A continuous adhesion-enhanced osteogenic pathway in artificial scaffold drives cellular infiltration and condensed mineralization for rapid bone regeneration

Abstract

Enhancing the cellular infiltration and mineralization capacity of bone scaffolds can effectively address the challenges of bone nonunion and the prolonged osteogenic repair cycle, particularly in the treatment of critical-sized bone defects. Conventional bone scaffolds, whether composed of inorganic materials or fabricated via 3D-printed titanium alloy, frequently hinder seamless cellular integration due to inherent structural discontinuities, such as granular interfaces or layer-by-layer striations. Here, we address this limitation by employing graphene, not merely as a filler, but as a continuous surface modifier within a 3D scaffold. Through an in-situ reduction-induced phase separation technique, we engineered a long-range, frost-like graphene surface at a low graphene concentration of 3.4 wt% in fabricated scaffold. The resulted unique architecture establishes a continuous pathway for cell migration, leading to significantly enhanced cellular adhesion, accelerated infiltration, rapid calcium deposition and bone ingrowth. We demonstrate that these pro-osteogenic effects are mediated through the modulation of genetic pathways related to ion channels and cell-extracellular matrix interactions. Furthermore, the scaffolds show excellent biocompatibility, integrating seamlessly into nascent bone tissue without eliciting inflammation or immune rejection. Thus, this strategy of constructing continuous cell-migration surfaces presents a promising and scalable platform for the regeneration of critical-sized bone defects.

Graphical abstract

Highlights

• •A continuous frost-like graphene pathway was constructed via in-situ reduction-induced phase separation.
• •Enhanced cellular adhesion and deep infiltration enabled by continuous cell migration pathways.
• •Accelerated calcium deposition and osteogenic mineralization through ion channel and cell-ECM interaction modulation.
• •Controlled metabolic clearance and excellent biocompatibility of graphene enable safe bone integration.

1. Introduction

In clinical orthopedics, delayed union or nonunion constitutes the most prevalent issues in the repair of bone defects of critical size and the rapid repair of critically sized bone defects still remains a major challenge [1,2]. Due to limited cell infiltration capacity, bone scaffolds often exhibit non-uniform calcium deposition. This inadequate mineralization, combined with delayed bone regeneration characterized by the formation of a thin layer of new bone along the defect margins, contributes to prolonged healing and represents a major challenge in the repair of critically sized bone defects [3,4]. In addition, artificial bone scaffolds must exhibit exceptional osteoinductive potential and calcium deposition capacity. This facilitates the effective remodeling of the scaffolds via mineralized nodule formation, thereby accelerating bone mass augmentation and ultimately achieving optimal osseointegration [[5], [6], [7]].

There are primarily two approaches to enhance the cell infiltration capacity of artificial bone scaffolds. The first is to augment the adhesion and migratory capabilities of cells, while the second is to improve the connectivity and permeability of the pore structure of the scaffolds. Cell adhesion and migration of MSCs are influenced by many factors, such as topology, roughness, functional groups, hydrophobicity and mechanical properties of the substrate [8,9]. Notably, these factors influence each other, so determining the effect of individual factor is very complicated and difficult. However, we can still discuss the effect of individual factor to some extent. For instance, surfaces with an oriented structure can guide cell orientation, and surfaces with other topologies including nanodots and nanopores with suitable roughness can improve cell adhesion [10,11]. In addition, surfaces with cell adhesion molecules (CAM) such as GFOGER (Gly–Phe–Hydroxyproline–Gly–Glu–Arg), KRSR (Lys-Arg-Ser-Arg), and RGD (Arg-Gly-Asp) or metal ions can also promote cell adhesion via CAM interaction [[12], [13], [14], [15]]. Substrates exhibiting viscoelastic properties in the low modulus range are more conducive to promoting cell migration [16]. The connectivity of the pores within the scaffold and its permeability exert an influence on the migratory pathways of cells and the scaffold's capacity for blood supply. Thus, the pore walls of osteogenic scaffolds should possess excellent connectivity, enabling cells to crawl and distribute uniformly across the entire scaffold. Simultaneously, the scaffold architecture should be highly permeable to facilitate the exchange of nutrients, thereby providing a more favorable microenvironment for osteoblasts [17].

In addition, both robust osteogenic induction and significant calcium deposition capabilities on the scaffold surface are essential. The calcium nodules secreted by osteoblasts that have been induced to differentiate and exhibit mineralization potential can be effectively anchored onto the scaffold surface. Acting as a bridge between cells and the scaffold, calcium nodules enhance local stiffness and provide physical cues that further stimulate osteogenic activity through mechanotransduction pathways [18]. Moreover, calcium deposition contribute to ion exchange processes, releasing calcium and phosphate ions that modulate pH and promote additional mineral deposition—a self-reinforcing cycle that supports progressive scaffold maturation [19,20]. Calcium nodule enables gradual replacement of the synthetic structure with native-like bone tissue, aligning with the physiological process of bone healing [21]. Ultimately, this dynamic interplay between cells and the biomaterial lays the essential groundwork for achieving stable bone integration and maintaining long-term bone homeostasis, both of which are critical for the functional restoration of damaged or defective bone segments.

Different from traditional artificial bone inorganic materials, graphene, a flexible two-dimensional (2D) material with high strength and very large specific surface area (SSA), may provide an effective solution to improve the cellular infiltrative ability and mineralization capacity of the scaffolds for critically sized bone defect treatment. Although graphene has been well studied in the field of tissue engineering owing to the ease of surface treatment [22,23], the biological toxicity of graphene is still controversial and the ability to promote cell adhesion or migration is also not so clear so far [24]. According to literature report, biofriendly macromolecular coatings on graphene can improve the biocompatibility considerably and broaden the applications of graphene in tissue engineering scaffolds [25,26]. For instance, a systematic long-term in vivo graphene/polyethylene glycol (PEG) biodistribution was studied, and the toxicity of PEGylated graphene was found to be negligible at a dose of 20 mg/kg [27]. Also, upon using graphene oxide (GO) as the graphene precursor for tissue engineering scaffolds, the biological toxicity was found to be significantly decreased with increasing GO reduction [28]. Thus, high levels of GO reduction and low content of graphene are preferable. However, the modification method of graphene for excellent cell adhesion, migration and osteogenic induction is still challenging and the long-range graphene pathway is also difficult to be constructed in traditional fabrication methods of graphene hydrogels or traditional graphene doping methods with a low content of graphene.

In this study, a one-step in-situ reduction-induced phase separation technique was developed, enabling the creation of a continuous graphene channel network without the need for biotoxic cross-linking agents. Unlike previously reported graphene-based scaffolds, where graphene often acts merely as a passive filler [25,26,29], our approach transforms the reduction process itself into a driving force for structural self-assembly. A key innovation lies in the formation of an interconnected, continuous graphene channel network throughout the scaffold—achieved at a remarkably low graphene content of only 3.4 wt%. This architecture not only provides stable electrical and mechanical pathways but also produces interconnected macroscopic pores, resulting in a biomimetic structure highly similar to natural cancellous bone. Concurrently, at the microscopic scale, chitosan (CS) undergoes synchronous microphase separation on the surface of reduced graphene, leading to the formation of abundant chitosan nanodots within the microstructure. Compared to conventional scaffolds such as hydroxyapatite (HA) and 3D-printed titanium alloys, this scaffold's stable porous architecture integrated with a continuously modified graphene surface offers distinct advantages: it provides essential anchor points and pathways that enhance cell adhesion and enable rapid, continuous cell migration, while simultaneously supporting efficient nutrient exchange. More importantly, the continuous graphene network actively regulates cell signaling pathways related to adhesion and ion channels, thereby significantly accelerating osteogenic induction and bone ingrowth—a functional leap beyond the primarily structural role of traditional materials. As a result, the composite scaffold demonstrates exceptional calcium deposition and osseointegration capability, leading to the complete healing of critically sized bone defects. Additionally, it was found that the modified graphene seamlessly integrated into newly-formed bone tissues without inducing inflammation or rejection responses. Therefore, this kind of porous composite scaffolds with long-range continuous modified graphene surfaces represent a highly promising clinical candidate for bone tissue engineering applications.

2. Materials and methods

2.1. Materials

HA nanopowder (>97%, particle size <100 nm (BET)), CS with a viscosity of 200∼ 400 mPa s and sodium ascorbate (99%) were purchased from Aladdin (Shanghai, China). Graphene flakes (99 wt%, 300 mesh) were from XFNANO (Jiangsu, China). Polyvinyl alcohol (PVA) (>99%, Mw = 146, 000 ∼ 186, 000, hydrolyzed) was purchased from Sigma-Aldrich (USA).

2.2. Preparation of CPH/rGO series, HA and 3D printing scaffolds

First, GO was synthesized using a modified Hummer's method [30]. Subsequently, 0, 0.6, 1.2, 1.8, and 2.4 % (m/v) of GO along with 1 g of HA powder were dispersed in 5 mL of 4% (v/v) acetic acid solution. The mixture was subjected to ultrasonic treatment followed by thorough magnetic stirring to ensure uniform dispersion. Next, 5 mL of aqueous PVA solutions with concentrations of 2, 6, 10, and 14 % (m/v) were separately introduced into the well-dispersed suspension under continuous stirring. Following this, 0.2 g of CS was gradually added under magnetic stirring until complete dissolution, resulting in a viscous and homogeneous precursor solution. Thereafter, 0.3 g of sodium ascorbate was incorporated, and the entire system was stirred for an additional 10 min. The composite precursor was then sealed and incubated at 60 °C for 12 h to facilitate self-assembly into composite hydrogels. The hydrogels gradually forms as the reduction degree of GO deepens, and no longer changes after 12 h when GO is completely reduced. The formed hydrogels were washed repeatedly with distilled water to remove residual acetic acid and any by-products. Finally, the washed hydrogels were lyophilized and stored for further characterization and testing. The final scaffold was designated as CPH/rGO, derived from the initial letters of the constituent raw material names. Samples labeled as CPH/rGO-X/Y were named according to the final PVA and rGO content (m/v) in the precursor formulations. The preparation protocol for CPH/rGO-X/0 is identical to that of CPH/rGO-X/Y, except for the omission of GO and acetic acid.

The fabrication protocol of the porous HA ceramics were adapted from our prior work [31]. Briefly, the CS solution was prepared by dissolving 0.3 g of CS powder in 10 mL of deionized water acidified with 200 μL of glacial acetic acid under magnetic stirring for 1 h. Then, HA powders were incorporated into separate CS solutions and stirred for another 2 h to form uniform slurries. Commercial melamine foam cubes (3 × 3 × 10 mm³) were fully infiltrated with these slurries. After removing the excess, the impregnated scaffolds were vacuum-dried at 60 °C for 12 h and subsequently sintered at 1350 °C for 2 h in a tube furnace to remove the polymer and consolidate the ceramic. The final ceramics were designated as HA scaffold. A photocurable slurry was prepared by mixing a bio-based photosensitive polylactic acid resin with 5 wt% HA powder, followed by high-speed stirring for 24 h. Using this slurry, 3D printing scaffolds were fabricated. The process involved first designing the corresponding 3D models and then printing them via a DLP-3D printer (nanoArch P150, BMF Precision Tech, China).

2.3. Preparation of rGO and rGO/CS plates

To facilitate the in vitro investigation and comparison of cell migration and calcium deposition behaviors on graphene surface decorated with CS nanodots, we also prepared the rGO and rGO/CS plates. Briefly, 0.1 wt% GO solution with a volume ratio of ethanol to water 3:7 was spin coated on the plastic plates at 3000 r/min. After drying at 60 °C, 0.1 wt% CS aqueous solution was added onto the GO plates for the reaction between GO and CS. Excess CS was washed with water after 1 h. Afterwards, the modified GO plates were reduced with 2 wt% sodium ascorbate under 60 °C for 12 h to get the rGO/CS plates. rGO plates were prepared with the same method for rGO/CS plates without using CS.

2.4. Characterizations

The morphology of the composite scaffolds was characterized using a JOEL JSM-5900LV field-emission scanning electron microscope (FESEM, Japan) at an accelerating voltage of 5 kV and EDS mapping was performed using Ultim Extreme (Oxford Instruments) at the voltage of 10 kV. The specific surface area (SSA) and pore size distribution of the composite scaffolds were obtained by BET measurements together with N₂ adsorption/desorption analyses conducted at 77 K using an Autosorb iQ/ASiQ adsorption analyzer (Quan-tachrome, USA). Dynamic rheological tests were carried out using a rotational rheometer (AR2000EX, TA instruments, USA) with the parallel-plate geometry (diameter of 25 mm) at 25 °C to evaluate the mechanical properties of the scaffolds. Fourier transform infrared (FTIR) spectra were obtained with a Thermo Nicolet 6700 FTIR spectrometer (Madison, WI, USA) at a resolution of 4 cm⁻¹ in reflection mode. Wide angle X-ray diffraction (WAXD) test was performed with a DX-1000 X-ray diffractometer (Dandong Fanyuan Instrument Co. LTD, China) at room temperature. Samples were scanned in diffraction angle 2θ = 3−80° at 2°/min using CuKα radiation (λ = 0.154056 nm) with a filament voltage of 40 kV and a current of 40 mA. Raman spectra were obtained using a LabRAM HR Raman Spectrometer (HORIBA Jobin-Yvon, France) with a laser at the excitation wavelength of 532 nm. XPS characterization was conducted using an XSAM800 spectrometer (Kratos Company, UK) with Al Kα radiation (hv = 1486.6 eV). AFM was carried out using an Icon AFM (Bruker Corporation, USA). The crystalline structures of newly-formed bone were measured by polarizing optical microscopy (POM, Olympus BX51, Japan) with a MicroPublisher 5.0 RTV CCD. TEM, HRTEM images and SAED patterns of calcium nodule on graphene sheets were obtained through TEM (Tecnai G2 F20 S-TWIN, FEI Company, USA). The cytoskeleton, cellular morphology and stained functional protein were observed using a laser confocal microscope (A1RMP+, Nikon, Japan). Cell migration was recorded by the time-lapse photography of the laser confocal microscope under the living cell system. The microplate reader (Eon, Biotek, USA) was employed to quantitatively determine the cell proliferation.

2.5. Migration of hMSC

The migration of hMSCs was examined through in-situ observation and the transwell permeability assay. hMSCs were seeded onto substrates of rGO, rGO/CS and uncoated tissue culture polystyrene as the blank control, respectively. After 24 h to allow for stable cell adhesion, samples were transferred to a confocal live-cell imaging system for time-lapse microscopy. Images were acquired at 60-s intervals over an 8-h period under controlled environmental conditions (37 °C, 5% CO₂). Cell migratory behavior was further evaluated using a Transwell permeability assay (diameter: 6.5 mm, pore size: 8 μm, polycarbonate membrane, Corning, USA). Cylindrical scaffolds—CPH/rGO-3/0, CPH/rGO-3/0.6, and 3D-printed constructs with distinct structural designs—were carefully placed on the bottom surface of the Transwell inserts for migration assessment. A suspension of 1 × 10⁵ hMSCs in 100 μL α-minimum essential medium (α-MEM, Gibco, USA) was loaded into the upper chamber, while 750 μL of complete growth medium was added to the lower well to establish a chemotactic gradient. Following a 24-h incubation period, hMSCs that had migrated through the scaffold matrix and reached the underside of the Transwell insert were fixed with paraformaldehyde and stained with Crystal Violet–Ammonium Oxalate Solution (1%, w/v, Solarbio, Beijing, China) for visualization and quantification.

2.6. In vitro studies of hMSCs and BMSCs

RAW cells Purchased from iCell bioscience Inc. (Shanghai China). hMSCs (OriCell® # HUXMA-01001) purchased from Cyagen Biosciences (Guangzhou, China) were used to detect the osteogenic induction capability of our scaffolds. We isolated primary BMSCs from bone marrow of rats. hMSCs were cultured on CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds in a 24-well plate. The cell proliferation assay was performed in the α-minimum essential medium (α-MEM, Gibco, USA) with 10 % fetal bovine serum (FBS, Gibco, USA) and 1 % antimicrobial of penicillin. The hMSCs proliferation in each well was also quantitatively determined after seeding for 1, 3, 5 days through CCK-8 Cell Proliferation and Cytotoxicity Assay Kit (Solarbio, Beijing, China).

The osteogenic induction medium was prepared with α-minimum essential medium (α-MEM, Gibco, USA) with 10 % fetal bovine serum (FBS, Gibco, USA), 1 % antimicrobial of penicillin, 8 nM β-Glycerol phosphate disodium (≧98 %, Solarbio, Beijing, China), 1 × 10⁻⁸ dexamethasone (≧98 %, Solarbio, Beijing, China) and 2 × 10⁻⁴ Vitamin C (≧ 99 %, Solarbio, Beijing, China), and osteogenic medium was changed every 2 days. After induction onto different plates and scaffolds, the hMSCs and BMSCs on plates or scaffolds were fixed for further test. For a better understanding of the calcium deposits and osteogenic differentiation of hMSCs after the induction, a calcium stain kit (Modified Alizarin Red S Method, Solarbio, Beijing, China) was employed to evaluate the calcium deposits of hMSCs on the plates and BCIP/NBT Alkaline Phosphatase Color Development Kit (Beyotime, Shanghai, China) was used to assess the ALP content of hMSCs.

Morphology of MSCs on scaffolds and plates was characterized by laser confocal microscope. hMSCs were seeded on the collagen coated cell slides with different scaffolds. After cell adhesion, hMSCs were stained by DAPI solution (10 μg/ml, ready-to-use, Solarbio, Beijing, China) and FITC-Phalloidin (Solarbio, Beijing, China) to investigate the morphology of the cells. The staining of OCN was performed with polyclonal antibody to osteocalcin (Proteintech, USA) and goat anti-rabbit IgG H&L (Abcam, USA) after the blocking of bovine serum albumin (BSA, BioFroxx, Germany). The integrated optical density of OCN was measured using Image J image analysis software.

For RT-qPCR, the total RNA was extracted using TRIzolTM (Invitrogen, USA) solution according to the manufacturer's protocol. Complementary DNA was synthesized by using the HiScript III RT SuperMix for qPCR (Vazyme, USA) in accordance with user manuals. Then quantitative real-time PCR was performed in triplicate by using AceQ Universal SYBR qPCR Master Mix (Vazyme, USA) for PCR reactions on an iCycler Real-Time Detection System (BioRad, USA). The relative amount of mRNA was normalized to house-keeping gene Glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Primers used for RT-qPCR are listed as follows. GAPDH: forward, 5′-GCTCTCTGCTCCTCCTGTTCG-3′, reverse, 5′-GCGAACACAT CCGGCCTGC-3’; SP7: forward, 5′-TCTCCATCTGCCTGACTCCT-3′, reverse, 5′-AGCGTAT GGCTTCTTTGTGC-3’; RUNX2: forward, 5′-GACTGTGGTTACCGTCATGGC-3′, reverse, 5′-ACTTGGTTTTTCATAACAGCGGA-3’; ALP: forward, 5′-GACCTCCTCGGAAGACA CTC-3′, reverse, 5′-TGAAGGGCTTCTTGTCTGTG-3’; COL1: forward, 5′-TCTAGACATGTT CAGCTTTGTGGAC-3′, reverse, 5′-TCTGTACGCAGGTGATTGGTG-3’; OCN: forward, 5′-AGCAAAGGTGCAGCCTTTGT-3′, reverse, 5′-GCGCCTGGGTCTCTTCACT-3’; BSP: forward, 5′-CAGGCCACGATATTATCTTTACA-3′, reverse, 5′-CTCCTCTTCTTCCTCCTCCT C-3’; DMP1: forward, 5′-CAGGAGCACAGGAAAAGGAG-3′, reverse, 5′-CTGGTGGTATCTT GGGCACT-3’; OPN: forward, 5′- CTCCATTGACTCGAACGACTC-3′, reverse, 5′-CAGGTC TGCGAAACTTCTTAGAT-3’; IL-1β: forward, 5′-GTAATGAAAGACGGCACACCC-3′, reverse, 5′-CAGGCTTGTGCTCTGCTTGTG-3’; IL-6: forward, 5′-CCCCAATTTCCAATGCTCTCC-3′, reverse, 5′-CGCACTAGGTTTGCCGAGTA-3’; ARG1: forward, 5′-CTGGGGATTGGCAAGGTGAT-3′, reverse, 5′-CAGCCCGTCGACATCAAAG-3’; CD206: forward, 5′-GGAGTGGCAGGTGGCTTATG-3′, reverse, 5′-CACTGCTCGTAATCAGCCTCC-3’. Total RNA was extracted using TRIzol from hMSCs that had been co-cultured with scaffolds or cultured on rGO and rGO/CS surfaces for 7 days, and then sent for sequencing. RNA sequencing analysis was supported by Novogene.

Cultured cells were washed with PBS and lysed using RIPA buffer supplemented with a protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific, Hudson, NH, USA). The cell lysates were then separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred for Western blot analysis. Horseradish peroxidase (HRP)-conjugated secondary antibodies were applied, and the blots were developed with an ECL solution. Signal detection and analysis were performed using a Bio-Rad ChemiDoc imaging system (Bio-Rad, Hercules, CA, USA). Primary antibodies used are KCNN3 (Bioss, BS-11569R), Integrin β1 (Servicebio, GB115173), ANK3 (Proteintech, 27980-1-AP), FAK (Servicebio, GB115455), MAPK (GB115455, GB11811), OCN (Affnity, DF7738), BSP (ABclonal, A20300).

2.7. In vivo studies

All procedures in the experiment were approved by the Medical Ethics Committee of West China Stomatological Hospital Sichuan University (WCHSIRB-D-2025-253). Male Sprague−Dawley (SD) rats (220 ± 10 g) were obtained from Huafukang Biotechnology Co., Ltd (Beijing, China) and randomly assigned to four groups corresponding to HA group, 3D Printing group, CPH/rGO-3/0 group and CPH/rGO-3/0.6 group (n = 6 per group). Cylindrical defects with a diameter of 3 mm and a depth of 4 mm in medial femoral joint were performed using a grinding drill. Then, different samples with proper sizes were stuffed to the defects. 3 months after surgery, the rats were sacrificed, and the femurs were harvested for further characterization.

The repair of bone defects was first studied through micro-CT (Quantum GX, PerkinElmer, USA). 3D reconstruction and quantitative analysis of osteogenic parameters were supported through the software Avatar 1.6.9.3 (Pingsheng Medical Co., Ltd, Shanghai, China). Then, the section of the defects was exposed through the wire cutting and investigated with SEM. In addition, the specimens were embedded in paraffin after decalcification and cut along the femur longitudinal for fluorescence labeling, immunohistochemistry and histological observation. Hematoxylin-Eosin (Solarbio, Beijing, China) Stain Kit, Masson's Trichrome Stain Kit (Solarbio, Beijing, China), OCN monoclonal antibody (ThermoFisher Scientific, USA) FREM1 and KCNN3 antibody (Bioss Antibodies, Beijing, China), ANK3 antibody (Proteintech, Wuhan, China), KCNH1 (Servicecbio, Wuhan, China) were used to stain the slice of specimens.

2.8. Simulation

Simulation of molecular dynamics was performed on the CASTEP module of MS2019 to evaluate the interaction between rGO and CS [32]. The adsorption energy (ΔE) was calculated according to ΔE = E_Total - (E_A + E_B), where E_Total, E_A and E_B are the total energy of the adsorption system, unit A and unit B respectively.

During reduction, the interaction energies among rGO increased because of the increasing п-п interaction. So, the effects of the interaction energies among rGO on the formation of porous structures in CS/PVA/rGO systems were carefully evaluated by a previously proposed lattice-based Monte Carlo simulation method [33,34]. The simulations employed a constant volume cubic box with dimensions of 120 × 120 × 120 under periodic boundary conditions, within which the molecular motions of polymer A (PVA), polymer B (CS) and part C (rGO) were simulated at a temperature of k_BT = 1. The lattice-based Monte Carlo simulations were conducted by choosing a monomer at random and attempting to move it to one of its six nearest neighbors. An attempted move is accepted if it obeys the excluded volume constraint, where only a single bead can occupy a single lattice site, and the polymers obey the chain connectivity constraint, which restricts the lengths of bonds connecting two neighboring monomers to the set of values {1, }.

Cell migration simulation was also performed with the same Monte Carlo method but with different models. 500 simulated cell balls with 1 eV interaction energy with scaffolds were placed on one side of the scaffolds. The normalized percentage of balls that pass through of different scaffolds were calculated after the dynamic process, and the proportion of red microspheres reaching the bottom of the scaffold within a defined timeframe is defined as the scaffold continuity degree.

2.9. Statistical analysis

All the data collected in this work were presented as mean ± standard deviation (SD). A normal distribution test was performed to determine whether a parametric or nonparametric test was conducted. Levene's test was applied before one-way ANOVA, and the F-test was applied before pairwise comparisons, with both confirming homoscedasticity. All graphic illustrations and statistical analyses were obtained using Origin 2021 software For comparison among multiple groups when appropriate one-way ANOVA followed by Tukey post hoc test was applied. With respect to results such as the comparison between the controls and the experimental groups, two-tailed Student's t-test was used to determine the significance of difference. p < 0.05 indicates a statistically significant difference, p < 0.01 indicates a highly significant statistical difference, and p < 0.001 indicates an extremely significant statistical difference. All fluorescent images were processed with Image Pro Plus 6.0 (Media Cybernetics, Rockville, MD, USA), or Image J (ImageJ software v1.51w). In the quantitative analysis of cellular immunofluorescence, the "Analyze Particles" function in Image J was used to quantify fluorescence intensity within regions of interest ≥500 pixels in area, following background subtraction and consistent thresholding across all samples. For quantitative analysis of staining intensity in histological and immunofluorescence assays of tissue sections, mean fluorescence intensity was calculated from square-shaped regions of interest (ROIs) containing 400 pixels to enable robust, comparative quantification. Specifically, 20 regions exhibiting distinct fluorescence intensity were selected as the ROIs for analysis.

3. Results and discussion

3.1. Synthesis and characterization of CPH/rGO-3/0.6 scaffold

Hierarchically porous structures are widely adopted in artificial bone tissue scaffolds, owing to its transport capabilities for cells and nutrients brought by the macropores and cell adhesion ability endowed by the microstructures on the macropore surface [35]. The preparation methods of hierarchically porous structures are multifarious, including 3D printing [[36], [37], [38]], etching [39] and templating methods [40] for macropores and in-situ construction [41,42], mineralization [43], and hydrothermal methods [44] for microstructures on macropore surface. Structures of two different size require two different preparation methods. Therefore, the preparation methods for hierarchically porous structures are generally complicated.

In this work, we developed a one-pot method for fabricating porous scaffolds. Initially, four components—short-chain CS, GO, long-chain PVA, and HA—were uniformly mixed to form a viscous slurry (Fig. 1a). In this mixture, short CS chains bonded to GO sheets via polar interactions. During the subsequent gentle reduction of GO, the GO sheets underwent stacking driven by π–π conjugation. Throughout this stacking process, the connective CS chains and hydrogen bonding between HA and CS promoted the formation of large aggregated sheets incorporating HA. Meanwhile, the long-chain PVA served as a structural linker between these large sheets, owing to its strong hydrogen bonding with CS. As GO reduction proceeded, porous scaffolds gradually assembled and precipitated from the slurry. The stacked graphene sheets, together with the cross-linking CS and PVA chains, ultimately established a long-range continuous graphene pathway within the scaffold, which is expected to provide an efficient route for rapid cell migration into the scaffold structure (Fig. 1a).

Fig. 1.

Diagram of preparation and function design of bone tissue scaffolds. (a) Schematic illustration of the formation mechanism of the self-assembled process of porous CPH/rGO-3/0.6 (CS/PVA/HA/rGO) composite scaffolds with continuous graphene surface. (b) Mechanism of cell adhesion and migration on the surface of the CPH/rGO-3/0.6 scaffolds and (c) the osteogenic differentiation and biomineralization of MSCs on the modified rGO surface in the porous CPH/rGO-3/0.6 scaffolds. (d) Schematic illustration of the process of CPH/rGO-3/0.6 artificial bone implantation and the rapid ingrowth of new bone.

Owing to the hydrogen bonding interaction between CS molecular chains and the polar functional groups on reduced graphene oxide (rGO), microphase separation of CS occurred on the surface of rGO because of the uneven distribution of functional groups of rGO. As a result, CS nanodots formed on the rGO sheets. The CS nanodots provided a great number of physical and chemical adhesion points for improved cell adhesion, migration and anchoring, promoting rapid bone ingrowth and repair (Fig. 1b). Furthermore, the polar CS nanodots also provided locations for calcium deposition. With osteogenic MSC differentiation, calcium nodules gradually formed. Thus, calcium nodules evenly deposited on the modified rGO and MSC surfaces (Fig. 1c). More importantly, with the calcium deposition, graphene sheets were gradually wrapped during repair without causing any inflammation or rejection (Fig. 1c). So, the hierarchically porous scaffolds prepared using this strategy exhibited great potential to repair critically sized bone defects.

The overall repair process exhibited by the CPH/rGO-3/0.6 scaffold is as follows (Fig. 1d): Upon implantation of the scaffold, osteogenic lineage cells can promptly spread into the interior of the scaffold along the continuous migration network provided by the scaffold (Fig. 1d). During osteogenic differentiation, calcium nodules are formed and deposited on the pore walls of the scaffold (Fig. 1d). Subsequently, the surface of the scaffold pore walls is rapidly reconstructed. New bone gradually develops and continuously ingrows into the interior of the scaffold along with the migration of cells and blood vessels (Fig. 1d). At this stage, chitosan, PVA, and HA within the scaffold are absorbed by the body, while the rGO sheets are encapsulated by the newly formed bone. Ultimately, a complete new bone structure is formed at the defect site (Fig. 1d).

The physical and chemical structures of the hierarchically porous scaffolds were thoroughly studied to establish the formation mechanism and analyze the osteogenesis performance. It can be observed that a pore structure with the pore sizes of 300∼400 μm was formed and HA particles mixed with CS were stuck in the middle of rGO sheets (Fig. 2a). A rough surface with many nanodots with diameter about 30 nm was also constructed (Fig. 2a). As a result, the CPH/rGO-3/0.6 scaffold with continuous graphene surface shows much higher SSA than that of CPH/rGO-3/0 without graphene (Fig. S1c). In contrast, the CPH/rGO-3/0 sample without rGO had many HA particles on the surface of the pore wall (Fig. S1a). Energy dispersive spectrometer (EDS) mapping and Ca and P contents on the surfaces further prove this result (Fig. S1a and b). The exposed HA particles cannot contribute to the mechanical strength improvement and even worse, and it may detach from the scaffold and cause inflammation after implantation. As shown in Fig. S2, after shaking in PBS at 37 °C for 30 days, the weight loss of CPH/rGO-3/0.6 scaffold was only 3.5 wt%, much lower than that of CPH/rGO-3/0 scaffold. Furthermore, the supernatant of CPH/rGO-3/0.6 was still clear (Fig. S2). These findings shows that CPH/rGO-3/0.6 scaffold remained stable and gradually degraded in vivo.

Fig. 2.

Characterization and formation mechanism of hierarchically porous scaffolds. (a) SEM and AFM images showing the macropore structure and nanodots on the pore wall of CPH/rGO-3. (b) Height, modulus and adhesion mapping of GO, rGO and rGO/CS. Structural stability of CPH/rGO-3/0 and CPH/rGO scaffolds with different content of PVA characterized through (c) Compressive modulus and (d) relaxation curves of CPH/rGO series scaffolds and cancellous bone (n = 5 per group. Data are expressed as mean ± SD). (e) Results of Monte Carlo simulations obtained for a system contained rGO, CS and PVA with different adsorption energy between rGO (E_rGO-rGO) and (f) sequence degree statistics of CS and PVA. (g) Molecular dynamics (MD) simulations of the interaction between rGO and CS, and the charge distributions in the interaction between CS and rGO with different oxygen-containing functional groups or pure graphene.

The chemical structures of the composite scaffold were also examined to determine the raw materials that were organically combined with each other. For comparison, rGO sample was prepared using the same reduction method and procedure for CPH/rGO-3. The absorption peaks at 1719 cm⁻¹ (C=O stretching vibrations) and 1213 cm⁻¹ (epoxy groups) shown in GO sample decreased in rGO sample, implying that GO was partially reduced (Fig. S3a) [45]. Both the absorption band corresponding to NH₂ vibration at 1590 cm⁻¹ in CS and the absorption band corresponding to C=O stretching of the carboxylic groups at 1719 cm⁻¹ in GO disappeared (Fig. S3a). Meanwhile, a new peak at 1639 cm⁻¹ formed in CPH/rGO-3/0.6 sample, indicating that the hydroxyl and amino groups on the CS molecular chains reacted with the carboxyl groups on the rGO sheets (Fig. S3a) [46]. A characteristic HA absorption peak at 1021 cm⁻¹ (P-O) was observed in CPH/rGO-3/0 and CPH/rGO-3/0.6 samples, clearly showing that HA was loaded in the composite scaffolds (Fig. S3a) [47].

In X-ray diffraction (XRD) spectra (Fig. S3b), characteristic absorption peaks of HA appeared in CPH/rGO-3/0 and CPH/rGO-3/0.6 samples, indicating the introduction of HA in composite scaffold, consistent with the Fourier-transform infrared (FTIR) results (Fig. S3a) [48,49]. The characteristic diffraction peak at 2θ = 9.56^o and 2θ = 24.15^o (GO and rGO, respectively) also proved the reduction of GO (Fig. S3b) [50]. In Raman spectra (Fig. S3c), the peaks at approximately 1350 and 1585 cm⁻¹ are attributable to D (symmetry A1g mode) and G bands (E_2g mode of sp² carbon atoms), respectively [51]. The G band of CPH/rGO-3/0.6 sample exhibited higher Raman shift, indicating that graphene with high strain formed in the pore structure of CPH/rGO-3/0.6 scaffold [52]. The increased D/G intensity ratio of rGO and GO also demonstrated the reduction of GO during the preparation of CPH/rGO-3/0.6 scaffold [53]. The type and number of functional groups on GO and rGO were obtained by X-ray photoecectron spectroscopy (XPS) spectra (Fig. S3d). XPS spectra of GO can be divided into four peaks, corresponding to the sp² carbons (C=C, 284.5 eV), hydroxyl groups (C-O, 286.3 eV), carbonyl groups (C=O, 287.4 eV), and carboxylate groups (O=C-O, 288.5 eV) (Fig. S3d) [54]. Following reduction, the peaks for the oxygen-containing functional groups decreased considerably and the peak of hydroxyls still remained to be the highest among the oxygen-containing functional groups (Fig. S3d).

The microtopology, microscopic modulus, and adhesion ability of CS modified rGO sheets were investigated using atomic force microscopy (AFM) (Fig. 2b). After reduction, the folds of graphene sheets became smaller, and the adhesion property related to micro physical and chemical structure decreased because of less polar functional groups on the surface of graphene sheets and smoother surface (Fig. 2b, Fig. S4). Moreover, sp² hybridization was completed after the reduction, and as a result, the modulus of rGO increased notably (Fig. 2b) [55]. After the introduction of CS, nanodots formed via the microphase separation of CS on the surface of rGO. The modulus of CS modified rGO decreased (Fig. 2b) compared with that of rGO, and moreover, the adhesion ability increased substantially because of numerous adhesion sites created by the CS nanodots and the hydrophilicity also increased after the modification (Fig. 2b). The mechanical properties of the scaffolds were examined via compression, relaxation, frequency sweep and cycle tests (Fig. 2c,d, Fig. S3e and f). It is noteworthy that CPH/rGO-3/0.6 exhibits a modulus comparable to that of native fresh porcine cancellous bone (Fig. 2c), coupled with a relatively low relaxation rate (Fig. 2d). These findings underscore the exceptional structural stability of the composite scaffold. When implanted into bone defect sites, CPH/rGO-3/0.6 can provide sustained mechanical support over an extended period, while simultaneously exerting osteogenic effects within the defect region. This combination of robust biomechanical performance and prolonged functional integrity makes CPH/rGO-3/0.6 a promising candidate for bone regeneration applications, where maintaining structural support throughout the healing process is critical. In addition, CPH/rGO samples show substantially higher storage modulus compared to CPH/rGO-3/0 without rGO because of the formation of stable pore structures. When proper content of PVA was included, CPH/rGO-3/0.6 samples exhibited the highest and most stable storage modulus with increasing frequency (Fig. S3e), implying that the most complete network structure was formed in CPH/rGO-3/0.6 [56]. The cycle test results are consistent with this finding (Fig. S3f).

The formation mechanism of the macropores was further examined using Monte Carlo (MC) simulation. As shown in Fig. 2e, the interaction between rGO sheets during reduction was the main driving force to form scaffolds from the mixed slurry. Thus, the thermodynamically stable state of the rGO-CS-PVA system with different bonding energy between rGO sheets (E_rGO-rGO) was calculated. With increasing E_rGO-rGO, the pores gradually formed and grew larger. The CS sequence degree also increased with increasing E_rGO-rGO (Fig. 2e and f), which indicated that continuous porous network formed gradually. In the sectional view of the system with E_rGO-rGO = 3 (Fig. 2e), rGO sheets bonded with CS molecule chains stacked together to create the space, and PVA chains connected the rGO/CS composite layers to form the network (Red circle in Fig. 2e). To further probe the influence of graphene content on the development of a continuous pore network during phase separation, molecular dynamics simulations were performed. At low graphene loadings, insufficient interfacial driving force impeded phase separation, thereby preventing the emergence of a porous architecture (Fig. S5). In contrast, excessive graphene loading led to severe sheet–sheet steric hindrance during the concurrent phase separation and reduction process, which likewise suppressed the formation of a well-defined, interconnected pore structure (Fig. S5). A well-balanced graphene content—sufficient to nucleate and stabilize pores yet low enough to avoid aggregation-induced kinetic trapping—was found to be critical for achieving an optimal, continuous porous network (Fig. S5).

The formation of chitosan nanodots on the surface of the CPH/rGO-3/0.6 scaffold is critical for cell adhesion, and the microphase separation mechanism underlying their formation has been investigated (Fig. 2g). The CASTEP module of Materials Studio 2019 was used to analyze the adsorption kinetics of the CS molecular chains onto rGO surface, as well as the charge distribution of CS molecular chains adsorbed on rGO surface with different kinds of oxygen-containing functional groups (Fig. 2g, Video S1-5). In the wake of the molecular dynamics process, the distance between the centroids of the CS molecular chain and the rGO sheet diminished from 11.141 Å to 6.300 Å, along with a negative adsorption energy of −1.08 eV, indicating that active adsorption could transpire between CS and rGO (Fig. 2g). Based on the adsorption energy, among all the oxygen-containing functional groups on rGO and pure graphene, hydroxyl and carboxyl groups exhibited higher adsorption ability to CS molecular chains (Fig. 2g). The intermingled electron cloud between rGO-OH and CS or rGO-COOH and CS were fused together, in consistence with the higher adsorption energy results (Fig. 2g). Therefore, CS nanodots were formed on the rGO sheets owing to the uneven adsorption energy distribution.

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.bioactmat.2026.02.026

The following are the Supplementary data related to this article.

3.2. The cellular infiltration capacity of the CPH/rGO-3/0.6 scaffold

To evaluate the cellular infiltration capacity of the CPH scaffold, initially, adhesion status of hMSC on rGO and rGO/CS substrates were studied. After cell adhesion of hMSCs on the surfaces of rGO and rGO/CS, most of hMSCs on rGO were fusiform (Fig. 3a). However, many hMSCs on rGO/CS spread their synapses and showed higher cell area than that of hMSCs on the surface of rGO (Fig. 3a and b). On the other hand, hMSCs on the surface of rGO/CS exhibited much higher adhesive capacity to the AFM probe (Fig. 3c and d). All these results meant hMSCs exhibited stronger adhesion ability on the surface of rGO/CS because the modified rGO sheets with CS nanodots can provide more anchor points for hMSCs.

Fig. 3.

Adhesion and migration of hMSC on rGO/CS surface and within CPH/rGO-3/0.6 scaffold. (a) Adhesion status and (b) cell area statistics of hMSCs on the surfaces of rGO and rGO/CS after seeding for 12h (n = 40 per group. Data are expressed as mean ± SD. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001). (c) AFM and adhesion images of hMSC on the surfaces of rGO and rGO/CS. (d) Force curve of hMSC on the surfaces of rGO and rGO/CS. (e) Adhesion mapping of rGO/CS with 3D bird's-eye view and energy dissipation of hMSC modified tip on different surface. (f) 8h migration trajectory of hMSC on Blank, rGO and rGO/CS surface after seeding for 12h. Statistics of (g) migration rate and (h) displacement of hMSC on Blank, rGO and rGO/CS surface (n = 18 per group. Data are expressed as mean ± SD. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001). (i) Simulation study of the transmissibility capacity of scaffolds with different structure and corresponding continuity. (j) The particle transmission ratio of different scaffolds. (k) Migration capability of hMSC in different scaffolds through the transwell permeability assay.

To further study the cell adhesion on rGO/CS surface, a hMSC modified tip was prepared (Fig. S6) and applied in the force-displacement (F-D) curve test of AFM (Fig. S7) according to previous report [57]. The adsorption energy between the sample and MSC can be obtained from the retraction process of the F-D curve [58]. It can be seen that MSC modified tips displayed the largest energy dissipation on the rGO/CS surface, indicating the best adhesion effect of MSC on the CS nanodot decorated rGO surface (Fig. 3e). Additionally, the MSC energy dissipation on the CS nanodot decorated rGO surface can be divided into two parts, a green part and a yellow part (Fig. 3e). The green one is the interaction between the rGO surface surrounding the nanodots and MSC, and the interaction between CS nanodots on rGO sheet and MSC displayed smaller energy dissipation in the yellow part (Fig. 3e).

The migration of MSC in the scaffolds notably influences the bone ingrowth and repair. Thus, the hMSC migration on different surfaces was elucidated (Fig. 3f). After seeding for 12h, hMSC attached and migrated on the scaffold surface. The migration trajectory, migration rate and displacement of 18 hMSCs on different substrates without mitosis were shown in Fig. 3f–h according to the videos of cell migration (Video S6-8 with a playback rate of 3000 times). It can be seen that the modified rGO/CS sheets can significantly increase the migration rate of hMSCs, which can be attributed to the topological and chemical structure of rGO/CS sample (Fig. 3g and h).

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.bioactmat.2026.02.026

The following are the Supplementary data related to this article.

hMSC migration in the scaffold was further studied using migration simulation (Fig. 3i, Fig. S8) and a modified transwell permeability assay. Four representative structures (Grid pattern (GRI); Lyophilized structure (LYO); Diamond structure (DIA); Lubricated diamond structure (LUB)), as well as structures of cancellous bone and CPH/rGO series scaffold were constructed (Fig. 3i, Figs. S8 and S9). Among the CPH/rGO scaffold series, CPH/rGO-3/0.6 exhibits the most well-developed porous continuous network and the highest degree of continuity (Fig. S9). On the other hand, both cancellous bone and CPH/rGO-3/0.6 scaffold structure had larger specific surface area than that of scaffolds with other structures (Fig. S8), which can provide a continuous migration surface for cells. CPH/rGO-3/0.6 and LUB scaffolds performed higher transmissibility because of the continuous and direct migration pathway of CPH/rGO-3/0.6 and LUB from the top to the bottom of the scaffolds (Fig. 3i and j). A modified transwell permeability assay for scaffolds was developed to explore the cellular permeability of the scaffolds, and hMSCs will migrate through different scaffolds and attach to the membrane of the transwell (Fig. S10). The results indicate that the CPH/rGO-3/0.6 scaffold demonstrated a strong capability of promoting cell migration throughout the entire scaffold, which was consistent with the results of migration simulation (Fig. 3k).

3.3. Investigation of the surface mineralization ability of CPH/rGO-3/0.6 scaffold

In order to explore the crystallization nucleation ability on the surface of modified graphene, we dropped supersaturated NaCl solutions onto the surfaces of different substrates. Then, we utilized a polarized light microscope to observe the nucleation phenomena on these various surfaces. The rGO/CS surface can enhance crystal nucleation (Fig. 4a). This property can facilitate hMSC-mediated calcium deposition on the surface of the scaffold and further promote the reconstruction of the scaffold.

Fig. 4.

Calcium deposition capacity of rGO/CS substrate and CPH/rGO-3/0.6 scaffold. (a) Crystallization on the surfaces of glass coverslip, rGO and rGO/CS. (b) Calcium nodules generated by hMSC on rGO and rGO/CS surfaces after 21 days of osteogenic induction. SEM images and EDS mapping of calcium nodules (c) on the surface of rGO/CS plate, (d) on the surface of hMSC and (e) in the hMSC cultured on the rGO/CS surface after 21 days of induction. (f) TEM images of calcium nodules generated by hMSCs on rGO and rGO/CS after 21 days of induction and the HRTEM image of calcium nodules generated by hMSCs and its SAED pattern. (g) SEM images of hMSCs on CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds after osteogenic induction for 7, 14 and 21 days and corresponding content of element Ca on 21 days. (h) SEM images of calcium deposition of hMSC on CPH/rGO-3/0.6 scaffolds after osteogenic induction for 21 days and corresponding C, O, Ca and P elemental mapping. (i) SEM images of calcium deposition of hMSC on CPH/rGO-3/0.6 scaffolds after osteogenic induction for 28 days and its corresponding C, O, Ca and P elemental mapping.

The artificial scaffolds can be gradually reconstructed by the attached cells to realize the bone connection and integration. Calcium deposition is a critical way to modify the surface of artificial scaffolds, and the calcium deposition rate is directly correlated to the repair rate [59]. hMSC were induced by osteogenic differentiation for 21 days on different surfaces, and the calcium deposition was investigated. It can be seen that numerous popcorn shaped calcium nodules formed on the CS nanodot decorated rGO surface (Fig. 4b and c). Moreover, many calcium nodules deposited on and in the hMSCs cultured on rGO/CS surface (Fig. 4d and e). That meant the high degree of osteogenic differentiation of hMSCs on rGO/CS and high ability of rGO/CS surface to promote cellular calcium deposition.

To further characterize the calcium nodule crystallization on rGO/CS, transmission electron microscopy (TEM) observation and selected area electron diffraction (SAED) (Fig. 4f) were performed. The weak polymorphic rings proved the semi-crystalline of deposited calcium nodules produced by differentiated hMSC on rGO and rGO/CS surfaces [49,60]. However, the brighter spots and diffraction rings of calcium nodules on rGO/CS surfaces implied higher crystallinity [49,60], which could be attributed to the highly differentiated hMSC and crystallization promoting ability of rGO/CS surfaces (Fig. 4f).

BMSCs were also applied to study the calcium deposition ability of modified rGO surface. After 21 days of induction, it can be seen that more calcium nodules deposited on the CS nanodot decorated rGO surface (Fig. S11). Moreover, the crystalline morphology of HA with a higher crystallinity could be observed on the rGO/CS surface through the brighter diffraction rings in SAED pattern (Fig. S12) [49,60]. Thus, the CS nanodot decorated rGO surface can also provide a suitable substrate for BMSCs to realize excellent calcium deposition.

In vitro studies of hMSC were performed to investigate the biocompatibility and osteogenesis promoting ability of different scaffolds. Cell counting kit-8 (CCK8) assays (Fig. S13) indicated that CPH/rGO series scaffolds showed no toxicity to the hMSCs, and the CPH/rGO-3/0.6 scaffold can promote hMSC proliferation, owing to improved cell adhesion ability and its hierarchically interconnected porous network. Calcium nodule deposition is an important evaluation criterion for osteogenic differentiation. hMSCs adhered well on both CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffold surfaces (Fig. 4g). However, after 21 days of differentiation induction, numerous calcium nodules were formed on the cell on CPH/rGO-3/0.6 scaffold, whereas, there is less calcium deposition on the CPH/rGO-3/0 surface (Fig. 4g and h). This implies that the surface of CPH/rGO-3/0.6 had good cell differentiation promotion ability and favored calcium deposition for biomineralization. The crystallization degree and the structure of calcium nodules produced by hMSCs were further studied using alizarin red staining (Fig. S14). After co-cultivation with different scaffolds for 21 days, CPH/rGO-3/0.6 displayed the largest number of calcium nodules on slides (Fig. S14). With further osteogenesis induction for 28 days, the calcium deposition of hMSCs became even more notable. Large calcium nodules formed on the hMSC and the calcium content increased substantially after induction for 28 days (Fig. 4i).

To investigate the dynamic balance between osteogenesis and osteoclastic bone resorption on biomaterial surfaces, hMSCs were first induced toward osteogenic lineage on rGO/CS and rGO substrates. Following 14 days of osteogenic induction—confirmed by the formation of mineralized calcium nodules—primary osteoclasts were co-cultured with the pre-osteoblast layers. Samples were harvested every three days for quantitative alizarin red S staining, and nodule morphology, and temporal stability were systematically assessed via bright-field microscopy and image-based morphometry. These analyses revealed that the rGO/CS and rGO surfaces supported a more robust and temporally stable osteogenic-osteoclastic coupling compared with the blank control (Fig. S15). Notably, rGO/CS exhibited significantly greater calcium deposition than rGO, indicating superior osteoinductive capacity.

3.4. The osteogenic inductive capacity and mechanisms of the CPH/rGO-3/0.6 scaffold

Given that scaffold immune rejection can significantly impair osteogenesis, we evaluated the immunocompatibility of CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds through in vitro co-culture with RAW macrophages and in vivo immunofluorescence staining at 1 month post-implantation. Both CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds elicited a transient, physiological acute immune response. As the culture period extended, a gradual transition from M1 to M2 polarization was observed in cells cultured with the scaffolds, especially CPH/rGO-3/0.6 (Fig. S16). This dynamic and timely shift from a pro-inflammatory (M1) to a pro-healing (M2) state underscores the modified material's excellent biocompatibility and favorable immunomodulatory properties [25,61,62]. These characteristics are crucial for supporting osseointegration and bone regeneration, which are fundamental to achieving functional bone repair. In vivo, IL-1β staining at 2 weeks showed that macrophage infiltration and IL-1β expression in the CPH/rGO-3/0.6 group were comparable to those in the biocompatible 3D printing scaffold, further confirming the absence of abnormal immunoinflammatory reactions (Fig. S17). However, the HA scaffold and the CPH/rGO-3/0 scaffold exhibited substantial immune cell infiltration and elevated expression of immune-related genes—attributed to HA powder leakage and impaired cellular infiltration, respectively (Fig. S17).

The hMSC shape is an effective way to determine the osteogenic differentiation [63]. The shape change is essential because it allows the cell to switch jobs. The original spindle shape is good for moving around, but a mature osteoblast needs to be a stationary builder. The new flat, polygonal shape lets the cell anchor firmly and gives it the space and structure to become a miniature factory, efficiently making and releasing bone matrix [64,65]. hMSCs co-cultured with different scaffolds were seeded on glass coverslips for cell morphology examination (Fig. 5a). In general, hMSCs maintained the spindle shape during the 21 days of induction. However, hMSCs with CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds gradually transferred from the spindle shape to triangle shape, especially with CPH/rGO-3/0.6 scaffold (Fig. 5a). Furthermore, fluorescence intensity of bone-related proteins of OCN in hMSCs with CPH/rGO-3/0.6 scaffold demonstrated the highest degree of osteogenic differentiation of hMSCs (Fig. 5a). Moreover, the hMSCs shape became polygonic and high expression of OCN fluorescent staining could be observed after further induction (Fig. 5b).

Fig. 5.

In vitro study of osteogenic capacity and mechanisms of the CPH/rGO-3/0.6 scaffold (a) Fluorescent staining of hMSCs grown on the surface of Blank, CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds for 7, 14 and 21 days and intensity statistics of osteocalcin (OCN) on 21 days (Cell nuclei of hMSCs were visualized using DAPI (blue); Cytoskeleton was stained with Phalloidin-FITC (green); OCN proteins were stained with Alexa Fluor 594 (red)) (n = 16, 12, 15 for Blank, CPH/rGO-3/0 and CPH/rGO-3/0.6 groups respectively. Data are expressed as mean ± SD. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001). (b) Fluorescent staining of MSCs grown on the surface of CPH/rGO-3/0.6 scaffold for 28 days. (c) Osteogenesis related genes expression of MSCs including alkaline phosphatase (ALP), type I collagen (COL-I), runt-related transcription factor 2 (Runx2), SP7 transcription factor (SP7), Bone sialoprotein (BSP), dentin matrix acidic phosphoprotein 1(DMP1), OCN and osteopontin (OPN) after 7, 14 and 21 days' incubation on CPH/rGO-3/0, CPH/rGO-3/0.6 scaffolds and Blank (n = 3 per group. Data are expressed as mean ± SD. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001). (d) OD value obtained from the ALP reagent of sample Blank, CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds after osteogenic induction of hMSC for 4, 8 and 12 days (n = 3 per group. Data are expressed as mean ± SD. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001). (e) Volcano map and (f) GO enrichment analysis of differentially expressed genes in hMSCs cultured on rGO/CS vs rGO and on CPH/rGO-3/0.6 vs CPH/rGO-3/0. (g) Hotmap of differentially expressed genes between rGO/CS and rGO samples, CPH/rGO-3/0.6 and CPH/rGO-3/0 scaffolds. (h) Western blot images of KCNN3, Integrin β1, ANK3, FAK, MAPK, OCN, and BSP following 14 days of osteogenic induction co-culture of hMSCs with rGO, rGO/CS, Blank. (i) Schematic diagram of osteogenic gene pathways mediated by CPH/rGO-3/0.6.

Related gene expression was determined by qPCR (Fig. 5c), which indicated that hMSCs on the CPH/rGO-3/0.6 scaffold exhibited a stronger osteogenic differentiation tendency in both the early lineage commitment and late mineralization stage [66]. Due to the remarkable cell adhesion ability, the critical osteogenic regulators of hMSCs on the CPH/rGO-3/0.6 scaffold increased significantly after 7 days (Fig. 5c). Moreover, the high biomineralization related gene expressions such as OCN and OPN at 21 days proved the high degree of differentiation of hMSCs on CPH/rGO-3/0.6 scaffold (Fig. 5c). However, at 14 days, no significant osteogenic advantage was observed, which might be because the osteogenic differentiation promoted by CPH/rGO-3/0.6 mainly manifested in the early stage by promoting cell adhesion and in the later stage by facilitating the deposition of calcium nodules during cell secretion. Furthermore, the qPCR results were highly consistent with the conclusion of calcium deposition and cell morphology. Meanwhile, results obtained from the ALP reagent during the incubation of hMSCs on different scaffolds (Fig. 5d) and ALP staining during the incubation of hMSCs on slides with different scaffolds (Fig. S18) also manifested the high osteogenic differentiation ability of CPH/rGO-3/0.6 scaffold.

Furthermore, from the research of biological mechanisms, there are a large number of differentially expressed genes between hMSCs on the surfaces of CPH/rGO-3/0.6 and CPH/rGO-3/0 or rGO and rGO/CS (Fig. 5e). Through the GO enrichment analysis of differentially expressed genes, the enrichment results of CPH/rGO-3/0.6 vs CPH/rGO-3/0 and rGO vs rGO/CS are highly consistent (Fig. 5f). Ion channel, interaction between ECM and cells, and cell morphology are main differences between hMSCs on the surfaces of CPH/rGO-3/0.6 and CPH/rGO-3/0 or rGO and rGO/CS (Fig. 5f and g). Potassium channel related genes KCNN3, potassium ion channel protein 1 (KCNH1) and cell adhesion related genes Ankyrin 3 (ANK3) and FRAS-related extracellular matrix 1 (FREM1) were selected as the important targets for subsequent biological mechanism validation (Fig. 5g, Fig. S19). Western blot analysis was performed to quantify protein expression levels (Fig. 5h, Fig. S20). Results demonstrated that the potassium channel–associated protein KCNN, the cell adhesion–associated proteins ANK3 and integrin β1, and the osteogenesis-associated proteins bone sialoprotein (BSP) and osteocalcin (OCN) were significantly upregulated in the rGO/CS group relative to the rGO group and in the CPH/rGO-3/0.6 group relative to the CPH/rGO-3/0 group (Fig. 5h, Fig. S20). Transcriptomic profiling further revealed that, compared with the blank control, both CPH/rGO-3/0.6 and CPH/rGO-3/0 scaffolds induced significant enrichment of gene sets associated with mitotic nuclear division and sister chromatid segregation—indicating enhanced proliferative capacity in hMSCs (Fig. S21). Consistent with these findings, cytotoxicity assays confirmed the absence of adverse effects on hMSC viability following exposure to either scaffold material (Fig. S13).

Based on transcriptomic profiling of the rGO/CS vs rGO and CPH/rGO-3/0.6 vs CPH/rGO-3/0 comparisons, we constructed a mechanistic pathway diagram illustrating how CPH/rGO-3/0.6 promotes osteogenic differentiation (Fig. 5i). CS-mediated graphene surface modification enhances cell adhesion signaling—partly through upregulation of the extracellular matrix protein FREM1—while concurrently inducing expression of potassium channel–associated genes KCNN3 and KCNH1. The synergistic modulation of integrin-mediated adhesion and membrane potential–associated ion channel activity converges to activate the MAPK–ERK and focal adhesion kinase (FAK) signaling axes. ANK3, identified as a key scaffolding component, stabilizes and sustains MAPK pathway activation, thereby driving transcriptional upregulation of canonical osteogenic effectors—including RUNX2, BSP, and OCN [67]. As a result, the modified rGO surface can greatly promote the adhesion of hMSCs through changing the state of cell adhesion, and increase the K⁺ channel activity and K⁺ transport simultaneously. The adjusted K⁺ channel will further play an important role to accelerate the cell migration and improve the osteogenic differentiation of hMSCs.

3.5. Repair of in-vivo critical bone defects by CPH/rGO-3/0.6 scaffold

The repair performance of the CPH/rGO-3/0.6 scaffold to critically sized bone defects was further studied in vivo. Cylindrical defects in the medial condyle of femur with a diameter 3 mm and height 4 mm was created (Fig. S22). To comprehensively evaluate systemic biosafety following the implantation of four distinct scaffold types in vivo, complete blood count analysis was carried out at 1 and 3 months post-implantation (Tables S1–8), followed by H&E staining of the heart, liver, spleen, lungs, and kidneys at the 3-month endpoint (Fig. S23). Results from both hematological and histopathological assessments consistently indicated an absence of significant systemic toxicity induced by either CPH/rGO-3/0.6 or CPH/rGO-3/0. After scaffold implantation for three months, the femur was removed and characterized. 3D images were reconstructed with Micro-CT and X-ray images (Fig. 6a). In the defect area of the blank group, no new bone formation occurred, demonstrating that this animal model pertains to the model of critical bone defect (Fig. 6a). The X-ray images showed that the white line formed around the defect area in the blank, CPH/rGO-3/0 and HA scaffold samples and further blocked the bone ingrowth, the high-density shadow could clearly be found in the defect area of CPH/rGO-3/0.6 scaffold (Fig. 6a). According to the 3D images, new bone formed a complete network in the defect area of sample CPH/rGO-3/0.6 (Fig. 6a).

Fig. 6.

Regeneration of bone defects with critical size. (a) 3D images reconstructed with Micro-CT and X-ray images of blank, CPH/rGO-3/0, CPH/rGO-3, HA and 3D Printing scaffolds after implantation for 3 months. (b) Statistics of osteogenic parameters based on Micro-CT (n = 6 per group. Data are expressed as mean ± SD. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001). H&E and Masson's staining of (c) entire defect area and (d) the junction between implanted scaffolds and native bone and inside of different scaffolds after implantation for 3 months. (e) Schematic illustrations of the ingrowth of new bone into different scaffolds. (f) H&E staining of CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffold and their crystallization characterized through POM and TEM after implantation for 3 months. (g) SEM images and EDS mapping of the entire implant area and images at high magnification of interface between defect area (D) and natural bone (B), and inside of the scaffolds. (h) SEM images and EDS mapping of interface between CPH/rGO-3/0.6 scaffold and new bone on tissue section. (i) Three-point bending tests of different scaffolds in the femoral hemisection model after implantation for 1 month (n = 5 per group. Data are expressed as mean ± SD. ns, no statistical significance. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001).

However, a large amount of new bone is formed around the defect site to constitute the thick bone white lines in the blank, CPH/rGO-3/0 and HA scaffold samples (Fig. 6a and b). Meanwhile, trabecular bone exhibiting higher bone mineral density (BMD), bone volume/total volume (BV/TV), and trabecular thickness (Tb. Th), as well as reduced trabecular separation (Tb. Sp), was observed in the CPH/rGO-3/0.6 scaffold, indicating enhanced osteogenic repair capability (Fig. 6b). This improvement can be attributed to the superior cell transmissibility, osteogenic inductivity, and mineralization capacity conferred by the continuous osteogenic channel network within the scaffold. Notably, despite the 3D-printing scaffold not forming an obvious bone white line owing to its outstanding through-hole structure, the total quantity of newly formed bone is far less than that of the CPH/rGO-3/0.6 scaffold (Fig. 6a and b). We longitudinally compared osteogenic progression across four scaffolds using X-ray and micro-CT at 1 month and histology at both 1 and 3 months. CPH/rGO-3/0.6 showed consistently superior osteogenesis—early (1 month) and late (3 months)—due to its strong initial cell adhesion, continuous porous network facilitating cell migration, and high capacity for calcium nodule deposition (Fig. S24).

Based on H&E and Masson's trichrome staining (Fig. 6c and d), both the CPH/rGO-3/0 and HA control groups exhibited the formation of a dense, thick bone layer around the defect periphery. This peripheral encapsulation acted as a barrier, substantially impeding inward cellular infiltration and new bone ingrowth, ultimately leading to incomplete osseous repair. Histologically, this is consistent with limited cell transmissibility and suboptimal calcium deposition kinetics on the implant surface. In contrast, the CPH/rGO-3/0.6 scaffold supported a markedly different healing pattern. No sharp boundary or encapsulating layer was observed between the host bone and the implant. Instead, the defect area was extensively filled with newly formed bone tissue that integrated seamlessly with the original bone. This superior osteointegration and abundant bone ingrowth can be directly attributed to the engineered, continuous cell-migration pathways within the 3D scaffold, which significantly enhanced cellular permeability and infiltration. In addition, one month after implantation, during the active period of osteogenesis, we performed vascular immunofluorescence staining on different scaffolds and found that only in the CPH/rGO-3/0.6 and 3D printing scaffolds did CD31 and EMCN positive blood vessels form inside (Fig. S25). This will greatly facilitate the exchange of nutrients at the defect site during the later osteogenesis process [68,69]. However, in the HA and CPH/rGO-3/0 scaffolds, only CD31 positive blood vessels formed at the scaffold edges (Fig. S25).

Furthermore, the modified graphene surface promoted rapid and effective calcium deposition (Fig. 6e), synergistically creating a highly osteoconductive microenvironment that facilitated direct bone regeneration within the scaffold structure. Preventing fibrous tissue encapsulation is also crucial for successful bone regeneration. The granular HA scaffold, which lacks an integrated structural framework, exhibited extensive fibrous tissue formation within the defect area due to its inability to effectively inhibit the ingrowth of rapidly proliferating fibrous tissue (Fig. 6c,d, Fig. S26). Although the CPH/rGO-3/0 scaffold possesses a integrated structural architecture, its internal microenvironment fails to promote efficient osteogenesis, leading to the development of fibrotic tissue within the scaffold (Fig. 6c,d, Fig. S26). In contrast, neither the CPH/rGO-3/0.6 scaffold nor the 3D-printed scaffold showed evident fibrotic tissue formation, indicating superior resistance to fibrous encapsulation and a more favorable environment for bone regeneration (Fig. 6c,d, Fig. S26).

Meanwhile, CPH/rGO-3/0.6 scaffold with a graphene interlayer can stabilize the scaffold structure and new bone; and thus, HA particles can be slowly degraded in-situ to support the Ca and P ion supply required for new bone formation. The presence of a bright spot indicative of the semicrystalline structure of new bone within the CPH/rGO-3/0.6 scaffold can be observed using a polarization microscope (POM) (Fig. 6f, Fig. S27). In contrast, no significant bright spots are detected in the original CPH/rGO-3/0.6 scaffolds (Fig. S28). To confirm the formation and ingrowth of new bone within the CPH/rGO-3/0.6 scaffold, detailed observations and corresponding polarized images are provided in Fig. 6f and Fig. S29. Extensive calcium deposition occurs on the graphene sheets and gradually reconstructs into new bone tissue. It can be observed that the newly formed calcium nodules are firmly anchored on the graphene surface from the TEM images (Fig. 6f), and the graphene sheets are tightly encapsulated by calcium nodules, thereby potentially reducing rejection and inflammation (Fig. 6f). Notably, even at the center of the CPH/rGO-3/0.6 scaffold, substantial calcium deposition and new bone formation are evident (Fig. 6f, Fig. S29).

SEM images and EDS mappings of the defective sections further demonstrated the best bone ingrowth and formation of new bone in CPH/rGO-3/0.6 scaffold (Fig. 6g). Only the CPH/rGO-3/0.6 sample exhibited no clear boundary between original bone and defect, and the content of element Ca was also the highest in the CPH/rGO-3/0.6 sample (Fig. 6g). On the tissue section of implanted CPH/rGO-3/0.6 (Fig. 6h), calcium deposited graphene will be gradually reconstructed into the new bone based on the CPH/rGO-3/0.6 scaffold. Graphene sheets can be well surrounded by the calcium deposition, which considerably decrease the impact of difficult graphene degradation (Fig. 6h).

To evaluate the efficacy of bone repair from the perspective of functional recovery, we implanted the four types of scaffolds using a hemi-transection defect model. Owing to its continuous, interconnected network of cell-migration channels and surface properties that facilitate calcium-nodule deposition, the CPH/rGO-3 scaffold group showed enhanced new bone formation. X-ray analysis further revealed superior integration between the CPH/rGO-3 scaffold and the host bone tissue (Fig. S30). Three-point bending tests performed one month after implantation demonstrated that the CPH/rGO-3 group exhibited the highest bending strength and flexural strain similar to that of the healthy group, indicating optimal restoration of bone mechanical function with this scaffold (Fig. 6i).

Immunohistochemical (IHC) and immunofluorescence (IF) staining of functional proteins were performed in vivo to further illustrate osteogenic differences and verify the in vitro biological mechanisms. The implantation area is closed, and only a few cells enter the CPH/rGO-3/0 scaffold after implantation for one month (Fig. 7a). However, there is a large amount of newly formed collagen fibers in the CPH/rGO-3/0.6 scaffold, and a large number of cells migrate into the interior of the CPH/rGO-3/0.6 scaffold (Fig. 7a). OCN IHC anf IF results also proved the occurrence of rapid bone ingrowth and new bone formation in CPH/rGO-3/0.6 scaffolds (Fig. 7b). Both inside of the CPH/rGO-3/0.6 and at its interface with the native bone, OCN expressed the high positivity with CPH/rGO-3/0.6 participation (Fig. 7b). Furthermore, the differences in IF staining of adhesion-related proteins FREM1 and ANK3, identified through in vitro differential gene expression analysis, suggest that CPH/rGO-3/0.6 can substantially enhance the adhesion and migration of MSCs, thereby further promoting new bone formation and ingrowth (Fig. 7c). Additionally, the activities of K⁺ channel-related proteins KCNN3 and KCNH1 were highly expressed in the CPH/rGO-3/0.6 scaffold (Fig. 7d), and the active K⁺ channels significantly promote cell migration and osteogenic differentiation. These IF staining results in vivo are consistent with the RNA sequencing data and in vitro experimental findings.

Fig. 7.

IHC and IF staining of functional proteins. (a) H&E and Masson's staining of defect area after implantation of CPH/rGO-3/0 and CPH/rGO-3/0.6 scaffolds for one month. (b) OCN IHC and IF staining of CPH/rGO-3/0 and CPH/rGO-3/0.6 after implantation for one month and the corresponding OD value and fluorescence intensity statistics. (c) IF staining of FREM1 and ANK3 in scaffolds after implanting for one months in rats and corresponding fluorescence intensity statistics. (d) IF staining of KCNN3 and KCNH1 in scaffolds after implanting for one months in rats and corresponding fluorescence intensity statistics. n = 20 per group. Data are expressed as mean ± SD. ∗ for p < 0.05; ∗∗ for p < 0.01; ∗∗∗ for p < 0.001.

4. Conclusion

A continuous, cell-guiding pathway based on chitosan-modified graphene was engineered within the 3D scaffold using an in-situ reduction-induced phase separation technique, achieving this microstructural control with only 3.4 wt% graphene content. Unlike conventional HA scaffolds or 3D-printed architectures, this modified scaffold significantly enhanced cellular infiltration and permeability. Furthermore, by actively regulating cell adhesion and ion channel-related signaling pathways, the scaffold dramatically improved the repair of critically sized bone defects, demonstrating superior osteogenic performance through this biointegrated design. Additionally, the pore wall surface fabricated by modified graphene exhibits outstanding capacity for calcium nodule deposition and in the later stages of repair, graphene could be enwrapped by deposited calcium nodules and new bone without presenting rejection and inflammation. This graphene-doped scaffold can leverage and eliminate the advantages and disadvantages of graphene respectively simultaneously to facilitate critical bone defect repair. Thus, this strategy offers a way to construct a reliable continuous cell migration pathway in the 3D scaffold for rapid bone formation and ingrowth, and also provides a highly potential clinical candidate for bone tissue engineering.

CRediT authorship contribution statement

Peng Yu: Writing – original draft, Validation, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Fei-Fei Li: Validation, Resources, Methodology, Investigation, Conceptualization. Fan-Yuan Yu: Software, Project administration, Methodology, Investigation, Data curation. Zheng-Min Zhang: Resources, Methodology, Investigation, Data curation, Conceptualization. Si-Yu Long: Software, Methodology, Investigation, Formal analysis, Conceptualization. Yun-Fei Tian: Visualization, Supervision, Software, Methodology, Investigation. Jing-Qiao Guo: Supervision, Project administration, Investigation, Funding acquisition, Conceptualization. Wei Yang: Writing – review & editing, Validation, Supervision, Software, Resources, Funding acquisition.

Ethics approval and consent to participate

All procedures in the experiment were approved by the Medical Ethics Committee of West China Stomatological Hospital Sichuan University (WCHSIRB-D-2025-253).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article.