In this work we explored further the idea that the endogenous BMP expression and activation of BMP-related pathways in hBMSC can be a key molecular signature of osteoinductive materials. Previously we showed that PLGA-based composites enriched with gel-derived bioactive glasses of either SiO2–CaO or SiO2–CaO–P2O5 systems were capable to induce BMP signalling and prompt osteogenesis of human BMSC when prepared in the forms of thin films and used as growth surfaces. Such biological response was achieved without any additional osteogenic stimuli but cell response depended partly on the CaO/SiO2 ratio of bioactive glass and its P2O5 content and only partly relied on the ions released from the material surface . We thus asked whether the osteoinductive potential of such materials is maintained upon 3D architecture and to what extent it relies on ions released from 3D structures. To address this, we chose two SBG/PLGA composites of different CaO/SiO2 ratio, both with and without P2O5 to examine in more details their potential to induce and maintain osteogenesis of hBMSC when prepared in the form of 3D porous scaffolds. As with our previous study , no other osteogenic stimuli were used besides the materials.
All studied SBG/PLGA composite scaffolds were bioactive as verified by SBF test and formed porous 3D structure. Regarding the latter, there were important differences in the effective porosity and pore volume distribution depending on the scaffold type, that could have contributed to ion exchange rate of the material surface as well as to cell response. Effective porosity was the highest for A1/PLGA scaffolds and these scaffolds released also the highest amounts of calcium ions in the first 24h of incubation in the culture medium. Addition of P2O5 to the SBG composition decreased A2/PLGA scaffolds porosity and Ca ions exchange rate, but still type A composite scaffolds were more porous vs. type S scaffolds, displaying a prevalent pore size fraction of about 100 µm. In the binary SiO2–CaO system, calcium oxide was found to depolymerize silicate structure, while in the ternary glasses the Ca2+ partition between the silicate part of the glass structure and the phosphate part suggest the increasing network connectivity in silica matrix, and thus enhanced chemical durability. The presence of orthophosphate/diphosphate groups charge-balanced by modifier cations Ca2+ results in decreased calcium release compared to binary system without P2O5. Overall, surface activity of type S scaffolds was lower in terms of ions exchange rate and these scaffolds were less porous, with the most significant changes for S2/PLGA, where the medium sizes pores about 50 µm were the dominant fraction. The structure of high silica S type glasses is more rigid and polymerized versus A type materials, and thus the reduction in solubility of S type glasses was observed. Besides composition and structure, the solubility of bioactive glasses in the SBG/PLGA composites could have been affected by the porosity of the scaffolds. The higher porosity results in better surface accessibility and improved ion release from glass particles. Notably, the scaffolds containing type A glasses displayed the same pattern of Ca and P exchange as respective films , but released relatively less Ca and adsorbed less P. In contrast, Ca release and P adsorption was slower for type S scaffolds, and they also adsorbed relatively less P vs. films. Both scaffolds and previously studied films released Si with the same pattern, but the scaffolds released Si faster and higher amounts vs. films. In the case of composite films, the surface of bioglass particles is more accessible (due to sedimentation), which directly contributes to both a massive release of Ca2+ from SBG and chemisorption of ions (Ca2+, PO43−) from the incubation solution/culture medium. This effect limits the release of Si ions due to the rapid formation of HAp layer, and in the case of cell culture, the uniform film surface promotes rapid cell adhesion which inhibits the dissolution of silica bond. Relatively lower dissolution of Ca2+ ions from 3D scaffolds, despite the larger surface area of interaction with the environment comes from the overall less exposed bioglass particles compared to the 2D forms. Furthermore, the lower supersaturation of the solution and slower surface changes promote further breaking of Si–O–Si bonds by the action of –OH groups and cause the dissolution of the glass network. This results in the release of soluble silica into the incubation solution. In addition, the adhesion of cells and organic molecules from the medium, due to the 3D structure, may not occur over the whole surface of composites. Thus, this indirect comparison of films  and scaffolds (this work) of the same composition suggests different surface activity of these materials, depending on the materials architecture
Regardless of the chemical composition of SBG/PLGA architecture, we confirmed that the scaffolds were suitable to support the growth of hBMSC and promoted cell-cell interaction as represented by Cx43 expression. Whereas the expression of Cx43 may indicate cell-cell communication, the expression of cFos can be indicative of cell-biomaterial surface interactions [24, 31, 32]. Notably, at day 3 culture, cFos expression was the highest for A1- and S2/PLGA composite scaffolds, but its expression was negligible for all studied composites at day 7 culture. This may reflect the process of cell responses after seeding to the scaffold, where cell-biomaterial interactions are most important at the initial culture stage and they are replaced by cell-cell interactions at further culture stages. In this view, human BMSC would best interact with the A1/PLGA and S2/PLGA surfaces as revealed by cFos expression at day 3 culture, followed by increased cell-cell communication on these and other scaffolds, as revealed by Cx43 expression at day 7. Although increased cFos expression may also reflect an early cells response to calcium ions , this is not a case for currently studied chemical compositions, as increased cFos expression was noted for both A1/PLGA and S2/PLGA, that differed markedly in the CaO/SiO2 ratio of SBGs as well as Ca release levels at the initial incubation stages in culture medium. It is also worth mentioning that, in the present study, the cells were seeded in a fibrin clot. This cell-seeding method determines different cell distribution on the 3D scaffolds (i.e. more spatial and closer to in vivo tissue complexities) compared to the uniform and flat cell distribution on 2D material. Moreover, during natural bone healing, broken bones are s by blood cells that also form blood clots at the site of bone damage [35, 36]. The latter also promotes the repair of damaged tissue. Therefore, the use of the fibrin-clot method of cell seeding on 3D scaffolds brings the processes taking place in the scaffolds closer to in vivo conditions vs 2D films where it is not necessary or even disadvantageous to use such method. Giving that the material surface activity and ion exchange rate with the physiological fluids may contribute to cell response, we investigated the potential of condition media (CM) harvested from hBMSC-loaded scaffolds to attract further human BMSC and activate BMP promoter in BRITER cells. We observed that human BMSC migration as well as luciferase activity of BRITER cells were significantly increased when the cells were exposed to CM collected from A1/PLGA scaffolds pre-cultured with BMSC (Fig. 5a, b). Notably, CM from empty scaffolds did not have any effects (Fig. 5c), which further supports the importance of studying the bioactive materials ion release profile and their biological action when they are loaded with cells rather than examining their surface activity as prepared. Considering that the A1/PLGA scaffolds released the highest levels of Ca and moderate levels of Si and they displayed the highest differential and cumulative pore volume, it is plausible to assume, that both the physical and chemical properties of these scaffolds contributed to hBMSC response and resulted in the release of chemotactic factors that consequently stimulated hBMSC migration and activated BMP response.
Bone morphogenetic proteins belong to chemotactic growth factors which can stimulate hBMSC migration  and the culture media collected from hBMSC loaded A1/PLGA scaffolds were the only ones that stimulated luciferase activity in BRITER cells . In fact, we detected high BMP-2 mRNA levels in hBMSC grown on A1- and A2/PLGA and high levels of BMP-6mRNA in hBMSC grown on S2/PLGA scaffolds. Furthermore, we observed the activation of BMP-related SMADs and Tak1, which were significantly elevated for cells grown on A1/PLGA (both SMAD1, 5, 8 and Tak1) and S2/PLGA (Tak1 only) scaffolds. Notably, the expression of BMP types had different pattern, depending on chemical composition of composites and the high BMP-2 expression for cells grown on A2/PLGA scaffolds was not sufficient to induce both canonical and noncanonical BMP signalling. It is plausible that additional to BMP-2 and − 6 BMP subsets could have contributed to the activation of the intracellular BMP signalling on A1- and S2/PLGA. Further exploration of Tak1- dependent signalling showed the activation of ERK on A1/PLGA scaffolds that corresponded well with high calcium release from these scaffolds. It has been documented that ERK can be activated by high extracellular calcium . In this work, higher calcium levels were released to the culture medium during the first 3 culture days on A1/PLGA materials vs. A2/PLGA and ERK activation on A1/PLGA was significantly higher than on A2/PLGA. In contrast, cells grown on S2/PLGA activated p38 and JNK. It has been recently showed that silica nanoparticles can activate p38 pathway in BEAS-2B and HBEC3-KT cells , JNK pathway in human bronchial epithelial cells  and murine RAW 264.7 macrophage cell line . These pathways could have contributed to some pro-inflammatory responses [42, 43] that prevented terminal differentiation of BMSC on our high-silica composites, as presented by low osteocalcin and bone sialoprotein levels. Furthermore, we observed higher levels of p-p38 and p-JNK in cells grown on S2/PLGA vs. S1/PLGA scaffolds that corresponded with higher silica release during first 3 culture days on S2/PLGA vs. S1/PLGA. Overall, it can be assumed that the presence of P2O5 decreases calcium release from A2/PLGA and increases silica release from S2/PLGA scaffolds, which is then reflected in opposite biological responses of high-calcium and high-silica materials with and without P2O5.
Examining early osteogenic transcription factors showed that the expression levels of Runx-2 and Osx were generally higher for cells cultured on the scaffolds without P2O5. This may indicate that the presence of P2O5 in the scaffolds slows down the induction of osteogenesis. Moreover, the expression of both Runx-2 and Osx was detected for high-calcium scaffolds only and in fact only cells grown on A1/PLGA scaffolds showed expression of late osteoblast markers such as osteocalcin and bone sialoprotein. This suggests that the induction of BMP signaling pathways together with the high expression of Runx2 and Osx is required to complete osteogenesis of cells cultured on SBG/PLGA composites. Similar to Osx expression pattern, expression of Cx43 at day 3 culture was higher for A1-, A2 and S1/PLGA scaffolds. This is consistent with literature showing important role of Osx in Cx43 expression . Importantly, we also observed relatively higher levels of Cx43 for cells cultured on the scaffolds without P2O5. Culturing cells for 7 days resulted in some increases in Cx43 expression for high-silica materials containing P2O5 (i.e. S2/PLGA), suggesting that in longer cultures phosphorus content in the high-silica scaffold may promote cell-cell communications. Overall, examination of typical osteoblast markers showed significant increases of collagen type I and osteonectin mRNA at day 7 culture, osteocalcin and bone sialoprotein at culture day 21 for A1/PLGA. Thus, these high-calcium composites without P2O5 seemed the best ones to induce and maintain osteogenesis. Although osteonectin and osteopontin mRNAs were elevated on silica-enriched composites, expression of all other markers on these composites was close to that on PLGA. Though, P2O5 presence in S2/PLGA composites were probably beneficial to induce the highest ALP activity in hBMSC.
Ectopic bone formation is a standard procedure to verify osteoinductivity of materials . Previously, we examined osteoinductive potential of selected scaffolds of SiO2–CaO–P2O5 system, namely A2/PLGA and S2/PLGA composites, and both scaffolds were more effective at inducing ectopic bone formation in a rabbit muscle compared to plain PLGA . Giving that our in present in vitro studies the scaffolds depleted of P2O5, i.e. A1/PLGA were the most effective at inducting hBMSC osteogenesis, it was plausible to verify its osteogenic potential in vivo along with the other studied S1/PLGA composite containing bioactive glass of SiO2–CaO system. When implanted, both A1/PLGA and S1/PLGA composite scaffolds performed well without causing any adverse tissue effects, but the morphometric data indicated more osteoid formation upon A1/PLGA implantation, which correlated well with the data obtained in vitro for this scaffold type. Altogether, in this work the PLGA-based scaffolds enriched with binary SiO2–CaO bioactive glasses emerge as promising candidates for bone tissue engineering and bone regeneration-related applications, especially that they may attract osteoprogenitor cells and drive their osteogenesis without additional pharmacological treatment. Moreover, P2O5 addition to bioactive glass composition significantly affects in vitro hBMSC responses, based on this and our previous study . Thus one may expect different outcomes upon their long-term implantation. Direct comparison of the in vivo effects of bioactive glasses enriched or depleted of P2O5 are necessary and this will be a subject of our further studies.