Role of magnesium oxide and strontium oxide as modifiers in silicate-based bioactive glasses: Effects on thermal behaviour, mechanical properties and in-vitro bioactivity

The composition of a CaO-rich silicate bioglass (BG_Ca-Mix, in mol%: 2.3 Na 2 O; 2.3 K 2 O; 45.6 CaO; 2.6 P 2 O 5 ; 47.2 SiO 2 ) was modified by replacing a fixed 10 mol% of CaO with MgO or SrO or fifty-fifty MgO-SrO. The thermal behavior of the modified glasses was accurately evaluated via differential thermal analysis (DTA), heating microscopy and direct sintering tests. The presence of MgO and/or SrO didn’t interfere with the thermal stability of the parent glass, since all the new glasses remained completely amorphous after sintering (treatment performed at 753 °C for the glass with MgO; at 750 °C with SrO; at 759 °C with MgO and SrO). The sintered samples achieved good mechanical properties, with a Young’s modulus ranging between 57.9 ± 6.7 for the MgO-SrO modified


A C C E P T E D M A N U S C R I P T
3 osteoblast turnover [3,4]. On account of its excellent bioactivity, 45S5 Bioglass ® is nowadays used in several clinical applications. In 2009, the same year of the 40 th anniversary of the first discovery of the glass, two sale records were set, since both the one millionth dose of bone graft product (NovaBone and Perioglas) and the one millionth tube of tooth paste containing 45S5 particulate (NovaMin) were sold [5].
However, the original glass formulation suffers from some drawbacks. In more detail, thermal processing of 45S5 Bioglass ® and related glasses, in order to obtain sintered materials, scaffolds and coatings, is known to promote a wide devitrification. The optimal temperature range to sinter 45S5 Bioglass ® (550 °C up to 610 °C) is indeed very close to its crystallization temperature (about 610 °C) and high temperature treatments are further required to complete the densification thanks to a second sintering step, between around 950 °C and 1100 °C. The crystallization may improve the mechanical properties of the final material, but it inhibits sintering, since it reduces the volume fraction of glass for viscous flow sintering. Moreover, crystallization may also affect the glass bioactivity, since, according to the available literature, the new crystal phases are less reactive and hence less bioactive than the amorphous counterpart. It is reported that devitrification of 45S5 Bioglass ® reduces the rate of conversion to HA , and the ability for a given bioactive glass to stimulate tissue regeneration at a cellular level is related to its rate of dissolution in a physiological environment and to the conversion to HA [6][7][8][9][10][11]. Finally, a partial crystallization can lead to instability, since the remaining amorphous areas degrade preferentially in vivo [12] and this fact may cause the implant to fail due to long-term instability of interphase boundaries.
Over the years, various attempts have been made to improve the sinterability of the glass through an appropriate change in composition [13,14]. For example, increasing the CaOto-Na2O ratio proved to be a valuable way to raise the crystallization temperature and hence to promote the sintering process, which could be completed at 800 °C instead of about 950-1100 °C, as usually observed for 45S5 Bioglass ® [15,16]. Additional investigations showed that the thermal behaviour of the glass could be modified further by partially or completely substituting the residual Na2O with K2O [17][18][19]. The modified

A C C E P T E D M A N U S C R I P T
4 formulations did not impair the biocompatibility of the glass, but they greatly favoured the fabrication of glass-based coatings, composites and scaffolds [20][21][22].
However, it is worth noting that the composition of a bioactive glass exerts a prominent effect not only on its thermal and mechanical behaviour, but also on its biological role. For example, it is known that the addition of as little as 3 wt% Al2O3 to 45S5 Bioglass ® is enough to restrain the bone-bonding ability of the glass [2,13]. As a matter of fact, the ionic dissolution products of bioactive glasses and related glass-ceramics are expected to play a specific role in relation to both osteogenesis and angiogenesis. As a consequence, the addition of specific elements such as magnesium, zinc, fluorine, copper, strontium and boron can be used to modulate the biological, as well as the mechanical performance of materials for bone replacement [23]. In particular, magnesium (Mg) is essential for numerous biological processes in the human body and it is one of the most important ions associated with biological apatites: for example, dentin and bone are composed by 1.11 wt% and 0.47 wt% of Mg, respectively [24]. Mg is known to stimulate osteoblast proliferation, differentiation and bone mineralization ability [25]; it was also reported that Mg regulates active calcium transport and activates phagocytosis, while in calcified tissues it is involved in the calcification process [23,26]. Mg deficiency may be related to reduced osteoclastic and osteoblastic activities, leading to bone fragility and/or decreased bone growth [25,27]. A comprehensive review describing the role of Mg in bone tissue engineering and the biological properties of Mg-containing bioactive glasses can be found in [28].
The importance of strontium (Sr) in the human body makes it a very interesting element as a component/dopant of bioactive glasses [23]. Strontium incorporation in bioactive glasses can accelerate bone-healing processes, stimulate osteoblasts and inhibit osteoclasts in vitro [29,30]; this means that the addition of Sr can be used to increase osteogenesis in vivo and, at the same time, to reduce bone resorption. For this reason Sr is considered a promising agent for treating osteoporosis [23] and it has been employed as the active ingredient of antiosteoporotic drugs such as strontium ranelate [31]. The positive effects of magnesium and strontium in terms of bone generation are witnessed by their diffused presence as

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T 6 quenched in room-temperature water to obtain a frit, which was subsequently dried overnight at 110 °C. The frit was then ground for 20 minutes in dry conditions in a porcelain jar and sieved to produce a powder with a final grain size lower than 67 μm.

Analysis of thermal behaviour
The glasses were characterized by Differential Thermal Analysis (DTA) using a Netzsch

Microstructural and mechanical characterization
After the thermal treatment, the samples were investigated by means of X-ray diffraction The sintered samples were cut and the cross sections were ground and polished according to a standard ceramographic procedure [36]. In order to determine the Vickers hardness of the sintered glasses, Vickers micro-indentation tests (Wolpert Group, Micro-Vickers Hardness Tester digital auto turret, Mod. 402MVD) were performed on the polished crosssections. A maximum load of 100 gf was applied for 15 s and at least 15 indents (clear and crack-free) were considered and analysed for each material [37].

A C C E P T E D M A N U S C R I P T
8 As for the local elastic modulus, a depth-sensing micro-indentation technique was applied, operating with an Open-Platform instrument (CSM Instruments, Peseux, Switzerland) equipped with a Vickers indenter tip. For each indentation, the applied load and the corresponding penetration depth were measured and the elastic modulus was calculated from the unloading part of the load-depth graph by means of the Oliver-Pharr method [38]. For each material, at least fifteen indentations were performed on the polished cross section using standard parameters: maximum applied load: 1 N; loading/unloading rate: 1.5 N/min; loading time at maximum load: 15 s.

Investigation of the in vitro bioactivity
The bioactivity of the sintered samples was investigated by soaking them in 25 ml of For the micro-Raman analysis, a Jobin-Yvon Raman Microscope spectrometer (Horiba Jobin-Yvon, Villeneuve D'Aseq, France) was used, with a 632.8 nm He-Ne diode laser source focused through a 100x objective.

A C C E P T E D M A N U S C R I P T
9 Bromodeoxyuridine (BrdU) assays were employed to investigate cell viability and proliferation. The cytotoxicity of the sintered glasses was evaluated through direct and indirect contact, in order to investigate possible cytotoxic effects of the glasses' eluates.
The samples (10 mm diameter sintered discs) were sterilized in ethylene oxide before biological tests.

Culture of MLO-Y4 cells
MLO-Y4 cells were grown as a confluent monolayer in Dulbecco's modified Eagle's medium (DMEM) containing L-Glutamine 2mM, 1 mM sodium pyruvate, penstreptomycin and 10% (v/v) FBS (fetal bovine serum -Invitrogen). The cells were cultured in 6-well plates for NR uptake assay (i.e. direct contact with the samples) and in 96-well plates containing the glasses' extracts for XTT and BrdU tests (i.e. indirect contact tests, see following paragraphs). The cells were maintained in an incubator at 37°C ± 1°C, 5.0 % ± 1 % CO2/air and 90 % ± 5 % humidity.

Preparation of the glasses' eluates
The samples were treated in centrifuge tubes, 6 cm 2 /ml area, each containing DMEM.
DMEM supplemented with 0.45% (v/v) of phenol solution, a cytotoxic agent, was employed as positive reference (CTRL +), while DMEM only was used as negative control (CTRL-). The flasks were incubated at 37°C for 5 days. Finally, the pH was measured and the eluates were filtered using a 0.22 μm filter.

NR uptake and morphological evaluations
MLO-Y4 cells were cultured in 6-well plates in direct contact with the sintered glasses, at 37°C ± 1°C, 5.0 % ± 1 % CO2/air and 90 % ± 5 % humidity. After 24h of incubation, the morphology of the cells was observed by means of an optical microscope (Nikon TMF,

A C C E P T E D M A N U S C R I P T
added to each well. The plates were incubated for 3 h at 37°C ± 1°C, 5.0 % ± 1 % CO2/air and 90 % ± 5 % humidity, then the NR solution was removed and the cells were washed to remove the excess NR solution. 100 μl of extraction solution (freshly prepared ethanol/acetic acid) was added to each well and the plate was incubated at room temperature for 20 min in order to extract NR from the cells. The test was repeated in triplicate for each glass. Finally, the amount of dye concentrated within the cells was evaluated by measuring the absorbance at 540 nm of all wells, including the CTRL+ and CTRL-ones, by means of a spectrophotometer (Multiscan RC by Thermolab system, Finland).

XTT test
Cells were grown in 96-well plates and then incubated at 37°C ± 1°C, 5.

Statistical analyses
One-way variance analysis (ANOVA) was employed to statistically treat the results, which are expressed as the mean ± standard deviation. Statistical differences among groups (p < 0.05) were established based on a t-test analysis, in which a two-population comparison was considered.

Thermal behaviour, sinterability and microstructural characterization
The results of the DTA investigations performed on the glasses are presented in Figure 1.
Independently of the glass composition, the curves are characterized by (1) a change in the baseline between 650°C and 680°C, when the glasses undergo a glass transition, and (2) a sharp exothermal peak between 860°C and 885°C, which is associated to the glass

A C C E P T E D M A N U S C R I P T
crystallization. While the BGMIX_Sr and BGMIX_MgSr DTA traces show similar trends, the BGMIX_Mg curve presents a second broad exothermal peak at higher temperature (960°C÷980°C), which could be ascribed to a further crystallization process. The characteristic temperatures of the glasses extrapolated by the DTA are listed in Table 2, together with the sintering (Ts) and the melting (Tm) temperatures defined via heating microscopy.
The values in Table 2 indicate that the glass transition temperature Tg of the modified glasses is lower than that of the original BG_Ca-Mix glass, while the trend is less clear for the crystallization temperature, since the peak crystallization temperature (Tc) is comparable for all the glasses, whereas the onset crystallization temperature Tc_onset is lower for the modified glasses than for the parent one. As far as the effect of Sr is concerned, a general decrease in Tg with increasing SrO addition was reported in [42], although a direct comparison is difficult because the glass composition employed by Lotfibakhshaiesh et al  [47,48]. However, it should be stressed that, in general, the role played by MgO in silicate glasses has not been completely clarified yet. In fact it has been reported that MgO (at least up to a certain compositional limit) acts as an intermediate oxide [26,49], while other investigations suggest that MgO mainly behaves as a glass network modifier [50]. Both the Mg-containing glasses developed in the present work, i.e.
BGMIX_Mg and BGMIX_MgSr, are characterized by a Tc rather similar to that of the unmodified parent glass, even if Tc_onset is slightly lower than that of the unmodified BG_Ca-Mix (Table 2) To conclude, the effect of the partial substitution of CaO with MgO and/or SrO on the thermal behaviour of the glasses cannot be generalized, since two opposing factors are involved: (1) an increase in the entropy of mixing, which is expected to promote the disordered glass state, thus retarding the crystallization process; (2) a weaker glass network, by which the crystallization would be promoted. As a consequence, the thermal behavior is governed both by the MgO/SrO ratio and by the specific glass composition.
The sintering temperature (Ts) identified by heating microscopy is also reported in Table 2.
Preliminary sintering tests were carried out to define adequate sintering conditions for each glass composition. The sintering attitude of the MgO and/or SrO modified glasses was estimated by measuring their volume shrinkage for different sintering conditions and

A C C E P T E D M A N U S C R I P T
14 by calculating a specific sinterability parameter SC (see Eq. 1) [34,51,52]. From a physical point of view, the sinterability parameter SC of a glass is representative of its sintering ability vs. its crystallization trend. In more detail, if Tc_onset < Ts and hence SC < 0, the glass crystallizes before sintering, which is expected to result in limited densification; vice versa, if Tc_onset > Ts and hence SC > 0, sintering occurs prior to crystallization and hence a good densification should be achieved [34]. Moreover, as a general trend, the greater is the value of Sc, the more independent are the kinetics of sintering and crystallization, which is likely to promote the sintering and compaction process of the glass [34]. On account of the results of the DTA and the heating microscopy reported in sinter. This fact is further confirmed by the remarkably strong volume shrinkage associated with sintering ( Table 2). This is indicative of the achievement of well-sintered glasses for all the compositions considered in the present contribution, independently of the specific modifier oxide introduced, MgO and/or SrO.
As an alternative to the previous sinterability parameter, the Hrubÿ parameter provides a first-hand idea of the stability of the glass against crystallization on heating and, vice versa, on its vitrifiability on cooling [53]. The Hrubÿ parameter KH is defined as where Tc_onset, Tg and Tm are the onset crystallization (on heating), the glass transition and the melting temperatures, respectively [53].
Again, introducing the values for Tc_onset, Tg and Tm as reported in

A C C E P T E D M A N U S C R I P T
of the modified glasses was confirmed by the XRD performed on the sintered glasses, which remained indeed completely amorphous (data not shown).
The cross section of the sintered glasses are reported in Figure 2. All the samples are adequately consolidated and possess a dense microstructure, as proved by the shrinkage data ( Table 2). This fact confirms the effectiveness of the sintering process which occurs in the investigated glasses. Some fine residual porosity can be observed in the cross sections for all the samples, but it has been widely reported in the literature that the presence of micro-porosity in the implant is not adverse from a biological point of view, because it can favour the osteointegration process [54,55].
In order to characterize further the thermal behaviour of the glasses, a set of samples was treated at higher temperature, i.e. 1000 °C, which is above Tc for all the glass compositions (in particular, for BGMIX_Mg it is above both crystallization temperatures). The XRD spectra, which refer to the samples sintered at 1000 °C for three hours, are reported in Figure 3. All the glasses showed the formation of CaSiO3, as previously observed for the parent glass BG_Ca-Mix [18,52]. Several investigations in the literature report that calcium silicate ceramics belonging to the CaSiO3 family have excellent bioactivity and biocompatibility [56][57][58]. Besides CaSiO3, other specific crystalline phases were observed for each glass composition (see Figure 3). However, the attribution of these secondary phases was difficult due to peak overlapping. As an additional analysis, the BGMIX_Mg glass was also fired at Tc (= 876°C) for three hours. As a matter of fact, since the DTA curve of BGMIX_Mg (Figure 1) was characterized by a first crystallization temperature at Tc = 876 °C and by a second broad crystallization peak between 960 °C and 980 °C, some differences might occur between the crystallization outcome of the BGMIX_Mg sample treated at Tc (= 876°C) and that obtained at T = 1000°C. The diffractogram acquired on the BGMIX_Mg sample fired at 876 °C is reported in the same Figure 3 for comparison purposes. It is worth noting that, even if they refer to two different temperature ranges, the two spectra associated to BGMIX_Mg, at 876 °C and at 1000°C, look qualitatively similar, being characterized by the peaks of CaSiO3 (though with different ICDD codes)

A C C E P T E D M A N U S C R I P T
16 and MgSiO3. The XRD peaks of the sample sintered at higher temperature are sharper, thus indicating a more extensive crystallization.

Mechanical characterization
As shown in Table 2 [59]. This is probably the result of the complicated interactions occurring among multiple modifier oxides.
As reported in the same

A C C E P T E D M A N U S C R I P T
17 but it should be kept in mind that the present contribution is a starting point, and further improvements to the sintering process and related mechanical properties are expected as a result of appropriate changes in the sintering times and, most of all, in the granulometric distribution of the initial powders.

In vitro bioactivity
According to the picture proposed by Larry Hench and co-workers [2], bioactive glasses chemically bond to bone through the formation of a surface film of hydroxycarbonate apatite (HCA), which mimics the mineral component of bone, i.e. the biological apatite.
HCA is able to support new bone tissue growth along the implant at the interface between bone and the implant itself. This property is called osteoconductivity [62]. The first five stages of the osteointegration process, which usually finish by the initial 24 hours, imply the release of soluble ionic species from the glass and the subsequent development of a bilayer deposit composed of hydrated silica (silica gel) and polycrystalline HCA on the surface of the implant. The following six stages require the intervention of growth factors and macrophages, the attachment, differentiation and proliferation of osteoblasts, and, to conclude, the generation and crystallization of the bone matrix [63]. The development of HCA highly depends on the glass composition, type (i.e. sol-gel vs melt) and morphology (pellets, powders, scaffolds, fibres, etc). At the beginning of the 90's, Kokubo et al. proposed to reproduce in vitro in SBF the in vivo HCA formation on the surface of a bioactive material [64]. Today the protocol developed by Kokubo and Takadama [39] is a widely diffused tool to preliminary investigate the bioactivity of new materials, although some criticisms have been expressed, in particular because (1) SBF tests look too simplistic to simulate the complexity of a dynamic biological environment, and (2) they may lead to false negative and false positive results [65].
Previous investigations demonstrated that the BG_Ca-Mix parent glass, both produced by melt quenching and by sol-gel method, showed a high in vitro bioactivity after sintering [18,66]. Figure 4 shows the surface of the new sintered glass samples after soaking in SBF for increasing times; BGMIX_Mg was considered as a representative example. After 1 day

A C C E P T E D M A N U S C R I P T
all the glasses have already started their dissolution and it is possible to observe the local formation of some globular deposits. For short immersion times, BGMIX_Mg seems to be the most bioactive composition; in particular, its surface is covered by white aggregates with the typical morphology of HA already after 3 days in SBF. A progressive increase in the amount of HA precipitation with increasing incubation time was observed for all the samples; after 7 and 14 days the surface of the sintered glasses looked indeed rather similar, being completely covered by HA. Apart from local fluctuations, the EDS analysis on the white aggregates measured a Ca/P ratio of about 1.7 (data not shown). This value is comparable to that of stoichiometric apatite (~1.67) [67]. The identification of the precipitates as HCA was confirmed by XRD and Raman analysis. Again, as a representative example, Figure 5 shows the XRD spectra of a BGMIX_Mg sample soaked in SBF for increasing times (for comparison purposes, the spectrum of the sintered BGMIX_Mg sample, not immersed in SBF, is included as well); since the results for The formation of HA and its chemical nature were investigated further by means of Raman spectroscopy, which is an useful tool to study the precipitation of apatite, as the

A C C E P T E D M A N U S C R I P T
19 Raman peaks associated to the vibration of the P-O group are particularly intense.
Moreover, it is possible to identify if the apatite is carbonated, thanks to the Raman response of the C-O groups. The Raman spectra acquired on the globular aggregates that grew on the BGMIX_Mg surface for increasing soaking times are presented in Figure 6.
Analogous results were obtained for BGMIX_Sr and BGMIX_MgSr (data not reported for brevity). It is possible to observe the typical Raman signals ascribable to apatite, i.e. two peaks at about 590 cm -1 and 430 cm -1 , together with a strong sharp peak at 960 cm -1 [69][70][71].
The peak at about 1070 cm -1 , which is related to the stretching of C-O groups, confirms that the apatite formed on the sample surface is carbonated [72].

Biological tests
The biological tests were performed on the BGMIX_Mg, BGMIX_Sr and BGMIX_MgSr sintered samples. As previously mentioned, two terms of comparison were also considered, namely (i) the BG_Ca-Mix (= parent glass) powders, which were pressed and sintered for 3 hours at Ts = 800 °C; and (ii) the commercially available Bioglass ® 45S5 powders, which were pressed and sintered for 3 hours at Ts = 1050 °C [17,18,35]. It is worth noting that all the biological tests were focused on sintered samples because the new glasses were formulated specifically to facilitate thermal processing, in view to fabricate scaffolds and other sintered devices.
As previously mentioned, even if soaking in SBF is commonly recognized as a simple and inexpensive way to investigate the apatite forming ability of new materials, several concerns exist about the real significance of this experimental approach, since SBF, being a solution which mimics the acellular and protein-free component of plasma, is unable to reproduce the dynamic physiological environment of the human body and its delicate equilibrium of trace elements [73,74]. Moreover, various false-positive and false-negative results have been described in the literature [75][76][77][78]. Besides this, it has been argued that the apatite-forming ability in vitro does not automatically imply the bioactivity in vivo [65,79].

A C C E P T E D M A N U S C R I P T
20 For these reasons, the cytotoxicity of the sintered glasses was tested with respect to MLO-Y4, which is an osteocite-like immortalized cell line isolated from murine long bones, whose properties reproduce those of primary osteocytes [80][81][82][83].
As shown in Figure 7, These outcomes are backed by the results of the viability test with NR uptake. As a matter of fact, the graph in Figure 8(a) shows that the NR uptake after 24 h of direct contact to the sintered glasses is comparable to CRTL−, which means that no significant decreases in lysosomal activity imputable to cytotoxic effects could be observed, since any alteration of the cell surface or the lysosomal membrane due to the action of xenobiotics is known to result in a decreased uptake and binding of NR [84].
The results of the XTT test of the MLO-Y4 cells cultured in eluates from the glasses are proposed in Figure 8 According to the BrdU test results in Figure 8(c), the sintered glasses did not interact negatively with the cell proliferation. Once more, the best results were achieved by the BGMIX_Mg sample, which surpassed by far the 45S5 sintered reference.

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T     samples treated at Ts (see Table 2 for specific temperatures) for 3 hours.

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T

A C C E P T E D M A N U S C R I P T
42 Highlights  The composition of a CaO-rich, K2O-containing silicate bioglass was modified:  A fixed 10 mol% of CaO was replaced with MgO or SrO or fifty-fifty MgO-SrO.  The sintered glasses showed a strong volume shrinkage with low residual porosity;  The samples showed good mechanical performance and apatite-forming ability in vitro;  The presence of such oxides, especially MgO, improves the samples' bioactivity.