Digestive-transformed titanium dioxide nanoparticles disrupt intestinal barrier via mitoxyperilysis mediated by mTOR-cytoskeleton axis dysregulation Authors: Chuxin Zhang, Yiping Liu, Xu Li, Jiaxing An, Wenjing Li, Zhilu Zhou, Yitao Yan, Juan Shao, Zhongyuan Guo Published: 2026 ABSTRACT AbstractTitanium dioxide nanoparticles (TiO2 NPs) are ubiquitous dietary additives, yet their impact on the intestinal barrier remains contentious, largely because conventional toxicological models overlook gastrointestinal digestive transformations. Characterization revealed that pristine TiO2 NPs exist as nanoscale aggregates, which, upon simulated digestion, lose electrostatic repulsion and expand into dense, biocorona-coated, micrometer-sized agglomerates enriched in nitrogen and phosphorus. While these transformed entities induce severe barrier dysfunction, the spatiotemporal mechanisms executing plasma membrane rupture remain a critical knowledge gap in nanotoxicology. Here, we uncover a non-canonical, spatially dependent cell death modality termed mitoxyperilysis as the core driver of TiO2 NPs-induced nanotoxicity, operating independently of classical apoptosis, ferroptosis, necroptosis, and pyroptosis. We demonstrate that exposure to digested TiO2 NPs agglomerates hyperactivates the protein kinase B (AKT)/mechanistic target of rapamycin complex 2 (mTORC2)-RhoA signaling axis, leading to widespread F-actin cytoskeletal paralysis. This structural breakdown dismantles intracellular spatial constraints, enabling ROS-emitting, dysfunctional mitochondria to aberrantly translocate and persistently tether to the plasma membrane, deploying high concentrations of mitochondria-derived ROS (mtROS) locally. Quantitative lipidomics of plasma membrane fractions reveals that this targeted mtROS assault induces severe depletion of membrane-stabilizing lipids and the explosive generation of oxidized lipid mediators, culminating in lytic membrane failure. Ultimately, this nanotoxicology study decodes a novel nano-mechanopathological paradigm dictated by mTOR-cytoskeleton dysregulation and localized mitochondrial lipid peroxidation, providing critical translational insights and therapeutic targets for mitigation.Keywords: Mitoxyperilysis, Digestive-transformed titanium dioxide nanoparticles, Plasma membrane lipidomics, Mitochondria-plasma membrane tethering, Cytoskeleton paralysis, Nanotoxicology Introduction Titanium dioxide nanoparticles (TiO2 NPs) are widely utilized in food additives, pharmaceutical formulations, and consumer products [[1], [2], [3]]. The gastrointestinal tract serves as income the primary route for continuous human exposure [4,5]. Within the gut, the intestinal epithelial barrier, consisting of a single layer of polarized cells connected by tight junction complexes, acts as the initial physical and immunological defense against exogenous nanomaterials [6]. Expanding upon our previous findings that oral exposure to TiO2 NPs induces intestinal epithelial cytotoxicity, disrupts key tight junction proteins such as zonula occludens-1 (ZO-1) and Occludin, and compromises overall barrier integrity [[7], [8], [9], [10]], we have established that this increased epithelial permeability facilitates the translocation of luminal antigens, contributing to local mucosal inflammation and systemic metabolic disturbances [11]. Despite the clear characterization of these macroscopic toxicological outcomes from our prior work, the precise cellular mechanisms and sequence of events that drive the physical rupture of crystalline and agglomerated TiO2 NPs particles across the intestinal epithelial plasma membrane remain incompletely understood. Nanoparticle-induced cytotoxicity is frequently attributed to canonical regulated cell death (RCD) pathways, including apoptosis, necroptosis, pyroptosis, and ferroptosis [12,13]. Concurrent with these pathways, the accumulation of reactive oxygen species (ROS), particularly mitochondria-derived ROS (mtROS), is a recognized and critical driver of nanomaterial toxicity [14]. However, explaining the rapid and severe plasma membrane disintegration induced by TiO2 NPs solely through these conventional mechanisms presents a significant spatial contradiction. Free radicals, such as superoxide anions, possess high biochemical reactivity, which inherently results in a very short half-life and a strictly limited intracellular diffusion radius [15]. Consequently, mtROS generated by mitochondria confined to the perinuclear region cannot efficiently traverse the viscous cytoplasm to inflict direct and extensive oxidative damage on the distant plasma membrane and apical tight junction complexes. For mtROS to execute such targeted peripheral damage, a spatial reorganization of the mitochondria must inevitably occur [16]. A recently described mitochondria-dependent lytic process, termed mitoxyperiosis, provides an important conceptual framework for understanding how mitochondrial dysfunction can directly culminate in plasma membrane lysis [16]. Under normal physiological conditions, the intracellular positioning, retention, and movement of mitochondria are tightly governed by the filamentous actin (F-actin) cytoskeleton [17]. The F-actin network maintains cellular architecture, regulates membrane dynamics, and acts as a physical scaffold that restricts random organelle motility [18]. Actin polymerization and cytoskeletal stability are primarily regulated by the small GTPase RhoA [19]. Furthermore, RhoA activity is heavily modulated by upstream kinase networks, including the protein kinase B (AKT)/mechanistic target of rapamycin complex 2 (mTORC2)-RhoA signaling axis, which translates environmental stress signals into cytoskeletal rearrangements [20]. Disruptions to this kinase signaling cascade can lead to RhoA inactivation, triggering global F-actin depolymerization [21]. We reason that such cytoskeletal impairment would remove the physical constraints on mitochondria, allowing their abnormal translocation to the cellular periphery and subsequent physical interaction with the plasma membrane. Based on this spatial and biochemical rationale, we propose that TiO2 NPs exposure perturbs the AKT/mTORC2-RhoA signaling axis in intestinal epithelial cells, leading to substantial F-actin cytoskeletal disassembly [22]. This loss of structural constraint permits dysfunctional, ROS-producing mitochondria to migrate and establish direct contact with the inner leaflet of the plasma membrane. The resulting localized accumulation of mtROS induces targeted lipid peroxidation, generating reactive lipid electrophiles that alter membrane fluidity and curvature [23,24]. This highly localized oxidative stress directly compromises plasma membrane integrity and causes epithelial barrier breakdown, operating independently of the executioner proteins associated with canonical programmed cell death pathways [25]. To systematically test this hypothesis, we integrated multi-omics, morphological, and pharmacological approaches. Initially, by combining whole-transcriptome sequencing (RNA-seq) with a comprehensive pharmacological exclusion matrix, we confirmed that TiO2 NPs-induced membrane rupture operates independently of canonical programmed cell death pathways. Subsequently, using biochemical and high-resolution imaging techniques, we demonstrated that the dysregulated AKT/mTORC2-RhoA axis paralyzes the F-actin cytoskeleton, thereby permitting the abnormal translocation of dysfunctional, ROS-emitting mitochondria to the plasma membrane. Finally, through mitochondria-targeted antioxidant interventions and quantitative lipidomics on purified plasma membrane fractions, we captured the direct chemical footprint of localized oxidative damage, hallmarked by structural phospholipid depletion and oxidized lipid accumulation. Ultimately, this study aims to elucidate the non-canonical, spatially driven mechanism of TiO2 NPs-induced intestinal toxicity, offering novel pathological insights and potential therapeutic targets for nanomaterial-induced epithelial barrier dysfunction. Materials and methods Cell culture and construction of the in vitro intestinal epithelial co-culture model An in vitro human intestinal epithelial co-culture model was constructed using Caco-2 and HT29-MTX cells to simulate the physiological barrier composed of absorptive enterocytes and mucus-secreting goblet cells. Both Caco-2 and HT29-MTX cell lines were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cells used in all experiments were maintained between passages 10 ∼20. Caco-2 cells were routinely cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 20% fetal bovine serum (FBS) and 1% penicillin-streptomycin, whereas HT29-MTX cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. All cells were maintained in a humidified incubator at 37 °C with 5% CO2. To establish an in vitro intestinal barrier model exhibiting both epithelial polarization and mucus secretion, Caco-2 and HT29-MTX cells in the logarithmic growth phase were mixed and seeded at a physiological ratio of 3:1 [10,[26], [27], [28]]. Depending on the experimental objectives, the cell mixture was seeded onto Transwell inserts (PET membrane, 0.4 μm pore size) to establish a polarized epithelial barrier for transepithelial electrical resistance (TEER) measurements, or into standard multi-well culture plates for routine exposure assays and biochemical sample collection. The co-cultured cells were maintained for 15 days post-seeding to allow for complete cellular differentiation, epithelial polarization, tight junction formation, and the establishment of a robust mucus layer. Fresh co-culture medium was replaced regularly during the differentiation period. Only fully matured co-culture models were used for subsequent experimental treatments [29,30]. Preparation of simulated gastrointestinal digesta and TiO2 NPs exposure solutions To simulate the transport and digestion environment of orally ingested TiO2 NPs, an in vitro static gastrointestinal digestion model was employed for sample pretreatment [31,32]. The simulated digestion process sequentially consisted of oral, gastric, and intestinal phases to recapitulate the enzymatic environment, ionic strength, pH conditions, and temperature fluctuations characteristic of each digestive stage. In the oral phase, samples were mixed with simulated saliva containing NaCl, KCl, KH2PO4, mucin, and α-amylase, and incubated at 37 °C for 5 min to initiate preliminary amylolysis. Subsequently, for the gastric phase, simulated gastric fluid (primarily comprising NaCl, HCl, and pepsin) was added to the oral digesta; the pH was adjusted to acidic gastric conditions, followed by incubation at 37 °C for 2 h. Finally, in the intestinal phase, simulated intestinal fluid containing bile salts and pancreatin was added. The system pH was adjusted and maintained at approximately 7.0, with further incubation at 37 °C for 2 h. Upon completion, the terminal intestinal digesta was collected and diluted with serum-free DMEM at a 1:3 (v/v) ratio. This dilution aimed to minimize non-specific matrix effects on cell viability while maintaining a consistent exposure background. The diluted end-products served as the exposure solutions for treating the matured Caco-2/HT29-MTX co-culture model. TiO2 NPs and sucrose were purchased from Sigma-Aldrich (St. Louis, MO, USA; cat. no. 718467 and S9378, respectively). The selection of TiO2 NPs exposure concentrations was based on a previously established physiologically relevant intestinal exposure framework. Previous studies estimated that humans may ingest approximately 1012–1014 dietary fine and ultrafine particles per day and normalized the estimated oral particle intake to the absorptive surface area of the small intestine and duodenum to define physiologically relevant in vitro exposure levels [33,34]. Building on this approach, 30 nm TiO2 NPs at 1.4 × 10−6 mg/mL and 1.4 × 10−4 mg/mL were subsequently used as medium and high physiologically relevant exposure levels, respectively [35]. Accordingly, 1.4 × 10−4 mg/mL was selected as the original working concentration in the present study. To further assess the concentration-response relationship, two lower concentrations, 3.5 × 10−5 and 7.0 × 10−5 mg/mL, corresponding to 25% and 50% of the original concentration, respectively, were additionally included. Both concentrations fall between the previously reported medium and high physiologically relevant TiO2 exposure levels [28,36]. The working concentration of sucrose was set at 36 mg/mL [26] to mimic the intestinal sugar exposure microenvironment resulting from the consumption of sugary beverages or high-sugar diets. Based on the experimental design, four treatment groups were established: blank digesta (Sucrose-/TiO2-), sucrose digesta (Sucrose+/TiO2-), TiO2 digesta (Sucrose-/TiO2+), and co-exposure digesta (Sucrose+/TiO2+). To ensure that the only variables among the groups were the presence of sucrose and TiO2 NPs, all samples underwent the exact same simulated digestion procedure described above. Physicochemical characterization of TiO2 NPs during simulated digestion To systematically evaluate the morphological and physicochemical alterations of TiO2 NPs across different digestive phases, samples from the pristine state, oral phase, gastric phase, and intestinal phase were collected and analyzed. The hydrodynamic diameter, polydispersity index (PDI), and Zeta potential of the nanoparticles were determined using a Zetasizer Pro (Malvern Panalytical, UK) at room temperature. Prior to the measurement, appropriately diluted samples were loaded into capillary cells to avoid air bubbles, and measurements were conducted based on dynamic light scattering (DLS) and electrophoretic mobility principles. For morphological observation, the diluted TiO2 NPs suspensions were drop-cast onto carbon-coated copper grids and allowed to air-dry naturally at room temperature. The micro- and nanoscale morphological transitions and agglomeration states of the particles were then imaged using a transmission electron microscope (TEM; JEM-F200, JEOL, Japan) operated at an accelerating voltage of 200 kV. Furthermore, to determine the surface chemical composition and visually confirm the formation of the digestive biocorona, energy-dispersive X-ray spectroscopy (EDS) elemental mapping was performed. Briefly, the regions of interest containing TiO2 NPs agglomerates were identified, and the elemental distribution patterns (including Carbon, Oxygen, Nitrogen, and Phosphorus) were collected using an EDX spectrometer (Xplore 30, Oxford Instruments plc, UK) coupled with the electron microscope. Data acquisition and spectral processing were conducted utilizing the AZtec software (Oxford Instruments), ensuring appropriate dead-time correction and sufficient counts for high-resolution mapping. Evaluation of epithelial cell injury and differentiation function Following TiO2 NPs exposure, the degree of cellular injury in the established in vitro intestinal epithelial model was systematically evaluated based on cellular metabolic activity, plasma membrane integrity, and epithelial differentiation function. Cell viability was assessed using the Cell Counting Kit-8 (CCK-8; Life-i-Lab, Shanghai, China; cat. no. AC11L054). After discarding the exposure solutions, fresh medium containing the CCK-8 working reagent was added to each well. Following incubation at 37 °C in the dark, the absorbance was measured at 450 nm using TECAN Infinite 200PRO microplate reader (Tecan, Männedorf, Switzerland) to calculate the relative cell viability of each treatment group. Plasma membrane integrity was evaluated by quantifying the release of lactate dehydrogenase (LDH) into the culture supernatants. The collected supernatants were analyzed using an LDH Cytotoxicity Assay Kit (LEAGENE, Beijing, China; cat. no. TE0158). The measured LDH release levels were normalized to quantitatively characterize the extent of plasma membrane damage. For the dose-response analysis, cells were exposed to TiO2 NPs at 3.5 × 10−5, 7.0 × 10−5, or 1.4 × 10−4 mg/mL, and cell viability and plasma membrane injury were evaluated using the CCK-8 assay and LDH release assay, respectively. The differentiation function of intestinal epithelial cells was evaluated by determining alkaline phosphatase (AKP/ALP) activity. After gentle washing with PBS and subsequent cell lysis, the AKP/ALP activity in the cell lysates was quantified using an AKP/ALP Activity Assay Kit (Boxbio, Beijing, China; cat. no. AKFA018M) to reflect alterations in the brush border differentiation characteristics induced by TiO2 NPs exposure. Assessment of epithelial barrier integrity and mucus secretion TEER was measured using a Millicell ERS-2 Volt-Ohm Meter (Merck Millipore, Burlington, MA, USA). The matured Caco-2/HT29-MTX co-culture models established in Transwell inserts were subjected to TiO2 NPs exposure. Prior to measurement, the electrodes were sterilized with 75% ethanol and equilibrated with PBS at 37 °C. The electrodes were then inserted into the apical and basolateral compartments, and resistance values were recorded upon stabilization. TEER values were calculated using the following equation: TEER (Ω·cm2) = (measured resistance− blank resistance) × membrane area. Mucus secretion by the intestinal epithelial cells was quantitatively assessed using Alcian Blue (AB) and Periodic Acid-Schiff (PAS) staining kits (Beyotime Biotechnology, Shanghai, China; cat. no. C0153M and C0142S), which correspond to acidic mucins and neutral/sialic acid-containing mucins, respectively. Following exposure, cells were washed with PBS, fixed, and stained. After AB staining, cells were washed until the background was clear, and bright-field images were captured. Subsequently, the stained cells were lysed with acetic acid, and the absorbance of the supernatant was measured at 630 nm. For PAS quantification, cells were lysed using 10% Triton X-100 in PBS. The cell lysates were reacted with 0.6% periodic acid, followed by color development with Schiff reagent in the dark. The absorbance was then measured at 550 nm. The contents of acidic and neutral mucins were determined against a porcine gastric mucin standard curve. Quantitative evaluation of ZO-1 and Occludin protein levels to assess changes in the tight junction complex was performed via Western blotting (detailed in Section 2.14 ). Measurement of oxidative stress and lipid peroxidation Intracellular ROS production was evaluated using the CellROX® Green Reagent (Solarbio, Beijing, China; cat. no. CA1410). Following exposure, cells were washed with PBS to remove residual treatments and incubated with the CellROX® Green working solution at 37 °C in the dark. After washing away unbound probes, fluorescence signals were captured and quantified. ROS levels were expressed as relative fluorescence intensity normalized to the control group. Cellular redox homeostasis was assessed using GSH/GSSG Assay Kit (Beyotime Biotechnology, Shanghai, China; cat. no. S0053). Cell lysates were prepared to measure total glutathione and oxidized glutathione (GSSG) contents, respectively. The concentration of reduced glutathione (GSH) and the GSH/GSSG ratio were calculated to reflect alterations in the intracellular antioxidant buffering capacity. For further validation, absolute GSH content was determined in parallel using a Glutathione Assay Kit (Boxbio, Beijing, China; cat. no. AKPR008M). Cells were lysed and centrifuged, and the supernatants were subjected to colorimetric reactions. GSH concentrations were quantified and normalized to total protein content. Intracellular superoxide anion (O2−) generation was measured utilizing a Superoxide Assay Kit (Boxbio; cat. no. AKAO008M). Collected cell samples were added to the reaction system, and absorbance changes were monitored to calculate relative O2− levels. The extent of lipid peroxidation was evaluated by quantifying malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE). MDA content was measured using a Malondialdehyde Assay Kit (Boxbio; cat. no. AKFA013M) via the thiobarbituric acid (TBA) colorimetric method. 4-HNE levels were determined using a competitive ELISA kit (Jianglai Biotechnology, Shanghai, China; cat. no. JL19014-96T). Cell lysates underwent standard incubation and color development procedures, and absorbance was recorded using a microplate reader. Final concentrations of MDA and 4-HNE were calculated against their respective standard curves to characterize TiO2 NPs-induced membrane lipid peroxidation. Mitochondrial respiration, bioenergetic status, and ultrastructural assessment The oxygen consumption rate (OCR) was monitored in real-time utilizing the Oxygraph-2k high-resolution respirometer (Oroboros Instruments, Innsbruck, Austria). Following exposure, Caco-2/HT29-MTX co-cultured cells were collected and resuspended in pre-warmed respiration buffer. After a stable basal oxygen flux was established, oligomycin, FCCP, and rotenone/antimycin A were sequentially injected into the chambers to determine ATP-linked respiration, maximal respiratory capacity, and non-mitochondrial oxygen consumption, respectively. All OCR data were corrected for residual oxygen consumption before calculating the respiratory parameters. Cellular energy homeostasis was evaluated by quantifying intracellular ATP and AMP levels. ATP content was measured using an ATP Assay Kit (Abbkine, Wuhan, China; cat. no. KTB1016). Cleared cell lysates were subjected to a luciferase-catalyzed luminescence reaction, and signals were recorded using microplate reader. AMP levels were determined using an AMP Assay Kit (Mokesha Biotechnology, China; cat. no. KT-99549) following standard fluorometric or colorimetric procedures. The concentrations of both nucleotides were calculated against their respective standard curves and normalized to total protein content. Mitochondrial membrane potential was assessed in situ using the JC-1 fluorescent probe (Servicebio, Wuhan, China; cat. no. G1515-100T). Washed cells were incubated with the JC-1 working solution at 37 °C in the dark. After washing away unbound probes, fluorescence images were acquired. The extent of mitochondrial membrane potential depolarization was characterized by the ratio of red fluorescence (J-aggregates) to green fluorescence (J-monomers). The accumulation of mtROS was quantified using the MitoSOX™ Green indicator (Thermo Fisher Scientific, Waltham, MA, USA; cat. no. M36005). Following staining and washing, single-cell fluorescence signals were acquired using BD FACSCanto II flow cytometer (BD Biosciences, San Jose, CA, USA). Data were analyzed using FlowJo software, and mtROS levels were expressed as the relative median fluorescence intensity. Ultrastructural damage to mitochondria was morphologically confirmed via transmission electron microscopy (TEM). Collected cell pellets were sequentially subjected to primary fixation with glutaraldehyde, post-fixation with osmium tetroxide, gradient ethanol dehydration, epoxy resin embedding, and ultramicrotomy. Following double-staining with uranyl acetate and lead citrate, the mitochondrial ultrastructure was observed under a TEM, with a specific focus on evaluating matrix swelling, cristae fragmentation or loss, vacuolization, and the integrity of the double-membrane architecture. Analysis of classical regulated cell death pathways To systematically investigate the cell death mechanisms induced by TiO2 NPs, four classical RCD pathways (apoptosis, ferroptosis, necroptosis, and pyroptosis) were evaluated using specific pharmacological inhibitors and positive inducers. All reagents (purchased from Aladdin, Shanghai, China) were diluted to working concentrations in serum-free DMEM, ensuring consistent final solvent concentrations across all experimental groups. Co-cultured Caco-2/HT29-MTX cells were pretreated with the respective inhibitors prior to exposure to TiO2 NPs or the corresponding positive induction systems. Subsequently, cell injury was evaluated via CCK-8 and LDH release assays, and protein samples were collected for Western blot analysis. The specific intervention and detection strategies for each pathway were designed as follows: (1) Apoptosis: Staurosporine (STS; cat. no. S102392) and Z-VAD(OH)-FMK (cat. no. Z413908) were utilized as the positive inducer and pan-caspase inhibitor, respectively, followed by the detection of Caspase-3 cleavage. (2) Ferroptosis: RSL3 (cat. no. R302648) was applied to induce ferroptosis, ferrostatin-1 (Fer-1; cat. no. F129882) served as the inhibitor, and glutathione peroxidase 4 (GPX4) expression was monitored. (3) Necroptosis: Cells were stimulated with a combined cocktail of TNF-α (cat. no. RP156007), SM-164 (cat. no. S413709), and Z-VAD(OH)-FMK, while necrostatin-1s (Nec-1s; cat. no. C275378) was used for inhibition, with subsequent assessment of MLKL phosphorylation (p-MLKL). (4) Pyroptosis: A lipopolysaccharide (LPS; cat. no. L386714) and nigericin (cat. no. N102401) co-treatment acted as the induction system, belnacasan (VX-765; cat. no. B275738) functioned as the Caspase-1 inhibitor, and gasdermin D (GSDMD) cleavage was examined. Detailed Western blot procedures are described in Section 2.14. Analysis of AKT/mTOR signaling, RhoA activity, and F-actin organization The activation status of the intracellular AKT/mTOR signaling pathway was evaluated via Western blot analysis. Following TiO2 NPs exposure or Torin-1 intervention, total protein was extracted from the co-cultured cells to determine the expression levels of p-AKT (Ser473) and total AKT. The densitometric ratio of p-AKT/AKT was calculated for statistical analysis. To elucidate the role of aberrant mTOR signaling, Torin-1 (Aladdin, Shanghai, China; cat. no. T129642), a specific mTOR inhibitor, was employed for blockade experiments. Cells were pretreated with Torin-1 (diluted in serum-free DMEM) prior to the addition of the TiO2 NPs exposure solution. Consistent final solvent concentrations were maintained across all groups, including the Torin-1 alone and vehicle control groups. Intracellular RhoA activity was quantified utilizing the RhoA G-LISA Activation Assay Kit (Cytoskeleton; supplied by Jianglai Biotechnology, Shanghai, China; cat. no. CSK-BK124-S). After treatment, cells were washed with ice-cold PBS and rapidly lysed. The cell lysates were then transferred to a pre-coated microplate. Following incubation with specific antibodies and a colorimetric development reaction, absorbance was measured using a microplate reader to quantify the levels of active, GTP-bound RhoA. Morphological alterations in the F-actin cytoskeleton were visualized using a Phalloidin fluorescent probe (Proteintech Group, Wuhan, China; cat. no. PF00001). Treated cells were fixed, permeabilized, and subsequently incubated with the Phalloidin working solution in the dark. After washing, images were captured using a confocal or fluorescence microscope. The structural integrity of F-actin was assessed by observing the arrangement of stress fibers and the distribution of actin at the cell periphery. Relative fluorescence intensity was quantified using ImageJ software. Isolation of plasma membrane fractions and detection of membrane-associated mitochondrial proteins Plasma membrane fractions were isolated and enriched utilizing a Plasma Membrane Protein Extraction Kit (BestBio, Shanghai, China; cat. no. P1203). Following exposure or pharmacological intervention, cells were washed with ice-cold PBS and harvested. The cell pellets were subjected to sequential lysis, homogenization, and differential centrifugation to obtain highly purified plasma membrane protein samples. To verify the aberrant tethering and distribution of mitochondria to the plasma membrane, the protein levels of voltage-dependent anion channel 1 (VDAC1) and translocase of outer mitochondrial membrane 20 (TOMM20) in the plasma membrane fractions were determined via Western blot analysis. Na+/K+-ATPase α1 (ATP1A1) was employed as a plasma membrane-specific marker and loading control to quantify the relative enrichment abundance of VDAC1 and TOMM20. In parallel, whole-cell total protein lysates from the corresponding treatment groups were extracted to compare the spatial distribution shifts of these mitochondria-associated proteins between the global cellular level and the local plasma membrane fraction. Antioxidant intervention experiments To validate the causal role of ROS in TiO2 NPs-induced cellular damage, pharmacological rescue experiments were conducted utilizing the broad-spectrum antioxidant N-acetyl-L-cysteine (NAC; MedChemExpress, Monmouth Junction, NJ, USA; cat. no. HY-B0215) and the mitochondria-targeted ROS scavenger Mito-TEMPO (Aladdin, Shanghai, China; cat. no. M414479). Both reagents were prepared as stock solutions and diluted in serum-free DMEM immediately prior to use, ensuring consistent final solvent concentrations across all experimental groups. Mature Caco-2/HT29-MTX co-cultures were initially pretreated with either NAC or Mito-TEMPO, followed by co-incubation with the TiO2 NPs exposure solution. Parallel experimental groups included the vehicle control, TiO2 NPs exposure alone, antioxidant alone, and the combined treatment groups. Following the exposure period, the mitigative effects of ROS scavenging on TiO2 NPs toxicity and cellular homeostasis were systematically evaluated across three dimensions: (1) cell survival profiles (CCK-8 and LDH release); (2) redox and lipid peroxidation homeostasis (ROS fluorescence, GSH/GSSG ratio, MDA, and 4-HNE contents); and (3) restoration of barrier integrity and aberrant mitochondrial distribution (TEER, along with Western blot quantification of ZO-1, Occludin, and mitochondrial markers). The specific operational and quantification procedures for all biochemical and molecular assays were conducted following the protocols detailed in preceding sections. RNA-seq and bioinformatics analysis To systematically delineate the transcriptomic remodeling induced by TiO2 NPs exposure, RNA-seq was performed on the co-cultured cells from both the Control and TiO2 NPs-exposed groups, with six independent biological replicates per group. Post-exposure, cell pellets were harvested and immediately lysed in TRIzol reagent (Invitrogen, Carlsbad, CA, USA) for total RNA extraction. RNA samples passing rigorous quality control for concentration, purity, and integrity were processed for sequencing library construction, followed by 150 bp paired-end high-throughput sequencing on an Illumina NovaSeq 6000 platform. Raw sequencing data were subjected to quality control using Cutadapt software to trim adapter sequences and filter out low-quality reads. The resulting high-quality clean reads were then aligned to the human reference genome utilizing Hisat2. Transcript assembly and expression quantification were performed using StringTie, with expression levels normalized as Fragments Per Kilobase Million (FPKM). Differential expression analysis was conducted on the raw read counts matrix utilizing the DESeq2 package. Differentially expressed genes (DEGs) were robustly defined by the thresholds of |log2(Fold Change)| > 1 and adjusted P-value < 0.05. To visualize the global shifts in gene expression profiles, clustered heatmaps and volcano plots were generated based on the DEGs. Furthermore, Gene Ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed on the DEGs to elucidate the principal biological processes and signaling pathways perturbed by TiO2 NPs exposure. Moreover, Gene Set Enrichment Analysis (GSEA) was employed to evaluate coordinated up- or down-regulation trends at the holistic gene set level, thereby corroborating critical transcriptional regulatory networks associated with cellular stress, metabolic reprogramming, and barrier disruption. Quantitative lipidomics of plasma membrane fractions To characterize the lipidomic remodeling of the plasma membrane following TiO2 NPs exposure, quantitative lipidomics was performed on purified plasma membrane fractions (n = 6 independent biological replicates per group). Membrane samples (isolated as described in Section 2.10 ) were thawed on ice and subjected to lipid extraction by adding 1 mL of an internal standard-containing extraction mixture (methyl tert-butyl ether:methanol = 3:1, v/v). Following 15 min of vortexing, 200 μL of ultrapure water was added to induce phase separation. The mixture was vortexed for 1 min and centrifuged at 12,000 rpm for 10 min at 4 °C. An aliquot of 500 μL from the upper organic phase was collected, evaporated to dryness, and reconstituted in 200 μL of a resuspension buffer (acetonitrile:isopropanol = 1:1, v/v). After further vortexing and centrifugation, the supernatant was subjected to LC-MS/MS analysis. Lipid profiling was conducted using an ExionLC™ AD UPLC system coupled with a QTRAP® 6500+ LC-MS/MS system (SCIEX, Framingham, MA, USA). Chromatographic separation was achieved on a Thermo Accucore™ C30 column (2.6 μm, 2.1 mm × 100 mm) maintained at 45 °C, with a 2 μL injection volume and a flow rate of 0.35 mL/min. Mobile phase A consisted of acetonitrile/water (60:40, v/v), and mobile phase B was acetonitrile/isopropanol (10:90, v/v); both contained 0.1% formic acid and 10 mmol/L ammonium formate. Mass spectrometry utilized an electrospray ionization (ESI) source operating in both positive (5500 V) and negative (−4500 V) ion modes. The source temperature was set to 500 °C, with ion source gases GS1, GS2, and curtain gas maintained at 45, 55, and 35 psi, respectively. Lipid species were detected using multiple reaction monitoring (MRM) with optimized declustering potential (DP) and collision energy (CE). Qualitative identification of lipids was performed referencing the MWDB self-built lipid database (MetWare), integrating retention times, precursor-to-product ion transitions, and MS response characteristics. Quantitative analysis was executed using the internal standard method, with chromatographic peak integration and calibration performed via Analyst 1.6.3 software. Pooled quality control (QC) samples were interspersed to evaluate instrument stability and data reproducibility. Following quality assurance, the lipid quantification matrix underwent normalization and unit variance scaling prior to principal component analysis (PCA), hierarchical cluster analysis (HCA), and orthogonal partial least squares-discriminant analysis (OPLS-DA). Differential lipids were subsequently screened based on variable importance in projection (VIP), fold change, and statistical significance (p-value/FDR), which were then utilized for downstream evaluations of lipid subclass composition, degree of unsaturation (double bond index, DBI), and KEGG pathway enrichment. Western blot analysis Whole-cell or plasma membrane protein samples were subjected to Western blot analysis. Following the respective treatments, cells were washed with ice-cold PBS and lysed using a lysis buffer containing protease and phosphatase inhibitors. Plasma membrane fractions, isolated as described in Section 2.10, were used directly for protein detection. Protein concentrations were determined using BCA assay. Subsequently, equivalent amounts of protein samples were mixed with loading buffer and denatured by heating. The protein samples were resolved by SDS-PAGE and transferred onto PVDF membranes. After transfer, the membranes were blocked with either 5% non-fat dry milk or 5% BSA, depending on the properties of the target proteins, and then incubated with primary antibodies overnight at 4 °C. The following day, after washing with TBST, the membranes were incubated with HRP-conjugated secondary antibodies at room temperature. Following another wash step, protein bands were visualized using an enhanced chemiluminescence (ECL) method, and signals were captured with a gel imaging system. The primary antibodies utilized in this study included p-AKT (Abways, Shanghai, China; cat. no. CY6569), AKT (Abways, cat. no. CY5551), ATP1A1 (Abways, cat. no. CY5159), VDAC1 (Abways, cat. no. CY5416), TOMM20 (Abways, cat. no. CY5527), mitofusin 2 (MFN2, Abways, cat. no. CY6638), phosphorylated mixed lineage kinase domain-like protein (p-MLKL, Abways, cat. no. CY5493), MLKL (Abways, cat. no. CY7146), GSDMD (Abways, cat. no. BY9011), Caspase-3/Cleaved Caspase-3 (Abways, cat. no. CY5501), GPX4 (Abways, cat. no. CY6959), ZO-1 (Proteintech, Wuhan, China; cat. no. 27260-1-AP), Occludin (Proteintech; cat. no. 21773-1-AP), and β-actin (Proteintech; cat. no. 20536-1-AP). All primary antibodies were diluted at 1:1000, except for β-actin, which was diluted at 1:5000. An HRP-conjugated goat anti-rabbit IgG (Proteintech; cat. no. RGAR001) was used as the secondary antibody at a dilution of 1:2000. Densitometric analysis of the protein bands was performed using ImageJ software. In the whole-cell protein samples, the expression levels of target proteins were normalized to β-actin as an internal control; phosphorylated proteins were normalized to their corresponding total proteins (i.e., p-AKT/AKT and p-MLKL/MLKL). In the plasma membrane protein samples, ATP1A1 served as the plasma membrane marker and normalization reference to calculate the relative expression levels of mitochondria-associated proteins, such as VDAC1 and TOMM20, within the plasma membrane fraction. Statistical analysis All data are expressed as the mean ± standard error of the mean (SEM). Statistical analyses and graphical presentations for routine experiments were performed using GraphPad Prism software (GraphPad Software, San Diego, CA, USA). Comparisons between two independent groups were analyzed using an unpaired Student's t-test, while a paired Student's t-test was employed for matched or paired data. For comparisons among multiple groups involving a single factor, a one-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test was utilized. In experiments involving two interactive factors, a two-way ANOVA was conducted to evaluate the main effects and their interactions, followed by Tukey's multiple comparisons test. Data derived from Western blot densitometry, fluorescence intensities, enzyme activities, lipid peroxidation indices, and barrier function assays were subjected to statistical analysis post-normalization. Unless otherwise specified, all experiments were performed with at least three independent biological replicates. The specific screening criteria for differential expression in RNA-seq and quantitative lipidomics data are detailed in their respective methodological sections. A P-value < 0.05 was considered statistically significant. Results Physicochemical transformation and biocorona formation of TiO2 NPs during simulated gastrointestinal digestion To elucidate the actual physicochemical state of TiO2 NPs prior to their interaction with the intestinal epithelium, we subjected the pristine nanoparticles to a standardized static in vitro gastrointestinal tract (GIT) model, sequentially simulating the oral, gastric, and intestinal phases (Fig. 1a). Dynamic light scattering (DLS) and electrophoretic mobility analyses revealed profound, phase-dependent alterations in the colloidal stability of TiO2 NPs (Fig. 1b and Table S1). Pristine TiO2 NPs exhibited a relatively small hydrodynamic diameter (192.49 ± 2.32 nm) with a negative surface charge (−22.42 ± 0.60 mV). However, upon continuous exposure to the digestive fluids, the absolute value of the Zeta potential progressively diminished toward neutrality (reaching −0.45 ± 0.31 mV in the intestinal phase, Table S1). This severe loss of electrostatic repulsion triggered massive, uncontrolled particle agglomeration. The hydrodynamic diameter surged to 875.8 ± 36.8 nm in the oral phase, peaked at 1658.2 ± 107.1 nm in the highly acidic gastric phase (pH 2.0), and ultimately stabilized at 1624.1 ± 142.1 nm as stable micrometer-scale agglomerates in the intestinal phase. Physicochemical transformation and biocorona formation of TiO2 NPs during simulated gastrointestinal digestion. (a) Schematic illustration of the static in vitro gastrointestinal tract (GIT) digestion model, sequentially simulating the oral, gastric, and intestinal phases. (b) Hydrodynamic size distribution histograms of TiO2 NPs at different digestive phases measured by dynamic light scattering (DLS). The inset text indicates the corresponding average diameter (d), polydispersity index (PDI), and Zeta potential. (c) Representative transmission electron microscopy (TEM) images illustrating the morphological changes and massive agglomeration of TiO2 NPs from loosely dispersed pristine state to dense monolithic chunks in the digestive fluids. Scale bars: 200 nm. (d) Energy-dispersive X-ray spectroscopy (EDS) elemental mapping of Carbon (C, yellow), Oxygen (O, cyan), Nitrogen (N, orange), and Phosphorus (P, magenta) on the surface of TiO2 NPs agglomerates, demonstrating the dynamic wrapping of biological matrices. Scale bars: 500 nm. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) This massive scale-up in physical dimensions was visually corroborated by transmission electron microscopy (TEM) (Fig. 1c), which demonstrated a clear transition from loosely dispersed nanoscale aggregates (pristine) to dense, micrometer-sized monolithic chunks in the digestive phases. Furthermore, to characterize the surface chemical modifications driving this agglomeration, energy-dispersive X-ray spectroscopy (EDS) elemental mapping was performed (Fig. 1d). Compared to the pristine state, TiO2 NPs agglomerates in the digested phases exhibited a progressive and intense surface enrichment of Nitrogen (N) and Phosphorus (P). The corresponding EDS spectra and quantitative composition analyses (Fig. S1a–e) confirmed the dynamic integration of these elements. The prominent acquisition of N and P signatures is highly indicative of the continuous wrapping of proteinaceous and lipid-rich biological matrices—such as mucin, pepsin, pancreatin, and bile salts—onto the TiO2 NPs surface, forming a complex digestive biocorona. Collectively, these characterizations confirm that the actual toxicological entities reaching the intestinal barrier are not pristine nanoparticles, but rather biocorona-coated, micrometer-scale enormous agglomerates, laying the physical foundation for the subsequent severe mechanical stress and membrane rupture observed in epithelial cells. Two-way ANOVA identifies TiO2 NPs as the primary inducer of multifaceted intestinal epithelial damage To distinguish the relative contributions of TiO2 NPs exposure and exogenous sucrose supplementation to intestinal epithelial damage, four treatment groups were established: a blank digestion fluid group (Sucrose−/TiO2−), a digestion fluid + sucrose group (Sucrose+/TiO2−), a TiO2 damage model group (Sucrose−/TiO2+), and a sucrose intervention group (Sucrose+/TiO2+). A two-way analysis of variance (ANOVA) was employed to evaluate the main effects of TiO2 NPs and sucrose, as well as their interactive effects, on multifaceted intestinal epithelial damage phenotypes (Fig. 2). TiO2 NPs exposure induces cytotoxicity, epithelial barrier dysfunction, impaired mucus secretion, and profound oxidative stress in intestinal epithelial cells. (a, b) Cell viability and membrane integrity were evaluated by CCK-8 assay (a) and lactate dehydrogenase (LDH) release (b), respectively. (c) alkaline phosphatase (AKP/ALP) activity was determined to assess intestinal epithelial differentiation. (d) TEER was measured to evaluate physical barrier integrity and permeability. (e–h) Mucus barrier function was assessed via Alcian blue (AB) and PAS staining, alongside corresponding quantitative analyses. AB staining (e, g) highlights acidic mucin secretion, whereas PAS staining (f, h) indicates neutral mucins. (i–k) Western blotting (i) and densitometric analyses (j, k) of the tight junction proteins ZO-1 and Occludin. (l, m) Representative fluorescence images (l) and quantitative analysis (m) of intracellular ROS accumulation. (n–p) Biomarkers of redox homeostasis and lipid peroxidation, including the GSH/GSSG ratio (n), intracellular superoxide anion (O2-) levels (o), and malondialdehyde (MDA) content (p), were quantified. Data are presented as mean ± SEM (n ≥ 3). Experimental groups: blank digestive fluid (Sucrose-/TiO2-), sucrose supplementation (Sucrose+/TiO2-), TiO2 NPs injury model (Sucrose-/TiO2+), and sucrose intervention (Sucrose+/TiO2+). Except for panel (d), quantitative data were analyzed by two-way ANOVA to assess the main effects of TiO2 NPs and sucrose and their interaction (TiO2 NPs × sucrose), followed by post hoc multiple-comparison testing. P values for the TiO2 NPs main effect, sucrose main effect, and TiO2 NPs × sucrose interaction are indicated in the corresponding panels. Different letters above the bars indicate significant differences between groups following post hoc comparisons (P < 0.05), whereas identical letters indicate no significant difference. Panel (d) was analyzed using a two-tailed unpaired Student's t-test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Cytotoxicity assays revealed that TiO2 NPs exposure exerted a significant main effect on cell viability (P < 0.0001), manifesting as a marked decrease in the cell survival rate. In contrast, neither the main effect of sucrose nor the TiO2 NPs × sucrose interaction reached statistical significance, suggesting that exogenous sucrose supplementation could not effectively reverse the TiO2 NPs-induced decline in cell viability (Fig. 2a). Furthermore, the LDH release assay demonstrated that TiO2 NPs treatment significantly elevated LDH levels in the supernatant (P < 0.0001), indicating compromised plasma membrane integrity. Conversely, sucrose treatment alone did not cause apparent LDH leakage, and sucrose supplementation failed to significantly alleviate the membrane damage induced by TiO2 NPs (Fig. 2b). To further evaluate the dose dependence of TiO2 NP-induced cytotoxicity, two lower exposure concentrations (3.5 × 10−5 and 7.0 × 10−5 mg/mL) were examined in addition to the original concentration of 1.4 × 10−4 mg/mL. Cell viability progressively decreased with increasing TiO2 NP concentration, whereas LDH release showed a corresponding concentration-dependent increase (Fig. S2). These results demonstrate a clear dose-dependent cytotoxic response across the tested exposure range. Subsequently, we evaluated intestinal epithelial differentiation and physical barrier functions. The AKP/ALP activity assay showed no significant difference between the Sucrose−/TiO2− and Sucrose+/TiO2− groups, whereas TiO2 NPs exposure significantly decreased AKP/ALP activity, indicating suppressed epithelial differentiation (Fig. 2c). TEER measurements revealed that TiO2 NPs treatment led to a dramatic drop in TEER, signifying increased epithelial permeability and compromised barrier integrity. Sucrose supplementation alone neither altered basal TEER levels nor effectively restored the TiO2 NPs-induced barrier dysfunction (Fig. 2d). Regarding the mucus barrier, both acidic and neutral mucopolysaccharide contents were markedly reduced following TiO2 NPs exposure, whereas sucrose treatment alone showed no significant difference compared to the blank control group (Fig. 2e and f). Alcian Blue staining further demonstrated a reduction in acidic mucus-positive areas in the TiO2 NPs group, implying impaired acidic mucus secretion (Fig. 2g). PAS staining showed weakened neutral mucus positivity and a thinned or discontinuous mucus layer structure in the TiO2 NPs group, while the Sucrose+/TiO2− group exhibited staining characteristics consistent with the blank control (Fig. 2h). These findings demonstrate that TiO2 NPs exposure significantly attenuates the intestinal mucus barrier, and this damage is not attributable to sucrose alone. At the structural level of tight junctions, Western blot analysis revealed that TiO2 NPs exposure substantially downregulated the protein expressions of Occludin and ZO-1 (Fig. 2i). Densitometric quantification further confirmed a significant decrease in Occludin and ZO-1 protein abundance after TiO2 NPs treatment, and sucrose supplementation failed to effectively restore this TiO2 NPs-induced downregulation (Fig. 2j and k). Corroborating the TEER decline, these results indicate that TiO2 NPs destabilizes the intestinal physical barrier by disrupting tight junction structures. Assessments of oxidative stress revealed a marked enhancement of intracellular ROS fluorescence signals following TiO2 NPs exposure, whereas the Sucrose−/TiO2− and Sucrose+/TiO2− groups maintained only low background fluorescence (Fig. 2l). Quantification of ROS fluorescence intensity further demonstrated that TiO2 NPs had a significant main effect on intracellular ROS accumulation (P < 0.0001), and sucrose supplementation failed to effectively suppress this TiO2 NPs-induced ROS elevation (Fig. 2m). Concurrently, TiO2 NPs treatment decreased the GSH/GSSG ratio and significantly increased intracellular superoxide anion levels, indicating an impaired cellular redox buffering capacity (Fig. 2n and o). Detection of MDA, a lipid peroxidation product, showed that TiO2 NPs exposure significantly elevated MDA contents (P < 0.0001), illustrating severe lipid oxidative damage induced by TiO2 NPs, a change that sucrose supplementation similarly could not effectively reverse (Fig. 2p). The two-way ANOVA results demonstrated that TiO2 NPs exposure exerted significant main effects across key damage indicators, including cell viability, LDH release, TEER decline, ROS accumulation, and MDA elevation. Conversely, the main effect of sucrose and the TiO2 NPs × sucrose interaction was generally non-significant. This establishes TiO2 NPs as the primary factor driving intestinal epithelial cytotoxicity, mucus barrier attenuation, tight junction disruption, and oxidative homeostasis disorder; exogenous sucrose supplementation cannot significantly rescue the multifaceted intestinal epithelial damage induced by TiO2 NPs. RNA-seq transcriptomics uncovers regulatory cues related to mitoxyperilysis To elucidate the molecular basis of TiO2 NPs-induced intestinal epithelial damage, we performed RNA-seq analysis on samples from the blank control and TiO2 NPs-exposed groups (Fig. 3). PCA revealed a distinct spatial separation between the two groups, alongside tight clustering of intra-group biological replicates, indicating that TiO2 NPs exposure triggered a robust and consistent global transcriptomic reprogramming (Fig. 3a). The sample correlation heatmap further confirmed the high intra-group correlation, ensuring the reliability of the subsequent differential expression analysis (Fig. S3a). TiO2 NPs exposure drives extensive transcriptomic remodeling and identifies key regulatory cues associated with Mitoxyperilysis. (a) Principal component analysis (PCA) score plot of RNA-seq data, demonstrating distinct transcriptomic separation between the Control and TiO2 NPs-exposed groups. (b) Volcano plot illustrating differentially expressed genes (DEGs). Red, yellow, and grey dots represent significantly upregulated, downregulated, and non-significant genes, respectively. Core regulatory genes associated with mitochondrial function (MT-TK), cytoskeletal regulation (PLEKHG7), lipid metabolism (MOGAT1), and barrier maintenance (MUC6) are explicitly annotated. (c) Gene Ontology (GO) enrichment scatter plot of DEGs. Bubble size corresponds to the gene count, and the color gradient reflects the adjusted p-value. (d) Hierarchical clustering heatmap of representative DEGs, visualizing the distinct expression signatures of candidate genes mapped to mitochondria, Rho GTPase/cytoskeleton, lipid transport, and tight junctions. (e) Chord diagram linking representative DEGs to their corresponding biological pathways, highlighting the involvement of the mTOR signaling pathway, small GTPase-mediated signal transduction, actin cytoskeleton organization, and phospholipid transport. (f–j) Gene Set Enrichment Analysis (GSEA) plots for specific functional modules: phospholipid transport (f), respirasome (g), mTOR signaling pathway (h), regulation of small GTPase-mediated signal transduction (i), and actin cytoskeleton organization (j). The plots display the running enrichment score, the position of gene set members in the ranked list, and the normalized enrichment score (NES), delineating the directional shift of these pathways upon TiO2 NPs exposure. Collectively, these transcriptomic signatures pinpoint mitochondrial respiratory failure, mTOR/small GTPase-mediated cytoskeletal dysregulation, and lipid transport perturbations as the foundational mechanisms driving TiO2 NPs toxicity. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) The differential expression analysis identified 173 upregulated and 254 downregulated genes. Volcano plot analysis demonstrated that TiO2 NPs exposure significantly altered a multitude of genes involved in mitochondrial function, cytoskeleton regulation, lipid metabolism, and the epithelial barrier. Notably, core regulatory genes such as the mitochondria-related MT-TK, the cytoskeletal regulatory hub PLEKHG7, the lipid metabolism-related MOGAT1, and the mucus barrier-associated MUC6 exhibited profound differential expression (Fig. 3b). This transcriptional shift suggested that TiO2 NPs simultaneously perturbs mitochondrial homeostasis, the cytoskeletal network, membrane lipid metabolism, and barrier maintenance. GO functional enrichment analysis revealed that the DEGs were primarily enriched in terms including respirasome assembly, mitochondrial electron transport, transmembrane transport, long-chain fatty acid binding, regulation of cholesterol efflux, microtubule-associated processes, calcium ion binding, phospholipid perturbation, extracellular matrix composition, and cell differentiation (Fig. 3c). KEGG pathway analysis further anchored these DEGs to signaling networks such as cytoskeleton in muscle cells, the calcium signaling pathway, focal adhesion, carbohydrate metabolism, and ABC transporters, corroborating the broad impact of TiO2 NPs on cytoskeletal architecture, calcium homeostasis, and substance transport (Fig. S3b). A representative DEG heatmap provided a visual panorama of these mechanism modules, demonstrating that candidate genes related to mitochondria, Rho GTPase/cytoskeleton regulation, lipid transport, and the barrier exhibited highly consistent expression shifts in the TiO2 NPs group, clearly differentiating the transcriptional signatures of the two groups (Fig. 3d). To delineate the cascade relationships between core genes and biological pathways, chord diagram analysis was employed. The representative DEGs were mapped primarily to key functional modules, including the mTOR signaling pathway, regulation of small GTPase mediated signal transduction, actin cytoskeleton organization, and phospholipid transport (Fig. 3e). The prominence of mTOR and small GTPase signals implied that TiO2 NPs exposure might disrupt the mTOR/Rho GTPase regulatory axis. The actin-related enrichment directly pointed to abnormal cytoskeletal organization, while anomalies in phospholipid transport hinted at the involvement of membrane lipid transport and phospholipid homeostasis in TiO2 NPs-induced membrane structural damage. Subsequently, GSEA was utilized to validate the global shift trends of these pathways at the whole-transcriptome level. In the TiO2 NPs group, the phospholipid transport gene set was positively enriched, confirming the activation or compensation of membrane lipid transport processes (Fig. 3f). Conversely, the respirasome gene set exhibited strong negative enrichment, indicating global suppression of genes related to the mitochondrial respiratory chain (Fig. 3g). Furthermore, significant enrichment alterations were observed in the mTOR signaling pathway (Fig. 3h), regulation of small GTPase mediated signal transduction (Fig. 3i), and actin cytoskeleton organization (Fig. 3j) gene sets, reinforcing the evidence for mTOR/small GTPase-mediated cytoskeletal collapse. Supplementary GSEA elaborated on this mechanistic landscape: mitochondrial energy metabolism gene sets, including the respirasome, mitochondrial electron transport, NADH dehydrogenase activity, and oxidative phosphorylation, were globally suppressed (Fig. S3c), concomitant with marked perturbations in small GTPase/Rho regulation and actin cytoskeleton assembly gene sets (Fig. S3d). Additionally, extensive impacts were observed on lipid networks encompassing phospholipid transport, cholesterol homeostasis, and sphingolipid/glycerolipid metabolism (Fig. S3e), as well as on structural barrier foundations (e.g., basement membrane, adherens junction, tight junction) (Fig. S3f) and calcium ion homeostasis (Fig. S3g). The RNA-seq transcriptomic results compellingly uncover a multidimensional transcriptional remodeling in intestinal epithelial cells triggered by TiO2 NPs exposure. This remodeling is precisely localized to mitochondrial respiratory dysfunction, mTOR/small GTPase-mediated cytoskeletal dysregulation, membrane lipid reprogramming, and the disintegration of epithelial junctional structures, thereby providing crucial transcriptional cues for the subsequent validation of the Mitoxyperilysis mechanism. Mitochondrial dysfunction precedes TiO2 NPs-induced plasma membrane rupture In light of the transcriptomic alterations in mitochondrial respiration, redox homeostasis, and energy metabolism, we further investigated the impact of TiO2 NPs exposure on mitochondrial function, the temporal progression of oxidative stress, and subcellular ultrastructure (Fig. 4). Oroboros O2k high-resolution respirometry revealed that TiO2 NPs exposure significantly diminished maximal respiratory capacity and respiratory reserve compared with the control, indicating an impairment of mitochondrial oxidative phosphorylation and adaptive respiratory compensation under stress (Fig. 4a). Concurrently, energy metabolism profiling showed a marked decline in both ATP and AMP levels, suggesting a profound disturbance in cellular nucleotide energy homeostasis (Fig. 4b and c). Furthermore, the rapid depletion of GSH reserves indicated a systemic failure of the endogenous antioxidant system (Fig. 4d). TiO2 NPs exposure induces mitochondrial respiratory dysfunction, redox homeostasis imbalance, and ultrastructural damage in intestinal epithelial cells. (a) Oxygen flux in Control and TiO2 NPs-treated cells measured via Oroboros O2k high-resolution respirometry. Sequential addition of oligomycin, FCCP, rotenone, and antimycin A was performed to determine basal respiration, maximal respiratory capacity, and spare respiratory capacity. (b, c) Quantification of intracellular ATP (b) and AMP (c) levels, reflecting the disruption of nucleotide energy homeostasis. (d) Intracellular glutathione (GSH) content, indicating the depletion of endogenous antioxidant reserves. (e, f) Time-course assessments of LDH release (e) and intracellular GSH depletion (f) during 0–4 h of TiO2 NPs exposure, illustrating the temporal onset of plasma membrane injury relative to antioxidant failure. (g) Representative JC-1 fluorescence images showing mitochondrial membrane potential changes; red aggregates denote polarized mitochondria, while green monomers signify depolarization. (h) Quantitative analysis of the JC-1 aggregate-to-monomer ratio, confirming mitochondrial membrane potential loss. (i, j) MitoSOX fluorescence quantification (i) and flow cytometry histograms (j) demonstrating the accumulation of mitochondrial superoxide anions. (k) Transmission electron microscopy (TEM) images revealing mitochondrial ultrastructural aberrations, including swelling, vacuolization, cristae loss, and abnormal tethering of damaged mitochondria toward the plasma membrane periphery in TiO2 NPs-exposed cells. Data are presented as Mean ± SEM (n ≥ 3). Statistical significance: *P < 0.05,**P < 0.01,***P < 0.001,****P < 0.0001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) To elucidate the causal sequence between oxidative defensive imbalance and plasma membrane integrity loss, we performed a time-course assessment. LDH release increased in a time-dependent manner, achieving statistical significance after 4 h of TiO2 NPs exposure (Fig. 4e). Notably, the decline in GSH levels occurred in the early stages of exposure, preceding the onset of significant LDH release (Fig. 4f). This temporal pattern strongly suggests that redox imbalance is a preceding event rather than a secondary consequence of plasma membrane rupture. Subsequent investigations using JC-1 staining and MitoSOX probes characterized the specific mitochondrial lesions. JC-1 imaging revealed a dramatic transition from red-aggregated fluorescence to green-monomeric fluorescence, confirming extensive depolarization of the mitochondrial membrane potential (Fig. 4g and h). MitoSOX flow cytometry further corroborated a significant accumulation of mitochondrial-derived superoxide anions, indicating an mtROS burst (Fig. 4i and j). TEM provided conclusive ultrastructural evidence: whereas control mitochondria maintained intact morphology and dense cristae, TiO2 NPs-exposed mitochondria exhibited severe swelling, vacuolization, and cristae loss, alongside a distinctive migration and tethering of damaged mitochondria toward the plasma membrane periphery (Fig. 4k). Collectively, these findings establish that TiO2 NPs-induced mitochondrial structural and functional collapse, accompanied by an mtROS burst, occurs chronologically prior to plasma membrane rupture. Having identified mitochondrial impairment as a critical early event in TiO2 NPs-induced cellular damage, we further investigated whether the subsequent plasma membrane rupture was dependent on classical programmed cell death pathways. Prior to the formal intervention experiments, we performed dose-response assessments for various pathway-specific inducers and inhibitors using the CCK-8 assay to establish optimal working concentrations for our intestinal epithelial cell model (Fig. S4). In the assessment of the apoptotic pathway, although the pan-caspase inhibitor Z-VAD-FMK robustly rescued STS-induced cytotoxicity, it failed to mitigate the TiO2 NPs-induced decline in cell viability or the increase in LDH release (Fig. 5a and b). Western blot analysis revealed that while STS induced significant expression of Cleaved Caspase-3, TiO2 NPs exposure failed to elicit comparable Caspase-3 activation, further confirming the irrelevance of apoptosis (Fig. 5c and d). In the ferroptosis pathway, although Fer-1 effectively alleviated cytotoxicity in the RSL3-induced model, it showed no protective effect against TiO2 NPs-induced cell viability loss or membrane leakage (Fig. 5e and f). Despite a slight decrease in GPX4 expression following TiO2 NPs exposure, Fer-1 could not rescue the cell membrane integrity (Fig. 5g and h). Similarly, in the necroptosis pathway, while Nec-1s successfully suppressed TSZ-induced MLKL phosphorylation and subsequent cell death, it failed to ameliorate the TiO2 NPs-induced cytotoxicity (Fig. 5i–l). Finally, in the pyroptosis pathway, although VX-765 significantly mitigated damage and GSDMD cleavage in the LPS + Nigericin-positive control, it failed to improve TiO2 NPs-induced cytotoxicity or prevent GSDMD processing (Fig. 5m–p). To further determine whether activation of classical RCD pathways preceded plasma membrane rupture, we performed time-course Western blot analyses during the first 4 h of TiO2 NPs exposure. No significant changes were observed in the CC3/C3, p-MLKL/MLKL, or C-GSDMD/GSDMD ratios, and GPX4 expression remained largely unchanged throughout the examined period (Fig. S5). Notably, LDH release was already significantly elevated at 4 h (Fig. 4e), whereas these classical RCD-associated markers remained unaltered. These findings indicate that plasma membrane rupture occurs without detectable concurrent activation of apoptosis, necroptosis, pyroptosis, or ferroptosis during the early injury phase. Classical regulated cell death (RCD) pathways are not the primary mechanisms mediating TiO2 NPs-induced plasma membrane injury. (a–d) Assessment of the apoptotic pathway: (a) CCK-8 cell viability and (b) LDH release assays evaluating the impact of the pan-caspase inhibitor Z-VAD-FMK on TiO2 NPs-induced injury; (c, d) Western blot analysis (c) and densitometric quantification (d) of the CC3/C3 ratio, demonstrating a lack of caspase-3 activation upon TiO2 NPs exposure. (e–h) Assessment of the ferroptosis pathway: (e) CCK-8 viability and (f) LDH release assays evaluating the effect of the ferroptosis inhibitor Fer-1; (g, h) Western blot (g) and densitometric quantification (h) of GPX4. (i–l) Assessment of the necroptosis pathway: (i) CCK-8 viability and (j) LDH release assays evaluating the effect of the necroptosis inhibitor Nec-1s; (k, l) Western blot (k) and densitometric quantification (l) of the p-MLKL/MLKL ratio, showing minimal MLKL phosphorylation in TiO2 NPs-exposed cells. (m–p) Assessment of the pyroptosis pathway: (m) CCK-8 viability and (n) LDH release assays evaluating the effect of the pyroptosis inhibitor VX-765; (o, p) Western blot (o) and densitometric quantification (p) of the cleaved-GSDMD (C-GSDMD)/GSDMD ratio.Experimental groups include Control, TiO2 NPs, Positive inducer, Inhibitor, and TiO2 NPs + Inhibitor. Positive controls for apoptosis, ferroptosis, necroptosis, and pyroptosis were induced by Staurosporine (STS), RSL3, TSZ, and LPS + Nigericin, respectively, with their corresponding specific inhibitors (Z-VAD-FMK, Fer-1, Nec-1s, and VX-765). Data are presented as Mean ± SEM (n ≥ 3). Statistical significance is denoted by asterisks: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Synthesizing the data from Fig. 4, Fig. 5, TiO2 NPs exposure is characterized by a rapid collapse of mitochondrial structural and functional integrity, an mtROS burst, and GSH depletion. Importantly, while specific inhibitors for apoptosis, ferroptosis, necroptosis, and pyroptosis successfully functioned in their respective positive control groups, they were entirely ineffective in preventing TiO2 NPs-induced cell death. Collectively, the pharmacological exclusion and temporal analyses indicate that apoptosis, ferroptosis, necroptosis, and pyroptosis are unlikely to serve as the primary drivers of TiO2 NPs-induced plasma membrane rupture during the early injury phase. The AKT/mTORC2-RhoA axis mediates TiO2 NPs-induced F-actin disassembly and plasma membrane injury Having established that TiO2 NPs-induced plasma membrane rupture is independent of classical programmed cell death pathways and linked to early mitochondrial dysfunction, we explored the regulatory mechanism underlying cytoskeletal instability. We focused on the AKT/mTORC2-RhoA signaling axis and verified its role in TiO2 NPs-induced cytoskeletal disruption and membrane damage using the mTOR inhibitor Torin-1 (Fig. 6). Aberrant AKT/mTORC2 signaling axis activation mediates TiO2 NPs-induced F-actin disassembly and plasma membrane injury. (a, b) Western blot (a) and densitometric quantification of the p-AKT/AKT ratio (b), confirming TiO2 NPs-induced activation of the AKT/mTORC2 axis. (c) Quantification of RhoA expression, showing TiO2 NPs-mediated suppression of RhoA-associated cytoskeletal regulation. (d) Representative confocal fluorescence images of F-actin stained with phalloidin (green) and DAPI (blue), visualizing the disruption of cytoskeletal architecture following TiO2 NPs exposure. (e) Quantitative analysis of relative F-actin fluorescence intensity. (f, g) Western blot (f) and densitometric analysis of the p-AKT/AKT ratio (g), demonstrating the inhibitory effect of Torin-1 on TiO2 NPs-induced AKT hyper-phosphorylation. (h) Quantification of F-actin fluorescence intensity, showing the restoration of cytoskeletal integrity by Torin-1 intervention. (i) Quantification of RhoA levels, indicating the rescue of TiO2 NPs-suppressed RhoA by Torin-1. (j, k) CCK-8 cell viability (j) and LDH release (k) assays assessing the impact of Torin-1 on mitigating TiO2 NPs-induced cytotoxicity and plasma membrane rupture. (l) Representative phalloidin-stained confocal images displaying the recovery of cytoskeletal morphology in the TiO2 NPs + Torin-1 group. (m) Intracellular GSH content, revealing that Torin-1 intervention partially mitigates TiO2 NPs-induced depletion of antioxidant reserves. Experimental groups: Control, Torin-1, TiO2 NPs, and TiO2 NPs + Torin-1. Data are presented as Mean ± SEM (n ≥ 3). Statistical significance is denoted by brackets and asterisks: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Western blot analysis demonstrated that TiO2 NPs exposure significantly elevated p-AKT (Ser473) levels compared to the control, while total AKT expression remained unchanged (Fig. 6a). Densitometric quantification confirmed a significantly increased p-AKT/AKT ratio, indicating the aberrant activation of mTORC2-related signaling (Fig. 6b). Concurrently, TiO2 NPs exposure significantly suppressed RhoA expression, suggesting a functional impairment of RhoA-mediated cytoskeletal regulation (Fig. 6c). Phalloidin staining revealed a collapse of the cytoskeletal architecture: Control cells exhibited continuous F-actin fibers and intact cell edges, whereas TiO2 NPs treatment led to reduced F-actin fluorescence, disorganized fibers, and retraction or disintegration of peripheral lamellipodia-like structures (Fig. 6d and e). To determine whether mTOR signaling is causally linked to these phenotypes, we utilized Torin-1 in rescue experiments. Western blot analysis revealed that Torin-1 significantly attenuated TiO2 NPs-induced AKT hyper-phosphorylation (Fig. 6f and g). Regarding cytoskeletal recovery, Torin-1 intervention effectively rescued the TiO2 NPs-induced reduction in F-actin fluorescence intensity, facilitating the reorganization of the cytoskeletal network (Fig. 6h). RhoA quantification further indicated that Torin-1 restored RhoA protein levels lost upon TiO2 NPs exposure, suggesting that mTOR activation is upstream of RhoA suppression (Fig. 6i). Furthermore, endpoint injury assessments showed that Torin-1 improved cell viability (Fig. 6j) and significantly reduced LDH release (Fig. 6k), indicating that blocking aberrant mTOR signaling reinforces plasma membrane integrity. Phalloidin confocal imaging corroborated these findings, demonstrating continuous F-actin fiber distribution and restored cell edges in the TiO2 NPs + Torin-1 group (Fig. 6l). Notably, Torin-1 also significantly restored intracellular GSH levels, implying that the dysregulation of the AKT/mTORC2-RhoA axis is tightly linked to the depletion of endogenous antioxidant reserves (Fig. 6m). TiO2 NPs exposure induces aberrant activation of the AKT/mTORC2 signaling axis accompanied by RhoA suppression, driving the disassembly of the F-actin cytoskeleton. Torin-1 rescue experiments underscore that the AKT/mTORC2-RhoA axis acts as a pivotal regulatory node connecting early mitochondrial impairment to subsequent plasma membrane rupture. Mitochondrial-plasma membrane tethering promotes mtROS-dependent lipid peroxidation and barrier dysfunction To investigate whether mtROS drive TiO2 NPs-induced plasma membrane lipid peroxidation and subsequent epithelial barrier damage, we employed the broad-spectrum antioxidant NAC and the mitochondria-targeted ROS scavenger Mito-TEMPO. We monitored the physical tethering between mitochondria and the plasma membrane, 4-HNE accumulation, and barrier functional indicators under antioxidant intervention (Fig. 7). Dose-response assays confirmed that non-cytotoxic working concentrations for both NAC and Mito-TEMPO were established for subsequent experiments (Fig. S6a and b). Targeted scavenging of mitochondrial ROS attenuates TiO2 NPs-induced plasma membrane lipid peroxidation and epithelial barrier disintegration. (a) Representative Western blot analysis of mitochondrial markers (VDAC1 and TOMM20) and the plasma membrane marker (ATP1A1) in purified plasma membrane fractions. (b, c) Densitometric quantification of VDAC1 (b) and TOMM20 (c) normalized to ATP1A1, indicating the aberrant accumulation of mitochondrial components at the plasma membrane. (d) Quantification of 4-HNE levels in the isolated plasma membrane fractions, demonstrating localized lipid peroxidation following TiO2 NPs exposure. (e–g) Mitigation of TiO2 NPs-induced cellular damage by the broad-spectrum antioxidant NAC and the mitochondria-targeted ROS scavenger Mito-TEMPO, as assessed by CCK-8 cell viability (e), LDH release (f), and TEER measurements for barrier integrity (g). (h, i) Quantification of relative intracellular ROS (h) and whole-cell 4-HNE levels (i), confirming the effective suppression of oxidative stress and lipid peroxidation by antioxidant interventions. Positive controls were included in (f) and (h) to validate assay responsiveness. (j) Representative Western blot analysis of tight junction proteins (ZO-1, Occludin) and mitochondrial proteins (MFN2, TOMM20) in whole-cell lysates. (k–n) Densitometric quantification of TOMM20 (k), MFN2 (l), Occludin (m), and ZO-1 (n) normalized to β-actin, illustrating the restoration of structural and mitochondrial proteins following ROS scavenging. Experimental groups: Control, TiO2 NPs, NAC, Mito-TEMPO, TiO2 NPs + NAC, and TiO2 NPs + Mito-TEMPO. Data are presented as Mean ± SEM (n ≥ 3). Statistical significance is denoted by brackets and asterisks: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Protein abundance analysis of purified plasma membrane fractions revealed that TiO2 NPs exposure significantly enriched mitochondrial markers, VDAC1 and TOMM20, within the plasma membrane (Fig. 7a). Densitometric quantification further confirmed increased VDAC1/ATP1A1 and TOMM20/ATP1A1 ratios, providing biochemical evidence for the aberrant tethering of mitochondria to the plasma membrane (Fig. 7b and c). Notably, 4-HNE levels within the plasma membrane fraction were drastically elevated in the TiO2 NPs group, indicating localized lipid peroxidation damage at the membrane surface (Fig. 7d). Subsequent rescue experiments demonstrated that both NAC and Mito-TEMPO significantly mitigated TiO2 NPs-induced cytotoxicity and plasma membrane rupture, as evidenced by improved cell viability and reduced LDH release (Fig. 7e and f). Furthermore, TEER measurements confirmed that antioxidant intervention partially rescued the epithelial barrier integrity compromised by TiO2 NPs (Fig. 7g). ROS quantification and 4-HNE assays showed that both antioxidants effectively quenched intracellular ROS bursts (Fig. 7h) and curtailed lipid peroxidation (Fig. 7i), while concomitantly restoring GSH levels depleted by TiO2 NPs, thus rectifying the redox imbalance (Fig. S7). Western blot analysis indicated that TiO2 NPs-induced downregulation of ZO-1, Occludin, MFN2, and TOMM20 was substantially alleviated by NAC or Mito-TEMPO treatment (Fig. 7j). Densitometric quantification corroborated that antioxidant intervention successfully reversed the loss of both mitochondria-associated (TOMM20, MFN2) and tight junction-related (Occludin, ZO-1) proteins (Fig. 7k–n). TiO2 NPs exposure promotes the aberrant accumulation of mitochondrial components at the plasma membrane, coupled with localized 4-HNE elevation, ROS accumulation, LDH leakage, and TEER decline. The rescue effects of NAC and Mito-TEMPO demonstrate that ROS—specifically mtROS—are the core effectors of TiO2 NPs-induced lipid peroxidation and epithelial barrier disintegration. These results support the conclusion that abnormal mitochondrial-plasma membrane tethering and the resulting localized oxidative damage constitute a critical pathogenic mechanism in TiO2 NPs-induced membrane rupture. Quantitative lipidomics reveals TiO2 NPs-induced plasma membrane lipid remodeling To delineate the chemical basis of TiO2 NPs-induced plasma membrane damage at the fundamental lipid molecular level, highly covered quantitative lipidomics was performed on purified plasma membrane fractions from the Control and TiO2 NPs-exposed groups (Fig. 8). The OPLS-DA score plot demonstrated a complete spatial separation of the two groups with robust model parameters (R2 = 0.782、R2 = 0.998、Q2 = 0.95), indicating a profound and global alteration of the plasma membrane lipidomic signature following TiO2 NPs exposure (Fig. 8a). Rigorous permutation testing confirmed a significant decrease in the predictive capability (Q2) after randomly shuffling group labels, validating the model's stability and mitigating the risk of overfitting (Fig. S8a). TiO2 NPs exposure drives extensive remodeling of the plasma membrane lipidome and disrupts localized lipid metabolic networks. (a) Orthogonal partial least squares-discriminant analysis (OPLS-DA) score plot of the plasma membrane lipidome, demonstrating distinct spatial separation between the Control and TiO2 NPs-exposed groups. (b) Volcano plot illustrating candidate differential lipids. Red, yellow, and grey dots represent significantly upregulated, downregulated, and non-significant lipids, respectively. Key oxidized lipid mediators, structural phospholipids, lysophospholipids, and free fatty acids are specifically annotated. (c) Hierarchical clustering heatmap of representative differential lipids, visualizing the distinct abundance signatures of mechanism-relevant lipid modules between the two groups. (d) Doughnut chart depicting the hierarchical class and subclass distribution of the candidate differential lipids. (e) Relative summed abundance of key structural and functional lipid subclasses (PC, PE, PA, PS, LPE, LPI, and FFA), normalized to the Control group. (f) Sankey diagram mapping representative differential lipids to their corresponding core metabolic networks, including glycerophospholipid, arachidonic acid, linoleic acid, glycerolipid, and sphingolipid metabolism. (g–k) Absolute quantification of representative single lipid molecules, revealing the accumulation of pro-inflammatory oxidative mediators PGE2 (g) and 12,13-EpOME (i) and the structural phospholipid PC(16:1_18:1) (h), alongside the catastrophic depletion of membrane-stabilizing lipids PS(16:0_22:6) (j) and LPE(22:6) (k) in TiO2 NPs-exposed cells. Data are presented as Mean ± SEM (n ≥ 3) unless otherwise indicated. Statistical significance is denoted by p-values or asterisks: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Abbreviations: OPLS-DA, orthogonal partial least squares-discriminant analysis; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PA, phosphatidic acid; PS, phosphatidylserine; LPE, lysophosphatidylethanolamine; LPI, lysophosphatidylinositol; FFA, free fatty acid. Collectively, these lipidomic alterations provide microscopic chemical evidence for TiO2 NPs-induced localized membrane lipid peroxidation and structural destabilization. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Differential lipid analysis identified 176 significant candidate differential lipids (32 upregulated and 144 downregulated). The volcano plot mapped the dramatic perturbations in oxidized lipid mediators and core structural lipids: notably, oxidized mediators such as PGE2 and 12,13-EpOME, alongside the structural phospholipid PC(16:1_18:1), accumulated remarkably in the TiO2 NPs group. Conversely, membrane homeostasis-associated lipids, including PS(16:0_22:6), the lysophospholipid LPE(22:6), and various free fatty acids (FFAs), exhibited catastrophic depletion (Fig. 8b). Hierarchical clustering heatmap of mechanism-relevant differential lipids distinctly demarcated the abundance boundaries between the two groups, implying systematic lipid reprogramming across modules encompassing oxidized lipid mediators, PC/PE/PA, PS, LPE/LPI, and FFAs (Fig. 8c). The lipid classification composition doughnut chart revealed that the candidate differential lipids were broadly distributed across major categories, including glycerophospholipids, glycerolipids, fatty acyls, and sphingolipids. This widespread distribution indicates that TiO2 NPs induces a massive reallocation of the entire plasma membrane lipid network rather than random depletion of single lipid species (Fig. 8d). Subsequent aggregation of total abundances for mechanism-relevant secondary lipid subclasses revealed a stark polarization: TiO2 NPs treatment significantly expanded the relative total pools of PC, PE, and PA, while drastically contracting the pools of PS, LPE, LPI, and FFAs (Fig. 8e). This dichotomy suggests that TiO2 NPs exposure simultaneously triggers aberrant structural phospholipid remodeling and erodes both membrane-stabilizing lipids and the lysophospholipid pools requisite for membrane repair. To investigate whether these alterations were accompanied by changes in the physical properties of the membrane lipids, we calculated the DBI for four major lipid classes (FFA, PE, PS, PC). Unexpectedly, no statistically significant differences in DBI levels were observed between the two groups for any of these classes (P = 0.186, 0.769, 0.379, 0.624, respectively) (Fig. S8b). This implies that TiO2 NPs-induced membrane remodeling destroys membrane homeostasis not merely by altering global lipid saturation, but rather by specifically consuming or enriching key subclasses and representative oxidized lipid molecules. To anchor these isolated lipid molecules to specific biochemical reaction networks, a differential lipid-metabolic pathway Sankey diagram was constructed. The representative differential lipids mapped highly to core pathways including glycerophospholipid metabolism, arachidonic acid metabolism, linoleic acid metabolism, and sphingolipid metabolism (Fig. 8f). Specifically, PGE2 and 12,13-EpOME were precisely mapped to the oxidative metabolism pathways of arachidonic acid and linoleic acid, respectively, whereas PC, PE, and PS dominated the glycerophospholipid metabolic network. This indicates that aberrant phospholipid remodeling coincides with intense lipid peroxidation immediately prior to membrane rupture. Absolute quantification of representative single molecules cemented this trend: pro-inflammatory oxidative mediators (PGE2 and 12,13-EpOME) and PC(16:1_18:1) surged in the TiO2 NPs group (Fig. 8g–i), whereas PS(16:0_22:6) and LPE(22:6), which are critical for maintaining membrane asymmetry and curvature, were significantly depleted (Fig. 8j–k). Quantitative lipidomics confirms that TiO2 NPs exposure induces deep remodeling of the plasma membrane lipidome. The central hallmarks of this process are the explosive accumulation of oxidized lipid mediators, the proportional dysregulation of structural phospholipids (e.g., PC, PE, PA), and the catastrophic loss of key membrane-stabilizing lipids (PS, LPE, LPI, FFA). Integrated with the prior findings of localized 4-HNE elevation at the plasma membrane and the robust rescue effect of antioxidants, this microscopic lipid molecular evidence perfectly closes the mechanistic loop, validating the model of TiO2 NPs-induced localized membrane lipid peroxidation (Mitoxyperilysis) and subsequent physical barrier disintegration. Discussion The intestinal epithelial barrier is remarkably vulnerable to ingested food additives. Notably, our simulated digestion characterizations revealed that pristine nanoscale TiO2 NPs undergoes profound physicochemical transformations, losing electrostatic repulsion to form dense, micrometer-scale agglomerates wrapped in a complex biological corona. While the barrier-disrupting capabilities of TiO2 NPs NPs are well-documented [37,38], the precise mechanisms leading to the ultimate physical tearing of the plasma membrane by these massive digestive entities have remained elusive. In the present study, we demonstrated that TiO2 NPs-induced intestinal epithelial barrier disintegration is not primarily driven by conventional systemic biochemical toxicity, but rather by a non-canonical, spatially driven mode of cell death: mitoxyperilysis (mitochondria-driven localized plasma membrane lipid peroxidation) [16]. Our findings delineate the pathological progression of this phenomenon, spanning from upstream kinase dysregulation to cytoskeletal paralysis and eventual localized lipid peroxidation. Accurately defining the mode of cell death remains a central challenge in nanotoxicology, given that nanomaterials frequently induce pleiotropic cellular stress responses [[39], [40], [41]]. Using a pharmacological exclusion matrix (Fig. 5), we found that apoptosis, ferroptosis, necroptosis, and pyroptosis are unlikely to represent the primary drivers of TiO2 NPs-induced membrane rupture. Although specific inhibitors (Z-VAD-FMK, Fer-1, Nec-1s, and VX-765) exhibited robust rescue effects in their respective positive control models, they failed to mitigate TiO2 NPs-induced LDH release or restore epithelial viability. Notably, classical ferroptosis is characterized by an iron-dependent lipid peroxidation cascade governed by the GPX4 regulatory axis [[42], [43], [44]]. The inability of ferrostatin-1 to prevent TiO2 NPs-induced cell lysis therefore indicates a mechanistic divergence from classical ferroptosis. This interpretation was further supported by the time-course analysis, in which LDH release became significantly elevated at 4 h, whereas the CC3/C3, p-MLKL/MLKL, and C-GSDMD/GSDMD ratios remained unchanged and GPX4 expression was largely stable during the same 0–4 h period (Fig. S5). Thus, plasma membrane rupture occurred without detectable concurrent activation of these classical regulated cell death pathways during the early injury phase. Although secondary activation at later stages cannot be excluded, these findings strongly argue against apoptosis, ferroptosis, necroptosis, or pyroptosis as the initiating mechanisms of TiO2 NPs-induced membrane rupture [45]. The present mechanism also shares conceptual features with the recently described mitoxyperiosis pathway, in which mitochondrial dysfunction directly contributes to plasma membrane lysis. However, the TiO2 NP-associated process identified here is distinguished by its upstream trigger and spatial organization: digestive-transformed TiO2 agglomerates perturb the mTOR/RhoA–F-actin axis, promote abnormal mitochondrial redistribution toward the plasma membrane, and generate spatially restricted membrane lipid peroxidation. We therefore regard mitoxyperilysis as a nanotoxicology-specific mitochondria-associated lytic mechanism that is mechanistically related to mitoxyperiosis rather than a completely unrelated cell-death program. Transcriptomic profiling (Fig. 3) directed our mechanistic focus toward profound perturbations in cytoskeletal regulation and mTOR signaling. It is highly plausible that the deposition of the aforementioned micrometer-scale, biocorona-coated TiO2 NPs agglomerates imposes severe physical and mechanical stress on the apical surface of enterocytes, acting as the upstream mechanosensory trigger for these signaling cascades. Subsequent respirometry and fluorescent staining revealed severe mitochondrial respiratory dysfunction accompanied by massive ROS accumulation (Fig. 4). To connect these phenomena, we investigated RhoA, the small GTPase governing actin polymerization and contractility [46,47]. Immunoblotting and fluorescent imaging (Fig. 6) demonstrated that TiO2 NPs exposure induced aberrant AKT phosphorylation, indicative of mTORC2 activation, and significantly suppressed RhoA. Biomechanically, RhoA orchestrates the assembly of actin stress fibers and maintains membrane tension [[48], [49], [50]]. Its profound suppression by the dysregulated mTORC2 axis inevitably culminates in widespread F-actin disassembly and the loss of cellular structural rigidity [51,52]. The mTOR inhibitor Torin-1 successfully restored the F-actin network architecture and rescued cell viability (Fig. 6), validating the AKT/mTORC2-RhoA axis as the upstream initiator of this toxicity. The consequent paralysis of the actin cytoskeleton creates the critical physical prerequisite for mitoxyperilysis, allowing mitochondria to escape their normal intracellular constraints and drift toward the cellular periphery [[53], [54], [55]]. Recent studies in other biological systems have also highlighted that engineered nanoparticles can induce oxidative responses and structural remodeling. Organo-nanoparticle formulations and silver nanoparticles have been shown to alter morphological and biochemical responses under abiotic stress conditions, while photocatalytic metal-based nanomaterials can generate controlled reactive oxygen species through surface-dependent mechanisms [[56], [57], [58]]. Although these studies were performed in agricultural and environmental models, they support the broader concept that surface-transformed nanoparticles can trigger localized oxidative stress and cellular structural alterations. This concept is consistent with our findings that digestive-transformed TiO2 NPs induce spatially restricted mtROS-associated membrane damage and cytoskeletal disruption in intestinal epithelial cells. The structural collapse of the cytoskeleton directly precipitates the abnormal subcellular redistribution of mitochondria. Our subcellular fractionation data (Fig. 7) provided biochemical evidence that mitochondrial markers (VDAC1 and TOMM20) were aberrantly enriched in the purified plasma membrane fraction following TiO2 NPs exposure. Without the F-actin network to dynamically regulate membrane protrusions and organelle retraction, translocated mitochondria remain trapped at the cell periphery. This mitochondrial redistribution toward the plasma membrane may create a highly localized oxidative microenvironment, permitting dysfunctional mitochondria to deploy mtROS directly at the membrane interface, thereby acting as proximal sources of oxidative stress [59,60]. The ability of the mitochondria-targeted antioxidant Mito-TEMPO to restore epithelial barrier function (TEER), block LDH leakage, and rescue tight junction proteins (Fig. 7) functionally corroborates that this membrane-localized mtROS serves as the direct executioner of membrane rupture. Although these findings collectively support a mitochondria-associated membrane injury model, direct visualization of dynamic mitochondria–plasma membrane contact formation was not performed in the present study. Therefore, the proposed mitochondrial tethering process should be interpreted as a mechanistically supported model based on biochemical fractionation, cytoskeletal remodeling, and spatial lipid peroxidation evidence. Future studies using live-cell imaging and super-resolution approaches will be valuable to directly characterize the spatial and temporal dynamics of mitochondria–plasma membrane interactions. The ultimate biochemical consequence of this targeted oxidative attack was captured by quantitative lipidomic analysis (Fig. 8). TiO2 NPs exposure provoked intense lipidomic reprogramming at the plasma membrane, hallmarked by the substantial generation of pro-inflammatory oxidized lipid mediators (e.g., PGE2, 12,13-EpOME) and a severe depletion of membrane-stabilizing lipids, particularly phosphatidylserine (PS) and lysophosphatidylethanolamine (LPE). PS is strictly maintained on the inner leaflet of the plasma membrane to preserve surface charge, while LPE provides critical flexibility for accommodating membrane curvature [[61], [62], [63], [64]]. Their rapid depletion compromises the fundamental biophysics of the lipid bilayer. Notably, the overall DBI of major lipid classes remained unaltered across groups (Fig. S7). This specific detail indicates that the physical destruction of the plasma membrane did not result from a systemic shift in global lipid saturation, but rather from a highly specific, localized consumption of key structural lipids driven by proximal mtROS [[65], [66], [67]]. The selective depletion of PS and LPE despite an unchanged DBI suggests that these lipid alterations may not simply reflect a generalized increase in membrane lipid unsaturation or nonspecific oxidation. Instead, these changes may arise from selective vulnerability and oxidation of specific phospholipid species, impaired membrane repair capacity, or altered phospholipid remodeling and trafficking pathways. PS and PE-related phospholipids are critical structural components of cellular membranes, and their homeostasis is tightly regulated by lipid synthesis, remodeling, and intracellular transport processes [68,69]. Oxidative modification of PS has also been shown to alter membrane properties and phospholipid signaling, highlighting the selective susceptibility of specific phospholipid species to oxidative stress [70]. One limitation of the present study is that the proposed mechanism was investigated primarily using an in vitro Caco-2/HT29-MTX intestinal epithelial co-culture model and has not yet been validated in vivo. Although this model provides a well-established platform for controlled mechanistic investigation of digestive-transformed TiO2 NPs-induced epithelial injury, it cannot fully recapitulate the complexity of the intestinal microenvironment, including interactions among epithelial cells, immune components, microbiota, and physiological exposure conditions. Therefore, future studies using appropriate animal models are warranted to further validate whether digestive-transformed TiO2 NPs induce intestinal barrier dysfunction, mucosal lipid peroxidation, and mitochondria-associated membrane damage in vivo, and to further confirm the proposed mitoxyperilysis mechanism under physiological exposure conditions. In summary, our study reveals a non-canonical, mitochondria-driven mechanism underlying TiO2 NPs-induced intestinal epithelial barrier dysfunction. Rather than operating through classical programmed cell death pathways, the massive, digestively transformed TiO2 NPs agglomerates disrupt the AKT/mTORC2-RhoA axis, causing F-actin cytoskeletal paralysis. This structural defect promotes abnormal mitochondrial redistribution toward the plasma membrane, triggering localized lipid peroxidation (mitoxyperilysis). These findings not only decode the specific mode of cell death induced by TiO2 NPs but also highlight the AKT/mTORC2-RhoA-actin signaling cascade and localized mtROS as promising therapeutic targets for preventing nanomaterial-induced intestinal toxicity. Conclusion In conclusion, this study identifies a non-canonical, spatially driven mode of cell death—mitoxyperilysis—as the fundamental mechanism underlying intestinal epithelial barrier disintegration following oral TiO2 NPs nanoparticle exposure. We demonstrate that upon interacting with the gut epithelium, TiO2 NPs agglomerates severely perturb the AKT/mTORC2-RhoA signaling axis, leading to the structural paralysis of the F-actin cytoskeleton. This loss of physical constraint permits dysfunctional mitochondria to aberrantly migrate and tether to the inner leaflet of the plasma membrane, thereby discharging high concentrations of mtROS directly at the cellular periphery. The ensuing targeted lipid peroxidation catastrophically depletes structural phospholipids and generates an abundance of oxidized lipid mediators. Ultimately, this localized oxidative burst culminates in physical membrane rupture and tight junction degradation, without detectable early activation of the classical regulated cell death pathways examined. This study not only resolves the mechanistic enigma of atypical membrane rupture in nanotoxicity but also underscores the critical importance of maintaining cytoskeletal-mitochondrial spatial homeostasis. Consequently, modulators targeting the AKT/mTORC2-RhoA axis or mitochondria-specific antioxidants hold significant promise as therapeutic interventions to safeguard against nanomaterial-induced intestinal barrier dysfunction and its associated inflammatory disorders. CRediT authorship contribution statement Chuxin Zhang: Writing – original draft, Validation, Software, Methodology, Investigation, Data curation. Yiping Liu: Writing – original draft, Validation, Software, Investigation, Data curation. Xu Li: Validation, Formal analysis. Jiaxing An: Formal analysis. Wenjing Li: Methodology. Zhilu Zhou: Validation. Yitao Yan: Visualization. Juan Shao: Methodology. Zhongyuan Guo: Writing – review & editing, Supervision, Funding acquisition, Formal analysis, Conceptualization. Funding statement This work was supported by the financial support was provided by the Fundamental Research Program of Shanxi Province (202303021211131); and The Doctoral Start-up Fund of Shanxi Province (SD2236). Declaration of competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103674. Supporting Information Supporting Information is available from the Online Library or from the author. Supplementary data The following is the Supplementary data to this article. Multimedia component 1 Data availability Data will be made available on request. References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement Data will be made available on request.