A comparative analysis of the cytocompatibility, protein adsorption, osteogenic and angiogenic properties of the 45S5- and S53P4-bioactive glass compositions

Despite their long history of application in orthopedics, the osteogenic and angiogenic properties as well as the cytocompatibility and protein adsorption of the 45S5- (in wt%: 45.0 SiO2, 24.5 Na2O, 24.5 CaO, 6.0 P2O5) and S53P4- (in wt%: 53.0 SiO2, 23.0 Na2O, 20.0 CaO, 4.0 P2O5) bioactive glass (BG) compositions have not yet been directly compared in one and the same experimental setting. In this study, the influence of morphologically equal granules of both BGs on proliferation, viability, osteogenic differentiation and angiogenic response of human bone-marrow-derived mesenchymal stromal cells (BMSCs) was assessed. Furthermore, their impact on vascular tube formation and adsorption of relevant proteins was evaluated. Both BGs showed excellent cytocompatibility and stimulated osteogenic differentiation of BMSCs. The 45S5-BG showed enhanced stimulation of bone morphogenic protein 2 (BMP2) gene expression and protein production compared to S53P4-BG. While gene expression and protein production of vascular endothelial growth factor (VEGF) were stimulated, both BGs had only limited influence on tubular network formation. 45S5-BG adsorbed a higher portion of proteins, namely BMP2 and VEGF, on its surface. In conclusion, both BGs show favorable properties with slight advantages for 45S5-BG. Since protein adsorption on BG surfaces is important for their biological performance, the composition of the proteome formed by osteogenic cells cultured on BGs should be analyzed in order to gain a deeper understanding of the mechanisms that are responsible for BG-mediated stimulation of osteogenic differentiation.


Introduction
The 45S5-(in wt%: 45.0 SiO 2 , 24.5 Na 2 O, 24.5 CaO, 6.0 P 2 O 5 ) and S53P4-(in wt%: 53.0 SiO 2 , 23.0 Na 2 O, 20.0 CaO, 4.0 P 2 O 5 ) bioactive glass (BG) compositions can be considered the most popular BG compositions for clinical application in orthopedic surgery [1,2].Mediated by the release of their ionic dissolution products, both BGs interact with surrounding cells and stimulate them towards development, such as osteogenic differentiation or angiogenic mediation [3][4][5][6][7][8][9].Furthermore, the conversion of the surface leading to the formation of carbonate-substituted hydroxyapatite (HCA) layer does not only make the BGs biocompatible but also allows them to bond to surrounding tissues [10,11].Eventually, the surface reactivity leads to the adsorption of proteins on the glass surface, facilitating the attachment of cells [12,13].
The 45S5-BG composition was introduced in 1969, after Prof. Larry Hench at the University of Florida (USA) had previously been informed about the problem of the rejection of implant materials by the human body by an army colonel [14].As a consequence, Hench postulated that the rejection of materials is mediated by the formation of scar tissue around the implants because the body recognizes the materials as foreign.A material that is able to form hydroxyapatite (the major inorganic component of bone) on its surface, might therefore not be rejected [14].These thoughts resulted in the development of the 45S5-BG composition [11].In 1981, the 45S5-BG composition received safety clearance by the United States Food and Drug Administration [11,15].In 1999, 45S5-BG was approved for clinical use in orthopedic bone grafting applications in the European Union-since then, it is frequently used in clinical routine exceeding 1 Million applications in bone grafting already in 2009 [11].Despite its favorable properties, 45S5-BG shows some limitations.For example, its tendency to crystallize during heat treatment limits the processability of the glass into three-dimensional structures such as scaffolds [16,17].Therefore, several alternative BG compositions have been developed over the years, from which the S53P4-BG composition was approved for orthopedic applications in the European Union in 2006 [18].Due to its higher network connectivity, S53P4-BG exhibits a broader sintering window compared to 45S5-BG combined with reduced dissolution behavior and lower bioactivity.This results in a slower development of the HCA-layer on the glass surface in comparison to 45S5-BG [18][19][20].Furthermore, the release of silicon ions from 45S5-BG, which is considered critical for the BGdriven stimulation of osteogenesis, initially exceeds the release from S53P4-BG [20].
However, despite their importance in clinical routine [11,21] and their thorough experimental characterization [11,14,18,20,22], there is a lack of comparative studies analyzing the general biological or-more specifically-the osteogenic and angiogenic properties of both BG compositions subjected to similar experimental conditions.It remains unclear if there are differences in the osteogenic and angiogenic response of osteogenic cells such as bone marrow derived mesenchymal stomal cells (BMSCs) when cultured in the presence of both BGs.Also, it is so far unknown how the surface of the BGs interacts with relevant osteogenic and angiogenic proteins.
Although there is a broad number of studies available evaluating the properties of both BGs individually [14,19,22], an actual comparison of the biological effects of the respective BG compositions based on the data available might not necessarily be possible: it is known that differences in the experimental settings, including but not limited to the use of different cell lines, cultivation settings, cultivation periods, evaluation methods, etc., have a significant impact on the obtained results with severe limitations for inter-study comparability [23].For example, it has been shown that the viability of commonly used 'osteoblast-like' cell lines after exposure to granules made of 45S5-BG differs significantly depending on the cell type: while the MG-63 cell line and BMSCs show similar viability after exposure, the viability of the Saos-2 cell line was significantly reduced when exposed to 45S5-BG in the same concentration [24].Therefore, in order to obtain data that provides evidence for an actual comparison of the biological effects of two BG compositions, both BG compositions need to be assessed in one and the same experimental setup.
This study thus aims to compare the cytocompatibility, protein adsorption, as well as the osteogenic and angiogenic properties of morphologically equal, amorphous particles made from the 45S5-and the S53P4-BG compositions using primary human BMSCs and selected proteins within a set of standardized experimental protocols.

BG production and characterization
The glasses were melted from analytical-grade reagents (Na 2 CO 3 , Sigma-Aldrich, Steinheim, Germany; CaCO 3 , Fluka, Buchs, Germany; CaHPO 4 •2H 2 O, Thermo Scientific, Dreieich, Germany; quartz sand for SiO 2 ; Merck, Darmstadt, Germany) in a platinum crucible at 1360 • C for 3 h.The melt was cast to a bar, annealed at 520 • C for 1 h, and cooled overnight.The bar was then crushed and remelted for homogeneity using the same parameters as in the first melting.To obtain particles, the final bars were subjected to crushing and milling using a jaw crusher and a planetary ball mill (Retsch, Haan, Germany).Subsequently, the particles were sieved by a vibratory sieve shaker (Retsch) using sieves of 25 and 45 µm to give a size fraction of 25-45 µm.The morphology of the particles was characterised via a scanning electron microscope (SEM, Auriga, Carl-Zeiss, Germany) at a voltage of 1.5 kV and 5 mm working distance.Particles were fixed to conductive aluminum tape before SEM observation.The particle size was measured with a laser diffraction particle size analyser (Mastersizer 3000, Malvern Panalytical, UK) under wet measurements in water.Gas sorption experiments were carried out to determine the surface area of the BG particles according to the Brunauer-Emmett-Teller theory (Surface Area Analyzer Nova 4000e, Quantachrome).Samples were previously degassed for 2.5 h under vacuum at 120 • C.

Experimental design of the study
BMSCs (fifth passage; 2 × 10 4 cells per cm 2 ) were cultivated in direct contact with preconditioned BGs in a concentration of 1, 2, and 3 mg ml −1 in cell culture medium (CCM) composed of Dulbecco's modified eagle medium (DMEM) high glucose, supplemented with 10% fetal calf serum (FCS) (both Thermo Fisher) and 100 µg ml −1 penicillin/streptomycin (Biochrom, Berlin, Germany).The impact of BGs on BMSC number, viability, osteogenic differentiation and angiogenic factor expression was assessed after seven (D7) and 21 (D21) days of incubation.The control group was cultured in the presence of pure CCM.For extracellular collagen quantification, BMSCs were not cultured in the physical presence to the BGs but in CCM containing their dissolution products.The angiogenic potential of 45S5-BG and S53P4-BG was further analyzed with tubular network formation assay of human umbilical cord vein endothelial cells (HUVECs) in the presence of media conditioned with supernatants of BMSCs cultured in the presence of the BGs.

Study ethics and cell origin
BMSCs were isolated from bone marrow of 10 patients that underwent elective hip replacement surgery at Heidelberg Orthopaedic University Hospital.To minimize inter-individual variability cells were pooled in passage 1 as previously described [25].Written consent from all patients was obtained prior to material collection.Harvesting and utilisation of BMSCs were approved by the responsible ethics committee of the Medical Faculty of the University of Heidelberg (S-340/2018).HUVECs were purchased from Promocell (Heidelberg, Germany).

Cell growth patterns and qualitative evaluation of cell viability
Cell morphology, viability, and growth patterns were visualized using a fluorescence microscopy-based live/dead assay as previously described [28].Cells were washed once with DPBS containing Ca 2+ and Mg 2+ (Thermo Fisher), followed by an incubation step with 6 µM (FDA; Sigma-Aldrich) in DPBS for 2 min at 37 • C for staining viable cells.FDA solution was removed, and cells were washed once with DPBS.Diluted propidium iodide (PI, Thermo Fisher; 0.2 µg ml −1 DPBS) was added for staining dead cells, immediately followed by imaging using a fluorescence microscope (Keyence BZ-X810; Keyence, Neu-Isenburg, Germany).

Combined alkaline phosphatase (ALP) activity and cell viability assay
The enzymatic activity of ALP represents a marker of osteogenic differentiation.A combined fluorescencebased assay was used to assess ALP activity together with cell viability, as recently published [29].In brief, BMSCs were washed with DPBS followed by staining with FDA (2 mg ml −1 ) for 5 min at 37 • C.After washing, cells were lysed with 0.5% Triton-X-100 (Sigma-Aldrich) for 5 min at 37 • C. Aliquots of the lysates were incubated with 100 µM 4-methylumbelliferone-phosphate (4-MUP, Thermo Fisher) for 15 min at 37 • C. The fluorescence intensity was measured at 485/535 nm (ex/em) for FDA and 355/460 nm (ex/em) for 4-MUP with a fluorescence microplate reader (OmegaReader).In order to calculate ALP activity, a standard made of recombinant Shrimp Alkaline Phosphatase (Thermo Fisher) was used.

mRNA expression level of selected osteogenic and angiogenic genes in-vitro
Gene expression levels of genes encoding for markers of osteogenic differentiation and angiogenic factors were examined by quantitative real-time polymerase chain reaction (qPCR).After collecting supernatants for protein analyses (see chapter 2.9), total RNA was isolated from the cells using the PureLink RNA Mini Kit (Thermo Fisher) following the manufacturer's instructions.Complementary DNA (cDNA) was synthesized using the high-capacity RNA-to-cDNA Kit (Thermo Fisher).The PowerUp SYBR Green Master Mix (Thermo Fisher) and the primers listed in table 1 were used to run the qPCR.Primer specificity was verified by melting curve analysis and agarose gel electrophoresis (data not shown).Relative expression of osteogenic and angiogenic genes was calculated using the ∆Ct method.Each target gene was related to tyrosine 3-monooxygenase/tryptophan 5monooxygenase activation protein zeta (YWHAZ) as endogenous control.

Evaluation of secretion of osteogenic and angiogenic proteins via enzyme-linked immunosorbent assay (ELISA)
The protein concentration of bone morphogenic protein 2 (BMP2), Osteopontin (OPN), and vascular endothelial growth factor (VEGF) was measured in cell culture supernatants using quantitative sandwich (ELISAs; Human BMP2 DuoSet, Human OPN DuoSet, Human VEGF DuoSet; all R&D Systems, Wiesbaden, Germany) as instructed by the manufacturer.The color reaction was quantified with an ELISA reader (OmegaReader) at a wavelength of 450 nm with wavelength correction at 540 nm.

Evaluation of collagen synthesis
To quantify collagen synthesis, Sirius Red staining was used as published previously [30].After a washing step with DPBS, cells were fixated with Roti HistoFix (Carl Roth, Karlsruhe, Germany) for 30 min.Cultures were washed thoroughly with distilled water and incubated with Sirius Red staining solution (1 mg ml −1 , Sirius Red F3BA (Waldeck, Münster, Germany) in 1.3% saturated aqueous picric acid (Sigma-Aldrich)) for 1 h.After a washing step with HCl (0.01 mol l −1 ), bound Sirius Red was eluted with NaOH (0.1 mol l −1 ; both Carl Roth).Aliquots of the eluted dye were transferred to a 96-well plate and absorbance was measured using a microplate photometer (OmegaReader) at a wavelength of 570 nm.For quantification of the measured values, a standard curve of Sirius Red was prepared.For normalization of the Sirius Red values on the protein content of the cultures, a 48-well plate was run in parallel for each time point.Cells were lysed in 1.0% Triton-X-100 (Sigma-Aldrich).Protein content in lysates was determined using the micro BCA protein assay kit (Thermo Fisher) according to the manufacturer's instructions.

Protein adsorption on BG particles
Protein adsorption is a crucial step towards functionalization of the BGs surface to facilitated not only cell adhesion but also to influence the behavior of adjacent cells [12].For protein adsorption analyses, BG particles were passivated in DMEM (Thermo Fisher), 100 µg ml −1 penicillin/streptomycin (Biochrom).Thereafter, particles were suspended in DMEM, 10% FCS (both Thermo Fisher), and 100 µg ml −1 penicillin/streptomycin (Biochrom).After incubation for 3 and 17 d at 37 • C, 5% CO 2 , a defined amount of BMP2 (750 pg ml −1 ) and VEGF (500 pg ml −1 ) protein (R&D Systems) was added to the medium and incubated for additional 4 d.After 7 and 21 d of incubation, supernatants of the particles were collected, centrifuged at 13 000 rpm at 4 • C to remove BG particles.BMP2 and VEGF in the samples were quantified using specific sandwich (ELISAs; Human BMP2 DuoSet, Human VEGF DuoSet; both R&D Systems) according to the instruction of the manufacturer.The color reaction was quantified with an ELISA reader (OmegaReader) at a wavelength of 450 nm with wavelength correction at 540 nm.DMEM, 10% FCS, 100 µg ml −1 penicillin/streptomycin supplemented with the same amount of either BMP2 or VEGF served as control.

Tube formation assay
HUVECs were cultivated in 200PRF medium (M200) containing low serum growth supplements (LSGSs) as recommended by the manufacturer.BMSCs were cultivated at 2 × 10 4 cells cm −2 in the presence of BGs (1 mg ml −1 , 2 mg ml −1 and 3 mg ml −1 ) in DMEM/10% FCS (both Thermo Fisher).24 h after seeding, the medium was exchanged to M200 containing 2% FCS (both Thermo Fisher).After incubation for three days, supernatants were collected and centrifuged at 4000 rpm for 10 min at 4 • C to remove cell debris and BG particles.Aliquots of the conditioned medium were stored at −80 • C until further use.For the HUVECs tube formation assay, 96-well plates were coated with 30 µl ice-cold geltrex reduced growth factor basement membrane matrix (Thermo Fisher) per well.The coated well plates were left on ice for 5 min to let the geltrex settle followed by incubation at room temperature for 10 min and thereafter at 37 • C for 30 min for gel crosslinking.HUVECs (passage 4) were suspended in a concentration of 2 × 10 5 cells ml −1 in the appropriate conditioned medium and 2 × 10 4 cells per well were seeded on the solidified gel.M200 with or without LSGS served as positive and negative controls, respectively.After 13 h, cells were stained with Calcein-AM (0,5 µg ml −1 ; Biotium, Fremont, CA, USA) for 15 min.Documentation of the tubular network formation by HUVECs was done by phase contrast microscopy immediately followed by of the calcein-stained tube network documentation using a fluorescence microscope (Keyence BZ-X810).The number of junctions and the total master segment length were quantified with the Angiogenesis Analyzer tool [31,32] in the ImageJ software.

Statistics
IBM SPSS (Version 25; IBM, Armonk, NY, USA) was used for statistical evaluation.Significances were calculated using the non-parametric Kruskal-Wallis test followed by the Mann-Whitney U test.Parameters showing statistically significant differences among different groups in the Kruskal-Wallis multigroup test were then analyzed for significant differences between individual groups.Differences were considered statistically significant for p ⩽ 0.05.All results are expressed as means ± standard deviation.Graphs in the manuscript were designed using Microsoft Excel (Microsoft, Redmond, WA, USA).

BG characterization
The morphology of the BG particles was evaluated through SEM images, where a polyhedral shape typical of milled glasses could be observed (figures 1(A) and (B)).Table 2 shows the values of particle size diameters describing the size distribution of the BG particles.Moreover, in terms of surface area, the measured values for both BG compositions showed no significant difference, resulting in 5.086 m 2 g −1 and 5.341 m 2 g −1 for 45S5-BG and S53P4-BG, respectively.

Cytocompatibility of the BGs
Both BGs showed excellent cytocompatibility (i.e.impact of the BGs on cell number, viability and growth patterns).Cell number was significantly enhanced at both time points by both BGs in all tested concentrations-at D21, 45S5-BG significantly outperformed the S53P4-composition (figure 2(A)).As observed for cell number, cell viability was increased in presence of the BGs (figure 2(B)).Cell viability was significantly higher in the S53P4-BG group at 2 mg ml −1 on D7 and at 3 mg ml −1 on D21 when compared to 45S5-BG, whilst the opposite could be observed at D21 for a concentration of 1 mg ml −1 .Cell morphology did qualitatively not differ between the groups (figure 2(C)).

Impact of the BGs on osteogenic differentiation
The ALP gene expression and enzymatic activity were either significantly downregulated or not significantly changed by the presence of the BGs (figures 3(A) and (B)).Among the two BG compositions, S53P4-BG induced a significantly higher ALP activity at D7 and at D21 when compared to the 45S5-BG group.
BMP2 gene expression was significantly upregulated by the presence of both BGs compared to the control group, with BMP2 gene expression levels in the 45S5-BG group exceeding those in the S53P4-BG group significantly at all concentrations and time points (figure 3(C)).Interestingly, the BMP2 protein concentrations did not reflect the gene expression levels.The BMP2 protein concentration in the supernatant of the BG-treated groups was decreasing with increasing BG concentration on D7 (figure 3(D)).On D21, the protein presence at BG concentrations of 1 mg ml −1 was significantly decreased compared to the control group.When comparing the BGs, BMP2 protein presence in cell culture supernatants was significantly higher in the 45S5-BG group at 1 mg ml −1 only on D7.
The OPN gene expression was significantly upregulated by 45S5-BG at 1 mg ml −1 and 3 mg ml −1 at D7 compared to the control group.On D21, both BGs at all concentrations significantly upregulated OPN gene expression.Comparing both BGs, the OPN gene expression in the S53P4-BG group at 2 and 3 mg ml −1 exceeded the expression levels in the 45S5-BG group on D21 (figure 3(E)).Comparable to the BMP2 protein presence, the amount of OPN protein in the supernatants of the BGs-treated groups was equal or significantly lower compared to the control group (figure 3(F)).OPN protein concentrations were significantly higher in the 1 mg ml −1 45S5-BG group on D21, whilst S53P4-BG increased protein presence at concentrations of 1 and 3 mg ml −1 significantly when compared to the levels within the corresponding 45S5-BG groups (figure 3(F)).
Coll IaI gene expression, one of the late markers of osteogenic differentiation, was downregulated by the presence of both BGs when compared to the control on D7, whereas no significant differences were observed between the BGs-treated groups and the control group on D21 (figure 3(G)).On D21, Coll IaI gene expression in S53P4-BG significantly outperformed the expression levels in the 45S5-BG  group at a concentration of 3 mg ml −1 .In terms of collagen protein deposition, 45S5-BG outperformed the control group significantly in a concentration of 3 mg ml −1 on both, D7 and D21 (figure 3(H)).

Impact of the BGs on angiogenic response of BMSCs and tubular network formation
Both BGs increased the gene expression patterns of VEGF, in parts to a significant extent (figure 4(A)).On D21, 45S5-BG outperformed S53P5-BG significantly at a concentration of 1 mg ml −1 whilst S53P4-BG outperformed 45S5-BG significantly in the highest concentration of 3 ml −1 .On the protein level, the presence of VEGF reached its maximum in the 45S5-BG group on D21 at a concentration of 1 mg ml −1 (figure 4(B)).However, the presence of the VEGF protein in the S53P4-BG was significantly higher compared to the 45S5-BG group at a concentration of 3 mg ml −1 on both, D7 and D21.The presence of both BGs had no or even a downregulating effect on tube formation of HUVECs, as seen in the segment lengths and numbers of

Adsorption of selected proteins on BG surface
Both BGs significantly reduced the presence of BMP2 when compared to the control group on D7 (figure 5(A)).When comparing D7 and D21, protein levels increased with increasing incubation time; however, the presence of both BGs as the higher concentrations still reduced the concentration of free BMP2 significantly.45S5-BG reduced the presence of BMP2 protein singificantly stronger than S53P4-BG at a concentration of 3 mg ml −1 .
In accordance with the observation with BMP2, VEGF concentration in the medium was significantly reduced by both BGs in all concentrations at D7 (figure 5(B)).On D21, protein presence was increased compared to D7.However, 45S5-BG at a concentration of 3 mg ml −1 still reduced VEGF presence in the medium significantly compared to both, the control group and the S53P4-group.

Discussion
The BG compositions 45S5 and S53P4 make a significant impact in orthopedic applications when it comes to the treatment of bone defects [11,21].Both BG compositions have been extensively studied regarding their chemical behavior, their processing properties and their biological functions [11,14,22].However, despite their relevance to the scientific community and frequent clinical utilization, there is no direct comparative data available comparing the biological properties of both BG compositions in one and the same setting.Therefore, in this study, morphologically equal particles of both BG compositions were directly compared analyzing their impact on cell vitality, proliferation, osteogenic differentiation, angiogenic response and protein adsorption.
Depending on concentration, composition, surface area, morphology and preconditioning protocols, BGs might exhibit toxic properties in cell culture settings [33,34].These properties are attributed to the high surface reactivity of BGs, especially in the early phases of contact with physiological fluids, and the changes in the surrounding milieu, especially due to a strong increase in local pH caused by a massive release of sodium ions [33,35].This reactivity was postulated as being a crucial step towards HCA-formation on the BG surface mediated via a pH-dependent (initiated by an alkaline environment) hydrolysis of silanol groups at the glass surface, which results in the formation of a silica-rich layer, an important step towards HCA formation [36][37][38].Both BGs showed excellent cytocompatibility lacking noteworthy differences between the compositions in this study.These results might be explained by the known resilience of BMSCs towards exposure to BGs [24,39].In contrast to other cell types such as osteosarcoma-derived osteoblast-like cell lines or certain bone tumor cells, BMSCs (that have been used in this study) tolerate exposure to 45S5-BG very well, while the viability of other cell types significantly decreases when exposed to the same concentration of 45S5-BG particles [24,39].Therefore, due to their crucial role in bone regeneration and their cytocompatibility with BGs, BMSCs can be considered as a cell type suitable for assessment of the biological impact of BGs [24,40].However, the in-vitro cytocompatibility evaluations exhibit only limited transferability to the in-vivo toxicity of biomaterials in general or BGs in particular: in static two-dimensional cell culture settings, the potential toxic properties of BGs might have a stronger impact on cell viability compared to the in-vivo situation [8,33,41,42].Reasons are, amongst others, the recruitment of new cells in the in-vivo situation or the differences in buffer capacity and the variations in metabolism [43].
Mediated by the release of their ionic dissolution products, BGs actively stimulate biological functions in their surroundings, including but not limited to osteogenic differentiation [7].This makes BGs attractive to be doped with selected ions exhibiting certain biological functions in order to tailor their properties towards the desired field of application [8,44,45].The influence of osteogenic differentiation of bioactive silicate glasses such as the 45S5and S53P4-compositions are attributed to the release of silicon ion species into the surrounding [4][5][6].Both BGs do not differ in regard to the oxides present in the composition, however, the portions of the constituents differs: while the SiO 2 portion in S53P4-BG (53.0 wt%) is higher compared to 45S5-BG (45.0 wt%), the 45S5-BG composition contains higher portions of Na 2 O (24.5 vs. 23.0wt%), CaO (24.5 vs. 20.0wt%) and P 2 O 5 (6.0 vs. 4.0 wt%) [20].Considering the release of silicon being determining for the osteogenic properties of the BGs, potential differences in the dissolution of the BGs and-as a consequence-in the release of their ionic dissolution products may led to differences in their osteogenic performance.When comparing the silicon ion release kinetics of both BGs, ion release from 45S5-BG initially exceeds the release from S53P4-BG [20].Furthermore, the development of the HCA-layer is faster in 45S5-BG compared to S53P4-BG [20]-the formation of the HCA layer on the glass surface is considered as the critical step towards tissue bonding and osteointegration of BGs [10].One marker for osteogenic differentiation in the selected experimental setting is ALP as it is produced by osteoblasts and its gene expression and enzymatic activity are used as a correlate for the differentiation of BMSCs towards osteoblastic lineage.In this study, the activity of ALP was upregulated by the presence of S53P4-BG at D21, whilst the gene expression levels were downregulated by both BGs at all time points and concentrations.One possible reason for these observations might be found in the chosen time points for analyses: the ALP activity is expected to peak after 10-14 d in comparable cell culture settings, whilst the activity is expected to increase until then and to decrease thereafter going in line with the chosen observation time points in this study [46].
BMP2 is a key player in osteogenic differentiation [47].BMP2 is used in orthopedic surgery for bone regeneration as a recombinant protein [48].While it has demonstrated to stimulate bone regeneration, its application can be associated with severe systemic side effects, especially when used in high concentrations [48].Furthermore, the half-life of recombinant BMP2 is within a range of minutes to hours rather short [49].Therefore, a constant and locally effective BG-mediated stimulation of local cells to increase BMP2 gene expression and protein production will improve local bone regeneration whilst avoiding excessive systemic side effects.The presence of both BGs leads to an upregulation of the BMP2 gene However, 45S5-BG seems to upregulate BMP2 gene expression to a greater extent when compared to S53P4-BG.The concentration of the BMP2 protein in cell culture supernatants is also favored by the presence of both BGs, again with a superior effect of 45S5-BG.The higher capacity of the 45S5-BG to induce BMP2 compared to the S53P4-BG observed in the present study is in good agreement with previously published data by Vallittu et al. [50].In their study, adipose-tissue derived mesenchymal cells cultivated for 48 h in the presence of fiber reinforced composites filled with 45S5-BG had a higher BMP2 gene expression compared to cells cultivated on composites containing S53P4-BG particles.The reasons for the differences between the two BG compositions with regard to the stimulation of both BMP2 gene expression and protein production cannot be answered based on the findings of this study.The most likely explanations are either the differences in silicon release [20]silicon is known to stimulate BMP2 gene expression [51]-or the differences regarding the surface reactivity of both glasses and consequently the formation of a HCA layer [20].Since BMP2 is a potent driver of early osteogenic differentiation and its presence is regulated by BG-driven mechanisms [52], future studies should focus on the mechanisms behind BGmediated upregulation of BMP2 gene expression and protein production.It is known that contact of BGs to bone precursor cells leads to an upregulation of the non-canonical (Smad-independent) signaling pathway leading to regulation of osteogenic differentiation [53][54][55][56]-these pathways are also regulated by BMP2 [55].Ojansivu and co-workers reported a mitogenactivated protein kinase (MAPK)-induction and subsequent regulation of osteogenic differentiation in human adipose stem cells to a direct contact of cells to the BG surface using discs made of S53P4-BG [53].In the present study, the stimulation of BMP-2 production was increased by contact of BMSCs to BGs-which might lead to synergistic effects.Interestingly, the direct contact of cells towards the BG surface seems to be crucial in order to induce certain biological effects, not only limited to osteogenic differentiation but also to induce cell death in bone tumor cells [39].These possible correlations should be investigated in greater detail in future studies.
Both collagen and OPN are parts of the extracellular osseous matrix: whilst collagen accounts for about 80% of the total proteins present in bone and is a relevant structure protein [57], OPN mediates the adhesion of cells towards mineralized tissues [58].The presence and gene expression of both collagen and OPN represent later stages of osteogenic differentiation in the present cell culture setting [46].Both BGs lead to an upregulation of the OPN gene activity, especially at D21.While the lowest concentration of 45S5-BG induces the most significant effects, OPN gene expression in the S53P4-BG group exceeds the expression in the 45S5-BG group at the higher concentrations.Comparable effects could be observed for the protein analyses.Expression of the Coll IaI gene, encoding for collagen, was stronger regulated by S53P4-BG, whilst the protein presence was slightly (non-significantly) higher in the 45S5-BG group.Both BGs seem to enhance the later stages of osteogenic differentiation.
Bone regeneration is dependent on angiogenesis [59].It is known that both 45S5-BG and S53P4-BG stimulate the presence of VEGF [3,9].VEGF is a key player in the regulation of angiogenesis during osteogenesis, for example by stimulating vessel ingrowth into zones of hypertrophic chondrocytes [60].In this study, both BGs stimulated RNA expression and the protein production of VEGF.When analyzing the impact of both BGs on the formation of a tubular network by HUVECs, a slight increase was found compared to the control group that was cultivated without pro-angiogenic supplements; comparable results for 45S5-BG have been reported before [61].
In this study, the strong upregulation of osteogenic and partially angiogenic gene expression does not seem to lead to an increased presence of the respective protein concentrations in the cell culture supernatants.Especially for BMP2 but also for OPN and VEGF, an upregulation of gene expression was observed in the BG-treated groups, whereas on protein level, it was equal or even downregulated in the groups incubated with the BGs.This was also observed at D21, however less pronounced.There are numerous possible explanations for discrepancies in gene expression and protein production-such as post-translational regulation processes [62] or elimination of inaccurately produced proteins [63].In this case, the most likely explanation comes from the mechanisms of BG surface reactions and bioreactivity.It is known that the gradual conversion processes on the BG surface eventually lead to adsorption and adhesion of proteins to the glass surface [10,12].These adsorbed proteins affect the biological properties of the materials, for example, they can regulate osteoblast adhesion [12,64].The local presence of pro-osteogenic or pro-angiogenic proteins might therefore regulate local cell metabolism towards bone formation or angiogenesis, respectively.In order to assess the influence of these effects on the results of protein concentration evaluation in the cell culture supernatants, we incubated the BGs in DMEM containing 10% FCS and finally added BMP2 and VEGF.The concentration of both proteins in the supernatants was significantly reduced by both BGs on D7.When comparing both BGs, 45S5-BG seems to be able to adsorb more protein in the later stages of incubation at a concentration of 3 mg ml −1 .These observations are crucial when it comes to the interpretation of the protein data: the observed differences in gene expression levels protein production can be most likely attributed to the BG-mediated adsorption (i.e.immobilization) of the proteins on the BG-surface.Therefore, the chosen assay analyzing the protein presence in cell culture supernatants will certainly underestimate the total protein concentration in culture consisting of the proteins in the supernatant and proteins bound to the BGs.Whilst there are data available on protein adsorption on BGs, most data focus on the adsorption of an undefined composition or number of proteins, e.g. when BGs are incubated with FCS or plasma [65][66][67].Other studies focus on the adsorption of single proteins such as fibronectin or albumin [68,69].However, analysis of the local proteome that develops during cultivation of cells on BG surfaces is certainly of high interest since the proteome is known to influence osteogenic differentiation processes in mesenchymal stromal cells [70].Therefore, future studies should focus on the in-depth protein analysis of BMSCs cultured on BG surfaces.

Conclusions
In this study, the osteogenic and angiogenic properties as well as the cytocompatibility and protein adsorption of the clinically relevant 45S5-BG and S53P4-BG compositions were comparatively analyzed.Morphologically equal particles of both BGs were exposed to BMSCs in a static cell culture setting.Both BGs showed excellent cytocompatibility and stimulated BMSCs towards osteogenic differentiation.However, the 45S5-BG composition showed stronger stimulation of both BMP2 gene expression and protein production.While gene expression and protein production of VEGF were stimulated, both BGs had only limited influence on the formation of a tubular network by HUVECs.The BGs were able to adsorb proteins on their surfaces, namely BMP2 and VEGF.The ability to adsorb proteins was superior in 45S5-BG.The strong ability of both BGs to adsorb proteins in the early phases of incubation aggravated the interpretation of the protein release into the supernatants.However, the local proteome on the surface of biomaterials is known to play an important role in the fate of cells located on these surfaces.Therefore, future studies should focus on analyzing the composition of the proteome formed by osteogenic cells cultured on BG surfaces in order to gain a deeper understanding of the mechanisms responsible for BG-mediated stimulation of osteogenic differentiation.

Table 2 .Figure 2 .
Figure 2. Influence of 45S5-BG and S53P4-BG on BMSC number, viability and morphology.Cells were cultivated with the indicated concentrations of the BGs for 7 (D7) or 21 (D21) days.Cell number (A) was analyzed by DAPI and viability (B) by FDA fluorescence intensity (FI) measurement (arbitrary unit, AU).Cell morphology and viability (C) were visualized by FDA/PI staining.(+) indicates significant differences between the respective group and the control group, ( * ) indicates significant differences between the BG groups in corresponding concentrations.Scale bars refer to 500 µm.

Figure 3 .
Figure 3. Impact the BGs on osteogenic differentiation.Cells were cultivated in presence of the BGs for 7 (D7) and 21 (D21) days.Gene expression of ALP (A), BMP2 (C), OPN (E), and Coll IaI (G) was analyzed by qPCR.The enzymatic activity of ALP (B) was measured by MUP conversion, protein concentration of BMP2 (D) and OPN (F) was quantified in the supernatants by ELISA assays.The collagen deposition (H) in the cultures was determined by Sirius Red staining and quantification.(+) indicates significant differences between the respective group and the control group, ( * ) indicates significant differences between the BG groups in corresponding concentrations.

Figure 4 .
Figure 4. Influence of the BGs on angiogenic factor expression and network formation.Cells were cultivated in presence of the BGs for 7 (D7) and 21 d (D21).Gene expression of VEGF (A) was analyzed by qPCR, VEGF protein concentration (B) was quantified in the supernatants by ELISA assay.(+) indicates significant differences between the respective group and the control group, ( * ) indicates significant differences between the BG groups in corresponding concentrations.Representative images showing the influence of BMSCs/BG-conditioned media on tubular network formation of HUVECs are shown in (C), scale bar refers to 1000 µm, woS = without supplements, wS = with supplements.Quantification was performed by analyzing the number of junctions (D) and total master segment length (E) using the Angiogenesis Analyzer tool for ImageJ.( • ) indicates significant differences between the respective group and the woS group.(+) indicates significant differences between the respective group and the M200 group.

Figure 5 .
Figure 5. BMP2 and VEGF adsorption on the BGs' surfaces.The concentration BMP2 (A) and VEGF (B) was quantified in the supernatants using ELISA assays at day 7 (D7) and day 21 (D21) after a 4-day exposure to the respective proteins.DMEM was used as control.(+) indicates significant differences between the respective group and the control group, ( * ) indicates significant differences between the BG groups in corresponding concentrations.