Dramatic Improvement of the Mechanical Strength of Silane-Modified Hydroxyapatite-Gelatin Composites via Processing with Cosolvent.

Bone tissue engineering (BTE) requires a sturdy biomaterial for scaffolds for restoration of large bone defects. Ideally, the scaffold should have a mechanical strength comparable to the natural bone in the implanted site. We show that adding cosolvent during the processing of our previously developed composite of hydroxyapatite-gelatin with a silane cross-linker can significantly affect its mechanical strength. When processed with tetrahydrofuran (THF) as the cosolvent, the new hydroxyapatite-gelatin composite can demonstrate almost twice the compressive strength (97 vs 195 MPa) and biaxial flexural strength (222 vs 431 MPa) of the previously developed hydroxyapatite-gelatin composite (i.e., processed without THF), respectively. We further confirm that this mechanical strength improvement is due to the improved morphology of both the enTMOS network and the composite. Furthermore, the addition of cosolvents does not appear to negatively impact the cell viability. Finally, the porous scaffold can be easily fabricated, and its compressive strength is around 11 MPa under dry conditions. All these results indicate that this new hydroxyapatite-gelatin composite is a promising material for BTE application.

Introduction

Bone tissue engineering (BTE), an emerging field to deal with bone defects, relies on using porous 3D scaffolds to provide temporary support for bone regeneration.[1] There are several key considerations in designing a scaffold for BTE, including biocompatibility, biodegradability, mechanical properties, scaffold architecture, and fabrication technique.[2] Among these, developing a resorbable scaffold with sufficient mechanical properties during the initial healing phase of bone regeneration is a significant challenge; this is because commonly used ceramics (e.g., tri-calcium phosphate) and polymers (e.g., poly-l-lactic acid) typically demonstrate inadequate mechanical strength.[3]

Currently, there are three major types of materials that are used for BTE, namely, polymers, ceramics, and composites.[4] Among these, composite materials have been shown to be a promising option because they demonstrated tunable mechanical strength, controllable degradation rate, and good biocompatibility.[5] In fact, bone itself is a composite material, consisting of an organic component (collagen) and an inorganic component (hydroxyapatite, HAp). Previously, Chang et al. developed a hydroxyapatite–gelatin (HAp–Gel) nanocomposite to mimic the structure of the nature bone.[6] Similar to the interaction between various components in the nature bone, HAp–Gel showed chemical bond formation between calcium cations of HAp nanocrystals and carboxyl anions of Gel macromolecules. Furthermore, HAp–Gel had a self-organized structure along the c-axis of crystalline HAp nanocrystals, which would account for the improved mechanical strength of this composite.[6] However, HAp–Gel was a particulate powder and hard to fabricate into scaffold. To make it formable, the Ko group used an aminosilane [i.e., bis[3-(trimethoxysilyl)propyl]ethylenediamine (enTMOS)] as the chemical linker to facilitate the binding and solidification of HAp–Gel and named such a composite as HAp–Gemosil (i.e., HAp–Gel modified with siloxane).[7] Compared to previously developed glutaraldehyde linker-treated HAp–Gel,[8] this new composite, HAp–Gemosil, demonstrated improved processability and acceptable cell compatibility. Still, the compressive strength of HAp–Gemosil was only ∼40% of the corresponding value achieved by the cortical bone (80 vs 205 MPa[9]) under the similar experimental condition, and the biaxial flexural strength of HAp–Gemosil was merely ∼18% of that achieved by the cortical bone (40 MPa vs 220 MPa[10]). To further improve the mechanical strength of HAp–Gemosil, we have attempted to incorporate biocompatible polymers to increase the toughness of HAp–Gemosil. For example, we designed a silane-functionalized poly(l-lactide-co-propargyl carbonate) copolymer to enhance the long-range interactions among different components by incorporating this copolymer within the HAp–Gemosil composite.[11] However, this polymer-enriched composite only improved the biaxial flexural strength from 40 to 60 MPa, which is still much lower than that of the natural bone.[11] More recently, we introduced polydopamine—a mussel adhesive protein-inspired material that has excellent coating and adhesion properties—into HAp–Gemosil, aiming to improve the adhesion between the hydrophilic HAp–Gel particles and the hydrophobic siloxane matrix. However, the compressive strength improvement was still limited, only from 80 to 100 MPa, which was caused by the lack of long-range interactions among different components in the composites.[12]

These results made us speculate that the main reason for the poor mechanical strength of HAp–Gemosil might be the intrinsically weak silane network in this composite. Given that the silane matrix is the major component of the HAp–Gemosil composite (∼46% by weight), engineering the silane network to dramatically improve its mechanical strength would be an alternative—perhaps an ideal—solution. The silane network formation is generally believed to go through a sol–gel process. This process starts with hydrolysis of alkoxide silane precursors to form silanol, followed by the condensation/polymerization between the silanol groups to form siloxane (Si–O–Si) linkages.[13] The gel structure (e.g., morphology and porosity), which decides the mechanical strength of the as-formed silane network, can be significantly influenced by the processing method and specific conditions, including pH, R ratio (R = [H2O]/alkoxide precursor),[14] solvents,[15] and drying process.[13] As a result, to produce a gel with desirable properties, the processing method and conditions need to be carefully optimized. Processing conditions become even more important when preparing composite materials that contain the silane network. For instance, an inhomogeneous mixing of different components in the composite can induce internal cracks during drying,[13,16] which can significantly decrease the final mechanical strength of such composites.

Among aforementioned various factors that could influence the gel structure, controlling the drying process of the sol to gel formation has been shown to have a crucial impact on the mechanical property of the final gel network.[13] To improve the drying while minimizing cracking, cosolvents[17] have often been employed. Theoretically, different cosolvents control different kinetics of hydrolysis and condensation, resulting in different morphologies of the silane network (Scheme ) with different physical properties.[15] Thus, we hypothesized that control of cosolvents in processing could further improve the mechanical properties of HAp–Gemosil. Indeed, we find that by choosing the appropriate cosolvent, as detailed below, the mechanical strength of HAp–Gemosil can be greatly improved. Compared with the original HAp–Gemosil, the new HAp–Gemosil composite—processed with tetrahydrofuran (THF) as the cosolvent—demonstrated almost twice the mechanical strength of the original one without THF (i.e., compressive strength: 97 vs 195 MPa and biaxial flexural strength: 222 vs 431 MPa). Furthermore, we show that the improvement of the mechanical strength is due to the improved morphology of the silane network and the uniform distribution of HAp–Gel in the enTMOS matrix.

Scheme 1

enTMOS Network Formed in Different Solvent Systems

Experimental Details

Materials

HAp–Gel and Ca(OH)2 powders were prepared by the method reported previously.[6] Ca(OH)2 with chlorhexidine (CHX) was prepared by doping 5–10% CHX into Ca(OH)2 powder. enTMOS (95%) and 62% enTMOS in methanol (MeOH) were purchased from Gelest, Inc. (Morrisville, PA, USA). Solvents, including methanol (MeOH), ethanol (EtOH), acetonitrile (CH3CN), THF, tetrahydropyran (THP), dimethyl sulfoxide (DMSO), dimethoxyethane (DME), and 1,4-dioxane were purchased from Alfa Aesar and used as received. Ammonium persulfate (APS) was purchased from Sigma-Aldrich and used as received.

Preparing the Composite

The method for making the composites was adapted from the previous report.[12] HAp–Gel (300 mg) and 200 mg of Ca(OH)2 with CHX powder were transferred into a mortar and ground into fine powder. Then, the powdered mixture was spread on a glass sheet, which was placed on a cold stage to maintain a depressed temperature of—20 °C. The premixed enTMOS solution, including 360 μL of cosolvent, 200 μL of MeOH, and 500 μL of 95% enTMOS, was then added into the powdered mixture quickly and mixed continuously with a spatula for 30 s. Then, 80 μL of 7.5% APS in MeOH/water (v/v = 3:7) was added to trigger the sol–gel reaction. After mixing thoroughly, the mixture was poured into a disk mold (size: diameter: 15.58 mm and thickness: 2.8 mm, designed for 3-point bending test), which lay on a smooth glass slide. Another smooth glass slide was carefully covered at the top of the mold to remove extra material. This “sandwich” structure (glass slides at the top and the bottom and material in the disk mold in between) was clamped and sealed into a plastic bag at room temperature for one week to let the composite dry slowly. Finally, the sample was further dried in an oven at 54 °C for 5 days before the flexural strength test. The reason why the temperature was set at 54 °C is because gelatin has been shown to degrade gradually around 100 °C, and collagen has been shown to denature to gelatin between 60 and 80 °C. Thus, we chose 54 °C to remove the remaining solvent as well as to avoid the degradation of gelatin or collagen. For comparison, composites without cosolvents, namely, all MeOH, were also prepared according to the same procedure described above. Furthermore, the previously reported HAp–Gemosil composite[7] made from 62% enTMOS was also repeated here as a reference. Additionally, cylinder-shaped samples with a 1:2 ratio of diameter (3.8 mm) to length (7.6 mm) were made according to our previous report with the same composition as described above.[12] To prepare a pure enTMOS network without HAp–Gel and Ca(OH)2 CHX, APS was directly added to the premixed enTMOS solution. After gelation, the same drying process was applied as described above.

Compressive and Biaxial Flexural Test

The testing procedures were performed according to methods established in our previous publication.[11] Cylinder-shaped samples (3.8 mm diameter by 7.6 mm length) or disk-shaped samples (13 mm diameter by 2.5 mm thickness) were used for the compression test and the 3-point bending test, respectively. Mechanical testing was performed on an Instron machine (model 4411, Instron Co., Norwood, MA, USA) with a cross-head speed of 0.5 mm/min. The same calculation method[11] was used to get compressive strength and biaxial flexural strength. A minimum of three samples for each group were used in all mechanical testing.

Fourier Transform Infrared

Bulk composites were ground into fine powder. Then, the fine powder was spread directly on the diamond crystal and analyzed by Fourier transform infrared (FTIR) using the attenuated total reflection mode.

29Si Solid-State NMR

29Si CPMAS NMR spectra were acquired on a Bruker DMX 360 MHz WB NMR spectrometer operating at 71.55 MHz for 29Si and at 360.13 MHz for 1H. A Q8M8 (i.e., octakis(trimethylsiloxy)silsesquioxane) sample was used to calibrate the proton pulse width and the silicon power level for the contact pulse (10 ms). A relaxation delay of 5 s and a sweep width of 27 kHz was used. Total number of scans ranged from 4 to 16k. Spectra were processed with 50 Hz line broadening and referenced to an external TMS sample.

Morphology Study

For scanning electron microscopy (SEM) analysis, a sample was carefully cut from the bulk disk-shaped composite. Then, the sample cross section was sputter-coated with gold in vacuum and imaged using Hitachi S-4700 cold cathode field emission SEM (Hitachi High-Technologies America, Inc.). Energy-dispersive X-ray spectroscopy (EDS) analysis was performed to examine different phases on sample’s cross section, which were analyzed by Inca operator software. For transmission electron microscopy (TEM) analysis, the bulk composite was ground into fine powder and suspended into MeOH with sonication. Then, a small drop of this solution was added to the TEM grid for imaging using JEOL 2010F-FasTEM (JEOL USA, Inc.).

3D Printing of the Composite

A 3D cylindrical porous template (diameter: 5 mm, height: 10 mm, pore size: 400 μm, and porosity: 50%) was designed using SolidWorks software (Dassault Systems SolidWorks Corp., Waltham, MA, USA). This template (stl format) was then used to print a 3D wax mold made of Solidcape Model (Solidscape Inc., Merrimack, NH, USA) with 0.16 mm2 trusses and 0.16 mm2 pore sizes (continuous space) using a Solidscape 3D printer (Solidscape Inc., Merrimack, NH, USA). Next, the composite mixture was prepared on the cold stage as described above and injected to the 3D-printed mold before the materials were let to solidify. After setting for 3–5 min, the wax mold and the composite were immersed in acetone for 15 min to remove the wax template and release the 3D porous composite-based scaffold, which is shown in Figure a. The porous scaffold was air-dried for one week, followed by drying at 52 °C for 4 days prior to the compression test. The same method and data analysis were used here to obtain the compressive strength. The compressive strength of the porous scaffold was averaged from at least three samples.

Figure 4

(a) Picture of the porous scaffold (pore size: 400 μm, porosity: 50%) and (b) mechanical strength of the porous scaffold processed with different cosolvents (student t-test, P = 0.0001).

Bulk Material Biocompatibility Study via RealTime-Glo MT Cell Viability Assay

Bulk disk samples (diameter: 6 mm and thickness: 1 mm) made from THF/MeOH or MeOH were leached in H2O for 3 days, followed by gas sterilization and balanced in the cell growth media overnight. The disk samples were then seeded with rat mesenchymal stem cells (rMSCs) in a Costar 48-well plate and incubated at 37 °C overnight to allow cells to attach. Then, the disk samples were transferred into a new 96-well opaque plate to prevent interwell interference during measurement, followed by the addition of NanoLuc luciferase and a cell-permeant prosubstrate and incubated at 37 °C. At predetermined time points (i.e., 0, 1, and 2 days), the illuminance of each well with the substrate was measured by a Cytation 5 cell imaging multi-mode reader. Each group with a minimum of three samples was tested for each time. The cell viability result was averaged from three separate measurements.

Results and Discussion

Mechanical Strength of HAp–Gemosil with Different Solvents: Observation

According to previous findings, on the one hand, low viscosity solvents can promote rapid hydrolysis which would likely promote fast condensation reactions to occur (because of the increased concentration of reactants for condensation), resulting in a condensed network.[15a] On the other hand, nonpolar aprotic solvent is preferred for a fast condensation reaction than polar protic and aprotic solvents. This is because hydrogen bonding and/or electrostatic interactions between polar solvents and nucleophilic substitution reaction intermediates would slow the rate of condensation.[15b] To form a dense network of silane and thereby achieve higher mechanical strength of such a network, one would need rapid hydrolysis and fast condensation reactions. Thus, nonpolar aprotic solvents with low viscosity would be ideal. To experimentally verify this hypothesis, we chose a number of cosolvents in this study (Table ). These cosolvents have different viscosity values and can be categorized into three different groups (i.e., polar protic, polar aprotic, and nonpolar aprotic). Moreover, the addition of cosolvents will ensure more homogenous mixing of the components in the composite.

Table 1

Biaxial Flexural Strength of the Composite with Different Cosolvents

type of cosolvent	entry	cosolvent/MeOH (v/v = 9:5)	viscosity of pure cosolvent	boiling point of cosolvent (°C)	biaxial flexural strength (MPa)
polar protic	1	MeOH	0.55	64.7	222.83 ± 54.80
	2	EtOH/MeOH	1.07	78.37	102.32 ± 34.53
polar aprotic	3	CH3CN/MeOH	0.37	81.3	520.70 ± 15.48
	4	DMSO/MeOH	1.99	189	138.01 ± 19.40
nonpolar aprotic	5	THF/MeOH	0.46	66	431.35 ± 58.72
	6	THP/MeOH	0.52	88	471.40 ± 44.83
	7	Dioxane/MeOH	1.18	101.1	291.05 ± 75.75
	8	DME/MeOH	1.1	85	408.73 ± 51.38

Our results in Table clearly demonstrate that the cosolvent indeed has a strong influence on the biaxial flexural strength of the composite. First, as we expected, among the three types of cosolvents we studied, aprotic ones—with the exception of DMSO—improve the mechanical strength of the composite more than protic ones. In some cases, we could even increase the mechanical strength over 100% when compared with that of the original HAp–Gemosil (prepared with MeOH only, entry 1 of Table ). For instance, when using CH3CN and THF as the cosolvents, the flexural strength of such composites increases from 222 MPa to around 520 and 431 MPa, respectively (entry 3 and entry 5 in Table ). This can be explained by the fact that THF and CH3CN are aprotic in nature, which can promote fast condensation reaction when compared with the protic ones (such as MeOH and EtOH). Moreover, THF and CH3CN also have relative lower viscosity values, which would likely promote fast hydrolysis reaction of enTMOS. For these reasons, the addition of THF or CH3CN would result in a dense silane network, as shown in Scheme with higher flexural strength. This can be further proved by the impressive improvement of compressive strength, which increases from 97 MPa when processed with pure MeOH to 195 MPa when adding THF as the cosolvent (Supporting Information Table S1, entries 2 and 4). However, if all MeOH was replaced by THF (i.e., THF only) or CH3CN (i.e., CH3CN only), the sol–gel reaction would be completed too fast (gelation time was reduced to 2 min from the original 5 min). The too fast cure rate would result in the incomplete condensation reaction, evidenced by the opaque color of pure enTMOS (Supporting Information Figure S5c,e), leading to lower mechanical strength of such a composite. For instance, the compressive strength and biaxial flexural strength of the THF-only composite would decrease from 195 to 72 MPa and 430 to 152 MPa, respectively (entry 4 and entry 5 in Table S1). Second, for cosolvents in the same category, the boiling point of the cosolvent also plays a role in the mechanical strength of the composite. For instance, composites made from dioxane/MeOH and DME/MeOH have different flexural strength, 291 vs 408 MPa (Table , entries 7 and 8). It appears that a close match of the boiling point of the cosolvent with that of MeOH (65 °C) would help the drying process. In this specific case, the boiling point of dioxane (101 °C) is too far from that of MeOH than the boiling point of DME (85 °C). It is likely that having a cosolvent with a boiling point similar to that of the primary solvent can lead to a more uniform solvent evaporation, which can help reduce the large cracks in the final composite and result in higher mechanical strength. Finally, viscosity of the cosolvent does not appear to have any correlation with the flexural strength of the final composite. This can be seen from cases of CH3CN/MeOH, THF/MeOH, and DME/MeOH. A range of viscosity values (0.37, 0.52, and 1.1) were observed, yet all cosolvents offered similarly high flexural strength of the final composites (over 400 MPa).

Hypothesis

We hypothesize that the effect of the cosolvent on the mechanical strength of the HAp–Gemosil composite is mainly due to the combined effects from hydrolysis and condensation of silanes that changed the morphology of the composite, rather than from cosolvent-induced chemical reactions/interactions between different components within the composite. Specifically, adding cosolvents into the processing solvent (i.e., MeOH) can have two functions during the composite formation. First, mixing different cosolvents with MeOH increases the total volume of the solvent when compared with the original processing method reported previously. The extra solvent would allow the various components, including HAp–Gel, Ca(OH)2, and enTMOS, to mix more homogenously. Second, the cosolvent can have a subtle impact on the morphology of the enTMOS network in the final composite (i.e., branched or condensed network, shown in Scheme ), likely caused by the different rates of hydrolysis and condensation when a different cosolvent is applied. As discussed earlier, the polarity and boiling point of the cosolvent can have a direct impact on these fundamental steps (i.e., hydrolysis and condensation) in the sol–gel process. Macroscopically, composites processed with an appropriate cosolvent would show less cracks/holes, which would benefit a higher mechanical strength. Finally, because the solvent can be completely removed during the drying process, the biocompatibility of these composites can be maintained in addition to the improved mechanical strength.

Fourier Transform Infrared Spectroscopy

To experimentally verify the hypothesis, we first applied FTIR spectroscopy to identify specific chemical bonds in the final gel structure and to monitor the sol–gel reaction process.[18] We started with pure enTMOS gels formed from different cosolvent/MeOH systems to probe whether the cosolvent reacted with enTMOS. Three representative cosolvent/MeOH systems from these three different types (Table ) were selected, THF/MeOH, CH3CN/MeOH, and MeOH only. First, the FTIR of a pure enTMOS network prepared with CH3CN/MeOH and THF/MeOH shows characteristic peaks for enTMOS gels (e.g., Si–O–Si stretching at 1029 and 1118 cm–1), indicating the formation of an aminosilica matrix. As expected, enTMOS gels made from different cosolvent/MeOH systems demonstrated almost identical absorption spectra (from 4000–400 cm–1) to that of the enTMOS gel made from MeOH only (Figure a). This indicates that the cosolvent did not react with the enTMOS during the sol–gel process. Moreover, the almost identical absorbance of peaks corresponding to the Si–O–Si bond for enTMOS processed with different solvents indicates that the total amount of the Si–O–Si bond is comparable across the samples. We also performed 29Si solid-state NMR to further investigate the variety of Si bonds (e.g., Si–O–Si and Si–OH) that were involved in the sol–gel process. As shown in the Supporting Information Figure S2, for all silane networks with different processing cosolvents, we observed two Si peaks. The signal at—65 ppm is corresponding to the formation of Si–O–Si, whereas the other signal at—58 ppm is attributed to the Si–OH structure from the incomplete condensation.[19] Again, the 29Si NMR spectra are very similar across the samples, implying a similar amount of Si species. Second, to examine whether the cosolvent would react with other components in the HAp–Gemosil composite, that is, HAp–Gel and Ca(OH)2 with CHX, we applied FTIR to study the HAp–Gemosil composite. Again, these FTIR spectra of composites made from different cosolvent/MeOH systems are almost identical to each other (Figure b), indicating that the cosolvent did not react with the remaining components [i.e., HAp–Gel and Ca(OH)2 with CHX] in the HAp–Gemosil system. Most importantly, we did not observe any cosolvent/MeOH characteristic absorptions in all FTIR spectra, indicating that the solvent was completely removed after drying. This is ideal for BTE application because the residual cosolvent (if any) could be toxic.

Figure 1

FTIR spectra of (a) enTMOS gels and (b) HAp–Gemosil composites prepared from different solvent systems.

Morphology

Results from the FTIR study rule out the possibility that the improvement of mechanical strength of these cosolvent-treated composites was from cosolvent-induced chemical interactions/reactions. Thus, it is very likely that the optimized morphology of the HAp–Gemosil composite is the main reason for the improved flexural strength. We next applied SEM to investigate the morphology (i.e., phase distribution and porosity) of the composite formed from different cosolvent/MeOH systems and of a pure enTMOS network. We chose to study the cross section of the composite made from two representative cosolvent/MeOH systems, namely, CH3CN/MeOH and THF/MeOH. We also investigated composites made from MeOH and THF only as the reference. From the cross-section images of these composites (Figure ), one can clearly observe that the composite formed from CH3CN/MeOH or THF/MeOH has a much smoother surface (less visual cracks), which agrees well with the greatly improved flexural strength of the composites (Table ). In contrast, the composite made from MeOH only has large cracks (Figure a). Furthermore, there are two distinct phases probed by EDS: one is enriched with HAp–Gel, while the other is enriched with the silsesquioxane phase (Supporting Information Figure S3). Comparing the phase distribution in composites made from MeOH only or THF only with that in composites made from CH3CN/MeOH or THF/MeOH, the enTMOS phases and HAp–Gel phases distribute more homogenously with the addition of a cosolvent (CH3CN or THF). We attribute this homogenous phase distribution to the increased total volume of the solvent with the addition of the cosolvent, which would help the homogenous mixing of different components.

Figure 2

SEM images of composites made from (a) MeOH only; (b) CH3CN/MeOH; (c) THF/MeOH; and (d) THF only. Inset: the physical appearance of the composite under that condition.

More importantly, these cosolvents can control the drying process of forming the HAp–Gemosil composite because the mixed solvent (cosolvent + MeOH) would have a slower evaporation rate than that of the pure MeOH. If this were the case, a further deviation of the boiling point of the cosolvent from that of the main solvent (i.e., MeOH) would have a more appreciable effect on the drying process and the morphology of the as-formed composite. Indeed, the composite made from dioxane/MeOH (Supporting Information Figure S4c) shows larger cracks than the one made from THF/MeOH. This can be ascribed to the fact that the boiling point of dioxane (101 °C) is too far from that of MeOH than that of THF (66 °C).

While our results show that mixing cosolvents into the main solvent can be beneficial to improving the morphology and mechanical strength of our composites, having pure cosolvents alone (i.e., cosolvent as the sole solvent) cannot offer the same improvement. This can be seen by comparing the composite made from cosolvent/MeOH with that made from THF only. The composite made with THF only shows less homogeneity and weaker mechanical strength (Figure d and entry 5 in Table S1). We speculate that rapid gelation would occur when pure THF was used, which would lead to the inhomogeneous mixing of different components in the composite.

Finally, we applied TEM to investigate more details about each component in the composite. In principle, if the cosolvent did not react with the HAp–Gel crystal, we should be able to observe intact HAp–Gel crystals. Indeed, TEM images of composites show that the HAp–Gel nanocrystals remain intact after the processing with cosolvents (Supporting Information Figure S6). This further proves that the improvement of mechanical strength by these cosolvents could be solely related to the optimized enTMOS matrix structure and the more homogenous mixing of different components in the composite.

Biocompatibility Study

Good biocompatibility of the composite is a prerequisite for BTE. A previous study has demonstrated that HAp–Gemosil has good biocompatibility.[7] In the current study, we add a cosolvent and CHX as additional components during the processing, and there is a possibility that the residual cosolvent and CHX could lead to cell toxicity. To investigate this possible pitfall, the following experiments were carried out. We chose THF/MeOH as the representative cosolvent/MeOH system and varied the amount of the cosolvent and CHX as independent variables. Experimentally, four different types of composites made from (a) MeOH and Ca(OH)2, (b) MeOH and Ca(OH)2 CHX, (c) THF/MeOH and Ca(OH)2, and (d) THF/MeOH and Ca(OH)2 CHX were incubated with cells. The viability of the potential transplanted cells in BTE application, rMSCs, was monitored by RealTime-Glo MT cell viability assay. We used this method to distinguish the direct toxicity monitored by the cells attached to the composites and the leaching toxicity measured by the cells in the well surrounding the bulk materials. However, a quick screening showed poor cell attachment on the bulk composite made from Ca(OH)2 with CHX and its surrounding area in the cell culture well, which indicated acutely direct and leaching cytotoxicity resulting from CHX. We thus focused on comparing the cell toxicity of group (a) and group (c) to understand the impact of THF on cell viability. As shown in Figure , the luminescent reading gradually increased from day 0 to day 1, suggesting that the addition of THF during the composite processing did not decrease the cell viability compared with the nontoxic composite processed with MeOH only. The decreased luminescence on day 2 may result from either the exhaustion of reagents or the confluency of cells on the composite. Importantly, we further discovered that even when we replaced Ca(OH)2 CHX with just Ca(OH)2, the compressive strength of composites (with or without CHX) was comparable (Supporting Information Table S2). Therefore, we recommend to replace Ca(OH)2 CHX with Ca(OH)2 for future use of the composites where good biocompatibility is required.

Figure 3

Cell viability test by RealTime-Glo MT cell viability assay (two way ANOVA analysis, P = 0.0012).

3D-Printed Scaffold Formation and Mechanical Testing

In addition to biocompatibility, other important considerations for scaffolds intended for BTE include the feasibility of forming porous structures yet still maintaining good mechanical strength. To this end, we fabricated a 3D porous structure with the HAp–Gemosil (THF/MeOH) composite via computer-aided design. A good control on the porous architecture can be achieved (Figure a). Moreover, porous scaffold processed with THF/MeOH has a compressive strength ∼11.33 ± 1.25 MPa, increased by ∼ 60% compared to the compressive strength of the scaffold processed with MeOH (6.94 ± 1.01 MPa) (Figure b). This increment further proves the effect of cosolvents in improving mechanical strength of the HAp–Gemosil composite. Furthermore, the achieved compressive strength of the scaffold processed with cosolvents is comparable to that (1–13 MPa) of the cancellous bone.[20] All these indicate a great potential of such a scaffold for BTE. Further, the in vivo test is in progress to access the feasibility of such scaffolds for biomedical applications.

Conclusions

In conclusion, we successfully improved the mechanical strength of our previously developed HAp–Gemosil composite significantly with the aid of selected cosolvents (e.g., THF, CH3CN, THP, etc.). We further demonstrated that the improvement of mechanical strength was not due to chemical interactions/reactions between the different components in HAp–Gemosil and cosolvents. Instead, adding cosolvents helped the enTMOS network formation and the composite processing. As a result, the likely uniformly cross-linked enTMOS matrix and the more homogenous composite would result in higher mechanical strength. Moreover, we showed that the cosolvent/MeOH could be completely removed from the composite during the drying process, thus having minimum impact on the biocompatibility of this new HAp–Gemosil composite. Finally, we demonstrated that the porous scaffold processed with cosolvents can be easily made yet can maintain good compressive strength. All these results point to that this newly modified HAp–Gemosil composite is a very promising candidate for BTE. Currently, we focus on evaluating the effect of this new scaffold for bone regeneration in vivo.