3D Printed SiOC(N) Ceramic Scaffolds for Bone Tissue Regeneration: Improved Osteogenic Differentiation of Human Bone Marrow-Derived Mesenchymal Stem Cells

Bone tissue engineering has developed significantly in recent years as there has been increasing demand for bone substitutes due to trauma, cancer, arthritis, and infections. The scaffolds for bone regeneration need to be mechanically stable and have a 3D architecture with interconnected pores. With the advances in additive manufacturing technology, these requirements can be fulfilled by 3D printing scaffolds with controlled geometry and porosity using a low-cost multistep process. The scaffolds, however, must also be bioactive to promote the environment for the cells to regenerate into bone tissue. To determine if a low-cost 3D printing method for bespoke SiOC(N) porous structures can regenerate bone, these structures were tested for osteointegration potential by using human mesenchymal stem cells (hMSCs). This includes checking the general biocompatibilities under the osteogenic differentiation environment (cell proliferation and metabolism). Moreover, cell morphology was observed by confocal microscopy, and gene expressions on typical osteogenic markers at different stages for bone formation were determined by real-time PCR. The results of the study showed the pore size of the scaffolds had a significant impact on differentiation. A certain range of pore size could stimulate osteogenic differentiation, thus promoting bone regrowth and regeneration.


Introduction
Silicon-based ceramic scaffolds for bone regeneration are one of the main strategies used for the treatment of bone loss and large-scale bone defects [1][2][3][4][5]. Current studies show that silicon is an essential element for bone development and formation [6,7]. Silicon-based materials play an important role in the surface bioactivity through the exchange of ions at the scaffold-tissue interface, which results in the formation of a layer, similar to the mineral phase of bone [8]. Silicon possesses similar properties to phosphorus when it comes to bone formation and development [9]. Previous research has shown that the expression of some osteogenesis-related genes, e.g., alkaline phosphatase (ALP), bone morphogenetic protein-2(BMP-2), and collagen type I (Col I), are affected by silicon [10]. Silicon also contributes to the promotion of early deposition of apatite, the growth of osteoblasts, and some genes that control the induction of cell cycles and progression which enhance osteogenesis [11]. Enhanced apatite mineralization ability, biocompatibility, and bioactivity are also seen. A (small pores) and 500 microns (large pores) to analyze the effect of pore size on bioactivity ( Figure 1). They were first printed on an open source RepRap-class FFF 3D printer (Lulzbot TAZ 6, Fargo Additive Manufacturing Equipment 3D, Fargo, ND, USA) using NinjaFlex TPU with a 0.15 mm nozzle. The scaffolds were printed with the pore sizes of 500 and 700 microns since the pyrolysis process results in 25-30% shrinkage. They were then impregnated with a solution of acetone, polysilazane preceramic polymer (Durazane 1800, Merck Gmbh, Darmstadt, Germany), and catalyst platinum divinyltetramethyldisiloxane complex, Pt 2% in xylene (CAS number: 68478-92-2, Sigma-Aldrich, St. Louis, MO, USA). The discs were then dried for 24 h in the air and pyrolyzed in a tube furnace (Gero tube furnace) at 1200 • C for 1 h in a nitrogen atmosphere with 400 cc/min flow. The obtained scaffolds were first rinsed with deionized water (DI water) and then sterilized by autoclave at 121 • C for 15 min.

Preparation of Ceramic Scaffolds
The samples were designed [44] (OnShape, PTC) as porous discs of 13 mm diameter and 1.5 mm thickness. The samples were designed with different pore sizes of 300 microns (small pores) and 500 microns (large pores) to analyze the effect of pore size on bioactivity ( Figure 1). They were first printed on an open source RepRap-class FFF 3D printer (Lulzbot TAZ 6, Fargo Additive Manufacturing Equipment 3D, Fargo, ND, USA) using Nin-jaFlex TPU with a 0.15 mm nozzle. The scaffolds were printed with the pore sizes of 500 and 700 microns since the pyrolysis process results in 25-30% shrinkage. They were then impregnated with a solution of acetone, polysilazane preceramic polymer (Durazane 1800, Merck Gmbh, Darmstadt, Germany), and catalyst platinum divinyltetramethyldisiloxane complex, Pt 2% in xylene (CAS number: 68478-92-2, Sigma-Aldrich, St. Louis, MO, USA). The discs were then dried for 24 h in the air and pyrolyzed in a tube furnace (Gero tube furnace) at 1200 °C for 1 h in a nitrogen atmosphere with 400 cc/min flow. The obtained scaffolds were first rinsed with deionized water (DI water) and then sterilized by autoclave at 121 °C for 15 min.

Silicon Release
The release of silicon ions was tested by submerging the samples in PBS solution (EuroClone). The samples were submerged into 5 mL of PBS solution and kept in the incubator at 37°C. At each time point, a 2 mL solution was extracted and replaced by a fresh PBS solution. The extracted solution was tested for silicon ions using ion coupled plasma employing optical emission spectroscopy (ICP-OES) (SPECTRO Analytical Instruments, Kleve, Germany). Eight replicates were used for each pore size at each time point. The final results were calculated as the total released amount.

Cell Culture
Human bone marrow-derived mesenchymal stem cell line (hMSCs, ATCC number: PCS-500-012) was cultured in -MEM medium supplemented with 10% Fetal Bovine Serum (FBS, EuroClone) and 1% Antibiotic/Antimycotic (AA, EuroClone), in a humidified atmosphere of 5% CO2 at 37 °C. The medium was changed every two days. Once they reached 70% confluence, the cells were detached by 1% trypsin-EDTA solution, counted, and re-suspended in standard medium with the concentration of 500,000 cells/mL.

Silicon Release
The release of silicon ions was tested by submerging the samples in PBS solution (EuroClone). The samples were submerged into 5 mL of PBS solution and kept in the incubator at 37 • C. At each time point, a 2 mL solution was extracted and replaced by a fresh PBS solution. The extracted solution was tested for silicon ions using ion coupled plasma employing optical emission spectroscopy (ICP-OES) (SPECTRO Analytical Instruments, Kleve, Germany). Eight replicates were used for each pore size at each time point. The final results were calculated as the total released amount.

Cell Culture
Human bone marrow-derived mesenchymal stem cell line (hMSCs, ATCC number: PCS-500-012) was cultured in α-MEM medium supplemented with 10% Fetal Bovine Serum (FBS, EuroClone) and 1% Antibiotic/Antimycotic (AA, EuroClone), in a humidified atmosphere of 5% CO 2 at 37 • C. The medium was changed every two days. Once they reached 70% confluence, the cells were detached by 1% trypsin-EDTA solution, counted, and re-suspended in standard medium with the concentration of 500,000 cells/mL.

Cell Seeding and Differentiation
After placing sterilized samples into 24-well plates, 0.6 mL cell suspensions (standard medium) were added directly to the samples (300,000 cell/well) and tissue culture plates (TCP), which were the control group (only cells without samples). The plates were incubated in a humidified atmosphere of 5% CO 2 at 37 • C to promote cell adhesion. Twenty-four hours after the seeding, the samples with cells were moved to new plates and the medium was switched into the differentiation medium: standard medium with 0.1 µM dexametha-sone (DEX), 0.1 mM ascorbic acid 2-phosphate (AP) and 10 mM β-Glycerophosphate (BGP). The differentiation medium was changed every two days until 21 days.

AlamarBlue Assay
The cells' viability activity on different pore sizes (300 µm and 500 µm) after 7 days, 14 days, and 21 days of culture was determined with AlamarBlue Cell Viability assay (Invitrogen, Carlsbad, CA, USA), which quantifies cellular metabolic activity. AlamarBlue reagent was added directly to each well at 10% of the cell culture medium volume. Then, the well plates were incubated at 37 • C in a humidified atmosphere with 5% CO 2 for 3 h. From each well, 100 µL of the solution was collected. The fluorescence signal was measured with a Tecan Infinite 200 microplate reader (Tecan Group, Männedorf, Switzerland) with an excitation wavelength of 560 nm and an emission wavelength of 590 nm. TCP was used as the control group and eight replicates were considered for each experimental condition.

DNA Quantification Assay
To evaluate cell proliferation on the different pore sizes (300 µm and 500 µm), a PicoGreen DNA quantification assay (Quant-iT PicoGreen dsDNA Assay, Invitrogen, Carlsbad, CA, USA) was used. TCP was used as the control group. After 7 days, 14 days, and 21 days of culture, the culture medium was removed, and the samples were washed with PBS. Samples were then covered with 300 µL of 0.05% Triton-X PBS solution and incubated at 37 • C for 1 h. Before analysis, the samples were sonicated for 10 s with a Hielscher ultrasonic homogenizer (UP400S, 400 W-24 kHz, cycle 1, amplitude 40%, from Hielscher Ultrasonics, Teltow, Germany). Subsequently, 100 µL supernatant of each sample was placed in a black 96-well plate and mixed with 100 µL of PicoGreen working solution, prepared following the manufacturer's instructions. Fluorescence intensity was measured with a Tecan Infinite 200 microplate reader (Tecan Group, Männedorf, Switzerland) using an excitation wavelength of 485 nm and an emission wavelength of 535 nm. A calibration curve was created using a double-stranded DNA standard provided by the kit and was used for the calculation of the DNA content. Finally, the approximate number of cells per sample was determined from DNA content by the conversion factor of 7.7 pg DNA per cell. Eight replicates were considered for each pore size at each time point.

Alkaline Phosphatase (ALP) Activity Assay
ALP activity of the cells on scaffolds with different pore sizes was evaluated by the Alkaline Phosphatase assay kit (abcam, Cambridge, UK). The preparation procedure of the samples is the same as for the DNA quantification assay, in which the supernatant was obtained after washing, incubating, and sonication. Non-fluorescent 4-methylumbelliferyl phosphate disodium salt (MUP) substrate and MUP reaction solution were prepared following the instructions of the manufacturer. The reaction wells were set up in a black 96-well plate by mixing supernatant with MUP reaction solution and stop solution, using the volume suggested by the instruction. A standard curve was also created using the ALP enzyme. The fluorescence intensity was measured at excitation wavelength 485 nm and emission wavelength 535 nm. The readings of the samples were applied to the standard curve to obtain the amount of MUP generated by the ALP sample, and the activity of ALP in the tested samples was calculated by dividing the amount of 4-MU by the volume of the sample. Eight replicates were considered for each pore size at each time point.

RNA Isolation
Total mRNA was isolated from the scaffolds directly by NucleoZLO reagent (MACHEREY-NAGEL, Düren, Germany) according to the protocol from the manufacturer. The isolated RNA of the samples was dissolved in 20 µL RNase-free water; the final concentration of RNA was determined by a NanoDrop (ND-1000 Spectrophotometer, Thermo Fisher Scientific, Waltham, MA, USA) and then diluted into 10 ng/µL.

cDNA Synthesis
The isolated RNA was reverse transcribed into cDNA by iScropt Reverse Transcription Supermix kit (BIO-RAD, Hercules, CA, USA). Then, 4 µL iScript RT Supermix was mixed with 16 µL of isolated RNA (total RNA 160 ng) for each reverse transcription reaction well. Then, the complete reaction mix was incubated in Bio-Rad CFX96 Touch (BIO-RAD, Hercules, CA, USA) using a thermal cycling protocol provided by the manufacturer.

Gene Expression by Quantitative Real-Time PCR (RT-qPCR)
The quantification of gene expression was performed by Bio-Rad CFX96 Touch (BIO-RAD, USA). SsoAdvanced Universal SYBR Green Supermix kit was used, and the primer assays used in this study were listed in Table 1. The tested samples were mixtures that consisted of 5 µL SsoAdvanced Universal SYBR Green Supermix, 0.5 µL primer, and 5 µL cDNA sample, which led to a final amount of 40 ng cDNA per well. The PCR amplification was carried out as follows: polymerase activation and DNA denaturation at 95 • C for 42 s, followed by 40 cycles at 60 • C and 30 s for each cycle. Then, the melt curve was performed between 95 to 65 • C with 0.5 • C increments at 2 to 5 s/step. The PCR results were relatively quantified with the comparative ∆∆C T method by CFX Manager Software, comparing to the housekeeping mRNA expression of glyceraldehyde-3 phosphate dehydrogenase (GAPDH). The analysis of each gene was processed in duplicate and there were eight replicates for each sample. Cell morphology and distribution were visualized by Oregon green phalloidin and 4 6diamidino-2-phenylindole (DAPI) staining. Oregon green phalloidin stains actin filaments of cytoskeleton resulting in green fluorescence while DAPI stains nuclei resulting in blue fluorescence. After 7 days, 14 days, and 21 days of culture, the cell-seeded samples were fixed with 4% paraformaldehyde, washed three times with PBS, and then were permeabilized using 0.2% Triton X-100 PBS solution for 30 min. After washing with PBS 3 times (15 min each time), cells were incubated in Oregon green phalloidin (5.0 µL/well) and DAPI (1.0 mL/well, 5.4 µL dilute in 25.0 mL PBS) solution for 1 h at room temperature. After three rinses with PBS, samples were observed using Zeiss LSM 510 Meta confocal laser scanning microscope.

Statistical Analysis
GraphPad Prism 9 (La Jolla, CA, USA) was used for statistical analysis for all the data obtained from each independent experiment. Where applicable, data were expressed as mean ± SD. The statistical analysis was performed by two-way ANOVA using the all-pair-wise multiple comparison procedure, in which * p < 0.05 were set as the level of significance.

Structural Characterization of SiOC(N) Ceramic Scaffolds
Two different pore sizes (500 µm and 300 µm) ( Figure 2) were selected to mimic the pore size of human bone. The theoretical surface area of the porous discs was calculated using OnShape and the results are shown in Table 2. The structure of the final products was observed by an optical microscope. Theoretically, the total surface of discs with small pores is 1.26 times larger and the top/bottom surface is 1.36 times larger, compared to the total surface area of discs with large pores.
wise multiple comparison procedure, in which * p < 0.05 were set as the level of significance.

Structural Characterization of SiOC(N) Ceramic Scaffolds
Two different pore sizes (500 µm and 300 µm) ( Figure 2) were selected to mimic the pore size of human bone. The theoretical surface area of the porous discs was calculated using OnShape and the results are shown in Table 2. The structure of the final products was observed by an optical microscope. Theoretically, the total surface of discs with small pores is 1.26 times larger and the top/bottom surface is 1.36 times larger, compared to the total surface area of discs with large pores.

Silicon Iron Release of SiOC(N) Ceramic Scaffolds
The release of silicon ion was monitored for 21 days by ICP-OES, and the total amount of the released silicon ion was calculated and plotted in Figure 3. Scaffolds with large pore size and small pore size had almost the same releasing curve and amount. Until day 10, the amount of released silicon kept increasing, and after 10 days of release, the released amount of Si ion reached a plateau.

Silicon Iron Release of SiOC(N) Ceramic Scaffolds
The release of silicon ion was monitored for 21 days by ICP-OES, and the total amount of the released silicon ion was calculated and plotted in Figure 3. Scaffolds with large pore size and small pore size had almost the same releasing curve and amount. Until day 10, the amount of released silicon kept increasing, and after 10 days of release, the released amount of Si ion reached a plateau.

Characterization of Proliferation, Metabolism of hMSCs during Osteogenic Differentiation
Cell proliferation and metabolic activity of hMSCs seeding on SiOC(N) ceramic scaffolds with different pore sizes were performed on day 7, day 14, and day 21 in the presence of the osteogenic medium, using PicoGreen DNA quantification assay and Alamar-Blue assay, respectively. The results are shown in Figure 4. The small pore size scaffolds induced a higher proliferation rate than the large ones. Interestingly, the cell number did not change over time in all the groups during 21 days of culture. Metabolic activity was normalized by cell numbers for each group at each time point. Except for day 7, in which the data of large and small pore size scaffolds showed no significant difference, cells cultured on the large pore size scaffold showed, in general, a higher metabolism activity than

Characterization of Proliferation, Metabolism of hMSCs during Osteogenic Differentiation
Cell proliferation and metabolic activity of hMSCs seeding on SiOC(N) ceramic scaffolds with different pore sizes were performed on day 7, day 14, and day 21 in the presence of the osteogenic medium, using PicoGreen DNA quantification assay and AlamarBlue assay, respectively. The results are shown in Figure 4. The small pore size scaffolds induced a higher proliferation rate than the large ones. Interestingly, the cell number did not change over time in all the groups during 21 days of culture. Metabolic activity was normalized by cell numbers for each group at each time point. Except for day 7, in which the data of large and small pore size scaffolds showed no significant difference, cells cultured on the large pore size scaffold showed, in general, a higher metabolism activity than those cultured of small pore size scaffold, in particular on day 21.
Cell proliferation and metabolic activity of hMSCs seeding on SiOC(N) ceramic scaffolds with different pore sizes were performed on day 7, day 14, and day 21 in the presence of the osteogenic medium, using PicoGreen DNA quantification assay and Alamar-Blue assay, respectively. The results are shown in Figure 4. The small pore size scaffolds induced a higher proliferation rate than the large ones. Interestingly, the cell number did not change over time in all the groups during 21 days of culture. Metabolic activity was normalized by cell numbers for each group at each time point. Except for day 7, in which the data of large and small pore size scaffolds showed no significant difference, cells cultured on the large pore size scaffold showed, in general, a higher metabolism activity than those cultured of small pore size scaffold, in particular on day 21.

Visualization of Cell Morphology and Distribution by Confocal Laser Scanning Microscopy
The confocal images were taken in both low and high magnification on scaffolds with different pore sizes ( Figure 5). Cell morphologies during the osteogenic differentiation were evaluated by staining the cell cytoskeleton with Oregon green phalloidin (green). In general, the scaffolds could promote very good cell adhesion during the cell culture period. In the first week, the adhered cells morphology exhibited a high degree of spreading with many cell/cell connections due to filopodia and lamellipodia formation (Figure 5b,d) in all groups, which was even more evident on small pore size scaffolds (Figure 5d). Instead, at 2-3 weeks of culture, adhered cells of all the groups penetrated in the pores, colonizing the 3D structure. A difference in cell behavior, however, was observed during the 3 weeks of culture. Despite a large number of cells on the scaffolds, the spatial organization of cells was quite different. On the large pore size scaffolds, the cells were aligning on the surfaces, but were not able to cross the empty space. However, the cellular structure of the small pore size scaffold was small enough for the cells to bridge the pore and the pores were almost filled by cells over time.

Characterization of ALP Activity of hMSCs during Osteogenic Differentiation
ALP production is widely used as a marker of bone cells because it is associated with osteoblastic differentiation. In this work, ALP activity was measured by the Alkaline Phosphatase assay kit, and the results are presented in Figure 6. In the first two weeks of differentiation, both groups had similar levels of ALP activity. A dramatic increase in the ALP activity occurred on day 21, which was more than two times higher than on day 7 and day 14, which is particularly relevant on the small pore size scaffold on day 21.

Characterization of ALP Activity of hMSCs during Osteogenic Differentiation
ALP production is widely used as a marker of bone cells because it is associated with osteoblastic differentiation. In this work, ALP activity was measured by the Alkaline Phosphatase assay kit, and the results are presented in Figure 6. In the first two weeks of differentiation, both groups had similar levels of ALP activity. A dramatic increase in the ALP activity occurred on day 21, which was more than two times higher than on day 7 and day 14, which is particularly relevant on the small pore size scaffold on day 21.

Expression of Osteogenic Marker Genes in the Presence of the SiOC(N) Ceramic Scaffold
The relative mRNA expression levels of the osteogenic markers: alkaline phosphatase (ALP), collagen type I (COL 1), runt-related transcription factor 2 (RUNX2), and osteonectin (SPARC) were monitored on day 7, 14, and 21 using RT-qPCR. The total RNA was isolated from cells seeded on SiOC(N) scaffolds with different pore sizes and cultured in the osteogenic medium. The results were presented in Figure 7 and the heat map of the gene expression overview were presented in Figure 8. Figure 6. ALP activity of hMSCs on large and small pore size SiOC(N) ceramic scaffold in osteogenic medium for 7, 14, and 21 days. n = 8; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001.

Expression of Osteogenic Marker Genes in the Presence of the SiOC(N) Ceramic Scaffold
The relative mRNA expression levels of the osteogenic markers: alkaline phosphatase (ALP), collagen type I (COL 1), runt-related transcription factor 2 (RUNX2), and osteonectin (SPARC) were monitored on day 7, 14, and 21 using RT-qPCR. The total RNA was isolated from cells seeded on SiOC(N) scaffolds with different pore sizes and cultured in the osteogenic medium. The results were presented in Figure 7 and the heat map of the gene expression overview were presented in Figure 8.  Generally, the gene expression trends of the selected markers were similar in both sample groups (Figure 7) and these trends were different from TCP during the 21 days of culture (Supplementary Materials, Figure S1). An increasing level of ALP expression was observed in both groups. It should be noted that the ALP expression of the small pore size scaffold was significantly higher at the first experimental time point (day 7), compared with the large pore size samples. Additionally, both groups reached a similar expression level of ALP on day 21. The expression of COL 1 for both groups showed a peak on day 14 and dropped almost to zero on day 21. No significant differences showed in the expression of COL 1 during 21 days between the two groups at the same time point. A downregulation of RUNX2 was observed in both groups during the three weeks, but the small pore size scaffold showed a higher level in the first week compared with the large pore size group. The expression of osteonectin (SPARC), for both groups, showed the same trend along with the three weeks of culture and similar levels. Generally, the gene expression trends of the selected markers were similar in both sample groups (Figure 7) and these trends were different from TCP during the 21 days of culture (Supplementary Materials, Figure S1). An increasing level of ALP expression was observed in both groups. It should be noted that the ALP expression of the small pore size scaffold was significantly higher at the first experimental time point (day 7), compared with the large pore size samples. Additionally, both groups reached a similar expression level of ALP on day 21. The expression of COL 1 for both groups showed a peak on day 14 and dropped almost to zero on day 21. No significant differences showed in the expression of COL 1 during 21 days between the two groups at the same time point. A downregulation of RUNX2 was observed in both groups during the three weeks, but the small pore size scaffold showed a higher level in the first week compared with the large pore size group. The expression of osteonectin (SPARC), for both groups, showed the same trend along with the three weeks of culture and similar levels.

Discussion
Tissue engineering is based on the use of instructive scaffolds that should provide 3D templates for initial cell attachment, proliferation, and subsequent tissue formation. Therefore, the materials used as well as the geometry of the scaffold have relevant impacts on the biological performance. In our previous study [40], the fabrication procedures were optimized and the physical, mechanical, chemical, and preliminary biological properties of the SiOC(N) ceramic scaffolds were already thoroughly characterized. The results demonstrated that the novel method eliminates the possibility of having pores or defects inside the struts with complete impregnation of the preceramic polymer in the TPU structures. Another issue with biomedical implant structures is that they are very expensive to manufacture. The research also showed that the scaffolds can be manufactured, with high reproducibility and industrial tolerances, at an affordable and thus accessible price. The preliminary biological studies proved the material to be non-cytotoxic, promote fast cell adhesion, and early-stage cell activations [40]. In the present study, the research is focused on the impact of the geometry of the scaffold on the osteogenic differentiation of hMSCs.
Different biological results were obtained by comparing two groups of scaffolds that are different in total surface areas and pore sizes, and, according to the literature, these

Discussion
Tissue engineering is based on the use of instructive scaffolds that should provide 3D templates for initial cell attachment, proliferation, and subsequent tissue formation. Therefore, the materials used as well as the geometry of the scaffold have relevant impacts on the biological performance. In our previous study [40], the fabrication procedures were optimized and the physical, mechanical, chemical, and preliminary biological properties of the SiOC(N) ceramic scaffolds were already thoroughly characterized. The results demonstrated that the novel method eliminates the possibility of having pores or defects inside the struts with complete impregnation of the preceramic polymer in the TPU structures. Another issue with biomedical implant structures is that they are very expensive to manufacture. The research also showed that the scaffolds can be manufactured, with high reproducibility and industrial tolerances, at an affordable and thus accessible price. The preliminary biological studies proved the material to be non-cytotoxic, promote fast cell adhesion, and early-stage cell activations [40]. In the present study, the research is focused on the impact of the geometry of the scaffold on the osteogenic differentiation of hMSCs.
Different biological results were obtained by comparing two groups of scaffolds that are different in total surface areas and pore sizes, and, according to the literature, these differences can affect the releasing amount of silicon ions. In the current work, despite the scaffold with the small pore size having a larger surface area (26% more) the results of silicon ion release showed that there is no significant difference in the release amount between the two groups ( Figure 3). This suggests that the only parameter that causes differences in cell behaviors is the pore size (in terms of morphology, proliferation, metabolic activity, ALP activity, and gene expression).
The ability of the materials to induce good cell adhesion and proliferation was confirmed with a higher cell number on small pore size scaffolds due to the larger surface area available. It is interesting to observe that, for both groups, cell numbers (Figure 4a) from one week to three weeks remained constant, demonstrating that the cells were differentiating in the very early stage (first week). As reported in the literature [45], during the differentiation process, cell proliferation and differentiation are two interdependent processes that have a counteracting relationship, and this correlated with the obtained results. Cell spatial organization, detected by confocal images (Figure 5), shows that the cells on the small pore size scaffolds were able to build a 3D network, migrating into the structure and connecting to each other crossing the pores (Figure 5d,h,i). The 3D cell distribution is a more physiological cell organization, promoting stem cell differentiation into the bone cells [46]. When the pore size is too large, the 3D structure is not able to drive cells forming the 3D organization, slowing down the differentiation. In fact, this is confirmed by the gene expression (Figure 7) that the cells on small pore size scaffolds were able to differentiate in an earlier stage, particularly in the expression of ALP (Figure 7a) and RUNX2 (Figure 7c). The geometry differences in the scaffolds also had an impact on the cell activities during the osteogenic differentiation. In the presented work, general and phenotypic-specific metabolism were determined by AlamarBlue assay and ALP production, respectively. It is reported in the literature that an increase in ALP activity should be associated with osteoblastic differentiation. ALP is thought to increase and then decrease when mineralization is well progressed [47]. The cells on small pore size scaffolds showed a lower general metabolic activity (Figure 4b), but higher ALP activity ( Figure 6) compared to cells on large pore size scaffolds, especially on day 21. The significant increase in this enzyme between the second and third week suggested that cells on small pore size scaffolds were shifting to a more differentiated state.
The mRNA levels of osteogenic genes (ALP, COL 1, RUNX2, and SPARC) were evaluated during the 21 days of differentiation, using RT-qPCR. The positive impact of Si-based materials on osteogenic differentiation has been investigated in numerous studies [48,49], in which the water-soluble silicon was proved to enhance osteoblast proliferation and differentiation under in vitro conditions. The investigated expression levels of osteoblastspecific marker genes in terms of osteoblast differentiation, have shown a special pattern that could help to have a perception of the functional bone construct (Figure 8). RUNX2, a member of the runt homology domain transcription factor family, plays a crucial role in osteoblast development. It is a major gene responsible for the early orientation of stem cells towards osteoblastic lineage and directly activates the transcription of genes such as osteocalcin, COL 1, and ALP. Moreover, it is typically downregulated in three or four weeks of culture, which is an important indicator of matrix maturation and mineralization [50]. In fact, there was a significant decrease in the expression of RUNX2 observed in both groups during the second and the third week, suggesting the acceleration of the differentiation. COL 1 plays an important role in biomineralization. It is expressed in high levels near the end of the proliferation state and during the period of matrix deposition. It is also known to decrease with time and an ongoing calcification of the bone tissue [51]. The increase in COL 1 expression showed a peak on day 14 and then dropped to a very low level in both scaffolds between the second and the third week. This observation is in accordance with the expression profiles reported in the literature for the osteogenic differentiation of hMSCs [52]. The ALP gene is first detected in osteoblast progenitor cells which are committed to differentiate into osteoblasts. The expression of ALP is dependent on the maturation stages of osteoblasts. It increases when mineralization is well progressed and then decreases in later stages [53]. In this case, upregulation of ALP expression was observed in both scaffolds in 21 days. Although, the expression level showed significant differences on day 21 between the samples with large pore size and small pore size. It is worth noting that hMSCs on small pore size expressed a significantly high level of ALP on day 7 compared to that on large pore size, suggesting the cells on the small pore size scaffolds had an earlier differentiation. Osteonectin (SPARC) is a phosphorylated glycoprotein that plays a role in regulating the initiation, and promotion of mineralization as well as crystal growth [54]. In this work, it is revealed that the expression of osteonectin is increased in hMSCs during differentiation on both groups, this may be due to the osteoprogenitor culture conditions.

Conclusions
In summary, the bioactivity and osteogenic differentiation ability of this open-source 3D printing process for SiOC(N) ceramic scaffolds with different pore sizes was investigated by in vitro cell culture of hMSCs. The results showed that the material of the scaffold has a good ability to release water-soluble silicon ions and the total amount of released silicon ions was not affected by the pore size. Moreover, the scaffolds were able to improve the differentiation while retaining the cell number. It has been demonstrated that the pore size of the scaffold has a strong impact on cell behaviors including cell number, metabolism, ALP activity, distribution, and the speed of osteogenic differentiation. This open-source 3D printing process for SiOC(N) ceramic scaffolds is promising and provides opportunities to have more complex and precise structural matrices with controllable bioactivity for bone regeneration applications. Future work, including in vivo testing, is warranted in order to improve the technology to be used as bone implants.