This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Calcium phosphate-based particles influence osteogenic maturationof human mesenchymal stem cells

L. Saldana a,b,1, S. Sanchez-Salcedo a,c, I. Izquierdo-Barba a,c, F. Bensiamar a,b,L. Munuera a,d, M. Vallet-Regı a,c, N. Vilaboa a,b,*

a Centro de Investigacion Biomedica en Red de Bioingenierıa, Biomateriales y Nanomedicina, CIBER-BBN, Spainb Unidad de Investigacion, Hospital Universitario La Paz, 28046 Madrid, Spain

c Departamento de Quımica Inorganica y Bioinorganica, Facultad de Farmacia, Universidad Complutense de Madrid, 28040 Madrid, Spaind Departamento de Cirugıa, Facultad de Medicina, Universidad Autonoma de Madrid, 28029 Madrid, Spain

Received 29 July 2008; received in revised form 20 November 2008; accepted 24 November 2008Available online 10 December 2008

Abstract

Biphasic calcium phosphates (BCPs) consist of a mixture of hydroxyapatite and b-tricalcium phosphate and are recommended asalternatives or additives to autogenous bone for orthopaedic and dental applications. There is clinical evidence showing particle releasefrom bioceramics, which might impair the ability of human mesenchymal stem cells (hMSC) from bone marrow to proliferate or matureinto a functional osteoblast phenotype. This study analyses the influence of BCP particles and their precursors, calcium-deficient apatite(CDA) particles, on in vitro hMSC behaviour. Both types of particles were efficiently internalized by hMSC. Cell viability, morphologyand actin cytoskeleton reorganization were unaffected by exposure of hMSC to BCP or CDA particles. Direct exposure to BCP particlesimpaired hMSC osteogenic differentiation and bone matrix mineralization to a lesser extent than CDA, as assayed by evaluation of alka-line phosphatase activity, osteopontin secretion and mineralized nodule formation. The ability of bioceramic particles to affect osteogenicmaturation through modification of soluble factors in media was assayed in an in vitro system that avoids direct cell–particle contact.Indirect exposure to CDA particles severely impaired hMSC osteogenic maturation owing to the uptake of Ca2+ from the culture media.Lower textural properties of BCP and the lack of calcium deficiency in its composition prevented Ca2+ uptake, allowing the developmentof a functional osteoblast phenotype.� 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Calcium phosphate; Bioceramics; Particles; Biocompatibility; Mesenchymal stem cells

1. Introduction

Biphasic calcium phosphates (BCPs) consist of a mix-ture of hydroxyapatite (HA) and b-tricalcium phosphate(b-TCP). For regenerative purposes, BCP show betterperformance than HA because of the higher solubility ofthe b-TCP compound, which facilitates subsequent bone

ingrowth in the implant [1]. BCPs are recommended foruse as alternatives or additives to autogenous bone fororthopaedic and dental applications [2,3]. Nowadays,BCPs used in clinical practice are available in dense andporous blocks, granulates and injectable particles withina polymer carrier [4]. Initial osseointegration of implantedbioceramics and ongoing bone ingrowth are initiated byrecruitment of mesenchymal stem cells, as precursors ofosteoblastic lineage, which then mature into fully func-tional, osteoid-producing osteoblasts [5]. Several clinicalstudies have shown particle release during the resorptionprocess of calcium phosphate-based bioceramics [6,7].More recently, it has been reported that the inflammatoryreaction occurring around implanted biphasic bioceramics

1742-7061/$ - see front matter � 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

doi:10.1016/j.actbio.2008.11.022

* Corresponding author. Address: Edificio I+D, Hospital La Paz, Paseode la Castellana 261, 28046 Madrid, Spain. Tel.: +34 912071512.

E-mail address: nvilaboa.hulp@salud.madrid.org (N. Vilaboa).1 Present address: Departamento de Especialidades Medicas, Facultad

de Medicina, Universidad de Alcala, 28871, Alcala de Henares, Madrid,Spain.

Available online at www.sciencedirect.com

Acta Biomaterialia 5 (2009) 1294–1305

www.elsevier.com/locate/actabiomat

Author's personal copy

may be related to micro-particle release due to insufficientsintering [8]. In vitro biocompatibility of calcium phos-phate-based particles on osteoblastic lineage cells has scar-cely been explored. Available data indicate that bioceramicparticles impaired mouse and human osteoblast prolifera-tion [9,10]. The response of human mesenchymal stem cellsas precursors of osteoblasts to particles released frombioceramics may be critical for successful bone regenera-tion. Metallic, polymeric and bioinert ceramic particlesundergo potential adverse effects on viability, proliferationand function of osteoblast precursors [11]. However, theeffects of bioceramic particles on human mesenchymal cellshave remained largely unexamined to date.

Osteoblasts respond directly to changes in Ca2+ concen-tration in bone microenvironment [12–14], and osteoblasticdifferentiation of mesenchymal stem cells is accompaniedby the expression of Ca2+ binding-proteins and Ca2+ incor-poration into the extracellular matrix [15]. Although it isknown that bioactive ceramics modify Ca2+ concentrationin biological media [16], there is a lack of informationaddressing the influence of these particles on mesenchymalstem cell maturation through modifications in the amountof Ca2+ available.

This study reports comparatively on the direct effect ofBCP particles and their precursors, calcium-deficient apatite(CDA), on human mesenchymal stem cell (hMSC) viabilityand their differentiation along the osteoblastic lineage. Italso examines the effect of exposure to both bioceramicparticles, when excluding direct cell–particle contact, onthe ability of hMSC to mature into a functional osteoblastphenotype.

2. Materials and methods

2.1. Particles

CDA powder was synthesized by the controlled crystal-lization method described elsewhere [16–18]. CDA particleswere prepared by aqueous precipitation from solutions of0.6 M (NH4)2HPO4 (Merck) and 0.92 M Ca(NO3)2�4H2O(Sigma, Madrid, Spain) under an air atmosphere. The for-mula of the CDA obtained was Ca10�x (HPO4)x(PO4)6�x

(OH)2�x, where x = 0.73. Finally, CDA was thermally trea-ted at 900 �C for 1 h, giving rise to a BCP. PowderX-ray diffraction (XRD) was carried out with a PhilipsX’Pert diffractometer using Cu Ka radiation. In order todetermine the quantitative phase composition, the XRDpatterns of CDA and BCP were refined by the Rietveldmethod using FullProf software [19]. Previously reportedstructural data for HA [20] and b-TCP [21] were used asan initial model for Rietveld refinements. Powder morphol-ogy was studied by scanning electron microscopy (SEM) ina JEOL 6400 Microscope-Oxford Pentafet super ATWsystem (Jeol, Herts, UK). Particle size distribution wasdetermined by a Malvern Mastersizer Type S (MalvernInstruments, Worcestershire, UK). The sample was ultraso-nicated 20 min before measurements. The specific surface

area was calculated using the BET method from a N2

adsorption isotherm in a ASAP2010 porosimeter (Microm-eritics Norcross, GA, USA). High-resolution transmissionelectron microscopy (HRTEM) studies were performed ina JEOL 3000 FEG electron microscope fitted with a doubletilting goniometer stage (±45o), operating at 300 kV (Cs0.6 mm, resolution 1.7 A). Fourier diffractograms were car-ried out using images of thin parts of a crystal and usingDigital Micrograph (Gatan, Pleasanton, USA). For TEMmeasurements, the sample was crushed in an agate mortar,dispersed in ethanol and deposited on a microgrid. The Ca/P ratio was determined by energy dispersive X-ray spec-trometry (EDS) analyses in an Oxford model ISIS analysercoupled to the TEM microscope. Commercially pure Tiparticles, obtained from Johnson Matthey (Ward Hill,MA, USA), were characterized in the course of previouslyreported work [22]. The mean equivalent circle diameter(ECD) sizes of Ti particles were 3.32 ± 2.39 lm (range 1–15 lm; 89% were lower than 7 lm).

CDA, BCP and Ti particles were weighed and sterilizedby incubation in isopropanol at room temperature, anddried under UV light in a laminar flow hood. Prior to theaddition to the cells, particles were resuspended in theappropriate culture medium (20 mg ml�1) and sonicatedat maximum power for 10 min in a bath sonicater (Bran-sonic 12, Branson Ultrasonidos SAE, Barcelona, Spain).

2.2. Cell culture and treatments

Purified hMSC (CD105+, CD29+, CD44+, CD14�,CD34� and CD45�) were purchased from Cambrex BioScience (Verviers, Belgium) and expanded in a definedmedium (Cambrex Bio Science) consisting of MSC basalmedium and SingleQuots growth supplements containingfetal bovine serum (FBS), L-glutamine and penicillin/strep-tomycin. In order to promote osteoblastic differentiation,hMSC were incubated in osteogenic induction medium(Cambrex Bio Science) consisting of MSC basal mediumand the SingleQuots osteogenic supplements containingFBS, L-glutamine, penicillin/streptomycin, dexamethasone,ascorbate and b-glycerophosphate, for up to 19 days. Toprevent nutrient exhaustion, osteogenic induction mediumwas partially replaced every 3 days with an equal volume offresh medium.

For particle treatments, cells were seeded and incubatedfor 24 h in growth medium. Then, cells were washed withPBS, and medium was replaced with fresh growth or oste-ogenic induction media. Particle suspensions at 20 mg ml�1

were prepared in either growth or osteogenic inductionmedia and were added to cells to achieve doses of 10 or50 ng cell�1, which are equivalent to 0.1 and 0.5 mg ml�1,respectively. For experiments involving microscopic exam-ination, doses of 10 ng cell�1 were assayed, as treatmentwith 50 ng cell�1 of particles interfered with visualizationof the cells. As controls, cells were subjected to the samemanipulations but incubated in the absence of particles.Particles and culture media used in this study contained

L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305 1295

Author's personal copy

levels of endotoxins <0.015 EU mL�1, as assayed by theSigma E-TOXATE assay for detection and semiquantifica-tion of endotoxins (Sigma).

2.3. Cell viability

Cells were seeded at a density of 5000 cells cm�2 in 24-wellplates, in a final volume of 1 ml, and incubated for 1 or 4 daysin the presence or absence of particles. Cell viability wasdetermined using the CellTiter-Glo assay (Promega, Leiden,The Netherlands), which measures intracellular ATP levels.CellTiter-Glo reagent was added to the culture mediumand cell lysis was induced by shaking for 2 min. The lumines-cent signal was quantified using a luminometer (WallacMicrobeta Trilux, Perkin-Elmer, Turkin, Finland).

2.4. Confocal microscopy

Cells were seeded at a density of 5000 cells cm�2 in 8-well chamber slides, in a final volume of 0.4 ml, and incu-bated for 1 or 4 days in the presence or absence of particles.Cell morphology was observed after fixation of cells with asolution of 2.5% glutaraldehyde in PBS using a spectralconfocal microscope (Leica TCS-SP2-AOBS, Leicamicrosystems, Heidelberg, GMBH, Germany). The auto-fluorescent signal of the biological samples due to glutaral-dehyde fixation was collected in the emission ranges495–590 nm (green) after excitation with laser line488 nm. In order to quantify the cell spreading, a total of30 cells randomly selected from five representative imagesper sample were manually outlined, and cell areas weremeasured using ImageJ v1.34 image analysis software. Tovisualize actin filaments, cells were fixed with 4% (w/v)formaldehyde in PBS, permeabilized with 0.1% TritonX-100 in PBS and stained with PBS containing 4 � 10�7

M phalloidine-TRITC (Sigma). Labelled actin was visual-ized under excitation with a laser lane of 561 nm and bycollecting the emission in the range 567–667 nm. Particleswere analysed by reflection under excitation with a laserline of 488 nm and by collecting the emission in the range480–500 nm. As controls, particles were also directly visu-alized by transmitted light. Overlaid images of the samefields showing particles observed by reflection and trans-mitted light were used to compare both methods of detec-tion. The presence of particles within the cells was directlyvisualized by reflection under excitation with a laser line of488 nm. To analyse the presence of particles within thecells, stacks of 0.5-lm optical sections spanning completedcells were recorded, and the confocal images were analysedusing LEICA LCS software, version 2.5 Build 1227. Onceit had been determined that CDA and BCP particles wereinternalized, their sizes were measured. Three-dimensional(3-D) confocal stacks are amenable to morphometric stud-ies by measuring the areas of a manually outlined region ofinterest (ROI) using suitable software, such as that used inthe present work [23,24]. The areas of intracellular reflec-tion regions, which correspond to internalized particles,

were measured by manually outlining a ROI around eachparticle on the projected maximum view of the confocalstack and then using the ‘quantify’ function of the LEICAsoftware LCS. The software is calibrated with the originaldata from the confocal stack collection to measure the realROI area and thereby calculate the corresponding ECD.The ECD was determined by quantitative analysis of repre-sentative images obtained from a minimum of five differentfields of view using the same magnifications. For eachexperimental condition, 30 individual cells were examinedto determine the number of internalized particles per cell.

2.5. Alkaline phosphatase activity, osteopontin secretion and

mineralized nodule formation

Cells were seeded at a density of 3000 cells cm�2 in 6-well plates in a final volume of 3 ml. One set of cells wasincubated in osteogenic medium in the presence or absenceof particles for 5, 12 or 19 days. A second set of cells under-went the same experimental manipulations, but particleswere placed on transwells (Corning, Life Sciences, MA,USA) which allow humoral contact of both cells and par-ticles through a microporous membrane with pore size0.4 lm, avoiding direct particle–cell contact. Cells culturedin growth medium for 5 days under the same experimentalconditions were used as controls. At the end of the incuba-tion periods, media were collected, centrifuged at 1200g

for 10 min, supplemented with a mixture of protease inhib-itors (17.5 lg ml�1 phenylmethylsulfonyl fluoride, 1 lg ml�1

pepstatin A, 2 lg ml�1 aprotinin, 50 lg ml�1 bacitracin, allfrom Sigma) and frozen at �80 �C. Cell layers were washedexhaustively with PBS and extracted with 5 � 10�2 M Tris–HCl pH 8.0, 5 � 10�1 M NaCl and 1% Triton X-100. Alka-line phosphatase (ALP) activity was assayed in cell layersby determining the release of p-nitrophenol from p-nitro-phenylphosphate at 37 �C and a pH of 10.5. Osteopontin(OPN) secretion was quantified in the media using a spe-cific enzyme immunoassay kit (R&D Systems, Abingdon,UK), with a sensitivity of 6 pg ml�1. ALP activity andOPN secretion data were normalized to the total proteinamount in cell layers, measured by the bicinchoninic acid(BCA) protein assay (Pierce, Rockford, IL, USA), usingBSA as standard. The degree of mineralization was deter-mined using Alizarin Red staining. Briefly, cells were fixedwith ethanol and stained with 4 � 10�2 M Alizarin Red indeionized water (adjusted to pH 4.2). Following rinsingwith PBS, the bound stain was eluted with 10% (w/v) cetyl-pyridinium chloride and the absorbance at 562 nm wasmeasured using a spectrofluorimeter. As Alizarin Redstains calcium phosphate-based materials, the degree ofmineralization was only performed in hMSC culturesexposed indirectly to particles.

2.6. Calcium measurement

Cells were seeded at a density of 3000 cells cm�2 in6-well plates, in a final volume of 3 ml, and incubated in

1296 L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305

Author's personal copy

osteogenic medium for 5, 12 or 19 days in either the pres-ence or absence of particles placed on transwells. As a con-trol, cells were cultured in growth media for 5 days andsubjected to the same experimental manipulations. A setof particles were placed in transwells and maintained inthe absence of cells. Ca2+ measurement in culture mediawas evaluated by the Quantichrom Calcium Assay (Bioas-say Systens, Haywasd, USA), which incorporates a phenol-sulphonephthalein dye that forms a stable blue colouredcomplex specifically with free Ca2+. The intensity of thecolour, measured at 612 nm, is directly proportional tothe Ca2+ concentration in the sample. The differencebetween Ca2+ amounts in media exposed or not to parti-

cles, in the absence of cells, corresponded to Ca2+ uptakeby particles. The resulting data were subtracted from thosemeasured in media from cultured cells exposed to particles,obtaining the corresponding values of Ca2+ uptake by cells.

2.7. Statistical analysis

The data are presented as means ± SD of several inde-pendent experiments. The differences between tested bioce-ramic particles were evaluated by an analysis of variancestatistical method. The post hoc test performed was Bon-ferroni’s test. The P-values <0.05 were considered statisti-cally significant.

Fig. 1. Experimental (symbols) and calculated (solid line) powder XRD patterns for CDA and BCP particles. The lower trace is the difference between theobserved and calculated patterns. The vertical lines mark the position of the calculated Bragg peaks for an apatite phase and b-TCP. SEM images andparticle size distribution of CDA and BCP particles.

L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305 1297

Author's personal copy

3. Results

3.1. Characterization of the particles

Fig. 1 displays the XRD patterns, the SEM micrographsand particle size distributions of CDA and BCP particles.CDA exhibited peaks corresponding to apatite phase.After calcination of CDA at 900 �C the characteristic peaksof HA and b-TCP could be detected. The phase composi-tion of BCP powder, calculated by the XRD Rietveld anal-ysis, corresponded to 73.1 wt.% of b-TCP and 26.9 wt.% ofHA. Micrographs showed aggregates of irregularly shapedCDA and BCP particles. The particle size distribution ofCDA particles corresponded to d10 = 1.0, d50 = 3.1 andd80 = 6.0 lm, and BCP corresponded to the valuesd10 = 1.0, d50 = 3.9 and d80 = 7.8 lm. The specific surfaceareas were 70.3 m2 g�1 for CDA and 8.0 m2 g�1 for BCP.

In order to gain more structural information about syn-thesized particles, a thorough HRTEM study was per-formed. The nanometre size range of the CDA samplewas confirmed, as shown in Fig. 2A, with needle-shapedparticles �20 nm long. Nevertheless, in the HRTEM the

measured periodicities as well as the corresponding FT dif-fractogram (Fourier transform) were in agreement with theapatite phase (Fig. 2B). The Ca/P ratio obtained by EDSanalyses had a value of 1.53. Regarding the BCP sample,the low magnification TEM image shows polygonal (b-TCP) and globular (HA) particles that are larger thanCDA particles owing to the calcination process at 900 �C(Fig. 2C). HRTEM images and FT diffractograms corre-sponding to the BCP sample confirm the presence of b-TCP phase with d-spacing at 0.42 and 0.34 nm (Fig. 2D)and apatite phase with d-spacing at 0.81 and 0.34 nm(Fig. 2E) [25,26]. The Ca/P ratios of b-TCP and HA, mea-sured by EDS analyses, are 1.51 and 1.67, respectively.

3.2. Cell viability

In all experimental conditions, hMSC viability increasedfor 1–4 days (Fig. 3). Compared with untreated cells, thepresence of either CDA or BCP particles did not modifycell viability during the observation period at any testeddose. Ti particles decreased this parameter in a dose-depen-dent fashion at both time points.

Fig. 2. Left: HRTEM study of CDA sample. (A) Low-resolution image showing the needle-like shape of the nanoparticles of CDA. (B) Highermagnification image and FT diffractogram showing the typical periodicities of apatite phase. Right: HRTEM study of BCP sample (treated at 900 �C).(C) Low-resolution magnification showing two kinds of particles (polygonal and globular shape). (D) Higher magnification and FT diffractogramcorresponding to polygonal particle showing typical d-spacing of b-TCP phase. (E) Higher magnification and FT diffratogram corresponding to globularparticle showing typical d-spacing of apatite phase.

1298 L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305

Author's personal copy

3.3. Internalization of CDA and BCP particles

A methodological approach to recognizing metallic par-ticles by reflection under excitation with a laser line of488 nm using a confocal microscope was recently described[23]. In order to investigate whether the same methodologycan be used to detect calcium phosphate particles, suspen-sions of CDA or BCP particles were mounted on micros-copy slides. Images of the same fields showing CDA andBCP particles observed by transmitted light (Fig. 4A andB) and reflection (Fig. 4C and D) are overlaid (Fig. 4Eand F), which indicates that bioceramic particles can be dis-tinguished by the reflection method. The intracellular pres-ence of particles in actin-stained cells was determined byvirtually sectioning the 3-D confocal stack along the XZ

and YZ planes (Fig. 4G and H). Reflection areas penetratedinside CDA or BCP particle-treated hMSC and showed athree-dimensional arrangement, as seen in the orthogonalprojections (Fig. 4I and J). The average sizes of internalizedparticles and the number of particles per cell were similar inhMSC exposed to CDA and BCP particles (Table 1).

3.4. Cell morphology and arrangement of actin cytoskeleton

Fig. 5 shows that hMSC exhibited an elongated shape thatwas unaffected by exposure to 10 ng cell�1 of CDA or BCPparticles. Cell spreading was not significantly modified afterculturing cells in the presence of particles. On untreated cells,actin filaments appeared organized in well-defined stressfibres and mostly oriented in a parallel direction followingthe main cellular axis. Actin network organization was notmodified by the presence of CDA or BCP particles (Fig. 6).

3.5. Osteogenic maturation of hMSC directly exposed toparticles

The ALP activity of untreated cells increased whenhMSC were cultured in osteogenic induction medium for

5 days, further increased up to 12 days and thereafterremained constant (Fig. 7). Treatment with particles atdoses of 10 ng cell�1 resulted in a delayed increase in ALPactivity. The delay was more pronounced after treatmentwith CDA particles. Treatment with 50 ng cell�1 of particlesdiminished ALP activity at all tested times. Enzymaticactivity at 12 and 19 days was higher after incubation withBCP than with CDA particles. OPN could not be detected

Fig. 3. Viability of hMSC exposed to particles. Cells were untreated (–) ortreated with 10 ( ) or 50 ng cell�1 ( ) of Ti, CDA or BCP particles for 1 or 4days. Each value represents means ± SD of three independent experiments.*P < 0.05 compared with untreated cells; #P < 0.05 compared with treat-ment with 10 ng Ti cell�1, at the corresponding tested time.

Fig. 4. Internalization of particles into hMSC. Cells were treated with10 ng cell�1 of CDA or BCP particles for 1 day. (A–F) Confocal images ofCDA or BCP particles. (A and B) Show transmitted light images ofparticles, (C and D) show images of particles obtained by reflection(green), and (E and F) show both images overlapping. Bars = 50 lm. (Gand H) Confocal maximum projections showing actin-stained cells (red)and particles detected by reflection (green). Bars = 30 lm. (I and J)Orthogonal projections along XZ (bottom panels) and YZ (right panels)at a random position.

Table 1Characterization of particles internalized into hMSCs exposed to 10 ngcell�1 of CDA or BCP for 1 day.

ECD (lm) Particles cell�1

CDA 5.39 ± 2.03 20 ± 5BCP 4.42 ± 2.46 19 ± 3

ECD of internalized particles and number of particles cell�1 into cellstreated with CDA or BCP particles. Results are expressed as means ± SD.

L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305 1299

Author's personal copy

in media from cells cultured for 5 days in growing or oste-ogenic media. OPN secretion was measurable after cultur-ing cells in osteogenic media for 12 days and increasedover time up to 19 days. Compared with untreated cells,CDA particles totally suppressed OPN secretion after 12days, while BCP diminished it independently of the dose.When the culture period was extended to 19 days, treatmentwith BCP particles diminished OPN secretion. At the sametime, OPN secretion from cells exposed to CDA particleswas only detected when applied at 10 ng cell�1, and mea-sured levels were lower than after incubation with BCP.

3.6. Osteogenic maturation of hMSC indirectly exposed to

particles

The ability of CDA and BCP particles to modulate oste-ogenic maturation of hMSC through modification of cul-

ture medium components was next investigated. To avoiddirect cell–particle contact, particles were placed on trans-wells. Compared with untreated cells, neither ALP activitynor OPN secretion was affected by exposure of cells to BCPparticles (Fig. 8A). Exposure of cells to 50 ng cell�1 ofCDA decreased ALP activity and OPN secretion at alltested times. Moreover, ALP activity also decreased afterexposure to 10 ng cell�1 of CDA for longer periods of 12and 19 days. The degree of mineralization of untreatedcells, evaluated by Alizarin Red staining, increased afterculturing cells in osteogenic media for 12 days and thengreatly increased over time up to 19 days (Fig. 8B). Micro-photographs in Fig. 8B show that calcified matrix forma-tion was unaffected by the presence of BCP particles,while incubation with CDA resulted in a substantiallylower degree of mineralization. Next analysed was whetherCDA particles could affect osteogenic maturation of hMSC

Fig. 5. Morphology of hMSC exposed to particles. Confocal maximum projections showing autofluorescence of fixed cells. Bars = 100 lm. Cells wereuntreated (–) or treated with 10 ng cell�1 of CDA or BCP particles for 1 ( ) or 4 ( ) days. The graph shows the cell area measurements and resultsrepresent means ± SD of three independent experiments.

1300 L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305

Author's personal copy

through modification of extracellular Ca2+ levels. Ca2+

concentration in media was unaffected by BCP particles,while it decreased after incubation with CDA in a dose-dependent manner (Fig. 9). CDA particles removed Ca2+

from media in a time- and dose-dependent manner (Table2). BCP particles did not remove Ca2+ from media underany experimental conditions assayed. Ca2+ uptake byuntreated cells was detected after incubation in osteogenicmedia for 12 days and increased over time (Table 3). Com-pared with untreated cells, exposure to CDA particlesdiminished Ca2+ uptake by cells in a time- and dose-depen-dent manner. The presence of BCP particles did not affectCa2+ uptake by cells, independently of the dose.

4. Discussion

Analysis of retrieval HA-coated prostheses indicatesthat bioactive ceramic dissolution during bone remodellingleads to the release of particles capable of modifying cellu-lar response [27]. In vitro studies using various kinds of cal-cium phosphate powders (b-TCP, HA or commercial andsintered b-dicalcium pyrophosphate) reported a composi-tion-dependent decrease in rat osteoblast growth [10]. Cal-cium phosphate cement (CPC) and b-TCP particles alsohighly diminished rat osteoblast viability [28]. Thus, parti-cles released from BCP substitutes might impair the viabil-ity of adjacent osteoblasts precursors, compromising thesuccess of bone regeneration. Bioceramic particles released

during the resorption of calcium phosphates may bephagocytosed by cells of the monocytic-macrophage line-age, triggering an inflammatory response [29,30]. Indeed,BCP particles implanted in quadriceps muscles of rats wereencapsulated by fibrous tissue infiltrated by numerous mac-rophages and giant cells [31]. In addition to professionalphagocytes, fibroblasts and osteoblasts efficiently internal-ize particles that compromise their cellular functions[23,32,33]. Particle endocytosis mediates the biologicalresponse of hMSC to titanium particles and is associatedwith reduced rates of cellular proliferation [34]. As neitherCDA nor BCP particles affected the viability of hMSC, itcould be speculated that these cells might not internalizebioceramic particles. 3-D confocal stacks are amenable tomorphometric studies by measurement of the areas of man-ually outlined particles using suitable software, such as thatemployed for the present work [22,23]. Confocal sectioningrevealed the presence of internalized CDA and BCP parti-cles in hMSC, and morphometrical analysis indicated thatboth bioceramics were internalized by these cells to sameextent, as indicated by the similar sizes and numbers ofinternalized particles. Interestingly, neither CDA norBCP particles affect the ability of osteoblast precursors toproliferate in an uncommitted stage, which supports theimproved bioreactivity of bioceramics as compared withmetallic particles. Although CDA and BCP particles exhi-bit different crystallochemical, textural and microstructuralcharacteristics, no change was detected associated with

Fig. 6. Actin cytoskeleton of hMSC exposed to particles. Cells were untreated (–) or treated with 10 ng cell�1 of CDA or BCP particles for 1 or 4 days.Confocal maximum projections showing actin-stained cells (red) and particles detected by reflection (green). Bars = 100 lm.

L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305 1301

Author's personal copy

decreased viability upon exposure of hMSC to these parti-cles, as cell morphology and actin cytoskeleton reorganiza-tion were not affected.

Prolonged exposure to phagocytosable metallic particlessubstantially inhibits hMSC differentiation into an osteo-blastic phenotype by reduction of bone matrix protein pro-duction and mineralization [35]. Information regarding theability of bioceramic particles to modulate osteoblastic dif-ferentiation markers remains scarce and far from conclu-sive. While increased levels of ALP production werefound after treatment of rat osteoblasts with CPC particles[28], b-TCP and HA particles highly decreased ALP activ-ity [10]. The data show that direct exposure to bioceramicparticles compromises hMSC differentiation into func-tional osteoblasts. Osteogenic maturation of hMSC wasimpaired after treatment with CDA and BCP particles, asindicated by the reduced levels of ALP activity and OPN

secretion measured in particle-treated hMSC. OPN is asecreted glycoprotein highly expressed in bone matrix thatmodulates the structure of mineralized matrix in vitro [36–39]. The data indicate that OPN secretion is more sensitiveto direct exposure to calcium phosphate-based particlesthan ALP activity. Interestingly, BCP particles affectedthe expression of both osteoblastic markers to a lesserextent than CDA particles. As sizes and numbers of inter-nalized CDA and BCP particles bioceramics were similar,differences could be attributed to changes in their composi-tion, microstructure or surface reactivity. Protein adsorp-tion to several types of HA particles was found tocorrelate directly to surface area [40]. The bioreactivity ofbioceramic particles is also influenced by crystal shape.Compared with spherical or irregular shapes, needle-likecrystals of HA particles were found to stimulate the stron-gest inflammatory response in cultures of human macro-phages [29]. Thus, the higher surface area and needle-likeshapes of CDA crystals may account for diminished osteo-blastic maturation upon direct particle–cell contact.

To investigate the remote effects of particles on thedevelopment of an osteoblastic phenotype, an experimentalapproach that avoids direct cell–particle contact was used.Expression of osteoblastic markers depends on fluctuationsin the extracellular free ionized Ca2+ concentration [13].The data show that BCP particles did not change Ca2+ lev-els in culture media. In this sense, Ca2+ concentration inculture media was unaffected upon dissolution of BCPsamples, and exposure of rat osteoblasts to conditionedmedia containing ionic products from these samples didnot modify rat osteoblast behaviour [41]. In agreementwith these data, indirect exposure of hMSC to BCP parti-cles did not modify the expression of the differentiationmarkers examined in the present work. However, CDAparticles decreased ALP activity, OPN secretion and min-eralized nodule formation in a dose-dependent manner.The methodology employed to determine mineralization,Alizarin Red staining, cannot distinguish between physio-logical and dystrophic calcification. However, mineraliza-tion data correlated well with ALP activity and OPNsecretion results, which supports the idea that the processdetected was mainly physiological.

The reduction in the capability of hMSC to differentiateinto osteoblastic lineage by the presence of CDA particlescan be explained by changes in Ca2+ levels in the culturemedia. Maeno et al. [14] reported that Ca2+ concentrationslower than 80 lg ml�1 decrease the mineralization abilityof mouse osteoblasts. The present data show that Ca2+

content of osteogenic media decreased to lower levels after5 days in contact with CDA particles, which paralleleddecreased ALP activity and OPN secretion. A decrease inCa2+ uptake by cell layers indirectly exposed to CDAwas associated with partial depletion of Ca2+ in culturemedium by particles. The evolution of Ca2+ content in cul-ture medium can be explained as a function of the phasecompositions, structural characteristics and textural prop-erties of particles. Effectively, CDA showed a Ca/P ratio

Fig. 7. ALP activity and OPN secretion of hMSC directly exposed toparticles. Cells were untreated (–) or treated with 10 ( ) or 50 ng cell�1 ( )of CDA or BCP particles, and cultured in growth or osteogenic media forthe indicated culture periods. The data are expressed as percentages of theabsorbance measured on untreated cells cultured for 19 days in osteogenicmedium, which was given the arbitrary value of 100. Each value representsmeans ± SD of three independent experiments. *P < 0.05 compared withtreated cells; #P < 0.05 compared with treatment with 10 ng cell�1;&P < 0.05 compared with treatment with BCP, at the correspondingtested time.

1302 L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305

Author's personal copy

of 1.53 which, in the presence of Ca2+, tends to increase to1.67, which corresponds to stoichiometric HA [42]. More-over, the higher surface area, the lower crystallinity andslightly lower particle size that CDA particles exhibit leadto calcium uptake to grow and nucleate new apatite crys-tals [43–45]. In concordance, Ca2+ depletion is detectedin the culture medium. BCP particles show higher particlesizes and a composition of stoichiometric HA and b-TCPphases without calcium deficiency in their microstructure.Thus, BCP exhibits a lower tendency to grow new apatitecrystals. In addition, b-TCP is a soluble phase that avoidsCa2+ depletion in media [16].

To assess bioceramic biocompatibility, a number ofin vitro studies using these materials in a bulk form havebeen performed with different cell types related to theosteoblastic phenotype. The behaviour of hMSC seededonto calcium phosphate-based scaffolds and cultured onthese substrates has also been investigated [46]. However,hMSC responses to BCP particles have not been carefullyaddressed to date. The present results indicate that directinteractions of hMSC with BCP particles of phagocytosa-ble size down-regulate in vitro osteoblastic maturation toa lesser extent than with CDA particles. In addition to crys-tallochemistry or textural properties, calcium deficiency of

Fig. 8. ALP activity, OPN secretion and mineralization of hMSC indirectly exposed to particles. (A and B) Cells were untreated (–) or treated with 10 ( )or 50 ng cell�1 ( ) of CDA or BCP particles placed in transwells and cultured in growth or osteogenic media for the indicated culture periods. The data areexpressed as percentages of the absorbance measured on untreated cells cultured for 19 days in osteogenic medium, which was given the arbitrary value of100. Each value represents means ± SD of three independent experiments. *P < 0.05 compared with untreated cells, at the corresponding tested time. Themicrophotographs in (B) show Alizarin red staining of hMSC cultured in growth media or in osteogenic media for 19 days in the presence or absence of50 ng cell�1 of CDA or BCP particles.

L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305 1303

Author's personal copy

bioceramic particles seems to be involved in impaired bonematrix maturation and mineralization by the uptake ofavailable Ca2+. Further research will determine whetherintracellular pools of Ca2+ are diminished upon cell inter-nalization of particles from calcium-deficient materials.

Acknowledgements

This work was supported by Grants Nos. MAT2006-12948-C04-02 and MAT2005-01486 from the Ministerio

de Educacion y Ciencia, Grant No. S-0505/MAT/000324from the Comunidad de Madrid and grants from Funda-cion Mutua Madrilena to LM and NV. The authors aregreatly indebted to Dolores Morales (Confocal MicroscopyLaboratory, Universidad Autonoma de Madrid) for excel-lent technical support.

References

[1] Daculsi G. Biphasic calcium phosphate concept applied to artificialbone, implant coating and injectable bone substitute. Biomaterials1998;19:1473–8.

[2] Schwartz C, Liss P, Jacquemaire B, Lecestre P, Frayssinet P. Biphasicsynthetic bone substitute use in orthopaedic and trauma surgery:clinical, radiological and histological results. J Mater Sci Mater Med1999;10:821–5.

[3] Weiss P, Layrolle P, Clergeau LP, Enckel B, Pilet P, Amouriq Y, et al.The safety and efficacy of an injectable bone substitute in dental socketsdemonstrated in a human clinical trial. Biomaterials 2007;28:3295–305.

[4] Daculsi G, Laboux O, Malard O, Weiss P. Current state of the art ofbiphasic calcium phosphate bioceramics. J Mater Sci Mater Med2003;14:195–200.

[5] Ciapetti G, Ambrosio L, Marletta G, Baldini N, Giunti A. Humanbone marrow stromal cells: in vitro expansion and differentiation forbone engineering. Biomaterials 2006;27:6150–60.

[6] Bloebaum RD, Beeks R, Dorr LD, Savory CG, DuPont JA,Hofmann AA. Complications with hydroxyapatite particulate sepa-ration in total hip arthroplasty. Clin Orthop 1994;298:19–26.

[7] Bloebaum RD, Zou L, Bachus KN, Shea KG, Hofmann AA, DunnHK. Analysis of particulates in acetabular components from patientwith osteolysis. Clin Orthop 1997;338:109–18.

[8] Lu J, Blary MC, Vavasseur S, Descamps M, Anselme K, Hardouin P.Relationship between bioceramics sintering and micro-particles-induced cellular damages. J Mater Sci Mater Med 2004;15:361–5.

[9] Alliot-Licht B, Gregoire M, Orly I, Menanteau J. Cellular activity ofosteoblasts in the presence of hydroxyapatite: an in vitro experiment.Biomaterials 1991;12:752–6.

[10] Sun JS, Tsuang YH, Liao CJ, Liu HC, Hang YS, Lin FH. The effectsof calcium phosphate particles on the growth of osteoblasts. J BiomedMater Res 1997;37:324–34.

[11] Goodman SB, Ma T, Chiu R, Ramachandran R, Smith RL. Effects oforthopaedic wear particles on osteoprogenitor cells. Biomaterials2006;27:6096–101.

[12] Duncan RL, Akanbi KA, Farach-Carson MC. Calcium signals andcalcium channels in osteoblastic cells. Semin Nephrol 1998;18:178–90.

[13] Dvorak MM, Siddiqua A, Ward DT, Carter DH, Dallas SL, NemethEF, et al. Physiological changes in extracellular calcium concentrationdirectly control osteoblast function in the absence of calciotropichormones. Proc Natl Acad Sci USA 2004;101:5140–5.

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

[15] Titorencu I, Jinga VV, Constantinescu E, Gafencu AV, Ciohodaru C,Manolescu I, et al. Proliferation, differentiation and characterization ofosteoblasts from human BM mesenchymal cells. Cytotherapy2007;9:682–96.

[16] Sanchez-Salcedo S, Izquierdo-Barba I, Arcos D, Vallet-Regı M. Invitro evaluation of potential calcium phosphate scaffolds for tissueengineering. Tissue Eng 2006;12:279–90.

[17] Vallet-Regı M. Ceramics for Medical applications. J Chem SocDalton Trans 2001;97:108.

[18] Vallet-Regı M. Bone repair and regeneration possibilities. Mat-WissWerkstofftech 2006;37:478–84.

[19] Rodrıguez-Carvajal J. Recent advances in magnetic structure deter-mination by neutron powder diffraction. Physica B 1993;192:55–69.

Fig. 9. Ca2+ concentrations in media of hMSC indirectly exposed toparticles. Cells were untreated (–) or treated with 10 ( ) or 50 ng cell�1 ( )of CDA or BCP particles placed in transwells and cultured in growth orosteogenic media for the indicated culture periods. The data are expressedas concentrations of Ca2+ (lg ml�1) measured on culture supernatants atthe end of the incubation periods. Each value represents means ± SD ofthree independent experiments. *P < 0.05 compared with untreated cells;#P < 0.05 compared with treatment with 10 ng cell�1, at the correspondingtested time.

Table 2Ca uptake (lg) by particles.

10 CDA 50 CDA 10 BCP 50 BCP

Growth ND 80 ± 20 ND NDOst. 5d ND 110 ± 10 ND NDOst. 12d 87 ± 7 232 ± 20* ND NDOst. 19d 234 ± 10# 425 ± 30*,# ND ND

Particles at doses of 10 ng cell�1 (10) or 50 ng cell�1 (50) soaked in growthor osteogenic medium (Ost.) for 5, 12 or 19 days.ND, not detected.* p < 0.05 compared to Ost. 5d.

# p < 0.05 compared to Ost. 12d.

Table 3Ca uptake (lg) by cells.

– 10 CDA 50 CDA 10 BCP 50 BCP

Growth ND ND ND ND NDOst. 56 ND ND ND ND NDOst. 12d 84 ± 7 78 ± 2 35 ± 2* 90 ± 4 88 ± 5Ost. 19d 115 ± 5* 20 ± 1*,# 25 ± 6*,# 118 ± 5* 117 ± 8*

hMSCs cultured in the absence (–) or presence of 10 ng cell�1 (10) or 50 ngcell�1 (50) of particles in growth or osteogenic medium (Ost.) for 5, 12 or19 days.ND, not detected.* p < 0.05 compared to Ost. 12d.

# p < 0.05 compared to (–).

1304 L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305

Author's personal copy

[20] Kay MI, Young RA, Posner AS. Crystal structure of hydroxyapatite.Nature 1964;204:1050–2.

[21] Yashima M, Sakai A, Kamiyama T, Hoshikawa A. Crystal structureanalysis of b-tricalcium phosphate Ca3(PO4)2 by neutron powderdiffraction. J Solid State Chem 2003;175:272–7.

[22] Valles G, Gonzalez-Melendi P, Gonzalez-Carrasco JL, Saldana L,Sanchez-Sabate E, Munuera L, et al. Differential inflammatorymacrophage response to rutile and titanium particles. Biomaterials2006;27:5199–211.

[23] Valles G, Gonzalez-Melendi P, Saldana L, Rodriguez M, Munuera L,Vilaboa N. Rutile and titanium particles differentially affect theproduction of osteoblastic local factors. J Biomed Mater Res A2008;84:324–36.

[24] Gonzalez-Melendi P, Ramırez C, Kumlehn J, Testillano PS, RisuenoMC. 3-D confocal and electron microscopy imaging define thedynamics and mechanisms of diploidisation at early stages of barleymicrospore-derived embryogenesis. Planta 2005;222:47–57.

[25] Richard M, Aguado E, Cottrel M, Dalcusi G. Ultrastructural andelectron diffraction of bone ceramics interfacil zone in coral andbiphasic implants. Calcif Tissue Int 1998;62:437–42.

[26] Leng Y, Chen J, Qu S. TEM study of calcium phosphate precipitationon HA/TCP ceramics. Biomaterials 2003;24:2125–31.

[27] Røkkum M, Reigstad A, Johansson CB. HA particles can be releasedfrom well-fixed HA-coated stems: histopathology of biopsies from 20hips 2–8 years after implantation. Acta Orthop Scand 2002;73:298–306.

[28] Pioletti DP, Takei H, Lin T, Van Landuyt P, Ma QJ, Kwon SY, et al.The effects of calcium phosphate cement particles on osteoblastfunctions. Biomaterials 2000;21:1103–14.

[29] Laquerriere P, Grandjean-Laquerriere A, Jallot E, Balossier G,Frayssinet P, Guenounou M. Importance of hydroxyapatite particlescharacteristics on cytokines production by human monocytes in vitro.Biomaterials 2003;24:2739–47.

[30] Grandjean-Laquerriere A, Tabary O, Jacquot J, Richard D, Frays-sinet P, Guenounou M, et al. Involvement of toll-like receptor 4 in theinflammatory reaction induced by hydroxyapatite particles. Bioma-terials 2007;28:400–4.

[31] Fellah BH, Josselin N, Chappard D, Weiss P, Layrolle P. Inflamma-tory reaction in rats muscle after implantation of biphasic calciumphosphate micro particles. J Mater Sci Mater Med 2007;18:287–94.

[32] Koreny T, Tunyogi-Csapo M, Gal I, Vermes C, Jacobs JJ, Glant TT.The role of fibroblasts and fibroblast-derived factors in periprostheticosteolysis. Arthritis Rheum 2006;54:3221–32.

[33] Annaz B, Hing KA, Kayser M, Buckland T, Di Silvio L. Anultrastructural study of cellular response to variation in porosity inphase-pure hydroxyapatite. J Microsc 2004;216:97–109.

[34] Okafor CC, Haleem-Smith H, Laqueriere P, Manner PA, Tuan RS.Particulate endocytosis mediates biological responses of humanmesenchymal stem cells to titanium wear debris. J Orthop Res2006;24:461–73.

[35] Wang ML, Nesti LJ, Tuli R, Lazatin J, Danielson KG, Sharkey PF,et al. Titanium particles suppress expression of osteoblastic pheno-type in human mesenchymal stem cells. J Orthop Res2002;20:1175–84.

[36] Alford AI, Hankenson KD. Matricellular proteins: extracellularmodulators of bone development, remodeling, and regeneration.Bone 2006;38:749–57.

[37] Yao KL, Todescan Jr R, Sodek J. Temporal changes in matrixprotein synthesis and mRNA expression during mineralized tissueformation by adult rat bone marrow cells in culture. J Bone MinerRes 1994;9:231–40.

[38] Liu YK, Uemura T, Nemoto A, Yabe T, Fujii N, Ushida T, et al.Osteopontin involvement in integrin-mediated cell signaling andregulation of expression of alkaline phosphatase during early differ-entiation of UMR cells. FEBS Lett 1997;420:112–6.

[39] Kojima H, Uede T, Uemura T. In vitro and in vivo effects ofthe overexpression of osteopontin on osteoblast differentiationusing a recombinant adenoviral vector. J Biochem 2004;136:377–86.

[40] Rouahi M, Champion E, Gallet O, Jada A, Anselme K. Physico-chemical characteristics and protein adsorption potential of hydroxy-apatite particles: influence on in vitro biocompatibility of ceramicsafter sintering. Colloids Surf B Biointerfaces 2006;47:10–9.

[41] Valerio P, Pereira MM, Goes AM, Leite MF. The effect of ionicproducts from bioactive glass dissolution on osteoblast proliferationand collagen production. Biomaterials 2004;25:2941–8.

[42] Rodrıguez-Lorenzo LM, Vallet-Regı M. Controlled crystallization ofcalcium phosphate apatites. Chem Mater 2000;12:2460–5.

[43] Frayssisnet P, Rouquet N, Fages J, Durand M, Vidalain PO, BonelG. The influence of sintering temperature on the proliferation offibroblastic cells in contact with HA-bioceramics. J Biomed MaterRes 1997;35:337–47.

[44] Radin SR, Ducheyne P. The effect of calcium phosphate ceramiccomposition and structure on in vitro behaviour. II. Precipitation. JBiomed Mater Res 1993;27:35–45.

[45] Guo Y, Zhou Y, Jia D, Meng Q. Fabrication and in vitrocharacterization of magnetic hydroxycarbonate apatite coatings withhierarchically porous structures. Acta Biomater 2008;4:923–31.

[46] Arinzeh TL, Tran T, Mcalary J, Daculsi G. A comparative study ofbiphasic calcium phosphate ceramics for human mesenchymal stem-cell-induced bone formation. Biomaterials 2005;26:3631–8.

L. Saldana et al. / Acta Biomaterialia 5 (2009) 1294–1305 1305