Hypoxia-activated PROTAC for dual inhibition of FAK and EGFR enables synergistic mechano-chemical cancer therapy Authors: Keli An, Yuan Li, Chunyu Cao, Ming Wang, Xiaohui Jia, Guorui Jin, Bo Cheng, Yuanbo Jia, Lin Wang, Hui Guo, Feng Xu, Kai Qu Published: 2026 ABSTRACT AbstractMechanical abnormalities in the tumor microenvironment (TME), such as matrix stiffening and hypoxia, drive tumor invasion and therapy resistance. Yet targeting these cues with spatial precision remains challenging. Here we present a hypoxia-responsive dual-warhead PROTAC (hrFP-E) that couples focal adhesion kinase (FAK) degradation with inhibition of epidermal growth factor receptor (EGFR), introducing a mechano-chemical therapy paradigm. hrFP-E is equipped with a nitroreductase (NTR)–sensitive motif activated in hypoxic regions to release a FAK degrader (FP) and an EGFR inhibitor (Erlotinib). In lung cancer models, hrFP-E induces near-complete FAK depletion (∼96–99%) and substantial EGFR pathway suppression (∼88%), resulting in synergistic anti-migration and anti-proliferation effects in both 2D and 3D assays. Mechanistically, FAK degradation stabilizes large focal-adhesions, alters actomyosin contractility, elevates ROS, and reduces collagen deposition, collectively reprogramming the TME. In vivo, hrFP-E achieves ∼66% tumor growth inhibition without overt toxicity while reducing tissue stiffness and collagen density. This platform is inherently modular and compatible with alternative oncogenic drivers and disease-specific gates. Our work establishes mechano-chemical therapeutics, spatiotemporally controlled degraders that rewire tumor mechanics alongside growth signaling, as a generalizable strategy for solid tumor treatment.Keywords: Tumor mechanical microenvironment, Mechanomedicine, Hypoxia-responsive PROTAC, Focal adhesion kinase Introduction Targeted cancer therapies against single oncogenic pathways have achieved notable successes but often succumb to treatment resistance due to the adaptive plasticity of tumor signaling networks[1]. In particular, inhibitors of the epidermal growth factor receptor (EGFR) – a prototypical oncogenic driver – prolong survival in EGFR-mutant cancers such as non-small cell lung cancer, yet nearly all patients eventually relapse as tumors bypass the blocked pathway[1,2]. One major compensatory mechanism is the activation of cell-adhesion-mediated signaling through focal adhesion kinase (FAK), a non-receptor tyrosine kinase that orchestrates mechanotransduction[3]. FAK is frequently overexpressed in aggressive tumors and correlates with poor prognosis[4,5]. Its activation drives tumor cell survival, invasion, metastasis, and even therapy resistance via both its kinase activity and scaffold protein interactions[6]. Notably, FAK signaling (for example, via integrin αVβ3) has been shown to confer resistance to EGFR tyrosine kinase inhibitors in EGFR-mutant cancers[2]. These findings underscore that concurrent targeting of growth signals (EGFR) and mechanotransduction pathways (FAK) could yield synergistic anti-tumor effects and prevent resistance that arises from pathway crosstalk. Combination therapies co-inhibiting multiple targets are conceptually attractive and indeed have shown improved efficacy in preclinical and clinical studies[[7], [8], [9]]. For example, dual blockade of the MAPK pathway and FAK achieved significant responses in early trials[10]. However, simultaneously targeting two signaling kinases can exacerbate systemic toxicity in normal cells[11,12]. An ideal solution would selectively activate the therapy in tumor tissue while sparing normal cells. As hallmark of solid tumors, hypoxia provides a valuable trigger for such selectivity[13]. Due to abnormal vasculature and high oxygen consumption, hypoxia creating a hostile TME that promotes invasion and treatment failure[[14], [15], [16]]. Indeed, extensive efforts over decades have aimed to exploit hypoxia with prodrugs that release cytotoxins only in low-oxygen zones[[17], [18], [19]]. Some hypoxia-activated prodrugs (e.g. nitroimidazole derivatives like evofosfamide) showed promise in preclinical models, but fails to significantly improve patient survivals in clinical trials[20]. Regarding to inhibitors of kinases, like FAK, one major reason is overlooking the scaffolding functions of kinase proteins[21]. These failures highlight the need for more thorough inhibitory strategy for hypoxic TME targeting therapy. Proteolysis-targeting chimeras (PROTACs) have emerged as a revolutionary approach to achieve targeted protein degradation by harnessing the ubiquitin–proteasome system[22]. Unlike conventional inhibitors that simply block an active site, PROTACs eliminate the entire target protein, thereby ablating all enzymatic and scaffolding functions[23]. PROTAC-mediated degradation of FAK is being actively explored in cancer treatment[[23], [24], [25]]. This approach can overcome certain drug-resistance mechanisms (such as mutation-amplified signaling or non-enzymatic oncogenic functions) and has demonstrated greater selectivity in some cases, such as NSCLC with KRAS mutations[26] and ovarian cancer with NRAS mutations[27]. Nitroreductase (NTR), an enzyme highly enriched in hypoxic regions but minimally expressed in normal tissue, has been utilized to selectively cleave NTR-sensitive linkers[28,29], thereby achieving localized release of PROTACs within hypoxic tumors and substantial suppression on “on-target” toxicity by systemic administration of PROTACs[22]. In this work, we developed a dual-target therapeutic platform, designated hrFP-E, which integrates a hypoxia-responsive (hr) FAK PROTACs (FP) with the EGFR inhibitor Erlotinib (E). We demonstrated that hrFP-E is exclusively activated by hypoxia to efficiently degrade FAK and show superior suppression of related mechanotransduction comparing to conventional FAKi. In addition, hrFP-E exerts synergistic anti-migration and anti-proliferation effects on NSCLC cells under hypoxia by simultaneously degrading FAK and inhibiting EGFR signaling. To our knowledge, this is the first example of a PROTAC engineered with a tumor-specific trigger to achieve spatiotemporally mechano-chemical inhibition. Finally, in a co-implantation mouse model of human NSCLC cells and stromal fibroblasts, hrFP-E significantly enhances therapeutic efficacy through mechano-chemical synergy in vivo. Our findings highlight a novel mechano-chemical combination therapy approach and provide a proof-of-concept for hypoxia-activated dual targeting of FAK and EGFR in tumor treatment. Results PROTAC-mediated FAK degradation shows more potency in downregulating FAK activity and migration of NSCLC cells under hypoxic TME We first synthesized the FAK PROTAC (FP), which effectively degrades FAK in a dose-dependent manner (Fig. 1a). Additionally, we assessed the half-inhibitory concentrations (IC50) of FAK degradation and inhibition under both hypoxic and normoxic conditions. We observed no correlation between degradation and inhibition of FAK. Despite FP exhibiting superior FAK degradation efficiency, its cellular inhibition efficiency is lower than that of FAKi under normoxic conditions (IC50=6.624 μM for FP vs. 4.311 μM for FAKi). However, under hypoxic conditions, the cytostatic efficiencies of both treatments align closely (IC50=1.782 μM for FP vs. 1.757 μM for FAKi) (Fig. S1a, b), suggesting that FP is particularly effective in hypoxic TME. Substantial distinctions between FAK degradation and FAK inhibition under hypoxic conditions. a, Western blotting of FAK levels after FP or FAKi treatment at the indicated doses for 12 h. b, Western blotting of HIF-1α, FAK, and pFAK protein levels under hypoxic and normoxic conditions. Cells were treated with FP and FAKi (1 μM) for 12 h. c, Representative images of wound healing (n = 6) in H1975 cell culture after FP and FAKi (0.5 μM) treatment under hypoxia (scale bar: 200 μm). d, Western blotting of N-cadherin, E-cadherin levels. Cells were treated with different concentrations of FP and FAKi for 12 h under hypoxia conditions. e, Western blotting of pMLC level. Cells were treated with different concentrations of FP and FAKi for 12 h under hypoxia conditions. f, Top: Representative heat maps of traction stress. Bottom: Quantification of maximum cumulative force and average traction stress in native, FAKi or FP-treated H1975 cells by MATLAB (from left to right n = 26, 32, 33 cells). g, Top: Representative confocal images of paxillin and F-actin in different treatment groups (1 μM, scale bar: 10 μm). Bottom: Corresponding quantification of skeleton width (n = 20 fields of view) and adhesion length (from left to right n = 161, 142, and 147 adhesion sites in 47, 42, and 40 cells) of the different treatment groups. h, Schematic of altered motility under hypoxic conditions and effects of FAK protein. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 in e (one-way ANOVA). The essential responsive player of hypoxia, HIF-1α, and its downstream genes (GLUT1, PKM) both appear increased expression under hypoxic conditions, which are reduced by both FAKi and FP treatment (Fig. 1b). Surprisingly, FAKi treatment increases the expression of FAK although it suppresses the activity of FAK; in contrast, FP treatment eliminates both FAK expression and activity as expected (Fig. 1b). Then, we assessed the impact of FAK degradation on cell migration via scratch-wound assay under both hypoxic (1% O2) and normoxic (20% O2) conditions. The result shows that FP treatment significantly reduces cell motility under both hypoxic (17% vs. 70% closure) (Fig. 1c) and normoxic conditions (25% vs. 69% closure) (Fig. S1c) within 48 h. However, there is lack of significant difference between FAKi treatment and control group under hypoxia condition. Next, we examined the expression levels of E-cadherin and N-cadherin since the ratio of E-cadherin over N-cadherin (E/N ratio) indicates the strength of cell-cell junctions (Fig. 1d). The increased E/N ratio indicates the augment of cell-cell junction and reduction of cell migration. Different from the decreased E/N ratio in FAKi-treated cells, the E/N ratio of FP-treated cells maintain unchanged under hypoxic conditions (Fig. S1d). Moreover, the FP-treated group shows an unexpected increase in cell contraction compared to the untreated group (Fig. 1f). In contrast, the FAKi-treated group shows no significant changes. The augment of cell contraction was also confirmed by the upregulation of the phosphorylated myosin light chain (pMLC) (Fig. 1e, Fig. S1e). The increase in traction force typically promotes cell migration; however, we observed the opposite outcome in the scratch-wound assay. To explain this observation, we assessed the focal adhesion (FA) dynamic and cytoskeleton structure of cells after treatment. We found that FP treatment significantly amplifies the number and size of FAs as revealed by immunofluorescent images of paxillin (Fig. 1g), consistent with the significantly increased expression of paxillin (over 4-fold) (Fig. S1f, g). In addition, FP-treated cells have thicker F-actin fibers than untreated and FAKi-treated cells, which are preferably distributed in the cell periphery colocalizing with enlarged FAs. The removal of FAK by FP may promote the formation of more stable FA plaques, which could potentially suppress cell migration despite an increase in cell traction force after FP treatment. These results collectively highlight the significant difference between FAK degradation and inhibition of cell functionality, particularly under varying oxygen conditions (Fig. 1h). Novel Design of “Y”-shaped hrFP-E exhibits good hypoxia responsiveness in vitro Since the elevated NTR expression is the iconic feature of hypoxia, we chose NTR to trigger the release of FP and Erlotinib (targeted therapeutics for EGFR mutation). We developed a hypoxia-responsive (hr) PROTAC compound termed hrFP-E, which precisely degrades FAK and inhibits EGFR in NSCLC tissues. This rational design shields the PROTAC activity by incorporating a tri-functional linker, which hinders the formation of the critical E3 ligase-substrate-PROTAC ternary complex. In the hypoxic TME, the elevated NTR activates hrFP-E, therefore synergistically skilling cancer cells in a mechano-chemical manner (Fig. 2a-b). hrFP-E effectively hinders the formation of ternary complexes and exhibits good stimulus responsiveness in vitro. a, NTR reduces the nitro group in nitrobenzene to an amino group and subsequently initiates 1,6-elimination and hydrolysis, resulting in complete uncaging to generate FP and EGFRNTR. b, Molecular docking results of Defactinib with FAK protein. The docking results revealed the formation of five hydrogen bonds at the docking position, involving amino acids 428, 429, 430, 435, and 506. c, Structural formulae, and molecular docking results of ErlotinibNTR and Erlotinib. d, Western blotting of EGFR and pEGFR protein levels in H1975 cells. Cells were treated with different concentrations of Erlotinib and ErlotinibNTR for 24 h. e, Molecular docking analysis using the standard AutoDock protocol. f, Equilibrium dissociation constants (KD) of different selected inhibitor molecules with FAK, CRBN, and EGFR. The data in e are presented as mean ± s.d. (n = 3 independent experiments). g, Schematic triggering of hrFP-E. h, HPLC traces of hrFP-E (20 μM, labeled by #) in the presence of triggers (NTR (5 μg ml-1), and the gradual release of FP (labeled by *) and EGFRX (labeled by ∨). i, Degradation kinetics of hrFP-E analyzed by GraphPad nonlinear fitting. All degradation experiments in h and i were performed in a working buffer of acetonitrile/PBS (v/v = 1/4) with pH=7.4 at 25°C. We designed the hypoxia-responsive component, ErlotinibNTR, by extending an NTR-responsive moiety (hypoxia sensitive, as depicted in yellow in Fig. 2a) from the amine-bearing linker diaminodipropylamine, which is further connected to Erlotinib (shown in purple in Fig. 2a). The binding energy of ErlotinibNTR is −5.87 kcal mol-1, indicating good affinity with EGFR as Erlotinib (−7.53 kcal mol-1) (Fig. 2c). As expected, ErlotinibNTR effectively reduces the protein levels of EGFR and phosphorylated EGFR (Fig. 2d). However, its IC50 value slightly increases from 20.84 μM (Erlotinib) to 25.82 μM due to the PROTAC arm (Fig. S2a), suggesting a slight decrease in potency due to the PROTAC arm. In summary, ErlotinibNTR maintains great functionality as an EGFR inhibitor. To elucidate the binding dynamics and activation mechanisms of hrFP-E, we examined the ability of hrFP-E’s extended group to inhibit ternary complex formation with target proteins (FAK, CRBN, and EGFR) via simulation assessments of affinity and activation. The results of molecular docking simulations indicated that FP exhibits a reduced binding affinity for FAK (−1.79 kcal mol-1) and CRBN (−3.47 kcal mol-1) compared to FAKi (−4.31 kcal mol-1) and thalidomide (−9.19 kcal mol-1) (Fig. 2e). Furthermore, hrFP-E shows unstable binding to FAK (7.16 kcal mol-1) and CRBN (5.83 kcal mol-1) protein (Fig. 2e). With surface plasmon resonance (SPR), we further observed significant decreases in the binding capabilities of hrFP-E’s extended group, with KD values ranging from 0.608 μM to 1.87 μM for FAK, and from 0.274 μM to 1.59 μM for CRBN (Fig. 2f and Fig. S2b). These findings are consistent with the molecular docking data and highlight a reduced affinity of FP for FAK. Similarly, hrFP-E exhibits both reduced binding affinity (810.56 kcal mol-1) and capability (5.86 μM) with the EGFR protein. Due to the unique catalytic mechanism of PROTACs, however, the diminished affinity of FP with FAK barely destroys the capability of FP to degrade FAK compared to FAKi. To verify the activation of hrFP-E by nitroreductase (NTR), we analyzed the resulting chemical transformations of hrFP-E with varying NTR doses using High-Performance Liquid Chromatography (HPLC) and Mass Spectrometry (MS). The NTR-induced reaction efficiently converts the nitro group of hrFP-E into an amino group, facilitating the release of the linked PROTAC and drug components through a self-immolative 1,6-elimination reaction. hrFP-E shows excellent selectivity and time-dependent degradation kinetics (Fig. 2g-i), with approximately 86% of FP (m/z = 924.30) and ErlotinibNTR (m/z = 644.26) released within 48 hours. These results indicate that hrFP-E can simultaneously perform drug release and protein degradation in a controlled, stable, and site-specific manner. hrFP-E activation specifically triggered by hypoxia offers synergistic therapeutic effects in vitro hrFP-E enters the cytoplasm via an endosomal/lysosomal way and performs sustained FAK degradation and EGFR inhibition (Fig. S3). To validate the hypoxia-specific activation of hrFP-E, we first assessed the de-caging efficiency in normoxic conditions. The caged hrFP-E does not alter the expression of cellular FAK, phosphorylated FAK (pFAK), or pEGFR proteins (Fig. 3a), nor does it affect cell functionality, such as migration and viability (Fig. 3c–e), indicating the inertia of hrFP-E in normoxic conditions. hrFP-E actively responds to hypoxia and degrades FAK and pFAK in a dose-dependent manner. Within 12 hours, degradation efficiency is observed at ∼99% for both FAK and pFAK at 5 μM concentration, ∼98% and ∼97% at 1 μM, and ∼96% and 94% at 0.5 μM, respectively (Fig. 3a). Additionally, hrFP-E inhibits pEGFR more effectively than Erlotinib alone, implying a synergistic therapeutic effect. Compared to only 45% inhibition of Erlotinib, hrFP-E achieves approximately 88% inhibition at 5 μM over 12 hours (Fig. 3a), which can be sustained for 72 hours under hypoxic conditions (Fig. S4). Hypoxia-specific activation of hrFP-E and synergistic therapeutic effects at the cellular level. a, Western blotting of FAK, pFAK, and pEGFR protein levels in H1975 cells. Cells were treated with different concentrations of hrFP-E, FP and Erlotinib for 12 h. b, Representative confocal images of ROS levels in H1975 cells after incubation with different formulations (scale bar: 50 μm). Right: quantification results of ROS levels. c, Representative images of wound healing (n = 6) in H1975 cell culture after PBS and hrFP-E (0.5 μM) treatment under hypoxia (1) and normoxia (0) (scale bar: 200 μm). d, Cell migration rate in 3D collagen culture model (0.5 μM, n = 77, 71, 66, 81). e, Cell viability after treatment with FP and hrFP-E under hypoxia (1) and normoxia (0) for 48 h (n = 4). f, Cell cycle after treatment with different formulations under different culture conditions. g, Results of collagen deposition assay in H1975/MRC5 cells. Cells were incubated with different formulations that mixed with TGF-β1 (5 ng mL-1). h, Cytoplasmic Young’s modulus of different formulations in H1975 cancer cells (n = 22 cells) was determined using AFM. Each data point is the average of at least 3 force profile measurements for a single cancer cell. Note: “0″ for 20% O2, “1″ for 1% O2. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 in b, c, d, f, h (one-way ANOVA). The hypoxia-induced activation of hrFP-E significantly reduces both two-dimensional (2D) (Fig. 3c), and three-dimensional (3D) cell migration (Fig. 3d). Concurrently, the reactive oxygen species (ROS) level increases substantially in hrFP-E treated group (∼31.6-fold) compared to the single-drug treated groups (FAKi ∼9.2-fold; FP ∼14.8-fold; Erlotinib ∼1.5-fold) (Fig. 3b). Since the level of ROS profoundly impacts cell growth, we next examined the effects of hrFP-E on cell proliferation and viability. The activated hrFP-E shows enhanced cytotoxicity with an IC50 of 2.6 μM and a decreased portion of cells in the S-G2-M phase, highlighting its superior efficacy under hypoxic conditions (Fig. 3e, F). Interestingly, we also observed that activated hrFP-E effectively reduces collagen deposition in a dose-dependent manner (Fig. 3g), which likely results from the significant downregulation of COL1A1 (type I collagen) by FP (Fig. S5). Additionally, hrFP-E treated cells display increased stiffness under hypoxic conditions (Fig. 3h), which could potentially inhibit metastasis. In summary, these results demonstrate the sensitive and specific responsiveness of hrFP-E to hypoxic TME and the synergistic effect of hrFP-E in enhancing the therapeutic response of NSCLC cells through FAK degradation. Hypoxia-specific activation of hrFP-E substantially suppresses tumor growth in vivo To evaluate the anti-tumor efficacy of hrFP-E in vivo, we conducted the animal study with nude mice xenografted with H1975 and MRC5 cells. The mice were randomized and treated with six cycles of hrFP-E or control treatments, administered intraperitoneally every three days over 21 days (Fig. 4a). Compared to the single-treatment groups, hrFP-E significantly inhibits tumor growth by approximately 66% without mortality or significant weight loss of mice throughout the study (Fig. 4b–d). Consistent with in vitro results, the reduction of tissue stiffness in hrFP-E and FP treated group is confirmed by decreased storage modulus (Fig. 4e). We further validated the negative regulation of hrFP-E on in vitro collagen production and deposition by Masson staining and observed obvious reduction of collagen fibers density after hrFP-E treatment in the tumor tissue (Fig. 4f). Histopathological staining confirms that repeated hrFP-E administration effectively curtails tumor cell proliferation (Fig. 4f) without notable damage in major organs (Fig. S6). In contrast, both Erlotinib and FAK inhibitors provoke evident hepatic injury in the liver (Fig. S6). The in vivo successful activation of hrFP-E and release of the active component is confirmed by the significant degradation of FAK (51% degradation rate) and inhibition of EGFR (75% inhibition rate) (Fig. 4g). These findings highlight the potential of hrFP-E as an effective and safe candidate for hypoxia-targeted tumor therapy. In vivo activation of hrFP-E to inhibit tumor growth. a, Schematic of treatment with different formulations (i.p.) (4 μmol kg-1) in BALB/c Nude mice bearing H1975 and MRC5 tumor. b, Tumor weight after treatment with different formulations (n = 6). c, Body weight curves of nude mice after intravenous injection of different formulations (n = 6). d, Tumor growth curves of nude mice after intravenous injection of different formulations (n = 6). e, Storage modulus of tumor after mouse execution (n = 6). f, Representative H&E staining of tumor tissues. Scale bars, 100 μm. g, Western blotting analysis of FAK and pEGFR protein levels in the tumor after six cycles of intraperitoneal administration. The data in b-e are presented as mean ± s.d. (n = 6 biologically independent animals), and the asterisks denote significance versus PBS group. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 in b-e. (one-way ANOVA). Discussion Hypoxia-responsive PROTAC achieves selective dual-target degradation in tumors without systemic activation Our development of a hypoxia-activated PROTAC showcases an innovative engineering solution for tumor specificity. By incorporating a bio-reductive trigger NTR, we created a prodrug that precisely control FAK degradation within hypoxic tumor (Fig. 5). This spatial control of PROTACs overcome the off-tumor toxicity, one of the major limitations of PROTAC therapy induced by the “uncontrolled” protein degradation in normal organs. The significance of this advance becomes clear considering prior combination-therapy challenges. Clinical studies combining broad pathway inhibitors have often required carefully titrated or intermittent dosing to manage toxicity[9]. In contrast, our hypoxia-triggered approach is designed to limit active drug to the tumor microenvironment, potentially avoiding such dose-limiting systemic toxicities. By coupling the hypoxia-activation with PROTAC-mediated protein degradation, we harness a more precise and catalytic degradation of target protein as far as it remains in the hypoxic tumor, amplifying its effect locally without continuously exposing normal cells. Our in vivo results validate that environment-specific PROTACs can widen the therapeutic window (Fig. 4). Interestingly, FP exhibited stronger inhibitory effects under hypoxic conditions than under normoxic conditions, possibly because it not only suppresses basal FAK activity but also counteracts hypoxia-enhanced FAK activation[30] and potential signaling compensation mediated by the non-enzymatic functions of FAK. From an engineering perspective, this approves the feasibility of building smart therapeutics that respond to tumor-specific chemical cues, a strategy that could be expanded to additional caging library that is responsive to other pathological cues (e.g., elevated ROS, phosphatase, H2S)[31]. Mechanism of multifunctional hypoxia-responsive dual functional PROTACs (hrFP-E) for controlled FAK protein degradation and EGFR inhibition. Under the triggers, hrFP-E would be cleaved into two parts, one continues to function as a protein degrader for controlled degradation of FAK by recruiting E3 ubiquitin ligase, the other as a kinase inhibitor for inhibiting EGFR phosphorylation. FAK degradation disrupts compensatory and non-kinase mechanisms for superior antitumor effects As the key mechanotransducer, FAK can respond to tumor mechanical microenvironment (TMME) and activate various signaling pathways to regulate the therapeutic response of tumor cells[6,32]. Although FAKi can effectively block the kinase-dependent signals, these inhibitors do not interfere with the non-kinase function of FAK, which has been recognized in the regulation of tumor cell therapy response[33]. A central finding of our work is that the complete removal of FAK via proteolysis (FP) yields superior disruption of tumor mechanotransduction compared to FAK kinase inhibition (FAKi) alone. First, compensatory mechanisms broadly exist in the scenario of single small-molecule inhibition. As observed in our study, FAKi-treated cells in hypoxia showed slightly decreased pFAK signaling and increased total FAK levels (perhaps due to reduced FAK auto-degradation or increased focal adhesion stability) (Fig. 1a). In contrast, PROTAC-mediated degradation eliminates FAK. It collapses the FA scaffolds, eliminating the FAK compensatory pathways that depend on FA scaffolds[34]. Moreover, by degrading FAK we also likely interfered the epigenetic regulation of FAK[35], something FAKi cannot achieve. This is supported by our in vitro data, which show that only the FP-treated tumor cells exhibited a marked upregulation of paxillin and pMLC (Fig. 1e, Fig. S1e,f), suggesting that the broader FAK functions were curtailed. Second, our data show that FAK inhibition fails to reverse the mechanically invasive phenotype of tumor cells under hypoxic conditions. FAKi-treated tumor cells maintained high cell stiffness, strong traction forces, and extensive focal adhesions, similar to untreated ones (Fig. 1f,g). This highlights the significant structural role of FAK in FA protein bundles, even without enzymatical activity. This is consistent with the findings of previous studies, which show that catalytically inactive FAK can remains at adhesion sites to regulate FA turnover[36]. In addition, these treatment-induced increases in cell contraction, actin bundle formation, and focal adhesion size may activate compensatory FAK-independent upstream pathways that promote HIF-1α expression, such as PI3K/AKT/mTOR[37] and MAPK/ERK[38], resulting in the upregulation of HIF-1α after FAKi and FP treatment (Fig. 1b). All these results underscore the importance of considering the non-enzymatic roles of target protein when designing therapeutics, indicating that protein degraders could succeed where inhibitors fall short. Dual inhibition of FAK and EGFR disrupts orthogonal oncogenic pathways for synergistic antitumor effects Although there are more than twenty FAKi in both the research and preclinical stage[39], inhibiting FAK alone fails to offer a significant therapeutic advantage[6]. Several preclinical trials have shown the enhanced efficacy of the combination of FAKi with other therapeutic agents, such as inhibitors of the cell cycle, MEK1/2, STAT3, BRAF, and immunomodulating antibodies[6]. Herein, our work demonstrates a synergistic attack on cancer cells by simultaneously targeting a growth signaling receptor (EGFR) and a mechanotransducer (FAK), particularly in the context of the hypoxic environment. EGFR and FAK operate in distinct but interlinked signaling networks, and co-inhibition proved to be markedly more effective than single-agent treatment in our models (Fig. 3, Fig. 4). Interestingly, the in vivo FAKi treatment indicates the crosstalk between EGFR and FAK signaling by reduces EGFR phosphorylation (Fig. 4g) through a Src-dependent mechanism[40]. Several teams have proposed various PROTAC-based formulas to target the FAK protein[[41], [42], [43], [44]], but no significant strides have been made in terms of synergistic efficacy of co-inhibition. Compared with FAKi, FP treatment exhibits inferior in vivo therapeutic effect (Fig. 4b), possibly resulting from FP-induced activation of survival and proliferation pathways that are not solely dependent on FAK. The increased focal adhesion size and cell contractility can both promote survival and proliferative signaling and thereby attenuate the antitumor efficacy of FP. For example, MAPK/ERK pathway can be activated on actin stress fibers in a contractility-dependent manner [45]. Integrin can induce EGFR phosphorylation and promote cancer cell proliferation[40]. While we have not conducted a thorough investigation into the molecular mechanisms of synergy by dual-inhibiting of FAK and EGFR, our data do suggest the synergistic effects of hrFP-E on lung cancer suppression arise from the inhibition of NSCLC cell migration (Fig. 4c,d) and proliferation (Fig. 4e,F) under hypoxic conditions. These insights reinforce the concept that biochemical and mechanobiological co-inhibitions can yield synergistic therapeutic effects by collapsing multiple pillars of tumor maintenance simultaneously. Our work exemplifies how principles from mechanobiology and biochemistry can be merged to create innovative therapies[[46], [47], [48]]. By targeting a mechanical signaling axis (integrin–FAK) in coordination with a chemical growth signal (EGFR), we embraced a mechano-chemical paradigm for cancer treatment. In practice, this could mean combining a targeted inhibitor (for a driver oncogene) with a mechanobiology-oriented therapy (such as a FAK degrader). The fact that our therapy reduces tumor ECM stiffness and FA formation is intriguing, as it suggests that we are not only killing tumor cells but also actively “reprogramming” TMME to a beneficial state that promotes the therapy responsiveness of tumor cells (Fig. 3g, h). This dual-target therapy benefits tumors with dense ECM or hypoxic cores, turning the stromal stiffness and harsh environment into the mechano-chemical vulnerability of tumor cells. Implications for drug design and bioengineering: a modular strategy for multi-target, context-specific therapeutics From an engineering and drug design standpoint, our study demonstrates a versatile framework for next-generation therapeutics. The hrFP-E molecule can be viewed as a modular assembly of functional components: a target-recognition ligand for FAK, a molecule inhibitor of EGFR, an E3 ligase recruiting ligand, and a hypoxia-sensitive linker that regulates activity (Fig. 5). Each module can be optimized or swapped – for instance, one could incorporate alternative inhibitors to eliminate activity of other oncogenic proteins (such as BRAF[49,50]), or use different hypoxia-labile groups with tunable trigger thresholds. This modularity means researchers could rapidly design PROTACs for different cancer types by plugging in the appropriate targeting ligands, all while using the same paradigm of hypoxia (or another tumor-specific factor) for activation. In our case, we achieved dual targeting by deploying hypoxia-activated inhibition of EGFR and proteolysis of FAK in the same context, but future designs might even tether two warheads for PROTAC degradation in one molecule if a suitable chemistry can be found. Our hrFP-E stands as a successful proof-of-concept in illustrating that even small-molecule therapeutics can be endowed with logic-gated activation. Similar with hypoxia-responsive nanoparticles currently under investigation[51,52], this opens new avenues for bioengineers: design activatable PROTACs, prodrugs, antibody-drug conjugates, and biomaterials in the presence of specific enzymes, pH levels, or metabolic states unique to disease tissue for personalized treatment. Clinical outlook: towards safer and more effective combination cancer therapies for patients Our findings carry encouraging implications for clinical translation and patient care. The synergistic efficacy and apparent safety of the hypoxia-activated dual degrader suggest that this strategy could overcome two perennial issues in oncology: insufficient efficacy of single-agent therapies and intolerable toxicity of combination regimens. Clinicians are increasingly aware of the need for combination approaches – as evidenced by recent trials adding FAK inhibitors or collagenase to targeted or immunotherapies[9,53], but have struggled with drug-related adverse events. A context-specific therapy could allow patients to receive aggressive multi-pathway treatment with fewer side effects, improving quality of life during treatment. While we confirmed that normal oxygenated tissues experience minimal drug activation (Fig. S6), careful toxicology studies in advanced models and non-human primates will be needed to rule out any subtle off-target effects for human dosing. Encouragingly, ARV-110, an androgen receptor PROTAC (NCT03888612)[54] and ARV-471, an estrogen receptor PROTAC (NCT04072952)[55], both have already entered human trials and shown favorable tolerability and clinical benefit in prostate cancer patients and breast cancer patients, respectively. These successes imply that hypoxia-activated PROTACs could become a template for future precision oncology drugs, offering new hope for patients with refractory tumors that currently evade best single-target treatments. Conclusion In summary, we introduce a hypoxia-activated, dual-warhead PROTAC that degrades FAK and inhibits EGFR to deliver mechano-chemical therapy with spatial precision. Using lung cancer as a demonstration, hrFP-E achieves near-complete FAK ablation under hypoxia, potent EGFR suppression, strong anti-migration/anti-proliferation effects, ECM softening, and in vivo efficacy without overt toxicity. The platform is generalizable to other oncogenic drivers and pathological gates, establishing a blueprint for mechanomedicine in solid tumors: rewire tumor mechanics and growth signaling together, where the tumor lives under hypoxia. Materials and methods General information All chemical reagents and solvents were purchased from Sigma-Aldrich (Shanghai, China), Fisher (MA, USA) or Energy Chemical (Shanghai, China), and used as received. All aqueous solutions were prepared using ultrapure water from a Milli-Q system (Millipore, USA). 1H NMR, 13C NMR were recorded on a Bruker Avance III 400 MHz spectrometer with solvent signals as internal references at room temperature. Chemical shifts (δ) were expressed in ppm and J values were given in Hz. ESI-HRMS was conducted on a WATERS I-Class VION IMS QTof (WATERS, USA), and the data were reported as mass-per-charge ratio m/z calculated and observed. SPR experiments were performed on a Biacore8K instrument (GE Healthcare, USA). Confocal laser scanning microscopy (CLSM) images were recorded using an Olympus FV3000 confocal microscope (Tokyo, Japan). Flow cytometry analysis was recorded using a CytoFLEX LX (BECKMAN COULTER, USA). CCK8 assay was recorded at 450 nm using a Synergy Mx monochromator-based multi-mode microplate reader (Biotek, USA). Chemiluminescence was detected using Ultra High Sensitivity ECL Kit (MCE, 5 mL) and imaged on a ChemiScope 3000mini Imaging System (Clinx) using automated exposure settings. Molecular docking methodology To understand the interaction of erlotinib, ErlotinibNTR with EGFR; srFPX-E, FP with FAK, CRBN or EGFR. Molecular docking simulation was conducted by Autodock4.2 software. For this aim, the crystal structure of EGFR (PDB ID 1M17), FAK (PDB ID:4Q9S) and CRBN (PDB ID:7BQV) was downloaded from Protein Data Bank (PDB). Prior to docking, protein structures were prepared by removing water molecules using PyMOL, RRID:SCR_000305 software. Following which, bond orders were assigned, and hydrogen atoms were added to the crystal structures. The chemical structures of the selected ligand molecules were converted into *. MOL2 format using the Marvin Sketch program. Autodock (RRID:SCR_012746) uses a semiempirical free energy force field to evaluate binding conformations of ligand while docking. The AutoDockTools was used for preparing protein and ligand parameters files. For ligands; the hydrogens, compute Gasteiger charges, and non-polar H atom were added which ensuring the total charge corresponds to the tautomeric state followed; and, at last torsion tree root and rotatable bonds were chosen. For macromolecules; hydrogens, compute Gasteiger charges, and merge non-polar H were added which was followed by Stouten atomic salvation parameters assignment; and at last flexible residues PDBQT in addition to the rigid PDBQT file were created. Docking of CRBN protein to ligands as an example, based on the dimensions of ligands, we created grid boxes covering the binding sites of the chosen protein ligands, the initial docking calculations were carried out for the FP ligand with the grid box (x, y, z = 108×106×114), center x = 28.269, center y = 16.064, center z = 21.804, exhaustiveness=8, parameters to optimize the docking calculation. For docking of FAK protein to ligand, the grid of size 126×126×126 was created with protein center (Centre coordinates = 9.7, 0.16, 15.1) as the grid protein center, exhaustiveness=8. For docking of EGFR protein to ligand, the grid of size 40×40×40 was created with protein center (Centre coordinates = 23.594, 1.054, 53.253) as the grid protein center, exhaustiveness=8. Here the center x, center y, and center z represent the center of the grid box and exhaustiveness represent the number of runs. Each docking calculation set was repeated using three distinct seeds while keeping the other default values. SPR binding study SPR experiments were performed on a BiacoreT200 instrument (GE Healthcare, USA) following the protocols. CRBN, FAK or EGFR proteins were immobilized on a Series S CM5 chip in an immobilization buffer (pH 4.0) containing 10 mM CH3COONa at 25 °C, that is, the surface of the CM5 chip was decorated with carboxyl groups and the chip was pretreated with EDC and NHS for 5 min to activate carboxyl groups. For binary studies, the final surface density of targeted protein was approximately 6500-7000 RU, (or 7000-8000 RU). All interaction experiments were performed in a running buffer (pH 7.4) containing 8.1 mM Na2HPO4, 137 mM NaCl, 2.68 mM KCl, 1.47 mM KH2PO4, 0.05% Tween 20, and 4% DMSO at 25 °C. Gradient concentrations of drugs (0-10 μM) were injected at a flow rate of 30 μL min-1. The association time was 120 s and the dissociation time was 180-600 s. Degradation of hrFP-E and urFP-E hrP-E (100 μL, 20 μM in MeCN/MES buffer (v/v = 1/4)) was added to MES (0.1 M, pH=6.5, 100 μL) containing NTR (N9284, Sigma-Aldrich) and NADPH (1 mM) and the mixture was incubated for indicated time. urFP-E (100 μL, 10 μM in MeCN/PBS (v/v = 1/4)) was transferred into a ½-dram borosilicate glass vial and irradiated on a UV transilluminator (365 nm) for indicated time and power. After degradation, the samples (10 μL) were collected, diluted into MeCN (90 μL) and then loaded on a Thermo Scientific™ UltiMate™ 3000 instrument. HPLC method: mobile phase A (water) and mobile phase B (MeCN), flow rate: 1 mL min-1, running time: 25 min, the gradient elution method: 95% B from 0 to 15 min, 95% to 5% B from 15 to 20 min. Detection wavelength: 254 nm. Column: HaloTM Protein C4 3.4 μm, 4.6 mm × 150 mm. Injection volume: 10 μL. Cell culture H1975, MRC-5 cell lines were obtained from the Institute of Cell Biology at the Chinese Academy of Sciences (Shanghai, China) or ATCC (USA). H1975 (RRID:CVCL_1511) cell line cultured in RPMI-1640 Medium (HyClone, USA), and MRC-5 (RRID:CVCL_0440) cell line cultured in Minimum Essential Medium (HyClone, USA), including 1% MEM Non-Essential Amino Acids Solution (Procell, China), 1% sodium pyruvate (Procell, China), and 1% glutaMAX-1 glutamine (Procell, China). All cultured media were supplemented with 10% (v/v) fetal bovine serum (FBS, Sigma-Aldrich, USA) and 1% (v/v) penicillin/streptomycin (Procell, China) and maintained at 37 °C in a 5% CO2 atmosphere. At the logarithmic growth phase, the cells were incubated with different compounds in cultured medium. To mimic the hypoxic microenvironment in solid tumors, cells were cultured in a Heracell VIOS 160i incubator (Thermo Scientific, USA) with 1% to 20% oxygen or by chemically inducing the generation of a hypoxic environment by CoCl2 (100 μg/mL). Evaluation of cytotoxicity, cell cycle and ROS levels In vitro cytotoxicity of all compounds was assessed by CCK8 assay. Briefly, H1975 (8 × 103 cells per well) cells were seeded in 96-well plates. After incubation in medium containing different compounds for the indicated time (48 h), the medium was replaced with fresh medium (200 μL) containing CCK8 (20 μL, 5 mg mL-1). After 1 h of incubation, the absorbance of the samples was measured using a Synergy Mx monochromator-based multi-mode microplate reader (Biotek, USA). The optical density (OD) was read at a wavelength of 450 nm. Relative cell viability was defined as ([OD]test/[OD]control) × 100%. Each experiment was repeated at three times. Cell cycle was tested by Cell Cycle and Apoptosis Analysis Kit (Beyotime, Shanghai), and analyzed using an Agilent NovoCyte Flow Cytometer (Agilent, CA, USA). ROS levels were evaluated by Reactive Oxygen Species Assay Kit (Beyotime, Shanghai), and analyzed using an Olympus FV3000 confocal microscope (Tokyo,Japan). Western blotting The protein expression levels were mainly assessed by Western blotting. In brief, cells were seeded in a 12-well plate at a density of 5 × 104 cells per well and cultured for 24 h before treatment. Then cells were treated with the test compounds or equivalent volume of DMSO (control group) and lysed at the stated time point. For lysis, cells were washed with cold PBS twice and lysed in RIPA Lysis Buffer containing protease inhibitor cocktail (Sigma-Aldrich, USA) at 1:100 dilution. Cell lysates were incubated on ice for 30 min with vortex every 5 min, then separated by centrifugation for 10 min at 12,000 rpm. The supernatants were stored at −80 °C. Protein concentrations were determined on a Synergy Mx monochromator-based multi-mode microplate reader (Biotek, USA) using a BCA Protein Kit (Thermo Scientific, USA). Total protein was separated by 10% SDS-PAGE and transferred to PVDF membranes (Millipore, USA). Blots were treated with 5% BSA (Millipore, USA) for 1 h and incubated with primary antibody, including rabbit mAb anti-FAK (Cell Signaling Technology Cat# 13009, RRID:AB_2798086), rabbit mAb anti-phospho-FAK (Cell Signaling Technology Cat# 3283, RRID:AB_2173659), rabbit mAb anti-EGFR (Cell Signaling Technology Cat# 4267, RRID:AB_2864406), rabbit mAb anti-phospho-EGFR (Cell Signaling Technology Cat# 3777, RRID:AB_2096270), rabbit mAb anti-E-cadherin (Cell Signaling Technology Cat# 3195, RRID:AB_2291471), rabbit mAb anti-N-cadherin (Cell Signaling Technology Cat# 13116, RRID:AB_2687616), rabbit mAb anti-Paxillin (Cell Signaling Technology Cat# 2542, RRID:AB_10693603), rabbit mAb anti-Phospho-Myosin Light Chain 2 (Thr18/Ser19) (Cell Signaling Technology Cat# 3674, RRID:AB_2147464), rabbit mAb anti-HIF-1α (D1S7W) (Cell Signaling Technology Cat# 36169, RRID:AB_2799095), mouse mAb anti-PKM2 (Proteintech Cat# 60268-1-Ig, RRID:AB_2881388), mouse mAb anti-GLUT1 (Proteintech Cat# 66290-1-Ig, RRID:AB_2881673), in 5 mL blocking buffer at 4 °C overnight with rotation, followed by incubation with anti-rabbit IgG HRP-linked secondary antibody (Cell Signaling Technology Cat# 7074, RRID:AB_2099233) or anti-mouse IgG HRP-linked secondary antibody (Cell Signaling Technology Cat# 7076, RRID:AB_330924) at room temperature with rotation for 1 h. Normalization was performed by blotting the same membranes with antibody rabbit mAb anti-GAPDH (Cell Signaling Technology Cat# 2118, RRID:AB_561053). The bands were detected using Ultra High Sensitivity ECL Kit (MCE, 5 mL) and imaged on a ChemiScope 3000mini Imaging System (Clinx) using automated exposure settings. Data are the average of three biological replicates unless indicated otherwise. Traction force microscopy To determinate the cellular traction, polyacrylamide gel (PA gel) precursor solutions (10% acrylamide solution; 0.3% bis-acrylamide solution; 0.2% photoinitiator LAP) were doped with fluorescent microspheres (0.2 μm diameter), at a concentration of 2% v/v (Invitrogen, F8811, USA), and then polymerization was initiated using blue light (465 nm) to form a cross-linked PA gel (Young’s modulus: 6425 Pa). The PA gels were treated with sulfo-SANPH (dissolved in pH 8.5 HEPES), washed three times with pH 8.5 HEPES, and incubated with 0.1 mg mL-1 (diluted in ddH2O) of rat tail collagen solution (Corning, 354236, USA) at 4 °C overnight to modify the PA gel surface. The collagen solution was removed and sterilized by UV light for 2 h prior to use. H1975 cells were spread evenly onto the collagen-modified PA gel surface at a density of 8000 cells mL-1 and incubated for 24 h for traction force microscopy (TFM) analysis. Phase contrast images of the cells and fluorescence images of the embedded fluorescent nanorods were captured using an Olympus FV3000 confocal microscope. Images of each cell taken before and after lysis of the cells with RIPA were analyzed with a previously reported MATLAB (RRID: SCR_001622) script[56] to obtain traction force maps and the average traction stress exerted by the cells on the underlying matrix. Atomic force microscopy AFM (Nanowizard Ⅳ biotype) was utilized to determine the changes in cell stiffness of the cells after treatment. The parameters of the probe used in this experiment were as follows: NP-O10 D Triangular, spring constant k value = 0.06 N m-1. And the cantilever beam probe was subjected to a sticky ball operation (silica bead of 6 μm diameter) using a previously reported method[57]. AFM was used to perform the force curve testing after completing the cantilever calibration in the approximate steps as follows: after locating the test cells under the AFM, force-displacement curves were obtained by aligning the cell nucleus region with the cantilever at a constant velocity of 2 μm s-1 and holding it on the surface of the cells of the region with a constant force of 0.8 nN. The obtained curves were systematically analyzed using the AFM test system, and the Hertz contact model was chosen for the calculation of the modulus for the fitting calculation. Evaluation of cell migration To evaluation the cell migration speeds in 2D, H1975 cells were seeded in 12-well plates at a density of 1 × 106 cells per well and cultured 12 h. After 12 h, the serum-free medium (1% FBS) was replaced, and after continuing the culture for 12 h, a sterile 10 μL lance tip was used to create a homogeneous scratch in each well, and the cells were washed twice with PBS to remove any floating or dead cells. This time point was considered to be 0 h. Cells were cultured in fresh medium containing test drugs or controls. To exclude serum interference without affecting cell viability, 1% FBS medium was used for all groups. Images of the scratched area after 0 h, 12 h, 24 h, 48 h, and 72 h were captured using a camera attached to a light microscope. The images were analyzed with ImageJ (RRID:SCR_003070) software and the scratch recovery area was quantified. The percentage of scratch healing (migration) was calculated by subtracting the remaining wound area at 48 h from the scratch area at 0 h. The percentage of scratch healing (migration) was calculated by subtracting the remaining wound area at 48 h from the scratch area at 0 h. To evaluation the cell migration speeds in 3D, H1975 cells were washed with PBS, digested with 0.05% trypsin and resuspended by centrifugation for spare parts. Collagen was neutralized in the following proportions: 100 μL 10× PBS; 11 μL 0.2 M NaOH; 789 μL cell suspension or ddH2O; and 100 μL collagen (10 mg mL-1) to obtain a collagen cell suspension with a final concentration of 1 mg mL-1 (in which the final density of the cells was 2 × 106 cells mL-1), and the above operation was performed on ice. The mixed collagen cell suspension was taken 100 μL in the bottom of a 4-compartment confocal dish (Cellvis, USA) and placed in a cell incubator (37°C, 5% CO2, 40 min) to continue gelation, and after the collagen became gelatinous, RPMI-1645 medium was added to continue the incubation for 12 h. Different preparations were added, and chemically induced hypoxic environment (100 μg mL-1 CoCl2) was used to produce the cells, taking into account that the cells were photographed overnight at a later stage. After 24 h of culture, 3D cell migration was tracked using the NucBlueTM live cell nuclear staining probe (Invitrogen, USA). Live cell migration was analyzed by imaging with a confocal microscope equipped with a live cell workstation at 37°C and 5% CO2, and images were analyzed using Trackmate (ImageJ plug-in). Data were processed using GraphPad Prism, RRID:SCR_002798. Differences between groups were analyzed by t-test and considered significant when P < 0.05. Collagen deposition assay To evaluation the tumor-derived collagen deposition, H1975 cells were seeded in 96-well plates at a density of 1 × 104 cells per well, and replaced with serum-free medium after 12 h. After 24 h of synchronization, the medium contain different formulations was added accordingly, and additionally the model and drug groups were treated with TGF-β1 (5 ng mL-1, Novoprotein, China) for 48 h. The positive drug group was treated with TGF-β1+SB431542 (TGF-β1 inhibitor, final concentration 5 μM), and the blank control group did not need to add TGF-β1 additionally. Cells were placed in different culture conditions, in which the hypoxia group were placed in the hypoxia incubator (Heracell VIOS 160i, Thermo Scientific, USA). After 48 h, cells were treated in accordance with the Sirius Red Assay Kit standardized procedure (Solarbio, China). Relative collagen deposition rate = A540 nm (drug administration group)/A540 nm (positive drug group). Four replicate-well experiments were performed in parallel for each group of samples. Therapeutic efficacy assessment in vivo All animals had free access to rat food and water. All the animal studies conformed to the ethical approval set by the Biomedical Research Ethics Committee, School of Life Science and Technology, Xi'an Jiaotong University (NO. 2024-23). For tumor xenograft models, H1975 cells (2.0 × 106) and MRC cells (5.0 × 106) were resuspended in 100 μL PBS, and injected subcutaneously into the flanks of male BALB/c Nude mice (RRID: IMSR_RJ: BALB-C-NUDE) (5-7 weeks). Tumor size was monitored with vernier calipers and the tumor volume was calculated as V=L×W2/2, where L and W were the length and width of the tumor, respectively. When the tumor volume reached 50 mm3, the tumor-bearing mice were weighed and randomly divided into different groups (6 mice per group). Thereafter, mice were injected intraperitoneally (i.p.) with different formulations in 100 μL of sterile PBS every three days for six times. Keeping the molar mass of the drugs consistent, the following doses were administered in each group: (1) PBS; (2) free erlotinib (1.7 mg kg-1); (3) free Defactinib (2.0 mg kg-1); (4) FP (3.7 mg kg-1); and (5) hrFP-E (6.4 mg kg-1). The size of the tumors and the weight of the mice were measured every three days. On day 21, all mice were euthanized. Major organs, including the heart, liver, spleen, lungs, and kidneys, as well as tumors, were collected, washed with cold PBS, and fixed in 10% neutral buffered formalin. For hematoxylin and eosin staining (H&E staining), formalin-fixed organs were embedded in paraffin blocks and observed with a light microscope (Olympus IX51, Japan). For Masson cell staining, fixed tumor sections were stained with Masson Trichrome Staining Kit (G1340, Solebo, China) according to the manufacturer’s experimental procedure and observed by light microscopy (Olympus IX51, Japan). To assess the changes in tumor stiffness after treatment, a rheometer (DHR3 Shear Rheometer) with a probe diameter of 8 mm was used. The storage modulus G’ and loss modulus G’’ were measured at 1% strain and 1 rad s-1 frequency. Where the magnitude of the storage modulus was used to respond to the changes in tumor stiffness. To assess the expression level of target proteins in tumor tissues after treatment, collected fresh tumor tissues (100 mg) were added to RIPA lysate (1 mL) and the tissues were sufficiently broken to release proteins using a tissue homogenizer (Scientz-24 High-Throughput Tissue Mills); the homogenized liquid was subjected to high-speed centrifugation at 4°C (14,000 g, 15 min), and the collected supernatant was transferred to a new centrifuge tube and stored in a −80°C refrigerator for backup. The expression levels of the target proteins in the tumor tissues were later evaluated by Western blotting. Statistical analysis Data are presented as mean±standard deviation (SD). The in vitro experiments were performed in three independent experiments. The in vivo experiments were performed with 8 mice for each group. The difference between the groups was analyzed by using a one-way ANOVA analysis with GraphPad Prism 9. P values: ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. CRediT authorship contribution statement Keli An: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Yuan Li: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization. Chunyu Cao: Validation, Methodology. Ming Wang: Validation, Methodology. Xiaohui Jia: Validation, Resources. Guorui Jin: Writing – review & editing, Visualization. Bo Cheng: Writing – review & editing, Visualization. Yuanbo Jia: Writing – review & editing, Visualization. Lin Wang: Writing – review & editing, Resources. Hui Guo: Writing – review & editing, Visualization, Resources, Funding acquisition. Feng Xu: Writing – review & editing, Writing – original draft, Visualization, Supervision, Project administration, Funding acquisition, Conceptualization. Kai Qu: Writing – review & editing, Writing – original draft, Visualization, Supervision, Project administration, Funding acquisition, Conceptualization. Declaration of competing interest All authors have approved the final version and declare no conflicts of interest. Acknowledgement This research was supported by the National Natural Science Foundation of China (12432015, 12102334, 12472321), the Key Research and Development Projects of Shaanxi Province of China (2023GXLH-033), and the National Science and Technology Major Project of the Ministry of Science and Technology of China (2025ZD0551500, 2025ZD0551501). Footnotes Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.neo.2026.101328. Contributor Information Lin Wang, Email: wanglin7209@muhn.edu.cn. Hui Guo, Email: guohui@xjthfh.edu.cn. Feng Xu, Email: fengxu@mail.xjtu.edu.cn. Kai Qu, Email: qukai001@xjtu.edu.cn. Supplementary materials References Associated Data Supplementary Materials