Inorganic–organic hybrid scaffolds forosteochondral regeneration

Dany J. Munoz-Pinto,1* Rebecca E. McMahon,1* Melissa A. Kanzelberger,1 Andrea C. Jimenez-Vergara,2

Melissa A. Grunlan,3 Mariah S. Hahn1,3

1Department of Chemical Engineering, Texas A&M University, College Station, Texas 778432Department of Materials Science and Engineering, Texas A&M University, College Station, Texas 778433Department of Biomedical Engineering, Texas A&M University, College Station, Texas 77843

Received 11 May 2009; revised 11 August 2009; accepted 26 August 2009Published online 2 February 2010 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.32695

Abstract: Ligament graft failure frequently results frompoor integration of the replacement tissue with associatedbone. Thus, the ability to regenerate the bone-ligamentosteochondral interface would be advantageous in liga-ment reconstruction. At the osteochondral interface, the tis-sue transitions from a bone-like matrix to fibrocartilage.Therefore, a scaffold which promotes a spatially regulatedtransition in cell behavior from osteoblast-like to chondro-cyte-like would be desirable. Previous research indicatesthat addition of inorganic components to organic scaffoldscan enhance the deposition of bone-like matrix by associ-ated osteoblasts. We therefore reasoned that a gradient inthe inorganic content of a hybrid inorganic–organic scaf-fold may induce an osteochondral-like transition in cellphenotype and matrix production. To test this hypothesis,hydrogels were prepared from poly(ethylene glycol) (PEG)and star poly(dimethylsiloxane) (PDMSstar). As anticipated,both the matrix deposition and phenotype of encapsulatedosteoblasts varied with scaffold inorganic content,although the directionality of this modulation was contrary

to expectation. Specifically, osteoblasts appeared to trans-differentiate into chondrocyte-like cells with increasingscaffold inorganic content, as indicated by increased chon-droitin sulfate and collagen type II production and by up-regulation of sox9, a transcription factor associated withchondrocytic differentiation. Furthermore, the depositionof bone-like matrix (collagen type I, calcium phosphate,and osteocalcin) decreased with increasing PDMSstar con-tent. The resistance of the PDMSstar-PEG scaffolds to pro-tein adsorption and/or the changes in gel modulus/meshstructure accompanying PDMSstar incorporation mayunderlie the unexpected increase in chondrocytic pheno-type with increasing inorganic content. Combined, thepresent results indicate that PDMSstar-PEG hybridgels may prove promising for osteochondral regeneration.� 2010 Wiley Periodicals, Inc. J Biomed Mater Res 94A:112–121, 2010

Key words: ligament tissue engineering; osteochondralinterface; inorganic–organic scaffolds

INTRODUCTION

The cruciate ligaments of the human knee with-stand a variety of tensile and torsional forces duringthe course of normal daily routine and athletic activ-ity. Damage can result in pain, joint instability anddysfunction, and eventual degenerative joint disease.1

As such, �150,000 surgical procedures are performedto treat injured anterior cruciate ligaments (ACLs)each year in the United States alone.2 Currently, auto-logous tissue is the preferred graft material for ACL

reconstruction.3 However, the limited supply of auto-logous tissue suitable for grafting and the risk of do-nor site morbidity complicate the use of autologousgrafts.4 Tissue engineering is an alternative approachfor ligament repair that may avoid many of limita-tions associated with autografts.

Although rapid progress has been achieved in lig-ament tissue engineering over the past decade, engi-neered ligaments have, thus far, generally failed toachieve mechanical properties sufficiently similar totheir native counterparts to serve as viable grafts.5 Inaddition, these engineered tissues usually lack theosteochondral interface critical to the appropriatetransfer of load between ligament and bone.5 Asgraft failure frequently results from poor integrationof the replacement tissue with associated bone,6 theability to regenerate the bone-ligament osteochondralinterface would be advantageous in ligament recon-struction.

*These authors contributed equally to this work.Correspondence to: M. S. Hahn; e-mail: mhahn@tamu.

eduContract grant sponsor: Texas Engineering Experimental

Station

� 2010 Wiley Periodicals, Inc.

At the osteochondral interface, the tissue transi-tions from a bone-like matrix to fibrocartilage-likematrix.7 Thus, a scaffold which promotes a spatially-regulated transition in cell behavior from osteoblast-like to chondrocyte-like would be desirable forosteochondral regeneration.8 For instance, a recentosteochondral tissue engineering study employed ascaffold containing a spatial gradient in levels of aretrovirus encoding for Runx2, an osteogenic tran-scription factor.9 This design induced a spatially-graded transdifferentiation of associated dermalfibroblasts into osteoblast-like cells. Similarly, anextensive body of literature indicates that appropri-ately tailoring scaffold inorganic composition (hy-droxyapatite and bioactive glass) can enhance itsosteoconductivity.10–13 Recently, silica-calcium phos-phate (CaP) composite scaffolds were shown toinduce increasing osteoblast alkaline phosphatase ac-tivity14 and osteocalcin expression15 with increasingsilica (decreasing CaP) content. Another work dem-onstrated elevated alkaline phosphatase activity inosteoblasts cultured on gelatin scaffolds of increasingsiloxane content.16 We therefore reasoned that a gra-dient in the siloxane content of a hybrid inorganic–organic scaffold may induce an osteochondral-liketransition in cell phenotype.

To test this hypothesis, we employed hydrogelscomposed of varying ratios of poly(ethylene glycol)(PEG) to star poly(dimethylsiloxane) (PDMSstar).PEG was selected for the organic component sincespatial gradients can readily be fabricated in hydro-gels prepared from diacrylate-derivatized PEG dueto its photoactivity. For instance, PEG hydrogelswith a continuous linear material property gradientcan be generated by simple adaptation of the equip-ment normally used to make gradient polyacryl-amide gels for electrophoresis.17 More complex spa-tial gradients or patterns can be fabricated usingphotolithographic techniques.18–20 PDMS was chosenover other polysiloxanes due to its known biocom-patibility and its widespread use in biomedicalapplications.21 In addition, a star conformation ofPDMS was selected over a linear form in order toreduce potential phase separation between thehydrophobic PDMS and hydrophilic PEG prior tohydrogel polymerization.22

In this study, rat calvarial osteoblasts wereencapsulated in PEG gels of varying PDMSstar con-tent, and the resulting modulation of cell behaviorwas examined. In assessing cell response, biochem-ical and histological analyses were conducted forextracellular matrix (ECM) components associatedwith mature bone (collagen type I, osteocalcin, andCaP) and fibrocartilage [collagen types I and II aswell as chondroitin sulfate proteoglycan (CSPG)].7

To gain insight into the signaling underlyingobserved cell responses, immunostaining for the

chondrogenic transcription factor sox923 was alsoperformed.

METHODS

Polymer synthesis

Synthesis of methacrylate-derivatizedPDMSstar (PDMSstar-MA)

Methacrylate-derivatized star PDMS (PDMSstar-MA) wasprepared via a two step synthetic strategy per a modifica-tion of the methodology validated in Grunlan et al.24 First,silane-terminated PDMSstar (PDMSstar-SiH) was preparedby the acid-catalyzed equilibration of octamethylcyclotetra-siloxane with tetrakis(dimethylsiloxane)silane. The desiredPDMSstar Mw of �14 kDa was achieved by appropriatelysetting the ratio of these two components.24 Briefly, octa-methylcyclotetrasiloxane (30 g, 0.10 mol) and tetrakis(di-methylsiloxane)silane (0.5 g, 1.5 mmol) were combined ina 100 mL round bottom flask and purged with N2. Triflicacid (15 lL) was then added and the reaction was allowedto stir for 1 h at 908C. After cooling, the pH was neutral-ized by combining with MgCO3 (0.5 g) and dichlorome-thane (DCM, 20 mL) and stirring for 2 h. After filtrationthrough a pad of Celite, the volatiles were removed underreduced pressure. In this way, PDMSstar-SiH was obtained(26.3 g, 86% yield) as a colorless liquid and subsequentlycharacterized by 1H NMR, 13C NMR, IR, and gel permea-tion chromatography (GPC).

The silane terminal groups of PDMSstar-SiH were thenconverted into photosensitive moieties by the hydrosilyla-tion reaction of PDMSstar-SiH and allyl methacrylate.25

Briefly, PDMSstar-SiH (6 g, 0.72 mmol) was combined with10 mL of dry toluene in a 100 mL round bottom flask andpurged with N2. After Karstedt’s catalyst (Pt-divinyltetra-methyldisiloxane complex in xylene, 2% Pt; 15 lL) wasadded, the reaction was heated under constant stirring to458C. Allyl methacrylate (0.39 mL, 2.9 mmol) was added tothis solution via an addition funnel over 15 min, afterwhich the reaction was heated to 908C and stirred over-night. Completion of the reaction was confirmed by thedisappearance of the Si��H (�2100 cm21) absorbance inthe IR spectrum. The reaction mixture was decolorized byrefluxing with activated carbon for 12 h. After filtration,the volatile side products were removed under reducedpressure. In this way, methacrylate-derivatized PDMSstar(PDMSstar-MA) was obtained (5.23 g, 82% yield) and sub-sequently characterized by 1H NMR.

Preparation of diacrylate-terminated PEG

Photosensitive acrylate groups were introduced to theterminal ends of linear PEG (Mw �3.4 kDa, Sigma) perestablished protocols.20 Briefly, PEG (10 g, 3.3 mmol) anddry DCM (50 mL) were combined in a 100 mL round bot-tom flask and purged with N2. Triethylamine (0.96 mL, 6.6mmol) was added slowly to the solution followed by the

INORGANIC–ORGANIC HYBRID SCAFFOLDS 113

Journal of Biomedical Materials Research Part A

dropwise addition of acryloyl chloride (1.08 mL, 13.2mmol). The reaction mixture was allowed to stir at roomtemperature overnight. Removal of HCl was accomplishedby washing the mixture with 2M K2CO3 and separatinginto aqueous and organic phases. The organic (DCM)phase was then dried with anhydrous MgSO4, and diacry-late-derivatized PEG (PEG-DA) was precipitated in diethylether and dried under vacuum. In this way, PEG-DA wasobtained and subsequently characterized by 1H NMR.

Synthesis of acrylate-derivatized celladhesion ligand

Cell adhesion peptide RGDS (American Peptide) wasreacted with acryloyl-PEG-N-hydroxysuccinimide (ACRL-PEG-NHS, Mw � 3.4 kDa, Nektar) at a 1:1 molar ratio for2 h in 50 mM sodium bicarbonate buffer, pH 8.5.19,20 Theproduct (ACRL-PEG-RGDS) was purified by dialysis, ly-ophilized, and stored at 2208C until use.

Hydrogel fabrication

Three distinct hydrogel precursor solutions were pre-pared by dissolving desired levels of PDMSstar-MA andPEG-DA in HEPES buffered saline (HBS; 10 mM HEPES,150 mM NaCl, pH 7.4). Each solution contained 10 weightpercent (wt %) total polymer comprised of one of the fol-lowing three wt ratios of PDMSstar-MA to PEG-DA: 0:100,1:99, and 5:95. Ten microliters of photoinitiator consistingof a 30 wt % solution of 2,2-dimethyl-2-phenyl-acetophe-none in N-vinylpyrrolidone was added per milliliter ofprecursor solution. The resulting solutions were mixed byvortex and then filtered (0.22 lm PES membrane, Milli-pore) immediately prior to being pipetted into 0.5-mm-thick transparent rectangular molds. In addition to per-forming a sterilization function, the filtration processserved to create a fine and stable dispersion of hydropho-bic PDMSstar-MA within the aqueous PEG-DA solution.The precursor solutions were polymerized by 2 min expo-sure to longwave UV light (Spectroline, �6 mW/cm2,365 nm).

Hydrogel characterization

The elemental composition, surface hydrophilicity, mod-ulus, mesh size, and equilibrium swelling of preparedhydrogels were characterized using established methods.The hydrogel samples were immersed in media or HBS forat least 24 h before all characterization steps, aside frombulk swelling assessments.

Hydrogel composition

To verify the incorporation of varying levels of PDMSstarinto the fabricated hydrogels, compositional analysis wasperformed using a Kratos Axis Ultra X-ray photoelectronspectrometer (XPS) with a monochromatized Mg Ka

source. Swollen hydrogels were transferred to dIH2O for

24 h and then dried under vacuum. The dried gel discswere then placed onto steel gravity mounts and loadedinto the XPS. Elemental atomic percent compositions wereobtained from survey spectra, which were performed from0 to 1100 eV. High-resolution analyses with pass energiesof 40 eV were performed at a take-off angle of 908. Thebinding energies were referenced to C1s peak at 285.0 eV.The raw data was quantified and analyzed using XPS PeakProcessing software.26

Contact angle and protein adsorption

The dependence of gel surface hydrophilicity and pro-tein adsorption on increased PDMSstar content was eval-uated. Static contact angle measurements were performedon swollen PDMSstar-PEG hydrogels using a CAM-200(KSV Instruments) contact angle measurement systemequipped with an autodispenser, video camera, and drop-shape analysis software. Briefly, 5 lL of dIH2O was dis-pensed onto each hydrogel surface, and the angle of thewater droplet relative to the surface was monitored for aperiod of 2 min. Five separate measurements were per-formed for each hydrogel formulation.

For protein adsorption analyses, four 6-mm diametersamples were harvested from swollen hydrogels of eachformulation. These samples were exposed to a 50 lg/mLsolution of AlexaFluor 488-labeled fibronectin in PBS. Fi-bronectin adsorption was selected for characterizationbased on previous bone tissue engineering studies.10,27 Fol-lowing 1 h at room temperature, the protein solutionswere removed and exchanged with PBS. One hour and12 h after removal of the protein solution, the fluorescenceof the hydrogel surfaces was monitored using a Zeiss Axi-overt 200 microscope and separately using a microplatereader at ex/em 488/532. Prior to each time point mea-surement, the PBS was removed and exchanged with freshPBS. A fibronectin standard curve was used to converteach fluorescence signal to a protein concentration.

Hydrogel swelling

To characterize equilibrium swelling, 8-mm diametersamples were cored from each PEG-DA hydrogel immedi-ately following polymerization and weighed. The sampleswere then transferred to HBS supplemented with 0.05 wt% sodium azide (HBS-azide). After 24 h, samples wereblotted and weighed. The swollen gels were then dried invacuo and their dry weight recorded. As PEG-DA hydro-gels are primarily water, the increase in weight with swel-ling can be directly related to the increase in gel volume

(V) with swelling, i.e., S 5 Vswollen

Vinitial

� �� swollen weight

initial weight

� �. S val-

ues were found to be 1.10 6 0.01, 1.07 6 0.01, and 1.116 0.01 for the 0:100, 1:99, and 5:95 gels, respectively.These ratios were used to calculate the amount of cellsand ACRL-PEG-RGDS to be added to each hydrogel pre-cursor solution so as to ensure the desired postswellingcell density and RGDS concentration. The dry weightmeasures were used to calculate the swelling ratio foreach formulation: R = (swollen weight/dry weight). Thisratio served as an indicator of gel crosslink density.

114 MUNOZ-PINTO ET AL.

Journal of Biomedical Materials Research Part A

Hydrogel mesh size

PEG-DA hydrogel mesh structure cannot be visualizedusing techniques such as conventional scanning electronmicroscopy (SEM).28 Thus, a variety of methods to esti-mate PEG-DA hydrogel mesh size have been developed,including correlations linking measurable quantities, suchas equilibrium hydrogel swelling and PEG-DA Mn, tomesh size.29,30 Although these correlations yield reasonablemesh size estimates for homopolymer hydrogels,29,30 thesecorrelations cannot readily be applied to PDMSstar-PEGhybrid hydrogels. Thus, in this study, hydrogel mesh sizewas characterized via dextran diffusion based on an adap-tation of the methodology of Watkins et al.31

A series of PDMSstar-PEG hydrogels were prepared,each containing 1 mM ACRL-PEG-RGDS postswelling.These gels were allowed to swell overnight at 378C inHBS-azide, after which 8 mm discs were cored from eachsample. FITC-labeled dextran (70 kDa, Sigma) was dis-solved at 0.05 mg/mL in HBS-azide and added at 1 mLper hydrogel disc. Dextran was then allowed to diffuseinto the hydrogels for 24 h at 378C. Each gel disc wasgently blotted and transferred to 1 mL fresh HBS-azide.Dextran that had penetrated into the hydrogels was thenpermitted to diffuse out into the surrounding solution at378C. After 24 h, the fluorescence of the HBS-azide solu-tion surrounding each disc was measured at ex/em 488/532. Dextran standard curves were used to convert eachfluorescence signal to microgram dextran. For each gelsample, these dextran readings were divided by gel weightto yield a quantitative indicator of hydrogel permissivity(C). These permissivity measures were used to estimatethe relative mesh size (n) of each hydrogel (x) as follows:

nx �Cx

C0:100 PDMS-PEO

� �:

Hydrogel mechanical properties

PDMSstar-PEG hydrogels were prepared containing 1mM ACRL-PEG-RGDS postswelling. Three 8-mm discswere cored from each swollen gel and mechanically testedunder unconstrained compression at room temperatureusing an Instron 3342. Following application of a 0.01 Npreload, each hydrogel was subjected to 10 lm cyclic com-pression (�1% cyclic strain) at 1 Hz. The compressivemodulus of each hydrogel was extracted from the result-ing stress-strain data.

Cell encapsulation studies

Cell expansion

Cryopreserved rat calvarial osteoblasts (Dominion Phar-makine) were thawed and expanded at 378C/5% CO2 inDulbecco’s Modified Eagles’ Media (DMEM, Hyclone) sup-plemented with 10% fetal bovine serum (FBS), and 100mU/mL penicillin and 100 mg/L streptomycin (Hyclone).

Cell encapsulation

Osteoblasts at passage 6 were harvested and combined.Precursor solutions containing 10 wt % total polymer wereprepared by dissolving PDMSstar-MA and PEG-DA atratios of 0:100, 1:99, and 5:95 in HBS followed by photoini-tiator. These wt ratios will be used to refer to the varioushydrogel formulations throughout the remainder of thetext. ACRL-PEG-RGDS was added to each precursor solu-tion to achieve 1 mM RGDS post-swelling. Each precursorsolution was sterile-filtered, immediately after which osteo-blasts were added at a postswelling density of �3 3 106

cells/mL. The hydrogel slabs were transferred to Omni-trays (Nunc) fitted with four sterile polycarbonate bars tosimultaneously prevent gel flotation and prevent gel con-tact with the tray bottom. Gels were maintained at 378C/5% CO2 in DMEM supplemented with 10% FBS and 100mU/mL penicillin and 100 mg/L streptomycin. Mediawas changed every 2 days.

Day 3 construct analyses

Following 3 days of culture to allow for completehydrogel equilibration, 8-mm diameter samples werecollected to characterize the initial mechanical proper-ties of the cell-laden gels. These measures were con-ducted as described earlier to assess the impact of cellson initial bulk-average hydrogel material properties(since hydrogel modulus, mesh size, and swelling areinterdependent).32

Day 28 construct biochemical analyses

After 28 days total culture time, a series of 8-mm sam-ples were collected from each hydrogel formulation forbiochemical and histological analyses. Samples harvestedfor biochemical analyses were weighed, flash-frozen in liq-uid nitrogen, and stored at 2808C.

DNA and total collagen biochemical analyses

Hydrogel samples were digested for 72 h at 378C in 1mL of 0.12M NaOH per 0.2 g hydrogel wet weight. Ali-quots of the NaOH hydrolyzed samples (n 5 3–6 per for-mulation) were neutralized and their DNA content deter-mined using the PicoGreen assay (Invitrogen).33 DNAmeasures were translated to cell number using a conver-sion factor of 6.6 pg DNA per cell. Calf thymus DNA(Sigma) served as a standard.

Levels of hydroxyproline were quantified as an indirectmeasure of total collagen. NaOH digested hydrogels (n 5

3–5 per formulation) were further hydrolyzed for 18 h at1108C in 6M HCl. The samples were then dried (Cen-trivap, Labconco) followed by resuspension in dIH2O andreaction with chloramine T and p-dimethylbenzaldehydereagents.34 Sample absorbance was read at 550 nm relativeto that of l-4-hydroxyproline (Sigma). Total collagen con-tent was estimated from measured hydroxyproline using

INORGANIC–ORGANIC HYBRID SCAFFOLDS 115

Journal of Biomedical Materials Research Part A

the collagen type I conversion factor of 0.13 grams hydrox-yproline per gram collagen.35 For each assay, the standardsused were subjected to the same association with PEG-DAand PDMSstar-MA and the same digestion conditions asthe samples.

Calcium and alkaline phosphatase analyses

Hydrogels (n 5 2 per formulation) were transferred to 2mL screw-cap microfuge tubes containing 1 mL of lysisbuffer from the EnzoLyte FDP Alkaline Phosphatase assaykit and 1 mL of 3.2 mm stainless steel beads. Each samplewas homogenized at 4800 rpm in a Bead-Beater homoge-nizer (Biospec) in 10 s cycles with 1 min intermediate cool-ing on ice. Two hundred microliter aliquots of each samplehomogenate were analyzed for alkaline phosphatase activ-ity using EnzoLyte kit reagents. Total calcium wasassessed from 5 lL aliquots of each sample homogenateusing the Calcium CPC liquid color kit (Stanbio).

Day 28 histological analyses

To gain further understanding of the effects of hydro-gel composition on the behavior of encapsulated cells,staining was conducted for ECM components associatedwith mature bone (collagen type I, osteocalcin, and CaP).Similarly, immunostaining for fibrocartilage-associatedECM components (collagen type I, collagen type II, andCSPG) and for chondrogenic transcription factor sox9was carried out. Samples harvested for histological anal-yses were fixed with 10% formalin for 30 min, embeddedin freezing media (Tissue-Tek), and cut into 35 lm sec-tions.

Immunostaining

Immunostaining was conducted according to standardprotocols. In brief, sections were blocked with peroxidase(Biocare Medical) for 30 min followed by Terminator (Bio-

Figure 1. Relative expression of (A) total collagen, (B) collagen type II, (C) collagen type I, and (D) chondroitin sulfateacross hydrogel formulations. *, significantly different from 0:100 gels; f, significantly different from 1:99 gels. Representa-tive images of staining for collagen type I and chondroitin sulfate are shown below the respective graphs. Scale bars applyto each image in the series and equal 50 lm. [Color figure can be viewed in the online issue, which is available atwww.interscience.wiley.com.]

116 MUNOZ-PINTO ET AL.

Journal of Biomedical Materials Research Part A

care Medical) for 10 min. Primary antibody for osteocalcin[FL-95, Santa Crux Biotechnology (SCBT)], collagen type I(Rockland Immunochemicals), collagen type II (Rockland),CSPG (cs-56, SCBT), or sox9 (H-90, SCBT) was thenapplied for 1 h following dilution in HBS. For the sox9antibody, sections were permeabilized before antibodyapplication with PBS containing 0.05% Triton-100X. Foreach antibody and hydrogel sample, 2–3 sections werestained. Bound primary antibody was detected by usingAP-/HRP-conjugated secondary antibodies (JacksonImmunochemicals) followed by application of chromogen(LabVision).

Van Kossa staining

To detect calcium deposits, sections were stained usinga standard van Kossa staining kit (American MasterTechScientific). In brief, rehydrated sections were rinsed withdIH2O, after which a 5% silver nitrate solution wasapplied. Sections were then exposed to full spectrum light

in a humidified chamber for 45 min. After rinsing withdIH2O, sections were exposed to a 5% sodium thiosulfatefor 2.5 min, briefly rinsed, and mounted.

Semiquantitative histological assessments

For intracellular transcription factor sox9, cell countswere conducted to semiquantitatively evaluate immuno-staining results. In addition, as deposited ECM remainedlocalized around the parent cells in each hydrogel formula-tion (Figs. 1 and 2), the relative ECM protein productionamong hydrogel formulations could also be evaluated bycell counts. For each stained section (containing �250cells), these semiquantitative analyses were conducted perestablished protocols.36–40 In brief, staining intensity, di,was recorded for each cell, i, on a scale of 0–3, with 0 5

‘‘no staining’’ and 3 5 ‘‘highest staining intensity amongall sections for that antibody’’. The cumulative staining in-tensity, d, for a given antibody in a particular section wascalculated as: d 5 (

Pdi)/(total cell number). The d values

Figure 2. Relative expression of (A) sox9, (B) alkaline phosphatase, (C) calcium, and (D) osteocalcin levels across hydro-gel formulations. *, significantly different from 0:100 gels; f, significantly different from 1:99 gels. Representative images ofstaining for collagen type I and chondroitin sulfate are shown below the respective graphs. Scale bars apply to each imagein the series and equal 50 lm. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

INORGANIC–ORGANIC HYBRID SCAFFOLDS 117

Journal of Biomedical Materials Research Part A

for all sections of a given hydrogel formulation were aver-aged to yield an overall sample average for each antibody.Immunostained samples were imaged using an Axiovert200 microscope (Zeiss).

Statistical analyses

Data are reported as mean 6 standard deviation. Com-parison of sample means was performed by ANOVA andTukey’s post hoc test (SPSS software), p < 0.05.

RESULTS

In the present work, rat calvarial osteoblasts wereencapsulated in a series of hybrid hydrogels ofincreasing inorganic PDMSstar content. Cell ECMproduction and phenotype were evaluated in all for-mulations after 28 days of culture.

Hydrogel material properties

Hydrogels with a 0:100, 1:99, or 5:95 wt ratio ofPDMSstar to PEG were prepared and their initialcomposition, mesh size, and mechanical propertiescharacterized. Results from these characterizationstudies are summarized in Table I. For each gel for-mulation, the mass percent of Si determined by XPSwas consistent with the value predicted from theknown Mw of both polymers and the wt ratio ofPDMSstar-to-PEG in the precursor solution. Thesedata support the successful incorporation ofPDMSstar into the hydrogel networks.

The equilibrium diffusion of dextran was used toevaluate the mesh size of the 1:99 and 5:95 gels rela-tive to the 0:100 gels (i.e., relative to pure PEGhydrogels). Gel mesh size increased directionallywith increasing PDMSstar content (Table I). In con-

trast, the compressive moduli of the 1:99 (p 5 0.018)and 5:95 (p 5 0.015) hydrogels were significantlylower than that of the 0:100 gel (Table I). Cell pres-ence did not significantly impact observed initialbulk modulus assessments (data not shown), asexpected given the relatively low initial cell densityemployed (�3 3 106 cells/mL).36 Furthermore, theswelling ratios, R, for the 0:100, 1:99, and 5:95 gelswere 8.75 6 0.03, 8.93 6 0.06, and 9.73 6 0.06,respectively. Each of these material property trendswas consistent with a reduced hydrogel crosslinkdensity with increasing PDMSstar content.

32

As scaffold hydrophilicity41 and protein adsorp-tion10,27 have been demonstrated to significantlyimpact osteoblast behavior, the surface hydrophilic-ity of and fibronectin adsorption on each hydrogelwas investigated. In agreement with the litera-ture,26,42 contact angle measures indicated thatPDMSstar-PEG hybrid scaffolds maintained thehydrophilic character of pure PEG gels (Table II).Similarly, the equilibrium (12 h) levels of fibronectinadsorption on the PDMSstar-containing gels werestatistically indistinguishable from those on pure,‘‘non-biofouling’’ PEG hydrogels (i.e., the 0:100PDMSstar-PEG gels, Table II).43

Cell ECM production and phenotype

Net cell density and ECM synthesis were eval-uated in each hydrogel formulation to probe theinfluence of PDMSstar levels on osteoblast behavior.Based on DNA measures, the cell densities in the0:100, 1:99, and 5:95 hydrogels at day 28 were 1.93 6

0.11, 1.93 6 0.25, and 2.15 6 0.19 million cells pergram, respectively. These data indicate that the netcell proliferation and loss over the time course of thestudy was similar across hydrogels and consistentwith previous tissue engineering studies using ‘‘non-degradable’’ PEG-DA gels.44 In contrast, osteoblasttotal collagen production showed marked variationswith hydrogel composition. Specifically, the total col-lagen levels in the 0:100 (p 5 0.049) and 5:95 (p 5

0.006) gels were �2 times greater than in the 1:99hydrogels [Fig. 1(A)].

TABLE IInitial Material Properties of PDMSstar-PEG

Hybrid Hydrogels

PDMSstar toPEG Weight

Ratio

RelativeMesh

Size (n)a,bModulus(kPa)b

Mass Percent Sic

Measured Predicted

0:100 1.0 6 0.1 167 6 8 0.0 0.01:99 1.1 6 0.1 142 6 4d 0.6 0.45:95 1.3 6 0.1d 142 6 11d 2.3 2.0

aTo give insight into the absolute mesh sizes representedby these relative mesh size values, the mesh size of the0:100 PDMSstar-PEG hydrogel was determined to be �14nm per Canal and Peppas.21

bHydrogels contained 1 mM ACRL-PEG-RGDS post-swelling.

cThe mass percent of Si is given relative to the totalmass of Si, C, and O.

dSignificantly different from 0:100 gels, P < 0.05.

TABLE IIPDMSstar-PEG Hydrogel Surface Hydrophilicity

and Protein Adsorption

PDMSstar toPEG Weight

Contact Angle (8)Fibronectin

Adsorption (ng/cm2)

Ratio 1 min 2 min 1 h 12 h

0:100 38.8 6 3.9 35.3 6 4.9 48.9 6 1.1 13.3 6 2.31:99 34.2 6 3.4 32.0 6 2.9 36.6 6 3.0 14.5 6 3.15:95 37.0 6 9.1 34.6 6 8.5 33.0 6 14.7 14.5 6 3.1

118 MUNOZ-PINTO ET AL.

Journal of Biomedical Materials Research Part A

To gain further insight into the observed total col-lagen results, immunostaining was conducted forcollagen type I (associated with mature bone and, toa lesser extent, with fibrocartilage) and for collagentype II (associated with fibrocartilage).7 These histo-logical analyses revealed a modulation of collagentypes I and II with gel formulation. Collagen type IIproduction was significantly higher in the 5:95hydrogels relative to the 0:100 (p 5 0.032) and 1:99(p 5 0.019) gels [Fig. 1(B)]. In contrast, collagen typeI expression appeared to decrease directionally withincreasing hydrogel inorganic content, although thistrend was not statistically significant [Fig. 1(C)].

CSPG and sox9 (associated with fibrocartilage) aswell as alkaline phosphatase, calcium deposits, andosteocalcin (mid-to-late term markers of bone tissueformation) were also assessed. CSPG levels were sig-nificantly higher in the 5:95 gels relative to the 0:100hydrogels [p 5 0.037, Fig. 1(D)]. In addition, sox9expression increased with an increase in hydrogelinorganic content from 0:100 to 5:95 [p 5 0.011, Fig.2(A)]. Although no significant differences in alkalinephosphatase expression were noted among formula-tions [Fig. 2(B)], van Kossa staining results revealedthe levels of deposited calcium to be markedlyhigher in the 0:100 PDMSstar-PEG gels than in the1:99 (p < 0.001) and 5:95 (p < 0.001) gels [Fig. 2(C)].This decrease in calcium deposition with increasinghydrogel inorganic content was further reflected inthe results from the CPC assay, which yielded 0.546 0.13, 0.20 6 0.14, and 0.18 6 0.05 ng calcium percell for the 0:100, 1:99, and 5:95 gels, respectively.Similarly, osteocalcin levels were reduced in the5:95 gels relative to both the 0:100 (p 5 0.008) and1:99 (p 5 0.027) hydrogels [Fig. 2(D)].

DISCUSSION

This study was designed to test the hypothesisthat a gradient in the siloxane content of a hybridinorganic–organic scaffold would induce a transitionin cell behavior from osteoblast-like to chondrocyte-like. Both the matrix deposition and phenotype ofencapsulated osteoblasts were indeed modulated byvarying scaffold inorganic content. In terms of cellECM production, increasing hydrogel PDMSstar con-tent was associated with increased production of col-lagen type II and CSPG. This increase in the synthe-sis of fibrocartilage-like ECM components wasaccompanied by a decrease in cell production ofbone-like ECM. Specifically, both osteocalcin and cal-cium deposition decreased with increasing inorganiccontent. To further evaluate the modulation of cellphenotype with scaffold composition, the relativeexpression of sox9, a transcription factor associated

with chondrocytic differentiation, was examined.Consistent with the trends in fibrocartilage-like ma-trix production, sox9 levels increased with increasingPDMSstar levels. Combined, the present results sug-gest that encapsulated osteoblasts were transdiffer-entiating into chondrocyte-like cells in the higherinorganic content hydrogels.

These trends in osteoblast response are contraryto those expected based on the literature, whichcollectively supports the idea of heightened osteoin-ductivity with increasing scaffold silica/siloxanecontent.14,15,45,46 It is possible that the observed dis-crepancies with previous literature arise from theabsence of calcium or phosphate in the initialPDMSstar-PEG hydrogels. For instance, Song et al.16

observed a significant increase in the alkaline phos-phatase levels produced by MC3T3-E1 cells culturedon gelatin-siloxane scaffolds relative to pure gelatincontrols only when the gelatin-siloxane hybrids wereprepared in the presence of CaCl2. However, Ninget al.14 observed increased osteoblast alkaline phos-phatase activity with an increase in the silica levels ofsilica-CaP composite scaffolds despite an associateddecrease in scaffold CaP content. Similarly, the high-est silica (lowest CaP) content scaffold examined byGupta et al.15 stimulated greater cell collagen type Iand osteocalcin expression than hydroxyapatite(Ca10(PO4)6(OH)2). Thus, the impact of the absence ofCaP in the initial PDMSstar-PEG gels is unclear.

Another potential source of the observed discrep-ancies between the current results and previous liter-ature is differences in the levels, identities, and ori-entations of scaffold absorbed proteins. It is nowwell established that cell responses to a particularscaffold depend not only on the initial scaffold com-position but also on the scaffold pretreatment,hydrophilicity, and resultant protein adsorp-tion.10,12,41,47 Both the contact angle and fibronectinadsorption profiles for the PDMSstar-PEG scaffoldssuggest that these hybrid scaffolds maintain the ‘‘bio-logical blank slate’’ character of pure PEG gels, inagreement with the literature.26,42,43 The term ‘‘bio-logical blank slate’’ refers that the fact that theadsorption of cell adhesive proteins on pure PEGgels is sufficiently low that cells cannot significantlyattach and spread onto these hydrogels without thespecific addition of cell adhesion ligands.43 This is incontrast to most silica-containing scaffolds, purePDMS substrates, and native polymer scaffolds (e.g.,gelatin) which generally adsorb significant levels bio-active proteins from serum.12,42,48 Thus, although theadhesion ligand RGDS was incorporated into thepresent PDMSstar-PEG formulations, the adsorptionof serum proteins conducive to bone formation waslikely limited.

A limitation of this study is that the observedmodulation of osteoblast responses cannot be defini-

INORGANIC–ORGANIC HYBRID SCAFFOLDS 119

Journal of Biomedical Materials Research Part A

tively attributed to the inorganic composition of thescaffolds alone. Specifically, alterations in the inor-ganic content of the PDMSstar-PEG gels resulted inconcomitant alterations in scaffold modulus andmesh size, both of which can influence cell behav-ior.34,36,49 Indeed, just as calcium deposition is nega-tively correlated with gel PDMSstar levels, it is alsonegatively correlated with mesh size and positivelycorrelated with modulus. Similarly, collagen type IIand CSPG are positively correlated with hydrogelPDMSstar content and scaffold mesh size, and nega-tively correlated with modulus. Thus, it is possiblethat scaffold modulus and/or mesh size could bethe primary compositional variables underlyingobserved cell behaviors. This limitation results fromthe Mw of the PDMSstar (�14 kDa) employed in thisstudy relative to that of PEG (�3.4 kDa). That beingsaid, the present results indicate that gradientPDMSstar-PEG hydrogels may be promising forosteochondral tissue engineering. Future studies willprobe the impact of a broader range of siloxane lev-els on osteoblast responses while employing aPDMSstar Mw that minimally alters base hydrogelmesh size and modulus.

CONCLUSIONS

In the present work, rat calvarial osteoblasts wereencapsulated in a series of PEG-based hydrogels ofincreasing PDMSstar content. Osteoblasts appeared totransdifferentiate into chondrocyte-like cells withincreasing scaffold inorganic content, as indicated byincreased chondroitin sulfate and collagen type IIproduction and by upregulation of sox9. Further-more, the synthesis of bone-like matrix (calciumdeposits and osteocalcin) decreased with increasingPDMSstar levels. Combined, the present results indi-cate that PDMSstar-PEG hybrid gels warrant furtherstudy as osteochondral regeneration matrices.

References

1. Woo S-Y, Hildebrand K, Watanabe N, Fenwick J, Papage-orgiou C, Wang J. Tissue engineering of ligament and tendonhealing. Clin Orthop Relat Res 1999;367(Suppl): S312–S323.

2. Cooper JA, Lu HH, Ko FK, Freeman JW, Laurencin CT.Fiber-based tissue-engineered scaffold for ligament replace-ment: design considerations and in vitro evaluation. Biomate-rials 2005;26:1523–1532.

3. Fu F, Bennett C, Lattermann C, Menetrey J, Lattermann C.Current trends in anterior cruciate ligament reconstruction,part I. Biology and biomechanics of reconstruction. Am JSports Med 1999;27:821–830.

4. Weitzel P, Richmond J, Altman G, Calabro T, Kaplan D.Future direction of the treatment of ACL ruptures. OrthopClin North Am 2002;33:653–661.

5. Vunjak-Novakovic G, Altman G, Horan R, Kaplan DL. TissueEngineering of Ligaments. Annu Rev Biomed Eng 2004;6:131–156.

6. Deehan DJ, Cawston TE. The biology of integration of the an-terior cruciate ligament. J Bone Joint Surg Br 2005;87:889–895.

7. Thomopoulos S, Williams GR, Gimbel JA, Favata M, Soslow-sky LJ. Variation of biomechanical, structural, and composi-tional properties along the tendon to bone insertion site.J Orthop Res 2003;21:413–419.

8. Yang PJ, Temenoff JS. Engineering orthopedic tissue interfa-ces. Tissue Eng Part B 2009;15:127–141.

9. Phillips JE, Burns KL, Le Doux JM, Guldberg RE, Garcia AJ.Engineering graded tissue interfaces. Proc Natl Acad Sci USA2008;105:12170–12175.

10. Ducheyne P, Qiu Q. Bioactive ceramics: The effect of surfacereactivity on bone formation and bone cell function. Biomate-rials 1999;20:2287–2303.

11. Kretlow JD, Mikos AG. Review: Mineralization of syntheticpolymer scaffolds for bone tissue engineering. Tissue Eng2007;13:927–938.

12. Garcia AJ, Ducheyne P, Boettiger D. Effect of surface reactionstage on fibronectin-mediated adhesion of osteoblast-like cellsto bioactive glass. J Biomed Mater Res 1998;40:48–56.

13. Khan Y, Yaszemski MJ, Mikos AG, Laurencin CT. Tissue en-gineering of bone: Material and matrix considerations. J BoneJoint Surg Am 2008;90(Suppl 1):36–42.

14. Ning CQ, Mehta J, El-Ghannam A. Effects of silica on thebioactivity of calcium phosphate composites in vitro. J MaterSci: Mater Med 2005;16:355–360.

15. Gupta G, El-Ghannam A, Kirakodu S, Khraisheh M, Zbib H.Enhancement of osteoblast gene expression by mechanicallycompatible porous Si-rich nanocomposite. J Biomed MaterRes B 2007;81:387–396.

16. Song J-H, Yoon B-H, Kim H-E, Kim H-W. Bioactive anddegradable hybridized nanofibers of gelatin-siloxane for boneregeneration. J Biomed Mater Res A 2008;84:875–884.

17. Delong SA, Moon JJ, West JL. Covalently immobilized gra-dients of bFGF on hydrogel scaffolds for directed cell migra-tion. Biomaterials 2005;26:3227–3234.

18. Hahn M, Miller J, West J. Three dimensional biochemical andbiomechanical patterning of hydrogels for guiding cell behav-ior. Adv Mater 2006;18:2679–2684.

19. Hahn M, Taite L, Moon J, Rowland M, Ruffino K, West J.Photolithographic patterning of polyethylene glycol hydro-gels. Biomaterials 2006;27:2519–2524.

20. Hahn MS, Miller JS, West JL. Laser scanning lithography forsurface micropatterning on hydrogels. Adv Mater 2005;17:2939–2942.

21. Mcdonald JC, Whitesides GM. Poly(dimethylsiloxane) as amaterial for fabricating microfluidic devices. Acc Chem Res2002;35:491–499.

22. Gong C, Frzchet JMJ. End functionalization of hyperbranchedpoly(siloxysilane): novel crosslinking agents and hyper-branched-linear star block copolymers. J Polym Sci Part A:Polym Chem 2000;38:2970–2978.

23. Panda DK, Miao D, Lefebvre V, Hendy GN, Goltzman D.The transcription factor Sox9 regulates cell cycle and differen-tiation genes in chondrocytic CFK2 cells. J Biol Chem 2001;276:41229–41236.

24. Grunlan MA, Lee NS, Mansfeld F, Kus E, Finlay JA, CallowJA, Callow ME, Weber WP. Minimally adhesive polymersurfaces (MAPS) prepared from star oligosiloxanes and staroligofluorosiloxanes. J Polym Sci Part A: Polym Chem 2006;44:2551–2566.

25. Boutevin B, Guida-Pietrasanta F, Ratsimihety JA. Synthesis ofphotocrosslinkable fluorinated polydimethylsiloxanes: Directintroduction of acrylic pendant groups via hydrosilylation.Polym Sci Part A: Polym Chem 2000;38:3722–3728.

120 MUNOZ-PINTO ET AL.

Journal of Biomedical Materials Research Part A

26. Murthy R, Shell CE, Grunlan MA. The influence of poly(eth-ylene oxide) grafting via siloxane tethers on protein adsorp-tion. Biomaterials 2009;30:2433–2439.

27. El-Ghannam AR. Advanced bioceramic composite for bonetissue engineering: Design principles and structure-bioactivityrelationship. J Biomed Mater Res Part A 2004;69:490–501.

28. Ford MC, Bertram JP, Hynes SR, Michaud M, Li Q, YoungM, Segal SS, Madri JA, Lavik EB. A macroporous hydrogelfor the coculture of neural progenitor and endothelial cells toform functional vascular networks in vivo. Proc Natl AcadSci USA 2006;103:2512–2517.

29. Canal T, Peppas NA. Correlation between mesh sizeand equilibrium degree of swelling of polymeric networks.J Biomed Mater Res 1989;23:1183–1193.

30. Mellott MB, Searcy K, Pishko MV. Release of protein fromhighly cross-linked hydrogels of poly(ethylene glycol) diacrylatefabricated by UV polymerization. Biomaterials 2001;22:929–941.

31. Watkins AW, Anseth KS. Investigation of molecular transportand distributions in poly(ethylene glycol) hydrogels withconfocal laser scanning microscopy. Macromolecules 2005;38:1326–1334.

32. Anseth KS, Metters AT, Bryant SJ, Martens PJ, Elisseeff JH,Bowman CN. In situ forming degradable networks and theirapplication in tissue engineering and drug delivery. J Con-trolled Release 2002;78:199–209.

33. Hahn M, Mchale M, Wang E, Schmedlen R, West J. Physio-logic pulsatile flow bioreactor conditioning of poly(ethyleneglycol)-based tissue engineered vascular grafts. Ann BiomedEng 2007;35:190–200.

34. Liao H, Munoz-Pinto D, Qu X, Hou Y, Grunlan M, Hahn M.Influence of hydrogel material properties on vocal fold fibro-blast extracellular matrix production and phenotype. ActaBiomater 2008;4:1161–1171.

35. Miller EJ, Gay S. Collagen—An Overview. Methods Enzymol1982;82:3–32.

36. Munoz-Pinto DJ, Bulick AS, Hahn MS. Uncoupled investiga-tion of scaffold modulus and mesh size on smooth musclecell behavior. J Biomed Mater Res A 2009;90:303–316.

37. Salinas CN, Anseth KS. The influence of the RGD peptidemotif and its contextual presentation in PEG gels on human

mesenchymal stem cell viability. J Tissue Eng Regen Med2008;2:296–304.

38. Burdick JA, Anseth KS. Photoencapsulation of osteoblasts ininjectable RGD-modified PEG hydrogels for bone tissue engi-neering. Biomaterials 2002;23:4315–4323.

39. Salinas CN, Anseth KS. The enhancement of chondrogenicdifferentiation of human mesenchymal stem cells by enzy-matically regulated RGD functionalities. Biomaterials 2008;29:2370–2377.

40. Benoit DSW, Schwartz MP, Durney AR, Anseth KS. Smallfunctional groups for controlled differentiation of hydrogel-encapsulated human mesenchymal stem cells. Nat Mater2008;7:816–823.

41. Lim JY, Shaughnessy MC, Zhou Z, Noh H, Vogler EA,Donahue HJ. Surface energy effects on osteoblast spatialgrowth and mineralization. Biomaterials 2008;29:1776–1784.

42. Murthy R, Cox CD, Hahn MS, Grunlan MA. Protein-resistantsilicones: Incorporation of poly(ethylene oxide) via siloxanetethers. Biomacromolecules 2007;8:3244–3252.

43. Gombotz WR, Wang GH, Horbett TA, Hoffman AS. Proteinadsorption to poly(ethylene oxide) surfaces. J Biomed MaterRes 1991;25:1547–1562.

44. Peyton S, Raub C, Keschrumrus V, Putnam A. The use ofpoly(ethylene glycol) hydrogels to investigate the impact ofECM chemistry and mechanics on smooth muscle cells. Bio-materials 2006;27:4881–4893.

45. Ren L, Tsuru K, Hayakawa S, Osaka A. In vitro evaluation ofosteoblast response to sol-gel derived gelatin-siloxanehybrids. J Sol-Gel Sci Technol 2003;26:1137–1140.

46. Ren L, Tsuru K, Hayakawa S, Osaka A. Novel approach tofabricate porous gelatin-siloxane hybrids for bone tissue engi-neering. Biomaterials 2002;23:4765–4773.

47. Anselme K. Osteoblast adhesion on biomaterials. Biomaterials2000;21:667–681.

48. Kim BS, Nikolovski J, Bonadio J, Smiley E, Mooney DJ. Engi-neered smooth muscle tissues: Regulating cell phenotypewith the scaffold. Exp Cell Res 1999;251:318–328.

49. Engler AJ, Sen S, Sweeney HL, Disher DE. Matrix elasticitydirects stem cell lineage specification. Cell 2006;126:677–689.

INORGANIC–ORGANIC HYBRID SCAFFOLDS 121

Journal of Biomedical Materials Research Part A