H₂S Self‐Supplied Micelles Reverse Tumor‐Immune Effector Cells Energy Metabolisms to Boost Breast Cancer Immunotherapy With Microenvironment Normalization

ABSTRACT

Reversing the immunosuppressive tumor microenvironment by targeting metabolic competition between tumors and immune effector cells could induce tumor starvation and enhance the activity of immune cells, representing a potential approach to boost tumor immunotherapy. However, its actual efficacy is limited by compensatory oxidative phosphorylation (OXPHOS) energy replenishment and low delivery efficiency. Herein, we report a hydrogen sulfide (H₂S)‐self‐supplying nanoplatform that orchestrates a dual blockade of glycolysis and OXPHOS for improved triple‐negative breast cancer (TNBC) immunotherapy. The micellar system, HA‐ADT@W, achieves tumor‐targeted delivery of a glycolysis inhibitor (WZB117) and H₂S continually released in GSH‐overexpressed tumor cells. This strategy concurrently suppresses glucose uptake in tumor cells by reversing the acidic tumor microenvironment (TME) and disrupts compensatory OXPHOS via H₂S‐mediated inhibition of cytochrome c oxidase. Consequently, we demonstrate a significant rewiring of tumor energy metabolism that not only induces immunogenic cell death with remodeling of the immunosuppressive TME but also alleviates nutrient constraints of immune effector cells, leading to enhanced infiltration and function of cytotoxic immune cells. This work exhibits a smart nanoplatform‐based H₂S self‐supplied micelle for reinforced TNBC immunotherapy via regulated metabolic competition between tumors and immune effector cells with TME normalization.

An H₂S self‐supplying nanomicelle is designed to enhance tumor immunotherapy by reversing tumor‐immune effector cells' energy metabolism. This system doubly blocks glycolysis and oxidative phosphorylation by inhibiting glucose uptake and responsively releasing H₂S. It significantly induces tumor starvation and enhances immunogenicity and competitively boosts the metabolism of immune effector cells with a remodeled immunosuppressive microenvironment, demonstrating a favorable tumor immunotherapy effect.

1. Introduction

Since tumor cells preferentially utilize glycolysis to rapidly consume available glucose while secreting lactate, thereby establishing an acidic, nutrient‐depleted microenvironment that directly suppresses immune effector cells function and infiltration [1, 2, 3], this metabolic reprogramming remains a formidable barrier to effective cancer immunotherapy. The strategy of redirecting the tumor‐dependent “fertilizer” glucose influx toward immune effector cells could directly induce tumor starvation and competitively enhance the metabolic activity of immune cells, with broad application prospects [4, 5, 6]. Typically, targeting glycolysis via glucose transporter 1 (GLUT1) inhibition could directly induce tumor starvation [7, 8, 9, 10], while this approach often triggers compensatory oxidative phosphorylation (OXPHOS) as an alternative energy‐generating pathway, maintaining tumor survival and proliferation, especially in triple‐negative breast cancer (TNBC) cells that primarily rely on OXPHOS for energy [11, 12]. Moreover, the inefficient delivery of metabolic inhibitors further restricts their translational potential [13]. Given the need for new vulnerability‐based strategies for difficult‐to‐treat TNBC with limited targeted options, developing an efficient delivery platform to simultaneously inhibit the addictive glycolysis and compensatory OXPHOS of tumors while competitively enhancing the glucose metabolism of immune cells is an early‐stage preclinical exploration to amplify tumor immunotherapy.

Hydrogen sulfide (H₂S) plays a key role in cellular bioenergetics and mitochondrial respiration by targeting cytochrome c oxidase (COX IV), the terminal enzyme in the electron transport chain [14, 15]. H₂S self‐supplying strategies therefore offer a promising approach for metabolic intervention in cancer, especially for disrupting mitochondrial energy metabolism [16, 17, 18]. However, the pharmacological effects of H₂S are highly dependent on the context and dose: while low or endogenous sulfide levels may support tumor adaptation, higher and localized exposure can cause metabolic stress and trigger antitumor responses [19, 20]. This makes the therapeutic window and safety margin critical considerations for translational development [21, 22]. Recent evidence indicates that H₂S also participates in glucose metabolic reprogramming and, under carefully engineered delivery conditions, may help remodel the tumor microenvironment and boost antitumor immunity [22, 23]. Nonetheless, the practical application of free H₂S donors remains limited due to their short half‐life, poor bioavailability, and lack of precise spatiotemporal control [23, 24, 25]. Therefore, achieving a sustained and controllable H₂S supply is essential not only for effective therapy but also as a pharmacological strategy to balance efficacy and safety. In this light, nanoplatform‐based delivery provides an effective means to improve tumor accumulation, control local release kinetics, and minimize sudden systemic exposure [25, 26]. Given the high metabolic plasticity of TNBC, combining a continuous H₂S supply with glycolytic intervention presents a feasible approach to simultaneously disrupt mitochondrial respiration and compensatory metabolic pathways, thereby enhancing TNBC immunotherapy through dual blockade of energy metabolism [22, 26].

Herein, a H₂S self‐supplied micelle (termed HA‐ADT@W) was constructed for TNBC immunotherapy via dual metabolic interference and TME normalization (Figure 1). Composed of hyaluronic acid (HA) modified with multiple H₂S donors (Food and Drug Administration‐approved anethole trithione, named as ADT‐OH) and loaded with the glucose uptake inhibitor WZB117, this micelle selectively targets CD44‐overexpressing TNBC cells. HA‐ADT@W possesses multiple merits: (1) HA‐ADT@W could efficiently deliver drugs to the tumors, where it was subsequently triggered by lysosomal acid to degrade and release ADT‐OH and WZB117. (2) ADT‐OH could continually release H₂S in response to elevated glutathione (GSH) levels, cutting off the key OXPHOS and damaging mitochondrial function via COX IV inhibition to deprive energy. (3) WZB117 blocked GLUT1‐mediated glucose uptake and glycolysis, achieving much more depletion of tumorigenic energy sources. (4) The comprehensive energy blockade created by the above dual metabolic interference not only induces tumor starvation and enhances immunogenic cell death (ICD), promoting the infiltration of immune effector cells and transforming the “cold” TNBC into an immunologically “hot” state, but also revitalizes immune effector cells through alleviating metabolic competition and normalizing the acidic immunosuppressive TME. By integrating multidimensional controls over metabolism, H₂S signaling and immune activity, this work exemplifies an early‐stage preclinical exploration to reinforce cancer immunotherapy through synergistic metabolic reprogramming and microenvironment remodeling.

FIGURE 1.

(a) Synthetic route and schematic illustration of H₂S self‐supplied HA‐ADT@W micelle. (b) The anti‐tumor mechanism of the therapeutic processes of the HA‐ADT@W nanosystem.

2. Results and Discussion

2.1. Synthesis and Characterization of HA‐ADT@W

To construct the functional micelle, HA‐ADT copolymer was first synthesized by conjugating ADT‐OH to the HA backbone via a one‐step esterification reaction, which was verified by ¹H NMR, ¹³C NMR, and FTIR spectra (Figures S1–S5), and the conjugated efficiency of ADT‐OH was calculated to be about 24.5% according to ¹H NMR spectra (Figure S3). Transmission electron microscope (TEM) suggested that the HA‐ADT micelle exhibited a uniformly spherical structure with a particle size of approximately 80 nm (Figure S6). After WZB117 loading, a similar monodisperse spherical morphology, particle size and zeta potential (about −20 mV) was also shown in HA‐ADT@W micelle (Figure 2a–c), which was consistent with dynamic light scattering (DLS) analysis. Meanwhile, the unique F element of WZB117 was also displayed in the HA‐ADT@W, detected by energy‐dispersive spectroscopy (EDS, Figure 2a), confirming the successful formation of micelles and WZB117 loading. Furthermore, X‐ray photoelectron spectroscopy (XPS) and the F 1s and S 2p signals observed in XPS fine spectra further confirmed the conjugation of ADT‐OH on HA‐ADT@W micelles (Figure 2d–f) and its fabrication. Moreover, the loading contents of WZB117 were calculated as 10%, according to the standard curve of absorbance spectra (Figure S7). Besides, the HA‐ADT@W micelles also displayed a low critical micelle concentration (CMC) of 35.48 µg/mL (Figure S8) and favorable biostability, revealed by the constant size in a 10% FBS environment for 6 days (Figure S9). The above results collectively confirmed the successful construction of the HA‐ADT@W nanosystem.

FIGURE 2.

Characterization of HA‐ADT@W nanosystem. (a) TEM and element mapping images of HA‐ADT@W (pH 7.4). (b) DLS analyses of HA‐ADT and HA‐ADT@W under different conditions. (c) Zeta potential of HA‐ADT and HA‐ADT@W. (d) XPS survey spectrum of the HA‐ADT@W. High‐resolution XPS spectra of (e) F 1s and (f) S 2p for HA‐ADT@W. (g) Schematic representation of the degradation of HA‐ADT@W in response to the low acidic stimuli. (h) TEM image of HA‐ADT@W (pH 5.5). Scale bar: 200 nm. (i) Cumulative WZB117 release of HA‐ADT@W after different treatments. (j) Schematic representation of GSH consumption of HA‐ADT@W. (k) GSH‐consuming ability of HA‐ADT@W with different concentrations. (l) Release curve of H₂S from HA‐ADT@W upon GSH.

As the key features of the functionalized HA‐ADT@W nanosystem, including the pH‐responsive disintegration and drug release, coupled with the GSH depletion and self‐supplied H₂S generation, were positively correlated with the design goals of drug delivery and tumor‐killing efficacy improvement, which were thus investigated before the antitumor study. On the one hand, the disintegration and drug release behavior of the nanosystem were confirmed by TEM, DLS, and UV–vis spectroscopy. As shown in Figure 2g,h, the spherical structures of HA‐ADT@W collapsed after pH 5.5 treatment (simulating the acidic environment of lysosomes), which was in accordance with the DLS result (Figure 2b), suggesting the pH‐sensitive disassembly trait. Further, a real‐time release assay was carried out to quantitatively verify the pH‐sensitive drug release behavior. As shown in Figure 2i, roughly 80% of WZB117 was released from the HA‐ADT@W nanosystem when exposed to pH 5.5 for 24 h, suggesting pH‐responsive drug release again. In contrast, a negligible release (below 15%) was seen in the control group (pH 7.4 physiological condition) over the course of 24 h, indicating the excellent biostability and minimal risk of drug leakage.

On the other hand, the GSH‐depletion ability of the nanosystem and the accompanying generation of H₂S were studied. As the sulfhydryl indicator 5,5′‐dithiobis(2‐nitrobenzoic acid) (DTNB) could react with GSH to form yellow 2‐nitro‐5‐mercaptobenzoic acid (TNB) with obvious absorption peak at 412 nm, and ADT‐OH from HA‐ADT@W could competitively react with GSH to produce H₂S, leading to the reduction of TNB, the decrease of characteristic peak of TNB thus was used to measure the GSH exhaustion capability of HA‐ADT@W (Figure 2j). As shown in Figure 2k, HA‐ADT@W significantly reduced TNB formation and exhibited a dose‐dependent behavior, attributed to effective GSH depletion. Simultaneously, the self‐supplied H₂S production of the nanosystem associated with GSH depletion was quantitatively detected using an H₂S content detection kit (Solarbio, China). Notably, HA‐ADT@W micelles were pre‐treated with GSH (10 mM, simulated tumor microenvironment) to completely unload ADT‐OH. As shown in Figure 2l, the release of H₂S from the HA‐ADT@W micelles peaked at approximately 50 µM within 18 h and continued for 24 h, suggesting the potent H₂S modulation ability of HA‐ADT@W, which supported the feasibility of constructing an H₂S self‐supplying nanosystem for metabolic interference and enhanced immunotherapy.

These results robustly demonstrated that the HA‐ADT@W nanosystem featured an intelligent “dual lock‐and‐key” release mechanism. The first level of response involved HA‐mediated tumor targeting and acidic pH‐triggered disintegration of the nanostructure and subsequent release of both WZB117 and ADT‐OH. The second level consisted of the overexpression of GSH in tumor cells, which led to the subsequent self‐generation of H₂S mediated by ADT‐OH. This cascade‐responsive characteristic ensured precise delivery and tumor cell‐specific release of WZB117 and H₂S, thereby laying a solid foundation for highly effective targeted cancer therapy.

2.2. Cytotoxicity and Endocytosis Efficacy of HA‐ADT@W Micelles

A cell viability study demonstrated that both WZB117 and WZB117‐loaded HA‐ADT@W micelles exhibited dose‐dependent cytotoxicity for 4T1 cells (Figure 3a). Furthermore, treatment groups showed different degrees of antitumor effect after 24 h of incubation, in an order of WZB117 < HA‐ADT < HA‐ADT@W (Figure 3b), which was respective attributed to the natural tumor killing features of WZB117 & ADT‐OH with significant synergy effect (combination index of IC₅₀ was 0.54, Figure S10) [27, 28], and improved delivery efficacy of micelles [29]. Moreover, the cytotoxicity trend was more pronounced at 48 h. These results collectively confirmed the superior antitumor effect in vitro of HA‐ADT@W micelles.

FIGURE 3.

In vitro cytotoxicity, uptake, and cargo effect of HA‐ADT@W in 4T1 cells. (a) Cytotoxicity of WZB117 or HA‐ADT@W with different concentrations against 4T1 cells after 24 h incubation. (b) Viability of 4T1 cells treated with different groups for 24 or 48 h. (c) CLSM images and (d) FCM data of 4T1 cells after incubation with Nile Red‐labeled WZB117 or HA‐ADT@W for 4 and 12 h, respectively. (e) Lysosomal escape images of 4T1 cells after treatment with Nile Red‐labeled HA‐ADT@W for various time intervals. (f) Apoptosis degree and (g) quantitative statistics of 4T1 cells after coincubation with different treatments for 24 h, detected by FCM. (h) CLSM images and (i) quantitative statistics of 4T1 cell‐based 3D tumor spheroids after different administrations for 12 h, stained with calcein‐AM/PI (green, cells live; red, cells dead). (j) Schematic diagram of the inhibition of tumor cell‐dependent glycolysis by WZB117. Extracellular (k) glucose and (l) lactate levels in 4T1 cells after diverse treatments for 24 h. (m) CLSM images of 4T1 cells stained with Thiol Tracker Violet with different treatments. (n) Intracellular H₂S level of 4T1 cells after diverse treatments for 24 h. (o) The GLUT1 and COX IV expression of 4T1 cells after administration. Significance analysis was determined using one‐way ANOVA: ^* p < 0.05, ^** p < 0.01, and ^*** p < 0.001.

To illustrate the enhanced uptake of micelles in vitro, 4T1 cells were incubated with Nile Red‐labeled WZB117 or HA‐ADT@W loaded with WZB117 for 4 and 12 h; the endocytosis effect was analyzed by CLSM and FCM. Compared to the control, both WZB117 and HA‐ADT@W displayed time‐dependent cellular uptake. Meanwhile, the accumulation amount of HA‐ADT@W in 4T1 cells was obviously higher than that of WZB117 despite incubation time (Figure 3c; Figure S11), resulting from the good targeting trait of HA [30, 31]. Furthermore, FCM analysis confirmed the improved uptake of HA‐ADT@W again (Figure 3d); it was an important aid in achieving effective tumor killing.

2.3. Lysosomal Escape, In Vitro Tumor Damage, Glycolysis, and COX IV Enzyme Inhibition of HA‐ADT@W Micelles

Since the precise lysosomal escape of active molecules WZB117 and ADT‐OH is the prerequisite for exerting their tumor damage [7, 32], the escape capacity of HA‐ADT@W micelles after endocytosis was investigated. As displayed in Figure 3e, the red fluorescence labeled HA‐ADT@W were mostly distributed at the edge of 4T1 cells after 1 h of co‐incubation, which was a signal for endocytosis; abundant yellow fluorescence overlapped with HA‐ADT@W and lysosomes marked with green appeared inside 4T1 cells when incubation time was extended to 4 h, indicating the successful capture of micelles by lysosomes; the separation behavior of green and red fluorescence clearly occurred when reached 12 h, revealing the successful lysosome escape of HA‐ADT@W micelles.

The tumor damage caused by the unloading of cargo of micelles was subsequently studied. The typical apoptosis was measured first at the cellular level. As shown in Figure 3f,g, WZB117 and HA‐ADT caused moderate apoptosis rates of 21.44% and 34.71%, owing to the inherent glucose metabolism suppression and OXPHOS destruction, plus the improved delivery efficacy of micelles, respectively. Importantly, HA‐ADT@W induced the highest level of apoptosis with 52.85%, implying a significant anticancer effect in vitro. It was likely related to the double work composed of glycolysis inhibition and OXPHOS interference caused by WZB117 and ADT‐OH in the HA‐ADT@W micelles.

The extent of tumor damage in vitro at the tissue level was subsequently investigated. The multicellular tumor spheroids (MCTS) were established and cultured with the above treatment groups for 12 h. Compared to control, WZB117 exhibited weak MCTS destruction, revealed by the strong green fluorescence (calcein‐AM, live cells, Figure 3h,i). Further, stronger, and the strongest damage respectively occurred in the HA‐ADT and HA‐ADT@W micelles treatment groups, which benefited from the introduction of HA‐modified micelle with active tumor targeting capability and conjugation/loading of antitumor molecules ADT‐OH and WZB117, leading to effective tumor damage via dual inhibition of OXPHOS and glycolysis metabolisms.

To clarify the intracellular induction effects of WZB117 and ADT‐OH from HA‐ADT@W micelles on glycolysis and OXPHOS interference, the directly associated key indexes, such as glucose uptake, lactate production, GSH consumption, ROS increasement and H₂S self‐generation regulated by the co‐work of the above two cargoes were investigated. As exhibited in Figure 3j–l, both the extracellular glucose and pH levels of the 4T1 cells incubation system increased significantly in WZB117 and WZB117‐loaded micelle compared to the control, and HA‐ADT@W exhibited the greatest effect, indicating effective inhibition of glycolysis, which was attributed to the natural blocking effect of WZB117 on glucose uptake and improved uptake efficiency mediated by the HA‐modified nanosystem [33]. Meanwhile, free WZB117 generated the moderated reduction of intracellular GSH content compared to control, shown by the decreased fluorescence intensity of GSH probe (Figure 3m; Figure S12), directly owing to GSH generation suppression induced by the blockade of glucose uptake and glycolysis [34, 35]. As ADT‐OH originated from the nanosystem could indeed consume GSH intracellularly to produce H₂S (Figure 3n; Figures S13 and S14), ADT‐OH conjugated HA‐ADT induced superior GSH consumption behavior, benefiting from the GSH scavenging mediated by ADT‐OH and high delivery efficiency by the tumor‐targeted micelle nanoplatform. Notably, the HA‐ADT@W group exhibited the lowest level of GSH, due to the further introduction of WZB117, which was also accompanied by the intracellular ROS levels rising sharply, manifested by the bright green fluorescence and the related quantitative statistics (Figures S15 and S16), confirming the H₂S self‐generation, GSH depletion, ROS enhancement, glucose uptake, and lactate production suppression, which were all directly serviced for glycolysis and OXPHOS interference.

To explore the molecular mechanism of HA‐ADT@W in the above metabolic intervention, western blotting was conducted. The results proved that GLUT1 expression in WZB117 and HA‐ADT@W groups was greatly lower than that in control and HA‐ADT groups (Figure 3o; Figure S17), indicating the inherent capability of WZB117 on the down‐regulation of GLUT1 and suppression of glycolytic activity. Meanwhile, COX IV expression and COX IV enzymatic activity in HA‐ADT and HA‐ADT@W remarkably decreased in comparison with the control and WZB117 groups (Figure 3o; Figure S18), evidencing the restrictive effects of ADT‐OH on COX IV, which could directly initiate OXPHOS inhibition. These results suggested that the HA‐ADT@W nanosystem could reduce the glycolytic and OXPHOS activity of tumor cells by inhibiting the expression of GLUT1 and COX IV.

2.4. Mitochondrial Destruction, ICD, and Antitumor Immune Response of HA‐ADT@W Micelles in vitro

The mitochondrial destruction at the subcellular level was further investigated to comprehensively illustrate the above good antitumor efficacy of the nanosystem, because it was closely associated with the reinforcement of tumor immunogenicity and anti‐tumor immune response [36]. First, HA‐ADT@W induced the most severe mitochondrial dysfunction compared with other groups, revealed by the mitochondrial membrane potential imbalance with the most intense green fluorescence and the weakest red (Figure 4a,b; Figure S19). It was consistent with the mitochondrial activity assay (Figure 4c). Briefly, the elongated mitochondria with strong red fluorescence were observed in 4T1 cells in the control and WZB117 groups, implying the intact morphology and maintained activity of the mitochondria. In contrast, HA‐ADT and HA‐ADT@W treatments led to shortened mitochondria emitting weak red fluorescence, suggesting the occurrence of mitochondrial damage. Moreover, biological TEM (Bio‐TEM) further confirmed the dramatic mitochondrial damage caused by HA‐ADT@W micelles (Figure 4d). In contrast to the normal morphology, severe mitochondrial structure damage represented by condensed cell nuclei, marginalized chromatin, ridge fracture, and membrane rupture was clearly observed in 4T1 cells treated with HA‐ADT@W. These results significantly illustrated from different levels that HA‐ADT@W micelles could effectively induce tumor damage and mitochondrial dysfunction in vitro.

FIGURE 4.

Mitochondrial damage and immunoregulation of HA‐ADT@W micelles in vitro. (a) Schematic illustration of HA‐ADT@W‐triggered mitochondrial dysfunction. (b) CLSM images of mitochondria stained with JC‐1 and (c) Mito‐Tracker Red in 4T1 cells after administration for 12 h. (d) Bio‐TEM images of mitochondria in 4T1 cells with different treatments. (e) Intracellular ATP level of 4T1 cells after diverse treatments for 24 h. (f) CLSM images and quantitative statistics of (g) CRT and (h) HMGB1 in 4T1 cells after different treatments for 12 h. (i) The representative FCM plots and (j) quantification analysis of the matured BMDC cells. (k) Schematic diagram of the maturation of BMDCs in co‐culture systems for 24 h. (l) The representative FCM plots of infiltrating CD4⁺ & CD8⁺ T cells. (m) Schematic diagram of the recruitment of spleen‐derived CD4⁺ T cells and CD8⁺ T cells in co‐culture systems. (n) Histograms of CD86 expression in RAW 264.7 cells after different treatments for 24 h. Significance analysis was determined using one‐way ANOVA: ^* p < 0.05 and ^*** p < 0.001.

As a typical result of dual inhibition of glycolysis and mitochondrial function, the lowest intracellular ATP level was observed in HA‐ADT@W‐treated 4T1 cells (Figure 4e), indicating sufficient energy blockage. Furthermore, the effective energy metabolism inhibition and comprehensive tumor damage mediated by HA‐ADT@W also induced the release of key damage‐associated molecular patterns (DAMPs), witnessed by the most pronounced calreticulin (CRT) expression on the surface and the lowest high mobility group box 1 (HMGB1) level in the 4T1 cells (Figure 4f–h), implying ICD enhancement. The quantitative analysis from FCM further confirmed the above results (Figure S20). As the release of DAMPs played a critical role in kick‐starting the cancer‐immunity cycle by stimulating the activation of antigen‐presenting cells (APCs), the in vitro antitumor immune response initiation represented by the maturation of APCs and immune effector cells (e.g., CD8⁺/CD4⁺ T cells) was studied. After incubation of the bone marrow‐derived dendritic cells (BMDCs) with 4T1 cells plus different treatments for 24 h, the highest amount of mature BMDCs was presented in the HA‐ADT@W group (Figure 4i–k), which was then further recruited with the largest amount of CD8⁺/CD4⁺ T cells (Figure 4l,m; Figure S21). This indicated that the ample DAMPs released from HA‐ADT@W micelles treated‐4T1 cells upon ICD accelerated the maturation of BMDCs and recruitment of immune effector cells, consequently achieving the activation and enhancement of antitumor immune response in vitro. Besides, the release of HMGB1 and the accompanying microenvironment caused by ICD also accelerated the polarization of M2‐type macrophage cells (Figure 4n; Figure S22), further boosting the antitumor immune response efficiency [37, 38]. These results fully demonstrated that the functionalized HA‐ADT@W nanosystem could effectively enhance tumor ICD and boost antitumor immune response through dual inhibition of glycolysis and oxidative phosphorylation.

2.5. Seahorse Analysis on Energy Metabolism Blockade

Seahorse analysis was further employed to reveal the dual metabolisms inhibition of glycolysis and OXPHOS induced by the nanosystem. For one thing, the gold indicator oxygen consumption rate (OCR) of mitochondrial OXPHOS was calculated according to the principle (Figure 5a,b): Briefly, various agents, such as the electron transport inhibitor antimycin A & rotenone, ATP synthase inhibitor (oligomycin), and uncoupling agent FCCP, are respectively added to calculate the OCR at specific stages of the standard procedure. Compared to control, the OCR value of 4T1 cells treated with WZB117, HA‐ADT, and HA‐ADT@W was significantly reduced (Figure 5c), implying effective OXPHOS suppression, which was further revealed by the reductions in basal respiration, maximal respiration, and ATP production (Figure 5d–f). The reason could be attributed to the indirect glycolysis blockade caused by WZB117 [39] and direct inhibition of COX IV activity in the respiratory chain by H₂S [14], which was produced by ADT‐OH from the nanosystem in response to GSH. Therefore, the tumor‐targeted HA‐ADT@W micelles integrating ADT‐OH prodrugs with glycolysis inhibitor WZB117 exhibited the highest OXPHOS inhibition and energy blocking among all groups.

FIGURE 5.

Seahorse analysis on the OXPHOS and glycolysis. (a) Schematic representation of the mitochondrial stress test protocol and (b) test curve. (c) OCR change of 4T1 after co‐incubation with different treatments for 24 h. Effect of the above treatments on the (d) basal respiration, (e) maximal respiration, and (f) ATP production of 4T1 cells. (g) Schematic diagram of the glycolysis stress test approach and (h) theoretical value variation curve of ECAR. (i) ECAR change of 4T1 after different treatments for 24 h. Effect of the above treatments on the glycolysis‐linked (j) glycolysis, (k) glycolytic capacity, and (l) glycolytic reserve of 4T1 cells. (m) Schematic diagram of the ECAR analysis of immune cells. (n) Schematic diagram of glucose metabolism by tumor cells and immune cells before and after regulation by HA‐ADT@W. (o) Change in ECAR of T cells in a co‐incubation system. Treatments effect on the glycolysis‐linked (p) glycolysis, (q) glycolytic capacity, and (r) glycolytic reserve of CD3⁺ T cells. (s) Change in ECAR of BMDCs in a co‐incubation system. Treatments effect on the glycolysis‐linked (t) glycolysis, (u) glycolytic capacity, and (v) glycolytic reserve of BMDCs. Data are represented as mean ± SD (n = 4 independent samples). Significance analysis was determined using one‐way ANOVA: ^* p < 0.05, ^** p < 0.01, and ^*** p < 0.001.

For another thing, glucose and two key inhibitors (2‐DG and oligomycin) were added at the special time interval referring to the protocol, and the glycolysis characterized extracellular acidification rate (ECAR) was measured (Figure 5g,h). As demonstrated in Figure 5i–l, glycolysis levels of 4T1 cells upon WZB117 were obviously decreased compared to control, owing to the natural uptake blockade of glucose; HA‐ADT treatment increased basal glycolysis, indicating that 4T1 cells upregulate glycolytic flux to compensate for H₂S‐induced OXPHOS impairment. However, the maximal glycolytic capacity was slightly reduced in the HA‐ADT‐treated group, suggesting that while glycolysis becomes essential for cellular survival under these conditions, the glycolytic reserve is limited [39]. This pattern is consistent with a state of metabolic stress where the glycolytic pathway operates at near‐maximal capacity basally, leaving insufficient reserve for further stimulation [40]; HA‐ADT@W micelles induced the most extensive glycolysis suppression in 4T1 cells, which directly benefited from the improved delivery efficiency of HA‐modified micelles with good tumor targeting capability. Besides, as the H₂S generated by ADT‐OH could disrupt the mitochondrial electron transport chain and force tumor cells to rely more heavily on glycolysis for energy production, the ADT‐OH‐conjugated HA‐ADT@W micelles with WZB117 loading could more efficiently suppress glycolysis, with increased glycolytic dependence and improved WZB117 bioavailability. Therefore, HA‐ADT@W micelles could effectively block energy metabolism and induce cancer damage via dual inhibition of tumor‐dependent glycolysis and OXPHOS.

Since the blocking of glycolysis mediated by HA‐ADT@W could allocate more fuel glucose to immune effector cells (e.g., CD3⁺ T cells) and APCs, it would significantly enhance the effectiveness of immunotherapy by competitively amplifying their metabolic activity of glycolysis. Typically, the above‐treated groups were used to culture with tumor cells in the upper chamber to interfere with their metabolism, and the metabolic activity of the immune effector cells and APCs in the lower chamber was observed using the Seahorse method (Figure 5m). As demonstrated in Figure 5n–v, glycolysis levels of both CD3⁺ T cells and BMDCs in the WZB117 group were significantly increased compared to control, due to the blockade and complete improvement of glucose uptake in 4T1 cells and CD3⁺ T cells or APCs, respectively. Further, HA‐ADT@W micelles induced the most extensive glycolysis enhancement in CD3⁺ T cells and BMDCs, which benefited from the improved delivery efficiency of HA‐modified micelles with the most potent glycolysis inhibition in 4T1 cells. These findings demonstrated that inhibiting glycolysis in 4T1 cells redirected glucose toward immune cells in the TME, which could shift competitively enhance their metabolic activity and remodel the immunosuppressive tumor microenvironment, consequently leading to reinforced antitumor immunotherapy.

2.6. RNA‐Sequencing and Metabolomics Analysis on the Mechanism of Energy Metabolism Blockade

RNA sequencing analysis was first conducted to evaluate the therapeutic mechanism of the nanosystem in 4T1 cells, and exhibited the obvious difference between HA‐ADT@W and control samples (Figure S23). Upon defining the significance parameter as p‐value < 0.05 and |log2 FC| > 1, the threshold of differentially expressed genes (DEGs) under HA‐ADT@W‐mediated 4T1 cells was 2754 (Figure 6a; Figure S24), of which 1437 genes were up‐regulated and 1317 down‐regulated, indicating a remarkable change of metabolism state. Importantly, the expression of the characteristic genes Ndufa12, Lhpp, Uqcr11, Aldh3b1, Ldhal6b, Eno3, Minpp1, Idh1, Acly, Adpgk, Uqcrc1, and Ppa2 was significantly downregulated after HA‐ADT@W treatment (Figure S25), which were all directly associated with OXPHOS and glycolysis [41], revealing the considerable suppression impact on energy metabolism pathways. Besides, the biological functions down‐regulation of above DEGs were significantly enriched in cell proliferation, DNA replication, ATP biosynthesis and metabolism, and mitochondrial function, detected by Gene Ontology (GO) analysis (Figure 6b,c; Figure S26), and up‐regulation of DEGs were obviously focused on p53 signaling pathways and apoptosis‐related processes, while down‐regulation were concentrated pathways in cancer, the biosynthesis and metabolic processes of biomacromolecules using Kyoto Encyclopedia of Genes and Genomes (KEGG) approach (Figures S27 and S28). These results indicated that HA‐ADT@W could effectively interfere with key energy metabolism in tumors and inhibit tumorigenesis.

FIGURE 6.

RNA sequencing and metabolomic analysis of 4T1 cells regulated by HA‐ADT@W. (a) Volcano plot of transcriptomic analysis of differentially expressed genes. (b, c) Down‐regulated GO enrichment analysis in HA‐ADT@W vs. the control group (n = 7 independent samples). (d) Partial least squares discriminant analysis of 4T1 cells with different administrations for 24 h. (e) The differential metabolites of the HA‐ADT@W compared with the control are represented by the volcano plot, the red and blue dots, respectively, indicating up‐ and down‐regulation of metabolites. (f) Bubble map of significant up/down‐regulation of metabolic pathways in HA‐ADT@W treated 4T1 cells, compared to control. (g) Hierarchical clustering heatmap results of differential metabolites between control and HA‐ADT@W. (h) Quantitative analysis of key metabolites related to aerobic glycolysis and OXPHOS pathways in 4T1 cells after different treatments. (i) Holistic trend analysis of metabolic pathways using differential metabolite abundances. (j) Schematic diagram of changes in intracellular metabolites of glycolysis and OXPHOS in 4T1 cells caused by HA‐ADT@W. Data are represented as mean ± SD (n = 6 independent samples). Significance analysis was determined using one‐way ANOVA: ^** p < 0.01 and ^*** p < 0.001.

The targeted metabolomic analysis was used to further reveal the suppression of glycolysis and OXPHOS metabolism pathways in tumor cells after administration. First, the sample separation of HA‐ADT@W and control groups was relatively small with a significant difference between them, indicating the stability and reliability (Figure 6d). Next, the metabolites of 4T1 cells after HA‐ADT@W treatment displayed the dramatic change compared with the control (Figure 6e), implying the indeed metabolic regulation effect. Importantly, HA‐ADT@W treatment obviously down‐regulated the glycolytic metabolism and its downstream TCA cycle (Figure 6f), as detected by enrichment analysis of the metabolic pathways. These results further clarified the effective suppression of glycolysis and OXPHOS energy metabolisms mediated by the nanosystem.

For the reason illustration, the levels of the typical glycolytic products, such as the D‐fructose 6‐phosphate, 3‐phosphoglycerate, and phosphoenolpyruvate, were dramatically down‐regulated (Figure 6g,h), proving the HA‐ADT@W‐caused GLUT1 occlusion indeed inhibited glycolysis (Figure 3o). Moreover, OXPHOS metabolism was also effectively limited by the HA‐ADT@W group (Figure 6f), which was attributed to the inhibition of COX IV by H₂S self‐generated by the nanosystem (Figure 3o), resulting in disruption of the electron transport respiratory chain and blockade of OXPHOS, revealed by the significantly down‐regulation of related metabolites, e.g., ATP, FMN, and TPP (Figure 6g,h). Notably, HA‐ADT@W not only significantly blocked TNBC‐dependent energy metabolisms involving OXPHOS and glycolysis by disrupting mitochondrial function and directly blocking GLUT1, but also obviously reduced the metabolisms of nucleic acid, protein, lipid, and other biological macromolecule synthesis (Figure 6i), consequently leading to superior tumor proliferation inhibition. The above data collectively demonstrated that HA‐ADT@W could effectively suppress the core energy metabolisms of TNBC at the root through reducing COX IV activity and blocking GLUT1 work (Figure 6j), displaying the potential tumor double starvation therapeutic effect in vivo.

2.7. In Vivo Antitumor Effect of HA‐ADT@W Micelles

4T1 tumor‐bearing mouse model was established to study the in vivo antitumor efficacy of HA‐ADT@W according to the protocol (Figure 7a). The biosafety was first assessed. After administration with HA‐ADT@W micelles for 1 and 7 days, all the liver & kidney functional indexes and blood biochemical parameters of mice showed no significant change compared to the control (Figure S29), indicating good biosafety. Further, HA‐ADT@W did not induce obvious body weight gain throughout the administration period compared with control (Figure S30). When the treatment window was extended to the end, the tissue damage was not observed in the major organs of mice (Figure S31), confirming the favorable biosafety of the nanosystem again. Considering the good accumulation of targeted HA‐ADT@W micelles at tumor sites was main factor for the good biosafety, its biodistribution was subsequently investigated. After post‐injection, HA‐ADT@W exhibited significantly higher accumulation in the tumor and lower in major normal tissues, compared to WZB117 (Figure 7b; Figure S32), attributed to the active targeting capability of HA, which was helpful for achieving the desired tumor therapeutic efficacy with good biosafety.

FIGURE 7.

In vivo antitumor evaluation of HA‐ADT@W nanosystem. (a) Schematic diagram of the anti‐tumor treatment regimen on 4T1 tumor‐bearing mice. (b) The biodistribution of Nile Red‐labeled WZB117 or HA‐ADT@W in vivo in major tissues after 6 h of injection. (c) Images of xenograft tumors after 0, 7, and 18 days of various treatments. (d) Tumor volume change curves of the mice in all groups. (e) The average tumor growth curves and (f) survival rate of mice in different groups (n = 6 independent samples). (g) H&E, TUNEL, and ki67 immunofluorescence images for tumor sections. Quantitative fluorescence intensity analysis of (h) ki67, (i) TUNEL, (j) COX IV, and (k) GLUT1. IFC images of (l) COX IV and (m) GLUT1 for tumor sections after various treatments. (n) The relative lactate level in the tumor tissues after different treatments. (o) Fluorescence imaging of tumor pH value in different groups by using SNARF‐1. Scale bar: 50 µm. Survival curves were compared using the log‐rank (Mantel‐Cox) test. Significance analysis was determined using one‐way ANOVA: ^** p < 0.01, ^*** p < 0.001.

Morphological analysis of tumor tissues directly suggested the antitumor degree of treatment groups (Figure 7c; Figure S33): HA‐ADT group exhibited a stronger tumor suppressive effect compared to the saline and WZB117 groups, while HA‐ADT@W displayed the greatest anti‐tumor effect. It was further confirmed by the real‐time monitoring of tumor volume (Figure 7d,e), indicating the effective tumor growth suppression of HA‐ADT@W micelles. Importantly, the survival of mice was also effectively extended to 45 days with a 67% survival ratio in the HA‐ADT@W group (Figure 7f), which was significantly better than that of other groups, confirming the excellent anti‐tumor effect in vivo of the nanosystem again. Moreover, the HA‐ADT@W group also induced the maximum tumor tissue damage and minimized proliferation efficiency, as revealed by the disrupted tissue's structure & prominent chromatin consolidation and extensive pink fluorescent dots overlapped the Cy3‐labeled broken DNA fragments and DAPI‐stained nuclei, and slight expression of the typical proliferation‐related antigen ki67 (green fluorescence) in H&E, TUNEL and immunofluorescence assays (Figure 7g–i), indicating the significant tumor damage. Besides, HA‐ADT@W induced a remarkable reduction of COX IV and GLUT1 expression (Figure 7j–m) and CRT expression and HMGB1 release enhancement with the highest level (Figures S34–S36), revealing the expected blockade of energy metabolism and ICD reinforcement in vivo. Notably, HA‐ADT@W treatment also significantly normalized the acidic TME, as revealed by the reduction of lactate and up‐regulation of pH in the microenvironment (Figure 7n,o; Figure S37). These results collectively confirmed the remarkable antitumor effect of HA‐ADT@W micelles at the tissue and molecular levels. The detailed reason could be interpreted as follows: (1) The HA‐ADT@W micelles significantly improve the bioavailability of WZB117 and ADT‐OH through active targeting mediated by HA; (2) The released WZB117 effectively inhibits tumor‐dependent glycolysis by cutting off glucose uptake; (3) The released ADT‐OH generates H₂S in situ to further amplify tumor starvation damage by inhibiting OXPHOS and inducing mitochondrial destruction, in response to the excess GSH in tumors; (4) The dual energy metabolism inhibition not only significantly accelerates tumor ICD and enhances antitumor immunomodulatory effects, but also effectively alleviates immune tolerance by normalizing the tumor microenvironment, ultimately resulting in an excellent antitumor therapeutic effect in vivo.

2.8. In Vivo Antitumor Immune Response, Anti‐Metastasis and Immune Memory of HA‐ADT@W Micelles

To intuitively illustrate the antitumor immunomodulatory effects of HA‐ADT@W, the proportion of various tumor‐infiltrating immune cells was quantitatively measured by FCM in 4T1 tumor‐bearing mice (Figure 8a). Based on the clear classification and gating strategy (Figure S38), the control group exhibited the inherent immune resistance, manifested by abundant immunosuppressive cells (M2‐TAMs, Tregs, and MDSCs) and a small number of immune‐promoting cells (CD8⁺/CD4⁺ T cells, M1‐TAMs, and mature DCs). Meanwhile, WZB117 and HA‐ADT prodrug micelles partly alleviated immune tolerance and enhanced antitumor immune response, as evidenced by the appropriately increased immune‐promoting cells and decreased immune‐suppressing cells compared with the control group (Figure 8b–m). This could interpret that the inhibition of glycolysis and OXPHOS generated by WZB117 and ADT‐OH prodrug not only enhances ICD through starving tumors, but also normalizes the tumor microenvironment and alleviates immune tolerance by reducing lactate production, generating H₂S stimulation, and competitively providing more glucose fuel for immune effector cells [42, 43, 44]. Importantly, the HA‐ADT@W group displayed the highest proportion of CD8⁺/CD4⁺ T cells, M1‐TAMs, and matured DCs, while the lowest proportion of M2‐TAMs, MDSCs, and Tregs, which was further confirmed by the significant upregulation of key inflammatory and antitumor immune factors (TNF‐α and IFN‐γ, Figure 8n–r), indicating the maximum antitumor immune response with minimum immune resistance.

FIGURE 8.

In vivo immune stimulation of HA‐ADT@W nanosystem. (a) Schematic illustration of the therapeutic profile of tumor‐bearing mice. (b‐m) FCM examination and the corresponding quantification analysis of matured DC cells (b,h), tumor‐infiltrating CD8⁺/CD4⁺ T cells (c,i), M1‐TAMs (d,j), M2‐TAMs (e,k), Tregs (f,l), and MDSCs (g,m) in tumor‐bearing mice after administration. (n) TNF‐α and IFN‐γ staining observation, corresponding quantitative analysis (o,p), and ELISA detection (q,r) of tumor tissues with different treatments. Scale bar: 50 µm. (s) Photographs and H&E staining images of the metastatic lung nodules in 4T1 tumor‐bearing mice. (t) Procedures of administration to investigate tumor recurrence. (u) Representative flow cytometric images of T_EM cells. Data are represented as mean ± SD (n = 5 biologically independent samples). Significance analysis was determined using one‐way ANOVA, ^* p < 0.05, ^** p < 0.01, and ^*** p < 0.001.

Moreover, the HA‐ADT@W nanosystem also displayed the strongest anti‐metastasis effect (Figure 8s), as revealed by the normal lung structure in H&E‐stained images and the lowest number of metastatic nodules (Figure S39), confirming the superior immune response regulated by the HA‐ADT@W nanosystem. This phenomenon could be explained as follows: first, the delivery efficacy and bioavailability of WZB117 and ADT‐OH were obviously enhanced based on HA‐modified micelles; second, WZB117 and ADT‐OH significantly blocked glucose uptake and OXPHOS of 4T1 cells, leading to dual tumor starvation with reinforced ICD, accompanied by enhanced recruitment of immune effector cells; Lastly, the closure of glucose influx into tumors also competitively elevated the glucose utilization rate in immune effector cells, as well as normalized the tumor microenvironment (upregulating pH, etc.) in conjunction with H₂S self‐production mediated by nanosystem, consequently leading to the augmenting antitumor response with reduced immune resistance. Besides, tumor recurrence and second immune response assays were employed to study the durability of protective immunity and the potential clinical application of the HA‐ADT@W nanosystem (Figure 8t). After secondary tumor cell implantation, mice pretreated with the nanosystem produced a large number of effector memory T cells (Tem, CD62L⁻CD44⁺) in the spleen on day 19 compared to control (nearly 5 times, Figure 8u; Figure S40), indicating a rapid and efficient secondary immune response. Importantly, the HA‐ADT@W group significantly prolonged the survival period with the highest survival rate of 83.3% within 45 days after secondary implantation of tumor cells (Figure S41), owing to the long‐lasting immune memory. These results comprehensively revealed that the H₂S self‐supplied HA‐ADT@W nanosystem could significantly reinforce TNBC immunotherapy via dual blockade of tumor‐dependent energy mechanisms and competitive enhancement of metabolism activity of immune effector cells with TME normalization, displaying a great tumor‐killing and metastasis‐suppression effect in vivo.

3. Conclusion

In short, we constructed a HA‐modified H₂S self‐supplied micellar nanosystem for enhanced TNBC immunotherapy via blocking the core energy mechanisms of tumors and improving the metabolism of immune effector cells with remodeling of the TME. With the help of the improved delivery efficacy of HA‐modified micelle, HA‐ADT@W could specifically unload WZB117 and ADT‐OH into tumor cells with enhanced bioavailability. Further, the released ADT‐OH could self‐generate H₂S in situ to suppress the dependent OXPHOS and induce mitochondrial destruction in response to the excess GSH. Meanwhile, the released WZB117 significantly blocked compensatory glycolytic metabolism by cutting off glucose uptake, thereby amplifying the therapeutic effect of tumor starvation and enhancing ICD. Importantly, the closure of glucose influx into tumors also competitively elevated the glucose utilization rate of immune effector cells and normalized TME (upregulating pH), in conjunction with H₂S self‐production, consequently leading to antitumor response reinforcement and metastasis suppression with immune memory response activation. Therefore, this work exhibits a promising H₂S self‐generated micelle nanosystem for boosted TNBC immunotherapy via reversing the competitive metabolism between tumors and immune effector cells with TME normalization.

4. Experimental Section/Methods

4.1. Synthesis of HA‐ADT Copolymer

The prodrug copolymer (denoted HA‐ADT) was synthesized by conjugating ADT‐OH to the HA backbone via a one‐step esterification reaction. Briefly, HA (20 mg) was completely dissolved in formamide (2 mL). The carboxyl groups of HA were then activated by the DMF solution (2 mL) containing DCC (12.8 mg) and DMAP (2.56 mg) for 1.5 h. Subsequently, ADT‐OH (12 mg) dissolved in 10 mL of DMF was added dropwise to the above solution under stirring at room temperature for 36 h. The mixture was dialyzed (MWCO 3500 Da) against ultrapure water for 72 h. Finally, the HA‐ADT was collected through filtration and lyophilization.

4.2. Synthesis of HA‐ADT@W Micelle

Typically, 10 mg of HA‐ADT polymer and 2 mg of WZB117 were dissolved in 2 mL of DMSO to achieve complete dispersion under stirring at room temperature for 4 h. The resulting solution was then added dropwise into 10 mL of distilled water under continuous stirring. After stirring for 24 h, the mixture was transferred into a dialysis bag (MWCO 3500 Da) and dialyzed for 36 h. Finally, the WZB117‐loaded micelles (named HA‐ADT@W) were obtained by lyophilization.

4.3. In Vitro Drug Release Behavior of HA‐ADT@W

HA‐ADT@W (6 mg) dispersed in PBS at pH 7.4 and 5.5 was transferred to centrifuge tubes at 37°C under stirring. Samples were centrifuged at designated time points (1, 2, 3, 4, 5, 6, 8, 12, and 24 h), and aliquots of the supernatant were collected to quantify the release level of WZB117 using fluorescence spectroscopy. Meanwhile, an equal volume of fresh PBS was added to maintain sink conditions.

4.4. Evaluation of Nanosystem Stability in Serum

HA‐ADT@W (1 mg mL⁻¹) was dissolved in PBS (pH 7.4) with 10% fetal bovine serum (FBS) at 37°C. At certain time intervals (0, 24, 48, 72, 96, 120 and 144 h), the samples were taken out, and the hydrodynamic size was detected by DLS.

4.5. In Vitro GSH Depletion

The GSH‐depleting capacity was evaluated using the DTNB (5,5’‐dithiobis(2‐nitrobenzoic acid)) assay. Briefly, 2 mL of HA‐ADT@W solution at various concentrations (0, 10, 20, 30, 40, 50, 100, 200 µg mL⁻¹) was mixed with 10 mM GSH under constant shaking at 37°C for 30 min. After centrifugation, the supernatants were incubated with 50 µL of ethanol with 2.5 mg mL⁻¹ DTNB. After reaction at room temperature for 5 min, the absorbance at 412 nm was measured using a UV–vis spectrophotometer to quantify residual GSH.

4.6. In Vitro H₂S Release Behavior

Briefly, HA‐ADT@W micelles were pre‐incubated in PBS solution (pH 5.5) with 10 mM GSH at 37°C for 24 h [45, 46, 47]. At the specified time intervals, the supernatant was collected to measure H₂S release using an H₂S content detection kit (Solarbio, China), following the manufacturer's instructions.

4.7. Cell Culture and Cytotoxicity Assay

4T1 cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium with 10% fetal bovine serum (FBS) plus 1% penicillin‐streptomycin. The culture condition was maintained at 37°C with 5% CO₂.

4T1 cells (1 × 10⁵ cells per well) seeded in 24‐well plates were treated with varying concentrations of WZB117 or HA‐ADT@W once the cell confluency reached approximately 60%–70%. After 24 h of incubation, cell viability was assessed using the Cell Counting Kit‐8 (CCK‐8) assay. In a separate experiment, 4T1 cells were treated with PBS, WZB117, HA‐ADT or HA‐ADT@W with a concentration of 5 µg mL⁻¹ WZB117 for 24 and 48 h, respectively. Cell viability was evaluated by the CCK‐8 protocol.

4.8. In Vitro Cellular Uptake

The uptake efficiency of the micelles was assessed in 4T1 cells using confocal laser scanning microscopy (CLSM) and flow cytometry (FCM). For one thing, 4T1 cells (1 × 10⁵) seeded in confocal dishes were incubated with PBS, Nile Red‐labeled WZB117, or HA‐ADT@W for 4 and 12 h, respectively. Next, the cells were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X‐100, followed by staining with Cell Light Talin‐GFP and 4′,6‐diamidino‐2‐phenylindole (DAPI) and imaging by CLSM (LSM 510 META Olympus). For other things, 4T1 cells (5 × 10⁵ cells per well) seeded in 6‐well plates were treated with the same formulations for 4 and 12 h. The cells were harvested to quantitatively analyze fluorescence intensity by FCM (FACS Calibur and Celesta, BD Biosciences).

4.9. Detection of Lysosome Escaping

4T1 cells (1 × 10⁵) seeded in confocal dishes were co‐incubated with Nile Red‐labeled HA‐ADT@W and LysoTracker Green at 37°C for different time points. Cells were then washed and imaged by CLSM.

4.10. Live/Dead Analysis of 3D Tumor Spheroids

A suspension of 1% (w/v) agarose in PBS was heated inside a high‐pressure autoclave and subsequently dispensed into 96‐well plates (50 µL per well). The plates were exposed to UV irradiation for 3 h and allowed to cool down. A suspension of 4T1 cells (1600 cells per well) was then seeded on the top layer of the agarose. On the seventh day, multicellular tumor spheroids (MCTSs) were formed [48]. The constructed MCTSs were incubated with PBS, WZB117, HA‐ADT and HA‐ADT@W at 37°C for 24 h, respectively. The MCTSs were then washed and stained with calcein‐AM and propidium iodide for 30 min. Last, the MCTSs were imaged using CLSM with Z‐stack scanning.

4.11. FCM Analysis of Cell Apoptosis and Western Blotting Assay

4T1 cells (5 × 10⁵ cells per well) planted in 6‐well plates were incubated with the above treatments for 24 h. The cells were then treated with the Annexin V‐FITC/PI kit according to the manufacturer's protocols. The apoptosis level was quantified by FCM.

For the western blotting assay, 4T1 cells were treated with the indicated formulations for 24 h. Cells were then lysed, and total protein was collected by centrifugation. The expression levels of GLUT1 and COX IV were analyzed by western blotting.

4.12. Detection of Extracellular Glucose, Lactic Acid, and Intracellular ROS

After coincubation with PBS, WZB117, HA‐ADT, or HA‐ADT@W for 24 h, the cell culture medium of 4T1 cells was measured by the Glucose Assay Kit and Lactate Assay Kit following the instructions, respectively.

After the above administrations for 24 h, 4T1 cells were incubated with DCFH‐DA probe (Beyotime, China) for 30 min and imaged by CLSM, based on the instructions.

4.13. Intracellular GSH Depletion and H₂S Detection

After the above administrations for 24 h, 4T1 cells were treated with a GSH detection kit, following the manufacturer's instructions. Additionally, after treatment with different formulations, the 4T1 cells were stained with Thiol Tracker Violet and imaged by CLSM to further monitor intracellular GSH depletion.

For H₂S detection, 4T1 cells (5 × 10⁵ cells per well) seeded in 6‐well plates were incubated with different formulations for 24 h. A commercial H₂S assay kit containing WSP‐5 probe was incubated with the above cells and used to assess the intracellular H₂S levels using fluorescence microscopy.

4.14. Detection of Mitochondrial Damage

The changes in mitochondrial membrane potential (Δ Ψm) were examined using the JC‐1 fluorescent probe. Typically, 4T1 cells were incubated with PBS, WZB117, HA‐ADT, or HA‐ADT@W for 12 h; a mitochondrial membrane potential assay kit (Beyotime, China) was then applied according to the manufacturer's protocol. The Δ Ψm disruption was evaluated by CLSM.

For mitochondrial distribution, 4T1 cells were treated with Mito Tracker Red CMXRos and visualized by CLSM after the above administrations for 12 h. For mitochondrial morphology observation, 4T1 cells (7 × 10⁵ cells) were first incubated with fresh medium with or without HA‐ADT@W (50 µg mL⁻¹) for 12 h. Next, cells were washed and harvested by centrifugation, followed by fixation, dehydration, embedding, and sectioning. Finally, mitochondrial ultrastructure was examined by bio‐transmission electron microscopy (Bio‐TEM).

4.15. Intracellular ATP Quantification

Briefly, 4T1 cells (5 × 10⁵ cells per well) seeded in 6‐well plates were treated with the above administrations for 24 h. The cells were then lysed, and the supernatant was obtained via centrifugation. The protein concentration in the supernatant was quantified, and the ATP levels were measured using a commercial ATP Assay Kit (Beyotime, China).

4.16. In Vitro Macrophage Polarization

Typically, RAW264.7 cells (3 × 10⁵ cells) seeded in the lower chamber of the transwell system were cultured with 10 ng mL⁻¹ IL‐4 for 12 h to induce M2 polarization. Meanwhile, 4T1 cells (1 × 10⁵ cells) treated with various formulations were seeded in the upper chamber. After 24 h of co‐incubation, RAW264.7 cells were collected, washed, stained with CD11b‐FITC, CD86‐PE, and F4/80‐APC antibodies, and analyzed by FCM.

4.17. Evaluation of BMDCs Maturation In Vitro

Generally, bone marrow cells isolated from 6‐week‐old female Balb/c mice were cultured in complete RPMI‐1640 medium supplemented with 20 ng mL⁻¹ GM‐CSF and 10 ng mL⁻¹ IL‐4 to harvest immature bone marrow‐derived dendritic cells (BMDCs). Immature BMDCs (3 × 10⁵ cells) were seeded in the lower chamber of the transwell system, while 4T1 cells (1 × 10⁵ cells) were treated with different formulations in the upper chamber. After 24 h of co‐incubation, BMDCs were collected and stained with CD11c‐FITC, CD86‐PE, and CD80‐APC antibodies. The maturation degree of BMDCs was assessed by FCM.

4.18. Evaluation of CD8⁺ T/CD4⁺ T Recruitment In Vitro

Immune cells were obtained from the spleens of 6‐week‐old Balb/c mice, undergoing steps such as grinding, filtering, lysis, and removal of red blood cells. Cells were then resuspended in RPMI‐1640 complete medium supplemented with IL‐2 (20 ng mL⁻¹) to maintain a single‐cell suspension, which was seeded into 6‐well plates (5 × 10⁵ cells per well). The BMDCs from the lower chamber were collected and added to 6‐well plates, where they were co‐cultured with the immune cells already seeded in the wells. After 24 h of incubation, the cells were collected and stained with CD3‐APC/Cy7, CD8‐PerCP/Cy5.5 and CD4‐FITC antibodies. Finally, the level of CD8⁺ T and CD4⁺ T cells was measured by FCM.

4.19. Measurement of CRT and HMGB1

4T1 cells were first co‐incubated with the above administrations for 12 h. For CLSM detection, cells were treated with 4% paraformaldehyde, 0.5% Triton X‐100, and 5% FBS. Subsequently, CRT and HMGB1 rabbit primary antibody was respectively incubated with the cells for 1 h, followed by staining with Cy3‐labeled goat anti‐rabbit secondary antibody and DAPI. Finally, the released CRT and HMGB1 were imaged by CLSM. For FCM analysis, 4T1 cells were collected, stained with CRT rabbit primary antibody plus Cy3‐labeled goat anti‐rabbit secondary antibody, and analyzed by FCM.

4.20. Transcriptomic Analysis

After treatment with PBS and HA‐ADT@W for 24 h, 4T1 cells were collected by centrifugation, and RNA was extracted using the RNAkey Reagent. RNA sequencing was performed by NovaSeq (Illumina, CA, USA). Data analysis was done by FastQC (version: 0.11.5).

4.21. Metabolomics Investigations

After the above treatments for 24 h, 4T1 cells were gently washed with pre‐cooled PBS, collected on ice with a cell scraper, and frozen with liquid nitrogen. Finally, the samples were sent to Bioprofile Technology Co., Ltd. (Shanghai, China) for metabolomic profiling.

4.22. Seahorse Assays

4T1 cells (40 000 cells per well) seeded into Seahorse XF 24‐well plates were cultured with the above treatments for 12 h. Next, the culture medium was removed and replaced with Seahorse XF assay medium containing the appropriate reagents for metabolic stress testing in accordance with the manufacturer's instructions. For extracellular acidification rate (ECAR) analysis, cells were incubated with 100 mM glucose, 10 µM oligomycin (Oligo), and 500 mM 2‐deoxy‐D‐glucose (2‐DG), based on prior optimization. For oxygen consumption rate (OCR) detection, 4T1 cells were treated with 1 µM oligomycin, 1 µM fluorocarbonyl cyanide phenylhydrazone (FCCP), and 0.5 µM rotenone/antimycin A (Rot/AA), based on prior optimization.

4.23. In Vivo Antitumor Study

Female Balb/c mice (4–6 weeks old) were purchased from the Beijing Institute of Drug Control. All animal experiments were performed in accordance with the guidelines approved by the Animal Care and Use Committee of Northwestern Polytechnical University (Approval No. 202401006). For the establishment of the 4T1 tumor‐bearing mouse model, mice were subcutaneously injected with 1 × 10⁶ 4T1 cells into the right hind limb. When tumor volumes reached ∼70 mm³, mice were randomly divided into four groups (n = 6) and injected intravenously with saline, WZB117, HA‐ADT, and HA‐ADT@W (equivalent to 10 mg kg⁻¹ WZB117), above administration was performed twice per week for 18 days. Tumor volumes were calculated as: Volume = (length × width²)/2.

4.24. H&E, TUNEL, and IFC Assays

After 18 days of treatment, the mice were euthanized. The tumor tissues and major organs (heart, liver, spleen, lung, kidney) were collected and processed into pathological sections. For hematoxylin and eosin (H&E) assay, tissue sections were stained accordingly and examined under a light microscope to assess morphological changes. For TUNEL analysis, tumor sections were stained using a One‐Step TUNEL Apoptosis Detection Kit (Beyotime, China) and imaged by CLSM. For immunofluorescence (IFC) study, tumor sections were blocked with 5% bovine serum albumin, incubated with GLUT1, COX IV, TNF‐α, IFN‐γ, CRT, HMGB1, or ki67 primary antibodies and Cy3 or Alexa Fluor 488‐conjugated goat anti‐rabbit IgG secondary antibodies (Immunofluorescence Staining Kit, Beyotime, China). Finally, the sections were stained with DAPI and observed by CLSM.

4.25. Quantification of Tumor‐Infiltrating Lymphocytes

Saline, WZB117, HA‐ADT, and HA‐ADT@W were intravenously injected into tumor‐bearing mice at a frequency of twice every 3 days. Subsequently, the tumor tissues were harvested and dissociated into single‐cell suspensions using a Tumor Dissociation Kit (Miltenyi Biotec), according to the manufacturer's instructions. Monodispersed lymphocytes were collected and stained with a Live/Dead viability dye for 30 min, followed by co‐staining with the appropriate antibodies for immune cell profiling at 4°C for another 30 min: CD8⁺ T cells (CD3⁺CD4⁻CD8⁺) and CD4⁺ T cells (CD3⁺CD4⁺CD8⁻) were identified using anti‐CD3‐APC/Cy7, anti‐CD4‐FITC and anti‐CD8‐PerCP/Cy5.5 antibodies; Regulatory T cells (Tregs, CD4⁺CD25⁺Foxp3⁺) were stained with anti‐CD4‐FITC, anti‐CD25‐APC and anti‐Foxp3‐PE; Matured dendritic cells (DCs, CD11c⁺CD80⁺CD86⁺) were labeled with anti‐CD11c‐FITC, anti‐CD80‐APC and anti‐CD86‐PE; M1‐type tumor‐associated macrophages (M1‐TAMs, CD11b⁺F4/80⁺CD86⁺) were stained using anti‐CD11b‐FITC, anti‐F4/80‐APC and anti‐CD86‐PE; M2‐TAMs (CD11b⁺F4/80⁺CD206⁺) were stained with anti‐CD11b‐FITC, anti‐F4/80‐APC and anti‐CD206‐PE; Myeloid‐derived suppressor cells (MDSCs, CD11b⁺Gr‐1⁺) were marked with anti‐CD11b‐FITC and anti‐Gr‐1‐PE. Lymphocyte populations were analyzed by FCM.

4.26. In Vivo Biodistribution

At 6 h post‐injection intravenously with Nile Red‐labeled WZB117 or HA‐ADT@W, the mice were euthanized, and tumor tissues along with major organs were harvested and observed using the animal live imaging system (PerkinElmer, USA).

4.27. Hematological and Biochemical Analysis

Healthy mice were intravenously injected with saline or HA‐ADT@W. After the final administration, whole blood samples were collected and transferred into anticoagulant tubes. Hematological parameters were analyzed using a routine blood test analyzer (Mindray, China). Additionally, after incubating whole blood at room temperature for 4 h, serum was obtained by centrifuging (1000 g, 15 min) to measure the indexes of liver/kidney function.

4.28. Detection of pH in Tumors

Tumor‐bearing mice were intravenously administered with the above treatments. After 24 h, SNARF‐1 (250 µg kg⁻¹) was injected into the mice via the tail vein. Next, tumor tissues were collected after 30 min and cut into halves for fluorescence imaging using an in vivo imaging system (PerkinElmer, USA).

4.29. Quantifying Memory T Cells with FCM Analysis

The splenocytes were collected from the administered mice and stained with anti‐CD8‐PerCP/Cy5.5, anti‐CD3‐APC/Cy7, anti‐CD62L‐APC, and anti‐CD44‐FITC antibodies, according to the manufacturer's protocols. Afterward, the amount of effector memory T cells (Tem, CD8⁺CD3⁺CD44⁺CD62L⁻) was analyzed by FCM.

4.30. In Vivo Anti‐Metastasis Effect

The lung metastasis model mice previously constructed were injected with the above treatment groups via the tail vein. The administrations were employed twice every 3 days and continued for 12 days. Finally, the mice were euthanized, and their lung tissues were collected, washed, photographed, and counted for nodules. Sections of lung tissues were stained with H&E and observed under a light microscope.

4.31. Statistical Analysis

All statistical analyses were performed using Origin 2022 and GraphPad Prism version 9.5. The values were expressed as means ± standard deviation for data that were normally distributed. The confidence levels of 95% and 99% were regarded as a significant difference. Comparison analysis between groups was conducted by one‐way ANOVA with Tukey's multiple comparisons test. Kaplan‐Meier survival curves were compared by the log‐rank (Mantel‐Cox) test. For all comparisons, p < 0.05 test was considered statistically significant (^* p < 0.05, ^** p < 0.01, ^*** p < 0.001).

Author Contributions

L.D. conceived the project. S.M., X.W., and Z.W. performed the experiments and analyzed the results. The manuscript was written by L.D. and S.M. All authors contributed to the general discussion and reviewed the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adma74041‐sup‐0001‐SuppMat.docx.

Data Availability Statement

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