Avobenzone induces intestinal injury by targeting ATG14 and impairing mitophagy Authors: Zi-Kang Gui, Zi-Jia Huang, Xiao-Yu Qi, Bao-Peng Wang, Jian-Rui Zhong, Yu-Jie Zeng, Yi-Ping Zhou, Qing-Yu He, Jing Zhang, Yang Wang Published: 2026 ABSTRACT AbstractBackgroundAvobenzone (Avo) is a widely used ultraviolet filter in sunscreens and facial skincare products. In addition to dermal exposure, chronic oral intake may also occur through daily-use products. However, the potential intestinal toxicity of Avo and its underlying mechanisms remain poorly understood.MethodsWe investigated the intestinal effects of Avo using repeated oral exposure mouse models, the Caco-2 intestinal epithelial cell line, human intestinal organoids, proteomic analysis, and mechanistic assays examining autophagy, mitophagy, mitochondrial stress responses, and Wnt/β-catenin signaling.ResultsAdministration of Avo (10–40 mg/kg) for two months induced intestinal epithelial injury in mice, as evidenced by reduced crypt number, shortened villi, impaired intestinal epithelial barrier, and decreased intestinal stem cell abundance. Avo also caused mitochondrial dysfunction, oxidative stress, and lipid accumulation. Mechanistically, Avo interacted with ATG14, a key component of the autophagy initiation complex, and promoted CUL3-mediated ubiquitination and proteasomal degradation of ATG14. Loss of ATG14 impaired mitophagy, leading to the accumulation of damaged mitochondria and activation of eIF2α-ATF5-associated mitochondrial unfolded protein response (UPRmt). This mitochondrial stress suppressed Wnt/β-catenin signaling, reduced maintenance of LGR5-positive intestinal stem cells, and ultimately disrupted epithelial regeneration. Importantly, pharmacological induction of autophagy with rapamycin or inhibition of stress signaling with ISRIB alleviated Avo-induced intestinal injury in mice. Similar effects were observed in human intestinal organoids.ConclusionsThese findings identify a previously unrecognized intestinal toxicity associated with Avo exposure and suggest that disruption of ATG14-dependent mitophagy and mitochondrial quality control is a key mechanism underlying this effect.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03211-5.Keywords: Avobenzone, Mitophagy, Intestinal organoid, ATG14, Wnt/β-catenin signaling Background The intestinal epithelium serves as a critical barrier that is constantly exposed to dietary and environmental toxicants [1–3]. These insults, along with intrinsic signaling pathways, play a pivotal role in regulating epithelial integrity and gut homeostasis [4]. As one of the most rapidly renewing tissues in the body, the intestinal epithelium is sustained by intestinal stem cells (ISCs) located in the crypts [5]. The Wnt/β-catenin signaling pathway is central to driving ISC self-renewal and epithelial regeneration [2, 4, 5]. Consequently, disruption of this pathway impairs epithelial renewal and barrier function, indicating the intestine’s vulnerability to toxicant-induced injury. Maintaining mitochondrial quality control is essential for intestinal epithelial homeostasis due to the high metabolic demands of proliferating crypt cells. Mitophagy, a form of selective autophagy responsible for removing damaged mitochondria, plays a crucial role in maintaining mitochondrial integrity and metabolic balance [6]. Defective mitophagy leads to the accumulation of dysfunctional mitochondria and excessive reactive oxygen species production, triggering mitochondrial stress responses and cellular damage [7–9]. Persistent mitochondrial stress can further activate pathways such as the mitochondrial unfolded protein response (UPRmt), contributing to cellular dysfunction and tissue degeneration [10]. Recent studies have shown that mitochondrial stress signaling can negatively regulate Wnt/β-catenin signaling and reduce the number of LGR5-positive intestinal stem cells [11]. These findings suggest that disruption of mitophagy-mediated mitochondrial quality control may represent an important mechanism linking environmental toxicants to intestinal epithelial dysfunction. Avobenzone (1-(4-methoxyphenyl)-3-(4-tert-butylphenyl) propane-1,3-dione, Avo) is one of the most widely used organic ultraviolet filters in sunscreen formulations, particularly facial skincare products [12]. Due to its extensive use in personal care products, Avo has been frequently detected in aquatic environments [13], with reported concentrations up to 721 ng/L in the surface waters of Hong Kong, up to 70 ng/L in the Arctic Ocean, suggesting its widespread environmental occurrence [14]. In addition to environmental contamination, human exposure to Avo occurs through dermal absorption from sunscreens and through oral routes, including inadvertent ingestion from lip products and hand-to-mouth contact. Based on conservative assumptions (3% formulation, 0.59% dermal absorption), the estimated systemic exposure dose (SED) from dermal sunscreen application ranges from approximately 0.083 –0.29 mg/kg/day [15]. However, additional oral exposure may occur through unintentional ingestion or consumption of contaminated seafood due to environmental bioaccumulation, highlighting the need to evaluate gastrointestinal effects of avobenzone at doses relevant to these non-dermal routes. Indeed, Avo has been detected in human plasma under maximal usage conditions [16], and has been reported to induce obesogenic phenotypes in human epidermal keratinocytes and mesenchymal stem cells through activation of PPARγ signaling [17]. Pharmacovigilance reports from the VigiAccess and FDA Adverse Event Reporting System databases indicate that there are increasing reported adverse drug reactions (ADRs) of Avo on the gastrointestinal tract, thus its potential adverse effects on human health have given rise to substantial concern. Notably, Avo has been reported to induce mitochondrial dysfunction-mediated apoptosis in human trophoblast cells [18], suggesting that impairment of mitochondrial quality control may represent a shared mechanism underlying Avo-induced cytotoxicity across cell types. However, whether and how Avo disrupts mitochondrial quality control in intestinal epithelial cells remains unknown. In the present study, we found that Avo disrupts intestinal epithelial homeostasis by interfering with mitochondrial quality control pathways. Using mouse models, human intestinal organoids, and molecular interaction analyses, we demonstrate that Avo targets ATG14, a key component of the autophagy initiation complex, promoting its CUL3-mediated ubiquitination and degradation. Loss of ATG14 impairs autophagosome formation and mitophagy, leading to accumulation of dysfunctional mitochondria and activation of mitochondrial stress responses. This mitochondrial stress subsequently suppresses Wnt/β-catenin signaling and compromises LGR5-positive intestinal stem cell maintenance. Importantly, pharmacological induction of autophagy by rapamycin and inhibition of stress signaling by ISRIB alleviate Avo-induced intestinal injury. These findings reveal a previously unrecognized mechanism by which Avo disrupts intestinal epithelial homeostasis and identify mitochondrial quality-control pathways as key mediators of intestinal toxicity induced by environmental chemicals. Methods Cell culture and chemical reagents The HEK293 human embryonic kidney cell line (RRID: CVCL_0045) and HEK293T (RRID: CVCL_0063) were obtained from the Xiamen Immocell Biotechnology Co., Ltd. (Xiamen, China). HEK293 and HEK293T cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, NY, USA) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin (Gibco). The human colorectal adenocarcinoma cell line Caco-2 (RRID: CVCL_0025) was obtained from the American Type Culture Collection (ATCC, Rockville, MD) and cultured in Minimum Essential Medium (MEM; Gibco) supplemented with 10% FBS, 1% non-essential amino acids (NEAA; Gibco), and 1% penicillin/streptomycin. All cells were maintained at 37 °C in a humidified atmosphere containing 5% CO₂. Avobenzone (CAS 70356-09-1), rapamycin (CAS 53123-88-9), and ISRIB (CAS 548470-11-7) were purchased from Selleck Chemicals (Houston, TX, USA) and dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich, MO, USA). Carbonyl cyanide m-chlorophenylhydrazone (CCCP, CAS 555-60-2) was acquired from TargetMol (Boston, MA, USA). Animal experiment Specific pathogen-free 4-week-old female C57BL/6J mice and nude mice were obtained from Guangdong GemPharmatech Co., Ltd. Animals were housed in a barrier facility under controlled conditions of 22 ± 2 °C, 55 ± 5% relative humidity, and a 12-h light/12-h dark cycle. After acclimatization, mice were randomly assigned to the indicated experimental groups. Treatment administration was not performed under blinded conditions. However, intestinal tissue sections were coded before evaluation, and histopathological assessment was performed by an experienced pathologist who was blinded to group allocation. All animal procedures were approved by the Institutional Animal Care and Use Committee and conducted in accordance with the approved protocol (IACUC-20231106-03 and IACUC-20241202-02). According to our preliminary animal study, 10 mg/kg per administration represented the lowest observed adverse effect level for intestinal tissues under subchronic exposure. This provides a crucial reference point for the long-term safety assessment of Avo, highlighting potential risks under conditions of sustained, high-intensity use. Thus, to explore dose–response (Experiment 1) relationships under repeated exposure conditions, three Avo doses were chosen for mice: low (10 mg/kg), medium (20 mg/kg), and high (40 mg/kg). Avo was dissolved in corn oil and administered by oral gavage every two days for 60 days, which corresponded to nominal time-averaged doses of 5, 10, and 20 mg/kg/day, respectively. To further contextualize the human relevance, the lowest tested daily dose (5 mg/kg/day) corresponds to a human equivalent dose of approximately 0.41 mg/kg/day for a 60 kg adult, based on body surface area normalization [19]. Although these doses exceed the estimated systemic exposure from typical dermal sunscreen use (estimated Systemic Exposure Dose of 0.083–0.29 mg/kg/day) [15], they are relevant for modeling oral ingestion, a plausible exposure route via unintentional intake or environmental bioaccumulation. Importantly, these doses represented only 1.1%–4.4% of the reported no-observed-adverse-effect level (NOAEL; 450 mg/kg/day) established in a subchronic dietary toxicity study in rats [20], thereby enabling the assessment of subclinical toxicity that might not be apparent at lower environmental concentrations. For Experiment 2 (combination-treatment study): C57BL/6J mice (n = 24) were treated for 60 days. The control group received corn oil gavage (3 mL/kg) with solvent (5% DMSO in saline, i.p.); the Avo monotreatment group received 20 mg/kg Avo gavage + solvent i.p.; the rapamycin combination group received 20 mg/kg Avo gavage + 5 mg/kg rapamycin i.p.; and the ISRIB combination group received 20 mg/kg Avo gavage + 5 mg/kg ISRIB i.p. Post-treatment processing matched Experiment 1. Plasmids and transfection Full-length Beclin1-3×Myc and ATG14-Flag constructs were generated by PCR amplification and cloned into the lentiviral pLVX-puro vector (Takara Bio, CA, USA) using the CloneExpress II One-Step Cloning Kit (Vazyme, Nanjing, China). For protein expression studies, Caco-2 cells and HEK293 were transfected with pLVX-Beclin1-3×Myc, pLVX-ATG14-Flag, or the empty pLVX-puro vector using Lipofectamine 3000 (Invitrogen, CA, USA) according to the manufacturer’s instructions. All plasmid constructs were verified by Sanger sequencing to confirm accurate insertion and preservation of the reading frames (Synbio Technologies, Suzhou, China). For ATG14 silencing, HEK293 and Caco-2 cells were transfected with the following siRNA oligonucleotides targeting human ATG14 mRNA using the same transfection method: #1, 5′-GCGAUUUCGUCUACUUCGA-3′; #2, 5′-GCAAGAUGAGGAUUGAACA-3′; and #3, 5′-GGGUCUGUGACGAUCACAA-3′ (Synbio Technologies). Cell viability assay Cells were seeded in 96-well plates at a density of 1 × 103 cells/well. After a 12 h incubation period, the cells were exposed to various specified concentrations of Avo for the designated time intervals of 24, 48, and 72 h. Subsequently, cell viability was assessed using the CCK-8 assay (TargetMol). The CCK-8 reagent was added to the cultured cells at 10% working concentration, and the plates were then incubated at 37 °C for 1.5 h. Finally, the absorbance was measured at 450 nm using a microplate reader (BioTek Instruments, Vermont, USA). Based on the half-maximal inhibitory concentration (IC₅₀) values determined by CCK-8 assay, 0–60 µM Avo for 24 h was selected for subsequent cellular experiments, the lower end concentration (20 µM) corresponds to the threshold where we began observing significant cellular responses, paralleling the lowest observed adverse effect level identified in our mouse model (10 mg/kg). Cell proliferation assay Cell proliferation was assessed using the BeyoClick™ EdU Cell Proliferation Kit (Beyotime, Shanghai, China). Cells were seeded in glass-bottom dishes at a density of 5 × 104 cells per well and incubated overnight at 37 °C. Subsequently, the cells were exposed to varying concentrations of Avo (0, 20, 40, and 60 µM) and cultured for an additional 48 h. Following the manufacturer’s instructions, the cells were incubated with 10 µM EdU at 37 °C for 90 min. Then, they were fixed with 4% paraformaldehyde (Beyotime), followed by permeabilization with 0.5% Triton X-100 (Beyotime). Next, the cells underwent fluorescent labeling via the click reaction solution in the dark for 30 min at room temperature. Finally, the cells were stained with Hoechst 33342 and imaged using a confocal laser scanning microscope (LSM900, Carl Zeiss, Jena, Germany) at a 20× magnification. Assessment of ROS Cells were seeded in 12-well plates and treated with the indicated concentrations of compounds for the specified durations. Intracellular ROS was assessed using the general redox-sensitive probe DCFH-DA (Beyotime), according to the manufacturer’s instructions. Briefly, cells were incubated with DCFH-DA in serum-free medium at 37 °C for 20–30 min in the dark, washed twice with PBS, and analyzed by flow cytometry (BD Biosciences, CA, USA). Mean fluorescence intensity was quantified using FlowJo software (BD Biosciences). H2O2-treated group was used as a positive control. DCFH-DA signals were used as supportive indicators of intracellular oxidative changes and were interpreted together with oxidative damage and mitochondrial functional assays. Confocal assay Confocal imaging assays were performed on HEK293-mCherry-EGFP-LC3B-Teton and HEK293 cells to measure intracellular LC3B levels, assess mitochondrial–LC3B colocalization, observe mitochondrial morphology changes and evaluate mitochondrial-lysosome colocalization. The cells were seeded in confocal dishes. To visualize the cell nuclei, they were stained with DAPI (Beyotime). For mitochondrial visualization, cells were stained with 100 nM MitoTracker Red (Thermo Fisher Scientific). To visualize lysosomes, cells were stained with 50 nM LysoTracker Green (Thermo Fisher Scientific). Cells were treated with different drug concentrations for varying durations, and imaged using a confocal laser scanning microscope (LSM900, Carl Zeiss). For mitochondrial morphology analysis, HEK293 cells were seeded in glass-bottom confocal dishes and treated with the indicated concentrations of Avo for 24 h after cell attachment. After staining with MitoTracker Red at 37 °C, the cells were washed twice with prewarmed phosphate-buffered saline (PBS) and then imaged using a confocal laser scanning microscope (LSM900, Carl Zeiss) for mitochondrial morphology analysis. Cells were classified as tubular (elongated, interconnected networks) or fragmented (short, punctate mitochondria) based on their predominant morphology [21, 22]. For each condition, three independent experiments were conducted, with 30 randomly selected cells analyzed per experiment. The percentage of cells in each category served as a single biological replicate for statistical analysis. Immunofluorescence detection of oxidative damage (4-HNE) Oxidative damage was assessed by immunofluorescence staining of 4-hydroxynonenal (4-HNE) in cells. Cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and blocked with blocking buffer. Samples were incubated with an anti-4-HNE primary antibody (Abcam, ab46545, 1:100) overnight at 4 °C, followed by incubation with the appropriate fluorophore-conjugated secondary antibody. Nuclei were counterstained with DAPI. Fluorescence images were acquired using an LSM900 confocal microscope (Carl Zeiss) equipped with a 40× objective, with identical acquisition settings applied across all groups. Fluorescence intensity was quantified under blinded conditions using ZEN software (Carl Zeiss). For each condition, 30 cells were randomly selected from multiple fields of view in each of three independent experiments. The mean fluorescence intensity of the 30 cells in each independent experiment was calculated, and the resulting three independent values were used for statistical analysis and graphical presentation. Measurement of mitochondrial membrane potential Mitochondrial membrane potential was assessed using tetramethylrhodamine methylester (TMRM, I34361, Invitrogen) staining, together with MitoTracker Green (M46750, Invitrogen). Following the indicated treatments, cells were first incubated with MitoTracker Green at 37 °C for 15 min to label mitochondrial mass. Cells were then incubated with TMRM for an additional 30 min to assess mitochondrial membrane potential. After staining, cells were washed with PBS and analyzed by flow cytometry. ΔΨm was expressed as the ratio of TMRM fluorescence to MitoTracker Green fluorescence. Where indicated, CCCP was used as a positive control for mitochondrial membrane potential dissipation. Measurement of the intracellular lipid droplet levels To measure intracellular lipid droplet levels, cells were treated with different concentrations of drugs for varying durations. Following the manufacturer’s instructions, the cells were stained with 10 µM BODIPY 493/503 dye (MCE, Shanghai, China) for 30 min and then washed twice with PBS. Detection was carried out using either a confocal laser scanning microscope (LSM900, Carl Zeiss) or a flow cytometer (BD Biosciences). To quantify intracellular FFA content, cells seeded in 12-well plates were exposed to different drug concentrations for the indicated durations. Samples were processed according to the standardized protocol of the Amplex Red Free Fatty Acid Assay Kit (Beyotime). Following cell lysis with BeyoLysis™ Buffer A and centrifugation at 12,000 g for 5 min, supernatants were incubated with the working solution for 30 min at 37 °C in the dark. Absorbance was measured at λ = 570 nm using a microplate reader (BioTek Instruments). Quantification was achieved by establishing a standard curve with serially diluted palmitic acid solutions (0–100 µM). Measurement of the autophagic flux and mitophagic flux The tandem fluorescent timer system relies on the differential stability of EGFP and mCherry under acidic conditions. While cargos in non-acidic compartments retain both signals, fusion with acidic lysosomes results in the quenching of EGFP and the persistence of mCherry [23]. For autophagy analysis, cells stably expressing mCherry-EGFP-LC3B reporter [24] were treated with Avo or EBSS. Fluorescence signals were visualized using a confocal laser scanning microscope (LSM 900, Carl Zeiss) to evaluate the transition of autophagosomes to autolysosomes. For mitophagy analysis, cells stably expressing COX8-EGFP-mCherry reporter were treated with Avo or CCCP (positive control). Samples were analyzed by flow cytometry (BD Biosciences), and the mitophagic flux was quantified based on the percentage of EGFP⁻/mCherry⁺ cells, representing mitochondria delivered to acidic lysosomal compartments. Dual luciferase reporter assay Wnt/β-catenin signaling pathway activation levels were detected using a dual-luciferase reporter gene system. HEK293 cells were co-transfected with the SuperTOPFlash-CP reporter plasmid (D2506, Beyotime) containing TCF/LEF response elements and the internal control plasmid pRL-TK (E6911, Beyotime) at a 10:1 mass ratio. After 48 h post-transfection, the indicated treatments were applied. Cells were lysed and the Beyotime Dual-Luciferase Reporter Assay Kit (RG027) was used to sequentially detect Firefly/Renilla luciferase activities (BioTek Synergy LX microplate reader). The ratio of firefly to Renilla was used to normalize the firefly luciferase values. Immunoblotting For immunoblotting, after centrifugation to obtain the cell pellet, an appropriate amount of lysis buffer (Cell Signaling Technology, MA, USA) was added. The cells were lysed on ice for 30 min, followed by ultrasonic treatment for 3 min. The mixture was then centrifuged at 12,000 g for 30 min at 4 °C. The supernatant was used to determine the protein concentration using a BCA assay kit (Thermo Fisher Scientific, MA, USA). Protein samples were separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to a PVDF membrane (Bio-Rad, CA, USA) via electroblotting. After blocking with 5% non-fat milk for 1.5 h, the membrane was incubated with the primary antibody overnight at 4 °C. Following three washes with Tween Tris-buffered saline (TBST), the membrane was incubated with the corresponding secondary antibody at room temperature for 2 h. After three additional washes with TBST, the membrane was imaged using Clarity Western ECL substrate (Bio-Rad) and the signals were detected with a Tanon 5200-Multi (Tanon Science & Technology Co., Ltd., Shanghai, China). The antibodies used were as follows: anti-p62 (Proteintech, 18420-1-AP, 1:10000), anti-LC3B (Abclonal, A19665, 1:2000), anti-β-Actin (Abclonal, AC026, 1:20000), anti-TIM23 (Proteintech, 11123-1-AP, 1:3000), anti-TOM20 (Proteintech, 11802-1-AP, 1:3000), anti-ATG14 (Proteintech, 19491-1-AP, 1:1000), anti-FLAG tag (MBL, M185-3L, 1:10000), anti-CUL3 (Proteintech, 11107-1-AP, 1:2000), anti-ubiquitin (Proteintech, 10201-2-AP, 1:3000), anti-MYC tag (Proteintech, 60003-2-Ig, 1:5000), anti-VPS34 (Proteintech, 12452-1-AP, 1:2000), anti-Beclin 1 (Proteintech, 11306-1-AP, 1:1000), anti-LONP1 (Proteintech, 15440-1-AP, 1:10000), anti-HSP60 (Proteintech, 15282-1-AP, 1:10000), anti-CLPP (Proteintech, 15698-1-AP, 1:5000), anti-LGR5 (Immunoway, YN2558, 1:1000), anti-phospho-eIF2α (Ser51) (Cell Signaling Technology, 3398S, 1:1,000), anti-eIF2α (Proteintech, 82936-8-RR, 1:3,000), anti-ATF4 (Proteintech, 60035-1-Ig, 1:3,000), anti-ATF5 (Proteintech, 67066-1-Ig, 1:3,000), anti-β-catenin (Cell Signaling Technology, 8480, 1:1,000), Goat anti-Mouse IgG(H+L)-HRP (Bioworld, BS12478, 1:5000) and Goat anti-Rabbit IgG (H + L)-HRP (Bioworld, BS13278, 1:5000). Immunoprecipitation For immunoprecipitation, cells were treated with different concentrations of Avo for 24 h. According to the manufacturer’s instructions, 1 mL of Western and IP cell lysis buffer (Beyotime) was added, and the cells were lysed on ice for 30 min. The lysate was then centrifuged at 12,000 g for 30 min at 4 °C, and the supernatant was collected. After measuring protein concentration using the BCA method, an appropriate amount of sample was taken and incubated with Nanobody Agarose Beads (AlpaLifeBio, Shenzhen, China) at 4 °C with gentle inversion for 2 h. The mixture was centrifuged at 4 °C, 1200 g for 3 min, and the supernatant was discarded. The beads were washed five times with PBST. Finally, an appropriate amount of loading buffer was added to resuspend the beads, and the mixture was heated in a 95 °C water bath for 10 min to release the immunoprecipitate from the beads. The supernatant was collected for analysis by Western blot. Recombinant ATG14 expression, purification, and fluorescence titration Recombinant ATG14 was expressed as an N-terminal GST-tagged fusion protein using a pGEX bacterial expression vector in Escherichia coli BL21 (DE3) cells. Bacteria were cultured in LB medium supplemented with ampicillin (100 µg/mL), and protein expression was induced with 0.2 mM IPTG at 16 °C overnight when the optical density at 600 nm (OD₆₀₀) reached approximately 0.6. Cells were harvested, washed with PBS, lysed by repeated freeze–thaw cycles, and clarified by centrifugation. GST-ATG14 protein was purified using glutathione affinity chromatography. The GST tag was subsequently removed by protease cleavage, and the tag-free ATG14 protein was further purified to remove the GST tag and protease. Purified ATG14 was concentrated by ultrafiltration and buffer-exchanged into PBS for subsequent assays. The interaction between Avo and ATG14 was analyzed by intrinsic fluorescence titration using a fluorescence spectrophotometer (F-7000, Hitachi, Japan). Tag-free ATG14 protein was diluted to a final concentration of 1 µM in PBS and titrated with increasing concentrations of Avo (0–6 µM). Avo was pre-dissolved in DMSO to generate a 100 mM stock solution and subsequently diluted with PBS, with the final DMSO concentration kept constant across all samples. Fluorescence emission spectra were recorded from 300 to 400 nm with an excitation wavelength of 280 nm. Fluorescence quenching (ΔF) was calculated relative to the ligand-free control, and binding curves were fitted to a one-site binding model to estimate the dissociation constant. Molecular docking The 3D structure of the ATG14 protein was predicted using AlphaFold based on its amino acid sequence obtained from the UniProt database (UniProt ID: Q6ZNE5). The 3D structure of Avo was retrieved from the PubChem database (PubChem CID: 51040). Protein and ligand structures were pre-processed using AutoDock Tools 1.5.7: polar hydrogen atoms were added and Gasteiger charges were calculated for the protein, with crystallographic water molecules removed where applicable, and the processed protein was saved in PDBQT format. Similarly, Gasteiger charges were added and rotatable bonds were defined for the Avo ligand, which was also saved in PDBQT format. Molecular docking calculations were performed using AutoDock Vina. The docking search space was defined by a grid box centered at coordinates X = -0.044, Y = -0.429, Z = 0.216 Å, with dimensions 38.25 Å × 36.75 Å × 37.5 Å. The docking procedure employed AutoDock Vina’s iterative local search algorithm and its built-in semi-empirical scoring function to optimize and evaluate ligand binding conformations. Following docking, the conformation with the lowest predicted binding free energy (ΔG) was selected as the most probable interaction mode. Finally, the docking results were visualized using PyMOL 2.7. Molecular dynamics simulations Molecular dynamics (MD) simulations were performed using the GROMACS 2026 software package to evaluate the stability and dynamic interactions of the protein–ligand complex. The protein receptor was parameterized using the CHARMM36 force field, whereas the force-field topology and parameters for the small-molecule ligand Avo were generated using the CGenFF server. The complex was placed in a periodic dodecahedral box and solvated with the TIP3P water model, with a minimum distance of 1.2 nm between the protein and the box boundaries. Appropriate numbers of Na⁺ and Cl⁻ ions were added to neutralize the net charge of the system and maintain a physiological ionic strength of 0.15 M. Following energy minimization using the steepest-descent algorithm, the system was equilibrated for 100 ps under the NVT ensemble using the velocity-rescaling thermostat at 298 K, followed by 100 ps under the NPT ensemble using Berendsen pressure coupling at 1 bar. Subsequently, a 100-ns production MD simulation was performed under the NPT ensemble using Parrinello–Rahman pressure coupling. After completion of the simulation, the root-mean-square deviation (RMSD) and radius of gyration (Rg) of the system were calculated using the gmx rms and gmx gyrate modules in GROMACS, respectively. RT-qPCR Total RNA was extracted using the HiPure Total RNA Kit (Magen, Guangzhou, China) according to the manufacturer’s instructions. The RNA was then reverse transcribed into cDNA using Hiscript III Reverse Transcriptase (Vazyme). Subsequently, qRT-PCR was performed to detect the relative expression of the target gene using ChamQ Blue Universal SYBR qPCR Master Mix (Vazyme). The primer sequences are as follows: ATG14: 5′-GCGCCAAATGCGTTCAGAG-3′ (forward) and 5′-AGTCGGCTTAACCTTTCCTTCT-3′ (reverse); GAPDH: 5′-GGAGCGAGATCCCTCCAAAAT-3′ (forward) and 5′-GGCTGTTGTCATACTTCTCATGG-3′ (reverse). Mass spectrometry and bioinformatics analyses Protein digestion and quantitative proteomics analysis based on data-independent acquisition (DIA) were performed as previously described. Briefly, HEK293 cells treated with Avo (40 µM) for 24 h were lysed using lysis buffer (Cell Signaling Technology). The proteins were then digested with trypsin (Life, MD, USA) at 37 °C for 16 h. The peptide fragments were vacuum freeze-dried, resuspended in anhydrous acetonitrile solution, and desalted using MonoTIP C18 (GL Sciences, Tokyo, Japan). Peptide samples were analyzed using an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific). The raw data were processed using Spectronaut software (Omicsolution, Shanghai, China), with a protein and peptide false discovery rate (FDR) set to 1%. Differentially expressed proteins (DEPs) were analyzed using the Kyoto Encyclopedia of Genes and Genomes (KEGG). Human intestinal organoid culture Human intestinal organoids were established from surgically resected intestinal tissues, which were obtained from patients with colorectal cancer. These tissues were taken from adjacent non‑tumor regions located more than 3 cm away from the tumor margins and were cultured as previously described [25]. In brief, the tissues were dissected and cleaned with PBS at 4 °C, the segments were incubated with 2 mM EDTA solution (pH 8.0) for 20 min at 4 °C, vortexed for 2 min and centrifuged at 110 g for 5 min. After the supernatant was removed, the intestinal crypts were gently flicked with DMEM/F-12 medium (Gibco), centrifuged at 110 g for 5 min, and resuspended with DMEM/F-12 medium. 10 µL intestinal crypt suspension and 60 µL Matrigel (R&D Systems) were mixed, added into a 24-well cell culture plate, and polymerized for 15 min at 37 °C. The intestinal crypts were then cultured in culture medium as previously described [25, 26]. The experiment was approved by institutional review board of Jinan University (Approval number: JNUKY-2023-0092). Organoids were dissociated into single-cell suspensions by TrypLE Express Enzyme (GIBCO; 37 °C, 15 min), cells were resuspended in pre-cooled Matrigel and seeded into 96-well plates (10 µL/well). After Matrigel polymerization, complete organoid medium (100 µL/well) was added. Cells were treated with Avo (0–100 µM) for 72 h, and cell viability was assessed by CCK-8 assay (Beyotime). The concentration of Avo before IC₅₀ (20 µM) was selected for subsequent experiments. For long-term studies, organoids were treated with Avo (20 µM) alone or in combination with rapamycin (100 nM) for two weeks. Organoid morphology was monitored by microscopy. Bright-field images were captured from five random fields per well, and organoids with a diameter > 20 μm were counted for quantification. For cell death assessment, cells were stained with propidium iodide (PI; 20 µg/mL; Solarbio, C0080, Beijing, China) at 37 °C for 20 min in the dark. After two PBS washes, fluorescence was measured (Ex 535 nm / Em 615 nm). The fluorescent multiplex immunohistochemistry of organoids was performed as we previously described [27]. Organoids from each patient were maintained as an independent organoid line, with three independent biological experiments performed per treatment condition. Immunofluorescence and immunohistochemistry Mouse intestinal samples and human intestinal organoids were prepared by washing with pre-chilled PBS, dissecting the middle segment, fixing in 4% formaldehyde, and sectioning using a cryostat (Leica Microsystems, Wetzlar, Germany). For immunohistochemistry, formalin-fixed, paraffin-embedded sections were dewaxed, rehydrated, underwent antigen retrieval, and endogenous peroxidase activity was quenched with H2O2. Sections were blocked with 5% serum for 30 min and incubated with the primary antibody (rabbit anti-Ki67, Abcam, 1:100) overnight at 4 °C, followed by a secondary antibody (HRP-conjugated anti-rabbit, Invitrogen, 1:200) for 2 h. For Hematoxylin-Eosin (H&E) staining, rehydrated sections were stained with Hematoxylin and Eosin (Beyotime, C0105) for 2 min, followed by washing. Images were captured and analyzed using Motic DSAssistant software. For β-catenin, HSP60, LONP1, and LGR5, ZO-1, and occludin staining, a TSA-based fluorescence kit (Shanghai Ruike Biotechnology Co., Ltd., China) was used as per the manufacturer’s instructions. Confocal imaging was performed at 10× magnification using a laser scanning microscope (LSM900, Carl Zeiss). For histological quantification, measurements from multiple sections and fields were averaged per mouse, and the resulting per-animal means (n = 6 per group) were used for group-level statistical comparisons. Statistical analysis All experiments were performed at least three times independently. Statistical analyses were conducted using GraphPad Prism software. For comparisons between two groups, an unpaired two-tailed Student’s t-test was used. For comparisons among multiple groups, one-way analysis of variance (ANOVA) was applied. For time-course or dose-response experiments involving two independent variables, two-way ANOVA was used. Quantitative data are expressed as mean ± standard deviation (SD). Error bars represent SD in all graphical representations. Statistical significance is denoted as follows: *p < 0.05, **p < 0.01, ***p < 0.001; ns indicates no significant difference. Results Avo exposure disrupts intestinal epithelial homeostasis in vitro and in vivo To explore potential adverse events associated with Avo exposure in humans, we first examined pharmacovigilance records from the VigiAccess and FDA Adverse Event Reporting System databases. Gastrointestinal events were observed in the pharmacovigilance data, albeit at a lower frequency compared to skin and general disorders (Figures S1A-B). These findings prompted us to investigate whether chronic Avo exposure could affect intestinal homeostasis. In a previous subchronic dietary toxicity study in Sprague–Dawley rats, the no-observed-adverse-effect level (NOAEL) of Avo was reported to be 450 mg/kg/day [15, 20]. Here, we established a dose-response gavage mouse model to evaluate its pathophysiological effects following long-term intake of increasing doses of Avo (0, 10, 20 and 40 mg/kg) every two days for two months (Fig. 1A). Histopathological analysis was performed to evaluate the effect of Avo on intestinal tissues. Hematoxylin-Eosin (H&E) staining demonstrated significant reductions in crypt number, intestinal villus length, and crypt height (Fig. 1B), while a shortened small intestinal length further indicated macroscopic structural abnormalities (Fig. 1C). Immunohistochemical (IHC) analysis showed that Avo exposure significantly downregulated the expression of the proliferative marker Ki67 (Fig. 1D) in crypts, suggesting impaired regenerative capacity of the intestinal epithelium. In addition, immunofluorescence staining revealed that Avo markedly decreased the expression of the tight junction proteins Occludin and ZO-1 in intestinal tissues (Figs. 1E-F), indicating disruption of epithelial barrier integrity. Consistently, in vitro experiments in Caco-2 and HEK293 cells showed that Avo suppressed cell viability in a time- and concentration-dependent manner, as determined by CCK-8 and EdU assays (Figs. 1G-H; Figures S1C-D). Collectively, these findings demonstrate that Avo treatment induces intestinal damage and suppresses cell viability. Avobenzone induces intestinal epithelial damage by inhibiting intestinal crypt cell proliferation. A Schematic of Avo administration in nude mice. Mice received Avo (0, 10, 20, or 40 mg/kg) by oral gavage every other day for 2 months. B Representative H&E-stained small intestinal sections and quantification of crypt density, villus length, and crypt height. Scale bar, 100 μm. C Representative small intestine images and length quantification. D Representative immunohistochemical staining of Ki-67 in intestinal crypts and quantification of Ki-67-positive cells per crypt. A total of 30 well-oriented villus–crypt units (B) or 30 crypts (D) were analyzed per mouse, and the mean value for each mouse was used for statistical analysis (n = 6 mice per group). E, F Representative immunofluorescence images and fluorescence intensity quantification of Occludin (E) and ZO-1 (F) expression in small intestinal tissues. Five randomly selected fields were analyzed for each mouse, and the mean value obtained from the five fields was used as one independent data point for statistical analysis (n = 6 mice per group). Scale bars, 50 μm. G Cell viability of Caco-2 cells treated with increasing concentrations of Avo for 24, 48, and 72 h was assessed by CCK-8 assay. H EdU incorporation in Caco-2 cells treated with Avo (0–60 µM) for 48 h. EdU-positive cells are shown in red and nuclei in blue (Hoechst 33342). Five randomly selected fields were analyzed per condition in each of three independent experiments, and the mean value from each experiment was used for statistical analysis (n = 3). Scale bar, 20 μm. Data are presented as means ± SD. For (B-F and H), differences among groups were analyzed using one-way ANOVA followed by Dunnett’s multiple comparison test versus the control group. For (G), statistical significance was determined by two‑way ANOVA followed by Tukey’s multiple comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001 Avo triggers mitochondrial dysfunction To dissect the molecular mechanism underlying Avo-induced intestinal damage, we performed data-independent acquisition mass spectrometry (DIA-MS) to profile differentially expressed proteins (DEPs) regulated by Avo (Fig. 2A). More than 8300 proteins were identified in each group and displayed normal distribution in protein expression (Figures S2A-B), suggesting high data reliability. Principal component analysis (PCA) of three biological replicates confirmed a clear separation between the control and Avo-treated groups (Fig. 2B). The Avo-regulated DEPs were visualized using a heatmap (Fig.2C), and KEGG pathway enrichment analysis of Avo-regulated DEPs revealed significant enrichment of lysosome and autophagy pathways, Wnt signaling pathway, fatty acid metabolism, and mitochondria-related pathways (TCA cycle, oxidative phosphorylation) (Fig. 2D), suggesting that perturbations in mitochondrial, lysosomal, and autophagic processes may contribute to Avo-induced cytotoxicity. To test this effect, we confirmed that Avo induced a dose-dependent increase in intracellular ROS (Fig. 2E) and enhanced lipid oxidative damage, as shown by 4-HNE staining (Fig. 2F). Furthermore, Avo-treated cells exhibited impaired mitochondrial function, evidenced by a significant loss of mitochondrial membrane potential (MMP) and increased mitochondrial fragmentation (Figs. 2G and Figure S2C). Since mitochondrial dysfunction is closely linked to lipotoxicity [28], we assessed intracellular lipid homeostasis. Avo exposure led to a significant accumulation of neutral lipid droplets and a dose-dependent increase in intracellular free fatty acid (FFA) content (Figures S2D-F). Collectively, these findings demonstrate that Avo disrupts mitochondrial homeostasis, leading to its anti-proliferative toxicity. Avobenzone induces mitochondrial dysfunction and lipid oxidation. A Schematic of the data-independent acquisition mass spectrometry (DIA-MS) workflow used to identify Avo-induced proteomic alterations in HEK293 cells. B Principal component analysis (PCA) showing sample clustering patterns. C Heatmap showing the expression patterns of Avo-regulated DEPs in the control and Avo-treated groups. D KEGG pathway enrichment analysis of all differentially expressed proteins (DEPs). E, F Caco-2 and HEK293 cells were treated with Avo at 0–60 µM for 24 h; H₂O₂ served as a positive control at 400 µM for Caco-2 cells and 200 µM for HEK293 cells for 30 min. Intracellular oxidative status was assessed using the general redox-sensitive probe DCFH-DA followed by flow cytometric analysis (E). Oxidative damage was assessed by immunofluorescence staining of 4-hydroxynonenal (4-HNE) (F). For quantitative analysis of immunofluorescence, 30 cells per condition were randomly selected from multiple fields of view in each independent experiment, and the mean fluorescence intensity values were calculated. G Caco-2 and HEK293 cells were treated with Avo (0, 20, 40, or 60 µM) for 24 h; CCCP (10 µM, 6 h) was used as a positive control. Mitochondrial membrane potential (ΔΨm) was assessed by TMRM staining, and mitochondrial mass was determined using MitoTracker Green. TMRM fluorescence was normalized to MitoTracker Green to obtain mitochondrial mass–corrected ΔΨm and analyzed by flow cytometry. Data are presented as mean ± SD. n = 3. Statistical significance for multiple groups was determined by one-way ANOVA followed by Dunnett’s multiple comparison test versus the control group. *p < 0.05, **p < 0.01, ***p < 0.001 Avo inhibits autophagic degradation of mitochondria Based on the DIA-MS result (Fig. 2D), we asked whether Avo plays a suppressive role in autophagy. Under nutrient-replete conditions, Avo alone did not markedly alter SQSTM1/p62 or LC3B-II levels (Figure S3A). However, under EBSS-induced starvation, Avo dose-dependently attenuated LC3B-II accumulation and SQSTM1/p62 degradation in both Caco-2 and HEK293 cells (Figs. 3A and Figure S3B). To dynamically monitor autophagic flux, we employed live-cell imaging of HEK293 cells expressing mCherry-EGFP-LC3B reporter and found that, upon starvation-induced autophagy activation via EBSS exposure, Avo dose-dependently reduced both mCherry+/EGFP+ (autophagosomes) and mCherry+/EGFP− (autolysosomes) puncta (Fig. 3B), indicating defective autophagosome formation induced by Avo. Mitophagy is a critical quality control mechanism that selectively eliminates dysfunctional mitochondria to safeguard cellular energy homeostasis [29]. Our DIA-MS result indicated that Avo-accumulated proteins mainly enriched in mitochondria (TCA cycle, Fig. 2D), we therefore hypothesized that Avo suppresses autophagic degradation of mitochondria. Mitophagy-specific analyses using mitochondrial uncoupler carbonyl cyanide m-chlorophenyl hydrazone (CCCP) showed that Avo substantially diminished CCCP-induced mitochondria–LC3B colocalization (Fig. 3C) and mitochondria–lysosome colocalization (Fig. 3D), suggesting Avo inhibits autophagosome formation and blocks the lysosomal clearance of mitochondria. To directly assess mitophagic flux rather than mitochondrial accumulation, we used the tandem COX8-EGFP-mCherry reporter, and found that Avo reduced the proportion of EGFP−/mCherry+ cells under basal conditions (Fig. 3E) and attenuated the CCCP-induced increase in this population in both Caco-2 and HEK293 cells (Fig. 3F), suggesting impaired mitophagic flux. Consistent with defective clearance of damaged mitochondria, Avo enhanced CCCP-induced ROS accumulation and aggravated mitochondrial membrane potential loss (Figs. 3G and H). Collectively, these results indicate that Avo impairs autophagic flux and mitophagy, leading to defective clearance of damaged mitochondria and enhanced oxidative stress. Avobenzone suppresses autophagic flux and mitophagy. A Western blot analysis of p62 and LC3B in Caco-2 and HEK293 cells treated with Avo (0–60 µM, 24 h), with or without EBSS starvation for 4 h (Caco-2) or 6 h (HEK293). B Confocal images of mCherry-EGFP-LC3B-expressing HEK293 cells treated with Avo (0–60 µM, 24 h), with or without EBSS starvation (6 h); green and red LC3B puncta per cell were quantified. C Mitochondrial colocalization with LC3B puncta in HEK293-EGFP-LC3B cells treated with Avo (0 or 40 µM, 24 h), with or without CCCP (10 µM, 6 h). Mitochondria were labeled with MitoTracker; intensity profiles along the indicated lines are shown. D Mitochondria–lysosome colocalization in HEK293 cells treated with Avo (0–60 µM, 24 h), followed by E64D (10 µM, 12 h). Mitochondria and lysosomes were labeled with MitoTracker (red) and LysoTracker (green), respectively, and colocalized puncta were quantified. For (B, D), 30 randomly selected cells per condition were analyzed from multiple fields in each of three independent experiments. Scale bars, 5 μm. E, F Flow cytometric analysis of mitophagy, indicated by EGFP−/mCherry+ cells, in Caco-2- and HEK293-mCherry-EGFP-COX8 cells. E Cells were treated with Avo (0–60 µM, 24 h), with CCCP (10 µM, 6 h) as a positive control. F Cells were treated with Avo for 24 h and then CCCP (10 µM, 6 h) to assess CCCP-induced mitophagy. G, H Caco-2 and HEK293 cells were treated with Avo (0–60 µM, 24 h), with or without CCCP (10 µM, 6 h). G Mitochondrial membrane potential was assessed by TMRM staining and normalized to mitochondrial mass measured using MitoTracker Green. H Intracellular ROS was assessed by DCFH-DA staining and flow cytometry. Data are presented as mean ± SD. Two-group comparisons used unpaired two-tailed Student’s t-test; multiple-group comparisons used one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001 Avo promotes proteasomal degradation of ATG14 and impairs autophagy initiation To investigate the molecular mechanism underlying Avo-mediated inhibition of autophagy initiation, we analyzed DIA-MS data of Avo with an autophagy-related proteome database (Fig. 4A). Heatmap clustering of Avo-altered DEPs revealed a significant downregulation of ATG14, a core component of the PI3KC3-C1 complex, following Avo treatment (Fig. 4B). Western blotting confirmed that Avo markedly reduced ATG14 protein levels in Caco-2 and HEK293 cells as well as in murine intestinal tissues (Figs. 4 C and S4A). We therefore investigated whether Avo has the potential to interact with ATG14 and thereby promote its degradation. Fluorescence titration showed dose-dependent quenching of ATG14 fluorescence by Avo, with an estimated dissociation constant (Kd) of approximately 3.39 µM, suggesting a potential interaction between Avo and ATG14 (Fig. 4D). Subsequent molecular dynamics simulations supported the stability of the predicted Avo–ATG14 complex (Figure S4B). Molecular docking predicted a favorable binding conformation with a calculated binding free energy of − 7.1 kcal/mol (Figure S4C). Taken together, these results indicate that Avo possesses the potential to bind ATG14. Since qPCR analysis revealed no change in ATG14 mRNA levels (Figure S4D), we speculated that Avo treatment resulted in ATG14 protein degradation. By using degradation inhibitors, we found that the proteasome inhibitor MG132, but not the lysosomal inhibitor bafilomycin A1 (Baf-A1), reversed Avo-induced ATG14 degradation (Figs. 4E and S4E). Cycloheximide chase assay showed that Avo significantly shortened the half-life of the ATG14 protein (Fig. 4F). The ZBTB16-Cullin3-Roc1 E3 ubiquitin ligase complex has been reported to modulate ubiquitination and degradation of ATG14 [30]. We therefore tested whether Cullin3 is responsible for Avo-induced ATG14 degradation. Co-immunoprecipitation assay showed enhanced interaction between Cullin3 and ATG14 along with increasing Avo treatment, which was accompanied by increased ubiquitination (Figs. 4G and S4F). Furthermore, Avo disrupted the interaction between autophagy initiation complex members VPS34 and Beclin 1, impairing PI3KC3-C1 assembly (Fig. 4H, Figures S4G and S4H). These results demonstrate that Avo inhibits autophagy by promoting CUL3-mediated ubiquitination and degradation of ATG14, thereby destabilizing the autophagosome initiation machinery. To determine whether ATG14 is involved in Avo-induced mitochondrial dysfunction, we performed ATG14 overexpression and knockdown experiments in Caco-2 and HEK293 cells. ATG14 overexpression significantly attenuated Avo-induced ROS accumulation and mitochondrial membrane potential loss. Conversely, ATG14 knockdown induced similar mitochondrial abnormalities, and Avo treatment did not further exacerbate ROS accumulation or mitochondrial membrane potential loss in ATG14-depleted cells (Figs. 4I-J, and S4I–L). In Caco-2 cells, the tandem COX8-EGFP-mCherry reporter showed that ATG14 overexpression restored the mitophagic flux suppressed by Avo, whereas ATG14 knockdown reduced mitophagic flux. Notably, Avo treatment did not further decrease mitophagic flux in ATG14-depleted cells (Fig. 4K). Collectively, these findings indicate that Avo-induced loss of ATG14 impairs mitophagy, resulting in mitochondrial dysfunction. Avobenzone induces ATG14 ubiquitination and degradation to inhibit mitophagy. A Venn diagram of proteins shared between DIA-MS and the Human Autophagy Database (HADb). B Heatmap of the 10 most significantly upregulated and downregulated autophagy-related proteins in Avo-treated samples. C Western blot analysis of ATG14 in Caco-2 and HEK293 cells treated with Avo (0–60 µM, 24 h) and small intestinal tissues from mice administered Avo (0–40 mg/kg). D Fluorescence titration of Avo binding to ATG14. Intrinsic emission spectra of ATG14 (1 µM) were recorded with increasing Avo concentrations (0–6 µM), and fluorescence quenching was fitted to a one-site binding model. E Western blot analysis of ATG14 in HEK293 cells treated with Avo (40 µM, 24 h), with or without MG132 (10 µM, 6 h) or Baf-A1 (100 nM, 6 h). F ATG14 degradation kinetics measured by western blotting after CHX chase (50 µg/mL). Relative half-life was calculated from ATG14/actin ratios. G Co-immunoprecipitation analysis of ATG14–CUL3 interaction and ATG14 ubiquitination in HEK293 cells treated with Avo (0–60 µM, 24 h) and MG132 (10 µM, 6 h). H Co-immunoprecipitation analysis of Beclin 1–VPS34 interaction in HEK293 cells treated with Avo (0–60 µM, 24 h). I–K Caco-2 cells with ATG14 overexpression or knockdown were treated with Avo as indicated. Intracellular ROS levels were assessed by DCFH-DA staining and flow cytometry (I). Mitochondrial membrane potential was determined by the TMRM/MitoTracker Green fluorescence ratio (J). Mitophagic flux was assessed using the tandem COX8-EGFP-mCherry reporter and quantified as the percentage of EGFP−/mCherry+ cells (K). Data are presented as mean ± SD. For comparisons between two groups, unpaired two-tailed Student’s t-test was used; for comparisons among multiple groups, one-way ANOVA was applied. For (F), statistical significance was determined by two-way ANOVA followed by Tukey’s multiple comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001; ns, not significant Avo induces ATF5-associated UPRmt response and suppresses Wnt/β-catenin signaling The mitochondrial unfolded protein response (UPRmt) is an adaptive stress mechanism triggered by mitochondrial proteostasis imbalance, which restores misfolded proteins through coordinated induction of chaperones and proteases to maintain mitochondrial function [31, 32]. Aberrant activation of UPRmt may disrupt Wnt signaling and exacerbate aging and inflammatory-related pathological processes [11]. In this study, we observed that the protein expression levels of canonical UPRmt markers LONP1 (a mitochondrial protease) and HSP60 (a chaperonin) were significantly upregulated in Avo-treated Caco-2 and HEK293 cells, as well as in small intestine tissues, while the mitochondrial protease complex subunit CLPP remained unaffected (Figs. 5A and S5A). This indicates that Avo selectively activates a subset of UPRmt pathways. Consistently, immunofluorescence analysis of mouse intestinal tissues further confirmed enhanced fluorescence intensity of LONP1 and HSP60 following Avo treatment (Fig. 5B). Collectively, these findings indicate that Avo induces a selective mitochondrial stress response characterized by increased LONP1 and HSP60 expression. This mitochondrial stress response may contribute to Avo-induced oxidative stress and mitophagy impairment, thereby forming a coordinated network of cellular toxicity. Avobenzone induces mitochondrial stress and suppresses Wnt/β-catenin signaling-mediated intestinal stem cell maintenance. A Western blot analysis of UPRmt markers LONP1, HSP60, and CLPP in Caco-2 and HEK293 cells treated with Avo (0–60 µM, 24 h) and small intestinal tissues from mice administered Avo (0–40 mg/kg). B Representative immunofluorescence images and quantification of HSP60 and LONP1 in small intestinal tissues from Avo-treated mice. Five randomly selected, non-overlapping fields were quantified for each mouse, and the mean value from each mouse was used for statistical analysis (n = 6 mice per group). Scale bar, 50 μm. C Western blot analysis of LGR5 in Caco-2 and HEK293 cells and mouse small intestinal tissues under the conditions described in (A). D Representative immunofluorescence images and quantification of LGR5 and β-catenin in small intestinal tissues from Avo-treated mice. Thirty randomly selected crypts per mouse were analyzed for LGR5, and five randomly selected fields per mouse were analyzed for β-catenin fluorescence intensity. The mean value per mouse served as one independent data point (n = 6 mice per group). Scale bar, 50 μm. E SuperTOPFlash reporter analysis of Wnt/β-catenin activity in HEK293 cells treated with the indicated Avo concentrations for 24 h; Renilla luciferase served as the internal control. F–H Western blot analysis in Caco-2 cells. F p-eIF2α, eIF2α, ATF4, ATF5, and β-catenin after Avo treatment (0–60 µM, 24 h). G p-eIF2α, eIF2α, ATF5, β-catenin, LONP1, HSP60, and LGR5 after Avo treatment (40 µM, 24 h), with or without ISRIB (500 nM, 12 h). H Proteins listed in (G) in cells transfected with ATG14-Flag or empty vector and treated with Avo (40 µM, 24 h). Data are presented as means ± SD. Statistical significance for multiple groups was determined by one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001; ns, not significant The maintenance of intestinal epithelial homeostasis relies on the continuous proliferation and differentiation capacity of LGR5-expressing intestinal stem cells (ISCs) [33, 34] regulated by Wnt/β-catenin signaling. Activation of the UPRmt-eIF2α signaling axis is recently reported to suppress Wnt/β-catenin signaling [11]. Here, our Western blotting revealed that Avo treatment significantly reduced LGR5 protein expression in both cell models and mouse small intestinal tissues (Figs. 5C and S5B). Moreover, the immunofluorescence intensities of LGR5 and β-catenin in mouse small intestinal tissues decreased following Avo treatment (Fig. 5D). TOPFlash reporter assay further confirmed that Wnt signaling was suppressed by Avo in a dose-dependent manner (Fig. 5E). We next examined whether this inhibition was associated with activation of mitochondrial stress signaling. Avo treatment dose-dependently increased the p-eIF2α and ATF5 expression without appreciably affecting total eIF2α or ATF4 levels, concomitant with a reduction in β-catenin expression (Figs. 5F and S5C). To determine whether mitochondrial stress was sufficient to suppress Wnt signaling independently of Avo, Caco-2 cells were treated with the mitochondrial uncoupler CCCP. CCCP recapitulated the Avo-induced molecular phenotype, as evidenced by increased p-eIF2α, ATF5, LONP1, and HSP60 levels and decreased β-catenin and LGR5 expression. Co-treatment with ISRIB attenuated the CCCP-induced stress response and restored β-catenin and LGR5 expression, supporting the involvement of ISR signaling in mitochondrial stress-mediated Wnt suppression (Figure S5D). Similarly, ISRIB attenuated Avo-induced increases in p-eIF2α, ATF5, LONP1, and HSP60 and partially restored β-catenin and LGR5 expression (Figs. 5G and S5E). Given that ISRIB primarily acts on eIF2B downstream of eIF2α phosphorylation [35], the reduced p-eIF2α signal observed here may represent an indirect consequence of ISR attenuation rather than direct inhibition of eIF2α phosphorylation. Importantly, ATG14 overexpression attenuated the Avo-induced UPRmt response and partially restored β-catenin and LGR5 levels (Figs. 5H and S5F), indicating that impaired ATG14-mediated mitophagy lies upstream of mitochondrial stress signaling and Wnt inhibition. Collectively, these findings support a mechanistic model in which Avo-induced impairment of ATG14-mediated mitophagy activates an eIF2α phosphorylation- and ATF5-associated UPRmt response, which contributes to suppression of the Wnt/β-catenin–LGR5 axis and impaired intestinal epithelial regeneration. Rapamycin and ISRIB attenuate Avo-induced toxicity in vitro Rapamycin, a classical inhibitor of mTORC1 signaling, regulates cellular metabolism and stress adaptation through the mTOR–autophagy axis [36]. In this study, we tested whether rapamycin-induced autophagy could alleviate Avo-induced toxicity. Rapamycin treatment attenuated Avo-induced ROS accumulation, mitochondrial membrane potential loss, and lipid droplet accumulation (Figs. 6A-B and Figures S6A-B). Because UPRmt-associated stress signaling may suppress Wnt signaling [11], we employed ISRIB, an integrated stress response inhibitor. ISRIB partially restored the Avo-reduced expression of LGR5, similar to rapamycin (Figs. 6C-F). Moreover, the TOPFlash luciferase reporter assay showed that the Avo-induced reduction in Wnt/β-catenin transcriptional activity was partially rescued by rapamycin and ISRIB (Fig. 6G), suggesting that autophagy induction and ISR inhibition may alleviate Avo-induced toxicity. Rapamycin and ISRIB alleviate avobenzone-induced mitochondrial damage. A–B Flow cytometric analysis and quantification of intracellular oxidative status (A) and mitochondrial membrane potential (ΔΨm) assessed by TMRM and normalized to mitochondrial mass using MitoTracker Green (B) in Caco-2 and HEK293 cells treated with Avo (0, 20, 40, or 60 µM) for 24 h, with or without rapamycin (100 nM, 12 h). C, D Western blot analysis (C) and densitometric quantification (D) of LGR5 expression in Caco-2 and HEK293 cells treated with the indicated concentrations of Avo, with or without rapamycin (100 nM, 12 h). E, F Western blot analysis (E) and densitometric quantification (F) of LGR5 expression in cells treated with Avo, with or without ISRIB (500 nM, 12 h). Densitometric quantification of LGR5 protein levels, expressed as fold changes relative to the corresponding 0 µM Avo control group. G Wnt/β-catenin pathway activity was assessed using a SuperTOPFlash dual-luciferase reporter assay in HEK293 cells treated with Avo (40 µM) for 24 h, with or without ISRIB (500 nM, 12 h) or rapamycin (100 nM, 12 h). All quantitative data are presented as mean ± SD from three independent experiments (n = 3). For (A, B, G), statistical significance was determined by two-way ANOVA. For (D and F), comparisons among multiple groups were analyzed by one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001; ns, not significant Rapamycin and ISRIB alleviate Avo-induced intestinal injury in vivo We next investigated whether administration of rapamycin or ISRIB could alleviate the Avo-induced intestinal epithelial damage in mice. As shown in Fig. 7A, C57BL/6J mice, a strain with higher expression levels of LGR5 and ATG14 in intestine than BALB/c (Figure S7A), were orally administered Avo (20 mg/kg) and rapamycin (5 mg/kg) or ISRIB (5 mg/kg) as indicated. Avo-treated mice exhibited reduced small intestine length, intestinal crypt number, crypt height, and villus length, which were restored by treatment with rapamycin and ISRIB (Fig. 7B-C). Rapamycin and ISRIB treatment in vivo attenuated the Avo-induced reduction in Ki67-positive cells within small-intestinal crypts induced by Avo exposure (Fig. 7D). Moreover, rapamycin and ISRIB treatment restored the reduced tight junction proteins Occludin and ZO-1 in Avo-treated intestinal tissues (Fig. 7E–F). Together, these findings support the involvement of impaired autophagy and UPRmt-associated stress signaling in Avo-induced small-intestinal injury. Rapamycin and ISRIB alleviate avobenzone-induced intestinal epithelial injury in mice. A Diagram depicting the experimental design for assessing the effects of Avo on the small intestine in mice treated with Control, Avo (20 mg/kg), Avo (20 mg/kg) + Rapa (5 mg/kg) and Avo (20 mg/kg) + ISRIB (5 mg/kg) for 2 months. B Representative images of the small intestine are displayed, and the length of the small intestine was quantified. C Representative hematoxylin and eosin (H&E)-stained sections of small intestinal tissues. Crypt density, villus length, and crypt height were quantified. Scale bar, 100 μm. D Representative immunohistochemical images of Ki-67 expression in intestinal crypts and quantification of Ki-67-positive cells per crypt. For (C), 30 well-oriented villus–crypt units were analyzed per mouse. For (D), 30 crypts per mouse were quantified. The mean value for each mouse was used for statistical analysis (n = 6 mice per group). Scale bar, 100 μm. E–F Representative immunofluorescence images and quantification of Occludin (E) and ZO-1 (F) expression in small intestinal tissues. Five randomly selected fields were analyzed for each mouse, and the mean fluorescence intensity obtained from the five fields was used for statistical analysis (n = 6 mice per group). Scale bars, 50 μm. Data are presented as means ± SD. Statistical significance was determined by one-way ANOVA, as appropriate. *p < 0.05, **p < 0.01, ***p < 0.001 Rapamycin alleviates the Avo-induced toxic effects in human intestinal organoids To evaluate the effects of avobenzone (Avo) on human intestinal organoids, organoids derived from two independent donors were treated with increasing concentrations of Avo for 72 h. CCK-8 assays showed that the cell viability of organoids was decreased by Avo in a dose-dependent manner (Fig. 8A). Consistently, PI fluorescence intensity increased in a concentration-dependent manner, suggesting that Avo induces cell death in human intestinal organoids (Fig. 8B). The protein level of the autophagy-related protein ATG14 and the intestinal stem cell marker LGR5 in organoids were reduced along with Avo treatment (Figs. 8C and S7B). More importantly, expression of COX8-EGFP-mCherry reporter in organoids further demonstrated that Avo attenuated CCCP-induced mitophagy (Fig. 8D). Collectively, these findings indicate that Avo impairs mitophagy and reduces the expression of the stem-cell-associated marker LGR5 in human intestinal organoids. Rapamycin alleviates avobenzone-impaired growth of human intestinal organoids. A–B Human intestinal organoids derived from two independent donors were treated with indicated concentrations of Avo for 72 h. A Organoid viability was assessed using a CCK-8 assay, and relative viability was normalized to the corresponding untreated control group. B Cell death was evaluated by propidium iodide (PI) staining, and relative PI fluorescence intensity was quantified. C Western blot analysis of ATG14 and LGR5 protein expression in human intestinal organoids derived from two independent donors treated with indicated concentrations of Avo for 72 h. D Representative confocal fluorescence images and quantitative analysis of mitophagic flux in human intestinal organoids expressing the COX8-EGFP-mCherry mitochondrial reporter. CCCP was used as a positive control for mitochondrial depolarization. The EGFP/mCherry fluorescence ratio was calculated to estimate mitochondrial delivery to acidic lysosomal compartments and mitophagic flux. Twenty randomly selected organoid regions of interest from each group were analyzed for fluorescence quantification. Scale bar, 50 μm. E Representative images of long-term cultured organoids with or without the addition of Avo (20 µM) and rapamycin (100 nM) for two weeks. For each of three independent experiments, five randomly selected fields were analyzed per condition, and the mean number of organoids per field was used for statistical analysis (n = 3). Scale bar, 100 μm. F Representative immunofluorescence images of HSP60, β-catenin, LGR5, and Ki67 in human intestinal organoids derived from two independent donors following the indicated treatments. Data are presented as means ± SD from three independent experiments performed using human intestinal organoids derived from two independent donors. Statistical significance was determined by one-way ANOVA, as appropriate. *p < 0.05, **p < 0.01, ***p < 0.001; ns, not significant Rapamycin, but not ISRIB, is an FDA-approved therapeutic drug that has been investigated for promoting health and longevity [37]. We therefore employed two independent human intestinal organoid lines to test the protective effect of rapamycin against Avo-induced cytotoxicity. After 2 weeks of exposure, Avo reduced organoid growth and organoid number, whereas cotreatment with rapamycin markedly attenuated these reductions (Fig. 8E). We further performed immunofluorescence staining for HSP60, β-catenin, LGR5, and Ki67 in organoids from the control, Avo, and Avo plus rapamycin groups after two weeks of treatment. Avo treatment decreased the fluorescence intensities of β-catenin, LGR5, and Ki67 and increased HSP60 fluorescence intensity, whereas these alterations were partially alleviated by rapamycin (Fig. 8F). These findings indicate that Avo-induced mitochondrial stress and Wnt/β-catenin suppression are associated with reduced stem-cell-associated marker expression and proliferative capacity in human intestinal organoids, and that these alterations are partially alleviated by rapamycin. Discussion Organelle-specific autophagy plays a central role in maintaining intestinal epithelial homeostasis, particularly within intestinal crypts where LGR5⁺ stem cells drive epithelial renewal. In the present study, we demonstrate that the widely used ultraviolet filter Avo, an emerging contaminant frequently detected in aquatic environments, disrupts intestinal epithelial homeostasis by impairing the self-renewal capacity of intestinal stem cells (ISCs). Using mouse models and intestinal organoids, we show that chronic Avo exposure suppresses Wnt/β-catenin signaling and leads to a reduction in LGR5⁺ stem cell abundance. Mechanistically, Avo interacts with ATG14, a key component of the PI3KC3-C1 autophagy complex, promoting its CUL3-mediated ubiquitination and proteasomal degradation. Loss of ATG14 disrupts organelle-specific autophagy, particularly mitophagy, resulting in accumulation of dysfunctional mitochondria. This mitochondrial stress activates an ATF5-associated UPRmt response that functionally contributes to the suppression of Wnt/β-catenin signaling and impaired ISC maintenance. Importantly, pharmacological induction of autophagy using rapamycin or inhibition of stress signaling using ISRIB significantly alleviated Avo-induced intestinal injury (Fig. 9), highlighting mitochondrial quality control as a critical mediator of Avo toxicity. Schematic model depicting avobenzone-induced mitophagy impairment and intestinal injury. Avo targets ATG14, a core component of the PI3KC3-C1 autophagy initiation complex, promoting its CUL3-mediated ubiquitination and subsequent proteasomal degradation. The depletion of ATG14 disrupts organelle-specific autophagy, particularly mitophagy, resulting in the accumulation of dysfunctional mitochondria and elevated mitochondrial ROS. This mitochondrial dysfunction triggers an ATF5-associated mitochondrial unfolded protein response (UPRmt), which functionally suppresses Wnt/β-catenin signaling and impairs intestinal stem cell (ISC) maintenance. Notably, pharmacological induction of autophagy via rapamycin or inhibition of stress signaling via ISRIB significantly alleviates Avo-induced intestinal injury, underscoring mitochondrial quality control as a critical mediator of Avo toxicity Maintenance of LGR5⁺ stem cell populations and proliferation of progenitor cells rely on a Wnt signaling gradient that peaks at the crypt base [34]. Disruption of this pathway is closely associated with impaired epithelial regeneration and intestinal injury. Previous studies have shown that environmental stressors, including irradiation or genetic disruption of Wnt signaling components, impair intestinal crypt regeneration and lead to epithelial damage [38]. Consistent with these observations, long-term oral exposure to Avo reduced crypt number and villus length in mice, accompanied by decreased Ki67⁺ proliferating cells and reduced LGR5 expression. Reduced Occludin and ZO-1 expression further indicated impaired intestinal barrier integrity, which was partially restored by rapamycin or ISRIB. Similar phenotypes were observed in intestinal organoids, indicating that Avo disrupts epithelial renewal by impairing Wnt/β-catenin–dependent ISC maintenance. Mitochondrial quality control is essential for intestinal stem cell function due to the high metabolic demand of proliferating crypt cells. Accumulation of damaged mitochondria triggers UPRmt, a stress adaptation pathway characterized by induction of mitochondrial chaperones and proteases such as LONP1, HSP60, and CLPP [11]. Persistent mitochondrial stress, in particular ATF5-dependent UPRmt [39], negatively regulates Wnt/β-catenin signaling and contributes to cellular dysfunction and aging [11, 40]. In the present study, Avo increased eIF2α phosphorylation and the expression of ATF5, LONP1, and HSP60 without affecting ATF4 or CLPP, indicating a selective ATF5-associated UPRmt response [11, 31]. Importantly, CCCP independently reproduced these molecular changes and reduced β-catenin and LGR5 expression, whereas ISRIB partially reversed the effects induced by either Avo or CCCP. Together, these findings support the conclusion that mitochondrial stress is sufficient to suppress the Wnt/β-catenin–LGR5 axis, at least partly through an eIF2α–ATF5-associated stress response [11]. Autophagy is a fundamental cellular process that maintains epithelial homeostasis by removing damaged organelles and macromolecules in response to stress [41]. Disruption of autophagy-related genes impairs intestinal barrier integrity and promotes inflammation. For example, loss of ATG16L1 leads to abnormal Paneth cell granule formation and increased susceptibility to intestinal inflammation [42]. Similarly, activation of autophagy in goblet cells alleviates ER stress and promotes mucus secretion, thereby protecting the intestinal epithelium [43]. In our study, long-term Avo exposure suppressed autophagy, particularly mitophagy, leading to mitochondrial membrane potential loss, ROS accumulation, lipid droplet formation, and UPRmt activation. Rapamycin-mediated induction of autophagy restored mitochondrial homeostasis and alleviated intestinal injury, indicating that impaired autophagy is a critical event in Avo-induced toxicity. Proteomic analysis identified ATG14 as a key molecular target of Avo. ATG14 is an essential component of the class III PI3K complex that regulates autophagosome formation and autophagosome–lysosome fusion [44, 45]. Previous studies have shown that ATG14 stability is regulated by ubiquitin-mediated degradation via E3 ligases such as CUL3 [30, 45]. Degradation of ATG14 has also been associated with lipid droplet accumulation and defects in organelle quality control [46]. In our study, Avo promoted CUL3-mediated ubiquitination and degradation of ATG14, resulting in impairment of PI3KC3 complex assembly and subsequent mitophagy blockade. ATG14 overexpression attenuated Avo-induced ROS accumulation, mitochondrial membrane potential loss, and mitophagy impairment, whereas ATG14 depletion recapitulated these mitochondrial defects. Moreover, Avo did not further exacerbate mitochondrial dysfunction in ATG14-deficient cells, suggesting that loss of ATG14 is essential for Avo-induced phenotypes. Importantly, restoration of ATG14 expression also attenuated Avo-induced eIF2α phosphorylation and the upregulation of ATF5, LONP1, and HSP60, and subsequently restored β-catenin and LGR5 expression. These findings establish the key regulatory role of the ATG14-mitophagy–UPRmt–Wnt signaling axis in Avo-induced intestinal epithelial dysfunction. Definitive validation of the direct binding between Avo and ATG14 warrants further studies. To further evaluate physiological relevance, we employed both mouse models and intestinal organoids. Similar intestinal injury phenotypes were observed in nude mice and C57BL/6J mice, suggesting that Avo-induced epithelial damage represents a robust phenotype that is not restricted to a specific mouse strain. Importantly, human intestinal organoids further confirmed that Avo directly impairs ISC self-renewal. Impaired mitophagy in organoids was observed using COX8-EGFP-mCherry reporter, and the Avo-induced mitochondrial stress and Wnt/β-catenin signaling suppression in organoids were visualized by Ki67 and LGR5 staining. These findings indicate that the mechanisms identified in mouse models are likely relevant to human intestinal physiology and highlight intestinal organoids as powerful platforms for investigating environmental toxicant-induced epithelial injury [47]. Considering the widespread occurrence of Avo in aquatic environments and its extensive use in personal care products, concerns have been raised regarding its environmental persistence and potential health risks [14]. Environmental monitoring studies have detected Avo in surface waters, sediments, and aquatic organisms [48]. Avo may accumulate in biological tissues, potentially resulting in higher local concentrations than those detected in environmental matrices [49]. Notably, the doses used in our C57BL/6J mouse model were substantially lower than the reported NOAEL in Sprague–Dawley rats (450 mg/kg/day) [15, 20], yet chronic Avo exposure still resulted in detectable intestinal epithelial injury. Further studies incorporating human biomonitoring data and clinical samples will be needed to better define the health risks associated with chronic exposure. Although our findings were supported by mouse models, intestinal epithelial cells, and human intestinal organoids, several limitations should be acknowledged. First, the experimental doses of Avo used in this study were designed for dose–response assessment and mechanistic investigation rather than a direct simulation of routine human exposure. The doses used in this study were higher than the estimated systemic exposures associated with sunscreen use, and oral gavage differs substantially from topical exposure in route, dosing pattern, bioavailability, and tissue-specific toxicokinetics [15, 16]. Moreover, Avo concentrations in plasma, intestinal tissues, and luminal contents were not measured, precluding direct toxicokinetic comparison with humans. Second, because Experiment 1 and Experiment 2 were performed in nude mice and C57BL/6J mice, respectively, the two in vivo datasets cannot be interpreted as a strain-matched comparison between immunodeficient and immunocompetent conditions. The observation that Avo induced mitochondrial dysfunction and epithelial damage in both models supports the robustness of the phenotype and suggests that epithelial toxicity may contribute substantially to Avo-induced intestinal injury, but it does not formally establish that this effect is independent of adaptive immunity. Third, although our data identify ATG14 as a target of Avo, the binding site and structural basis of this interaction warrant further investigation. Moreover, Avo-induced mitochondrial dysfunction may involve targets beyond ATG14, highlighting the need for further study. Finally, because rapamycin and ISRIB have effects beyond mitophagy and mitochondrial stress signaling, their protective actions cannot be attributed exclusively to the proposed pathway. Accordingly, these findings should be interpreted primarily as mechanistic hazard-identification evidence, and future studies using physiologically relevant exposure models, intestinal-specific systems, direct target-validation approaches, and quantitative toxicokinetic analyses will be required to define their human health relevance. Conclusions In conclusion, our study identifies a previously unrecognized intestinal injury associated with Avo exposure. By targeting ATG14 and impairing mitophagy-dependent mitochondrial quality control, Avo disrupts intestinal stem cell homeostasis and epithelial regeneration. These findings highlight mitochondrial quality control as a critical pathway linking environmental UV filter exposure to intestinal epithelial dysfunction and suggest that modulation of autophagy and mitochondrial stress responses may represent promising strategies for mitigating toxicity induced by personal care product–derived contaminants. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements Not applicable. Abbreviations Avobenzone Adverse drug reactions Free fatty acid Intestinal stem cells Mitochondrial membrane potential Mitochondrial unfolded protein response Authors’ contributions Z.-K.G. conceived and designed the study and performed most of the experiments. Z.-J.H. and X.-Y.Q. performed the animal experiments. B.-P.W. contributed to data analysis. J.-R.Z. performed cellular and molecular experiments and assisted with data validation. Y.-J.Z. performed the animal experiments. Y.-P.Z. contributed to data analysis. Q.-Y.H., J.Z., and Y.W. supervised the project, revised the manuscript, and provided critical intellectual input. All authors read and approved the final manuscript. Funding This work was supported by the National Natural Science Foundation of China (82372691), the National Key Research and Development Program of China (2023YFF1204600 and 2022YFA1304604), the Guangdong Basic and Applied Basic Research Foundation (2024B1515020029 and 2026A1515012568), and the Fundamental Research Funds for the Central Universities (21626714, 21626102). Data availability All data associated with this study are present in the paper or the Supplementary Materials. Declarations Ethics approval and consent to participate All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Jinan University and conducted in accordance with the approved protocols (IACUC-20231106-03 and IACUC-20241202-02). Human intestinal tissues used for organoid establishment were obtained from surgically resected specimens from patients with colorectal cancer. The use of human tissues was approved by the Institutional Review Board of Jinan University (Approval number: JNUKY-2023-0092). Written informed consent was obtained from all participants prior to tissue collection. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Qing-Yu He, tqyhe@email.jnu.edu.cn. Jing Zhang, Email: zj6410@jnu.edu.cn. Yang Wang, Email: wangyang0507@jnu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement All data associated with this study are present in the paper or the Supplementary Materials.