A sustained NAD⁺ supplementation-biosynthesis nanoplatform for metabolic restoration in aged bone regeneration

Abstract

Increasing NAD⁺ levels has demonstrated promising therapeutic potential for treating aging-related skeletal disorders. However, for existing bone defects in aged individuals, it remains a formidable challenge to achieve localized and sustained NAD⁺ restoration while overcoming poor NAD⁺ delivery efficiency and impaired endogenous biosynthesis in senescent cells. Herein, this study developed a sustained NAD⁺ supplementation-biosynthesis nanoplatform, N/S@M@P, to stimulate aged bone regeneration. This platform employs dual-mesoporous silica nanoparticles to highly load NAD⁺ and controllably load NAD⁺ biosynthesis activator, establishing a “supplementation-biosynthesis” strategy for rapid NAD⁺ replenishment and sustained NAD⁺ metabolic restoration in senescent bone marrow-derived mesenchymal stromal cells (BMSCs). Meanwhile, N/S@M enhanced cellular uptake efficiency by 41.0% through regulating the endocytic pathways of senescent BMSCs. After incorporation into an injectable hydrogel scaffold, N/S@M@P enabled sustained particle release over 14 days, supporting prolonged metabolic restoration. This strategy increased the NAD⁺/NADH ratio by 11.2-fold and increased ATP production by 3.25-fold in senescent BMSCs. In aged bone defects, N/S@M@P reduced the proportion of senescent BMSCs to 21.2% at day 7 and increased the bone volume fraction (BV/TV) by 119% at 4 weeks. These findings demonstrate that this nanoplatform can effectively restore the functions of senescent BMSCs, providing a promising therapeutic strategy for aging-related bone regeneration.

Graphical abstract

Current NAD⁺ restoration strategies remain limited by inefficient intracellular delivery, transient metabolic compensation, and impaired endogenous NAD⁺ biosynthesis in senescent cells, which restrict their efficacy in aged bone regeneration. To address these challenges, we developed an injectable dual-mesoporous silica nanoparticle-based nanocomposite hydrogel platform, N/S@M@P, for the co-delivery of NAD⁺ and the NAMPT activator SBI. This platform established a complementary “supplementation–biosynthesis” strategy by combining rapid exogenous NAD⁺ replenishment with SBI-mediated reinforcement of the endogenous NAD⁺ salvage pathway. In senescent BMSCs, N/S@M@P increased the NAD⁺/NADH ratio by 11.2-fold and enhanced ATP production by 3.25-fold. In aged bone defects, N/S@M@P reduced the proportion of p16⁺LepR⁺ cells among LepR⁺ cells to 21.2% at day 7 and increased BV/TV by 119% at 4 weeks, thereby promoting metabolic restoration and structural bone regeneration. These findings highlight NAD⁺ metabolic restoration as an effective strategy for improving the regenerative capacity of senescent BMSCs and provide a nanotechnology-mediated approach for aged bone regeneration.

Highlights

• •N/S@M@P co-delivers NAD⁺ and NAMPT activator to enable sustained supplementation-biosynthesis NAD⁺ restoration.
• •N/S@M@P restores NAD⁺/NADH homeostasis in senescent BMSCs, improveing mitochondrial function andparticle internalization.
• •N/S@M@P alleviates senescent BMSC dysfunction and restores osteogenic differentiation, thereby promoting aged bone regeneration.

1. Introduction

Aging triggers progressive deterioration of skeletal regenerative capacity, leading to osteoporosis and impaired fracture healing in the elderly, which have become a growing clinical challenge amid global demographic shifts [1,2]. There are significant distinctions between age-related osteoporosis and bone regeneration disorders in aging individuals in terms of pathological mechanisms and treatment goals. The former primarily arises from a chronic imbalance in bone remodeling homeostasis [3,4], whereas the latter is fundamentally driven by the functional decline of bone marrow-derived mesenchymal stromal cells (BMSCs) and systemic dysregulation of the local regenerative microenvironment, resulting in the failure to mount an effective regenerative response to skeletal injuries such as fractures or bone defects [5,6]. Consequently, therapeutic strategies for age-related bone regeneration disorders should extend beyond conventional anti-resorptive approaches and instead target the regulation of BMSC senescence and the restoration of regenerative potential[[7], [8], [9]].

Recent research advances indicate that the depletion of nicotinamide adenine dinucleotide (NAD⁺), a critical metabolic cofactor in senescent BMSCs, is the core link driving the above pathological process[[10], [11], [12]]. Its deficiency can lead to mitochondrial dysfunction, decreased activity of deacetylases such as Sirtuin 1 (SIRT1), and abnormal epigenetic reprogramming, which collectively impair the energy metabolism and signaling networks essential for osteogenic differentiation [13,14]. Therefore, restoring the metabolic homeostasis of NAD⁺ in senescent BMSCs has become a key therapeutic strategy for intervening in age-related bone regeneration disorders [15].

Current mainstream strategy for restoring NAD⁺ homeostasis mainly relies on the supplementation of NAD⁺ precursors, such as nicotinamide mononucleotide (NMN) and nicotinamide riboside (NR), which have shown certain efficacy in multiple aging-related disorders [16,17]. However, NAD⁺ precursors require intracellular enzymatic conversion before generating functional NAD⁺, and their efficacy is limited by the decline in the activity of rate-limiting enzyme nicotinamide phosphoribosyl transferase (NAMPT) in senescent cells [18,19]. NAMPT converts nicotinamide into NMN, which is subsequently converted into NAD⁺ by nicotinamide mononucleotide adenylyltransferases (NMNATs). Therefore, NAMPT is essential for maintaining endogenous NAD⁺ biosynthesis and also plays an important role in endochondral ossification [20,21]. Accordingly, age-related dysfunction of NAMPT limits the efficiency and persistence of the NAD⁺ precursor-based recovery process in senescent BMSCs.

Direct delivery of NAD⁺ represents a more straightforward strategy for rapidly replenishing intracellular NAD⁺ pools. However, free NAD⁺ is difficult to efficiently utilize because of its poor membrane permeability, susceptibility to extracellular degradation, and limited intracellular retention [22,23]. Moreover, aged bone regeneration is a prolonged process, during which NAD⁺ is continuously consumed by NAD⁺-dependent enzymes involved in mitochondrial metabolism, DNA damage repair, and stress responses [24,25]. Thus, direct NAD⁺ supplementation alone may provide only transient metabolic compensation if the endogenous NAD⁺ recycling machinery remains impaired. These limitations indicate that an ideal therapeutic strategy should not only deliver bioactive NAD⁺ into senescent BMSCs, but also reinforce endogenous NAD⁺ regeneration to sustain long-term NAD⁺ homeostasis. Recent studies have shown that nanocarrier-mediated NAD⁺ delivery can improve tissue repair by restoring mitochondrial function and regulating the regenerative microenvironment [22,26]. However, most existing NAD⁺ delivery systems mainly focus on single-cargo NAD⁺ supplementation and remain limited by insufficient loading capacity, delivery efficiency, and the difficulty of co-delivering NAD⁺ with hydrophobic synergistic regulators. This situation makes it difficult to meet the specific requirements of the aged bone regeneration process, especially the regenerative and differentiating functions of senescent BMSCs.

To overcome these, we attempt to develop a “supplementation-biosynthesis” strategy by integrating exogenous NAD⁺ delivery with SBI-797812 (SBI)-mediated NAMPT activation. Together, these two processes provide rapid metabolic compensation and reinforce endogenous NAD⁺ regeneration. Therefore, NAD⁺ supplementation and SBI-mediated NAMPT activation were integrated to achieve NAD⁺ restoration through complementary mechanisms, in which exogenous NAD⁺ provides rapid metabolic compensation, while SBI reinforces the salvage pathway to maintain sustained NAD⁺ regeneration in senescent BMSCs with impaired endogenous recycling capacity and persistent metabolic stress.

To implement this strategy, a sustained NAD⁺ supplementation-biosynthesis nanoplatform (N/S@M@P) was developed for aged bone regeneration (Scheme 1a). This system employed dual-mesoporous silica nanoparticles (DMSNs) to the co-deliver hydrophilic NAD⁺ and hydrophobic SBI. Notably, rather than being designed to match the molecular sizes of NAD⁺ and SBI, the unique dual-mesoporous structure of DMSNs was introduced to generate heterogeneous loading domains with differentiated microenvironments for cargos of distinct physicochemical properties. Meanwhile, an injectable amino-functionalized PEGylated poly(glycerol sebacate)-poly(γ-glutamic acid) hydrogel matrix (PEGSN-PGA, P) was used as a locally retained delivery reservoir. In this system, the hydrogel component P mainly enabled the controlled and sustained release of N/S@M, while the nanoparticle component M contributed to the stable loading, protection, and intracellular delivery of NAD⁺ and SBI. Thus, the combined delivery architecture was designed to achieve sustained metabolic intervention and prolonged functional recovery of senescent BMSCs.

Scheme 1.

Mechanism of the N/S@M@P nanocomposite hydrogel for restoring NAD⁺ metabolism and promoting regeneration of aged bone defects. a) Schematic illustration of the construction of the N/S@M@P nanocomposite hydrogel. Nicotinamide adenine dinucleotide (NAD⁺) and the nicotinamide phosphoribosyltransferase (NAMPT) activator SBI-797812 (SBI) were co-encapsulated within aminated dual-mesoporous silica particles (N-DMSNs, M) and integrated with an amino-functionalized PEGylated poly(glycerol sebacate)-poly(γ-glutamic acid) hydrogel matrix (PEGSN-PGA, P) to construct an injectable nanocomposite system N/S@M@P. b) The injectable hydrogel enabled in situ filling at aged bone defect sites with controlled release of therapeutic nanoparticles. c) Following the release and uptake of N/S@M by senescent BMSCs, N/S@M restored NAD⁺ metabolic homeostasis through a “supplementation-biosynthesis” strategy. Exogenous NAD⁺ supplementation provided rapid metabolic compensation to replenish the intracellular NAD⁺ pool, while SBI-mediated NAMPT activation reinforced the salvage pathway to maintain sustained NAD⁺ regeneration. This complementary metabolic intervention improved the NAD⁺/NADH redox balance, enhanced ATP production, and restored mitochondrial function. The recovered NAD⁺ metabolism further supported SIRT1-related downstream regulation, thereby attenuating DNA damage-associated stress, alleviating cell-cycle arrest, and reducing senescence-associated secretory phenotype (SASP)-related inflammatory responses. Ultimately, this local sustained delivery strategy improved the osteogenic function of senescent BMSCs and promoted aged bone regeneration.

After local administration into aged bone defects, N/S@M@P enabled in-situ filling of aged bone defect sites and sustained release of therapeutic nanoparticles (Scheme 1b). Following uptake by senescent BMSCs, exogenous NAD⁺ supplementation and SBI-mediated NAMPT activation jointly restored NAD⁺ metabolic homeostasis, as evidenced by an increased NAD⁺/NADH ratio, enhanced ATP production, and improved mitochondrial function (Scheme 1c). This metabolic restoration further activated NAD⁺-dependent downstream signaling, including SIRT1-related pathways, thereby promoting DNA damage repair, alleviating cell cycle arrest, and attenuating senescence-associated secretory phenotype (SASP)-related inflammatory responses. Ultimately, this hydrogel-based local sustained delivery strategy restored the osteogenic function of senescent BMSCs and promoted bone regeneration in aged bone defect models.

2. Results and discussion

2.1. Preparation and characterization of dual-mesoporous silica nanoparticles with co-loaded NAD⁺ and SBI (N/S@M)

According to the previously reported protocol, DMSNs were synthesized using two organic templates, polystyrene-b-poly(acrylic acid) (PS₁₀₀-b-PAA₁₇) and cetyltrimethylammonium bromide (CTAB) [27,28] (Figs. S1 and S2). Transmission electron microscopy (TEM) analysis revealed that DMSNs possessed a well-developed internal pore structure, offering abundant adsorption sites for effective drug loading (Fig. S3).

Firstly, to evaluate the basic adsorption performance of DMSNs towards NAD⁺, preliminary experiments were conducted. Different concentrations of NAD⁺ solutions were directly mixed with a fixed amount of DMSNs in deionized water. The results indicated that the loading efficiency of this direct-mixing method was relatively low, not exceeding 4% at the highest (Fig. S4). However, when the mass ratio of NAD⁺ to DMSNs was within the range of 1:2 to 3:4, the NAD⁺ loading capacity increased by 21.9 % and 16.8 % respectively compared to the 1:4 condition, suggesting that there was an optimal loading interaction within this ratio range, providing key parameter references for subsequent method optimization.

Given that NAD⁺ carries a negative charge, to enhance the loading capacity and loading efficiency, DMSNs were subjected to amino functionalization via modification with (3-aminopropyl) trimethoxysilane (APTMS), resulting in positively charged nanoparticles (N-DMSNs) for enhanced electrostatic adsorption (N-DMSNs) (Fig. S5–S7). Meanwhile, NAD⁺ and SBI loading was achieved using the co-precipitation method (Fig. 1a). Specifically, an appropriate volume of particle suspension was initially introduced into the good solvent for the drug (solvent 1), and ultrasonic stirring or similar techniques were employed to ensure adequate electrostatic adsorption. Subsequently, a large volume of poor solvent was rapidly introduced (solvent 2), inducing drug self-assembly within the particle pores and restricting it within these pores, thereby achieving efficient drug loading.

Fig. 1.

Preparation and characterization of co-loaded nanoparticles (N/S@M) and the modulation of uptake behavior in senescent BMSCs. a) Schematic illustration of the preparation steps for the dual-loaded nanoparticles (N/S@M) using the co-precipitation method. The N-DMSN (M) and the drug (NAD⁺ (N) or SBI (S)) were thoroughly mixed in an optimized ratio in a good solvent for the drug (Solvent 1). Subsequently, a defined volume of the corresponding poor solvent was added to induce co-precipitation of the drug and particles (Solvent 2). After centrifugation, washing and freeze-drying, the drug-loaded particles were finally obtained. b) Transmission electron microscope (TEM) image of M. Scale bar: 50 nm. c) Pore size distribution curves of M. d) Dynamic size distributions of M and N/S@M nanoparticles. e) Zeta potentials of M and N/S@M nanoparticles. f) NAD⁺ and SBI release profile from N/S@M in phosphate buffered saline (PBS, pH 7.4) at 37 °C. g) Uptake ratio of senescent to young BMSCs following a 4-h co-culture with FITC-labeled particles. h) Relative uptake in the presence of multiple endocytosis inhibitors by flow cytometry analysis, normalized to the corresponding inhibitor-free control in young or senescent BMSCs. i) Colocalization analysis of nanoparticles and uptake-related proteins. Scale bar: 20 μm. j) Oxygen consumption rate (OCR) in BMSCs with the treatment. k) Schematic illustration of N/S@M-mediated metabolic restoration, senescence alleviation and enhanced nanoparticle internalization in senescent BMSCs.

Compared with polyethyleneimine (PEI)-modification (P-DMSNs), APTMS-modified particles demonstrated superior loading performance (Fig. S8). This may be attributed to the longer chain length of PEI modification, which tended to block the entrances of particle surface pores upon the addition of a poor solvent like ethanol (EtOH), thereby restricting drug molecule entry. The prepared N-DMSNs, denoted as M, featured two distinct pore sizes of 2.7 nm and 10.7 nm and exhibited a thin surface shell layer structure (Fig. 1b and c). This spatial separation provided a structural basis for the differentiated loading of hydrophilic NAD⁺ and hydrophobic SBI.

Additionally, considering the substantial polarity difference between water and EtOH, treating EtOH as a poor solvent can achieve a higher NAD⁺ loading capacity than with methanol (MeOH) (Fig. S8). To investigate the influence of the ratio of deionized water (H₂O) (the good solvent, Solvent 1) to EtOH (the poor solvent, Solvent 2) in the co-precipitation method on the loading capacity and loading efficiency of NAD⁺, the system compared the loading effects under different volume ratios. The results showed that as the proportion of H₂O increased, the loading capacity of NAD⁺ was inhibited (Fig. S9). Therefore, the final volume ratio of H₂O to EtOH was set at 1:9. After the above systematic optimization, the loading capacity of NAD⁺ increased to 35.7 ± 2.2 wt%, and the loading efficiency reached 47.7 ± 2.9% (Fig. S10).

During the SBI loading process, dimethyl sulfoxide (DMSO) was chosen as the good solvent and H₂O as the poor solvent, with a ratio of 9:1. The correlation between the concentration of added SBI and the loading efficiency was investigated, providing a basis for subsequent precise control and in-depth analysis (Fig. S11).

Dynamic light scattering (DLS) results indicated that the average hydrodynamic size of blank particle M was approximately 289.3 nm, while that of dual-loaded particle N/S@M was 360.0 nm, showing a slight increase (Fig. 1d). Compared with M, the surface charge of N/S@M decreased from +28.2 ± 0.8 mV to +8.9 ± 1.8 mV, approaching a more neutral state (Fig. 1e). These results indicate the successful loading of NAD⁺ and SBI. In vitro release kinetics under physiological conditions (37 °C, pH 7.4) showed burst release profiles for both NAD⁺ and SBI within 12 h (Fig. 1f), indicating that the drugs could be promptly released from N/S@M to achieve rapid intervention.

2.2. Optimization of the proportion of the N/S@M and investigation of the cellular uptake into senescent BMSCs

Based on previous studies, an in vitro model of young and senescent BMSCs was established using conditioned medium to evaluate the biological properties of materials (Fig. S12) [16,29,30]. The elevated senescence-associated β-galactosidase (SA-β-gal), a significantly reduced proliferation ability, and an impaired osteogenic differentiation ability, along with significantly upregulated canonical senescence markers (p16, p21, p53) and decreased osteogenic indicators (Alpl, Runx2, Col1) at the molecular level (Fig. S13), demonstrated the successful establishment of the senescent BMSCs by treatment with aged-conditioned medium (A-CM). These A-CM-treated BMSCs were defined as the A-ctrl group, while BMSCs treated with young-conditioned medium (Y-CM) were defined as the Y-ctrl group.

The proliferation capacity and osteogenic differentiation potential of senescent BMSCs under the intervention of particles were evaluated to determine the effective working concentration of the dual-loaded particle N/S@M and the optimal ratio of the dual-loaded drugs. Under A-CM treatment, the senescent BMSC proliferation assays demonstrated that at a concentration of 300 μg mL⁻¹, particle M exhibited negligible inhibitory effects on BMSC proliferation, whereas N@M displayed a markedly enhanced pro-proliferative effect (Fig. S14a and b). At this concentration, the optimal loading capacity of SBI in S@M was determined to be approximately 2 wt% through screening (Fig. S14c). To further justify the optimal loading ratio of NAD⁺ and SBI, N/S@M particles with different SBI loading were prepared under the same particle concentration and fixed NAD⁺ loading capacity, and further confirmed that N/S@M with 2 wt% SBI loading showed the most favorable effects on senescent BMSC proliferation and osteogenic differentiation (Fig. S14d–f).

Based on the optimal loading ratio of NAD⁺ and SBI in N/S@M, the corresponding single-agent and dual-agent treatments were further evaluated in senescent BMSCs. It was found that at this selected loading ratio, the dual-loaded particle N/S@M significantly promoted the proliferation ability and osteogenic differentiation potential of senescent BMSCs (Fig. S15). Compared with the corresponding single-agent treatments, the combined intervention of NAD⁺ and SBI exhibited a more pronounced effect on restoring the proliferation ability and osteogenic differentiation potential of senescent BMSCs. In addition, comparison between the Free N/S group and the N/S@M group showed that the delivery function of the particles did not impair these beneficial effects and further improved them. In subsequent studies, the loading capacity of NAD⁺ and SBI in M were respectively set at approximately 36 wt% and 2 wt%.

Cellular senescence is associated with dysregulated endocytosis, which may compromise the efficacy of nanoparticle-based drug delivery [31,32]. This study systematically evaluated the uptake of different nanoparticle formulations, including blank particles (M), SBI-loaded particles (S@M), NAD⁺-loaded particles (N@M), and dual-loaded particles (N/S@M), in senescent and young BMSCs. Results demonstrated that relative to the uptake of the same particles by young BMSCs, senescent BMSCs internalized blank particle M at only 0.61 ± 0.01 of the level observed in young cells, whereas the uptake of dual-loaded N/S@M reached 0.86 ± 0.01, representing a 41.0% increase over M (Fig. 1g). Notably, this enhanced uptake could not be simply attributed to more favorable physicochemical properties of N/S@M. Since the dual-loaded particles exhibited a larger hydrated size and a less positively charged surface from the DLS and zeta potential analysis, neither of which would typically be expected to promote cellular internalization, the improved uptake in senescent BMSCs is more likely associated with cargo-induced biological regulation, particularly modulation of cellular metabolic activity, rather than with intrinsic particle physicochemical characteristics.

To investigate differences in nanoparticle uptake routes between senescent and young BMSCs, four inhibitors were used to interfere with distinct endocytic pathways: amiloride for macropinocytosis, methyl-β-cyclodextrin (MβCD) for cholesterol-dependent lipid raft-associated endocytosis, genistein for caveolae-related endocytosis, and chlorpromazine (CPZ) for clathrin-mediated endocytosis [33,34]. In young BMSCs, MβCD significantly inhibited the uptake of all particles, regardless of drug loading, indicating that cholesterol-dependent lipid raft-associated endocytosis is the predominant uptake route in young cells (Fig. 1h–S16). In contrast, senescent BMSCs exhibited distinct inhibitor-response patterns (Fig. 1h–S17). For blank particle M, uptake was mainly suppressed by genistein and CPZ, suggesting that caveolae-related and clathrin-mediated endocytosis are the principal pathways involved in its internalization. By comparison, uptake of the drug-loaded particles S@M, N@M, and N/S@M was predominantly inhibited by MβCD and genistein, indicating that drug loading shifted the preferred uptake route of senescent BMSCs toward cholesterol-dependent lipid-rafts-associated and caveolae-related endocytosis. Amiloride showed no evident inhibitory effect under the present conditions. Notably, uptake values in several inhibitor-treated groups exceeded 100%, which is likely due to compensatory upregulation of alternative routes. Therefore, assignment of the major uptake route was based primarily on the relative inhibitory pattern across different inhibitors, rather than on the absolute uptake percentage in any single treatment group [35]. Collectively, these results suggest that cellular senescence alters the preferred nanoparticle uptake routes of BMSCs, while cargo loading can further reshape the underlying endocytic pathways engaged in senescent cells.

To further support the inhibitor-based analysis of endocytic pathway preference, colocalization analysis was performed using ‌fluorescein isothiocyanate‌ (FITC)-labeled nanoparticles and uptake-related protein signals detected by immunofluorescence staining using Alexa Fluor 555 (AF555)-conjugated secondary antibodies. Colocalization was quantified as the fraction of FITC⁺ signals overlapping with AF555⁺ signals (AF555⁺FITC⁺/FITC⁺) (Fig. 1i). In young BMSCs, nanoparticles showed a relatively high degree of colocalization with flotillin-1, a marker associated with cholesterol-dependent lipid raft domains. This ratio decreased in senescent BMSCs treated with blank nanoparticles, but increased when incubated with drug-loaded nanoparticles. By contrast, colocalization with clathrin heavy chain (CHC) showed the opposite trend. These results were generally consistent with the inhibitor-response profiles, suggesting drug loading reshaped the uptake preference of senescent BMSCs. Specifically, drug loading reduced the CHC-associated uptake route characteristic of senescent cells and increased the involvement of a flotillin-1-related, cholesterol-dependent lipid raft pathway, making the uptake route more like that of young BMSCs.

Given the potential link between endocytic preference and cellular metabolism, the oxygen consumption rate (OCR) analysis was subsequently performed (Fig. 1j). The results showed that A-CM induction markedly impaired mitochondrial respiration in BMSCs, as evidenced by a decrease in OCR from 50.36 ± 2.42 in the Y-ctrl group to 27.56 ± 2.25 in the A-ctrl group, corresponding to a 45.3% reduction. Compared with the A-ctrl group, S@M and N@M treatment increased OCR levels to 51.19 ± 5.75 and 57.70 ± 3.07, corresponding to increases of 85.7% and 109.4%, respectively. Notably, N/S@M further elevated the OCR level to 70.70 ± 4.20, representing a 156.5% increase relative to A-ctrl and demonstrating a more pronounced effect on mitochondrial respiratory recovery. These results indicate that NAD⁺/SBI-based intervention effectively alleviated the bioenergetic impairment of senescent BMSCs, with the dual-delivery system producing the most pronounced recovery of mitochondrial respiration.

Consequently, these findings suggest that the enhanced internalization of drug-loaded nanoparticles in senescent BMSCs is not merely a consequence of particle formulation, but is also associated with payload-induced changes in cellular state. NAD⁺/SBI dual delivery partially alleviated senescence-associated bioenergetic dysfunction and was accompanied by a concurrent shift in endocytic preference away from the senescence-associated pathway toward a more youth-like pathway. This may establish a favorable feedback loop in which metabolic improvement is accompanied by a more permissive nanoparticle uptake, thereby further facilitating therapeutic intervention in senescent BMSCs (Fig. 1k). Although the present results do not establish a direct mechanistic link between metabolic recovery and endocytic remodeling, they support the concept that senescent cells retain a degree of functional plasticity that can be modulated by appropriate metabolic intervention. From a translational perspective, these observations further imply that improving nanotherapeutic delivery in senescent cells may require not only optimization of nanoparticle properties, but also consideration of the biological state of the cells themselves.

2.3. Preparation and characterization of injectable PEGSN-PGA (P) hydrogel loaded with N/S@M nanoparticles (N/S@M@P)

In the treatment of bone defects, there is an urgent need for a material that exhibits excellent biocompatibility and sufficient mechanical adaptability to effectively fill and support the damaged area. To address this, injectable hydrogels, as a minimally invasive application material, can serve as a supportive scaffold for bone regeneration and act as a multifunctional delivery matrix for the sustained delivery of therapeutic agents to the defect site [36]. To enable prolonged NAD⁺ metabolic intervention at senescent bone defects, this study integrated the nanoparticles N/S@M with amino-functionalized PEGylated poly (glycerol sebacate) (amino-functionalized PEGS, PEGSN) and poly (γ-glutamic acid) (PGA) to formulate an injectable hydrogel precursor, crosslinked with 1-(3-Dimethylaminopropyl)-3-ethylcarbodiimide (EDC) and N-Hydroxy succinimide (NHS) to construct the nanocomposite hydrogel scaffold N/S@M@P (Fig. 2a). The hydrogel substrate demonstrated excellent tissue adhesion and mechanical resilience while providing sustained local delivery of therapeutic particles at the defect site. The synthesis of PEGS and PEGSN was based on previous research work [37] (Fig. S18a and b). Compared with the ¹H NMR spectrum of PEGS, the 1H NMR spectrum of PEGSN showed new peaks from the amino group and neighboring methylene group protons at 8.34 (peak g) and 3.83 ppm (peak f), respectively (Fig. S18c and d). To ensure that the hydrogel simultaneously possesses excellent injectability and supporting performance, the molecular weight of the synthesized PEGSN was precisely controlled at approximately 18,000 (Fig. S19).

Fig. 2.

Preparation and characterization of the injectable hydrogel (N/S@M@P) for restoring NAD(H) homeostasis and mitochondrial function in senescent BMSCs. a) Synthesis route of injectable N/S@M@P involves the reaction of N/S@M, PEGSN, and PGA under the action of the crosslinking agents 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) and N-Hydroxy succinimide (NHS). b) Photographs of the injectable N/S@M@P showing shape adaptability to complex contours. The gelation time was approximately 1 min. Scale bar: 5 mm. c) Representative SEM images of N/S@M@P. Scale bar: 20 μm. d) Fluorescence image of FITC-labeled N/S@M (N/S@M-FITC) uniformly dispersed in P. Scale bar: 20 μm. e) Release profile of N/S@M-FITC particles from P obtained in PBS (pH 7.4) at 37 °C. f) Remaining weight profile of N/S@M@P in PBS (pH 7.4) at 37 °C and representative image of the hydrogel after 14 days of implantation in an aged bone defect. g, h) Intracellular NAD(H) levels and NAD⁺/NADH ratio in BMSCs after treatment. i) Quantification of intracellular ATP level in BMSCs after treatment. j) Representative MitoSOX Red and JC-1 fluorescence staining. Scale bar: 50 μm. k, l) Quantitative analysis of MitoSOX Red and JC-1 fluorescence staining.

The gelation time of the hydrogel was approximately 60 s, which was suitable for injection molding and could be applied to complex mold structures (Fig. 2b–S20). In all groups, the storage modulus (G′) of the hydrogel increased over time and eventually exceeded the loss modulus (G″), and stabilized after 3 min, indicating that all precursors formed stable hydrogels in the presence of the crosslinker (Fig. S21a). Throughout the entire frequency test range, the G′ values of the P and N/S@M@P were always higher than their corresponding G″ values, indicating that the cured hydrogels were stable viscoelastic solids (Fig. S21b). The hydrogel's porous architecture endowed it with mechanical resilience, demonstrating full elastic recovery post-compression (Fig. 2c–S22). Confocal 3D imaging verified homogeneous nanoparticle dispersion within the hydrogel matrix when loaded with FITC-labeled particles (Fig. 2d), achieving sustained particle and NAD⁺ release over 300 h (Fig. 2e–S23). This result indicated that the hydrogel can mediate the long-term delivery of therapeutic particles, thereby enabling sustained intervention. The degradation curve results showed that N/S@M@P degrades uniformly and slowly in the simulated environment over a period exceeding 14 days, which is consistent with the in vivo results (Fig. 2f). Together, these properties ensured persistent mechanical support at the defect site, critical for maintaining scaffold integrity during early osteogenic remodeling under load-bearing conditions.

The binding of the amino groups on the N/S@M surface to the carboxyl groups of the PGA chain significantly inhibited the sudden release of nanoparticles. The incorporation of N/S@M resulted in a reduced swelling ratio of the hydrogel (Fig. S24). The establishment of covalent bonds between amino-functionalized nanoparticles and carboxyl groups on PGA chains significantly enhanced interfacial adhesion, as demonstrated by a 1.9-fold elevation in lap-shear strength compared to the P group (Fig. S25). Moreover, this molecular interaction substantially improved mechanical robustness under compressive loads, with ultimate compressive strength and compressive modulus increasing by 1.1-fold and 4.1-fold, respectively (Fig. S26). These are pivotal for ensuring long-term mechanical stability in load-bearing bone defect microenvironments, where sustained structural integrity is critical to resisting physiological stresses during osteogenesis. Collectively, these results established N/S@M@P as a multifunctional platform, enabling efficient adhesion and mechanical support at bone defect sites while facilitating sustained and controllable NAD⁺ delivery.

The impaired regeneration of aged bone defects is closely associated with the accumulation of senescent BMSCs and the diminished capacity for proliferation and differentiation [38,39]. Based on osteogenic differentiation analysis with senescent BMSCs, the optimal incorporation concentration of nanoparticles in the hydrogel precursor solution was determined to be 3 mg mL⁻¹ (Fig. S27). Live/dead staining further confirmed the good biocompatibility of N/S@M@P toward both young and senescent BMSCs (Figs. S28 and S29).

2.4. N/S@M@P restored NAD(H) homeostasis and alleviated mitochondrial dysfunction in senescent BMSCs

NAD⁺ metabolism is closely associated with intracellular redox homeostasis, mitochondrial bioenergetics, and cellular aging, while NAD(H) cycling provides an important metabolic basis for ATP generation and energy conversion [[40], [41], [42]]. Based on this rationale, intracellular total NAD(H) levels and the NAD⁺/NADH ratio were measured as two complementary indicators to evaluate whether N/S@M@P could restore NAD(H) metabolic homeostasis in BMSCs cultured under A-CM (Fig. 2g and h). Compared with young BMSCs, senescent BMSCs exhibited a marked decline in intracellular NAD(H) levels and NAD⁺/NADH ratio, indicating disruption of intracellular redox homeostasis. Treatment with either S@M@P or N@M@P partially alleviated this metabolic imbalance, while the combined N/S@M@P treatment produced the most pronounced effect. Specifically, N/S@M@P increased intracellular NAD(H) levels to 1.28-fold of those in the A-ctrl group, and elevated the NAD⁺/NADH ratio from 0.37 ± 0.11 to 4.13 ± 0.46, corresponding to an 11.2-fold increase. Consistently, N/S@M@P treatment increased ATP levels from 2.04 ± 0.17 in the A-ctrl group to 6.63 ± 0.20, corresponding to a 3.25-fold increase and exceeding the level observed in young BMSCs (Fig. 2i). These results suggest that N/S@M@P effectively improved mitochondrial bioenergetic output after NAD(H) metabolic intervention in BMSCs, providing a metabolic basis for functional recovery.

Mitochondrial function was complementarily assessed to evaluate the recovery of senescent BMSC dysfunction. Fluorescence microscopy quantification revealed that mitochondrial oxidative stress was markedly elevated in senescent BMSCs, with the MitoSOX fluorescence intensity increasing from 12.99 ± 4.15 in the Y-ctrl group to 139.10 ± 14.85 in the A-ctrl group, corresponding to an approximately 10.7-fold increase (Fig. 2j and k). Notably, N/S@M@P treatment produced the most pronounced reduction in mitochondrial oxidative stress, decreasing MitoSOX fluorescence to 26.54 ± 6.91, corresponding to an 80.9% reduction compared with the A-ctrl group. This effect was stronger than that observed with either S@M@P or N@M@P alone, further supporting the advantage of the combined NAD⁺ supplementation and NAMPT activation strategy. Consistently, flow cytometry analysis independently confirmed the marked elevation of mitochondrial reactive oxygen species (ROS) in senescent BMSCs and its effective attenuation after treatment, with N/S@M@P showing the strongest inhibitory effect (Fig. S30a). In parallel, mitochondrial membrane potential (MMP) was evaluated using the JC-1 probe. The JC-1 aggregates/monomers ratio decreased from 2.51 ± 0.14 in young BMSCs to 0.62 ± 0.05 in senescent BMSCs, indicating a 75.4% reduction in MMP compared with the Y-ctrl group (Fig. 2j–l). Treatment with S@M@P or N@M@P increased the MMP by 1.77-fold and 1.87-fold, respectively, relative to the A-ctrl group, whereas N/S@M@P further increased the MMP to 1.95 ± 0.11, corresponding to a 3.16-fold increase over the A-ctrl group. Although the MMP was not fully restored to the young-cell level, N/S@M@P markedly rescued mitochondrial depolarization and achieved the most effective recovery among all treatment groups. Similar trends were further confirmed by flow cytometry analysis (Fig. S30b). Together, these findings demonstrate that N/S@M@P effectively alleviated mitochondrial oxidative stress and restored mitochondrial membrane potential in senescent BMSCs, further supporting its role in improving mitochondrial function through NAD(H) metabolic intervention.

2.5. N/S@M@P alleviates BMSC senescence and improves the proliferation and differentiation of senescent BMSCs in vitro

Since NAD⁺ metabolism is closely involved in cellular senescence and BMSC functional maintenance [24,43], we further investigated whether N/S@M@P-mediated metabolic restoration could alleviate BMSC senescence and improve the proliferative and osteogenic capacities of senescent BMSCs. As a rate-limiting enzyme in the NAD⁺ salvage pathway, NAMPT has been reported to suppress BMSC senescence through NAD⁺/SIRT1 signaling, whereas NAMPT inhibition reduces intracellular NAD⁺ content, NAD⁺/NADH ratio, and SIRT1 activity [20,25]. Consistently, exogenous NAD⁺ supplementation has been shown to delay BMSC senescence through SIRT1-related signaling, and knockdown of Sirt1 weakens the protective effects of NAD⁺ in senescent BMSCs [44]. In addition, SIRT1, as an NAD⁺-dependent deacetylase, is involved in DNA damage responses, p53-related senescence regulation, and epigenetic repression of senescence-associated secretory factors [45,46]. Therefore, this work examined whether N/S@M@P could modulate NAMPT-dependent NAD⁺ biosynthesis and SIRT1-mediated senescence regulation, thereby improving senescence-associated functional impairment in BMSCs.

Reverse transcription quantitative PCR (RT-qPCR) analysis showed that senescent BMSCs exhibited suppressed Nampt and Sirt1 expression compared with young BMSCs, accompanied by increased expression of p21 and p16 (Fig. 3a and b). SBI-containing treatment markedly upregulated Nampt, with S@M@P and N/S@M@P inducing 3.24-fold and 4.13-fold increases relative to the A-ctrl group, respectively. Meanwhile, both N@M@P and S@M@P increased Sirt1 expression and reduced p16 and p21 expression to varying degrees. Notably, N/S@M@P showed the strongest regulatory effect, increasing Sirt1 expression by 3.15-fold while decreasing p21 and p16 expression by 36.0% and 56.4%, respectively, compared with the A-ctrl group. These results suggest that N/S@M@P modulated NAMPT-dependent NAD⁺ salvage metabolism and SIRT1-mediated senescence regulation, thereby alleviating senescence-associated cell-cycle arrest in BMSCs.

Fig. 3.

N/S@M@P alleviates the senescence of BMSCs in vitro. a, b) Reverse transcription quantitative PCR (RT-qPCR) analysis of NAD⁺ metabolism-related genes (Nampt, Sirt1) and cellular senescence–associated genes (p21, p16) in young and senescent BMSCs with the treatment of hydrogels. c) Representative senescence-associated β-galactosidase (SA-β-gal) staining images in BMSCs with the treatment for 3 days. Scale bar: 200 μm. d) Representative immunofluorescence images of p53 and γH2A.X in BMSCs with the treatment. Scale bar: 50 μm. e, f) Quantitative analysis of p53 and γH2A.X immunofluorescence staining. g) Heat map representing the Rat Cytokine Antibody Array analysis in senescent BMSCs with the treatment. h) Cell-cycle analysis of G0/G1, S and G2/M phase in BMSCs with the treatment of hydrogels for 1 day after starvation.

Consistent with the transcriptional changes in senescence-associated regulators, senescent BMSCs showed markedly elevated p53 expression and increased SA-β-gal activity (Fig. 3c–e). After treatment with S@M@P or N@M@P, these senescence-associated signals were reduced, with the most pronounced effect observed in the N/S@M@P group. γH2A.X immunofluorescence was further performed to evaluate DNA damage-associated stress (Fig. 3d–f). Compared with the Y-ctrl group, senescent BMSCs exhibited increased nuclear γH2A.X-positive foci, which were substantially reduced after N/S@M@P treatment. A similar decreasing trend in γH2A.X protein level was further observed by western blotting (Fig. S31). These results indicate that restoration of NAD(H) homeostasis by N/S@M@P attenuated DNA damage-associated stress and thereby alleviated senescence phenotypes in BMSCs.

Furthermore, through the detection of cytokines in the supernatant of senescent BMSCs after material intervention, it was found that N/S@M@P downregulated multiple factors including VEGF, OPN, Fetuin A, Galectin-1, Cystatin C, IGFBP-3, OPG, CCL2 and CCL20 (Fig. 3g–S32). These factors have been widely reported to be involved in the aging-related or inflammatory regulatory networks. CCL2 and CCL20 are core chemokines that recruit immune cells and drive chronic inflammation [47]. IGFBP-3 is considered an important marker of senescence [48]. Fetuin-A is closely related to metabolic inflammation and is elevated in age-related diseases [49]. Cystatin C is an important marker reflecting renal function and associated with aging [50]. These data are interpreted as evidence that N/S@M@P modulates the inflammatory and senescence-associated secretory microenvironment of senescent BMSCs, rather than as complete suppression of all SASP factors.

Compared with young BMSCs, the cell cycle of senescent BMSCs undergoes significant changes, characterized by an increased proportion of cells in the G0/G1 phase and a decrease in the proportion of cells in the S phase (Fig. 3h–S33). The experimental results showed that N@M@P effectively alleviated the G0/G1 phase arrest in senescent BMSCs, while S@M@P modestly increased the proportion of cells in the S phase. In addition, N/S@M@P combined the advantages of two systems and showed a stronger effect, increasing the proportion of S-phase cells by 2.5-fold compared with the A-ctrl group and showing a further enhancement of 1.4- to 1.7-fold relative to the S@M@P and N@M@P groups. This suggested that modulating NAD(H) homeostasis may provide a feasible approach for alleviating senescence-associated proliferative arrest.

To evaluate whether the attenuation of senescence-associated phenotypes was accompanied by functional recovery, the proliferation and osteogenic differentiation of senescent BMSCs were examined at different time points (Fig. 4a). On days 1, 4, and 7, N/S@M@P significantly promoted the proliferation ability of senescent BMSCs (Fig. 4b). To further determine whether dual NAD⁺ supplementation and NAMPT-related regulation could restore the osteogenic potential of senescent BMSCs, osteogenic gene expression, alkaline phosphatase (ALP) analysis, and alizarin red S (ARS) staining were used for detection under osteogenic induction conditions. After 7 days of osteogenic induction, RT-qPCR analysis showed that the expression levels of Alpl, Runx2, and Col I were significantly increased in the N/S@M@P group compared with the A-ctrl group, by 2.2-, 1.4-, and 2.6-fold, respectively (Fig. 4c). Consistently, ALP and ARS staining showed that N@M@P and S@M@P partially restored the osteogenic function of senescent BMSCs, while N/S@M@P exhibited the strongest effect, increasing ALP and ARS levels by 1.7- and 2.5-fold, respectively (Fig. 4d–f). These findings suggest that dual regulation of NAD(H) homeostasis improves both the proliferative capacity and osteogenic potential of senescent BMSCs. Moreover, the enhanced effects observed in the N/S@M@P group further suggest that the dual-drug delivery strategy, integrating exogenous NAD⁺ supplementation with SBI-mediated NAMPT regulation, provides a more effective approach for restoring NAD(H) homeostasis and improving senescent BMSC function than either intervention alone.

Fig. 4.

N/S@M@P improved the proliferative and osteogenic function of senescent BMSCs in vitro. a) Proliferation and osteogenic differentiation assessment in senescent BMSCs at different time points with the treatment of material. b) Quantitative proliferation analysis of senescent BMSCs on days 1, 4, and 7 post-treatments with hydrogels. c) RT-qPCR analysis of Alpl, Runx2 and Col1 in senescent BMSCs after osteogenic induction for 7 days. d) Representative alkaline phosphatase (ALP) staining and alizarin red S (ARS) staining images in senescent BMSCs after osteogenic induction for 14 and 21 days, respectively. Scale bar: 200 μm. e, f) ALP activity and ARS staining quantification in senescent BMSCs after osteogenic induction for 14 and 21 days, respectively. g) Quantitative proliferation analysis of senescent BMSCs on day 3 post-treatment. h) Quantitative ALP activity of senescent BMSCs after 7 days of osteogenic induction. i) Representative ALP staining images of senescent BMSCs after 7 days of osteogenic induction. Scale bar: 200 μm.

To further verify the functional involvement of NAMPT- and SIRT1-related signaling in N/S@M@P-mediated BMSC restoration, pharmacological inhibition assays were performed using the NAMPT inhibitor FK866 and the SIRT1 inhibitor EX-527. Senescent BMSCs were assigned to the A-ctrl, N/S@M@P, N/S@M@P + FK866, and N/S@M@P + EX-527 groups. Cell proliferation was evaluated after 3 days of treatment, and osteogenic differentiation was assessed by ALP analysis after 7 days of osteogenic induction (Fig. 4g–i). As shown, N/S@M@P significantly promoted the proliferation and osteogenic differentiation of senescent BMSCs, whereas FK866 or EX-527 treatment substantially attenuated these effects. These results indicate that inhibition of either NAMPT-mediated NAD⁺ salvage regulation or SIRT1-related signaling partially compromised the restorative effects of N/S@M@P on senescent BMSC proliferation and osteogenic potential. Together with the observed recovery of NAD(H) homeostasis and mitochondrial function, these findings support the involvement of NAMPT-mediated NAD⁺ salvage regulation and SIRT1-related signaling in N/S@M@P-mediated functional restoration of senescent BMSCs.

To clarify the role of hydrogel component P in the system, the blank hydrogel (P), drug-loaded particles (N/S@M), and drug-loaded nanocomposite hydrogel (N/S@M@P) were co-cultured with senescent BMSCs. For comparison, the amount of hydrogel and the equivalent drug dose were matched across the relevant groups. Osteogenic differentiation of BMSCs was assessed at 7 days and 14 days. As shown in Fig. S34, both P and N/S@M improved the osteogenic differentiation of senescent BMSCs, while N/S@M@P exhibited the strongest pro-osteogenic effect. Notably, this advantage became more evident with prolonged culture time. By day 14, the osteogenic activity in the N/S@M@P group reached 1.8-fold that of the A-ctrl group, 1.3-fold that of the P group, and 1.2-fold that of the N/S@M group. These results, together with the release profile, suggest that P mainly contributes by providing a supportive local matrix and prolonging the presentation of N/S@M, thereby enhancing the long-term therapeutic effect.

The earlier comparison between Free N/S and N/S@M in Fig. S15 had already suggested that particle-mediated delivery preserved the beneficial effects of the drug combination and provided additional benefit, offering preliminary support for the functional contribution of M in the integrated system. To further analyze this contribution, the blank nanocomposite hydrogel (M@P), the hydrogel containing free NAD⁺ and SBI (Free N/S@P), and drug-loaded nanocomposite hydrogel (N/S@M@P) were co-cultured with senescent BMSCs. As demonstrated by the 3-day SA-β-gal staining results and their quantitative analysis (Fig. S35a and b), both free and particle-formulated NAD⁺/SBI reduced the proportion of SA-β-gal-positive cells, indicating that the two drugs could alleviate senescence-related phenotypes at the early stage regardless of the delivery format. However, in a longer-term assessment, the 14-day ALP analysis showed that Free N/S@P failed to produce a sustained improvement in osteogenic differentiation, whereas N/S@M@P remained markedly effective (Fig. S35a and c). These findings indicate that M is essential for stabilizing and retaining the payloads, thereby preserving drug availability and enabling prolonged biological activity during the recovery of senescent BMSCs.

Overall, the hydrogel component P mainly promoted the controlled slow release of N/S@M. In parallel, the nanoparticle component M contributed to the stability of NAD⁺ and SBI. The combined incorporation of P and M in N/S@M@P was therefore essential for promoting the prolonged functional recovery of senescent BMSCs.

To further examine whether the therapeutic effect of N/S@M@P was restricted to the aged-serum-induced senescent BMSC model, two additional senescence models were established using H₂O₂ treatment and UV irradiation (Figs. S36 and S37). Notably, N/S@M@P treatment reduced the proportion of SA-β-gal-positive cells and enhanced ALP staining in both H₂O₂- and UV-induced senescent BMSCs. These results further confirmed that the restorative effect of N/S@M@P was not limited to a single senescence model, supporting the robustness and broader applicability of this NAD⁺ metabolic intervention platform.

2.6. Integrated regulation of inflammatory signaling and redox homeostasis contributes to the functional restoration of senescent BMSCs by N/S@M@P

To thoroughly investigate the molecular mechanism underlying the N/S@M@P-mediated long-term functional repair of senescent BMSCs, RNA sequencing analysis was performed in this study. The experimental design was as follows: BMSCs cultured with Y-CM served as the young control group (Y group), BMSCs cultured with A-CM constituted the senescent control group (A group), and senescent BMSCs treated with N/S@M@P formed the experimental group (AT group). Principal component analysis (PCA) revealed that the gene expression profiles of the three groups exhibited significant inter-group variation, while intra-group consistency was observed among biological replicates within each group, thereby objectively reflecting the high reproducibility and reliability of the data (Fig. 5a). Venn diagram analysis identified 268 core differentially expressed genes (DEGs) shared between the Y vs A and AT vs A comparison groups, accounting for 55.6% of the total DEGs in the AT vs A group (Fig. 5b). These shared DEGs may represent key gene sets associated with the transition between young and senescent states and the therapeutic regulation induced by N/S@M@P. Notably, 108 DEGs were specifically present in both the Y vs A and AT vs A comparisons but absent in the AT vs Y comparison, suggesting that N/S@M@P treatment could restore part of the senescence-associated transcriptional changes toward a youthful expression pattern. In contrast, the 160 DEGs common to all three comparisons (Y vs A, AT vs A, and AT vs Y) represented potential gene clusters that were partially regulated by N/S@M@P but not fully restored to the young state.

Fig. 5.

Transcriptomic analysis reveals the potential mechanism underlying N/S@M@P-mediated functional recovery of senescent BMSCs. a) Principal component analysis (PCA) of gene expression profiles in young BMSCs (Y group), senescent BMSCs (A group), and senescent BMSCs treated with N/S@M@P (AT group). b) Venn diagram illustrating differentially expressed genes (DEGs) among the Y vs A, AT vs A, and AT vs Y comparisons. c) Heat map illustrating the expression values of significantly DEGs within the set (YvsA∩ATvsA)\ATvsY. |log2FC|>1, p < 0.05. d) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis and e) Gene Ontology (GO) enrichment analysis of senescent BMSCs following treatment with N/S@M@P. f, g) GSEA plots comparing BMSCs in the AT versus A comparison in immune response and positive regulation of cytokine production. h) Heat map representing the DEGs expression values for AT versus A in pathways associated with immune response regulation, oxidative stress homeostasis, and NADPH metabolic processes.

Heatmap visualization analysis of these gene clusters further showed that N/S@M@P mainly suppressed the aberrant upregulation of senescence-associated genes in A-CM-induced senescent BMSCs (Fig. 5c–S38). Further Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses of DEGs in the AT vs A comparison revealed that the regulated genes were significantly associated with inflammatory and immune-related pathways (Fig. 5d and e). KEGG analysis showed enrichment in the chemokine signaling pathway and cytokine-cytokine receptor interaction, while GO analysis indicated that DEGs were mainly involved in immune response regulation, oxidative stress homeostasis, and NADPH metabolic processes. These biological processes were closely associated with SASP, oxidative damage, and immune dysregulation during cellular senescence.

The results of the gene set enrichment analysis (GSEA) indicated that inflammatory and oxidative stress-related pathways were significantly suppressed in the AT group compared with the A group (Fig. 5f and g). Specifically, N/S@M@P treatment downregulated genes associated with immune response (Ccl6, Pf4, Prg4, Nrros, Tlr2, RT1-DMa, Ccl12, Sash3, Tnfsf13, and Irf8) and cytokine production (Ccdc88b, Slc11a1, Nfam1, Clec5a, and Scimp), as well as genes involved in NADPH oxidase activation (Cybb, Ncf1, Ncf2, and Ncf4) and superoxide anion generation (Itgam, Clec7a, and Syk) (Fig. 5h). These findings suggested that N/S@M@P not only improves NAD⁺-related metabolic homeostasis but also attenuates inflammatory signaling and oxidative stress at the transcriptional level.

In the field of aging biology, metabolic dysregulation, chronic low-grade inflammation, and immune dysfunction are closely interconnected during cellular senescence [51,52]. Cellular senescence and SASP can amplify inflammatory signaling and impair tissue homeostasis, whereas persistent inflammation can further aggravate senescence-related functional decline. In parallel, NAD⁺ metabolism is closely associated with stromal cell function, oxidative stress regulation, and inflammatory responses. Consistently, RNA sequencing analysis showed that N/S@M@P partially reshaped the senescence-associated transcriptional profile of BMSCs, with major changes involving NAD⁺-related metabolic homeostasis, NADPH oxidase-associated oxidative stress, and inflammatory/immune regulatory programs. These findings provide transcriptomic evidence supporting the involvement of metabolism-associated redox and inflammatory regulation in N/S@M@P-mediated restoration of senescent BMSC function.

2.7. N/S@M@P restores and promotes bone regeneration in aged mice in vivo

The integration of NAD⁺ metabolism with cellular senescence represents a promising yet underexplored frontier for aging interventions [[53], [54], [55]], while systemic NAD⁺ supplementation has shown efficacy in age-related pathologies [23,56,57]. To investigate the therapeutic efficacy of N/S@M@P in aged bone regeneration, this study established a femoral defect model in 20-month-old male C57BL/6 mice based on the previous work [29,58]. A standardized 1.2 mm diameter defect was surgically induced in the distal femur, followed by implantation of hydrogels with distinct compositions: M@P, S@M@P, N@M@P, N/S@M@P, or vehicle control. The effect of delivering NAD⁺ and SBI was evaluated by detecting the NAD⁺/NADH ratio of the defect site. Meanwhile, the recruitment of BMSCs and their senescence status were assessed using tissue immunofluorescence analysis. Bone regeneration dynamics were longitudinally monitored at 4- and 8-weeks post-surgery through quantitative micro-computed tomography (micro-CT) and histomorphometric analysis (Fig. 6a).

Fig. 6.

In vivo retention and senescence regulation of N/S@M@P in aged bone defects. a) Timeline of the in vivo experiment. b) In vivo fluorescence imaging of Cy7-labeled M@P implanted at the bone defect site. Ctrl: aged mice before surgery. c) Quantitative analysis of the relative radiant efficiency in the region of interest. d) The relative NAD⁺/NADH ratio of aged bone defects on day 7 and 14 after implantation. Ctrl: bone defect without implantation. e) Representative immunofluorescence images of LepR and p16 staining in femoral defect regions on day 7. Green fluorescence: LepR-positive (LepR⁺) cells; red fluorescence: p16-positive (p16⁺) cells. Scale bar: 100 μm. f) Quantitative analysis of LepR⁺ area. g) Quantitative analysis of the proportion of p16⁺LepR⁺ cells among LepR⁺ cells of aged bone defects.

To evaluate the biosafety of the N/S@M@P platform, this study performed both early-stage serum biochemical analysis and long-term histopathological assessment after implantation. At 7 days post-implantation, serum biochemical analysis showed no obvious abnormalities in liver- and kidney-function-related indicators, including blood urea nitrogen (BUN), creatinine (CREA), uric acid (UA), alkaline phosphatase (ALP), alanine aminotransferase (ALT), and aspartate aminotransferase (AST), in the M@P and N/S@M@P groups, with all values remaining within the reference ranges (Fig. S39). In addition, RT-qPCR analysis of bone defect and peri-defect tissues showed that M@P and N/S@M@P implantation did not induce excessive local inflammatory activation at the early stage after surgery (Fig. S40). At 4 weeks post-implantation, hematoxylin and eosin (H&E) staining of major organs, including the heart, liver, spleen, lungs, and kidneys, revealed no evident tissue damage, inflammatory cell infiltration, or pathological alterations in the N/S@M@P group relative to the Ctrl groups (Fig. S41).

To assess the in vivo retention and release behavior of the hydrogel-based system, Cy7-labeled M@P was implanted at the aged bone defect site and monitored by fluorescence imaging. The Cy7 signal was mainly localized at the implantation site and gradually decreased over time, while region-of-interest quantification showed that detectable fluorescence was maintained for up to 14 days (Fig. 6b and c). These results indicated that the hydrogel matrix enabled sustained local retention and slow release of nanoparticles at the aged bone defect site. Moreover, ex vivo biodistribution imaging at 7 days after subcutaneous hydrogel injection showed no obvious Cy7 fluorescence accumulation in major organs, suggesting no evident systemic retention of released nanoparticles during the observation period (Fig. S42).

Based on this sustained local retention behavior, the NAD⁺/NADH ratio at aged bone defect sites was further examined on days 7 and 14 post-implantation (Fig. 6d). On day 7, implantation of S@M@P and N@M@P increased the NAD⁺/NADH ratio to 3.73 ± 1.10-fold and 5.42 ± 0.97-fold of that in the Ctrl group, respectively. Notably, the N/S@M@P group showed the most pronounced improvement, with the NAD⁺/NADH ratio reaching 10.38 ± 2.26-fold of that in the Ctrl group (Fig. 6d). On day 14, although the intergroup differences became smaller along with material release, degradation, and new bone formation, the N/S@M@P group still maintained a higher NAD⁺/NADH ratio than the other groups, indicating a sustained metabolic regulatory effect at the defect site. Immunofluorescence analysis of bone tissue on day 7 post-implantation showed that NAD⁺ and SBI delivery increased the LepR-positive (LepR⁺) area at aged bone defect sites (Fig. 6e). Among all groups, the dual-loaded hydrogel N/S@M@P exhibited the strongest regulatory effect, as evidenced by a 63.3% increase in LepR⁺ area and a reduction in the proportion of senescent LepR⁺ cells, defined as p16⁺LepR⁺/LepR⁺ cells, to 21.2% compared with the Ctrl group (Fig. 6f and g). These results suggested that effective local NAD⁺ delivery and metabolic intervention promoted LepR⁺ cell recruitment and alleviated cellular senescence at aged bone defect sites.

High-resolution micro-CT scanning was performed to observe and quantify the regenerated bone at four weeks post-injection (Fig. 7a). Among them, bone regeneration was significantly better in the subgroups with added S@M and N@M than in the unfilled control group. According to morphometric analysis, the N/S@M@P group showed the highest bone volume fraction (BV/TV, %; 49.6% ± 9.3%), approximately 2.19-fold that of the control group (22.6% ± 2.2%) (Fig. 7b). The N/S@M@P group also showed the highest trabecular number (Tb. N, mm⁻¹; 4.00 ± 0.47 mm⁻¹) and the highest bone mineral density (BMD, g cm⁻³; 1.30 g cm⁻³ ± 0.02 g cm⁻³) (Fig. 7c and d). Histological analysis using H&E and Masson's trichrome staining at the site of the defect further supported the micro-CT findings. The N/S@M@P group exhibited a greater amount of trabecular bone (TB) and a denser new bone (NB) structure (Fig. 7e). To further evaluate the tissue-level mechanism, immunohistochemical staining of SIRT1 and osteocalcin (OCN) was performed in femoral defect samples (Fig. 7f–h). Compared with the Ctrl group, N/S@M@P significantly increased the expression of SIRT1 and OCN by 2.51 ± 0.26-fold and 3.06 ± 0.60-fold, respectively. These results indicated that N/S@M@P not only promoted structural bone repair, but also enhanced NAD⁺-related metabolic regulation and osteogenic activity at the tissue level.

Fig. 7.

N/S@M@P promotes structural and functional regeneration of aged bone defects in vivo. a) Representative 3D reconstructed micro-CT images at 4 weeks after implantation. The green circles indicate the defect sites. Scale bar: 1 mm. b-d) Quantification of bone volume fraction (BV/TV; %), trabecular number (Tb. N, mm⁻¹) and bone mineral density (BMD; g cm⁻³) at the defect sites derived from micro-CT at 4 weeks. e) Representative hematoxylin and eosin (H&E) and Masson's trichrome staining images of the defect in the femur at 4 weeks. TB: trabecular bone; NB: new bone. Scale bar: 100 μm. f) Immunohistochemistry images of SIRT1 and osteocalcin (OCN) in femoral defect samples. Brown staining: SIRT1-or OCN-positive signals; Blue staining: cell nucleus. Scale bar: 100 μm. g, h) Quantitative analysis of SIRT1 and OCN expression in femoral defect samples.

To further evaluate the long-term bone repair, micro-CT and histological analyses were performed at 8 weeks after injection. As shown by micro-CT images, eight weeks after injection, the N/S@M@P group was near-completely repaired, while the other groups remained in the active bone regeneration stage (Fig. S43a). Quantitative analysis further confirmed the superior bone regeneration in the N/S@M@P group, as evidenced by significantly increased BV/TV and BMD compared with the Ctrl group at 8 weeks (Fig. S43b). Moreover, according to the staining results of the tissue sections, it was observed that a distinct Haversian system (HS) emerged in the N/S@M@P group (Fig. S43c). These results indicated advanced maturation and remodeling of the regenerated bone tissue by 8 weeks post-injection.

To further distinguish the contribution of the nano-delivery system from that of the drug combination itself, an equivalent free NAD⁺ and free SBI hydrogel group (Free N/S@P) was added for comparison. At 4 weeks after implantation, micro-CT analysis showed that N/S@M@P induced more pronounced bone regeneration than Free N/S@P, indicating that the nanoparticle-mediated co-delivery system further improved therapeutic efficacy by enhancing drug stability, sustained release, and intracellular delivery (Fig. S44).

3. Conclusions

By integration of dual-mesoporous silica nanoparticles and an injectable hydrogel matrix, N/S@M@P, this study developed a NAD⁺ supplementation-biosynthesis nanoplatform to provide a metabolic intervention strategy for aged bone regeneration. By co-delivering NAD⁺ and the NAMPT activator SBI, this platform could realize a sustained NAD⁺ metabolic restoration, local particle retention, and sustained release in senescent BMSCs and aged bone defect sites.

Functionally, N/S@M@P restored NAD⁺/NADH homeostasis, improved mitochondrial bioenergetic function, delayed cellular senescence, and promoted the functional recovery of BMSCs. At the tissue level, this strategy sustained an increased NAD⁺/NADH ratio at the defect site and reduced the proportion of p16⁺LepR⁺ cells within the LepR⁺ cell population ultimately contributing to structural and functional repair of aged bone defects. These findings demonstrate that NAD⁺ metabolic restoration is an effective strategy for improving the regenerative capacity of senescent BMSCs and suggest that nanotechnology-mediated metabolic intervention provides an effective strategy for aging-related bone regeneration.

Future studies should focus on optimizing the targeting ability and release behavior of the platform, especially through the incorporation of ligands that specifically recognize senescent cells or osteogenic progenitors. Together with long-term biocompatibility evaluation, scalable fabrication, and standardized quality control, these efforts will be important for advancing this strategy toward clinical translation. Overall, this work bridges aging biology, nanotechnology, and regenerative medicine, and provides a promising framework for developing metabolism-based therapies against age-related tissue regeneration disorders.

4. Methods

4.1. Preparation of amino-functionalized dual-mesoporous silica nanoparticles (N-DMSNs)

In beaker A, 50 mg of PS₁₀₀-b-PAA₁₇ was dissolved in 10 mL of tetrahydrofuran. In beaker B, 200 mg of cetyltrimethylammonium bromide (CTAB) was dissolved in 40 mL of deionized water, followed by the addition of 0.5 mL of ammonia water. In beaker C, 325 μL of tetraethyl orthosilicate (TEOS) was dispersed in 80 mL of ethanol. The contents of beaker A were rapidly poured into beaker B under sustained stirring at 700 rpm and balanced for 15 s, after which the mixture from beaker C was subsequently added. The mixture was kept at 35 °C for 18 h. The product was centrifuged, washed, dried, and calcined in a nitrogen atmosphere at 550 °C to obtain DMSNs. To obtain amino-functionalized DMSNs (N-DMSNs), DMSNs were dispersed in toluene with (3-aminopropyl)trimethoxysilane (APTMS) at a ratio of 10 mg: 75 μL, and refluxed at 60 °C under inert nitrogen for 24 h to obtain N-DMSNs.

4.2. Immobilization of NAD⁺ and SBI on N-DMSNs (N/S@M)

N-DMSNs were dispersed at a concentration of 10 mg mL⁻¹ into NAD ⁺ aqueous solution (7.5 mg mL⁻¹). The mixture was sonicated for 30 min to ensure uniform dispersion, followed by stirring for 4 h to allow sufficient interaction between NAD⁺ and N-DMSNs. Then, ethanol with a volume nine times that of the deionized water was rapidly added to the mixture. After stirring for 12 h, the mixture was centrifuged at 13,000 rpm to collect the precipitate and the first supernatant. The precipitate was washed with a quantitative amount of deionized water and then centrifuged again at 13,000 rpm to collect the precipitate and the second supernatant. Finally, the precipitate was freeze-dried to obtain N-DMSNs loaded with NAD⁺ (N@M). The loading capacity and efficiency of NAD⁺ was quantitatively determined by measuring the difference between the initial amount of NAD⁺ input and the residual NAD⁺ content in the two supernatants using an ultraviolet-visible (UV-Vis) spectrophotometer (Molecular Devices, SpectraMax M3).

The SBI loading procedure is analogous to that of NAD⁺, with the following specific adjustments: the aqueous solution of NAD⁺ is replaced with the DMSO solution of SBI, 9-fold the volume of ethanol is substituted with deionized water, and the washing solution is changed to a DMSO/deionized water mixture (v/v = 1:9). Quantitative determination is performed using high-performance liquid chromatography (HPLC, Agilent, Agilent 1260). All other operational steps are consistent with those in the NAD⁺ loading process. Thus, N-DMSNs loaded with SBI (S@M) as well as N-DMSNs co-loaded with SBI and NAD⁺ (N/S@M) were successfully prepared.

The loading capacity (LC, wt%) and loading efficiency (LE, %) of NAD⁺ or SBI were calculated according to the following equations:

$$m_{loaded}=m_{initial}-m_{free}$$

$$LC\left(wt\%\right)=\frac{m_{loaded}}{m_M}\times 100\%$$

$$LE(\%)=\frac{m_{loaded}}{m_{initial}}\times 100\%$$

where $m_{initial}$ is the initially added mass of NAD⁺ or SBI, $m_{free}$ is the mass of unloaded NAD⁺ or SBI in the supernatant and washing solutions, $m_{loaded}$ is the mass of NAD⁺ or SBI loaded into M, and $m_M$ is the mass of blank M used for loading.

4.3. Preparation and characterization of injectable N/S@M@P hydrogel

PEGSN was prepared as previously described [37], which is with a molecular weight of approximately 1.8 × 10⁴ Da. PEGSN (200 mg) was dissolved and thoroughly mixed with PGA (100 mg, M_w = 5.0 × 10⁵ Da) in a 1 mL N/S@M solution. Before injection, 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC·HCl, 3.5 mol L⁻¹, 50 μL) and N-hydroxy succinimide (NHS, 3.5 mol L⁻¹, 50 μL) was added uniformly to the mixture, resulting in the formation of the N/S@M@P hydrogel. The gelation kinetics and stability of the hydrogel were measured by the HAAKE MARS III rotational rheometer (P20 TiLS, 20 mm in diameter). The ultimate compressive stress-strain tests and lap shear tests of the hydrogel were performed using an electronic universal testing machine (SANS, CMT250) equipped with a 20 N load cell to systematically assess its compressive strength, strain at fracture, and adhesion performance on bone tissue. For the lap shear test, 300 μL of the sample was injected into the adhesion area, with the adhesion area fixed at 25 mm in width and 10 mm in length.

4.4. Establishment of the senescent BMSCs model

Primary BMSCs were isolated and purified from the femurs of 1-month-old male Sprague Dawley (SD) rats. Conditioned media (CM) were formulated by supplementing α-MEM basal medium with 15% serum obtained from either 1-month-old (Young-serum, Y-CM) or 24-month-old (Aged-serum, A-CM) male SD rats, as previously described [16,29,30]. To establish comparative models of young and senescent BMSCs, primary BMSCs were expanded through three passages in complete growth medium (α-MEM containing 10% fetal bovine serum and 1% penicillin-streptomycin), followed by 48-h induction with respective CM formulations. The induced cells were used for subsequent experimental analysis.

4.5. Analysis of cellular uptake

BMSCs were seeded in 6-well culture plates at a density of 2 × 10⁴ cells mL⁻¹ and induced with CM (2 mL well⁻¹) for 2 days. To investigate the similarities and differences in the uptake mechanisms between senescent and young BMSCs, the induced BMSCs were preincubated with various inhibitors: amiloride (10 μg mL⁻¹) to inhibit micropinocytosis, MβCD (5 mg mL⁻¹) to inhibit cholesterol-dependent lipid raft-associated endocytosis, genistein (50 μg mL⁻¹) to inhibit caveolae-related endocytosis, and CPZ (5 μg mL⁻¹) to inhibit clathrin-mediated endocytosis. Following a 2-h preincubation with these inhibitors, FITC-labeled particles (300 μg mL⁻¹ in M equivalents) were added. After an additional 4-h incubation, the BMSCs were harvested, and fluorescence intensities were quantified using flow cytometry (Beckman, CytoFLEX s).

4.6. Detection of NAD⁺/NADH ratio

BMSCs were seeded into 6-well culture plates at a density of 2 × 10⁴ cells mL⁻¹ and cultured with CM for 2 days. Thereafter, the cells were treated with hydrogels for 3 days and lysed for NAD⁺/NADH analysis. The intracellular NAD(H) level and the NAD⁺/NADH ratio were measured using an Amplite Fluorimetric NAD/NADH Ratio Assay Kit (AAT Bioquest, 15263) according to the manufacturer's instructions. For in vivo studies, bone tissues encompassing the defect sites and adjacent peri-defect regions was harvested on day 7 and 14 following defect induction and material implantation. The tissue samples were homogenized thoroughly at low temperature, followed by addition of lysis buffer at the recommended proportion. The supernatant obtained after centrifugation of the lysis products was used for the determination of the NAD⁺/NADH ratio.

4.7. Evaluation of mitochondrial function

Following 2 days of induction with CM and 3 days of co-culture with hydrogels, the mitochondrial functional recovery effect of the hydrogels on senescent BMSCs was assessed and compared to that in young BMSCs. ATP concentration was quantified using a commercial assay kit (Beyotime, S0027), following the manufacturer's protocol. Luminescence intensity was measured using a SpectraMax M3 multimode microplate reader (Molecular Devices). Mitochondrial membrane potential and mitochondrial superoxide levels in BMSCs post-treatment were evaluated using JC-1 (Beyotime, C2003) and MitoSO™ Red (Beyotime, S0061), respectively. The results were analyzed by high-resolution slide scanner (Evident, APX100) and flow cytometry (Beckman, CytoFLEX s).

4.8. RNA sequencing

To unravel the mechanism of the effect of N/S@M@P hydrogel on senescent BMSCs, senescent BMSCs treated with N/S@M@P for 3 days were collected. Young BMSCs and untreated senescent BMSCs served as control groups. Total RNA was extracted using RNAiso Plus (TaKaRa, 9108). Subsequently, RNA purification, library preparation, and sequencing were conducted by Hangzhou LC-Bio Technology Co., Ltd. Bioinformatic analysis was performed using the OmicStudio tools at https://www.omicstudio.cn/tool. All genes with |log fold change| > 1 and a p-value of <0.05 were selected as DEGs and analyzed through Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation and Gene Ontology (GO) term enrichment analysis.

4.9. In vivo therapeutic efficacy for aged bone regeneration

Composite hydrogels were implanted into distal femoral defects (1.2 mm diameter) in 20-month-old male C57BL/6 mice, and the bone repair efficacy was assessed in vivo. At 4- and 8-week postoperative intervals, femurs were collected and osteogenesis at defect sites was assessed through micro-CT and histological analysis.

4.10. Statistical analysis and reproducibility

All data are expressed as mean ± standard deviation. Unless otherwise specified, all experiments were independently repeated at least three times. Statistical analyses were executed using GraphPad Prism 10 (v10.0.3). Group comparisons were evaluated using unpaired, two-tailed Student's t-tests, one-way ANOVA followed by Tukey's multiple comparison test, or two-way ANOVA followed by Tukey's multiple comparison test for experiments involving two independent variables, as indicated in the figure legends. Statistical significance was defined as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, with ns denoting non-significant differences. No animals or data were excluded from the analyses. No randomization method was used. The investigators were not blinded to allocation during experiments and outcome assessments, but the experiments were performed in appropriate biological replication by independent personnel to avoid bias.

CRediT authorship contribution statement

Fangru Xie: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis, Data curation, Conceptualization. Zirui He: Writing – review & editing, Methodology, Formal analysis, Data curation. Shiyu Xu: Writing – review & editing, Methodology, Formal analysis, Data curation. Xiaoqiao Bai: Validation, Formal analysis, Data curation. Xuan Wang: Validation, Formal analysis, Data curation. Fan Zhang: Validation, Formal analysis, Data curation. Yuan Yuan: Writing – review & editing, Resources, Project administration, Funding acquisition, Conceptualization. Changsheng Liu: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition. Dechao Niu: Writing – review & editing, Writing – original draft, Resources, Project administration, Funding acquisition, Conceptualization.

Ethics approval and consent to participate

All procedures were conducted in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All experimental protocols were approved by East China University of Science and Technology (Approval number: Ecust-2023-043).

Declaration of competing interest

The authors declare no competing interests.

Appendix A. Supplementary data

The following is the Supplementary data to this article: