Migrasomes constrained by a homologous-targeting photodynamic nanoplatform: enhancing intratumoral CD8⁺ T-cell-associated antitumor immunity in oral squamous cell carcinoma

Abstract

Photodynamic therapy (PDT) is an effective adjunct treatment for oral squamous cell carcinoma (OSCC). Enhancing photosensitizer targeting and inducing effective cytotoxic T-cell responses through photoimmunotherapy have become key strategies to improve PDT efficacy. Migrasomes, as vesicular structures assembled by TSPAN4 and cholesterol microdomains, are implicated in immune escape and are emerging as sensitization targets for PDT. Here, we report a biomimetic nanoplatform, MOF-919@CCM, that combines enhanced tumor-cell membrane adhesion with light-controlled cholesterol degradation. Cloaking with a homologous cancer-cell membrane (CCM) imparts specific adhesion to tumor cells and improves targeted delivery of the photosensitizer. Moreover, the transition-metal nodes of MOF-919 exhibit peroxidase- and catalase-like activities that alleviate tumor hypoxia and, under laser irradiation, effectively reduce cellular cholesterol levels. Experiments further revealed that PDT based on MOF-919@CCM markedly suppresses migrasome formation via effective degradation of cholesterol and promotes CD8⁺ T-cell infiltration and cytotoxic activity against tumor cells. This work develops a targeted PDT approach using MOF-919@CCM and provides a new strategy for the immunotherapy of OSCC.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04037-6.

Introduction

Photodynamic therapy (PDT), as a significant adjunctive treatment for tumors, offers nonsurgical therapeutic options for patients with early-stage malignancies and holds promise for functional preservation and extended survival in advanced disease [1, 2]. Current evidence indicates that the efficacy of adjunct therapies such as PDT relies on robust and durable antitumor T-cell responses [3, 4]. However, existing photosensitizers are constrained by the hypoxic, acidic, and immunosuppressive characteristics of the tumor microenvironment, making it difficult to fully induce effective T-cell responses. The efficacy of PDT is further limited by the inherently poor depth of tissue penetration, collectively restricting its broad clinical application [5]. Therefore, enhancing tumor-targeting efficiency within the complex physicochemical properties of the tumor microenvironment (TME) and amplifying PDT-induced antitumor T-cell immunity constitute effective strategies for further improving PDT efficacy.

Migrasomes contribute to resistance against T-cell–based adjuvant therapies across multiple cancers, including head and neck squamous cell carcinoma (HNSCC), and are regarded as potential prognostic and predictive indicators of the immune response [6–10]. Migrasomes are pomegranate-like organelles (0.5–3 μm) that arise during cell migration through the cooperative action of TSPAN4 and cholesterol within membrane microdomains, forming structures packed with numerous microvesicles [11, 12]. Tumor cells employ migrasomes to execute a migrasome-dependent quality-control process that actively removes damaged organelles, maintains intracellular homeostasis, and thereby enhances resistance to physicochemical stress, such as drug cytotoxicity [13, 14]. Migrasomes produced by highly migratory tumor cells can carry signals that promote an immunosuppressive microenvironment and serve as key mediators of communication between tumor cells, stromal cells, and immune cells [15–17]. Tumor-derived migrasomes facilitate invasion of surrounding tissues and bone by epithelial or gland-derived cancers through the induction of neovascularization and abnormal osteoclast differentiation [18–20]; in parallel, they foster an immunosuppressive milieu and establish a premetastatic niche via extracellular-matrix remodeling and changes in tumor-cell motility [21]. The resulting reduction in T-cell cytotoxicity and abundance is closely associated with the failure of adjuvant cancer therapies. At present, small-molecule drugs or inhibitors that specifically target migrasomes have yet to be identified. Given that migrasome formation depends on TSPAN4 assembly and plasma-membrane mechanics maintained by appropriate cholesterol levels, disrupting cellular cholesterol homeostasis and suppressing TSPAN4 expression have emerged as effective strategies to reduce migrasome production [22, 23]. Such interventions provide a feasible approach to enhance local T-cell killing and increase T-cell abundance by limiting migrasome formation in tumor cells.

While PDT can oxidize membrane cholesterol, the principal photoproduct cholesterol 5α-hydroperoxide (5α-OOH) is short-lived in the highly reducing tumor microenvironment and is readily eliminated by a two-electron reduction pathway catalyzed by intracellular systems such as glutathione peroxidase 4 (GPX4) [24–26]. Moreover, tumor hypoxia suppresses the chain propagation of lipid peroxidation, leading to sustained cholesterol depletion [27]. Notably, transition metal catalysts with peroxidase-like activity convert cholesterol 5α-OOH via a one-electron reduction pathway to the cholesterol alkoxyl radical (Cho·) [28]. This highly reactive radical initiates post-illumination chain peroxidation, increases LOOH/ChOOH formation, and amplifies membrane lipid damage [29]. Among current photosensitizer platforms, bimetallic metal–organic frameworks (MOFs) possess distinctive enzyme-mimetic activities. Their catalase-like activity decomposes endogenous H₂O₂ to O₂, thereby alleviating tumor hypoxia. While their peroxidase-like activity promotes chain peroxidation of lipid hydroperoxides, enhancing membrane cholesterol depletion [30]. In addition, cancer cell membrane (CCM) cloaking imparts photosensitizer homotypic adhesion and immune camouflage, markedly increasing tumor uptake and localizing reactive oxygen species (ROS) generation near cholesterol-rich lipid raft domains [31, 32]. It has indicated Membrane-localized ROS more readily induce cholesterol peroxidation and alter membrane viscosity [33, 34], which holds promise for suppressing migrasome biogenesis and weakening migrasome-mediated immune shielding, thereby potentially strengthening T-cell responses.

Herein, we present a multifunctional photodynamic nanoplatform, MOF-919@CCM, that combines a Cu/Al bimetallic catalytic core with CCM cloaking (Scheme 1), integrating two key functions: (i) an enzyme-mimetic Cu/Al core that alleviates intratumoral hypoxia and amplifies ROS generation to reinforce PDT efficacy; and (ii) a CCM coating that confers homotypic adhesion and immune camouflage, enhances tumor accumulation, and lowers membrane cholesterol under irradiation, which suppresses migrasome formation and augments CD8⁺ T-cell cytotoxicity. In our models, indirect inhibition of migrasome formation elicited durable and effective cytotoxic CD8⁺ T-cell responses, suggesting a feasible strategy to mitigate immunosuppressive barriers in the TME. Collectively, these findings broaden the prospects of PDT as an adjunct therapy for OSCC and provide preliminary validation of migrasomes as actionable targets, offering a path toward improved survival and functional preservation.

Scheme 1.

Schematic illustration of the biomimetic platform and its antitumor mechanism. MOF-919-Cu-Al is assembledfromH₂PyC and metal nodes and then cloaked with a homologous cancer cell membrane to yield MOF-919@CCM, which confers homotypic adhesion. After administration, the nanoplatform preferentially adhered to OSCC. Under laser irradiation, MOF-919@CCM exhibits enzyme-mimetic activity that relieves hypoxia while lowering cellular cholesterol, thereby disrupting TSPAN4-cholesterol microdomains and inhibits migrasome formation. The resulting loss of migrasomes shielding promotes intratumoral CD8⁺T-cell infiltration and cytotoxicity, remodels the immune microenvironment, and leads to tumor regression

Materials and methods

Synthesis of MOF-919 (MOF-919-Cu-Al)

MOF-919-Cu-Al was synthesized based on a previously reported method with slight modifications to reduce the particle size and enlarge pore diameter [35]. Specifically, 39 mg of H₂PyC, 207.1 mg of Cu (NO₃)₂·3H₂O, and 43.1 mg of Al(NO₃)₃·6H₂O were dissolved in 10 mL of DMF. The solution was stirred for 10 min and then ultrasonicated for 10 min before being transferred into a Teflon-lined stainless-steel autoclave. The sealed vessel was heated at 120 °C for 12 h and then allowed to cool naturally to room temperature. The as-synthesized MOF-919 samples were immersed in DMF for three days, and the solvent was changed approximately every 8 h using fresh DMF. The DMF molecules in the samples were subsequently replaced with anhydrous ethanol, which was also refreshed every 8 h with fresh ethanol. The resulting solids were then soaked in acetone for 10 h, followed by vacuum drying for 2 h. Finally, the dried powder samples were heated under vacuum at 120 °C for 18 h to obtain activated mesoporous MOF-919 samples.

Instrumentation

Transmission electron microscopy (TEM) was performed using a JEM-2100F (JEOL, Japan) for tissue imaging and an HT7800 (Hitachi, Japan) for material characterization. X-ray photoelectron spectroscopy (XPS) was conducted on an ESCALAB 250Xi spectrometer (Thermo Fisher Scientific, USA). Thermogravimetric analysis (TGA) was carried out on a TG 209 F3 analyzer (NETZSCH, Germany). Fourier transform infrared (FT-IR) spectra were acquired using a TENSOR27 spectrometer (Bruker, Germany). Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) was performed using a Agilent 720-ES (Varian, USA). High-content imaging analysis was performed on an Operetta CLS system (PerkinElmer, USA). Atomic force microscopy (AFM) measurements were taken on a Dimension Fastscan Bio instrument (Bruker, Germany). Laser scanning confocal microscopy was conducted on an FV3000 microscope (Olympus, Japan). In vivo fluorescence imaging was performed using an IVIS Spectrum system (PerkinElmer, USA). Digital whole-slide images were acquired with an Aperio AT2 scanner (Leica Biosystems, USA). Gas adsorption analyses were performed with a BELSORP-Max instrument (MicrotracBEL, Japan). Paraffin sections were prepared using a Thermo Shandon Finesse 325 microtome (Thermo Fisher Scientific, USA). Flow cytometry (FCM) was performed on an LSRFortessa (BD Biosciences, USA) and a CytoFLEX (Beckman Coulter, USA) cytometer.

Coating with homologous cancer cell membrane and SDS-PAGE analysis

Cancer cell membranes were extracted from MOC1 cells using a membrane protein extraction kit (P0033, Beyotime) according to the manufacturer’s instructions. To coat the nanoparticles with the cell membranes, a membrane suspension (1 mg/mL, 500 µL) and a MOF-919 nanoparticle solution (1 mg/mL, 500 µL) were mixed by pipetting, followed by sonication in an ice-water bath (180 W, 3 min). The resulting suspension was sequentially extruded through porous polycarbonate membranes (pore sizes: 1000, 800, 400, and 200 nm). Excess membrane debris was removed by centrifugation (1000 × g, 10 min, 4 °C). The final membrane-coated nanoparticles (MOF-919@CCM) were obtained and stored at 4 °C for further use. The protein compositions of the isolated MOC1 cells, CCM vesicles, and MOF-919@CCM nanoparticles were analyzed by SDS‒PAGE, followed by Coomassie Brilliant Blue staining.

Statistical analysis

Data were presented as mean ± standard deviation (SD), with sample size (n ≥ 3) specified in the figure legends. Student’s t test was used for comparisons between two groups. For multiple comparisons involving more than two groups, one-way ANOVA followed by Tukey’s multiple comparison tests was employed. The specific statistical tests and corresponding P values were provided in the figure legends. Statistical significance was indicated in the figures as *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant. All of the flow cytometry analyses were performed using FlowJo 10.8.1. Data analysis was conducted via GraphPad Prism 9.0.

Results and discussion

Construction and functional characterization of a biomimetic catalytic nanoplatform MOF-919@CCM

Specifically, MOF-919-Cu-Al (MOF-919), synthesized from Cu/Al bimetallic nodes and H₂PyC ligands [35, 36], exhibits intrinsic catalytic activity and a prominent size effect [37, 38]. Coating with a homologous CCM markedly improved the targeting ability of MOF-919@CCM in complex tumor microenvironments. (Fig. 1A). TEM images (Fig. 1B-D) revealed that MOF-919 possessed a quasi-spherical morphology with well-defined polyhedral edges (≈ 200 nm). By precisely optimizing its size, our MOF-919 is expected to exhibit tumor accumulation and retention via the EPR effect compared with the previously reported larger MOF-919. The extracted CCM vesicles exhibited unilamellar membrane structures, and the final MOF-919@CCM particles presented an increased diameter (≈ 250 nm) with a distinct membrane layer, indicating successful biomimetic coating. Elemental mapping (Fig. 1E) showed uniform distribution of Cu, Al, C, and O, whereas energy-dispersive spectroscopy (EDS) (Fig. S1) confirmed the incorporation of Cu and Al via characteristic Cu K_α/K_β and Al peaks. Elevated C, N, and O signals are attributed to organic ligands and membrane components, confirming the integration of functional components and successful biomimetic modification. The porous structure of MOF-919 was confirmed through nitrogen adsorption–desorption analysis, which exhibited a characteristic type IV isotherm with a distinct hysteresis loop. The material exhibited a high Brunauer‒Emmett–Teller (BET) surface area of 1,538 m²/g and a total pore volume of 1.725 cm³/g. Barret-Joyner-Halenda (BJH) and density functional theory (DFT) analyses revealed a predominant pore diameter of approximately 3.86 nm and a corresponding pore volume of 0.871 cm³/g (Fig. S2A, B). Compared with previously reported MOF-919 frameworks, the Cu/Al hybrid exhibited an improved specific surface area and pore size, indicating superior potential for catalytic efficiency and catalytic diffusion [35, 39, 40].

Fig. 1.

Synthesis and characterization of MOF-919@CCM. (A) Schematic illustration of the synthetic process of MOF-919@CCM. (B-D) Representative TEM images of (B) MOF-919, (C) CCM vesicles, and (D) MOF-919@CCM nanoparticles. Scale bars: 100 nm. (E) Elemental mapping of MOF-919@CCM showing a uniform distribution of Cu, Al, C and O across the nanostructure. Scale bars: 100 nm. (F) DLS size distributions of MOF-919 and MOF-919@CCM (n = 3). (G) Zeta potential measurements of MOF-919 and MOF-919@CCM (n = 3). (H, I) Time-dependent changes in zeta potential (H) and DLS size distributions (I) of MOF-919@CCM in PBS containing 10% FBS at 37 °C over 7 days (n = 3). (J, K) DLS size distributions of MOF-919@CCM in PBS at 37 °C under pH 7.4 (J) and pH 6.8 (K) over 24 h (n = 3). (L, M) DLS size distributions of MOF-919@CCM in PBS at 37 °C with 1× PBS (L) and 2× PBS (M) over 24 h (n = 3) (N) FT-IR spectra of MOF-919 and MOF-919@CCM. (O) TGA analysis of MOF-919. (P) Analysis of the protein compositions of Ⅰ: isolated membrane (MOC1 cells), Ⅱ: CCM, and Ⅲ: MOF-919@CCM by SDS-PAGE. Data are presented as mean ± SD

Dynamic light scattering (DLS) measurements indicated that MOF-919 had an average size of 194.5 nm and a zeta potential of + 24.7 mV (Fig. 1F, G). After the CCM coating, the average hydrodynamic diameter increased to 266.7 nm and the zeta potential shifted to −16.5 mV, confirming successful coating and charge shielding by the cell membrane. The hydrodynamic diameter measured by DLS was slightly larger than the size observed by TEM (≈ 250 nm). This deviation can be attributed to the fact that DLS accounts for the soft membrane shell and the surrounding hydration layer of particles in solution, whereas TEM reveals the projected dimensions of particles in a dehydrated state after staining and drying [41]. These physicochemical changes underscore efficient membrane fusion and imply the favorable tumor-targeting capability of the biomimetic nanoplatform. We further evaluated the stability of MOF-919@CCM under physiologically relevant conditions. As shown in Fig. 1H-M, MOF-919@CCM maintained a nearly constant hydrodynamic size and zeta potential for up to 7 days in PBS containing 10% FBS, and showed good stability under different pH and ionic strength over 48 h. These results indicated that MOF-919@CCM preserves its structural integrity and is well suited for application within the tumor microenvironment. We further examined the potential premature leakage of Cu ions from MOF-919@CCM under physiological conditions. MOF-919@CCM (1 mg/mL) was incubated in PBS (pH 7.4) for 48 h in the absence of H₂O₂, and Cu-ion release was quantified by ICP-OES. ICP-OES analysis revealed minimal Cu leakage at 1, 6, 12, 24, and 48 h, with the Cu concentration in the supernatant reaching only 7.01 ppb after 48 h (Fig. S3). This high stability is likely attributable to the CCM shell, which shields the MOF-919 core from the aqueous environment and thereby suppresses decomposition and unintended Cu-ion release. These results confirm that MOF-919@CCM exhibits excellent structural stability under normal physiological conditions, indicating a negligible risk of systemic toxicity prior to activation.

XPS analysis confirmed the presence of C, O, Al, and Cu, supporting the successful formation of the framework (Fig. S4). FT-IR was further employed to verify the functional groups and confirm the successful construction of the MOF-based nanosystem (Fig. 1N). MOF-919 exhibited characteristic peaks at 1625 cm⁻¹ and 1385 cm⁻¹, corresponding to asymmetric and symmetric –COO^- stretching vibrations, indicating metal–ligand coordination. An additional signal at approximately 3433 cm⁻¹ was assigned to O–H stretching vibrations, confirming the presence of terminal hydroxyl groups. After CCM loading, a new C = O peak at ≈ 1690 cm⁻¹ appeared, which enhanced the O–H/C–H signals, along with additional membrane-derived bands at ≈ 1250 cm⁻¹ and 1080 cm⁻¹, confirming successful CCM coating. Notably, no significant shift in COO^- bands was observed, indicating that the MOF framework remained structurally intact. The thermal behavior of the nanoplatform was assessed by TGA of MOF-919 (Fig. 1O). MOF-919 showed a two-step weight loss with a final residue of 43.24%, which was attributed to the thermal decomposition of organic ligands and retention of the metal–organic framework. SDS‒PAGE analysis (Fig. 1P) revealed that the protein bands from the isolated membrane (lane I), CCM vesicles (lane II), and MOF-919@CCM (lane III) exhibited similar profiles, indicating effective retention of natural membrane proteins during the coating process.

The TME is characterized by mild acidity, hypoxia, redox imbalance, and elevated levels of endogenous H₂O₂ [40]. These pathophysiological conditions not only promote tumor progression but also present unique opportunities for the rational development of stimuli-responsive cancer therapies [42–44]. In response, we systematically investigated the enzyme-mimicking catalytic performance of our MOF-based platform under TME-mimetic conditions, leveraging the synergistic benefits of catalytically active metal centers and a highly porous framework. First, we evaluated the peroxidase-like (POD-like) activity of MOF-919. As shown in Fig. 2A, under various pH conditions, the material effectively catalyzed the oxidation of 3,3’,5,5’-tetramethylbenzidine (TMB) by H₂O₂, resulting in a significant time-dependent increase within a certain period of time. This observation indicates the formation of oxidized TMB (oxTMB) and preliminarily confirms the POD-like catalytic activity of MOF-919. To further explore its catalytic behavior under different physicochemical conditions, we examined the effects of pH, temperature, and substrate concentration. As shown in Fig. 2B, the catalytic efficiency peaked at pH 5.0, with respectable activity still retained at pH 5.5 and 6.8, and was markedly suppressed under neutral and alkaline conditions. This pH-dependent profile highlights the selective activation of MOF-919 in the mildly acidic tumor microenvironment and indicates its biosafety for normal tissues. Figure 2C and D show that the enzyme-like catalytic activity of MOF-919 increased with increasing temperature. The activity reached a considerable level at 37 °C, suggesting promising potential for in vivo applications. Furthermore, as shown in Fig. 2E, MOF-919 strongly depended on the H₂O₂ concentration, indicating its substrate-responsive catalytic behavior. To further elucidate the underlying mechanism, a methylene blue (MB) degradation assay was performed (Fig. 2F), confirming that MOF-919 catalyzed the production of highly cytotoxic hydroxyl radicals (·OH) from H₂O₂. These reactive species are considered key intermediates in peroxidase-like catalytic processes [45]. In addition to exhibiting POD-like reactivity, MOF-919 also exhibited catalase (CAT)-like behavior. As shown in Fig. 2G, the catalase-like activity was evaluated by monitoring the generation of dissolved O₂ from H₂O₂. The O₂ concentration increased steadily over time and clearly exhibited a concentration-dependent trend. Specifically, with 75 µg·Ml⁻¹ MOF-919, the O₂ level reached 6.72 mg·L⁻¹ at 300 s, confirming its efficient catalytic decomposition of H₂O₂ into O₂. This catalytic capacity was positively correlated with the MOF-919 concentration, suggesting its potential for relieving tumor hypoxia. Moreover, MOF-919 demonstrated excellent photodynamic performance under 670 nm laser irradiation. As shown in Fig. 2H, the fluorescence intensity of the singlet oxygen sensor green (SOSG) probe gradually increased, indicating the continuous generation of singlet oxygen (¹O₂).

Fig. 2.

Multi-enzymatic catalytic properties and photodynamic reactivity of MOF-919. (A) Time-dependent absorbance of MOF-919 as a peroxidase-like catalyst against pH. (B) Peroxidase-like catalytic activity of the MOF-919 as a function of pH (n = 3). (C) Time-dependent absorbance of MOF-919 as a peroxidase-like catalyst as a function of temperature. (D) Peroxidase-like catalytic activity of MOF-919 against temperature (n = 3). (E) Peroxidase-like catalytic activity of the MOF against the H₂O₂ concentration. (F) Hydroxyl radical generation detected by the MB assay. (G) Catalase-like activity of MOF-919 evaluated by measuring dissolved oxygen generation at different concentrations. (H) Time-dependent ¹O₂ generation under 670 nm laser irradiation detected by the SOSG probe. (I) Catalytic activity of MOF-919 toward cholesterol at different concentrations in the presence of laser irradiation and TMB Data are presented as mean ± SD

Prior studies indicate that a significant mechanism of photodynamic cytotoxicity is the ¹O₂–mediated oxidation of cholesterol in cell membranes [46, 47]. Given the efficient production of ¹O₂ by MOF-919 upon laser irradiation, its ability to mediate cholesterol oxidation under photodynamic conditions was investigated. A cascade reaction system using TMB as the chromogenic substrate was employed. As shown in Fig. 2I, the control group (MOF-919 + laser + TMB) exhibited negligible absorbance at 652 nm, indicating that the system itself did not significantly oxidize TMB. Upon the addition of cholesterol, however, a marked increase in absorbance was observed, which was positively correlated with the cholesterol concentration (0–200 µM). This can be attributed to a cascade mechanism in which photodynamically generated ¹O₂ oxidizes cholesterol into lipid hydroperoxides (e.g., 5α/5β-OOH) [48, 49], which are subsequently catalyzed by the peroxidase-like activity of MOF-919 to oxidize TMB, yielding the blue-colored product oxTMB [45]. These results confirm that MOF-919 enables cholesterol oxidation through the synergistic effect of the photodynamic generation of ¹O₂ and enzyme-mimicking catalysis.

Under a simulated TME, MOF-919@CCM exhibited POD-like catalysis, CAT-like oxygen production, and photo-triggered ROS generation. These capabilities establish a cascade for ROS amplification that effectively oxidizes cholesterol upon laser irradiation. This cascade enables MOF-919@CCM to increase ROS generation and cholesterol consumption, highlighting its potential for significantly enhancing PDT efficacy.

Homotypic targeting and enhanced photodynamic therapy enabled by MOF-919@CCM

Building upon the catalytic framework established in Fig.2, we next evaluated the photodynamic therapeutic potential of MOF-919@CCM in MOC1 cells, focusing on its tumor-homing ability and ability to induce intracellular oxidative stress. To assess the toxicity of MOF-919 and MOF-919@CCM, MOC1 cells were incubated with various concentrations of MOF-919 and MOF-919@CCM under dark conditions. As shown in Fig. S5, all formulations exhibited negligible cytotoxicity, indicating excellent biocompatibility. To evaluate the cellular uptake and homotypic targeting behavior of MOF-919@CCM, the MOF-919 nanocore was coated with DiI-labeled (red fluorescence) MOC1 cell membranes to construct MOF-919@CCM nanoparticles. The membranes of MOC1, human oral keratinocyte (HOK), and MTCQ1 cells were subsequently labeled with DiO (green fluorescence). After co-incubation of MOF-919@CCM with the three labeled cell lines, confocal microscopy revealed marked uptake of MOF-919@CCM by homologous MOC1 cells compared with HOK and MTCQ1 cells (Fig. 3A). This selective uptake is attributed to the functional membrane proteins and adhesion molecules retained on the homologous CCM coating, which facilitates specific recognition and efficient endocytosis by parent tumor cells. This homotypic targeting mechanism aligns with recent findings that membrane-camouflaged nanoplatforms are preferentially internalized by their originating cancer cells via membrane protein-mediated recognition [50].

Fig. 3.

Homotypic tumor targeting and intracellular ROS amplification under laser activation enable enhanced PDT. (A) Confocal laser scanning microscopy (CLSM) images of different cell lines (MOC1, HOK, and MTCQ1) after incubation with MOF-919@CCM. MOC1 cells were homologous to the CCM used in the MOF-919@CCM coating. Nuclei: DAPI (cyan); cell membrane: DiO (green); CCM (MOC1 cells): Dil (red); MOF-919 signal: Cy5.5 (blue). Scale bar: 10 μm. (B) Bio-TEM images of MOC1 cells incubated with MOF-919@CCM at different time points. The yellow and red dashed boxes indicate regions shown at higher magnification. The blue arrows highlight internalized nanozymes. Scale bars, left: 5 μm; middle: 1 μm; right: 0.5 μm. (C) Representative FCM profiles showing the time-dependent cellular uptake of MOF-919@CCM by MOC1 cells. (D) Quantitative analysis of fluorescence intensity at different incubation time points (n = 3).(E) Cell viability of MOC1 cells incubated with MOF-919 and MOF-919@CCM at different concentrations with laser (n = 3). (+) indicates laser treatment. (n = 3) (F, G) Live/dead staining and quantitative analysis of MOC1 cells treated with MOF-919@CCM under laser irradiation. The cells were stained with calcein-AM (green, live cells) and PI (red, dead cells) and imaged by fluorescence microscopy. Scale bar: 150 μm (n = 3). (H) Quantitative analysis of the ROS fluorescence intensity measured by FCM (n = 3). Statistical significance was calculated via one-way ANOVA followed by Tukey’s multiple comparison test. Data are presented as mean ± SD.*p < 0.05, **p < 0.01, ***p < 0.001

Bio-transmission electron microscopy (Bio-TEM) further confirmed the time-dependent cellular accumulation and cytoplasmic localization of the nanoplatform (Fig. 3B). To further quantitatively demonstrate the enhanced cellular uptake, we performed FCM analysis. As shown in Fig. 3C, D and Fig. S6, MOF-919@CCM displayed significantly higher cellular internalization compared to MOF-919. This result indicates that cancer cell membrane coating facilitates active endocytic uptake and effective cytosolic delivery, laying the foundation for subsequent photodynamic and catalytic activity. Under laser irradiation, both MOF-919 and MOF-919@CCM exhibited dose-dependent cytotoxicity in MOC1 cells, with MOF-919@CCM showing superior antitumor efficiency, attributed to its enhanced homotypic recognition and cellular internalization (Fig. 3E). Live/dead staining further confirmed that MOF‑919@CCM under laser irradiation induced superior, concentration‑dependent cell death compared to MOF-919, with a clear concentration-dependent increase in cell death (Fig. 3F, G and Fig. S7A, B). The intracellular O₂ level under hypoxic conditions was assessed by the fluorescence indicator [Ru(dpp)₃]Cl₂, whose fluorescence can be quenched by O₂. Compared with those in the other groups, the fluorescence of MOF-919@CCM-treated cells was weaker, indicating that MOF-919@CCM can relieve tumor hypoxia (Fig. S8). H₂DCFDA-based ROS quantification revealed 15.28-fold and 12.59-fold increases in intracellular ROS in the irradiated MOF-919@CCM group compared with those in the control and laser-only groups (Figs. 3H and Fig. S9). This enhanced photodynamic effect is attributed to the peroxidase- and catalase-like activities of the MOF-919 core, which facilitate the in situ conversion of H₂O₂ to both O₂ and •OH and subsequently promote ¹O₂ generation under laser. Collectively, these results elucidate a mechanism that integrates homotypic tumor targeting, intracellular oxygen self-supply, and cascade ROS amplification, highlighting MOF-919@CCM as a promising photodynamic nanozyme platform capable of overcoming the dual challenges of tumor selectivity and hypoxia-induced PDT resistance.

Identification and characterization of migrasomes in OSCC

The migrasome is a key organelle that enables tumor cells to maintain self-quality control. It supports malignant invasion and metastasis by eliminating damaged intracellular components and transmitting immunosuppressive signals [51]. In addition to TSPAN4 expression, membrane cholesterol content is a critical regulator of migrasome biogenesis, as evidenced by prior studies demonstrating that the cholesterol scavenger Mβ-CD can effectively suppress migrasome formation. These findings suggest that MOF-919@CCM-mediated photodynamic therapy, which is capable of depleting membrane cholesterol, may represent a novel migrasome suppression strategy. This study investigated the correlation between migrasomes and tumor cell invasiveness in OSCC and then evaluated the effects of MOF-919@CCM-mediated PDT on migrasome formation. TEM of murine OSCC tissues revealed the presence of migrasomes exhibiting characteristic “pomegranate-like” vesicle structures (Fig. 4A). Recent research has demonstrated that PIGK is enriched in migrasomes and serves as a specific marker [52]. Immunofluorescence analysis further demonstrated that OSCC cells can generate migrasomes in vivo, which exhibit spatial colocalization with PD-L1⁺ tumor cells (Fig. 4B), suggesting that migrasomes are derived primarily from OSCC cells and are closely associated with immunosuppressive regions within the TME. These findings are consistent with previous reports indicating that migrasomes can carry PD-L1 on their surface and facilitate immune evasion [53]. Our findings suggest that OSCC may employ migrasomes as intercellular carriers for PD-L1 delivery, thereby modulating immune responses in the microenvironment. Therefore, we directly observed the formation of migrasomes in different OSCC cell lines and validated their association with OSCC invasion through transwell and high-content imaging assays. Isolation of migrasomes from MOC1 and MTCQ1 cells via TEM revealed classic membrane-bound structures containing varying numbers of smaller vesicles. AFM imaging clearly demonstrated that these migrasomes were intimately attached to retraction fibers projecting from the cell membrane (Fig. 4C-F). TSPAN4 overexpression has been shown to markedly increase migrasome formation [54]. Using wheat germ agglutinin (WGA) as a specific probe for migrasome detection [55], we further demonstrated that TSPAN4 overexpression significantly promoted migrasome generation and enhanced cell migration in both cell lines. Conversely, TSPAN4 knockdown markedly reduced migrasome abundance and suppressed cell migration (Fig. 4G-K and Fig. S10A-D). These results collectively revealed a positive correlation between migrasome and the invasive capacity of OSCC cells. High-content imaging was employed to dynamically capture the gradual extrusion and release of migrasome-like vesicles from membrane protrusions, which exhibited a marked time dependence throughout the process (Fig. 4L, M and Fig. S11A, B). OSCC cells overexpressing TSPAN4 migrated faster and produced more migrasomes, whereas TSPAN4 knockdown yielded the opposite phenotype. Notably, some migrasomes were sequentially arrayed along individual retraction fibers, indicating that migrasome formation is a dynamic, continuous process tightly coupled to cell migration (Video S1-S4). Collectively, these data support TSPAN4 as a positive regulator of migrasome biogenesis in OSCC and suggest a mechanistic link to enhanced motility behavior. Collectively, these findings demonstrate that OSCC cells can produce migrasomes, which are intimately associated with the establishment of an immunosuppressive TME. Moreover, the strong correlation between migrasomes and inhibitory immune checkpoint molecules implies that adjuvant therapies in OSCC could achieve therapeutic effects by altering migrasome abundance, prompting further therapeutic exploration.

Fig. 4.

Migrasome Formation in OSCC Characterized by Morphological and Dynamic Imaging. (A) Images of TEM with tumor tissues of OSCC mice, with magnified image of migrasomes on the right. Yellow star emphasizes the cytoplasm of tumor cells (brown tint), green star emphasizes the nucleus of tumor cells (blue tint) and red arrows emphasize migrasomes (green tint). (B) Immunostaining images of PanCK (cyan), PD-L1(yellow), PIGK (green) and DAPI (blue) in OSCC tissue, with magnified images of migrasomes on the right. White arrows emphasize migrasomes. Scale bar: 40 μm. (C) TEM images of negatively stained migrasomes isolated from MOC1 cells. Scale bar: 0.5 μm. (D) AFM images of migrasomes in MOC1 cells. Yellow arrows indicate retraction fibers; white circles indicate migrasomes structures. Scale bars: 4 μm and 1 μm. (E) TEM images of negatively stained migrasomes isolated from MTCQ1 cells. Scale bar: 200 nm. (F) AFM images of migrasomes in MTCQ1 cells. Yellow arrows indicate retraction fibers; white circles indicate migrasomes structures. Scale bars: 4 μm and 1 μm. (G) Schematic of migrasomes isolation procedure. (H) Images of migrasomes in control and TSPAN4 knockdown (top) MOC1 cells. Scale bar: 10 μm; Transwell assay assessing the impact of siTSPAN4 on the invasiveness of MOC1 cells (bottom). Scale bar: 50 μm. (I) Quantitative analysis of cell invasion in MOC1 cells and TSPAN4-overexpressing (n = 3). (J) Images of migrasomes in control and TSPAN4-overexpressing (top) MOC1 cells. Scale bar: 10 μm; Transwell invasion assay of TSPAN4-overexpressing MOC1 cells (bottom). Scale bar: 50 μm. (K) Quantitative analysis of cell invasion in siTSPAN4 MOC1 cells (n = 3). (L) Migrasome formation of TSPAN4-overexpressing MOC1 cells captured by high content imaging. Scale bar: 5 μm. (M) Migrasome formation of siTSPAN4 MOC1 cells captured by high content imaging. Scale bar: 5 μm. Statistical significance was calculated using Student’s t test or one-way ANOVA followed by Tukey’s multiple comparison test. Data are presented as mean ± SD. ***p < 0.001. ns, not significant

PDT inhibits TSPAN4–mediated migrasome formation in OSCC

Prior studies have shown that MOF-919@CCM-mediated PDT can oxidize plasma-membrane cholesterol. This oxidation leads to cholesterol depletion [56]. Additionally, cholesterol depletion has been shown to destabilize the lipid raft microdomains essential for migrasome assembly. We quantified membrane cholesterol using a cholera toxin B subunit (CTB) fluorescent probe. Only the MOF-919@CCM (+) group exhibited a marked decrease in CTB fluorescence, indicating the decomposition of cholesterol (Fig. 5A). Cholesterol is essential for migrasome formation. Its depletion tends to compromise migrasome structural integrity [54]. We subsequently evaluated the effect of MOF-919@CCM-mediated PDT on migrasome generation in OSCC. MOF-919@CCM with laser significantly reduced cell migration (Fig. S12) and migrasome formation under non-cytotoxic conditions (Fig. 5B, C), suggesting that this combined approach interferes with migrasome biogenesis. TSPAN4-enriched microdomains on retraction fibers cooperate with the cytoskeleton to assemble macrodomains that swell to form migrasomes, and migrasomes are rich in immunosuppressive ligands such as PD-L1 that promote immune evasion [57]. Immunofluorescence analysis of PD-L1 and WGA/F-actin–labeled migrasome structures revealed colocalization in TSPAN4-overexpressing cells (Fig. 5D-H), which was markedly attenuated after PDT. Flow cytometry further confirmed a decrease in PD-L1 expression under PDT (Fig. 5I). Taken together, these data indicate that MOF-919@CCM–PDT suppresses PD-L1, which is correlated with the disruption of migrasome integrity.

Fig. 5.

PDT Disrupts the Migrasome and Reverses Immune Suppression in OSCC (A) AF488-CTB-labeled membrane lipid rafts and DAPI-stained nuclei were visible to CLSM; scale bar: 10 μm. (B) Migrasomes were identified via WGA-based immunolabeling and quantified from 20 individual cells per group(n = 3). (C) Western blot analysis of the expression levels of migrasomes marker proteins in tumor cells subjected to different treatments. (D-H) Confocal laser scanning microscopy (CLSM) images of MOC1 cells and colocalization of F-actin and PD-L1 at distinct subregions of migrasomes under different treatments, as indicated by white lines. F-actin was stained with rhodamine-phalloidin (red), nuclei were stained with Hoechst (blue), and migrasomes were labeled with WGA (green) and PD-L1 (purple). Scale bars: 10 μm. (I) Representative FCM profiles and semi-quantification of PD-L1 expression after various treatments (n = 3). (J) Schematic illustration of the co-culture process of pretreated MOC1 cells and activated T cells in vitro. (K) Activated CD8⁺ T cells were cocultured with PBS-treated, OE TSPAN4-treated, MOF-919@CCM-treated, OE TSPAN4 + MOF-919@CCM(+)-treated and MOF-919@CCM(+)-treated MOC1 cells (E: T ratio = 1:1, 5:1, 10:1, 20:1). The resulting mixture was used for the lactate dehydrogenase (LDH) release assay (n = 3). (L) Apoptosis ratios of MOC1 cells treated with activated CD8⁺ T cells (E: T ratio = 10:1) (n = 3). (M, N) The release of (M) TNF-α and (N) IFN-γ cytokines in the supernatant of the coculture assay (E: T ratio = 10:1) was detected via ELISA (n = 3). Statistical significance was calculated via one-way ANOVA or two-way ANOVA followed by Tukey’s multiple comparison test. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant

PDT enhances CD8⁺ T-Cell cytotoxicity via disruption of migrasome-associated immune suppression

Evidence indicates that tumor-derived migrasomes promote immune evasion by modulating immune cell function and fostering an immunosuppressive microenvironment, which is correlated with T-cell dysfunction and poor prognosis [8, 58, 59]. Further studies revealed that PD-L1–bearing migrasomes can be internalized by neighboring cells and increase surface PD-L1, thereby suppressing CD8⁺ T-cell-mediated immune responses. Therefore, we evaluated how MOF-919@CCM with a laser influences CD8⁺ T-cell antitumor activity by attenuating or reversing migrasome-mediated immune suppression. A coculture system was established in which preactivated murine CD8⁺ T cells were incubated for 5 h with MOC1 cells subjected to subsequent treatments (control, OE TSPAN4, MOF-919@CCM, OE TSPAN4 + MOF-919@CCM(+), and MOF-919@CCM(+)). To assess cytotoxicity, the cells were coincubated at different effector-to-target ratios (E: T = 1:1, 5:1, 10:1, 20:1), and lactate dehydrogenase (LDH) release (Fig. 5J). As shown in Fig. 5K, the OE TSPAN4 group exhibited significantly reduced CD8⁺ T-cell–mediated tumor cell lysis, suggesting that migrasomes suppress CD8⁺ T-cell cytotoxicity. Upon PDT, T cells are unleashed from suppression and efficiently lyse the tumor cells. FCM analysis showed that total tumor cell apoptosis in the PDT group (Q2 + Q3, E: T = 10:1) reached 59.3%, which was significantly greater than that in the control and OE TSPAN4 groups (Fig. 5L and Fig. S13), which was consistent with the LDH cytotoxicity assay.

Moreover, the levels of IFN-γ and TNF-α in the co-culture supernatants increased 3.0-fold and 5.4-fold, respectively, compared with those in the control group (Fig. 5M, N), further confirming that CD8⁺ T cells have effector functions. Collectively, these findings demonstrate that PDT enhances CTL cytotoxicity by reducing the presentation of migrasomes. This finding potentially reveals a novel mechanism that contributes to the antitumor efficacy of PDT.

Antitumor effect of MOF-919@CCM in an OSCC tumor model

To further evaluate the therapeutic potential of MOF-919@CCM(+) in vitro, a subcutaneous OSCC model was established (Fig. 6A). The subsequent biodistribution was tracked using an IVIS imaging system. In the MOF-919 group, the fluorescence signal at the tumor site was relatively weak and declined rapidly over time, whereas MOF-919@CCM produced a much stronger and more sustained fluorescence signal from 12 to 48 h post-injection, indicating markedly enhanced tumor-targeting capability and retention (Fig. 6B). This performance can be attributed to the homotypic recognition of the membrane coating and the enhanced passive targeting by the unique characteristics of the tumor microenvironment, both of which collectively facilitate tumor penetration and retention [60]. On day 7, tumor-bearing mice were randomly divided into five groups to evaluate therapeutic efficacy and probe the functional role of migrasomes: control (G1), migrasomes (G2), MOF-919@CCM (G3), MOF-919@CCM(+) (G4), and MOF-919@CCM(+) + migrasomes groups (G5), where (+) denotes groups that received laser irradiation. The MOF-919@CCM suspension was administered intravenously to groups G3–G5, and groups G4 and G5 received local laser irradiation on two consecutive days. To directly assess the contribution of migrasomes to treatment outcome, purified migrasomes were injected intraperitoneally every other day for two weeks into groups G2 and G5 (the migrasomes group and MOF-919@CCM(+) + migrasomes group, respectively). In this design, group G2 evaluated the effect of exogenous migrasomes alone, whereas group G5 served as a mechanistic rescue experiment, in which exogenous migrasomes were used to recreate a migrasome-rich tumor microenvironment after MOF-919@CCM(+) treatment and to test whether restoring migrasomes could attenuate its antitumor and immunostimulatory effects. Tumor images (Fig. 6C) revealed that the MOF-919@CCM(+) group produced the greatest tumor growth inhibition. Conversely, adding exogenous migrasomes (MOF-919@CCM(+) + migrasomes group) partially reversed this therapeutic efficacy, implicating migrasomes as mediators of immune suppression. Notably, the migrasomes group did not significantly increase tumor growth relative to the control group. We speculate that this may be related to an intrinsically “cold” immune phenotype in the untreated tumors, with limited CD8⁺ T-cell infiltration and consequently few effector cells that can be further suppressed by additional migrasomes. By contrast, in MOF-919@CCM(+)-treated tumors, PDT is expected to convert the microenvironment into a more immunologically active state with increased recruitment and activation of CD8⁺ T cells, making these tumors more susceptible to the negative regulatory effects of exogenous migrasomes and helping to explain the partial loss of tumor control observed in the MOF-919@CCM(+) + migrasomes group.

Fig. 6.

In vivo antitumor efficacy. (A) Schematic illustration of the therapeutic protocol used in the OSCC model. C57BL/6 mice were randomly divided into five groups: G1, control; G2, migrasomes; G3, MOF-919@CCM; G4, MOF-919@CCM(+); and G5, MOF-919@CCM(+) + migrasomes. On day 7, the nanoplatform was intravenously administered and irradiated for 2 consecutive days. Intraperitoneal injections of purified migrasomes were given to groups 2 and 5 every other day for two weeks. (+) indicates laser treatment. (B) Fluorescence images of the mice at 1, 12, 24 h and 48 h after intravenous injection of MOF-919@CCM (n = 3). (C) Representative gross observation images of the tumor tissues in different groups at the endpoint (n = 5). (D) Relative tumor volume growth curves under different treatments. (E) Tumor weight under different treatments in each group (n = 5). (F) Curves of body weight under different treatments in each group (n = 5). (G) H&E staining of major organs to evaluate biosafety. Scale bar: 100 μm. Statistical significance was calculated via one-way ANOVA and two-way ANOVA followed by Tukey’s multiple comparison test. Data are presented as mean ± SD. ***p < 0.001, ns, not significant

Tumor volume monitoring (Fig. 6D) and tumor weight measurements (Fig. 6E) further supported the superior antitumor activity of MOF-919@CCM(+). Consistently, IHC staining of hypoxia-inducible factor-1α (HIF-1α) showed reduced expression in tumors from MOF-919@CCM-treated mice and an even more pronounced decrease in the MOF-919@CCM(+) group compared with the control and migrasomes groups, whereas co-administration of migrasomes partially restored HIF-1α levels (Fig. S14), indicating that MOF-919@CCM, particularly under laser, effectively alleviates the hypoxic tumor microenvironment. Importantly, all the mice maintained stable body weights (Fig. 6F), and no histopathological abnormalities were observed in major organs (Fig. 6G), supporting the biosafety and translational potential of this strategy.

Suppression of migrasome formation and local immune escape by PDT in vivo

Given that MOF-919@CCM plus laser irradiation could effectively enhance CTL cytotoxicity by inhibiting migrasomes in vitro, we further evaluated the immunostimulatory effects and mechanisms of MOF-919@CCM in vivo. As expected, TEM revealed obvious migrasome-like structures (0.5–3.0 μm, containing smaller vesicles) surrounding the tumor cells in the control group. In contrast, these structures were substantially diminished in MOF-919@CCM(+) treated tumors (Fig. 7A, B). Western blot analysis further demonstrated a significant reduction in migrasome marker (TSPAN4 and PIGK) expression, accompanied by a marked increase in CD8 expression, suggesting enhanced CD8⁺ T-cell infiltration. A moderate decrease in PanCK cells was also observed, which was consistent with the suppression of tumor epithelial cell proliferation (Fig. 7C). These data confirm that MOF-919@CCM(+) effectively suppresses migrasome formation and promotes antitumor immunity. To further elucidate the immunomodulatory mechanisms of this treatment, we performed multiplexed immunohistochemical (mIHC) staining to assess the intratumoral distributions of CD8⁺ T cells, PIGK (a migrasome marker), PD-L1, and PanCK (a tumor cell marker). In the control and MOF-919@CCM groups (Fig. 6D and Fig. S15), imaging revealed the expression of PIGK and PD-L1 within tumor tissues, with limited CD8 T-cell infiltration. Further analysis revealed that a subset of tumor cells co-expressed PIGK and PD-L1, suggesting that these cells are possibly migrasome-releasing cells. Several surrounding PIGK⁺ spots further indicated that migrasomes were released into the tumor microenvironment (Fig. 7E, F). Notably, these PIGK⁺PD-L1⁺ tumor cells were spatially adjacent to CD8⁺ T cells (Fig. 7G, H), suggesting that migrasome-delivered PD-L1 mediates apoptosis or exhaustion in CD8⁺ T cells. In contrast, MOF-919@CCM-treated tumors presented markedly reduced PIGK and PD-L1 expression, accompanied by increased CD8⁺ T-cell infiltration (Fig. 7I), suggesting that this therapeutic approach effectively alleviates TME-associated immune suppression. Quantitative analyses further confirmed a significant decrease in PIGK⁺, PD-L1⁺, and double-positive tumor cells, along with a marked increase in CD8⁺ T-cell density (Fig. 7J-M). These findings indicate that MOF-919@CCM-mediated PDT effectively restrained tumor growth and enhanced antitumor immunity by suppressing migrasome formation and enhancing intratumoral CD8⁺ T-cell infiltration and cytotoxicity, thereby alleviating tumor immune suppression. Nevertheless, to further advance this therapeutic strategy, we acknowledge the need for more systematic and in-depth analysis of the tumor immune microenvironment. Future studies will establish clinically relevant orthotopic OSCC models in the oral cavity and employ techniques such as multiparameter flow cytometry and single-cell analyses to comprehensively evaluate the antitumor efficacy of MOF-919@CCM in the native oral mucosa and its effects on immune cell subsets in both the tumor microenvironment and draining cervical lymph nodes (including metastatic foci). The long-term therapeutic potential and safety of this strategy will also require rigorous validation in diverse animal models.

Fig. 7.

MOF-919@CCM-Mediated PDT Alleviates Tumor Immunosuppression by Suppressing Migrasome Formation. (A) Representative TEM images of tumor tissues from the control and MOF-919@CCM (+) groups, with magnified images of migrasomes on the right. Yellow stars emphasize the cytoplasm of tumor cells (purple tint), green stars emphasize the nucleus of tumor cells (blue tint), white arrows emphasize migrasomes (green tint), and red arrows emphasize retraction fibers. (B) Structure of a migrasomes. (C) Western blot analysis of TSPAN4, PIGK, CD8, and PanCK expression in tumor tissues. (D) Upper panel: Representative multiplex immunofluorescence images of tumors from the control group, showing the expression of DAPI (blue), PIGK (green), PD-L1 (yellow), CD8 (red), and PanCK (purple). Lower panel: Magnified views of the boxed regions. Scale bars: 200 and 100 μm. (E, F) Representative images from the control group showing spatial colocalization of PD-L1⁺, PIGK⁺, and PanCK⁺ tumor cells. Scale bar: 25 μm. The yellow arrow indicates PD-L1⁺ PIGK⁺ PanCK⁺ triple-positive tumor cells; the red arrows mark dispersed PIGK⁺ migrasomes. (G, H) Representative images from the control group showing spatial colocalization of PD-L1⁺, PIGK⁺, and PanCK⁺ tumor cells. Scale bar: 25 μm. Yellow arrows indicate PD-L1⁺ PIGK⁺ PanCK⁺ triple-positive tumor cells; white arrows indicate dispersed CD8⁺ T cells. (I) Corresponding immunofluorescence images from the MOF-919@CCM (+) group. (J-M) Cell counts per mm² for different cell types: (J) CD8⁺ T cells, (K) PD-L1⁺ tumor cells (tumor cells were labeled with PanCK), (L) PIGK⁺ PanCK⁺ tumor cells and (M) PIGK⁺ PD-L1⁺PanCK⁺ triple-positive tumor cells (n = 3). Statistical significance was calculated via Student’s t test. Data are presented as mean ± SD. *p < 0.05, **p < 0.01

Conclusion

In this study, we developed a biomimetic nanoplatform (MOF-919@CCM) capable of homotypic tumor-targeted delivery and enzyme-mimicking activity. Upon laser irradiation, this system effectively alleviated tumor hypoxia and reduced cholesterol levels. We further demonstrated that OSCC progression is accompanied by increased cellular invasiveness and migrasome generation. By disrupting cholesterol homeostasis in tumor cell membranes, MOF-919@CCM-mediated PDT significantly suppressed migrasome formation in OSCC models. This leads to increased CD8⁺ T-cell infiltration and enhanced CD8⁺ T-cell cytotoxic function, thereby alleviating local CD8⁺ T-cell-associated immunosuppression and improving therapeutic outcomes. Overall, this work proposes a migrasome-oriented photodynamic immunotherapy strategy. Accordingly, MOF-919@CCM serves as a representative photodynamic nanozyme platform that restrains migrasome-driven tumor progression and potentiates antitumor CD8⁺ T-cell responses by modulating membrane cholesterol, while its mesoporous architecture is well suited for further drug loading. These findings provide a promising approach to addressing the current limitations of PDT in OSCC treatment and offer insights into the rational design of future photodynamic nanoplatforms. They may also have potential clinical implications for minimally invasive early-stage therapy and function-preserving strategies in advanced OSCC.

Abbreviations

PDT

Photodynamic therapy

OSCC

Oral squamous cell carcinoma

CCM

Cancer-cell membrane

TME

Tumor microenvironment

HNSCC

Head and neck squamous cell carcinoma

5α-OOH

5α-hydroperoxide

GPX4

Glutathione peroxidase 4

Cho·

Cholesterol alkoxyl radical

MOFs

Metal-organic frameworks

ROS

Reactive Oxygen Species

TEM

Transmission electron microscopy

XPS

X-ray photoelectron spectroscopy

TGA

Thermogravimetric analysis

FT-IR

Fourier transform infrared

AFM

Atomic force microscopy

FCM

Flow cytometry

EDS

Energy-dispersive spectroscopy

BET

Brunauer‒Emmett–Teller

BJH

Barret-Joyner-Halenda

DFT

Density functional theory

DLS

Dynamic light scattering

POD-like

Peroxidase-like

TMB

3,3’,5,5’-tetramethylbenzidine

oxTMB

Oxidized TMB

MB

Methylene blue

·OH

Hydroxyl radicals

CAT

Catalase

SOSG

Singlet oxygen sensor green

¹O₂

Singlet oxygen

HOK

Human oral keratinocyte

Bio-TEM

Bio-transmission electron microscopy

CLSM

Confocal laser scanning microscopy

WGA

Wheat germ agglutinin

CTB

Cholera toxin subunit B

LDH

Lactate dehydrogenase

H&E

Hematoxylin and eosin

mIHC

Multiplexed immunohistochemical

Author contributions

W.Z., Y.H., and D.Y. conceived and designed the study, provided critical reading and editing, and resourced the project. L.Z. performed most of the experiments, performed the statistical analysis for all the data, and wrote the original draft. H.G. performed the histological analysis. W.J. helped with the schematic illustration. F.S. participated in the characterization experiments of the materials. X.C. and K.P. participated in the animal feeding experiments and collected and analyzed tissues from the mice. R.L., K.L., and J.W. polished and edited this article. W.Z., Y.H., and D.Y. supervised the project. All the authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (no. 82373255, 82571048, and 82501230), the Natural Science Foundation of Guangdong Province (no. 2024A1515012918 and 2025A1515010769), and the Guangdong Medical Science Foundation (no. 557A2025238).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All animal experimental procedures were conducted in strict accordance with relevant laws and regulations and approved by the Institutional Animal Care and Use Committee (IACUC) of Sun Yat-Sen University (Reference: SYSU-IACUC- 2023002545).

Consent for publication

Written informed consent for publication was obtained from all participants involved in this study.

Competing interests

The authors declare no competing interests.