Bone 45 (2009) 617–626

Contents lists available at ScienceDirect

Bone

j ourna l homepage: www.e lsev ie r.com/ locate /bone

Effect of chitosan particles and dexamethasone on human bone marrow stromal cellosteogenesis and angiogenic factor secretion☆

Jessica Guzmán-Morales a, Hani El-Gabalawy b, Minh H. Pham a, Nicolas Tran-Khanh a, Marc D. McKee c,William Wu d, Michael Centola e, Caroline D. Hoemann a,f,⁎a Department of Chemical Engineering, École Polytechnique, Montréal, QC, Canadab Rheumatic Diseases Research Laboratory, University of Manitoba, Winnipeg, MB, Canadac Faculty of Dentistry, and Department of Anatomy and Cell Biology, McGill University, Montréal, QC, Canadad Bone and Joint Center, Henry Ford Hospital, Detroit MI, USAe Department of Microbiology and Immunology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USAf Institute of Biomedical Engineering, École Polytechnique, Montréal, QC, Canada

☆ Sources of funding: Operating grants from Canadia(CIHR, MOP144440), Natural Sciences and Engineerin(NSERC, 262874-03), and salary support from the FondsQuébec (FRSQ) to JGM, MDM, and CDH. The authors ha⁎ Corresponding author. Department of Chemical Eng

Montréal, QC, Canada H3C 3A7. Fax: +1 514 340 2980.E-mail address: caroline.hoemann@polymtl.ca (C.D.

8756-3282/$ – see front matter © 2009 Elsevier Inc. Adoi:10.1016/j.bone.2009.06.014

a b s t r a c t

a r t i c l e i n f o

Article history:Received 9 April 2009Revised 8 June 2009Accepted 13 June 2009Available online 18 June 2009

Edited by: J. Aubin

Keywords:AngiogenesisBone marrow stromal cellsChitin/chitosanDexamethasoneOsteogenesis

Chitosan is a polysaccharide scaffold used to enhance cartilage repair during treatments involving bonemarrow stimulation, and it is reported to increase angiogenesis and osteogenesis in vivo. Here, we tested thehypotheses that addition of chitosan particles to the media of human bone marrow stromal cell (BMSC)cultures stimulates osteogenesis by promoting osteoblastic differentiation and by favoring the release ofangiogenic factors in vitro. Confluent BMSCs were cultured for 3 weeks with 16% fetal bovine serum, ascorbate-2-phosphate and disodium β-glycerol phosphate, in the absence or presence of dexamethasone, an anti-inflammatory glucocorticoid commonly used as an inducer of BMSC osteoblast differentiation in vitro. Asexpected, dexamethasone slowed cell division, stimulated alkaline phosphatase activity and enhanced matrixmineralization. Added chitosan particles accumulated intra- and extracellularly and, while not affecting mostosteogenic features, they inhibited osteocalcin release to the media at day 14 and interfered with mineralizedmatrix deposition. Interestingly, dexamethasone promoted cell attachment and suppressed the release andactivation of matrix metalloprotease-2 (MMP-2). While chitosan particles had no effect on the release ofangiogenic factors, dexamethasone significantly inhibited (pb0.05 to pb0.0001) the release of vascularendothelial growth factor (VEGF), granulocyte-macrophage colony stimulating factor (GM-CSF), tumornecrosis factor-alpha (TNF-α), interleukins 1β, 4, 6, and 10 (IL-1β, IL-4, IL-6, IL-10), and a host of otherinflammatory factors that were constitutively secreted by BMSCs. These results demonstrate that chitosanparticles alone are not sufficient to promote osteoblast differentiation of BMSCs in vitro, and suggest thatchitosan promotes osteogenesis in vivo through indirect mechanisms. Our data further show that continuousaddition of dexamethasone promotes osteoblastic differentiation in vitro partly by inhibiting gelatinase activityand by suppressing inflammatory cytokines which result in increased cell attachment and cell cycle exit.

© 2009 Elsevier Inc. All rights reserved.

Introduction

Repair of damaged trabecular bone is an essential feature ofarticular cartilage repair strategies involving microdrilling or micro-fracture of the subchondral bone plate [1]. Chitosan is a polysacchar-ide biomaterial composed of glucosamine with variable levels of N-acetyl glucosamine that is biocompatible, cationic and adhesive,biodegradable, and angiogenic when implanted in bleeding wounds

n Institutes of Health Researchg Research Council of Canadade la Recherche sur la Santé duve no conflicts to declare.ineering, École Polytechnique,

Hoemann).

ll rights reserved.

[2]. We previously reported using animal cartilage repair models thatchitosan-stabilized blood clot implants, when solidified in situ overmicrofracture cartilage defects, elicit more trabecular bone and hya-line cartilage repair compared to surgery-only controls [1,3,4]. Otherstudies have reported that insertion of an imidazole-modified chito-san sponge in osteochondral drill holes in sheep condyles led to amore complete bone repair after 20 and 40 days, compared to drilledcontrols [5], and that chitosan powder applied to canine bone frac-tures accelerated repair by approximately 1 week in veterinary prac-tice [6]. Thus, current evidence indicates that chitosan in physicalcontact with bone marrow can stimulate osteogenesis in vivo.

Osteogenesis during fracture repair occurs through endochondralossification under hypoxic conditions, or through new woven bonedeposition in vascularized granulation tissues [7,8]. New woven bonesynthesis takes place through an appositional growth mechanism in

618 J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

which osteoblasts differentiate and assemble on pre-existing lamellarbone, and deposit new layers of a type I collagen-rich extracellularmatrix which subsequently mineralizes. To clarify the role ofbiomaterials in stimulating fracture repair, in vitro osteogenesis assaysusing various primary cell sources can be used; these assays tend tomimic events that occur during new woven bone synthesis [9]. Invitro, osteoblast differentiation was reported to be maintained orenhanced when osteoblast precursors were directly seeded onchitosan-coated culture dishes [10], chitosan sponges [11], or cross-linked chitosan membranes [12]. However, there is currently noevidence that osteoblasts assemble upon, and directly adhere to,chitosan implanted in vivo. Indeed, the cell-based mechanisms bywhich chitosan stimulates new bone growth remain unclear.

The purpose of this study was to further elucidate the molecularand cellular basis of trabecular bone repair by chitosan-glycerolphosphate/blood implants by an in vitro approach. Since the local cellpopulation contributing to trabecular bone repair is bone marrow-derived, we used human bone marrow stromal cells (BMSCs), whichare non-haematopoietic cells of mesenchymal origin with thepotential to differentiate into chondrocyte, adipocyte and osteoblastlineages. BMSCs, in contrast to calvaria-derived osteoblast precursorsor MC3T3-E1 cells, require dexamethasone (Dex) or other osteogenicinducers to undergo osteoblast differentiation in vitro [9,13–17].Biomaterials are frequently tested for osteogenic potential by seedingosteoblast precursors on biomaterial-coated culture dishes or solidbiomaterial substrates; however, the chitosan present in the hybridchitosan-glycerol phosphate/blood clot implants, as we previouslyreported, is in a particulate form [1,18]. Therefore, we tested thehypothesis that exposure of BMSCs to chitosan particles could directlystimulate osteoblast differentiation and matrix mineralization after3 weeks of in vitro culture, a standard endpoint for in vitroosteogenesis assays [9]. Three weeks in vitro correspond to atimepoint in our in vivo rabbit cartilage repair model where chitosanparticles are nearly completely cleared from the wound site, andconsiderable bone remodeling and new woven bone synthesis havetaken place in the microdrill holes beneath the implant [4]. Finally,given that chitosan stimulates the formation of angiogenic granula-tion tissues [2,4,19], we also tested the hypothesis that chitosanstimulates bone repair indirectly by inducing BMSCs to produceangiogenic factors.

Materials and methods

Medical-grade chitosan (80.6% degree of deacetylation, b0.2%w/wprotein, b500 EU/g) was provided by BioSyntech (Laval, QC, Canada).Autoclave-sterile 0.5 mg/mL chitosan HCl (pH 5.6, Mn=176 kDa, andpolydispersity (Mw/Mn) PDI=1.4) and 0.22 μm filter-sterile rhoda-mine isothiocyanate (RITC)-chitosan HCl (pH 5.6, 80.6% DDA,Mn=144 kDa, PDI=1.3, 0.5% mol/mol RITC/chitosan [20]) werestored in aliquots at −80 °C. Dex, disodium β-glycerol phosphate,ascorbate-2-phosphate, 2-hydroxypropyl β-cyclodextrin, chloramineT hydrate, 4-(dimethylamino) benzaldehyde (Ehrlich's Reagent), calfthymus DNA, trans-4-hydroxy-L-proline, gelatin, Alizarin red andalkaline phosphatase substrate kit were purchased from Sigma-Aldrich (Oakville, ON, Canada). Alpha-Minimal Essential Medium(α-MEM) and penicillin–streptomycin were from Invitrogen (Bur-lington, ON, Canada), and fetal bovine serum (FBS) from AtlantaBiologics (Product No. S115500, Lot. CO136; Atlanta, GA, USA). Humanbone marrow stromal cells (BMSCs) were purchased from the TulaneCenter for Gene Therapy (NewOrleans, LA, USA) under institutionally-approved protocols. BMSCs were obtained by iliac crest aspirates fromhealthy consented donors (N=5, 22–27 years old, 1 female, 3 male, 1unknown), and characterized by the supplier at passage 2 using flowcytometry as being over 98% positive for stem cell/endothelial cellmarkers and cell adhesion receptors (CD44, CD90, CD166, CD105, CD49c,CD59, CD147), 9–22%CD184+(CXCR4/SDF-1 receptor), 5–12%CD49b+,

2–19% CD106+, less than 2% positive for hematopoietic markers (CD34,CD36, CD45, CD117, CD14), and weakly positive for HLA-1:ABC. Proteinmultiplex bead array kitswere fromBio-Rad (25-plex, Hercules, CA, USA)and Invitrogen (25 Cytokine 25-Plex AB Bead Kit-HU, and Growth Factor4-PlexABBeadKit-HU, BioSource International). ELISAkits includedMid-Tact human osteocalcin (Biomedical Technologies, Stoughton, MA, USA),human VEGF (DuoKit: 121 and 165 VEGF isoforms) and total MMP-2(R&D Sciences, Cedarlane, Burlington, ON, Canada). Mouse monoclonalanti-CD105 clone SN6h was from Dako (Cedarlane). Purified humanosteonectin/SPARC and mouse anti-human osteonectin antibody werefrom Haematologic Technologies (Essex Junction, VT, USA). Horse-radish peroxidase (HRP)-coupled anti-mouse antibody was fromVector Laboratories (Burlington, ON, Canada), and chemiluminescentreagent from Roche (Lumilight, Mississauga, ON, Canada) or Amersham(ECL Plus, GE Healthcare, Mississauga, ON, Canada).

Osteogenesis assay

Human BMSCs were received as frozen cryovials of P1 cells thatwere thawed and passaged at subconfluency, then seeded subcon-fluently at passage 3 (P3, N=4) or 5 (P5, N=1) in 6, 6-well plates. Atconfluency, one 6-well plate was harvested as the initial culturecondition (day 0), and the remaining plates were cultured for 3 weeksin complete culture media (CCM, 1 plate), mineralizing media (MM, 2plates), or osteogenic media (OSM, 2 plates). CCM consisted, assuggested by the supplier, of α-MEM, 16% FBS, 100 U/mL penicillinand 100 μg/mL streptomycin. MM consisted of CCM with 5 or 10 mMdisodium β-glycerol phosphate (GP), 30 μg/mL L-ascorbic acid-2-phosphate, and 0.26 μg/mL 2-hydroxypropyl β-cyclodextrin, tocontrol for the presence of the carrier in Dex. OSM consisted of MMwith 10 nM (N=3donors) or 100 nMDex (N=2donors). Mediawerechanged twice weekly for 3 weeks using freshly thawed FBS andfreshly prepared filter-sterile ascorbate-2-phosphate. In pilot studies,50 μg/mL, but not 5 μg/mL, chitosan provoked cell apoptosis after3 weeks of bi-weekly administration (unpublished data). Therefore,culture wells were treated by adding 5 μg/mL chitosan particles ateach media change, directly pipetting 10 μL of 0.5 mg/mL liquidchitosan into each mL of media, and swirling to disperse the particleswhich rapidly precipitated after their introduction into the media.RITC-chitosan was added in the same fashion to one well of eachchitosan-treated plate to track chitosan scaffold fate. 4-day condi-tionedmediawere collected at weekly intervals, and stored at−20 °C.After 3 weeks of culture, completely detached cell nodules thatdeveloped in cultures without Dex were blotted off excess media andweighed to obtain wet mass of detached cells. In other experiments,cells were similarly cultured for 2 weeks in 24-well plates.

Confocal microscopy and histology of cultures with fluorescent chitosanscaffold

After 3 weeks of culture, the wells of monolayer cells treated withRITC-chitosan particles, and matching untreated wells, were incu-bated for 30 min at 37 °C in α-MEM containing 1 μg/mL calcein AM(viable green cytosol) and 2 μg/mL Hoechst 33342 (viable cell bluenuclei) (Molecular Probes, Invitrogen, Mississauga, ON, Canada). Thelabeling media were replaced by α-MEM, and live imaging wasperformed with a LSM 510 META Axioplan 2 confocal scanningmicroscope equipped with an apochromat 63×/0.9 NA water-immersion objective (Carl Zeiss, Germany). Following confocalimaging, the monolayers were fixed in 4% paraformaldehyde/100 mM sodium cacodylate pH 7.2, dehydrated through a gradedseries of ethanol solutions, folded, and embedded in LR White acrylicplastic resin. 1-μm-thick sections were stained with von Kossa reagent(5% w/w silver nitrate in water) for phosphate deposition as anindication of mineralization, counter-stained with Toluidine blue, anddigital images acquired with bright-field optics and confocal imaging.

Fig. 1. Evaluation of the human bone marrow stromal cell differentiation state.Representative Western blots of cell extracts (BMSCs) for stem cell marker CD105 atday 0 (D0) and after 3 weeks of culture (A), and cell culture media (CM) for matureosteoblast marker osteonectin (B) at D0 and after 2 weeks of culture in completeculture media (CCM), mineralizing media (MM), or dexamethasone (Dex)-containingosteogenic media (OSM). Some cultures were also treated with 5 μg/mL of chitosanparticles (+chi) at each media change. (C and D) ELISA for quantitative osteocalcinrelease into conditioned media (mean±SD). Grey boxes in C and D represent therange of osteocalcin values (min–max) measured in culture media alone. ⁎pb0.05, MMand OSM vs MM+chi and OSM+chi at day 14.

619J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

In situ staining for alkaline phosphatase and mineral deposition

After 3 weeks of culture, 2 wells from each 6-well plate were fixedfor 1 h in 4% paraformaldehyde/100mM cacodylate pH 7.2, and rinsedin PBS. One well was incubated in the dark with gentle agitation for10 min at room temperature with 5-Bromo-4-Chloro-3-Inolyl Phos-phate/Nitro Blue Tetrazolium (BCIP-NBT) alkaline phosphatase sub-strate, washed twice with PBS containing 20 mM EDTA, and once withPBS. The other well was incubated for 20 min in Alizarin red (1% w/vin ddH2O, brought to pH 4.2 with ammonium hydroxide), then rinsedrepeatedly with ddH2O.

Biochemical analyses

After 3 weeks of culture, 2 wells from each 6-well plate wereaspirated of media and all tissue (monolayer as well as tissue nodules)collected in 1 mL of 4 M guanidine hydrochloride/50 mM Tris–HCl pH7.2 and the cell nuclei were further disrupted by storing at−80 °C.Wedeliberately used guanidine hydrochloride instead of guanidineisothiocyanate extraction buffer, to suppress chitosan solubilizationwhich we previously found to interfere with DNA quantitation [21].Samples were then thawed, vortexed and centrifuged; and the clearedsupernatant were submitted to the Hoechst 33258 fluorescent DNAassay as previously described [21,22]. Samples and standards fromsheared calf thymus were assayed in TEN buffer (10 mM Tris, 150 mMNaCl, 1 mM EDTA pH 7.2) with 0.2 μg/mL Hoechst 33258, using blackFluoroNunc 96-well plates and a Gemini Molecular Dynamicsfluorescent plate reader (Sunnyvale, CA, USA) with excitation360 nm/emission 460 nm. Hydroxyproline determination was per-formed as previously described [21,22]. Briefly, guanidine-insolublepellets were rinsed with 75% ethanol then hydrolyzed at 110 °C in6 N HCl, oxidized with chloramine T, and reacted with Ehrlich'sreagent. Samples were read at OD560 against a standard curve of 0.1to 5 μg/mL trans-4-hydroxy-L-proline. Collagen mass per well wascalculated assuming 12% w/w hydroxyproline/collagen.

Proteomic analysis of osteocalcin, inflammatory and angiogenic factorsin culture media

Conditioned media from day=0, 7, 14, and 21 (N=4 distinctdonors) were analyzed by enzyme-linked immunosorbent assay(ELISA) for VEGF; at day 0, 7, and 14 for osteocalcin; and at day 0and 14 for MMP-2. VEGF levels in conditionedmediawere normalizedfor each donor to express the data as fold-change from day 0.Conditioned media were also collected from P3 to P5 monolayercultures in 24-well plates from 3 distinct donors at day 0 or day 14 andanalyzed by protein multiplex bead array using a Bio-Plex array reader(Bio-Rad Laboratories, Hercules, CA, USA) which uses Luminexfluorescent bead-based technology, and broad sensitivity rangestandards (BioSource International) ranging between 1.95 and32,000 pg/mL as previously described [23].

Western blots for cell differentiation markers

Proteins in guanidine hydrochloride cell extracts were precipitatedin ethanol and resolubilized in 8 M urea to analyze CD105 expression.Conditioned media from cultures of N=4 distinct donors wasanalyzed at day 0 and 14 for osteonectin. Equal volumes of cellextracts or conditionedmediawere separated by 5% or 10% acrylamidesodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) to analyze CD105/endoglin (non-reducing gel) or osteonectin/SPARC (reducing gel), respectively. Positive controls included humanskin fibroblasts (CD105) and 5 ng purified human osteonectin.Proteins were transferred to polyvinylidene fluoride (PVDF) mem-branes (Millipore, MA, USA). Membranes were blocked with 5% w/vmilk powder in TBST (50 mM Tris, 150 mM NaCl, 0.1% Triton X-100) at

50 °C for 1 h, and incubated with 0.3 μg/mL anti-CD105 or 0.4 μg/mLanti-osteonectin, at 4 °C overnight in 0.5% w/vmilk/TBST, followed by0.2 μg/mL (CD105) or 0.5 μg/mL (osteonectin) HRP-conjugated anti-mouse antibodies. Bands were visualized by chemiluminescence.

Gelatin zymography

A previously described method for gelatin zymography [24] wasused to analyze 20 μL of culture media or cell-conditioned media (day0, day 14), and positive controls that included 5 ng of active humanrecombinant MMP-2 (Calbiochem, CA, USA) or conditioned mediafrom human OA chondrocytes stimulated with 5 ng/mL IL-1β.Samples were resolved under non-reducing conditions on 10%polyacrylamide gels co-polymerized with 1 mg/mL gelatin, renaturedin 2.5% Triton X-100 for 30 min at room temperature, incubated for24 h at 37 °C in digestion buffer (50 mM Tris, 0.2 M sodium chloride,

620 J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

5mM calcium chloride, 0.2% Brij, pH 7.6), stainedwith 0.5% CoomassieBlue R-250 and destained for contrast. Proteolysis by gelatinases wasdetected as clear bands against a blue background of stained gelatin.

Statistical analyses

The student t-test was used to analyze the effect of Dex andchitosan on collagen, DNA content, and the mass of detached tissuenodules (N=5 each condition, mean±SD). Statistica's (StatSoft,Tulsa, OK, USA) main effects analysis of variance (ANOVA) with TukeyHSD post-hoc analysis was used to determine the effect of Dex andtime on VEGF (N=4 donors with or without chitosan per timepoint,mean±SEM); and the effect of Dex and chitosan on total MMP-2(N=4 donors, mean±SD) and cytokine levels in conditioned media(N=3donors, mean±SD). The effect of chitosan on osteocalcin levelsat day 14 in conditioned media from MM and OSM cultures (N=4donors, mean±SD) was evaluated by the General Linear Model(GLM). pb0.05 was considered as significant.

Results

Effects of dexamethasone and chitosan particles on osteogenicdifferentiation

To analyze multiple parameters within each culture, we expandedBMSCs to passage 3 or 5 (P3–P5). Since human BMSCs can lose theirdifferentiation potential when expanded in vitro [25], we evaluatedthe state of cell differentiation in our model. At day 0, BMSCs from 4distinct donors expressed both a stem cell marker (CD105/endoglin),

Fig. 2. Osteogenic induction by dexamethasone. In situ staining for alkaline phosphatase (A3 weeks of culture in CCM, MM or OSM, with or without chitosan. OSM enhanced ALP activityparticles had no effect. Results from 2 out of 5 different donors analyzed are shown. White

and low levels of the mature osteoblast marker osteonectin/SPARC(Figs. 1A, B). Serum is known to contain 5–10 ng/mL osteocalcin [26],and osteocalcin levels in conditioned media at day 0 were roughlyequivalent to background levels detected in media alone containing16% FBS (grey boxes, Figs. 1C, D). Altogether, these data suggest thatthe cells at the beginning of our assay consisted in partly differentiatedmesenchymal cells mixedwith a precursor cell population. Expressionlevels of CD105 remained steady during the 3-week culture (Fig. 1A).After 2 weeks of culture, secreted osteonectin was detected in mediaof all cultures (Fig. 1B), except for OSM with 100 nM Dex whereosteonectin was sporadically suppressed (data not shown). Osteocal-cin levels in conditioned media were widely variable among the 4donors, but tended to increase during the first 2 weeks of culture in allmedia conditions (Figs. 1C, D). Among the various differentiationmarkers examined, the sole effect of added chitosan particles was toinhibit osteocalcin release at 2 weeks of culture in mineralizing media(MM) and OSM (p=0.043, Figs. 1C, D).

Dexamethasone (Dex) present in OSM stimulated osteoblastdifferentiation of confluent human BMSCs after 3 weeks of culture,as scored by the upregulation of alkaline phosphatase (ALP) activity,and accumulation of a diffusely stained calcified matrix (Fig. 2). Theintensity of ALP and Alizarin red staining varied according to donor.Dex also inhibited cell proliferation (pb0.05, Fig. 3A) and stimulated aminor increase in insoluble collagen deposition during the 3-weekculture period (Fig. 3B). Added chitosan particles had no effect on cellproliferation, or collagen deposition (Fig. 3). In OSM, chitosansporadically increased ALP activity (1 out of 5 donors, data notshown), without intensifying matrix calcification (OSM vs OSM+chitosan, Fig. 2). Therefore, chitosan particles did not consistently

LP) activity, and mineralized matrix by Alizarin red staining for calcium (Ca++) after, mineralization and monolayer adhesion to the culture dish; while addition of chitosanarrows highlight monolayer detachment in the absence of Dex.

Fig. 3. Cell proliferation and collagen content. Biochemical assays of cell extracts at day0 (D0) and after 3 weeks of culture in CCM, MM and OSM showed accumulation ofDNA (A) and collagen mass measured by hydroxyproline content (B) (mean±SD). InOSM less cell proliferation occurred and collagen content tended to be higher than inmedia without Dex. Chitosan had no influence on DNA or collagen accumulation.⁎pb0.05, MM vs OSM; ^pb0.05, MM+chi vs OSM+chi.

621J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

enhance osteoblastic differentiation either in the absence or presenceof Dex.

After 3 weeks of culture in MM with ascorbate-2-phosphate andβ-glycerol phosphate, BMSCs were spindle-shaped with oval nuclei(Fig. 4A), in contrast to BMSCs cultured with Dex in OSM which weremore rounded (Fig. 4B) indicating an “osteoblast-like” phenotype.Live confocal microscopy revealed that RITC-chitosan particlesaccumulated throughout the monolayers that were comprised oforthogonally organized cells 2 to 5 cell layers thick (data not shown).RITC-chitosan particles were evenly distributed over the cells withno loss of cell viability and no clear alteration in cell morphology(Figs. 4A, B). Chitosan particles accumulated in the extracellularmatrix, and became internalized by BMSCs in MM and OSM (Fig. 4,

Fig. 4. RITC-chitosan particle distribution throughout the BMSC monolayers. Representativ(blue — live cell nuclei) after 3 weeks of culture revealed a spindle-shaped morphologyosteogenic media (OSM, B). RITC-chitosan particles were homogenously distributed and acmagnification 63×/0.9.

arrowheads and arrows). In cultures carried out in OSM, a variablelevel of phosphate mineralization was observed for the differentdonors (data not shown). Less mineral phosphate tended toaccumulate in chitosan-treated monolayers (Fig. 5B vs A), whichwas consistent with a slightly diminished Alizarin red stain (seeDonor 2, Fig. 2). RITC-chitosan particles (Fig. 5C) were mutuallyexclusive of mineral deposits (Fig. 5D). To summarize, the major effectof chitosan particles on in vitro osteogenesis was to suppress theincrease in osteocalcin release, and to interfere with the deposition ofmineralized collagen where extracellular chitosan deposits occurred.

Effects of dexamethasone and chitosan particles on cell attachment andgelatinase activity

BMSCs cultured for 1 to 2 weeks without Dex tended to detachfrom the culture dishes and form tissue nodules, as indicated by thewhite arrows in Fig. 2. After 3 weeks of culture, detached tissuenodules had an average 9 mg of wet mass (Fig. 6A). Dex suppressedcell detachment (pb0.0001), and this effect was not altered by addedchitosan particles (Figs. 2 and 6A).

Gelatinases are collagen fibril-degrading enzymes also termedmatrix metalloproteases (MMP-2 and MMP-9), which were pre-viously found to be expressed by human BMSCs [27]. In 4 BMSC donorcultures, MMP-2 released to conditioned media increased after2 weeks in both CCM and MM, while in OSM with Dex, total MMP-2(pro- and active) release into conditioned media was inhibited(pb0.0001, Fig. 6B). Chitosan particles had no effect on MMP-2secretion in any culture condition (Fig. 6B). As revealed by zymogra-phy, continuous addition of Dex suppressed all gelatinase activity inconditionedmedia, while chitosan had no effect (Fig. 7, lanes 7 and 8).Interestingly, high levels of pro-MMP-9 were detected in media alonecontaining 16% FBS (Fig. 7, lanes 1 and 2). These data allow us toconclude that human BMSCs constitutively expressed MMP-2 andgelatinase-activating convertases that were suppressed by continuousaddition of Dex.

Effect of dexamethasone and chitosan particles on VEGF andinflammatory cytokine release

Since chitosan particles did not directly stimulate osteogenesis invitro, and did not influence cell attachment, we tested the hypothesisthat chitosan indirectly induces osteogenesis by stimulating BMSCs tosecrete the angiogenic factor VEGF. In the absence of Dex, VEGFisoforms 165 and 121 were released to the media at day 0, and theselevels tripled by day 7, remaining elevated through day 21 (Fig. 8A,

e confocal images of cells stained with calcein AM (green — live cells) and Hoechstof BMSCs cultured in mineralizing media (MM, A) and a more rounded shape in

cumulated intra-(arrows) and extracellularly (arrowheads) in BMSC cultures. Original

Fig. 5. Mineralization of BMSCs cultured in OSM in the absence or presence of RITC-chitosan. Histological sections of folded monolayers cultured for 3 weeks in OSM without (A) orwith (B–D) addition of RITC-chitosan followed by von Kossa staining and counterstaining with Toluidine blue. Less mineral accumulated in chitosan-treatedmonolayers than in non-treated cultures as seen in representative bright-field images (B vs A). Extracellular deposits of RITC-chitosan (black arrows) were mutually exclusive of mineralized matrix areas(white arrows) as shown in the merged confocal image (D) of the fluorescent RITC-chitosan in C and the field shown in B. Original magnification A and B: 40×/0.7; C and D: 40×/1.2.

Fig. 6. BMSC monolayer detachment and total MMP-2 release. Significantly less wetmass of monolayer tissue detached from the culture dishes after 3 weeks was found inOSM compared to CCM and MM; chitosan had no effect on these values (A). ELISAquantification of MMP-2 (both pro- and active forms) in cell-conditionedmedia (CM) atday 0 (D0) and day 14 of culture in different media (B), showed that Dex inhibited totalMMP-2 release into media, while chitosan had no effect (mean±SD).

622 J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

CCM and MM). The 3-fold induction in VEGF secretion from day 0 today 7 roughly paralleled a 4-fold increase in DNA content thatoccurred during 3 weeks of culture (Fig. 3A). Continuous addition ofDex suppressed VEGF release between day 7 to 21 (pb0.0001, Fig. 8A,OSM), despite a doubling in DNA content in OSM over 3 weeks ofculture (Fig. 3A). Chitosan particles had no effect on VEGF release byBMSCs (Fig. 8).

In addition to secreting high quantities of VEGF (Fig. 8B), BMSCswere found to constitutively secrete inflammatory factors includingvery high levels of IL-6 and IL-8/CXCL8 (Figs. 9A and B), and lowerquantities of TNF-α, IL-1β, GM-CSF, MIP-1α and β, eotaxin/CCL11, IFN-α and γ, IP-10/CCL10, and a host of interleukins (Figs. 9C–H). Dexcollectively suppressed the secretion of all inflammatory chemokinesand interleukins (pb0.05 to pb0.001, Fig. 9), while chitosan particleshad no influence on the release of these soluble factors. FBSinflammatory factors, if present, had negligible cross-reactivity withthe anti-human antibodies (M, Fig. 9). We only analyzed inflammatoryfactor levels after 2 weeks of culture, however given the constant effectof Dex on suppressing VEGF, it is most probable that inflammatoryfactor levels were also continuously suppressed by Dex throughout the3 week culture. Altogether, our data showed that a primary effect ofdexamethasone in osteogenic cultures was to suppress gelatinaseactivation, inflammatory factor, and angiogenic factor release.

Discussion

Chitosan particles did not promote in vitro osteogenesis of humanBMSCs cultured in mineralizing media, either in the absence or thepresence of Dex. Given that chitosan did not diminish ALP activity orcollagen accumulation, the reduced osteocalcin at 2 weeks inchitosan-treated cultures could potentially be related to increased

Fig. 7. BMSC gelatinase activity. Representative zymogram of media-only, and BMSC 4-day conditioned media at day 0 (D0) or day 14 (D14) in CCM, MM or OSM. BMSCs releasedpro-MMP-2 at D0 and active MMP-2 at D14 in CCM and MM. Dex in OSM inhibited all gelatinases. No effect of chitosan was observed. Conditioned media of IL-1β-stimulatedchondrocytes served as positive control (+C) on a gel run in parallel. Pro-MMP-9 was present in media with 16% FBS.

623J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

osteocalcin binding to the cell monolayer. In cultures with continuousDex, chitosan particles themselves did not mineralize; they accumu-lated intra- and extracellularly where they were space-filling andexcluded the deposition of mineralized collagen fibrils. Repeatedadministration of chitosan particles did result in many chitosan-cellcomplexes (Fig. 4B) that may have sterically interfered with interac-tions between osteoblasts, collagen fibrils, and accessory proteinsneeded for mineral nucleation, maturation or stability.

We only tested one dose of chitosan (5 μg/mL bi-weekly for3 weeks); and given the inhibitory effects seen on osteocalcin releaseand mineralization, we believe it most improbable that higher orlower concentrations of chitosan particles would stimulate in vitroosteogenesis. Our results seem to contradict previous reports thatchitosan-coated petri dishes stimulate osteoblast differentiation [10].However when using biomaterials as cell differentiation scaffolds, itis known that the form and size of the materials used can influencecell behavior [28]. Our data show that cells react differently with

Fig. 8. Vascular endothelial growth factor secretion by BMSCs. VEGF121 and VEGF165levels in conditioned media (CM) were analyzed by (A) ELISA and (B) multiplexproteomic array. Dex inhibited a time-dependent increase in VEGF release to mediaand chitosan particles had no effect. ⁎⁎pb0.005 and ⁎⁎⁎pb0.0001: OSM with andwithout chitosan vs all other culture conditions. Data are shown as mean±SEM (A) ormean±SD (B).

particulates compared to coatings. Given recent data showing thatamine-modified plastic culture dishes that promote cell attachmentalso stimulate osteoblast differentiation [29], it is possible thatchitosan-coated petris stimulate osteoblast differentiation through anon-specific increase in cell adhesion. In this study, chitosan addedin particulate form did not increase attachment of cells to the tissue-culture plastic. Chitosan particles also failed to stimulate the releaseof angiogenic factors. Altogether, these data allow us to reject thehypothesis that chitosan particles promote osteogenesis by directlystimulating BMSCs to differentiate or to release angiogenic factors.

In the present study, ALP activity was induced and a mineralizedmatrix was formed to variable extent in passaged human BMSCmonolayers from 5 different donors, when cultured in the continuouspresence of Dex for 3 weeks. In our assay, osteocalcin and osteonectinlevels measured at 2 weeks in conditioned media were unrelated tomatrix mineralization observed at 3 weeks, although it should benoted that these factors were not normalized to DNA content. It wassuggested before [14,30] that soluble osteocalcin may not be afunctional marker for osteogenic differentiation of Dex-treatedhuman BMSC cultures. This is in contrast to rat osteoblast precursorswhich most frequently show increased osteocalcin in parallel withmineralization [31,32].

Our data further clarify the role of Dex as an inducer of in vitroosteogenesis of human BMSCs. In our cultures, and as previouslyreported [9,16], mineralizing media was insufficient to induce in vitroosteogenesis of BMSCs. BMSCs behave differently from calvarialprimary osteoblast cells and MC3T3-E1 pre-osteoblast cells whichexpress ALP and mineralize in media containing β-glycerophosphateand ascorbic acid without Dex [33]. The role of Dex in BMSCdifferentiation has yet to be fully understood. Our data provide newevidence that Dex promotes osteoblast differentiation in part bysuppressing inflammatory factors (i.e., IL-1β, Fig. 9C), that drivegelatinase activity (Fig. 7), thereby promoting cell-substrate attach-ment and differentiation (Figs. 2 and 6A). Accordingly, Dex can inhibitgelatinase activity in a variety of cell types including rat vascularsmooth muscle cells [34] and rat brain endothelial cells [35]. Thenotion that the inhibition of gelatinase activity promotes osteogenesisis supported by recent studies showing that MC3T3-E1 osteoblastdifferentiation is correlated with low levels of MMP-2 [36], and thatdownregulation of MT1-MMP, a transmembrane metalloproteasecapable of activating gelatinases, is required for mineralization of ratpre-osteoblasts [37]. It should be noted that the use of 16% FBS in ourassay provided large quantities of pro-MMP-9 which may have beenactivated by human BMSCs, leading to higher cell detachment thanpreviously observed in other investigations using the typical 10% FBSconcentration.

Gelatinases also have a role in the regulation of inflammation andneovascularization. MMP-2 may also play a role in the resolution of aninflammatory response [38] andMMP-9 releases VEGF from its matrix

Fig. 9. Inflammatory factor release by BMSCs. The concentration of inflammatory chemokines and interleukins in conditioned media was quantified by multiplex proteomic array atday 0 (D0) or day 14 in MM or OSM in the absence or presence of chitosan. The negative control consisted in culture media alone (M). Asterisks show significant inhibitory effects ofDex (pb0.05 to pb0.0001, OSM vs corresponding value in MM). Data are represented as mean±SD. Added chitosan particles did not affect inflammatory factor release.

624 J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

stores [39]. From our BMSC-conditioned media we were able to detectVEGF121 and VEGF165, both soluble isoforms of VEGF-A, the pre-dominant form of VEGF within developing bone [40]. While bothisoforms are diffusible, VEGF165 can bind to heparin with high affinity[40]. Since Dex inhibited gelatinase activity, it is possible that the low

levels of VEGF seen in Dex-treated cultures were attributable tosequestration in the extracellular matrix. Nonetheless, even if VEGF165remains partly sequestered in the matrix, our results still convincinglyshow that chitosan particles neither promoted nor inhibited solubleVEGF-A release by BMSCs. It was recently shown that VEGF is abun-

625J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

dantly released from human BMSCs, after as many as 10 cell passages[41]. Our results are not consistent with a previous report showingthat Dex stimulates VEGF121 in human BMSCs [42], although oursupplier of BMSC primary cells and the media used for cell expansionwere different from this previous study. Conversely, our data are inagreement with previous studies in which VEGF protein wasdecreased by Dex in rodent osteoblast cell lines [43,44] — an effectwhich has been related to Dex-induced osteonecrosis in vivo [45]. Theimplications of Dex inhibiting VEGF release by BMSCs merit furtherinvestigation, especially given the possibility that VEGF has a directeffect on differentiation, migration, proliferation, and survival ofosteoblast-like cells [46].

Consistent with its properties as an anti-angiogenic and anti-inflammatory drug [47], Dex inhibited BMSC release of severalinflammatory factors including IL-6. It was previously shown thathuman BMSCs express IL-6, IL-11 and leukemia inhibitory factor (LIF)[30,48], and their common receptor gp130 [49,50]. Exogenous LIF ispro-mitogenic [51], and suppresses rat calvarial cell osteogenesis [52].Furthermore, IL-1β and TNF-α were recently reported to inhibitosteogenic differentiation of mouse mesenchymal stem cells [53]. Inthe present study, human BMSCs expressed high levels of IL-6, IL-1β,TNF-α, and a variety of other inflammatory chemokines andinterleukins. In vitro, BMSCs survive during Dex-induced deprivationof mitogenic autocrine factors by virtue of other anti-apoptotic factorsbeing provided as constituents of 15% fetal bovine serum [54,55].Thus, uniform suppression of inflammatory factors by Dex should bebeneficial to in vitro osteogenesis, because cell cycle arrest can occurwithout leading to apoptosis, but instead, to cell differentiation.

In conclusion, our results demonstrate that Dex-induced osteo-genic differentiation of human BMSCs in vitro is related to activitiesthat slow cell division and promote substrate cell attachment. Addedchitosan particles interfered with matrix mineralization withoutaltering cell adhesion or angiogenic/inflammatory factor release byprimary BMSCs. Collectively these data indicate that a directinteraction between chitosan particles and BMSCs is not sufficient topromote osteogenesis. Thus, the chitosan particles present inchitosan-glycerol phosphate/blood implants may promote osteogen-esis in vivo through indirect mechanisms which warrant furtherinvestigation.

Acknowledgments

We thank BioSyntech for supplying chitosan, Dr. Marc Thibault forgenerating confocal bright-field images, and Thuan Nguyen, Gene-viève Lavallée, Lydia Malynowsky andMagdalena Pasierb for technicalassistance. Funding was provided by operating grants from theCanadian Institutes of Health Research (MOP-144440-BME, CDH andHEG) and the National Engineering Sciences and Research Council(262874-03, CDH). Salary support to CDH, MDM and JGM wasprovided by the Fonds de la Recherche en Santé du Québec.

References

[1] Hoemann CD, Sun J, McKee MD, Chevrier A, Rossomacha E, Rivard GE, et al.Chitosan-glycerol phosphate/blood implants elicit hyaline cartilage repairintegrated with porous subchondral bone in microdrilled rabbit defects.Osteoarthr Cartil 2007;15:78–89.

[2] Shigemasa Y, Minami S. Applications of chitin and chitosan for biomaterials.Biotechnol Genet Eng Rev 1996;13:383–420.

[3] Hoemann CD, Hurtig M, Rossomacha E, Sun J, Chevrier A, Shive MS, et al. Chitosan-glycerol phosphate/blood implants improve hyaline cartilage repair in ovinemicrofracture defects. J Bone Joint Surg Am 2005;87:2671–86.

[4] Chevrier A, Hoemann CD, Sun J, Buschmann MD. Chitosan-glycerol phosphate/blood implants increase cell recruitment, transient vascularization and sub-chondral bone remodeling in drilled cartilage defects. Osteoarthr Cartil 2007;15:316–27.

[5] Muzzarelli RA, Mattioli-Belmonte M, Tietz C, Biagini R, Ferioli G, Brunelli MA, et al.Stimulatory effect on bone-formation exerted by a modified chitosan. Biomaterials1994;15:1075–81.

[6] Khanal DR, Choontanom P, Okamoto Y, Minami S, Rakshit SK, ChandrakrachangS, et al. Management of fracture with chitosan in dogs. Indian Vet J 2000;77:1085–9.

[7] Davies JE, Hosseini MM. Histodynamics of endosseous wound healing. In: DaviesJE, editor. Bone engineering. Em Squared Inc: Toronto; 2000. p. 1–14.

[8] Carano RA, Filvaroff EH. Angiogenesis and bone repair. Drug Discov Today 2003;8:980–9.

[9] Hoemann CD, El-GabalawyH, McKeeMD. In vitro osteogenesis assays: influence ofthe primary cell source on alkaline phosphatase activity andmineralization. PatholBiol 2008;57:318–23.

[10] Klokkevold PR, Vandemark L, Kenney EB, Bernard GW. Osteogenesis enhanced bychitosan (poly-N-acetyl glucosaminoglycan) in vitro. J Periodontol 1996;67:1170–5.

[11] Seol YJ, Lee JY, Park YJ, Ku Y, Rhyu IC, Lee SJ, et al. Chitosan sponges as tissueengineering scaffolds for bone formation. Biotechnol Lett 2004;26:1037–41.

[12] Amaral IF, Lamghari M, Sousa SR, Sampaio P, Barbosa MA. Rat bone marrowstromal cell osteogenic differentiation and fibronectin adsorption on chitosanmembranes: the effect of the degree of acetylation. J Biomed Mater Res Part A2005;75:387–97.

[13] Beresford JN, Bennett JH, Devlin C, Leboy PS, Owen ME. Evidence for an inverserelationship between the differentiation of adipocytic and osteogenic cells in ratmarrow stromal cell-cultures. J Cell Sci 1992;102:341–51.

[14] Cheng SL, Yang JW, Rifas L, Zhang SF, Avioli LV. Differentiation of human bone-marrow osteogenic stromal cells in vitro — induction of the osteoblast phenotypeby dexamethasone. Endocrinology 1994;134:277–86.

[15] Gregory CA, Gunn WG, Peister A, Prockop DJ. An Alizarin red-based assay ofmineralization by adherent cells in culture: comparison with cetylpyridiniumchloride extraction. Anal Biochem 2004;329:77–84.

[16] Chang PL, Blair HC, Zhao XC, Chien YW, Chen D, Tilden AB, et al. Comparison of fetaland adult marrow stromal cells in osteogenesis with and without glucocorticoids.Connect Tissue Res 2006;47:67–76.

[17] Coelho MJ, Cabral AT, Fernandes MH. Human bone cell cultures in biocompatibilitytesting. Part I: osteoblastic differentiation of serially passaged human bonemarrow cells cultured in alpha-MEM and in DMEM. Biomaterials 2000;21:1087–94.

[18] Iliescu M, Hoemann CD, Shive MS, Chenite A, Buschmann MD. Ultrastructure ofhybrid chitosan-glycerol phosphate blood clots by environmental scanningelectron microscopy. Microsc Res Tech 2007;71:236–47.

[19] Malette WG, Quigley HJ, Gaines RD, Johnson ND, Rainer WG. Chitosan: a newhemostatic. Ann Thorac Surg 1983;36:55–8.

[20] Ma O, Lavertu M, Sun J, Nguyen S, Buschmann MD, Winnik F, et al. Precisederivatization of structurally distinct chitosans with Rhodamine B isothiocyanate.Carbohydr Polym 2008;72:616–24.

[21] Hoemann CD, Sun J, Chrzanowski V, Buschmann MDA. multivalent assay to detectglycosaminoglycan, protein, collagen, RNA, and DNA content in milligram samplesof cartilage or hydrogel-based repair cartilage. Anal Biochem 2002;300:1–10.

[22] Hoemann C. Molecular and biochemical assays of cartilage components. In: DeCeuninck F, Sabatini M, Pastoureau P, editors. Cartilage and osteoarthritis. Totowa,NJ: Humana Press; 2004. p. 127–56.

[23] Alex P, Szodoray P, Knowlton N, Dozmorov IM, TurnerM, FrankMB, et al. Multiplexserum cytokine monitoring as a prognostic tool in rheumatoid arthritis. Clin ExpRheumatol 2007;25:584–92.

[24] Kleiner DE, Stetler-Stevenson WG. Quantitative zymography: detection ofpicogram quantities of gelatinases. Anal Biochem 1994;218:325–9.

[25] Aslan H, Zilberman Y, Kandel L, Liebergall M, Oskouian RJ, Gazit D, et al. Osteogenicdifferentiation of noncultured immunoisolated bone marrow-derived CD105+cells. Stem Cells 2006;24:1728–37.

[26] Lee AJ, Hodges S, Eastell R. Measurement of osteocalcin. Ann Clin Biochem2000;37:432–46.

[27] Son BR, Marquez-Curtis LA, Kucia M, Wysoczynski M, Turner AR, Ratajczak J,et al. Migration of bone marrow and cord blood mesenchymal stem cells invitro is regulated by stromal-derived factor-1-CXCR4 and hepatocyte growthfactor-c-met axes and involves matrix metalloproteinases. Stem Cells 2006;24:1254–64.

[28] Burdick JA, Vunjak-Novakovic G. Review: engineered microenvironments forcontrolled stem cell differentiation. Tissue Eng Part A 2009;15:205–19.

[29] Curran JM, Chen R, Hunt JA. The guidance of human mesenchymal stem celldifferentiation in vitro by controlled modifications to the cell substrate.Biomaterials 2006;27:4783–93.

[30] Jaiswal N, Haynesworth SE, Caplan AI, Bruder SP. Osteogenic differentiation ofpurified, culture-expanded humanmesenchymal stem cells in vitro. J Cell Biochem1997;64:295–312.

[31] Shalhoub V, Conlon D, Tassinari M, Quinn C, Partridge N, Stein GS, et al.Glucocorticoids promote development of the osteoblast phenotype by selectivelymodulating expression of cell growth and differentiation associated genes. J CellBiochem 1992;50:425–40.

[32] Peter SJ, Liang CR, Kim DJ, Widmer MS, Mikos AG. Osteoblastic phenotype of ratmarrow stromal cells cultured in the presence of dexamethasone, beta-glycerolphosphate, and L-ascorbic acid. J Cell Biochem 1998;71:55–62.

[33] Yohay DA, Zhang J, Thrailkill KM, Arthur JM, Quarles LD. Role of serum in thedevelopmental expression of alkaline phosphatase in MC3T3-E1 osteoblasts. J CellPhysiol 1994;158:467–75.

[34] Pross C, Farooq MM, Angle N, Lane JS, Cerveira JJ, Xavier AE, et al. Dexamethasoneinhibits vascular smooth muscle cell migration via modulation of matrixmetalloproteinase activity. J Surg Res 2002;102:57–62.

[35] Harkness KA, Adamson P, Sussman JD, Davies-Jones GA, Greenwood J, Woodroofe

626 J. Guzmán-Morales et al. / Bone 45 (2009) 617–626

MN. Dexamethasone regulation of matrix metalloproteinase expression in CNSvascular endothelium. Brain 2000;123:698–709.

[36] Bergeron E, Marquis ME, Chretien I, Faucheux N. Differentiation of preosteoblastsusing a delivery system with BMPs and bioactive glass microspheres. J Mater SciMater Med 2007;18:255–63.

[37] Manduca P, Castagnino A, Lombardini D, Marchisio S, Soldano S, Ulivi V, et al. Roleof MT1-MMP in the osteogenic differentiation. Bone 2009;44:251–65.

[38] McQuibban GA, Gong JH, Wong JP, Wallace JL, Clark-Lewis I, Overall CM. Matrixmetalloproteinase processing of monocyte chemoattractant proteins generates CCchemokine receptor antagonists with anti-inflammatory properties in vivo. Blood2002;100:1160–7.

[39] Bergers G, Brekken R, McMahon G, Vu TH, Itoh T, Tamaki K, et al. Matrixmetalloproteinase-9 triggers the angiogenic switch during carcinogenesis. Nat CellBiol 2000;2:737–44.

[40] Evans KD, Oberbauer AM. Spatiotemporal localization of VEGF-A isoforms in themouse postnatal growth plate. Anat Rec (Hoboken) 2008;291:6–13.

[41] Kagiwada H, Yashiki T, Ohshima A, Tadokoro M, Nagaya N, Ohgushi H. Humanmesenchymal stem cells as a stable source of VEGF-producing cells. J Tissue EngRegen Med 2008;2:184–9.

[42] Furumatsu T, Shen ZN, Kawai A, Nishida K, Manabe H, Oohashi T, et al. Vascularendothelial growth factor principally acts as the main angiogenic factor in theearly stage of human osteoblastogenesis. J Biochem 2003;133:633–9.

[43] Li X, Jin L, Cui Q, Wang GJ, Balian G. Steroid effects on osteogenesis throughmesenchymal cell gene expression. Osteoporos Int 2005;16:101–8.

[44] Harada S, Rodan SB, Rodan GA. Expression and regulation of vascular endothelialgrowth factor in osteoblasts. Clin Orthop 1995:76–80.

[45] Yang L, Boyd K, Kaste SC, Kamdem Kamdem L, Rahija RJ, Relling MV. A mousemodel for glucocorticoid-induced osteonecrosis: effect of a steroid holiday.J Orthop Res 2009;27:169–75.

[46] Villars F, Bordenave L, Bareille R, Amedee J. Effect of human endothelial cells on

human bone marrow stromal cell phenotype: role of VEGF? J Cell Biochem2000;79:672–85.

[47] Badruddoja MA, Krouwer HG, Rand SD, Rebro KJ, Pathak AP, Schmainda KM.Antiangiogenic effects of dexamethasone in 9L gliosarcoma assessed by MRIcerebral blood volume maps. Neuro Oncol 2003;5:235–43.

[48] Kim CH, Cheng SL, Kim GS. Effects of dexamethasone on proliferation, activity, andcytokine secretion of normal human bone marrow stromal cells: possiblemechanisms of glucocorticoid-induced bone loss. J Endocrinol 1999;162:371–9.

[49] Taga T, Kishimoto T. gp130 and the interleukin-6 family of cytokines. Annu RevImmunol 1997;15:797–819.

[50] Bellido T, Stahl N, Farruggella TJ, Borba V, Yancopoulos GD, Manolagas SC.Detection of receptors for interleukin-6, interleukin-11, leukemia inhibitory factor,oncostatin M, and ciliary neurotrophic factor in bonemarrow stromal/osteoblasticcells. J Clin Invest 1996;97:431–7.

[51] Evans DB, Gerber B, Feyen JHM. Recombinant human leukemia inhibitory factor ismitogenic for human bone-derived osteoblast-like cells. Biochem Biophys ResCommun 1994;199:220–6.

[52] Malaval L, Gupta AK, Aubin JE. Leukemia inhibitory factor inhibits osteogenicdifferentiation in rat calvaria cell cultures. Endocrinology 1995;136:1411–8.

[53] Lacey DC, Simmons PJ, Graves SE, Hamilton JA. Proinflammatory cytokines inhibitosteogenic differentiation from stem cells: implications for bone repair duringinflammation. Osteoarthr Cartil 2009;17:735–42.

[54] Hung SC, Pochampally RR, Chen SC, Hsu SC, Prockop DJ. Angiogenic effects ofhuman multipotent stromal cell conditioned medium activate the PI3K-Aktpathway in hypoxic endothelial cells to inhibit apoptosis, increase survival, andstimulate angiogenesis. Stem Cells 2007;25:2363–70.

[55] Alonso V, de Gortazar AR, Ardura JA, Andrade-Zapata I, Alvarez-Arroyo MV, EsbritP. Parathyroid hormone-related protein (107–139) increases human osteoblasticcell survival by activation of vascular endothelial growth factor receptor-2. J CellPhysiol 2008;217:717–27.