3D bioprinted autologous bone particle scaffolds for cranioplasty promote bone regeneration with both implanted and native BMSCs

Although autologous bone (AB) grafting is considered to be the gold standard for cranioplasty, unresolved problems remain, such as surgical-site infections and bone flap absorption. In this study, an AB scaffold was constructed via three-dimensional (3D) bedside-bioprinting technology and used for cranioplasty. To simulate the skull structure, a polycaprolactone shell was designed as an external lamina, and 3D-printed AB and a bone marrow-derived mesenchymal stem cell (BMSC) hydrogel was used to mimic cancellous bone for bone regeneration. Our in vitro results showed that the scaffold exhibited excellent cellular affinity and promoted osteogenic differentiation of BMSCs in both two-dimensional and 3D culture systems. The scaffold was implanted in beagle dog cranial defects for up to 9 months, and the scaffold promoted new bone and osteoid formation. Further in vivo studies indicated that transplanted BMSCs differentiated into vascular endothelium, cartilage, and bone tissues, whereas native BMSCs were recruited into the defect. The results of this study provide a method for bedside bioprinting of a cranioplasty scaffold for bone regeneration, which opens up another window for clinical applications of 3D printing in the future.


Introduction
Decompressive craniectomy is a procedure that is performed to alleviate increased intracranial pressure resulting from different causes (e.g. head injury, hypertensive cerebral hemorrhage, etc) that can lead to postoperative cranial defects [1]. Moreover, wars and terrorist activities can cause craniofacial trauma, resulting in craniofacial or fronto-orbital cranial defects [2,3]. These cranial defects seriously affect a patient's physical and psychological state. Timely and early cranioplasty can restore aesthetics, protect the brain tissue, and solve psychological problems [4][5][6]. However, selecting appropriate patient-specific cranioplasty materials is crucial for the success of surgery and a favorable prognosis. Previous findings showed that currently used cranioplasty materials (such as autologous bone (AB) grafts, titanium plates, and calcium phosphate) can partially meet the repair requirements [6,7]. Among them, AB grafts with non-immunogenic properties, good biocompatibility, suitable hardness, and heat insulation can reduce the financial pressure on patients, can provide considerable shape, and are the gold standard for cranioplasty. However, some problems associated with AB grafts are unresolved, such as surgical-site infection (9.2%) and aseptic bone resorption (19.7%) [8]. Therefore, making full use of AB grafts and effectively solving the complications has become an urgent problem to solve.
The ideal cranioplasty materials for treating cranial defects should have good biocompatibility, nonmagnetic properties, anti-infection properties, low thermal conductivity, high mechanical strength, and plasticity [9,10]. However, the cranioplasty materials commonly used in clinical practice, such as titanium plates and polyetheretherketone, cannot meet the needs of bone regeneration. The advantages of threedimensional (3D) printing in the field of regenerative medicine include an improved ability to build specific complex shapes and the potential of providing better regeneration effects through the addition of bioactive components [11,12] instead of bone resorption.
The motivation for the present research came from our observations in daily clinical practice that many AB flaps are discarded after large bone flap decompressive craniectomy or debridement, which is quite wasteful. However, AB particles combined with 3D-bioprinting technology can be used to create a cell-laden cranioplasty material to promote bone regeneration. Nevertheless, owing to the limitations of 3D-printing processing, it is very important to select suitable printing materials that can retain biological activity. In our previous cranioplasty experiments, we combined poly (ε-caprolactone) (PCL) with micron-sized AB for use with multilevel customized 3D-printing strategies, which confirmed the capacity of AB to promote the osteogenic differentiation of stem cells [13]. However, this method had drawbacks similar to those of most current 3D-bioprinting experiments. Currently, 3D implants are usually first printed on the bench and then applied to experimental animals after in vitro culture. This approach prone to various problems: (1) cell-unfriendly solvents used for postprocessing of printed scaffolds limit the combination of cell printing and AB printing; (2) the structures of bioactive components are easily damaged during in vitro culture; and (3) maintaining a highly sterile environment during in vitro culture is difficult [14][15][16][17]. To better meet the requirements of clinical applications, in this study, the implants were used immediately after printing without additional postprocessing. To explore new possibilities for increasing the clinical utility of 3D-bioprinting technology, we developed bedside bioprinting (BBP) as a novel 3D-bioprinting scenario. A 3D-printed AB cranioplasty-composite scaffold laden with autologous bone marrow-derived mesenchymal stem cells (BMSCs) was used to quickly perform BBP during beagle dog cranioplasty. Namely, the 'soft and hardware assembly method' was adopted by assembling and cross-printing 'soft' and 'hard' materials, in order to provide biological cells and biological factors that can promote osteogenesis, maintain short-term mechanical support, and solve the problem of AB resorption. Additionally, the mechanism of osteogenesis was investigated.

Experimental animals
All animal procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication number 85- 23, revised 1996) and approved by the Air Force Medical University (approval number IACUC-20210120). For this study, male beagle dogs (18-20 months old, 15-17 kg) were obtained from Dilepu Biomedical Co., Ltd (Xi'an, China, license number SCXK 2019-002) and maintained under clean conditions. The animals were housed under a controlled temperature of 25 • C± 1 • C, a relative humidity of 40%-70%, and a 12 h light/dark cycle, with ad libitum access to food and water.

Experimental groups
In vitro studies were performed using both twodimensional (2D) and 3D culture systems with same conditions, in addition to scaffolds in 3D culture systems. The 2D and 3D systems are divided into BMSC and AB/BMSC two groups, respectively. Beagle dogs were randomized into the following five treatment groups (n = 3/group): (1) cranial defect without cranioplasty (Ctrl group); (2) cranioplasty using a PCL shell (PCL group); (3) cranioplasty using a PCL shell and an AB scaffold (PCL/AB group); (4) cranioplasty using a PCL shell and a BMSC scaffold (PCL/BMSC group); and (5) cranioplasty using a PCL shell, AB, and BMSC scaffold (PCL/AB/BMSC group), as illustrated figure 1.

Canine model of cranial defects
After anesthetization with 2%-3% isoflurane (Hengrui, Lianyungang, China) in oxygen, the beagle dogs were placed in a sterile console in a prone position. A 2 cm diameter full-thickness cranial defect was made in the flat part of the head which was 2.5 cm to the right of the midline and 2.5 cm behind the anterior fontanelle (supplemental figure 2(A)). Subsequently, a separate layer suture was made, and the wound was disinfected. Postoperative cranial bone flaps were marked according to the numbers assigned to the dogs, collected, and stored at −80 • C for AB preparation.

BMSC preparation
BMSC isolation was performed in accordance with a published method [13]. After anesthetization with 2%-3% isoflurane (Hengrui) in oxygen, the beagle dogs were placed in a sterile console in a prone position. A 2 cm incision was made in the thigh of each experimental animal, and the femur was exposed after the muscles and fascia were separated. A hole with a diameter of 0.5 cm at the flat part of the femur was drilled into the bone marrow cavity, and 8 ml of bone marrow was extracted using a syringe rinsed with heparin (1000 U l −1 ). Subsequently, the hole was sealed with bone waxer full-thickness cranial de and the surgical wound was sutured and disinfected with iodophor. BMSCs were obtained by density-gradient centrifugation with Histopaque-1077 (10771, Sigma) [13] and according to the numbers assigned to the dogs. After isolation, primary BMSCs were cultured at 37 • C in 5% CO 2 and subcultured to passage 3 before carrying out the following experiments.

AB preparation
AB were prepared as described previously [13]. Briefly, postoperative cranial bone flaps from beagle dogs were crushed into 0.5 × 0.5 cm 2 pieces and dialyzed with ultrapure water for 24 h, and the water was refreshed every 8 h. The cleaned bone fragments were lyophilized at −80 • C and 1 Pa for 24 h (LYOQUEST-85, Telstar, Spain). The bone fragments were then placed into a grinding jar, pre-cooled with liquid nitrogen for 10 min, and ground into AB in a vibration grinder (GRINDER GT200, Beijing Grinder Instrument Co., Ltd, China).

Rheological analysis
A MCR 302 rheometer (Anton Paar, Austria) with a parallel plate (PP25) was used for rheological examination to characterize the thermosensitivity and shear thinning behavior of inks. The gap distance was 0.5 mm and 300 µl stock solutions (37 • C) was added for each test. To test the thermosensitivity, a temperature sweep from 37 to 4 • C in oscillation mode (1% strain, 1.5 Hz) was performed to observe the changes of the storage (G ′ ) and loss modulus (G ′′ ) over time. To test the shear thinning behavior of hydrogels, the added stock solutions were kept at 15 • C for 15 min to gelation and then the oscillation amplitude sweep (0.1%-2000% strain, 1.5 Hz) was performed.

Extrusion experiments
First, the independent inks were transferred into respective syringes, and then loaded to the 3D printer. The temperature controller was set at 15 • C for Alg-AB and Alg-Gel-AB ink, and 20 • C for Alg-Gel ink, respectively. After 20-25 min incubation, the extrusion began with a speed of 0.07 ml min −1 and the extrusion process was recorded by the smartphone.

Printing path design and 3D printing
3D printing path of the hybrid scaffold was performed using a multi-nozzle bioprinter (Livprint Norm, Medprin, China), as illustrated in supplemental figure 3(A). Briefly, the AB ink and BMSC ink were designed to print alternatively and parallelly within the same layer. And the nozzle rose 0.2 mm for printing the next layer. The printing platform was maintained at 4 • C. A 25G nozzle was used for BMSC bioink printing at 19 • C-21 • C, whereas a 22G nozzle was used for AB ink printing at 15 • C. The linear speed for printing was held constant at 5 mm s −1 , and the extrusion speed was 0.07 ml min −1 . After printing, the scaffolds were cross-linked by immersion in a 3% calcium chloride solution for 3 min. To better distinguish alternately printed microfilaments, a demonstration scaffold was designed with 4-fold magnification of the in vitro scaffold (supplemental figures 3(B)-(D)). The implant scaffolds for the in vivo experiments were designed as 20 × 20 × 2.1 mm square grids or cylinders printed using BMSC bioink and AB ink (figure 2(C)). For the PCL scaffold printing, the printing parameters were set as below: 40% infill, 27 G printing nozzle, 0.2 mm layer height and 10 mm s −1 printing speed.

Cell viability of BMSCs within scaffolds
BSMC survival in scaffolds was measured using the Calcein-AM/PI Double Stain Kit (40747ES76, Yeasen, China). After incubation at 37 • C for 15 min, the scaffolds were photographed under a fluorescent microscope (FV10i; Olympus, Tokyo, Japan) in the same setting. After incubation at 37 • C for 45 min, the scaffolds were photographed under a confocal fluorescence microscope (FV1000; Olympus).

Scanning-electron microscopy (SEM) analysis
The morphologies of the scaffolds were observed by SEM. Scaffolds were fixed using 2.5% glutaraldehyde and dehydrated using gradient alcohol. After drying to the critical point, the samples were sprayed with Pt, and cell images were captured using SEM.

Quantitative real-time polymerase chain reaction (PCR) analysis
A PCR assay was performed to detect osteogenic mRNA expression. The scaffolds were dissolved in 55 mM trisodium citrate and 20 mM ethylenediaminetetraacetic acid (EDTA), and the BMSCs in the scaffolds were collected. RNA was extracted from BMSCs using the TRIzol reagent (15596026, Thermo Fisher, USA). RNA was then transformed into complementary DNA using a reverse transcription kit (11123ES60, Yeasen). Real-time PCR was performed using an RT-qPCR SYBR Green Kit (11143ES70, Yeasen) and the Applied Biosystems 7500 Real-Time PCR Detection system. The primers used for quantitative real-time PCR are shown in supplementary table 1, and GAPDH expression was detected as a reference gene.

Alkaline phosphatase (ALP)-activity assay
ALP-activity assays were performed to evaluate osteogenic differentiation according to the specifications of ALP Assay Kit (P0321S, Beyotime, China).
The scaffolds were dissolved in 55 mM trisodium citrate and 20 mM EDTA, and the BMSCs in the scaffolds were collected. To remove the dissolved remnants of trisodium citrate and EDTA, BMSCs were washed with 1 × phosphate buffered saline (PBS) three times before detection. The BMSCs were incubated at 37 • C for 10 min in detection buffer (provided in the kit). Subsequently, the absorbance was determined at 405 nm, and ALP activity was calculated.

Cranioplasty
In each case, the skull cranioplasty was performed 3 months after the craniectomy. For the BBP operation, our team was divided into two groups of team members, where one group performed 3D printing and the other group performed the cranioplasty surgery. After the 3D scaffolds were printed in the operating room, they were immediately implanted into the cranial defects of the beagle dogs. Attention was paid to the temperature and sterile conditions during the procedure. For cranioplasty, after anesthetization with 2%-3% isoflurane (Hengrui) in oxygen, the beagle dogs were placed on a sterile console in a prone position. A flap was made to cover the cranial defects before different cranioplasty materials (including the 3D-bioprinted scaffolds) were implanted into the cranial defects of different groups and fixed with sutures. The periosteum was continuously sutured for cranioplasty-material fixation, and the wound was closed and disinfected (supplemental figures 2(B) and (C)). The beagle dogs were sacrificed 3 or 9 months after cranioplasty, and each calvaria with a graft was harvested, rinsed with normal saline, and fixed in 4% paraformaldehyde for subsequent tests.

Micro-computed tomography (micro-CT) testing
Micro-CT testing was performed to evaluate bone regeneration. The specimens were scanned and imaged using a micro-CT scanner (ZKKS-MCT-Sharp; China). Reconstruction was performed with isotropic 20 µm voxels using a ZKKS-Micro-CT 3.0 instrument (ZKKS, China) and analyzed using ZKKS-Micro-CT 4.1 (ZKKS). Micro-CT scans were performed 1 cm outside the perimeter of the cranial defect and the region of interest (ROI) was examined as a 2 cm diameter circle. And the implanted area of the sample was set as the ROI. After all the samples were analyzed, the threshold was adjusted to optimally reproduce the bone structure in the ROI. The part of the ROI where the x-ray absorptivity was greater than the threshold was identified as the bone. The 3D parameters of the trabecular bone analysis were the bone volume per tissue volume (BV/TV; %), trabecular number (Tb.N; 1 mm −1 ), trabecular spacing (Tb.Sp; mm).

Computed tomography (CT) testing
CT scans were performed with all animals in different groups using a multi-detector CT scanner (760, United Imaging, China), and the images were archived in the Digital Imaging and Communications in Medicine format. Axial CT images were taken using 1 mm thick slices. Animals were fasted overnight and anesthetized with 2%-3% isoflurane in oxygen, after which CT examinations were performed in the supine position. Radiant software was used to measure the axial CT value at each level of the cranioplasty area, and the average was used as the CT value of the sample. ImageJ software was used to measure the maximum cross-sectional area of the cranioplasty area of the reconstructed 3D CT image at the same observation angle and to calculate the percentage of the cranioplasty area to obtain the relative defect area.
To ensure repeatability and consistency of the measurements, three researchers performed the measurements independently.

Enzyme-linked immunosorbent assay (ELISA) analysis
SDF1 ELISAs were performed to detect the expression of SDF1. The process was performed using the Mouse CXCL12/SDF-1 ELISA Kit (KE10049, Proteintech), and 100 µl of each standard and sample was added to the appropriate wells. A cover seal was used to seal each plate by firmly pressing it onto the top of the microwells. The plates were incubated for 2 h at 37 • C in a humid environment. After the wells were washed, 100 µl of 1 × Detection Antibody Solution (provided in the kit) was added, and the plates were incubated for 1 h at 37 • C. After treatment with the horseradish peroxidase-conjugated antibody and TMB substrate solution, stop solution was added, and the absorbance of the plate was read on a microplate reader at a wavelength of 450 nm.

Statistical analysis
The above experiments were independently repeated at least three times, and the results are shown as the mean ± the standard error of the mean (SE). When analyzing three sets of variables, Brown-Forsythe tests were used to test the homogeneity of variances, and one-way analysis of variance and Tukey's posthoc tests were used to analyze statistical differences. Student's two-tailed t-test was used to compare differences between two groups. Differences were analyzed using GraphPad Prism 8.0 and considered statistically significant at a threshold of p < 0.05. Power analysis was performed using PASS 15 software (NCSS, USA) and the results are shown in supplemental table 2.

Characterization of BMSCs
BMSCs were obtained from beagle dogs by centrifuging and purifying animal femur bone marrow blood, and optical images of the cells are shown in supplemental figure 1(A). HLA-DR, CD34, CD45, CD90 and other CD proteins are important surface markers for BMSCs, and flow cytometric analysis revealed the following expression profile for the extracted BMSCs: HLA-DR-/CD19-/CD11b-/CD45-/CD34-/CD73+/CD90+/CD105+, which corresponds to the characteristics of stem cells (supplemental figure  1(B)). Next, differentiation identification was performed with the BMSCs. The BMSCs were cultured in osteogenic differentiation medium for 2 weeks before ALP staining and were found to be strongly positive for ALP (supplemental figure 1(C)). Alizarin Red staining was performed 1 month after the osteogenic culture, and multiple calcified nodules were observed (supplemental figure 1(C)). These results suggest that BMSCs can underwent osteogenic differentiation. In addition, chondrogenic and adipogenic differentiation were observed after the BMSCs microspheres were cultured in chondrogenic differentiation medium for 1 month to detect their chondrogenic differentiation ability. Alcian Blue staining revealed that the BMSCs exhibited chondrocyte characteristics. Numerous lipid droplets were found after 1 month of in vitro adipogenic-induced culture (supplemental figure 1(C)). These results show that the BMSCs generated in this study had the characteristics of BMSCs and were isolated with good purity.

Ink printability and AB-bioprinted scaffolds
The rheological properties and extrusion experiment were conducted to assess the ink printability. The result of temperature sweeps was shown in figure 2(A) (i) where G ′ kept larger than G ′′ in the Alg-AB group whereas G ′′ increased to surpass G ′ in both Alg-Gel and Alg-Gel-AB groups as the temperature decreased. Three types of stock solution were set at 15 • C for 15 min and then the strain sweeps were executed. The results of strain sweeps in figure 2(A) (ii) showed that G ′ increased at the low-strain region (0.1% to ≈150%) and dramatically decreased at the highstrain region (≈150%-2000%) in Alg-Gel and Alg-Gel-AB groups (gelation has initiated). The transition point of G ′ and G ′′ appeared in Alg-Gel-AB groups (≈270%) and in Alg-Gel (≈300%) whereas no transition point appeared in Alg-AB group where the G ′′ maintained higher than G ′ . The results of extrusion experiments showed that the Alg-AB ink broke to an irregular droplet after extrusion whereas the Alg-Gel-AB and Alg-Gel ink formed continuous fibers (as shown in figure 2(B)). A multi-nozzle 3D bioprinter was used for scaffold manufacturing. As shown in supplemental figure 3(A), the green lines represent the print path of the BMSC bioink, and the blue lines represent the print path of the AB ink. A grid structure was constructed via interleaved layerby-layer printing. The white structure represents the AB microfilament, and the red structure represents the BMSC microfilament. As shown in figure 2(C), the printed scaffold was self-supporting with a distinct layer in the vertical direction. The opaque ABprinted structure and the transparent AB -bioprinted structure were observed under a light microscope, which revealed encapsulated BMSCs in transparent structures (figure 2(D)). SEM analysis showed that the BMSCs on the surface of the scaffold were spherical after printing. Because the cells were encapsulated within the hydrogel, they could not be directly observed, and only a rounded protrusion wrapped in the microfilaments was visible (figure 2(E)). In addition, several wrinkles were present in the AB hydrogels, in contrast to the smooth BMSC hydrogels (figure 2(E)). Complete AB were observed by SEM (figure 2(E)), indicating that the structure of the AB remained as intact as possible.

The 3D-bioprinted AB scaffolds possessed good cellular affinity
The 3D-printed AB/BMSC scaffolds was cultured in vitro for 21 d, and the cellular affinity was determined. To confirm the viability of the BMSCs within the scaffold, calcein-AM/PI staining was performed. Multiple live BMSCs (green) were found in the scaffold, accounting for the vast majority of the cells (figures 3(A) and (B)). Only a few dead BMSCs (red) were observed during the 21 d of in vitro culture, indicating that the BMSCs were healthy after printing. However, on days 14 and 21 of in vitro culture, the number of live BMSCs decreased, although the number of dead BMSCs did not increase ( figure 3(A)). Several BMSCs migrated from the scaffold to the bottom of the culture dish. The cellular morphology in the 3D-printed hydrogel scaffold was observed via SEM, and the results indicated that the BMSCs within the hydrogel were clearly visible, indicating that BMSCs survived and remained in good condition during the 21 d period of in vitro culture ( figure 3(A)).

AB promoted the osteogenic differentiation of BMSCs in vitro
As a control, the osteogenic differentiation effect of AB was evaluated using a 2D culture system ( figure 4(A)). ALP staining demonstrated the feasibility of promoting osteogenic BMSC differentiation ( figure 4(B)). Moreover, PCR analysis indicated that the AB partially promoted the osteogenic differentiation of BMSCs in the 2D culture system (figure 4(C)). For the 3D culture system, we printed BMSC and AB/BMSC hybrid scaffolds, digested the scaffolds and collected the cells after 2 weeks of culture in osteogenic induction medium, and performed follow-up experiments ( figure 4(D)). We found that the ALP activity in the AB/BMSC group was higher than that in the BMSC group ( figure 4(E)). In addition, PCR was performed to detect the mRNA-expression levels of other osteogenic markers (including OCN, BMP2, and RUNX2). The OCN, BMP2, and RUNX2 mRNA-expression data paralleled the ALP activities (figure 4(F)), demonstrating that AB promoted osteogenic differentiation within the AB/BMSC scaffold.

AB-bioprinted scaffolds promoted in vivo bone regeneration 3 months after cranioplasty
To observe the conditions of the scaffolds at an early stage, we performed sampling and histological staining 3 months after cranioplasty (figure 5). HE staining, Masson's trichrome staining, and Safranin O-Fast Green staining showed different degrees of tissue repair in the defects of all experimental groups. The defects were covered by fibrous tissue in the control group, whereas osteoid tissues and new bone tissues grew in the defects in the other experimental groups. Punctate, mature bone tissues were observed in the PCL/AB/BMSC group (figure 5), indicating that the scaffolds were laden with BMSCs and AB had good in vivo bone regeneration potential. In other groups, only various degrees of maturity were observed in terms of osteoid structures, rather than mature bone tissues. To further detect the bone-regeneration ability, osteogenesisand chondrogenesis-related immunohistochemical staining was performed ( figure 6). We detected expression of the cartilage markers, Acan and Col2, in the osteoid region by immunohistochemistry and found that they were expressed at higher levels in the PCL/AB and PCL/BMSC groups than in the PCL group, demonstrating that AB and encapsulated BMSCs promoted chondrogenesis. The expression levels of Acan and Col2 were highest in the PCL/-AB/BMSC group ( figure 6). In addition, immunohistochemical staining for OCN and Col1 (ossification markers) showed that BMSCs and AB could promote Col1 expression. As a calcification marker, OCN did not change significantly during the early repair stages in all groups ( figure 6). The immunohistochemicalstaining results for OCN and Col1 were similar to those obtained with other stains; that is, composite scaffolds laden with BMSCs and AB better promoted in vivo-defect repair through bone and cartilage regeneration than that of PCL/BMSC and PCL/AB scaffolds.

AB and BMSC-composite scaffolds promoted in vivo bone regeneration 9 months after cranioplasty
In addition, we examined the effect of in vivo bone regeneration 9 months after cranioplasty to observe the long-term repair conditions of the scaffolds. HE staining revealed light red and dark red osteoid structures in the Ctrl group, and Masson and Safranin O-Fast Green staining confirmed that these structures were immature osteoid structures (figure 7). No mature bone tissue was observed in the PCL group, with only osteoid structures found in the defects, similar to that in the Ctrl group. In the PCL/BMSC group, HE staining revealed local formation of massive dark red bone tissue. Masson and Safranin O-Fast green staining confirmed that the bone tissue had tissue characteristics representative of mature bone (figure 7). In the PCL/AB group, new bone tissue was only visible around the native bone, and Masson and Safranin O-Fast Green staining indicated that the maturity of degree of this bone tissue was inconsistent. In the PCL/AB/BMSC group, HE staining showed that dark red bone tissue covered most of the defect, and a large block of new bone was found in the dark red area. Masson's trichrome and Safranin O-Fast Green staining showed that the surface and deep bone tissues were more mature (figure 7). The above results showed that the PCL/AB/BMSC scaffolds formed massive mature bone-tissue structures within 9 months.
Micro-CT has been widely used in bone tissue-engineering research as the 'gold standard' for evaluating bone morphology and microarchitectures [21,22]. Micro-CT scans were performed 1 cm outside the perimeter of the cranial defect and the ROI was delineated as a circle with a 2 cm diameter ( figure 8(A)). The relative bone volume fractions of the cranioplasty groups were higher than that of the Ctrl group, indicating that both PCL and other cranioplasty materials can promote bone repair in vivo. The relative bone volume fractions in the PCL/AB and PCL/BMSC groups were higher than that in the PCL group, indicating that adding AB or BMSCs to the scaffolds could also promote bone regeneration in vivo ( figure 8(B)). As shown in figure 8(B), the relative bone volume fraction in the PCL/AB/BMSC group was higher than those in the PCL/AB and PCL/BMSC groups, indicating that AB combined with BMSCs synergistically promoted in vivo osteogenesis. In terms of the number of trabecular bones and the degree of separation, the differences among the groups were not as obvious as the relative bone volume fractions (figures 8(C) and (D)). In conclusion, the above-mentioned histological staining and micro-CT experiments showed that the composite scaffold promoted bone regeneration at 9 months in vivo. In addition, CT images were obtained 7 months after cranioplasty. As shown in supplemental figure 4, at 3 months after cranioplasty, the CT values for the cranioplasty areas in the PCL/-AB/BMSC, PCL/BMSC, and PCL/AB groups were higher than that of the control group, and the relative defect area also decreased (p < 0.05). However, the CT value and relative defect area did not change in the PCL group (p < 0.05). At 7 months after cranioplasty, the CT value in the PCL/AB/BMSC group was higher than those of the PCL/BMSC and PCL/AB groups, and the relative defect area became smaller (p < 0.05). The CT-scanning data provided another kind of evidence that the 3D-bioprinted scaffolds, especially the PCL/AB/BMSC scaffold, promoted bone regeneration.

BMSCs in scaffolds differentiated into vascular endothelium, cartilage, and bone tissues
Before the scaffolds were transplanted into beagles, the BMSCs were pre-transfected for stem cell tracking. Nine months after implanting Green fluorescent protein (GFP)-labeled BMSC scaffolds, the materials were obtained and the slices were stained. First, CD31 (a vascular endothelial marker) and GFP colocalization was analyzed, which showed that the transplanted BMSCs had different degrees of survival ( figure 9(A)). Only a small number of CD31-positive vascular endothelial cells were observed in the PCL group, indicating that the vascular endothelium was spontaneously generated without implanting BMSCs. The number of CD31-and GFP-positive cells in the PCL/BMSC group was higher than that in the PCL group, and some GFP-labeled BMSCs showed CD31 positivity, suggesting that some BMSCs differentiated into vascular endothelium tissue in vivo ( figure 9(A)). More BMSCs survived and were CD31-positive in the PCL/AB/BMSC group than in the PCL/AB and PCL/BMSC groups, indicating that AB promoted the proliferation and differentiation of stem cells in vivo. To verify that the implanted BSMCs could differentiate into bone tissue, we co-stained Col1-positive cells for GFP. Indeed, the BMSCs differentiated into bone tissue, indicating that the encapsulated BMSCs were in good condition and could survive in situ differentiation in vivo. In the PCL/BMSC group, a small number of GFP-positive cells were observed in Col1 red-stained new bone regions, indicating that BMSCs Figure 6. Immunohistochemistry analysis of bone regeneration in cranial defects at 3 months post-surgery. Immunohistochemistry images of osteogenic-related proteins (OCN and Col1) and chondrogenic-related proteins (Acan and Col2). The mean immunohistochemical density was recorded as the integral optical density per unit area. The data shown are presented as the mean ± SE (n = 3). * p < 0.05, ns: p ⩾ 0.05 IOD: integrated optical density. underwent osteogenic differentiation in vivo and generated some of the new bones, although some of the new bones was not generated from the implanted BMSCs ( figure 9(B)). The number of Col1-positive cells in the PCL/AB/BMSC group was higher than that in the PCL/BMSC group, indicating that AB promote osteogenic BMSC differentiation in vivo. In addition, we co-stained the scaffolds with Col2 and GFP and found results characteristic of vascular endothelial and bone tissues, where BMSCs in the scaffolds could differentiate into Col2-positive chondrocytes in vivo (figure 9(C)). Numerous chondrocytes were found in the PCL/AB/BMSC group that differentiated from GFP-labeled BMSCs, indicating that the PCL/AB/BMSC scaffolds promoted chondrogenesis well ( figure 9(C)). These results show that BMSCs successfully differentiated into vascular endothelium, cartilage, and bone tissues in vivo and that AB promoted these differentiation processes.

AB bioprinted scaffolds recruited native BMSCs
The results presented above showed that, although the implanted BMSCs could differentiate into bone tissue and participate in bone regeneration, some bone tissues did not differentiate from the transplanted BMSCs, indicating that they were derived from an endogenous source. To clarify whether ABbioprinted scaffolds recruited native stem cells, we co-stained BMSCs with CD90 and CD105, which are stem cell markers. In the PCL/BMSC group, many GFP+ implanted BMSCs survived, and some CD90+/CD105+/GFP-cells were scattered around the GFP+ cells, which were recruited native stem cells (figures 10(A) and (B)). Compared with that of the PCL/BMSC and PCL/AB group, more native stem cells (CD90+/CD105+/GFP-cells) were encapsulated around GFP+ cells in the PCL/AB/BMSC group (figures 10(A) and (B)), indicating that addition of AB and BMSCs jointly promoted native stem cell recruitment. SDF1 is a chemokine that is responsible for stem cell migration and recruitment. Therefore, we stained tissue sections for SDF1 and found that adding BMSCs and AB increased its expression ( figure 11(A)). These results show that bioprinted scaffolds with BMSCs and AB can promote SDF1 expression in vivo and recruit native stem cells. Therefore, we constructed a co-culture system of in vitro scaffolds and BMSCs, with the BMSCs or AB/BMSC scaffold placed in the lower chamber and BMSCs placed in the upper chamber for 14 d of incubation in osteogenic medium ( figure 11(B)). After 14 d, the old medium was replaced with fresh medium, and after 40 h of co-culture, the cell supernatant in the upper chamber was aspirated, and SDF1 expression was detected by ELISA analysis. The addition of AB promoted SDF1 expression in this 3D BMSC culture system ( figure 11(C)). Crystal violet staining of the Transwell polycarbonate membranes was also performed after 40 h of co-incubation, revealing that the BMSCs migrated from the upper chamber to the lower chamber. We found that BMSC migration in the AB/BMSC group was greater than that in the BMSCs group, indicating that the addition of AB promoted the migration ( figure 11(D)). In summary, the above in vivo and in vitro experiments jointly show that AB/BMSC-bioprinted scaffolds can recruit native BMSCs.

Anti-inflammatory effect of AB bioprinted scaffolds
It was found that PCL/BMSC, PCL/AB and PCL/-AB/BMSC reduced inflammation compared with that of PCL group ( figure 12). In addition, the expression of anti-inflammatory cytokines (IL-10, IL-4) was increased in PCL/BMSC, PCL/AB and PCL/-AB/BMSC group (figure 12). All these results implied that BMSCs and AB in scaffolds possessed antiinflammatory effects.

Discussion
To meet the requirements of clinical applications, we implemented the BBP strategy to fabricate a bionic scaffold for cranioplasty. Before the technology matures and is applied, there are some key technical issues that need to be studied and verified, including printing materials, cells, material-cell binding methods, special equipment for printing, production process, process methods, performance evaluation, etc for bedside applications. We conducted preliminary research on the above problems, carried out relevant analysis, explored the relevant printing parameters and conditions, and verified the effect after printing. This work is a proof-of-concept study, focusing on the key technical issues, feasibility and important parameters of the new concept, and it is necessary to conduct a more comprehensive and systematic study of BBP for us in the future.
In our BBP strategy, the cranioplasty materials were individualized including AB and BMSC, which decreased the transplant rejection to the maximum extent and promoted new bone and osteoid formation, and even with native BMSCs recruitment and anti-inflammatory function. Moreover, since our scaffolds are made up of bioactive ingredients, not a foreign body, therefore, wound exudates, allergies, infections, etc resulted from foreign body cranioplasty materials such as titanium plates can be reduced. In addition, because our scaffolds can promote osteogenesis, in this way, can inevitably grow and fuse with the bone around the defect, so there will be no loosening or even shifting of cranioplasty materials. This occurs in clinical cases where cranioplasty materials are physically immobilized (e.g. skull locks, titanium nails, etc.) [23]. Furthermore, BBP sharply reduced the cost of cranioplasty. The cost of manually molded titanium implants was nearly 2000$, while the customized titanium and PEEK implants needed at least 30 000$, resulting in economic burden for patients and their families [24,25]. The cost of cranioplasty materials for BBP mainly included the reagents of   autogenous BMSCs extracting and culturing, consumable materials of 3D printing, and the labor cost. In the end, 6 weeks-6 months after decompressive craniectomy is generally considered as the proper time for cranioplasty in clinical practice [26,27]. In our BBP strategy, the preparatory time was 6 weeks-2 months for in vitro stem cells culture and AB particles and the printing time was less than 1 h. While the preparatory time for current clinical cranioplasty materials (titanium and PEEK) was 1 d-1 week for customization and aseptization. Both cranioplasty strategies can meet the clinical requirement and complete the customized implant (table 1). Although more cranioplasty strategies such as extracellular matrix (ECM) -inspired 3D printing, injectable cell-loaded hydrogel and cell-seeding scaffolds [27][28][29] have been developed recently, ECM-inspired 3D printing could not be used as stem cell-delivery vehicles; injectable cell-loaded hydrogel has not the ability to provide 3D architecture and mechanical stability; cell-seeding scaffolds increased the risk of infection when in vitro culture. The invention of BBP broke through the limitation of these strategies, for our BBP is performed surgically, which is important to reduce infection risk. And our BBP is available for both cell delivery and supporting ability.
At present, the difficulty of bone regeneration is mainly due to the lack of ideal stem cells and an internal environment that guides osteogenesis [30][31][32]. Our countermeasure was to induce bone regeneration by using BMSCs and micron-sized AB. As one of the most important types of stem cells in adults, BMSCs have the characteristics of a strong proliferation capacity, stability, easy extraction, and can be differentiated into osteoblasts [33]. AB grafts can promote bone regeneration through osteoconduction and osteoinduction [13]. The boneregeneration effect of AB is mainly due to their ability to induce osteogenic differentiation of stem cells, as verified in previous experiments by our group [13]. Previous data revealed that adding AB to scaffolds could promote bone regeneration [34][35][36]. In our previous study, we also found that AB could promote bone regeneration in skull defects [13], but the underlying mechanism was unclear. Here, we found that adding AB not only promoted the osteogenic differentiation of implanted BMSCs, but also recruited native BMSCs. In our experiments, sodium alginate mixed with AB showed good printability. Moreover, after printing the BMSCs into a composite scaffold, we found that the scaffold promoted BMSC osteogenesis in vivo and in vitro, confirming the osteoinductive effect of the AB. To enhance the ability of AB to promote stem cell differentiation, we also simultaneously printed the stem cell scaffold-AB scaffold cross-layer into a grid structure, with two adjacent layers of stem cells and AB in the scaffold being parallel to each other. In addition, the use of AB can minimize autoimmune rejection [37], preserve the biological activity, be easily accessible, and have high patient acceptance. BMSCs have been widely used in tissue engineering and regenerative medicine. In both animal and human studies, BMSCs have shown good capacity for promoting bone regeneration [38][39][40], although controversy remains over how BMSCs might contribute to bone regeneration in mechanistic terms. One theory is that BMSCs produce bioactive components that contribute to bone regeneration (mainly through paracrine activities after transplantation into tissues), and the other theory is that BMSCs can directly differentiate into osteoblasts to repair defective bone tissues [41]. Our research showed that BMSCs transplanted into the body can promote bone regeneration through in situ differentiation. Nine months after transplantation into animals, we found that GFP-labeled BMSCs differentiated into vascular endothelial, cartilage, and bone tissues, indicating that BMSCs in scaffolds can be directly differentiated into bone tissue to stimulate bone regeneration. These results are similar to those of other studies in which transplanted BMSCs promoted bone regeneration through osteogenic and chondrogenic differentiation [42][43][44] However, because the proliferation and differentiation abilities of BMSCs is often affected by various factors, such as the extraction method, extraction site, extraction age, number of passages, and culture method, the conclusions of relevant studies on BMSC transplantation have been inconsistent [39,41]. In the future, it will be necessary for relevant organizations to unify the experimental processes and standards for such research. In addition, we also found that the RUNX2 comparison between BMSCs and BP/BMSCs showed no significant difference. As the significant osteogenic transcription factor, RUNX2 is mainly expressed in preosteoblasts and immature osteoblasts [45,46]. In the early stage of osteogenesis, RUNX2 promotes the genetic expression of ALP, OCN and Col1. Conversely, RUNX2 inhibits expression of osteogenic gene in the final stage [45,46]. Hence, the RUNX2 is the early biomarker of osteogenesis. Our previous study found that RUNX2 reached the maximal level in one week, and sharply decreased in two weeks [13]. It was also reported that there was no difference of RUNX2 expression in comparison with control group in 1-3 weeks [47,48]. We inferred that the BMSCs in this study arrived in the later stage of osteogenesis and reduced the expression of RUNX2.
We believed the advantages of the BBP strategy is the cooperation of stem cell and AB and the capacity for immediate clinical application after fabrication. To introduce the AB into cell printing, not only the printing process should be cell-friendly but also the postprocessing of printed implants should be nontoxic, rapid prototyping and convenient. Based our previous studies, alginate/gelatin composite hydrogel enables cell-laden functional structures printing with simple and rapid crosslinking [18,49]. So we applied the alginate/gelatin for BMSC printing. Inspired by the previous report which succeed in printing alginate/gelatin/hydroxyapatite composites [50], the introduction of gelatin and alginate may be adaptive for AB printing. However, the introduction of AB to cell ink directly may cause unavoidable damage to cells induced by the physical contact during the preparation and extrusion processes. Therefore, in this study, we applied the hybrid scaffold printing method where multi-nozzle devices were used to print the AB ink and cell ink, independently (See in supplemental figure 3). And we applied the alginate and gelatin hydrogel as the binder for AB printing, where the gelatin/alginate hydrogel was used for cell printing so that the BMSC and AB can be directly printed as integrated implants for immediate clinical application after printing and crosslinking. Before in vivo experiments, this hybrid scaffold printing strategy was explored and established systematically. We conducted the rheologic examination of Alg-Gel-AB ink and found that Alg-Gel-AB ink could gel at a low temperature and exhibit shear-thinning properties after gelling (as shown in figure 2(A)). Similar to the Alg-Gel ink, the Alg-Gel-AB ink can also form a continuous fiber after extrusion (as shown in figure 2(B)). And then we adopted the Alg-Gel-AB ink and Alg-Gel ink to print hybrid scaffolds and these resultant scaffolds maintained a stable structure after crosslinking (as shown in figure 2(C)). Overall, the thermosensitive gelatin can function as a disperser to enhance the printability of AB while the alginate can form a stable hydrogel scaffold after crosslinking with calcium ions (Ca 2+ ). Compared to the period of days required for post-processing, the hydrogel-based scaffold required only a few minutes for preparation, which greatly increases the feasibility of the BBP strategy. Gelatin and alginate have been widely used in clinical practice as hemostatic materials for drug delivery, wound repair, tissue engineering, and interventional therapy [51][52][53][54]. Therefore, we determined the feasibility of the this hybrid scaffold printing method in BBP strategy. However, gelatin-alginate has the disadvantage of insufficient mechanical strength. Although the skull does not play a load-bearing role as a support bone, the cranioplasty material should maintain sufficient strength to protect the brain. Therefore, we used PCL, which has been approved by the FDA for clinical applications [55,56], as the shell of the hydrogel grid scaffold. In fact, our design was aimed to enable the BBP strategy for cranioplasty and achieve bone regeneration by fabricating a bionic scaffold containing autogenous AB and BMSCs, and structures that simulate the physiological structures of flat bones. The PCL scaffold located in the outer layer simulated the outer plate of the skull, and the hydrogel located in the inner layer simulated the inner plate and diploe, which are believed to participate in bone regeneration and repair.
During the course of our experiments, we found an interesting phenomenon that only part of the new bone formed in a defect originated from BMSCs within the scaffold, and some bone tissues of unknown origin were also observed around the GFP-positive cells. We speculate that these bone tissues may have differentiated from native BMSCs. Perhaps our printed scaffold promoted the recruitment of native BMSCs. To examine this possibility, we stained cells for CD90, CD105, and GFP and found that a numerous native BMSCs were indeed recruited to the defects, indicating that our printed scaffold promoted bone regeneration through the direct differentiation of implanted BMSCs and the recruitment of native BMSCs. The effect of BMSCs on cell migration has been demonstrated in several studies [57,58]. The addition of ECM and osteogenetically active components can promote BMSC recruitment in vivo and in vitro [59,60]. In this study, both AB and BMSCs from the implanted scaffold could release bioactive components, which may help explain their recruitment of native BMSCs. SDF1 is considered the primary chemokinetic factor that promotes BMSC recruitment [58], and exogenous supplementation with SDF1 has been shown to promote the recruitment of native BMSCs in vivo and in vivo [57,61]. Interestingly, our in vivo SDF1 staining and ex vivo SDF1-expression assays conducted in this study demonstrated that SDF1 expression was promoted after the introduction of our printed scaffold, which plausibly led to native BMSC recruitment. Our micro-CT-and CT-scanning results showed that bone regeneration and bone formation occurred along the original bone edge toward the center, which may be related to the recruitment of native BMSCs. In addition, we detected pro-inflammatory and antiinflammatory cytokines, and found that PCL/BMSC, PCL/AB and PCL/AB/BMSC group reduced inflammation in comparison with PCL group. This result implied that BMSCs and AB in scaffolds possessed anti-inflammatory effect, which was consistent with past research [19,20].
This study has a couple of limitations. First, only GFP cell tracking and immunofluorescence were applied to verify that BMSCs differentiated into other tissues. More in vivo cell-tracking technologies need to be used in future research. Second, CT images could not be performed at 9 months after cranioplasty. Hence, further in vivo data are still necessary before conducting clinical experiments.
In this study, we constructed a new bioprinted scaffold for use in beagle cranioplasty that promoted in situ bone regeneration. More importantly, as a preclinical study of 3D printing applied to human patients with cranial defects, we integrated autologous AB and BMSC composite printing to perform on-site BBP during cranioplasty. The printed scaffold was placed into the cranial defect of the corresponding beagle immediately after printing. This approach opens up possibilities for clinical applications of 3D printing in the future.

Conclusion
A preclinical exploration of the BBP strategy was applied to generate bionic implantable scaffolds composed of AB particles and BMSCs for beagle cranioplasty in this study. Our findings support the following conclusions: (1) An implantable scaffold composed of AB and BMSCs was successfully printed using the BBP strategy. (2) The viability of BMSCs within the hybrid scaffold was maintained for 21 d in culture. (3) The feasibility of osteogenic BMSC differentiation in 3D culture systems was demonstrated. (4) The PCL/AB/BMSC group facilitated AB regeneration at 3 months after cranioplasty, accompanied by high expression of osteogenic and chondrogenic markers in osteoid structures. Nine months after cranioplasty, different degrees of new bone generation were observed in the PCL/BMSC and PCL/AB/BMSC groups, and the PCL/AB/BMSC scaffold showed the best osteogenic-regeneration efficiency. (5) In situ differentiation of transplanted BMSCs was detected. The recruitment of native BMSCs by the AB/BMSC scaffold was demonstrated. (6) The AB/BMSC scaffold jointly promoted osteogenic regeneration by mediating the differentiation of transplanted BMSCs and by stimulating the recruitment of native BMSCs. (7) The addition of BMSCs and AB in scaffolds partly inhibited the inflammation.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.

Conflict of interest
The authors declare that they have no competing interests.

Declarations
All experimental procedures were approved by the Institutional Animal Care and Use Committee of Air Force Military Medical University (No. IACUC-20 210120).