Metal-phenolic nanocapsules enable a self-amplifying cuproptosis–STING cascade for synergistic cancer immunotherapy

Abstract

Immunosuppressive tumor microenvironment remains a major obstacle to effective cancer immunotherapy, largely due to insufficient initiation and amplification of antitumor immune responses. Herein, we report a mechanism-driven nanotherapeutic strategy that establishes a self-amplifying cuproptosis–STING cascade to overcome tumor immune resistance. The multifunctional copper/manganese-phenolic nanocapsules (HLCM@Cap) undergo pH-responsive release in the acidic tumor microenvironment, enabling efficient intratumoral copper accumulation and triggering cuproptosis characterized by mitochondrial dysfunction and proteotoxic stress. The resulting release of mitochondrial DNA activates the cGAS-STING pathway, while concurrently released Mn²⁺ further amplifies STING signaling. Meanwhile, Mn²⁺ also enables T₁-weighted magnetic resonance imaging for real-time monitoring of intratumoral nanocapsule accumulation and release, allowing optimization of the administration window. To counteract tumor adaptive resistance, a Wnt/β-catenin inhibitor is incorporated to suppress glycolytic reprogramming and copper efflux, thereby enhancing intracellular copper toxicity and metabolic stress. This coordinated regulation forms a positive feedback loop that reinforces STING activation through persistent damage-associated signaling. Consequently, the cascade promotes dendritic cell maturation, enhances CD8⁺ T cell infiltration, remodels the immunosuppressive tumor microenvironment, and induces durable immune memory. In a 4T1 tumor model, HLCM@Cap achieves significant antitumor and antimetastatic effects, which are further enhanced in combination with αPD-L1 therapy. Overall, this work presents a self-amplifying cuproptosis–STING cascade to convert immunologically “cold” tumors into “hot” tumors, offering a promising and translatable strategy for synergistic cancer immunotherapy.

Graphical abstract

Highlights

• •A self-amplifying cuproptosis–STING cascade overcomes tumor immune resistance.
• •A pH-responsive nanocapsule platform (HLCM@Cap) enables tumor-targeted copper delivery and T₁-enhanced MR imaging.
• •Targeting Wnt/β-catenin signaling amplifies copper toxicity and disrupts tumor metabolic adaptation.
• •HLCM@Cap synergizes with αPD-L1 to remodel the tumor microenvironment and enhance antitumor and antimetastatic efficacy.

1. Introduction

Cuproptosis is a recently elucidated form of immunogenic cell death (ICD) driven by intracellular copper accumulation [1]. Unlike conventional cytotoxic mechanisms, cuproptosis simultaneously induces direct tumor cell death and stimulates antitumor immunity, rendering it a promising mechanism for breaking tumor immune resistance in immunosuppressive solid tumors, such as triple-negative breast cancer (TNBC), where poor response to immunotherapy remains a significant clinical challenge. Mechanistically, copper ions directly bind to lipoylated components of the tricarboxylic acid (TCA) cycle within the mitochondrial respiratory chain, inducing aggregation of lipoylated proteins and downregulation of iron-sulfur (Fe-S) cluster proteins, thereby triggering lethal proteotoxic stress [[1], [2], [3], [4]]. Concurrently, mitochondrial dysfunction leads to the release of mitochondrial DNA (mtDNA) into the cytosol. This mtDNA is sensed by cyclic GMP-AMP synthase (cGAS), which activates the stimulator of interferon genes (STING) pathway, resulting in the robust production of type I interferons (IFN-I) and pro-inflammatory cytokines [[5], [6], [7]]. The activation of cGAS-STING signaling pathway promotes dendritic cell (DC) maturation and enhances antigen presentation, ultimately facilitating CD8⁺ T cell-mediated antitumor immunity [[8], [9], [10]]. Therefore, the cuproptosis–STING cascade represents a unique therapeutic modality that integrates direct cytotoxicity with innate immune activation [[11], [12], [13], [14], [15]].

Despite its therapeutic promise, the efficacy of cuproptosis is frequently compromised by intrinsic tumor resistance mechanisms, which impair the execution of this immunogenic cascade. Tumor cells maintain copper homeostasis by upregulating copper-exporting transporters such as ATP7B, thereby reducing intracellular copper accumulation [16,17]. In parallel, metabolic reprogramming toward glycolysis provides an alternative energy supply that alleviates mitochondrial stress and supports tumor survival under copper-induced damage [[18], [19], [20]]. These adaptive responses collectively attenuate cuproptosis-mediated cytotoxicity, suppress mtDNA release and limit the activation of the cGAS-STING pathway, thereby maintaining the immunosuppressive tumor microenvironment (TME) phenotype. Consequently, strategies that simultaneously enhance intracellular copper toxicity and disrupt metabolic adaptation are essential to fully exploit the therapeutic potential of the cuproptosis–STING cascade.

The Wnt/β-catenin signaling pathway is a central regulator of tumor immune evasion, metabolic rewiring, and therapeutic resistance [21,22]. Aberrant activation of this pathway is closely associated with enhanced copper transport and impaired antitumor immunity. Accordingly, we propose that inhibition of the Wnt/β-catenin pathway renders cells more vulnerable to mitochondrial damage under stress conditions, thereby increasing the sensitivity of tumor cells to copper-induced cell death. Moreover, previous studies have shown that both Wnt/β-catenin inhibition and cGAS-STING activation can suppress tumor glycolytic metabolism by regulating glycolysis-related genes and inducing metabolic reprogramming [[23], [24], [25]], which may further constrain tumor metabolic plasticity. These findings provide a strong mechanistic rationale for employing Wnt/β-catenin inhibitors, such as LF3 [26], to overcome tumor resistance and preserve the efficient operation of the cuproptosis-STING-immune activation cascade.

Herein, we developed hyaluronic acid (HA)-modified copper/manganese-phenolic nanocapsules loaded with LF3 (denoted as HLCM@Cap) for pH-responsive release, tumor-targeted copper delivery, and synergistic modulation of metabolic and signaling pathways. Polyphenol-mediated reduction of Cu²⁺ to the more cytotoxic Cu⁺ induces mitochondrial copper overload to amplify cuproptosis. Simultaneously, LF3 inhibits the Wnt/β-catenin pathway, suppressing glycolysis and inhibiting copper efflux to counteract key resistance mechanisms. The resulting release of damaged mtDNA, in concert with Mn²⁺ as a potent STING agonist, synergistically enhances cGAS-STING activation [[27], [28], [29], [30]], which further augments immune stimulation and metabolic inhibition. Notably, the incorporation of Mn²⁺ also provides effective T₁-weighted magnetic resonance (MR) imaging contrast [[31], [32], [33]], allowing for real-time visualization of intratumoral nanocapsule accumulation and therapeutic responses. Collectively, this integrated system establishes a self-amplifying cuproptosis–STING cascade that converts immunosuppressive TNBC into an immunoactivated phenotype, demonstrating robust synergy with αPD-L1 immunotherapy. This approach thus constitutes a multifunctional theranostic platform for synergistic cancer immunotherapy (Scheme 1).

Scheme 1.

Schematic illustration of HLCM@Cap fabrication and the underlying mechanism of synergistic antitumor immunotherapy.

2. Results and discussion

2.1. Preparation and characterization of metal-phenolic nanocapsules

The synthesis processes for the preparation of LF3-loaded copper/manganese-phenolic nanocapsules are schematically illustrated in Scheme 1. The metal-phenolic nanocapsules were synthesized using zeolitic imidazolate framework-8 (ZIF-8) as a sacrificial template according to previously reported methods [34,35]. The formation of copper/manganese-phenolic nanocapsules (CM@Cap) was driven primarily by coordination interaction between metal ions and phenolic compounds [36,37]. Subsequently, LF3 was encapsulated into CM@Cap to obtain LCM@Cap. The loading of LF3 was achieved via non-covalent interactions with polyphenols, including π-π stacking, cation-π interaction, and hydrogen bonding (Fig. 1a and b). The hollow structure of the nanocapsules was confirmed by transmission electron microscopy (TEM) (Fig. 1c, Fig. S1) and scanning electron microscopy (SEM) images (Fig. S2). TEM mapping analysis of LCM@Cap revealed a well-matched elemental distribution of Cu, Mn, O, C, S, and N, where the presence of S and N signals confirmed the successful loading of LF3 (Fig. 1d). Dynamic light scattering (DLS) measurements indicated hydrodynamic sizes of 165.2 nm for Mn@Cap, 231.2 nm for Cu@Cap, 201.3 nm for LC@Cap, and 186.3 nm for LCM@Cap (Fig. 1e), and polydispersity index (PDI) values of all formulations were below 0.20 (Fig. S3). ζ-Potential analysis showed that all prepared capsules exhibited a negative surface charge (Fig. 1f). Furthermore, the sizes of LCM@cap did not change significantly after incubation with water, 5% glucose or culture medium (10% FBS), indicating their good colloidal stability (Fig. S4).

Fig. 1.

Preparation and characterization of metal-phenolic nanocapsules. (a) Chemical structure of LF3. (b) Driving forces for the loading of LF3 into metal-polyphenol nanocapsules, including π-π stacking, cation-π interaction, and hydrogen bonding. (c) TEM images and the corresponding statistical size distribution of LCM@Cap and HLCM@Cap. Scale bars: 100 nm. (d) TEM mapping of LCM@Cap. Scale bars: 100 nm. (e) Size distribution of ZIF-8, Mn@Cap, Cu@Cap, LC@Cap, LCM@Cap and HLCM@Cap in vitro. (f) ζ-Potentials of ZIF-8, Mn@Cap, Cu@Cap, LC@Cap, LCM@Cap and HLCM@Cap. (g) Cu 2p XPS spectra of LCM@Cap. (h) Mn 2p XPS spectra of LCM@Cap. (i) S 2p XPS spectra of LCM@Cap. (j) Cumulative release profiles of Cu²⁺ and Mn²⁺ from LCM@Cap at pH 5.5 and pH 7.4.

To further investigate the chemical composition and structure of nanocapsules, powder X-ray diffraction (PXRD) analysis revealed that LCM@Cap was in an amorphous phase, confirming the absence of the crystalline structure originally present in the ZIF-8 template (Fig. S5). X-ray photoelectron spectroscopy (XPS) spectra were employed to verify the elemental composition of LCM@Cap, aligning with the TEM mapping results (Fig. 1g–i). The full-scan XPS spectrum confirmed the presence of Cu, Mn, and S (Fig. S6). In the high-resolution Cu 2p spectrum, prominent peaks were observed at binding energies of 934.35 eV for Cu 2p_3/2 and 954.3 eV for Cu 2p_1/2 [38]. Additionally, the peak profile indicated a partial reduction of Cu²⁺ to Cu⁺, which facilitates the induction of cuproptosis (Fig. 1g). In the Mn 2p region, the characteristic binding energy peaks for Mn 2p_3/2 at 641.40 eV and Mn 2p_1/2 at 652.9 eV were observed, confirming the presence of Mn²⁺ (Fig. 1h) [39]. Meanwhile, the O 1s spectrum featured distinct peaks at 532.65 eV (Cu-O) and 533.45 eV (Mn-O), verifying the formation of a metal-phenolic coordination network (Fig. S7) [40]. The appearance of S 2p peaks at 162.0 eV (S 2p_3/2) and 163.55 eV (S 2p_1/2) substantiated the successful loading of LF3 (Fig. 1i). Additionally, the pH-responsive release profiles of LCM@Cap were also evaluated. As shown in Fig. 1j, approximately 64.8% of Cu²⁺ was released within 36 h at pH 5.5, compared to less than 30% at pH 7.4. Similarly, the cumulative release of Mn²⁺ reached 42.4% at pH 5.5, significantly higher than the 12.7% release at pH 7.4, demonstrating the pH-responsive nature of the metal-phenolic nanocapsules.

2.2. Targeting properties of the nanocapsules

To achieve active tumor targeting, HA was conjugated to the surface of LCM@Cap (denoted as HLCM@Cap) to facilitate binding to CD44 receptors, which are overexpressed on tumor cells [41,42]. The successful surface modification was confirmed by confocal laser scanning microscopy (CLSM) colocalization analysis using Cy5.5-labeled HA and FITC-labeled LCM@Cap (Fig. S8), as well as by the detectable change in the ζ-potential of the nanocapsules (Fig. 1f). The conjugation of HA improved the dispersibility of the capsules and enhanced hydrophilicity, resulting in a more regular morphology for HLCM@Cap (Fig. 1c). Importantly, the HA conjugation did not significantly alter the hydrodynamic size (183.1 nm, Fig. 1e) and other characteristics of the nanocapsules (Figs. S4, S5 and S9). Quantitative characterization of HLCM@Cap revealed loading contents of 5.84–6.13% for LF3, 6.22–6.46% for Cu²⁺, 5.57–6.20% for Mn²⁺, and 0.83–1.05% for HA. These loading levels met the requirements for subsequent experiments (Figs. S10 and S11).

To systematically evaluate the tumor-targeting ability of the nanocapsules, cellular uptake assays were first performed. A time-dependent comparison between FITC-labeled LCM@Cap and FITC-labeled HLCM@Cap revealed that the uptake of HLCM@Cap by 4T1 cells was significantly higher than that of unmodified LCM@Cap. This enhanced uptake was confirmed by both flow cytometry (Fig. 2a) and fluorescence imaging (Fig. 2b). Further evidence of efficient intracellular delivery was provided by live-cell fluorescence imaging (Fig. S12) and MR imaging (Fig. S13). As shown in Fig. S14, HLCM@Cap is progressively internalized by 4T1 cells and subsequently escapes from lysosomes. This feature helps prevent intracellular degradation, maintain effective intracellular concentrations, and support its intended biological function [43].

Fig. 2.

Targeting performance and in vivo imaging evaluation of nanocapsules. (a) Flow cytometry analysis of cellular uptake of LCM@Cap and HLCM@Cap by 4T1 cells. (b) Representative fluorescence images of 4T1 cells after 12 h incubation with FITC-labeled LCM@Cap and HLCM@Cap. Scale bar: 50 μm. (c) IVIS fluorescence images of 4T1 tumor-bearing mice before and after intravenous administration of Cy5.5-labeled BSA (control), LCM@Cap, or HLCM@Cap. The dashed circles indicate tumor locations. (d) Mean change in tumor fluorescence intensity relative to 0 h. (e) Ex vivo IVIS images of major organs (heart, liver, spleen, lung and kidney) and tumors harvested 24 h post-injection. (f) T₁-weighted MR imaging of nanocapsules at different concentrations. (g) Corresponding T₁ mapping images of the nanocapsules. (h) Longitudinal relaxation rate (T₁⁻¹) of nanocapsules as a function of Mn²⁺ concentration. (i) Schematic illustration of the experimental setup for MR scanning. (j) In vivo T₁-weighted MR images of a 4T1 tumor-bearing mouse before and after HLCM@Cap administration. The dashed circles indicate the tumor location. (k) Quantification of relative T₁ signal intensity from the tumor region in five experimental mice. Data are presented as mean ± SD. Statistical significance was determined by the Student's t-test. (∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

In vivo tumor-targeting capability was subsequently assessed using the In Vivo Imaging System (IVIS). The results demonstrated that Cy5.5-labeled LCM@Cap accumulated in tumor tissue through the enhanced permeability and retention (EPR) effect, with fluorescence intensity peaking at 1–2 h post-injection (Fig. 2c and d). The Cy5.5-labeled HLCM@Cap exhibited markedly enhanced tumor-targeting capability. This was evidenced by a more rapid increase in fluorescence intensity, a higher peak signal at approximately 4 h, and substantially stronger fluorescence accumulation at the tumor site at both 24 h (Fig. 2e, Fig. S15) and 48 h (Fig. S16) post-injection. These findings indicate that HLCM@Cap possesses superior tumor-targeting specificity, accumulation efficiency, and tumor retention time. Complementing these imaging results, ICP-MS analysis performed 12 h post-injection revealed significantly higher Cu accumulation in tumors of the HLCM@Cap group, while the hepatic Cu levels were reduced relative to those in the LCM@Cap group (Fig. S17). These results indicate enhanced tumor targeting efficacy and improved biosafety profile of the HA-modified nanocapsules.

To further verify the tumor-targeting specificity and MR imaging potential of HLCM@Cap, we evaluated its T₁-weighted imaging performance and the underlying mechanism. The comparison of T₁-weighted imaging of the nanocapsules at different concentrations showed that HLC@Cap exhibited no noticeable T₁ contrast, while the addition of Mn in HLCM@Cap resulted in a significant enhancement of T₁ contrast (Fig. 2f). This indicates that Mn²⁺ ions play a key role in T₁-weighted imaging. Further T₁ mapping analysis revealed that higher Mn²⁺ concentrations were associated with lower T₁ values (Fig. 2g), revealing a strong linear relationship between the longitudinal relaxation rate (T₁⁻¹) and Mn²⁺ concentration. This was attributed to the higher Mn content, which strengthened paramagnetism, shortened the T₁ relaxation time of surrounding water protons, and ultimately enhanced T₁-weighted MR signals. Under pH 5.5 conditions, the longitudinal relaxivity (r₁) of HLCM@Cap was 21.70, significantly higher than the r₁ of 8.89 under neutral conditions (Fig. 2h). This enhancement was mainly attributed to accelerated nanocapsule decomposition under acidic conditions, which promoted Mn²⁺ release and exposure of active sites. The liberated Mn²⁺ interacted more efficiently with water molecules via enhanced hydration and inner-sphere water exchange, thereby improving T₁ relaxivity [44,45]. Subsequently, the T₂-weighted MR imaging performance of HLCM@Cap was further evaluated (Fig. S18). The calculated r₂/r₁ value at pH 5.5 was 1.70, confirming that HLCM@Cap is an effective T₁ MR contrast agent. The above findings indicate that the Mn-based nanocapsules exhibit favorable T₁ contrast in the acidic TME, with the contrast further enhanced by intratumoral accumulation of the nanocapsules and subsequent Mn²⁺ release. Such imaging properties enable evaluation of drug uptake and monitoring of therapeutic outcomes.

To evaluate in vivo tumor MR imaging performance, mice were scanned at different time intervals following tail vein injection of HLCM@Cap (Fig. 2i). In the tumor region, particularly within highly vascularized areas, an initial rapid signal enhancement was observed, followed by a washout phase. As the nanocapsules accumulated within the tumor and released Mn²⁺ ions, the T₁ signal progressively increased, reaching a peak approximately 8 h post-injection (Fig. 2j and k, Fig. S19). Notably, this MR signal peak was observed later than the IVIS fluorescence peak at 4 h. This delay can be attributed to the fact that MR contrast enhancement depends on the effective release and subsequent hydration of Mn²⁺ ions, which is a time-consuming process. Importantly, Mn²⁺-mediated T₁-weighted imaging enables visualization of the nanocapsule targeting process, reflecting intratumoral drug release and accumulation, thereby providing preliminary guidance for optimizing the administration window of αPD-L1 to achieve maximal synergistic therapeutic efficacy. Collectively, these findings confirm the effective tumor targeting properties of HLCM@Cap and highlight its potential for imaging-guided cancer therapy.

2.3. Cuproptosis-enhancing effect and metabolic regulatory mechanism of HLCM@Cap

Cuproptosis is a distinct form of regulated cell death driven by copper-induced proteotoxic stress within mitochondria (Fig. 3a). Fig. S17 has confirmed that the enhanced tumor-targeting capability of HLCM@Cap enables sufficient intratumoral copper accumulation. To elucidate the contribution of individual components within the nanocapsules to cuproptosis activation, 4T1 cells were divided into control, HMn@Cap, HCu@Cap, HLC@Cap, and HLCM@Cap groups. Based on CCK-8 assays, a copper concentration of 1.5 μg mL⁻¹ was selected for subsequent experiments, as it induced pronounced cytotoxicity in 4T1 cells (Fig. S20) while preserving the viability of 3T3 cells (Fig. S21), ensuring a tumor-selective therapeutic window. Then, Western blot analysis of key cuproptosis-associated proteins revealed marked downregulation of Fe-S cluster proteins, including ACO2, LIAS, and FDX1, in the HCu@Cap, HLC@Cap, and HLCM@Cap groups (Fig. 3b and c), collectively confirming the induction of cuproptosis. Notably, nanocapsules loaded with LF3 and Mn²⁺ ions, particularly HLCM@Cap, elicited a more pronounced cuproptosis response than HCu@Cap, indicating that LF3-mediated Wnt/β-catenin inhibition and Mn-assisted signaling modulation synergistically amplify the efficiency of cuproptosis.

Fig. 3.

Potentiation of cuproptosis induced by HLCM@Cap. (a) Schematic illustration of the mechanism by which HLCM@Cap enhances cuproptosis. (b) Western blot analysis of cuproptosis-related proteins. (c) Quantification of the protein expression levels from panel b. (d) Western blot analysis of proteins associated with Wnt/β-catenin pathway (ATP7B, β-catenin, p-AKT, Cyclin D1). (e) Western blot analysis of proteins associated with glycolysis pathway (HK2, LDHA). (f) Quantitative PCR analysis of mRNA expression levels of ATP7B, β-catenin and Cyclin D1. (g) Quantitative PCR analysis of mRNA expression levels of HK2 and LDHA. (h) Quantification of cellular glucose uptake, reflected by 2-DG6P production, after different treatments. (i) Quantification of cellular ATP production after different treatments. (j) Quantification of lactate production after different treatments. (k) Representative TEM images showing mitochondrial ultrastructural changes in 4T1 cells. (l) Fluorescence images of MPTP in 4T1 cells after different treatments. Scale bar: 50 μm. (m) JC-1 fluorescence staining revealing changes in mitochondrial membrane potential of 4T1 cells after different treatments. Scale bar: 50 μm. Data are presented as mean ± SD. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

To further dissect the resistance mechanisms that constrain cuproptosis efficacy, adaptive cellular responses to copper overload were investigated. The excess intracellular copper activates the Wnt/β-catenin pathway through PDK1 binding [17] and simultaneously disrupts Fe-S cluster proteins, impairing mitochondrial respiration and driving compensatory glycolytic reprogramming (Fig. 3a). Consistent with this hypothesis, HCu@Cap group exhibited elevated expression of Wnt/β-catenin pathway components (β-catenin, p-AKT, Cyclin D1), glycolysis-related markers (HK2, LDHA, HIF-1α), and the copper efflux transporter ATP7B (Fig. 3d–g, Figs. S22 and S23), indicating activation of copper efflux and metabolic plasticity to mitigate copper-induced cytotoxic stress. In contrast, LF3-containing nanocapsules (HLC@Cap and HLCM@Cap) effectively suppressed Wnt/β-catenin signaling, as evidenced by reduced expression of related genes and proteins, together with downregulation of Cyclin D1 (Fig. 3d, Fig. S23a–c). Downregulation of Cyclin D1 can also induce G1 phase cell cycle arrest and suppress tumor cell proliferation. Meanwhile, ATP7B gene (Fig. 3f) and protein (Fig. S23d) expression decreased, leading to limited copper efflux and sustained intracellular copper accumulation. Moreover, the combined action of LF3 and Mn²⁺ profoundly inhibited glycolytic metabolism, as reflected by decreased expression of glycolysis-related genes (Fig. 3g, Fig. S22) and proteins (Fig. S23e and f), reduced glucose uptake (Fig. 3h), decreased ATP production (Fig. 3i) and lower lactate release (Fig. 3j). The inhibition of glycolysis limits the metabolic reprogramming ability of tumor cells, increases cellular stress, and further accelerates the cuproptosis process. Importantly, inhibition of tumor glycolysis alleviates the metabolic competition imposed on immune cells, thereby enhancing antitumor immunity [46]. These findings suggest that HLCM@Cap does not merely induce cuproptosis, but fundamentally disrupts the adaptive response that tumor cells employ to escape cuproptosis, thereby establishing a self-amplifying positive feedback loop.

Mitochondrial structural and functional alterations further substantiated effective cuproptosis induction. TEM images revealed that mitochondria in untreated 4T1 cells exhibited a short rod-like morphology, whereas those in the HLCM@Cap-treated group were markedly swollen, showed increased membrane electron density, and displayed blurred or disrupted cristae (Fig. 3k). Detection of mitochondrial permeability transition pore (MPTP) activity showed a significant increase in mitochondrial membrane permeability (Fig. 3l), and JC-1 staining indicated a clear drop in mitochondrial membrane potential (Fig. 3m, Fig. S24). These results demonstrate that HLCM@Cap effectively disrupts copper homeostasis and metabolic adaptability, induces severe mitochondrial dysfunction, and robustly drives cuproptotic tumor cell death.

2.4. HLCM@Cap synergistically activates the cGAS-STING pathway

As described above, HLCM@Cap-induced cuproptosis causes severe mitochondrial damage and membrane rupture. The subsequent damaged mtDNA release, which is recognized by cGAS and thereby activates the STING pathway (Fig. 4a). To verify the mitochondrial origin of the damaged DNA, 8-hydroxy-2-deoxyguanosine (8-OHdG) staining was performed in combination with the mitochondrial marker TOMM20. The pronounced co-localization between 8-OHdG and TOMM20 confirmed that the majority of damaged DNA originated from mitochondria (Fig. 4b). Subsequent mtDNA-specific staining further demonstrated its effective release into the cytosol following HLCM@Cap treatment. Structured illumination microscopy (SIM) images revealed distinct mitochondrial dynamic responses to different levels of HLCM@Cap stimulation (Fig. 4c). In untreated 4T1 cells, mitochondria exhibited a short rod-like morphology, consistent with the high metabolic demand and proliferative capacity of aggressive tumor cells. Low-dose HLCM@Cap treatment induced mitochondrial elongation and fusion, indicative of an adaptive stress response that limits rapid cell division. In contrast, high-dose HLCM@Cap treatment resulted in progressive mitochondrial swelling, cristae disruption, membrane rupture, and extensive mtDNA release, ultimately leading to mitochondrial vacuolization. These observations suggest a dose-dependent mitochondrial stress threshold, beyond which adaptive responses collapse and irreversible mitochondrial damage ensues. Consistently, gene set enrichment analysis (GSEA) revealed significant enrichment of the cytosolic DNA-sensing pathway (Fig. S25).

Fig. 4.

Cuproptosis synergistically activates the cGAS-STING pathway via mitochondrial DNA release. (a) Schematic illustration of cuproptosis-mediated activation of the cGAS-STING pathway activation and subsequent immune responses. (b) Immunofluorescence staining of oxidatively damaged DNA (8-OHdG, green) in 4T1 cells. Mitochondria and nuclei were stained with TOMM20 (red) and DAPI (blue), respectively. Scale bar: 50 μm. (c) SIM fluorescence images showing mtDNA release within 4T1 cells. Mitochondria (red) were labeled with MitoTracker, mtDNA (green) with SYBR Gold, and nuclei (blue) with Hoechst 33342. Scale bar: 10 μm. (d) Western blot analysis of key proteins in the cGAS-STING pathway. (e–h) Quantitative analysis of protein expression levels from panel d. (i,j) ELISA quantification of extracellular IFN-β (i) and TNF-α (j) secretion by 4T1 cells at 24 h post-treatments. (k) Western blot: the expression of key proteins associated with cuproptosis and glycolysis after H-151 treatment. (l) Volcano plot of DEGs between the HLCM@Cap group and the control group. (m) KEGG pathway enrichment analysis of DEGs. (n) Heatmap analysis of DEGs. Data are presented as mean ± SD. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test (∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001).

PCR and Western blot analyses demonstrated that cuproptosis-induced mtDNA release, together with Mn²⁺ released from the nanocapsules, synergistically activated cGAS-STING signaling. This was evidenced by increased cGAS expression at both the mRNA (Fig. S26) and protein levels (Fig. 4d and e), along with elevated phosphorylation of key downstream signaling proteins, including p-TBK1 (Fig. 4f), p-IRF3 (Fig. 4g), and p-STING (Fig. 4h). Fig. S27 further demonstrated that Mn-mediated activation of the cGAS-STING pathway was mainly achieved by enhancing the sensitivity of damaged DNA sensing. After the addition of DNase I, the activation of the cGAS-STING pathway induced by HMn@Cap was markedly attenuated and showed no significant difference compared with the control group. Notably, this dual-input mechanism integrates mtDNA as a damage-associated molecular pattern and Mn²⁺ as a signal amplifier, thereby enabling robust and sustained STING activation. Correspondingly, Enzyme-linked immunosorbent assay (ELISA) revealed markedly increased secretion of IFN-β (Fig. 4i) and TNF-α (Fig. 4j), indicating activation of the inflammatory response and remodeling of the tumor immune microenvironment.

Activation of the cGAS-STING pathway could enhance cuproptosis sensitivity by suppressing glycolytic metabolism in tumor cells, thereby forming a positive-feedback loop within the cuproptosis-STING cascade. This hypothesis was further supported by the H151 inhibition experiments. As shown in Fig. 4k and Fig. S28, HLCM@Cap treatment markedly downregulated the expression of glycolysis-related proteins, induced oligomerization of the cuproptosis-associated protein DLAT, and reduced the expression of the Fe-S cluster proteins FDX1. These results suggested that HLCM@Cap effectively inhibited glycolysis and promoted cuproptosis. However, after treatment with the cGAS-STING pathway inhibitor H151, both glycolytic suppression and cuproptosis induction were significantly attenuated. Collectively, these findings further demonstrate that cGAS-STING activation is involved in metabolic regulation and contributes to enhanced cuproptosis sensitivity.

To further elucidate the potential therapeutic mechanism of HLCM@Cap, transcriptomic analysis of 4T1 cells after HLCM@Cap treatment was performed. Differential gene expression analysis revealed 4474 significantly altered genes (|log (FC)| ≥1 and Q value < 0.05) (Fig. 4l). KEGG pathway enrichment analysis revealed coordinated modulation of cancer-associated signaling pathways, cell fate regulation, metabolic reprogramming, and immune and inflammatory responses (Fig. 4m). In particular, enrichment of cytosolic DNA sensing, T cell receptor signaling, and the PD-1/PD-L1 immune checkpoint pathway provides a mechanistic basis for the observed synergy between HLCM@Cap and immune checkpoint blockade. A heatmap of differentially expressed genes (DEGs) illustrated a pronounced transcriptional reprogramming induced by HLCM@Cap, with stress and immune related gene clusters being strongly enriched, whereas genes associated with proliferation, invasion, and metabolic fitness were markedly suppressed (Fig. 4n). The coordinated enrichment of these pathways indicates that HLCM@Cap exerts antitumor activity through a multi-level mechanism involving mitochondrial and metabolic stress, induction of programmed cell death, suppression of oncogenic growth signaling, and remodeling of the tumor immune microenvironment.

These transcriptomic findings were independently confirmed by flow cytometry and fluorescence staining. SIM fluorescence imaging confirmed enhanced mitochondrial swelling and reactive oxygen species (ROS) generation (Fig. 5a), and flow cytometry analysis revealed a rightward shift in the intracellular ROS peak (Fig. 5b and c), indicating a pronounced oxidative stress response. This effect was further amplified by metal-catalyzed Fenton-like reactions involving released Cu and Mn ions [47,48], which lead to the generation of highly reactive hydroxyl radicals (•OH) (Fig. S29), thereby exacerbating oxidative stress and cellular damage in tumor cells. The combined effects of cuproptosis, STING pathway activation, and ROS generation further induced DNA damage (Fig. 5d and e) and apoptosis (Fig. 5f), collectively contributing to ICD. This was evidenced by robust calreticulin (CRT) exposure on the tumor cell surface (Fig. 5g and h). CRT functions as an “eat-me” signal and binds specifically to LDL receptor-related protein-1 (LRP1) on DCs, facilitating tumor antigen uptake and processing [49]. In addition, HMGB1 translocated from the nucleus to the cytoplasm (Fig. 5i), while intracellular ATP levels decreased (Fig. 3i) and extracellular ATP release increased (Fig. S30), further confirming the induction of ICD. GSEA also demonstrated a pronounced enrichment of the antigen processing and presentation pathway (Fig. S31).

Fig. 5.

HLCM@Cap activates the cGAS-STING pathway, inducing cell damage and promoting BMDC maturation. (a) Fluorescence images of mitochondrial ROS in 4T1 cells detected using MitoSOX Red. Mitochondria were stained with MitoTracker (green) and nuclei were counterstained with Hoechst 33342 (blue). Scale bar: 10 μm. (b) Representative flow cytometry plots of DCFH-DA staining for intracellular ROS detection. (c) Flow cytometric quantification of intracellular ROS levels after different treatments. (d,e) Representative immunofluorescence images and corresponding quantitative analysis of γH2AX expression in 4T1 cells subjected to different treatments. Scale bar: 50 μm. (f) Representative flow cytometry plots of Annexin V-FITC/propidium iodide (PI) staining for apoptosis analysis in 4T1 cells following the indicated treatments. (g,h) Representative immunofluorescence images and corresponding quantitative analysis of CRT exposure in 4T1 cells after different treatments. Scale bar: 50 μm. (i) Representative immunofluorescence images of HMGB1 in 4T1 cells after different treatments. Scale bar: 50 μm. (j) Schematic illustration of co-culture of BMDCs with HLCM@Cap-treated 4T1 cells. (k) Flow cytometric analysis of BMDC maturation after co-culture with treated 4T1 cells. Data are presented as mean ± SD. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test (∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Finally, to evaluate whether STING activation promotes DC maturation, BMDCs were co-cultured in vitro with tumor cells treated by the metal-phenolic nanocapsules (Fig. 5j). Flow cytometry analysis revealed a markedly increased proportion of CD80⁺/CD86⁺ DCs in the HLCM@Cap group (Fig. 5k, Fig. S32), indicating enhanced DC maturation. Together, these results demonstrate that HLCM@Cap triggers DC activation through a coordinated mechanism, thereby providing a supportive basis for subsequent adaptive antitumor immune responses.

2.5. In vivo synergistic antitumor efficacy

Given the importance of biosafety for in vivo applications and clinical translation, systemic toxicity should be carefully evaluated [50]. Serum TNF-α, IL-6, and IFN-β levels were slightly elevated at 72 h post-administration but showed no significant differences compared with the control group (Fig. S33), indicating no obvious systemic inflammatory toxicity. Blood biochemical parameters related to liver and kidney function (ALT, AST, BUN, CREA) showed no significant differences between metal-phenolic nanocapsules treatment and control groups (Fig. S34), and body weights remained stable throughout the treatment period (Fig. S35). Hematoxylin and eosin (H&E) staining of major organs revealed no noticeable pathological abnormalities (Fig. S36), confirming good biocompatibility. Importantly, blood Cu and Mn levels reached peak concentrations at 1 h post-injection and gradually decreased thereafter, with a marked decline observed within 24 h (Fig. S37). In addition, fecal Cu analysis demonstrated efficient post-treatment excretion, with peak Cu levels observed at 24 h followed by gradual clearance (Fig. S38). This favorable metabolic profile reduces the risk of long-term metal accumulation and supports the preliminary biosafety and translational promise of HLCM@Cap as a nanotherapeutic platform.

Inspired by the strong tumor-targeting and cytotoxic effects of HLCM@Cap observed in vitro, a 4T1 TNBC tumor-bearing mouse model was established to evaluate its in vivo therapeutic efficacy and synergistic antitumor effects in combination with αPD-L1 immunotherapy. Briefly, 2 × 10⁶ 4T1 cells were inoculated into the right back of BALB/c mice. Once tumor volumes reached approximately 60 mm³, tumor-bearing mice were randomly divided into six groups (n = 5): control (5% glucose solution), HMn@Cap, HCu@Cap, HLC@Cap, HLCM@Cap and HLCM@Cap+αPD-L1. MR imaging features preliminarily demonstrated efficient intratumoral accumulation and release of nanocapsules within the 8–24 h window (Fig. 2j), which may facilitate immune pathway activation and immune cell recruitment, thereby providing preliminary guidance for the dosing interval between HLCM@Cap and αPD-L1. Nanocapsules (100 μL, 5 mg kg⁻¹) were administered intravenously on days 1, 4, 7, and 10, while αPD-L1 (100 μL, 10 mg kg⁻¹) was administered intraperitoneally on days 2, 5, 8, and 11 (<24 h interval). Mice were sacrificed on day 15 for tumor collection and subsequent analyses (Fig. 6a).

Fig. 6.

In vivo antitumor efficacy of different treatments. (a) Schematic illustration of the experimental design for evaluating antitumor efficacy. (b) Representative photographs of excised tumors from each group. (c,d) Tumor volume growth curves (c) and tumor weights (d) after treatment with 5% glucose solution, HMn@Cap, HCu@Cap, HLC@Cap, HLCM@Cap and HLCM@Cap+αPD-L1 (n = 5). (e) Representative images of tumor tissue sections stained with H&E, TUNEL, and Ki67. Scale bars: 100 μm. (f) Immunofluorescence staining of LIAS in tumor tissues. Scale bar: 1 mm. (g–i) Quantitative analysis of TUNEL (g), Ki67 (h), and LIAS (i) expression in panels e and f. Data are presented as mean ± SD. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test (∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Dynamic tumor volume monitoring (Fig. 6b and c) showed rapid tumor progression in the control group. HMn@Cap and HCu@Cap groups exhibited moderate tumor inhibitory. HLC@Cap and HLCM@Cap groups exhibited substantially enhanced antitumor efficacy, whereas HLCM@Cap+αPD-L1 group showing the most potent tumor growth suppression. At the experimental endpoint, the tumor volumes were 1210.28 ± 185.50 mm³ in the control group, 1054.76 ± 170.16 mm³ in the HMn@Cap group, 743.40 ± 97.16 mm³ in the HCu@Cap group, 354.10 ± 97.26 mm³ in the HLC@Cap group, 250.38 ± 48.54 mm³ in the HLCM@Cap group, and 122.68 ± 43.48 mm³ in the HLCM@Cap+αPD-L1 group. Statistical analysis indicated significant differences among the groups (Fig. 6c, P < 0.05). Tumor weight measurements further confirmed this trend, as the HLCM@Cap+αPD-L1 group exhibited the lowest tumor weight (0.30 ± 0.09 g), significantly lower than the control (1.14 ± 0.15 g), HMn@Cap (0.97 ± 0.10 g), HCu@Cap (0.88 ± 0.07 g), HLC@Cap (0.65 ± 0.09 g) and HLCM@Cap (0.56 ± 0.07 g) groups (Fig. 6d). These results highlight the essential role of multicomponent synergy within HLCM@Cap, integrating copper overload, metabolic suppression, and immune activation, and further demonstrate that HLCM@Cap effectively sensitizes tumors to immune checkpoint blockade.

Histopathological and immunohistochemistry analyses further clarified the mechanism of action (Fig. 6e). Significant tumor necrosis was observed in the HLCM@Cap and HLCM@Cap+αPD-L1 groups by H&E staining. Consistently, tumors treated with HLCM@Cap combined with αPD-L1 exhibited significantly increased TUNEL-positive apoptotic cells (Fig. 6e and g) and reduced Ki67-positive proliferating cells (Fig. 6e and h), indicating effective suppression of tumor proliferation and promotion of tumor cell death. To verify the induction of cuproptosis in vivo, immunohistochemical staining was performed to assess the expression of the Fe-S cluster protein LIAS, a key marker of cuproptosis. LIAS expression showed a stepwise decrease across the HCu@Cap, HLC@Cap, and HLCM@Cap groups (Fig. 6f and i), indicating progressively enhanced cuproptosis and confirming a cooperative effect among the components of the nanocapsules.

2.6. HLCM@Cap combined with αPD-L1 remodels the tumor immune microenvironment

In vitro studies have demonstrated that HLCM@Cap effectively activate the cGAS-STING pathway and promote DC maturation (Fig. 5j). Subsequently, mature DCs enhance tumor antigen presentation and promote CD8⁺ T cell activation, leading to effective tumor cell killing (Fig. 7a). Based on this immune-activating capability, we hypothesized that combining HLCM@Cap with the immune checkpoint inhibitor αPD-L1 would further relieve immune evasion, enhance antigen recognition, and improve the immunosuppressive TME.

Fig. 7.

HLCM@Cap combined with αPD-L1 blockade induces immune activation in vivo. (a) Schematic illustration of tumor ICD-induced activation of the cGAS-STING pathway and subsequent immune responses. (b) Immunofluorescence staining of CRT and immunohistochemical staining of p-STING in tumor tissues. Scale bars: 100 μm. (c) Representative flow cytometry plots of tumor-infiltrating CD8⁺ T cells in 4T1 tumor tissues. (d) Quantification of tumor-infiltrating CD8⁺ T cells in 4T1 tumors. (e) Representative flow cytometry plots of T_reg cells (CD25⁺Foxp3⁺) in 4T1 tumor tissues. (f) Quantification of tumor-infiltrating T_reg cells in 4T1 tumors. (g) Representative flow cytometry plots of MDSCs (CD11b⁺Ly6G⁺) in 4T1 tumor tissues. (h) Quantification of tumor-infiltrating MDSCs in 4T1 tumors. Data are presented as mean ± SD. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test (∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Histological analysis of tumor tissues confirmed effective induction of ICD, as evidenced by increased CRT exposure (Fig. 7b). Meanwhile, the level of p-STING was markedly increased, suggesting activation of the STING signaling pathway. Notably, among all treatment groups, HLCM@Cap combined with αPD-L1 elicited the strongest activation of immune-related indicators, demonstrating the superior immunotherapeutic efficacy of the combination strategy.

To quantify immune activation within the TME, tumor-associated immune cell subsets were analyzed using flow cytometry. The proportion of mature DCs in both the tumors (Fig. S39) and tumor-draining lymph nodes (Fig. S40) was significantly increased following HLCM@Cap treatment, particularly in combination with αPD-L1. HLCM@Cap treatment, especially when combined with αPD-L1, significantly increased the proportion of mature DCs in TNBC-bearing mice compared with the control and monotherapy groups.

As DC maturation directly supports T cell priming, tumor-infiltrating T cell populations were further examined. CD8⁺ T cells, known as cytotoxic T lymphocytes (CTLs), are key effector cells mediating tumor-specific immunity. Flow cytometry results showed that HLCM@Cap+αPD-L1 significantly increased the infiltration of CD8⁺ cytotoxic T lymphocytes in tumors (Fig. 7c and d). In parallel, the proportion of tumor-infiltrating immunosuppressive regulatory T cells (T_reg, defined as CD25⁺Foxp3⁺ cells) was markedly reduced (Fig. 7e and f), and myeloid-derived suppressor cells (MDSCs, defined as CD11b⁺Ly6G⁺ cells) also displayed a downward trend (Fig. 7g and h). Given that immunosuppressive tumors respond poorly to immune checkpoint blockade due to limited immune cell infiltration [51,52], these results demonstrate that HLCM@Cap effectively remodels the tumor immune microenvironment by inducing ICD and activating cGAS-STING signaling, thereby converting immunologically “cold” tumors into “hot” tumors and sensitizing them to αPD-L1 immunotherapy. To investigate whether the mice developed immune memory after treatment, flow cytometry analysis was performed on tumor and spleen tissues to quantitatively assess CD8⁺ effector memory T cells (T_em, CD44⁺CD62L⁻) and CD8⁺ central memory T cells (T_cm, CD44⁺CD62L⁺). Following HLCM@Cap treatment, the proportions of both CD8⁺ Tem and CD8⁺ Tcm cells were markedly increased, with a more pronounced enhancement observed when combined with αPD-L1 therapy. This was reflected by a significant increase in T_em cells within the tumor (Fig. S41) and T_cm cells within the spleen (Fig. S42). The expansion of these memory T cell subsets is critical for sustained immune surveillance, enabling rapid recognition and elimination of disseminated tumor cells. Collectively, these results demonstrate that HLCM@Cap synergized with αPD-L1 not only suppresses primary tumor growth but also establishes durable antitumor immune memory, thereby providing long-term protection against recurrence and metastasis.

2.7. In vivo inhibition of tumor metastasis

Lung metastasis represents a major route of distant dissemination in malignant tumors and is closely associated with epithelial-mesenchymal transition (EMT), TME remodeling, and impaired immune surveillance [53]. TNBC is particularly prone to lung metastasis [54]. Our results demonstrated that combination therapy of HLCM@Cap and αPD-L1 markedly suppressed primary TNBC tumor growth and enhanced antitumor immunity through TME modulation. To further evaluate the inhibitory effect of the combination therapy on distant metastatic progression, a 4T1 lung metastasis mouse model was established. Tumor-bearing mice were randomly divided into four groups: control, αPD-L1 monotherapy, HLCM@Cap monotherapy, and HLCM@Cap+αPD-L1 combination therapy, and the observation period was extended to 30 days (Fig. 8a).

Fig. 8.

In vivo antitumor metastatic effect. (a) Schematic illustration of the experimental protocol for assessing the inhibition of lung metastasis formation in vivo. (b) Representative photographs of lung metastatic nodules and corresponding panoramic and magnified H&E-stained images of lung metastatic lesions. Scale bar: 500 μm. (c) Quantification of metastatic tumor nodules in the lungs (n = 5). (d) Survival curves of mice receiving different treatments. Data are presented as mean ± SD. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test (∗∗p < 0.01, ∗∗∗∗p < 0.0001).

At the endpoint, lung tissues were harvested, fixed in Bouin's solution and subjected to H&E staining (Fig. 8b). Metastatic nodules were quantified in a blinded manner, and survival was recorded. The combination group exhibited significantly fewer and smaller metastatic nodules (3.8 ± 2.05) compared with the control (68.0 ± 7.25), αPD-L1 (37.4 ± 7.40), and HLCM@Cap (17.2 ± 4.15) groups (Fig. 8c). In addition, combination treatment markedly prolonged survival (Fig. 8d). These findings demonstrate that HLCM@Cap combined with αPD-L1 effectively suppresses metastatic progression. This effect is likely associated with enhanced systemic antitumor immunity and the establishment of durable immune memory.

3. Conclusion

In summary, we developed multifunctional LF3-loaded metal-phenolic nanocapsules (HLCM@Cap) that enable tumor-targeted delivery and pH-responsive release, while integrating T₁-weighted MR imaging with immunometabolic therapy. Through a synergistic multi-mechanistic design, this nanoplatform promotes intracellular copper accumulation and metabolic stress, thereby establishing a self-amplifying cuproptosis–STING cascade. By coordinately modulating copper metabolism and tumor metabolic adaptation, HLCM@Cap effectively amplifies innate and adaptive immune responses, promotes TME remodeling, and induces durable antitumor immune memory. In addition, MR imaging capacity enables real-time monitoring of drug delivery and provides preliminary guidance for optimizing the administration window of αPD-L1. The combination therapy of HLCM@Cap and αPD-L1 further improves the therapeutic efficacy of synergistic cancer immunotherapy, effectively suppressing primary tumor growth and distant metastasis. Overall, this work establishes a mechanism-driven strategy that couples cuproptosis with innate immune activation and highlights a generalizable approach for overcoming immune resistance in solid tumors.

4. Materials and methods

4.1. Materials

Tannic acid (TA) was purchased from Sigma-Aldrich (USA). Zn (NO₃)₂·6H₂O and CuSO₄∙5H₂O were bought from Sinopharm Chemical Reagent Co. Ltd. (China). 2-Methylimidazole (2-MIM) was purchased from Beijing Jianqiang Weiye Technology Co., Ltd (China). 8-arm-PEG_40k-OH was purchased from SINOPEG (China). MnCl₂·4H₂O was purchased from Beijing J&K Technology Co., Ltd (China). 3-Morpholinopropanesulfoinc Acid (MOPS) and LF3 were purchased from Shanghai Aladdin Technology Co., Ltd (China). Ethylenediaminetetraacetic acid disodium salt (EDTA-2Na) was purchased from Shanghai Yuanye Bio-Technology Co., Ltd (China). Tris (hydroxymethyl) aminomethane (Tris-HCl) was purchased from Shanghai Titan Technology Co., Ltd (China). Polyetherimide (PEI) was purchased from Sigma-Aldrich (Shanghai) Trading Co., Ltd. Hyaluronic Acid (HA) was purchased from Beijing Mreda Technology Co., Ltd. Anti-mouse programmed death ligand 1 (aPD-L1) was obtained from Starter-Bio (S0B0593, Hangzhou, CN).

4.2. Preparation of metal-phenolic nanocapsules

Preparation of ZIF-8 NPs: ZIF-8 NPs with an average diameter of 150 nm were synthesized using PEG as a mineralizer. Briefly, 10 mg of 8-arm-PEG-OH (40 kDa) was completely dissolved in 3 mL of a 2-MIM solution (13.1 mg mL⁻¹), followed by mixing with 1 mL of a Zn (NO₃)₂ solution (11.9 mg mL⁻¹) and stirring for 40 min. The resulting ZIF-8 NPs were washed three times with water by centrifugation (15000g, 5 min).

Preparation of metal-phenolic nanocapsules: metal-phenolic nanocapsules were prepared by a previously reported method with slight modification. The steps to obtain LCM@Cap are as follows, CuSO₄∙5H₂O solution (10 mmol L⁻¹, 0.5 mL), MnCl₂·4H₂O (10 mmol L⁻¹, 0.5 mL) and TA solution (3.53 mmol L⁻¹, 1 mL) were mixed vigorously with a vortex mixer. The resulting mixture of TA and Cu²⁺/Mn²⁺ was centrifuged and the precipitate in the solution was removed by a syringe filter with a pore size of 0.22 μm. The supernatant was added to the suspension of ZIF-8 NPs (4 mg of ZIF-8 NPs in 300 μL of water), followed by vigorous stirring for 30 min. Subsequently, the pellets were washed three times with 1 mL of water via centrifugation to remove the excess TA and Cu²⁺/Mn²⁺. Tris–HCl buffer (1 mL, pH 8.0, 50 mmol L⁻¹) was added to the suspension of ZIF-8 NPs to adjust the pH, followed by a brief sonication after each step. To obtain metal-phenolic nanocapsule, the pellets were washed with 1 mL of EDTA solution (pH 8.0, 50 mmol L⁻¹) and water three times by centrifugation. To load LF3, the obtained nanocapsules were dispersed in 400 μL of aqueous solution, followed by the addition of 100 μL of LF3 (5 mg mL⁻¹ in DMSO). After reacting for 4 h, the mixture was centrifuged and washed three times to obtain LCM@Cap. Other metal-phenolic nanocapsule were prepared according to the above method.

Preparation of HA-modified nanocapsules: to prepare HLCM@Cap, LCM@Cap was dispersed in 1 mL of aqueous solution, PEI was added and reacted for 4 h (PLCM@Cap), then centrifuged and washed to remove free PEI. HA was then added and reacted for 4 h to obtain HLCM@Cap. Other HA-modified nanocapsule were prepared according to the above method.

4.3. Cellular uptake

4T1 cells were seeded at a density of 3 × 10⁵ cells/well in 6-well plates and cultured for 24 h. FITC fluorescence-labeled LCM@Cap and HLCM@Cap were incubated with cells for 1, 3, 6, 12 h respectively. After that, the cells were digested with trypsin, rinsed with PBS, and then analyzed by flow cytometry. Additional, 4T1 cells were seeded on glass slide at a density of 3 × 10⁴ cells/well in 24-well plates. The fluorescent nanoparticles were co-incubated with the cells, and then stained with YF594-Phalloidin or lysotracker RED. Finally, cells were stained with DAPI, followed by fluorescence imaging.

4.4. Mitochondrial fluorescent probe

4T1 cells were seeded in 24-well plates or confocal dishes and grown to approximately 70% confluency. Cells were then treated with PBS, HMn@Cap, HCu@Cap, HLC@Cap, or HLCM@Cap for 24 h. Fluorescent assays were performed using commercially available probe kits according to the manufacturers’ instructions. DCFH-DA was used to detect mitochondrial ROS. SYBR Gold was used to detect mtDNA. A Mitochondrial Permeability Transition Pore (MPTP) Assay Kit was used to assess mitochondrial membrane permeability, and a JC-1 Assay Kit was used to evaluate mitochondrial membrane potential. Finally, the fluorescence signals were observed and analyzed using CLSM, SIM, or flow cytometry.

4.5. In vivo therapeutic effects evaluation

To investigate the therapeutic efficacy of HLCM@Cap, 4T1 tumor-bearing mice were treated when the average tumor volume reached approximately 60 mm³. Mice carrying 4T1 tumors were randomly divided into six groups (n = 5): Control (5% glucose, 100 μL, i.v.), HMn@Cap (100 μL, 5 mg kg⁻¹, i.v.), HCu@Cap (100 μL, 5 mg kg⁻¹, i.v.), HLC@Cap (100 μL, 5 mg kg⁻¹, i.v.), HLCM@Cap (100 μL, 5 mg kg⁻¹, i.v.), and HLCM@Cap+αPD-L1 (HLCM@Cap, 5 mg kg⁻¹, i.v.; αPD-L1, 10 mg kg⁻¹, i.p.). Administration was performed every 3 days for a total of 4 injections. Mouse body weight and tumor size were measured every three days. After 15 days of treatment, all mice were euthanized, and their tumors were weighed, photographed and then part of the tumor tissue was cut and fixed in 10% neutral buffered formalin. The tumors were processed into paraffin using standard methods and sectioned for histological analysis. Sections were stained with H&E, Ki67 (1:500), TUNEL, LIAS (1:200), CRT (1:100), and p-STING (1:200), and scanned with VS200.

4.6. Single-cell preparation and flow cytometry

Tumor tissues, tumor-draining lymph nodes, and spleens were harvested from mice and processed immediately to obtain single-cell suspensions. Tumor tissues were minced into small fragments and enzymatically digested in RPMI-1640 medium containing collagenase IV (1 mg mL⁻¹) and DNase I (0.1 mg mL⁻¹) at 37°C for 30–45 min with gentle agitation. The digested tissues were then filtered through a 70 μm cell strainer to remove debris and obtain single-cell suspensions. Lymph nodes were mechanically dissociated by gently grinding through a 70 μm cell strainer using the plunger of a syringe, followed by washing with cold PBS. Spleens were similarly dissociated through a 70 μm cell strainer, and red blood cells were lysed using RBC lysis buffer for 2–5 min at room temperature. The reaction was stopped by adding excess PBS, followed by centrifugation and washing. All cell suspensions were centrifuged at 300 × g for 5 min at 4°C, washed twice with cold PBS, and resuspended in staining buffer. The obtained single-cell suspensions were stained with fluorochrome-conjugated antibodies according to standard protocols and subsequently analyzed by flow cytometry.

4.7. In vivo evaluation of lung metastasis

Mice were randomly divided into four groups: Control (5% glucose, 100 μL, i.v.), αPD-L1 (100 μL, 10 mg kg⁻¹, i.p.), HLCM@Cap (100 μL, 5 mg kg⁻¹, i.v.), and HLCM@Cap+αPD-L1 (HLCM@Cap, 100 μL, 5 mg kg⁻¹, i.v.; αPD-L1, 10 mg kg⁻¹, i.p.). To establish the lung metastasis model, 1 × 10⁵ 4T1 tumor cells were intravenously injected following the completion of drug treatment. On day 30, mice were euthanized, and the lungs were harvested, fixed, and subjected to H&E staining for histopathological analysis.

4.8. Statistical analysis

All experiments were independently repeated at least three times. Data are presented as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism 10.0. For comparisons between two groups, Student's t-test was applied, while one-way ANOVA followed by Tukey's post hoc test was used for multiple-group comparisons. A p-value <0.05 was considered statistically significant.

Ethics approval and consent to participate

All procedures were conducted in accordance with the Guidelines for the Welfare and Usage of Lab Animals of the China National Institutes of Health and were approved by the Ethical Committee on Scientific Research of Shandong University Qilu Hospital (DWLL-202500191).

CRediT authorship contribution statement

Jie Yu: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Writing – original draft. Ruyue Zhang: Data curation, Formal analysis, Investigation, Writing – original draft. Zeyi Sun: Data curation, Formal analysis, Investigation, Methodology. Kanaparedu P.C. Sekhar: Data curation, Software, Writing – review & editing. Weikai Sun: Formal analysis, Methodology, Software. Xiaoqing Yang: Resources, Supervision, Visualization, Writing – review & editing. Zhiliang Gao: Methodology, Project administration, Software, Supervision, Writing – review & editing. Jiwei Cui: Funding acquisition, Project administration, Supervision, Validation, Writing – review & editing. Dexin Yu: Funding acquisition, Project administration, Resources, Supervision, Validation, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: