procedures and protocols were approved by the Army Experimental Animal Management Committee (Ethics Approval No. AMUMEC2020962). We all confirmed that all animal experiments and procedures were conducted in accordance with the National Guidelines for the Care and Use of Laboratory Animals.

Author contributions

Yanyan Cen (Funding acquisition, Investigation, Validation, Writing—original draft), Yalan Xiong (Investigation), Chenxu Liu (Data curation), Pan Wang (Investigation), Ting Zhu (Investigation), Fei Li (Methodology, Supervision), Haigang Zhang (Supervision), Wenxuan Cao (Data curation), Youcai Deng (Project administration, Supervision), and Xichun Pan (Funding acquisition, Project administration, Resources, Writing—review & editing).

Conflicts of interest

The authors declare no competing interests.