NK cell biomimetic MOF nanoplatform disrupts hypoxia adaption for complete tumor ablation.

Introduction
Triple-negative breast cancer (TNBC), defined by the absence of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) expression, represents 10–20% of all breast cancer diagnoses [1, 2]. This aggressive subtype is marked by early onset, frequent visceral metastasis, high recurrence rates, and dismal prognoses [3–5]. The lack of targetable receptors, coupled with pronounced tumor heterogeneity, confines systemic treatment largely to conventional chemotherapy [6, 7]. However, chemotherapy often yields transient responses, rapid relapse, and chemoresistance, particularly in metastatic disease, highlighting the urgent need for innovative therapeutic paradigms [8, 9].
Immunotherapy, notably immune checkpoint inhibitors (ICIs), has revolutionized oncology, with approvals across 20 + cancer types since 2011 [10]. While breast cancers are traditionally considered immunologically quiescent, TNBC exhibits relatively higher immunogenicity, positioning it as a candidate for immune-based strategies. Natural killer (NK) cells mediate innate antitumor immunity by recognizing stress ligands (e.g., CD112, CD155) via receptors like CD226 and NKG2D [11, 12]. Recent advances in chimeric antigen receptor (CAR)-NK cell therapy highlight NK cells’ clinical potential for solid tumor treatment, particularly their intrinsic safety profile (e.g., reduced cytokine release syndrome) and MHC-independent cytotoxicity [13, 14]. While CAR-NK exemplifies engineered NK cell applications, leveraging native NK membrane proteins (e.g., CD226, NKG2D) retains these advantages while circumventing the complexity of live-cell therapies [15, 16].
However, the immunosuppressive tumor microenvironment (TME), characterized by hypoxia, aberrant cytokine networks, and tumor-derived soluble factors, impairs NK cell viability and function [17–19]. Compounding this, tumors evade NK surveillance by downregulating or shedding ligands (e.g., MICA/B, CD112/155), necessitating strategies to enhance NK cell targeting and TME adaptability [20, 21].
Photodynamic therapy (PDT), a minimally invasive modality, faces parallel challenges: its oxygen-dependence limits efficacy in hypoxic TNBC. Hypoxia not only curtails ROS production but also fosters therapeutic resistance [22–25]. Further limitations include poor tumor-specific PS accumulation and off-target effects, necessitating innovations to enhance oxygen availability and targeted PS delivery [26, 27]. While existing strategies to address hypoxia in PDT have shown promise, certain limitations warrant consideration. Hyperbaric oxygen therapy, while clinically viable for certain applications, presents practical limitations in treating deeper tumors and requires careful safety monitoring [28, 29]. Catalytic nanosystems utilizing tumor-derived H2O2 represent an innovative approach, their efficacy may be influenced by the variable H2O2 concentrations naturally present in tumor microenvironments [30–34]. Alternative approaches targeting TME normalization through VEGF inhibition or vascular remodeling agents show more sustained effects, but the narrow therapeutic window between vascular normalization and excessive pruning frequently leads to paradoxical hypoxia exacerbation [35, 36]. In addition, Oxygen-carrier systems, such as perfluorocarbon emulsions or hemoglobin-based nanoparticles, exhibit favorable oxygenation capacity in preclinical studies, their clinical translation may be constrained by suboptimal pharmacokinetics and potential safety considerations [37–40]. These observations collectively underscore the importance of developing next-generation solutions that combine adaptive TME modulation and targeted activation mechanisms to optimize therapeutic outcomes [41, 42].
To address these limitations, we developed NKAMP (NK cell membrane-camouflaged, Aptamer-engineered, and Methoxyestradiol-PCN-224 Integrated Nanosystem), a biomimetic nanoplatform, which combine: (1) PCN-224 (porphyrin MOF photosensitizer) loaded with 2-methoxyestradiol (2ME, HIF-1α inhibitor) for enhanced PDT; (2) NK cell membrane serves as a stealth and innate targeting layer, providing immune evasion to prolong systemic circulation while enabling broad-spectrum tumor recognition via preserved receptors (e.g., CD226, NKG2D mediated ligand-specific recognition of TNBC cells) that bind stress-induced ligands on cancer cells. (3) MUC1 aptamers (2×C18-Apt) for active tumor homing via selectively binding overexpressed mucin-1 glycoproteins on TNBC surfaces, Fig. 1. NKAMP inhibits hypoxia adaptation (via 2ME) to synergistically enhance PDT (via photosensitizer PCN-224), while the hierarchical targeting approach significantly enhances robustness against tumor microenvironment heterogeneity and adaptive resistance.
The as-designed NKAMP overcomes triple clinical barriers (physical barrier, hypoxia barrier, and immune evasion barrier) by integrating biomimetic targeting, HIF-1α pathway inhibition, and PDT to simultaneously rescue 2ME pharmacokinetics via tumor-directed delivery, reverse hypoxia resistance through HIF-1α suppression, and bypass MHC I-mediated immune evasion with dual-targeting (NK membrane receptors + MUC1 aptamers). This study advances precision oncology by offering a multifunctional solution to TNBC’s intertwined challenges of hypoxia, immune evasion, and therapeutic resistance.

Results and discussions

Nanodrug fabrication
PCN-224, a porphyrin-based metal-organic framework (MOF) photosensitizer, was synthesized following an optimized protocol (Figure S1, Supporting Information). The resulting nanoparticles exhibited uniform size distribution (85 ± 5 nm) and morphology, as confirmed by dynamic light scattering (DLS) and transmission electron microscopy (TEM), Fig. 2A. X-ray diffraction (XRD) pattern suggest that PCN-224 retains good crystallinity (Fig. S1D). PCN-224 exhibited good biocompatibility with a panel of cell lines (MCF10A, MDA-MB-231, MCF-7, and T47D), while demonstrating photodynamic activity under laser irradiation, producing reactive oxygen species (ROS) to kill cancer cells (Figure S2-S4). To overcome hypoxia-induced treatment resistance, the anti-HIF-1α drug 2-methoxyestradiol (2ME) was loaded into PCN-224 via passive diffusion under continuous agitation (20 °C, 12 h), yielding the 2ME@PCN-224 complex (denoted as MP). Liquid chromatography-mass spectrometry (LC-MS) analysis (Figure S5) revealed that a PCN-224-to-2ME mass ratio of 2:1 achieved optimal drug-loading equilibrium, with a loading efficiency of 6.46% (Fig. 2B).
NK cell membranes were isolated through a series of freeze-thaw cycles in liquid nitrogen, followed by differential centrifugation to remove intracellular components. Membrane integrity and protein content were verified using a bicinchoninic acid (BCA) assay. The absorbance of NK cell membrane suspensions remained unchanged over one week (Figure S6), confirming stability of the isolated NK membrane and preserved bioactivity of surface proteins. The purified NK cell membranes were then homogenized and fused with MP nanoparticles via sequential vortex mixing and extrusion through a polycarbonate membrane (200 nm pore size) using a liposome extruder, resulting in NK membrane-coated MP nanoparticles (denoted as NKMP).
To enhance tumor-specific targeting, NKMP was further functionalized with an DNA aptamer designed to recognize MUC1, a glycoprotein tumor marker overexpressed in triple-negative breast cancer (TNBC). To this end, C18-alkylated MUC1 aptamer (Apt-2×C18) was synthesized via linking two C18 chains to the 5’-end of the aptamer (Fig. 2C top panel). The hydrophobic C18 chains spontaneously inserted into the lipid bilayer of the NK membrane via hydrophobic interactions (Fig. 2C bottom panel), while the MUC1-binding aptamer remained exposed on the nanoparticle surface. For in vivo tracking, the aptamer was conjugated with fluorescent dye Cy5, enabling real-time visualization of tumor accumulation. The final construct, designated NKAMP, is supposed to integrate (1) dual-targeting via NK cell membrane proteins (CD226/NKG2D) and MUC1 aptamers, (2) 2ME-driven HIF-1α inhibition for disrupting hypoxia adaption and enhancing ROS generation, and (3) PCN-224-mediated photodynamic activity.
TEM image revealed that the NKAMP nanodrug exhibited a uniform spherical morphology with excellent dispersibility (Fig. 2D), with an average diameter of 95 ± 5 nm (Fig. 2E). A noticeable shell was enveloped on each PCN-224 particle, yielding an increase in size compared to bare PCN-224, while maintaining an optimal nanoscale dimension for enhanced vascular permeation. UV-vis spectroscopy analysis identified characteristic absorption peaks for PCN-224, MP and the final NKAMP construct at 420 nm (Soret band) and 520 nm, 550 nm, 580 nm, and 650 nm (Q-band), corresponding to the porphyrin ligand’s π-π* transitions (Fig. 2F, note: The spectral shoulder observed at approximately 670 nm can be attributed to the aggregation of the organic linker, tetra(4-carboxyphenyl)porphine (TCPP)), confirming that the drug-loading and membrane-coating processes did not compromise the photophysical properties of PCN-224. Notably, the absorbance at 650 nm, within the phototherapeutic window (600–850 nm), ensures effective photodynamic therapy (PDT).

The colloidal stability of NKAMP was confirmed by zeta potential (ζ) measurements. A consistent value of − 34.28 ± 0.15 mV was maintained over seven days (Figure S7), demonstrating excellent stability without aggregation. Furthermore, the successive change in ζ-potential from − 10.2 mV for bare PCN-224 to − 34.28 mV for NKAMP provides strong evidence for the successful fabrication of the biomimetic nanoplatform. This highly negative surface charge ensures colloidal stability by electrostatic repulsion, underscoring the formulation’s suitability for in vivo applications. The drug release efficiency of the nanodrug reached 51.17% after 24 h of incubation in PBS without 650 nm light irradiation, and increased to 73.65% under light irradiation, demonstrating significantly enhanced release upon light treatment (Fig. 2G).
Flow cytometry analysis was performed to validate the preservation of NK cell membrane proteins on NKAMP nanoparticles (Fig. 2H and S8). The nanoconstructs were probed with fluorescently labeled antibodies targeting four critical NK receptors: CD56 (a characteristic NK cell marker), CD226 and NKG2D (essential for tumor recognition through binding to stress-induced ligands CD112/CD155 and MICA/B, respectively), and CD16 (mediating antibody-dependent cellular cytotoxicity, ADCC). Comparison with native NK cells (Fig. 2Hi) revealed that NKAMP nanodrug maintained positive expression of all four proteins (Fig. 2Hii). This confirmed that the membrane-coating process preserved both the structural integrity and functional activity of NK cell-derived surface proteins. Consequently, NKAMP retains the dual advantage of inheriting NK cells’ tumor-targeting specificity while maintaining their capacity for immune activation in the tumor microenvironment.

Enhanced cellular internalization and ROS generation of the nanodrug
Conventional bottom-up chemically synthesized nanomaterials frequently suffer from rapid immune clearance and poor biodistribution profiles due to host recognition mechanisms. In contrast, our top-down biomimetic approach utilizing natural killer (NK) cell membrane encapsulation addresses these limitations through biologically inspired stealth properties [43]. Derived from human peripheral blood, NK cell membranes provide native surface markers that significantly reduce immunogenicity while prolonging systemic circulation, critical factors for effective tumor accumulation [44]. Previous studies have established that such biological coatings promote cellular internalization through membrane fusion mechanisms[45, 46, 51, 52].
We systematically evaluated the cellular uptake kinetics via employing Cy5-labeled PCN-224 and NKAMP incubated with MDA-MB-231 triple-negative breast cancer cells, with quantitative assessment by flow cytometry (Fig. 3A). The analysis demonstrated greater fluorescence intensity after 1 h incubation (4700 a.u. vs. 1000 a.u. Figure 3A left panel), indicating significantly enhanced cellular internalization of NKAMP compared to uncoated PCN-224. This uptake advantage became more pronounced over time. At 12 h, NKAMP loaded cells reaching 1.78-fold higher intensity than that of PCN-224 loaded cells (8000 a.u. vs. 4700 a.u., Fig. 3A right panel), unequivocally demonstrating the superior internalization propensity of the NK cell membrane-coated nanodrug.

To provide more nuanced understanding of the internalization process, we established a novel internalization factor (IF) metric that categorizes nanodrug particles into three distinct states: fully internalized (IF > 1), membrane-bound (0 < IF < 1), and uninternalized (Fig. 3B). Quantitative IF analysis at 1 h and 12 h post-incubation revealed that the proportion of fully internalized NKAMP nanoparticles (IF > 1) was substantially greater than that of the control group (Fig. 3C). For example, at 12 h, 92% NKAMP treated cells harboring fully internalized nanodrug (IF > 1); while this number is only 75% for PCN-224 treated cells. This data provides rigorous confirmation that the biomimetic NK cell membrane coating markedly enhances cellular uptake efficiency through improved biological recognition and internalization kinetics.
These findings collectively demonstrate that NK cell membrane cloaking actively promotes tumor cell internalization through biomimetic recognition mechanisms. The enhanced endocytosis kinetics, coupled with prolonged circulation, provide a dual mechanism for improved therapeutic delivery to tumor sites.
Subsequently, we investigated the reactive oxygen species (ROS) generation kinetics of the nanodrugs. DCFH-DA-loaded cells were treated with PCN-224 or NKAMP and subjected to continuous irradiation with a 655 nm laser (60 mW/cm²). As shown in Fig. 4A(i), fluorescence intensity increased with prolonged irradiation, confirming enhanced ROS generation over time. Notably, NKAMP-treated cells exhibited significantly higher fluorescence intensity than PCN-224 at 10, 15, and 20 min (p < 0.05), indicating that 2-methoxyestradiol (2ME) potentiates ROS production during photodynamic therapy.
To further evaluate the sustained ROS generation capacity between the two materials under controlled conditions, we maintained a constant irradiation duration (30 mW/cm², 10 min) and measured DCF fluorescence intensity at sequential time intervals (0, 5, 10, 15, 20, 25 and 30 min post-irradiation) (Fig. 4A (ii)). We observed that both NKAMP and PCN-224 sustained ROS production throughout the 30-min monitoring period, with PCN-224 reaching saturation at 20 min, while NKAMP continued generating ROS until 25 min, indicating prolonged active ROS generation duration in the drug-loaded formulation. At 15 min, 20 min, 25 and 30 min, NKAMP exhibited significantly higher fluorescence intensity (p < 0.01), showing greater ROS levels compared to PCN-224 at corresponding timepoints. The prolonged ROS generation window coupled with consistently elevated ROS levels confirms that 2Me not only sustains the duration of photodynamic activity but also amplifies its oxidative output.
We attribute this enhanced ROS production to a dual-source mechanism: the direct photodynamic ROS burst from PCN-224 and the intrinsic pharmacological ROS induction by 2ME via suppressing HIF-1α function. The sustained high levels of oxidative stress are a consequence of the disrupted antioxidant defenses resulting from HIF-1α inhibition. To confirm that 2ME functions as a HIF-1α inhibitor, we performed HIF-1α immunofluorescence staining in MDA-MB-231 cells treated with PCN-224, MP, NKMP, or NKAMP. Laser confocal imaging (Fig. 4B(i)) revealed strong HIF-1α expression in the control and PCN-224 groups, whereas significantly reduced signals were observed in 2ME-loaded groups (MP, NKMP, and NKAMP). Mean fluorescence intensity (MFI) quantification confirmed distinct downregulation of HIF-1α expression (71.88% reduction compared to control, Fig. 4B(ii)) upon 2ME treatment.
HIF-1α is a master regulator of hypoxia. HIF-1α suppression by 2ME disrupts cellular antioxidant defenses and amplifies ROS activity through multiple pathways (Fig. 4C). First, Mitochondrial ETC (electron transfer chain) dysfunction: HIF-1α inhibition impairs ETC function, causing electron leakage at Complex I (FMN→O2⁻) and Complex III (Q•⁻→O2⁻), leading to increased ROS production [47]; (2) Antioxidant Suppression: HIF-1α inhibition reduces Nrf2 activity, decreasing SOD2, catalase and GSH, compromising ROS clearance capacity [48, 49]; (3) Iron Dysregulation: HIF-1α inhibition disrupts iron homeostasis, elevating free Fe2+ that catalyzes Fenton reactions (H2O2→•OH), enhancing oxidative damage by ROS [50]. Our data confirmed NKAMP treatment indeed induced ferroptosis primarily upregulating free Fe2+ (figure S9). These mechanisms collectively create a self-reinforcing oxidative cascade that overwhelms cellular repair systems.
We further examined whether amplified ROS triggers mitochondrial apoptosis using JC-1 staining to monitor mitochondrial membrane potential (ΔΨm, Fig. 4D). JC-1 dye forms J-aggregates emitting orange-red fluorescence in polarized mitochondria but shifts to green-fluorescent monomers during ΔΨm depolarization. Control cells maintained 92% red fluorescence, while NKAMP and PCN-224 treatments following laser irradiation (655 nm, 30 mW/cm², 10 min) resulted in decreased red fluorescence to 28% and 49% respectively, with concurrent increases in green fluorescence to 72% versus 51% (p < 0.001), demonstrating NKAMP’s superior ΔΨm disruption capacity. These results, obtained from n = 3 replicate experiments with high reproducibility (SD < 5%), confirm that NKAMP enhances PCN-224’s photodynamic effects by potentiating ROS-mediated mitochondrial damage, leading to more efficient ΔΨm collapse and significantly increased cytotoxicity.

In vivo biodistribution and targeting efficacy
NKAMP was engineered with a dual targeting approach as illustrated in Fig. 5A. The dual functionalization is a strategic design to create a robust and redundant targeting system that counters a key tumor immune evasion mechanism. On the one hand, protein CD226 and NKG2D on NK cell membrane mediate tumor-specific recognition through stress-induced ligand interactions: CD226 binds CD112/CD155 while NKG2D engages MHC class I-related molecules (MIC-A/B in humans), Fig. 5Ai, both selectively upregulated by oncogenic/DNA damage stress in cancer cells. Their combinatorial signaling exploits tumor-specific ligand patterns absent in healthy cells, and forms high-avidity immune synapses through receptor clustering, ensuring discrimination between normal and malignant cells. In addition, tumors frequently downregulate MHC class I molecules to escape T-cell surveillance. Crucially, many stress-induced ligands recognized by NK cell receptors (e.g., NKG2D) are often co-downregulated or shed by aggressive tumors. This can render the innate targeting of the NK membrane partially ineffective. To overcome tumor immune evasion mediated by MHC class I downregulation in cancer cells, a DNA-based fluorescent aptamer Apt specifically recognizing mucin 1 (MUC1), a pan-adenocarcinoma biomarker overexpressed in most of breast malignancies (Figure S10) and is often independent of MHC I status, was functionally anchored onto NK membrane via two C18 chain, Fig. 5Aii. The hydrophobic C18 chains spontaneously inserted into the lipid bilayer of the NK membrane via hydrophobic interactions, while the MUC1-binding aptamer remained exposed on the nanoparticle surface. In this way, dual-strength targeting (NK membrane + aptamer) ensures highly tumor-specific drug delivery.
Female nude Balb/c mice were selected as the animal model to evaluate the in vivo targeting effect of the nanodrug on breast cancer cells. PCN-224, NKMP and NKAMP nanodrugs were venously injected into tumor-bearing mice, respectively. Live-animal fluorescence imaging was performed after 2 h, 4 h, 8 h, 12 h, 24 h and 48 h. The mean fluorescence intensities (MFI) at all time points revealed preferential tumor accumulation of all formulations, with NKAMP demonstrating significantly higher tumor fluorescence intensity than PCN-224 and NKMP, Fig. 5B (n = 3), validating the dual-targeting advantage conferred by NK membrane camouflage and DNA aptamer functionalization. Tracking the fluorescence intensity at different times (Fig. 5C) showed that the fluorescence intensity reached the highest at 8 h after injection. With the extension of time, the fluorescence gradually decreased, indicating that NKAMP can be excreted from the body along with its own metabolism and has high biocompatibility. At 48 h post-injection, the NKAMP group exhibited a significantly higher mean fluorescence intensity (MFI) than both the NKMP and PCN-224 groups, with a striking 3.04-fold greater tumor accumulation compared to PCN-224 alone (p < 0.05). This demonstrates NK cell membrane camouflage enables deep tumor stroma penetration, superior tumor retention and enhanced active targeting capability of the NKAMP nanodrug.

in vivo Organ imaging (Fig. 5D) and quantitative biodistribution analysis (Fig. 5E) also demonstrated tumor-specific accumulation following the hierarchy: NKAMP > NKMP > PCN-224 (MFIs were 62.25 ± 0.5, 42.25 ± 0.7, and 21.56 ± 0.3 respectively). Notably, NKAMP achieved the critical threshold of tumor accumulation surpassing hepatic uptake (tumor-to-liver ratio ~ 3.5), a key indicator for minimizing off-target toxicity. This comparative profile highlights NKAMP’s dual-targeting superiority over NKMP’s single-targeting (NK membrane only) and PCN-224’s passive EPR-dependent accumulation. The synergistic integration of NK membrane-mediated immune evasion with aptamer-directed MUC1 recognition enables precise biological navigation, collectively addressing both systemic clearance and tumor penetration challenges.
The biodistribution data revealed predominant accumulation of NKAMP in the liver and spleen, with negligible signal detected in the kidneys. This pattern suggests that the nanoplatform is primarily cleared via the hepatobiliary system, a common route for nanoparticles with a hydrodynamic diameter exceeding the renal filtration cutoff (~ 5–6 nm). The absence of renal clearance is advantageous, as it prevents rapid elimination from the bloodstream, thereby favoring prolonged circulation and enhanced tumor targeting through the EPR effect.

PDT therapy of TNBC with NKAMP nanodrug
Before conducting photodynamic therapy (PDT) with the fabricated nanodrug in an animal model, cellular-based PDT was performed. Under dark conditions, PCN-224, MP, and NKAMP exhibited excellent biocompatibility in cytotoxicity assessments, maintaining > 90% cell viability across all tested nanodrugs (Fig. 6A PCN-224, MP, and NKAMP lines and Figure S4). Upon laser irradiation (655 nm, 60 mW/cm², 10 min), dose-dependent cytotoxicity emerged. At 40 µg/mL, the cell viabilities of PCN-224, MP, and NKAMP were 81.72%, 63.76%, and 59.12%, respectively; while at 100 µg/mL, they decreased to 37.50%, 18.68%, and 17.15%, respectively. The PDT effects of MP and NKAMP were significantly higher than those of PCN-224 (Fig. 6A, PCN-224 + laser, MP + laser, NKAMP + laser lines), indicating that the introduction of 2 ME enhances the PDT effect, which echoes the previous reactive oxygen species generation experiment. Notably, NKAMP retained MP’s cytotoxicity, confirming that the membrane coating preserves the 2Me/PCN-224 synergy without interference. Comparative analysis of cell viability between the MP and NKAMP groups revealed slightly enhanced cytotoxicity in the NKAMP group across all concentration gradients. This effect can be attributed to key mechanism of enhanced cellular internalization of the NKAMP formulation, resulting in higher effective intracellular drug concentrations.
Subsequently, we evaluated the in vivo therapeutic efficacy of PDT in tumor-bearing mouse models. Female nude Balb/c mice were used as the animal model, with tumors inoculated on the dorsal aspect of the right hind limb. The treatment regimen is illustrated in Fig. 6B. As shown in Fig. 6C, in the absence of laser irradiation, only NKMP and NKAMP demonstrated significant tumor growth inhibition (final tumor volumes: 610.7 ± 0.6 mm³ and 506.2 ± 2.1 mm³, respectively, versus PBS control: 1298.8 ± 2.3 mm³). Notably, NKAMP exhibited superior therapeutic efficacy compared to NKMP (p < 0.01), which can be attributed to its dual-targeting mechanism and the immunomodulatory properties of NK membranes (Note: While the immunotherapeutic effects of NK membranes have been well-documented in previous studies [45, 46, 51, 52], they were not specifically investigated in this work). Importantly, MP nanoparticles showed no therapeutic effect without laser activation, confirming the safety advantage of 2ME encapsulation in porous PCN-224 nanocarriers.
Following laser irradiation (655 nm, 200 mW/cm², 10 min), all nanodrug-treated groups (PCN-224, MP, NKMP, and NKAMP) demonstrated significant photodynamic tumor inhibition compared to the PBS control group, which showed marginal tumor volume increase (Fig. 6C, PCN-224, MP, NKMP, NKAMP lines). Quantitative analysis revealed superior therapeutic efficacy in the NK membrane-coated groups (NKMP and NKAMP) relative to both PCN-224 and MP treatments (Fig. 6C, PCN-224 + laser, MP + laser, NKMP + laser, NKAMP + laser lines).While PCN-224 and MP groups showed moderate tumor growth suppression (final tumor volumes: 504.7 ± 0.5 mm³ and 424.2 ± 1.1 mm³ vs. PBS control: 1298.8 ± 2.3 mm³ at day 14, p < 0.01), the NKAMP and NKMP formulations produced marked tumor ablation, demonstrating superior therapeutic outcomes (p < 0.05). This enhanced efficacy can be primarily attributed to the targeting and therapeutic effects mediated by the NK cell membrane coating. Natural killer (NK) cell membranes exhibit remarkable therapeutic potential by harnessing the innate biological intelligence of NK cells to achieve targeted tumor suppression [45, 46]. These membranes naturally retain the critical targeting receptors and immune-modulating proteins of their parent cells, allowing them to selectively seek out and bind to malignant cells through stress ligand recognition while avoiding healthy tissues. Beyond simple targeting, NK membranes actively participate in tumor suppression through their embedded signaling molecules that can trigger immunogenic cell death and modulate the tumor microenvironment [51, 52]. When engineered into nanoplatforms like NKAMP, these membranes synergistically combine their inherent biological functions with loaded therapeutic agents, creating a multifaceted attack on tumors while maintaining superior safety compared to live cell therapies. This biomimetic approach capitalizes on nature’s own defense mechanisms while overcoming many limitations of conventional immunotherapies.
Notably, this effect was more pronounced in vivo than in cellular experiments (in vitro), likely due to the comprehensive immune response in live animals. Notably, NKAMP demonstrated the most potent ablation effect, reducing tumor volumes to 41.3% of initial size (69.2% for NKMP), representing 31.7-fold greater shrinkage than PBS group (p < 0.001). The superior therapeutic outcomes of NKAMP is attributed to the enhanced tumor targeting capability conferred by the anti-MUC1 aptamer modification.
Comparing the PCN-224 and MP groups, MP group showed significant higher therapeutic efficiency, indicating the evident advantage of 2ME loading on PCN-224. To sum up, NKAMP leaded to 96.16% tumor volume reduction with no systemic toxicity, achieving 10.9-fold greater tumor reduction (volume) than PCN-224.
Post-treatment analysis included tumor dissection and mass quantification, with representative tumor photographs presented in Fig. 6D. Quantitative measurements revealed significantly reduced tumor weights in all laser-irradiated groups compared to their non-irradiated counterparts (Fig. 6E). Particularly, the NKAMP group exhibited the most pronounced reduction in tumor mass (PCN-224: 0.42 ± 0.03 g versus NKAMP 0.04 ± 0.003 g), with excised tumors showing characteristic yellowish-white discoloration - a morphological indicator of complete growth arrest and irreversible tumor damage that suggests durable therapeutic effects without recurrence potential.
Physiological monitoring data demonstrated excellent treatment tolerability, as evidenced by stable body weights throughout the study period (Fig. 6F). Whole-animal photographic documentation of treatment outcomes is provided in Figure S11, confirming the absence of systemic toxicity while maintaining therapeutic efficacy.

Conclusions
In this study, we successfully developed a biomimetic nanoplatform (NKAMP) that synergistically integrates targeted photodynamic therapy with immunotherapy for precision breast cancer treatment. The engineered nanotherapeutic agent combines: (1) a porphyrin-based metal-organic framework (PCN-224) as the photosensitizer core, (2) 2-methoxyestradiol (2ME) as a HIF-1α inhibitor to breaking the HIF-1α-induced therapeutic resistance loop, (3) NK cell membrane coating for immune-compatible targeting, and (4) MUC1-specific DNA aptamers (2×C18AptM) for enhanced tumor specificity. Key advancements of this study include:

A delivery system overcoming triple-barrier via (1) Physical barrier penetration: NK cell membrane camouflage enables deep tumor stroma penetration, achieving 3.04-fold higher accumulation than conventional nanosystems. (2) Hypoxia barrier disruption: 2ME release suppresses HIF-1α by 71.88%, disrupting the hypoxic response pathway and sensitizing cells to PDT under hypoxic conditions. (3) Immune evasion barrier: Synergistic targeting via preserved NK membrane receptors (CD226/NKG2D) and MUC1 aptamers bypasses tumor defense mechanisms (e.g., MHC I downregulation).

Self-Reinforcing Therapeutic Efficacy: The proposed NKAMP platform triggers mitochondrial apoptosis (72% ΔΨm loss) and achieves 96.16% tumor regression in vivo, the highest reported for PDT in TNBC models; It also exhibits zero systemic toxicity with > 90% metabolic clearance within 72 h.

These results position NKAMP as a promising next-generation therapeutic that overcomes key limitations of conventional breast cancer treatments. Its advantages stem from dual-targeting precision (integrating immune recognition with molecular targeting), a self-amplifying therapeutic mechanism (simultaneously inhibiting hypoxia adaptation and enhancing ROS), and an excellent safety profile. The platform’s modular design also suggests broad potential, as it could be adapted to other cancers by substituting targeting moieties. It is important to acknowledge, however, that several challenges remain on the path to clinical translation, including: (1) the complexity of its multi-step fabrication, which may pose challenges for scalable GMP production; (2) the need for further investigation into its long-term biodistribution, chronic toxicity, and detailed immune activation mechanisms in vivo; and (3) the requirement to validate its efficacy beyond the TNBC models demonstrated here.

Materials and methods

Cell culture
NK cells (ATCC) were maintained in α-MEM complete medium (Cobioer) under standard culture conditions. Upon reaching a minimum density of 1 × 10^6 cells with ≥ 90% viability (determined by trypan blue exclusion), cells were processed for membrane isolation. All subsequent steps were performed under chilled conditions using pre-cooled (4 °C) equipment and buffers.
The cell suspension was first washed twice with ice-cold PBS (600 ×g, 5 min initial spin followed by 600 ×g for 1 min) to remove culture medium components. The washed cell pellet was then resuspended in 1 mL of Membrane and Cytosol Protein Extraction Kit Buffer A (Beyotime) containing freshly added 100 mM PMSF protease inhibitor (100 µL per 1 mL buffer). After 15 min of ice incubation to allow membrane destabilization, complete cell lysis was achieved through three cycles of alternating liquid nitrogen immersion (30 s) and room temperature thawing (2 min).
The lysate was clarified by low-speed centrifugation (700 ×g, 10 min, 4 °C) to remove nuclei and cellular debris. The resulting post-nuclear supernatant was ultracentrifuged (14,000 ×g, 30 min, 4 °C) to pellet the crude membrane fraction. Membrane protein concentration was quantified using a BCA Protein Assay Kit (Merck Millipore) according to manufacturer’s instructions, with absorbance measurements referenced against a freshly prepared BSA standard curve (0.2–2.0.2.0 mg/mL range).

Nanodrug synthesis
The reagents required for material synthesis (zirconium oxychloride octahydrate, 5,10,15, 20-tetris (4-carboxylphenyl) porphyrin, benzoic acid and 2-methoxyestradiol) were all purchased from Maclin.

PCN-224
ZrOCl2·8H2O (30 mg), H2TCPP (10 mg), and benzoic acid (300 mg) were dissolved in DMF (2 mL). The resulting mixture was transferred to a polytetrafluoroethylene-lined autoclave and reacted at 120 °C for 24 h. Upon completion, the reaction mixture was cooled to room temperature and centrifuged to isolate the crude PCN-224 product. To achieve uniform nanoparticle size, the obtained PCN-224 dispersion was sequentially subjected to: (1) probe sonication (360 W, 5 min), followed by (2) water bath sonication (10 min) to enhance dispersion. Finally, a stable PCN-224 nanoparticle suspension was obtained by filtration through a 200 nm polycarbonate membrane.

MP
2-Methylestradiol (2ME, 3 mg) was dissolved in DMSO (1 mL) to prepare a 10 mM stock solution. This 2ME solution was mixed with the pre-synthesized PCN-224 solution at mass ratios (2ME : PCN-224) of 2:1, 1:1, 1:2, and 1:4. Each mixture was incubated in a light-protected constant temperature shaker at 20 °C for 12 h. Then the product was collected by centrifugation at 14,000 × g for 30 min. The resulting pellet was washed three times with DMSO to remove unbound drug, yielding 2-methylestradiol-loaded PCN-224 nanoparticles (designated MP).

NKMP
Isolated NK cells were dispersed in PBS to obtain a membrane suspension (5 mg/mL) and stored for subsequent use. For coating, 20 µL of the membrane suspension was diluted in 480 µL ultrapure water, then mixed with 500 µL of MP solution (0.2 mg/mL). The mixture was homogenized and repeatedly extruded through a 200 nm polycarbonate membrane using a liposome extruder to yield NK cell membrane-coated MP particles (designated NKMP).

NKAMP
A DNA aptamer specifically targeting MUC1 protein was designed (sequence: 5’-GGCTATAGCACATGGGTAAAACGAC-Cy5-3’, denoted 2×C18AptM). The aptamer’s 3’ terminus was conjugated to the fluorophore Cy5, while its 5’ terminus was modified with two C18 alkyl chains. This dual C18 modification enables spontaneous insertion into the lipid bilayer of NK cell membranes, creating a membrane-anchored fluorescent DNA aptamer. The 2×C18AptM was then incorporated into the NKMP nanoparticle surface via post-insertion, yielding DNA aptamer-functionalized NK cell membrane-coated nanoparticles (designated NKAMP).

Flow cytometry analysis
Surface expression of key markers (CD16, CD56, CD226, NKG2D) on NKAMP nanoparticles was evaluated by flow cytometry using immunofluorescence labeling. Samples were incubated with primary antibodies (anti-CD16, anti-CD56, anti-CD226, anti-NKG2D; Abcam) followed by species-matched fluorescent secondary antibodies (Thermo Fisher Scientific). Mean fluorescence intensity (MFI) was quantified to assess marker retention.
MDA-MB-231 cells (ATCC) in logarithmic growth phase were counted and seeded in 35 mm culture dishes at 6 × 105 cells/dish. Following 24 h incubation at 37 °C/5% CO2, PCN-224 and NKAMP nanoparticles were dispersed in PBS and labeled with Cy5 fluorescent dye (Macklin) under dark conditions: 20 min room temperature incubation with agitation, then 10 min low-speed vortexing. Unbound dye was removed by centrifugation (14000 × g, 4 °C, 30 min), with subsequent triple PBS washing of the pellet. Labeled nanoparticles were resuspended in fresh medium at 40 µg/mL and added to cell monolayers. After 1–12 h incubation at 37 °C, cells were gently washed twice with PBS, trypsinized, and resuspended in PBS. Cellular uptake was quantified by flow cytometry (λEx = 642 nm; Cy5 channel). IDEAS software analysis determined nanoparticle internalization using the Internalization Factor (IF) metric。.

ROS measurement
MDA-MB-231 cells in logarithmic phase were counted and seeded in 96-well plates at 1 × 104 cells/well. After 24 h incubation, the medium was aspirated and replaced with 40 µg/mL PCN-224, NKAMP dispersions respectively. Following 6 h incubation at 37 °C, treatments were removed and cells washed twice with PBS. Cells were loaded with 10 µM DCFH-DA fluorescent probe (Beyotime) for 20 min at 37 °C, then washed twice to remove extracellular probe. Each well was irradiated with 655 nm laser (60 mW/cm2) for 0, 5, 10, 15, or 20 min. ROS-generated fluorescence intensity was measured at 520 nm using a microplate reader.

Mitochondrial potential
MDA-MB-231 cells in logarithmic growth phase were trypsinized, counted, and seeded in 35 mm dishes at 6 × 105 cells/dish. After overnight incubation (16–24 h) at 37 °C/5% CO2, cells were treated with 40 µg/mL dispersions of PCN-224 or NKAMP nanoparticles for 6 h. Uninternalized nanoparticles were removed by triple PBS washing. Following medium replacement with fresh complete medium, cells were irradiated with a 655 nm laser (60 mW/cm2) for 10 min. Post-irradiation cultures were maintained for 12 h prior to trypsinization. Mitochondrial membrane potential was assessed using JC-1 staining (Solarbio) per manufacturer’s protocol, with flow cytometric analysis of red/green fluorescence ratios.

Live animal imaging
All animal experimental protocols were performed in accordance with the requirements of the Animal Care Center and the Ethics Committee of Beijing University of Technology, China. (HS202109001) and in accordance with the National Law on the Use of Laboratory Animals (China).
Female Balb/c nude mice (4–5 weeks old, Cancer Hospital Chinese Academy of Medical Sciences) were used to establish breast cancer xenografts. MDA-MB-231 cells were centrifuged (2,000 × g, 10 min), washed thrice with saline, and resuspended in PBS (1 × 107 cells/mL). 0.2 mL suspension was subcutaneously injected into the right hind limb. Tumor growth was monitored until reaching 200 mm³ (volume = 0.5 × length × width2). PCN-224, NKMP and NKAMP nanoparticles were dispersed in PBS (1 mg/mL). Tumor-bearing mice received intravenous injections of 200 µL nanoparticle suspensions via tail vein. Under stable anesthesia (2% isoflurane/O2), in vivo fluorescence imaging (IVIS Spectrum, PerkinElmer) was performed at 2, 4, 6, 8, 24, and 48 h post-injection. Fluorescence intensity (ROI = tumor site) was quantified using Living Image® software.

In vitro and in vivo PDT
In vitro cytotoxicity under laser irradiation was assessed via MTS assay. MDA-MB-231 cells were seeded in duplicate 96-well plates (1 × 104 cells/well) and incubated for 24 h (37 °C, 5% CO2). After aspirating the medium, cells were gently washed with PBS. PCN-224, MP, NKMP, and NKAMP nanoparticles were dispersed in PBS at gradient concentrations (10, 20, 40, 60, 80, 100 µg/mL) using probe sonication (3 min, 40 kHz), then added to plates. After incubation for 6 h, remove the unbound particles and replace with fresh medium. Plates were designated as light-protected incubation and 665 nm irradiation (60 mW/cm2, 10 min). After 24 h re-incubation, medium was aspirated and cells washed. Cell viability was quantified using the CellTiter 96® AQueous Non-Radioactive Cell Proliferation Assay (Promega, G5421) per manufacturer’s instructions. Absorbance at 490 nm was measured on a microplate reader (BioTek Synergy H1).
For in vivo PDT, mice bearing orthotopic tumors were randomly grouped when the tumor volume reached ~ 100 mm³. The tumor region was precisely exposed to a 650 nm laser at a power density of 200 mW/cm² for 10 min. The wavelength of 650 nm was chosen for the reason of its precise match with the strong Q-band absorption peak of PCN-224, coupled with its optimal tissue penetration within the phototherapeutic window. The power density was verified at the tumor surface before each treatment using a calibrated photodiode power meter. 200 µL suspension (1 mg/mL) of PCN-224, MP, NKMP, or NKAMP was injected into tumor-bearing mice. After 24 h, the mice were anesthetized, and the tumor site was irradiated with the laser. This irradiation procedure was repeated twice at 5-day intervals. Subsequently, the mice were monitored for 15 days, with body weight and tumor volume measured every other day. Finally, the mice were euthanized, and the subcutaneous tumors were excised for photographic documentation.

Supplementary Information