How the Sirt1-Piezo1 Pathway Helps Bone Growth and Healing in Mice

As an initial step to investigate potential role of the mechanosensitive ion channel Piezo1 in regulation of osteogenesis and fracture healing, we determined its expression in the process of fracture healing process. First, we examined the expression changes of Piezo1 during fracture healing. We first constructed a mouse model of intramedullary nail fixation of femoral fracture, and collected samples at 1/2/4 weeks after surgery. Results from IF staining showed that, 1 week after surgery, there was significant expression of Piezo1, Osx and Col2a1 in the cartilage callus area (Figure 1A‐D), and Piezo1 and Osx expression levels were increased during the healing process (Figure 1A‐D). We then established a mouse femur distractive osteogenesis (DO) model and analyzed the samples after 6 weeks of distraction. Results showed that the expression levels of Piezo1 in cartilage callus were higher than those in resting or proliferating chondrocytes in the growth plate, and were positively correlated with osteogenic protein (Figure 1A,E,F). Moreover, six weeks after DO, the expression levels of cartilage marker protein decreased as osteogenic protein levels increased (Figure 1A,E,G). To further investigate the dynamic changes of Piezo1 expression during fracture callus formation, we quantified Piezo1 expression in chondrocytes at single‐cell resolution, we reanalyzed a previously published single‐cell sequencing dataset, GSE150291, which was derived from single‐cell RNA sequencing of non‐hematopoietic cells extracted from the bone marrow of 4‐month‐old mice. We characterized the differentiation of mesenchymal stem/stromal cells (MSCs) with RNA velocity (Figure S1A,B, Supporting Information). The Piezo family members Piezo1 and Piezo2 showed distinct expression patterns in chondrocytes (Figure S1C, Supporting Information). At single cell level, the expression of Piezo1 was highly and specifically enriched in chondrocyte lineage, while that of Piezo2 was absent in chondrocytes (Figure S1D,E, Supporting Information). We next analyzed the single‐cell RNA sequencing dataset GSE154247. In this study, femurs were harvested at post‐fracture day 14 (PFD14) and from uninjured controls. Single‐cell suspensions were prepared from both control and PFD14 bones, and non‐hematopoietic stromal cells were isolated for single‐cell RNA sequencing. As revealed by single cell profiling, the proportion of chondrocytes was dramatically expanded at post fracture day 14, when compared with that in control (Figure S1F‐H, Supporting Information). Furthermore, differential gene expression analysis showed 844 up‐regulated genes and 2489 down‐regulated genes in chondrocytes, including those involved in chondrocyte differentiation and ossification (Figure S1I, Supporting Information). The expression of endochondral ossification‐related genes, including those encoding osterix (Osx), Col1a1, Alp and Vegfa, was up‐regulated in chondrocytes after fracture (Figure S1J‐M, Supporting Information). Piezo1 expression was up‐regulated at post fracture day 14 (Figure S1N, Supporting Information). We confirmed that the expression of PIEZO1 was up‐regulated in human cartilage callus compared to that in the control tissue (Figure S1O,P, Supporting Information).

To further investigate whether Piezo1 is involved in fracture healing, we constructed femur fracture models in mice with Piezo1 deletion in chondrocytes. Four days after surgery, TAM (100 mg kg⁻¹ body weight) was injected into Piezo1^flox/flox; Aggrecan^CreERT2 mice to induce Cre expression to delete Piezo1 in chondrocytes (cKO) (Figure 2A). Sex‐ and age‐matched Piezo1^flox/flox mice injected with TAM were used as controls in this study. Results showed that Piezo1 expression was drastically reduced in callus chondrocytes of cKO mice, as revealed by IF staining (Figure 2B). X‐ray, µCT and 3D reconstruction analyses were performed on the fracture sites of mice at 4 weeks post fracture. Results showed that the cKO group had significantly smaller callus than that in the control group, (Figure 2C). When compared to control group, the callus in the cKO group was less mineralized and displayed larger unmineralized gaps (Figure 2D‐G). SO/FG staining showed that at 2 and 4 weeks post fracture, the cKO group had fewer bone trabeculae and more cartilage tissue (Figure 2H), as well as the lower proportion of Runx2 positive cells in callus (Figure 2I,J). Collectively, these results demonstrate that Piezo1 loss in chondrocytes impairs callus mineralization and delays fracture healing by primarily impairing the endochondral ossification.

We further investigated the role of Piezo1 in chondrocytes during DO process. We successfully established the DO model in mice. Stretching started 1 week post surgery and lasted for 1 week. TAM injection began 1 day before stretch to induce Cre expression and Piezo1 deletion in callus chondrocytes (Figure 2K). X‐ray analysis of stretch sites in mice was performed 0/2/6 weeks after stretch surgery (Figure 2L). Six weeks after surgery, samples were collected for µCT and 3D reconstruction analyses. cKO group showed reduced bone mass in the distraction gap (Figure 2M‐Q). cKO group had a large amount of fibrous tissue in the distraction gap, and the bone trabeculae did not close the gap (Figure 2R). Piezo1 ablation decreased the proportion of Osx‐, Ocn‐, Bmp2‐ and Vegf‐positive cells in the callus (Figure 2S‐W). Thus, we demonstrate that Piezo1 loss in chondrocytes inhibits mechanically induced osteogenesis in mice.

Piezo1 loss in chondrocytes increased the width of the growth plate (Figure , Supporting Information), inhibited the endochondral ossification of the growth plate (Figure , Supporting Information), and increased the osteoclast formation (Figure , Supporting Information). S2A‐E S2H‐K S2C,F,G

We next investigated the effect of Piezo1 activation in the callus on fracture healing. Yoda1 is a selective activator of Piezo1. We first utilized the femoral fracture model in 12‐week‐old C57BL/6 male mice. Yoda1 or equal volume of solvent was injected into the fracture gap, and samples were collected for analysis 4 weeks post fracture (Figure 3A). Results revealed that Yoda1 treatment promoted fracture healing by enhancing callus mineralization and increasing the trabecular bone volume and mineral density (Figure 3B‐I). Yoda1 increased expression of osteogenic proteins, such as Runx2 (Figure 3J,K). We also investigated the effect of Piezo1 activation on osteogenesis in the mouse DO model. Yoda1 or equal volume of solvent was applied to the distraction gap in 12‐week‐old C57BL/6 male mice, and samples were collected for analysis 45 days post operation (Figure 3L). Results showed that Yoda1 treatment increased callus mineralization, thickened bone trabeculae, accelerated closing of the fracture gap (Figure 3M‐R), and increased the expression of vascular formation related proteins (Figure 3S,T)

In our effort to explore post‐translational modification regulation of Piezo1 activation, we firstly predicted the presence of potential acetylation sites in Piezo1 molecule through Deep‐PLA website (http://deeppla.omicsbio.info/↗) and then studied the effects of the deacetylation of Piezo1 on its activity. The acetylation level of Piezo1 was elevated by inhibiting the activity of deacetylase, indicating the presence of acetylation modification in Piezo1 (Figure 4A). Next, we investigated the effect of the deacetylation activator Sirt1 on Piezo1 acetylation. We found that Sirt1 can interact with Piezo1 either directly or indirectly, and Sirt1 is capable of deacetylating Piezo1. (Figure 4B,C). Co‐IP experiments revealed that Sirt1 bound to Piezo1 (Figure 4D,E). SRT2104(SRT), a specific activator of Sirt1, reduced the acetylation of Piezo1 (Figure 4F). Moreover, we found that both overexpression and pharmacological activation of Sirt1 dramatically increased the level of Piezo1 protein in a dose‐dependent manner (Figure 4G,H). Based on the subcellular colocalization analysis technique in light microscopy,^[22, 23^] IF imaging revealed that colocalization of Piezo1 and Sirt1 (Figure 4I,J). We determined the effect of SRT on Piezo1‐mediated calcium flow. The results showed that both SRT and Yoda1 similarly induced cellular calcium flow, both of which was dramatically lower in calcium‐free ringer solution than those in calcium‐containing ringer solution (Figure 4K,L), suggesting that significant amount of the calcium flow induced by either activator is through the cell membrane. Importantly, siRNA knockdown of Piezo1 expression dramatically reduced SRT‐ or Yoda1‐induced elevations in calcium flow (Figure 4M,N). Furthermore, the increase of calcium flow induced by SRT, but not Yoda1, was inhibited by 3‐TYP (a Sirt deacetylase inhibitor), but not by Panobinostat (a pan‐HDAC deacetylase inhibitor) (Figure 4O‐Q), suggesting that the Sirt rather than HDAC pathway is involved in SRT activation of Piezo1. Next, we constructed the plasmid L1341K/L1344K with mutations at key sites of Piezo1 mechanical force transduction^[24^] and found that Yoda1 no longer activated Piezo1 calcium flow in cells transfected with L1341K/L1344K plasmid, while these Piezo1 mutations had no marked effect on SRT activated calcium flow (Figure 4R). These results suggest that the calcium flow caused by Piezo1 deacetylation is not via the conventional mechanical stress. We predicted the interaction between Piezo1 and Sirt1 and the structure of the complex using AlphaFold 2 and visualized it with ChimeraX, which demonstrated that Sirt1 forms stable hydrogen bonds with the lysine residues K1525 and K2174 in Piezo1. K1525, located near the beam, forms three hydrogen bonds with Sirt1, while K2174 is situated in the C‐terminal region, forms one hydrogen bond with Sirt1. Both the beam and C‐terminus are important for mechanical transduction of Piezo1 (Figure S3A, Supporting Information). We next determined whether Sirt1 can deacetylate these specific lysine residues in Piezo1. Through multiple sequence alignment analyses, we found that K2174 of Piezo1 is highly conserved across different species, while K1525 shows relatively poor conservation (Figure S3B, Supporting Information). In protein acetylation studies, lysine mutations to arginine can maintain deacetylation, whereas mutations to glutamine preserve a neutral state. Next, we performed structural superposition and sequence alignment of Piezo1 with lysine mutations at both sites or Sirt1 intervention using ChimeraX 1.18. The results indicated that Sirt1 binding to Piezo1 can induce spatial conformational changes in the central pore (Figures S3C and S4, Supporting Information). After the K1525R/Q mutations, significant conformational changes were observed in the central pore of Piezo1. K1525Q, when combined with Sirt1, also led to noticeable changes in the central pore's conformation. In contrast, no significant conformational changes were observed in Piezo1 after Sirt1 intervention alone, suggesting that the K1525Q mutation does not affect Sirt1's influence on Piezo1 activity, indicating that K1525 may not be the key site for Sirt1 modulation of Piezo1 activation (Figures S3C and S4, Supporting Information). Following the K2147R mutation, a significant change in the central pore's spatial conformation was observed in Piezo1, whereas no significant change occurred after K2174Q mutation. Moreover, when K2174Q was combined with Sirt1 intervention, only minor conformational changes were observed compared to Piezo1 or Piezo1 with Sirt1 intervention alone. This suggests that K2174 may be the key site for Sirt1‐mediated activation of Piezo1(Figures S3C and S4, Supporting Information).

In addition, our results suggest that Piezo1 activation suppresses Sirt1 expression (Figure , Supporting Information). These findings may reflect the complex regulatory mechanisms underlying fracture healing. S5A

Given the critical role of Sirt1 in activating Piezo1 as demonstrated above, we further investigated whether Piezo1 is required for Sirt1 promotion of bone healing. SRT2104 (SRT) is a highly specific Sirt1 activator that has been demonstrated to have favorable therapeutic effects on various conditions, such as neurodegenerative diseases, cardiovascular diseases, musculoskeletal diseases, and other diseases.^[25^] The therapeutic use of SRT is restricted in part due to its poor solubility in water, which largely limits its bioavailability. Oral administration is the most widely used route to treat various diseases due to its convenience, high patient acceptance, cost‐effectiveness, and high safety.^[26^] To overcome the limitations of traditional oral formulations, such as short residence time in the digestive system, limited targeting ability, and water insolubility and low bioavailability of biologic drugs, based on our previous studies,^[21^] we have developed a yeast drug delivery platform that can be orally administrated and selectively targeted to the inflammatory fracture sites as a high effective payload and sustainable release for fracture treatment. The oral YC‐SRT was constructed in three main sequential modification steps: cationic nanoparticles (NP) synthesis, NP self‐deposition, and surface enteric coating (Figure S6A, Supporting Information). Initially, yeast microcapsules (YCs) with ≈5.0 µm in diameter were purified from S. cerevisiae by acidic and alkaline extraction to remove the cytoplasm and cell‐wall polysaccharides. Next, cationic PEI‐SRT NPs were prepared by self‐assembly of bPEI and SRT, and their shape, size and ζ‐potentials were determined using SEM and dynamic light scattering (DLS), respectively (Figure S6B‐F, Supporting Information). To accomplish the drug loading to YCs, cationic PEI‐SRT NPs were loaded into negatively charged YCs by electrostatic deposition. Then, the YC‐SRT was characterized by confocal fluorescence imaging using cyanine (Cy5.5)‐labeled SRT. The merged images verified that PEI‐SRT loaded cationic nanoparticles were efficiently packaged into YCs (Figure S6G, Supporting Information). The encapsulation efficiency of the YC‐SRT was 50.9%, and the loading contents of the YC‐SRT was 31.7 µg mg⁻¹ (Figure S6H, Supporting Information). These results indicate that YCs are high‐efficient carriers for SRT. We further determined the drug release curves of YC and YC‐SRT at pH 6.6 (simulating the acidic microenvironment of the fracture site) and pH 7.4 (simulating the normal bone microenvironment), respectively. YC‐SRT exhibited a relatively stable drug release (≈6%) within 100 h at pH 7.4. In acidic environments (pH 6.6), it exhibited a more rapid and effective drug release (≈12%) (Figure S6I, Supporting Information). We next determined the effect of YC‐SRT on fracture healing in mice. YC‐SRT were given orally post operation and analyzed 4 weeks post operation. YC‐SRT did not affect the weight and histology of major organs in mice (Figure S7, Supporting Information). Importantly, the fluorescence signal of YC‐SRT was detected only at the fracture site, which peaked on the 7th day and maintained for at least 14 days (Figure 5A,B). These results demonstrate that targeted delivery of SRT to the fracture site via oral yeast microcapsules has high efficiency and sustained release.

We next utilized our novel deliver system developed above to determine the effect of Sirt1 activation on fracture healing in mice; if so, does this require Piezo1 in chondrocytes. To this end, control and cKO mice were randomly divided into 2 groups for fracture surgery and intervention, and samples were collected 4 weeks later (Figure 5C,D). Results showed that YC‐SRT increased in the callus mineralization, bone mineral density, trabecular thickness and density, and the expression levels of osteogenic genes, when compared to those in untreated control mice. Piezo1 loss reduced the values of all above parameters (Figure 5E‐I). Importantly, the effect of SRT on promoting fracture healing was largely attenuated by Piezo1 deficiency in mice (Figure 5E‐I). At the molecular level, Sirt1 activation increased expression of Runx2 and Osx, which was abolished by Piezo1 deficiency (Figure 5J‐N).

Resveratrol (RSV) is a natural activator of Sirt1, and its efficacy, safety and pharmacokinetics have been well documented in more than 244 clinical trials.^[27^] In view of the ameliorative effect of RSV on bone mass loss, we first examined the intervention effect of RSV on bone healing. Due to the rapid metabolism and poor bioavailability of RSV, which limits its therapeutic use,^[27^] we developed a yeast microcapsule‐based drug delivery platform that can be orally absorbed and targeted at fracture sites, which is closer to clinical application. Oral YC‐RSV was constructed as described in Figure S6J (Supporting Information). Briefly, cationic PEI‐RSV NPs were firstly prepared by self‐assembly of bPEI and RSV, and their shape, size and ζ‐potentials were determined (Figure S6K‐O, Supporting Information). The cationic PEI‐RSV NPs were then loaded into negatively charged YCs by electrostatic deposition. Afterward, the YC‐RSV was characterized by confocal fluorescence imaging showing that PEI‐RSV loaded cationic nanoparticles was efficiently packaged into YCs (Figure S6P, Supporting Information). The encapsulation efficiency of the YC‐RSV was 73.1%, and the loading contents of the YC‐RSV was 51.2 µg mg⁻¹ (Figure S6Q, Supporting Information). YC‐RSV exhibited relatively stable drug release (≈7%) within 100 h at pH 7.4. In acidic environments (pH 6.6), it exhibited more rapid and effective drug release (≈13.5%) (Figure S6R, Supporting Information). Finally, we determined the effect of RSV on fracture healing in mice. YC‐RSV was given orally post operation and analyzed 4 weeks post operation. Neither YC nor YC‐RSV displayed any detrimental effects on major organs, including heart, liver, spleen, lung and kidney (Figure S7, Supporting Information). While YC alone had no significant effect on fracture healing, YC‐RSV promoted fracture healing by promoting the callus mineralization and largely increased osteogenesis across the fracture space (Figure 6A‐G). RSV treatment increased expression of Osx and Runx2 in the callus in mice (Figure 6H‐J). Notably, RSV also elevated the Sirt1 expression in the callus (Figure 6H,K).

In view of the promoting effect of oral administration of RSV on fracture healing, we finally examined the intervention effect of local administration of RSV into the fracture site on fracture healing. We developed an injectable drug loaded hydrogel platform as a high payload and sustainable release for fracture treatment. In the presence of H₂O₂ and HRP, the oxidative polymerization of RSV with catechol groups in Gelda molecules formed RSV@Gelda (gelatin‐dopamine hydrogel‐based delivery system loaded with resveratrol) (Figure 7A). We first synthesized Gelda by grafting dopamine groups onto the gelatin skeleton. 1H nuclear magnetic resonance (1H NMR) observed a specific peak of 6.5–7 ppm in Gelda and dopamine, but not in the gelatin spectrum, indicating that dopamine was successfully grafted onto the gelatin skeleton (Figure 7B). As shown in Figure 7C, Gelda remained liquid after dissolution and flowed when tilted. It formed a gel after brief mixing with HRP and H₂O₂, and did not flow when tilted (Figure 7C). Results from SEM revealed that both RSV@Gelda and Gelda hydrogels showed porous structures, but RSV@Gelda had a denser microscopic pore structure (Figure 7D). We then analyzed the surface functional groups of the hydrogels by FTIR spectroscopy. For all hydrogels, peak values were observed at 3270 cm⁻¹ (phenol O‐H stretching) and 1245 cm⁻¹ (amine C‐N stretching). At 1155 cm⁻¹ band, phenolic compounds containing carbonyl group was found. The formation of the (HN‐CO) bond was shown at 1631 cm⁻¹ band, confirming that dopamine forms a chemical bond with gelatin^[28, 29^] (Figure 7E). The freeze‐dried RSV@Gelda hydrogel significantly reduced the swelling rate compared with GelDa hydrogel, indicating that RSV crosslinking promoted the hydrogel to have a denser internal structure (Figure 7F). Lines of evidence have shown that ROS is overproduced in bone defect region.^[30^] To simulate the release environment of hydrogels in vivo, in addition to detecting the RSV release efficiency of hydrogels in PBS, we also studied the effect of ROS on the RSV release of hydrogels. Sustained drug release experiments showed that RSV@Gelda hydrogel could slowly release RSV for up to 28 days without being affected by H₂O₂ (Figure 7G).

We finally determined the effect of local administration of the RSV@Gelda on fracture healing in mice. Gelda and RSV@Gelda were locally injected to the fracture site during the operation and analyzed 4 weeks after operation. Both Gelda and RSV@Gelda did not exhibit any organ and cell toxicity (Figure S8, Supporting Information). The hydrogel alone had no marked effect on fracture healing. Importantly, RSV@Gelda promoted expression of osteogenic proteins and fracture healing by promoting the callus mineralization and increasing trabeculae volume across the fracture gap in mice (Figure 7H‐N).

To characterize gene expression at single cell level, two published single cell RNA sequencing (scRNA‐seq) dataset, GSE150291 and GSE154247 were investigated.^[44, 45^] The raw sequencing data at FASTQ format was downloaded, and was subsequently processed with Cell Ranger for expression matrices. For RNA velocity, cells of MSC, OLC, and chondrocyte were adopted in scVelo analysis.

Six patients who underwent internal fixation surgery for fractures at the Affiliated Hospital of Shandong University of Traditional Chinese Medicine 2‐6 weeks after the fracture due to displacement or non‐union of the fracture, callus tissues and paired bone tissues excised during surgery were collected, and the information of the specimens was shown in Table(Supporting Information). All specimens obtained patient consent and approval from the Medical Ethics Committee of the Affiliated Hospital of Shandong University of Traditional Chinese Medicine (2024‐165‐KY). S2

Piezo1^flox/flox mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Aggrecan^CreERT2 mice was previously described.^[34^] Both mice were in stable C57BL/6 background for more than ten generations. It crossed Piezo1^flox/flox mice with Aggrecan^CreERT2 mice to generate Piezo1^flox/flox; Aggrecan^CreERT2 mice to delete Piezo1 in chondrocytes as previously described.^[46^] In vivo studies were conducted on 12‐week‐old mice. Piezo1^flox/flox; Aggrecan^CreERT2 mice were intraperitoneally injected with tamoxifen (TAM, 1 mg/10 g body weight; Sigma–Aldrich) 4 days after fracture surgery or 7 days after DO surgery^[47, 48^] for 5 consecutive days to delete Piezo1 in chondrocytes,^[49^] which as conditional knockout (cKO) group. Piezo1^flox/flox mice were used as the control group, and an equal amount of TAM were injected at the same time. To minimize animal use and experimental variability,^[50, 51^] male mice was used in this study to avoid confounding effects of estrogen fluctuations on fracture healing, and all mice were housed under standard conditions in the specific pathogen free (SPF) experimental animal center of the Southern University of Science and Technology (SUSTech). All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of SUSTech (SUSTech‐JY202203002‐202408A1). All relevant guidelines for mouse experiments were adhered to in this study.

Baker's yeast (S. cerevisiae) was provided by Leadfree Management Co. Ltd. (Shanghai, China). Resveratrol (RSV) and SRT2104 (SRT) were obtained from the MedChemExpress (MCE). The commercial Cy5‐NHS ester and branched polyethyleneimine (bPEI, molecular (weight = 25 kDa) were purchased from Sigma–Aldrich (Missouri, USA). All other chemicals and reagents used were of analytical grade. Hollow YCs were purified from baker's yeast through alkaline and acid extraction methods.^[52^] As our previously described,^[21^] baker's yeast was dissolved in 1 M sodium hydroxide (NaOH), and the solution was heated at 80 °C for 1 h. After centrifugation at 3000 g for 10 min, the pellet was rinsed twice with deionized water. Then, the sample was dispersed in hydrochloric acid (HCl) solution at pH 4.0 and in curated at 60 °C for 1 h. Subsequently, the obtained sample was rinsed with isopropyl alcohol and acetone, and the YCs were collected and dried at room temperature. Cationic NPs were prepared according to the self‐assembly method described previously.^[53^] In short, RSV or SRT was mixed with bPEI in anhydrous dimethyl sulfoxide (DMSO) at a weight concentration of 10 mg mL⁻¹ for 2 h, and then dialyzed with deionized water at 37 °C for 24 h to obtain the PEI‐SRT/RSV cationic nanoparticles.

First, 10 mg of YCs and 100 µL of PEI‐RSV/SRT solution were resuspended in deionized water. The suspension was poured into a 5 mL centrifuge tube and heated in a 37 °C water bath for 2 h to obtain the YC‐RSV/SRT. After the reaction, unloaded PEI‐RSV/SRT was removed by centrifugation and washing with deionized water. The PEI‐RSV/SRT was cleaned up again after overnight at 4 °C. Considering oral administration, glycerin was added to YC‐RSV/SRT for the purpose of lubricating the esophagus. To further identify of YC‐RSV/SRT, SEM and TEM images of the YC‐RSV/SRT were obtained using SUPRA55 instrument (Zeiss, Germany) and JEM‐1400Plus microscope (JEOL, Japan). RSV or Cy5.5 probe labeled SRT was imaged and analyzed using a STELLARIS 5 confocal microscope (LEICA, Germany) to further verify that RSV/SRT was successfully loaded into YCs. Meanwhile, the particle size and zeta potential of YC‐RSV or YC‐SRT as well as PEI‐RSV and PEI‐SRT particles were determined using a using Malvern Zetasizer Nano ZS instrument. Next, the drug release profile of YC‐RSV/SRT was evaluated in a simulated normal(pH6.6) and fracture site microenvironment^[54, 55, 56^] (pH 7.4) added with 0.2% Triton X‐100 surfactant. The concentration of SRT and RSV in the buffer was quantified by fluorescence spectroscopy and UV spectrophotometry, respectively.

Gelatin‐Dopamine (Gelda) was synthesized as previously described.^[57^] First, 10 g of gelatin (Sigma–Aldrich) was dissolved in 1 L of deionized water at 40 °C until it was completely dissolved under N₂. N‐(3‐dimethylaminopropyl)‐N'‐ethylcarbodiimide hydrochloride (EDC, Aladdin) (4 g, 20.87 mmol), N‐hydroxysuccinimide (NHS, Aladdin) (4 g, 34.76 mmol), and dopamine hydrochloride (Sigma–Aldrich) (6 g, 31.64 mmol) were added to the gelatin solution successively, and the chemical reaction was carried out for 6 h at 50 °C and pH ≈5. Then the reaction solution was put into a dialysis bag (MWCO = 8000–14000 Da) and immersed in aqueous solution of NaCl (20 × 10⁻³ M) and HCl (833 × 10⁻⁶ M) at 40 °C for 3 days to remove by‐products such as EDC, NHS and dopamine with molecules smaller than 8 kDa, and then soaked in deionized water for 3 h to remove NaCl. The purified Gelda was collected and freeze‐dried for 3–5 days, and the final product was stored at −20 °C. The molecular structure of Gelda was confirmed by ¹H‐NMR (Bruker Ascend 600).

At 37 °C, Gelda (150 mg) was dissolved in PBS (0.7 mL) and then added to resveratrol (RSV) (0.1 mL) solution for polymerization. Then the polymer solution was fully mixed with HRP(Aladdin) (0.1 mL, 3.2 mg·mL⁻¹) solution and H₂O₂ solution (0.1 mL, 1–5 wt.%) and stood for hydrogel crosslinking.^[58^] The gelation time was determined by vial tilting method. The final concentrations of Gelda, HRP, H₂O₂ and RSV were 15 wt.%, 0.32mg·mL⁻¹, 0.1–0.5 wt.% and 1–10 mM, respectively. The structures and compositions of the hydrogels were detected by Fourier transform infrared spectroscopy (FTIR, Thermo‐Nicolet Nexus 670 ATR‐IR spectrometer). The hydrogel samples were prepared and swollen to saturation, then freeze‐dried and sliced for microstructure analysis. Field emission SEM and energy dispersive X‐ray spectroscopy were used to observed and photographed the microstructure of Gelda hydrogel and RSV@Gelda hydrogel after sputtering gold. The freeze‐dried hydrogel samples were weighed and recorded as W₀, and then immersed in PBS water at 37 °C for 24 h, using filter paper to remove excess surface moisture and the weight of each swelling sample was measured and recorded it as W_t. The swelling capacity of the hydrogel was assessed as: Swelling ratio = (W_t ‐W₀)/ W₀ × 100 %. To test the release time‐concentration curve of RSV, 0.2 mL RSV@Gelda hydrogel was soaked in 2 mL PBS and placed it on a shaker at 37 °C. At the scheduled time, PBS extracts were collected and replaced with fresh PBS for sustained release. The concentration of RSV in solution was determined by High Performance Liquid Chromatography (HPLC).

The surgery was performed as previously described.^[59^] Briefly, after anesthetizing the mice, the left femur was sterilized, then a transverse fracture was formed at the mid‐shaft with a wire saw, and the fracture was fixed by inserting a sterilized wire into the medullary cavity. Micro‐computed tomography (µCT) was performed at every week and samples were taken 4 weeks post‐fracture for histological analysis.

The mouse model of DO of the femur was modified according to the previously reported protocols.^[60^] In brief, after being anaesthetized with isoflurane, all animals were subjected to a right femur transverse osteotomy procedure at the mid‐shaft near the fibula‐tibia junction under sterile condition. An external fixator (RISYSTEM AG, SWITZERLAND) was applied to fix proximal and distal segments of the osteotomy site. An osteotomy was also made in the right fibula. Surgical incisions were then sutured sequentially. All mice were randomized into two groups with 6 in each group. All groups had bone lengthening started at day 7 postoperatively with 0.25 mm increments every 12 h, twice a day for seven days, for a total lengthening distance of 3.50 mm. Animal wellbeing and condition of surgical site were checked every three days. Samples were taken 6 weeks post‐surgery for histological analysis.

The mice were given orally 100 mg kg⁻¹ YC‐SRT or YC‐RSV 1 day post fracture operation, once every 2 days, for three consecutive times. Four weeks post‐fracture, euthanize the mice for histological analysis. The other group injected with the same amount of PBS or YC were used as the control.

The mice were given 500 ul GelDa or RSV@GelDa during fracture surgery. 4 weeks post‐fracture, euthanize the mice for histological analysis. The other group injected with the same amount of resveratrol free hydrogel (GelDa) were used as the control. The callus region injection of Yoda1 or the same amount of solvent were performed as previously described,^[34^] 50 µg mL⁻¹, 2 µL/time, once every two days, for 5 consecutive times, all the injections were given on the 4 days post fracture or 7 days post DO surgery.

Samples were harvested at different time points post‐fracture or distraction, and fixed in 4% paraformaldehyde (PFA). Micro‐CT analyses were performed using a high‐resolution µCT scanner (Bruker SkyScan 1176, Karlsruhe, Germany). The 3D reconstruction of the fracture callus was performed using CT‐Vox 2.1. Three‐dimensional morphological parameters were recorded and calculated using CTAN 1.12 software (Bruker).^[61^]

Femur samples were fixed in 4% PFA at 4 °C for 24 h, decalcified with 10% EDTA (pH 7.4) for 3 weeks at 4 °C and paraffin‐embedded and 5‐µm thick sagittal sections were cut. Hematoxylin‐eosin (H/E), Safranin O‐Fast green (SO/FG) and TRAP staining was performed as previously described.^[62, 63, 64^]

Specimens were prepared as previously described.^[65, 66^] For immunohistochemistry (IHC), After antigen‐retrieved, BSA blocking and overnight incubation of the primary antibody, and incubated with the secondary antibody coupled with horseradish peroxidase (HRP), stained with DAB (Abcam), and restained with hematoxylin. Finally, images were taken under a microscope. For immunofluorescence (IF), After antigen‐retrieved overnight, permeabilized with 0.2% Triton X‐100, blocked with 2% BSA blocking and overnight incubation of the primary antibody, and incubated with anti‐rabbit Alexa Fluor 488 (Invitrogen) or anti‐mouse Alexa Fluor 568 (Invitrogen) secondary antibodies for 1 h at room temperature. The fluorescence signal was determined using a confocal microscope (Leica SP8 Confocal Microscopy System). Quantitative analysis of IHC and IF staining was performed in a double‐blind manner.

Before the intervention of all compounds in cells, cell activity tests were conducted to screen for safe concentrations for use. The detection method was as previously described.^[67^]

Small interfering RNA (siRNA) oligonucleotides and untargeted interfering control siRNA for mouse Piezo1 were purchased from GenePharma. The cell transfection method is as described above.^[68^] The oligonucleotide sequence of Piezo1 siRNA is: For: GCUUGCUAGAACUUCACGUTT; Rev: ACGUGAAGUUCUAGCAAGCTT.

HEK293T cells were seeded in a six‐cell plate, and after 60% fusion, 2.5 µg plasmid was transfected with lipo3000 (Thermo Fisher Scientific, Inc.) in Opti MEM (Thermo Fisher Scientific, Inc.). Eight hours after transfection, culture with fresh culture medium and then continue for 48 h before corresponding treatment.

The mouse Piezo1 plasmid was purchased from MIAOLING BIOLOGY. Site‐directed mutagenesis of human and mouse Piezo1 was undertaken using a custom protocol with the high‐fidelity polymerase.

Fluid‐induced flow shear stress (FSS) was performed with Streamer System STR‐4000 (Flexcell International Corporation, Burlington, NC, USA) as previously described.^[69^] For FSS treatment group, the culture slips were transferred into a parallel plat flow chamber and cells were exposed to 5 dyne cm⁻² fluid flow for 2 h. For static controls, ATDC5 were kept in incubator without any further treatment. Cell and protein samples were collected after FSS treatment.

Co‐immunoprecipitation (Co‐IP) assay was performed as previously described.^[45^] Briefly, Cells were transfected with corresponding expression plasmids. After 24 h, the cells were subjected to IP lysis buffer (Thermo Fisher) containing protease inhibitor cocktail and TSA/NAM to extract protein supernatant, and then incubated overnight with the corresponding primary antibody and Protein A/G magnetic beads at 4 °C. The Dynabheads‐antigen‐antibody complex was collected using Magnetic Stand (Thermo Fisher). The complex was washed with IP buffer for three times, re‐suspended with 1×loading buffer, boiled at 95 °C for 5 min, and perform Western blotting.

Western blot analysis was performed as previously described.^[70^] Extract cellular proteins, sequentially undergo electrophoresis, membrane transfer, incubation with primary and secondary antibodies, and detect protein expression with enhanced chemiluminescence kit (ECL kit, Bio‐Rad, USA).

Cell fluorescence was measured under excitation light of 340 and 380 nm as described.^[34^] The fold change of [Ca²⁺]i was expressed as the 340/380 ratio before and after treatment. Cells with a [Ca²⁺]i fold change greater than 1.05 were defined as responsive cells.

AlphaFold was capable of predicting the combined structures of complexes, including proteins, nucleic acids, small molecules, ions, and modified residues, triggering a revolution in the modeling of protein structures and their interactions.^[71^] It first used AlphaFold 2 to predict the structure of Piezo1 under different intervention conditions, then employed ChimeraX 1.9 for protein structure visualization to analyze the positions and lengths of hydrogen bonds. Additionally, Chimera 1.18 was used to perform structural superimposition and sequence alignment of different Piezo1 conformations to analyze the potential molecular mechanism by which Sirt1 deacetylation activates Piezo1.^[72^]

All experiments were performed at least three times. All data were presented as mean ± standard deviation and analyzed or plotted using GraphPad Prism 8.0. Differences between the two groups were analyzed using Student's t‐test. For comparisons between more than two groups, one‐way analysis of variance (ANOVA) and Tukey's or Dunnett's multiple comparison test were used. The value of P < 0.05 was considered statistically significant.