The marine sponges were calcinated in a furnace (FornoCeramica, Portugal) at 800°C for 6 h in order to remove the whole organic components thereby extracting the inorganic part. Prior to the calcination process, the samples were cut into pieces and washed repeatedly with ultrapure water.

The biosilicas obtained from the different marine sponges were immersed in SBF. The SBF was prepared according to the procedure described by Kokubo and Takadama (2006). The initial concentrations of calcium and phosphorus ions in SBF were 2.5 and 1 mmol/L, respectively. The samples were immersed in SBF using plastic tubes, under a powder-to-liquid ratio of 1.5 mg: 1 mL (Jones et al., 2001; Yu et al., 2012; Greasley et al., 2016), using aliquots of 15 mg of each biosilica per 10 mL of SBF. Samples were placed in a thermostatic water bath at 37°C, for 6 h, 1, 3, 7, 14, and 28 days under constant agitation (i.e., 60 rpm). Three samples of each biosilica were used per test, with average values reported. At each time point the SBF was separated from the solid and used to determine the Ca and P concentrations. The solid was washed with distilled water, dried at 37°C and assessed microscopically.

The cytotoxicity of the biosilicas was evaluated following an established protocol based on ISO/EN 10 993 standard (ISO 10993-5, 2009) guidelines, using an immortalized mouse lung fibroblast cell line (L929) and Sarcoma cell line (SaOs2) purchased from the European Collection of Cell Cultures. In brief, biosilicas (100, 50, and 10 mg/mL; 3 samples from the biosilica obtained from the same sponge were used throughout the analysis) were incubated in DMEM culture medium (Dulbecco’s Modified Eagle’s Medium—low glucose) with 10% heat-inactivated fetal bovine serum (FBS; Biochrom AG, Germany) and 1% antibiotic/antimycotic solution (containing a final concentration of penicillin 100 units/mL and streptomycin 100 mg/mL; Gibco, United Kingdom), for 24 h in a water bath at 37°C and 60 rpm. The leachables of each sample were collected and filtered using a 0.22 μm syringe filter. L929 and SaOs2 cells (11,000 cells.cm^–2) were cultured in a 96-well plate in DMEM and after 24 h, the media were replaced by the collected biosilica extracts. Cell viability was evaluated by AlamarBlue assay after 24 and 72 h. The percentage of reduced AlamarBlue was calculated according to the manufacturer instructions (Bio-Rad). The results presented were normalized to the absorbance obtained for control cells cultured in DMEM (positive control) and correspond to the average of three measurements (±standard deviation).

The data obtained from the ICP analysis and cytotoxicity tests were expressed as mean ± standard deviation (SD). The statistical analysis of the cytotoxicity study was performed by using the Shapiro-Wilk test (p < 0.05) to check if the data presented a normal distribution. The data did not follow a normal distribution and, as a consequence, were analyzed using the Mann-Whitney test. The statistically significant differences are represented with ^∗ for p < 0.05 and ^∗∗ for p < 0.01.

Biosilica structures were obtained after calcination of the sponges (Figures 1A,B, before and after treatment) as described in the methods section. In all the cases, the biosilicas were fragile 3D structures composed of siliceous spicules with different sizes and shapes (Figure 1B). The elimination of the organic matrix resulted in an amount of loose spicules, indicating that the spicules were embedded into the organic matrix which, upon calcination, lost its solid support. Moreover, SEM analysis confirmed that all the spicules were cracked after the calcination step, probably due to the degradation of the organic components, such as proteins (e.g., silicateins and collagen) that occurs during the heating process (Figure 1C). The microscopic observation of the biosilicas obtained from GA and GB revealed the presence of different types of spicules, such as sterrasters (microscleres, spherical) and dichotriaenes (megascleres), which have been previously reported and reviewed (Cárdenas et al., 2013).

Thermogravimetric analysis showed that the organic portion of the marine sponges were removed upon calcination at 800°C during 6 h, with the silica content of these sponges accounting to about 50–60% of their dry mass (Supplementary Figure 1, example of TG and DTG curves obtained for GB). The weight loss observed in the lower temperature region (i.e., T_max ∼ 100°C, weight loss ∼ 8%) is attributed to water desorption. Further weight loss in the temperature region between 200 and 544°C with a total mass loss of about 40% corresponds to the thermal decomposition of the sponge’s organic components.

EDS, XRD, and FTIR analysis showed that the inorganic part of the marine sponges is mainly composed of silicates, compatible with biosilica spicules. However, EDS data revealed that they also contained traces of other elements, such as Ca, Na, P, Mg, K, and S. Moreover, Al was also found, but only in the biosilicas obtained from GB, SN, AI, and PV (Table 1). The elemental composition of the biosilicas demonstrates their inherent ability to bind to other essential cations, such as Ca²⁺, Mg²⁺, K⁺, Na⁺, and Al³⁺. In their FTIR spectra (Supplementary Figure 2), the shoulder at apx. 1,650 cm^–1 can be attributed to adsorbed water. The bands at 1,099 and 790 cm^–1 are assigned to the O-Si-O asymmetrical and symmetrical stretching vibrations, respectively, from the Si-O-Si network. The broad band at ∼3,425 cm^–1 corresponds to the stretching vibrations of −OH from the Si-OH moieties (Khan et al., 2017).

The XRD powder patterns (Figure 2) of the biosilicas obtained upon calcination revealed the presence of the most intense peak at 22° 2θ and less intense peaks between 28 and 65° 2θ, which correspond to Cristobalite crystalline phase (COD 9015087) (Xue et al., 2015), confirming that the main structural inorganic component of the sponges are silicates.

To assess the bioactivity of the generated biosilicas (herein referred as the ability of the biosilicas to promote the deposition of CaP layer on their surface), samples were immersed in SBF for different timeframes. At each timepoint, the solution was filtered and analyzed by ICP to determine the Ca and P concentrations. After separation of the biosilica materials from the SBF solution, their surface morphology was analyzed by SEM, while EDS was used to quantify the Ca and P concentrations present in the inorganic structures formed on the surface of the biosilicas.

ICP analysis of the SBF solutions, at the different timepoints of the evaluation of the bioactivity (Figure 3), demonstrated that the concentration of Ca and P in the SBF solution increased for several samples up to day 1 (increase of Ca concentration for the biosilicas generated from GA, SN, AI, and increase of P concentration for GA-based biosilica). This could be explained by the diffusion of Ca and P ions present in the biosilica samples into the SBF. Afterwards, for the GB-, SN-, AI-, and PV-derived biosilicas, a decrease of Ca and P concentration was observed between day 1 and day 7, followed by a stabilization up to day 28, showing that these ions were responsible for the formation of a CaP layer on their surfaces. The GA-based biosilicas had an overall higher consumption of Ca and P from SBF leading to its deposition on the biosilica surface between day 3 and day 28 (as shown in the SEM analysis, Figure 4).

The formation of the CaP layer on the biosilica surface was monitored by SEM and EDS. SEM observation showed that, after immersion in SBF for 28 days, GA-based biosilicas presented the formation of cauliflower structures typical of hydroxyapatite (HA) on the surface of both types of spicules (Figure 4). Moreover, the EDS analysis (Table 2) showed that the detected CaP structures present a Ca/P ratio of ∼1.7, which is similar to the one expected to be present in HA, confirming its presence in the biosilica surface (Bailey et al., 2009). In contrast to these results, GB-based biosilicas, after immersion in SBF for 7 days, presented dispersed CaP structures only on the rod-like spicules, that noticeably decrease through the 28 days of the evaluation of bioactivity (Figure 4).

In the case of the spicules obtained from the SN, PV-, and AI-based biosilicas the formation of dispersed CaP deposits was observed on their surface throughout the whole 28 days of testing. However, especially in the case of the SN-derived biosilica, an additional reduction of CaP deposits was detected after 28 days of immersion (Figure 5), confirming that the generated CaP structures are not stable over longer timeframes.

In general, the SEM/EDS data are consistent with the results obtained by ICP, revealing a significant and stable increase in Ca and P content only on the surface of GA-based biosilica after immersion in SBF for 28 days (Figure 6 and Table 2). In fact, it has been previously reported that the bioactivity of biosilicas is correlated with their chemical composition (Barros et al., 2016), where the presence of Ca, Mg, and P promotes the nucleation and growth of HA on their surface, while biosilicas that do not contain Ca, Mg, nor P, did not show any inherent bioactivity. We did not confirm this trend in the biosilicas obtained from the species herein studied, as all of them presented similar values of Ca/Mg/P concentrations, however, it is relevant to point out that the GA-based biosilica (the one with higher bioactivity) is, in fact, the one that had the higher concentration of P in its initial composition (Table 1). Our results give a strong evidence that the GA-based biosilica should be able to promote the regeneration of bone tissue and, consequently, it is highly relevant for the development of BTE strategies. However, this evaluation was executed using chemical methodologies that do not guarantee per si the cytocompatibility of the tested biosilica formulations.

To guarantee that the proposed formulations can indeed be used under the biological environment, we tested the cytotoxicity of the developed sponge-based biosilicas. L929 and SaOS2 cells were cultured in the presence of extracts from the biosilicas generated from the sponges GA, GB, SN, AI, and PV. In the cases of GB, SN, AI, and PV, a cellular metabolic activity was detected between 65 and 100%, demonstrating the non-cytotoxic behavior of the biosilica originating from these sponges (Figure 7). The GA-based biosilica was non-cytotoxic only at a concentration of 10 mg/mL, being toxic for the highest concentrations, i.e., 50 and 100 mg/mL. The non-cytotoxic behavior of the herein tested biosilicas (which is concentration dependent in the case of GA) suggest their potential use in the BTE. Of note, the metabolic activity of the SaOs2, when in contact with GB− (10 mg/mL) and SN-based (100 mg/mL) biosilicas after 72 h, is higher than the control experiment.

Barros et al. (2016) previously evaluated the suitability of different biosilicas as substitutes of synthetic Bioglass, by assessing the metabolic activity of L929 cells in the presence of extracts from biosilicas of the species Petrosia ficiformis, Agelas oroides, and Chondrosia reniformis. In all the cases the cell metabolic activity was between 80 and 100% confirming their non-cytotoxic character. In addition, Gabbai-Armelin et al. (2019) showed that extracts from biosilica obtained from Tedania ignis and Dragmacidon reticulatum marine sponges, at a concentration of 50 mg/mL, promoted the increment of cell viability. Our results showed that the presence of biosilica from the GB, SN, AI, and PV sponges is non-cytotoxic. The only biosilica with limited cell compatibility (i.e., the GA-derived biosilica) demonstrates a higher concentration of CaP structures on their surface after immersion in SBF. A possible explanation for the observed toxicity is the increment of pH generated from the release of cations from the biosilica structure, as has been observed for Bioglass (Brückner et al., 2016). In fact, we observed an increase in the pH of the medium after 3 days of incubation in the presence of biosilica particles, showing the highest value for GA-based biosilica (Supplementary Table 1). This result is consistent with the higher susceptibility of the cells to the extracts obtained from this type of biosilica, especially at higher concentrations, i.e., 100 and 50 mg/mL. In fact, this type of observation has been also reported by Wanandi et al. (2017) showing that the release of cations leads to the alkalinization of the culture medium, which, in turn, causes cell death. Importantly, when reducing the concentration of the biosilica to 10 mg/mL, the cell viability is not affected. Moreover, the concentration of cations in the medium after 3 days of incubation (Supplementary Table 2), shows a significant reduction of P and Ca content in the medium (due to the deposition of CaP) while significantly increasing the Si content (a known promoter of osteogenesis) (Götz et al., 2019). These differences in the composition of the medium is consistent with the higher bioactivity of GA-derived biosilica and its ability to promote new bone formation. Overall, our results indicate that the GA-derived biosilica are suitable to be further tested in the context of BTE and regeneration.