Available online at www.sciencedirect.com

Acta Biomaterialia 5 (2009) 3536–3547

www.elsevier.com/locate/actabiomat

Enhanced osteocalcin expression by osteoblast-like cells(MC3T3-E1) exposed to bioactive coating glass

(SiO2–CaO–P2O5–MgO–K2O–Na2O system) ions

V.G. Varanasi a,*, E. Saiz b, P.M. Loomer a, B. Ancheta a, N. Uritani a, S.P. Ho a,b,A.P. Tomsia b, S.J. Marshall a, G.W. Marshall a

a Division of Biomaterials and Bioengineering, University of California, San Francisco, CA 94143–0758, USAb Material Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA

Received 5 August 2008; received in revised form 15 April 2009; accepted 21 May 2009Available online 2 June 2009

Abstract

This study tested the hypothesis that bioactive coating glass (SiO2–CaO–P2O5–MgO–K2O–Na2O system), used for implant coatings,enhanced the induction of collagen type 1 synthesis and in turn enhanced the expression of downstream markers alkaline phosphatase,Runx2 and osteocalcin during osteoblast differentiation. The ions from experimental bioactive glass (6P53-b) and commercial BioglassTM

(45S5) were added to osteoblast-like MC3T3-E1 subclone 4 cultures as a supplemented ion extract (glass conditioned medium (GCM)).Ion extracts contained significantly higher concentrations of Si and Ca (Si, 47.9 ± 10.4 ppm; Ca, 69.8 ± 14.0 for 45S5; Si,33.4 ± 3.8 ppm; Ca, 57.1 ± 2.8 ppm for 6P53-b) compared with the control extract (Si < 0.1 ppm, Ca 49.0 ppm in a-MEM) (ANOVA,p < 0.05). Cell proliferation rate was enhanced (1.5� control) within the first 3 days after adding 45S5 and 6P53-b GCM. MC3T3-E1subclone 4 cultures were then studied for their response to the addition of test media (GCM and control medium along with ascorbicacid (AA; 50 ppm)). Each GCM + AA treatment enhanced collagen type 1 synthesis as observed in both gene expression results (day1, Col1a1, 45S5 GCM + AA: 3� control + AA; 6P53-b GCM + AA: 4� control + AA; day 5, Col1a2, 45S5 GCM + AA: 3.15� con-trol + AA; 6P53-b GCM + AA: 2.35� control + AA) and in histological studies (Picrosirius stain) throughout the time course of earlydifferentiation. Continued addition of each GCM and AA treatment led to enhanced expression of alkaline phosphatase (1.4� con-trol + AA after 5 days, 2� control + AA after 10 days), Runx2 (2� control + AA after 7 days) and osteocalcin gene (day 3, 45S5GCM + AA: 14� control + AA; day 5, 6P53-b GCM + AA: 19� control + AA) and protein expression (40�–70� control + AA after6 days). These results indicated the enhanced effect of bioactive glass ions on key osteogenic markers important for the bone healingprocess.� 2009 Published by Elsevier Ltd. on behalf of Acta Materialia Inc.

Keywords: Bioactive glass ions; Osteogenesis; Osteoblasts; Silicon; Calcium

1. Introduction

Implants are used in dental applications to replace miss-ing teeth, or in orthopedic and craniofacial applications toreplace lost bone. These implants must restore the physio-logical structure and function while also facilitating com-

1742-7061/$ - see front matter � 2009 Published by Elsevier Ltd. on behalf o

doi:10.1016/j.actbio.2009.05.035

* Corresponding author. Tel.: +1 415 476 0917; fax: +1 415 476 0858.E-mail address: venu.varanasi@ucsf.edu (V.G. Varanasi).

plete bone apposition [1]. Currently, Ti implants are usedsuccessfully for tooth replacement in mandibular or maxil-lary bone (86% and 76%, respectively [2]) and nearly 93%for bone replacement in the cranium [3]. However, fasterosteointegration and improved implant–bone attachmentare desired to improve implant success rates and longevity.

Various coating technologies to improve the implant–bone interface have been attempted. Hydroxyapatite(HA) was used for a number of years. However, delamina-tion of HA occurred at the implant–ceramic interface [4] or

f Acta Materialia Inc.

V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547 3537

bone–ceramic interface [5]. Commercial Bioglass [6,7] iswidely known for its benefits in various bone substitutes[8] and periodontal procedures [9], however, it is difficultto use as an implant coating because it cracks at the Ti–glass interface when cyclically loaded [10]. Both HA andBioglass cracking is primarily due to a large thermal expan-sion mismatch with Ti [10].

Improvement in glass coating technology led to thedevelopment of a family of bioactive glasses (50–59 wt.%SiO2) to enhance the osteointegration potential of Ti [10–17] or as polymer–bioactive glass composite scaffolds forhard and soft tissue regeneration [18]. By doping the glasswith additional constituents and partial substitution ofCaO with MgO and Na2O with K2O (Table 1, 6P53-b vs.45S5), bioactive coating glass had improved adhesion toTi alloy during the coating process. In previous in vitro

studies, these glasses facilitated direct mineralized tissueattachment (compared with mechanical attachment of min-eralized tissue to Ti surfaces [16]). The bioactive glass coat-ing formed a HA surface layer which facilitated the directbond to bone [16,19].

In general, the surface of the glass material and its corro-sion behavior in the physiological environment influence itsapposition to bone. Another study [20] attempted to deter-mine the separate impact of the initially fabricated bioactiveglass surface and the initial dissolution of ions from thatsurface. The goal was to determine whether additional sur-face treatment was required to promote a favorable osteo-blast response. It was found that the initial in vitro

dissolution of the glass was rapid, which increased the pHof the in vitro environment to which osteoblasts wereexposed. This increased pH appeared to decrease theamount of cell proliferation and alkaline phosphatase activ-ity [20]. For these reasons, the glass materials were pre-soaked (or conditioned) in simulated body fluid (SBF) fora period of 10 days. This conditioning period (1) stabilizedthe in vitro pH to near physiological (7.0–7.4) [20], (2)reduced exposure of osteoblasts to contaminants from thefabrication process [21], and (3) induced the formation ofa HA surface layer for mineralized tissue attachment. Theenhanced marker expression observed by Foppiano et al.[22] in fact occurred when the bioactive glass was condi-tioned in vitro. This study focuses on the corrosion of ionicproducts (isolated from the bioactive glass surface and afterthis initial conditioning period) to discover whether theyalso influence the osteoblasts’ cell cycle.

Interestingly, They have suggested that these ions mayplay an active role in osteoblast behavior. They have sug-gested that these ions may alter the expression of osteoblastdifferentiation markers associated with bone matrix forma-

Table 1Composition (wt.%) of BioglassTM (45S5) and experimental bioactiveglass (6P53-b).

SiO2 Na2O K2O MgO CaO P2O5

6P53-b (LBL) 52.7 10.3 2.8 10.2 18.0 6.045S5 (Mo-Sci) 45.0 24.5 24.5 6.0

tion [23–32], while other studies showed that individualions may alter osteoblast function [33–37]. For example,commercial Bioglass has been shown to influence Runx2and osteocalcin expression [38], which are expressed withinthe first 10 days of the differentiation compartment of theosteoblast cell cycle. Furthermore, Foppiano et al. [22]found that these dissolution products promoted the up-reg-ulation of Runx2. Yet, no significant effort has beenattempted to connect the enhanced gene expression (within7 days) as it impacts downstream matrix protein expression(within 10 days) by osteoblasts under the influence of thesebioactive coating glass corrosion products. To determinethe effect of these ions on osteoblasts, a materials extractwas isolated as a dissolved product from the bioactive glasssurface.

This study tests the hypothesis that bioactive glass ionsenhance osteoblast differentiation. The aims of this studyare (1) to demonstrate that specific osteoblast markersare enhanced in the presence of bioactive coating glass(6P53-b) ions; (2) to demonstrate that similar osteoblastbehavior is observed for commercially available Bio-glassTM (45S5) ions; (3) to determine a corresponding effecton gene, protein and extracellular matrix expression ofosteogenic biomarkers.

To study the effect on osteoblasts of ionic products of bio-active glass dissolution, a well-characterized cell line,MC3T3-E1 subclone 4, was used. This cell line mimics osteo-blast progenitors in that it expresses markers associated withdifferentiation into a mineralizing phenotype. MC3T3-E1subclone 4 cells proliferate to confluence within 6–7 daysafter seeding (40,000–50,000 cell cm�2). Once confluent(�2–2.5� initial seeding density within 6–7 days), these cellscan be induced to differentiate by the addition of (ascorbicacid) AA and glycerol 2-phosphate. Ascorbic acid is usedto induce collagen type 1 synthesis, while glycerol 2-phos-phate is used to differentiate this cell line fully into a miner-alizing phenotype. Since collagen type 1 forms the biologicalsupport to which mineralized tissue binds (for later bone for-mation) [39–42], the focus is on the effect that the ionic prod-ucts of bioactive glass dissolution have on collagen type 1synthesis (Col1a1 and Col1a2 chains) and downstreamexpression of other key differentiation markers, alkalinephosphatase, core binding factor a (cbfa/Runx2) and osteo-calcin. Runx2 is a key transcription factor associated withearly expression of the osteoblast phenotype [43]. Alkalinephosphatase is a key dephosphorylating enzyme expressedby osteoblasts to turn over expressed collagen into a formthat is amenable for bone matrix formation [35,39–42]. Oste-ocalcin is a key non-collagenous protein which binds extra-cellular calcium to bone matrix [44].

2. Materials and methods

2.1. Study design

Pertaining to the goals described above, three experi-ments were designed to ascertain the effect of the ionic

3538 V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547

products of bioactive glass dissolution in vitro. First,in vitro dissolution experiments were conducted to deter-mine the ion constituency released from conditioned bioac-tive glass surfaces. Second, in vitro proliferationexperiments were conducted to determine the effect thatthis ion constituency had on osteoblast growth. Finally,in vitro differentiation experiments were conducted todetermine the effect that this ion constituency had on osteo-blast differentiation, and in particular, the effect that thision constituency had on the induction of collagen type 1,Runx2, alkaline phosphatase and osteocalcin by MC3T3-E1 subclone 4 cells. Gene, protein and extracellular matrixexpression of these markers were assayed or visualized inthis work.

2.2. In vitro dissolution experiments

2.2.1. Bioactive glass specimen preparation

Preparation of experimental bioactive coating glass(6P53-b, Table 1) was performed as described previously[16]. In brief, 6P53-b powders were commercially pur-chased (SEM-COM, Toledo, OH), placed into a Pt cruci-ble, and melted in air for 5 h between 1400 and 1500 �C.The melt was cast in a pre-heated (200 �C) graphite moldyielding glass bars and annealed (500 �C, 6 h). Theannealed bars were then cut into square samples(�1 � 1 � 0.2 cm) with a low-speed diamond saw (Isomet,Buehler Ltd., Lake Bluff, IL) and sterilized by irradiation(290 rad min�1). All glass specimens (45S5 and 6P53-b)were then soaked for 10 days in SBF prior to study in vitro.

2.2.2. Ion extract preparation

After SBF conditioning, 45S5 and 6P53-b glass speci-mens were removed, ultrasonicated in acetone and 2-pro-panol, and then dried overnight in a desiccator. Thesespecimens were then soaked in a-MEM (Invitrogen, Carls-bad, CA) at a volume: surface area of 3 ml cm�2, containedin 0.2 lm filter capped centrifuge tubes and stored in anincubator (37 �C, 5% CO2, 100% RH) for 2 days. The med-ium was then recovered as an ion extract, while glass spec-imens were removed. Additional sterile filtration wasperformed using a 0.2-lm sterile nylon filter cup.

2.2.3. ICP–MS analysis of ion extracts

The ion extracts (45S5 + a-MEM, 6P53-b + a-MEM)and control samples (a-MEM) were analyzed using induc-tively coupled plasma mass spectrometry (ICP–MS). ICP–MS analysis was performed by introducing ion extract ali-quots into a mass spectrometer (250 amu 7500a, AgilentTechnologies, Palo Alto, CA). Solutions were sprayedthrough a high solid type nebulizer as a plasma into a ther-moelectrically controlled spray chamber. This instrumenthas four mass-flow controllers (plasma, auxiliary, carriergas lines) and also contains a 27.12 MHz solid state (1600W) ICP source on a three-stage vacuum system. An ionoptic system (Omega II off-axis lens system) was used todirect the plasma through the quadrupole mass filter

(hyperbolic cross-sectional rod system, 3 MHz) and ontothe electron multiplier detector. Data acquisition was per-formed using a simultaneous dual-mode electron multiplier(nine-order dynamic range, 100 ls dwell time). The resultsof the analysis were then compared with a standard foridentification and quantitation of ion concentration. Theaverage and standard deviation of the ion concentrationsper set of samples were reported. The results of the ICP–MS analysis were reported as absolute ion concentrations.The units of ion concentration were ppm.

2.3. In vitro osteoblast proliferation experiments

2.3.1. Glass conditioned medium preparationThe ion extracts were supplemented with fetal bovine

serum (FBS, 10% by volume; Hyclone, Logan, UT) andpenicillin–streptomycin (pen-strep, 100 U ml�1 or 1% byvolume) to make glass conditioned medium (GCM). Thissupplementation means that cells were exposed to ion con-centrations at 89% of that reported for ICP–MS results ofion extracts and a-MEM. Basal medium was prepared bysupplementing a-MEM, with 10% FBS and 1% pen-strep.The above medium preparation was used for proliferationstudies.

2.4. Cell proliferation experiment

Prior to all experiments, MC3T3-E1 subclone 4 cells(passage 25–30) were cultured in 150 cm2 flasks. Cells wereseeded at a density of 50,000 cells cm�2. After the cell linedoubling time (12–16 h), these cells were synchronized(a-MEM, 1% FBS, 1% pen-strep) for an additional 48 h.The medium was then replaced, and cells were culturedin 45S5 and 6P53-b GCM treatments and basal media(control) treatments. No AA or glycerol 2-phosphate wasused. Cell cultures for each treatment were conducted intriplicate. These cells were cultured in four 96-well platesfor 1, 3, 5 and 7 days.

Cells were also assayed to determine whether prolifera-tion occurs during experiments, for which AA was added(to induce osteoblast collagen synthesis). To perform thisexperiment, the cells were cultured in GCM and basal med-ium with AA (50 ppm) addition. This experiment was con-ducted in combination with the previously describedproliferation experiment (without AA addition). Basalmedium that was not treated with AA was used as a con-trol treatment. Cells were seeded and arranged in wellplates as described above.

2.4.1. Cell proliferation assay

At the desired time point, each well plate was removedfrom incubation to assay for viable cell numbers. GCMand control media treatments were exchanged with the cellproliferation assay treatment (a-MEM, 10% FBS, 1% pen-strep, 317 lg ml�1 MTS reagent as described by the manu-facturer’s protocol) for 4 h in an incubator (37 �C, 5% CO2,100% RH). The color of the formazen product released by

Table 2Gene expression assay primers and probes (from ABI).

Gene Accession No. Gene bank mRNAs Amplicon length

GAPDH NM_008084.2 31 107AKP2 NM_007431.1 8 90OCN NM_031368.3 6 89RUNX2 NM_009820.2 2 115COL1a1 NM_007742.3 8 89COL1a2 NM_00773.2 18 140

V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547 3539

the cells was then analyzed using a spectrophotometer(490 nm, SpectraMax Plus, Molecular Devices, San Jose,CA) to quantify cell numbers relative to standard curves.

The MTS Assay (Promega, Madison, WI) is a colorimet-ric assay which is used for measuring the metabolic activityof viable cells during proliferation. The assay reagentcontains a tetrazolium compound [3-(4,5-dimethylthia-zol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophe-nyl)-2H-tetrazolium,innersalt;MTS] and an electroncoupling reagent (phenazine ethosulfate; PES). Thesereagents are combined into one solution, in which the tetra-zolium product is bioreduced by cells into a formazen prod-uct, which colors the solution and indicates the numbers ofviable cells. This assay has shown similar sensitivity to viablecell metabolic activity compared with the MTT assay [45].Standard curves for the use of this assay kit for this cell linewere used to determine cell numbers.

2.5. In vitro osteoblast differentiation experiments

2.5.1. GCM treatments and cell culture experiments

The cells were cultured in 150 cm2 flasks (passage 25–30)prior to seeding (50,000 cells cm�2) in 6-well tissue cultureplastic plates and incubated overnight. After the cell linedoubling time (12–16 h), these cells were synchronized (a-MEM, 1% FBS, 1% pen-strep) for an additional 48 h.For differentiation studies, synchronization media wasreplaced with 45S5 and 6P53-b GCM and basal media withadditional supplementation of AA (50 lg ml�1 (ppm)) wasused. The control treatment was basal medium with AA(50 ppm) supplement.

Protein concentrations of cell lysates were assayed atintervals of 1, 3, 5, 6 and 10 days. The first week of timepoints was compared with gene expression during this sametime period. The last time point was chosen to examine arelatively long-term effect of GCM and AA treatments onosteocalcin and alkaline phosphatase production. Allexperiments were conducted in triplicate, and samples wereassayed twice for confirmation. All results of analyzed pro-teins were normalized to total protein concentration. Eachassay contains an internal reference standard for determi-nation of concentration.

2.6. Protein expression assays

At the desired time point for assay, cultured cells werecollected in lysis buffer (CelLytic-M, Sigma Co., St. Louis,MO) and lysed according to the manufacturer’s protocol.Total protein quantification (BCA protocol, Pierce Bio-technology, Rockford, IL) was used to measure total cellu-lar protein collected from lysates. The results of this assaywere used to normalize alkaline phosphatase and osteocal-cin assay results.

The concentration of alkaline phosphatase (Procedure104) and osteocalcin (Mouse Osteocalcin EIA Kit, Bio-medical Technologies Inc., Stoughton, MA) were mea-sured. Adjustment of Procedure 104 was performed by

reaction of sample alkaline phosphatase with preincubated(37 �C) nitrophenyl phosphate (0.4% w:v) and amino-2-methyl-1-propanol (1.5 M, pH 10). The reaction was thenstopped using 0.05 N NaOH. The reaction produces inor-ganic phosphate and nitrophenol [46], which produces ayellow color, and its intensity is assayed using a spectro-photometer (400–420 nm) [46].

2.7. Reverse transcriptase polymerase chain reaction

To quantify levels of gene expression, quantitativereverse transcriptase polymerase chain reaction (qRT-PCR) was used. Cells were cultured as described aboveand pelleted at the selected time point (days 1, 3, 5 and7). The cells were lysed for their total RNA using theRNeasy Mini Kit (Quiagen, Valencia, CA). Extracts wereconverted to cDNA using reverse transcriptase (ReverseTranscription System, Promega, Madison, WI) accordingto the manufacturer’s protocol. Absorbance (A) measure-ments of mRNA and cDNA samples were performed usinga full-spectrum UV/vis nanodrop volume analyzer (ND-1000, Nanodrop Technologies, Wilmington, DE). Concen-tration of total RNA or total cDNA were measured at260 nm (A260). The measurement of nucleic acid absor-bance at A260/A280 is commonly used to assess the purityof nucleic acids after retrieval from cell and tissue cultures.The purity range used for this work was between 1.8 and2.0, which is recommended from the manufacturer’s proto-col. All cDNA samples were diluted to the same concentra-tion (100 ng ll�1).

For quantitative PCR, samples were mixed withreagents for PCR as follows (final concentration): cDNAsample (10%), FastStart Taqman Master Mix (Rox, 2�,50%) (Roche Applied Sciences, Mannheim, Germany), for-ward primer (900 nM), reverse primer (900 nM), andhydrolysis probe (250 nM) mixture (Table 2) (10%), andPCR grade water (Amgen Inc., South San Francisco,CA). Sample reaction was performed using a real-timePCR machine (ABI7500, Applied Biosystems Inc., FosterCity, CA).

All amplification was compared with an internal house-keeping gene (glyceraldehyde 3-phosphate dehydrogenaseor GAPDH) for relative quantification. Data acquisitionwas performed using the Applied Biosystems 7500 thermalcycler software package. Statistical analysis was conductedon amplification curves (for all samples, controls and

3540 V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547

blanks) using the SigmaPlot software package (Systat Soft-ware Inc., San Jose, CA). Statistical analysis on amplifica-tion curves was performed using sigmoidal curve fitting(following the procedure outlined by Qiu et al. [47]),Regression of these sigmoidal fits was performed using a4-parameter model offered by the SigmaPlot softwarepackage. The strength of the non-linear fit (R2) and thepresence of an adequate plateau of the amplification curvewere used as criteria for which curves were evaluated fortheir threshold cycle (CT). Amplification curves withR2 < 0.95 were not used.

Standard curves were used to evaluate serial dilutions ofsamples for gene amplification efficiency. Relative compar-ison of CT values was calculated so long as the expressionof this gene did not change as a function of time or treat-ment. Statistical evaluation among replicates was evaluatedto ensure that error was within 1%. Thus, when using thedelta-delta CT method, the variance in the data is associ-ated with the concentration of original target cDNA rela-tive to the concentration of original internal referencegene cDNA.

Genes that were studied are given in Table 2. Geneexpression was studied within the first 7 days of differenti-ation, because this period marks the early expression of dif-ferentiation of the osteoblast phenotype.

2.8. Histology

Cells were seeded (50,000 cells cm�2) onto glass coverslips seated in 6-well plates as described above. Cells weretreated in 6P53-b GCM and basal media with additionalsupplementation of AA (50 ppm). The experiment was car-ried out over 14 days, during which samples were removedfrom culture on 2, 9 and 14 day time points. Samples weretransferred to a fresh well plate, washed twice in PBS andfixed using Bouin’s fixative solution (Richard-Allen Scien-tific, Kalamazoo, MI) under ambient conditions for 1 h,with excess fixative rinsed off with deionized water. Bouin’sfixative is used to enhance the binding of Picrosirius stain-ing of collagen fibrils while using a Fastgreen backgroundtissue stain. Cell layers were then stained with 0.1% Fast-green (Sigma Inc.) for 30 min, with excess staining removedusing glacial acetic acid (American Master Tech ScientificInc., Lodi, CA) for 30 min. Fastgreen (light green or forestgreen, depending on tissue density) was used as a counter-stain for Picrosirius stain, and its intensity increased withincreasing extracellular matrix tissue density. Cell layerswere then stained using Picrosirius stain for 1 h, with excess

Table 3Ion extract concentration (mg l�1).

Si4+ Na+ K+

6P53-b + a-MEM 33.4 ± 3.8* 4458 ± 186 29845S5 + a-MEM 47.9 ± 10.4* 4168 ± 404 299a-MEMa 0.1 4300 325

* Significant difference relative to control (a-MEM) ion concentrations, p <a a-MEM ion concentrations were within 1% error. The same lot of a-MEM

stain washed off using deionized water. Picrosirius stainingstains collagen fibrils and is enhanced using polarized lightowed to the birefringence of collagen. Cell layers were thensequentially alcohol dehydrated (70–100% ethanol) for 30 sto remove excess water prior to imaging. Cell layers wereimaged using an optical microscope (Olympus BX51,Tokyo, Japan) with a CCD camera and Image Pro soft-ware. When imaged using polarized light, Picrosirius stain(on a fast green counterstain) stains collagen type 1 red,orange or yellow.

2.9. Statistics

Data analysis was conducted to determine the statisticalsignificance (p < 0.05). Two-way and one-way analysis ofvariance (ANOVA) testing was used to compare the treat-ments (45S5 GCM, 6P53-b GCM, control) and temporaleffects on MC3T3-E1 subclone 4 cells. For gene-relatedstudies, genes expressed as a result of exposure of cells todifferent treatments were reported relative to the controltreatment (100%). All between-group comparisons weremade using Tukey’s analysis. Comparisons betweena-MEM, 45S5 and 6P53-b ion extract ion concentrationswere analyzed for statistical significance using one-wayANOVA with Tukey’s analysis for between-groupcomparisons.

3. Results

3.1. Ion concentration analysis

Ion concentrations in the cell culture media changed as aresult of bioactive glass dissolution in vitro (without addi-tion of pen-strep and FBS). ICP–MS results (Table 3)showed (1) significantly increased concentrations of Si(p < 0.05 for both 45S5 and 6P53-b GCM) and Ca(p < 0.05 for 45S5 GCM) and (2) Na, K, Mg and phos-phate concentrations were not significantly changed ascompared with a-MEM. Although 45S5 ion extracts hadhigher Si and Ca ion concentrations than 6P53-b ionextracts, no significant differences were found.

3.2. GCM effect on osteoblast proliferation

Direct exposure of GCM enhanced the proliferation rateof pre-osteoblasts. Results (Fig. 1) of proliferation experi-ments showed increased numbers of cells in both GCMwithin 24–72 h (after cell synchronization (t = 0 days),

Ca2+ Mg2+ P043�

.3 ± 34.9 57.1 ± 2.8* 32.7 ± 5.4 32.6 ± 9.3

.8 ± 26.6 69.8 ± 14.0 27.8 ± 2.5 23.6 ± 4.7

.0 49.0 37.6 28.8

0.05.was used for all experiments.

Days1 3 5 7

Cel

l Den

sity

(ce

lls c

m -

2)

104

105

106

6P53-b GCM45S5 GCMcontrol

*

* ** *

Fig. 1. Effect of GCM on osteoblast proliferation (ANOVA, p < 0.05, *

denotes statistical significance). The control treatment in this experimentwas a-MEM (Table 3) with 10% FBS and 1% pen-strep supplements.GCM treatments were 6P53-b and 45S5 ion extracts (Table 3) with 10%FBS and 1% pen-strep supplements. GCM treatments appeared toincrease cell proliferation rate within 72 h after initial exposure of cellsto GCM and control treatments.

Days1 3 5 7

Co

l1α1

mR

NA

rel

ativ

e ex

pre

ssio

n, %

of

con

tro

l

0

100

200

300

400

500

6006P53-b GCM+AA45S5 GCM+AAcontrol+AA*

*

*

*

*

* *

Fig. 2. Collagen type 1a1 expression in the presence of GCM and AAcontrol- and AA-treated cells (ANOVA, p < 0.05, * denotes statisticalsignificance). The control in this experiment was a-MEM (Table 3) with10% FBS, 1% pen-strep and 50 ppm AA supplements. GCM treatmentswere 6P53-b and 45S5 ion extracts (Table 3) with the same supplementsused in control treatments. qPCR results were reported as Col1a1expression relative to an internal reference gene (GAPDH) and relative tocontrol treatments.

Days1 3 5 7

Co

l1α

2 m

RN

A r

elat

ive

exp

ress

ion

, % o

f co

ntr

ol

0

100

200

300

4006P53-b GCM+AA45S5 GCM+AAcontrol+AA

*

*

*

*

*

*

Fig. 3. Col1a2 expression in the presence of GCM and AA control- andAA-treated cells (ANOVA, p < 0.05, * denotes statistical significance).The control in this experiment was a-MEM (Table 3) with 10% FBS, 1%pen-strep and 50 ppm AA supplements. GCM treatments were 6P53-b and45S5 ion extracts (Table 3) with the same supplements used in controltreatments. qPCR results were reported as Col1a2 expression relative toan internal reference gene (GAPDH) and relative to control treatments.

V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547 3541

p < 0.05). For example, 1 day after treatment, cell numberswere significantly higher in 45S5 and 6P53-b GCM (1.2�and 1.3� of control, respectively) and a similar result wasseen on day 3 (1.4� and 1.5� of control, respectively). Sig-nificant differences were not observed after 72 h, with anobserved plateau of the cell density after 7 days in culturefor all treatments (probably due to limited MTS reagentexposure and/or reduced metabolic activity associated withcell overcrowding within each well in the well plate).

3.3. GCM effect on osteoblast differentiation

It is well known that AA induces the expression of col-lagen type 1 in MC3T3-E1 subclone 4 cells. Collagen type 1expression (Col1a1 and Col1a2) was enhanced in AA- andGCM-treated cultures compared with AA- and control-treated cultures. For example, Col1a1 was maximallyincreased after 1 day of exposure to GCM + AA (45S5,4� control + AA; 6P53-b, 3� control + AA; Fig. 2) whileCol1a2 expression was maximally increased after 5 days ofGCM + AA exposure (45S5, 3.2� control + AA; 6P53-b,2.4� control + AA; Fig. 3). Besides this maximal expres-sion, overall collagen type 1 expression was observedthroughout differentiation.

This enhanced collagen expression was observed extra-cellularly. Picrosirius staining of cell layers (Fig. 4) showedgreater density of collagen fibers in 6P53-b and AA treat-ments compared with control and AA treatments. Forexample, after 2 days of treatment, higher amounts offibrillar collagen were observed in cells treated in 6P53-b

GCM and AA treatments (Fig. 4d), whereas very littlefibrillar collagen was observed in control- and AA-treated

Fig. 4. Optical micrographs showing extracellular matrix collagen as a result of MC3T3-E1 subclone 4 exposure to (a–c) control media and AA treatmentsand (d–f) 6P53-b GCM and AA treatments treatments over a 14-day period. Cells were seeded (50,000 cells cm�2) onto glass cover slips and cultured for 2,9 and 14 days in their respective treatments. Cells were fixed in Bouin’s fixative for 1 h, followed by Fastgreen and Picrosirius stains for 30 min each.Imaging shows the presence of type 1 collagen in cell layers. A greater presence of such fibers was observed in 6P53-b GCM- and AA-treated cellscompared with control media and AA treatments over the time course of the experiment. Arrows indicate the presence of fibrillar collagen. Note:

background Fastgreen stained more intensely in more dense tissue layers, hence, the reason for the difference in green background staining. Further,background staining of blue, blue-green and green are typical for fast green treatments coupled with Picrosirius follow-on staining.

Days1 3 5 10μ g

/mL

alk

alin

e p

ho

sph

atas

e / m

g/m

L t

ota

l pro

tein

0

1

2

3

4

5

6

76p53b GCM+AA45s5 GCM+AAcontrol+AA

* *

*

*

Fig. 5. Effect of GCM and control treatments on osteoblast alkalinephosphatase expression (ANOVA, p < 0.05, * denotes statistical signifi-cance). The control in this experiment was a-MEM (Table 3) with 10%FBS, 1% pen-strep and 50 ppm AA supplements. GCM treatments were6P53-b and 45S5 ion extracts (Table 3) with the same supplements used incontrol treatments. Results were reported as alkaline phosphataseconcentrations normalized to total protein concentrations.

3542 V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547

cells (Fig. 4a). Throughout the time course of this experi-ment (14 days), a higher density of fibrillar collagen wasobserved for 6P53-b GCM- and AA-treated cell cultures(Fig. 4e (9 days), Fig. 4f (14 days)) compared with control-and AA-treated cells (Fig. 4b (9 days), Fig. 4c (14 days)).Also note the ‘‘layering” of collagen fibrils seen inFig. 4d (2 days of 6P53-b and AA treatment) comparedwith Fig. 4a (2 days of control and AA treatment). The lay-ers of collagen fibrils appeared to align approximatelyorthogonally to each other. This appearance is owed tothe birefringent nature of collagen fibrils when stained withPicrosirius staining. Such layering of collagen fibrils wasnot apparent in control- and AA-treated cells after 2 days,but this layering of collagen was observed after 9 and14 days (Fig. 4b and c) in culture. For 6P53-b- and AA-treated cells, layering of collagen happened as early as2 days in culture (Fig. 4d) and illustrates the enhancedeffect that bioactive glass ions have on collagen synthesis.

Collagen expression and AA treatment are prerequisitesfor downstream expression of alkaline phosphatase, Runx2and osteocalcin expression [39,40,48]. Since the bioactiveglass corrosion products enhanced collagen type 1 expres-sion, it is reasonable to predict that downstream expressionof these markers occurs. For alkaline phosphatase expres-sion, GCM- and AA-treated cells induced a significantincrease in expression compared with control- and AA-treated cells. For example, after 5 and 10 days of exposure,a 1.4� and 2� increase in alkaline phosphatase expression,was seen in GCM and AA-treated cells compared with con-trol and AA treated cells (Fig. 5).

GCM and AA treatments also enhanced Runx2 expres-sion. Runx2, a transcription factor expressed by osteo-blasts during early differentiation [49], was significantly

Days3 5 7

Ru

nx2

mR

NA

rel

ativ

e ex

pre

ssio

n, %

of

con

tro

l

0

50

100

150

200

250

3006P53-b GCM+AA45S5 GCM+AAcontrol+AA

* *

* *

Fig. 6. Gene expression of Runx2 during the time course of differentiation(ANOVA, p < 0.05, * denotes statistical significance). The control in thisexperiment was a-MEM (Table 3) with 10% FBS, 1% pen-strep and50 ppm AA supplements. GCM treatments were 6P53-b and 45S5 ionextracts (Table 3) with the same supplements used in control treatments.qPCR results were reported as Runx2 expression relative to an internalreference gene (GAPDH) and relative to control treatments. (Amplifica-tion results for day 1 were not observed and are therefore not reportedhere.)

Days1 3 5 6 10

μg/m

L o

steo

calc

in /

mg

/mL

to

tal p

rote

in

0

200

400

600

800

1000

1200

1400

1600

1800

20006P53-b GCM+AA45S5 GCM+AAcontrol+AA

*

*

*

*

**

*

Fig. 7. Effect of GCM and AA and control and AA treatments onosteoblast osteocalcin expression (ANOVA, p < 0.05, * denotes statisticalsignificance). The control in this experiment was a-MEM (Table 3) with10% FBS, 1% pen-strep and 50 ppm AA supplements. GCM treatmentswere 6P53-b and 45S5 ion extracts (Table 3) with the same supplementsused in control treatments. Results were reported as osteocalcin concen-trations normalized to total protein concentrations.

Days1 3 5 7O

steo

calc

in m

RN

A r

elat

ive

exp

ress

ion

, % o

f co

ntr

ol

0

500

1000

1500

2000

25006P53-b GCM+AA45S5 GCM+AAcontrol+AA

* *

*

*

**

*

Fig. 8. Osteocalcin gene expression in the presence of GCM and AA andcontrol- and AA-treated cell cultures over 7 days during the course ofdifferentiation (ANOVA, p < 0.05, * denotes statistical significance). Thecontrol in this experiment was a-MEM (Table 3) with 10% FBS, 1% pen-strep and 50 ppm AA supplements. GCM treatments were 6P53-b and45S5 ion extracts (Table 3) with the same supplements used in controltreatments. qPCR results were reported as Runx2 expression relative to aninternal reference gene (GAPDH) and relative to control treatments.

V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547 3543

increased (>2� control- and AA-treated cells) in GCM-and AA-treated cells (Fig. 6).

GCM treatments also enhanced osteocalcin expression.For example, the expression of osteocalcin protein peaked(Fig. 7) for cells treated with 6P53-b GCM and AA treat-ments (40� control and AA treatments after 10 days)and 45S5 GCM and AA treatments (70� control andAA treatments after 6 days). This significantly increasedprotein expression of osteocalcin coincided with the geneexpression of osteocalcin within 7 days of treatment. Oste-ocalcin gene expression reached a maximum after 3 days in6P53-b GCM and AA treatments (14� control and AAtreatments) and after 5 days in 45S5 GCM and AA treat-ments (19� control and AA treatments) (Fig. 8). These sig-nificant increases in osteocalcin gene and proteinexpression indicate the enhanced effect of GCM on differ-entiating MC3T3-E1.4 cells.

4. Discussion

This study tested the hypothesis that bioactive coatingglass ions enhanced the expression of genes and proteinsassociated with osteoblast differentiation (ECM). Theresults from this study indicated that osteoblast behaviorwas enhanced by these ions derived from both bioactivecoating glasses and commercial Bioglass, and that thisenhanced behavior occurred at the gene, protein and extra-cellular matrix levels.

3544 V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547

The dissolution of bioactive glass 6P53-b in a-MEM wasfound to be similar to that of Bioglass [7]. The dissolutionof both glass materials (after an initial 10-day conditioningperiod) led to significantly higher concentrations of Si andCa released in vitro. The 10-day in vitro glass conditioningperiod was needed to prevent MC3T3-E1 subclone 4 cul-tures from incidental exposure to higher than physiologicalin vitro pH, as was found in previous work [19,20]. Withoutin vitro glass conditioning, the GCM takes on a basic pH,which was found to decrease alkaline phosphatase expres-sion [20]. Besides the in vitro release of Si and Ca ions intoa-MEM, no significant increase in alkali ions wasobserved, which meant that exposure of cells to increasedpH was averted. This change in ion release characteristicswith in vitro immersion time follows a heterogeneous disso-lution mechanism. First, rapid ion exchange of alkali ions(Na+, K+) with aqueous hydrogen (H+) occurred (diffu-sion-limited regime). This was followed by a surface-reac-tion limited release of Si and/or Ca. This change in ionrelease mechanism coincides with the formation of a por-ous, partially crystalline HA surface layer (as confirmedby X-ray diffraction, scanning electron microscopy andelectron dispersive spectroscopy analysis [19]). In a similarglass system (57 wt.% SiO2), Saiz et al. [10] reported contin-ued silicon and calcium dissolution over a 30-day periodin vitro. They found that, after 10 days of conditioning inSBF, Ca2+ concentrations increased (15–20 ppm higherthan SBF Ca concentration after 8 days of soaking) andSi4+ concentrations increased (40–50 ppm higher thanSBF Si concentration after 8 days of soaking) [10]. Dissolu-tion over an additional 2-day soak in a-MEM (to makeGCM) showed continued dissolution of Si and Ca, consis-tent with the observed trend reported by Saiz et al. [10].

MC3T3-E1 subclone 4 cells were observed to have a sig-nificantly higher cell density within 72 h after seeding(Fig. 1). This meant that the growth rate of cells was higherduring this 72-h period in GCM-treated cells comparedwith control-treated cells. Although this increase is consid-ered significant, it may be relatively small, considering thatthe cell density approached the same maximum value forall treatments. Thus, the effect that GCM had on cellgrowth rate may induce only a slight decrease in theamount of time needed for this cell line to reach confluence.

Before discussing the results of GCM treatments onosteoblast differentiation, it is important first to considerthe timeline of events that occurs during the MC3T3-E1subclone 4 cell cycle during differentiation (reviewed byXiao et al. [50]). The events of osteogenesis take place inthe presence of a collagenous matrix (if one is not present,cells can be exposed to AA to induce such a matrix) [50].The binding of these cells to this collagenous matrix isa2b1 integrin mediated [50]. Integrin binding at the osteo-blast membrane induces the expression of mitogen-acti-vated protein kinases (MAPK) which transduce signals tothe osteoblast nucleus [50]. These MAPK phosphorylateand activate alkaline phosphatase and Runx2, which thenbinds to the promoter region of downstream genes such

as osteocalcin [50]. This timeline can occur within the first10 days of MC3T3-E1 subclone 4 differentiation.

The ionic products of bioactive coating glass dissolutionhad a significant impact on osteoblast differentiation. Col-lagen type 1 gene expression (Figs. 2 and 3) and extracellu-lar matrix formation (Fig. 4) was enhanced over a period of14 days. Considering the timeline of marker expression thatoccurs with these MC3T3-E1 subclone 4 osteoblasts (pre-sented above), the increased collagen expression probablyled to increased downstream expression of alkaline phos-phatase (Fig. 5), Runx2 (Fig. 6) and osteocalcin (Figs. 7and 8).

There were similarities and differences between this studyand that of Foppiano et al. [22] using a similar glass compo-sition. On the one hand, this work and that of Foppianoet al. [22] differed on collagen type 1 expression. Foppianoet al. [22] found that collagen type 1 (Col1a1) was expressedat a slightly lower level (80% of control and AA treatment)after 7 days of exposure of MC3T3-E1 subclone 4 cells to a57 wt.% SiO2 GCM and AA treatment. Under similar con-ditions in this work, Col1a1 was 1.4� control and AA treat-ment in 6P53-b GCM and AA treatment, and 0.9� controland AA treatment in 45S5 GCM and AA treatment (Fig. 2).Considering that the present results show markedly higherexpression level increases (>2� control and AA treatment),the small changes in expression may not greatly influenceosteoblast induction of collagen synthesis. Thus, a 20%decrease or increase in relative expression may not appre-ciably alter downstream matrix production. A greaterimpact on collagen type 1 expression was observed at earliertime points; thus, the result observed by Foppiano et al. [22]was probably due to the down-regulation of collagen once asignificant number of fibrils were present in the ECM after5 days of exposure to GCM and AA treatments.

On the other hand, this work and that of Foppiano et al.[22] are in agreement with regard to the GCM and AAeffect on Runx2 expression. Foppiano et al. [22] found thatincreased expression of Runx2 (2� control and AA treat-ment) was observed for MC3T3-E1 subclone 4 cellsexposed to their GCM and AA treatment after 7 days. Sim-ilar results were observed in the present study, in whichRunx2 expression was increased (2� control and AA treat-ment) in GCM and AA treatments (Fig. 6). The increasedexpression of Runx2 was probably due to the increasedpresence of a collagen matrix and collagen type 1 geneexpression.

In addition to the enhanced osteocalcin expression thatoccurred as a result of MC3T3-E1 subclone 4 cells’ expo-sure to GCM and AA treatments (protein, Fig. 7; mRNA,Fig. 8), two unexpected results occurred. First, enhancedosteocalcin expression occurred after 3–6 days of GCMand AA treatments compared with control and AA treat-ments. Second, this enhanced osteocalcin expression wasmarkedly higher than the enhanced expression observedfor collagen type 1, alkaline phosphatase and Runx2. Thisenhanced osteocalcin expression appeared in line with theenhanced collagen type 1 expression, but appeared out of

V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547 3545

step with the increased Runx2 expression (7 days of GCMand AA treatments). These results could be explained bythe increased cellular production and faster attachment toa collagenous ECM when these cells are exposed toGCM and AA treatments compared with control- andAA-treated cells. Considering the timeline above, the earlyappearance of enhanced osteocalcin expression could berelated to enhanced integrin-mediated MAPK expressionas a result of an enhanced collagenous matrix, formedwhen these cells were exposed to GCM and AA treatments.The enhanced osteocalcin expression also suggested thatthese cell-signaling mechanisms could also mediate osteo-calcin production along with Runx2 expression and subse-quent promotion of osteocalcin.

The enhanced behavior of pre-osteoblasts in the pres-ence of GCM is probably due to the increased Ca and Simedia concentrations. On the one hand, the role of Ca iswell known in osteoblasts, as evident in the vast array ofcommercially available dietary supplements and pharma-ceuticals used to treat patients with low bone density orosteoporosis. Maeno et al. [33] found that increased Caconcentrations (8 mM higher than that of a-MEM Ca con-centration) also enhanced the expression of osteocalcin (2�control) in primary mouse osteoblast monolayer cultures.In the present work, the relative increased Ca concentra-tions for 6P53-b and 45S5 GCM were 8 and 20 ppm (0.2and 0.5 mM, respectively) and may have led, in part, toincreased levels of osteocalcin. The increased Ca concentra-tion may trigger mechanisms in which MC3T3-E1 cellsrespond via a calcium receptor [51–54].

On the other hand, the role of Si on osteoblasts is stilluncertain. In the present work, increased Si concentrationsfor 6P53-b and 45S5 GCM (1.2 and 1.7 mM, respectively)may have had a greater impact on the observed enhancedosteocalcin expression, Runx2 and collagen expression. InMG-63 osteosarcoma cultures, Reffitt et al. [34] found thatincreased Si ion concentrations (10–50 lM, 3 days after cellconfluence) increased the relative expression of collagentype I (1.8� control), alkaline phosphatase (1.5� control)and osteocalcin (1.5� control). Further studies or evidenceare lacking in showing a direct effect of Si on osteoblasts.The present study shows a general trend in that increasedSi concentrations (45S5 GCM had higher levels of Si com-pared with 6P53-b GCM) had an effect on the expression ofsome osteogenic markers; however, no statistically signifi-cant differences were found. The effect Si has on osteoblastswill be the focus of future work. Furthermore, possible syn-ergistic effects of Ca and Si have yet to be explored.

This work establishes enhanced behavior of osteoblastsin the presence of GCM and AA treatments. A follow-upstudy on mineralization is currently under way to determinethe effect that these ions have on osteoblast mineralization.

This study raises clinically relevant issues with regard tofuture implant preparation prior to implantation. One suchissue is the necessity of the conditioning period to avoidpotentially adverse localized conditions near the implantcoating-bone interface. Although an increased in vitro pH

could suggest a potentially harmful situation physiologi-cally, the initial rapid dissolution physiologically may notbe deleterious, since the physiological environment ishighly buffered and frequently replenished. Since commer-cial Bioglass is already in clinical use, the use of the coatingglass on a Ti implant may be suitable. Still, pre-clinical andclinical trials must be pursued to ensure the possible use ofthese materials as coatings for Ti.

5. Conclusions

Enhanced expression of four key osteogenic markerswas found for cultured cells in the presence of GCM andAA treatments. Markedly higher expression of osteocalcinprotein was observed, and coincided with increased osteo-calcin gene expression. The increased osteocalcin expres-sion was probably due to the increased expression ofcollagen type 1 at both the gene and extracellular matrixlevels. GCM and AA treatments also enhanced alkalinephosphatase and Runx2 expression. The enhanced expres-sion most likely occurred as a result of the increased con-centrations of both Ca and Si in the GCM from bothbioactive glasses.

Acknowledgements

The authors would like to thank the following contribu-tors to this paper: Tiffany Vallortigara, Janet Wong, KellyLeong and Garrett Porteous. The authors would also liketo thank the following for their kind advice related to theabove work: Linda Prentice, Larry Watanabe, GraceNonomura, Steven Lee, Michael Carillo, Dr. Stuart Gan-sky, Professor Pamela DenBeston, Dr. Yuan Zhang, Dr.James Chen, Dr. Stefan Habelitz, Dr. Kuniko Saeki andDr. Huynh Tri. In addition, the authors thank UC DavisICP–MS facility for solution and GCM analyses. Finally,the authors appreciate the financial support by theNational Institutes of Health/National Institute of Dentaland Craniofacial Research Grants K25 DE018230Varanasi (PI) and R01 DE11289 Tomsia (PI).

Appendix

Figures with essential colour discrimination. Certainfigures in this article, particularly Figures 1–7, are difficultto interpret in black and white. The full colour images canbe found in the on-line version, at 10.1016/j.actbio.2009.05.035.

References

[1] Sawyer JD. The quest for nature’s match. In: American CeramicSociety Bulletin; 2007. p. 25–9.

[2] Adell R, Lekholm U, Rockler B, Branemark PI. A 15-year study ofosseointegrated implants in the treatment of the edentulous jaw. Int JOral Surg 1981;10(6):387–416.

[3] Miles BA, Sinn DP, Gion GG. Experience with cranial implant-basedprosthetic reconstruction. J Craniofac Surg 2006;17(5):889–97.

3546 V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547

[4] Baltag L, Watanabe K, Kusakari H, Taguchi N, Miyakawa O,Kobayashi M, et al. Long-term changes of hydroxyapatite-coateddental implants. J Biomed Mater Res 2000;53(1):76–85.

[5] Lee TM, Yang CY, Chang E, Tsai RS. Comparison of plasma-sprayed hydroxyapatite coatings and zirconia-reinforced hydroxyap-atite composite coatings: in vivo study. J Biomed Mater Res A2004;71A(4):652–60.

[6] Hench LL, Wilson J. Surface-active biomaterials. Science1984;226(4675):630.

[7] Hench L. Bioceramics. J Am Ceram Soc 1998;81(7):1705.[8] Moreira-Gonzalez A, Lobocki C, Barakat K, Andrus L, Bradford M,

Gilsdorf M, et al. Evaluation of 45S5 bioactive glass combined as abone substitute in the reconstruction of critical size calvarial defects inrabbits. J Craniofac Surg 2005;16(1):63–70.

[9] Wilson J, Low SB. Bioactive ceramics for periodontal treatment –comparative-studies in the Patus monkey. J Appl Biomater1992;3(2):123–9.

[10] Saiz E, Goldman M, Gomez-Vega JM, Tomsia AP, Marshall GW,Marshall SJ. In vitro behavior of silicate glass coatings on Ti6Al4V.Biomaterials 2002;23:3749–56.

[11] Lopez-Esteban S, Saiz E, Fujino S, Oku T, Suganuma K, Tomsia AP.Bioactive glass coatings for orthopedic metallic implants. J EurCeram Soc 2003;23(15):2921–30.

[12] Pazo A, Saiz E, Tomsia AP. Silicate glass coatings on Ti-basedimplants. Acta Mater 1998;46(7):2551–8.

[13] Oku T, Suganuma K, Wallenberg LR, Tomsia AP, Gomez-Vega JM,Saiz E. Structural characterization of the metal/glass interface inbioactive glass coatings on Ti–6Al–4V. J Mater Sci Mater Med2001;12(5):413–7.

[14] Gomez-Vega JM, Saiz E, Tomsia AP. Glass-based coatings fortitanium implant alloys. J Biomed Mater Res1999;46(4):549–59.

[15] Fujino S, Morinaga K, Saiz E, Tomsia AP. Bioactive glass coatingson Co–Cr implant alloys. Glass Sci Technol 2002;75:221–6.

[16] Foppiano S, Marshall SJ, Marshall GW, Saiz E, Tomsia AP. Theinfluence of novel bioactive glasses on in vitro osteoblast behavior. JBiomed Mater Res A 2004;71A(2):242–9.

[17] Bloyer DR, Gomez-Vega JM, Saiz E, McNaney JM, Cannon RM,Tomsia AP. Fabrication and characterization of a bioactive glasscoating on titanium implant alloys. Acta Mater 1999;47(15–16):4221–4.

[18] Russias J, Saiz E, Deville S, Gryn K, Liu G, Nalla RK, et al.Fabrication and in vitro characterization of three-dimensionalorganic/inorganic scaffolds by robocasting. J Biomed Mater Res A2007;83A(2):434–45.

[19] Varanasi VG, Vallortigara T, Loomer PM, Saiz E, Tomsia AP,Marshall SJ, et al. Improving biomaterials from a cellular pointof view. In: Materials research society symposium proceedings,2006. San Francisco, CA: Materials Research Society; 2006. p.BB03.02.

[20] Varansi VG, Saiz E, Loomer PM, Uritani N, Ancheta BT, TomsiaAP, et al. On the in vitro behavior and osteoblast response tounconditioned bioactive coating glass and commercial bioglass. JBiomed Mater Res B, submitted for publication.

[21] Foppiano S, Marshall SJ, Saiz E, Tomsia AP, Marshall GW.Functionally graded bioactive coatings: reproducibility and stabilityof the coating under cell culture conditions. Acta Biomater2006;3(5):765–71.

[22] Foppiano S, Marshall SJ, Marshall GW, Saiz E, Tomsia AP.Bioactive glass coatings affect the behavior of osteoblast-like cells.Acta Biomater 2007;3(5):765–71.

[23] Xynos ID, Hukkanen MVJ, Hench LL, Polak JM. Bioglass 45S5induces mineralisation in osteoblast cultures in vitro. J Pathol1999;187:38a.

[24] Xynos ID, Hukkanen MVJ, Batten JJ, Buttery LD, Hench LL, PolakJM. Bioglass (R) 45S5 stimulates osteoblast turnover and enhancesbone formation in vitro: implications and applications for bone tissueengineering. Calcif Tissue Int 2000;67(4):321–9.

[25] Xynos ID, Edgar AJ, Ramachandran M, Buttery LDK, Hench LL,Polak JM. Biochemical characterisation and gene expression profilingof human trabecular bone derived osteoblasts. J Pathol 2001;193:31a.

[26] Xynos ID, Edgar AJ, Buttery LDK, Hench LL, Polak JM. Gene-expression profiling of human osteoblasts following treatment withthe ionic products of Bioglass (R) 45S5 dissolution. J Biomed MaterRes 2001;55(2):151–7.

[27] Xynos ID, Edgar AJ, Buttery LDK, Hench LL, Polak JM. Ionicproducts of bioactive glass dissolution increase proliferation ofhuman osteoblasts and induce insulin-like growth factor II mRNAexpression and protein synthesis. Biochem Biophys Res Commun2000;276(2):461–5.

[28] Jones JR, Sepulveda P, Hench LL. Dose-dependent behavior ofbioactive glass dissolution. J Biomed Mater Res 2001;58(6):720–6.

[29] Jones JR, Hench LL. Regeneration of trabecular bone using porousceramics. Curr Opin Solid State Mater Sci 2003;7(4–5):301–7.

[30] Hench LL, Polak JM. Third-generation biomedical materials. Science2002;295(5557):1014.

[31] Bielby RC, Pryce RS, Hench LL, Polak JM. Enhanced derivation ofosteogenic cells from murine embryonic stem cells after treatmentwith ionic dissolution products of 58S bioactive sol–gel glass. TissueEng 2005;11(3–4):479–88.

[32] Bielby RC, Christodoulou IS, Pryce RS, Radford WJP, Hench LL,Polak JM. Time- and concentration-dependent effects of dissolutionproducts of 58S sol–gel bioactive glass on proliferation and differen-tiation of murine and human osteoblasts. Tissue Eng 2004;10(7–8):1018–26.

[33] Maeno S, Niki Y, Matsumoto H, Morioka H, Yatabe T, FunayamaA, et al. The effect of calcium ion concentration on osteoblastviability, proliferation and differentiation in monolayer and 3Dculture. Biomaterials 2005;26(23):4847–55.

[34] Reffitt DM, Ogston N, Jugdaohsingh R, Cheung HFJ, Evans BAJ,Thompson RPH, et al. Orthosilicic acid stimulates collagen type 1synthesis and osteoblastic differentiation in human osteoblast-likecells in vitro. Bone 2003;32(2):127–35.

[35] Beck GR, Zerler B, Moran E. Phosphate is a specific signal forinduction of osteopontin gene expression. Proc Natl Acad Sci USA2000;97(15):8352–7.

[36] Zreiqat H, Shakibaei ME, Evans P, Knabe C, Howlett CR.Mechanisms of magnesium-stimulated adhesion of osteoblastic cellsto commonly used orthopaedic implants. J Bone Miner Res2001;16:S328.

[37] Zreiqat H, Howlett CR, Zannettino A, Evans P, Schulze-Tanzil G,Knabe C, et al. Mechanisms of magnesium-stimulated adhesion ofosteoblastic cells to commonly used orthopaedic implants. J BiomedMater Res 2002;62(2):175–84.

[38] Hattar S, Asselin A, Greenspan D, Oboeuf M, Berdal A, Sautier JM.Potential of biomimetic surfaces to promote in vitro osteoblast-likecell differentiation. Biomaterials 2005;26(8):839–48.

[39] Franceschi RT, Iyer BS. Relationship between collagen-synthesis andexpression of the osteoblast phenotype in Mc3t3-E1 cells. J BoneMiner Res 1992;7(2):235–46.

[40] Franceschi RT, Iyer BS, Cui YQ. Effects of ascorbic-acid on collagenmatrix formation and osteoblast differentiation in murine Mc3t3-E1cells. J Bone Miner Res 1994;9(6):843–54.

[41] Franceschi RT, Wilson JX, Dixon SJ. Requirement for Na+-dependent ascorbic-acid transport in osteoblast function. Am JPhysiol Cell Physiol 1995;37(6):C1430–9.

[42] Quarles LD, Yohay DA, Lever LW, Caton R, Wenstrup RJ. Distinctproliferative and differentiated stages of murine Mc3t3-E1 cells inculture – an invitro model of osteoblast development. J Bone MinerRes 1992;7(6):683–92.

[43] Byers BA, Pavlath GK, Murphy TJ, Karsenty G, Garcia AJ. Cell-type-dependent up-regulation of in vitro mineralization after overex-pression of the osteoblast-specific transcription factor Runx2/Cbfa1.J Bone Miner Res 2002;17(11):1931–44.

[44] Xiao GZ, Cui YQ, Ducy P, Karsenty G, Franceschi RT. Ascorbicacid-dependent activation of the osteocalcin promoter in MC3T3-E1

V.G. Varanasi et al. / Acta Biomaterialia 5 (2009) 3536–3547 3547

preosteoblasts: requirement for collagen matrix synthesis and thepresence of an intact OSE2 sequence. Mol Endocrinol1997;11(8):1103–13.

[45] Rotter BA, Thompson BK, Clarkin S, Owen TC. Rapid colorimetricbioassay for screening of fusarium mycotoxins. Nat Toxins1993;1:303–7.

[46] Sestini S, Notarantonio L, Cerboni B, Alessandrini C, Fimiani M,Nannelli P, et al. In vitro toxicity evaluation of silver soldering,electrical resistance, and laser welding of orthodontic wires. Eur JOrthod 2006;28(6):567–72.

[47] Qiu H, Durand K, Rabinovitch-Chable H, Rigaud M, Gazaille V,Clavere P, et al. Gene expression of HIF-1 alpha and XRCC4measured in human samples by real-time RT-PCR using thesigmoidal curve-fitting method. Biotechniques 2007;42(3):355–62.

[48] D’Alonzo RC, Kowalski AJ, Denhardt DT, Nickols GA, PartridgeNC. Regulation of collagenase-3 and osteocalcin gene expression bycollagen and osteopontin in differentiating MC3T3-E1 cells. J BiolChem 2002;277(27):24788–98.

[49] Xiao GZ, Jiang D, Ge CX, Zhao ZR, Lai YM, Boules H, et al.Cooperative interactions between activating transcription factor 4and Runx2/Cbfa1 stimulate osteoblast-specific osteocalcin geneexpression. J Biol Chem 2005;280(35):30689–96.

[50] Xiao GZ, Gopalakrishnan R, Jiang D, Reith E, Benson MD,Franceschi RT. Bone morphogenetic proteins, extracellular matrix,

and mitogen-activated protein kinase signaling pathways are requiredfor osteoblast-specific gene expression and differentiation in MC3T3-E1 cells. J Bone Miner Res 2002;17(1):101–10.

[51] Yamaguchi T, Chattopadhyay N, Kifor O, Butters RR, Sugimoto T,Brown EM. Mouse osteoblastic cell line (MC3T3-E1) expressesextracellular calcium (Ca-0(2+))-sensing receptor and its agonistsstimulate chemotaxis and proliferation of MC3T3-E1 cells. J BoneMiner Res 1998;13(10):1530–8.

[52] Yamaguchi T, Chattopadhyay N, Kifor O, Sanders JL, Brown EM.Activation of p42/44 and p38 mitogen-activated protein kinases byextracellular calcium-sensing receptor agonists induces mitogenicresponses in the mouse osteoblastic MC3T3-E1 cell line. BiochemBiophys Res Commun 2000;279(2):363–8.

[53] Yamauchi M, Yamaguchi T, Kaji H, Sugimoto T, Chihara K.Involvement of calcium-sensing receptor in osteoblastic differentia-tion of mouse MC3T3-E1 cells. Am J Physiol Endocrinol Metab2005;288(3):E608–16.

[54] Yamauchi M, Yamaguchi T, Sowa H, Yano S, Kaji H, Sugimoto T,et al. The extracellular calcium-sensing receptor (CaR) is indispens-able for expression of alkaline phosphatase (ALP) and osteocalcin(OC) as well as mineralization in mouse osteoblastic MC3T3-E1 cells.J Bone Miner Res 2003;18:S289.