CNC–PNIPAM Drug Delivery System Improves Endometrial Receptivity via Natural Killer Cell Immune Tolerance in Endometriosis

Advanced Hub Main Navigation Menu CNC–PNIPAM Drug Delivery System Improves Endometrial Receptivity via Natural Killer Cell Immune Tolerance in Endometriosis Abstract Endometriosis (EM) results in pelvic adhesions and chronic inflammation, which significantly reduce embryo implantation and the clinical pregnancy rates of in vitro fertilization (IVF), affecting 190 million women worldwide. Traditional hormonal and surgical treatments alter ovarian and endometrial function, thereby hindering IVF. An urgent need exists for a localized treatment strategy that does not interfere with IVF procedures, aiming to enhance endometrial receptivity. To address this, we developed an innovative localized drug delivery system that simultaneously targets immune dysregulation and enhances endometrial function. This study reports a dual-drug delivery system based on poly(N-isopropylacrylamide) (PNIPAM) combined with nanocrystalline cellulose (CNC) for the local release of levonorgestrel (LNG) and botropase (BTP) to reprogram uterine natural killer cells and treat EM, overcoming previous drawbacks. The experimental results confirm stable CNC and PNIPAM structures and controlled the release of the composite drug. At concentrations of 0.5 mg/L LNG and 100 mg/L BTP, the release system achieved the highest CD16–/CD16+ ratio and the most pronounced upregulation of autophagy-related molecules in uNK cells in our in vitro assays, indicating a shift in uNK cells toward an immune-tolerant state. In a mouse model of EM, the CNC–PNIPAM drug delivery system significantly improve endometrial receptivity. The CNC–PNIPAM drug delivery system demonstrates the medical potential of the local release of composite drugs, offering a new immunomodulatory approach for the treatment of EM and broader medical applications. 1 Introduction Endometriosis (EM), a chronic inflammatory gynecological disorder affecting approximately 10% (190 million) of women of reproductive age, is a major contributor to infertility. The condition profoundly impacts female fertility by creating a hostile pelvic environment, characterized by adhesions and chronic inflammation, which can disrupt ovulation, sperm transport, fertilization, and, most critically, embryo implantation [1, 2]. This is reflected in the significantly reduced rates of embryo implantation (21%–54%), clinical pregnancy (21%), and cumulative live birth (18%) observed in EM patients undergoing in vitro fertilization (IVF) [3-6]. A key pathophysiological feature underlying implantation failure in EM is impaired endometrial receptivity, the transient period when the endometrium acquires a functional state conducive to blastocyst attachment [7, 8]. The establishment of receptivity is a complex process intricately regulated by immune cells, particularly uterine Natural Killer (uNK) cells. Unlike their peripheral blood counterparts, the predominant uNK cell population (CD56brightCD16-) is not cytotoxic but is instrumental in orchestrating placental development and fostering immune tolerance at the maternal-fetal interface [9, 10]. In EM, this delicate balance is disrupted, with evidence suggesting an aberrant increase in the cytotoxic CD16+ uNK subset and a dysregulated immune milieu that contributes to the refractory endometrium [11, 12]. Concurrently, defects in cellular processes like autophagy, a crucial mechanism for cellular homeostasis, have been implicated in the pathogenesis of EM, further compromising endometrial function [13-16]. Therefore, a therapeutic strategy capable of locally reprogramming uNK cells toward a tolerant phenotype and restoring endometrial health is highly desirable for managing EM-associated infertility. Current EM management relies on surgical intervention and hormonal therapies (e.g., GnRH agonists). However, these approaches present significant drawbacks for women seeking pregnancy. Surgery can diminish ovarian reserve [17], while systemic hormonal suppression of the hypothalamic-pituitary-ovarian axis inhibits ovulation and causes endometrial atrophy, effectively preventing conception and necessitating a delay in IVF treatment [18]. To circumvent these systemic effects, localized drug delivery systems such as the levonorgestrel-releasing intrauterine system (LNG-IUS; Mirena) have been explored. We previously investigated the LNG-IUS in a clinical setting, finding that it could potentially improve the local uterine environment without suppressing systemic ovulation [19]. Despite this advantage, the LNG-IUS has several inherent limitations: 1) a fixed, nonadjustable drug release profile; 2) a common side effect of irregular vaginal bleeding that impacts patient quality of life; 3) a rigid, noncustomizable physical size unsuitable for all uterine anatomies; and crucially, and 4) the capacity to deliver only a single drug (LNG), precluding combination therapy aimed at multiple pathological pathways. To address these shortcomings, an ideal localized therapeutic platform for EM should embody the following characteristics: 1) the capability for tunable and sustained drug release; 2) the versatility to co-deliver multiple therapeutic agents, including both hydrophobic and hydrophilic drugs to target different mechanisms (e.g., progestins and anti-inflammatory/hemostatic agents); and 3) possess injectability and adaptable mechanical properties to conform to the uterine cavity. The thermosensitive poly(N-isopropylacrylamide) (PNIPAM), provides a platform for sustained drug release. PNIPAM-based hydrogels undergo a reversible sol-gel transition near body temperature, making them promising candidates for injectable drug depots [20, 21]. However, the clinical translation of pure PNIPAM hydrogels is hampered by inherent limitations, including poor mechanical strength, limited drug-loading capacity, and slow biodegradation. To overcome these challenges, we introduced nanocrystalline cellulose (CNC) as a functional nanomaterial reinforcer. CNC can be incorporated into polymer matrices to form nanocomposites with significantly enhanced mechanical properties [22]. Furthermore, surface-modified CNC can act as a nanocarrier for hydrophobic drugs, promoting their uniform distribution within the hydrophilic PNIPAM matrix and facilitating a more controlled release profile [18]. Herein, we developed an innovative, injectable, and biodegradable CNC-PNIPAM nanocomposite hydrogel for the intrauterine co-delivery of levonorgestrel (LNG) and botropase (BTP). This dual-drug system is designed to synergistically enhance endometrial receptivity in EM. LNG, a potent progestin, provides localized hormonal control, while BTP, a snake venom-derived protease with purported anti-inflammatory and hemostatic properties, is included to address the aberrant bleeding and inflammatory microenvironment. We hypothesize that this locally applied combination, released in a sustained manner from our advanced material platform, can effectively reprogram uNK cells toward an immune-tolerant state and rectify the molecular signatures of defective receptivity, thereby offering a novel and promising strategy to improve IVF outcomes for women with EM (Figure 1). 2 Results 2.1 Development of the CNC–PNIPAM Drug Delivery System We developed an optimized method for preparing CNC-PNIPAM hydrogels with sustained drug release capabilities. Scanning electron microscopy (SEM) characterization revealed that CTAB-modified CNC nanoparticles exhibited a short rod-shaped morphology (106 ± 16 nm in length) (Figure 2A,B), with Dynamic light scattering (DLS) confirming a negative surface charge (−55 mV) (Figure 2D). Water contact angle measurements were conducted to evaluate the hydrophilicity of the gel. The initial contact angle was 58.45°, which rapidly decreased to 37.52° within 10 s (Figure 2C). This significant decrease and the low final contact angle confirm the excellent hydrophilic nature of the material. The CTAB modification enabled effective loading of hydrophobic LNG into the aqueous-phase hydrogel system. Additional SEM images are provided in Figure S1. The resulting hydrogels displayed an interconnected porous structure with uniform distribution of both drug particles and cellulose nanocrystals (CNCs) within the polymer network. Increasing CNC content reduced micropore diameters due to enhanced physical crosslinking, attributed to additional hydrogen bonding between CNC hydroxyl groups and PNIPAM chains. FTIR analysis verified these interactions through a characteristic shift in the –OH stretching vibration from 3550 to 3420 cm−1 with increasing CNC content (Figure 2E). These hydrogen bonds not only stabilized the PNIPAM network but also improved mechanical properties through efficient load transfer. The combination of PNIPAM's inherent biocompatibility and prolonged circulation characteristics with this enhanced mechanical stability makes these nanocomposite hydrogels particularly promising for therapeutic delivery applications. The viscoelastic properties of the gels were characterized. When the storage modulus (G’) exceeds the loss modulus (G”), the material behaves as an elastic solid, meaning it can maintain its structural integrity and resist deformation without collapsing. All three gel samples maintained this solid-like, stable state within a strain amplitude of 10%. Since the intended application as a drug carrier does not subject the gel to larger deformations, its stable performance within this 10% strain range satisfactorily meets the mechanical requirements for this purpose (Figure 2G). Figure 2H illustrates the fabrication process of the CNC-PNIPAM hybrid hydrogel. The synthesis proceeds through in situ free radical polymerization of NIPAM monomers with CNCs serving as reinforcing templates. This process yields a thermosensitive network capable of being loaded with therapeutic agents for drug delivery applications. 2.2 The Sustained Release Curve for CNC–PNIPAM All hydrogel formulations demonstrated excellent drug loading capacities (>0.02 μg/mL) with high encapsulation efficiencies (Figure 2F). Release kinetics studies revealed an initial burst release phase, attributed to rapid diffusion of surface-associated LNG into the surrounding buffer due to the steep concentration gradient. This was followed by a slower, sustained release phase as the system transitioned to diffusion-controlled kinetics through the hydrogel matrix. 2.3 Assessment of LNG-IUD's Early Impact on Human Endometrium Immunohistochemistry showed increased expression of autophagy markers (Beclin-1, LC3B) in endometrial epithelial cells after LNG-IUD treatment, indicating heightened abundance of autophagy markers versus baseline (Figure S2A). This suggests a potential restoration of autophagy-related function in EM patients. Stromal (vimentin+) and epithelial (CK7+) cells isolated from fresh endometrial tissue (Figure S2B) exhibited elevated integrin αvβ3 levels post-LNG by flow cytometry. LNG exposure also expanded NK and T cell populations in endometrial tissue, though macrophage (Mφ) numbers were unaffected (Figure S2C). NK cells profiling demonstrated a preferential increase in CD16– subsets within the uterine cavity, while endometrial stromal and epithelial compartments displayed a rise in CD16+ NK cells, further skewing the CD16–/CD16+ balance (Figure S2C). Additionally, LNG treatment upregulated key endometrial receptivity genes (progesterone receptor (PGR), Indian Hedgehog (IHH), IL-18, PTGS2, ITGAV) (Figure S2D). 2.4 In Vitro Evaluation of CNC–PNIPAM Efficacy 2.4.1 Optimization of Drug Combination within the Tested Range The CD16− phenotype, a critical marker of decidual NK cells implicated in maternal-fetal immune tolerance, exhibited concentration-dependent modulation under varying drug conditions. Notably, 100 mg/L BTP induced a significantly higher proportion of CD16− NK cells compared to both lower (50 mg/L) and higher (200 mg/L) concentrations (Figure 3A). This pattern was further amplified by LNG co-treatment, with 0.5 and 1 mg/L LNG showing a more pronounced effect over untreated controls or 2 mg/L LNG. The CD16−/CD16+ ratio followed an identical concentration-response relationship (Figure 3B), reinforcing the dose-specific effects. In parallel experiments screening 12 material ratios on endometrial stromal cells, western blot analysis revealed striking concentration-dependent effects on key functional markers. Autophagy regulators (LC3B, Beclin-1), the proliferation marker Ki-67, and adhesion mediator αv0β3 integrin all showed maximal expression at 100 mg/L BTP, with intermediate and minimal levels at 200 and 50 mg/L BTP, respectively (Figure 3C). This most effective combination within the tested range was further validated through polymerase chain reaction (PCR) analysis of five endometrial receptivity markers, which mirrored the protein expression patterns (Figure 3D). Collectively, across the parameters we measured, the combination of 100 mg/L BTP with 0.5–1 mg/L LNG yielded the most favorable outcomes among all tested formulations. 2.4.2 Effect of CNC–PNIPAM on In Vitro Cell Lines Co-culture of decidual NK (dNK) cells and human endometrial stromal cells (HESCs) with 100 mg/L BTP revealed distinct modulation of cytokine and adhesion molecule expression. Flow cytometric analysis demonstrated that NK cell activity-associated cytokines were differentially regulated upon exposure to 0.5 mg/L LNG, with CD107a and interferon (IFN)-γ showing marked upregulation, whereas granzyme B remained unaffected across all concentrations (Figure 4A). Notably, adhesion factor profiling on dNK cells indicated a concentration-dependent response, with CD62E and CD54 peaking at 0.5 mg/L LNG, while CD106 expression remained unchanged regardless of treatment (Figure 4B). These findings imply a potential shift toward immune tolerance in NK cells under LNG exposure. Similarly, HESCs exhibited LNG-dependent alterations in adhesion molecule expression. Strikingly, CD62E, CD54, and CD106 levels were maximally elevated at 0.5 mg/L LNG, with CD106 additionally showing sustained high expression even at 1.0 mg/L LNG (Figure 4C). Collectively, these data suggest that LNG modulates both NK cell functionality and stromal cell adhesion properties, with 0.5 mg/L representing a critical concentration for maximal effector responses. 2.4.3 CNC–PNIPAM System Enhances Endometrial Receptivity In Vivo Intrauterine delivery of CNC–PNIPAM loaded with 100 mg/L BTP and 0.5 mg/L LNG markedly restored endometrial receptivity in a murine model. Twelve C57BL/6 mice were stratified into four cohorts (n = 3 per group): untreated healthy controls (Ctrl), healthy mice receiving drug treatment (Ctrl-treated), EM-modeled mice, and EM mice undergoing therapeutic intervention (EMs + treatment). A single 0.1 mL dose was administered transcervically to ensure localized uterine delivery. Notably, drug exposure attenuated uNK cell cytotoxicity, as evidenced by reduced CD107a and IFN-γ levels in both healthy and EM mice (Figure 5A). Intriguingly, granzyme B suppression occurred exclusively in healthy animals, suggesting a differential immunomodulatory effect in pathological versus physiological states. Strikingly, endometrial receptivity gene expression was universally elevated post-treatment, as quantified by reverse transcription (RT-PCR) analysis of five key molecular markers (Figure 5B). Immunohistochemical profiling revealed broad endometrial activation, with significant upregulation of autophagy markers (LC3B, Beclin-1), proliferative driver Ki-67, and the adhesion mediator αvβ3 integrin (Figure 5C). Importantly, baseline expression of receptivity-associated proteins remained consistently higher in healthy controls than in EM mice, underscoring the disease-associated impairment in endometrial function. 3 Discussion Our study demonstrates the successful development of a thermoresponsive CNC-PNIPAM nanocomposite hydrogel for dual-drug delivery, exhibiting optimized mechanical properties and controlled release kinetics for both hydrophobic LNG and hydrophilic BTP. The incorporation of CNC nanoparticles significantly enhanced the hydrogel's mechanical strength while maintaining its characteristic thermosensitivity, with the lower critical solution temperature transition at 32°C enabling precise temperature-controlled drug release, a critical feature for intrauterine applications. The molecular markers selected for evaluating endometrial receptivity are well-established indicators of the window of implantation. Integrin αvβ3 (ITGAV/ITGB3) facilitates embryo adhesion to the endometrial epithelium [23-27]. The PGR is essential for mediating the effects of progesterone in preparing the endometrium for implantation. IHH, a downstream target of PGR, is a critical signaling molecule in uterine receptivity and embryo-uterine crosstalk. Additionally, we assessed interleukin-18 (IL-18) and prostaglandin-endoperoxide synthase 2 (PTGS2/COX-2), which are involved in the inflammatory and vascular changes necessary for successful implantation. The coordinated expression of these molecules is indicative of a receptive state, and their dysregulation has been consistently reported in EM-associated infertility [7, 8, 23]. The differential behavior of uterine NK (uNK) cells versus peripheral blood NK (pbNK) cells proved particularly relevant to our therapeutic strategy. While pbNKs display characteristic CD56dimCD16+ cytotoxic phenotypes, the predominant CD56brightCD16− uNK population plays crucial roles in endometrial remodeling and immune tolerance. Notably, we identified that the minor CD56dimCD16+ uNK subset may contribute to EM pathogenesis through cytotoxic effects on endometrial cells, suggesting their modulation as a potential therapeutic target. Through systematic concentration screening, we established that 0.5 mg/L LNG combined with 100 mg/L BTP optimally modulated the endometrial microenvironment. This combination maximized the CD16−/CD16+ uNK ratio while significantly upregulating autophagy markers (Beclin-1, LC3B) and proliferation indicators (Ki-67), all previously shown to be deficient in EM patients. The treatment concurrently enhanced integrin αvβ3 expression and improved endometrial receptivity, as evidenced by upregulation of at least 10 key molecular markers (BMP2, CEBPB, WNT4, etc.). Mechanistic studies revealed that the optimized formulation effectively shifted NK cell activity from cytotoxicity to immune tolerance, as demonstrated by reduced CD107a and IFN-γ expression. Flow cytometry analysis further showed selective upregulation of adhesion molecules CD62E and CD54, known mediators of endometrial immune cell interactions. These in vitro findings were corroborated in our EM mouse model, where treatment consistently improved autophagy markers while reducing inflammatory cytokines and enhancing receptivity factors. The drug concentrations employed in the murine EM model were selected based on their superior efficacy in our in vitro screening. Notably, the local concentration of LNG (0.5 mg/L) is comparable to that delivered by clinically used levonorgestrel-releasing systems, supporting its physiological relevance [19]. The concentration of BTP (100 mg/L) was chosen as it represented a critical threshold for modulating NK cell phenotype and receptivity-associated molecules in our cellular assays. A key question arising from our findings is whether the observed therapeutic benefits are attributable to LNG, BTP, or their synergistic action. While this study was designed to test the efficacy of the combined system, our data offer insights into this interplay. The concentration-dependent effects shown in Figure 3 demonstrate that the most robust and coordinated improvements in uNK cell phenotype (CD16–/CD16+ ratio), autophagy markers, and receptivity genes were achieved only with the specific combination of 0.5 mg/L LNG and 100 mg/L BTP. LNG, as a progestin, is a known immunomodulator, but its efficacy can be limited in the inflammatory milieu of EM. We postulate that BTP, by mitigating local hemorrhage and inflammation, may normalize the endometrial microenvironment. This ‘priming’ effect could, in turn, sensitize uNK and stromal cells to the immunomodulatory actions of LNG, resulting in a synergistic restoration of receptivity. This proposed synergy between a hormonal agent and a hemostatic/anti-inflammatory enzyme represents a novel therapeutic paradigm for EM that warrants dedicated future investigation with single-agent controlled studies. While our data demonstrate a clear association between LNG-BTP treatment, enhanced autophagy, improved integrin expression, and a shift in uNK cell phenotype toward tolerance, the precise signaling pathways that orchestrate these changes warrant further investigation. Based on existing literature, it is plausible that the observed effects are mediated through the modulation of key signaling nodes. For instance, the promotion of autophagy and resolution of inflammation are often linked to the inhibition of the mTOR/PI3K/Akt pathway. Conversely, the upregulation of integrin αvβ3 and other adhesion molecules is a hallmark of integrin-mediated focal adhesion signaling, which can be influenced by both extracellular matrix composition and intracellular autophagy flux. A compelling future direction will be to investigate whether the LNG-BTP combination acts through these pathways to create a coordinated signaling network that ultimately reprograms the endometrial microenvironment and uNK cell function in EM. This study, however, has certain limitations. The sample size in the animal model was relatively small, and the long-term biodegradation profile of the hydrogel was not evaluated. Furthermore, it is also important to note that while the thermosensitive nature of PNIPAM is central to its injectable depot function, the specific role of temperature-mediated release in vivo requires further elucidation in future studies. Future work will focus on investigating the precise signaling pathways involved in NK cell reprogramming and conducting long-term safety and efficacy evaluations in larger animal models. While the CNC-PNIPAM platform shows promising efficacy, its translational potential requires careful consideration of long-term biocompatibility. An acknowledged limitation of this study is the absence of data on the in vivo degradation fate of the hydrogel and potential chronic tissue response. Pure PNIPAM is criticized for its slow biodegradation and risk of accumulation. In our design, the incorporation of CNC aims to mitigate this by creating a more hydrophilic and mechanically robust composite network, which is anticipated to exhibit more favorable degradation kinetics, though this requires formal validation in future studies. Dedicated investigations tracking the hydrogel's residence time, degradation products, and associated immune response (e.g., foreign body reaction) over extended periods (≥28 days) are a critical next step. Furthermore, regarding the safety of Botropase (BTP), its localized intrauterine delivery is a key feature that minimizes systemic exposure. The uterinefirst -pass effect and the expected low bioavailability into the systemic circulation substantially reduce the risk of off-target systemic coagulation. The action of BTP is designed to be confined to the endometrial tissue, where it can exert its hemostatic and potential anti-inflammatory effects precisely at the site of endometriotic pathology. Looking toward clinical translation, the most feasible administration route for our CNC-PNIPAM system is minimally invasive intrauterine injection. This approach leverages the injectable nature of the sol at room temperature and its rapid gelation at body temperature to form a localized drug depot. This strategy offers distinct advantages: (1) it ensures high drug bioavailability at the endometrial target site while minimizing systemic exposure and off-target effects; (2) it is highly compatible with IVF treatment cycles, as it can be administered in a single, brief procedure (e.g., during hysteroscopy or prior to embryo transfer) without the need for surgical implantation or removal that could disrupt the cycle; and (3) it provides superior adaptability to individual uterine anatomy compared to rigid intrauterine devices, potentially improving patient comfort and overcoming the issue of device-size mismatch. Future development will focus on optimizing the injection protocol and confirming the safety and retention duration of the hydrogel in larger animal models. Regarding the clinical dosing strategy, our in vitro release profile suggests the system can provide sustained drug delivery for approximately 10–14 days. This supports the feasibility of a single intrauterine administration timed to precede embryo transfer in an IVF cycle. Such a regimen is designed to ensure optimal drug exposure throughout the critical window of endometrial receptivity. This single-dose approach stands in contrast to daily oral medications and aligns with the discrete timeline of assisted reproduction, offering significant patient convenience. Future work will focus on validating this release duration in vivo and correlating it with the maintenance of therapeutic efficacy. Despite the promising efficacy, the translational path for this CNC-PNIPAM system is accompanied by several defined challenges that represent the focus of our ongoing development. First, a comprehensive safety assessment is paramount. While the incorporation of biodegradable CNC is designed to improve the profile of PNIPAM, dedicated long-term studies in large animals are required to fully evaluate biodegradation, biodistribution, and chronic local and systemic toxicity. Second, from a regulatory perspective, this platform constitutes a combination product (drug-device), necessitating a complex approval pathway that demands rigorous characterization and manufacturing control under GMP conditions. Finally, scalable and reproducible manufacturing of both pharma-grade CNC and the sterile hydrogel formulation presents significant engineering hurdles. Addressing these challenges through targeted future studies will be essential to advancing this technology toward clinical application. This study focuses on the efficacy of a novel drug delivery platform; however, we acknowledge that a comprehensive assessment of its long-term biodegradation and systemic safety profile is an essential next step. While our initial in vivo observations showed no adverse effects, future work will include detailed histopathological examination of reproductive organs and major viscera (e.g., liver, kidney, spleen) following extended exposure, coupled with serum biochemistry panels to monitor organ function and systemic inflammation. Furthermore, dedicated studies on the hydrogel's degradation kinetics and the clearance pathways of its components are warranted to fully address the potential for in vivo accumulation, a known challenge with PNIPAM-based systems. The incorporation of biodegradable CNC is a strategic step toward this goal, and its impact on the overall safety profile will be a central focus of our subsequent research. The convergence of these findings suggests that our dual-drug hydrogel system achieves three critical therapeutic effects: (1) restoration of defective autophagy processes in endometrial cells, (2) modulation of uNK cell populations toward immune-tolerant phenotypes, and (3) improvement of endometrial receptivity markers. The temperature-responsive drug release properties coupled with these biological effects position this system as a promising approach for localized EM therapy, potentially overcoming limitations of current systemic treatments. 4 Conclusion In summary, we developed an improved method for preparing hydrogels and generated a novel CNC–PNIPAM drug delivery system with sustained release. This system, which combined modified CNC nanoparticles with a PNIPAM matrix, to enable the co-delivery of hydrophilic and hydrophobic drugs, providing controlled release and excellent injectability. We also evaluated the apparent performance and drug release properties of this system, explored the most promising formulation based on our screening criteria, and assessed its effects on human dNK cells, endometrial cells, and uNK cells function in an in vivo EM mouse model, as well as its impact on endometrial receptivity. This study provides a novel strategy for the treatment of EM. 5 Experimental Section 5.1 Reagents and Cell Lines The sodium hydroxide, sulfuric acid, potassium persulfate, N,N,N′,N′-Tetramethylethylenediamine were purchased from Shanghai Sinopharm Chemical Reagent Co., Ltd. The N-Isopropylacrylamide, Cetyltrimethylammonium bromide, and DMSO were purchased from Shanghai Aladdin Biochemical Co., Ltd. The human endometrial stromal cell lines were purchased from Procell Corporation (Wuhan, China).10% normal goat serum and fetal bovine serum were purchased from Gibco (California, USA). DMEM/F12 medium and RPMI/1640 medium were purchased from Hyclone (Utah, USA). Penicillin-streptomycin double antibody, and Tryptase containing 0.25% EDTA were purchased from NCMbio (Suzhou, China). Percoll, collagenase type IV were purchased from Sigma (St. Louis, USA). Total RNA was extracted using the TRIzol reagent from Invitrogen (California, USA, 15596018). Subsequently, RNA underwent reverse transcription to generate cDNA employing the PrimeScript RT reagent Kit from Takara (Kyoto, Japan, RR036A). RT-PCR was conducted utilizing SYBR Premix Ex Taq from Takara (Kyoto, Japan, RR820A). Rapid lysis buffer was purchased from (RIPA; Beyotime, P0013B) containing 1% phenylmethanesulfonyl fluoride (PMSF; Beyotime, ST506). Protein concentrations were quantified using the BCA protein assay kit (Beyotime, P0012).T SDS-PAGE gels were purchased from (Epizyme Biotech, PG113) using a Miniprotein III system (Bio-Rad, 1658033). PBST solution was purchased from Beyotime (Shanghai, China, P0222). Peroxidase-conjugated goat anti-mouse IgG secondary antibody was purchased from Bioworld Technology (1:5000; BS10003). Immobilon Western Chemiluminescent HRP Substrate Kit was purchased from (Millipore, WBKLS0100). Xylene and absolute ethanol were purchased from Sinopharm Group Chemical Reagent Co., LTD (Shanghai, China). EDTA antigen repair solution, spontaneous fluorescence quencher, fluorescent secondary antibody, DAPI dye, anti-fluorescence quenching sealing tablet, and environmentally friendly dewaxing solution were purchased from Servicebio (Wuhan, China). Flow cytometry antibodies, Fixation Buffer, Perm Wash Buffer, and Cell Activation Cocktail (with Brefeldin A) were purchased from Biolegend (California, USA). NK cell isolation kits were purchased from Miltenyi (Shanghai, China). 5.2 Preparation of CNCs CNCs were prepared by treating commercially available cotton with sodium hydroxide, followed by concentrated sulfuric acid. The reaction mixture was heated to 65°C for approximately 80 min, then diluted with water and allowed to stand overnight. The suspension was subsequently washed and centrifuged (4000 rpm) and dialyzed for 3 days. The obtained CNCs suspension was diluted to 0.4 wt%, slowly added to the CTAB solution, and heated at 3 h (60°C). The mixture was stirred overnight, followed by centrifuging (4000 rpm, 10 min), and the hydrophobic drug levonorgestrel was added to the modified CNC suspension. 5.3 Preparation of PNIPAM Cross-linked CNCs Drug Delivery System Cationic-modified CNCs were combined with NIPAM at a 5:0.5 mass ratio (CNCs solution:NIPAM) along with 0.05 g each of BIS and APS, using TEMED as a catalyst to incorporate CNCs as reinforcing phases within crosslinked PNIPAM for hydrophobic drug loading. Dual drug loading was achieved by either adding hydrophilic drugs like botropase during polymerization or through post-synthesis swelling absorption, where the partially hydrated PNIPAM matrix spontaneously uptakes aqueous drug solutions until reaching saturation. Hydrophilic drug incorporation was performed by incubating drug solutions with pre-formed hydrogels containing modified CNCs and levonorgestrel for 24 h under sealed conditions. The mechanical properties of the resulting hydrogels were precisely tuned by varying CNC concentrations to achieve optimal modulus ranges for biomedical applications. 5.4 In Vitro Drug Release Study The drug release profile of Levonorgestrel (LNG) from the CNC-PNIPAM hydrogel was investigated. The drug-loaded hydrogel was immersed in phosphate-buffered saline (PBS, pH 7.4) and incubated at 37°C under constant agitation. At predetermined time intervals, samples of the release medium were withdrawn and replaced with an equal volume of fresh pre-warmed PBS to maintain sink conditions. The concentration of released LNG in the samples was quantified using high-performance liquid chromatography. The cumulative drug release was calculated and plotted as a function of time. 5.5 Tissue Processing and Cell Isolation Endometrial tissue was rinsed with PBS to remove blood contaminants, then minced into fragments using ophthalmic scissors in a 5 mL glass vial. The tissue fragments were digested in 2 mL collagenase IV (1 mg/mL) with 0.1 mL penicillin-streptomycin solution at 37°C with shaking (1500 rpm) for 20–30 min. The digestate was sequentially filtered through 40 μm and 70 μm cell strainers, and the filtrate was centrifuged (1500 rpm, 4°C, 8 min). Cell Separation and Culture: The pellet was resuspended in DMEM/F12 supplemented with 10% fetal bovine serum (FBS) and plated. After 24 h of culture, adherent endometrial stromal cells (ESCs) were separated from nonadherent endometrial immunocytes (EICs) in suspension. EICs were collected by centrifugation (1000 rpm, 4°C, 5 min) and maintained in RPMI/1640 with 10% FBS at 37°C with 5% CO2. Adherent ESCs were detached using 0.25% EDTA-trypsin for 5 min, neutralized with complete medium, and centrifuged (1000 rpm, 4°C, 5 min) before resuspension in fresh DMEM/F12 with 10% FBS. Cells were passaged at 80%–90% confluence and cultured under standard conditions (37°C, 5% CO2). 5.6 The Selection of Endometrial NK Cells Prepare cells and determine the cell count using a cell counting plate; Suspend every 10 cells in 40 µL magnetic-activated cell sorting (MACS) buffer. Add 10 µL NK cell Biotin-cocktail per 10 cells. Mix thoroughly by pipetting and incubate at 2°C–8°C for 5 min. Then, add 30 µL MACS buffer and 20 µL NK cell MicroBead Cocktail per 10 cells. Mix again by pipetting, and incubate at 2°C–8°C for 10 min. For magnetic bead separation, place the MS sorting column on a magnetic field sorting frame of MACS. Pre-rinse the column, the sorting column, with 500 µL MACS buffer. The cell suspension was added to the center of the column. Collecting the liquid passing through the sorting column, which contains unlabeled cells, represents the enriched NK cell population. Wash the sorting column with 500 µL MACS buffer and combine it with the effluent from the previous step. 5.7 FCM After treatment with Levonorgestrel (LNG) and Batropase (BTP), or co-culture with control eESCs (ESC-NK) or PPD-pretreated eESCs (ESC-PPD-NK), dNK cells were collected, and the expression levels of CD107a, granzymeB, IFN-γ, CD62E, CD106, and CD54 (all from Biolegend, USA) were analyzed by FCM according to the manufacturer's instructions. In addition, the expression levels of CD62E, CD106, and CD54 (all from Biolgend) in HESCs were also analyzed by FCM. Isotypic IgG antibodies were used as negative controls. The samples were analyzed using a FACS-Calibur flow cytometer (Becton Dickinson, USA) and Cellquest software (Becton Dickinson). Statistical analysis was conducted using isotype-matched controls as references. 5.8 Real time (RT)-PCR Total RNA was extracted using the TRIzol reagent (Invitrogen, 15596018). Subsequently, RNA underwent reverse transcription to generate cDNA employing the PrimeScript RT reagent Kit (Takara, RR036A). RT-PCR was conducted utilizing SYBR Premix Ex Taq (Takara, RR820A) and analyzed via an ABI Prism 7900 Fast Sequence Detection system (Thermo Fisher Scientific). The fold change in transcriptional expression for the aforementioned genes was determined employing the 2− ΔΔCt method, with each sample being analyzed in triplicate wells. Relative mRNA expression levels were normalized to GAPDH. 5.9 Immunohistochemistry The decidual and endometrial tissues were fixed, embedded in paraffin, sectioned to a 5 µm thickness, and subsequently rehydrated using a graded ethanol series (Merck). To inhibit endogenous peroxidase activity and prevent nonspecific antibody binding, the tissues were treated with 3% H2O2 for 30 min and then incubated with 10% normal goat serum (Gibco) diluted in TBS for 30 min at room temperature. Following this, the sections were incubated overnight at 4°C with a rabbit monoclonal antibody against HO-1 (ab52947, 1:2000; Abcam) or a rabbit IgG isotype control. The sections were subsequently washed three times with TBS and stained with the secondary antibody at 37°C for 60 min. Lastly, the sections were developed using 3,3-diaminobiphenylamine (Sigma) and counterstained with hematoxylin. Images were captured using an Olympus BX51 fluorescence microscope. 5.10 Western Blotting HESCs were lysed in high-efficiency cell tissue rapid lysis buffer (RIPA; Beyotime, P0013B) containing 1% phenylmethanesulfonyl fluoride (PMSF; Beyotime, ST506). The lysates were collected with a cell scraper and centrifuged at 12 000× g for 30 min at 4°C. Protein concentrations were quantified using the BCA protein assay kit (Beyotime, P0012). The lysates were boiled at 95°C for 10 min for long-term preservation. A total of 25 µg of protein was separated by SDS-PAGE (Epizyme Biotech, PG113) using a Miniprotein III system (Bio-Rad, 1658033), and transferred to PVDF membranes (Millipore, ISEQ00010) for 90 min. Membranes were blocked with 5% skimmed milk in PBST (Beyotime, P0222) at room temperature for 1 h. The PVDF membranes were incubated overnight with the primary antibody at 4°C. Then, the membranes were washed with PBST solution (Beyotime, P0222) and subsequently incubated at room temperature for 1 h with peroxidase-conjugated goat anti-mouse IgG secondary antibody (1:5000; Bioworld Technology, BS10003). Finally, membranes were washed three times. The PVDF membranes were developed using the Immobilon Western Chemiluminescent HRP Substrate Kit (Millipore, WBKLS0100). 5.11 EMS Mice Model The animal experiments in this study were approved by the Animal Care Committee of Shanghai Ninth Hospital, Shanghai Jiao Tong University School of Medicine (SH9H-2020-A270-1). Five-week-old female C57BL/6 mice were obtained from the Jie Si Jie Experimental Animal Company and were housed under SPF-level conditions with unrestricted access to food and water. An EM model was established by intraperitoneally injecting uterine tissue fragments from donors into recipient mice, with a donor-to-recipient ratio of 1:2. One week after the fragment transplant, half of the recipient mice (EMS mice model) or control mice were randomly assigned to treatment and non-treatment groups. Slow-release drug-carrying materials were injected vaginally into the treatment groups of mice. After four weeks, the mice were euthanized by CO2 asphyxiation, and the ectopic tissues were removed for subsequent assessment. 5.12 Isolation of Murine Uterine Cells For the differential analysis of uNK cells, a two-step isolation protocol was employed to separate cells from the uterine cavity and the endometrial tissue proper. - Uterine Lavage: To obtain cells from the uterine cavity lumen, the uterine horn was carefully exposed post-euthanasia. A 30-gauge blunt needle connected to a 1 mL syringe containing 0.5 mL of ice-cold PBS was inserted into the anterior end of the uterine horn. The PBS was gently flushed through the lumen and collected from the posterior end. This lavage process was repeated twice. The collected lavage fluid was centrifuged (500× g, 5 min, 4°C) to pellet the luminal cells. - Endometrial Tissue Digestion: The lavaged uterine horn was then transferred to a Petri dish, longitudinally dissected, and minced into approximately 1 mm3 fragments. The tissue fragments were transferred to a conical tube containing 2 mL of digestion solution (Collagenase Type IV, 1 mg/mL; DNase I, 0.1 mg/mL in DMEM/F12 medium) and incubated at 37°C with constant shaking at 150 rpm for 45–60 min. The digest was vortexed every 15 min. The reaction was stopped by adding an equal volume of complete DMEM/F12 medium supplemented with 10% FBS. The cell suspension was sequentially passed through 70 and 40 μm cell strainers to remove debris and obtain a single-cell suspension from the endometrial tissue. - Cell pellets from both the lavage and tissue digestion were resuspended separately and subjected to immune cell enrichment using a 40%/70% discontinuous Percoll density gradient centrifugation (800× g, 20 min, room temperature, with low brake). The interphase layer, containing the mononuclear cells, was carefully collected. NK cells were subsequently isolated from this population using a mouse NK Cell Isolation Kit via MACS according to the manufacturer's instructions. The purity of the isolated NK cells was consistently >90% as verified by flow cytometry. 5.13 Intrauterine Drug Administration in Mice Fourteen days post-modeling, intrauterine drug delivery was performed in mice under isoflurane anesthesia. The animals were positioned supine with the perineal area shaved to ensure aseptic conditions. A specialized blunt-tipped catheter was gently inserted approximately 2–3 mm into the vaginal orifice along the posterolateral axis, and 20–50 μL of drug solution was infused slowly and uniformly to minimize reflux. The catheter was retained for 1–2 s post-injection before withdrawal, and the mouse was briefly maintained vertically to facilitate solution retention. This volume was selected to match the physiological capacity of the murine uterine lumen while avoiding overdistension. The blunt-tipped catheter design and controlled insertion depth aimed to reduce the risk of tissue trauma. Post-procedure, mice were singly housed and monitored for 7 days to ensure recovery and model stability before subsequent cellular extraction, thereby maintaining experimental consistency and animal welfare. All reagents used in the above experiments are listed in the Tables S1–S5. 5.14 Statistical Analysis All experiments were repeated in triplicate at a minimum. Data analysis was conducted using GraphPad Prism version 8. Statistical analyses were performed using unpaired t-tests, Mann–Whitney U tests, or one-way ANOVA for the various parameters. Data that followed a normal distribution are presented as the mean ± standard deviation (SD). Statistical significance was defined as p < 0.05. Supporting Information Additional supporting information can be found online in the Supporting Information section. Supporting Fig. S1: SEM images for CNC nanoparticles (5 samples). Supporting Fig. S2: Assessment of LNG-IUD's Early Impact on Human Endometrium. Supporting Table S1: Materials for synthetic CNC. Supporting Table S2: Cell Culture Reagents. Supporting Table S3: Molecular Biology Reagents. Supporting Table S4: Immunology and Immunohistochemistry Reagents. Supporting Table S5: Flow Cytometry Reagents. Author Contributions Z.L. was involved in the data collection and statistical analysis and wrote the article. Y.J.W. did a cell and mouse experiment. Y.L. designed the nano material. X.H. and X.Y.L concentrate on the releasing drug system. Y. W. and Y.S acted as the senior/consulting author and revised the manuscript. All authors have approved the final version of the manuscript. Acknowledgments All procedures involving human participants were performed in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study was approved by the Ethics Committee of Shanghai Ninth Hospital, Shanghai Jiao Tong University School of Medicine (Reference No. SH9H-2020-TK121-1). The authors gratefully acknowledge all the staff of the Department of Assisted Reproduction in Shanghai Ninth People's Hospital for their cooperation and support. This work was supported by the National Nature Science Foundation of China (No. 31770989). Funding National Nature Science Foundation of China (31770989). Conflicts of Interest The authors have no conflicts of interest to disclose. Data Availability Statement The datasets generated and analyzed during the current study, including those for drug release, flow cytometry, qPCR, and histology quantification, are available from the corresponding author on reasonable request.