Silica–gelatin hybrid sol–gel coatings: A proteomic study with biocompatibility implications

Osseointegration, including the foreign body reaction to biomaterials, is an immune‐modulated, multifactorial, and complex healing process in which various cells and mediators are involved. The buildup of the osseointegration process is immunological and inflammation‐driven, often triggered by the adsorption of proteins on the surfaces of the biomaterials and complement activation. New strategies for improving osseointegration use coatings as vehicles for osteogenic biomolecules delivery from implants. Natural polymers, such as gelatin, can mimic Collagen I and enhance the biocompatibility of a material. In this experimental study, two different base sol–gel formulations and their combination with gelatin were applied as coatings on sandblasted, acid‐etched titanium substrates, and their biological potential as osteogenic biomaterials was tested. We examined the proteins adsorbed onto each surface and their in vitro and in vivo effects. In vitro results showed an improvement in cell proliferation and mineralization in gelatin‐containing samples. In vivo testing showed the presence of a looser connective tissue layer in those coatings with substantially more complement activation proteins adsorbed, especially those containing gelatin. Vitronectin and FETUA, proteins associated with mineralization process, were significantly more adsorbed in gelatin coatings.


| INTRODUCTION
The regenerative processes in the bone entail responses to continuous biological challenges. The sequence of bone induction and conduction events, involving various types of cells and signaling pathways in a determined order, is necessary to achieve an ideal regeneration (Dimitriou, Jones, McGonagle, & Giannoudis, 2011). The implants used in the field of bone regeneration have been continuously studied and optimized since the last century. The results of implantation depend largely on the deposition of signaling proteins onto the surface of a biomaterial and will define the magnitude and type of the reaction (especially inflammatory, immune, and coagulation) of the host to the foreign body implantation (Wilson, Clegg, Leavesley, & Pearcy, 2005).
The complement cascade is involved in a variety of physiological and pathophysiological processes, apart from its role as an immune effector. This cascade also regulates the cellular turnover, healing, proliferation, and regeneration (Rutkowski et al., 2010). The disproportionate long-term effects are generally interpreted as implant rejection events. These responses involve mostly uncontrolled blood coagulation processes, the development of infection, and the formation of immune structures (e.g., fibrous capsule) surrounding the foreign body and infected or damaged tissue.
Biomaterials are manufactured and tested to improve the life quality of the patient by minimizing the impact of the implanted foreign body and achieving the recovery in the shortest time possible (Boehler, Graham, & Shea, 2011;Duffield, Lupher, Thannickal, & Wynn, 2008). Assessing the viability of biomaterials involves a battery of extensive tests before the final product can be released to the market; these tests normally entail both in vitro and in vivo procedures. In vivo testing is the ideal standard in new biomaterial trials as it examines their effects on a living organism. However, in vitro testing is the first and necessary step to exclude the immediate damage to the organism; it can also help to avoid the ethical problems and minimize the costs. However, the in vivo results often do not reflect the in vitro outcomes. This problem is invariably emphasized by the experts in the field of regenerative bone engineering. Thus, new approaches and tools are needed to avoid detrimental side effect and predict the efficacy of biomaterials (Hulsart-Billström et al., 2016).
Silica sol-gel hybrid materials are often used in biomedical applications due to the relative ease of controlling their degradation kinetics and the network pore size. These materials degrade by releasing silicon compounds in the Si(OH) 4 form to the surrounding microenvironment, providing a good osteogenic setting for the new bone tissue formation (Juan-Díaz et al., 2014;Juan-Díaz et al., 2016;Romero-Gavilán et al., 2016). Moreover, the process itself results in a good grade of purity at low temperatures. Gelatin is occasionally embedded in the surface of the biomaterial to favour biocompatibility, cell adhesion, proliferation, and differentiation (K. Y. Chen et al., 2009;Shue, Yufeng, & Mony, 2012) because it can mimic the chemical and biological functions of Collagen I in a living organism (Chan, Stahl, Li, & Yu, 2015).
In a previous work, sol-gel compositions with gelatine physically or chemically entrapped were synthetized, achieving different physico-chemical properties (Martinez-Iba ez et al., 2018). The protein adsorption onto these gelatin-silica networks was studied with quartz crystal microbalance using monoprotein solutions, and distinct affinities were detected when the gelatin was present (Martínez-Ibá ez et al., 2017).
In this experimental study, two sol-gel coating bases and their silicon-gelatin correspondents were applied as coatings on sandblasted, acid-etched titanium disc substrates/implants (SAE-Ti), and their biological potential as biomaterials was tested. The compositions with gelatin and their respective base materials were incubated with human serum, simulating a more real setup. Their effects on the adsorbed protein layer using mass spectrometry (liquid chromatography coupled with tandem mass spectrometry) were examined, and the in vitro and in vivo outcomes were studied.
To obtain the sandblasted, acid-etched (SAE) Ti, the discs were abraded with 4-μm aluminium oxide particles and acid-etched by submersion in sulfuric acid for 1 hr to simulate a moderately rough implant surface. Discs were then washed in acetone, ethanol, and 18.2-Ω purified water (for 20 min in each liquid) in an ultrasonic bath and dried under vacuum. Finally, all Ti discs were sterilized using ultraviolet radiation (UV).

| Sol-gel synthesis and sample preparation
The silica-gelatin hybrid coatings were obtained through the sol-gel route. The precursor alkoxysilanes used were methyltrimethoxysilane

| Physicochemical characterization of coated titanium discs
A mechanical profilometer Dektak 6 M (Veeco, NY, USA) was used to determine the roughness. Two coated discs of each composition were tested. Three measurements were performed for each disc to obtain the average values of the Ra parameter. The contact angle was measured using an automatic contact angle meter OCA 20 (Dataphysics Instruments, Filderstadt, Germany). Ten microlitres of ultrapure water W04 were deposited on the sol-gel coated surface at a dosing rate of 27.5 μl/s at room temperature. Contact angles were determined using SCA 20 software. Five discs of each material were studied, after depositing two drops on each disc. The surface topography of the coatings was characterized using scanning electron microscopy (SEM) employing the Leica-Zeiss LEO equipment under vacuum (Leica, Wetzlar, Germany). Platinum sputtering was applied to make the samples more conductive for the SEM observations.

| In vitro assays
MC3T3-E1 (mouse calvaria osteosarcoma cell line) cells were cultured on the sol-gel coated titanium discs at a concentration of 1 × 10 4 cells/well, in Dulbecco's modified Eagle's medium with phenol red (Gibco-Life Technologies, Grand Island, NY, USA), 1% 100× penicillin/streptomycin (Biowest Inc., Riverside, KS, USA), and 10% fetal bovine serum (Gibco-Life Technologies, Grand Island, NY, USA). After incubation for 24 hr at 37°C in a humidified (95%) atmosphere of 5% CO 2 , the medium was replaced with an osteogenic medium composed of Dulbecco's modified Eagle's medium with phenol red 1×, 1% penicillin/streptomycin, 10% fetal bovine serum, 1% ascorbic acid (5 mg/ ml), and 0.21% β-glycerol phosphate and incubated again under the same conditions. The culture medium was changed every 48 hr. In each plate, a well with cells at the same concentration (1 × 10 4 cells) was used as a control of culture conditions. The biomaterial cytotoxicity was assessed following the ISO 10993-5 norm, measured by spectrophotometry, by contact of the material extract with the cell line. The 96-Cell Titter Proliferation Assay (Promega®, Madison, WI, USA) was employed to measure the cell viability after 24-hr incubation of the cells with the extract. We used one negative control (empty cell well) and a positive control with latex, known to be toxic to the cells. Seventy percent cell viability was the limit below which a biomaterial was considered cytotoxic.
For measuring cell proliferation, the commercial cell viability assay alamar Blue® (Invitrogen-Thermo Fisher Scientific, Waltham, MA, USA) was used. This kit measures the cell viability on the basis of a redox reaction with resazurin. The cells were cultured in wells with the discs (three replicates per treatment) and examined following the manufacturer's protocol after 4, 8, and 14 days. The percentage of reduced resazurin was used to evaluate cell proliferation.
To obtain the samples for total protein measurement Bicinchoninic acid assay (BCA) and alkaline phosphatase (ALP) activity, the culture medium was removed from the wells, the wells were washed three times with 1× Dulbecco's phosphate-buffered saline (DPBS) and 100 μl of lysis buffer (0.2% Triton X-100, 10 mM Tris-HCl, pH 7.2) were added to each well, obtaining the cell lysate.
After being kept on ice for 10 min, the lysate was sonicated and centrifuged for 7 min at 13,300 rpm, and the supernatant was used to measure the total protein content and the ALP activity. Each sample was pipetted in triplicate (5 μl/well).
The total protein content was calculated from a standard curve for bovine albumin and expressed as μg/μl, following the manufacturer's instructions, using the colorimetric measurement of BCA at 570 nm on a microplate reader Multiskan FC® (Thermo Scientific®).
The conversion of p-nitrophenylphosphate to p-nitrophenol was used to assess the ALP activity. Sample aliquots of 0.1 ml were used to carry out the assay. One hundred microlitres of pnitrophenylphosphate (1 mg/ml) in substrate buffer (50 mM glycine, 1 mM MgCl 2 , pH 10.5) was added to the 100 μl of the supernatant obtained from the lysate. After 2 hr of incubation in the dark (37°C, 5% CO 2 ), absorbance was measured using a microplate reader at a wavelength of 405 nm. ALP activity was obtained from a standard curve obtained using different solutions of p-nitrophenol and 0.02 mM sodium hydroxide. Results were presented as mmol of pnitrophenol/hour (mmol PNP/hr), and data were expressed as ALP activity normalized by the total protein content (μg/μl) obtained using Pierce BCA assay kit (Thermo Fisher Scientific, Waltham, MA, USA) after 7 and 14 days.

| Statistical analysis
Data were submitted to one-way analysis of variance and to a Newman-Keuls multiple comparison post-test, when appropriate. Differences with p ≤ 0.05 were considered statistically significant.

| In vivo experimentation
To assess the in vivo behaviour to the selected coatings, coated dental implants were surgically placed in the tibia of New Zealand rabbits We used 20 rabbits, five for each material, with weights between 2,000 and 3,000 g, of the age near the physical closure (indicative of an adequate bone volume). The implantation period for the experimental model was 2 weeks. Implants were inserted in both left and right proximal tibiae, each animal receiving two implants (one control sample and one test sample). Animals were sedated (chlorpromazine hydrochloride) and prepared for surgery and then anaesthetized (ketamine hydrochloride). A coetaneous incision was made in the implantation site in the proximal tibia. The periosteum was removed, and the osteotomy was performed using a low revolution micromotor and drills of successive diameters of 2, 2.8, and 3.2 mm, with continuous irrigation. Implants were placed by press-fit, and surgical wound was sutured by tissue planes, washed with saline, and covered with plastic spray dressing (Nobecutan, Inibsa Laboratories, Barcelona, Spain).
After each implantation period, the animal was euthanized by carbon monoxide inhalation, and the implant screws were retrieved to study the surrounding tissues.
Samples for histological examination were processed following the method described by Peris, Prat, Comin, Dejoz, and Vera (1993).
Briefly, the samples were embedded in methyl methacrylate, and 25-30-μm-thick sections were obtained using EXAKTtechnique (EXAKT Technologies, Inc., Oklahoma, USA). For optical microscopy examination, all the sections were stained using Gomori Trichrome solution.

| Adsorbed protein layer
Sol-gel-coated titanium discs were incubated in a 24-well plate for 180 min in a humidified atmosphere (37°C, 5% CO 2 ), after the addition of 2 ml of human blood serum from male AB plasma (Sigma-Aldrich, St. Louis, MO, USA).
The serum was removed, and, to eliminate the nonadsorbed proteins, the discs were rinsed five times with ddH 2 O and once with 100 mM NaCl, 50 mM Tris-HCl, pH 7.0. The adsorbed protein layer was collected by washing the discs in 0.5 M triethylammonium bicarbonate buffer with 4% of sodium dodecyl sulfate and 100 mM of dithiothreitol. The experimental method was adopted from a study by Kaneko et al. (2011). Four replicates for each biomaterial were obtained. The total protein content of the serum employed to this study was quantified before the experiment (Pierce BCA assay kit; Thermo Fisher Scientific, Waltham, MA, USA), obtaining a value of 51 mg/ml.

| Proteomic analysis
Proteomic analysis was performed as described by Romero-Gavilán et al. (2017), with minor variations. Briefly, the same amount of sample (2/10 of the eluted material) was loaded in each lane for the same experimental conditions. The eluted protein was resolved in polyacrylamide gels; the gels were cut into slices. Each of these slices was digested with trypsin and loaded onto a nanoACQUITY UPLC system connected online to a SYNAPT G2-Si MS System (Waters, Milford, MA, USA). Differential protein analysis was carried out using Progenesis software (Nonlinear Dynamics, Newcastle, UK) as described before , and the functional annotation of the proteins was performed using PANTHER (www. pantherdb.org/) and DAVID Go annotation programs (https://david. ncifcrf.gov/). Uniprot (http://www.uniprot.org/) nomenclature was adopted to name the proteins without the ending "_HUMAN".

| Synthesis and physicochemical characterization
Our chosen synthesis parameters allowed us to obtain different materials, all with homogenous surfaces, as can be seen in the SEM micrographs (Figure 1). The 70M30T coating had different morphology in comparison with 35M35G30T. However, no differences in morphology were detected when gelatin was incorporated in the networks.
This agrees with the data obtained using the mechanical profilometer.
70M30T and 70M30T-GEL materials had an average surface roughness (Ra) of 0.77 ± 0.13 and 0.79 ± 0.07 μm, respectively. The compositions 35M35G30T and 35M35G30T-GEL exhibited lower roughness (Ra of 0.51 ± 0.14 μm and 0.58 ± 0.15 μm, respectively). The contact angle measurements (Figure 2) gave similar values for 70M30T and 35M35G30T coatings. However, the addition of gelatin caused a decrease in wettability on both materials.

| In vitro assays
None of the biomaterials tested was cytotoxic. After 7 days of incubation, we found no differences between the ALP activities for the examined materials or even between these materials and SAE-Ti.
Interestingly, after 14 days, we observed a significant increase in the ALP activity on the material 35M35G30T-GEL in comparison with the other formulations even though this material had the lowest activity after 7 days (Figure 3). After 14 days, cell proliferation increased slightly on the formulations with gelatin in comparison with their base materials; the cultures grown on the formulation 70M30T-GEL showed higher levels of proliferation than the control cells. We noted that the proliferative potential of 70M30T base material was better than the proliferative properties of 35M35G30T.

| Proteomic analysis
The proteins eluted from each biomaterial were studied using liquid chromatography coupled with tandem mass spectrometry. The Progenesis QI software was employed to compare the characteristic proteins adhering to the different surfaces. One hundred seventyone proteins were detected and quantified for each surface coating.
The comparison of identified proteins on the 70M30T and 35M35G30T materials displays six proteins more absorbed onto the 35M35G30T coating (Table 1) Similarly, Table 3 shows the comparison between the compositions 35M35G30T and 35M35G30T-GEL. In this case, nine proteins were identified as more abundant on the composition with gelatin; CFAD, CO6, CRP, CO8B, and APOM were among them. However, the levels of adhering IGJ, CATD, HORN, and FCN2 were significantly higher for the 35M35G30T biomaterial. The Gene Ontology (GO) functional classification of the proteins was performed using PANTHER system. Figure 7a and 7b shows the biological functions of the proteins differentially adsorbed onto 35M35G30T and 35M35G30T-GEL, respectively. It is noteworthy that although the 35M35G30T proteins were only involved in four functions, after adding gelatin, we found differentially adhering proteins participating in nine biological processes.
Moreover, for the formulations with gelatin, the proportion of functions associated with immune system processes increased from 14% to 20%. The comparison between Tables 2 and 3 gives us the unique common differentially adhering protein common to the two materials with gelatin, a complement protein C08B. The comparison between the gelatin compositions is presented in Table 4. Sixteen proteins were preferentially adsorbed onto the 35M35G30T-GEL coating (e.g., S10A9, CFAD, CRP, SAMP, C1QC, and VTNC), whereas only four were more abundant on the 70M30T-GEL (APOA, CLUS, APOA5, and IGJ).

| DISCUSSION
The implantation of bone biomaterials triggers an immediate host response, provided by the immune system. Multidirectional pathways  Chen et al., 2015). The initial layer of proteins adsorbed onto the biomaterial surface will ultimately define its biocompatibility, triggering, among other processes, coagulation, immune, and angiogenesis signaling cascades. Hence, each biomaterial, depending on its chemical and physical composition, conformation, and intrinsic characteristics, can adsorb distinct sets and quantities of proteins to its surface. Titanium has been widely used as the base material for implants because of its bioinertia and osteoconductive characteristics (Buser et al., 1991). Nowadays, various coatings are deposited onto this material to confer bioactive properties that enhance and accelerate the osseointegration in a living organism (Jones, 2001) (Kim et al., 2012;Lei et al., 2013;Lei et al., 2014;Mahony et al., 2014), improving the in vitro results (Y. Takahashi, Yamamoto, & Tabata, 2005).
Gelatin was incorporated in both sol-gel base compositions (70M30T and 35M35G30T). In 70M30T, the gelatin is kept in the silica network due to hydrogen bonds between amino and carboxyl groups from gelatin and silanol groups (Smitha, Mukundan, Krishna Pillai, & Warrier, 2007). However, in 35M35G30T, it is anchored to the structure through covalent bonds formed by the reaction with the epoxy ring of the GPTMS precursor (Mahony et al., 2014).
The main chemical difference between the tested base materials (70M30T and 35M35G30T) is the presence of the GPTMS organic groups in the 35M35G30T. In general, physico-chemical results display a decrease on roughness when GPTMS is added, whereas the incorporation of gelatin in both base materials shows an increase on hydrophobicity, possibly due to the special distribution of hydrophobic groups at the surface. Regardless of these differences, the in vitro results show nonsignificant or non-existent divergences between both base materials. The only exceptions were the increased cell proliferation on the 70M30T-GEL samples (result that it is described in the bibliography, Lei et al., 2013;Lei et al., 2014), and the significant increase in the ALP activity on 35M35G30T-GEL ( Figure 3). However, it is worth mentioning the in vitro strategy adopted (using a single immortalized cell line) is not the ideal to simulate the whole in vivo setup of an implantation procedure, in which various biological systems and cues are involved, but it is the generally accepted standard for testing biomaterials nowadays. This in vitro setup gives the experimentation some clues about the material influence directly on the osteoblastic cell behaviour but does not consider parameters such as the immune response and coagulative systems, which is something to take into account as a future perspective (Kohli et al., 2018).
Regarding proteomic analysis, it is interesting to observe the correlation between the base material 35M35G30T and 70M30T  Interestingly, as it is displayed on Table 2, it is notable the slightly greater adsorption of C1QA when the base material 70M30T is supplemented with gelatin, a possible explanation for the existance of a very thin sheet of fibrous connective tissue (Figure 4c). At the same time, it is clear a greater adsorption of FINC on the material with gelatin; FINC is regarded as a protein involved on cellular adhesion and proliferative processes (Ruoslahti, 1984;Sottile, Hocking, & Swiatek, 1998). Notably, the proteins FETUA, CO8B, and C1QC were more abundant on the 70M30T-GEL than on its base material. C1QC and CO8B are proinflammatory proteins.
However, it is interesting to note that FETUA has been described as a modulator of macrophage opsonization, displaying anti-inflammatory activity, among its other functions. Moreover, this protein has a role in the fibril mineralization and may promote bone tissue formation (Manolakis et al., 2017).
The comparison and characterization of the layers of proteins formed on 35M35G30T-and 35M35G30T-GEL-coated surfaces (Table 3) showed more proteins related to the immune system adsorbed onto the 35M35G30T-GEL material (Figure 7), in particular, the proteins CO6, CFAD, CO8B, and CRP, a complement system activator (Murphy, Travers, & Walport, 2008). Hence, in this aspect and at   Applying these criteria, the data obtained from the proteomic analysis show a decrease on these ratios, in particular on the materials incorporating gelatin. For example, the ratio VTNC/CRP on the 70M30T material reaches a value of 76.05, whereas the same material supplemented with gelatin as a value of 56.71. The same finding is observed with the 35M35G30T material, although not having such high values.
This might be related with the differences between the base materials. Having into account these ratios, an appropriate equilibrium between anti-and proinflammatory proteins can be desirable to avoid a chronic inflammatory response and fibrotic tissue formation. The analysis of this ratio support the in vivo results, as it is observable the increased inflammatory reaction by the greater presence of multinucleated giant cells around both gelatin-supplemented coatings. This might be explained due to the smaller ratio due to the greater adsorption of CRP on them. This fact comes to reinforce the potential of proteomic analysis when addressing material biocompatibility, as documented in a previous study (Araújo-Gomes et al., 2017).

| CONCLUSIONS
In summary, it has been shown that the base material 35M35G30T may induce overall a higher immune response than the other base material 70M30T in vivo. Although in vitro results are not concordant with this behaviour, proteomic analysis show effectively more adsorption of proteins related to the immune/inflammatory response on the base material 35M35G30T. The in vivo behaviour displays that 70M30T base material produces a lower immune response at the period tested that increases when adding gelatine whereas the 35M35G30T formulation induces a higher response that also increases when gelatin is added. Overall, the addition of gelatin on each material's matrix provides an even greater immune response, supported by the fact of the adsorption of having more proinflammatory proteins adsorbed on the gelatin-silica hybrid sol-gel formulations, in particular the CRP, a great activator of the complement cascade. On the equilibrium between pro-and anti-inflammatory adsorbed proteins may reside the key for a prediction of in vivo outcome.