Osteoarthritis and Cartilage (2010) 18, 249e256

ª 2009 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.joca.2009.09.002

av and b1 integrins regulate dynamic compression-induced proteoglycansynthesis in 3D gel culture by distinct complementary pathwaysD. H. Chaiy, E. C. Arnerz, D. W. Griggsz and A. J. Grodzinskyyxk*yBiological Engineering Department, Massachusetts Institute of Technology, United StateszPfizer, Inc., United StatesxMechanical Engineering Department, Massachusetts Institute of Technology, United StateskElectrical Engineering Department, Massachusetts Institute of Technology, United States

Summary

Objective: Our goal was to test the hypothesis that specific integrin receptors regulate chondrocyte biosynthetic response to dynamic com-pression at early times in 3D gel culture, during initial evolution of the pericellular matrix, but prior to significant accumulation of further-removed matrix. The study was motivated by increased use of dynamic loading, in vitro, for early stimulation of tissue engineered cartilage,and the need to understand the effects of loading, in vivo, at early times after implantation of constructs.

Methods: Bovine articular chondrocytes were seeded in 2% agarose gels (15� 106 cells/mL) and incubated for 18 h with and without the pres-ence of specific integrin blockers (small-molecule peptidomimetics, function-blocking antibodies, and RGD-containing disintegrins). Sampleswere then subjected to a 24-h dynamic compression regime found previously to stimulate chondrocyte biosynthesis in 3D gel as well ascartilage explant culture (1 Hz, 2.5% dynamic strain amplitude, 7% static offset strain). At the end of loading, proteoglycan (PG) synthesis(35S-sulfate incorporation), protein synthesis (3H-proline incorporation), DNA content (Hoechst dye 33258) and total glycosaminoglycan(GAG) content (dimethyl methylene blue (DMMB) dye binding) were assessed.

Results: Consistent with previous studies, dynamic compression increased PG synthesis and total GAG accumulation compared to free-swell-ing controls. Blocking avb3 abolished this response, independent of effects on controls, while blocking b1 abolished the relative changes insynthesis when changes in free-swelling synthesis rates were observed.

Conclusions: This study suggests that both avb3 and b1 play a role in pathways that regulate stimulation of PG synthesis and accumulation bydynamic compression, but through distinct complementary mechanisms.ª 2009 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Key words: Integrin, Mechanical stimulation, Dynamic compression, Chondrocyte, Proteoglycan synthesis, Agarose culture,Mechanotransduction.

Introduction

In vivo, articular cartilage experiences a combination ofcompressive, tensile, and shear forces, both dynamic andstatic in nature1e3. These mechanical forces can inducea variety of macroscopic signals including changes in intra-tissue pH and osmotic pressure, hydrostatic pressure gradi-ents, fluid flow, streaming potentials and current, andmechanical deformation4e6, which are sensed by chondro-cytes and can regulate cell behavior. In vivo, static immobi-lization or reduced joint-loading leads to a loss ofglycosaminoglycan (GAG) content and decreased proteo-glycan (PG) synthesis, which can be partially rescued by re-mobilization7,8. In vitro, radially unconfined dynamiccompression at frequencies greater than 0.001 Hz hasbeen shown to increase chondrocyte biosynthesis of PGand protein in both explant and 3D agarose gel culture mod-els9e12, while static compressive loading can lead to de-creases in biosynthesis13e16. In vitro, static and dynamic

*Address correspondence and reprint requests to: Alan J.Grodzinsky, MIT, NE47-377, 77 Massachusetts Avenue,Cambridge, MA 02139, United States. Tel: 1-(617)-253-4969;Fax: 1-(617)-258-5239; E-mail: alg@mit.edu

Received 17 July 2008; revision accepted 9 September 2009.

249

loading cause activation of mitogen-activated protein(MAP) kinase pathway and ion channel signaling cascades,which result in time-varying changes in gene transcription ofmatrix proteins, catabolic enzymes, and transcriptionfactors17e20.

Recent research on the transduction of mechanical signalsinto changes in cell behavior has begun to elucidate the roleof integrin interactions with the extracellular matrix (ECM).Chondrocytes express a1b1, a2b1, a3b1, a5b1, avb3 andavb521e23. This expression can change with local microenvi-ronment, mechanical stimulation, and during osteoarthri-tis24e26. Integrins play a role in adhesion, cell survival,regulating matrix metabolism, and in chondrocyte responseto mechanical stimuli (see 27e29 for reviews). b1, a5b1,avb5 integrins have been shown to mediate chondrocyte ad-hesion to cut cartilage surfaces21,30. In monolayer studies,blocking a5b1 integrin interactions led to increased matrixmetalloproteinase-13 (MMP-13) activation and cellular apo-ptosis23,31. Upregulation of aggrecan mRNA and suppres-sion of matrix metalloproteinase-3 (MMP-3) mRNA levelsby dynamic stretching of monolayer chondrocytes wasshown to involve a b1 integrin-dependent pathway as wellas stretch-activated ion channels and autocrine/paracrinestimulation by interleukin-4 (IL-4)32,33. Other studies have

HN

NH

OCO2H

O

N

NH

N

NH

HO

Cl

Br

OH .HCl

Fig. 1. Chemical structure for PF001 (previously cited as S24740).PF001 is a synthetic RGD peptidomimetic that acts as an integ-rin-binding antagonist with broad specificity to av and a5 integrins.

250 D. H. Chai et al.: Integrins in chondrocyte mechanotransduction

also identified a b1 integrin-dependent translocation of pro-tein kinase C alpha (PKCa) to the cell membrane, increasedassociation of intracellular receptor for activated C-kinase1(RACK1) and PKCa with b1 integrin after mechanical stimu-lation34, and interactions between the integrin associatedprotein (CD47/IAP) and a5 integrin35, as possible down-stream signaling mechanisms. More recently, monolayerstudies demonstrated that blocking with anti-avb5 antibodycould also abolish chondrocyte responses to dynamicstretching36.

The use of physiologic dynamic compression as a stimu-lant for cartilage regeneration in tissue engineering hasbeen demonstrated by several groups10,11,19,37e41 and hasprompted studies of the mechanisms by which cells in tissueengineered constructs transduce signals under physiologicalloading conditions. While some studies suggest a role for ionchannels, especially calcium channels, in regulating sulfatedglycosaminoglycan (sGAG) synthesis in response to dy-namic compression42,43, studies in 3D culture models havesuggested that the roles are more complicated than that elu-cidated in monolayer culture models42. Monolayer studieshave provided insights into signaling pathways involved inchondrocyte mechanotransduction, but the role of cell inter-actions with newly synthesized and assembled matrix mac-romolecules in 3D geometries is less well understood.

The goal of this study was to examine the role of integrinsin chondrocyte response to dynamic compression by cellsin 3D agarose gel culture at early times in culture, during ini-tial evolution of the pericellular matrix, but prior to significantaccumulation of further-removed ECM. Chondrocyte cul-tures in agarose at early times, with little pericellular matrixpresent, can respond to dynamic compression with in-creased sulfate incorporation and sGAG accumula-tion10,11,39. They also have the added benefit of permittingthe use of multiple methods for perturbing integrinematrixinteractions, including the use of blocking antibodies similarto those used in monolayer studies, which have diffusivelimitations in native cartilage tissue.

Methods

CELL HARVEST AND CULTURE

Chondrocytes were isolated from the femoral condyle cartilage of 2- to3-week old bovine calves (Research 87, Marlborough, MA) by sequentialdigestion in 0.2% pronase (Protease type XIV, Sigma) and 0.025% Collage-nase-P (Roche), as described previously44. Cells were seeded in 2% aga-rose (low melting-temperature, Invitrogen) at concentrations of 15 millioncells/mL using a stainless steel casting frame38,45, in a slab geometry ap-proximately 1.6 mm thick. 4 mm diameter disks were cored from the slab us-ing a dermal punch and cultured in 1% ITS-supplemented feed medium (highglucose Dulbecco’s Modified Eagles Medium (DMEM), 0.1 mM nonessentialamino acids, 0.4 mM proline, 100 U/mL PSA e penicillin, streptomycin, am-photeracin, 10 mg/mL ascorbate).

INTEGRIN-BLOCKING COMPOUNDS

Table IMolecular weight and relative specificities for PF001, PF002,PF003. IC50s were measured using specific integrin-transfectedHEK 293 cell-adhesion assays as in Ref. 46. Data obtained from

Pfizer, Inc. PF001 was previously cited as S24746

Mol Wt IC50 (nM)

avb3 avb5 a5b1

PF001 (S247) 569.8 0.40 1.50 64PF002 388.9 179 1660 1.23PF003 681.7 0.627 1.38 8940

An array of integrin-blocking compounds was used including small-mole-cule peptidomimetics, function-blocking antibodies, and RGD-containing dis-integrins. Small-molecule compounds have the advantage of rapid diffusiontimes, which allow for more spatial uniformity of treatment. PF001 (previouslyreferenced as S24746), PF002, PF003 are synthetic peptidomimetics (ob-tained from Pfizer; Fig. 1) of the conserved amino acid motif RGD (arginineeglycineeaspartic acid) and are potent in vitro antagonists (with IC50s on theorder of 1 nM) of ligand interaction with specific integrins. Their molecularweights and potencies against selected integrins as measured in integrin-overexpressing cell-adhesion assays are summarized in Table I. PF001 isa relatively broad-spectrum blocker while PF002 and PF003 are more spe-cific blockers of a5b1 and avb3 integrins, respectively.

To test activity and toxicity of the peptidomimetic compounds, a cell-adhe-sion assay was performed using an RGD-conjugated comb copolymer surfacewhich promotes integrin-mediated adhesion and prevents non-specific

adhesion47e49. Prepared surfaces were obtained50 in which cover slips werespin-coated with a poly (methyl methacrylate)-graft-poly (ethylene oxide),PMMA-g-PEO, a comb copolymer with a fraction of PEO side chains function-alized with maleimide groups using N-(p-maleimidophenyl) isocyanate(PMPI), and conjugated with a PHSRN-K-GRGDSP peptide (RGD peptideþfibronectin synergy site). The RGD-containing peptide was presented indispersed and clustered conformations. Further details of polymer synthesis,peptide synthesis, and surface preparation are described in Kuhlmanet al.48. Cover slips were placed at the bottom of a 24-well plate and held inplace by silicon rings. After isolation, 25,000 chondrocytes were seeded perwell and given 2 h to attach. PF001, PF002, PF003 were then added at con-centrations of 0, 50, 100, or 200 mM. Cells were cultured for 7 days, with me-dium changes every other day. Cells were imaged daily using an invertedmicroscope to qualitatively observe adhesion and spreading. After 7 days,cells were stained with fluorescein diacetate (FDA)/ethidium bromide (EtBR)(live/dead assessment) and imaged using an inverted UV microscope.

Echistatin (Sigma) is a 5.4 kDa disintegrin peptide that non-selectivelyblocks the activities of several integrins including avb3, a5b1, and aIIbb3with IC50s on the order of 10 nM51. Function-blocking antibodies to avb3(MAB1976, clone LM609), av (MAB1953, clone P3G8), a5 (MAB1956, cloneP1D6), and b1 (MAB1965, clone JB1A) integrins were obtained from Chem-icon. Cross-reactivity of blocking antibodies MAB1976, MAB1956, andMAB1965 to bovine integrins have been previously demonstrated52e54 andare supported by dot-blot analysis (Fig. 1S). Echistatin31 and the blocking an-tibodies21,23,34,55,56 have been previously used on chondrocytes with no re-ported cell death. All antibody preps had endotoxin levels below 0.5 EU/mL(LAL method, Lonza QCL-1000 test kit).

MECHANICAL LOADING

On day of casting, 3e4 chondrocyte-seeded agarose disks maintained infree-swelling culture were pre-incubated for 18 h in one of the following con-ditions: (1) untreated control (feed medium), (2) 200 mM PF001, PF002, orPF003, (3) 5 mg/mL integrin-blocking antibody (each antibody used in sepa-rate experiments), (4) 1 mM echistatin. Parallel sets of disks from treatmentconditions (1)e(4) were subjected to dynamic compression. Immediatelyprior to loading, 5 mCi/mL of 35S-sulfate and 10 mCi/mL of 3H-proline wereadded to the medium. Disks were then subjected to a 24-h, 1 Hz continuous,sinusoidal unconfined dynamic compression at 2.5% dynamic strain ampli-tude (in displacement feedback control) superimposed on a 7% static offsetstrain, using a non-porous polysulfone loading chamber and platens in an in-cubator-housed loading apparatus57. Free-swelling cultures over the same24-h incubation period served as controls. Throughout the applied compres-sion, the total harmonic distortion of the resulting continuously measured dy-namic load signal was <25%. The applied cyclic displacement waveformand resulting load waveform were similar to that shown in Ref. 37. Uponcompletion of loading, plugs were washed 3� 20 minutes in phosphate buff-ered saline (PBS) with 142 mg/mL sodium sulfate and 50 mg/mL L-proline toremove free radiolabel, and digested in 1 mL Proteinase K (0.1 mg/mL,Roche) at 60�C overnight (w16 h). Experiments were repeated using chon-drocytes isolated from 1e6 animals (as noted below).

35S-sulfate

251Osteoarthritis and Cartilage Vol. 18, No. 2

BIOCHEMICAL MEASUREMENTS

lortnoclle

GAG

******

***+

++

+

1.6

1.4

1.2

Radiolabel incorporation rates were measured by scintillation count-ing58,59. sGAG content in the proteinase K digests and collected mediumwas assayed by DMMB dye binding9. DNA was quantified by Hoechst33258 dye-binding assay60.

wseerf

d

1.0

DATA ANALYSIS

Free swell Dynamic compression

untreated control

PF001 PF002 PF003 untreated control

PF001 PF002 PF003

etaertnuot

evitaleR

0.8

0.6

0.4

0.2

0.0

Radiolabel incorporation and sGAG content data were normalized to DNAcontent to account for disk-to-disk variations in cell number. Data were fur-ther normalized to averaged free-swell controls for each animal to accountfor baseline animal-to-animal variability. All data are shown as mean� 95%confidence interval (CI). The number of observations varied with treatment,as detailed in the figure legends. A ShapiroeWilk test was used to test fornormality of data. One-way and 2-way ANOVA were used to test for the ef-fect of treatment (mechanical and/or pharmacological) followed by post-hocBonferoni pairwise comparisons to correct for multiple comparisons frommultiple treatments. A P-value< 0.05 was considered significant. Statisticalanalysis was performed using Systat 12 software (Richmond, CA).

Fig. 2. Biosynthetic response of chondrocytes in agarose culture to24 h of dynamic compression (1.0 Hz, 2.5% amplitude) in the pres-

Results ence of small-molecule integrin blockers PF001, PF002, PF003(200 mM). Data were normalized by the averaged untreated free- ACTIVITY AND TOXICITY OF PF001, PF002, PF003

swell control for each animal. Data shown as mean � 95% CI,n¼ 6e19 disks from 2e6 animals (PF001 n ¼ 6, PF002 n¼ 6,PF003 n¼ 19). Sulfate incorporation as measured by radiolabel in-corporation. GAG accumulation as measured by DMMB dye-binding assay. ***P< 0.0005 relative to free-swell, þP< 0.0005

relative to untreated controls.

RGD-conjugated comb copolymer surfaces that can pro-mote binding by the avb3 or a5b1 integrins were used toconfirm activity of PF001, PF002, and PF003 on chondro-cytes. By day 6 in culture, untreated cells seeded ontoRGD-conjugated surfaces showed a spread morphology,while cells treated with 100 mM PF001, PF002, or PF003 re-mained rounded on the RGD-conjugated surfaces through-out the 7-day treatment (Supplementary Fig. 2S). Treatmentwith 1 mM echistatin served as an assay control. At concen-trations up to 200 mM of each compound over 7 days, therewas no qualitative increase in ethidium bromide-stainedcells (approximately 10%) (Fig. 2S) compared to untreatedcontrols or quantitative changes in DNA content (Fig. 3S),suggesting no adverse effects on cell viability.

EFFECTS OF SMALL-MOLECULE BLOCKERS ON GAG

BIOSYNTHESIS AND ACCUMULATION WITH DYNAMIC

COMPRESSION OF AGAROSE GEL CULTURES

To test the role of integrineECM interactions in chondro-cyte biosynthetic response to dynamic compression, aga-rose cultures were incubated with small-moleculepeptidomimetics of the RGD-binding sequence recognizedby integrins such as avb3 and a5b1 to block these interac-tions and subjected to a 24-h continuous, unconfined dy-namic compression. At the end of the loading period,untreated free-swelling cultures incorporated an averageof 192 pmol sulfate/mg DNA/h and accumulated 7.38 mgGAG/mg DNA. Treatment with PF001, PF002, or PF003did not alter DNA content (Supplementary Fig. 3S). DNAcontent was used to normalize data for all subsequent anal-yses. In free-swelling controls, PF001 and PF002 did notsignificantly change sulfate incorporation or sulfated GAGcontent, while PF003 (200 mM) resulted in a 27% decreasein sulfate incorporation (P< 0.0005 compared to untreatedcontrol using post-hoc Bonferoni pairwise comparison)and 17% decrease GAG content (P< 0.0005 vs untreatedcontrol) (Fig. 2). Free-swell levels for sulfate incorporationand GAG content under all treatment conditions can befound in Supplementary Materials (Fig. 5S).

Consistent with previous studies10,11,39, low amplitude(<10% strain amplitude) dynamic compression at 1 Hz fre-quency increased sulfate incorporation rates by approxi-mately 25% (to 243 pmol sulfate/mg DNA/h, P< 0.0005 vsfree-swell control) after 24 h of compression in untreated

agarose gel plugs (Fig. 2). GAG loss to the medium wasminimal: <10% of the total GAG content (i.e., agarosedisk plus medium combined) or approximately 0.7 mgGAG/mg DNA, was lost to the medium over 24 h in free-swell culture. While GAG loss to medium increased byw75% to 1.2 mg GAG/mg DNA with dynamic compression,possibly due to increased transport, enhanced synthesiscaused a net increase in GAG accumulation within the disksafter 24 h of dynamic compression by w14% (to 8.51 mgGAG/mg DNA, P< 0.0005 vs free-swell control) (Fig. 2).Since the experiment was conducted during early times inculture and GAG loss was minimal, the majority of theGAG content was newly synthesized and GAG accumula-tion mirrored sulfate incorporation trends.

PF001-treated samples responded to dynamic compres-sion with similar increases in sulfate incorporation (27%,P< 0.0005 compared to treated free-swell control in post-hoc Bonferoni comparison) and GAG accumulation (15%,P¼ 0.089 vs treated free-swell control, P¼ 0.017 uncor-rected) (Fig. 2). PF002-treated samples responded to dy-namic compression with slight increases in sulfateincorporation (12%, P¼ 0.16 vs treated free-swell control,P¼ 0.03 uncorrected) and GAG accumulation (6%). In con-trast, PF003-treated samples showed no stimulation of sul-fate incorporation or GAG accumulation with dynamiccompression (Fig. 2). None of the treatments (PF001,PF002, or PF003) affected GAG loss to the medium. Takentogether, markers of PG synthesis (sulfate incorporationand GAG content) were responsive to dynamic compres-sion and sensitive to the integrin blocker PF003, the mostselective inhibitor of avb3 function. A dose response exper-iment up to the efficacious concentration of 200 mM withPF003 was then conducted: PG synthesis response to dy-namic compression was measured at 0, 100, 150, 200 mMPF003. Free-swelling disks showed a graded decrease insulfate incorporation and GAG accumulation with increas-ing concentrations of PF003 [Fig. 4S(A, B)]. Stimulation ofPG synthesis (sulfate incorporation and GAG accumulation)

252 D. H. Chai et al.: Integrins in chondrocyte mechanotransduction

by dynamic compression relative to free-swell also variedwith increasing concentrations of PF003, with a slight(10%, P¼ 0.8 compared to treated free-swell in Bonferonipost-hoc pairwise comparison, P¼ 0.15 uncorrected) in-crease in sulfate incorporation and GAG accumulation at100 mM, and no stimulation by 150 mM PF003 (Fig. 4S).

BROAD-SPECTRUM BLOCKERS PF001 AND ECHISTATIN

SHOWED LITTLE EFFECT ON RESPONSE TO DYNAMIC

COMPRESSION

Echistatin is a disintegrin containing the RGD-motif that hasbroad specificity to integrins51. In free-swelling controls, echis-tatin decreased sulfate incorporation by 43% (P< 0.0005compared to untreated control in Bonferoni post-hoc pairwisecomparison) and GAG content by 21% (P< 0.0005 vs un-treated control) [Fig. 3(A)], and increased GAG loss to mediumby77%. Incontrast,PF001 had noeffect on radiolabel incorpo-ration or GAG accumulation in free-swelling conditions. WithPF001 treatment, dynamic compression increased sulfate in-corporation by w20% (P< 0.0005 compared to treated free-swell) and slightly increased GAG content (P¼ 0.071 vstreated free-swell, P¼ 0.02 uncorrected). In contrast, withechistatin treatment, dynamic compression increased sulfateincorporation in treated samples by w20% (P¼ 0.125,P¼ 0.032 uncorrected), and dynamic compression did notstimulate GAG accumulation [Fig. 3(B)]. This may be due toGAG loss to medium, not seen in PF001-treated samples,since the total GAG content (agarose disk þ medium) in-creased by &sim;9% after 24 h dynamic compression in thepresence of echistatin.

DIFFERENTIAL EFFECTS OF INTEGRIN BLOCKERS IN

FREE-SWELLING CULTURE

To further test the hypothesis that blocking specific integ-rineECM interactions can disrupt chondrocyte response todynamic compression, a series of blocking antibodieswere used in parallel experiments. Antibodies are larger in

llews

eerfot

evitalernoisserp

mocci

manydot

esnopseR 0.

0.

0.

0.

1.

1.

1.

untreated control

PF001 echistatin

lortnocdetaertnu

otevitale

R

***

***

35S-sulfateGAG

Free-swell response1.2

A B

1.0

0.8

0.6

0.4

0.2

0.0

Fig. 3. Effects of broad-spectrum integrin antagonists PF001 and echistalation of agarose cultures (B). Data shown as mean� 95% CI. (A) Sulfatcontrols. n¼ 6e51 samples from 2e11 animals (control n¼ 51, PF001 n¼PG synthesis as indicated by sulfate incorporation and GAG content wn¼ 6e13 samples from 2e4 animals (control n¼ 13, PF001 n¼ 6, ec

free-swe

size (w150 kDa), but in agarose culture at early times be-fore much accumulation of ECM, antibodies can still diffusein relatively easily61,62. avb3 blocker PF003 decreased sul-fate incorporation by approximately 30% (P< 0.0005 com-pared to untreated control by Bonferoni pairwisecomparisons) and GAG accumulation by 18% (P< 0.0005vs untreated control) in free-swelling culture [Fig. 4(A)].avb3 blocking antibody clone LM609 showed a slight butnon-significant decrease in sulfate incorporation at 5 mg/mL[Fig. 4(A)]. av blocking antibodies showed no effects onsulfate incorporation, GAG accumulation, or GAG loss infree-swelling culture [Fig. 4(A)]. In contrast, a5b1 blockerPF002 showed a slight increase in sulfate incorporation(9%) and GAG accumulation (6%) under free-swelling con-ditions (Fig. 5), and blocking antibodies to b1 (5 mg/mL,10 mg/mL) integrins increased sulfate incorporation rates,and GAG accumulation by 20e40% (P< 0.0005, vs un-treated control) [Fig. 5(A)]. None of the blocking antibodieshad significant affects on GAG loss to the medium.

aVb3 AND b1 BLOCKING ANTIBODIES, NOT PF002 OR a5

BLOCKING ANTIBODIES, ABOLISHED PG SYNTHESIS

RESPONSE TO COMPRESSION

In untreated samples, dynamic compression increasedsulfate incorporation rates by approximately 20%(P< 0.0005 compared to free-swell control by Bonferonipost-hoc comparison) and GAG accumulation by 14%(P< 0.0005 vs free-swell control). Treatment with PF003,avb3 blocking antibodies, and av blocking antibodies abro-gated this response to dynamic compression (P< 0.0005by ANOVA) [Fig. 4(B)]. In separate experiments withPF002 and a5 blocking antibody-treated samples, dynamiccompression resulted in a 12e15% increase in sulfate in-corporation (P¼ 0.079, 0.023, respectively vs treated free-swell controls, P¼ 0.011, 0.006 uncorrected), while GAGaccumulation increased by only 5e10% [Fig. 5(B)]. b1blocking antibody-treated samples showed no stimulationof sulfate incorporation or GAG accumulation by dynamic

0

2

4

6

8

1

2

4

6

untreated control

PF001 echistatin

******

*** 35S-sulfateGAG

Response to Dynamic Compression

tin in free-swelling culture (A) and on dynamic compression stimu-e incorporation and GAG content in free-swell relative to untreated

6, echistatin n¼ 7), ***P< 0.0005 relative to untreated control. (B)ith dynamic compression relative to treated free-swell controls.

histatin n¼ 7), ***P< 0.0005 relative to 1 (no change relative toll).

untreated control

PF003 5 μg/mLαvβ3

5 μg/mLαv

10 μg/mLαv

lortnocdetaertnu

otevitaler

esnopseR

******

35S-sulfate

GAG

Free-swell Response1.4

A

B

1.2

1.0

0.8

0.6

0.4

0.2

0.0

0

0.2

0.4

0.6

0.8

1

1.2

1.4

untreated control

PF003 5 μg/mLαvβ3

5 μg/mLαv

10 μg/mLαv

llews

eerfot

evitalernoisserp

mocci

manydot

esnopseR

******

35S-sulfateGAG

Response to Dynamic Compression

Fig. 4. Effects of blocking avb3 integrins on biosynthetic behavior infree-swelling culture and in response to dynamic compression.Data shown as mean � 95% CI. (A) Sulfate incorporation andGAG content in free-swelling agarose cultures relative to untreatedcontrols. n¼ 3e51 samples from 1e11 animals (control n¼ 51,PF003 n ¼ 29, avb3 n ¼ 16, 5 mg/mL av n ¼ 12, 10 mg/mL avn¼ 3), ***P< 0.0005 relative to untreated control. (B) PG synthesis(sulfate incorporation and GAG accumulation) after 24 h of dynamiccompression relative to respective treated free-swell controls.n ¼ 3e19 samples from 1e6 animals (control n ¼ 19, PF003n ¼ 19, avb3 n ¼ 3, 5 mg/mL av n ¼ 9, 10 mg/mL av n ¼ 3),

***P< 0.0005 relative to 1 (no change compared to free-swell).

lortnocdetaertnu

otevitaler

esnopseR

untreated control

PF002 5 g/mLα5

5 g/mLβ β1

10 g/mL1

******

35S-sulfateGAG

Free-swell Response1.8

A

B

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

******

0

0.2

0.4

0.6

0.8

1

1.2

1.4

untreated control

PF002 5 g/mL5

5 g/mL1

10 g/mL1

***** *35S-sulfateGAG

Response to Dynamic Compression

μ μ μ

μ μ μα β β

llews

eerfot

evitalernoisserp

mocci

manydot

esnopseR

Fig. 5. Effects of blocking a5b1 integrins on agarose cultures infree-swelling culture and in response to dynamic compression.Data shown as mean � 95% CI. (A) Sulfate incorporation andGAG content in free-swelling agarose cultures relative to untreatedcontrols. n¼ 3e51 samples from 1e11 animals (control n¼ 51,PF002 n ¼ 6, a5 n ¼ 6, 5 mg/mL b1 n ¼ 6, 10 mg/mL b1 n ¼ 3),***P< 0.0005 relative to untreated controls. (B) PG synthesis (sul-fate incorporation and GAG content) response after 24 h of dy-namic compression relative to respective treated free-swellcontrols. n ¼ 3e12 samples from 1e4 animals (control n¼ 12,PF002 n ¼ 6, a5 n ¼ 6, 5 mg/mL b1 n ¼ 6, 10 mg/mL b1 n ¼ 3).***P< 0.0005, **P¼ 0.001, *P¼ 0.023 relative to 1 (no change

compared to free-swell).

253Osteoarthritis and Cartilage Vol. 18, No. 2

compression [Fig. 5(B)]. None of these effects describedwere due to changes in GAG loss to medium. Whileblockers of avb3 and b1 integrins both appear to abrogatethe response to dynamic compression, they appear to beacting through distinct pathways considering their opposingeffects in free-swelling culture.

Discussion

Dynamic compression and other mechanical stimuli havebeen increasingly used in tissue engineering to promote de-velopment of cartilage constructs through increasing ECMcontent and mechanical properties37,38, and the use of imma-ture chondrocytes in tissue engineering and cartilage repairhas been shown to be an effective cell source for tissue engi-neering, with greater activity than adult chondrocytes63e65.Even at early times in culture, when little pericellular matrixis present, chondrocyte cultures in agarose can respond todynamic compression with increased sulfate incorporationand sGAG accumulation10,11,39. This response to 24 h ofcontinuous dynamic compression increased with number of

days in free-swell culture10. In long-term studies of the effectsof dynamic loading, the presence of a pre-elaborated pericel-lular matrix (either by seeding chondrons initially or culturingfor 2 weeks prior to loading) did not alter the stimulatoryeffects of extended dynamic loading, which increased withloading duration37. This suggests that interactions betweenthe cell and its surrounding matrix developed during dynamiccompression may play a greater role than pre-existing inter-actions. The goal of this study was to examine the role ofintegrineECM interactions in the response of chondrocytesto dynamic compression at early times in culture usinga 3D agarose culture of immature bovine chondrocytes asa model system. An added benefit of studying such interac-tions at early times is the ability to compare multiple integrinblockers, including antibodies, without the complicating issueof diffusion and penetration of antibodies into a dense tissuematrix. The results of this study suggest that multiple integ-rins (b1, avb3) appear to play a role in mechanotransductionand the chondrocyte’s ability to sense its local microenviron-ment; however these integrins appear to play opposing orcomplementary roles.

254 D. H. Chai et al.: Integrins in chondrocyte mechanotransduction

In the present study, blocking b1 integrin function with block-ing antibodies, or blocking avb3 integrins with either small-molecule antagonists or blocking antibodies, abolished PGstimulation by dynamic compression (as measured by sulfateincorporation or sGAG accumulation). While the concentrationof the small-molecule antagonists used in these functional as-says were much higher than the IC50s reported in Table I, pre-vious studies have confirmed the observation that higherconcentrations are necessary to see functional response, es-pecially in 3D culture models66. Our results using blocking an-tibodies support the specificity of these small-molecules aswell. As previously shown10,11,39, 24 h continuous unconfineddynamic compression stimulated PG synthesis at days 1e2 inculture. Measurable amounts of sGAG were accumulated inconstructs by the end of culture. Previous studies have alsoshown that a pericellular matrix begins developing within 4 hafter isolation67 and can be visualized at the cell surface onday 2 in agarose culture68. The b1 integrin subunit can associ-ate with a large number of differentially expressed alpha sub-units to form integrins with distinct ligand binding and cellsignaling characteristics. Echistatin and the RGD peptidomi-metics used in this study are expected to inhibit only the subsetof b1-containing integrins that interact with their ligands via theRGD sequence. avb3 and a5b1 are both RGD-recognizing in-tegrins. While the main binding partners for avb3 and a5b1 arevitronectin and fibronectin, respectively, avb3 has also beenshown to bind to fibronectin, fibrinogen, osteopontin, and col-lagen69. Blocking a5b1 specifically using the RGD peptidomi-metic PF002 or a5 blocking antibodies may partially decreasePG stimulation by dynamic compression, but this effect wasnot significant. Other potential b1 integrins include collagen re-ceptors a1b1 and a2b1, as well as a3b1 and a10b1.

While blocking both avb3 and b1 integrins prevented stim-ulation of PG synthesis by dynamic compression, these re-sults appeared to be acting through independentmechanisms. In this study, blocking b1 integrins in free-swell-ing culture with blocking antibodies resulted in a significantupregulation of sulfate incorporation, while blocking avb3 in-tegrins with small-molecule compounds resulted in a downregulation. Blocking avb3 integrins with antibodies resultedin no detectable change in basal sulfate incorporation. Previ-ous studies have suggested that avb3 and a5b1 have modu-lating roles in articular chondrocytes56. In those studies,treatment with anti-a5b1 antibody JBS5 induced a pro-in-flammatory response (upregulation of NO, prostaglandinE2 (PGE2), IL-6, IL-8, and IL-1b) in both normal and osteoar-thritic cartilage as well as bovine articular chondrocytes,while treatment with avb3 blocking antibody LM609 de-creased these pro-inflammatory signals and could regulatethe a5b1 response in a dominant-negative fashion56. Otherstudies demonstrated similar responses when blocking withanti-a5b1 antibodies or treating with fibronectin fragments23.In addition, studies have shown that blocking a1b1 and a2b1with blocking antibodies can stimulate MMP-13 productionthrough the MAP kinase pathway similar to treatment witha5b1 blocking antibodies or fibronectin fragments23,70,although their roles in mechanotransduction had not previ-ously been investigated to our knowledge. These studiesdemonstrate that the use of blocking antibodies can inducecellular responses by disrupting integrin signaling or causingabnormal signaling. While the stimulation of PG synthesis byblocking b1 integrins in agarose culture might appear to be incontradiction to these previous studies, it is possible that thestimulation observed, here, is part of increased turnover thathas been described previously with hyaluronan (HA) oligo-saccharide treatment of cartilage71. In addition, only blockersof b1 integrins that induced an increase in biosynthesis in

free-swelling controls resulted in abolishment of theresponse to dynamic compression, and the effects of block-ing antibodies were distinct from that of small-molecule an-tagonists, which likely block integrin interactions throughdifferent mechanisms. Taken together, these studies sug-gest that blocking b1 integrins with blocking antibodies maydecrease response to dynamic compression through a pro-inflammatory pathway or an analogous pathway resultingfrom abnormal integrin signaling.

Consistent with previous studies, we observed that treat-ment with different types of avb3 blockers can result in dis-tinctly different free-swelling responses56. However, allspecific blockers of avb3 integrins affected stimulation ofPG synthesis by dynamic compression. Thus, blockingavb3 appears to be inhibiting a signaling response directly re-sulting from dynamic compression, independent of effects infree-swelling culture. Finally, our observation that treatmentwith relatively broad-spectrum blockers (PF001 and echista-tin) that target RGD-binding integrins, and which bind to a5b1and avb3 with similar affinities, did not affect stimulation of PGsynthesis by dynamic compression, suggests a modulatoryrole for a5b1 and avb3 as seen in previous studies56.

It is important to note that previous studies of the role ofa5b1 integrins in chondrocyte mechanotransduction fo-cused on chondrocyte monolayer cultures23,31e35. Thepresent study utilizes 3D gel culture, and the results in 3Dsuggest a more complex view in which multiple integrinsmay be playing a role in regulating chondrocyte responseto compression within the gel construct. This hypothesis isconsistent with research on the role of ion channel signal-ing, where more complex interactions were observed in3D compared to 2D culture43. A variety of mechanical sig-nals are present during physiological (3D) dynamic com-pression (e.g., fluid flow, pressure gradients, streamingpotentials, deformation)4e6. Even at early times in culture,the presence of PGs would contribute to streaming poten-tials and osmotic gradients, while cell and pericellular matrixdeformation72 and fluid flow may affect receptoreECM in-teractions. With the multitude of signals, it is conceivablethat individual ligandesurface receptor interactions maybe sensing different mechanical signals, mechanisms forwhich are beginning to be elucidated73. Finally, further stud-ies on downstream signaling events may shed light on howupstream mechanical signals may interact, and allow forbetter understanding as to how mechanical stimulationmay be utilized to optimize ECM development in cartilagetissue engineering applications.

Conflict of interest

E.C.A. is a former employee and stockholder of Pfizer,Inc.; D.W.G. is an employee and stockholder of Pfizer, Inc.

Acknowledgements

Funded in part by National Institutes of Health (NIH)/Na-tional Institute of Arthritis Musculoskeletal and Skin Dis-eases (NIAMS) Grant AR33236 and Pfizer, Inc, andGraduate Student Fellowships from National Science Foun-dation (NSF) and National Defense Science and Engineer-ing Graduate (NDSEG).

Supplementary material

Supplementary data associated with this article can befound in the online version, at doi:10.1016/j.joca.2009.09.002.

255Osteoarthritis and Cartilage Vol. 18, No. 2

References

1. Li G, Van de Velde SK, Bingham JT. Validation of a non-invasive fluo-roscopic imaging technique for the measurement of dynamic kneejoint motion. J Biomech 2008;41(7):1616e22.

2. Hodge WA, Fijan RS, Carlson KL, Burgess RG, Harris WH, Mann RW.Contact pressures in the human hip joint measured in vivo. Proc NatlAcad Sci U S A 1986 May;83(9):2879e83.

3. Herberhold C, Faber S, Stammberger T, Steinlechner M, Putz R,Englmeier KH, et al. In situ measurement of articular cartilage defor-mation in intact femoropatellar joints under static loading. J Biomech1999 Dec;32(12):1287e95.

4. Frank EH, Grodzinsky AJ. Cartilage electromechanics e II. A continuummodel of cartilage electrokinetics and correlation with experiments.J Biomech 1987;20(6):629e39.

5. Mak AF. Unconfined compression of hydrated viscoelastic tissues ea biphasic poroviscoelastic analysis. Biorheology 1986;23(4):371e83.

6. Mow VC, Holmes MH, Lai WM. Fluid transport and mechanical proper-ties of articular cartilage: a review. J Biomech 1984;17(5):377e94.

7. Jurvelin J, Kiviranta I, Saamanen AM, Tammi M, Helminen HJ. Partialrestoration of immobilization-induced softening of canine articular car-tilage after remobilization of the knee (stifle) joint. J Orthop Res 1989;7(3):352e8.

8. Behrens F, Kraft EL, Oegema Jr TR. Biochemical changes in articularcartilage after joint immobilization by casting or external fixation. J Or-thop Res 1989;7(3):335e43.

9. Sah RLY, Kim YJ, Doong JYH, Grodzinsky AJ, Plaas AHK, Sandy JD.Biosynthetic response of cartilage explants to dynamic compression.J Orthopaed Res 1989 Sep;7(5):619e36.

10. Buschmann MD, Gluzband YA, Grodzinsky AJ, Hunziker EB. Mechani-cal compression modulates matrix biosynthesis in chondrocyte/agar-ose culture. J Cell Sci 1995 Apr;108(Pt 4):1497e508.

11. Lee DA, Bader DL. Compressive strains at physiological frequencies in-fluence the metabolism of chondrocytes seeded in agarose. J OrthopRes 1997 Mar;15(2):181e8.

12. Kelly TA, Ng KW, Wang CC, Ateshian GA, Hung CT. Spatial and tem-poral development of chondrocyte-seeded agarose constructs infree-swelling and dynamically loaded cultures. J Biomech 2006;39(8):1489e97.

13. Kiviranta I, Jurvelin J, Tammi M, Saamanen AM, Helminen HJ. Weightbearing controls glycosaminoglycan concentration and articular carti-lage thickness in the knee joints of young beagle dogs. ArthritisRheum 1987 Jul;30(7):801e9.

14. Gray ML, Pizzanelli AM, Grodzinsky AJ, Lee RC. Mechanical and phys-iochemical determinants of the chondrocyte biosynthetic response.J Orthop Res 1988;6(6):777e92.

15. Caterson B, Lowther DA. Changes in metabolism of proteoglycans fromsheep articular-cartilage in response to mechanical-stress. BiochimBiophys Acta 1978;540(3):412e22.

16. Kim YJ, Sah RL, Grodzinsky AJ, Plaas AH, Sandy JD. Mechanical reg-ulation of cartilage biosynthetic behavior: physical stimuli. Arch Bio-chem Biophys 1994 May 15;311(1):1e12.

17. Fanning PJ, Emkey G, Smith RJ, Grodzinsky AJ, Szasz N, Trippel SB.Mechanical regulation of mitogen-activated protein kinase signaling inarticular cartilage. J Biol Chem 2003 Dec 19;278(51):50940e8.

18. Fitzgerald JB, Jin M, Dean D, Wood DJ, Zheng MH, Grodzinsky AJ. Me-chanical compression of cartilage explants induces multiple time-de-pendent gene expression patterns and involves intracellular calciumand cyclic AMP. J Biol Chem 2004 May 7;279(19):19502e11.

19. De Croos JN, Dhaliwal SS, Grynpas MD, Pilliar RM, Kandel RA. Cycliccompressive mechanical stimulation induces sequential catabolicand anabolic gene changes in chondrocytes resulting in increasedextracellular matrix accumulation. Matrix Biol 2006 Aug;25(6):323e31.

20. Fitzgerald JB, Jin M, Chai DH, Siparsky P, Fanning P, Grodzinsky AJ.Shear- and compression-induced chondrocyte transcription requiresMAPK activation in cartilage explants. J Biol Chem 2008 Mar 14;283(11):6735e43.

21. Kurtis MS, Schmidt TA, Bugbee WD, Loeser RF, Sah RL. Integrin-medi-ated adhesion of human articular chondrocytes to cartilage. ArthritisRheum 2003 Jan;48(1):110e8.

22. Salter DM, Hughes DE, Simpson R, Gardner DL. Integrin expression byhuman articular chondrocytes. Br J Rheumatol 1992 Apr;31(4):231e4.

23. Forsyth CB, Pulai J, Loeser RF. Fibronectin fragments and blocking an-tibodies to alpha2beta1 and alpha5beta1 integrins stimulate mitogen-activated protein kinase signaling and increase collagenase 3 (matrixmetalloproteinase 13) production by human articular chondrocytes.Arthritis Rheum 2002 Sep;46(9):2368e76.

24. Lucchinetti E, Bhargava MM, Torzilli PA. The effect of mechanical loadon integrin subunits alpha5 and beta1 in chondrocytes from matureand immature cartilage explants. Cell Tissue Res 2004 Mar;315(3):385e91.

25. Kim SJ, Kim EJ, Kim YH, Hahn SB, Lee JW. The modulation of integrinexpression by the extracellular matrix in articular chondrocytes. Yon-sei Med J 2003 Jun 30;44(3):493e501.

26. Lapadula G, Iannone F, Zuccaro C, Grattagliano V, Covelli M, Patella V,et al. Integrin expression on chondrocytes: correlations with the de-gree of cartilage damage in human osteoarthritis. Clin Exp Rheumatol1997 MayeJun;15(3):247e54.

27. Millward-Sadler SJ, Salter DM. Integrin-dependent signal cascades inchondrocyte mechanotransduction. Ann Biomed Eng 2004 Mar;32(3):435e46.

28. Knudson W, Loeser RF. CD44 and integrin matrix receptors participatein cartilage homeostasis. Cell Mol Life Sci 2002 Jan;59(1):36e44.

29. Mobasheri A, Carter SD, Martin-Vasallo P, Shakibaei M. Integrins andstretch activated ion channels; putative components of functionalcell surface mechanoreceptors in articular chondrocytes. Cell BiolInt 2002;26(1):1e18.

30. Kurtis MS, Tu BP, Gaya OA, Mollenhauer J, Knudson W, Loeser RF,et al. Mechanisms of chondrocyte adhesion to cartilage: role ofbeta1-integrins, CD44, and annexin V. J Orthop Res 2001 Nov;19(6):1122e30.

31. Pulai JI, Del Carlo Jr M, Loeser RF. The alpha5beta1 integrin providesmatrix survival signals for normal and osteoarthritic human articularchondrocytes in vitro. Arthritis Rheum 2002 Jun;46(6):1528e35.

32. Millward-Sadler SJ, Wright MO, Lee H, Nishida K, Caldwell H, Nuki G,et al. Integrin-regulated secretion of interleukin 4: a novel pathwayof mechanotransduction in human articular chondrocytes. J Cell Biol1999 Apr 5;145(1):183e9.

33. Millward-Sadler SJ, Wright MO, Davies LW, Nuki G, Salter DM. Mecha-notransduction via integrins and interleukin-4 results in altered aggre-can and matrix metalloproteinase 3 gene expression in normal, butnot osteoarthritic, human articular chondrocytes. Arthritis Rheum2000 Sep;43(9):2091e9.

34. Lee HS, Millward-Sadler SJ, Wright MO, Nuki G, Al-Jamal R, Salter DM.Activation of integrin-RACK1/PKCalpha signalling in human articularchondrocyte mechanotransduction. Osteoarthritis Cartilage 2002Nov;10(11):890e7.

35. Orazizadeh M, Lee HS, Groenendijk B, Sadler SJ, Wright MO,Lindberg FP, et al. CD47 associates with alpha 5 integrin and regu-lates responses of human articular chondrocytes to mechanical stim-ulation in an in vitro model. Arthritis Res Ther 2008 Jan 10;10(1):R4.

36. Holledge MM, Millward-Sadler SJ, Nuki G, Salter DM. Mechanical regu-lation of proteoglycan synthesis in normal and osteoarthritic humanarticular chondrocytes e roles for alpha5 and alphaVbeta5 integrins.Biorheology 2008;45(3e4):275e88.

37. Kelly TA, Wang CC, Mauck RL, Ateshian GA, Hung CT. Role of cell-as-sociated matrix in the development of free-swelling and dynamicallyloaded chondrocyte-seeded agarose gels. Biorheology 2004;41(3e4):223e37.

38. Kisiday JD, Jin M, DiMicco MA, Kurz B, Grodzinsky AJ. Effects ofdynamic compressive loading on chondrocyte biosynthesis in self-assembling peptide scaffolds. J Biomech 2004 May;37(5):595e604.

39. Chowdhury TT, Bader DL, Shelton JC, Lee DA. Temporal regulation ofchondrocyte metabolism in agarose constructs subjected to dynamiccompression. Arch Biochem Biophys 2003 Sep 1;417(1):105e11.

40. Demarteau O, Wendt D, Braccini A, Jakob M, Schafer D, Heberer M,et al. Dynamic compression of cartilage constructs engineered fromexpanded human articular chondrocytes. Biochem Biophys Res Com-mun 2003 Oct 17;310(2):580e8.

41. Mauck RL, Byers BA, Yuan X, Tuan RS. Regulation of cartilaginousECM gene transcription by chondrocytes and MSCs in 3D culture inresponse to dynamic loading. Biomech Model Mechanobiol 2006May 12.

42. Pingguan-Murphy B, Lee DA, Bader DL, Knight MM. Activation of chon-drocytes calcium signalling by dynamic compression is independentof number of cycles. Arch Biochem Biophys 2005 Dec 1;444(1):45e51.

43. Mouw JK, Imler SM, Levenston ME. Ion-channel regulation of chondro-cyte matrix synthesis in 3D culture under static and dynamic compres-sion. Biomech Model Mechanobiol 2006 Jun 10.

44. Ragan PM, Chin VI, Hung HH, Masuda K, Thonar EJ, Arner EC, et al.Chondrocyte extracellular matrix synthesis and turnover are influ-enced by static compression in a new alginate disk culture system.Arch Biochem Biophys 2000 Nov 15;383(2):256e64.

45. Kisiday J, Jin M, Kurz B, Hung H, Semino C, Zhang S, et al. Self-assem-bling peptide hydrogel fosters chondrocyte extracellular matrix pro-duction and cell division: implications for cartilage tissue repair. ProcNatl Acad Sci U S A 2002 Jul 23;99(15):9996e10001.

46. Shannon KE, Keene JL, Settle SL, Duffin TD, Nickols MA, Westlin M,et al. Anti-metastatic properties of RGD-peptidomimetic agentsS137 and S247. Clin Exp Metastasis 2004;21(2):129e38.

47. Irvine DJ, Mayes AM, Griffith LG. Nanoscale clustering of RGD peptidesat surfaces using Comb polymers. 1. Synthesis and characterizationof Comb thin films. Biomacromolecules 2001 Spring;2(1):85e94.

256 D. H. Chai et al.: Integrins in chondrocyte mechanotransduction

48. Kuhlman W, Taniguchi I, Griffith LG, Mayes AM. Interplay between PEOtether length and ligand spacing governs cell spreading on RGD-mod-ified PMMA-g-PEO comb copolymers. Biomacromolecules 2007 Oct;8(10):3206e13.

49. Koo LY, Irvine DJ, Mayes AM, Lauffenburger DA, Griffith LG. Co-regu-lation of cell adhesion by nanoscale RGD organization and mechani-cal stimulus. J Cell Sci 2002 Apr 1;115(Pt 7):1423e33.

50. Kindly provided by Maria Ufret from the Griffith Lab at MIT.51. Pfaff M, McLane MA, Beviglia L, Niewiarowski S, Timpl R. Comparison

of disintegrins with limited variation in the RGD loop in their binding topurified integrins alpha IIb beta 3, alpha V beta 3 and alpha 5 beta 1and in cell adhesion inhibition. Cell Adhes Commun 1994 Dec;2(6):491e501.

52. Duque H, LaRocco M, Golde WT, Baxt B. Interactions of foot-and-mouthdisease virus with soluble bovine alphaVbeta3 and alphaVbeta6 in-tegrins. J Virol 2004 Sep;78(18):9773e81.

53. Radel C, Rizzo V. Integrin mechanotransduction stimulates caveolin-1phosphorylation and recruitment of Csk to mediate actin reorganiza-tion. Am J Physiol Heart Circ Physiol 2005 Feb;288(2):H936e45.

54. Funderburgh JL, Funderburgh ML, Mann MM, Corpuz L, Roth MR. Pro-teoglycan expression during transforming growth factor beta-inducedkeratocyte-myofibroblast transdifferentiation. J Biol Chem 2001 Nov23;276(47):44173e8.

55. Genes NG, Rowley JA, Mooney DJ, Bonassar LJ. Effect of substratemechanics on chondrocyte adhesion to modified alginate surfaces.Arch Biochem Biophys 2004 Feb 15;422(2):161e7.

56. Attur MG, Dave MN, Clancy RM, Patel IR, Abramson SB, Amin AR.Functional genomic analysis in arthritis-affected cartilage: yin-yangregulation of inflammatory mediators by alpha 5 beta 1 and alpha Vbeta 3 integrins. J Immunol 2000 Mar 1;164(5):2684e91.

57. Frank EH, Jin M, Loening AM, Levenston ME, Grodzinsky AJ. A versa-tile shear and compression apparatus for mechanical stimulation oftissue culture explants. J Biomech 2000 Nov;33(11):1523e7.

58. Hascall VC, Handley CJ, McQuillan DJ, Hascall GK, Robinson HC,Lowther DA. The effect of serum on biosynthesis of proteoglycansby bovine articular cartilage in culture. Arch Biochem Biophys 1983Jul 1;224(1):206e23.

59. Sah RL, Doong JY, Grodzinsky AJ, Plaas AH, Sandy JD. Effects of com-pression on the loss of newly synthesized proteoglycans and proteinsfrom cartilage explants. Arch Biochem Biophys 1991 Apr;286(1):20e9.

60. Kim YJ, Sah RL, Doong JY, Grodzinsky AJ. Fluorometric assay of DNAin cartilage explants using Hoechst 33258. Anal Biochem 1988 Oct;174(1):168e76.

61. Leddy HA, Awad HA, Guilak F. Molecular diffusion in tissue-engineeredcartilage constructs: effects of scaffold material, time, and culture con-ditions. J Biomed Mater Res B Appl Biomater 2004 Aug 15;70(2):397e406.

62. De Rosa E, Urciuolo F, Borselli C, Gerbasio D, Imparato G, Netti PA.Time and space evolution of transport properties in agaroseechon-drocyte constructs. Tissue Eng 2006 Aug;12(8):2193e201.

63. Hidaka C, Cheng C, Alexandre D, Bhargava M, Torzilli PA. Maturationaldifferences in superficial and deep zone articular chondrocytes. CellTissue Res 2006 Jan;323(1):127e35.

64. Lu Y, Adkisson HD, Bogdanske J, Kalscheur V, Maloney W, Cheung R,et al. In vivo transplantation of neonatal ovine neocartilage allografts:determining the effectiveness of tissue transglutaminase. J Knee Surg2005 Jan;18(1):31e42.

65. Feder J, Adkisson H, Kizer N, Hruska K, Cheugn R, Grodzinsky A, et al.The promise of chondral repair using neocartilage. In: Sandell L,Grodzinsky A, Eds. Tissue Engineering in Musculoskeletal ClinicalPractice. Rosemont: American Academy of Orthopaedic Surgeons;2004:219e26.

66. Engleman VW, Nickols GA, Ross FP, Horton MA, Griggs DW, Settle SL,et al. A peptidomimetic antagonist of the alpha(v)beta3 integrin inhibitsbone resorption in vitro and prevents osteoporosis in vivo. J Clin In-vest 1997 May 1;99(9):2284e92.

67. Goldberg RL, Toole BP. Pericellular coat of chick embryo chondrocytes:structural role of hyaluronate. J Cell Biol 1984 Dec;99(6):2114e22.

68. Dimicco MA, Kisiday JD, Gong H, Grodzinsky AJ. Structure of pericellu-lar matrix around agarose-embedded chondrocytes. OsteoarthritisCartilage 2007 Oct;15(10):1207e16.

69. Hynes RO. Integrins: versatility, modulation, and signaling in cell adhe-sion. Cell 1992 Apr 3;69(1):11e25.

70. Ronziere MC, Aubert-Foucher E, Gouttenoire J, Bernaud J, Herbage D,Mallein-Gerin F. Integrin alpha1beta1 mediates collagen induction ofMMP-13 expression in MC615 chondrocytes. Biochim Biophys Acta2005 Oct 30;1746(1):55e64.

71. KnudsonW, Casey B, NishidaY, Eger W, KuettnerKE, Knudson CB. Hyalur-onan oligosaccharides perturb cartilage matrix homeostasis and inducechondrocytic chondrolysis. Arthritis Rheum 2000 May;43(5):1165e74.

72. Knight MM, Ghori SA, Lee DA, Bader DL. Measurement of the deforma-tion of isolated chondrocytes in agarose subjected to cyclic compres-sion. Med Eng Phys 1998 NoveDec;20(9):684e8.

73. Jalali S, del Pozo MA, Chen K, Miao H, Li Y, Schwartz MA, et al. Integ-rin-mediated mechanotransduction requires its dynamic interactionwith specific extracellular matrix (ECM) ligands. Proc Natl Acad SciU S A 2001 Jan 30;98(3):1042e6.