Enhanced osteoprogenitor elongated collagen fiber matrix formation by bioactive glass ionic silicon dependent on Sp7 (osterix) transcription

Abstract

Bioactive glasses release ions, those enhance osteoblast collagen matrix synthesis and osteogenic marker expression during bone healing. Collagen matrix density and osteogenic marker expression depend on osteogenic transcription factors, (e.g., Osterix (OSX)). We hypothesize that enhanced expression and formation of collagen by Si⁴⁺ depends on enhanced expression of OSX transcription. Experimental bioactive glass (6P53-b) and commercial Bio-glass™ (45S5) were dissolved in basal medium to make glass conditioned medium (GCM). ICP-MS analysis was used to measure bioactive glass ion release rates. MC3T3-E1 cells were cultured for 20 days, and gene expression and extracellular matrix collagen formation was analyzed. In a separate study, siRNA was used to determine the effect of OSX knock-down on impacting the effect of Si⁴⁺ on osteogenic markers and matrix collagen formation. Each bioactive glass exhibited similar ion release rates for all ions, except Mg²⁺ released by 6P53-b. Gene expression results showed that GCM markedly enhanced many osteogenic markers, and 45S5 GCM showed higher levels of expression and collagen matrix fiber bundle density than 6P53-b GCM. Upon knockdown of OSX transcription, collagen type 5, alkaline phosphatase, and matrix density were not enhanced as compared to wild type cells. This study illustrates that the enhancement of elongated collagen fiber matrix formation by Si⁴⁺ depends on OSX transcription.

INTRODUCTION

Ti/Ti-alloy implants are becoming increasingly used for structural bone restoration in patients suffering from severe bone loss or injury. Upon implantation, the implant is used to mechanically support the surrounding bone. However, these materials have limited lifetimes because they do not facilitate biological mineral ingress.¹ Because of the poor attachment, the incidence of inflammation and fibrosis can increase.² This can cause eventual implant loosening and probable rejection, thus, a “bone-like” surface modification could improve implant-bony attachment.

Previous attempts to coat Ti to facilitate fewer rejections have not greatly improved their bony attachment. Two examples of such attempts include hydroxyapatite (HA)^3–7 and Bioglass™.⁸ HA coatings on Ti showed enhanced mineral attachment to the applied layer, however, catastrophic failure occurred at the Ti-mineral interface.^9–15 This failure was mainly due to the poor quality of HA formed by the plasma spray process. The mineral layer that results was highly soluble (22–62% amorphous calcium phosphate) and incorporated multiple calcium phosphate phases (in addition to the HA overlay).^16–23 Hench and colleagues²⁴ developed a glass material that releases calcium and phosphate ions to cells (as raw materials for osteoblast mineralization) while forming a bone-like HA surface for mineralized tissue attachment. However, coatings involving Bioglass resulted in delamination at the metal-glass/ceramic interface. This delamination was owed to cracking of the glass during the coating process and thermal expansion mismatch between Bioglass™ and Ti.²⁵

By modifying the composition of commercial Bioglass™, family of bioactive glasses (50–59 wt % SiO₂) that can be used to form a functionally graded interface between the implant and bone has been developed.²⁶ These materials form a thin and adherent Ti–Si bond between the glass and TiO₂ layer during an enameling process.^26–28 The bioactive features that define the bioactivity of the glass matrix are its ability to form a hydroxyapatite surface layer and the release of ions to cells for mitogenic activity.²⁹ These newer bioactive glasses could provide the same bioactive features as commercial Bioglass,³⁰ while also adhering well to Ti even during loading conditions. Therefore, the effect that these newer glasses have on osteogenesis will be explored here.

For long-term osteointegration, these materials must stimulate and support newly formed bone. Stimulation involves the up-regulation of collagenous, enzymatic, and matrix molecules that promote collagen and bone matrix development. Matrix enzymes, such as alkaline phosphatase, lysil oxidase, and matrix metaloproteineases, are expressed to convert the collagen matrix into a form amenable for matrix protein and mineral incorporation. Table I lists several of these markers involved in osteogenesis. More recent finding have shown that antioxidant metalloenzymes, such as superoxide dismutase (Sod1),⁴¹ are involved in the mechanisms of collagen cross-linking enzymes.⁴³ If these materials are to be a successful coating for Ti implants, they must stimulate new bone formation by stimulating the expression of these biomarkers and enhancing the density of collagenous matrices during osteogenesis.

TABLE I.

Genes Expressed by MC3T3-E1 Subclone 4 Cells During Differentiation

Gene	Name	Primary Function	Reference
Col1α1	Collagen type 1	Provides backbone of bone matrix	31
VEGFA	Vascular endothelial growth factor	Involved in “cross-talk” between osteoblasts and endothelial cells for vascular tissue formation during bone development	32
TGFβ1	Transforming growth factor	Involved in bone metabolism regulation	33
CDH11	Cadherin	Fundamental role for the development and maintenance of bone by mediating cellular crosstalk between osteogenic cells and by For Peer Review providing targets for the sorting and migration of osteogenic precursors toward the bone surface.	34
AKP2	Alkaline Phosphatase	Involved in turnover of collagen type 1 during bone matrix development	35
SMAD1	Smad1	Involved in intracellular signaling during bone development	36
SPP1	Osteopontin	Involved in regulating calcification during bone matrix development	37
MMP9	Matrix metalloprotease	Involved in regulation of growth plate maturation and bone formation	38
MMP2	Matrix metalloprotease	ECM enzyme involved in bone metabolism	39
LOX	Lysyl oxidase	Catalyzes enzymatic step for collagen- elastin cross-linking, which is required for bone formation.	40
DMP1	Dentin matrix Protein	Nuclear transcription factor involved in osteogenic gene expression	41

It has been speculated that Si⁴⁺ plays a role during collagen matrix synthesis. Collagens are critical for the proper formation of tissues and organs in our body. Type I collagen (Col(I)) is the fibrillar collagen comprising up to 90% of the collagens found in bone.⁴⁴ Type V collagen (Col(V)) is a member of the fibrillar collagen group and is present in tissues where Col(I) is present. Col(V) is involved in the control of collagen fibril dimension (diameter) and assembly of collagen matrix.⁴⁵ In our previous results^30,46 as well as reports by others^47,48 Si⁴⁺ (released as an ionic product €from bioactive glasses) was linked to enhanced Col(I) expression, collagen matrix formation, and mineralized tissue formation. Interestingly, collagen fiber bundles were found to be dense, elongated, and thick when osteoblast-like cells were exposed to Si⁴⁺.⁴⁹ Yet, determination of the mechanistic role that Si⁴⁺ has on osteogenesis is still needed to isolate a Si⁴⁺ dose that maximally up-regulates collagen and other osteogenic markers.

To better understand the impact of Si⁴⁺ on collagen synthesis, a closer examination of up-stream osteogenesis regulators is needed. One such upstream regulator, Osterix (OSX/SP7), is a zinc finger-containing transcription factor belonging to the Sp/KLF (Kruppel Like Factor) family of transcription factors involved in bone development.⁵⁰ It is distinctly expressed in osteoblasts and in all developing bones and is essential for the commitment of preosteoblastic cell differentiation into a mineralizing phenotype.⁵⁰ OSX triggers downstream expression of other collagenous (Col(V)) and noncollagenous (ALP) markers that indicate the mineralizing osteoblast phenotype.^51,52 Furthermore, it has been reported that OSX upregulates collagen matrix density during osteoblast differentiation.⁵³ Osteoblast differentiation has been shown to be arrested in OSX null mice, resulting in the absence of bone formation.⁵⁴ Overall, current studies indicate the indispensable role of OSX in osteogenesis. However, no study to date has determined if Si⁴⁺ enhances collagen matrix density via enhanced osteogenic regulators, such as OSX, and this will be studied here.

Thus, we will test the hypothesis that bioactive glasses release ions that enhance osteogenic marker expression and elongated collagen matrix density via Si⁴⁺ enhancement of osterix osteogenic transcription. The aims of this study are to (1) determine the effect of bioactive glass ionic products of dissolution on osteogenic marker expression and collagen matrix densification and (2) determine if such enhancement depends on Si⁴⁺ enhancement of osterix transcription factor expression. The goal of this study is to demonstrate that osterix gene expression is required for Si⁴⁺ enhancement of elongated collagen matrix density.

MATERIALS AND METHODS

Study design

Pertaining to the above aims, particles of bioactive glass (45S5 and 6P53-b) were dissolved in cell culture medium. The glass conditioned medium (GCM) were measured for ion concentrations using inductively coupled plasma mass spectrometry (ICP-MS). Each GCM and control medium was added to differentiating MC3T3-E1 osteoblasts. Cells were cultured for 20 days, and extracted for RNA to measure gene expression using quantitative polymerase chain reaction (qPCR). In parallel studies, cell layers were stained histologically for ECM collagen matrix formation. Statistical comparisons were made between each GCM and control treatment. For the second aim, OSX-null and wild type control osteoblasts were studied to elucidate the possible mechanistic effect of Si⁴⁺. The Osterix gene in MC3T3-E1 cells was inhibited using siRNA. Gene expression and histological staining were conducted as described above for OSX and collagen expression.

GCM preparation

Melt-derived experimental bioactive glass (6P53-b) and commercial Bioglass (45S5) were used in this study. The bioactive glass powder specimens were commercially purchased in a specific particle size range of 100–300 µm (Mo-Sci Specialty Products L.L.C., North Rolla, MO). The glass particles were soaked in alpha-minimum essential medium (α-MEM, Invitrogen, Carlsbad, CA for 72 h at 0.016 g mL⁻¹ in an incubator (37°C, 5% CO₂, 100% RH). These ion extracts were collected, sterile filtered (0.2 mm), and measured for pH changes using a standard pH meter while ion concentrations were measured using ICP-MS. Each ion extract was then supplemented (by volume) with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (pen-strep), 50 ppm ascorbic acid, and 10 mM glycerol-2-phosphate to make 45S5 and 6P53-b GCM.

ICP-MS

Ion extract samples were diluted 1:100 in 3% nitric acid. Ion concentrations were measured as described previously³⁰ and performed at the University of California ICP-MS Facility in Davis, CA using a 250 amu range ICP-MS machine (Agilent Technologies, Palo Alto, CA).

Cell culture

Prior to experiments, cells (MC3T3-E1 subclone 4, ATCC, Manassas, VA) were grown in control media and in 150 cm² flasks to obtain the desired amount of cells for experiments. Cells were expanded to passage 25–30 with media changes every 2–3 days. Cells were pelleted, counted (using a standard hemacytometer and an optical microscope (Nikon TE300, Nikon Inc., Tokyo, Japan)), and seeded (50,000 cells cm⁻²) into six-well plates. The cells were allowed to attach and then synchronized (using α-MEM, 1% FBS, 1% pen-strep) for 48 h.

Experiments were initiated by exchanging synchronization media with GCM and control treatments. All treatments were further supplemented with AA (50 ppm) and β-GP (10 mM) to induce differentiation. A parallel experiment was conducted in which cells were cultured on glass cover slips (not bioactive glasses) for histological staining. Several time points were used over the course of osteoblast differentiation (~30 days).

qRT-PCR

In quantifying levels of gene expression, qPCR was used. Cells cultured in each treatment were lysed at the time points specified above to collect mRNA (RNeasy Mini Kit, Qiagen, Valencia, CA) and converted to cDNA using reverse transcriptase (Reverse Transcription System, Promega, Madison, WI). A full-spectrum UV–vis nanodrop volumetric analyzer (ND-1000, Nanodrop Technologies, Wilmington, DE) was used to obtain the absorbance (A) measurements of mRNA and cDNA samples. The mRNA and cDNA concentrations were measured at 260 nm (A₂₆₀) and a nucleic acid absorbance at A₂₆₀/A₂₈₀ (1.8–2.1) was used to determine quality of samples (advised by manufacture’s protocol). cDNA samples were all diluted to a concentration of 100 ng µL⁻¹ prior to PCR amplification.

After conversion, cDNA samples were prepared for amplification with the following reagents in these specified concentrations (final concentration): cDNA sample (10% by volume), FastStart Taqman Master Mix (Rox, 2×, 50%) (Roche Applied Sciences, Mannheim, Germany), forward primer (900 nM), reverse primer (900 nM), and hydrolysis probe (250 nM), and PCR grade water (Amgen, South San Francisco, CA). All reaction volumes did not exceed 10 µL. Reactions were executed using a real-time PCR machine (ABI7900, Applied Biosystems, Foster City, CA).

PCR amplification was evaluated using relative quantification by comparison to an internal housekeeping gene (glyceraldehyde 3-phosphate dehydrogenase or GAPDH), which was not affected by treatment. PCR amplification curves were collected using the SDS thermal cycler software package and regressed for their threshold cycle (C_T) using the sigmoidal curve fitting method⁵⁵ [SigmaPlot software package (Systat Software, San Jose, CA)]. Relative quantification of gene expression was evaluated using the ΔΔC_T method (as described previously.⁶¹ Serial dilutions of cDNA samples and sigmoidal curve fitting analysis were used to confirm amplification efficiency.

Collagen visualization

Cells-on-glass cover slip samples were removed from culture after 6 and 20 days. Samples were moved to fresh six-well plate, washed twice in phosphate buffered saline (PBS), fixed with Bouin’s solution (Richard-Allen Scientific, Kalama-zoo, MI) under ambient conditions for 1 h and rinsed with deionized water to remove excess fixative. Bouin’s fixative enhances binding of Picrosirius stain to collagen fiber bundles. Fastgreen stain (0.1%, Sigma) was then used to stain cell layers for 30 min followed by a treatment of glacial acetic acid (American Master Tech Scientific, Lodi, CA) for 30 min to remove excess staining. Fastgreen stains cell layers aniline blue or green with darker color intensity indicating increased tissue density. The Picrosirius stain (American Master Tech Scientific Inc.) was then applied to cell layers for 1 h, followed by a wash with deionized water. A sequential alcohol dehydration (70–100% ethanol) for 30 s followed before imaging. The collagen fiber bundles stained by Picrosirius staining were imaged using polarized light microscopy (due to the birefringent nature of collagen, which appears yellow, orange, or red). An optical microscope (Olympus BX51, Tokyo, Japan) equipped with a CCD camera and Image Pro software was used to capture images of the cell layers under polarized light.

Silicon ion treatments

To determine the influence of ionic silicon on osteoblast function, treatments were made such that the ion concentrations used represented the lowest dissolved ion concentration from ICP-MS results. The source of ionic Si was sodium meta-silicate (Sigma, St. Louis, MO). Additional supplementation of treatments with ascorbic acid and glycerol-2-phosphate were used to make treatments for osteoblasts. Control media was used as described above.

OSX siRNA transfection

The siRNA cocktails targeting mouse Sp7/Osterix were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). MC3T3-E1 subclone 4 cells were synchronized with αMEM, 1% FBS, and 1% pen-strep for 48 h. Cells were then transfected using Lipofectamine (Invitrogen) according to the manufacturer’s protocol with the transfection agent concentration scaled proportionately to the cell number. After 5–7 h of transfection at 37°C in a CO₂ incubator with growth media (no pen-strep), cells were exposed to 24 h of growth media followed by their respective treatments. Cells were lysed 72 h after the treatments were introduced and analyzed for gene expression using qRT-PCR. Histological analysis for collagen fiber bundle formation was performed 6 days after treatment.

Statistics

Statistical comparison of gene expression results was analyzed using GraphPad Prism 4. Two-way analysis of variance (ANOVA) followed by Bonferroni testing was used for an overall comparison of the independent factors, time and treatment, with the dependent factor, gene expression. For individual and combined ion studies, one-way ANOVA and Tukey’s analysis were used for experimental and control comparisons. Statistical comparisons shown are relative to corresponding control group. A power value of p < 0.05 was used to report statistical significance. All data analyses were calculated from triplicate sampling per experiment.

RESULTS

Experimental bioactive coating glass ion release kinetics

A generalized kinetic model was used to determine the mass transfer coefficient as it related to bioactive glass ion release. The generalized model (Sepulveda et al.⁵⁶) describes the time-dependent release of ions in vitro by,

$$\frac{d{c}_i}{dt}=k{t}^{-1/2}$$ (1)

where c_i is the ion concentration (ppm or mg L⁻¹), k_i is the overall mass transfer coefficient (ppm hr ^−½or mg L ^{−1 hr −½}), and t is time (hours). The overall mass transfer coefficient, k, depends on many factors including the diffusivity, the glass surface reaction rate constants associated with ion formation, the reaction rate constants involved in the reprecipitation of hydroxyapatite, the degree of super-saturation, the boundary layer thickness, the initial ion concentration, and the diffusion path length.

The ion release from the glass particles follows a diffusion-controlled behavior. At 24–48 h, the release of calcium and phosphate ions reaches an apex [6P53-b: Fig. 1(A,B); 45S5: Fig. 1(C,D)]. The release of Si⁴⁺ and Mg²⁺ did not reach an apex in ion concentration. The Mg ion release rate had a t^−½ dependence, giving mass transfer coefficients [according to Eq. (1)] of 1.6 ± 0.1 mg L⁻¹ hr^−½. Similarly, Si ion release had an estimated mass transfer coefficient of 4.5 ± 0.2 mg L⁻¹ hr^−½ (45S5) and 5.0 ±€ 0.1 mg L⁻¹ hr^−½ (6P53-b). In addition, pH measurements showed that the OH⁻ concentration increased with time, yielding mass transfer coefficients of 6.4 ± 0.9 × 10 ⁻⁷ moles OH⁻ L⁻¹ h ^−½ (6P53-b) and 8.2 ± 0.9 × 10 ⁻⁷ moles OH⁻ L⁻¹ h ^−½ (45S5).

FIGURE 1.

Measurement of ion concentrations for bioactive glass dissolution. (A) Ion concentration and (B) pH measurements [release rate: 6.4 ± 0.9 × 10⁻⁷ moles OH– L⁻¹ h^−½.]] for calcium, total phosphate, silicon, and magnesium released from 6P53-b particulates in α-MEM. (C) Ion concentration and (D) pH measurements [(8.2 ± 0.4 × 10⁻⁷ moles OH– L⁻¹ h^2−½.)] for calcium, total phosphate, silicon, and magnesium released from 45S5 particulates in α-MEM. (E) Calculations of the degree of saturation (DS) for 45S5 and 6P53-b ion concentrations and (F) 45S5 and 6P53-b GCM and control GCM during in vitro cell culture studies (concentrations given in A and C). For DS 5 14, the in vitro environment is considered super-saturated and the migration of calcium, phosphate, and hydroxyl ions from the liquid phase to the surface is likely. At DS = 12, the solution is supersaturated, however, heterogeneous nucleation is more likely to occur. For DS < 0, the solution is considered stable and these ions remain soluble.

This apex in calcium and phosphate ion concentration caused an increase in the degree of saturation [Fig.1(E,F)]. The increase in these ion concentrations resulted in the degree of saturation, which indicated the favorability of homogeneously or heterogeneously nucleated hydroxyapatite formation.^57,58 As the dissolved ion concentration increased, the degree of saturation also increased.

Effect of GCM treatments on osteoblast-specific gene expression

Growth factors.

Both 45S5 and 6P53-b GCM enhanced growth factor gene expression. For example, the 45S5 GCM markedly enhanced the gene expression of VEGFA in MC3T3-E1 subclone 4 at day 2 (5× control) and day 6 [2× control, Fig. 2(A)]. The gene expression of TGFb1 [Fig. 2(B), 2.5× control] was significantly enhanced in these cells by the 6P53-b GCM treatment at day 6. TGFβ1 involved in the regulation of collagenous and non-collagenous markers during osteogenic differentiation.⁵⁹

FIGURE 2.

Gene expression of VEGFA and TGFb(I) Effect of 45S5 and 6P53-b GCM on MC3T3-E1 subclone 4 gene expression of (A) vascular endothelial growth factor (VEGFA), and (B) transforming growth factor (TGFb(I)) relative to control treatments. Data presented represent the means ± S.D. of triplicate experiments. Significant differences between treatments were determined by One-Way ANOVA with Dunnett posttest comparison. *p < 0.05, **p < 0.01, ***p < 0.005 compared to control.

Matrix proteins.

The 45S5 GCM enhanced the gene expression levels of CDH11 [Fig. 3(A)], SPP1 [Fig. 3(B), and DMP1 Fig. 3(C)] in MC3T3-E1 subclone 4 cells. Gene expression for CDH11 was four times greater on day 6 and five times greater on day 20 relative to the corresponding controls [Fig. 3(A)]. Similarly, SPP1 gene expression was moderately enhanced on days 6 (2.5× control) and 20 [3.5× control; Fig. 3(B)]. While 45S5 GCM treatment had an impact on the matrix protein DMP1 by 2 fold (day 6), 6P53-b showed a marked enhancement day 20 [16× control, Fig. 3(C)].

FIGURE 3.

Gene expression of CDH11, SPP1, and DMP1 Effect of 45s5 and 6P53-b GCM on MC3T3-E1 subclone 4 gene expression of (A) cadherin 11 (CDH11), (B) sialo phosphoprotein (SPP1), and (C) dentin matrix protein (DMP1) relative to control treatments. Data presented represent the means ± S.D. of triplicate experiments. Significant differences between treatments were determined by One-Way ANOVA with Dunnett posttest comparison. *p < 0.05, **p < 0.01, ***p < 0.005 compared to control.

Matrix protease.

MMP2 and MMP9 were utilized to study the effect of bioglass ion release on matrix proteinases. The data from MMP2 gene expression showed slight significant differences by both 45S5 and 6P53-b GCM from the control and each other on day 6 [2x, Fig. 4(A)]. Statistical differences were not observed afterward. Conversely, not only were the mRNA levels of MMP9 enhanced by the 45S5 GCM treatment on day 2 (2.5×), but they were markedly upregulated on day 10 by ~25-fold, with a greater enhancement on day 20 (45×) [Fig. 4(B)].

FIGURE 4.

Gene expression of MMP(II) and MMP(IX) Effect of 45s5 and 6P53-b GCM on MC3T3-E1 subclone 4 gene expression of (A) metalomatrix protein II (MMP(II), and (B) metalomatrix protein IX (MMP(IX)) relative to control treatments. Data presented represent the means ± S.D. of triplicate experiments. Significant differences between treatments were determined by One-Way ANOVA with Dunnett posttest comparison. *p < 0.05, **p < 0.01, ***p < 0.005 compared to control.

Collagen gene expression.

GCM-treated MC3T3-E1 subclone 4 cells showed increased collagen expression [Col(I)α 1 Fig. 5(A)], Col(I)α2 [Fig. 5(B)]. Compared to control treated cells, Col(I)α1 was significantly enhanced at day 10 for 45S5 GCM treated cells (3× control) and 6P53-b GCM treated cells (1.2× control). For Col(I)α2, its expression was initially enhanced on day 2 (2.0× control) and maximally enhanced in 45S5 GCM-treated cells at 14 days (3.5× control) and at 20 days in 6P53-b GCM treated cells (1.5× control). In addition, continued enhanced collagen expression was observed throughout the time course of the experiment for both GCM treatments. Although not observed in this work, enhancement of Col(I) expression was reported as early as 1 day during differentiation.^30,39

FIGURE 5.

Gene expression of type I collagen effect of 45S5 and 6P53-b GCM treatments on MC3T3-E1 subclone 4 collagen type I collagen expression (Col(I)α1 (A), Col(I)α2 (B)) relative to control treatments. Data presented represent the means ± S.D. of triplicate experiments. Significant differences between treatments were determined by One-Way ANOVA with Dunnett posttest comparison. *p < 0.05, **p < 0.01, ***p < 0.005 compared to control.

Metalloenzyme expression.

Non-collagenous genes including ALP, LOX, and Sod1 were enhanced in GCM-treated MC3T3-E1 subclone 4 cells. Sod1 expression [Fig. 6(C)] was enhanced in both GCM treatments at day 2 (6P53-b GCM: 2.0× control, 45S5 GCM: 3.5× control). Sod1 is a regulator of collagen cross-linking through the regulation of LOX.⁴³

FIGURE 6.

Gene expression of LOX, ALP, and Sod1 Effect of 45S5 and 6P53-b GCM treatments on MC3T3-E1 subclone 4 expression of (A) lysil oxidase (LOX), (B) alkaline phosphatase (ALP), and (C) superoxide dismutase (Sod1) relative to control treatments. Data presented represent the means ± S.D. of triplicate experiments. Significant differences between treatments were determined by one-way ANOVA with Dunnett posttest comparison. *p < 0.05, **p < 0.01, ***p < 0.005 compared to control.

LOX expression [Fig. 6(A)] was also enhanced on day 2 (37× control) and day 20 (25× control) in 45S5 treated cells. For 6P53-b GCM, LOX enhancement occurred after 20 days (2.5× control). Lysyl oxidase is responsible for the cross-linking of collagen fibrils to increase their length.⁶⁰

Alkaline phosphatase, a dephosphorylating enzyme involved in collagen turnover, plays a vital role in the formation of hydroxyapatite and mineralized tissues.³⁵ Concerning its expression [Fig. 6(B)], 45S5 GCM treatments significantly enhanced its expression on day 2 (3× control), and onward to day 6 (5× control), while 6P53b GCM treated cells had maximal upregulation of ALP on day 6 (3.5× control).

Extracellular matrix collagen formation.

The enhanced gene expression by GCM treatments led to enhanced collagen matrix formation. 45S5 GCM treatments enhanced MC3T3-E1 subclone 4 collagen formation (Fig. 7), which was imaged using polarized light microscopy. 45S5 GCM treatments [Fig. 7(C,F)] exhibited a higher density of orthogonally arranged, elongated collagen fiber bundles as compared to 6P53-b GCM treatments [Fig. 7(B,E)] or control treatments [Fig. 7(A,D)].

FIGURE 7.

Collagen visualization after 6 days (A–C) and 20 days (D–F) culture Optical micrographs displaying extracellular matrix collagen as a result of MC3T3-E1 subclone 4 exposure to (A), (D) control media + AA + βGP treatment, 6P53-b GCM treatment, (B), (E) and 45S5 GCM treatment, (C), (F) over a 6-day and 20-day period, respectively. White and black arrows indicate elongated collagen fibers aligned in orthogonally oriented layers.

Si⁴⁺ enhances ALP and col(V) expression and collagen synthesis via OSX enhancement

The cells were analyzed for the gene expression of OSX, its down-stream biomarkers (ALP and Col(V)α3), and collagen matrix formation in WT osteoblasts. The results from the gene expression studies of OSX [Fig. 8(A)] showed an increase in OSX mRNA by Si⁴⁺ treatments vs. control treatments in. Si⁴⁺ treatments also significantly increased the gene expression of ALP [Fig. 8(B)] and Col(V)α3 [Fig. 8(C)] vs. control treatments. Collagen staining after 6 days post 100 µM silicon treatment [Fig. 8(D)] showed more dense and elongated collagen fibers relative to control [Fig. 8(E)]. These results indicate that Si⁴⁺ enhances OSX expression, its downstream markers ALP and Col(V), and density of elongated collagen fiber bundles in ECM collagen matrices.

FIGURE 8.

Results of 100 µM Si additive for OSX, ALP, and Col(V)a3 expression and collagen formation. Relative gene expression of OSX, ALP, and Col(V)α3 by MC3T3-E1 subclone 4 cells treated with control media and 100 µM Si for 72 h. Optical micrograph showing extracellular collagen fiber as result of MC3T3-E1 subclone 4 exposure to basal media (D) and 100 µM silicon after 6 days (E). *** indicates p < 0.005.

OSX deficiency limits Si⁴⁺ enhancement of ALP, col(V), and collagen synthesis

Knock-down studies were conducted using siRNA of OSX/Sp7. MC-3T3 cells were treated with control medium and 100 µM Si⁴⁺ in OSX-null osteoblasts for 3 days. OSX knock-down was first confirmed for both treatments in which siRNA induced a >80% decrease in gene expression of OSX [Fig. 9(A)], OSX-null osteoblasts also showed a significant decreases (≥75%) in mRNA levels of ALP in both treatments [Fig. 9(B)]. Interestingly, although Col(V)α3 mRNA was produced in statistically lower amounts than control treated WT osteoblasts, Col(V)α3 expression was produced in significantly greater amounts in Si⁴⁺ treated OSX-null cells vs. control treated OSX-null cells [Fig. 9(C)].

FIGURE 9.

OSX knock-down results for gene expression and collagen formation Relative gene expression of OSX, ALP, and Col(V)α3 by MC3T3-E1 subclone 4 cells treated with control media, OSX siRNA, and 100 µM Si + OSX siRNA for 72 h. Optical micrograph showing extracellular collagen fiber as result of MC3T3-E1 subclone 4 after day 6 post-treatment. MC3T3-E1 subclone 4 cells were exposed to (D) 100 µM silicon + OSX siRNA or (E) OSX siRNA. The presence of fibrillar collagen is seen as the greenish yellow streaks.

An analogous trend was also observed in ECM collagen matrix formation. Both control-treated and Si^4+-treated OSX-null cells had a lower density of elongated collagen fiber bundles in ECM as compared to these same treatments in WT osteoblasts. However, the addition of silicon ions to OSX siRNA-treated cells resulted in a greater density of fiber bundles [Fig. 9(D)] compared to the sparse amounts of short extracellular collagen in cells treated with OSX siRNA-alone [Fig. 9(E)]. Overall, these results indicate the dependence of Si⁴⁺ on OSX expression for the enhancement of ALP and Col(V) expression and collagen matrix formation in osteoblasts.

DISCUSSION

In this study, it was found that the two bioactive glasses studied here (45S5, 6P53-b) enhanced collagen formation during osteogenesis. Several families of growth factors, transcription factors, and enzymes associated with osteogenesis were enhanced. Matrix protein expression of osteocalcin was also enhanced and showed a dependence on glass dosage. Both glasses exhibited similar dissolution behavior (Si⁴⁺, Ca²⁺, PO₄³⁻) with the exception that 6P53-b released additional Mg²⁺.

The equilibrium reaction for HA formation is:

$$C{a}_{10}{(P{O}_4)}_6{(OH)}_{2(s)}=10C{a}_{(aq)}^{2\pm }\pm 6P{O}_{4(aq)}^{3-}\pm 2O{H}_{(aq)}^{-}$$ (2)

As the concentrations of these ions increased during glass dissolution (24–48 h), the likelihood of HA formation increased [Fig. 1(E)], which was also observed by Orme et al.⁶¹ Because the likelihood of HA formation was high, this caused the depletion in calcium and phosphate ion concentrations between 48 and 72 h of dissolution. Sepulveda et al.⁵⁶ observed a similar trend for 45S5 dissolution after 24 h.

According to the equilibrium constant [Eq. (3)] for the reaction in Eq. (2),

$$K_{\text{SP}}=C_{C{a}^{2\pm }}^{10}\cdot C_{P{O}_4^{3-}}^6\cdot C_{O{H}^{-}}^2\cdot {\text{γ}}_{C{a}^{2\pm }}^{10}\cdot {\text{γ}}_{P{O}_4^{3-}}^6\cdot {\text{γ}}_{O{H}^{-}}^2,$$ (3)

the ion activities are couple to the ion concentrations that result in a shift in the equilibrium reaction toward HA formation. These ion activities can be described by the well-known Pitzer relation that can be applied to all ions dissolved in the media. This model accounts for short and long range interactions for ions of varying size and charge. This model is valid for media (blood plasma, cell culture media, simulated body fluid) where the ionic strength has been well-documented to be above 0.1 mol kg⁻¹ and is considered a saturated solution as shown in Figure 1(E,F). At this ionic strength, the extended Debye–Huckel relation is not valid because of increasing short and long-range interactions with increasing ionic strength.

Si⁴⁺ release was slower than Ca²⁺ and PO₄³⁻ release, while Mg²⁺ release was even more tepid. These results confirmed the work of Saiz et al. where it was found that Si⁴⁺ release was approximately an order of magnitude higher than that of Mg²⁺ release.¹ The tepid Mg²⁺ release rate could be owed to (1) increased tendency of chelation within the glass structure owed to the Ca²⁺ and Mg²⁺ presence based on their field electron strength, and (2) the ability of these dopants to slow water molecule ingress through the glass matrix, thus slowing the formation of a hydration layer and improving glass durability.^62,63

In general, 45S5 GCM treatments enhanced the expression of several osteogenic markers. The enhanced markers included collagens (Col1α1, Col1α2), matrix proteins (DMP1, SPP1, CDH11), growth factors (VEGFA, TGFβ1), and matrix enzymes (LOX, ALP, MMP2, MMP9). This up-regulated effect was probably owed to the presence of Si⁴⁺ and Ca²⁺. It is well-known that calcium receptors on cells stimulate many cellular functions including osteogenesis. Thus, the effect of Ca²⁺ on osteogenesis probably acts through this receptor.

In contrast, 6P53-b GCM induced more modest enhancements in the expression of these same osteogenic markers. These modest enhancements were probably owed to the down-regulatory effect of Mg²⁺ in 6P53-b GCM. In this work, it was found that the difference in the ions released by each glass was the release of Mg²⁺ from 6P53-b while all other ion constituencies and concentrations were approximately the same. Moreover, it was found that Mg²⁺ individually down-regulates alkaline phosphatase expression. When Mg²⁺ was combined with Si⁴⁺ or Ca²⁺, it was seen to limit their enhanced effects, which probably explains a similar trend when these ions are released together from 6P53-b. Interestingly, 6P53-b GCM did induce a slight up-regulation of Sod1, this did not result in enhanced expression of LOX and collagen matrix fiber bundle density. Thus, Mg²⁺ likely acts as a negative regulator of osteogenic marker expression when released from the bioactive glass matrix.

Yet, the role of Si⁴⁺ in osteogenesis has not been determined. On the basis of our results, 45S5 GCM had increased density of elongated collagen fiber bundles after 6 days as compared to 6P53-b GCM and control treated osteoblasts. In this work, we isolated the effect of Si⁴⁺ on osteogenic marker expression involved in ECM collagen formation. When Si⁴⁺ was added to wild type osteoblasts, it enhanced the expression of osteogenic markers osterix, alkaline phosphatase, and collagen type V. Moreover, we found that the density of collagen matrices was markedly more dense than control treated cells. Because collagen matrices cannot form and alkaline phosphatase and Col(V) depend on osterix expression (as was rationalized above and confirmed in this work), it was hypothesized that Si⁴⁺ may enhance collagen matrix formation via the osterix pathway.

Upon knockdown of Sp7 (osterix), we found that alkaline phosphatase, and Col(V) expressions were markedly decreased in both control and Si⁴⁺ treated cells. This contributed to a partial reduction in elongated collagen fiber density. These results indicate that enhanced alkaline phosphatase expression, Col(V) expression, and density of elongated collagen fibers by Si⁴⁺ depends on enhanced osterix expression. Despite a reduction in the density of elongated collagen fibers, fiber elongation may be dependent on other transcription factors. Other markers that may be present such as Sp1, Runx2, ATF4, or other transcription factors that play a role in osteogenesis may also be involved in collagen fiber elongation. Our future work will focus on resolving this mechanism in full to isolate the specific transcription factor dependence of Si⁴⁺ enhancement of collagen matrix formation.

As an implant coating material, these MgO-doped bioactive glasses may have more specific applications. It may be required that these glasses have a reduced MgO content for their use in bone regeneration. Still, the release of Mg²⁺ from their matrices has some beneficial effects. Mg²⁺ is a regulator of many osteogenic markers involved in bone regeneration, thus, implicating it as an ion more involved in sorbtive properties of bone matrix. In fact, tight regulation of the Ca²⁺/Mg²⁺ ratio greatly impacts bone regeneration. Thus, as an implant coating, genetic design of the bioactive glasses is important for not only forming new bone around the implant, but enhancing remodeling processes that further strengthen the resultant apposed bone matrix.

CONCLUSIONS

Pertinent to the aims of this study, it was found that bioactive glasses release ions that enhance osteoblast differentiation. These materials control the release of Si and Ca, which were found to enhance collagen gene expression and matrix formation. The enhanced formation of densely packed collagen fiber bundles was attributed to the enhanced expression of regulators of osteogenic collagen fiber crosslinking, elongation, and dephosphorylation. It was also found that Mg release from bioactive glass dampens these enhanced effects by Si and Ca.