3D printed mesoporous bioactive glass,bioglass 45S5,and β-TCP scaffolds for regenerative medicine: A comparative in vitro study

Abstract

BACKGROUND:

While autografts to date remain the “gold standard” for bone void fillers, synthetic bone grafts have garnered attention due to their favorable advantages such as ability to be tailored in terms of their physical and chemical properties. Bioactive glass (BG), an inorganic material, has the capacity to form a strong bond with bone by forming a bone-like apatite surface, enhancing osteogenesis. Coupled with additive manufacturing (3D printing) it is possible to maximize bone regenerative properties of the BG.

OBJECTIVE:

The objective of this study was to synthesize and characterize 3D printed mesoporous bioactive glass (MBG), BG 45S5, and compare to β-Tricalcium phosphate (β-TCP) based scaffolds; test cell viability and osteogenic differentiation on human osteoprogenitor cells in vitro.

METHODS:

MBG, BG 45S5, and β-TCP were fabricated into colloidal gel suspensions, tested with a rheometer, and manufactured into scaffolds using a 3D direct-write micro-printer. The materials were characterized in terms of microstructure and composition with Thermogravimetric Analyzer/Differential Scanning Calorimeter (TGA/DSC), Fourier Transform Infrared Spectroscopy (FTIR), X-ray Diffraction (XRD), Micro-Computed Tomography (μ-CT), Scanning Electron Microscopy (SEM), Energy Dispersive X-ray Spectroscopy (EDS), and Mattauch–Herzog-Inductively Coupled Plasma-Mass Spectrometry (MH-ICP-MS).

RESULTS:

Scaffolds were tested for cell proliferation and osteogenic differentiation using human osteoprogenitor cells. Osteogenic media was used for differentiation, and immunocytochemistry for osteogenic markers Runx-2, Collagen-I, and Osteocalcin. The cell viability results after 7 days of culture yielded significantly higher (p < 0.05) results in β-TCP scaffolds compared to BG 45S5 and MBG groups.

CONCLUSION:

All materials expressed osteogenic markers after 21 days of culture in expansion and osteogenic media.

Keywords

3D printing bioglass 45S5 mesoporous bioactive glass β-TCP scaffolds

1. Introduction

Bone is the second most commonly transplanted tissue due to bone defects [1], caused by high impact trauma, osteoporosis, congenital diseases, or resection due to infection or bone tumor [1–4]. Furthermore, bone defects can be classified as non-critical or critical, where the later does not have the capacity to heal spontaneously, ultimately requiring surgical intervention [4]. Surgical procedures to rehabilitate the defect utilize grafts which may be natural (autografts or allografts) or synthetic (alloplasts) [5]. Autologous bone grafts, harvested from the patient, are considered the “gold standard”, due to their osteogenic, osseoconductive and osseoinductive capacity [6]. While autografts are the preferred option, they are associated with respective disadvantages such as secondary surgical site, restricted shape, limited bone quantity, absorption over time and risk of infection [2,5–8]. Due to their limitations, focus has shifted to alloplasts, consisting of various osseoconductive synthetic biomaterials to be utilized for bone regeneration [1,5,8–10]. While alloplasts are available in nearly unlimited quantities, their primary disadvantage is degree of osseoinductivity [5,8].

An array of biomaterials have been evaluated and considered to be utilized as osseoconductive scaffolds (bone void fillers) for bone regeneration [1]. Calcium-phosphate based bioceramics have excellent osseoconductivity properties and facilitate protein adhesion, in addition to being degradable [5,7,11,12]. For example, hydroxyapatite has been one of the most widely investigated, calcium-phosphate based ceramics, due to its favorable composition, similar to the inorganic part of the bone [6]. Although it is biocompatible and osseoinductive, its slow degradation rate (∼1% per year in bulk form) interferes with bone regeneration [6]. As an alternative, β-TCP which has a faster degradation rate has shown encouraging results in terms of bone regeneration and scaffold resorption [6,7].

An alternative is BG, which is a synthetic vitreous structure with bioactive properties. The material has the ability to form strong and fast bonds with bone through the formation of a bone-like apatite surface that increases bone attachment and formation [1,13]. Since the discovery of BG 45S5 in the late 1960s there have been several different compositions introduced [14,15]. A newer type of BG, MBG, presents ordered mesoporous channels with pores ranging from 2 to 20 nm; different from conventional BGs which have no mesoporous component [16,17]. MBG has yielded improved results in terms of bone regeneration in comparison to non-MBGs [18] due to the increased surface area, pore volume, and ordered mesoporous structure [3,18–21].

An important mechanism for BG osteoinduction is the controlled ion release from the degrading BG, with the concentration of soluble Ca²⁺ and Si⁴⁺ from the scaffold being critical for cells to proliferate and differentiate [22]. BG by-products released as the material dissolves can generate up-regulation of seven families of genes in osteoblasts [23,24] and enhance the expression of osteogenic markers such as Collagen type I, ALP, osteocalcin, and Runx-2 [9]. The bioactive properties of BGs are improved with the incorporation of elements such as Sr, Zn, Mg, and B that add angiogenic and antibacterial properties [1].

Studies have established that osseoconductivity is affected by physical properties ranging from the macro-level to the nanometer level with respect to the geometric design [6]. Direct-write technology, an additive manufacturing (AM) process, allows for the fabrication of patient specific devices for tissue regeneration by the extrusion of colloidal gels [8]. 3D printed devices/scaffolds have the capacity to provide an ideal template for cell proliferation and differentiation with an interconnected pore structure that supports nutrient transportation, regenerating bone defects [18]. In summary, not only are the material properties important, but also the spatial configuration, which is achievable through 3D printing. The shape, porosity, and size of the scaffolds can be tailored, giving the material an advantage for osseoconduction in comparison to their bulk or particulate counterparts. The objective of this research was to synthesize and characterize 3D printed MBG, BG 45S5, and β-TCP scaffolds; test cell viability and osteogenic differentiation on human osteoprogenitor cells in vitro. The hypothesis was that BG scaffolds would yield superior results in comparison to β-TCP scaffolds with respect to cell proliferation and osteogenic differentiation.

2. Materials and methods

2.1. Colloidal gel preparation

2.1.1. 𝛽-TCP colloidal gel

Raw β-TCP powder was calcined at 800 °C for 11 hours (h) and attrition milled with 1-mm zirconia milling media in distilled water. The colloid gel ink was formulated with deionized water (DI-H₂O), hydroxypropyl methylcellulose, ammonium polyacrylate, polyethyleneimine and the calcined and milled β-TCP powder. A volume fraction of ∼0.45 of β-TCP was obtained by combining β-TCP powder and ammonium polyacrylate dispersant (Darvan 821A, 40% ammonium polyacrylate aqueous solution), stabilizing the ceramic particles in the DI-H₂O, using ∼15 mg of dispersant per gram of ceramic. After the suspension was mixed, a viscosifying agent, hydroxypropylmethylcellulose (F4M), and polyethyleneimine 10% by weight solution were added [8].

2.1.2. BG 45S5 colloidal gel

The BG45S5 powder (45 wt% SiO₂, 6 wt% P₂O₅, 24.5 wt% CaO, and 24.5 wt% Na₂O) was formulated following the protocol described by Pirayesh and Nychka [25] with tetraethyl orthosilicate (Si(OC₄H₉)₄, 99.99%), triethyl phosphate (P(C₂H₅O)₃, 99.5%), calcium nitrate tetrahydrate ((Ca(NO₃)₂⋅4H₂O, 99.60%) and sodium nitrate (NaNO₃, 100.40%). First tetraethyl orthosilicate (TEOS; Si(OC₄H₉)₄, 99.99%) and nitric acid (HNO₃ (1 M) was used as a catalyst) were added to a glass beaker and stirred for 60 min at room temperature. Afterwards, triethyl phosphate (TEP; P(C₂H₅O)₃, 99.5%), calcium nitrate tetrahydrate and sodium nitrate were added separately in 45 min intervals with continuous magnetic stirring [25]. The suspension was dried at 70 °C (24 h), followed by heating to 120 °C (48 h until dry) and finally calcined at 700 °C in a dry furnace for 24 h (heating rate 80 °C/h, cooling rate 300 °C/h. The calcined BG 45S5 was milled with ethyl alcohol and 5-mm zirconia-milling media [25]. To obtain the BG 45S5 colloid gel for 3D printing, the powder was suspended in DI-H₂O and combined with carboxymethyl cellulose (CMC, average Molecular weight ∼250,000) to act as a binder, dispersant, and gelation agent following the protocol described by Eqtesadi et al. [26] where a 45 vol% BG 45S5 suspension in DI-H₂O was combined with 1% CMC [26].

2.1.3. MBG colloidal gel

The MBG powder was formulated using 16 g of P123 (nonionic block copolymer) as a structure-directing agent, 5.6 g of calcium nitrate tetrahydrate ((Ca(NO₃)₂⋅4H₂O, 99.60%), 26.8 g of TEOS, 2.92 g of TEP and 4 g of HCl (0.5 M) based on the protocol from [27], for a final Si/Ca/P molar ratio of 80:15:5. First the P123 and HCl were added to 240 g of ethanol and stirred for 60 min at room temperature. The other required chemicals were added in 45 min intervals in the following order: calcium nitrate tetrahydrate, TEOS, and TEP. The solution was left under constant stirring for 24 h to formulate the sol. The sol was afterward dried for 72 h at 40 °C in the oven for evaporation-induced self-assembly (EISA). Lastly, it was calcined at 600 °C for 6 h (heating rate 80 °C/h, cooling rate 300 °C/h) and 400 mesh sieved [27,28]. The MBG paste was made with an MBG/SA ratio of 2:1 of MBG powder, with sodium alginate (SA) dissolved in DI-H₂O (1:10 water mass ratio) following the protocol described by [27].

2.2. Rheological analysis and evaluation

Approximately 1 g of sample of each colloidal gel (β-TCP, BG 45S5, and MBG) was loaded to the rheometer (Discovery HR-2, TA Instruments New Castle, DE, USA) equipped with a 20 mm parallel, steel Peltier plate. The water reservoir of the cover plate was filled to minimize evaporation. The equipment was programmed for an oscillation amplitude test with frequency set at 1 Hz using a logarithmic sweep stress of 1–10⁴ Pa (all data were collected at 25°C).

2.3. 3D printing

Scaffolds were designed to have 10 mm diameter and a ∼2.5 mm height with 490 μm pore spacing and 510 μm struts (Fig. 1) via computer assisted design (RoboCAD 4.5, 3D Inks LLC, Tulsa, OK, USA) and fabricated using a custom-built 3D direct-write micro-printer (3D Inks LLC, Tulsa, OK, USA). The respective colloid gels were loaded into a 3-mL syringe equipped with a 510 μm-diameter extrusion nozzle. The scaffolds were printed in low viscosity paraffin oil (β-TCP and BG 45S5) or air (MBG).

Fig. 1.

Representative CAD image.

2.4. Post processing: sintering and/or lyophilization

After 3D printing, β-TCP scaffolds were removed from the oil, left to partially dry in a low temperature oven at 80°C, before being sintered to 1100°C in a dry furnace to eliminate impurities and densify the constructs [7]. The 3D printed BG 45S5 scaffolds were removed from the bath and left to dry at room temperature for 24–48 h, then were thermally treated to remove the CMC at 450°C and sintered to 1000°C (25–450°C heating rate 60°C/h, 450°C for 2 h, 450–1000°C heating rate 120°C/h, 1000°C for 1 h, 1000–25°C, cooling rate 300°C/h) [26]. Following the protocol described by Luo et al., the 3D-printed MBG scaffolds were immersed in 500 mM CaCl₂ solution for 10 h for the crosslinking of SA [29]. Afterwards the scaffolds were washed 3 times in DI-H₂O, frozen at −80°C for at least 12 h and freeze-dried for 24 h.

2.5. Material characterization methods

2.5.1. Thermogravimetric analysis/differential scanner calorimetry

The TGA/DSC (Model SDT Q600, TA Instruments, New Castle, DE, USA) analysis was performed with approximately 6.0 (±2) mg of each material in an Al₂O₃ crucible. The analysis was carried between 20°C to 1100°C, at a rate of 10°C/min in zero hydrocarbon air (80/20 N₂/O₂ ratio), using a flow rate of 100 mL/min [8].

2.5.2. X-ray diffraction

The ground powder of each material was loaded onto a Si substrate, zero background, flat sample holder. All scans used a curved crystal monochromator with the following settings: 45 kV, 40 mA, step size 0.013° (2θ) at 99.45 s per step, and a range from 20–80° (PanAlytical X’Pert powder X-Ray diffractometer, Malvern Panalytical, Almelo, The Netherlands).

2.5.3. Fourier transform infrared spectrometer

1 mg of grounded sample powder was mixed with 250 mg of KBr (IR grade) and pressed at 10,000 psi under vacuum using a hydraulic press (Pike Technology-Crush IR, Madison WI, USA) to fabricate pellets. The pellets were scanned by Nicolet FTIR Magna 550 series II (Thermo Electron Corporation, Madison, WI, USA), covering the range from 4000–400 cm⁻¹.

2.5.4. Micro-computed topography

Three scaffolds of MBG (lyophilized), BG 45S5 (sintered), and β-TCP (sintered) were scanned using a micro-CT (micro-CT 40, Scanco Medical AG, Bruettisellen, Switzerland). The scaffolds were individually mounted to a 16 mm diameter specimen holder. Additionally, 3 hydrated MBG scaffolds were embedded in polymethyl methacrylate (PMMA) and individually mounted to a 30 mm diameter specimen holder. The settings for all scans were: 70 kVp, 114 μA, 8 Watts, BH at 200 mg HA/cm², FOV/diameter 16.4 mm, and voxel size 8 μm. For the embedded MBG scaffolds, the FOV/diameter and voxel size were 30.7 mm and 15 μm, respectively. The DICOM files were then transferred to Amira (version 6.3.0) for reconstruction and volume analysis.

2.5.5. Field emission-scanning electron microscopy

The Zeiss Gemini 300 FE-SEM (Carl Zeiss Microscopy, Oberkochen, Germany) was used in SE mode to observe the surface morphology of β-TCP, BG 45S5, and MBG scaffolds. Scaffolds of the three materials were also embedded in PMMA, cut and polished to obtain the cross-sections and observe strut shape and inner architecture of the scaffolds.

2.5.6. Energy dispersive X-ray spectroscopy

A Bruker Xflash 6|10 (Bruker, Berlin, Germany), attached to the Hitachi S-3500N SEM (Hitachi High-Tech, Clarksburg, MD) was operated with 20 kV and 15 mm working distance. The β-TCP, BG 45S5, and MBG scaffolds were each mounted on aluminum stubs (Electron Microscopy Sciences, USA) and coated (Denton Vacuum Desk V Sputter Coater, Somerset, NJ, USA) with a thin layer of carbon for 30 s at 50 mA) for elemental analysis.

2.5.7. Mattauch–Herzog-inductively coupled plasma-mass spectrometry

A dissolution test was performed with conditioned (described on in vitro experiments) and non-conditioned MBG and BG 45S5 scaffolds in 20X (weight) DI-H₂O at 37°C and 5% CO₂ to obtain the concentration of Ca, P, Si, and Na at different time points of 24, 48, and 72 h. All groups were tested in triplicate, and each sample tested three times. The extracted medium was used undiluted and diluted with ultrapure water (18.2 MΩ.cm, ELGA, Purelab^® Ultrapure Water Purification Systems, Buckinghamshire, UK) prior to the analysis. The analysis was performed by simultaneous inductively coupled plasma-mass spectrometry, known as an Mattauch–Herzog-ICP-MS (MH-ICP-MS) instrument (SPECTRO MS, SPECTRO Analytical Instruments GmbH, Kleve, Germany), previously described by Bäuchle et al. [30]. Both diluted and undiluted samples were acidified to 2% (vol/vol) with 65% Suprapur nitric acid (analytical-reagent grade, Merck, Darmstadt, Germany) and transferred with a Cetac autosampler (ASX-560, Teledyne Cetac Technologies, Omaha, NE, USA) to the MH-ICP-MS instrument. TraceCERT^® periodic table Mix 1 for ICP, diluted Merck VI multi-element standard (Millipore-Sigma, Darmstadt, Germany), as well as single element standard solutions of the four target elements (Inorganic Ventures, Christiansburg, VA, USA), were analyzed as controls for analysis repeatability. The concentrations used were in the range of parts per million (ppm) and parts per billion (ppb).

2.6. In vitro experiments

MBG and BG 45S5 scaffolds to be used for cell culture experiments were sterilized using gamma irradiation for 30 min at 27.797 Gy, 125 kV, 5 mA (CellRad X-Ray Irradiator, Faxitron, Tucson, AZ, USA). β-TCP scaffolds were autoclaved (30-min sterilization cycle) prior to cell culture use. Since bioactive glass initially releases a high burst of ions, both BG and MBG scaffolds were preconditioned to avoid cell death due to pH changes. MBG scaffolds were treated in DI-H₂O at 37 °C and 5% CO₂, static conditions (refreshed after 24 h, and left for an additional 72 h), followed by immersing in Dulbecco’s Modified Eagle Medium (DMEM) high glucose, L-glutamine (Gibco^{^TM}) for 24 h. BG 45S5 scaffolds were treated following the protocol described by Boccardi et al. [22]. Individual scaffolds were set in a 6-well plate with 4 mL of the DMEM + 30 mM N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid (HEPES; Gibco^{^TM}) and incubated at 37°C and 5% CO₂, static conditions, with daily media changes. At day 9±1, 1 M HCl (sterile-filtered) was used to decrease and adjust the pH to be slightly less than 8.0 (measured with a pH meter-Mettler Toledo FiveEasy Plus). Once the pH was below 8.0, the scaffolds were washed three times using Dulbecco’s phosphate-buffered saline (DPBS 1X, no calcium, no magnesium, Gibco^{^TM}), immersed in DMEM for 1 h, and then transferred to 24-well plates.

After receiving approval from NYU Langone Health’s Institutional Review Board (NYU IRB#: s18-01579), human osteoprogenitor (hOP) cells were harvested from unused hard tissue in young patients following surgery. The tissues were divided into small pieces and cultured in 6-well plates at 37 °C and 5% CO₂ to obtain monolayer cultures. hOP cells were expanded with DMEM, high glucose, L-glutamine, 10% Embryonic Stem Cell Qualified Fetal Bovine Serum (ES-FBS), 1% Antibiotic-Antimycotic (all from Gibco^{^TM}) until ∼80% confluence. 50,000 cells per scaffold were seeded in 24-well plates (n = 3 for 3 cell lines) and cultured for 7 days (cell viability assay) in expansion media with 25 mM HEPES and 21 days (for osteogenic differentiation) in both expansion and osteogenic media. PrestoBlue^{^TM} Cell Viability Reagent (Invitrogen) was used for the fluorescence assay at 24, 48, 72, and 168 h. The cells with the PrestoBlue^{^TM} Reagent were incubated for 15 min (at 37°C and 5% CO₂) followed by the fluorescence reading performed with a microplate reader (FilterMax F5, Molecular Devices, San Jose, CA, USA). After day 7, the cells were fixed with 4% paraformaldehyde (Sigma-Aldrich) overnight at 4°C and stained with Phalloidin-iFluor 488 Reagent (Abcam) 1:1000 and diamidino-2-phenylindole (DAPI; from Abcam) 1:10,000 for fluorescence imaging (DM IL LED, Leica Microsystems, Buffalo Grove, IL).

Osteogenic media was composed of expansion medium (with 25 mM HEPES), 100 mM β-glycerophosphate (Sigma-Aldrich), 0.1 μM dexamethasone (Sigma-Aldrich), 100 μg/mL ascorbate-2-phosphate (Sigma-Aldrich). At the end of week 3, the cells were fixed in 4% paraformaldehyde, permeabilized for 1 h with blocking buffer (10% ES-FBS, 3% Bovine Serum Albumin (Boston BioProducts), 0.2% Triton-X in DPBS 1X) and immunocytochemistry with three osteogenic markers was performed: Runx-2 1:50 (mouse IgG monoclonal primary antibody, abcam)/Donkey anti-Mouse IgG (H+L), Highly Cross-Adsorbed Secondary Antibody Alexa Fluor 594 1:200, Collagen-I 1:100 (rabbit IgG polyclonal primary antibody, abcam)/Donkey anti-Rabbit IgG, (H+L) Highly Cross-Adsorbed Secondary Antibody Alexa Fluor 488 1:400, Osteocalcin 1:100 (mouse IgG monoclonal primary antibody, Invitrogen)/Donkey anti-Mouse IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 1:200. DAPI (1:10,000) was used to stain the nuclei.

2.7. Statistical analysis

Levene’s test for homogeneity of variances, one-way ANOVA with 95% confidence interval and Tukey’s post hoc test was performed in SPSS v.27 to compare the 3 materials (β-TCP, BG 45S5, and MBG) based on the free volume% from the micro-CT analysis (Amira). A Mixed Model Analysis and Pairwise comparisons with Least Significant Difference Post hoc test at 95% confidence was performed for the cell viability assay with the criteria time point-material and time point-material-cell line.

3. Results

3.1. Characterization of β-TCP, BG 45S5, and MBG

Oscillation (shear) stress 𝜏 increased (logarithmically), storage modulus G’ is maintained constant in the linear viscoelastic region until reaching yield value, crossing to the non-linear region where G’ decreased rapidly (Fig. 2). The BG 45S5 colloid gel yielded the highest storage modulus (>10⁵ Pa), while the MBG colloid gel gave the second highest with ∼10⁴ Pa, and β-TCP gel presented the lowest storage modulus. The higher storage modulus for MBG and BG 45S5 inks indicates that they are a more viscous gel in comparison to the β-TCP ink. All three colloidal gels are within the adequate range for extrusion, being able to be extruded through a thin nozzle while maintaining their shape; i.e. its shear thinning.

Fig. 2.

Shear storage modulus versus oscillation stress (log–log) for MBG, Bioglass 45S5 and β-TCP colloid gels.

After the rheological test, the scaffolds were 3D printed and imaged with a brightfield microscope (Fig. 3) and with SEM (Fig. 4) after sintering (BG 45S5 and β-TCP scaffolds), and after lyophilizing and hydration (MBG scaffolds). The SEM images (Fig. 4) of the different scaffolds demonstrated that the scaffolds retained their cylindrical strut shape and square pore spacing on the outer surface and on the inner architecture (as observed with the cross-sections). The surface roughness is also observed, where the MBG scaffold has a rougher surface compared to the other two materials, followed by the BG 45S5 scaffold and finally β-TCP which has a smoother surface.

Fig. 3.

(A) β-TCP sintered scaffold. (B) Bioglass 45S5 sintered scaffold. (C) Lyophilized Mesoporous Bioactive Glass scaffold. (D) Hydrated Mesoporous Bioactive Glass scaffold after lyophilizing.

Fig. 4.

(A) Mesoporous Bioactive Glass scaffold. (A1–A2) Cross-section of Mesoporous Bioactive Glass scaffold. (B) Bioglass 45S5 scaffold. (B1–B2) Cross-section of Bioglass 45S5 scaffold. (C) β-TCP scaffold. (C1–C2) Cross-section of β-TCP scaffold.

The elemental analysis confirmed the presence of calcium and phosphate in β-TCP, the presence of Ca, P, Si, and Na in BG 45S5, and the presence of Ca, P, Si, and Cl in MBG (Fig. 5). While MBG does contain Na, and the MH-ICP-MS analysis demonstrated sodium release, it was not detected on the EDS. The β-TCP XRD spectra (Fig. 6A) indicates the three characteristic peaks representative of β-TCP. The BG 45S5 XRD spectra (Fig. 6A) shows characteristic peaks for Ca₃Na₆O₁₈Si₆, while the MBG spectra, indicates that it is amorphous (calcined only up to 600 °C) with no presence of definitive peaks.

Fig. 5.

Atom% versus elements present in MBG, Bioglass 45S5, and β-TCP scaffolds (sintered/lyophilized respectively). Error bars represent absolute error (%).

Fig. 6.

XRD and FTIR spectra. (A) Intensity (counts) versus angle 2θ for β-TCP (crystalline), Bioglass 45S5 (crystalline) and MBG (amorphous). (B) Absorbance Intensity versus wavenumber (cm⁻¹) for the three materials.

The FTIR spectra corresponding to β-TCP (Fig. 6B) shows the presence of PO₄ (wavenumbers ∼1100–1040 cm⁻¹ and ∼600–560 cm⁻¹), as well as P₂O₇ groups (wavenumbers 850–600 cm⁻¹) with P–O stretching bonds [8,31]. The BG 45S5 spectra (Fig. 6B) is similar to the results obtained by Pirayesh et al. [25] and Vafa et al. [32], which indicate the presence of Si–O stretching bonds (∼1039 cm⁻¹, ∼930 cm⁻¹, ∼729 cm⁻¹, ∼696 cm⁻¹), Si–O bending bonds (∼460 cm⁻¹), and crystalline P–O bending bonds (∼520 cm⁻¹, ∼620 cm⁻¹) in the material. At ∼1450 cm⁻¹ there is another peak that corresponds to residual carbonate groups from the precursors [32]. The MBG spectra (Fig. 6B) is similar to the spectra obtained by Wu et al. [16] and Shah et al. [33], indicating the presence of Si–O stretching/bending bonds, P–O bending bonds, O–H bonds, and carbonate groups. TGA/DSC (Fig. 7) analysis for both BG 45S5 and β-TCP (green-state) showed changes corresponding to water evaporation (100°C) and binder burn out (∼450°C). The TGA/DSC for MBG yielded a different profile due to the higher organic content.

Fig. 7.

TGA and DSC for green-state materials (A) Bioglass 45S5. (B) β-TCP. (C) MBG.

The 3D reconstructed DICOM files of the β-TCP, BG 45S5, and MBG scaffolds are illustrated in Fig. 8A–D. The four groups (all analyzed in triplicates: β-TCP, BG 45S5, MBG lyophilized, and MBG hydrated after lyophilizing) were statistically homogeneous (Levene’s test p > 0.05), with statistically different volumes (p < 0.05) between subgroups (Fig. 8E). The lyophilized MBG scaffold was statistically different compared to the other scaffolds (subgroup C, p < 0.05), while the BG 45S5 scaffold (subgroups A, B) was not statistically different compared to the hydrated MBG (subgroup B) and β-TCP scaffolds (subgroup A).

Fig. 8.

3D reconstructions in Amira for the different scaffolds. Error bars represent 95% confidence interval (CI). (A) Bioglass 45S5 scaffold. (B) β-TCP scaffold. (C) MBG scaffold (lyophilized). (D) MBG scaffold (hydrated after lyophilizing). (E) Average free volume% of β-TCP, Bioglass 45S5, and MBG scaffolds.

The results obtained before and after the preconditioning treatment indicated a significant decrease in the scaffold ion release of both materials (Fig. 9), especially sodium and silicon in BG 45S5 scaffolds, and sodium and calcium in MBG scaffolds.

Fig. 9.

MH-ICP-MS results for ion concentration from extracted media. Error bars represent 95% CI. PT = Preconditioning Treatment. (A–D) Ion concentration for Bioglass 45S5 scaffolds and (E–H) Ion concentration for MBG scaffolds.

3.2. In vitro study of β-TCP, BG 45S5, and MBG scaffolds

As presented in Fig. 10, the hOP cells in all materials and control displayed normal cell morphology throughout the 7-day culture period, although the number of cells changed between materials. After the 7 days, the cells were stained with phalloidin and DAPI. The cells showed normal cell morphology (spindle-shaped cells and regular nuclei) in all materials as well as in the control.

Fig. 10.

(A1–D1) Brightfield microscopy of hOP cells after 7 days of culture. Each image shows a pore in the scaffold, with the borders outlined for reference. (A2–D2) hOP cells stained with phalloidin iFluor 488 and DAPI. (A) MBG scaffold. (B) β-TCP scaffold. (C) Bioglass 45S5 scaffold. (D) Control. (E) Cell viability (% of hOP control cells without material) of the three different scaffolds in each time point. Letters represent statistically homogenous subgroups for each time point, error bars represent 95% CI.

The cell viability assay was performed after 24, 48, 72, and 168 h (7 days). hOP cells cultured without material were considered the positive control, represented as 100% cell viability to compare with cells cultured on the β-TCP, MBG, and BG 45S5 scaffolds. The results obtained were analyzed separately for each time point to compare between the different materials (cell lines combined Fig. 10E). After 24 h, there was no significant difference between materials (p > 0.05). After 48 and 72 h, the BG scaffolds were statistically different (p < 0.001) compared to the MBG and β-TCP scaffolds. After 168 h, the β-TCP scaffolds were statistically different −(p < 0.001) compared to the MBG and BG scaffolds, while between MBG and BG scaffolds there was no significant difference (p > 0.05).

The results were also analyzed individually with each cell line (figures not shown) due to normal variability between cell lines (different donors). The results for each cell line yielded no significant differences (p > 0.05) between materials after 24 h. After 48 h, there were differences in cell viability between cell lines. BG scaffolds were statistically different compared to MBG (p = 0.009 cell line 1, p = 0.019 cell line 2) and β-TCP (p = 0.022 cell line 1, p = 0.002 cell line 3) scaffolds on cell lines 1 and 3, while on cell line 2 there was only significant difference (p < 0.001) between BG and β-TCP cell viability. BG and MBG for cell line 2 was not statistically significant (p = 0.051). After 72 h of incubation, BG yielded statistically significant differences with MBG (p = 0.026, p < 0.001, p = 0.015 respectively for cell lines 1, 2, and 3) and β-TCP (p = 0.003, p < 0.0001, p < 0.001 respectively) on all cell lines. Additionally, MBG yielded a significant difference with β-TCP on cell lines 2 (p = 0.014) and 3 (p = 0.025). After 7 days (168 h), cell viability for β-TCP scaffolds yielded a significant difference with MBG (p < 0.001, p = 0.002, and p = 0.029, respectively) and BG scaffolds (p < 0.001, p = 0.012, and p < 0.001, respectively) on all cell lines. Cell line 3 also yielded statistically different (p = 0.037) results between MBG and BG.

Cells with the three different scaffolds and control were cultivated for 21 days in expansion and osteogenic media. After 21 days, immunocytochemistry (ICC) was performed. All groups (β-TCP, MBG, BG, and control) cultured in osteogenic media (Fig. 11) and expansion media (Fig. 12) presented high expression of early osteogenic markers Runx-2 and Collagen-I. Osteocalcin expression (late-stage osteogenic marker) presented low expression in hOP cells cultured with the scaffolds, nonetheless it was present in both osteogenic and expansion media groups with scaffolds. For the control group the expression of Osteocalcin could neither be clearly identified in osteogenic nor in expansion media.

Fig. 11.

Fluorescence images of hOP cells cultivated in osteogenic media for 21 days with Runx-2, Collagen-I, Osteocalcin, DAPI on (A) MBG scaffolds (B) Bioglass 45S5 scaffolds (C) β-TCP scaffolds (D) control without material. (1) Runx-2 and DAPI (2) Collagen-I and DAPI (3) Osteocalcin and DAPI. Scale bars measure 100 μm.

Fig. 12.

Fluorescence images of hOP cells cultivated in expansion media for 21 days with Runx-2, Collagen-I, Osteocalcin, DAPI on (A) MBG scaffolds (B) Bioglass 45S5 scaffolds (C) β-TCP scaffolds (D) control without material. (1) Runx-2 and DAPI (2) Collagen-I and DAPI (3) Osteocalcin and DAPI. Scale bar measures 100 μm.

4. Discussion

Diverse biomaterials have been extensively studied to replace the need for autologous bone grafts, the “gold standard”, due to limitations on the latter, and the advantage of synthetic bone grafts to tailor its physical and chemical properties for bone regeneration [1]. Thus, the objective of this study was to synthesize, characterize, and test cell viability and osteogenic differentiation in vitro of 3D printed MBG, BG 45S5, and β-TCP scaffolds.

The cell viability results demonstrated significantly higher (p < 0.05) viability in β-TCP scaffolds compared to MBG and BG scaffolds after 7 days of culture. Therefore, the research hypothesis corresponding to cell viability was rejected, since MBG and BG scaffolds did not yield superior results relative to β-TCP. The research hypothesis aimed to compare osteogenic differentiation between β-TCP and the two bioactive glass groups could not be addressed. After 21 days of culture, hOP cells incubated with the three different materials were positive for Runx-2 and Collagen-I in immunocytochemistry. To determine whether β-TCP, MBG, or BG scaffolds yield superior results for osteogenic differentiation, qPCR with quantitative results will be needed.

3D printing allows to strategically design porosity, shape, and size of the scaffolds to support cell proliferation, differentiation, and nutrient transportation [18]. To successfully obtain 3D printed scaffolds from a colloid gel, the “ink” must present the capacity of shear thinning, meaning that the gel must be able to be extruded through a thin nozzle and maintain its shape after extrusion [26]. The rheometric results for β-TCP, MBG, and BG colloid gels determined the presence of this property (high storage modulus G’ and high shear oscillation stress 𝜏). SEM images of the scaffolds confirmed they maintained the architecture of struts and pores after sintering/lyophilizing and the higher porosity of MBG scaffolds compared to BG 45S5 and β-TCP.

The 3D reconstructions from micro-CT scans demonstrated that the macro-porosity of the scaffolds yielded a free volume from 43.2% (β-TCP) to 44.5% (BG 45S5) to 62.7% (lyophilized MBG, while hydrated the free volume decreased to 51.5%). The high free volume% given by 3D printing versus the bulk material achieves a higher surface area [15]. A higher surface area increases the ion release, which is an important factor to consider, especially with 3D printed bioactive glasses that lead to a faster ion exchange, increasing alkalization of the solution [34]. BG scaffolds need and should be preconditioned prior to in vitro experimentation due to the rapid pH increase that surpasses physiological values [22,35] when exposed to biological fluids, causing cell cytotoxicity [34]. This pH change is due to the initial burst ion release, mainly by the exchange of Na⁺ and Ca²⁺ from the BG with H⁺ and H₃O⁺ from the solution [34]. The morphology (powder, bulk, porous scaffold), BG composition, BG synthesis method, culture type and medium used are factors that affect the results on the preconditioning treatment conditions [34].

The EDS analysis confirmed the presence of the respective elements in β-TCP (Ca and P), BG 45S5 (Ca, P, Si, and Na), and MBG scaffolds (Ca, P, Si, and Cl). The elemental analysis (except Cl) were performed by MH-ICP-MS after a dissolution test. The analysis of MBG and BG 45S5 scaffolds verified the high concentration of ions (Na⁺, Si⁴⁺, and Ca²⁺) released in the first 24–48 h without a preconditioning treatment. After preconditioning the scaffolds, the concentration of ions released to the medium decreased considerably; comparing between the 24 h results, the concentrations decreased 10–30 times (∼10x, ∼27x Na⁺ in MBG and BG scaffolds respectively, ∼30x Ca²⁺ in MBG and ∼10x Si⁴⁺ in BG), and the pH lowered to physiological values. Xie et al. studied the dissolution behavior of Ca²⁺ and phosphorus in β-TCP when immersed at 37 °C in Simulated Body Fluid. The concentration of ions released after 24 h was ∼4 mg/L for Ca²⁺ and ∼1.5 mg/L for phosphorous [31]. Compared to β-TCP, bioactive glass has higher dissolution rates.

The composition of each material was studied with XRD, TGA/DSC, and FTIR. The XRD results confirmed that both BG 45S5 and β-TCP are crystalline materials, while MBG is amorphous. Previous research on β-TCP evinced the same 3 characteristic XRD peaks and similar TGA/DSC results [8]. Research based on BG 45S5 by Cacciotti et al. and Motealleh et al. supported the XRD results, with most of the representative peaks corresponding to CaNa₂O₆Si₂ [35,36]. The TGA profile obtained for green-state BG 45S5 is similar to the TGA obtained by Pirayesh and Nychka, with marked weight loss events associated first to water and silanol removal (60°–150°C), second to precursor and catalyst condensation (∼230°C), and third to crystallization of BG [25]. The TGA from this study showed an additional event around 450°C corresponding to the binder removal. Previous literature based on MBG demonstrated the characteristic amorphous spectra in XRD [3,28]. Shah et al. obtained a TGA/DSC profile similar to the results obtained for MBG in this study [33]. The weight loss was associated to water removal, followed by Pluronic degradation and removal [33].

FTIR spectra of β-TCP showed phosphate (PO₄) and pyrophosphate (P₂O₇) groups, as reported previously in the literature [8]. FTIR spectra of BG 45S5 indicated characteristic peaks of Bioglass corresponding to Si–O bending and stretching bonds, and P–O bending, as described in literature [25,32,36]. The carbonate group in the spectra has been associated to residuals from the precursors [32]. The spectra corresponding to MBG presents the same peaks of Si–O, P–O, O–H, and carbonate groups that characterize the material as previous studies have reported [16,33].

The in vitro cell culture experiments yielded no significant differences (p > 0.05) between the different materials after 24 h. Up to day 3 (72 h), MBG performed as well as β-TCP (p > 0.05), before decreasing at 7 days. After 7 days, there was a significantly higher cell viability (p < 0.05) for β-TCP scaffolds compared to BG 45S5 and MBG. BG 45S5 performed statistically lower than β-TCP and MBG at 48, 72, and 168 h. The lower cell viability in BG can be associated with the fast degradation of the scaffolds and in the case of MBG, the disintegration of the scaffolds as well. Previous studies reported by Motealleh et al. observed a similar trend when comparing cell proliferation on cells cultured with amorphous and crystalline BG 45S5 scaffolds versus hydroxyapatite (HA) scaffolds [35]. The faster degradation of BG compared to HA or β-TCP, leads to a more variable environment with higher ion release and pH changes [35]. Also, amorphous BG degrades at a faster rate than crystalline BG, producing a less stable environment which lead to a decrease in cell proliferation in the study by Motealleh et al. [35].

When comparing cell viability with individual cell lines, there were statistically significant (p < 0.05) changes within the same materials. This can be attributed to donor variability, as previously reported in literature on primary cell lines. A study by Herrmann et al. determined there are inter- and intra-donor differences in cell phenotypes that were observed during trilineage differentiation, which could not be attributed to age or gender [37], while Vigilante et al. demonstrated that cell proliferation, cell clumping, and cell area are phenotypic variations correlated to different genes [38].

Collagen-I is associated with extracellular matrix (ECM) formation in early stages of osteoblast differentiation, while Osteocalcin is associated with mineralization in later stages of osteogenesis when osteoblasts are already differentiated (mature osteoblasts) [39]. This study shows that MBG, BG, and β-TCP scaffolds stimulate osteogenesis and accelerated the process of mineralization without the need of osteogenic factors in the media. The control cells showed early osteoblast differentiation (associated with Collagen-I and Runx-2), but no mature osteoblasts (associated with osteocalcin) after 21 days of culture on expansion and osteogenic media. The results are in accordance with previous research with BG related materials [39,40]. An additional factor to consider is that MBG scaffolds after 7 days of cell culture started to disintegrate after swelling with DMEM. A possible reason is that the SA may have been incompletely cross-linked, causing destabilization of the scaffolds.

Despite the lower cell viability results obtained for BG 45S5 and MBG scaffolds compared to β-TCP in vitro, all three materials induced osteogenic differentiation. More studies are required in order to better evaluate BG since the faster degradation rates in an in vivo kinetic environment may yield better results compared to β-TCP than in vitro static studies.

5. Conclusion

BG 45S5, MBG, and β-TCP colloid gels were successfully 3D printed and characterized in terms of physico-chemical properties. The scaffolds were highly porous and maintained their architecture after sintering/lyophilizing. In vitro testing with human osteoprogenitor cells yielded higher cell viability after 7 days of culture for β-TCP scaffolds when compared to BG 45S5 and MBG. All materials expressed osteogenic markers Runx-2, osteocalcin and Collagen-I after 21 days of culture in osteogenic and expansion media. To obtain quantitative results on osteogenic differentiation between the materials, qPCR should be included in future studies, as well as on the in vivo behavior of MBG and BG 45S5 scaffolds.

Footnotes

Acknowledgements

The authors thank the National Agency for Research and Development (ANID), Scholarship Program, Magister Becas Chile, 2019 – 73200579. The Zeiss Gemini 300 FE-SEM was provided courtesy of the National Institutes of Health S 10 Shared Instrumentation Program, grant number 1S10OD026989-01.

Conflict of interest

The authors declare no conflict of interest.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

Ribas

R.G.

Schatkoski

V.M.

Montanheiro

T.L.d.A.

de Menezes

B.R.C.

Stegemann

Leite

D.M.G.

, Current advances in bone tissue engineering concerning ceramic and bioglass scaffolds: A review, Ceramics International 45(17) (2019), 21051–21061.

Wiese

and Pape

H.C.

, Bone defects caused by high-energy injuries, bone loss, infected nonunions, and nonunions, Orthop Clin North Am 41(1) (2010), 1–4, table of contents.

Wei

Huang

Zhu

Zhang

and Zhu

, 3D printing of mesoporous bioactive glass/silk fibroin composite scaffolds for bone tissue engineering, Mater Sci Eng C Mater Biol Appl 103 (2019), 109731.

Schemitsch

E.H.

, Size matters: Defining critical in bone defect size! J Orthop Trauma 31(Suppl 5) (2017), S20–S22.

Wang

and Yeung

K.W.K.

, Bone grafts and biomaterials substitutes for bone defect repair: A review, Bioact Mater 2(4) (2017), 224–247.

Lopez

C.D.

Witek

Torroni

Flores

R.L.

Demissie

D.B.

Young

, The role of 3D printing in treating craniomaxillofacial congenital anomalies, Birth Defects Res 110(13) (2018), 1055–1064.

Wang

M.M.

Flores

R.L.

Witek

Torroni

Ibrahim

Wang

, Dipyridamole-loaded 3D-printed bioceramic scaffolds stimulate pediatric bone regeneration in vivo without disruption of craniofacial growth through facial maturity, Sci Rep 9(1) (2019), 18439.

Tovar

Witek

Atria

Sobieraj

Bowers

Lopez

C.D.

, Form and functional repair of long bone using 3D-printed bioactive scaffolds, J Tissue Eng Regen Med 12(9) (2018), 1986–1999.

Varanasi

V.G.

Owyoung

J.B.

Saiz

Marshall

S.J.

Marshall

G.W.

and Loomer

P.M.

, The ionic products of bioactive glass particle dissolution enhance periodontal ligament fibroblast osteocalcin expression and enhance early mineralized tissue development, J Biomed Mater Res A 98(2) (2011), 177–184.

10.

Bari

Bloise

Fiorilli

Novajra

Vallet-Regi

Bruni

, Copper-containing mesoporous bioactive glass nanoparticles as multifunctional agent for bone regeneration, Acta Biomater 55 (2017), 493–504.

11.

LeGeros

R.Z.

, Calcium phosphate-based osteoinductive materials, Chem Rev 108(11) (2008), 4742–4753.

12.

Lopez

C.D.

Diaz-Siso

J.R.

Witek

Bekisz

J.M.

Cronstein

B.N.

Torroni

, Three dimensionally printed bioactive ceramic scaffold osseoconduction across critical-sized mandibular defects, J Surg Res 223 (2018), 115–122.

13.

Lehman

L.F.C.

de Noronha

M.S.

Diniz

I.M.A.

da Costa

E.S.R.M.F.

Andrade

A.L.

de Sousa Lima

L.F.

, Bioactive glass containing 90% SiO₂ in hard tissue engineering: An in vitro and in vivo characterization study, J Tissue Eng Regen Med 13(9) (2019), 1651–1663.

14.

Kowal

T.J.

Hahn

N.C.

Eider

Marzillier

J.Y.

Fodera

D.M.

Thamma

, New bioactive glass scaffolds with exceptional qualities for bone tissue regeneration: Response of osteoblasts and osteoclasts, Biomed Mater 13(2) (2018), 025005.

15.

El-Sayed

M.M.

Mostafa

A.A.

Gaafar

A.M.

El Hotaby

Hamzawy

E.M.

El-Okaily

M.S.

, In vitro kinetic investigations on the bioactivity and cytocompatibility of bioactive glasses prepared via melting and sol–gel techniques for bone regeneration applications, Biomed Mater 12(1) (2017), 015029.

16.

Luo

Cuniberti

Xiao

and Gelinsky

, Three-dimensional printing of hierarchical and tough mesoporous bioactive glass scaffolds with a controllable pore architecture, excellent mechanical strength and mineralization ability, Acta Biomater 7(6) (2011), 2644–2650.

17.

Nescakova

Zheng

Liverani

Nawaz

Galuskova

Kankova

, Multifunctional zinc ion doped sol–gel derived mesoporous bioactive glass nanoparticles for biomedical applications, Bioact Mater 4 (2019), 312–321.

18.

Wang

Zhang

Pang

Xiao

Jia

, Mesoporous bioactive glass-coated 3D printed borosilicate bioactive glass scaffolds for improving repair of bone defects, Int J Biol Sci 14(4) (2018), 471–484.

19.

Baino

, Bioactive glasses – when glass science and technology meet regenerative medicine, Ceramics International 44(13) (2018), 14953–14966.

20.

Pei

Zhu

Xin

Zhao

, Three dimensional printing of calcium sulfate and mesoporous bioactive glass scaffolds for improving bone regeneration in vitro and in vivo, Sci Rep 7 (2017), 42556.

21.

and Chang

, Mesoporous bioactive glasses: Structure characteristics, drug/growth factor delivery and bone regeneration application, Interface Focus 2(3) (2012), 292–306.

22.

Boccardi

Philippart

Melli

Altomare

De Nardo

Novajra

, Bioactivity and mechanical stability of 45S5 bioactive glass scaffolds based on natural marine sponges, Ann Biomed Eng 44(6) (2016), 1881–1893.

23.

Nommeots-Nomm

Labbaf

Devlin

Todd

Geng

Solanki

A.K.

, Highly degradable porous melt-derived bioactive glass foam scaffolds for bone regeneration, Acta Biomater 57 (2017), 449–461.

24.

Hoppe

Guldal

N.S.

and Boccaccini

A.R.

, A review of the biological response to ionic dissolution products from bioactive glasses and glass-ceramics, Biomaterials 32(11) (2011), 2757–2774.

25.

Pirayesh

and Nychka

J.A.

, Sol–gel synthesis of bioactive glass-ceramic 45S5 and its in vitro dissolution and mineralization behavior, J Am Ceram Soc 96(5) (2013), 1643–1650.

26.

Eqtesadi

Motealleh

Miranda

Lemos

Rebelo

and Ferreira

J.M.F.

, A simple recipe for direct writing complex 45S5 bioglasss 3D scaffolds, Materials Letters 93 (2013), 68–71.

27.

Zhu

Tian

Wei

and Zhu

, 3D printing of layered mesoporous bioactive glass/sodium alginate–sodium alginate scaffolds with controllable dual-drug release behaviors, Biomed Mater 14(6) (2019), 065011.

28.

Pei

Zhu

Zhao

and Zhu

, Three-dimensional printing of tricalcium silicate/mesoporous bioactive glass cement scaffolds for bone regeneration, J Mater Chem B 4(46) (2016), 7452–7463.

29.

Luo

Lode

and Gelinsky

, Hierarchical mesoporous bioactive glass/alginate composite scaffolds fabricated by three-dimensional plotting for bone tissue engineering, Biofabrication 5 (2013), 015005.

30.

Bauchle

Ludecke

Rabieh

Calnek

and Bromage

T.G.

, Quantification of 71 detected elements from Li to U for aqueous samples by simultaneous-inductively coupled plasma-mass spectrometry, Rsc Adv 8(65) (2018), 37008–37020.

31.

Xie

H.Y.

Deng

Yang

W.Z.

Liao

and Long

, Preparation, characterization and in vitro dissolution behavior of porous biphasic alpha/beta-tricalcium phosphate bioceramics, Mat Sci Eng C-Mater 59 (2016), 1007–1015.

32.

Vafa

Bazargan-Lari

and Bahrololoom

M.E.

, Synthesis of 45S5 bioactive glass-ceramic using the sol–gel method, catalyzed by low concentration acetic acid extracted from homemade vinegar, J Mater Res Technol 10 (2021), 1427–1436.

33.

Shah

A.T.

Ain

Chaudhry

A.A.

Khan

A.F.

Iqbal

Ahmad

, A study of the effect of precursors on physical and biological properties of mesoporous bioactive glass, Journal of Materials Science 50(4) (2015), 1794–1804.

34.

Ciraldo

F.E.

Boccardi

Melli

Westhauser

and Boccaccini

A.R.

, Tackling bioactive glass excessive in vitro bioreactivity: preconditioning approaches for cell culture tests, Acta Biomater 75 (2018), 3–10.

35.

Motealleh

Eqtesadi

Civantos

Pajares

and Miranda

, Robocast 45S5 bioglass scaffolds: in vitro behavior, Journal of Materials Science (52) (2017), 9179–9191.

36.

Cacciotti

Lombardi

Bianco

Ravaglioli

and Montanaro

, Sol–gel derived 45S5 bioglass: Synthesis, microstructural evolution and thermal behaviour, J Mater Sci Mater Med 23(8) (2012), 1849–1866.

37.

Herrmann

Hildebrand

Menzel

Fahy

Alini

Lang

, Phenotypic characterization of bone marrow mononuclear cells and derived stromal cell populations from human iliac crest, vertebral body and femoral head, International Journal of Molecular Sciences 20(14) (2019); 3454.

38.

Vigilante

Laddach

Moens

Meleckyte

Leha

Ghahramani

, Identifying extrinsic versus intrinsic drivers of variation in cell behavior in human iPSC lines from healthy donors, Cell Rep 26(8) (2019), 2078-+.

39.

Barrioni

B.R.

Norris

Naruphontjirakul

Jones

J.R.

and Pereira

M.M.

, Osteogenic potential of sol–gel bioactive glasses containing manganese, J Mater Sci Mater Med 30(7) (2019), 86.

40.

Naruphontjirakul

Porter

A.E.

and Jones

J.R.

, In vitro osteogenesis by intracellular uptake of strontium containing bioactive glass nanoparticles, Acta Biomater 66 (2018), 67–80.