mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α...

12
Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1a,25- dihydroxyvitamin D3 and BMP2 via two specic vitamin D receptors Jiaxuan Chen 1, Christopher R. Dosier 2, Jung Hwa Park 3 , Subhendu De 1 , Robert E. Guldberg 2 , Barbara D. Boyan 1,4 * and Zvi Schwartz 1,4 1 Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, GA, USA 2 Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA, USA 3 School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA, USA 4 Department of Biomedical Engineering, Virginia Commonwealth University, Richmond, VA, USA Abstract 1a,25-Dihydroxyvitamin D3 [1a,25(OH)2D3] and bone morphogenetic protein-2 (BMP2) are both used to stimulate osteoblastic differentiation. 1a,25(OH)2D3 regulates osteoblasts through classical steroid hormone receptor mechanisms and through rapid responses that are mediated by two receptors, the traditional vitamin D receptor (VDR) and protein disulphide isomerase family A member 3 (Pdia3). The interaction between 1a,25(OH)2D3 and BMP2, especially in three-dimensional (3D) culture, and the roles of the two vitamin D receptors in this interaction are not well understood. We treated wild-type (WT), Pdia3-silenced (Sh-Pdia3) and VDR-silenced (Sh-VDR) pre-osteoblastic MC3T3-E1 cells with either 1a,25(OH)2D3, or BMP2, or with 1a,25(OH)2D3 and BMP2 together, and measured osteoblast marker expression in 2D culture and mineralization in a 3D poly(«-caprolactone)collagen scaffold model. Quantitative PCR showed that silencing Pdia3 or VDR had a differential effect on baseline expres- sion of osteoblast markers. 1a,25(OH)2D3 + BMP2 caused a synergistic increase in osteoblast marker expression in WT cells, while silencing either Pdia3 or VDR attenuated this effect. 1a,25 (OH)2D3 + BMP2 also caused a synergistic increase in Dlx5 in both silenced cell lines. Micro-computed tomography (mCT) showed that the mineralized volume of untreated Sh-Pdia3 and Sh-VDR 3D cultures was greater than that of WT. 1a,25(OH)2D3 reduced mineral in WT and Sh-VDR cultures; BMP2 increased mineralization; and 1a,25(OH)2D3 + BMP2 caused a synergistic increase, but only in WT cultures. SEM showed that mineralized matrix morphology in 3D cultures differed for silenced cells compared to WT cells. These data indicate a synergistic crosstalk between 1a,25(OH)2D3 and BMP2 toward osteogenesis and mineral deposition, involving both VDR and Pdia3. Copyright © 2013 John Wiley & Sons, Ltd. Received 9 April 2013; Accepted 16 April 2013 Supporting information may be found in the online version of this article. Keywords polycaprolactone scaffolds; osteoblasts; mineralization; 1a,25(OH)2D3; BMP2; Pdia3; VDR 1. Introduction The vitamin D metabolite 1a,25-dihydroxyvitamin D3 [1a,25(OH)2D3] is known for its role in maintaining calcium and phosphate homeostasis. Mineralization of growth plate cartilage and bone is reduced in 1a,25(OH) *Correspondence to: B. D. Boyan, School of Engineering, Virginia Commonwealth University, 601 West Main Street, Richmond, VA 23284, USA. E-mail: [email protected] J. Chen and C. R. Dosier are co-rst authors of this manuscript. Copyright © 2013 John Wiley & Sons, Ltd. JOURNAL OF TISSUE ENGINEERING AND REGENERATIVE MEDICINE RESEARCH ARTICLE J Tissue Eng Regen Med (2013) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/term.1770

Upload: zvi

Post on 17-Feb-2017

216 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

Mineralization of three-dimensional osteoblastcultures is enhanced by the interaction of 1a,25-dihydroxyvitamin D3 and BMP2 via two specificvitamin D receptorsJiaxuan Chen1†, Christopher R. Dosier2†, Jung Hwa Park3, Subhendu De1, Robert E. Guldberg2,Barbara D. Boyan1,4* and Zvi Schwartz1,41Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, GA, USA2Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA, USA3School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA, USA4Department of Biomedical Engineering, Virginia Commonwealth University, Richmond, VA, USA

Abstract

1a,25-Dihydroxyvitamin D3 [1a,25(OH)2D3] and bone morphogenetic protein-2 (BMP2) are bothused to stimulate osteoblastic differentiation. 1a,25(OH)2D3 regulates osteoblasts through classicalsteroid hormone receptormechanisms and through rapid responses that aremediated by two receptors,the traditional vitamin D receptor (VDR) and protein disulphide isomerase family A member 3 (Pdia3).The interaction between 1a,25(OH)2D3 and BMP2, especially in three-dimensional (3D) culture, andthe roles of the two vitaminD receptors in this interaction are not well understood.We treatedwild-type(WT), Pdia3-silenced (Sh-Pdia3) and VDR-silenced (Sh-VDR) pre-osteoblastic MC3T3-E1 cells witheither 1a,25(OH)2D3, or BMP2, or with 1a,25(OH)2D3 and BMP2 together, and measured osteoblastmarker expression in 2D culture and mineralization in a 3D poly(«-caprolactone)–collagen scaffoldmodel. Quantitative PCR showed that silencing Pdia3 or VDR had a differential effect on baseline expres-sion of osteoblast markers. 1a,25(OH)2D3+BMP2 caused a synergistic increase in osteoblastmarker expression in WT cells, while silencing either Pdia3 or VDR attenuated this effect. 1a,25(OH)2D3+BMP2 also caused a synergistic increase in Dlx5 in both silenced cell lines. Micro-computedtomography (mCT) showed that the mineralized volume of untreated Sh-Pdia3 and Sh-VDR 3D cultureswas greater than that of WT. 1a,25(OH)2D3 reduced mineral in WT and Sh-VDR cultures; BMP2increased mineralization; and 1a,25(OH)2D3+BMP2 caused a synergistic increase, but only in WTcultures. SEM showed that mineralized matrix morphology in 3D cultures differed for silenced cellscompared to WT cells. These data indicate a synergistic crosstalk between 1a,25(OH)2D3 and BMP2toward osteogenesis and mineral deposition, involving both VDR and Pdia3. Copyright © 2013 JohnWiley & Sons, Ltd.

Received 9 April 2013; Accepted 16 April 2013

Supporting information may be found in the online version of this article.

Keywords polycaprolactone scaffolds; osteoblasts; mineralization; 1a,25(OH)2D3; BMP2; Pdia3; VDR

1. Introduction

The vitamin D metabolite 1a,25-dihydroxyvitamin D3[1a,25(OH)2D3] is known for its role in maintainingcalcium and phosphate homeostasis. Mineralization ofgrowth plate cartilage and bone is reduced in 1a,25(OH)

*Correspondence to: B. D. Boyan, School of Engineering,Virginia Commonwealth University, 601 West Main Street,Richmond, VA 23284, USA. E-mail: [email protected]†J. Chen and C. R. Dosier are co-first authors of this manuscript.

Copyright © 2013 John Wiley & Sons, Ltd.

JOURNAL OF TISSUE ENGINEERING AND REGENERATIVE MEDICINE RESEARCH ARTICLEJ Tissue Eng Regen Med (2013)Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/term.1770

Page 2: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

2D3 deficiency, leading to skeletal deformities associatedwith rickets (Holick, 2006, 2007). In addition to its sys-temic effects on mineral ion homeostasis, 1a,25(OH)2D3has effects on development of these tissues (Chen et al.,2009; Lee et al., 2011), indicating that skeletal cellspossess receptors for this secosteroid. Two receptors for1,25(OH)2D3 have been identified in osteoblasts: thecanonical nuclear vitamin D receptor (VDR) (Huhtakangaset al., 2004; Walters et al., 1982) and a plasma membranereceptor, protein disulphide isomerase family A member 3(Pdia3) (Boyan et al., 2002b; Chen et al., 2010). Both VDRand Pdia3 contribute to rapid membrane-associated signal-ling (Chen et al., 2010; Mizwicki et al., 2004; Zanello andNorman, 2004), leading to altered gene expression, in addi-tion to the traditional role of the VDR (Breen et al., 1994;Kraichely and MacDonald, 1998).

The role that 1a,25(OH)2D3 plays in mineralization iscomplex. Mice lacking a functional VDR exhibit rickets(Li et al., 1997; Yoshizawa et al., 1997), which can behealed by restoring serum Ca++ content through diet(Amling et al., 1999). Although mineralization of thegrowth plate and bone matrix is restored, growth plateanomalies remain (Chen et al., 2009; Lee et al., 2011),indicating that VDR-dependent signalling is involved inmore aspects of skeletal development than mineral iontransport. Moreover, VDR–/– mice possess Pdia3 and oste-oblasts from VDR–/– mice retain Pdia3-dependent rapidresponses to 1a,25(OH)2D3 (Boyan et al., 2003). Globalknockout of Pdia3 is embryologically lethal, but Pdia3+/–

heterozygous mice exhibit a bone phenotype with onlyminor effects (Wang et al., 2010). Thus, both receptorsplay important roles in bone development, but theirindependent contributions to osteoblast differentiationand mineralization have not been compared in one modelsystem.

In culture, 1a,25(OH)2D3 has been shown to bothincrease (Halvorsen et al., 2001; Matsumoto et al., 1985,1991) and decrease (Ecarot and Desbarats, 1999; Lynchet al., 1995; Slater et al., 1994) calcium phosphate deposi-tion, suggesting that the effects of 1a,25(OH)2D3 maydepend on culture conditions, cell source or cell matura-tion stage. In addition to 1a,25(OH)2D3, cell culturemodels examining osteoblast differentiation frequentlyinclude factors to stimulate osteoblast differentiation, suchas dexamethasone and b-glycerophosphate (Bonewaldet al., 2003), as well as bone morphogenetic protein-2(BMP2) (Boyan et al., 2002a). These media additives canalter the effects of 1a,25(OH)2D3 in different ways. Forexample, addition of transforming growth factor b1(TGFb1) to osteoblast cultures modulates the effect of1a,25(OH)2D3 on osteoblast differentiation, increasingalkaline phosphatase activity but inhibiting osteocalcinproduction (Bonewald et al., 1992). While these studiesdemonstrate that osteoblast differentiation in two-dimensional (2D) cultures is regulated by multipleinteracting factors, few studies have been done toinvestigate the interaction between 1a,25(OH)2D3 andBMP2 signalling pathways in either 2D or three-dimensional (3D) systems.

To better understand the roles of Pdia3 and VDR inmediating the interaction of 1a,25(OH)2D3 and BMP2in the regulation of osteoblast differentiation, we tookadvantage of a 3D cell culture model in which we wereable to examine the ability of osteoblasts stably silencedfor Pdia3 or VDR to support calcium phosphate deposi-tion. In this model, MC3T3-E1 cells were cultured onpoly(e-caprolactone) (PCL)–collagen scaffolds. The PCLprovided a consistent porosity, and the collagen facilitatedcell adhesion and retention within the scaffold (Dosieret al., 2012). We first investigated the interaction of1a,25(OH)2D3 and BMP2 through the expression of oste-oblast markers in 2D cultures and then studied the rolesof Pdia3 and VDR in mediating this interaction. Finally,we studied the roles of Pdia3 and VDR in mediating1a,25(OH)2D3 and BMP2 stimulated calcification by oste-oblasts in the 3D model.

2. Materials and methods

2.1. Pdia3/VDR silencing

An MC3T3-E1 cell line with> 80% silencing of Pdia3mRNA and protein (Sh-Pdia3) was previously established(Chen et al., 2010). The same approachwas used to developMC3T3-E1 cells stably silenced for VDR (Sh-VDR). Briefly,VDR short hairpin RNA probes were designed to targetthe mouse Vdr mRNA (NM_009504) and five differentsequences were generated and incorporated into lentivirusparticles (MISSION™ shRNA, Sigma-Aldrich, St. Louis,MO, USA). MC3T3-E1 cells (CRL-2593, ATCC, Manassas,VA, USA) were plated at a density of 20 000 cells/cm2 in a24-well plate in a-minimal essential medium (a-MEM)supplemented with 10% fetal bovine serum (FBS) and 1%penicillin–streptomycin (P/S). After 24h, the medium waschanged to a-MEM supplemented with 10% FBS, 1% P/Sand 8mg/ml hexadimethrine bromide and transduced withlentivirus particles at a multiplicity of infection (MOI) of7.5. Cells containing shRNAs or empty vectors were selectedby culturing the cells for 2weeks in medium containing2.0mg/ml puromycin. Loss of Vdr expressionwas quantifiedby real-time PCR. The clone with the highest silencing ratewas chosen from the five different shRNA-transfectedclones. The selected Sh-VDR cell line exhibited a 90%reduction in Vdr mRNA and a comparable reduction inVDR protein (data not shown).

2.2. 2D cell culture

Wild-type (WT), Sh-Pdia3 and Sh-VDR MC3T3-E1 cellswere plated at 20 000 cells/cm2 in T75 flasks with fullmedium (a-MEM supplemented with 10% FBS, 1% P/S,with or without 2.0 mg/ml puromycin). After 48 h, fullmedium was changed to full medium containing 50mg/mlvitamin C to enable cross-linking of type I collagen in theextracellular matrix.

J. Chen et al.

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 3: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

Ten days after plating, WT, Sh-Pdia3 and Sh-VDRMC3T3-E1 cells were treated with vehicle or 10–8M1a,25(OH)2D3 (Biomol, Plymouth Meeting, PA, USA)for 15min to activate Pdia3-dependent signalling viaprotein kinase C (PKC) and ERK1/2 mitogen-activatedprotein kinase (MAPK) (Chen et al., 2010). Alternatively,the medium was replaced by fresh medium with or with-out 50 ng/ml recombinant human BMP2 (B3555, Sigma-Aldrich) for 12 h to stimulate osteoblastic gene expression(Hassan et al., 2006; Ulsamer et al., 2008). This resultedin cultures treated with medium containing no additives(I); cultures treated with medium containing 1a,25(OH)2D3, followed by medium without BMP2 (II); culturestreated with medium alone for 15min, followed bymedium containing BMP2 (III); and cultures treated withmedium containing 1a,25(OH)2D3 for 15min followed bymedium containing BMP2 (IV). The cell layers were lysedwith TRIzol (Invitrogen, Carlsbad, CA, USA) 12 h later toharvest RNA (Table 1).

2.2.1. Gene expression

RNA was reverse-transcribed into cDNA, using the high-capacity cDNA reverse transcription kit (AppliedBiosystems, Carlsbad, CA, USA) according to the manu-facturer’s directions. In order to determine whether therewas crosstalk at the expression level amongst Pdia3, VDRand BMP2, mRNAs for Pdia3 (Pdia3) and VDR (Vdr), aswell as BMP2 (Bmp2) and the BMP2 inhibitor Noggin(Nog) were measured in WT, Sh-Pdia3 and Sh-VDR cells.mRNAs for alkaline phosphatase (Alpl), osteocalcin (Bglap)and osteopontin (Spp1) were measured as indicators of os-teoblast differentiation. In addition, others have shown thatBMP2 directly upregulates Distal-less homeobox5 expres-sion mRNA (Dlx5) and indirectly upregulates Runt-related

transcription factor 2 mRNA (Runx2) (Lee et al., 2003;Ulsamer et al., 2008). Therefore, these two transcriptionfactors, which are also associatedwith osteoblastic differen-tiation, were measured.

Oligonucleotide primers to the targeted genes weredesigned using Beacon Designer 7.0 software (PremierBiosoft International, Palo Alto, CA, USA). A homologyblast search was performed within the mammaliangenome to exclude the possibility of sequence similarity.Primers were synthesized by Eurofins MWG Operon(Huntsville, AL, USA) and are shown in Table 2. Real-time PCR was performed using SYBR Green SuperMix,the Veriti 96-well Thermal Cycler and Step One software(Applied Biosystems). Data were normalized to theendogenous reference gene, glyceraldehyde 3-phosphatedehydrogenase (Gapdh).

2.2.2. Alkaline phosphatase activity

Changes in alkaline phosphatase specific activity wereused as an outcome measure of osteoblast differentiation.Wild-type, Sh-Pdia3 and Sh-VDR cells were treated asdescribed above. At the end of the 12 h BMP2 treatment,the medium was replaced by full medium for another12 h. The cell layers were washed twice with cold PBSand lysed in 0.05% Triton X-100. Alkaline phosphatase-specific activity was measured as the release of p-nitrophenolfrom p-nitrophenylphosphate at pH 10.2, as describedpreviously (Martin et al., 1995).

2.3. 3D culture model

2.3.1. PCL–collagen 3D scaffold preparation

PCL scaffolds were prepared as previously described(Dosier et al., 2012; Peister et al., 2009). Sheets (100� 100 � 9mm) of medical grade poly(e-caprolactone)(Osteopore International, Singapore) with 85% poros-ity were cut with a 5mm internal diameter biopsypunch to yield a cylindrical scaffold. The scaffolds weretreated briefly with 5M sodium hydroxide in order toroughen the surface to facilitate cell attachment; thescaffolds were then washed three times with sterile wa-ter and sterilized overnight via 70% ethanol evaporation.

Table 1. Treatment conditions

Group 1,25D3 BMP2

I – –

II – +III + –

IV + +

Table 2. Oligonucleotide primers for real-time PCR of mRNAs for Pdia3, VDR (Vdr), BMP2 (Bmp2), Noggin (Nog), alkaline phospha-tase (Alpl), osteocalcin (Bglap), osteopontin (Spp1), Dlx5, Runx2 and GAPDH

Gene Forward primer Reverse primer

Pdia3 CGA TGT GTT GGA ACT GAC G TTC ATA CTC AGG GGC AAG CVdr AGG CAG GCA GAA GAG ATG AG AGG GAT GAT GGG TAG GTT GTGBmp2 TGG GTT TGT GGT GGA AGT G TCGTTTGTGGAGCGGATGNog GCC AGC ACT ATC TAC ACA TCC CAG CAG CGT CTC GTT CAGAlpl GTG GGC ATT GTG ACT ACC GGT GGC ATC TCG TTA TCCBglap TCT CTC TGC TCA CTC TGA GTC TGT TCA CTA CCT TAT TGCSpp1 AAC TCT TCC AAG CAA TTC C TCT CAT CAG ACT CAT CCGDlx5 TCA GGA ATC GCC AAC TTT G CCA TAA GAA GCA GAG GTA GGRunx2 CCG CCA CCA CTC ACT ACC GAT AGG ATG CTG ACG AAG TAC CGapdh TTC AAC GGC ACA GTC AAG G TCT CGC TCC TGG AAG ATG G

Regulation of 3D osteoblast cultures

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 4: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

Sterile PCL scaffolds were washed three times with excesssterile water and then were placed into a sterile cus-tom mould. Sterile rat tail type I collagen solution(Trevigen, Gaithersburg, MD, USA) was diluted withsterile filtered 0.05% acetic acid to 1.5mg/ml, neutral-ized with sterile filtered 1M sodium bicarbonate andaseptically pipetted into the mould to fill the pores of thescaffold. The PCL–collagen gel constructs were then placedin a –80�C freezer for 1 h prior to being lyophilized over-night. Using a sterile scaffold holder, the lyophilized con-structs were placed into 24-well low-attachment cellculture plates (Corning, Lowell, MA, USA); they were thenwrapped with parafilm and stored at room temperatureuntil cell seeding. The porosity of the PCL–collagenconstructs was not determined.

2.3.2. 3D cell culture

In order to study mineralization in this 3D system,WT, Sh-Pdia3 and Sh-VDR MC3T3-E1 cells wereplated at 20 000 cells/cm2 in T75 flasks with full medium(a-MEM supplemented with 10% FBS, 1% P/S, with orwithout 2.0 mg/ml puromycin). At confluence the cellswere trypsinized, counted and reconstituted in fullmedium at a density of 3 � 104 cells/ml. 100 ml of thecell suspension was pipetted onto the tops of thePCL/collagen constructs and cells were allowed toattach to the surface. The cells interact with the colla-gen, forming a network spanning the struts of the scaf-fold. Cells also attach to the roughened PCL surface(Dosier et al., 2012; Erdman et al., 2012; Peisteret al., 2011). After a 1 h incubation period, full mediumwas added to the culture wells so that the cell–scaffoldconstructs were submerged. After 24 h, the cultureswere treated with vehicle (ethanol) or 10–8 M 1a,25(OH)2D3 in full medium for 15min. The medium waschanged to osteogenic medium consisting of a-MEMsupplemented with 16% FBS (Atlanta Biologics,Lawrenceville, GA, USA), 1% P/S, 50 mg/ml ascorbic acid2-phosphate (Sigma-Aldrich), 6mM b-glycerophosphate(Sigma-Aldrich) and 1nM dexamethasone (Sigma-Aldrich).These culture conditions were previously demonstratedto be optimal for long-term cell culture in this system(Dosier et al., 2012; Erdman et al., 2012; Peisteret al., 2011). One half of the cultures were treated with50ng/ml recombinant human BMP2 (Sigma-Aldrich). Theosteogenic medium � BMP2 was changed three timesweekly during cell culture. At each change, the same15min transient treatment of 1a,25(OH)2D3 was applied.Cell–scaffold constructs were cultured dynamically on anorbital shaker (Stovall Life Scientific, Greensboro, NC,USA) at a rate of 6.5 rpm in a 5% CO2 incubator. At 4 and8weeks, mineralized matrix of the cell–scaffold constructswas determined via mCT imaging. At the end of 8weeks,samples were fixed in 10% neutral buffered formalin for24h twice and further processed for scanning electronmicroscopy (SEM) and surface analysis by X-ray photonmicroscopy (XPS), as described below.

2.4. Analysis of 3D constructs

2.4.1. mCT imaging

In order to determine the volume of mineral, cell–scaffoldconstructs were removed aseptically from culture at 4 and8weeks and placed in custom tubes for mCT scanning.Mineral volume was determined using a VivaCT scanner(Scanco Medical, Brüttisellen, Switzerland) at 55 kVp,109mA, 1024 mm scaling and a 200ms integration time.The constructs were evaluated with a threshold of 80,with a filter width of 1.2 and a filter support of 1.0. Thetotal volume of the mineralized construct was thendetermined.

2.4.2. Scanning electron microscopy

In order to study the morphology of the mineral, theconstructs were first sectioned transversely, coated withgold (thickness� 8nm) and the cut surface was imagedusing a Hitachi 4700 scanning electron microscope(SEM; Hitachi High Technologies America, USA) with anaccelerating voltage of 12 kV.

2.4.3. X-ray photoelectron spectroscopy

The chemical composition of the cut surface was deter-mined by XPS. Adult mouse femoral bone was used as apositive control. XPS (Thermo K-a, Thermo FisherScientific, MA, USA) was used under ultra-high vacuum(< 10–9 Torr) with a monochromatic Al Ka X-ray source(hn=1486.6 eV, 90� take-off angle). The XPS spectrawere evaluated using Thermo Advantage 4.43 softwarepackage. The distribution of Ca (347 eV) and P (133 eV)on the surface of the constructs was obtained by XPSchemical mapping.

2.5. Statistical analysis

All experiments were repeated at least once to ensurevalidity of the data. The data presented are from a singlerepresentative experiment. Each data point represents themean� standard error (SE) for six independent cellcultures from a single experiment. Significance was deter-mined by one-way analysis of variance and post hoctesting performed using Bonferroni’s modification ofStudent’s t-test for multiple comparisons. p≤0.05 wasconsidered significant.

In some cases, repeated experiments had differentbaseline levels. In order to compare the results, we usedthe ratios of treatment to control from each experiment.The value for each sample in the treated group wasdivided by the mean of the control group. Each data pointrepresents mean� SE for six normalized values and adashed line with value equal to one represented the con-trol. Due to the non-normal distribution, significancewas determined using the Mann–Whitney test; p≤ 0.05was considered significant.

J. Chen et al.

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 5: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

3. Results

3.1. Gene expression

Baseline mRNA expression differed among the three celllines. Silencing Pdia3 reduced Pdia3 expression by 80%compared to wild-type MC3T3-E1 cells (Figure 1a) buthad no effect on expression of Vdr (Figure 1b). Similarly,silencing Vdr had no effect on Pdia3 expression butreduced Vdr levels by 90%. Bmp2 levels increased in Sh-Pdia3 cells by 500% and in Sh-VDR cells by 300%(Figure 1c). Expression of Nog was also affected in thesilenced cells. There was a 200% increase in Sh-Pdia3cells and a 1000% increase in Sh-VDR cells (Figure 1d).Baseline levels of mRNAs for Runx2 were unaffected by

silencing either Pdia3 or Vdr (Figure 1e). Whereas Dlx5did not change in Sh-Pdia3 cells compared to WT, expres-sion was reduced in the Sh-VDR cells (Figure 1f). Baselinelevels for mRNAs associated with osteoblast differentiationwere sensitive to silencing of specific receptors for1a,25(OH)2D3. Sh-Pdia3 cells had reduced levels of Alpl,whereas Alpl was increased in Sh-VDR cells (Figure 1g).Alkaline phosphatase-specific activity was affected in acomparable manner (Figure 1h). No mRNA for Bglap wasdetected in Sh-Pdia3 cells but expression was markedlyincreased in Sh-VDR cells (Figure 1i). In contrast, silencingPdia3 had no effect on Spp1 but expression in Sh-VDR cellswas markedly reduced compared to WT (Figure 1j).

mRNA levels were differentially affected by 1a,25(OH)2D3 and BMP2, either alone or in combination. Regard-less of whether or not cells were silenced for Pdia3 or

Figure 1. Effect of silencing Pdia3 or VDR on gene expression and alkaline phosphatase activity. WT, Sh-Pdia3 and Sh-VDR MC3T3-E1cells were treated with full medium containing the 1a,25(OH)2D3 vehicle (ethanol) for 15min, followed by replacing the mediumwith medium containing vehicle. mRNA was harvested 12h later and real-time PCR was performed: Pdia3 (A), Vdr (B), Bmp2 (C),Nog (D), Runx2 (E), Dlx5 (F), Alpl (G), Bglap (I) and Spp1 (J). Alkaline phosphatase activity was measured at 24h (H): $p<0.05,Sh-Pdia3 or Sh-VDR vs WT; %p<0.05, Sh-VDR vs Sh-Pdia3

Figure 2. Effect of 1a,25(OH)2D3 and BMP2 on gene expression and alkaline phosphatase activity in WT, Sh-Pdia3 and Sh-VDRMC3T3-E1 cells. WT, Sh-Pdia3 and Sh-VDR MC3T3-E1 cells were treated with full medium containing the 1a,25(OH)2D3 vehicle(ethanol) or 10–8M 1a,25(OH)2D3 for 15min, followed by replacing the medium with medium with vehicle (ethanol) or 50ng/mlBMP2. mRNA was harvested 12h later and real-time PCR was performed: Pdia3 (A), Vdr (B), Bmp2 (C), Nog (D), Runx2 (E), Dlx5(F), Alpl (G), Bglap (I) and Spp1 (J). Alkaline phosphatase activity was measured at 24h (H): $p<0.05, Sh-Pdia3 or Sh-VDR vs WT.To evaluate the effect of the treatments independently of differences in baselines, values from treated groups [1a,25(OH)2D3,BMP2 or both] were divided by values from the vehicle-only control group. Data presented are the resulting treatment/control ratios:*p<0.05, treatments [1a,25(OH)2D3, BMP2 or 1a,25(OH)2D3+BMP2] vs control; #p<0.05, [1a,25(OH)2D3 or 1a,25(OH)2D3+BMP2]vs BMP2 alone; &p<0.05, 1a,25(OH)2D3+BMP2 vs 1a,25(OH)2D3 alone

Regulation of 3D osteoblast cultures

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 6: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

Vdr, no treatment altered Pdia3 expression compared totheir baseline levels (Figure 2a). Similarly, neither treat-ment with 1a,25(OH)2D3 alone nor BMP2 alone alteredVdr, but the combination caused a> two-fold increase inVdr expression in WT cells but had no effect on thesilenced cells (Figure 2b). 1a,25(OH)2D3 alone increasedBmp2 expression in WTcells, whereas BMP2 had no effecton any of the cell lines (Figure 2c). Similarly, 1a,25(OH)2D3 increased Nog expression in WT cells only andBMP2 had no effect (Figure 2d). Expression of Runx2was not affected by any of the treatments in any of the celllines (Figure 2e). Dlx5 was not affected by 1a,25(OH)2D3alone but BMP2 caused a small increase in WT and Sh-Pdia3 cells. When BMP2 was used following 1a,25(OH)2D3 treatment, Dlx5 was increased in WT cells and to agreater extend in both of the silenced cell lines(Figure 2f). Both 1a,25(OH)2D3 and BMP2 alone causedsmall increases in Alpl expression in WT cells and therewas a synergistic increase in Alpl in WTand Sh-Pdia3 cellswhen the combination treatment was used (Figure 2g).This was correlated with increased activity in the WT andSh-Pdia3 cells treated with the combination of 1a,25(OH)2D3 and BMP2 (Figure 2h). Bglap expressionwas increased

over baseline only in cultures treated with 1a,25(OH)2D3plus BMP2 (Figure 2i). It was absent in Sh-Pdia3 cellsregardless of treatment and reduced compared to baselinein Sh-VDR cells. Spp1 was increased by 1a,25(OH)2D3 inWT cells and Sh-Pdia3 cells and to a comparable extent inWT and Sh-Pdia3 cells treated with 1a,25(OH)2D3 plusBMP2 (Figure 2j). 1a,25(OH)2D3 also increased Spp1 inSh-VDR cells, but to a lesser extent. By itself BMP2 had noeffect on Spp1.

3.2. Mineralization in 3D scaffolds

3D reconstruction of m-CT scans of the cell/scaffoldconstructs showed that mineral deposition was regulatedby 1a,25(OH)2D3 and BMP2 in a differential manner,depending on cell type (Figure 3). AlthoughWTconstructswere not mineralized at 4weeks, Sh-Pdia3 constructs werewell mineralized at that time. Treatment with 1a,25(OH)2D3 alone had no effect on WT and Sh-Pdia3 cells but re-duced mineral deposition in Sh-VDR cells. BMP2 increasedmineral deposition in the WT constructs and the Sh-VDRconstructs, but did not increase mineral deposition in

Figure 3. Effect of 1a,25(OH)2D3 and BMP2 onmineralized volume in 3D PCL scaffolds. Cells were cultured and treated as previouslydescribed. At 4 and 8weeks, mCT was used to measure the mineralized volume. An image of the representative sample that had theclosest value to the mean is shown. The mean mineralized volume (mm3) � SEM for the group is provided under the image: #,p<0.05, Sh-Pdia3 and Sh-VDR vs WT under same treatment; &, p<0.05, BMP2 vs control; *, p<0.05, 1a,25(OH)2D3 vs control

J. Chen et al.

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 7: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

Sh-Pdia3 constructs. Treatment of the constructs withBMP2 and 1a,25(OH)2D3 had no additional effect butSh-VDR constructs exhibited reduced mineral contentcompared to the constructs treated with BMP2 alone.

At 8weeks, these effects were more pronounced. WTconstructs were more mineralized than at 4weeks; thiswas increased by BMP2 alone. 1a,25(OH)2D3 and BMP2together generated a synergistic increase in mineral depo-sition. Sh-Pdia3 constructs had the greatest mineral con-tent and this was not affected by treatment with 1a,25(OH)2D3 alone, BMP2 alone or both factors. Sh-VDRconstructs also exhibited greater mineral deposition at8weeks but this was reduced by treatment with 1a,25(OH)2D3. Treatment with BMP2 increased mineral depo-sition over that seen in untreated Sh-VDR cells and it alsoreduced the inhibitory effect of 1a,25(OH)2D3.

These qualitative observations were confirmed by mea-surement of mineralized volume. At 4weeks there was noevidence of mineral in WT cultures and only low levels ofmineral in Sh-VDR cultures, whereas Sh-Pdia3 cultureswere well calcified (Figure 4a). At 8weeks, mineralizedvolume of the WT scaffolds was low, whereas calciumphosphate deposits in Sh-Pdia3 cultures and Sh-VDRcultures were extensive and to a comparable extent(Figure 4b). Treatment with 1a,25(OH)2D3 reducedmineralized volume in WT and Sh-VDR constructs at4weeks (Figure 4c), whereas BMP2 increased minera-lized volume in these same culture. 1a,25(OH)2D3 and

BMP2 in combination caused a synergistic increase inWT constructs only. At 8weeks, 1a,25(OH)2D3 continuedto suppress mineralized volume of sh-VDR constructs(Figure 4d). BMP2 increased mineral in WT constructsand caused a small increase over baseline in Sh-VDR con-structs. However 1a,25(OH)2D3 and BMP2 in combina-tion resulted in a synergistic increase in WT constructsbut the stimulatory effect of BMP2 on Sh-VDR constructswas abrogated.

The morphology of the 3D cultures varied with the cellline and treatment regimen. Low magnification SEMs ofthe bisected scaffolds showed that cells and extracellularmatrix were present throughout (see supporting informa-tion, Figure S1). WT cells generated a smooth plate-likestructure compared to the very porous net-like structuresthat were seen in both Sh-Pdia3 and Sh-VDR constructs.High-magnification SEMs of the bisected surface of scaf-folds containing wild-type cells treated with 1a,25(OH)2D3 or BMP2 showed that aggregated clusters wererandomly distributed (Figure 5). In contrast, WTcells thatwere treated with both 1a,25(OH)2D3 and BMP2 exhibiteda morphology with mineralized areas under 1mm in lengthevenly covering all of the surface. Globular aggregated fea-tures approximately 10–15mm in diameter were observedin most of the scaffolds containing Sh-Pdia3 cells. Scaffoldscontaining Sh-VDR cells had densely packed sphericalfeatures embedded in continuously formed layers similarto those seen in Sh-Pdia3 constructs, whether the cells were

Figure 4. Effect of 1a,25(OH)2D3 and BMP2 on mineralized volume in 3D PCL scaffold. Cells were cultured and treated as describedabove. At 4 and 8weeks, mCTwas used to measure the mineralized volume. The mineralized volumes without treatment are shown in(A; 4weeks) and (B; 8weeks). To evaluate the effect of the treatments independently of differences in baselines, values from treatedgroups [1a,25(OH)2D3, BMP2 or both] were divided by values from the vehicle-only control group. Data presented are the resultingtreatment/control ratios for 4 (C) and 8 (D) weeks: *, p<0.05, treatments [1a,25(OH)2D3, BMP2 or 1a,25(OH)2D3+BMP2] vs con-trol; #, p<0.05, 1a,25(OH)2D3 or 1a,25(OH)2D3+BMP2 vs BMP2 alone; &, p<0.05, 1a,25(OH)2D3+BMP2 vs 1a,25(OH)2D3 alone

Regulation of 3D osteoblast cultures

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 8: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

treated with vehicle or 1a,25(OH)2D3. SEM images ofdecalcified Sh-Pdia3 constructs exhibited a fibrous ma-trix structure without any globular features, indicatingthat these structures might be mineral deposits (seesupporting information, Figure S2).

Oxygen (O), carbon (C), nitrogen (N), and calcium(Ca), and phosphorus (P) were detected in all samplesas well as in native bone (see supporting information,Figure S3a). Ca:P ratios for all constructs varied from0.8 to 0.9, similar to native bone. Interestingly, therewas a difference in the distribution of the depositedelements. Wild-type constructs exhibited localizeddeposits of Ca and P. The locations of Ca and P werepartially overlapping. The distribution was not changed,whether they were treated with BMP2 alone or with1a,25(OH)2D3 and BMP2. The Sh-Pdia3 constructs hadmore diffuse and more colocalized Ca and P depositsthroughout compared to WT constructs, with or withoutBMP2 treatment (see supporting information, FigureS3b). The deposits of Ca and P in the femur were alsomore diffused compared to WT constructs.

4. Discussion

The results of this study demonstrate that BMP2 and1a,25(OH)2D3 act in concert to enhance osteoblast differ-entiation in both 2D and 3D cultures. The intent of theexperimental design was to examine the potential role ofrapid responses to 1a,25(OH)2D3 in osteoblast differenti-ation by exposing the cultures to a 15min pulse of thesecosteroid at each change of medium. This sort of expo-sure has been shown to induce effects in osteoblasts inprevious publications (Chen et al., 2010; Doroudi et al.,2012). This is different from the previous study, where os-teoblast differentiation was primed by exposure to BMP2prior to 24 h treatment with 1a,25(OH)2D3 (Schwartzet al., 1998). We found that even short-duration treatmentwith 1a,25(OH)2D3 was sufficient to stimulate osteoblastdifferentiation in 2D culture and regulate mineralizationin 3D culture, indicating that membrane-associated rapidsignalling was involved.

As anticipated, Sh-Pdia3 cells had reduced levels ofPdia3 mRNA and Sh-VDR cells had reduced levels of

Figure 5. Effect of 1a,25(OH)2D3 and BMP2 on morphology of the mineralized matrix at high magnification. Cells were cultured andtreated as described previously. At 8weeks, the constructs were bisected and the cut surface examined by SEM (magnification =�2000). The images were digitally enlarged from the original image to show the details; scale bar=1mm

J. Chen et al.

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 9: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

Vdr. Moreover, reduced expression of one receptor did nothave an appreciable effect on expression of the otherreceptor. In contrast, diminished levels of both receptorscaused marked increase in Bmp2 and its inhibitor Nog.Furthermore, baseline levels of mRNAs for osteoblastmarker genes varied with the cell line. Whereas cellssilenced for Pdia3 had reduced expression of mRNAs foralkaline phosphatase and osteocalcin compared to wild-type MC3T3-E1 cells, VDR-silenced cells had higher levelsof these mRNAs. In contrast, Sh-Pdia3 cells had normallevels of mRNA for osteopontin whereas Sh-VDR cellshad markedly reduced levels of mRNAs for this extracellu-lar matrix protein. These observations suggest that signal-ling pathways mediated by these receptors differentiallyregulate these mRNAs. We did not use charcoal-strippedserum for these experiments due to the length of theculture; as a result, the cells were exposed to the complexmilieu of growth factors and hormones typical of serum,including low levels of 1a,25(OH)2D3 (10–11–10–13M)(Schwartz et al., 1989). Even so, the marked differencesbetween the two cell lines indicate different roles for eachreceptor, which are evident when only one receptor isfunctional.

The 2D culture studies demonstrated that the rapidsignalling pathways activated by 1a,25(OH)2D3 involveboth receptors. WT MC3T3-E1 cells exhibited increasedmRNAs for Alpl, Bglap and Spp1. Importantly, Bmp2 andNog were also upregulated. Others have shown thatexpression of Bmp2 can be upregulated by PKC activators,PMA and phorbol ester (Helvering et al., 2000). Given thefact that 1a,25(OH)2D3 causes a rapid increase in PKCin MC3T3-E1 cells (Chen et al., 2010), it is likely that1a,25(OH)2D3 increased Bmp2 expression through PKC-dependent rapid responses.

Reduced expression of either Pdia3 or Vdr blocked thestimulatory effects of the secosteroid on all mRNAs exam-ined, except Spp1. Expression of osteopontin is under theregulation of the vitamin D response element (VDRE)(Nishikawa et al., 1993); thus, it is not surprising thatthe effect on Spp1 was greater in Sh-VDR cells than inSh-Pdia3 cells, but in neither cell line was 1a,25 (OH)2D3-dependent Spp1 expression completely inhibited.Reduced expression of Pdia3 completely abrogatedexpression of mRNAs for osteocalcin. These observationssuggest that Pdia3 action is downstream of VDR forSpp1 but upstream of VDR signalling for Bglap expression,and that at least some of the stimulatory effects of 1a,25(OH)2D3 on osteoblast differentiation may be mediatedby BMP2. Interestingly, neither Runx2 nor Dlx5 was af-fected by 1a,25(OH)2D3, although both have been associ-ated with BMP2 signalling (Lee et al., 2003).

Our results based on 2D cultures demonstrate that 1a,25(OH)2D3 and BMP2 enhance osteoblastic differentiation,as has been shown by others (Jorgensen et al., 2004), butdata also exist showing that BMP2 can act to inhibit 1a,25(OH)2D3-stimulated osteocalcin expression (Kawasakiet al., 1998). BMP2 dose is a critical element in determiningthe outcome. We used a low dose of 50ng/ml, whereas thelatter study used a 10-fold high concentration of 500ng/ml.

By itself, BMP2 had no effect on expression of Pdia3,Vdr, Bmp2, or Nog in WT cells or in either of the silencedcell lines. Similarly, BMP2 did not affect expression ofRunx2, Bglap or Spp1. In contrast, BMP2 caused a smallbut significant increase in Dlx5, and the effect was compa-rable in WT, Sh-Pdia3 and Sh-VDR cells. BMP2 alsocaused a small increase in Alpl in WTcells, but not in alka-line phosphatase activity. Even though BMP2 did not stim-ulate Bglap in WT cells, levels of this mRNA were actuallyreduced in Sh-VDR cells. Taken together, these observa-tions suggest that, under the experimental conditionsused for the 2D cultures, BMP2 exposure alone was insuf-ficient to stimulate osteoblastic differentiation to theextent normally associated with BMP2 action, unless thecells were first primed with 1a,25(OH)2D3.

The synergistic increase in alkaline phosphatase mRNAexpression and activity observed when WT cells wereprimed with 1a,25(OH)2D3 and then treated with BMP2supports this hypothesis. This was most likely due to thesynergistic increase in Vdr expression, indicating thatenhanced response to 1a,25(OH)2D3 precedes enhancedresponse to BMP2. The combination treatment did notaffect expression of Pdia3, Nog, Runx2 or Bmp2 expres-sion. The expression of Dlx5 in the WT cells with the com-bined treatment was comparable to Dlx5 in cells treatedwith BMP2 alone. We did not assess expression ofBMP2 receptor subunits or of components of the BMP2-dependent SMAD signalling pathway, so it is possible thatBmp2 was increased, but the cells might not have beencompetent to respond to endogenously generated protein.

Importantly, Vdr expression was not sensitive to BMP2in the Sh-Pdia3 cells or in the Sh-VDR cells, either whentreated with BMP2 alone or in combination with 1a,25(OH)2D3. Vdr expression is regulated via VDR directly(Zella et al., 2010) and via mechanisms resulting in activa-tion of the ERK1/2 ERK1/2 family of mitogen-activatedprotein kinases (MAPK), such as those due to Pdia3-dependent signalling (Schwartz et al., 2002). Even whenVDREs are present in the promoter, as they are forosteocalcin and osteopontin (Nishikawa et al., 1993;Owen et al., 1990), Pdia3 signalling plays an importantrole in regulating their expression by 1a,25(OH)2D3,either alone or in combination with BMP2.

The loss of a differentiated osteoblast phenotype in Sh-Pdia3 cells may be due in part to its effect on Vdr expres-sion. Reduced levels of Pdia3 blocked the increase in Vdrcaused by the combination treatment and reduced Bmp2expression to levels below baseline, although even base-line levels of Bmp2 were significantly greater in the Sh-Pdia3 cells than in WT. However, reduced levels of VDRalso reduced Bmp2, indicating that the interaction ofthe two signalling pathways is complex. Interestingly,VDR was found to interact with Pdia3 to initiate rapidresponses in fibroblasts (Sequeira et al., 2012).

There appears to be feedback involved as well. To oursurprise, in MC3T3-E1 cells treated with 1a,25(OH)2D3and BMP2, Dlx5 levels in Sh-Pdia3 were increased 0.5-foldover levels in WT cells and 2.7-fold over baseline; inSh-VDR cells Dlx5 was increased by one-fold over WT

Regulation of 3D osteoblast cultures

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 10: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

and 3.4-fold over baseline. This result suggests thatDlx5 is normally suppressed by 1a,25(OH)2D3 viaPdia3- and VDR-dependent pathways. The reverse istrue for alkaline phosphatase. Alpl expression dependson both Pdia3 and VDR, but the stimulatory effect ofthe combination on enzyme activity is mediated byVDR only.

Taken together, our results support the hypothesis thatthere is crosstalk between the signalling pathways acti-vated by each ligand. 1a,25(OH)2D3 activates Pdia3-dependent PKC, PLC and PLA2 signalling in osteoblasts(Boyan et al., 1999; Schwartz et al., 1999; Schwartzet al., 2001), as well as ERK1/2 MAPK (Schwartz et al.,2002). Others have shown that 1a,25(OH)2D3 canenhance BMP2 signalling by increasing the phosphoryla-tion of receptor-regulated Smads through PKCa (Leeet al., 2007; Lee et al., 2006) and MAPKs may also interactwith Smads to modulate the interaction of BMP2 and1a,25(OH)2D3 pathways. In addition, VDR mediates theregulation of osteoblasts, including VDRE-dependentgene expression, aswell as rapid activation of transcaltachiaat the plasma membrane via voltage-gated ion channels,providing another means of modulating rapid responses to1a,25(OH)2D3 and downstream responses to BMP2.

We previously reported WT MC3T3-E1 cells did notexhibit significant mineral deposition in 2D culture, withor without 1a,25(OH)2D3 treatment(Chen et al., 2010).The more physiologically relevant 3D system provides amore robust osteogenic screening method enabling us todetect difference in response to 1a,25(OH)2D3 and BMP2.The chemistry, dynamic nature and media are differentbetween the twomodels, and these can influence the exper-imental outcomes. Therefore, information gained from the2D study does not provide a definitive mechanism forinterpreting the 3D results. The 3D cell culture model en-abled us to examine the interactive effects of 1a,25(OH)2D3 and BMP2 on biomineralization. We used the same ex-perimental design as for the 2D cultures, exposing the cul-tures to a 15min pulse of 1a,25(OH)2D3 at each mediumchange prior to adding BMP2. FBS was present in the cul-tures to ensure that the cells could proliferate. Because ofthe low levels of 1a,25(OH)2D3 in the medium, due tothe FBS, it is not possible to ascribe the effects of thehormone, either alone or in combination with BMP2, torapid responses per se. Even with this shortcoming,however, the study shows definitively that mineral deposi-tion is regulated by 1a,25(OH)2D3 and BMP2 in an interac-tive manner.

By itself, BMP2 treatment increased mineral depositionin WT constructs and this effect was enhanced by thepulse treatment with 1a,25(OH)2D3. The medium usedfor these studies contains b-glycerophosphate, a substrate foralkaline phosphatase. Alkaline phosphatase was upregulatedby both factors, thereby releasing free phosphate andincreasing the potential for calcium phosphate deposition(Hiwada and Wachsmuth, 1974). Interestingly, 1a,25(OH)2D3 alone inhibited mineral deposition in the WTcultures, supporting previous studies showing that 1a,25(OH)2D3 blocks terminal differentiation of osteoblasts

(Ecarot and Desbarats, 1999; Lynch et al., 1995; Slateret al., 1994).

Others have observed enhanced osteogenic potential in2D cultures of osteoblasts from VDR knockout mice (Sooyet al., 2005). In our study, we found that suppression ofeither Pdia3 or VDR resulted in a dysregulation of mineraldeposition in the 3D cultures. In the Sh-Pdia3 constructs,the amount of mineral was greater than in either WTor inSh-VDR constructs at 4weeks and more mineral waspresent in the Sh-VDR constructs at 4weeks than in WTconstructs. At 8weeks, mineral volume in Sh-Pdia3constructs and Sh-VDR constructs was comparableand> 20-fold greater than in WT cultures. The increasein mineral in the Sh-Pdia3 cultures may have been dueto increased Alpl expression and corresponding increasein enzyme activity, resulting in an increase in inorganicphosphate production due to hydrolysis of b-glycerophosphate(Anagnostou et al., 1996). In addition, Sh-Pdia3 cultureslacked Bglap expression and osteocalcin provides aninhibitory control on crystal size during calcium phosphatedeposition in the extracellular matrix (Romberg et al.,1986). Reduced levels of osteocalcin may have played a rolein the Sh-VDR constructs as well. These data suggest that,by depletion of either of the two receptors, osteoblasts canescape from the inhibitory effect of 1a,25(OH)2D3 on theirterminal differentiation and mineralization.

The 20-fold increase in mineralized volume seen in 3Dcultures of Sh-Pdia3 and Sh-VDR cells compared to WTcell constructs was greater than the 10-fold increase inmineral seen in cultures treated with BMP2. This was anunexpected finding. These data suggest that the inhibitionin calcium phosphate deposition due to 1a,25(OH)2D3may be a more dominant regulatory control than BMP2-dependent stimulation of mineral formation. If this isthe case, mineral content of constructs cultured underthe combination regimen should not be greater than inthe constructs treated with the secosteroid alone. This isthe case. At 4weeks in the absence of BMP2, Sh-Pdia3constructs were mineralized to a comparable extent,whether or not they were treated with 1a,25(OH)2D3.Treatment with BMP2 alone or BMP2+1a,25(OH)2D3increased mineral deposition in WT and Sh-VDR cultures,but the amount of mineral did not exceed that seen in Sh-Pdia3 constructs. These relationships were also evident at8weeks. However, 1a,25(OH)2D3 enhanced mineral inWT cultures and decreased it in Sh-VDR cultures. Theseresults provide strong circumstantial evidence that bothreceptors are required for the crosstalk with BMP2 tooccur. Interestingly, in Sh-VDR cells, 1a,25(OH)2D3 hada strong inhibitory effect on mineralization at both 4 and8weeks, unlike what was seen in WT cells. We believethat these data suggest the existence of a 1a,25(OH)2D3-induced VDR independent inhibitory pathway inmineralization.

With regard to BMP2, our data demonstrated that theeffect of BMP2 treatment remained intact in Sh-VDR cells.Therefore, as expected, VDR does not participate in ef-fects induced by BMP2 alone. However, to our surprise,in Sh-Pdia3 cells, not only is the synergistic increase

J. Chen et al.

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 11: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

absent but also the stimulatory effect of BMP2 does nottake place. These results suggest that Pdia3 may be impor-tant to the proper functioning of BMP2 in osteoblastmineralization.

In summary, we found that the combination treatmentof 1a,25(OH)2D3 and BMP2 combined treatment syner-gistically increased osteoblast associated gene expressionin the 2D culture. This effect is mediated through bothVDR and Pdia3, with VDR having a dominant role. Inthe 3D study, WT cells treated with both factors alsoexhibited a synergistic increase in the mineralizedvolume, and a difference in morphology of the mineraldeposition. Silencing of either receptor strongly increasedthe basal level of mineralization. Silencing Pdia3 resultedin no response to both factors, whereas silencing VDRcaused a significant decrease in mineralized volume after1a,25(OH)2D3 treatment, which reveals the differentnature of the two receptors in mediating osteoblast mineral-ization. Based on these results, we believe that 1a,25(OH)2D3 may be a valuable additive to BMP2, which is alreadyused clinically. Moreover, due to the dominating role ofPdia3 and VDR in regulating osteoblast mineralization, theymay serve as potential targets to promote osteogenesis andmineralization for bone tissue-engineering purposes.

Acknowledgements

We would like to thank Dr Rene Olivares-Navarrete, Sharon L.Hyzy, Reyhaan A. Chaudhri and Asia A. Bailey for technicalassistance. This work was supported by the Price Gilbert Jr.Foundation and the US Department of Defense.

Conflict of interest

The authors have declared that there is no conflict ofinterest.

Supporting information on the internet

The following supporting information may be found in theonline version of this article:

Figure S1. Effect of 1α,25(OH)2D3 and BMP2 on themorphology of mineralized matrix at high magnification

Figure S2. Effect of decalcification on the morphology ofmineralized matrix

Figure S3. Effect of 1α,25(OH)2D3 and BMP2 on Ca andP deposition

References

Amling M, Priemel M, Holzmann T et al.1999; Rescue of the skeletal phenotypeof vitamin D receptor-ablated mice inthe setting of normal mineral ionhomeostasis: formal histomorphometricand biomechanical analyses. Endocrino-logy 140: 4982–4987.

Anagnostou F, Plas C, Nefussi JR et al. 1996;Role of bGP-derived Pi in mineralizationvia ecto-alkaline phosphatase in cul-tured fetal calvaria cells. J Cell Biochem62: 262–274.

Bonewald LF, Harris SE, Rosser J et al. 2003;von Kossa staining alone is not sufficientto confirm that mineralization in vitrorepresents bone formation. Calcif TissueInt 72: 537–547.

Bonewald LF, Kester MB, Schwartz Z et al.1992; Effects of combining transforminggrowth factor-b and 1,25-dihydroxyvitaminD3 on differentiation of a humanosteosarcoma (MG-63). J Biol Chem267: 8943–8949.

Boyan BD, Bonewald LF, Paschalis EP et al.2002a; Osteoblast-mediated mineraldeposition in culture is dependent onsurface microtopography. Calcif Tissue Int71: 519–529.

Boyan BD, Bonewald LF, Sylvia VL et al.2002b; Evidence for distinct membranereceptors for 1a,25(OH)2D3 and 24R,25-(OH)2D3 in osteoblasts. Steroids 67: 235–246.

Boyan BD, Sylvia VL, Dean DD et al. 1999;1a,25-(OH)2D3 modulates growth platechondrocytes via membrane receptor-mediated protein kinase C by amechanism that involves changes inphospholipid metabolism and the actionof arachidonic acid and PGE2. Steroids64: 129–136.

Boyan BD, Sylvia VL, McKinney N et al.2003; Membrane actions of vitamin Dmetabolites 1a,25(OH)2D3 and 24R,25(OH)2D3 are retained in growth platecartilage cells from vitamin D recep-tor knockout mice. J Cell Biochem90: 1207–1223.

Breen EC, van Wijnen AJ, Lian JB et al. 1994;In vivo occupancy of the vitamin D respon-sive element in the osteocalcin gene sup-ports vitamin D-dependent transcriptionalupregulation in intact cells. Proc Natl AcadSci USA 91: 12902–12906.

Chen J, Lee CS, Coleman RM et al. 2009; For-mation of tethers linking the epiphysis andmetaphysis is regulated by vitamin dreceptor-mediated signaling. Calcif TissueInt 85: 134–145.

Chen J,Olivares-NavarreteR,WangYet al. 2010;Protein-disulphide isomerase-associated 3(Pdia3) mediates the membrane response to1,25-dihydroxyvitamin D3 in osteoblasts. JBiol Chem 285: 37041–37050.

Doroudi M, Schwartz Z, Boyan BD. 2012;Phospholipase A2 activating protein is re-quired for 1a,25-dihydroxyvitamin D3 de-pendent rapid activation of protein kinaseC via Pdia3. J Steroid Biochem Mol Biol132: 48–56.

Dosier CR, Erdman CP, Park JH et al. 2012;Resveratrol effect on osteogenic differenti-ation of rat and human adipose derivedstem cells in a 3D culture environment. JMech Behav Biomed Mater 11: 112–122.

Ecarot B, Desbarats M. 1999; 1,25-(OH)2D3downregulates expression of Phex, amarker of the mature osteoblast. Endocri-nology 140: 1192–1199.

Erdman CP, Dosier CR, Olivares-Navarrete Ret al. 2012; Effects of resveratrol on

enrichment of adipose-derived stem cellsand their differentiation to osteoblasts intwo- and three-dimensional cultures. J Tis-sue Eng Regen Med 6(suppl 3): s34–46.

Halvorsen YD, Franklin D, Bond AL et al.2001; Extracellular matrix mineralizationand osteoblast gene expression by humanadipose tissue-derived stromal cells. TissueEng 7: 729–741.

Hassan MQ, Tare RS, Lee SH et al. 2006;BMP2 commitment to the osteogeniclineage involves activation of Runx2 byDLX3 and a homeodomain transcrip-tional network. J Biol Chem 281:40515–40526.

Helvering LM, Sharp RL, Ou X et al. 2000;Regulation of the promoters for the hu-man bone morphogenetic protein 2 and 4genes. Gene 256: 123–138.

Hiwada K, Wachsmuth ED. 1974; Catalyticproperties of alkaline phosphatase frompig kidney. Biochem J 141: 283–291.

Holick MF. 2006; Resurrection of vitaminD deficiency and rickets. J Clin Invest116: 2062–2072.

Holick MF. 2007; Vitamin D deficiency. NEngl J Med 357: 266–281.

Huhtakangas JA, Olivera CJ, Bishop JEet al. 2004; The vitamin D receptor ispresent in caveolae-enriched plasmamembranes and binds 1a,25(OH)2-vita-min D3 in vivo and in vitro. MolEndocrinol 18: 2660–2671.

Jorgensen NR, Henriksen Z, Sorensen OHet al. 2004; Dexamethasone, BMP-2, and1,25-dihydroxyvitamin D enhance a moredifferentiated osteoblast phenotype:validation of an in vitro model for humanbone marrow-derived primary osteoblasts.Steroids 69: 219–226.

Regulation of 3D osteoblast cultures

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term

Page 12: Mineralization of three-dimensional osteoblast cultures is enhanced by the interaction of 1 α ,25-dihydroxyvitamin D3 and BMP2 via two specific vitamin D receptors

Kawasaki K, Aihara M, Honmo J et al. 1998;Effects of recombinant human bone mor-phogenetic protein-2 on differentiation ofcells isolated from human bone, muscle,and skin. Bone 23: 223–231.

Kraichely DM, MacDonald PN. 1998; Tran-scriptional activation through the vitaminD receptor in osteoblasts. Front Biosci3: d821–833.

Lee CS, Chen J, Wang Y et al. 2011; Coordi-nated tether formation in anatomically dis-tinct mice growth centers is dependent ona functional vitamin D receptor and istightly linked to three-dimensional tissuemorphology. Bone 49: 419–427.

Lee HJ, Ji Y, Paul S et al. 2007; Activation ofbone morphogenetic protein signaling by aGemini vitamin D3 analogue is mediatedby Ras/protein kinase Ca. Cancer Res67: 11840–11847.

Lee HJ, Wislocki A, Goodman C et al. 2006; Anovel vitamin D derivative activates bonemorphogenetic protein signaling inMCF10 breast epithelial cells. MolPharmacol 69: 1840–1848.

Lee MH, Kim YJ, Kim HJ et al. 2003; BMP-2-induced Runx2 expression is mediated byDlx5, and TGFb1 opposes the BMP-2-induced osteoblast differentiation by sup-pression of Dlx5 expression. J Biol Chem278: 34387–34394.

Li YC, Pirro AE, Amling M et al. 1997;Targeted ablation of the vitamin D recep-tor: an animal model of vitamin D-dependent rickets type II with alopecia.Proc Natl Acad Sci USA 94: 9831–9835.

Lynch MP, Stein JL, Stein GS et al. 1995; Theinfluence of type I collagen on the develop-ment and maintenance of the osteoblastphenotype in primary and passaged ratcalvarial osteoblasts: modification of ex-pression of genes supporting cell growth,adhesion, and extracellular matrix miner-alization. Exp Cell Res 216: 35–45.

Martin JY, Schwartz Z, Hummert TW et al.1995; Effect of titanium surface roughnesson proliferation, differentiation, and pro-tein synthesis of human osteoblast-likecells (MG63). J Biomed Mater Res 29:389–401.

Matsumoto T, Igarashi C, Takeuchi Y et al.1991; Stimulation by 1,25-dihydroxyvitaminD3 of in vitro mineralization induced byosteoblast-like MC3T3-E1 cells. Bone 12:27–32.

Matsumoto T, Kawanobe Y, Morita K et al.1985; Effect of 1,25-dihydroxyvitamin D3on phospholipid metabolism in a clonal

osteoblast-like rat osteogenic sarcoma cellline. J Biol Chem 260: 13704–13709.

Mizwicki MT, Keidel D, Bula CM et al.2004; Identification of an alternativeligand-binding pocket in the nuclearvitamin D receptor and its functionalimportance in 1a,25(OH)2-vitamin D3signaling. Proc Natl Acad Sci USA101: 12876–12881.

Nishikawa J, Matsumoto M, Sakoda Ket al. 1993; Vitamin D receptor zinc fin-ger region binds to a direct repeat as adimer and discriminates the spacingnumber between each half-site. J BiolChem 268: 19739–19743.

Owen TA, Bortell R, Yocum SA et al. 1990;Coordinate occupancy of AP-1 sites in thevitamin D-responsive and CCAAT box ele-ments by Fos-Jun in the osteocalcin gene:model for phenotype suppression oftranscription. Proc Natl Acad Sci USA87: 9990–9994.

Peister A, Deutsch ER, Kolambkar Y et al.2009; Amniotic fluid stem cells producerobust mineral deposits on biodegradablescaffolds. Tissue Eng Part A 15: 3129–3138.

Peister A, Woodruff MA, Prince JJ et al. 2011;Cell sourcing for bone tissue engineering:amniotic fluid stem cells have a delayed,robust differentiation compared to mesen-chymal stem cells. Stem Cell Res 7: 17–27.

Romberg RW, Werness PG, Riggs BL et al.1986; Inhibition of hydroxyapatite crystalgrowth by bone-specific and othercalcium-binding proteins. Biochemistry25: 1176–1180.

Schwartz Z, Ehland H, Sylvia VL et al. 2002;1a,25-dihydroxyvitamin D3 and 24R,25-dihydroxyvitamin D3 modulate growthplate chondrocyte physiology via proteinkinase C-dependent phosphorylation ofextracellular signal-regulated kinase 1/2mitogen-activated protein kinase. Endocri-nology 143: 2775–2786.

Schwartz Z, Gilley RM, Sylvia VL et al. 1999;Prostaglandins mediate the effects of 1,25-(OH)2D3 and 24,25-(OH)2D3 on growthplate chondrocytes in a metabolite-specific and cell maturation-dependentmanner. Bone 24: 475–484.

Schwartz Z, Schlader DL, Ramirez V et al.1989; Effects of vitamin D metabolites oncollagen production and cell proliferationof growth zone and resting zone cartilagecells in vitro. J Bone Miner Res 4: 199–207.

Schwartz Z, Sylvia VL, Liu Y et al. 1998;Treatment of resting zone chondrocyteswith bone morphogenetic protein-2

induces maturation into a phenotype char-acteristic of growth zone chondrocytes bydownregulating responsiveness to 24,25(OH)2D3 and upregulating responsivenessto 1,25-(OH)2D3. Endocrine 9: 273–280.

Schwartz Z, Sylvia VL, Luna MH et al. 2001;The effect of 24R,25-(OH)2D3 on proteinkinase C activity in chondrocytes ismediated by phospholipase D whereasthe effect of 1a,25-(OH)(2)D(3) ismediated by phospholipase C. Steroids66: 683–694.

Sequeira VB, Rybchyn MS, Tongkao-On Wet al. 2012; The role of the vitamin Dreceptor and ERp57 in photoprotection by1a,25-dihydroxyvitamin D3.Mol Endocrinol26: 574–582.

Slater M, Patava J, Mason RS. 1994; Role ofchondroitin sulphate glycosaminoglycansin mineralizing osteoblast-like cells: effectsof hormonal manipulation. J Bone MinerRes 9: 161–169.

Sooy K, Sabbagh Y, Demay MB. 2005; Osteo-blasts lacking the vitamin D receptordisplay enhanced osteogenic potentialin vitro. J Cell Biochem 94: 81–87.

Ulsamer A, Ortuno MJ, Ruiz S et al. 2008;BMP-2 induces Osterix expressionthrough up-regulation of Dlx5 and itsphosphorylation by p38. J Biol Chem283: 3816–3826.

Walters MR, Rosen DM, Norman AW et al.1982; 1,25-Dihydroxyvitamin D receptorsin an established bone cell line. Correla-tion with biochemical responses. J BiolChem 257: 7481–7484.

Wang Y, Chen J, Lee CS et al. 2010;Disruption of Pdia3 gene results in boneabnormality and affects 1a,25-dihydroxy-vitamin D3-induced rapid activationof PKC. J Steroid Biochem Mol Biol121: 257–260.

Yoshizawa T, Handa Y, Uematsu Y et al. 1997;Mice lacking the vitamin D receptorexhibit impaired bone formation, uterinehypoplasia and growth retardation afterweaning. Nat Genet 16: 391–396.

Zanello LP, Norman AW. 2004; Rapidmodulation of osteoblast ion channelresponses by 1a,25(OH)2-vitamin D3 re-quires the presence of a functional vita-min D nuclear receptor. Proc Natl AcadSci USA 101: 1589–1594.

Zella LA, Meyer MB, Nerenz RD et al.2010; Multifunctional enhancers regulatemouse and human vitamin D receptorgene transcription. Mol Endocrinol 24:128–147.

J. Chen et al.

Copyright © 2013 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2013)DOI: 10.1002/term