Switchable Enzyme‐Regulated ECM‐Integrin‐Cholesterol Signaling Orchestrate PD‐L1 Dual Destabilization for Boosting Photo‐Immunotherapy

ABSTRACT

Although PD‐1/PD‐L1 blockade therapy has shown clinical success, its efficacy in solid tumors remains limited by the immunosuppressive tumor microenvironment. The tumor extracellular matrix (ECM), which forms a dense physical barrier and activates integrin‐mediated mechanotransduction to upregulate PD‐L1, while cholesterol‐enriched membrane domains stabilize integrin clustering and PD‐L1 localization, forming a self‐reinforcing ECM‐integrin‐cholesterol signaling network that drives immune evasion. Here, we designed a switchable natural enzyme nanoplatform with dual safety locks, co‐delivering papain and cholesterol synthesis inhibitor simvastatin to disrupt this network at multiple levels. Papain remains inactive during systemic circulation and intracellular trafficking, ensuring biosafety, but is selectively activated by tumor‐released glutathione and boosted by localized photothermal heating to precisely degrade ECM, suppressing integrin signaling and PD‐L1 transcription, while simvastatin‐driven cholesterol depletion amplifies this effect by destabilizing integrin clusters and promoting PD‐L1 degradation. Combined with phototherapy‐induced immunogenic cell death, this strategy remodels ECM, enhances T cell infiltration, overcomes checkpoint blockade resistance, and eradicates advanced tumors (∼500 mm³).

Limited efficacy of PD‐1/PD‐L1 therapy in solid tumors stems from the extracellular matrix (ECM) barrier and the “ECM–integrin–cholesterol” axis. We developed a switchable enzyme nanoplatform with dual safety locks co‐delivering papain and simvastatin to suppress PD‐L1. Combined with phototherapy‐induced immunogenic cell death, this strategy remodels ECM, enhances T cell infiltration and eradicates advanced tumors.

1. Introduction

Cancer immunotherapy has revolutionized oncology by harnessing the host immune system to combat tumors. However, tumor cells evade immune surveillance through multiple mechanisms, including secretion of immunosuppressive factors and upregulation of immune checkpoint proteins, which have driven the development of immune checkpoint blockade (ICB) therapies [1, 2, 3]. For instance, monoclonal antibodies targeting programmed cell death protein 1 (PD‐1) and its ligand PD‐L1 have demonstrated clinical efficacy by enhancing tumor cell recognition by effector T cells, significantly improving overall survival in multiple malignancies [4, 5, 6]. However, the clinical benefits remain limited by suboptimal response rates and acquired resistance, largely driven by adaptive upregulation of PD‐L1 expression within tumors [7, 8]. Recent studies reveal that multiple oncogenic transcription factors, including Hippo Yes‐associated protein (YAP), nuclear factor kappa‐B (NF‐κB), c‐Jun, and MYC, can drive PD‐L1 transcription, promoting immune evasion and limiting ICB efficacy [9, 10]. Thus, a comprehensive strategy that simultaneously blocks PD‐1/PD‐L1 interactions, suppresses PD‐L1 transcription, and promotes the degradation of membrane‐bound PD‐L1 is urgently needed to overcome resistance and enhance antitumor immunity.

Beyond immune checkpoint signaling, tumors establish a physical barrier to immune infiltration through extracellular matrix (ECM) remodeling. The ECM forms a dense structural network of fibrous proteins, proteoglycans, and glycosaminoglycans that support tumor growth and facilitates bidirectional cell‐ECM communication via integrin‐mediated mechanotransduction [11, 12, 13, 14]. Increased ECM stiffness activates integrin‐mediated mechanotransduction, driving the oncogenic activation of signal transducer and activator of transcription 1 (STAT1) [15], YAP [16, 17], NF‐κB [18], c‐Jun [19, 20] and MYC [19], which in turn enhances PD‐L1 transcription and establishes a feedforward loop that reinforces tumor immune evasion, while simultaneously receiving feedback signals from tumor cells that further enhance matrix rigidity. Meanwhile, cholesterol‐rich lipid rafts in the plasma membrane stabilize integrin clustering and maintain the membrane localization of PD‐L1, supporting both mechanosignaling and immune evasion [21, 22, 23, 24]. These three elements, ECM stiffness, integrin mechanotransduction, and membrane cholesterol are tightly interconnected, forming a self‐reinforcing ECM‐integrin‐cholesterol signaling network. In this network, a stiff ECM enhances integrin activation, integrin signaling drives PD‐L1 transcription, and cholesterol‐rich domains stabilize both integrins and PD‐L1 on the membrane, collectively amplifying immune suppression. This interdependence explains the limited and heterogeneous clinical responses observed with therapies that target only individual ECM components or specific integrin subtypes. Therefore, a multiplexed strategy that simultaneously disrupts ECM structure, integrin signaling, and cholesterol‐mediated stabilization is required to overcome this barrier and suppress PD‐L1 at multiple regulatory levels.

Phototherapeutic approaches, including photothermal therapy (PTT) and photodynamic therapy (PDT), have emerged as promising immunomodulatory strategies by inducing immunogenic cell death (ICD), and initiating antitumor immune activation [25, 26]. Moreover, reactive oxygen species (ROS) generated during PDT can directly oxidize and degrade key ECM proteins such as collagen and fibronectin, softening the tumor matrix and facilitating immune cell infiltration [27]. However, natural enzymes such as papain (PAP), which can directly degrade ECM proteins and soften the tumor matrix, are limited by uncontrolled activity and off‐target toxicity. To address these challenges, we developed a switchable natural enzyme nanoplatform (SI@PAP) with dual safety locks that integrates the natural hydrolase PAP, the cholesterol synthesis rate‐limiting enzyme HMG‐CoA reductase (HMGCR) inhibitor simvastatin (Siv), and the photothermal agent indocyanine green (ICG) (Figure 1a). The enzymatic activity of PAP remains strictly suppressed during systemic circulation and intracellular trafficking, ensuring biosafety. Upon reaching the tumor microenvironment and being internalized by tumor cells, SI@PAP undergoes intracellular disassembly to safely and efficiently release ICG and Siv. The liberated ICG mediates photothermal effects upon NIR irradiation, while Siv induces apoptosis, together triggering ICD to initiate local immune activation (Figure 1b, Step 1). ICD further promotes the release of glutathione (GSH) from dying tumor cells, generating a tumor‐specific, GSH‐enriched microenvironment that selectively cleaves the disulfide bonds within SI@PAP, thereby reactivating the enzymatic activity of PAP, which is further boosted by localized photothermal heating, enabling precise ECM degradation and matrix softening to suppress integrin mechanotransduction and reduce PD‐L1 transcription (Figure 1b, Step 2). Simultaneously, Siv‐mediated cholesterol depletion destabilizes integrin clusters and membrane‐bound PD‐L1, promoting its degradation and amplifying the inhibitory effect on the ECM‐integrin‐cholesterol signaling network. Together with phototherapy‐induced ICD, this multifaceted strategy drives deep T cell infiltration and establishes a self‐amplifying antitumor immune cycle (Figure 1b, Step 3). In this work, we demonstrate a holistic therapeutic approach that dismantles both the physical and biochemical barriers of the tumor microenvironment by targeting the ECM‐integrin‐cholesterol axis. This strategy achieves dual PD‐L1 suppression at both transcriptional and post‐translational levels, while simultaneously promoting ICD to amplify immune activation. By integrating TME remodeling with checkpoint regulation, our approach offers a promising avenue for overcoming immune resistance, achieving durable remission, and preventing recurrence in advanced solid tumors.

FIGURE 1.

Design and mechanism of the switchable hydrolase‐based nanoplatform (SI@PAP) targeting the ECM‐integrin‐cholesterol signaling network. (a) Construction of SI@PAP by co‐assembling the natural PAP with the cholesterol synthesis inhibitor Siv and the photothermal agent ICG, followed by stabilization through disulfide bonds. (b) Proposed mechanism of action. SI@PAP accumulates in tumors and is selectively activated by GSH. NIR irradiation triggers ICG‐mediated photothermal effects, while Siv induces apoptosis, together eliciting ICD (Step 1). ICD‐released GSH reactivates PAP to degrade ECM and soften the matrix, suppressing integrin‐driven mechanotransduction and PD‐L1 transcription (Step 2). Simultaneously, Siv depletes membrane cholesterol, disrupting lipid rafts and destabilizing integrin clusters and PD‐L1. These combined effects dismantle the ECM‐integrin‐cholesterol signaling network, promoting cytotoxic T cell infiltration and establishing a self‐amplifying antitumor immune cycle (Step 3).

2. Results and Discussion

2.1. Preparation and Characterization of a Dual‐Lock, Enzyme‐Activity‐Switchable Nanoplatform to Disrupt ECM‐Integrin‐Cholesterol Network

To validate our design concept, we constructed PAP‐based nanovehicles co‐loaded with Siv and ICG through a self‐assembly process, in which intermolecular disulfide bonds stabilized the vesicular structure (Figure 2a). Upon GSH addition, disulfide bonds within PAP were cleaved to generate free thiols, increasing hydrogen bond donors and molecular binding sites, which promoted the assembly of PAP with small molecules. To elucidate the specific mechanism by which GSH influences the drug‐loading process, we simulated the drug‐loading process using PAP and GSH‐PAP through the global adsorption of small molecules onto proteins. After 200 ns of molecular dynamics (MD) simulation, both systems formed distinct “drug‐protein” assemblies (Figure 2b–e). MD simulations showed that GSH treatment enhanced the binding of Siv and ICG to PAP, increasing hydrogen bonding by 20% and 63%, respectively (Figure S1a,b), and improving the simulated drug‐loading capacity by 44.5% for Siv and 150% for ICG (Figure 2f). Upon oxidation, the free thiols reformed interprotein disulfide bonds, driving the self‐assembly of stable nanovesicles (Figure S2a). Transmission electron microscopy (TEM) revealed that SI@PAP nanovesicles exhibited a monodisperse and homogeneous spherical distribution with an average size of approximately 120 nm (Figure 2g), while hydrodynamic light scattering confirmed a hydrated size of ∼129 nm and a positive surface charge of ∼24.0 mV (Figure S3a,b). The hydrodynamic size remained stable under physiological conditions upon 6 days incubation (Figure S3c), indicating excellent stability. Furthermore, the hydrated size of SI@PAP showed no significant change after lyophilization (Figure S3d), suggesting that the lyophilized formulation is highly stable and suitable for transportation and storage. The encapsulation efficiencies of Siv and ICG were 11.8% and 8.35%, respectively, as determined by high‐performance liquid chromatography (HPLC).

FIGURE 2.

Preparation and characterization of SI@PAP. (a) Illustration of SI@PAP preparation. MD snapshots of (b) PAP and (c) GSH‐PAP with Siv and ICG during the initial state. MD snapshots of (d) PAP and (e) GSH‐PAP with Siv and ICG during 200 ns simulation. (f) Drug loading ratio of PAP/GSH‐PAP toward Siv/ICG (200 ns). (g) TEM image with size distribution (inset) of SI@PAP. (h) Release profile of Siv from SI@PAP under different conditions (n = 3). (i) 2D Density heatmap for PAP (yellow) interacting with Siv (blue) and ICG (purple). (j) 2D Density heatmap for GSH‐PAP (yellow) interacting with Siv (blue) and ICG (purple). (k) The average interactions between the PAP/GSH‐PAP and Siv/ICG. (l) Enzyme activity of PAP and SI@PAP with or without GSH at 1 h (n = 3). (m) Enzyme activity of PAP at different pH values. (n) Temperature curves of SI@PAP solutions at different concentrations under 808 nm laser irradiation (1.0 W cm⁻²). (o) Temperature of SI@PAP aqueous solutions (35 µg ml⁻¹) under three cycles of laser ON/OFF at 1.0 W cm⁻². (p) Enzyme activity of PAP at different temperatures. Data represents the mean ± s.d.

Subsequently, we investigated the release profile of Siv under the high GSH concentration and weakly acidic pH using Ultra‐high‐performance liquid chromatography quadrupole‐orbitrap high‐resolution mass spectrometer (UHPLC‐Q‐Orbitrap HRMS). Our findings revealed a markedly accelerated release of Siv under high‐GSH and acidic conditions (Figure 2h). However, owing to the short half‐life of ICG in aqueous solutions [28, 29], its release profile could not be precisely quantified by UHPLC‐Q‐Orbitrap HRMS. To gather evidence for ICG release, we examined the drug release from the prepared SI@PAP nanovesicles using TEM. The TEM images indicated that upon incubation with high GSH, the nanovesicles underwent disintegration, forming smaller spheres that resembled PAP but exhibited higher contrast (Figure S2a,b), suggesting that some drug molecules remained bound to the released PAP. To discern the species of drug molecules, we constructed a 2D density distribution heat map to characterize the positions of Siv and ICG relative to PAP following GSH treatment. The heat map visualization revealed distinct spatial distributions, with ICG maintaining closer proximity to PAP compared to Siv after GSH addition (Figure 2i,j), which was further confirmed through radial distribution function analysis (Figure S1c,d). Subsequently, we calculated the molecular interactions between PAP and Siv or ICG following GSH treatment. The interaction analyses demonstrated a significant increase in both electrostatic (49.5%) and van der Waals (76.7%) interactions between PAP and ICG following GSH treatment (Figure 2k). In contrast, the interactions between PAP and Siv exhibited decreased interaction strengths, showing 13.4% and 10.7% reductions in electrostatic and van der Waals interactions, respectively (Figure 2k). These findings suggest that the differential interaction strengths following GSH treatment contribute to distinct drug distribution: the enhanced molecular interactions between PAP and ICG promote ICG retention, while the weakened interactions facilitate Siv release. This differential release ensures rapid cholesterol depletion by Siv while retaining ICG for photothermal activation. TEM images under weakly acidic pH conditions demonstrated a gradual release of small‐molecule drugs and a 36 h incubation resulted in the formation of hollow nanoparticles (Figure S2c), indicating that a weakly acidic environment facilitates the release of both drugs. To elucidate the underlying mechanisms of this phenomenon, we conducted MD analysis to examine the root‐mean‐square deviation (RMSD) of Siv and ICG under different pH conditions. MD analysis demonstrated that RMSD exhibited an inverse relationship with pH within a specific range (Figure S1e,f), suggesting that the release of both Siv and ICG is significantly enhanced in the weakly acidic environment.

Given the risk of off‐target proteolysis with systemic enzyme delivery [30], we engineered a dual safety lock to precisely regulate PAP activity. Under physiological GSH levels (<20 µm), SI@PAP remained enzymatically inactive, as confirmed by a 2,4,6‐trinitrobenzene sulfonic acid‐based protease activity assay (Figure 2l) and circular dichroism (Figure S3e). Given the difference in extracellular (∼2–20 µm) and intracellular (∼2–10 mm) GSH concentrations in tumor microenvironment [31, 32], PAP activity was restored only at GSH concentrations ≥60 µm (Figure 2l), a threshold substantially higher than in normal tissues but achievable in tumors following immunogenic cell death (ICD) and localized GSH release (Figure S3f–h). Once internalized into lysosomes (pH 4.5–5.0), PAP was denatured and irreversibly inactivated (Figure 2m) [33], further ensuring spatially confined ECM degradation. This GSH‐responsive “switch” mechanism enables matrix remodeling to occur only in the tumor extracellular milieu, thereby avoiding systemic toxicity.

SI@PAP exhibited robust photothermal properties. Upon 808 nm near‐infrared (NIR) irradiation, the temperature of SI@PAP solutions rose in a concentration‐ and power‐dependent manner with consistent heating profiles over repeated cycles (Figure 2n,o and Figure S3i). Quantitative analysis revealed a photothermal conversion efficiency of 52.45% (Figure S3j,k), indicating robust photothermal capability. Notably, PAP enzymatic activity increased within the therapeutic PTT temperature window (50°C–70°C), indicating that local photothermal heating further boosts enzymatic ECM degradation (Figure 2p). In addition, ICG within SI@PAP generated singlet oxygen, verified by the oxidation‐dependent decrease of 1,3‐diphenylisobenzofuran absorbance (Figure S3l), supporting a secondary photodynamic effect.

Together, these findings demonstrate that SI@PAP maintains stability and enzymatic quiescence during circulation, undergoes selective activation in GSH‐enriched tumor microenvironment, and leverages photothermal stimulation to enhance localized ECM degradation. This design enables simultaneous cholesterol depletion, integrin signaling suppression, and precise ECM softening, establishing a foundation for dismantling the ECM‐integrin‐cholesterol network and overcoming immune evasion.

2.2. Dual Regulatory Role of SI@PAP in Cholesterol‐Dependent PD‐L1 Destabilization and Integrin‐FAK Mechanotransduction Remodeling

To elucidate the intracellular behavior and signaling consequences of SI@PAP, we first visualized its uptake using confocal fluorescence microscopy by tracking the ICG‐derived red fluorescence. 4T1 tumor cells treated with SI@PAP showed markedly stronger intracellular fluorescence compared with free ICG, indicating that the PAP‐based nanovehicle substantially enhanced cellular internalization (Figure 3b). Importantly, SI@PAP displayed negligible cytotoxicity toward normal 3T3 fibroblasts, which was attributed to the low intracellular GSH levels in normal cells and the rapid denaturation of PAP within lysosomes, highlighting the biocompatibility and tumor selectivity of this design (Figure 3c).

FIGURE 3.

SI@PAP degrades the intracellular PD‐L1 and blocks the Integrin signaling pathway by cholesterol synthesis inhibition. (a) Schematic illustration of SI@PAP‐induced degradation of PD‐L1 and integrin by cholesterol synthesis inhibition. (b) Confocal image of intracellular ICG (red) distribution in 4T1 tumor cells. (c) 3T3 cell viability after different treatments (n = 4). (d) Cholesterol levels of 4T1 cells after different treatment (n = 3). (e) WB analysis of PD‐L1 protein in 4T1 cells treated with CHOL at different concentration. (f) WB analysis of PD‐L1 protein in 4T1 cells after treatment with PBS, Siv1 (1 µg ml⁻¹), Siv 4 (4 µg ml⁻¹) and SI@PAP (Siv: 4 µg ml⁻¹). (g) Confocal image of PD‐L1 distribution in 4T1 tumor cells. (h) Confocal image of membrane lipid rafts in 4T1 cells. (i) WB analysis of Integrin and FAK protein in 4T1 cells. (j) AFM image of living 4T1 cells. (k) Quantification for the measurements of nanoindentation (n = 10). Data are analysed by two tailed Student T‐test and shown as the mean ± s.d. ^* p < 0.05; ^** p < 0.01; ^*** p < 0.001; ^**** p < 0.0001.

Given that cholesterol is a central structural component of lipid rafts that stabilizes membrane proteins such as PD‐L1 and integrins [21, 24], we hypothesized that its depletion would promote ubiquitin‐mediated degradation of PD‐L1 and modulate the cellular abundance of mechanotransduction‐related proteins, including integrins and focal adhesion kinase (FAK) (Figure 3a). To confirm this hypothesis, we assessed the capability of Siv in reducing the levels of cholesterol by inhibiting the function of HMGCR [34]. We quantitatively assessed the cholesterol levels by fluorescence measurement of Amplex Red reagent in the presence of cholesterol oxidation‐produced hydrogen peroxide and horseradish peroxidase, which revealed a significant Siv‐induced reduction in total cellular cholesterol content (Figure 3d). When oxidizing cellular membrane‐associated cholesterol with cholesterol oxidase to detect the intracellular cholesterol content and then subtracting it from the total cholesterol concentration to obtain the content of plasma membrane cholesterol, we observed the same tendency (Figure 3d). Consistent with in vitro results, we also observed a decreased cholesterol content within Siv or SI@PAP‐treated 4T1 tumors (Figure S4a).

Given the well‐established role of cholesterol in stabilizing PD‐L1 expression on the cell membrane, we systematically evaluated PD‐L1 degradation following cholesterol modulation in the 4T1 cell line. Western blot (WB) analysis demonstrated that cholesterol supplementation upregulated PD‐L1 expression, whereas Siv or SI@PAP treatment led to significant PD‐L1 downregulation, with the most substantial reduction observed in the SI@PAP + L group, likely due to photothermal‐enhanced Siv release (Figure 3e,f). Immunofluorescence staining corroborated these findings and revealed that cholesterol enrichment preserved PD‐L1 at the cell membrane, while SI@PAP+L treatment caused marked PD‐L1 loss (Figure 3g). Since cholesterol‐rich lipid rafts provide a signaling hub for integrin clustering, we next examined raft organization using Alexa Fluor 488‐conjugated cholera toxin subunit B. Confocal imaging revealed that cholesterol supplementation increased raft integrity, whereas Siv or SI@PAP treatment disrupted raft continuity, with the most pronounced disruption again seen in the SI@PAP + L group (Figure 3h). In parallel, WB analysis showed that Siv‐mediated cholesterol depletion led to significant downregulation of integrins and FAK, two key mediators of mechanotransduction, confirming that lipid raft destabilization impairs the ECM‐integrin signaling axis (Figure 3i). Finally, to directly assess the biophysical consequences of cholesterol modulation, we used atomic force microscopy (AFM) to measure cortical stiffness of live 4T1 cells. Cholesterol supplementation reduced cortical stiffness by 47%, whereas SI@PAP treatment increased it by 2.34‐fold compared to control (Figure 3j,k). Together, these data reveal a multi‐tiered mechanism in which SI@PAP‐mediated cholesterol depletion destabilizes lipid rafts, triggers ubiquitin‐mediated degradation of PD‐L1 and integrins, and increases membrane stiffness, collectively dismantling the ECM–integrin–cholesterol signaling network.

2.3. SI@PAP Remodels ECM And Suppresses Integrin‐FAK Signaling via Dual Actions on Matrix Hydrolysis and Cholesterol Metabolism

Tumor ECM serves as both a survival environment for tumor cells and a physical barrier that impedes oxygen diffusion and nanoparticle penetration. More crucially, it influences tumor progression via cell‐ECM signaling, making modulation of the ECM an essential anti‐tumor strategy [12, 35]. However, the ECM exhibits dynamic remodeling under tumor regulation, necessitating interventions that can disrupt this regulatory feedback loop. Our approach leverages two complementary mechanisms: (1) PAP, a natural hydrolase that degrades ECM proteins to soften the matrix, and (2) Siv, which disrupts cholesterol‐dependent membrane microdomains through cholesterol depletion, leading to functional alterations in lipid raft‐localized mechanosensors, particularly integrin complexes and FAK‐mediated adhesion signaling [24]. By coupling these complementary mechanisms, SI@PAP remodels the tumor microenvironment at both the biochemical and biomechanical levels, weakening ECM‐mediated tumor signaling and immune evasion (Figure 4a).

FIGURE 4.

SI@PAP modulates tumor ECM to inhibit multiple oncogenic signaling pathways and block the synthesis of the immunosuppressive molecule PD‐L1. (a) Schematic diagram depicting tumor ECM‐induced activation of multiple oncogenic signaling pathways by mechanical transduction protein integrin. (b) Immunofluorescence images of Coll‐I (green), Coll‐III (red), Fibronectin (green) and Laminin (red) in 4T1 primary tumor sections. (c) Elastic modulus of 4T1 tumor tissues (n = 5). (d) 4T1 tumor opening analysis (n = 5) after different treatments. (e) Confocal image of YAP (red) and PD‐L1 (green) in 4T1 primary tumor sections. (f) Volcano plot showing key differentially expressed genes between SI@PAP+L and Control group. Significantly enriched (g) GO biological processes and (h) KEGG pathways between SI@PAP+L and Control group. (i) Clustered heatmap of integrin‐related genes in 4T1 tumor tissues between SI@PAP+L and Control group. (j) Quantification of pathway‐related gene expression between SI@PAP+L and Control group (n = 3). Data are analyzed by two tailed Student t‐test and shown as the mean ± s.d. ^* p < 0.05; ^** p < 0.01; ^*** p < 0.001; ^**** p < 0.0001.

Immunofluorescence staining confirmed that PAP treatment markedly degraded the backbone ECM proteins collagen I (Col‐I) and collagen III (Col‐III), which are predominantly secreted by tumor cells [36], as well as interstitial components such as fibronectin and laminin. Siv treatment also reduced Col‐I/III deposition, although indirectly, by inhibiting cholesterol‐driven integrin signaling that regulates ECM synthesis (Figure 4b). Concurrently, photothermal effects induced by ICG (ICG+L) also reduced the content of intracellular collagen (Figure 4b). Photothermal activation (ICG+L) further amplified these effects, with the SI@PAP and SI@PAP+L groups showing the most pronounced loss of intact ECM fibers and the greatest accumulation of fragmented protein structures (Figure 4b). We noted that exposure of PAP induced a downregulation of ECM components fibronectin and laminin, accompanying the morphology changes from intact fiber structure to protein fragments, while Siv has no effect (Figure 4b). Thus, PAP enzymatically disassembles the ECM, whereas Siv suppresses ECM production by impairing tumor cell–ECM signaling, demonstrating a cooperative effect on matrix remodeling.

To confirm that degrading matrix proteins and modulating cell‐ECM signaling inevitably soften ECM, we measured the tissue stiffness of 4T1 tumors following treatment with PBS (Control group), SI@PAP and SI@PAP+L by elastic modulus measurements and tumor opening analysis. Both treatment groups exhibited significant tumor softening compared to the control group (Figure 4c,d and Figure S4b). Because YAP acts as a key mechanosensor that translocate to the nucleus in response to matrix stiffening, ECM softening led to cytoplasmic retention of YAP and a consequent reduction in PD‐L1 transcription (Figure 4e) [37]. Confocal imaging further showed decreased nuclear localization of multiple oncogenic transcription factors, including STAT1, MYC, NF‐κB, and c‐Jun, following SI@PAP or SI@PAP+L treatment (Figure S5). These changes collectively suppressed PD‐L1 expression, thereby enhancing antitumor immune responses. In addition, ECM softening alleviated tumor hypoxia, as reflected by reduced expression of the oxygen‐sensitive transcription factor HIF‐1α (Figure S4c) [38], which may further potentiate oxygen‐dependent PDT efficacy.

To dissect the molecular basis of these effects, we performed transcriptomic sequencing of 4T1 tumors treated with PBS or SI@PAP+L. This analysis revealed extensive transcriptional reprogramming, with 9303 genes differentially expressed (fold change >2, p<0.05), including 5300 upregulated and 4003 downregulated genes (Figure 4f). Gene ontology (GO) analysis revealed that gene differential expression was enriched in the processes related to cholesterol metabolism, ECM organization, and immune response (Figure 4g). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis demonstrated substantial modulation of key oncogenic signaling pathways, including RAS, PI3K‐AKT, NF‐κB, and JAK‐STAT (Figure 4h), which regulate PD‐L1 transcription through nuclear translocation of factors such as STAT1, c‐Jun, NF‐κB, and MYC. q‐PCR assay demonstrated significant downregulation of these signaling factors, along with reduced expression of HMGCR and components of the integrin‐FAK mechanotransduction pathway (Figure 4j). In particular, different types of integrin expression were downregulated (Figure 4i). Therefore, we deduced that ECM modulation affects multiple intracellular signaling pathways through integrin‐mediated transduction, ultimately inhibiting oncogene activation and PD‐L1 production. Moreover, cholesterol depletion not only disrupts lipid raft integrity but also broadly attenuates integrin‐mediated mechanotransduction, thereby dismantling a central signaling network that links ECM mechanics to tumor immune evasion.

2.4. Synergistic Cholesterol Depletion and ECM Remodeling by SI@PAP Potentiates Photo‐Immunotherapy

To further validate the therapeutic potential of SI@PAP and its ability to disrupt the ECM‐integrin‐cholesterol signaling axis, we first examined its photodynamic effect in vitro. ROS generation was monitored using the oxidation‐sensitive probe 2′,7′‐dichlorodihydrofluorescein diacetate (DCFH‐DA), which is intracellularly de‐esterified to non‐fluorescent DCFH and subsequently oxidized by ROS into fluorescent 2′,7′‐dichlorofluorescein. Upon laser irradiation, 4T1 cells treated with SI@PAP displayed markedly higher ROS production compared with free ICG treatment (Figure 5a), indicating that PAP‐based delivery enhances intracellular ICG accumulation and potentiates PDT. Beyond enhancing phototherapy, SI@PAP significantly improved the intracellular delivery of Siv, as evidenced by the enhanced cytotoxicity toward 4T1 cells relative to free Siv (Figure 5b). This cytotoxicity was attributed to cholesterol depletion–mediated apoptosis, confirmed by robust TUNEL staining of treated cells (Figure S4d). The combination of Siv‐induced apoptosis and photothermal activation in the SI@PAP+L group resulted in the greatest reduction of cell viability, demonstrating strong synergistic therapeutic efficacy. We next investigated the in vivo tumor‐targeting ability of SI@PAP using the intrinsic fluorescence of ICG to track nanoparticle biodistribution in BALB/c mice bearing 4T1 tumors. Following intravenous injection, SI@PAP displayed significantly higher tumor accumulation and prolonged retention compared to free ICG (Figure 5c). Ex vivo fluorescence imaging of dissected tumors confirmed persistent SI@PAP accumulation for up to 72 h post‐injection (Figure 5d). Biodistribution analysis further validated the efficient tumor homing and retention of the PAP‐based vehicle, providing a favorable foundation for subsequent therapeutic studies.

FIGURE 5.

Infusion of SI@PAP mediates solid tumor clearance in primary early‐stage and advanced‐stage tumor models. (a) Confocal image of intracellular ROS levels in 4T1 tumor cells. (b) Viability of 4T1 cells after different treatment (n = 4). (c) Time‐dependent fluorescence imaging of 4T1 tumor‐bearing mice. (d) Fluorescence imaging of the main organs after i.v. injection for 72 h. (e) Operation schedule of the treatments and evaluations in early‐stage tumor model. (f) Tumor volume changes in early‐stage tumor model. (n = 6). (g) Survival curves of mice in early‐stage tumor model. (h) Operation schedule of the treatments and evaluations in advanced‐stage tumor model. (i) Tumor volume and (j) tumor weight change in advanced‐stage tumor model (n = 5). (k) HE, TUNEL and KI67 staining in advanced‐stage tumor model. Data are analyzed by two tailed Student t‐test and shown as the mean ± s.d. ^* p < 0.05; ^** p < 0.01; ^*** p < 0.001; ^**** p < 0.0001.

To evaluate antitumor efficacy in vivo, we established two treatment models using subcutaneous 4T1 tumors at different stages: early‐stage (120 mm³) and advanced‐stage (500 mm³) tumors. Tumor‐bearing mice were divided into six treatment groups randomly (Control, PAP, ICG+L, Siv, SI@PAP and SI@PAP+L) and received intravenous injections (dose of PAP, Siv and ICG was 35, 5, and 3.75 mg·kg⁻¹, respectively) every other 2 days for a total of two treatments. Laser irradiation for 5 min at a power density of 1.0 W cm⁻² was administered after injection of ICG or SI@PAP for 24 h. Tumor growth was monitored every 2 days, with tissues collected for analysis on day 22 or 25, respectively (Figure 5e,h). The tumor exhibited rapid progression in mice following treatment with PBS (Control group), with only marginal growth inhibition observed in groups following treatment with PAP, Siv and ICG+L (Figure 5f,i). In contrast, both SI@PAP and SI@PAP+L treatments demonstrated remarkable tumor suppression. Notably, the SI@PAP+L group achieved complete tumor regression in the early‐stage model (120 mm³) (Figure 5f) and a 76% inhibition rate in the advanced‐stage model (500 mm³) (Figure 5g, Figure S4f). Mice with complete tumor regression showed significantly prolonged survival, maintaining a 50% survival rate at 60 days post‐treatment (Figure 5g). Hematoxylin‐eosin (HE) staining, TUNEL staining and KI67 staining results confirmed that SI@PAP+L induced extensive tumor tissue damage and inhibited cell proliferation (Figure 5k). Importantly, we did not observe obvious changes in body weight during treatment (Figure S4e,g), and HE analysis of major organs revealed no pathological changes (Figure S6), indicating favorable biosafety. Blood biochemical and hematological analyses showed no significant differences between SI@PAP+L and Control groups (Figure S7), further supporting the system's safety profile. Collectively, these results demonstrate that SI@PAP efficiently targets tumors, depletes cholesterol to disrupt integrin‐mediated signaling, and synergizes with photothermal stimulation to achieve potent tumor ECM remodeling and immune checkpoint suppression. By integrating spatially confined enzymatic ECM degradation with cholesterol‐dependent signaling disruption, SI@PAP not only eradicates early‐stage tumors but also significantly inhibits advanced‐stage tumor progression, offering a promising therapeutic strategy for precise and effective cancer immunotherapy.

2.5. SI@PAP‐Mediated ECM Degradation Induces Immune Cell Infiltration and Long‐Term Antitumor Immune Response

To systematically reveal the immune‐centric molecular circuitry governing SI@PAP's antitumor response, we first investigated its ability to induce ICD. As established in the literature, both phototherapy and apoptosis can trigger ICD, characterized by sequential ecto‐calreticulin (CRT) exposure, high mobility group protein 1 (HMGB1) translocation, and adenosine triphosphate (ATP) release, which constitute damage‐associated molecular patterns (DAMPs) triad for promoting the activation of antigen‐presenting cells, particularly dendritic cells (DCs), thereby initiating antitumor immune responses [25]. We systematically evaluated ICD induction by SI@PAP through multiple approaches. Immunofluorescence staining and flow cytometry were employed to quantify CRT exposure and HMGB1 translocation, while ATP release was measured using a luciferase‐based bioluminescence assay. As expected, both ICG‐mediated phototherapy and Siv‐induced apoptosis significantly increased CRT exposure, as evidenced by enhanced green fluorescence intensity (Figure S8a) and increased CRT‐positive 4T1 cell populations (Figure 6a). Concurrently, we observed substantial HMGB1 translocation from the nucleus to the extracellular space following treatment with SI@PAP+L, demonstrated by decreased intracellular fluorescence (Figure S8b) and reduced HMGB1‐positive cell populations (Figure 6b). These changes were accompanied by elevated ATP secretion (Figure 6c). The SI@PAP+L treatment group exhibited the most pronounced effects, with maximal upregulation of CRT exposure, HMGB1 translocation, and ATP secretion (Figure 6a–c and Figure S8a,b). Additionally, immunofluorescence staining of tumor tissues further confirmed that SI@PAP+L treatment significantly promoted the surface exposure of CRT and the extracellular release of HMGB1 (Figure S8c). These synergistic effects suggest that ECM degradation and cholesterol depletion cooperatively enhance tumor immunogenicity, thereby amplifying ICD.

FIGURE 6.

SI@PAP mediates antitumor immune response by tumor ECM modulation and cholesterol synthesis inhibition against tumor rechallenge and metastasis. Average frequencies of (a) CRT⁺ and (b) HMGB1⁺ cells among 4T1 tumor cells (n = 3). (c) Quantification of ATP secretion from 4T1 tumors (n = 3). (d) Evaluation of mature DCs after incubation with pre‐treated 4T1 cell fragments. (e) Operation schedule of the treatments, immune response evaluation and rechallenge evaluation in 4T1 tumor‐bearing mice. The relative abundance of (f) mature DCs, (g) helper T cells, (h) CTLs, and (i) Tregs in 4T1 primary tumors (n = 4). Average frequencies of (j) T_CM and (k) T_EM among lymphocyte in 4T1 primary tumors (n = 3). Average frequencies of (l) T_CM and (m) T_EM among lymphocyte in spleens (n = 3). (n) Confocal imaging of CD4⁺ T cells (red) and CD8⁺ T cells (green) in 4T1 primary tumor sections. (o) 4T1 cell viability after different treatment. (p) Tumor volume changes in rechallenge tumor model (n = 5). (q) Indian ink staining and (r) HE staining for the lung in the lung metastasis model. Data are analyzed by two tailed Student t‐test and shown as the mean ± s.d. ^* p < 0.05; ^** p < 0.01; ^*** p < 0.001; ^**** p < 0.0001.

We next investigated downstream immune activation. Immature DCs, derived from cytokine‐stimulated murine bone marrow cells, were co‐cultured with 4T1 tumor cell fragments pre‐treated with PBS (Control group), ICG, ICG+L, SI@PAP and SI@PAP+L, and DC maturation was characterized by cell surface markers CD11C, CD80, and CD86 revealed by flow cytometry measurement, which demonstrates a significant enhancement in DC maturation following stimulation with 4T1 tumor cell fragments pre‐treated with ICG+L, SI@PAP and SI@PAP+L. Notably, the SI@PAP+L treatment group exhibited the most pronounced effect, demonstrating an approximately 1.9‐fold increase in DC maturation relative to the control group (Figure 6d). These results demonstrate the efficiency of combination of cholesterol depletion and phototherapeutic in enhancing tumor immunogenicity and stimulating DCs maturation. To further investigate ICD induction in tumor tissue, we performed comprehensive immune cell analysis using flow cytometry on tumor specimens collected from 4T1 tumor‐bearing mice one day post‐treatment (day 25). The tumors following treatment with SI@PAP+L exhibited a significant increase in mature antigen‐presenting DCs compared to other treatment groups (Figure 6f and Figure S10d), which can be attributed to the synergistic effects of phototherapy and ECM modulation. Moreover, flow cytometric analysis revealed a substantial increase in both cytotoxic T lymphocytes (CTLs) and helper T cells within the tumor microenvironment following treatment with SI@PAP and SI@PAP+L compared to the control group (Figure 6g,h and Figure S10e). Notably, this enhancement was more pronounced than that observed with nanomedicines lacking ECM degradation capability (Figure S9a–e), underscoring the critical role of ECM remodeling in facilitating immune cell infiltration. Immunofluorescence staining further confirmed increased penetration depth of CTLs and helper T cells in SI@PAP and SI@PAP+L‐treated groups compared to the control group (Figure 6n).

The cytotoxic of infiltrating CTLs is significantly influenced by the immunosuppressive tumor microenvironment. Regulatory T cells (Tregs) subsets, in concert with IL‐10/TGF‐β‐dominated cytokine milieus, contribute to tumor immune escape by inhibiting DC and CTL function, thereby maintaining immune tolerance [39]. As established in the literature, oncogenic signaling pathways such as NF‐κB can promote immune evasion by enhancing Treg activation and immunosuppressive cytokine production [39]. Consequently, simultaneous inhibition of multiple oncogenic signaling pathways represents a promising therapeutic strategy for reversing the tumor immunosuppressive microenvironment. As anticipated, SI@PAP demonstrated significant regulatory effects on oncogenic signaling pathways, leading to a substantial reduction in both the tumor‐infiltrating Treg population (Figure 6i and Figure S10f) and the levels of IL‐10 and TGF‐β (Figure S11a,b), thereby reversing immunosuppressive signaling and reactivating effector CTLs. This immune reprogramming was accompanied by elevated production of pro‐inflammatory cytokines TNF‐α and IFN‐γ (Figure S11c,d). These findings collectively demonstrate that SI@PAP effectively induces immune responses against tumor by converting the tumor microenvironment from an immunologically suppressed “cold” into an immunologically active “hot” phenotype. Moreover, the cytotoxic activity of CTLs is primarily mediated through the secretion of cytotoxic granules containing perforin (PRF) and granzyme B (GzmB) [40, 41]. PRF forms transmembrane pores in target cells, facilitating GzmB entry and subsequent induction of apoptosis (Figure 6o) [40]. To investigate the relationship between membrane stiffness and PRF efficacy, we examined tumor cell susceptibility to PRF‐mediated cytotoxicity under conditions of cholesterol supplementation and depletion. Cholesterol supplementation significantly reduced PRF‐induced cytotoxicity, while SI@PAP‐treated 4T1 cells showed increased susceptibility to PRF‐mediated killing (Figure 6o). These results highlight the role of the ECM‐integrin‐cholesterol axis in regulating CTL function, linking physical ECM remodeling to intracellular signaling modulation.

Beyond immediate antitumor effects, SI@PAP also established durable immune memory. Flow cytometry revealed substantial expansion of both central memory (T_CM) and effector memory (T_EM) T cell subsets in tumors and spleens of treated mice (Figure 6j–m and Figure S12b–e), indicating the establishment of robust anti‐tumor immune memory that could potentially prevent tumor recurrence, which was proved through establishing a secondary tumor model by reinoculating 4T1 tumor cells into long‐term survivors from the SI@PAP+L group (Figure 6e). Age‐matched naive BALB/c mice receiving 4T1 tumor cell implantation served as the control group. Notably, SI@PAP+L‐treated survivors exhibited significantly delayed tumor growth, with 40% showing complete tumor rejection, compared to control mice, which demonstrated rapid tumor progression (Figure 6p). The rechallenge evaluation revealed that SI@PAP+L treatment can establish long‐lasting anti‐tumor immunity response for long‐term tumor regression. Given the rich vascularity and susceptibility to metastatic dissemination of the lungs [42], we further investigated SI@PAP's potential to inhibit tumor metastasis using a metastasis model of lung, which was made by intravenously administrating 4T1 tumor cells into the long‐term survivors from the SI@PAP+L group (Figure 6e). Age‐matched naive BALB/c mice receiving 4T1 tumor cell intravenously served as the control group. Both India ink staining and HE analysis confirmed that SI@PAP+L treatment completely inhibited metastatic progression, while control mice exhibited extensive pulmonary metastases (Figure 6q,r). This demonstrates that SI@PAP not only eradicates primary tumors but also induces systemic immune surveillance that prevents metastatic spread.

3. Conclusion

In summary, our study successfully designed and validated a dual‐safety‐lock natural enzyme nanoplatform, SI@PAP, presenting an innovative strategy to overcome tumor immune resistance. Our central contribution lies in elucidating and synergistically targeting a previously underappreciated “ECM‐integrin‐cholesterol” signaling network within the immunosuppressive tumor microenvironment. Specifically, we delineated a key mechanism whereby ECM stiffness orchestrates immune suppression by promoting PD‐L1 transcription via the integrin‐FAK mechanotransduction pathway, which activates oncogenic transcription factors like YAP. Concurrently, we demonstrated that cholesterol metabolism sustains immune evasion by stabilizing both PD‐L1 and integrins within lipid rafts at the post‐translational level. Capitalizing on these insights, the SI@PAP platform employs photo thermally‐activated enzymatic ECM softening to disrupt this pro‐tumorigenic mechanical signaling loop, thereby transcriptionally suppressing PD‐L1. In parallel, simvastatin‐mediated cholesterol depletion disrupts lipid raft integrity, achieving post‐translational destabilization of membrane‐bound PD‐L1. This “full‐spectrum” PD‐L1 regulation, combined with phototherapy‐induced immunogenic cell death, effectively converted immunologically “cold” tumors into “hot” ones, inducing durable tumor regression and establishing robust immune memory even in advanced tumor models.

Funding

This work was supported by the National Science Fund for Distinguished Young Scholars (22025406), the National Natural Science Foundation of China (22204162), and the Natural Science Foundation of Tianjin City (24JCYBJC01810).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll73877‐sup‐0001‐SuppMat.docx.

Data Availability Statement

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