Targeting mitochondrial-ER homeostasis via autophagy inhibition with celastrol-based nanotherapy for triple-negative breast cancer

Abstract

The functional crosstalk between mitochondria and the endoplasmic reticulum (ER) serves as a critical adaptive mechanism in cancer cells, wherein mitochondrial damage-induced ER stress can paradoxically activate protective mitophagy to restore cellular homeostasis and limit therapeutic efficacy. To subvert this self-repair cycle and amplify immunogenic cell death (ICD), we engineered a mitochondria-targeted biomimetic nanoplatform (Cel-Ca/CQ@OMM) for tumor-selective co-delivery of Celastrol (Cel) and Chloroquine (CQ). The nanosystem leverages homologous mitochondrial membrane functionalization to achieve precise subcellular localization. Celastrol coordinates with calcium ions to form a complex (Cel-Ca) that induces Ca²⁺ overload and reactive oxygen species (ROS) burst, thereby damaging mitochondria and concomitantly triggering lipophagy as a compensatory survival response. The ER, upon contact with damaged mitochondria, activates mitophagy to clear these organelles. By suppressing autophagic flux, CQ simultaneously abrogates both reparative mitophagy and adaptive lipophagy. The resulting cumulative accumulation of damaged mitochondria and lipid droplets perpetuates ER-mitochondria interaction and imposes metabolic burden on the ER, establishing a vicious cycle that progressively amplifies cellular stress through positive feedback regulation, thereby steering the cell toward apoptotic elimination. This dual-inhibition intervention disrupts mitochondrial-ER homeostasis, leading to exacerbated ER stress, enhanced damage-associated molecular pattern (DAMP) release, and robust CD8⁺ T cell-mediated antitumor immunity. This study highlights the amplification of ICD by synergistically blocking the key adaptive pathways of mitophagy and lipophagy, providing a promising approach for Triple-negative breast cancer (TNBC) treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04769-5.

Introduction

Triple-negative breast cancer (TNBC) is associated with the highest recurrence and mortality rates among breast cancer subtypes [1]. Conventional therapies often yield limited clinical benefits [2, 3], highlighting the urgent need for targeted treatment strategies that offer greater precision and personalization [4].

Mitochondrial-targeted therapy has recently emerged as a promising approach, leveraging the central role of mitochondria in regulating cellular metabolism and programmed cell death [5, 6]. However, the effectiveness of this strategy is intrinsically limited by sophisticated inter-organelle communication networks, particularly the dynamic crosstalk between mitochondria and the ER [7]. These organelles maintain functional connectivity through specialized mitochondria-associated ER membranes (MAMs), which serve as critical platforms for Ca²⁺ exchange, lipid transfer, and metabolic coordination [8–10]. The mitochondrial-ER relationship demonstrates remarkable bidirectionality. While mitochondrial damage initiates ER stress through Ca²⁺ dysregulation and ROS production [11], the ER actively participates in mitochondrial quality control by sensing damage and coordinating mitophagic responses. Through membrane contact sites, the ER facilitates the encapsulation of damaged mitochondria by phagophore membranes, thereby activating protective mitophagy to restore cellular homeostasis [12–14]. This ER-mediated quality control represents a fundamental adaptive mechanism that cancer cells exploit to limit therapy-induced mitochondrial damage. This vicious cycle of mutual amplification culminates in severe cellular stress and commits cells to apoptosis.

Celastrol, a natural bioactive compound, is a potent inducer of mitochondrial dysfunction. It promotes the uptake of Ca²⁺ by mitochondria, leading to Ca²⁺ overload and the opening of the mitochondrial permeability transition pore, which subsequently induces ROS generation and mitochondrial damage [15, 16]. Chloroquine, an autophagy inhibitor, blocks lysosomal degradation and prevents the clearance of damaged mitochondria via mitophagy [17]. It has also been reported that inhibition of mitophagy can enhance endoplasmic reticulum stress, disrupt the mitochondria-ER homeostasis, and further prolong the organelle stress state [18].

Furthermore, beyond organellar quality control, cancer cells employ metabolic autophagy pathways such as lipophagy—the selective degradation of lipid droplets—to maintain lipid homeostasis and energy supply under stress. Inhibition of lipophagy leads to intracellular lipid accumulation, which can induce lipotoxicity, disrupt ER membrane fluidity and function, and thereby provoke ER stress [19]. Notably, Celastrol has been reported to activate both mitophagy and lipophagy as compensatory survival responses [20, 21]. We hypothesize that Chloroquine, by blocking autophagic flux, concurrently inhibits these two critical pathways. The resultant accumulation of both damaged mitochondria and unmetabolized lipids creates a dual assault on cellular homeostasis, synergistically amplifying ER stress beyond the level achievable by targeting either pathway alone.

Despite the promise of mitochondrial-targeted agents, their clinical translation has been hampered by poor bioavailability, off-target toxicity, and limited tumor-specific accumulation [5]. Conventional nanocarriers often lack subcellular precision and fail to replicate natural targeting mechanisms. In contrast, membrane-coated nanoparticles offer distinct advantages: they inherit source-specific homing properties through retained surface proteins, exhibit enhanced biocompatibility [22–24]. Moreover, such biomimetic systems can bypass lysosomal sequestration, augment tumor accumulation through homologous targeting, and potentiate immune activation through native antigen presentation [25]. These biomimetic nanosystems, functionalized with cell membrane coatings, are capable of not only preserving the physicochemical properties of the encapsulated nanomaterials but also inheriting the intrinsic biological or immunological characteristics of the cell membrane [26]. However, the application of the subcellular membrane’s biological functions in drug delivery systems remains underexplored, particularly considering that mitochondria frequently undergo fusion processes [27]. Therefore, we propose the development of a drug delivery system modified with mitochondrial membrane proteins derived from tumor cells, which may enhance the mitochondrial targeting capability of nanoparticles.

In summary, this study reports the synthesis of a Celastrol-Calcium complex (Cel-Ca) through the coordination interaction between Celastrol and Calcium ions. The resulting complex is capable of inducing Calcium overload and a subsequent burst of ROS. It is subsequently loaded with the classical autophagy inhibitor CQ via hydrophobic interactions and π − π stacking [28, 29], and further encapsulated with homologous mitochondrial membranes for targeting the mitochondria of tumor cells, thereby effectively blocking the autophagy pathway, enhancing ER stress, and synergistically inducing ICD (Scheme 1). This design exploits natural mitochondrial homing capacity for precise subcellular localization, concurrently provoking mitochondrial injury and associated metabolic stress that activate both protective mitophagy and compensatory lipophagy. By blocking these dual autophagy-mediated repair pathways, CQ effectively disrupts ER-mitochondria homeostasis and lipid metabolism, leading to synergistically amplified ER stress and ICD (Scheme 1). We demonstrate that this targeted perturbation overcomes cellular repair and adaptive programs, leading to amplified ICD and robust antitumor immunity in TNBC.

Scheme 1.

Manufacture and application of Cel-Ca/CQ@OMM. (A) Preparation of Cel-Ca/CQ@OMM. (B) Application of Cel-Ca/CQ@OMM in TNBC treatment.

Materials and methods

Materials

Celastrol (purity ≥ 98%), chloroquine (purity ≥ 98%), and calcium chloride were obtained from Bido Pharmaceutical (Shanghai, China), Polyvinylpyrrolidone K30 and RhodamineB were provided by Macklin (Shanghai, China). Mitochondria lsolation Kit was obtained from Adamas life (Shanghai, China). Dio, ATP and HMGB1 kit were obtained from Solarbio (Beijing, China). Annexin V-FITC/PI Apoptosis Kit and Bicinchoninic acid (BCA) protein assay kit were gained from Vazyme (Nanjing, China). Indocyanine green (ICG) was procured from MCE (America). The ROS Detection Kit, Mito-Tracker, Lyso-Tracker, Enhanced Mitochondrial Membrane Potential Detection Kit, and Apoptosis Detection Kit were procured from Beyotime (Shanghai, China).

Antibodies: The antibodies used for western blotting (WB), flow cytometry (FCM), and immunofluorescence (IF) are summarized in Tables S1.

Cell lines and Experimental animals: 4T1^Fluc cells were maintained in DMEM medium supplemented with 10% fetal bovine serum under a humidified atmosphere of 5% CO₂ at 37 °C in a cell culture incubator. Experimental mice (female, BALB/c, 6–8 weeks) were obtained from the Animal Experimental Research Center of Zhejiang Chinese Medical University. All experimental procedures strictly complied with the guidelines approved by the Animal Care and Use Committee of Zhejiang Chinese Medical University, in accordance with ethical permit No. IACUC-20230821-09.

Extraction and identification of mitochondrial membrane

To isolate the mitochondrial membrane, we initially isolated the mitochondria. In brief, 4T1 cells were grown to complete confluence, detached using a cell scraper, and subsequently washed three times with 1× PBS (centrifugation at 800 g). The cells were then treated with a mitochondrial extraction reagent and homogenized twenty times on ice throughout the procedure. According to the manufacturer’s instructions for the extraction kit, the precipitate obtained through sequential differential centrifugation was identified as mitochondria. Then the purified mitochondria were subsequently resuspended in PBS (10 mM, pH 7.4) and incubated for 30 min (centrifugation at 105,000 g, 60 min). The supernatant was carefully discarded, and the pellet was resuspended in a sucrose-Tris-HCl buffer solution (centrifugation at 115,000 g, 15 min). The supernatant was removed, and the sample was centrifuged at 105,000 g for 60 min. The resulting supernatant was discarded, and the pellet was resuspended in an appropriate sucrose Tris-HCl buffer solution. Discontinuous sucrose density gradient solutions (25.2%, 37.7% and 51.7%) were sequentially layered in an ultrafiltration tube. And then the sample solution was gently overlaid onto the interface of the 25.2% sucrose solution, followed by ultracentrifugation at 165,000 g for 45 min. Subsequently, the interfacial solution was collected, and its protein concentration was quantified using the BCA method. Mitochondrial membranes were subsequently stored at -20 °C in isotonic solutions supplemented with protease and phosphatase inhibitors.

The particle size and zeta potential of the extracted membrane vesicles were systematically characterized using dynamic light scattering (DLS, Zeta Sizer Nano ZS, Malvern, UK).

For SDS-Page analysis, mitochondrial membranes were extracted and subsequently loaded into each well for electrophoretic separation, followed by harvesting for western blot analysis. For protein visualization, the western blots were stained using Coomassie Blue staining solution. To demonstrate the existence of the mitochondrial membrane, the levels of MFN-2 proteins were quantitatively analyzed by Western blotting.

Preparation and characterization of Cel-Ca/CQ@OMM NPs

Firstly, Celastrol-calcium complex (Cel-Ca) was successfully synthesized by mixing Cel and CaCl₂ solutions at various molar ratios. It was observed that the particle size and dispersity of the complex was optimized under a molar ratio of 2:1. Subsequently, 1.6 mg of CQ and 20 mg of PVP were added to the ethanol solution containing Cel-Ca. The mixture was stirred at 500 rpm for 24 h at room temperature. Thereafter, the reaction solution was transferred to 2 mL of water and stirred for an additional 4 h. Finally, the supernatant was removed by centrifugation at 500 g for 10 min, and the precipitate was collected by centrifugation at 12,000 g for 30 min. The precipitate was then washed, re-centrifuged, dispersed in water, and stored at 4 °C for future use. For membrane coating, the resulting vesicles were integrated with nanoparticles at varying mass ratios. Subsequently, the mixture was co-extruded through a 200 nm polycarbonate membrane and centrifuged to purify for further application. In order to prepare Cel-Ca / Rho, Rho was used instead of CQ, the ratio of Cel-Ca : Rho was 1 : 0.05.

The particle size distribution and zeta potential of the nanoparticles were characterized using dynamic light scattering (DLS). For stability evaluation, the particle size and polydispersity index (PDI) of the nanoparticles were systematically monitored via DLS during incubation at 37 °C in phosphate-buffered saline (PBS) or 10% fetal bovine serum (FBS). Simultaneously, the morphology was systematically characterized using transmission electron microscopy (TEM, SU8010, Hitachi Ltd. Japan). The drug loading yield was quantified using UV spectroscopy (GENESYS 180, Thermo Fisher Technologies, USA) at a wavelength of 425 nm. Drug loading efficiency (DL%) and encapsulation efficiency (EE%) were calculated using the following equation: DL% = (Amount of drug loaded in NPs / Weight of NPs) × 100%, EE% = (Amount of drug loaded in NPs / Total amount of drug input) × 100%. Fourier Transform Infrared (FT-IR) (Nicolet Nexus 470, Thermo Electron) spectroscopy was used to detect Cel, CQ, Cel-Ca/CQ, and the mixtures of Cel and CQ.

In vitro release of Cel from Cel-Ca/CQ@OMM NPs

The release of Cel from Cel-loaded NPs (Cel-Ca/CQ NPs and Cel-Ca/CQ@OMM NPs) in pH 7.4, 6.5 and 5.0 media was examined by dialysis. In brief, 1.0 mL of the Cel-loaded NPs solution (n = 3) was carefully placed into a dialysis bag with a molecular weight cutoff of 5000. The dialysis bag was subsequently immersed in 20 mL of the specified pH release medium and stirred at 37 °C with a speed of 100 rpm. At predetermined time intervals, 2 mL of the release medium was withdrawn for sampling and replenished with an equivalent volume of fresh blank medium. The concentration of Cel was subsequently quantified using UV absorption spectroscopy.

Cellular uptake and subcellular distribution

For cellular uptake analysis, 4T1 cells were incubated with Cel-Ca/Rho and Cel-Ca/Rho@OMM for 4 h under controlled experimental conditions. Subsequently, the 4T1 cells were harvested, and the uptake efficiency was quantitatively analyzed using FCM.

For subcellular distribution, 4T1 cells (5 × 10⁴ cells/well) were seeded onto coverslips and subsequently incubated with Cel-Ca/Rho and Cel-Ca/Rho @OMM for 4 hours under controlled conditions. Subsequently, the cells were washed and stained with Mitotracker Green for 30 min. Following this, the cells were washed again, fixed with 4% paraformaldehyde for 15 min. Finally, fluorescence images were analyzed by super-resolution microscopy. Rho fluorescence was detected at an excitation wavelength of 525 nm and an emission wavelength of 609 nm, while MitoTracker Green fluorescence was detected at an excitation wavelength of 488 nm and an emission wavelength of 561 nm.

Cytotoxic assay

For cytotoxicity evaluation, MCF-10 A cells(6 × 10³ cells/well) were seeded into 96-well plates and cultured overnight under standard conditions. Subsequently, the cells were treated with Cel, Cel-Ca NPs, Cel-Ca/CQ NPs, and Cel-Ca/CQ@OMM NPs at predetermined concentrations for 24 h under controlled experimental conditions. After incubation, cell viability was assessed using a CCK-8 assay according to the manufacturer’s instructions. Briefly, 10 µL of CCK-8 solution was added to each well and incubated for an additional 1–2 h at 37 °C. The absorbance at 450 nm was measured using a microplate reader. Cell viability was calculated as a percentage relative to the untreated control group.

Assessment of mitochondrial dysfunction

ROS production was quantified using 2’, 7’-dichlorofluorescein diacetate (DCFH-DA). 4T1 cells were incubated with Cel, Cel-Ca-NP, Cel-Ca/CQ-NP, and Cel-Ca/CQ@OMM-NP (all at an equivalent Cel concentration of 1.6 µM) for 24 h. Subsequently, the cells were washed twice, and the fluorescence intensity of intracellular ROS was analyzed via fluorescence microscopy (LSM 880, Zeiss, Germany). To investigate mitochondrial membrane potential, 4T1 cells were treated for 24 h with the preparations described above. Subsequently, the cells were harvested and incubated with the JC-1 staining working solution for 30 min at 37 °C. Finally, the cells were washed and analyzed quantitatively using fluorescence microscopy.

In vitro cytotoxicity and apoptosis

The extent of apoptosis was quantified using the Annexin V-PE/7-AAD apoptosis assay. Specifically, 4T1 cells (3 × 10⁵ cells/well) were cultured for 24 h in the presence of the aforementioned preparations. After incubation, the cells were collected, resuspended in binding buffer, and subsequently stained with Annexin V-PE and 7-AAD solutions. The samples were then promptly analyzed by FCM. 4T1 cells (3,000 cells/well) were cultured with each of the four preparations at predetermined concentrations for 24–48 h. Subsequently, MTT solution was added to each well and cultured at 37 °C for 3–4 h. The culture medium was carefully aspirated, and DMSO was added to dissolve the formazan crystals.

Evaluation of autophagy inhibition in vitro

4T1 cells (1 × 10⁵ cells/well) were seeded on coverslips and cultured overnight. The following day, the cells were treated with the four nanopreparations described above. Subsequently, the cells were incubated with Mitotracker Red or Lysotracker Red for 30 min, followed by fixation with 4% paraformaldehyde. Next, the cells were incubated with the primary antibody against LC3B overnight at 4 °C. This was followed by incubation with an Alexa Fluor 488-conjugated secondary antibody (sheep anti-rabbit) for 1 h at room temperature. The cells were then washed three times with PBS and stained with DAPI for 5 minutes to label nuclei. Finally, colocalization between mitochondria or lysosomes and LC3B-positive autophagosomes was visualized by CLSM.

4T1 cells (2 × 10⁵ cells/well) were seeded in 6-well plates and subsequently treated with PBS, Cel, Cel-Ca-NP, Cel-Ca/CQ-NP, and Cel-Ca/CQ@OMM-NP for 24 h. The levels of LC3B and TOM20 proteins were quantitatively analyzed by Western blotting. The intracellular expression levels of ATG5 and ATG7 were quantified by RT-qPCR. Total RNA was extracted from tumor cells using the TRIzol reagent, followed by reverse transcription and quantitative PCR analysis performed with the SYBR Green Premix qPCR kit. Data normalization was conducted using the ΔΔCt method, with GAPDH and β-actin as reference genes. The primer sequences employed in this study are presented in Table S2.

Molecular regulators of ER-mitochondria interaction in vitro

4T1 cells (2 × 105 cells/well) were seeded in 6-well plates and subsequently treated with PBS, Cel, Cel-Ca NPs, Cel-Ca/CQ NPs, and Cel-Ca/CQ@OMM NPs for 24 h. The levels of LC3B and TOM20 proteins were quantitatively analyzed by Western blotting.

ICD effect evaluation

4T1 cells (5000 cells /well) were seeded into 24-well plates and cultured for 24 h under standard culture conditions. To detect CRT markers on the cell surface, cells were incubated with free Cel, Cel-Ca-NP, Cel-Ca/CQ-NP, and Cel-Ca/CQ@OMM-NP at a concentration of 1.6 µM for 4 h under controlled conditions. Subsequently, all groups were fixed, permeabilized, washed, and incubated with the CRT primary antibody (diluted at 1:2000) for 1 h under gentle agitation. To detect intracellular HMGB1, cells were treated with free Cel, Cel-Ca-NP, Cel-Ca/CQ-NP, and Cel-Ca/CQ@OMM-NP at a concentration of 1.2 µM for 12 h under strictly controlled conditions. Primary HMGB1 antibody (diluted at 1:200) was added to the cells and incubated under appropriate conditions for 1.5 h. Subsequently, the cells were incubated with a secondary anti-rabbit antibody (diluted at 1:200) for an additional 1.5 h. Finally, the cells were mounted using an anti-fluorescence quencher containing DAPI for nuclear staining. CLSM was employed to capture immunofluorescence images. CRT-positive cells were analyzed by flow cytometry. The amounts of HMGB1 and ATP released into the supernatant were quantified following the protocols outlined in the HMGB1 ELISA kit and ATP detection kit instructions.

To investigate the in vivo effects of ICD on checkpoint CRT, tumors were harvested at the conclusion of the experiment and subjected to further analysis using immunofluorescence techniques. Initially, the tissue sections were washed and fixed. Subsequently, they were incubated with an anti-CRT antibody overnight at 4 °C, followed by a 1.5-hour incubation with a FITC-labeled secondary antibody. Finally, the samples were mounted with an anti-fade reagent containing DAPI and analyzed using a high-resolution slide scanner.

In vivo distribution of nanoparticles

The 4T1 tumor-bearing mouse model was successfully established, and various ICG-loaded nanoparticles, including Cel-Ca/ICG-NPs and Cel-Ca/ICG@OMM NPs. When the tumors reached 600 mm^³, the mice were administered intravenous treatments of ICG or ICG/NP. Subsequently, the mice were imaged at predetermined time intervals using an imaging system (Caliper Life Sciences, USA). Mice were euthanized 24 h post-injection, and tumors as well as major organ tissues were harvested for ex vivo fluorescence imaging under standardized conditions.

Similarly, mice were administered two injections of ICG-NPs via the tail vein. Four hours later, mitochondria were isolated according to the manufacturer’s protocol using the isolation kit, suspended in PBS, and their fluorescence intensity was quantified by FCM.

In vivo anti-tumor efficiency of nanoparticles

For the efficacy study of Cel-Ca/CQ@OMM-NPs, a tumor-bearing Balb/c mice model was established by inoculating 4T1 cells (8 × 10⁵ cells) into the third mammary fat pad of female Balb/c mice. Once the tumor size reached approximately 80 mm³, the tumor-bearing mice (n = 5) were intravenously administered with saline, Cel, Cel-Ca-NP, Cel-Ca/CQ-NP, and Cel-Ca/CQ@OMM-NP (equivalent dose of 1 mg/kg Cel) every two days. Afterwards, mice were sacrificed on day 22. The tumor inhibition rate was determined using the following formula: Tumor inhibition rate (%) = (tumor weight in saline group - tumor weight in treatment group) / (tumor weight in saline group) × 100%. Throughout the experiment, both tumor volume and body weight were measured daily. The tumor volume was calculated using the formula: Tumor volume = width² × length×0.5. Finally, tumors and major organs were excised and prepared for subsequent H&E staining and ROS staining. Concurrently, the survival duration of each mouse was meticulously monitored, and the corresponding survival curve was documented.

In-vivo safety evaluation

Upon completion of the treatment, a whole blood sample was collected for routine hematological analysis. Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels were measured to evaluate potential hepatotoxicity, while serum blood urea nitrogen (BUN) and creatinine (CREA) levels were assessed to determine renal function.

Immune status investigation

The tumor microenvironment was analyzed by FCM. The infiltration of immune cells was evaluated, including CD4⁺ T cells (CD3⁺ CD4⁺), CD8⁺ T cells (CD3⁺ CD8⁺), regulatory T cells (CD3⁺ CD4⁺ Foxp3⁺), and myeloid-derived suppressor cells (CD11b⁺ Gr-1⁺). Additionally, dendritic cells (CD11c⁺ MHC II⁺) as well as M1 macrophages (CD11b⁺ F4/80⁺ CD86⁺) and M2 macrophages (CD11b⁺ F4/80⁺ CD206⁺) were assessed. Furthermore, indicators of T cell function, such as granzyme B (CD8⁺ GZMB⁺) and interferon-γ (CD8⁺ IFN-γ⁺), were also analyzed. First, the dissociation buffer was added to the tumor tissue, followed by mechanical disruption using surgical scissors. Subsequently, the resulting single-cell suspension was centrifuged, and the cells were resuspended in an appropriate staining buffer. Then, surface marker expression of immune cells was analyzed by FCM.

Statistical analysis

The results are expressed as mean±standard deviation (SD). For statistical comparisons between two independent groups, Student’s t-test was employed. For comparisons involving multiple groups, one-way analysis of variance (ANOVA) was conducted. All statistical analyses were performed using SPSS 22.0 software (SPSS, Inc., USA). A p value of less than 0.05 was considered statistically significant.

Results and discussion

Preparation and characterization of Cel-Ca/CQ@OMM NPs

Herein, we encapsulated chloroquine within the Cel-Ca complex and subsequently coated the nanoparticles with the mitochondrial membrane through a coextrusion process as described (Fig. 1A). First, the Celastrol-calcium complex was obtained by mixing the Celastrol and CaCl₂ solution (Fig. 1B). Subsequently, CQ was loaded into the complex, resulting in nanoparticles characterized by uniform particle size and excellent dispersion, as shown in Fig. 1C. Also, the UV − vis absorption peak of Cel-Ca/CQ roughly overlapped with CQ (330 nm), showing that CQ was effectively incorporated into the complex. Moreover, the absorption peak of Celastrol (425 nm) exhibited a shift, this suggests that complex was effective assembly (Fig. 1D). The coordination bond between Cel and Ca was further investigated using Fourier transform infrared (FTIR) spectroscopy. As illustrated in Fig. 1E, the C = O stretching vibration peak at 1730 cm^− 1 in Cel was observed to decrease in intensity and transform into a broader peak and the physical mixture remained unchanged, suggesting the formation of a coordination interaction between the C = O group and Ca²⁺. We subsequently coated the Cel-Ca/CQ NPs with varying proportions of mitochondrial outer membrane. At a ratio of 1:1, the potential of the NPs closely resembled that of the native mitochondrial membrane, while the particle size of the NPs was marginally larger than that of the extruded vesicles derived from the mitochondrial membrane. (Fig. 1F-G). In addition, the EE% and DL% of Cel and CQ in the formulations were determined, and the detailed data are summarized in Table 1. The data showed that the encapsulation efficiency and drug loading capacity of both Cel and CQ in Cel-Ca/CQ@OMM NPs were higher than those in Cel-Ca/CQ NPs.

Fig. 1.

Preparation and characterization of mitochondrial membrane coated nanoparticles. (A) Illustration of mitochondrial membrane coated nanoparticles. (B-C) Particle sizes and TEM images of Cel-Ca complex (Cel-Ca) and Cel-Ca/CQ NP. (D) UV − vis spectra of Cel, CQ, and Cel-Ca/CQ. (E) Fourier transform infrared spectroscopy (FTIR) spectra of Cel, CQ, and Cel-Ca/CQ. (F-G) Size and zeta potential of Cel-Ca/CQ@OMM NP. (H)TEM image of Cel-Ca/CQ@OMM NP. (I) Fluorescence imaging of nanoparticles in 4T1 cells. (J) SDS-Page of mitochondrial membrane and mitochondrial membrane coated nanoparticles (Cel-Ca/CQ@OMM NP). (K) WB analysis of mitochondrial membrane MFN-2. (L) Cumulative release profiles of nanoparticles in PBS. (M) Stability of Cel-Ca/CQ@OMM NP in 24 h at 37 ℃. Each bar denotes mean ± SD (n = 3)

Table 1.

Encapsulation efficiency and drug loading of cel and CQ in Cel-Ca-CQ@OMM nanoparticles at different mass rations (n=3)

Vehicle	Cel (EE%)	Cel (DL%)	CQ (EE%)	CQ (DL%)
Cel-ca/CQ	70.27±2.02	13.64±1.28	80.14±2.64	18.64±2.18
Cel-Ca/CQ@OMM	83.21±1.39	15.74±1.26	90.21±1.46	24.54±1.58

TEM analysis confirmed that the mitochondrial outer membranes were adsorbed onto the surface of the nanoparticles (Fig. 1H). As shown in Fig. 1I, the fluorescence imaging of Cel-Ca/Rho@OMM NPs in 4T1 cells further confirmed that the surface of the preparation successfully covered the mitochondrial membrane and could be taken up by 4T1 cells together. In addition, both OMM and Cel-Ca/CQ@OMM NPs exhibited identical protein expression profiles in SDS-Page analysis, which suggests the preservation of the functional integrity of the OMM during preparation (Fig. 1J). Western blot analysis was also conducted to detect MFN-2, a well-established outer mitochondrial membrane marker protein, in the membrane fraction and in the membrane-coated nanoparticles. The results show clear enrichment of MFN-2 in both samples (Fig. 1K), supporting the successful incorporation of mitochondrial membrane components into the nanoparticles. Then, we detected the release of Cel-Ca / CQ @ OMM NPs in different pH media(Fig. S1. (A)). Cel-Ca/CQ@OMM NPs are more likely to undergo cleavage in a weakly acidic environment. Moreover, the coating on Cel-Ca/CQ@OMM NPs enhances their sustained-release performance (Fig. 1L). After incubation in PBS supplemented with 10% FBS for 24 h at 37 °C, the coated nanoparticles demonstrated significantly enhanced stability in comparison to the control group (Fig. 1M). These findings suggest that Cel-Ca/CQ@OMM NPs exhibits superior encapsulation efficiency and stability, as well as a more sustained release profile in comparison to Cel-Ca/CQ NPs. These properties may potentially enhance the efficacy of Cel-Ca/CQ@OMM NPs, thereby enabling more precise mitochondrial targeting at tumor sites.

Cel-Ca/CQ@OMM NPs causes mitochondrial damage and initiates apoptosis

When targeted to mitochondria, Cel-Ca/CQ@OMM NPs could generate mass ROS and also cause mitochondrial damage. To evaluate mitochondrial targeting, we performed super-resolution fluorescence microscopy (Fig. 2A, Fig. S2. (A-B)). The results demonstrate that membrane-coated nanoparticles show significantly higher colocalization with mitochondria compared to non-coated nanoparticles, indicating enhanced mitochondrial accumulation. These findings provide more direct evidence for the improved mitochondrial targeting capability of OMM-coated nanoparticles. To clarify the endocytic mechanism, we then conducted inhibitor studies using filipin, amiloride, and chlorpromazine (Fig. 2B). Using Cel-Ca/Rho as a control, we observed that Cel-Ca/Rho@OMM exhibited significantly enhanced cellular uptake. Notably, filipin had minimal effect on the uptake of Cel-Ca/Rho@OMM, while amiloride partially inhibited internalization, and chlorpromazine exerted the strongest inhibitory effect. These results suggest that the uptake of Cel-Ca/Rho@OMM is primarily mediated through clathrin-dependent endocytosis, with a partial contribution from micropinocytosis. Meanwhile, comparing to Cel-Ca/CQ NPs, membrane coated nanoparticles induced a significant increase in ROS levels in 4T1 cells (Fig. 2C). Furthermore, as shown in Fig. 2D, the reduction in mitochondrial membrane potential further corroborates the occurrence of mitochondrial damage. Subsequently, we proceeded to assess apoptosis in response to mitochondrial damage and found that the coated nanoparticles exhibited greater efficacy in inducing both early and late apoptosis (Fig. 2E-F). As illustrated in Fig. 2F, Cel-Ca/CQ@OMM NPs is internalized via endocytosis and subsequently targeted to mitochondria, leading to an increase in ROS, which induces mitochondrial damage and further triggers apoptosis. To further verify whether the observed cytotoxicity was dependent on mitochondrial function, rescue experiments were subsequently performed using both CCCP and MitoTEMPO. Pretreatment with CCCP, a mitochondrial membrane potential uncoupler that impairs mitochondrial function, partially alleviated the cytotoxic effects induced by both Cel-Ca/CQ NPs and Cel-Ca/CQ@OMM NPs, resulting in a significant recovery of cell viability (Fig. S3. (A)). Similarly, pretreatment with the mitochondria-targeted ROS scavenger MitoTEMPO markedly attenuated nanoparticle-induced cell death (Fig. S3. (B)). Consistent with these observations, MitoTEMPO significantly reduced intracellular ROS accumulation and partially restored mitochondrial membrane potential, as demonstrated by ROS and JC-1 analyses (Fig. S4 (A-B)). These results indicate that mitochondrial dysfunction and mitochondrial ROS generation are essential upstream events mediating the antitumor activity of Cel-Ca/CQ@OMM NPs, thereby providing functional evidence supporting the mitochondria-dependent mechanism of action. To further evaluate the potential off-target effects associated with autophagy inhibition, we assessed the cytotoxicity of different formulations in MCF-10 A cells. Notably, CQ alone exhibited negligible cytotoxicity under the tested conditions. In contrast, Cel, Cel-Ca NPs, Cel-Ca/CQ NPs, and Cel-Ca/CQ@OMM NPs showed a gradual increase in cytotoxicity. However, the overall cell viability reduction remained limited (approximately 20–30%) and was consistently lower than that observed in 4T1 cells (Fig. 2G).

Fig. 2.

Cel-Ca/CQ@OMM NPs accumulate at the mitochondria of tumor cells and cause damage. (A) Super-resolution images of Cel-Ca / Rho and Cel-Ca / Rho @ OMM nanoparticles endocytosis. (B) Cellular uptake of Cel-Ca/Rho and Cel-Ca/Rho@OMM NP by FCM. (C-D) Fluorescence imaging of ROS and Mitochondrial membrane potential changes in 4T1 cells after different treatments. Scale bar: 20 μm. (E-F) Cell apoptosis and cell viability analysis of 4T1 cells after different treatments. (G) Cytotoxicity of different nanoparticles on normal breast cells MCF-10A. All results were presented as mean ± SD, (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

Cel-Ca/CQ@OMM NPs attenuated compensatory mitophagy

In response to reactive oxygen species, nutrient deprivation, cellular senescence, or other external stimuli, tumor cells initiate mitophagy to eliminate damaged mitochondria [30]. Both general autophagy and mitochondria-specific autophagy further remove dysfunctional or impaired mitochondria, thereby mitigating tumor cell death [31]. As illustrated in Fig. 3A, Cel-Ca/CQ@OMM NPs inhibit the clearance of damaged mitochondria by impairing lysosomal function and reducing lysosomal quantity. The colocalization analysis of LC3B (green) and mitochondria (red) in confocal microscopy images (Fig. 3B) demonstrated that the Cel-Ca complex induces mitophagy. Moreover, the addition of CQ led to an increased accumulation of LC3B, thereby enhancing their colocalization with mitochondria. This effect is likely attributable to the inhibition of lysosomal degradation, which results in the further accumulation of LC3B. Moreover, the co-localization of LC3B (green) and lysosomes (red) was markedly diminished following the addition of CQ, suggesting the inhibition of mitophagy (Fig. 3C), and the colocalization of mitochondria (green) and lysosomes (red)was weakened, which also proved the down-regulation of autophagy (Fig. 3D). However, free CQ did not induce a significant change in LC3 colocalization under the tested conditions, whereas Cel-Ca/CQ NPs markedly enhanced LC3 accumulation (Fig. 3B), suggesting more effective inhibition of autophagic flux via nanoparticle-mediated delivery.

Fig. 3.

Compensatory mitophagy is attenuated by Cel-Ca/CQ@OMM NPs. (A) Schematic illustration of Cel-Ca/CQ@OMM NPs inhibiting compensatory mitophagy. (B) Confocal images of autophagic substrate LC3B in 4T1 cells treated with different nanoparticles and colocalization of autophagic substrate LC3B (green) with mitochondria (red) in 4T1 cells. (C) Colocalization laser confocal fluorescence (yellow) of mitochondria (green) and lysosomes (red) in 4T1 cells after different treatments. (D) Confocal images of autophagic substrate LC3B in 4T1 cells treated with different nanoparticles and colocalization of autophagic substrate LC3B (green) with lysosomes (red) in 4T1 cells. (E-H) WB analysis of the autophagic substrate (LC3B) and mitochondrial membrane Protein (TOM20) expression in 4T1 cell treated with different nanoparticles. (I) The mRNA expression of ATG5 and ATG7 in 4T1 cell treated with different nanoparticles. (J) The expression of MFN-2 protein in 4T1 cells treated with different nanoparticles. (K) Bio-TEM images of lysosomes and mitochondria in 4T1 cells after different treatments. Scale bar: 1 μm. All results were presented as mean ± SD, (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

As is well known, the activation of autophagy results in enhanced autophagosome formation, thereby increasing the LC3-II/LC3-I ratio. Subsequently, the autophagosomes is internalized and degraded by lysosomes [32, 33]. Theoretically, the inhibition of lysosomal function and autophagosome-lysosome fusion by CQ could lead to a substantial increase in the LC3-II/LC3-I ratio [34]. WB analysis was employed to further investigate the autophagy pathway. As illustrated in Fig. 3E-H, Cel-Ca complex markedly increased the LC3-II/LC3-I ratio compared with the control group, indicating the initiation of autophagy. Moreover, the upregulation of TOM20 levels following the addition of CQ suggested the inhibition of mitophagy. The reduced expression of the autophagy-related genes ATG5 and ATG7 suggests a potential decline in compensatory autophagy (Fig. 3I).

To further explore the molecular regulation of ER-mitochondria interactions, we examined the expression of MFN-2, a key tethering protein that mediates physical and functional coupling between the endoplasmic reticulum and mitochondria. As shown in Fig. 3J, treatment with Cel resulted in a noticeable reduction in MFN-2 expression, which was further exacerbated by Cel-Ca NPs, suggesting that enhanced mitochondrial stress disrupts ER-mitochondria contact sites. This disruption is likely associated with impaired Ca²⁺ transfer and altered inter-organelle signaling, thereby contributing to mitochondrial dysfunction and ER stress. Interestingly, the addition of CQ partially restored MFN-2 expression in the Cel-Ca/CQ NPs group, and a more pronounced recovery was observed in the Cel-Ca/CQ@OMM NPs group. This effect may be attributed to the inhibition of autophagic degradation, which prevents excessive loss of mitochondrial-associated membrane components, including MFN-2. The improved preservation of MFN-2 in the OMM-coated group suggests that membrane engineering may help stabilize ER-mitochondria interactions under stress conditions. Collectively, these findings indicate that Cel-Ca-induced mitochondrial damage disrupts ER-mitochondria coupling, while CQ-mediated autophagy inhibition, particularly in the context of OMM coating, partially preserves MFN-2 expression and modulates inter-organelle communication. This dynamic regulation may contribute to the amplification of ER stress and apoptotic signaling observed in our system. Collectively, these findings indicate that Cel-Ca-induced mitochondrial damage disrupts ER-mitochondria coupling, while CQ-mediated autophagy inhibition, particularly in the context of OMM coating, partially preserves MFN-2 expression and modulates inter-organelle communication. This dynamic regulation may contribute to the amplification of ER stress and apoptotic signaling observed in our system.

When mitochondria and autophagosomes were examined using Bio-TEM (Fig. 3K), it was evident that the mitochondrial cristae and double membrane were well-preserved in the control group, and no autophagosomes were observed. In contrast, following treatment with Cel-Ca/CQ@OMM NPs, the mitochondria exhibited severe damage and accumulated in substantial amounts. Notably, autophagosomes were also detected in 4T1 cells treated with Cel-Ca/CQ@OMM NPs. In conclusion, the Cel-Ca/CQ@OMM NPs significantly enhanced the Cel-Ca complex-induced apoptosis by effectively inhibiting mitophagy.

Damaged mitochondria and accumulation of lipid droplets together induce ER stress, further amplifying ICD

Inhibition of mitophagy results in an accumulation of damaged mitochondria, When mitochondria are damaged, their capacity for calcium uptake diminishes, leading to an abnormal elevation in calcium concentration surrounding the endoplasmic reticulum [35]. This disturbance further compromises calcium homeostasis within the endoplasmic reticulum and ultimately aggravates endoplasmic reticulum stress [36]. Moreover, lipid droplets cannot be degraded normally, and lipids cannot be turnover in time, which aggravates the deposition and metabolic burden of lipids in the ER [37, 38]. Using Bodipy 493/503 staining for neutral lipids and co-localization analysis with the autophagosome marker LC3B, we found that Cel-Ca treatment increased the co-localization signal, indicating lipophagy activation. However, after the addition of chloroquine, undegraded autophagosomes persisted due to the blocking of autophagic flow, resulting in the accumulation of LC3-II on the autophagosome membrane, and the colocalization of LC3 with lipid droplets was further elevated (Fig. 4A). Consequently, this inhibition led to a marked accumulation of intracellular lipid droplets (Fig. S5). WB analysis further showed that the expression levels of ER stress markers GRP78 and p-PERK/ total PERK were significantly up-regulated after treatment with cel-ca /CQ@OMM NPs (Fig. 4B-C). Upon treating 4T1 cells with Cel-Ca/CQ@OMM NPs for 24 h and staining with a calcium ion fluorescent probe, it was observed that treatment with an autophagy inhibitor resulted in an increase in lipid content (Fig. S5). This finding confirms that ER stress was effectively induced under these experimental conditions.

Fig. 4.

Mitochondrial damage induces ER stress and further amplifies ICD. (A) Confocal images of lipid droplets in 4T1 cells treated with different nanoparticles and colocalization of autophagic substrate LC3B (red) with BODIPY (green) in 4T1 cells. (B-C) WB analysis of the GRP78 expression in 4T1 cell after different treatments. (D-E) The CLSM images of CRT and HMGB1 (Green). Scale bar = 50 μm. (F-G) HMGB1 and ATP release of 4T1 cell after treated with different nanoparticles. (H) Quantitative FCM analysis of CRT treated with different nanoparticles. All results were presented as mean ± SD, (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

ER stress can facilitate the release of DAMPs via mechanisms such as CRT exposure, ATP secretion, and HMGB1 release [39]. This process subsequently activates DCs and triggers the initiation of adaptive immune responses [40]. We next investigated the ICD-induction potential of Cel-Ca/CQ@OMM NPs. The surface exposure of CRT in 4T1 cells was evaluated using CLSM and FCM. As shown in Fig. 4D and E, CLSM images revealed that cells treated with Cel-Ca/CQ@OMM-NPs exhibited markedly enhanced green fluorescence compared to the Cel group. Furthermore, FCM demonstrated that treatment with Cel-Ca/CQ@OMM-NPs significantly elevated the proportion of CRT⁺ cell ratio, which was two times higher than that of the control group (Fig. 4H). ICD is frequently linked to the release of ATP and HMGB1. The levels of ATP secretion and HMGB1 release were quantified. CLSM and ELISA assays showed that HMGB1 was released from the nuclei of 4T1 cells treated with Cel-Ca/CQ@OMM-NPs. Notably, Cel-Ca/CQ@OMM-NPs enhanced the release of HMGB1 three-fold compared to the control group (Fig. 4F). In addition, the treatment with Cel-Ca/CQ@OMM-NPs led to the highest level of ATP release compared to all other groups in the study (Fig. 4G). Notably, treatment with Cel-Ca/CQ@OMM-NPs in tumor tissues led to a markedly higher staining intensity of CRT and HMGB1 compared to free Cel. This clearly indicates that the efficacy of inducing ICD is more pronounced in vivo (Fig. S6).

Taken together, these findings demonstrate that Cel-Ca/CQ@OMM-NPs induces ER stress and significantly enhances the ICD effect by inhibiting mitophagy and lipophagy, thereby highlighting its promising role in anti-tumor immunity.

Cel-Ca/CQ@OMM NPs enhanced mitochondrial targeting efficacy at the tumor site in 4T1^Fluc tumor-bearing mice

To investigate the tumor-targeting capability of nanoparticles, ICG was encapsulated into a variety of nanoparticle formulations. In vivo imaging experiments (Fig. 5A) demonstrated that the fluorescence of Cel-Ca/ICG@OMM NPs was detected at the tumor site as early as 4 h post-injection and remained observable until 24 h, significantly surpassing the fluorescence intensity of Cel-Ca/ICG NPs. This is further substantiated by its quantitative outcomes (Fig. 5C). Additionally, in Fig. 5B and D, in vivo imaging system successfully captured the fluorescence signals from the dissected tumor and major organs. Notably, the fluorescence intensity of Cel-Ca/ICG@OMM NPs in the tumor was 1.36-fold higher than that of Cel-Ca/ICG NPs.

Fig. 5.

In vivo homologous tumor targeting effects, distribution of Cel-Ca/CQ@OMM NPs. (A) Real-time fluorescence imaging of 4T1 tumor-bearing Balb/c mice (n = 3) treated with ICG-loaded nanoparticles at 2 h, 4 h, 8 h and 24 h after intravenous injection. (B) Ex vivo fluorescence images of dissected tumors and organs (from left to right: heart, liver, lung, spleen, kidney and tumor tissue) from sacrificed 4T1 tumor-bearing Balb/c mice at 4 h post-injection. (C-D) Quantitative analysis of fluorescence signal for 4T1 tumor-bearing Balb/c mice and dissected tissues. (E)Schematic illustration of mitochondrial segregation in tumor tissue. (F) Quantitative FCM analysis of mitochondria in tumor tissue with ICG-loaded nanoparticles at 4 h. (G) In vivo fluorescence imaging of mice simultaneously bearing 4T1 and CT26 tumors after being injected (i.v.) with ICG-loaded nanoparticles (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

Next, we further validated the targeting capability of Cel-Ca/ICG@OMM NPs to mitochondria at the tumor site. As illustrated in Fig. 5E, we promptly isolated mitochondria from tumor tissues using commercially available kits according to the manufacturer’s instructions, the fluorescence intensity of Cel-Ca/ICG@OMM NPs was measured by flow cytometry. The fluorescence intensity of Cel-Ca/ICG@OMM NPs was approximately twice that of Cel-Ca/ICG NPs without OMM (Fig. 5F).

To further assess its in vivo targeting capability, ICG-labeled Cel-Ca/ICG@OMM NPs was intravenously administered into mice that were simultaneously inoculated with two distinct types of tumor xenografts: 4T1 on the left flank and CT26 on the right flank. A significantly stronger fluorescence signal was observed in 4T1 tumors compared with CT26 tumors (Fig. 5G). These results demonstrate the capacity of Cel-Ca/ICG@OMM NPs to specifically target and recognize both cognate tumors and tumor cells, which is essential for achieving effective drug delivery.

In vivo antitumor efficacy of Cel-Ca/CQ@OMM NPs

Inspired by the aforementioned positive outcomes, we subsequently assessed the in vivo anticancer properties of Cel-Ca/CQ@OMM NPs. 4T1-bearing mice were randomly divided into 5 groups and treated with different dosage forms when the tumor volume of tumor-bearing mice reached approximately 80 mm³. Treatments included Treatments included Control, Cel (1 mg/kg, ip), Cel- Ca (Cel, 1 mg/kg, iv), Cel-Ca/CQ NPs (Cel, 1 mg/kg, iv), Cel-Ca/CQ@OMM NPs (Cel,1 mg/kg, iv) (Fig. 6A). In vivo bioluminescence imaging (BLI) demonstrated that the administration of Cel-Ca/CQ@OMM NPs markedly suppressed tumor progression (Fig. 6B-C).

Fig. 6.

Antitumor effect of Cel-Ca/CQ@OMM NPs in vivo. (A) Treatment scheme to 4T1 tumor-bearing mice. (B) BLI images of tumor in 4T1^Fluc tumor-bearing mice at 1, 7, 14, 21 days (n = 5). (C) BLI ratio curve of 4T1^Fluc-bearing mice (n = 5). (D) Tumor volume changes and (G) Weights of tumors changes after treatments (n = 5). (E) Photographs of tumors after treatment (n = 5). (F) Tumor inhibition ratio (n = 5) and (H) survival curves of mice (n = 5). (I-J) H&E and ROS staining images of tumors (scale bar = 50 μm). *P < 0.05, **P < 0.01, ***P < 0.001

In addition, mice were monitored for changes in tumor size and body weight at 2-day intervals throughout the 21-day experimental period (Fig. 6D and S7). After 21 days, the mice were humanely euthanized. As illustrated in Fig. 6E-G, tumor growth was observed to be rapid in the Control, Cel, and Cel-Ca groups, suggesting minimal antitumor effects. In contrast, both Cel-Ca/CQ NPs and Cel-Ca/CQ@OMM NPs demonstrated varying degrees of tumor regression, indicating potential therapeutic efficacy. Notably, the tumor inhibition rate achieved with Cel-Ca/CQ@OMM NPs treatment was significantly higher at approximately 63.2%, compared to the rate of approximately 18.4% observed with Cel treatment alone. Furthermore, the median survival time was substantially prolonged in the Cel-Ca/CQ@OMM NPs treatment group in comparison to other group (Fig. 6H). Histological analysis via H&E staining confirmed that tumor cells treated with Cel-Ca/CQ@OMM NPs demonstrated a significantly higher rate of apoptosis compared to those in the other groups (Fig. 6I). Moreover, ROS staining of the frozen sections of tumor tissue suggested that the increase in ROS levels was likely to further induce apoptosis (Fig. 6J). Overall, these findings highlight the synergistic potential of Cel combined with the use of autophagy inhibitors to improve the treatment of TNBC.

Finally, the safety profile of the nanoparticles was comprehensively assessed. First, Fig. S7 indicate that there was no significant change in the body weight of mice during the treatment period. Moreover, hematoxylin and eosin (H&E) staining images reveal no apparent damage to major organs (Fig. S8). Additionally, blood biochemical indices presented in Fig. S9 further confirm the absence of liver and kidney injury.

Systemic anti-tumor immune response induced by Cel-Ca/CQ@OMM NPs

Given that treatment with Cel-Ca/CQ@OMM NPs delays tumor progression, we systematically investigated the mechanistic underpinnings of this synergistic antitumor effect. A comprehensive FCM was conducted to elucidate the impact on the immune cell population within tumors from 4T1-bearing mice. Our analysis focused on DC maturation, a critical indicator of immune activation. We evaluated the frequency of mature DCs (CD11c⁺ CD86⁺ CD80⁺) in tumor tissue on day 21 post-treatment (Fig. 7A). Notably, the mature DCs was significantly promoted by Cel treatment alone and was further enhanced by the addition of CQ, suggesting that the inhibition of autophagy can potentiate ICD in vivo. Cel-Ca/CQ@OMM NPs significantly increased the population of mature DCs by approximately 3-fold compared to the control group. The pronounced maturation of DCs induced by Cel-Ca/CQ@OMM NPs further underscores the synergistic effect of mitochondrial targeting in augmenting immune activation.

Fig. 7.

Cel-Ca/CQ@OMM NPs improve TME in the in 4T1^Fluc tumor-bearing mice. (A) FCM results of matured DC in tumor (gated from CD11c⁺). (B) Quantitative results by FCM of CD8⁺, (C) GZMB, (D) CD4⁺, (E) M1/M2 ratio and (F) IFN-γ expression in CD8⁺ T cells in tumor (n = 3). (G-H) Quantitative FCM analysis of Tregs and MDSCs in tumor (n = 3). (I) IF images of tumor tissues after various treatments (n = 3). Nucleus = blue, CD8 = red. Scale bars=100 μm. *p < 0.05, **p < 0.01, ***p < 0.001

DCs are crucial for pathogen recognition and the activation of adaptive immunity, particularly in the context of T lymphocyte stimulation [40]. FCM analysis demonstrated that the proportion of CD8⁺ T cells in the Cel-Ca/CQ@OMM NPs treatment group was 14.32%, indicating a 2.8-fold increase compared with the control group (Fig. 7B). In addition, the administration of Cel-Ca/CQ@OMM NPs significantly enhanced the functional indicators of CD8⁺T cells, as evidenced by higher levels of IFN-γ and GZMB compared to other treatment groups (Fig. 7C, F). The treatment also significantly activated CD4⁺ helper T cells, thereby indicating a strong T-cell-mediated immune response (Fig. 7D). Moreover, treatment with Cel-Ca/CQ@OMM NPs significantly decreased the populations of two immunosuppressive cell types, namely MDSCs and Tregs, compared to the control group (Fig. 7G-H). As illustrated in Fig. 7E, the introduction of CQ resulted in autophagy dysfunction and simultaneously inhibited macrophage polarization, thereby further potentiating the ICD effect.

IF images further confirmed that Cel-Ca/CQ@OMM NPs markedly enhanced the infiltration of CD8⁺ T cells in tumor tissues (Fig. 7I). Taken together, our findings indicate that treatment with Ca/CQ@OMM NPs can markedly activate DCs and T lymphocytes, synergistically suppress autophagy, and elicit a robust ICD effect, thereby providing an effective therapeutic strategy for TNBC.

Discussion

In this study, we developed a mitochondria-targeted nanoplatform (Cel-Ca/CQ@OMM) that induces mitochondrial damage, amplifies ER stress, and blocks mitophagy, ultimately promoting apoptotic and immunogenic cell death. Emerging evidence suggests that mitophagy plays a context-dependent role in tumor progression. While excessive mitophagy may lead to metabolic collapse, basal mitophagy—primarily regulated by pathways such as PINK1/Parkin—functions as a cytoprotective mechanism that removes damaged mitochondria and maintains redox homeostasis [41]. Therefore, inhibition of mitophagy, as achieved in our system, may prevent tumor cells from adapting to mitochondrial stress, thereby converting reversible damage into irreversible cytotoxic signaling.

Importantly, our findings also highlight the interplay between mitochondrial dysfunction and ER stress in regulating apoptosis. Sustained mitochondrial damage and Ca²⁺ overload disrupt ER-mitochondria crosstalk, leading to persistent ER stress and activation of pro-apoptotic signaling pathways. In addition, inhibition of autophagic flux by chloroquine further amplifies cellular stress by preventing the clearance of damaged organelles, forming a feed-forward loop that enhances cell death.

From an immunological perspective, although our study demonstrates DC maturation and immune activation, antitumor immunity is a highly coordinated process involving multiple immune cell populations. Recent studies emphasize the importance of tumor microenvironment complexity, including the roles of T cells, macrophages, and immune regulatory networks in determining therapeutic outcomes [42–44]. Therefore, while our data support the induction of immunogenic cell death, further comprehensive immune profiling will be necessary to fully elucidate long-term antitumor immunity and immune memory formation.

In addition, several limitations should be acknowledged. First, although our data support a mitochondria-dependent mechanism, validation in ρ0 cells lacking functional mitochondria would further strengthen this conclusion. Second, while moderate cytotoxicity toward non-aggressive cells was observed, a more comprehensive evaluation using normal cell lines and in vivo toxicity models is required. Third, although the OMM-coated nanoplatform demonstrates enhanced efficacy, its structural complexity may pose challenges for large-scale production and clinical translation. Future studies should focus on simplifying the formulation, improving reproducibility, and establishing standardized manufacturing protocols to facilitate potential clinical application [45].

Overall, our study provides a strategy for disrupting mitochondrial quality control to enhance cancer therapy, while also highlighting the need for further investigation into immune responses and translational feasibility.

Conclusions

In this study, we have developed a biomimetic mitochondrial-targeted nanoplatform that effectively disrupts the mitochondrial-ER homeostasis through coordinated delivery of Celastrol and Chloroquine. By simultaneously inducing mitochondrial Ca²⁺ overload to provoke integrated organelle and metabolic stress and inhibiting compensatory autophagy, this strategy comprehensively blocks key survival pathways, including mitophagy and the triggered lipophagy, thereby establishing a self-amplifying cycle of cellular stress. The resultant severe disruption of cellular homeostasis significantly enhances immunogenic cell death through amplified DAMPs release and robust CD8⁺ T cell activation. Our findings not only demonstrate an effective nanotherapeutic approach against TNBC but also establish a new paradigm for comprehensively targeting the adaptive interplay between organelle communication and metabolism in cancer therapy.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

Cel

Celastrol

TNBC

Triple-negative breast cancer

CQ

Chloroquine

ER

Endoplasmic reticulum

ICD

Immunogenic cell death

ROS

Reactive oxygen species

DAMP

Damage-associated molecular pattern

MAMs

Mitochondria-associated ER membranes

DLS

Dynamic light scattering

BCA

Bicinchoninic acid

ICG

Indocyanine green

WB

Western blotting

FCM

Flow cytometry

IF

Immunofluorescence

ATP

Adenosine triphosphate

HMGB1

High mobility group protein box-1

CRT

Calreticulin

LC3

Microtuble-associated protein light chain 3

TOM20

Translocase of outer mitochondrial membrane 20 homolog

PERK

Protein Kinase R-like Endoplasmic Reticulum Kinase

GRP78

Glucose regulated protein 78

PDI

Polydispersity index

H&E

Hematoxylin and eosin

AST

Aminotransferase

ALT

Alanine aminotransferase

BUN

Blood urea nitrogen

CRE

creatinine

PVDF

Polyvinylidene difluoride

Author contributions

**, Yuhao Ye, Xin Zhang, Qi Tong** : Writing – review & editing, Writing – original draft, Project administration, Methodology, Investigation, Formal analysis, Data curation. **Qiong Xie: ** Methodology, Visualization, Investigation. **Xuanming Gong，Siqi Du, Peiqin Gao: ** Validation, Formal analysis, Data curation, Visualization. **Qiushuang Li, Jigang Piao, Yang Xiong** : Conceptualization, Writing- Reviewing and Editing, Resources, Project administration, Methodology, Investigation, Funding acquisition.

Funding

This work was supported by the National Natural Science Foundation of Zhejiang Province (grant number LZ26H290001), Zhejiang Province Traditional Chinese Medicine Science and Technology Project (grant number 2026ZL1003), Natural Science Foundation of Hangzhou (grant number 2024SZRZDC200001).

Data availability

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article.

Declarations

Ethics approval and consent to participate

All experimental procedures strictly complied with the guidelines approved by the Animal Care and Use Committee of Zhejiang Chinese Medical University, in accordance with ethical permit No. IACUC-20230821-09.

Competing interests

The authors declare no competing interests.