Mitochondria-targeted photodynamic nanoparticles boost antitumor immunity by suppressing mitophagy in osteosarcoma

Abstract

Osteosarcoma (OS) responds poorly to immunotherapy owing to its highly immunosuppressive phenotype. Photodynamic therapy (PDT) can induce mitochondrial damage by generating reactive oxygen species (ROS), thereby triggering immunogenic cell death (ICD) and activating antitumor immunity. However, mitochondrial damage readily activates mitophagy, which attenuates oxidative stress and compromises therapeutic efficacy. In this study, we construct a multifunctional nanoparticle (TPSM@IT-4Cl), which co-loads the photosensitizer IT-4Cl and the mitochondrial fission inhibitor Mdivi-1 and can target mitochondria. TPSM@IT-4Cl is selectively delivered to the mitochondria of tumor cells and releases drugs in a glutathione (GSH)-responsive manner within a high-GSH microenvironment. Under localized light irradiation, TPSM@IT-4Cl efficiently generates ROS via IT-4Cl to induce mitochondrial damage, while the released Mdivi-1 inhibits mitochondrial fission and thereby indirectly interferes with mitophagy, ultimately amplifying the efficacy of PDT. Both in vitro and in vivo studies indicated that the resulting ICD remodels the tumor immune microenvironment and elicits potent anti-tumor immunity. Moreover, TPSM@IT-4Cl exhibits significant antitumor efficacy in OS patient-derived xenograft (PDX) models, highlighting its translational potential. Collectively, we developed a mitochondria-targeted photodynamic nanoparticle with concomitant mitophagy inhibition, which may provide a feasible strategy to overcome the limitations of immunotherapy in OS.

Graphical abstract

Highlights

• •TPSM@IT-4Cl enables mitochondria-targeted and GSH-responsive co-delivery of IT-4Cl and Mdivi-1.
• •Upon light irradiation, TPSM@IT-4Cl induces mitochondrial oxidative damage in osteosarcoma cells.
• •Mitophagy inhibition further amplifies PDT-induced oxidative stress and mitochondrial injury.
• •Enhanced oxidative damage promotes ICD and remodels the immunosuppressive tumor microenvironment.
• •The PDX model provides strong support for future preclinical studies and translational applications.

1. Introduction

Osteosarcoma (OS) is one of the most common primary malignant bone tumors among adolescents [[1], [2], [3]]. Currently, the standard multimodal treatment strategy, based on the combination of surgery and multi-agent chemotherapy, continues to face several challenges, including high recurrence rates and chemoresistance [[4], [5], [6]]. In recent years, immunotherapeutic strategies have provided breakthrough efficacy in various solid tumors; however, their overall clinical benefit in OS remains unsatisfactory [[7], [8], [9]]. A key reason is that OS generally exhibits an “immune-cold” phenotype, characterized by insufficient tumor antigen presentation, low efficiency of T-cell priming, and an immunosuppressive tumor microenvironment (ITME), all of which undermine effective antitumor immune responses [[10], [11], [12], [13]]. Therefore, simultaneously enhancing tumor immunogenicity, promoting antigen presentation, and remodeling the immunosuppressive microenvironment in OS represents a major challenge for enhancing the efficacy of immunotherapy [14,15].

Photodynamic therapy (PDT) is an effective approach to enhance anti-tumor immune responses [[16], [17], [18]]. Upon light irradiation (L), it generates reactive oxygen species (ROS), kills tumor cells, and also induces immunogenic cell death (ICD) with the release of multiple immunostimulatory molecules; thus, PDT is regarded as a treatment modality with “immune-sensitizing” potential [[19], [20], [21], [22]]. However, the immune-activating effects of PDT in OS are often insufficient to establish durable and systemic immune protection [23,24]. This limitation may be closely associated with the short lifetime and limited diffusion distance of ROS and the rapid activation of adaptive cytoprotective mechanisms in tumor cells [[25], [26], [27], [28], [29]].

Mitochondria are central hubs for maintaining intracellular redox homeostasis and are among the organelles most directly exposed to and most sensitive to ROS [30,31]. While ROS induces mitochondrial damage, it simultaneously activates the mitochondrial quality-control system. Particularly, it activates mitophagy, a selective form of autophagy [[32], [33], [34]]. Mitophagy refers to the process through which cells recognize, sequester, and degrade damaged mitochondria via the autophagy-lysosome pathway, thereby preserving mitochondrial homeostasis [33,35]. During mitophagy, damaged mitochondria are labeled for clearance. For example, mitochondrial depolarization stabilizes PINK1 on the outer membrane, which recruits/activates Parkin and triggers ubiquitination of outer-membrane proteins [36,37]. Then, these ubiquitinated substrates will be linked by autophagy receptors (e.g., p62) to LC3-positive autophagosomal membranes. The ubiquitinated substrates will be degraded following fusion with lysosomes [38,39]. This process reduces sustained ROS accumulation and attenuates oxidative stress, thereby compromising the efficacy of PDT [30,40]. Therefore, inhibition of mitophagy may represent a rational strategy to amplify ROS-driven oxidative stress and improve the outcomes of photo-immunotherapy in OS [[41], [42], [43]].

In summary, we constructed a mitochondria-targeted nanodelivery particle featuring glutathione (GSH)-responsive release in the high-GSH tumor microenvironment. This platform co-loaded the photosensitizer IT-4Cl and the mitochondrial fission inhibitor Mdivi-1. Owing to its typical A-D-A conjugated structure, IT-4Cl exhibits favorable near-infrared absorption and efficient singlet oxygen (¹O₂) generation, enabling effective induction of mitochondrial oxidative damage under near-infrared excitation [44]. Meanwhile, Mdivi-1 inhibits mitochondrial fission and thereby interferes with mitophagy, reducing tumor cell resistance to oxidative stress, further aggravating mitochondrial damage, promoting cell death, and facilitating the release of immunogenic molecules [45].

To achieve this goal, we covalently conjugated the mitochondrial fission inhibitor Mdivi-1 to TPP-PEG-SH (TPS) via a disulfide bond, thereby synthesizing a GSH-responsive polymer, TPP-PEG-SS-Mdivi-1 (TPSM), which enabled stimuli-triggered release in the high-GSH milieu of OS. Triphenylphosphonium (TPP), a typical lipophilic cationic mitochondria-targeting moiety, can readily penetrate biological membranes and preferentially accumulate in mitochondria driven by the high mitochondrial membrane potential (MMP), thereby conferring mitochondrial targeting capability to the polymer [46]. Finally, IT-4Cl was incorporated through self-assembly to generate TPSM@IT-4Cl nanoparticles(Scheme 1A). After intravenous administration, these nanoparticles passively accumulated in the tumor region via the enhanced permeability and retention (EPR) effect. Following cellular uptake, they achieved mitochondrial targeting mediated by the TPP moiety. The high concentration of GSH in mitochondria triggers disulfide bond cleavage in the polymer, leading to the rapid release of both the photosensitizer and the inhibitor. On the one hand, the released IT-4Cl generates ROS under light irradiation, thereby inducing mitochondrial damage. On the other hand, the released Mdivi-1 suppresses mitophagy, thereby reducing tumor cell tolerance to oxidative stress, exacerbating mitochondrial damage, and ultimately inducing ICD. Exposure to multiple tumor-associated antigens(TAAs) promotes BMDC maturation, which facilitates CD8⁺ T cell differentiation and infiltration, remodels the immunosuppressive microenvironment of OS, and finally activates antitumor immune responses (Scheme 1B).

Scheme 1.

Schematic illustration of the design of TPSM@IT-4Cl and its photodynamic antitumor immunity mechanism. A) The mitochondrial fission inhibitor Mdivi-1 was conjugated to the TPS backbone via a disulfide bond to construct a GSH-responsive polymer (TPSM), which was subsequently co-assembled with the photosensitizer IT-4Cl to form TPSM@IT-4Cl nanoparticles. B) TPSM@IT-4Cl preferentially accumulates in mitochondria of OS cells and undergoes GSH-triggered disassembly/release in the reductive microenvironment. Upon light irradiation, IT-4Cl generates ROS to induce mitochondrial damage, while the released Mdivi-1 suppresses mitophagy and prevents the clearance of damaged mitochondria, leading to sustained oxidative stress. This cascade potentiates ICD and ultimately amplifies antitumor immune responses.

2. Results

2.1. Preparation and characterization of TPSM@IT-4Cl

First, TPS was covalently conjugated to the mitochondrial fission inhibitor Mdivi-1 via a disulfide bond to obtain the polymer TPSM. The detailed synthetic procedure is presented in Fig. S1. The ¹H NMR spectrum revealed two characteristic peaks of Mdivi-1: the methoxy group (–OCH₃) at 3.8 ppm and the thiol group (–SH) at 13.2 ppm. Compared to TPS and TPP-PEG-SS-Py, the –OCH₃ signal appeared in TPSM, indicating that Mdivi-1 was successfully conjugated to TPS. Meanwhile, the other characteristic –SH peak of Mdivi-1 disappeared in TPSM, confirming disulfide bond formation and validating the successful construction of the polymer TPSM(Fig. S2).

Next, the photosensitizer IT-4Cl was loaded via self-assembly to obtain TPSM@IT-4Cl nanoparticles. Scanning electron microscopy (SEM) showed that the nanoparticles were well dispersed with a uniform spherical morphology (Fig. 1A). Dynamic light scattering (DLS) measurements indicated a hydrodynamic diameter of 75.1 ± 1.8 nm, a polydispersity index (PDI) of 0.15 ± 0.03, and a zeta potential of +14.5 mV (Fig. 1B).

Fig. 1.

Characterization and in vitro performance evaluation of TPSM@IT-4Cl nanoparticles. A) SEM image of TPSM@IT-4Cl. B) Hydrodynamic diameter, PDI, and zeta potential of TPSM@IT-4Cl measured by DLS. C) UV–vis absorption spectra of TPS, Mdivi-1, TPSM, IT-4Cl, and TPSM@IT-4Cl. D) Fluorescence emission spectra of IT-4Cl and TPSM@IT-4Cl. E) Time-dependent absorbance changes of TPSM@IT-4Cl under continuous 735 nm laser irradiation (0.5 W/cm²) for 10 min. F) Comparison of the time-dependent normalized absorbance changes of TPSM@IT-4Cl and ICG under continuous 735 nm laser irradiation (0.5 W/cm²) for 10 min. G) Time-dependent absorbance changes of TPSM@IT-4Cl in aqueous solution over 7 days. H) Time-dependent fluorescence changes of SOSG in the presence of TPSM@IT-4Cl under 735 nm laser irradiation (0.5 W/cm²). I) Time-dependent fluorescence changes of DHR123 in the presence of TPSM@IT-4Cl under 735 nm laser irradiation (0.5 W/cm²). J) CLSM images showing the co-localization of TPSM@IT-4Cl with mitochondria (Mito-Tracker) in HOS cells at different incubation times. K) FCM analysis of the time-dependent cellular uptake of TPSM@IT-4Cl by HOS cells.

UV–vis absorption spectra indicated that TPSM displayed a characteristic absorption peak at 277 nm, whereas IT-4Cl exhibited a characteristic absorption peak at 677 nm. Notably, TPSM@IT-4Cl exhibited absorption peaks at both 277 nm and 754 nm. The peak at 277 nm corresponded to the characteristic absorption of TPSM, while the peak at 754 nm was red-shifted compared to that of free IT-4Cl. This finding suggests that the aggregation state of IT-4Cl changed after encapsulation and confirms successful loading (Fig. 1C). Fluorescence spectra indicated that IT-4Cl had a maximum emission at 752 nm, whereas TPSM@IT-4Cl exhibited a maximum emission peak at 814 nm, indicating that TPSM@IT-4Cl retained the fluorescent properties of IT-4Cl with a longer emission wavelength (Fig. 1D).

Subsequently, we evaluated the photostability of TPSM@IT-4Cl (Fig. 1E and F; Fig. S3A). After continuous light irradiation for 10 min, no obvious change in absorbance was observed for the nanoparticles. In contrast, indocyanine green (ICG) exhibited a marked decrease in absorbance after 10 min of 735 nm laser irradiation, indicating that TPSM@IT-4Cl possesses excellent photostability. In addition, no obvious fluctuation in absorbance was observed after TPSM@IT-4Cl was stored in aqueous solution for 7 days, demonstrating good stability of the nanoparticles in water (Fig. 1G; Fig. S3B). After incubation in PBS for 7 days and in culture medium containing 10% fetal bovine serum for 48 h, both the particle size and PDI of TPSM@IT-4Cl remained relatively stable, with no obvious increase in particle size or aggregation (Fig. S3C and D). These results indicate that TPSM@IT-4Cl possesses good colloidal stability under physiologically relevant conditions.

The ROS-generating capability of TPSM@IT-4Cl nanoparticles in aqueous solution was assessed using Singlet Oxygen Sensor Green (SOSG, for detecting ¹O₂), dihydrorhodamine 123 (DHR123, for detecting O₂•^-), and terephthalic acid (TA, for detecting •OH). In the presence of TPSM@IT-4Cl, the fluorescence intensities of SOSG and DHR123 markedly increased over time after 735 nm laser irradiation, whereas the fluorescence signals of SOSG and DHR123 exhibited no obvious changes in water alone (Fig. 1H and I; Fig. S4A–F). Moreover, compared to ICG, TPSM@IT-4Cl showed superior capacity to generate ¹O₂ and O₂•^-. However, no •OH production was detected in the TPSM@IT-4Cl aqueous solution under 735 nm laser irradiation (Fig. S4G and H).

Disulfide bonds have been widely reported as major GSH-responsive chemical structures [47]. To evaluate the GSH-responsive release behavior of TPSM@IT-4Cl, we simulated the low-GSH environment of blood circulation and the high-GSH intracellular environment in vitro. Under the 10 μM GSH condition, TPSM@IT-4Cl exhibited slow drug release with relatively low cumulative release. The cumulative release rates of IT-4Cl and Mdivi-1 within 48 h were 18.3% and 14.2%, respectively, suggesting that the nanoparticles remained relatively stable under low-reductive conditions. In contrast, under the 10 mM GSH condition, drug release was markedly accelerated, and the cumulative release rates of IT-4Cl and Mdivi-1 within 48 h increased to 70.8% and 61.2%, respectively. These results indicate that the high-GSH environment effectively triggered disulfide bond cleavage and nanoparticle disassembly, thereby promoting drug release (Fig. S5A and B).

To investigate the mitochondrial targeting capability and cellular uptake behavior of the nanoparticles, we loaded the nanoparticles with a FITC fluorescent surrogate and examined their colocalization with the mitochondrial probe Mito-Tracker in HOS cells. With prolonged incubation, the nanoparticle-associated intracellular fluorescence signal gradually increased, and pronounced yellow colocalization fluorescence was observed. Quantitative colocalization analysis showed that the Pearson's correlation coefficient (PCC) and overlap coefficient (OLC) were 0.78 and 0.82, respectively (Fig. 1J; Fig. S6A). Subsequently, we performed subcellular fractionation experiments. Western blot analysis showed that GAPDH was mainly detected in the cytosolic fraction, whereas TOM20 was predominantly enriched in the mitochondrial fraction, confirming the successful separation of cytosolic and mitochondrial fractions (Fig. S6B). Quantitative analysis showed that the drug contents in the cytosolic and mitochondrial fractions were 25.7% and 74.2%, respectively, further confirming the excellent mitochondrial targeting capability of the nanoparticles (Fig. S6C). Cellular uptake was then evaluated by flow cytometry (FCM), and the results were consistent with the confocal observations (Fig. 1K). Together, these fluorescence colocalization, subcellular fractionation, and FCM results demonstrate that TPSM@IT-4Cl can be efficiently internalized by OS cells and preferentially accumulate in mitochondria.

2.2. In vitro cytotoxicity and ROS-generating capability of TPSM@IT-4Cl

To evaluate the in vitro antitumor activity of TPSM@IT-4Cl, we first performed a CCK-8 assay to determine its effects on the proliferation of three OS cell lines over 24 h. The results of CCK-8 showed that Mdivi-1 at 1, 2, and 5 μM did not significantly affect the proliferation of the three OS cell lines (Fig. 2A). At 10 μM, free Mdivi-1, TPSM, and TPSM@IT-4Cl only slightly inhibited the viability of HOS cells. In the PDT-alone group (TPS@IT-4Cl + L), cell viability was approximately 65%, whereas in the combined treatment group (TPSM@IT-4Cl + L), cell viability decreased to 30%, indicating that inhibition of mitophagy can enhance the cytotoxic effects of PDT. Although 20 μM led to stronger inhibition, previous studies have indicated that Mdivi-1 at 20 μM can interfere with mitochondrial oxidative metabolism, thereby introducing metabolic confounding in addition to mitophagy inhibition [48]. Therefore, 10 μM Mdivi-1 was administered in subsequent experiments to inhibit mitophagy [49]. Consistent results were also observed in K7M2 and MG63 OS cell lines.

Fig. 2.

In vitro cytotoxicity and ROS-generating capability mediated by TPSM@IT-4Cl + L. A) Cell viability of HOS, MG63, and K7M2 cells after different treatments, assessed by the CCK-8 assay. B) Calcein-AM (green, live)/PI (red, dead) staining to evaluate treatment-induced cell killing in HOS cells. C) Quantification of the percentage of dead cells in Calcein-AM/PI staining. D) Apoptosis of HOS cells was analyzed by Annexin V/PI staining across groups. E) Quantification of the percentage of apoptotic cells in Annexin V/PI staining. F) Representative fluorescence images of intracellular ROS in HOS cells after different treatments. G) Quantification of relative fluorescence intensity of intracellular ROS. H) FCM analysis of intracellular ROS levels in HOS cells after different treatments. I) Quantification of intracellular ROS levels measured by FCM. J) Representative fluorescence images of mtROS after different treatments. K) Quantification of relative fluorescence intensity of mtROS. L) FCM analysis of mtROS levels after different treatments. M) Quantification of mtROS levels measured by FCM. Data are presented as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA. ns, not significant; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 vs. the PBS group.

Since the therapeutic effect of TPSM@IT-4Cl may arise from both drug combination and nanoplatform-mediated delivery, we further compared TPSM@IT-4Cl + L with a simple physical mixture of free IT-4Cl and free Mdivi-1 under light irradiation. The CCK-8 results showed that, compared with the PBS group, cell viability was significantly reduced in the IT-4Cl + L, IT-4Cl + Mdivi-1 + L, and TPSM@IT-4Cl + L groups, with viability values of 63.5%, 36.3%, and 16.8%, respectively (p < 0.0001) (Fig. S7A). Apoptosis-related Western blot analysis further showed that, compared with the IT-4Cl + Mdivi-1 + L group, TPSM@IT-4Cl + L markedly downregulated Bcl-2 expression and upregulated BAX expression (p < 0.0001) (Fig. S7B–D). These results indicate that the therapeutic advantage of TPSM@IT-4Cl is not merely derived from the simple combination of IT-4Cl and Mdivi-1, but is closely associated with nanoplatform-mediated co-delivery, mitochondrial targeting, and GSH-responsive release.

Calcein-AM/PI double staining was also conducted to assess cell viability (Fig. 2B and C; Fig. S8). The TPSM@IT-4Cl + L group exhibited the strongest cytotoxic effect, evidenced by the highest proportion of PI (red)-positive cells, reaching approximately 82.3%. In addition, FCM analysis indicated that the apoptosis rate of HOS cells increased to 57.9% after treatment with TPSM@IT-4Cl + L, which was significantly higher than that of the TPS@IT-4Cl + L group (35.3%) (p < 0.0001) (Fig. 2D and E). Collectively, these findings suggest that the combination of PDT and mitophagy inhibition confers superior anti-tumor activity compared to PDT alone.

Given that the cytotoxic effect of PDT primarily depends on ROS generation upon light irradiation, we evaluated intracellular ROS levels after different treatments using the fluorescent probe DCFH-DA. Compared to the control group, both TPS@IT-4Cl + L and TPSM@IT-4Cl + L produced markedly enhanced green fluorescence signals after light activation, indicating efficient ROS generation by the photosensitizer under irradiation (Fig. 2F and G). Notably, the fluorescence intensity exhibited an additional “burst-like” increase in the TPSM@IT-4Cl + L group and reached 2.39-fold that of the TPS@IT-4Cl + L group, with a significant difference (p < 0.0001). We hypothesized that this difference is associated with the negative-feedback regulation of oxidative stress by mitophagy, which may reduce ROS levels in the PDT-alone group. FCM analysis also supported this conclusion (Fig. 2H and I).

Mitochondria are central hubs of intracellular oxidative stress and represent one of the most direct subcellular targets of ROS generation during PDT [50,51]. Therefore, precise delivery of photosensitizers to mitochondria may enable more efficient oxidative damage at the subcellular level. Thus, we used the superoxide-specific fluorescent probe MitoSOX Red to detect mitochondrial ROS (mtROS) and to evaluate mitochondrial oxidative stress in HOS cells after different treatments. Upon light irradiation, MitoSOX fluorescence increased substantially in the TPS@IT-4Cl + L and TPSM@IT-4Cl + L groups, consistent with pronounced mitochondrial superoxide production (Fig. 2J and K). Notably, the signal was the strongest in the TPSM@IT-4Cl + L group, reaching 1.79-fold that of the TPS@IT-4Cl + L group, suggesting that it amplified mitochondrial oxidative stress. FCM analysis also confirmed these findings (Fig. 2L and M).

To determine whether TPSM@IT-4Cl could induce sustained oxidative stress, we performed a time-resolved mtROS generation analysis. The results showed that the TPS@IT-4Cl + L group exhibited an early increase in mtROS after light irradiation, followed by a gradual decline. In contrast, the TPSM@IT-4Cl + L group not only induced stronger mtROS generation at early time points but also maintained a higher oxidative stress level at later time points (Fig. S9). At the endpoint of detection, the mtROS level in the TPSM@IT-4Cl + L group was approximately 1.63-fold higher than that in the TPS@IT-4Cl + L group (p < 0.0001). These results suggest that Mdivi-1-mediated mitophagy inhibition may impair the clearance of damaged mitochondria, thereby promoting sustained mtROS accumulation.

2.3. Mechanistic investigation of the anti-tumor properties of TPSM@IT-4Cl

To elucidate the potential anti-tumor mechanisms of PDT, high-throughput transcriptomic sequencing (RNA-seq) was performed on OS HOS cells treated with PBS or TPSM@IT-4Cl + L, and systematically analyzed changes in their mRNA expression profile. In total, 19132 mRNA transcripts were detected. Using |Log2(fold change)| ≥ 1 and adj. p < 0.05 as the screening criteria, 8255 differentially expressed genes (DEGs) were identified, including 4080 upregulated and 4175 downregulated genes (Fig. 3A). To clarify the biological significance of these DEGs, we conducted Gene Ontology (GO) enrichment analysis covering biological process (BP), cellular component (CC), and molecular function (MF). The DEGs were mainly enriched in terms associated with responses to stress and pressure, mitochondria-associated components, and programmed cell death (Fig. 3B). These findings suggest that ROS generated by TPSM@IT-4Cl + L in response to light irradiation markedly disrupted mitochondrial homeostasis in tumor cells and was accompanied by activation of programmed cell death-related processes.

Fig. 3.

Transcriptomic profiling of mRNA expression following TPSM@IT-4Cl-mediated PDT. A) Volcano plot showing DEGs in the TPSM@IT-4Cl + L group compared with the control group, including upregulated and downregulated genes. (criteria: |log2FC| ≥ 1 and adj. p < 0.05) B) GO enrichment analysis of DEGs. C) KEGG pathway enrichment analysis of DEGs. D) Venn diagram showing the overlap between DEGs and mitophagy-related genes (MRGs). E) Heatmap of normalized expression patterns for genes overlapping between DEGs and MRGs. F) Bar plot of representative DEGs associated with the PINK1/Parkin signaling pathway. Statistical analysis was performed using Student's t-test. ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 vs. the PBS group (n = 3).

In addition, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis indicated that following treatment with PDT, the DEGs were closely associated with mitophagy, the TNF signaling pathway, and the Toll-like receptor signaling pathway (Fig. 3C). Among all DEGs, genes linked to mitophagy were screened and 45 mitophagy-related DEGs were obtained by intersecting gene sets using a Venn diagram (Fig. 3D). Their expression patterns were then visualized as a heatmap, including key genes, such as PINK1 (encoding PINK1), PRKN (encoding Parkin), SQSTM1 (encoding p62), and MAP1LC3B (encoding LC3B) (Fig. 3E). These genes were significantly enriched in the PINK1/Parkin-mediated mitophagy pathway. This pathway plays a critical role in the oxidative damage response by coordinating the recognition and clearance of damaged mitochondria to maintain mitochondrial quality control (Fig. 3F).

In summary, at the transcriptomic level, RNA-seq data indicated that compared to the control group, TPSM@IT-4Cl + L led to more severe oxidative damage after PDT and was accompanied by activation of mitophagy-related pathways. Mitophagy is an important defensive mechanism through which tumor cells mitigate oxidative stress and maintain mitochondrial homeostasis. Mitophagy was also a key process for targeted intervention in this study. Therefore, the inhibitory effects of TPSM@IT-4Cl + L on mitophagy and autophagic flux still need to be systematically validated in subsequent in vitro experiments, such as analyses of the expression of key proteins, colocalization studies, and biological transmission electron microscopy (Bio-TEM).

2.4. Functional evaluation of mitochondrial damage and mitophagy inhibition by TPSM@IT-4Cl

Mitochondria-targeted PDT can directly damage mitochondria [52]. We used a JC-1 assay kit to measure changes in MMP, as a decreased MMP typically reflects mitochondrial dysfunction and membrane depolarization. In JC-1 stainingafter light irradiation, both the TPS@IT-4Cl + L and TPSM@IT-4Cl + L groups exhibited pronounced MMP depolarization, manifested by decreased red aggregated fluorescence and increased green monomer fluorescence (Fig. 4A and B). Notably, the TPSM@IT-4Cl + L group showed a pronounced shift toward JC-1 green fluorescence with a concomitant reduction in red fluorescence, consistent with substantial MMP depolarization and severe mitochondrial impairment. Consistently, FCM showed that this group had the highest green/red fluorescence ratio, confirming that combination therapy significantly exacerbated mitochondrial damage and enhanced PDT-mediated cytotoxicity (Fig. 4C and D).

Fig. 4.

TPSM@IT-4Cl + L damages mitochondria and suppresses mitophagy, thereby amplifying oxidative injury. A) JC-1 fluorescence imaging showing changes in MMP in HOS cells after different treatments. B) Green/red JC-1 signal ratios derived from the imaging data, used as an index of mitochondrial depolarization. C) FCM analysis of JC-1 staining showing MMP changes in HOS cells after different treatments. D) Green/red JC-1 ratios calculated from the flow-cytometry results. E) Western blot analysis of mitophagy-related proteins (PINK1, Parkin, p62, and LC3B) after different treatments. F) Densitometric quantification of mitophagy-related proteins normalized to GAPDH. G) MitoTracker-labeled mitochondria and LC3B immunofluorescence staining showing their co-localization to evaluate mitophagy after different treatments. H) Quantification of the relative fluorescence intensity of LC3B. I) Bio-TEM images showing mitochondrial ultrastructural alterations in HOS cells after different treatments; red arrows indicate mitophagosome formation. J) Schematic illustration by Figdraw depicting that TPSM@IT-4Cl + L induces mitochondrial damage and inhibits mitophagy, thereby amplifying oxidative injury. Data are presented as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA. ns, not significant; ∗∗∗∗p < 0.0001 vs. the PBS group.

Mitophagy represents a protective mechanism through which cells eliminate damaged mitochondria following oxidative damage [53,54]. It typically reduces the ROS burden, prevents the accumulation of oxidative damage, and limits the injury. Therefore, when administering anti-tumor treatments, such as PDT, mitophagy may undermine therapeutic efficacy by removing damaged mitochondria. Mitophagy is a selective form of autophagy regulated by multiple signaling axes, among which the PINK1/Parkin pathway represents the canonical mechanism for the clearance of damaged mitochondria [55,56]. Our earlier transcriptome profiling indicated a pronounced enrichment of PINK1/Parkin-associated signatures in the TPSM@IT-4Cl group relative to PBS. Therefore, using Western blotting, we validated mitophagy and its key related proteins, including PINK1, Parkin, p62, and LC3B.

Western blot analysis revealed markedly elevated PINK1 and Parkin expression in the PDT-treated groups (TPS@IT-4Cl + L and TPSM@IT-4Cl + L), suggesting that PDT-induced mitochondrial injury activated PINK1/Parkin-dependent mitophagy (Fig. 4E). LC3B-II and p62 are also closely associated with mitophagy. An increased LC3B-II/I ratio is a classical molecular indicator of enhanced autophagosome formation, whereas p62, as a selective autophagy receptor, is negatively correlated with mitophagy [57,58]. Quantitative analysis of Western blotting data showed that compared to the TPS@IT-4Cl + L group, the TPSM@IT-4Cl + L group exhibited a significantly higher relative expression level of p62 (p < 0.0001) and a significantly decreased LC3B-II/I ratio (p < 0.0001) (Fig. 4F). To further evaluate mitochondria-specific changes, we examined Tom20, a mitochondrial outer membrane protein commonly used as a marker of mitochondrial mass. Changes in Tom20 expression can partly reflect the clearance or retention of damaged mitochondria. Compared with the TPS@IT-4Cl + L group, Tom20 expression was relatively increased in the TPSM@IT-4Cl + L group (p < 0.01), suggesting that Mdivi-1-mediated inhibition of mitophagy-associated clearance may impair the removal of PDT-damaged mitochondria and promote their intracellular retention (Fig. S10A and B).

Together, the accumulation of p62, the decreased LC3B-II/I ratio, and the relatively preserved Tom20 level suggest that TPSM@IT-4Cl + L impairs mitophagy-associated mitochondrial clearance in the context of PDT-induced mitochondrial damage, thereby promoting the retention of damaged mitochondria and amplifying oxidative injury.

To confirm that the observed autophagy was mitophagy, we conducted immunofluorescence staining of LC3B (green) and labeled mitochondria with the fluorescent probe Mito-Tracker (red), thereby assessing their colocalization. Marked colocalization (yellow) was observed in both PDT-related treatment groups (Fig. 4G and H). However, compared to the TPS@IT-4Cl + L group, the TPSM@IT-4Cl + L group exhibited significantly reduced colocalization, and LC3B expression was markedly decreased (p < 0.0001). These results suggest that mitochondria-targeted PDT robustly activates mitophagy, whereas Mdivi-1 significantly suppresses mitophagy.

To validate the occurrence of mitophagy at the ultrastructural level, HOS cells were examined by Bio-TEM. In the TPS@IT-4Cl + L group, multiple mitochondria were encapsulated by double-membrane structures, forming typical mitophagosome-like structures (red arrows) (Fig. 4I; Fig. S10C). This finding indicates that PDT-induced mitochondrial damage initiated mitophagy-mediated clearance. Meanwhile, some membranous vesicles exhibited autolysosome-like changes, with heterogeneous electron density and blurred structures. This may represent residual morphologies when the engulfed mitochondria entered the degradation stage. In contrast, these mitochondria-encapsulating structures were markedly fewer in the TPSM@IT-4Cl + L group. Although more damaged mitochondria accumulated in the cytoplasm in this group, characterized by swelling and indistinct or disrupted cristae (Fig. S10D). Consistent with the results of Western blotting and immunofluorescence, Mdivi-1 significantly inhibited mitophagy after PDT, thereby promoting the sustained accumulation of PDT-induced mitochondrial damage and oxidative stress (Fig. 4J). Collectively, the results of Bio-TEM provide morphological evidence supporting the synergistic enhancement of therapeutic efficacy.

Collectively, mitochondria-targeted PDT induces mitochondrial dysfunction and activates PINK1/Parkin-associated mitophagy, which may help clear damaged mitochondria and limit oxidative stress. In the TPSM@IT-4Cl + L group, Mdivi-1 may interfere with mitochondrial fission and mitophagy-related clearance, leading to persistent accumulation of damaged mitochondria. Considering Mdivi-1's non-specific effects on mitochondrial dynamics and metabolism, its role should be seen as a broader modulation of mitochondrial damage responses rather than selective mitophagy inhibition.

2.5. TPSM@IT-4Cl-induced ICD and immune cell modulation in vitro

Accumulating evidence indicates that PDT enhances antitumor immunity by initiating ICD. ICD is commonly characterized by surface translocation of calreticulin (CRT) together with extracellular release of high mobility group box 1 (HMGB1) following its nuclear mobilization [59,60]. To evaluate the ICD-inducing capacity of different treatments, we first assessed CRT exposure and HMGB1 release. Immunofluorescence analysis showed that CRT exposure was significantly more pronounced in the TPSM@IT-4Cl + L group than in the TPS@IT-4Cl + L group (p < 0.0001) (Fig. 5A and C). HMGB1 is primarily localized in the nucleus in resting conditions, whereas it can translocate to the cytoplasm or the extracellular space during ICD. Immunofluorescence staining showed that TPS@IT-4Cl + L induced moderate extranuclear translocation of HMGB1 (Fig. 5B and D). In contrast, TPSM@IT-4Cl + L more strongly induced HMGB1 release, evidenced by a markedly reduced nuclear HMGB1 signal (p < 0.0001). HMGB1 levels in the cell culture supernatant were also quantified using enzyme-linked immunosorbent assay (ELISA) (Fig. S11). TPSM@IT-4Cl + L group produced the greatest HMGB1 secretion, approximately doubling the level observed for TPS@IT-4Cl + L group (p < 0.0001). Western blotting showed that the relative protein levels of CRT and HMGB1 in the TPSM@IT-4Cl + L group were 1.62-fold and 0.32-fold those of the TPS@IT-4Cl + L group, respectively (Fig. 5E–G).

Fig. 5.

TPSM@IT-4Cl + L induces ICD in vitro and promotes BMDC maturation. A) CLSM images showing CRT exposure in HOS cells after different treatments. B) CLSM images showing HMGB1 release in HOS cells after different treatments. C) Quantification of the relative fluorescence intensity of CRT. D) Quantification of the relative fluorescence intensity of HMGB1. E) Western blot analysis of CRT and HMGB1 expression in HOS cells after different treatments. F) Densitometric quantification of CRT normalized to GAPDH. G) Densitometric quantification of HMGB1 normalized to GAPDH. H) Schematic illustration of the co-culture of differently treated K7M2 cells with BMDCs by Figdraw. I) Quantification of BMDC maturation after co-culture with differently treated K7M2 cells. J) Representative FCM plots showing BMDC (CD11c⁺ CD80⁺ CD86⁺) maturation induced by differently treated K7M2 cells. Data are presented as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA. ns, not significant; ∗∗∗∗p < 0.0001 vs. the PBS group.

To further compare the ICD-inducing capacity of the TPSM@IT-4Cl co-delivery nanoplatform with that of a simple free drug combination, we examined CRT exposure and HMGB1 release after treatment with IT-4Cl + L, Mdivi-1 + IT-4Cl + L, and TPSM@IT-4Cl + L. Compared with the PBS group, all three treatment groups markedly increased CRT exposure and HMGB1 release (p < 0.0001). Notably, compared with the Mdivi-1 + IT-4Cl + L group, TPSM@IT-4Cl + L increased CRT fluorescence intensity to 1.45-fold and reduced nuclear HMGB1 fluorescence intensity to 0.65-fold (Fig. S12A–D), indicating stronger CRT exposure and HMGB1 release. These results suggest that TPSM@IT-4Cl has a stronger ICD-inducing capacity than the simple combination of free IT-4Cl and Mdivi-1.

In this study, the enhanced therapeutic and ICD-inducing effects observed in the TPSM@IT-4Cl + L group may be associated with impaired PINK1/Parkin-mediated mitophagy-associated clearance triggered by damaged mitochondria; however, potential effects of Mdivi-1 on mitochondrial dynamics and metabolic stress cannot be completely excluded. To further evaluate the role of PINK1/Parkin-mediated mitophagy in this process, we knocked down PRKN in HOS cells using siRNA to establish a si-PRKN OS cell model. Western blot analysis showed that Parkin expression was markedly reduced in the si-PRKN-2 group compared with the si-NC group, confirming successful PRKN knockdown (Fig. S13A). ROS analysis showed that, compared with si-NC cells, TPS@IT-4Cl + L induced higher ROS levels in the si-PRKN background (p < 0.0001) (Fig. S13B and C). In addition, LC3B fluorescence intensity in the TPS@IT-4Cl + L group decreased from 99.7% under the si-NC condition to 54.3% under the si-PRKN condition (p < 0.0001), suggesting that PRKN knockdown attenuated PINK1/Parkin-related mitophagy (Fig. S13D and G). Meanwhile, PDT-induced ICD was further enhanced in the si-PRKN background, as indicated by increased CRT exposure and enhanced HMGB1 release (Figs. S13E, F, H, I). These results suggest that inhibition of Parkin-mediated mitophagy can partially mimic the enhancing effect of Mdivi-1 on PDT, supporting the role of PINK1/Parkin-mediated mitophagy as a cytoprotective response after PDT-induced mitochondrial damage.

Collectively, these results indicate that TPSM@IT-4Cl + L amplifies PDT-induced oxidative stress and ICD more effectively than PDT alone or the simple free drug combination. The PRKN knockdown results further support that PINK1/Parkin-mediated mitophagy acts as a cytoprotective response after PDT-induced mitochondrial damage, and that interference with this process contributes to enhanced ICD induction.

Given that dendritic cells (DCs) are professional antigen-presenting cells and that DC maturation is essential for T cell–driven antitumor immunity, we examined whether ICD could facilitate DC maturation [61,62]. Bone marrow-derived DCs (BMDCs) were generated from C57BL/6 mouse bone marrow and subsequently co-incubated with K7M2 cells that had been preconditioned under the indicated regimens. The proportion of mature BMDCs (CD11c⁺ CD80⁺ CD86⁺) was analyzed using FCM (Fig. 5H). Compared to the PBS group, the PDT-related treatment groups exhibited a significant increase in BMDC maturation (Fig. 5I and J). Notably, the TPSM@IT-4Cl + L group reached 31.7% BMDC maturation, which was significantly higher than 17.7% BMDC maturation observed in the TPS@IT-4Cl + L group (p < 0.0001). ELISA analysis of BMDC culture supernatants revealed significantly elevated TNF-α and IFN-γ levels in the TPSM@IT-4Cl + L group compared with the control group (p < 0.0001) (Fig. S14).

Since macrophages are also an important component of the OS immune microenvironment, we further examined whether tumor cells subjected to different treatments could influence macrophage phenotypes. RAW264.7 cells were co-cultured with K7M2 cells pretreated under different conditions, and macrophage phenotypic changes were assessed by immunofluorescence staining. The results showed that, compared with the TPS@IT-4Cl + L group, the TPSM@IT-4Cl + L group induced the highest iNOS/CD206 fluorescence intensity ratio in macrophages, reaching 1.62 times that of the TPS@IT-4Cl + L group, suggesting that this treatment more effectively promoted a shift toward a pro-inflammatory M1-like phenotype (Fig. S15A and B). Collectively, these results indicate that PDT can effectively induce ICD in OS cells, and Mdivi-1-mediated inhibition of mitophagy can potentiate PDT-induced oxidative damage, thereby enhancing ICD, promoting BMDC maturation, and inducing macrophage phenotypic changes associated with a more immunostimulatory state.

2.6. In vivo biodistribution and safety evaluation of TPSM@IT-4Cl

Effective penetration and sustained retention of nanomedicines at the tumor site are important prerequisites for their antitumor activity [63]. To investigate the in vivo delivery performance of TPSM@IT-4Cl and its accumulation capacity in OS, we established a K7M2 tibial orthotopic tumor-bearing mouse model and intravenously injected each mouse with the nanoparticles at a dose equivalent to 2.5 mg/kg Mdivi-1 for in vivo fluorescence imaging-based tracking (Fig. 6A) [64]. The results showed that the fluorescence signal in the tumor region gradually increased after administration and reached its maximum at 24 h (Fig. 6B and C). Then, the signal declined but remained detectable at 48 h, indicating relatively prolonged retention of the nanoparticles at the tumor site. Ex vivo fluorescence imaging of organs at 48 h also confirmed a strong signal in tumor tissues, demonstrating effective nanoparticle accumulation in the tumor (Fig. 6D). In addition to the tumor, fluorescence was mainly distributed in the liver, suggesting that hepatic metabolism/clearance is the primary route of elimination. A secondary signal was also detected in the spleen, which may be associated with the uptake and retention of nanoparticles by the mononuclear phagocyte system (MPS). Overall, TPSM@IT-4Cl exhibited favorable lesion accumulation and extended retention in the orthotopic model of OS, providing a solid delivery basis for subsequent in vivo studies.

Fig. 6.

In vivo biodistribution of TPSM@IT-4Cl and its antitumor efficacy in an orthotopic OS model. A) Schematic illustration of orthotopic K7M2 OS model establishment and the treatment regimen, created with BioGDP.com [65]. B)In vivo fluorescence imaging at various time points following intravenous administration of TPSM@IT-4Cl. C) Quantification of fluorescence signals in the tumor region at different time points after injection. D)Ex vivo fluorescence imaging of major organs and tumors: heart (H), liver (Li), spleen (S), lung (L), kidney (K), intestines(I), and tumor (T). E) Body weight changes of mice in different treatment groups during the treatment period. F) Tumor volume changes of mice in different treatment groups during the treatment period. G) Tumor weights of mice in different treatment groups. H) Representative photographs of orthotopic tumors from mice in different treatment groups. I) H&E staining of tumor tissues from different treatment groups. J) TUNEL staining of tumor tissues from different treatment groups. Data are presented as mean ± SD (n = 5). Statistical significance was determined by one-way ANOVA. ns, not significant; ∗∗∗∗p < 0.0001 vs. the PBS group.

To evaluate the in vivo circulation behavior of the nanoparticles, we measured the fluorescence signal in mouse serum at different time points after administration and calculated the apparent circulation half-life of TPSM@IT-4Cl based on the serum fluorescence intensity–time curve. The results showed that the serum fluorescence signal reached its peak at 10 min post-injection, then gradually decreased and became relatively stable after 12 h (Fig. S16A and B). Based on the fluorescence decay curve, the fluorescence-derived apparent circulation half-life of TPSM@IT-4Cl was calculated to be 7.55 h.

We conducted a hemolysis assay to assess hemocompatibility and evaluate the systemic safety of this nanoparticle after intravenous administration and its interactions with blood components. The negative control (PBS) induced almost no hemolysis, whereas the positive control (distilled water) led to complete hemolysis. The supernatants remained largely clear after treatment with the nanoparticle at concentrations ranging from 2 to 40 μg/mL(Fig. S17A). The slight color change observed in the high-concentration groups was mainly attributable to the intrinsic color of the material rather than hemolysis (Fig. S17B). Quantitative analysis showed that the hemolysis rate only slightly increased with increasing concentrations. Notably, the hemolysis rate remained <5% across all groups, indicating good hemocompatibility of this nanoparticle (Figure S17C).

Regarding systemic toxicity, under 735 nm laser irradiation at 0.1 W/cm² for 10 min, all mice exhibited good general condition during the treatment period, with no obvious adverse reactions. Furthermore, the body weight curves showed no downward trend. (Fig. 6E). At the study endpoint, the heart, liver, spleen, lungs, and kidneys were collected and subjected to histopathological analysis (Fig. S18). H&E staining revealed no apparent structural damage, necrotic lesions, or inflammatory infiltration. Together with the results of hemocompatibility, these findings collectively suggest that TPSM@IT-4Cl exhibits favorable tolerance and biosafety in vivo under the applied dose and administration regimen, supporting its application as a nanoparticle for anti-tumor therapy.

2.7. In vivo anti-tumor efficacy of TPSM@IT-4Cl

We established a tibial orthotopic mouse model of OS following the same protocol described above to evaluate the in vivo anti-tumor efficacy of the nanoparticle (Fig. 6A). The dose and route of administration of the nanoparticles were the same as described above. The irradiation condition was set as 735 nm laser irradiation at 0.1 W/cm² for 10 min. To further determine the optimal irradiation time point for ROS generation after nanoparticle administration, we examined intratumoral ROS generation after laser irradiation at different time points after injection, including 1, 2, 4, 12, 24, and 48 h. At earlier time points, ROS generation in tumor tissues was relatively low, likely due to limited nanoparticle accumulation at the tumor site. As TPSM@IT-4Cl gradually accumulated in the tumor region, intratumoral ROS generation increased and reached a relatively high level at 24 h post-injection (Fig. S19). This trend was generally consistent with the fluorescence-based tumor accumulation profile. These results suggest that 24 h post-injection provides both favorable tumor accumulation and strong ROS-generating capability, supporting its selection as the irradiation time point for subsequent in vivo therapeutic experiments.

Based on this optimized irradiation schedule, we next evaluated the in vivo antitumor efficacy of TPSM@IT-4Cl + L in the orthotopic OS model. Tumor growth inhibition curves showed that the tumor inhibition rate was 51.32% in the TPS@IT-4Cl + L group, whereas 75.03% in the TPSM@IT-4Cl + L group, indicating superior therapeutic efficacy in the combined treatment group (Fig. 6F). Mice were sacrificed on day 25, and tumors were excised and weighed (Fig. 6G). The results showed that the average tumor weight was 0.84 ± 0.07 g in the TPSM@IT-4Cl + L group, which was significantly lower than that in the PBS (2.41 ± 0.26 g) and TPS@IT-4Cl + L (1.40 ± 0.17 g) groups. Representative images of excised tumors at the endpoint indicated a substantial reduction in tumor burden in the TPSM@IT-4Cl + L group (Fig. 6H). Collectively, these results suggest that TPSM@IT-4Cl + L possesses significant tumor-suppressive activity in vivo.

H&E staining was conducted on tumor tissues harvested at the endpoint to histologically validate the therapeutic response (Fig. 6I). In the control group, tumor cells were densely packed, exhibited hyperchromatic nuclei, and showed pronounced atypia. In contrast, both the TPS@IT-4Cl + L and TPSM@IT-4Cl + L groups displayed more extensive disruption of the tissue architecture and larger necrotic areas, characterized by reduced cellular density, blurred cellular contours, and nuclear pyknosis/karyorrhexis. Notably, compared to the TPS@IT-4Cl + L group, the TPSM@IT-4Cl + L group showed an expanded necrotic area and more severe necrosis, suggesting that combination therapy more effectively destroys tumor tissues.

Next, TUNEL staining was conducted to evaluate apoptosis in tumor tissues (Fig. 6J). Compared to the control group, PDT-related treatment groups showed markedly enhanced TUNEL-positive signals. Notably, the TPSM@IT-4Cl + L group exhibited the strongest TUNEL-positive staining, suggesting that this nanoparticle effectively promotes tumor cell apoptosis and enhances anti-tumor efficacy in vivo.

The results of in vivo immunostaining for mitophagy- and ICD-related markers were consistent with in vitro findings (Fig. S20). Compared to the TPS@IT-4Cl + L group, the TPSM@IT-4Cl + L group showed markedly decreased fluorescence intensities of LC3B and HMGB1, along with significantly increased fluorescence intensities of p62 and CRT. This finding suggests that after PDT-induced mitochondrial damage, co-delivery of Mdivi-1 effectively blocked mitophagy and enhanced the release of immunogenic molecules, thereby inducing stronger ICD. The reductions in tumor vol/wt and enhanced histological necrosis and apoptosis collectively suggest that through mitophagy inhibition and ICD amplification, this nanoparticle promoted in vivo anti-tumor benefits in an orthotopic model of OS.

2.8. In vivo immune activation analysis of TPSM@IT-4Cl

As shown in in vitro studies, TPSM@IT-4Cl + L effectively induced ICD, promoted the release of immunogenic molecules, and facilitated BMDC maturation, thereby enhancing anti-tumor immune responses. Building on these findings, we systematically analyzed immune cell compositions in tumor tissues, tumor-draining lymph nodes (TDLNs), and spleens in the immunocompetent K7M2 orthotopic OS model to elucidate how TPSM@IT-4Cl + L remodels the tumor immune microenvironment. First, the maturation status of DCs in TDLNs (CD11c⁺ CD80⁺ CD86⁺) was evaluated using FCM. The results showed that the proportion of mature DCs was 18.5%–21.8% in the PBS and non-irradiated groups (Fig. 7A and D). In contrast, the proportion of mature DCs was 30.6% in the TPS@IT-4Cl + L group, which was significantly higher than that in the control group (p < 0.05). Notably, the proportion of mature DCs was 39.6% in the TPSM@IT-4Cl + L group, which was markedly higher than that in the TPS@IT-4Cl + L group (p < 0.01), indicating that combination treatment further promoted antigen presentation-associated immune priming and enhanced subsequent tumor-specific immune responses.

Fig. 7.

In vivo immune activation induced by TPSM@IT-4Cl + L in the orthotopic OS mouse model. A) FCM analysis of mature DCs (CD11c⁺ CD80⁺ CD86⁺) in lymph nodes from mice receiving different treatments. B) FCM analysis of activated CD8⁺ T cells (CD3⁺ CD8⁺ CD69⁺) in the spleen from mice receiving different treatments. C) FCM analysis of CD8⁺ T cells (CD45⁺ CD3⁺ CD8⁺) in tumor tissues from mice receiving different treatments. D) Percentage of mature DCs in lymph nodes. E) Percentage of activated CD8⁺ T cells in the spleen. F) Percentage of CD8⁺ T cells in tumor tissues. G) Ratio of M1/M2 macrophages in tumor tissues. H) Immunofluorescence images of CD8 staining in tumor sections. I) Schematic illustration of PDT-triggered and amplified antitumor immune responses by Figdraw. Data are presented as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA. ns, not significant; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 vs. the PBS group.

Antigen uptake and presentation mediated by DCs are key processes driving T-cell activation and effector differentiation. Given that the spleen, as an important secondary lymphoid organ, serves as a central site for naïve T-cell activation, clonal expansion, and effector T-cell generation, we analyzed the activation status of splenic CD8⁺ T cells [66]. FCM showed that after treatment with TPSM@IT-4Cl + L, the proportion of activated CD8⁺ T cells (CD3⁺ CD8⁺ CD69⁺) was significantly increased in the spleen, indicating that the nanoparticle effectively promoted CD8⁺ T cell activation and functional priming at the systemic immune level (Fig. 7B and E).

We also quantitatively analyzed CD8⁺ T cell infiltration in tumor tissues to evaluate the regulatory effects of the treatment on the local tumor immune microenvironment. The results showed that after treatment with TPSM@IT-4Cl + L, the proportion of CD8⁺ T cells (CD45⁺ CD3⁺ CD8⁺) increased to 31.9%, which was markedly higher than that in the TPS@IT-4Cl + L group (23.3%) (p < 0.0001) (Fig. 7C and F). These findings suggest that this therapeutic strategy effectively attenuated the ITME and promoted effector T cell recruitment to tumor lesions.

Since tumor-associated macrophages (TAMs) play dual “pro-/anti-tumor” roles during tumor progression, we investigated whether this nanoparticle can drive TAM repolarization from an immunosuppressive M2-like phenotype toward a pro-inflammatory M1-like phenotype. FCM-based phenotyping of intratumoral macrophages showed that compared to TPS@IT-4Cl + L, TPSM@IT-4Cl + L increased the proportion of M1 macrophages (CD45⁺ F4/80⁺ CD80⁺) from 27.0% to 35.8% (p < 0.0001) and concomitantly decreased the proportion of M2 macrophages (CD45⁺ F4/80⁺ CD206⁺) from 5.83% to 3.42% (p < 0.05) (Fig. S21A-D). Moreover, the M1/M2 macrophage ratio increased significantly (p < 0.0001), suggesting that the nanoparticles mitigated ITME and skewed it toward an antitumor phenotype (Fig. 7G).

Meanwhile, after treatment with TPSM@IT-4Cl + L, the serum levels of IL-6, TNF-α, CXCL10, and IFN-γ showed the greatest increase and were significantly higher than those in the other groups (p < 0.0001) (Fig. S22A-D). Immunostaining of tumor tissues showed markedly enhanced fluorescence of CD8⁺ T cells, which was consistent with the results of FCM (Fig. 7H).

In summary, in the immunocompetent K7M2 orthotopic OS model, PDT induced ICD in tumor cells, thereby promoting the release and exposure of damage-associated molecular patterns (DAMPs). These DAMPs subsequently drive DC maturation and promote a phenotypic shift of TAMs toward an M1-like profile. Ultimately, these processes promote CD8⁺ T cell differentiation and enhance their infiltration into tumors, synergistically amplifying antitumor immune responses at both local and systemic levels (Fig. 7I).

2.9. Evaluation of prolonged survival in tumor-bearing Nude mice

We established a patient-derived xenograft (PDX) model to assess the in vivo therapeutic benefit of TPSM@IT-4Cl + L and validate its anti-tumor efficacy in a clinically relevant human tumor context. The relevant information regarding the PDX model is summarized in Table S2. The experimental workflow is illustrated in Fig. 8A. The tumor growth curves showed that the TPSM@IT-4Cl + L group consistently exhibited the strongest inhibition of tumor growth throughout the treatment period (Fig. 8B and C). By day 48, the mean tumor volume was only 813.8 mm³ in this group. Survival analysis indicated that the TPSM@IT-4Cl + L group achieved a 100% survival rate at the end of follow-up, which was clearly superior to the other groups (Fig. 8D). Mice in the PBS, free Mdivi-1, TPSM, and TPSM@IT-4Cl groups all reached humane endpoints within 4 weeks. The TPS@IT-4Cl + L group also reached humane endpoints within 5-6 weeks. Collectively, the results indicated that TPSM@IT-4Cl + L markedly delayed tumor progression and conferred a clear survival benefit in the highly clinically relevant PDX model, suggesting outstanding in vivo therapeutic advantages.

Fig. 8.

Antitumor efficacy of TPSM@IT-4Cl + L in a PDX model. A) Schematic illustration of PDX establishment and the treatment regimen, created with BioGDP.com [65]. B) Tumor volume growth curves of tumor-bearing mice in each group. C) Tumor volume changes of all tumor-bearing mice. D) Kaplan–Meier (KM) survival curves of different treatment groups. E) H&E staining of tumor tissues from different treatment groups. F) Ki67 immunohistochemical staining of tumor tissues from different treatment groups. G) TUNEL staining of tumor tissues from different treatment groups. (n = 5).

Next, tumor tissues were harvested for pathological evaluation when mice reached humane endpoints. Based on H&E staining, the TPSM@IT-4Cl + L group exhibited more extensive necrosis and disruption of tissue architecture (Fig. 8E). Ki67 staining indicated reduced proliferative activity of tumor cells, and TUNEL staining revealed increased apoptosis (Fig. 8F and G). These histopathological changes were consistent with the results of the survival curve, suggesting that TPSM@IT-4Cl + L confers a significant in vivo anti-tumor therapeutic advantage and provides a strong evidence base for advancing subsequent preclinical and translational studies.

3. Discussion

OS remains challenging to treat due to limited drug delivery, therapy resistance, and poor responsiveness to immunotherapy. Although OS is generally considered immune-cold, with limited T-cell infiltration and immunosuppressive microenvironments, enhancing tumor immunogenicity and modulating immune responses remain feasible strategies [67].

Here, we developed a mitochondria-targeted, GSH-responsive photodynamic nanoplatform, TPSM@IT-4Cl, to coordinate mitochondrial PDT with interference of damaged mitochondria clearance. IT-4Cl, with a typical A-D-A conjugated structure, enables efficient ROS generation under near-infrared irradiation, while the TPP⁺ moiety drives mitochondrial enrichment of the nanoparticles through its lipophilic cationic property and mitochondrial membrane potential-dependent accumulation. TPSM@IT-4Cl carries a moderately positive surface charge, but it maintained good colloidal stability in PBS and serum-containing medium and showed no obvious basal cytotoxicity in the absence of light irradiation. Previous studies have also shown that positively charged nanoparticles may enhance cell membrane adhesion and endocytosis [68]. Together, these findings support that this level of surface charge did not markedly compromise formulation stability or biosafety. Meanwhile, the disulfide linkage further endows the system with GSH-responsive release capability, allowing it to remain relatively stable under low-GSH conditions while promoting drug release in the intracellular high-GSH environment.

Self-assembly loading preserves IT-4Cl's photophysical properties and allows synchronized co-delivery with Mdivi-1. Free drug controls showed enhanced cytotoxicity and ICD induction with IT-4Cl + Mdivi-1 + L compared to IT-4Cl + L alone, but TPSM@IT-4Cl + L produced stronger effects, indicating platform-mediated co-delivery, mitochondrial targeting, and GSH-responsive release contribute significantly beyond simple drug combination.

Mechanistically, TPSM@IT-4Cl + L induces mitochondrial membrane potential collapse, mtROS accumulation, and structural damage. PDT-activated PINK1/Parkin-mediated mitophagy provides cytoprotection by clearing damaged mitochondria. Mdivi-1 may interfere with mitophagy-associated clearance, prolonging damaged mitochondria accumulation and enhancing oxidative stress and ICD. PRKN knockdown further confirms the contribution of Parkin-mediated clearance to PDT-induced cytoprotective responses. Notably, Mdivi-1 is not a selective mitophagy inhibitor and may affect mitochondrial dynamics, metabolism, and other stress pathways; its role here is better interpreted as modulating mitochondrial damage responses and clearance rather than specific mitophagy inhibition.

TPSM@IT-4Cl + L effectively induced ICD, as evidenced by enhanced CRT exposure and HMGB1 release, and further promoted BMDC maturation and an M1-like phenotypic shift in RAW264.7 macrophages. In the immunocompetent K7M2 orthotopic OS model, TPSM@IT-4Cl + L inhibited tumor growth, induced tumor cell apoptosis, and was accompanied by changes in immune cell composition, including increased DC maturation, enhanced CD8⁺ T-cell activation and tumor infiltration, and altered macrophage phenotypes. These findings suggest that TPSM@IT-4Cl + L may contribute to antitumor efficacy through ICD induction and immune microenvironment modulation. Further studies involving CD8⁺ T-cell depletion, tumor-specific T-cell response assays, adoptive transfer experiments, and functional evaluation of macrophages, such as cytokine secretion, and antigen presentation capacity, are still needed to clarify the direct contribution of immune cell function to this therapeutic process [69].

In PDX models, TPSM@IT-4Cl + L inhibited tumor growth and altered Ki67 and TUNEL markers, reflecting direct tumor cell killing via mitochondrial PDT and Mdivi-1-mediated interference. Immune conclusions are derived from the immunocompetent K7M2 model, while PDX serves as complementary evidence for human OS tissue efficacy.

Limitations should also be noted. Although the 735 nm laser used in this study lies within the NIR region and has better tissue penetration than visible light—and has been applied in previous PDT studies for OS [70]—attenuation by dense bone and surrounding muscle may still limit ROS generation in deep-seated OS lesions [71]. In addition, our biodistribution and circulation analyses relied mainly on fluorescence tracing, which provides apparent rather than absolute pharmacokinetic parameters. Future translation will require optimized light delivery strategies and validated HPLC/LC-MS-based pharmacokinetic studies.

In summary, TPSM@IT-4Cl integrates mitochondrial targeting, GSH-responsive release, PDT-induced oxidative damage, and modulation of damaged mitochondria clearance, enhancing oxidative stress and ICD, and inducing immune-associated antitumor responses. While clinical translation requires further optimization, this study provides a nanoplatform strategy to enhance immunogenicity and therapeutic vulnerability in OS.

4. Conclusion

In conclusion, TPSM@IT-4Cl integrates mitochondria-targeted PDT with interference of damaged mitochondria clearance. Through GSH-responsive release of IT-4Cl and Mdivi-1, this nanoplatform induces mtROS generation under light irradiation and disrupts the protective clearance of PDT-damaged mitochondria, thereby amplifying oxidative injury and ICD. In vivo studies showed that TPSM@IT-4Cl exhibited potent antitumor activity, induced immune-associated antitumor responses, and showed favorable biosafety. Overall, this study supports TPSM@IT-4Cl as a mitochondria-targeted photoimmunotherapy nanoplatform with potential applicability in OS treatment.

CRediT authorship contribution statement

Qing Deng: Writing – review & editing, Writing – original draft, Project administration, Methodology, Formal analysis, Data curation. Jinsong Li: Writing – review & editing, Writing – original draft, Supervision, Methodology, Formal analysis. Zhaochen Tong: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis. Lingpu Zhang: Methodology. Yuanyu Tang: Methodology, Formal analysis. Junyan Liu: Methodology, Formal analysis. Dong Wang: Methodology, Formal analysis. Jin Zeng: Methodology, Formal analysis. Zixin Li: Methodology, Formal analysis. Yueqiang Zhang: Formal analysis. Fangmin Wang: Formal analysis. Xuanxuan Li: Formal analysis. Jinrong Zeng: Formal analysis. Yi Peng: Formal analysis. Weiguo Wang: Formal analysis. Jinglei Miao: Formal analysis. Dabao Xu: Writing – review & editing, Writing – original draft, Supervision, Methodology, Data curation. Kun Shang: Writing – review & editing, Writing – original draft, Supervision, Methodology, Data curation. Minhuan Lan: Writing – review & editing, Writing – original draft, Supervision, Methodology, Data curation. Shijie Chen: Writing – review & editing, Writing – original draft, Supervision, Resources, Methodology, Funding acquisition, Data curation, Conceptualization.

Data availability statement

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

Ethics approval and consent to participate

This study involved the use of human osteosarcoma tissue samples obtained from patients undergoing surgical resection. Written informed consent was obtained from all patients or their legal guardians. The use of human tissue for establishing patient-derived xenograft (PDX) models was approved by the Ethics Committee of The Third Xiangya Hospital of Central South University (Approval No.: Kuai 26121). All animal experiments were performed in accordance with the institutional guidelines for animal care and use. No direct human experimentation (e.g., clinical trial or patient intervention) was conducted in this study.

Declaration of competing interest

The authors declare no conflict of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: