Three-dimensional poly(1,8-octanediol–co-citrate) scaffold pore shape and permeability

Three-dimensional poly(1,8-octanediol–co-citrate)scaffold pore shape and permeability effects on sub-cutaneous in vivo chondrogenesis using primary chondrocytesClaire G.Jeong a ,Huina Zhang a ,Scott J.Hollister a ,b ,c ,⇑aDepartment of Biomedical Engineering,University of Michigan,Ann Arbor,MI 48109-2125,USA bDepartment of Mechanical Engineering,University of Michigan,Ann Arbor,MI 48109-2125,USA cDepartment of Surgery,University of Michigan,Ann Arbor,MI 48109-0329,USAa r t i c l e i n f o Article history:Received 16April 2010Received in revised form 29July 2010Accepted 26August 2010Available online 31August 2010Keywords:Poly(1,8-octanediol–co-citrate)scaffold Pore shape PermeabilityMechanical properties Cartilagea b s t r a c tThe objective of this study was to evaluate the coupled effects of three-dimensional poly(1,8-octanediol–co-citrate)(POC)scaffold pore shape and permeability on chondrogenesis using primary chondrocytes in vivo.Chondrogenesis was characterized as cartilage matrix formation by sulfated glycosaminoglycan (sGAG)quantification,relative mRNA expression of the cartilage-related proteins collagen types I,II and X,aggrecan and matrix metalloproteinases 13and 3and the compressive mechanical properties of the tissue/scaffold construct.A low permeability design with a spherical pore shape showed a significantly greater increase in cartilage matrix formation over 6weeks in vivo than a high permeability design with a cubical pore shape.This increase in cartilage matrix synthesis corresponded with increases in mechan-ical compressive nonlinear elastic properties and histological data demonstrating darker red Safranin-O staining.There was higher mRNA expression for both cartilage-specific proteins and matrix degradation proteins in the high permeability design,resulting in overall less sGAG retained in the high permeability scaffold compared with the low permeability scaffold.Controlled POC scaffolds with a spherical pore shape and low permeability correlated with significantly increased cartilage matrix production using primary seeded chondrocytes.These results indicate that the low permeability design with a spherical pore shape provided a better microenvironment for chondrogenesis than the high permeability design with a cubical pore shape.Thus,scaffold architecture and material design may have a significant impact on the success of matrix-based clinical cartilage repair strategies.Ó2010Published by Elsevier Ltd.on behalf of Acta Materialia Inc.1.IntroductionArticular cartilage has very limited self-repair capability due to its low cellularity and vascularization.Current therapeutic strate-gies,such as microfracture,osteochondral transplantation and autologous chondrocyte implantation,have been applied with rel-ative success for more than a decade,yet are still limited to small defect sizes and suffer from donor site morbidity,restricting their clinical application [1–5].These limitations have increased interest in tissue engineering approaches combining degradable biomate-rial scaffolds with or without cell therapy.While it is widely pos-tulated that scaffolds play a major role in the success or failure of cartilage repair,there is very limited data on how scaffold architecture and materials should be designed to enhance cartilage repair using tissue engineered approaches.Specifically,the mechanical and mass transport environments provided to seeded cells or host cells by implanted scaffolds may significantly affect cell activity,dictating the outcome of cartilage repair.These envi-ronments are determined by scaffold structural parameters,including pore geometry,pore size,porosity,pore interconnectivi-ty,etc.To create scaffolds that enhance chondrogenesis we must first be able to test hypotheses concerning how scaffold design affects chondrogenesis,which requires fabricating scaffolds with controlled architectures.Here we have used a biocompatible,biodegradable elastomer poly(1,8-octanediol–co-citrate)(POC)scaffold and cultured chondrocytes in vivo to examine the coupled effects of designed pore shape and permeability on matrix produc-tion,mRNA gene expression and differentiation of chondrocytes in vivo after sub-cutaneous implantation,as well as the resultant mechanical property changes in scaffold/tissue constructs due to tissue formation and scaffold degradation.POC has been shown to support cell attachment and proliferation of and matrix produc-tion by chondrocytes [6],while our own recent work [7]has dem-onstrated that POC enhances in vitro chondrogenesis over other1742-7061/$-see front matter Ó2010Published by Elsevier Ltd.on behalf of Acta Materialia Inc.doi:10.1016/j.actbio.2010.08.027⇑Corresponding author at:Department of Biomedical Engineering,University ofMichigan,Ann Arbor,MI 48109-2125,USA.Tel.:+17346479962.E-mail address:scottho@ (S.J.Hollister).biodegradable polymers of the same design.Moreover,our previ-ously engineered scaffolds had compressive mechanical properties similar to native articular cartilage(Table2)[8].Our scaffold designs were rigorously fabricated such that only pore shape and scaffold permeability varied in a controlled manner and other factors such as pore size,pore interconnectivity,porosity and scaffold surface area were kept constant between the two dif-ferent designs.Our previous work[7,9–11]demonstrated that a low permeability design with a spherical pore shape promoted chondrogenesis in vitro on different materials using chondrocytes. The purpose of this paper was to investigate whether low perme-ability spherical pore designs also supported enhanced chondro-genesis in vivo.Specifically,we tested the hypothesis that POC scaffolds designed with a spherical pore shape and low permeabil-ity would enhance chondrogenesis in vivo in a sub-cutaneous model.Chondrogenesis in chondrocyte-seeded POC scaffolds was assessed as cartilaginous matrix production,cartilage-specific gene expression and composite tissue/scaffold compressive mechanical properties,including how tissue formation and scaffold degrada-tion interacted in determining thefinal construct mechanics.For this purpose wefirst cultured seeded primary chondrocytes in de-signed POC scaffolds using collagen I(Col I)/hyaluronic acid(HyA) hydrogels as a cell carrier for1week in vitro.These cell/scaffold constructs were then implanted for6weeks in vivo in a sub-cuta-neous nude mouse model[12].2.Materials and methods2.1.Collagen I/hyaluronic acid hydrogelHigh concentration Col I hydrogels and HyA were purchased from BD Bioscience Discovery Labs(San Jose,CA)and Hyalogic LLC(Edwardsville,KS),respectively.High concentration collagen I hydrogels were diluted with0.02N sterile acetic acid to a concen-tration of6mg mlÀ1and5wt.%hyaluronic acid was added based on our previous results showing that a Col I gel with5%HyA en-hanced chondrogenesis[12].2.2.Synthesis of pre-poly(1,8-octanediol–co-citrate)All chemicals were purchased from Sigma–Aldrich(Milwaukee, WI).The pre-polymer(pPOC)was synthesized as previously described[13].Briefly,equimolar amounts of citric acid and 1,8-octanediol were added and the mixture was melted at 160–165°C for15min under aflow of nitrogen gas with stirring and then further lowered to140°C for40min to create the pre-polymer.2.3.Scaffold design and fabricationThe three-dimensional(3D)POC scaffold architecture was de-signed using previous methods and software[12,14–16].Porous POC scaffolds(6.35mm diameter,4.0mm height)with900l m interconnected spherical or cubical pores(porosity spherical50% S50),cubical62%(C62);permeability high(C62)=13.5Âlow (S50))were designed using custom IDL programs(RSI,Boulder, CO).The details of POC scaffold fabrication were as previously re-ported[13,16].In brief,wax molds with a designed architecture were built using a Solidscape Patternmaster™,from which solid free-form fabricated hydroxyapatite(HA)molds were then pre-pared.pPOC was then cured in the HA molds to create the two de-signed scaffold architectures[17].pPOC was poured into the wells of a Teflon container and the HA molds were embedded within the pPOC.The pPOC/HA mold/Teflon container unit was cured at 100°C for1day,followed by curing at100°C for3more days un-der high vacuum(À30in.Hg).The HA mold was removed by incu-bation in a decalcifying reagent(RDO,APEX Engineering Products, Plainfield,IL)for10h,followed by incubation in water(Milli-Q Water Purification System,Billerica,MA)for24h to obtain thefinal porous POC scaffolds,which were dried at room temperature for 24h before autoclaving.2.4.Mechanical testsPorous scaffolds(n=4–6per design per time),gel-filled scaffolds (n=4–6per design per time)or scaffolds with tissue in-growth (after in vitro pre-culture(0week in vivo)or3or6weeks of in vivo sub-cutaneous implantation,n=4–6per design per time) were tested in unconfined compression at a displacement rate of 2mm minÀ1equivalent to a strain rate of0.0063sÀ1using a large strain to applied stretch ratio of E=0.5(k2À1),where k is the stretch ratio(<1for compression)and E is the large strain.(Alliance RT/30 electromechanical test frame,50N load cell with0.5%error range, MTS Systems,Eden Prairie,MN).Data were collected using Test-Works4software.MATLAB(The MathWorks Inc.,Natick,MA)soft-ware(LSQNONLIN)was used tofit a nonlinear elasticity model r=A[e B eÀ1]to the experimental data,where r is thefirst Piola–Kirchoff stress,e is the large strain and A and B are model coefficients. Tangent moduli(=AB e B e)were calculated at10%strain from thefit.2.5.Porosity and permeability measurementsSeven scaffolds per design were scanned using a MS-130high resolution micro-computer tomography(micro-CT)scanner(GE Medical Systems,Toronto,Canada)at19l m voxel resolution be-fore mechanical testing to avoid any compression artifacts.Scaffold permeability(n=6–7)with and without the HyA/Col I(6mg mlÀ1) composite gel and for tissue/scaffold constructs after6week in vivo implantation was measured using a permeability test set-up[9,13].Briefly,a custom permeability chamber was designed to apply a constant hydraulic pressure on a scaffold contained in-side the chamber that forces water toflow through the chamber and into a secondary container placed on top of a scale connected to a computer recording data at3Hz.A customized LabView pro-gram based on an equation derived from Bernoulli’s equation and Darcy’s law[18]was used to compute permeability.The permeability of the scaffolds with the hydrogel was mea-sured to mimic freshly loaded cell conditions in vitro for compari-son with the permeability of scaffolds with tissue in-growth after 6weeks.Primary porcine chondrocytes were seeded into3D scaf-folds byfirst suspending the cells in medium with the composite HyA/Col I gel and then pushing the gel into the3D scaffolds[12]. The gelation procedure was as follows:770l l of Col I(6mg mlÀ1) with77l l of HyA(stock concentration3mg mlÀ1in1.5M sodium chloride,molecular weight 2.4–3MDa)were well mixed,after which the pH of the suspension was increased by the addition of 11l l of0.5N sodium hydroxide with220mg mlÀ1sodium bicar-bonate to initiate gelation,then the hydrogel mixtures were dripped onto pre-prepared sterile scaffolds until the scaffolds were fully soaked andfilled with gel to the top surface,followed by incu-bation at37°C for at least30min to further solidify the gels.The 50%and62%porous scaffolds were seeded with121and150l l of cell/gel mixture,respectively,in order to maintain the same cell density per volume.2.6.In vitro pre-culture and in vivo implantationPorcine chondrocytes(pChon)were isolated from the full depth of the joints of domestic pigs and seeded onto scaffolds following previous methods[12].Cells were resuspended at a density of 30Â106cells mlÀ1in770l l of HyA/Col I composite gel with506 C.G.Jeong et al./Acta Biomaterialia7(2011)505–51450l l of culture medium.The remaining steps were as previously described(see Section2.5).24scaffolds per design(15per design for6weeks in vivo implantation and9per design for1week in vitro pre-culture)seeded with pChon were cultured in chondro-genic medium(Dulbecco’s modified Eagle’s medium(DMEM)sup-plemented with10%fetal bovine serum(FBS),1%penicillin/ streptomycin(Gibco),50mg mlÀ12-phospho-L-ascorbic acid(Sig-ma),0.4mM proline(Sigma),5mg mlÀ1insulin(Gibco),0.1mM non-essential amino acids(Gibco))in12-well plates for1week be-fore in vivo implantation.Chondrocytes were cultured under gen-tle agitation on an orbital shaker and the medium was changed every other day.All POC scaffolds were sterilized in an autoclave and pre-soaked in DMEM for24h,then briefly rinsed with phos-phate-buffered saline(PBS)prior to cell seeding.The cell cultures were maintained in a water-jacketed incubator equilibrated with 5%CO2at37°C.To observe mechanical stiffness changes over time due to degradation,gel-filled scaffolds with no cells(n=4–6per design at each time point)were also subjected to the same in vitro conditions to maintain the same conditions as for the cell/gel-filled scaffolds.After1week of culture30cell-seeded POC scaffolds(15per design)and20gel-filled POC scaffolds(no cells)(n=4per design6weeks)were removed,finely diced and immediately placed in 1ml of pre-prepared papain solution(10U mgÀ1papain(Sigma Aldrich No.P4762),1ÂPBS,5mM cysteine HCl,5mM EDTA,pH 6.0,mixed for2h at37°C and thenfiltered).Scaffolds were di-gested in papain solution for24h at60°C then immediately stored atÀ20°C.The digested tissue/scaffold solution was analyzed by dimethylmethylene blue(DMMB)assay.Briefly,10l of sample was mixed with200l of DMMB reagent and absorbance was read in a plate reader(MultiSkan Spectrum,Thermo,Waltham,MA)at 525nm.A standard curve was established from chondroitin6-sul-fate from shark(Sigma C4384)to determine sample absorbance for [19,20].The total sGAG was normalized to DNA content,which was measured using Hoechst dye33258(Sigma DNA-QF).In brief,10l of digested sample was added to200l of pre-prepared Hoechst solution and read with excitation at355nm and emission at 460nm(Fluoroskan Ascent FL,Thermo,Waltham,MA)in a96-well plate.Readings were compared with standard curves from calf thy-mus DNA(Sigma DNA-QF)[21].2.8.Quantitative polymerase chain reaction(PCR)C.G.Jeong et al./Acta Biomaterialia7(2011)505–514507primer was obtained by qtPCR with serially diluted cDNA sample mixtures.Samples were prepared using a Taqman universal PCR master mix(Applied Biosystems)and custom designed porcine primers.The level of gene expression was calculated with standard samples and normalized to GAPDHfirst and the low permeability design(S50).2.9.HistologyConstructs(n=3per design)at each time point werefixed in 10%buffered formalin overnight,dehydrated in a series of graded ethanol and embedded in paraffin.Tissue sections were stained with hematoxylin and eosin(H&E)for scaffolds without cells or with Safranin-O with Fast Green counterstaining for scaffolds with cells,to assess cell and tissue distribution,cell morphology and sGAG production.The cross-sectional and longitudinal views of eight to ten slides(4sections per slide)were obtained from the center of each scaffold[12].2.10.Statistical analysisData are expressed as means±standard deviation.The statisti-cal significance among different designs or time points was calcu-lated using linear regression and one-way ANOVA with Tukey’s post hoc comparison or Student’s t-test using SPSS software(SPSS for Windows,Release14.0.,SPSS Inc.,Chicago).Data were taken to be significant when a P value of0.05or less was obtained.3.Results3.1.Scaffold design,permeability,and mechanical characterization3D scaffolds were fabricated from POC using solid free-form fabrication methods.The scaffolds were designed with either a spherical or cubical pore shape,with resultant significant differ-ences in permeability.Pore size and regular interconnectivity were maintained between scaffold designs(Fig.1A and Table1).For descriptive purposes the scaffold with a cubical pore design and 62%porosity is designated C62and the scaffold with a spherical pore design and50%porosity is designated S50.By varying the pore shape and porosity scaffolds could be fabricated with signifi-cantly different permeabilities(t-test,P60.05)between designs, with S50having low permeability(3.51±0.95Â10À7m4N sÀ1) and C62having high permeability,13.6times that of S50 (47.4±1.15Â10À7m4N sÀ1)The difference in permeability was far greater than the difference in porosity(C62=S50Â1.2).Sur-face area was also not significantly different between the designs (S50=288±38vs.C62=243±15mm2).Thus,among design parameters permeability and pore shape exhibited the greatest variations between designs.With Col I/HyA hydrogel the scaffold permeability values all de-creased from the original scaffold permeability,as expected,yet con-tinued to exhibit a similar differential trend between designs.Due to collagen gel degradation after1–2weeks the permeability without gel most likely represents the permeability of scaffolds during the 6weeks of in vivo implantation without tissue in-growth,whereas the permeability with the gel represents the scaffolds at0weeks with initial cell seeding.As most of collagen gel was degraded within a week during tissue formation,scaffold permeability changed dynamically during the6week in vivo implantation period with gel and POC degradation,tissue in-growth from seeded chondro-cytes and tissue formation around the scaffold by mouse cells.The permeabilities measured for tissues grown for6weeks after in vivo implantation for both designs were significantly lower than the permeabilities with the gel,indicating extensive growth of tissue into the scaffolds during in vivo implantation.The gel-filled scaffold permeability for the S50low permeability design(1.72±0.45Â10À7m4N sÀ1)was3.3times higher than the permeability of the scaffold with tissue in-growth(0.52±0.01Â10À7m4N sÀ1), whereas that for the C62gel-filled high permeability design (4.14±0.73Â10À7m4N sÀ1)was8.1times higher than the perme-ability of the scaffold with tissue in-growth(0.51±0.01Â10À7m4N sÀ1).However,there was no significant difference in per-meability between tissue/scaffold constructs between designs after 6weeks in vivo implantation.In order to observe mechanical stiffness changes due to scaffold degradation and tissue formation in vivo we implanted gel-filled scaffolds(which were also cultured in vitro for1week before implantation)in mice and measured their stiffness at0,3and 6weeks in vivo to determine changes in effective nonlinear scaf-fold stiffness.Table2summarizes the compressive tangent moduli at10%strain for different scaffold designs after applying the non-linear modelfit for gel-filled or tissue grown scaffolds at each time point.For both designs the tangent modulus increased from0to3 or6weeks when not seeded with cells(Fig.2B).The low perme-ability design demonstrated the greatest increase over the6week implantation period(9times vs.3.4times compared with the high permeability design).There were no significant changes in tangent modulus from3to6weeks for either design(Fig.2B).Unlike the case without cells,both designs with cells showed roughly the same increase(3.8times)in tangent modulus from0to6weeks in vivo.When comparing cases with and without cells,scaffolds with cells showed higher tangent moduli for both designs.In Fig.2A the stress–strain curve for the low permeability de-sign(S50)shifts up from0to3weeks with increased stress at each strain,then shifts down from3to6weeks.The curve for the high permeability design(C62)increased from0to3week and from3 to6weeks.C62(high)showed less nonlinear behavior than S50 (low)at the3week time point,but at the6week time point C62 (high)showed more nonlinear behavior,due to possible increased tissue matrix deposition by host cells within the pores,coupled with slower scaffold degradation.When cells were present both designs increased in nonlinear stiffness over6weeks and showed more nonlinear behavior after implantation,due to tissue formation inside the pores.However, when comparing scaffolds with cells vs.without cells at6weeks, the low permeability design(S50)showed a larger increase in non-linear stiffness than the high permeability design(C62).3.2.Matrix production and mRNA expression before and after in vivo implantationThe low permeability design(S50)showed a significant increase (P60.05)in matrix formation(sGAG/DNA content)over6week in vivo implantation,whereas the high permeability design(C62) did not(Fig.3).An increase in sGAG/DNA content implies that aTable1Scaffold descriptions a.Design(n=8)S50,Low C62,HighPorosity(%)50±1.6262±2.36Permeability without gel(10À7m4N sÀ1)b 3.51±0.9547.4±1.15Permeability with gel(10À7m4N sÀ1)b 1.72±0.45 4.14±0.73Permeability with tissues(in vivo)(10À7m4N sÀ1)c0.52±0.010.51±0.01Surface area(mm2)c288±38243±15Pore shape Spherical CubicalPore diameter900l m900l ma Data are expressed as means±standard deviation.b n=4per design,t-test,P60.05,statistically significant(after6weeks in vivoimplantation).c Not statistically significant(t-test,P60.05).508 C.G.Jeong et al./Acta Biomaterialia7(2011)505–514single cell is more geared towards the chondrocytic phenotype with a higher matrix production/maintenance rate.Even though it was not significantly different,the low permeability design (S50)showed22%higher matrix production than the high perme-ability design(C62)at6weeks,which was the opposite of the re-sults at0weeks(immediately after the1week in vitro pre-culture),where the high permeability design formed35%more ma-trix than the low permeability design.Table2Tangent moduli(MPa)at10%strain(n=4–6)a.In vivo implantation c S50,low b C62,high bWithout cells With cells Without cells With cells0weeks0.059±0.0190.192±0.0470.054±0.0150.122±0.022 3weeks0.531±0.1590.185±0.0266weeks0.525±0.1730.701±0.1310.202±0.0970.467±0.092a Data are expressed as means±standard deviation.b For statistical significance between groups and conditions see Fig.2b and d.c Note that all of the scaffolds were pre-cultured for1week in vitro under the same conditions before in vivo implantation,whether cells were present or not.ACC.G.Jeong et al./Acta Biomaterialia7(2011)505–514509All mRNA expression for the low permeability design except Col X(Fig.4)showed higher ratios than the high permeability design at 0week in vivo(Fig.4A).The ratio of Col II gene expression to Col I gene expression(Col II/Col I),known as the‘‘differentiation index”(DI),and aggrecan expression for the low permeability design(S50) in particular were significantly higher than expression for the high permeability design(C62).A higher DI value indicates a more chondrocytic genotype,while a lower value indicates morefibro-blastic gene expression[22].Type II collagen and aggrecan are typ-ically used as biomarkers of differentiated chondrocytes and cartilaginous tissues[23–25],whereas type I and type X collagen are typically used as markers of dedifferentiated or terminally dif-ferentiated(hypertrophic)chondrocytes,respectively[26,27].In contrast,the high permeability design(C62)showed higher values than the low permeability design(S50)after6weeks in vivo implantation for all expressed genes(Fig.4B).Type I colla-gen and aggrecan expression for C62were significantly higher(4 times)than those for S50.Thus,the results for all gene expression were higher for the C62design after6weeks in vivo,whereas all gene expression was higher for the S50design after1week in vitro culture(Fig.4).Since total sGAG(Fig.3)is a result of not only matrix formation and secretion but also matrix degradation and sGAG leaching out of the scaffold,it is also important to investigate mRNA expression of matrix degradation proteins.MMP-13and MMP-3play critical roles in extracellular matrix(ECM)degradation in cartilage. MMP-13appears to be the primary collagenase of articular carti-lage and is also critical for cartilage turnover and chondrocyte hypertrophy in the growth plate[28,29].In addition to collagen, MMP-13also degrades the proteoglycan molecule aggrecan,giving it a dual role in matrix destruction.[30–32].MMP-3is elevated in arthritis,and degrades non-collagen matrix components of the joints.Both MMP-13and MMP-3mRNA expression for the low permeability design(S50)were significantly higher than for the high permeability design(C62)at0week,which then reversed after6weeks in vivo implantation.Over6weeks the expression of both MMPs greatly increased for the high permeability design, whereas the low permeability design maintained relatively con-stant MMP mRNA expression.In general,mRNA expression levels were high for the low permeability design at0week whereas mRNA expression levels were high for the high permeability design at6week in vivo.3.3.HistologySafranin-O staining(Fig.5)supported the sGAG quantification data(Fig.3)in that more intense and darker sGAG staining inside the scaffolds was observed after6weeks for both designs com-pared with the0week time point.For the0week time point there was no discernible difference in sGAG between scaffold designs. However,after6weeks implantation the low permeability design (S50)showed increased staining over the entire scaffold,with dar-ker sGAG staining particularly near the pore necking area(denoted by arrows),compared with the high permeability design(C62).The innermost parts of the low permeability scaffolds,both cross-sec-tionally and longitudinally,had rich Safranin-O staining,demon-strating that sGAGs were formed and maintained inside the scaffolds.The chondrocytic phenotype was easily observed,with lacunae,for both designs at6weeks.H&E stained images(Fig.6)of scaffolds without seeded cells at 3and6weeks showed that tissue formed outside the scaffold but not in the center pore structures at3weeks.Further infiltration510 C.G.Jeong et al./Acta Biomaterialia7(2011)505–514denote tissues formed at the edge of the scaffolds while the asterisks denote the central parts of the scaffolds.was seen at6weeks.This infiltrated tissue was likelyfibrous tissue resulting from in-growth of surrounding host tissuefibroblasts,as there was no positive staining of Safranin-O/Fast Green for either design at any time point(data not shown).The combination offi-brous tissue formation with scaffold degradation likely dictated the change in mechanical properties of the gel-filled,cell-free pared with S50,C62,High showed deeper cell infiltra-tion into the central part of the pore architecture(asterisks)at 6weeks,likely due to the higher permeability allowing greater and more rapid cell migration.4.DiscussionThe focus of this study was to determine the effects of3D POC controlled scaffold pore shape and permeability on chondrogenesis in vivo.Our previous in vitro studies[7,9]demonstrated that scaf-folds with lower permeability and spherical pore shapes increased cartilage matrix production and gene expression.The question then is whether scaffold permeability and pore shape influence chondrogenesis in vivo.Although the ultimate test of in vivo chondrogenesis would be an articular defect,such a defect would also entail mechanical compression and stimulus effects on chondrogenesis,confounding the investigation of pore shape and permeability influences on chondrogenesis.Thus,we chose a sub-cutaneous mouse model for in vivo chondrogenesis,to better isolate the pore shape and permeability affects on chondrogenesis using scaffolds with different controlled designs manufactured from the same ing the same scaffold material also helps isolate design effects,since we have also shown that the scaf-fold material significantly influences chondrogenesis[7].In order to evaluate the effects of POC scaffold pore architec-tures and the resulting permeability on tissues and the quality of engineered cartilage/scaffold constructs we examined constructs before and after in vivo implantation.Specifically,in this way we would be able to tell how scaffold architecture and the resulting permeability may affect chondrogenesis differently in vitro and in vivo.Some studies have reported that cells are predominantly distributed in the outer or surface zones of scaffolds when cultured in vitro with static seeding[33].Thus we tested pre-cultured POC/ cell constructs in vitro before implantation to ensure that cells were deposited in the scaffold pores and synthesized cartilage ECM(Figs.3and5).Both designs showed an ability to form matrix (Fig.3)and to synthesize articular cartilage-specific proteins (Fig.4)before in vivo implantation.In the in vitro pre-culture period there were no significant dif-ferences between the designs in terms of matrix production (Fig.3).The high permeability design(C62)produced slightly more matrix.However,with respect to relative mRNA expression,the low permeability design(S50)had relatively high mRNA expres-sion for all proteins except Col X.Of note,the differentiation index (Col II/Col I)and aggrecan expression were significantly higher. This can be explained by the relative permeability.With initial cell seeding and with a regular supply of nutrients during1week of in vitro pre-culture cells in both designs were exposed to excessive nutrients,but permeabilities that differed in relative magnitude between the designs,regulated by the permeability and pore shape (Table1).The low permeability design had higher expression of Col II and aggrecan and a higher Col II/Col I ratio,which are indicators of differentiated chondrocytes,closely related to sGAG formation, than the high permeability design(Fig.4).In addition,both MMP-13and MMP-3expression levels for the low permeability design(S50)were almost10times higher than for the high perme-ability design(C62),indicating active matrix remodeling.Hence, the overall sGAG/DNA content at0weeks for the low permeability design showed lower values than that for the high permeability de-sign,since sGAG/DNA would be a product of sGAG secretion/pro-duction inside the scaffolds,matrix degradation and possible sGAG loss by leaching out of the scaffolds.At6weeks after implantation,however,the results were the opposite of the short-term1week in vitro results.At6weeks after implantation the low permeability design(S50)produced more matrix than the high permeability design(C62)and demonstrated a significant increase in matrix production compared with the 1week in vitro data(Fig.3).However,with respect to relative mRNA expression,the high permeability design(C62)had relatively high mRNA expression for all proteins,with Col I and aggrecan expression being significantly higher(Fig.4B).This again can be explained by scaffold permeability.The environment of native articular cartilage is well known for its low permeability, avascularization(thus low nutrient supply)and hypoxic condi-tions.Unlike the in vitro conditions,there is probably limited chondrogenic nutrient supply around sub-cutaneous scaffolds in vivo.The low permeability design with a spherical pore shape likely allows even lower nutrient exchange from the outside.The scaffold permeability differences resulted from a balance between an increase in permeability due to gel and POC degradation and a decrease in permeability due to tissue in-growth.We hypothesize that the permeability difference between designs would be main-tained or possibly increase,allowing increased sGAG retention by the low permeability S50design.In addition,even though mRNA expression of relevant cartilage-like tissue formation proteins such as Col II and aggrecan were lower after6weeks in vivo for the low permeability design,the activities of MMP-13and MMP-3were also lower(2-fold lower for MMP-13and 10-fold lower for MMP-3).Furthermore,mRNA expression of both MMPs remained relatively constant over6weeks after implantation for the low permeability design(Fig.4B).This suggests that the low permeability design may be better at retaining sGAG due to its lim-ited nutrientflux,coupled with lower matrix degradation.These re-sults are in line with the histological data(Fig.5),clearly showing darker staining in the wider area of the middle of the low permeabil-ity scaffolds.There was especially intense sGAG staining near the necks of the spherical pores(Fig.5,arrows),further support for the postulate that the spherical pore shape helps retain cartilage matrix. Regardless of the scaffold pore architecture and permeability,Figs.5 and6show that the pores of the POC scaffold in vivo appeared to be filled with cartilaginous tissue produced by the implanted chondro-cytes,and the scaffold maintained its architecture after6weeks implantation.Along with the biochemical and histological assessments, mechanical assessments,such as changes in the nonlinear mechan-ical properties of entire tissue/scaffold constructs before and after implantation,are important,as mechanical performance is another important factor determining the success of regenerated tissues. Brown et al.[34]recently characterized the nonlinear hyperelastic behavior of20mm discs of bovine patella articular cartilage in unconfined compression at strain rates of0.1and0.025sÀ1.They noted that articular cartilage exhibited nonlinear elastic behavior at both strain rates,although the behavior stiffened at the higher strain rate as a characteristic of poroelastic/biphasic materials like cartilage.Fitting our one-dimensional nonlinear constitutive mod-el to the data of Brown et al.yielded a range of A coefficients from 0.64to1.72,and B coefficients from4.77to6.62.Our scaffold re-sults with cells for the S50designs yielded A=0.145±0.05and B=3.55±0.56,while the C62design yielded A=0.068±0.015 and B=4.44±0.30.With this nonlinear model a higher A value yielded a higher initial tangent stiffness,while a higher B value yielded more nonlinearity and stress paring these results suggests that the engineered cartilage had a nonlinearity similar to the lower range of the bovine results of Brown et al.[34],but the initial tangent stiffness was less.This indicates that512 C.G.Jeong et al./Acta Biomaterialia7(2011)505–514。