Surface Modified β-Tricalcium phosphate enhanced stem cell osteogenic differentiation in vitro and bone regeneration in vivo

A major number of studies have demonstrated Beta-tricalcium phosphate (β-TCP) biocompatibility, bioactivity, and osteoconductivity characteristics in bone regeneration. The aim of this research was to enhance β-TCP's biocompatibility, and evaluate its physicochemical properties by argon glow discharge plasma (GDP) plasma surface treatment without modifying its surface. Treated β-TCP was analyzed by scanning electron microscopy (SEM), energy-dispersive spectrometry, X-ray photoelectron spectroscopy (XPS), X-ray diffraction analysis, and Fourier transform infrared spectroscopy characterization. To evaluate treated β-TCP biocompatibility and osteoblastic differentiation, water-soluble tetrazolium salts-1 (WST-1), immunofluorescence, alkaline phosphatase (ALP) assay, and quantitative real-time polymerase chain reaction (QPCR) were done using human mesenchymal stem cells (hMSCs). The results indicated a slight enhancement of the β-TCP by GDP sputtering, which resulted in a higher Ca/P ratio (2.05) than the control. Furthermore, when compared with control β-TCP, we observed an improvement of WST-1 on all days (p < 0.05) as well as of ALP activity (day 7, p < 0.05), with up-regulation of ALP, osteocalcin, and Osteoprotegerin osteogenic genes in cells cultured with the treated β-TCP. XPS and SEM results indicated that treated β-TCP’s surface was not modified. In vivo, micro-computed tomography and histomorphometric analysis indicated that the β-TCP test managed to regenerate more new bone than the untreated β-TCP and control defects at 8 weeks (p < 0.05). Argon GDP treatment is a viable method for removing macro and micro particles of < 7 μm in size from β-TCP bigger particles surfaces and therefore improving its biocompatibility with slight surface roughness modification, enhancing hMSCs proliferation, osteoblastic differentiation, and stimulating more new bone formation.

Immunofluorescence. The hMSCs were prepared for immunofluorescence microscopy on days 1, 3, and 7, in 24-well Petri dishes (NUNCLON; Roskilde, Denmark) as previously described in cell culture. The hMSCs were washed using 2× phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde for 15 min. After fixation cells were washed three times with PBS for 10 min. Cells were permeabilized using 0.2% Triton X-100 for 20 min, washed three more times, blocked with 1% goat serum in PBS for 1 h, and incubated overnight with primary antibodies in 0.1% goat serum at 4 °C. The cells were washed three times and incubated with secondary antibodies in 0.1% goat serum for 2 h at ambient temperature. After 3× washes, samples were quenched with 0.5% (wt/vol) Sudan Black B (Sigma-Aldrich, St. Louis, MO, USA) for 10 min, nuclear counterstaining was performed using 300 μl of DAPI (0.1 μg/mL) for 10 min. When phalloidin staining was performed, 200 μl of phalloidin solution (methanol-based stock solution diluted in PBS once to obtain a 100-μl final concentration) was added for 15 min after staining with the secondary antibodies and two washing steps with PBS. Residual phalloidin removal was performed before mounting. The same methodology was utilized for three-dimensional (3D) immunofluorescence microscopy after cells were cultured in TCP control or β-TCP plasma-treated particles for 24 h. The immunofluorescent-labeled samples were placed on glass slides and viewed using the Olympus Fluoview FV-1000 confocal laser-scanning microscope (Olympus, Japan), equipped with a 40× oil objective. The fluorescence images from DAPI and Phalloidin were merged using Leica LAS X software.
Alkaline phosphatase assay. Alkaline phosphatase (ALP) activity was determined by modifying the previously reported methods 16 . After cell culture media was suctioned and DMEM + β-TCP control or β-TCP plasma treated media were added to the test wells. On days 1, 3, and 7, hMSCs were washed twice with PBS and resuspended in 300 μl of Triton-100 0.05%. The cells underwent three cycles of 5 min at 37 °C and 5 min at − 4 °C. Afterward, using the Thermo Scientific 1-Step p-nitrophenyl phosphate disodium salt (PNPP) protocol. 100 µL of the 1-Step PNPP was added to each 96-well plate and gently mixed. Following incubation at room temperature for 30 min. Next, the reaction was stopped by adding 0.4 M of NaOH, and the plate was read at a wavelength of 405 nm in the Multiskan GO microplate spectrophotometer assay reader (Thermo Fisher Scientific).
Real-time polymerase chain reaction. The cultures of hMSCs were completed on days 0 (only for control), 1, 3, and 7, as previously described, in cell cultures after the media were suctioned, and DMEM + β-TCP control or β-TCP plasma-treated media were added to the test wells. Total RNA was extracted using the Novel Total RNA Mini Kit (NOVELGENE, Molecular Biotech, Taiwan) under the conditions recommended by the manufacturer. The cells were trypsinized, harvested, and resuspended; subjected to cell lysis and RNA binding; and washed and eluted, as previously described 18,19 .
Subsequently, gene expression levels were normalized to the expression of the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase. The analysis results were expressed as time-course gene changes relative to the cell's genes cultured in DMEM only, and the calibrator sample representing the amount of transcript, was expressed on day 0 20 . After the design of multiple primers, ALP, OC, CatK, Rank, RankL, OPG using the Primer-BLAST from the U.S. National Library of Medicine. Reactions were run using 2 μl of cDNA in a 20-μL reaction volume on LightCycler 96 system (Roche Molecular Systems, Inc., Pleasanton, CA, USA) with Fast SYBR Green Master Mix (Thermo Fisher Scientific, Vilnius, Lithuania) ( Table 1). The reaction was repeated for 45 cycles; each cycle consisted of denaturing at 95 °C for 15 s and annealing, synthesis at 60 °C for 1 min and extension at 72 °C for 30 s, as per the manufacturer's instructions. The relative amounts of the transcript of the tested genes were normalized using a human glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Posterior quantification was performed using the delta-delta calculation method 16,21 . In vivo analysis. Surgical procedure. This in vivo animal experiment was conducted based on the AR-RIVE guidelines 22  Following an intramuscular anesthesia injection of Zoletil 50 (50 mg/mL) into the gluteal region at a dose of 15 mg/kg, the calvarial region was shaved and disinfected with iodine. Next, on the periphery of the calvaria for hemostatic and local anesthetic, 1.8 mL of 2% lidocaine with epinephrine 1/100,000 was injected.
A 2-cm long full depth incision was made in the linea media of the calvaria starting midway between the base of the ears. The pericranium was separated with a periosteal elevator from the outer table of the cranial vault 23 .
To prevent brain damage, the parietal bone was perforated three times using a 6.0-mm sterile trephine (3I Implant Innovation, Palm Beach Gardens, FL, USA) 24 . The first defect was filled with β-TCP control, and the second with β-TCP plasma-treated media. The third defect was left to heal as an unfilled control. Each animal was monitored closely until it fully recovery from anesthesia. The food intake, stool and urine output, and behavior of animals were monitored until they were sacrificed [25][26][27] .
Sample preparation. Five randomly selected animals were sacrificed at 2, 6, and 8 weeks after surgery were euthanized with intramuscular injection of Zoletil 50 (50 mg/mL) at 15 mg/kg and later CO 2 asphyxiation for 10 min. The monolithic blocks were extracted and immediately fixed in 10% formaldehyde for Micro CT, histological and histomorphometrical analysis.
Micro-computed tomography scanning of new bone formation. Sample blocks were prepared in formalin, and micro-computed tomography (micro-CT) scanning analyses were performed within 2 weeks using SkyScan 1076 (SKYSCAN, Antwerp, Belgium). After setting the micro-CT images, coronal images of the upper peripheral areas of the defect were saved in the database. To measure the tissue area/bone area, two-dimensional morphological analyses were performed. Thus, binary selections of samples from the morphometric analyses were made according to gray-scale density between units 20 and 80. The morphometric analyses were performed using SkyScan 1076 data-viewer software according to the manufacturer's instructions.
Histomorphometric analysis. Sections from all paraffin-embedded tissues were routinely stained with hematoxylin and eosin (H&E) and were simultaneously processed to reduce internal staining variations. The optical images of two mid-sections crossing the center of the calvaria defects were used to perform histomorphometrical analysis. For each histological section, the area occupied by the growing bone was identified and measured at a magnification of 200× using the ImageJ software, developed by the National Institutes of Health (NIH IMAGE, Bethesda, MD, USA) . These values were used to calculate the percentage of bone area/tissue area.

Statistical analyses.
All experiments shown in this study were conducted independently, and the results were presented as means ± standard deviation. Microsoft Excel Professional Plus 2016 (Microsoft Software, Redmond, WA, USA) was used for all quantitative statistical analyses. After the statistical analyses, the differences among the groups were compared and considered significant at *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001. The two-tailed Student t-test was used to compare between groups, respectively.

Results
SEM surface morphological observations. The GDP-treated β-TCP and nontreated β-TCP surfaces presented similar surface morphology (Fig. 1A,B). SEM results determined the surface morphology of the β-TCP test was not significantly affected by the argon GDP treatment. The GDP-treated β-TCP surface (20% ± 4.38% R a ) sample resembled that of the nontreated β-TCP sample (24% ± 4.93% R a ), but with slightly smoother rough surfaces within the structures (Fig. 1C,D). Furthermore, the number of β-TCP macro and micro particles measuring < 7 μm that were homogeneously distributed on a larger particle surface decreased after GDP treatment (Fig. 1A,B).  CTT CCT GAA AGA GGA TTGG GCC AGT ACT TGG GGT CTT TCT   OC  TCC TTT GGG GTT TGG CCT AC  CCA GCC TCC AGC ACT GTT TA   RANKL  ACT GGC CTC TCA CCT TTT CTG  AGC CAT CCA CCA TCG CTT TC   CatK  ACC CAC GGG AAG CAG TAC AA  GGC CTC AAG ATT ATG GAC AGAAA   OPG  CTG GAA CCC CAG AGC GAA AT  GCC TCC TCA CAC AGG GTA AC   RANK  GAA GGT GGA CTG GCT ACC AC  TTT CCT TCC CCT CCC CAG    www.nature.com/scientificreports/ X-ray photoelectron spectroscopy. The XPS analysis results of the surface chemistry and atomic concentrations of the specimens appear in Table 3. The mean (C1s) values of the surfaces were 18.85% ± 0.20% for the β-TCP test sample and 16.97% ± 0.21% for the β-TCP control sample. Moreover, the (O1s) values for the β-TCP test were 50.60% ± 0.54%, and β-TCP control 53.11% ± 0.66%, respectively. The (P2p) value was 11.63% ± 0.25% in the β-TCP control sample and 11.33% ± 0.09% in the β-TCP test sample. The mean (Ca2p) values were 15.48% ± 0.23% in the β-TCP test sample and 16.55% ± 0.24% in the β-TCP control sample (Table 3; Fig. 3).
XRD analyses. XRD measurements revealed the highly crystalline characteristic of β-TCP in general, as particle grafts before and after GDP treatment, presented similar sharp peak patterns. The β-TCP in both materials is responsible for the higher-intensity peaks (Fig. 4).
FTIR characterization. β-TCP test and β-TCP control particles revealed FTIR spectra results to agree with the XRD analysis. In terms of characteristics, there were no other types of crystalline phase in all samples in   www.nature.com/scientificreports/ addition to the apatite. Similar pronounced peaks close to 470-700 cm −1 were found in both β-TCP test and β-TCP control particles 28 . Furthermore, 543, 604, 1043, and 1120 cm −1 absorption band peaks belong to phosphatase peaks (Fig. 5).
Cell morphology. Morphological analysis via phalloidin/DAPI immunofluorescence staining of hMSCs revealed healthy growth and spreading on β-TCP control and GDP-treated β-TCP test. Most of the cells had a cytoplastic spindle shape and extended morphology with filopodial extensions. Cells spread and proliferated, making possible the formation of relatively thin continuous monolayers at 1, 3, and 7 days within some small   www.nature.com/scientificreports/ particles of the β-TCP control and β-TCP test present within the media in 2D cultured and same results after 24 h in 3D culture (Fig. 7).
Real-time polymerase chain reaction. We analyzed gene expression related to the osteogenic differentiation and function of the hMSCs. RNA was isolated separately from both the cells cultured in DMEM media mixed with β-TCP or β-TCP plasma treated. After a culture of 7 days, there was a significant difference in the expression of crucial osteoclast gene marker RankL and about two times CatK relative gene expression of hMSCs cultured in DMEM-β-TCP control and β-TCP plasma treated over cells cultured in DMEM only. There were also differences in the expression of important genes related to osteoblastic differentiation, such as ALP, Rank, Osteoprotegerin (OPG), and osteocalcin (OC; Fig. 8). ALP was expressed differently on cells cultured with β-TCP plasma-treated media: approximately four times higher relative gene expression over control and  www.nature.com/scientificreports/ β-TCP control cells. However, after 7 days, the expressions of OC and OPG were similar between cells cultured in DMEM media mixed with β-TCP or β-TCP plasma treated, but both values were similarly significantly higher than those of control cells (Fig. 9).

Histomorphometric analysis.
In the second week, it was visible in all the groups predominantly granulation tissue with slightly woven bone surrounded by inflammatory cells and osteoblasts. In the defects in which both types of β-TCP were used, direct contact between the bone and particles was minimum. Immature bone was mainly found at the defect's borders, with β-TCP particles serving as scaffolds for the newly regenerated woven bone. The control group regenerated 9.91% ± 3.24% new bone, close to the 9.11% ± 5.88% of β-TCP control, Though, the β-TCP test had the statistically significantly highest bone regeneration with 20.28% ± 9.85% (P < 0.05, Table 5; Fig. 11).
Higher new bone regeneration on the surface of some β-TCP test and control particles was observed at 6 weeks. Biomaterial granules with more advanced osteogenic activity than previous weeks was observed on the defect borders; areas where β-TCP test and control granules had more new bone within the particles. In contrast to previous weeks, β-TCP control generated 25.46% ± 14.03% more new bone, which was only statistically significantly different from control group (P < 0.05; control group: 14.36% ± 8.48% new bone). Also, the β-TCP test produced 18.59% ± 7.56% new bone, higher than the control, but with no statistically significant difference (P < 0.05, Table 5; Fig. 11).
Histology slides demonstrated resorption of the grafts' biomaterials at week 8 with similar new bone formation tendency as in week 2. The β-TCP test (29.67% ± 8.95%) had the highest new bone formation (P < 0.05). This new bone was mature in some areas and differentiated from control defects because of graft particles that created a    Table 5; Fig. 11).

Discussion
The β-TCP control samples used in the present work were identical to those used in different in vivo and in vitro clinical studies [2][3][4][5]29 , including a 10-year retrospective study that demonstrated reduced crestal bone loss around implants placed immediately after tooth extraction and grafted with β-TCP in extraction sockets 30 . In this previous study, 61 implants were evaluated in which immediate implant was made in fresh extraction sockets with β-TCP placement. The authors showed that β-TCP as a grafting material in immediate implant placement resulted in less bone loss around the implants in 72.1% 30 . Another prospective, multicenter randomized clinical trial evaluated histologic and histomorphometric results between autogenous bone, which is still considered the gold standard, and β-TCP control to avoid donor-site morbidity during bilateral sinus elevation. The study revealed that β-TCP control is a satisfactory graft material, even in the absence of autogenous bone 31 .
Previous studies have demonstrated that TCP surfaces can contain intrinsic impurities present in the starting material 32 . Previous guided tissue regeneration studies have attempted to improve β-TCP qualities 33 , including placing importance on TCP bioceramics surface cleaning for better performance. Four main effects of reactive gas surface treatment include surface cleaning, ablation or etching, surface chemical functionalization, and cross-linking. Each of the effects is present to some degree in a reactive gas process; the extent and degree of the effects depend on the process, energy and frequency, gas chemistry, reactor design, and operating parameters 34 . The GDP treatment on biomaterials can affect the mechanical and biological properties of ceramics during long-term evaluation 35 . Thus, β-TCP biocompatibility, bioactivity, and osteoconductivity characteristics in the present study were improved through surface treatment with argon GDP at 80 W, 13.56 MHz, 100 mTorr, for 15 min, and 10 mm distance of argon plasma to graft granules.
The apatite crystals and organic matrix occupied approximately one-and three-quarters of the bone volume, respectively, which were calculated using bone composition and the densities of the components 36 . These proportions show the importance of the physical and chemical characterization of β-TCP. Based on the SEM results (Fig. 1), the GDP-treated β-TCP surface (20 ± 4.38 R a ) sample had a slightly less roughness than the nontreated β-TCP sample (24 ± 4.93 R a ), which is typical for TCP sintered at temperatures of at least 900 °C and can provide good anchorage for cells 37,38 . Seidenstuecker et al. used a Keyence 3D laser scanning microscope (Keyence VK-X210) for measuring a particle's surface roughness with the same β-TCP biomaterial used in the current study 39 . The surface roughness was 26.08 R a , which was similar to that of the nontreated β-TCP sample. In addition, the number of β-TCP macro and micro particles measuring < 7 μm on larger β-TCP particles surfaces decreased. Previous studies have shown that extremely small particles may be immediately resorbed for bone formation 40 . Thus, the removal of these particles after GDP surface treatment is advantageous.
Other results, such as for EDS (Table 2; Fig. 2), demonstrated an increase in oxygen on the β-TCP test surface with 29.7 wt%, whereas the β-TCP control had 24.8 wt%. Despite having used the argon element to bombard the particle graft during the GDP surface treatment, the present study reached similar results to the findings of Roh and his team 41 . In the latter study, the effect of oxygen plasma treatment and the addition of nano-HA and β-TCP on the 3D poly(lactide-co-glycolide) (PLGA) scaffold was investigated. After treatment with oxygen plasma, the hydrophilicity of the scaffold surface increased, and its surface became rough because of the presence of oxygen functional groups on the scaffold surface. Base on their results, oxygen plasma-treated 3D PLGA/ nano-HA/β-TCP scaffolds showed higher bioactivity when compared to those of the control 41 . Regardless of www.nature.com/scientificreports/ these similar outcomes, plasma modification occurs in a particular mechanism for each gas studied, indicating that further studies are necessary 42 .
In addition to EDS, the results of the XRD analyses showed that the β-TCP test material exhibited a higher Ca/P ratio (2.05) than the β-TCP control material (Ca/P = 1.62). Nonetheless, both β-TCP samples had higher ratios than the expected stoichiometric TCP Ca/P ratio of 1.5 reported elsewhere 43 . A possible explanation for the higher Ca/P ratios in both β-TCP materials is the biomaterial sintering process. At the same time, the higher Ca/P ratio of the β-TCP test material compared with that of the control material can be associated only with the plasma sputtering process, which helps eliminate macro and micro particles of < 7 μm in size present on the sample's surface. The same characteristic peaks of dominant crystalline phases for both β-TCP materials are evident in Fig. 4, in which the two different expected stoichiometric TCP phases, α and β, which are less stable but more soluble than HA in aqueous environments, exhibit the same intense sharp peaks with the same peak width, indicating that both materials exhibit the same high crystallinity. In contrast to the poorly crystallized apatites of human bone 36 , both β-TCP samples exhibited greater crystallinity. This characteristic can reduce osteoclastic activity through the saturation of calcium ions that are dissolved continuously on the β-TCP surface [44][45][46] . This decrease in osteoclastic activity was seen in our study with down-regulation of RankL osteoclast-related gene expressions after 7 days (Fig. 9).
FTIR analysis supported the XRD findings by showing that both materials are apatites with no other crystalline phase, and both particles presented equal sharp peak patterns attributed to phosphatase band peaks. The XPS results indicated that the number of Si ions increased from 1.53% ± 0.46% in the β-TCP control to 2.138% ± 0.04% in the GDP-treated β-TCP. Previously, the addition of Si ions in hydroxyapatite when used as a biomaterial in bone grafting improved the bioactivity and enhancement of bone growth 47 . This finding could explain why WST-1 had statistically significant, superior, and faster cell proliferation, as observed in the β-TCP test sample compared with the β-TCP control sample after 7 days of testing. Using phalloidin/DAPI immunofluorescence staining, the morphological analysis of these cells indicated cells were elongated in appearance when cultured in 2D and 3D. In addition, the presence of Si ions could also explain the ALP results. After 3 days at all time sets, the ALP activity from the hMSCs cultured in β-TCP test media showed a greater increase than that from the hMSCs cultured in β-TCP control media. These results indicate that the plasma surface was able to increase the osteogenic differentiation of hMSCs in vitro.
Mesenchymal stem cells (MSCs) are multipotent cells that act as precursors to osteoblasts. The proliferation and osteogenic differentiation of MSCs are important for the maintenance of osteoblasts. ALP is an early osteoblast marker in the early stage of MSC osteogenic differentiation. However, OC is a gene expressed by mature osteoblasts at later stages of differentiation 48 . Osteoclastogenesis is a multistep process mainly under the control of the essential molecules RankL, RANK, and OPG, which makes the RankL/OPG ratio a major determinant of bone volume and health 49 . To analyze hMSCs osteogenic differentiation and Osteoclastogenesis, we performed quantitative real-time polymerase chain reaction, targeting established markers. We found up-regulation of ALP, OC, and OPG osteogenic genes relative to the control cells in β-TCP control media genes. RANK signaling presented significantly higher relative gene expression in β-TCP test media cell cultured over the other two groups, with a reduction observed after 7 days. These results are in accordance with the findings of Chen et al., who demonstrated that RANK is expressed in bone marrow MSCs and is decreased during osteogenic differentiation. RANK silencing significantly promotes, whereas overexpression suppresses, the osteoblast differentiation of bone marrow MSCs in vitro 50 . This suppression effect on osteoclast differentiation was partially attributed to OPG expression. We found a significant difference in the expression of key osteoclast gene markers RankL and CatK. The former was significantly higher in cells cultured with β-TCP test media than in those cultured with β-TCP control media and media alone and was later reduced to less than that of the control alone. CatK doubled in cells cultured with β-TCP control and β-TCP test media as compared with those cultured with the control media after 7 days. These results indicated that hMSCs mainly indicated osteogenic differentiation expressing a mixture of early and mature osteoblasts after 7 days; at the same time, there was a minimum differentiation in Osteoclastogenesis from some cells. In terms of functionality, the cells cultured with β-TCP test media demonstrated a major feature of osteoblasts, which is to precipitate calcium. Differentiated osteoblasts from the progenitor cells can produce ALP and deposit calcium, which is consistent with bone differentiation 51 . In our study, cells cultured with β-TCP test media after 7 days produced higher ALP than the other cells cultured with β-TCP control and β-TCP test media (P < 0.05).
The β-TCP test particle grafts degraded from 2 to 8 weeks of healing, showing similar behavior to that of the β-TCP control. In their study, Sohn et al. evaluated rabbit calvaria defects over a healing period of 2 to 4 weeks for the early phase of healing and of > 8 weeks for the late phase of healing 52 . Thus, our results can be classified as a medium phase of healing. Similar to the findings of Schaller et al. 53 in rabbit calvarial models, woven bone formation was first observed at 2 weeks of healing in all samples. At 8 weeks, we observed primarily lamellar new bone in the peripheral area. In the central region, both the woven and lamellar bones were seen. The β-TCP test and control particles were in close contact with the new bone but yielded greater defect closure at 8 weeks of healing than control defects did (Figs. 10, 11). However, the new bone formation through osteoconduction was greater for β-TCP test defects than that for β-TCP control (Tables 4, 5).

Conclusions
The ALP results indicated that the proliferation of hMSCs was substantially higher in argon plasma treated β-TCP than β-TCP control. These results indicate a slight enhancement of the β-TCP received by the argon GDP surface treatment. The in vitro and in vivo tests of the present study showed that the GDP-treated β-TCP surface material has better biocompatibility with cells than the control material, which can possibly lead to more bioactive and osteoconductive material in clinical studies. Further studies involving the connections between the β-TCP test