Cells perceive the distribution differences of adhesion ligands and consequently establish cellular polarization to regulate osteogenic differentiation Authors: Jinsheng Li, Ming Yang, Guowen Duan, Jianhao Huang, Tao Shen, Jie Cui, Keke Huang, Qiting Ge, Liming Yang, Jingyi Sheng, Jie Weng, Tailin Guo, Zhihong Xu, Jizi Liu, Ning Gu, Qing Jiang Published: 2026 ABSTRACT AbstractEngineering implant topography has emerged as a promising strategy to promote bone regeneration in complex fractures and large bone defects. Among such topographies, TiO2 nanotubes serve as a model topography and have been shown, within a certain range, to positively regulate osteogenic differentiation through diameter-dependent effects. However, the underlying mechanisms remains fragmentary. Here, we revealed how cells perceived nanotube interfaces and identified the intracellular force-based mechanotransduction that arose from interface perception. We unexpectedly found that cellular perception of nanotube interfaces depended on diameter-associated topographical cues that induced differential distribution of adhesive ligands. Cells engaged these adhesive ligands to modulate focal adhesion (FA) organization, with small and many FAs forming on the small-diameter nanotubes (30 nm), whereas fewer but larger FAs formed on the large-diameter nanotubes (100 nm). Fewer but larger FAs regulated cytoskeletal assembly, generating greater intracellular force and enhancing cellular polarization. Furthermore, large-diameter nanotubes promoted nuclear pore deformation and YAP nuclear translocation, leading to enhanced osteogenic differentiation both in vitro and in vivo. Together, our findings suggest that nanotube diameter-dependent geometry regulates the spatial presentation of adhesive ligands and subsequently influences FA maturation, cellular polarization, and YAP-associated mechanotransduction. This study provides mechanistic insight into how implant nanotopography modulates osteogenic responses and offers a theoretical basis for the rational design of osteogenic implant surfaces.Keywords: Nanotube-based topography, Adhesive ligand distribution, Focal adhesion maturation, Cellular polarization, Osteogenic mechanotransduction Introduction As global life expectancy continues to rise and populations age, the prevalence of musculoskeletal disorders—including fractures, bone metastases and osteoporosis—has increased substantially, leading to growing demands and costs for bone-related medical treatments. Currently, surgical interventions for fractures and joint replacements primarily rely on titanium (Ti) and Ti alloy implants [1]. However, Ti and Ti alloy implants still face clinical challenges, including prolonged osseointegration times and the risk of loosening, which may ultimately lead to implant failure. Therefore, there is an urgent need to develop bioactive Ti implants capable of achieving faster and more controlled osseointegration. As osseointegration occurs at the bone–implant interface, surface topography modification has emerged as an effective strategy, among which TiO2 nanotubes fabricated by anodic oxidation are widely used [2]. The nanotubes are vertically aligned and orderly distributed on the material surface. Previous studies have suggested that TiO2 nanotubes mimic key nanoscale features of the bone extracellular matrix, including dimensions comparable to collagen fibrils [[3], [4], [5]] and hydroxyapatite nanocrystals [6], thereby providing biomimetic topographical cues for cell–material interactions. This topography significantly enhances the osteogenic differentiation of precursor cell (such as MC3T3-E1 cells [7], bone marrow-derived mesenchymal stem cells [8,9]and calvarial osteoblasts [10]), thereby promoting tissue integration. For translational medicine research, nanotubes are commonly produced in simple aqueous electrolyte systems, yielding diameters in the 20–120 nm range [[11], [12], [13]]. Interestingly, within this range, a widely accepted view is that the osteogenic capacity of TiO2 nanotubes increases with increasing tube diameter [14,15]. Several studies have attempted to explain this positive correlation between nanotube diameter and osteogenic differentiation. Taken together with previous studies and our findings, the expression of multiple factors, including XB130 [16], P2Y6 [17], POSTN [18] and RNA RMRP [19], was dependent on nanotube diameter, leading to activation of downstream signal pathways that modulate osteogenesis. However, it is evident that these studies remain largely phenomenological, as many of these factors have primarily been identified by RNA sequencing. However, despite extensive studies on diameter-dependent osteogenic responses, the interfacial mechanisms by which cells perceive nanotopographies of different diameters and translate these cues into distinct biological responses remain incompletely understood. Upon implantation, biomaterials become rapidly coated with proteins, particularly adhesion-related extracellular matrix (ECM) components, which mediate integrin-dependent cell adhesion. Cell adhesion is fundamental to downstream cellular processes, including proliferation, migration, and differentiation [20,21]. Because the interaction between the cell membrane and nanotopography is non-specific, ECM-derived adhesive ligands (e.g., RGD motifs) are required for stable cell-material attachment [22]. Previous work revealed that 110 nm nanotubes adsorbed more protein than 30 nm nanotubes, thereby influencing the diameter-dependent regulation of oxidative stress responses and osteogenic differentiation in MC3T3-E1 cells [23]. Increasing tube diameter enhanced fibrinogen adsorption while inhibiting BSA adsorption, thereby modulating platelet and endothelial cell adhesion [24]. Although ECM-derived adhesive ligands from serum are known to exert important biological effects on cells, how diameter-associated nanotopographical cues regulate the organization of interfacial adhesive ligands, and how these adhesive signals are subsequently converted into cellular responses, remain poorly understood. Although nanotube diameter is generally considered an important design parameter governing osteogenic responses, it should be noted that diameter itself does not directly interact with cells as an isolated variable. Instead, changes in nanotube diameter generate distinct nanotopographical features, including alterations in the effective top surface area, wettability, and roughness, which subsequently influence the adsorption, distribution, and presentation of interfacial adhesive ligands [25,26]. These diameter-dependent differences in adhesive ligand organization may serve as a biological intermediary that converts nanotopographical signals into cell-perceivable adhesive cues. Therefore, the regulation of osteogenic differentiation by TiO2 nanotube diameter likely involves a hierarchical process in which diameter-dependent nanotube geometry determines adhesive ligand-mediated cell adhesion patterns, followed by modulation of focal adhesion (FA) assembly, and downstream signaling pathways. However, how nanotopography-regulated adhesive ligand organization determines cell adhesion architecture and downstream cellular responses remains poorly understood. Over the last decade, in vitro studies have increasingly highlighted the role of mechanical signals in regulating stem cell differentiation, with mechanical loading broadly categorized into active and passive modes. Active mechanical loading involves conventional mechanical stimuli, such as tensile, compressive and shear force, and has been shown to affect cell lineage commitment and proliferation [27,28]. In contrast, passive mechanical loading arises from changes in intracellular force elicited by stem cell responses to subtle surface topographies and inherent material properties [29,30]. Although topographical cues have been reported to influence cell adhesion, cytoskeletal organization, traction force generation, and mechanotransduction signaling [31,32], how nanotopography regulates intracellular force and its relationship with osteogenic differentiation remain incompletely understood. In this study, we generated a series of TiO2 nanotube surfaces with defined diameters (30–100 nm) using different anodization voltages and investigated how nanotube geometry regulates cell behavior. We found that nanotube diameter was associated with distinct spatial presentation patterns of serum-derived adhesive proteins. These differences influenced FA organization, resulting in numerous small FAs on small-diameter nanotubes (30 nm) and fewer but more mature FAs on large-diameter nanotubes (100 nm). The more mature FAs promoted cytoskeletal assembly and enhanced cellular polarization. Using a Talin-TS biosensor [33], we further visualized changes in intracellular force. Increased intracellular force was accompanied by nuclear pore deformation and enhanced YAP nuclear translocation, ultimately promoting osteogenic differentiation in vitro and in vivo. These findings provide mechanistic insight into how implant nanotopography regulates cell behavior and may contribute to the rational design of osteogenic implant surfaces. Results Voltage-dependent nanotubular topographies positively regulate osteogenic efficacy To determine how nanotubes with diverse diameters influence the osteogenic differentiation of the preosteoblastic cell line MC3T3-E1, we employed a fluoride-assisted field-dissolution anodization system. Nanotubes with different diameters were fabricated on Ti substrates by adjusting the applied voltage (Fig. 1A). Ti samples anodized at 5 (NT 5), 10 (NT 10) and 20 V (NT 20) displayed voltage-dependent colour changes from blue to brown (Fig. 1B). The flat (F) surface remained featureless, whereas anodization produced nanotubes with mean diameters of ∼30 nm (NT 5, 33.20 ± 7.53 nm), ∼60 nm (NT 10, 65.90 ± 9.77 nm) and ∼100 nm (NT 20, 100.20 ± 12.55 nm) (Fig. 1B and C). The osteogenic efficacy of MC3T3-E1 cells exhibits a positive dependence on the nanotube topographical diameter. (A) Flowchart of Ti sheet preparation and its application in cell experiments. (B) Characterization of Ti sheets using digital images and SEM micrographs. (C) Quantification of nanotube diameter (n = 20). (D) SEM images of cells after critical-point drying on different samples on day 3. (E) Quantification of cell roundness. Roundness was calculated as 4 × area/(π × long axis [2]) (n = 20). (F) Osteogenic gene expression of cells on different substrates at day 7 of osteogenic induction (n = 3, experiments were independently repeated n times with three biological replicates). (G) WB assay for osteogenic protein in cells on various surfaces on day 14 under osteogenic conditions. (H) Expression analysis for (G) (n = 3). Interestingly, after 3 days of MC3T3-E1 cell culture on these substrates, a large number of uniformly distributed cells on F and NT 5, and these cells exhibit a relatively high degree of circularity. With increasing nanotube diameter, cells exhibited progressively elongated morphology accompanied by pronounced pseudopodial extension, indicating enhanced cellular polarization (Fig. 1D and E). We next evaluated the osteogenic effects of the different nanotube diameters. After 7 days of osteogenic induction, expression of osteogenesis-related genes, including Alpl (alkaline phosphatase), Spp1 (secreted phosphoprotein 1, osteopontin), Runx2 (runt-related transcription factor 2), and Bglap2 (bone gamma-carboxyglutamate protein 2, osteocalcin), was upregulated by the nanotubular topographies compared with F, with the induction efficacy following the trend NT 20 > NT 10 > NT 5 > F (Fig. 1F). After 14 days, the upregulation of ALPL, SPP1, RUNX2, and BGLAP2 protein levels remained positively correlated with nanotube diameter (Fig. 1G and H). This trend persisted under normal culture conditions, although the overall levels were slightly lower than those observed under osteogenic induction (Figs. S1 and S2). Collectively, these findings demonstrate that osteogenic differentiation increased progressively with increasing nanotube diameter, both at the gene and protein levels. Topography-induced spatial presentation of RGD-containing adhesive ligands regulates cell morphology Given the diameter-dependent effects of nanotubes on cell morphology, we hypothesized that topographical cues drive their osteogenic activity. Upon implantation, protein adsorption from blood and tissues rapidly forms a corona, influencing subsequent cell adhesion [34]. To model this, Ti samples were incubated in 10 % foetal bovine serum (FBS). F adsorbed the least protein (∼5 mg), whereas NT surfaces adsorbed over twice as much, with NT 20 slightly higher than NT 5 and NT 10 (Fig. S3). Proteins on F were evenly distributed, while on nanotubes they localized to tube rims and formed mesh-like structures with pore sizes dependent on nanotube diameter (Fig. 2A, Fig. S4). Quantitative analysis showed that the percentage of the effective top surface area before protein adsorption and the top-surface protein coverage after protein adsorption exhibited similar decreasing trends with increasing nanotube diameter (Fig. 2B and C). Although NT 20 showed a slightly higher total amount of adsorbed protein than NT 5 and NT 10 (Fig. S3), its reduced effective top surface area was accompanied by a lower protein coverage on the nanotube top surface, indicating that the adhesive interface available for cell attachment became progressively smaller as nanotube diameter increased. Interfacial topography influences MC3T3-E1 morphology by regulating the distribution of adhesion ligands. (A) SEM images of FBS adsorption on different samples. (B) Quantification of the percentage of the effective top surface area before protein adsorption (n = 3). (C) Quantification of the top-surface protein coverage after protein adsorption (n = 3). (D) Representative immunogold electron microscopy images of fibronectin on different samples. (E) SEM images of pseudopodia on each group of samples. (F) Representative fluorescence images of F-actin staining of cells at different time points under conditions with or without FBS. (G) Representative fluorescence images of F-actin staining showing cell morphology following pre-incubation with the RGD inhibitor Cyclo(-RGDfk) (0.5 mM) under FBS-coated conditions or with RGD (7.5 mg/mL) under FBS-free conditions. (H) Quantification of single cell area after 2 h of culture on each sample in the presence of FBS (n = 50). (I) Quantification of single cell area after 2 h of culture on each sample group in the absence of FBS (n = 50). (J) Quantification of single cell area after 1 day of culture on each sample under either FBS or FBS-free conditions (n = 50). To further evaluate factors that may influence protein adsorption, surface wettability, roughness, and nanotube wall thicknes were characterized. As shown in Figs. S5 and S6, F exhibited the lowest surface roughness and relatively poorer hydrophilicity. No significant differences in wettability were observed among the nanotube groups, whereas the introduction of nanotube topography was accompanied by an increase in surface roughness compared with the flat surface. Although NT 20 showed a slightly higher roughness than NT 10, the difference was not statistically significant. Nanotube wall thickness increased with nanotube diameter, reaching 19.48 ± 3.21 nm in NT 20 (Fig. S7). However, the increase in wall thickness was insufficient to compensate for the progressive reduction in the effective top surface area associated with increasing nanotube diameter. These results indicate that, although nanotube diameter was accompanied by changes in several surface characteristics, the reduction in top-surface protein coverage was more closely associated with the geometrical decrease in the effective top surface area. To further visualize the spatial presentation of serum-derived adhesive ligands, colloidal gold immunoelectron microscopy was performed using a fibronectin-specific antibody. Previous proteomic studies have shown that fibronectin, vitronectin, and laminin are major serum-derived adhesion-related proteins, all of which contain integrin-recognized RGD motifs [[35], [36], [37], [38]]. Fibronectin was therefore selected as a representative adhesive ligand because it is one of the predominant serum-derived extracellular matrix proteins and contains a well-characterized RGD sequence within its FIII10 domain [39]. Immunogold labeling revealed that fibronectin was randomly distributed on the flat surface, whereas on nanotube surfaces gold particles preferentially localized along the nanotube rims, forming diameter-dependent distribution patterns (Fig. 2D). Because the nanotube diameter of NT 5 was relatively small, the overall distribution pattern remained visually similar to that on the F. As nanotube diameter increased, gold particles exhibited a progressively more clustered distribution along the tube walls. We further quantified the immunogold-labeled fibronectin distribution from the immunoelectron microscopy images. The results showed that the total number of gold particles within the entire projected areas was comparable among TiO2 nanotube surfaces with different diameters (Fig. S8A). However, after normalization to the effective top surface area of nanotubes, the gold particle density increased with increasing nanotube diameter (Fig. S8B). This increase reflects the reduced effective top surface area available for ligand presentation and the consequent enrichment of gold-labeled fibronectin signals at the nanotube top rim regions, resulting in a more localized ligand presentation pattern. These findings suggest that nanotube diameter-dependent topographical features regulate the spatial organization of functional adhesive ligands rather than simply altering their overall abundance. It should be noted, however, that individual gold particles cannot be interpreted as representing individual fibronectin molecules because immunogold labeling does not establish a one-to-one correspondence between gold particles and antigen molecules. Fibronectin immunogold labeling was therefore used as a surrogate indicator of the spatial presentation of RGD-containing adhesive ligands rather than a direct quantification of individual fibronectin molecules. Next, differences in pseudopodium formation were examined across the samples (Fig. 2E–S9). Cells on F and NT 5 exhibited broader pseudopodia with extensive spreading. Yellow arrows highlight the random connection of pseudopodium edges to protein patches. Beginning with NT 10, pseudopodia became progressively slender and elongate, arranging themselves along the tube wall. The pseudopodia on NT 20 were noticeably slenderer, with an average width of only 30 nm, and similarly arranged along the tube walls (Fig. S10). These observations suggest that the spatial presentation of adhesive ligands influences pseudopodium extension. To evaluate the contribution of adhesive ligands, nanotube surfaces were pre-coated in PBS with or without FBS for 2 h, and seeded with cells in FBS-free medium. Under FBS-coated conditions, cells spread within 2 h (Fig. 2F), with no significant differences in area initially, though a decreasing trend was observed with increasing nanotube diameter (Fig. 2H). After 1 day, cells on F and NT 5 remained round and well-spread, whereas larger nanotubes induced elongated shapes with reduced area (Fig. 2F and J). In the absence of FBS, cells remained round, non-adherent, and largely non-viable (Fig. 2I–S11). To further determine whether these responses were mediated by RGD-dependent adhesion, the integrin-binding peptide inhibitor Cyclo(-RGDfk) and exogenous RGD coating were applied. Under FBS-coated conditions, Cyclo(-RGDfk) markedly inhibited cell spreading, whereas coating nanotube surfaces with RGD under FBS-free conditions restored cell spreading (Fig. 2G). Moreover, coating nanotubes with RGD alone reproduced the same distribution of adhesive ligands and the same pseudopodium extension patterns observed with FBS (Fig. S12). Together with the immunogold localization of fibronectin, these results indicate that the diameter-dependent cellular response is primarily mediated by differences in the spatial presentation of RGD-containing adhesive ligands. Differential distribution of adhesive sites promotes focal adhesion (FA) maturation and cytoskeletal oriented alignment To investigate whether the topography-dependent differences in cell morphology were accompanied by mechanotransduction-associated transcriptional responses, Ankrd1, a well-established mechanosensitive target gene [40,41], was first examined at different time points. Significant differences in Ankrd1 expression on different surfaces first became apparent on day 3, with the highest expression observed in NT 20 (Fig. S13). Therefore, transcriptomic analysis was performed on day 3, when cell spreading had matured and mechanotransduction-associated transcriptional differences had become distinguishable among the different groups. Transcriptomic analysis showed that the number of differentially expressed genes (DEGs) increased with nanotube diameter (Fig. 3A). Gene Ontology (GO) enrichment revealed that DEGs in NT 10 and NT 20 vs F were associated with FA assembly, cytoskeletal organization, and pseudopodium formation (Fig. 3B). Notably, NT 20 vs F also began to exhibit enrichment in the biological processe related to bone development. Topography mediates the formation of differential focal adhesions (FAs) and the oriented assembly of the cytoskeleton. (A) Volcano plot of DEGs in each group. (B) GO analysis of differential genes in different groups. (C) Paxillin (PXN) immunofluorescence in MC3T3-E1 cells after 3 days of culture on different substrates. (D) Quantification of single-FA area, single-FA length, number of FA in each cell and percentage of mature FA (n = 20 cells). (E) Immunofluorescence detection of cytoskeleton. Based on these observations, we focused on assessing FA development. Paxillin (PXN), a scaffolding protein within FA, recruits FA-associated proteins and downstream signaling components, and is widely used as a FA marker [42]. After 3 days of culture, nanotube diameter resulted in distinct differences in FA characteristics: (1) FA area and length increased with diameter; (2) the number of FAs per cell decreased as diameter increased; (3) the number of mature FAs (>1 μm) [43] was positively correlated with diameter; (4) FA localized mainly at the cell edges (Fig. 3C, D, S14). These results emphasize the topography-dependent differential maturation of FA, which has been largely overlooked. Large-diameter nanotubes promoted the formation of fewer but larger FAs, whereas small-diameter nanotubes produced numerous but smaller FAs. Interestingly, despite these distributional differences, the overall expression level of PXN did not differ significantly across groups (Fig. S15). Another intriguing finding was that although cytoskeletal expression was similar across groups (Fig. S16), circumferentially arranged stress fibers was showed on F and NT 5, where cells spread uniformly (Fig. 3E). In contrast, with increasing nanotube diameter, cells became slenderer, and stress fibers adopted a pronounced anisotropic alignment along the long axis of the cell, indicating enhanced cellular polarization. Differential expression analysis of mechanosensitive genes further showed that NT 20 markedly upregulated mechanical response genes [44] (Fig. S17). These results suggest that although FA and cytoskeletal protein levels remain relatively constant across groups, as expected for housekeeping components that maintain cellular structure, diameter-dependent nanotube topography substantially altered their spatial organization and orientation, which may contribute to distinct mechanotransduction outcomes. Mature FAs generate intrinsic mechanical force through polarized stress fibers assembly FA maturation involves larger FA size and phosphorylation of FA-enriched proteins. Focal adhesion kinase (FAK), a kinase localized at FAs, undergoes autophosphorylation as its local concentration increases, thereby activating downstream signaling pathways [45]. Our results showed that total FAK protein levels did not differ significantly among groups (Fig. 4B, S18). However, similar to PXN, FAK displayed striking differences in spatial distribution. Larger nanotube diameters produced fewer but larger FAK clusters, whereas smaller nanotubes yielded a greater number of smaller clusters (Fig. 4A–S19). Phosphorylation analysis further revealed that FAK phosphorylation increased with nanotube diameter. In parallel, the phosphorylation level of myosin light chain II (MLC II) also increased with increasing diameter (Fig. 4B–S20). To clarify the relationship between FAK and MLC II, inhibition of FAK phosphorylation with Defactinib (Def) reduced MLC II phosphorylation, indicating that MLC II acts downstream of FAK. Mature FAs promote intracellular force generation through the p-FAK (Phosphorylated FAK)/p-MLC II signaling axis. (A) Immunofluorescence images of FAK and p-FAK in MC3T3-E1 cells cultured for 3 days on different samples (n = 20 cells). (B) WB assay for FAK, p-FAK, MLC II and p-MLC in cells on various surfaces (n = 3). (C) The structure diagram of Talin-TS biosensor. The donor is EGFP and the acceptor is TagRFP, linked by an extensible elastic spider silk protein embedded within the Talin structure. Although the Talin-TS biosensor is localized at FAs, the FLIM-FRET analysis was performed using whole-cell regions of interest from individual cells and was not restricted to individual FA regions. (D) The fluorescence lifetime (FLIM) at different degrees of fluorescence resonance energy transfer (FRET) was detected. FLIM bar represented the color indication of fluorescence lifetime from 1 ns to 3 ns. (E) Statistical analysis of fluorescence lifetime (n = 6 cells). (F) Schematic illustration of intracellular force generation. Increased MLC II phosphorylation and stress fiber assembly are generally regarded as indicators of enhanced intracellular mechanical force; however, direct visualization of these force changes has remained challenging. To quantitatively assess intracellular force, we transfected cells with the Talin-TS biosensor plasmid. This sensor reports mechanical force through changes in fluorescence resonance energy transfer (FRET) efficiency (Fig. 4C). FRET efficiency is inversely proportional to the sixth power of donor-acceptor distance, while donor fluorescence lifetime is positively proportional to the sixth power of this distance [46]. Based on this principle, cells cultured on NT 20 exhibited the highest donor fluorescence lifetime, whereas cells on F showed the lowest (Fig. 4D and E). Inhibition of phosphorylated MLC II using Blebbistatin (Ble) effectively suppressed intracellular force, resulting in a marked reduction in fluorescence lifetime of the donor. These findings provide direct evidence that large-diameter nanotubes (NT 20) induce greater intracellular force. Mature FAs promotes FAK autophosphorylation, which activates downstream signaling pathways of p-MLC II and serves as a key driver of intracellular force generation (Fig. 4F). Moreover, these results demonstrate that topographical cues alone, without any externally applied force, are sufficient to drive spontaneous changes in intracellular mechanical force. Polarization-enhanced intracellular force promotes nuclear pore stretching and osteogenic differentiation The upregulation of mechanosensitive genes (Fig. S17) suggested that cytoskeletal tension was transmitted into the nucleus. 3D reconstruction revealed that nuclear shape became increasingly oriented and deformed with larger nanotube diameters (Fig. 5A, S21). Nuclei on NT 20 were slenderer in planar area, and exhibited stronger orientation, whereas nuclei on F retained a disk-like morphology (Fig. 5B and C). Increased intracellular force is accompanied by nuclear pore deformation, enhanced YAP nuclear localization, and osteogenic differentiation. (A) Laser confocal fluorescence images of MC3T3-E1 stained for F-actin (green), and nucleus (blue) after 3 days of culture. (B) Statistical analysis of nuclear area, nuclear height, and nuclear volume (n = 20 cells) and reconstruction of 3D schematic. (C) Three-dimensional schematic illustration showing nuclear morphology. (D) TEM images of nuclear pores. (E) Nuclear pore size in cells on different samples (n = 40 nuclear pores from 15 cells per condition). (F) Representative images of YAP immunostaining. (G) Nuclear/cytosolic YAP ratios in cells on different samples (n = 12 cells). (H) Representative micrographs of MC3T3-E1 stained with the ALPL after being cultured on the related substrates for 14 days. (I) Determination of ALPL activity (n = 3). (J) WB assay for nucleus YAP and cytosol YAP in cells on various surfaces on day 3. (K) Quantification of nuclear and cytosolic YAP protein levels using ImageJ (n = 3). Cells were located by upright microscopy, sectioned along their long axis, and nuclear cross-sections were examined (Fig. 5D, E, S22). We observed a clear trend of increasing nuclear pore size with increasing nanotube diameter. Disruption of the cytoskeleton, inhibition of FAK signaling, or suppression of MLC II–mediated contractility all led to reduced pore size. Nuclear pore dilation can affect passive transport of transcription factors [47]. YAP, a key mechanosensitive transcriptional coactivator involved in mechanotransduction and osteogenic differentiation [48], showed increased nuclear localization with larger nanotube diameters, while inhibition of mechanical signaling strongly reduced its nuclear translocation (Fig. 5F, G, J, K). Consistent with previous reports, blocking the interaction between YAP and TEAD significantly impaired osteogenic differentiation (Fig. 5H, I, S23). Our results indicate that diameter-dependent nanotopographical cues regulate intracellular force generation, which is accompanied by nuclear pore deformation, increased YAP nuclear localization, and enhanced osteogenic differentiation. Large-diameter nanotubes promote in vivo bone formation Representative images showed that rod-shaped implants exhibited surface colours consistent with the corresponding Ti sheets (Fig. 6A). In the NT groups, the surfaces were uniformly and extensively covered with nanotube structures, and the tube diameters corresponded exactly to their matched Ti sheet groups regardless of which side of the rod was examined. The Ti rods were implanted into the rat distal femur intercondylar fossa (Fig. 6B), with no postoperative deaths or infections. The rat femoral condyle implantation model is widely used to evaluate the influence of implant surface modifications on peri-implant bone remodeling and osseointegration [49,50]. This model reasonably mimics the biological events occurring at the bone–implant interface following clinical implantation of orthopedic and dental implants. Diameter-dependent enhancement of osteogenic differentiation and osseointegration in rats. (A) Digital photos and SEM photos of each group of Ti rod, sample diameter 2 mm, length 8 mm. (B) Schematic diagram of implantation experiment. (C) Two months after implantation of different samples, digital photos of distal femurs. (D) SEM images of the implanted rods after the push-out test 2 months after implantation. (E) 3D reconstruction images of the Ti rod and the bone tissue around the Ti rod 2 months after implantation. (F) Analysis of push-out force in different samples (n = 6). (G) Representative histological sections 2 months after implantation at 1.5 mm below the growth plate (n = 6). (H) Representative immunofluorescence images of YAP expression and DAPI nuclear staining in peri-implant bone tissue 14 days after implantation in the different groups (n = 6). After two months, all implants remained in place. The drilling sites of NT 10 and NT 20 were fully closed and covered by white tissues, with fascia already regenerated on top (Fig. 6C). NT 5 showed a moderate degree of closure, whereas in the F group, the drilling site still exhibited a blood-scab window with incomplete closure of the white tissue. Push-out testing revealed higher average force in the NT groups than F, with NT 20 significantly greater than F (Fig. 6F–S24). Retrieved implants showed bone fragments and dense trabeculae increasing with nanotube diameter, while F had minimal bone fragments (Fig. 6D). NT groups exhibited excellent interfacial bonding between the implants and newly formed bone. Micro-CT (Fig. 6E–S25) and histology (Fig. 6G) showed that NT 10 and NT 20 implants had extensive, dense trabeculae and superior bone coverage, whereas F exhibited sparse bone tissue and fibrous encapsulation. NT 20 displayed the greatest osseointegration, followed by NT 10 and NT 5, demonstrating that nanotube topography promotes bone formation and reduces fibrous encapsulation in a diameter-dependent manner. To further examine whether the in vitro mechanotransduction pathway was recapitulated in vivo, immunofluorescence staining was performed on peri-implant bone tissue 14 days after implantation. Representative images revealed increased nuclear localization of YAP in the NT 10 and NT 20 groups compared with the F group, accompanied by enhanced RUNX2 expression in localized regions (Fig. 6H–S26). Together, these findings indicate that the enhanced osteogenic differentiation and osseointegration induced by large-diameter nanotubes in vitro can also be recapitulated in vivo. Discussion Interface topography has emerged as a powerful regulator of cell behavior, influencing cell morphology, adhesion, proliferation, and differentiation through intracellular signaling pathways [51]. However, the initial events by which cells perceive topographical cues and convert them into biological responses remain incompletely understood. In the present study, TiO2 nanotubes were employed as a model nanotopographical system. TiO2 nanotubes were selected not only because of their widespread application as implant surface modifications [52], but also because they provide a well-controlled model system for investigating diameter-dependent cellular responses. Previous studies have consistently reported a positive correlation between nanotube diameter and osteogenic differentiation within a defined size range [14,15,53]; however, the mechanisms underlying this diameter-dependent osteogenic effect remain largely unclear. Importantly, nanotube diameter can be systematically adjusted while maintaining the same substrate composition, allowing the influence of nanoscale geometry to be examined with minimal interference from changes in material chemistry [54,55]. Therefore, TiO2 nanotubes provide a suitable platform for exploring the fundamental mechanisms responsible for diameter-dependent osteogenic responses. Previous studies have shown that micro-island substrates with defined RGD spacings exhibit a critical ligand spacing of approximately 70 nm for effective cell spreading, corresponding to the characteristic length scale of α-actinin; beyond this threshold, α-actinin assembly is impaired [56]. Guided by this observation, we employed nanotube topographical cues with diameters of 30, 60 and 100 nm. Although cell spreading progressively decreased with increasing tube diameter, osteogenic differentiation showed a positive dependence on tube diameter rather than peaking at 60 nm. To further investigate this discrepancy, we examined the distribution of FBS-derived proteins and RGD ligands on the nanotopographical interfaces. Unlike engineered surfaces with predefined and uniform ligand spacing [56,57], adhesion ligands on nanotube surfaces were preferentially distributed along the nanotube rims, forming mesh-like structures with diameter-dependent pore sizes. This heterogeneous ligand organization provided spatially distinct adhesion sites and more closely reflected the native adhesive microenvironment. Colloidal gold immunoelectron microscopy (Fig. 2D) revealed that FBS-derived adhesive ligands (fibronectin) were preferentially localized along the nanotube rims, forming a diameter-dependent distribution pattern. Previous studies have proposed that fibronectin adsorption onto titanium dioxide is mediated by divalent cations and that nanoscale convex edge features provide energetically favorable sites for protein adsorption [[58], [59], [60]]. Accordingly, the top surface of the nanotube rims is likely to serve as the primary protein-presenting interface accessible to cells and may therefore represent the predominant site for integrin-mediated cell adhesion. It should be noted that increasing nanotube diameter was accompanied by changes in several structural and surface characteristics, including wall thickness, surface roughness, wettability, and effective top surface area. Our geometric analysis (Fig. 2B, C, S7) showed that enlargement of nanotube diameter substantially reduced the effective top surface area available for protein presentation and cell interaction. Although nanotube wall thickness also increased with diameter, its contribution to the available top surface area was relatively small and did not compensate for the loss of effective interface. Likewise, while the introduction of nanotube topography increased surface roughness (Fig. S6) compared with the flat surface, roughness reached a plateau between NT 10 and NT 20, and no significant differences in wettability (Fig. S5) were observed among the nanotube groups. In contrast, both the effective top surface area and the corresponding top-surface protein coverage continued to decrease progressively with increasing nanotube diameter. Collectively, these observations suggest that protein presentation is primarily governed by the overall diameter-dependent nanotube geometry, with the geometrical reduction in effective top surface area playing a dominant role in determining the protein-presenting interface directly accessible to cells. Together with the preferential enrichment of fibronectin along the nanotube rims, this diameter-dependent geometrically regulated adhesive interface may provide the initial topographical cue governing focal adhesion organization, cellular polarization, intracellular force generation, and subsequent mechanotransduction. Based on these observations, we propose a possible model (Fig. 7) that the interface topography regulates the distribution of adhesion proteins and thus provides differential adhesion sites for cells. Cells perceive different distribution of adhesion sites and adopt two ways to form FA. 1. Perceiving the uniform distribution of ligands on the small diameter, forming uniform dense FAs. 2. Perceiving the uneven distribution of ligands on the large diameter, forming more mature but less FAs. In other words, when adhesive ligands are presented in a more homogeneous spatial distribution, cells tend to form FAs in a more uniform manner, whereas heterogeneous spatial presentation of adhesive ligands promotes the localized assembly of larger FAs, a phenomenon that has not been extensively described in the literature. Proposed model illustrating how TiO2 nanotube topography regulates osteogenic differentiation. Adsorbed adhesion-related proteins were preferentially distributed along the nanotube walls, with their distribution pattern varying with nanotube diameter. Cells perceived these differences in adhesive protein presentation and exhibited distinct focal adhesion (FA) organization. Small-diameter nanotubes promoted the formation of numerous small FAs, whereas large-diameter nanotubes promoted the formation of fewer but larger FAs. These differences in FA organization regulated cytoskeletal assembly, increased intracellular force, and enhanced cellular polarization. Increased intracellular force was accompanied by nuclear pore deformation and enhanced YAP nuclear translocation, ultimately promoting osteogenic differentiation. We propose that the distinct FA architectures associated with nanotube diameter arise from the combined effects of diameter-dependent nanotopographical features and ECM-derived adhesive ligand organization. With increasing nanotube diameter, the effective top surface area for protein adsorption and the top-surface protein coverage gradually decrease. However, immunogold electron microscopy revealed that gold-labeled fibronectin signals were preferentially distributed along the nanotube rims, and larger-diameter nanotubes exhibited increased local clustering of these ligands at the tube boundaries (Fig. 2D–S8). These findings suggest that, although the overall ligand abundance is not increased, larger-diameter nanotubes provide fewer but more concentrated adhesive ligand presentation sites due to their specific nanotopographical geometry. Previous studies have shown that local clustering of extracellular matrix ligands facilitates integrin clustering and FA maturation [[61], [62], [63]]. Such localized ligand enrichment may facilitate integrin clustering and FA maturation, thereby promoting the formation of larger and more mature FAs. However, because the present study did not directly correlate fibronectin distribution with individual FAs at the same locations, this interpretation remains speculative and should be regarded as a proposed mechanistic model rather than direct experimental evidence. In addition, nanotube diameter alters the physical constraints experienced by cells during adhesion formation. Previous studies have suggested that the effective spacing required for FA assembly is approximately 70 nm [56,64]. On smaller-diameter nanotubes, cells can bridge adjacent nanotube structures and establish relatively distributed adhesion sites, allowing cell attachment to be maintained through the formation of multiple small FAs. In contrast, larger-diameter nanotubes impose greater nanoscale spacing constraints, limiting the ability of cells to form adhesion structures across individual nanotube gaps. Under these conditions, cells preferentially develop fewer but larger and more mature focal adhesions at spatially favorable adhesion regions, which may facilitate enhanced integrin clustering and force transmission. These diameter-dependent modes of FA assembly further influence downstream mechanotransduction. The formation of mature FAs on larger-diameter nanotubes promotes enhanced FAK phosphorylation and activation of downstream signaling pathways, resulting in increased cytoskeletal tension and intracellular force generation [65]. Consequently, cells on larger-diameter nanotubes exhibit enhanced polarization compared with cells on smaller-diameter nanotubes, where adhesion sites are more uniformly distributed and intracellular forces are relatively lower. Therefore, nanotube diameter-dependent topographical features regulates both the spatial presentation of adhesive ligands and the physical constraints for FA assembly, leading to distinct FA maturation patterns, intracellular mechanical states, and ultimately different cellular phenotypes. FAs are large protein complexes formed by integrins, scaffold proteins, and signaling proteins. As FAs mature, they organize the F-actin cytoskeleton and generate intracellular force, promoting cell proliferation [66]. However, the molecular regulation of this process and its dependence on nanoscale topography remain unclear. Intracellular force is mediated by MLC II-driven stress fiber assembly, with Thr18/Ser19 phosphorylation activating ATPase activity to generate contractility [67]. In this study, we found that mature FAs recruited more FAK, and although total FAK levels were unchanged, Tyr397 phosphorylation increased significantly. FAK activation involves dimerization and conformational changes exposing autophosphorylation sites, while FAK in turn promotes MLC phosphorylation [68,69]. These results indicate that FA organization is closely associated with intracellular force generation and cellular polarization. Interface topography has been shown to drive mechanotransduction [70,71], but measuring the force sensed by individual mechanosensitive units remains challenging. Techniques such as atomic force microscopy and optical or magnetic tweezers provide subcellular or single-protein resolution, yet often outside physiological contexts. Using Talin-TS, which detects force up to 6 pN [33,72], we observed that cells on topological interfaces generated spontaneous force changes regulated by FA maturity, providing direct evidence that interface topography regulates talin tension and intracellular mechanical states. The formation of fewer but larger FAs was accompanied by increased intracellular force, enhanced cytoskeletal alignment, and increased cellular polarization. These changes were associated with nuclear pore deformation and enhanced YAP nuclear translocation, ultimately promoting osteogenic differentiation, consistent with previous observations on soft and stiff hydrogels, consistent with findings on soft and stiff hydrogels [73]. It should be noted that direct visualization and quantitative measurement of nuclear pore size in intact cells remain technically challenging, and TEM therefore provides only morphological evidence of apparent nuclear pore deformation. Furthermore, the diameter-dependent nanotube topography had the same positive effect on promoting osteogenesis through rat femur implantation experiments. Together, these results establish that interface topography regulates adhesive protein presentation, FA organization, cellular polarization, and YAP-dependent osteogenic differentiation, offering guidance for the design of biomaterials. Conclusion Interface topography plays an important role in regulating cellular mechanotransduction and subsequent gene expression. In this study, using TiO2 nanotubes with defined diameters as a model system, we found that diameter-dependent nanotube geometry influenced the spatial presentation of adsorbed adhesive proteins on the material surface. These differences were associated with distinct FA organization, characterized by numerous small FAs on small-diameter nanotubes and fewer but larger FAs on large-diameter nanotubes. The altered FA organization was accompanied by increased intracellular force, enhanced cytoskeletal alignment, and greater cellular polarization, which facilitated YAP nuclear translocation and promoted osteogenic differentiation both in vitro and in vivo. Our findings provide a mechanistic framework linking nanotopography, adhesive protein presentation, FA organization, and osteogenic mechanotransduction. This work offers new insights into how cells perceive implant interfaces and may provide guidance for the rational design of bioactive implant surfaces. Material and methods Preparation and characterization of TiO2 nanotubes with different diameters Pure Ti sheet (Dongguan Jiazhi Metal Products Co., Ltd., China, thickness 0.05 mm, purity 99.999 %) was polished to the mirror and cut into the required size. The oxide film on the surface of the Ti sheet was removed by ultrasonic pickling for 60 s in a pickling solution containing 2 wt % hydrofluoric acid and 20 wt % nitric acid. The electrolytic cell was poured into an electrolyte containing 15 wt % phosphoric acid and 0.9 wt % hydrofluoric acid. Ti and graphite sheets were used as the anode and cathode, respectively, with a spacing of 5 cm. Electrolytic treatment was performed for 1 h under three different voltage conditions (5 V, 10 V, 20 V) (KR-15001, Kingrang, China). At the same time, a magnetic stirrer is used to stir to ensure uniform electrolyte composition. The TiO2 nanotube interfaces obtained by this step are named NT 5, NT 10 and NT 20, respectively. The Ti surface smoothed by pickling treatment was used as a control, named F. To enhance the hardness and corrosion resistance of TiO2 nanotubes, it is necessary to keep it at 450 °C for 5 h to convert it into anatase crystal form. We have verified that this process does not affect the size change of nanotubes. For animal experiments, a total of 72 cylindrical Ti implants (8 mm long and 2 mm in diameter) were anodized for nanotubular surfaces. It is worth noting that F still referred to the untreated Ti rod after pickling, and NT 5, NT 10 and NT 20 also referred to the samples prepared at 5 V, 10 V and 20 V, respectively. The nanostructured surface was sputtered with Au, and the structure of the nanotubes was scanned and analyzed using a scanning electron microscope (SEM, Gmini 300, Zeiss, Germany). Measurements of surface wettability and roughness Surface wettability was evaluated by measuring the static water contact angle using a contact angle goniometer (DSA100, KRÜSS, Germany). A 10 μL droplet of deionized water was gently deposited onto the sample surface using the instrument's automated dispensing system, and images were immediately captured for contact angle analysis. Three independent measurements were performed on each sample at different locations, and the average value was used for statistical analysis. Surface morphology and roughness were characterized by atomic force microscopy (AFM, NanoWizard IV, JPK, Germany) operated in tapping mode under ambient conditions (25 °C, 10 % relative humidity). Three-dimensional surface topography images were acquired, and the root mean square roughness (Rq) was calculated using the instrument software. For each independent sample, at least three randomly selected regions were analyzed, and the average Rq value was used for statistical analysis. Detection of protein adsorption performance To simulate cell culture, PBS (Hyclone, USA) was used to prepare 10 % (v/v, about 7.5 mg/mL) FBS solution or 7.5 mg/mL RGD (Selleck, USA) solution. Each group of samples was placed in 24-well plates, followed by the addition of 200 μL of FBS solution per well. The supernatant was discarded at 2 h and 24 h respectively. The samples were then placed in a freeze dryer for drying and subsequently examined under SEM. In order to determine the accurate content of FBS adsorbed by different samples, BCA (Beyotime, China) kit was used for detection. The amount of adsorbed protein = the amount of total protein – supernatant protein. The total protein is the protein content of 10 % FBS stock solution, and the amount of supernatant protein is the amount of protein not adsorbed after the experiment. Immunogold labeling and scanning electron microscopy To visualize the spatial distribution of serum-derived adhesive proteins on different nanotopographical interfaces, the samples were first incubated in 10 % foetal bovine serum (FBS) for 1 day at 37 °C to allow protein adsorption. After gentle rinsing with PBS to remove loosely bound proteins, the samples were incubated overnight at 4 °C with a primary antibody against fibronectin (Supplementary Table S2). After washing with PBS, the samples were incubated with a gold nanoparticle-conjugated secondary antibody. The labeled samples were washed thoroughly with PBS, air-dried at room temperature, and directly observed using a SEM (Gmini 300, Zeiss, Germany). Representative images were acquired using backscattered electron imaging mode to evaluate the spatial distribution of immunogold-labeled fibronectin on different nanotube surfaces. Cell culture and osteogenic differentiation MC3T3-E1 cells (ATCC, USA) were seeded on various substrates at 3 × 103 cells/cm2 and cultured in α-MEM supplemented with 10 % FBS (Hyclone) and 1 % penicillin–streptomycin (Hyclone). Osteogenic differentiation was induced by adding 100 nM dexamethasone, 50 μM ascorbic acid, and 2.5 mM β-glycerophosphate, with medium replacement every 3 days for 7 or 14 days. After 24 h, inhibitors were applied as follows: Cyclo(-RGDfk) 0.5 mM (Selleck), Blebbistatin 5 μM (MCE, USA), Cytochalasin D 2 μM (Selleck), Defactinib 5 μM (MCE), and Verteporfin (VP) 2 μM (MCE). For ALPL staining, cells were fixed with 4 % paraformaldehyde and stained using a commercial ALPL staining kit according to the manufacturer's instructions (Beyotime, China). Representative images were acquired using a light microscope. For ALPL activity assay, cells were lysed, and alkaline phosphatase activity was determined using a commercial ALPL activity assay kit (Beyotime, China) according to the manufacturer's instructions. Total protein concentration was measured using a BCA protein assay kit, and ALPL activity was normalized to the total protein content. Absorbance at 405 nm was recorded using a microplate reader. Morphological observation and adhesion activity analysis of MC3T3-E1 cells MC3T3-E1 cells were seeded onto the samples at a density of 3 × 103 cells/cm2 and cultured for 3 d. For morphological observation, samples were rinsed with PBS, fixed in 2.5 % glutaraldehyde for 2 h, and sequentially dehydrated in graded ethanol solutions (30–100 %). After CO2 critical point drying (EM CPD300, Leica, Germany), cell morphology was examined by SEM. To assess the role of adhesion ligands, samples were pre-incubated with PBS containing either 10 % FBS or 7.5 mg/mL RGD for 2 h MC3T3-E1 cells were then seeded in serum-free αMEM after PBS washing. Then, cells were fixed in 4 % paraformaldehyde and permeabilized with 0.1 % Triton X-100, blocked with 5 % BSA, and stained with phalloidin probes (Supplementary Table S4). Nuclei were counterstained with DAPI. Fluorescence images were acquired using a confocal laser scanning microscope (FluoView FV1000, Olympus, Japan). Cell viability was evaluated after 24 h using Calcein AM/PI staining (Beyotime), followed by fluorescence imaging (CKX53, Olympus, Japan). Live and dead cells were visualized as green (Ex/Em: 494/517 nm) and red fluorescence (Ex/Em: 535/617 nm), respectively. Quantification was performed using ImageJ. Transmission electron microscopy (TEM) examination Cells on the samples were fixed in 2.5 % glutaraldehyde in PBS at 4 °C overnight and post-fixed in 1 % osmium tetroxide for 2 h at room temperature. After dehydration in graded ethanol, samples were embedded in epoxy resin (Electron Microscopy Sciences, USA). Cells were identified by upright light microscopy and sectioned along the long axis into ultrathin sections. Sections were stained with uranyl acetate and lead citrate, and ultrastructure was examined by TEM (HT7800, HITACHI, Japan). Real-time quantitative PCR (RTqPCR) Total RNA was isolated using the RNeasy™ Animal RNA Extraction Kit (Beyotime) following the manufacturer's protocol. Genomic DNA was removed and reverse transcription was performed using the PrimeScript™ RT kit (TaKaRa, Japan) with 1 μg of RNA. Quantitative PCR was conducted using IQ SYBR Green Supermix on a CFX96™ Real-Time PCR system (BIO-RAD, USA). Relative gene expression was calculated using the 2−ΔΔCT method and normalized to Gapdh. Primer sequences are provided in Supplementary Table S1. Protein extraction and Western blot (WB) examination Total protein was extracted by lysing cells in RIPA buffer (Boster, USA) after washing with ice-cold PBS. Cytoplasmic and nuclear fractions were prepared using a nuclear protein extraction kit (Solarbio, China) according to the manufacturer's instructions. Briefly, cells were collected by scraping and centrifugation, followed by cytoplasmic protein extraction and high-speed centrifugation (12,000–16,000 g, 4 °C, 10 min) to separate cytoplasmic (supernatant) and nuclear (pellet) proteins. Protein concentrations were determined by BCA assay. Equal amounts of protein were separated by SDS–PAGE, transferred to PVDF membranes, and blocked with 5 % skim milk in TBST. Membranes were incubated with primary antibodies at 4 °C overnight, followed by HRP-conjugated secondary antibodies at room temperature for 2 h. Immunoreactive bands were detected using an ECL chemiluminescence system (iBright FL1500, Invitrogen, USA) and quantified with ImageJ. Antibody information is provided in Supplementary Table S3, and protein levels were normalized to GAPDH. Immunofluorescence staining and confocal imaging MC3T3-E1 cells were seeded on different surfaces, washed with PBS, and fixed in 4 % paraformaldehyde. Cells were permeabilized with 0.1 % Triton X-100, blocked with 5 % BSA, and incubated sequentially with primary antibodies and fluorescent secondary antibodies (Supplementary Table S4). Nuclei were counterstained with DAPI. Fluorescence images were acquired using a confocal laser scanning microscope (FluoView FV1000, Olympus, Japan). Fluorescence intensity was quantified using ImageJ from five randomly selected high-power fields. Nuclear regions were defined by DAPI staining, and corresponding regions on the FITC channel were used to quantify nuclear YAP fluorescence. Cytoplasmic fluorescence was measured from non-nuclear regions under identical background conditions, and nuclear-to-cytoplasmic fluorescence intensity ratios were calculated. For cytoskeletal analysis, all images were acquired in 16-bit format and processed using ImageJ. Plasmids transient transfection and FRET imaging MC3T3-E1 cells were transiently transfected with the talin tension sensor Talin-TS (Addgene #83376, USA) by electroporation (520 V; Celetrix SP100, China). FRET efficiency was determined from donor fluorescence lifetime using fluorescence lifetime imaging microscopy (FLIM). FLIM–FRET measurements were performed on a confocal microscope (STELLARIS 8, Leica, Germany) equipped with Lightning and STED modes using time-correlated single-photon counting. EGFP was excited at 488 nm and detected at 527–555 nm; tagRFP was excited at 561 nm and detected at 615–670 nm. The FRET channel was excited at 488 nm and detected at 615–670 nm. Images were acquired with an integration time of 20 s using a cooled hybrid photomultiplier tube. Cells were imaged in sealed chambers, and laser power was adjusted to avoid pixel saturation. FLIM-FRET measurements were performed using the whole-cell region of interest of individual cells to assess the overall intracellular force state induced by different nanotopographical cues. Although the Talin-TS biosensor is localized at focal adhesion sites, the FLIM-FRET analysis was performed using whole-cell regions of interest from individual cells and was not restricted to individual focal adhesion regions. Cell transcriptomics analysis After 3 d of culture, total RNA was extracted using the TRIzol method. Subsequent RNA sequencing and library preparation were performed by Wuhan Kangce Technology Co., Ltd. RNA purity was assessed by NanoDrop™ OneC (A260/A280), concentration was measured using Qubit 3.0 with Qubit™ RNA assay kits, and RNA integrity was evaluated using the RNA integrity number (RIN), with all samples showing RIN ≥7.0. Stranded mRNA libraries were constructed from 2 μg of total RNA using the KC™ Stranded mRNA Library Prep Kit for Illumina® following the manufacturer's protocol. Libraries with insert sizes of 200–500 bp were enriched and sequenced on a DNBSEQ-T7 platform. Read counts were generated using FeatureCounts software, and gene expression levels were normalized as FPKM (Fragments Per Kilobase of transcript per Million mapped reads). Differentially expressed genes were identified using the edgeR package, with a threshold of P < 0.05 and |fold change| > 2. Gene ontology enrichment analysis was performed using KOBAS. All experiments were conducted with three independent biological replicates per group. Anterior femoral intercondylar notch defect model and surgery in rats 72 male Sprague-Dawley (SD) rats (2 months old, weighing about 300 g) were randomly divided into 4 groups. Rats were housed under controlled conditions (25 °C, 55 % humidity, 12 h light/dark cycle) with ad libitum access to food and water. All procedures were approved the Ethical Review Board of Southwest Jiaotong University (Approval No. SWJTU-2403-NSFC-053). All animal care and experiments were conducted in accordance with the standards of the Animal Research Committee of Southwest Jiaotong University and in accordance with international animal welfare guidelines. In short, all animals were anesthetized with isoflurane inhalation. After skin preparation, the surgical knife cut the left knee joint about 1 cm incision. The femur was exposed by blunt separation. A 2 mm diameter hole was drilled vertically from the intercondylar fossa along the bone axis, and the implant was placed in the hole. Animals were implanted unilaterally, and penicillin potassium injection (80000 units/rat, 0.1 mL, EASE PHARMACEUTICAL, China) was continuously injected 3 days after surgery to control infection. Rats were subsequently assigned to four experimental groups: F, NT 5, NT 10, and NT 20. Implant push-out force test After 2 months of implantation, the rats were euthanized by cervical dislocation, and the femur was taken out. Biomechanical tests were performed on 6 femoral samples in each group to evaluate the implant-bone bonding strength. The femur sample was vertically fixed on the test bench of the electronic universal testing machine (CMT5504/5105, MTS, USA), and the implant was pushed from the epiphysis with a needle tube, and the mechanical peak when pushing the implant was recorded. Micro-CT imaging evaluation Six samples per group were scanned using a high-resolution micro-CT system (Scanco Medical AG, Switzerland). Scans were performed at 70 kV, 120 μA, and 700 ms integration time, with an isotropic voxel size of 10 μm. The multi-level thresholds were applied to distinguish bone (threshold = 200) and implant (threshold = 700), and a Gaussian filter (σ = 1.2) was used to reduce noise. For each sample, 20 slices were analyzed from 1.5 mm below the growth plate to the distal end of the implant. A 250 μm circular region surrounding the implant defined the volume of interest. Cancellous bone parameters such as bone volume fraction (BV/TV), trabecular number (Tb. N), trabecular spacing (Tb. Sp), connectivity density (ConnD), and bone mineral density (BMD) were quantified. Histological and immunofluorescence analyses Femurs were fixed in 4 % paraformaldehyde for 3 days and decalcified in 15 % EDTA, with solution refreshed every 3 days until complete. Samples were dehydrated through graded ethanol (70–100 %) and cleared in a 1:1 mixture of xylene and ethanol before paraffin embedding. Continuous 8 μm sections were prepared. For hematoxylin-eosin (HE) staining, sections were dewaxed, rehydrated, and stained according to the manufacturer's protocol (Servicebio, China). Masson and Van Gieson (VG) staining were performed following standard procedures (Servicebio). Sections were imaged using a Pannoramic 250 digital slide scanner (3DHISTECH, Hungary). Low-magnification images were used to assess overall tissue morphology, and high-magnification images were collected within a 250 μm annular region surrounding the implant to evaluate local tissue response. For immunofluorescence analyses, femoral specimens were harvested 14 days after implantation, fixed in 4 % paraformaldehyde, decalcified in 10 % EDTA solution, dehydrated through a graded ethanol series, and embedded in paraffin. 8 μm paraffin sections were prepared. After deparaffinization, rehydration, and antigen retrieval, the sections were permeabilized with 0.1 % Triton X-100 and blocked with 5 % BSA for 1 h at room temperature. The sections were incubated overnight at 4 °C with primary antibodies against YAP or RUNX2 (Supplementary Table S4), followed by the corresponding fluorescent secondary antibodies for 1 h at room temperature in the dark. Cell nuclei were counterstained with DAPI. Fluorescence images were acquired using a fluorescence microscope (CKX53, Olympus, Japan) and analyzed using ImageJ software. Statistical analysis Data are presented as mean ± standard deviation. Data normality was assessed using the Shapiro-Wilk test. For data that did not conform to a normal distribution, non-parametric statistical analyses were performed using the Kruskal-Wallis test. Comparisons between two groups were performed using unpaired t-tests. For three or more groups, one-way ANOVA followed by Tukey's multiple comparison test was applied. When variance was unequal, Brown-Forsythe and Welch ANOVA with Dunnett's T3 test was used. All analyses were performed using GraphPad Prism 9 (USA), and only p values less than 0.05 are shown. Ethics approval and consent to participate All in vivo experiments used SPF-grade male SD rats (Chengdu Dossy Experimental Animals Co., Ltd.) and were approved by the Medical Ethics Committee of Southwest Jiaotong University (Approval No. SWJTU-2403-NSFC-053), in accordance with the National Research Council Guidelines for the Care and Use of Laboratory Animals. Data availability statement The data that support the findings of this study are available from the corresponding author upon reasonable request. CRediT authorship contribution statement Jinsheng Li: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Ming Yang: Data curation, Formal analysis, Investigation, Methodology. Guowen Duan: Data curation, Formal analysis, Investigation, Methodology. Jianhao Huang: Software, Visualization. Tao Shen: Software, Visualization. Jie Cui: Methodology, Validation. Keke Huang: Methodology, Validation. Qiting Ge: Methodology, Validation. Liming Yang: Methodology, Validation. Jingyi Sheng: Conceptualization. Jie Weng: Conceptualization, Writing – review & editing. Tailin Guo: Funding acquisition, Project administration, Resources. Zhihong Xu: Funding acquisition, Project administration, Resources. Jizi Liu: Funding acquisition, Methodology, Project administration, Resources, Validation, Writing – review & editing. Ning Gu: Funding acquisition, Project administration, Resources, Supervision. Qing Jiang: Funding acquisition, Project administration, Resources, Supervision. Declaration of competing interest The authors declare no competing financial or personal interests. The manuscript represents original work that has not been published previously and is not under consideration for publication elsewhere. All authors have approved the manuscript and agree with its submission to Bioactive Materials. Acknowledgements The research was supported by National Key Research and Development Project(2021YFA1201404),National Natural Science Foundation of China (92368201, 32071343), Basic Research Program of Jiangsu (BK20253053,BK20232023 ) Nanjing Municipal Science and Technology Bureau (202305001, 202205020), AI & AI for Science Project of Nanjing Universityand Nanjing Drum Tower Hospital Youth Cultivation Fund 2024-JCYJ-QP-87. Footnotes Peer review under the responsibility of editorial board of Bioactive Materials. Supplementary data to this article can be found online at https://doi.org/10.1016/j.bioactmat.2026.07.031. Contributor Information Tailin Guo, Email: tlguo@home.swjtu.edu.cn. Zhihong Xu, Email: Xuzhihongjoint@hotmail.com. Jizi Liu, Email: jzliu@nju.edu.cn. Ning Gu, Email: guning@nju.edu.cn. Qing Jiang, Email: qingj@nju.edu.cn. Supplementary data The following are the Supplementary data to this article: Multimedia component 1 Multimedia component 2 References Associated Data Supplementary Materials Multimedia component 1 Multimedia component 2 Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.