Osteoarthritis and Cartilage (2008) 16, 855e864

ª 2007 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.joca.2007.12.005

InternationalCartilageRepairSociety

Chondroitin sulphate impedes the migration of a sub-populationof articular cartilage chondrocytesL. C. Davies B.Sc., Ph.D., Dr.a, E. J. Blain B.Sc., Ph.D., Dr., B. Caterson B.Sc., Ph.D., Professorand V. C. Duance B.Sc., Ph.D., Professor*Connective Tissue Biology Laboratories, School of Biosciences, Cardiff University,Museum Avenue, Cardiff, CF10 3US, UK

Summary

Objective: To determine whether chondroitin sulphate (CS) impedes the migration of primary articular chondrocytes.

Design: Articular chondrocytes were isolated from young and skeletally mature bovine animals. Boyden chambers were used to quantify chon-drocyte migration on aggrecan in the presence and absence of CS chains. A novel in vitro model of cell migration into articular cartilage ex-plants was designed to visualise and quantify the migration of labelled chondrocytes into cartilage matrix which had been treated withchondroitinase ABC to remove CS chains present.

Results: A consistent trend of increased migration with both age groups of a sub-population of chondrocytes was demonstrated on aggrecanin the absence of CS. These data were supported by results from the in vitro model of chondrocyte migration which demonstrated increasingnumbers of a chondrocyte sub-population from both age groups of cartilage migrating into the chondroitinase ABC digested cartilage explantswith time in culture. Minimal migration of these chondrocytes was demonstrated into phosphate buffered saline (PBS) treated control explants.

Conclusions: We confirm that a sub-population of chondrocytes isolated from both young and skeletally mature articular cartilages have theability to migrate. We also demonstrate that CS chains inhibit the migration of these articular chondrocytes and that their removal by chon-droitinase ABC digestion enhances the migration of these chondrocytes. Such findings may provide a clinical application for improvingcell-based cartilage repair strategies by enhancing integration between endogenous and repair tissue.ª 2007 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Key words: Chondroitin sulphate, Chondrocyte, Migration, Aggrecan, Glycosaminoglycans, Cartilage.

Abbreviations: ACI autologous chondrocyte implantation, sGAG sulphated glycosaminoglycan, DMEM Dulbecco’s Modified Eagle’s Medium,ITS insulinetransferrineselenium, CS chondroitin sulphate, DMMB dimethylmethylene blue, PET polyethylene terephthalate, TAE Tris acetateEDTA, BSA bovine serum albumin, PBS phosphate buffered saline, PBS-T PBS/Tween 20�, FITC fluorescein isothiocyanate, CMFDA50chloromethylfluorescein diacetate, DMSO dimethylsulfoxide.

Introduction

Articular cartilage, once injured, does not have the ability toinitiate a repair response. This fact is partially attributed tothe static nature of the resident chondrocytes. Cell-basedmethods of surgical repair, such as autologous chondrocyteimplantation (ACI)1 have provided a means for the success-ful generation of a hyaline-like repair tissue with the correctbiomechanical characteristics of articular cartilage. Themajor limitation of this technique, as demonstrated by biop-sies taken from the repair site, is the lack of lateral integra-tion between repair tissue and endogenous cartilage2,3.This subsequently leads to degeneration of the repair

aCurrent address: Department of Oral Surgery, Medicine andPathology, Wound Healing Group, School of Dentistry, CardiffUniversity, Heath Park, Cardiff, CF14 4XY, UK.

*Address correspondence and reprint requests to: ProfessorV. C. Duance, Connective Tissue Biology Laboratories, Schoolof Biosciences, Cardiff University, Museum Avenue, Cardiff,CF10 3US, UK. Tel: 44-0-2920-874111; Fax: 44-0-2920-874594;E-mail: duance@cardiff.ac.uk

Received 30 August 2007; revision accepted 11 December 2007.

855

tissue4,5. The reason for this lack of integration is unknownbut may be due, in part, to the lack of migration of the im-planted chondrocytes and/or the matrix macromoleculesat the wound edge to which the cells are attempting to ad-here and migrate on.

The ability of isolated chondrocytes to migrate has notbeen widely studied6e9 and most have not taken into consid-eration the nature of the chondrocyte phenotype as well asthe established fact that once plated and passaged, chon-drocytes will ‘‘de-differentiate’’ into fibroblast-like cells10,11.It has previously been demonstrated that chondrocytes iso-lated from newborn calves do have the capacity to migrateon fibronectin6. However, to date, there have been no re-ports on the migratory ability of primary chondrocytes iso-lated from skeletally mature articular cartilage, which is ofparticular importance as cell-based methods of surgical re-pair generally use autologous chondrocytes isolated fromadult articular cartilage. The large aggregating proteoglycan,aggrecan, is generally considered to be anti-adhesive due tothe presence of its negatively charged sulphated glycosami-noglycan (sGAG) side chains12e17 and inhibits the migrationof numerous cell types18e21. Surprisingly, there have beenno studies on the effect of aggrecan on chondrocyte migra-tion. As aggrecan synthesis is known to be up-regulated at

856 L. C. Davies et al.: CS impedes migration in a sub-population of chondrocytes

the wound edge of injured articular cartilage22,23 we hy-pothesise that this matrix macromolecule may have an inhib-itory effect on the migration and integration of implantedchondrocytes into the damaged tissue.

The effects of enzymatically removing proteoglycansfrom the wound surface to enhance adhesion of implantedchondrocytes have been previously investigated21,24,25 butnone have studied whether these adherent chondrocytesgo on to migrate into the digested tissue and adopt the char-acteristics of the resident cells. Interestingly, some of thebest results have been seen using the enzyme chondroiti-nase ABC, an endolytic sGAG lyase, which directly targetsthe sGAG chains, without damage to the actual architectureof the cartilage21,24,25.

In this study we demonstrate, for the first time, that a sub-population of primary chondrocytes isolated from bothyoung and skeletally mature articular cartilages have theability to migrate. Using both a Boyden chamber model ofcell migration and an in vitro full thickness articular cartilageexplant model we also confirm that chondroitin sulphate(CS) impedes the migration of articular chondrocytes andthat prolonged digestion of cartilage with chondroitinaseABC promotes the migration of isolated chondrocytes intothe digested matrix.

Materials and methods

All reagents were purchased from Sigma (Poole, UK) unless otherwisestated.

ISOLATION OF PRIMARY ARTICULAR CARTILAGE

CHONDROCYTES

Primary chondrocytes were isolated from the medial and lateral condylesof the metacarpophalangeal joint of 7 day and 18e24-month-old bovine an-imals within 12 h of slaughter, as described previously26. Chondrocytes werecultured in Dulbecco’s Modified Eagle’s Medium with Hams F12 (DMEM/Hams F-12-Glutamax-1�, Invitrogen, Paisley, UK) supplemented with100 U/ml penicillin, 100 mg/ml streptomycin (both Invitrogen, Paisley, UK),50 mg/ml ascorbate-2-phosphate and 1� insulinetransferrinesodium sele-nite (ITS) (Invitrogen, Paisley, UK). Supplementation of the media with ITShas previously been shown to maintain the chondrocytic phenotype in mono-layer culture27.

GENERATION OF A CHONDROITINASE ABC DIGESTED

AGGRECAN SUBSTRATE

Monomeric aggrecan, A1D1 fraction, purified from bovine nasal cartilagewas digested with chondroitinase ABC at a concentration of 0.001 U/10 mg ofsGAG for 24 h at 37�C with agitation. Loss of the CS side chains was as-sessed by both the dimethylmethylene blue (DMMB) colorimetric assay28

and immunochemical slot blotting using the monoclonal antibodies 1-B-5and 2-B-6 (raised against chondroitinase ABC generated chondroitin-0 and4-sulphate stubs, respectively)29,30.

ASSESSMENT OF CELL MIGRATION: BOYDEN CHAMBER

ASSAY

Boyden chamber porous membranes [8 mm pore size polyethylene tere-phthalate (PET) membrane (Becton Dickinson, USA)] were coated bothsides with 50 mg/ml of either purified bovine nasal cartilage aggrecan orchondroitinase ABC treated aggrecan, in 0.1 M Tris-acetate buffer (TAE),pH 6.8, for 2 h at room temperature. Remaining binding sites were blockedwith 0.1% (w/v) bovine serum albumin (BSA) diluted in phosphate bufferedsaline (PBS; pH 7.4) for 1 h at room temperature and washed with PBS priorto storage at 4�C.

Isolated chondrocytes were seeded onto the upper side of the membraneat a density of 2� 105 cells/well as described previously6. Media was addedto the bottom of the well through the sampling port. Chondrocytes wereallowed to adhere to the coated membrane and migration assessed overa period of 1e24 h.

Non-motile cells were removed from the upper surface of the membrane.The number of migrated chondrocytes i.e., attached to the underside of the

membrane, was quantified by staining the cells with 0.1% (w/v) crystal violetdye as described previously6. Stained cells were visualised by inverting theplate inserts and viewing using an upright light microscope (Laborlux 12,Leitz DRMB, Leica, Germany) connected to a digital camera (CoolsnapRS Photometrics, USA). Images from three random visual fields, of0.25 mm2, in each well were recorded using the Coolsnap image captureprogram version 1.2 (Roper Scientific Inc., USA). The average number of mi-grated cells within this 0.25-mm2 field was counted, and the total number ofmigrated cells was calculated (surface area of Boyden chamber membraneis 30 mm2). The mean� standard error mean (S.E.M.) was calculated for eachgroup of samples and all data were assessed for normal distribution usingthe AndersoneDarling test (P> 0.05). Statistical significance was calculatedusing a one-way analysis of variance (ANOVA) and Bonferroni post-testwhere P< 0.05.

DEVELOPMENT OF AN IN VITRO ARTICULAR CARTILAGE

MODEL

Full thickness articular cartilage explants (5 mm� 4 mm) were excisedfrom the metacarpophalangeal joints of 7-day-old calves using a sharp scalpelblade and embedded in 2% agarose (Flowgen, UK) equilibrated with media.The full depth explant was orientated such that only the cut surface was left ex-posed (Fig. 1). Chondroitinase ABC (1 U/ml in PBS) was added to the exposedsurface and incubated for periods up to 24 h at 37�C. Following incubation, thesolution was removed and the explant washed three times in PBS.

IMMUNOHISTOCHEMISTRY

Cartilage explants were orientated at 90� to the cut chondroitinase ABCtreated edge in order to obtain full depth cryosections (10 mm) (Bright Instru-ment Co Ltd., UK), which were mounted on poly-L-lysine coated microscopeslides (BDH Laboratory Supplies, UK). Sections were rehydrated in PBS/0.001% Tween (PBS-T) 20� for 10 min, before blocking with normal wholegoat serum (1:20 dilution; DakoCytomation, Glostrup, Denmark) for 30 min atroom temperature. Serum was replaced with primary antibodies recognisingchondroitinase-generated ‘‘CS-stubs’’ chondroitin-0 and 4-sulphate29,30,using 1-B-5 and 2-B-6, respectively, and incubated overnight at 4�C in ahumidified chamber. Sections were washed in PBS-T and incubated withan anti-mouse fluorescein isothiocyanate (FITC)-conjugated secondary anti-body raised in goat (1:50 dilution; DakoCytomation, Glostrup, Denmark) for1 h at room temperature. After repeated washes in PBS-T, sections weremounted in Vectashield� containing propidium iodide (Vector LaboratoriesInc., CA, USA). Sections were viewed using an upright fluorescent micro-scope (Laborlux 12, Leitz DRMB, Leica, Germany) at 40� magnification.

CELL DEATH ASSAY: INCORPORATION OF ETHIDIUM

HOMODIMER

The effect of digesting cartilage explants with chondroitinase ABC (1 U/ml, 8 h) on chondrocyte viability was assessed using a cell death assay.Explants treated with chondroitinase ABC or PBS (controls) were incubatedwith 2 mM ethidium homodimer in PBS for 1.5 h at 37�C. Explants werewashed in PBS three times over 15 min before snap freezing in liquid nitro-gen and cryosections visualised using a fluorescence microscope at 20�magnification.

FLUORESCENCE LABELLING OF ISOLATED ARTICULAR

CARTILAGE CHONDROCYTES

50chloromethylfluorescein diacetate (CMFDA) (Molecular Probes EuropeBV, The Netherlands) was dissolved in anhydrous dimethylsulfoxide(DMSO) and diluted in sterile PBS to provide a working concentration of1.94 mM. Chondrocytes (1� 107) were resuspended in CMFDA and incu-bated at 37�C for 1 h. Labelled chondrocytes were centrifuged (1800 RPM,10 min), resuspended in culture media and seeded at 2� 106 cells/ml directlyonto the chondroitinase ABC treated and control (PBS treated) explants em-bedded in agarose as described above and cultured for periods up to 28 days.

QUANTIFICATION OF CELL MIGRATION INTO ARTICULAR

CARTILAGE EXPLANTS

The number of CMFDA-labelled chondrocytes that had migrated into thechondroitinase ABC digested and control articular cartilage explants wasquantified. Immunohistochemistry with antibodies 2-B-6 and 1-B-5 was per-formed using the same protocol as previously described, except that sectionswere blocked with whole donkey serum as a donkey anti-mouse Texas Red�

conjugated secondary antibody (1:50 dilution; GE Healthcare UK Ltd., UK)was used to visualise the CS-stubs. Sections were mounted in Vectashield�.

Fig. 1. Schematic representation of the in vitro cartilage explant model for measuring chondrocyte migration and integration. The explant wasexcised from the bovine metacarpophalangeal joint (1) and cut using a sharp scalpel blade (2). The excised cartilage was orientated to givea full thickness exposed cut edge for chondroitinase ABC treatment (3) and embedded in agarose leaving only the cut edge exposed and flush

with the surrounding agarose (4).

857Osteoarthritis and Cartilage Vol. 16, No. 8

Grids of 0.1 mm2 were used to quantify the migration of labelled cells into theexplants over the 28-day culture period. The total number of labelled cells pres-ent within the superficial, transitional and deep zones of the explants wascounted. The mean� S.E.M. was calculated for each group of samples andall data were assessed for normal distribution using the AndersoneDarlingtest (P> 0.05). Statistical significance was calculated using a two-samplet test.

The superficial zone has been defined as the region extending from thearticular surface to a depth of 100 mm into the tissue. The deep zone hasbeen defined for this purpose as the region extending 100 mm above the startof the subchondral bone.

Results

A SUB-POPULATION OF ARTICULAR CHONDROCYTES

DEMONSTRATE INCREASED MIGRATION ON

CHONDROITINASE ABC DIGESTED AGGRECAN

The ability of young and mature chondrocytes to migrateon aggrecan core protein e in the presence or absence ofCS chains, was assessed using the Boyden Chambermodel (Fig. 2). Of the chondrocytes seeded onto the upper-most side of the chamber membranes, only 8% migrated tothe lower side of the membrane irrespective of chondrocytesource. Both young [Fig. 2(A)] and mature [Fig. 2(B)] chon-drocyte populations were able to migrate on aggrecan. In-creased migration, for both age groups, was observed onthe chondroitinase ABC treated aggrecan when comparedto the native aggrecan substrate, however, this was onlysignificant in the young cells after 24 h (P< 0.05).

EXTENDED DIGESTION OF ARTICULAR CARTILAGE

IS NECESSARY TO REMOVE CS FROM THE CUT EDGE

In order to achieve sufficient digestion within all the carti-lage zones for the migration and integration of transplantedcells without causing permanent damage to the cartilage,optimisation of chondroitinase ABC concentration and incu-bation time for digestion of the cartilage explants were per-formed. Explants embedded in agarose were digested forup to 24 h with 1 U/ml enzyme and the degree of digestionascertained immunohistochemically using the ‘‘CS-stub’’neo-epitope antibodies 1-B-5 and 2-B-6 (Fig. 3).

The rate and amount of digestion between zones var-ied due to the differing concentrations of proteoglycanswithin these zones. Digestion within the superficial zoneof the cartilage explants was more extensive comparedto the other zones of the cartilage [Fig. 3(AeD)]. Chon-droitinase ABC treatment of the tissue for 1 h did not ex-tend the area of digestion beyond the cut surface withinthe transitional [Fig. 3(E)] and deep zones [Fig. 3(I)]. At4 h, further digestion was seen within the superficial layer[Fig. 3(B)], but within the deeper layers digestion of thecartilage was still seen to be minimal [Fig. 3(F) and (J)].An incubation period of 8 h resulted in sufficient digestionthrough all zones of the tissue. The depth of digestionfrom the exposed surface had increased within all zones,although digestion was far more limited in the transitional[Fig. 3(G)] and deep zones [Fig. 3(K)] relative to thesuperficial zone of the cartilage explant [Fig. 3(C)]. Aftera 24-h incubation period, complete digestion of the CSchains was observed within the superficial zone as evi-denced by a blanket of chondroitinase ABC generatedCS-stubs [Fig. 3(D)]. Increased staining was also ob-served in the transitional [Fig. 3(H)] and deep zone[Fig. 3(L)]. To preserve a balance between digestionand tissue damage, an optimal concentration of 1 U/mlchondroitinase ABC for an incubation period of 8 h waschosen for subsequent experiments using this in vitromodel for chondrocyte migration studies.

CHONDROITINASE ABC DIGESTION OF ARTICULAR

CARTILAGE DOES NOT INDUCE CELL DEATH

To determine the extent of cell death in the explants afterincubation with chondroitinase ABC, digested explantswere labelled with ethidium homodimer which is only takenup by dead cells (red). Cell death was observed at the cutedge and in the deep zone of the chondrocytes after diges-tion with chondroitinase ABC for 8 h [Fig. 4(A)]. The amountof cell death in explants treated with PBS was utilised asa negative control; cell death was only observed at thecut edge and in the deep zone [Fig. 4(B)] indicating that

0

20

40

60

80

100

120

140

160

Incubation Time (Hours)

Averag

e N

um

ber o

f M

ig

ra

ted

Ch

on

dro

cytes / 0.25m

m2

A

B

0

20

40

60

80

100

120

140

160

1 18 24Incubation Time (Hours)

Averag

e N

um

ber o

f M

ig

rated

Ch

on

dro

cytes / 0.25m

m2

Native Aggrecan Chondroitinase ABC Digested Aggrecan

Native Aggrecan Chondroitinase ABC Digested Aggrecan

6

1 18 246

*

Fig. 2. Numbers of chondrocytes isolated from (A) young and (B) mature articular cartilages migrating on native and chondroitinase ABCdigested aggrecan over 24 h. Values are expressed �S.E.M. (n¼ 3) *P< 0.05.

858 L. C. Davies et al.: CS impedes migration in a sub-population of chondrocytes

chondroitinase ABC did not directly induce any additionalcell death.

A SUB-POPULATION OF ARTICULAR CHONDROCYTES

MIGRATE INTO CHONDROITINASE ABC DIGESTED

EXPLANTS IN A TIME DEPENDENT MANNER

To determine the extent of cell migration into the chondroi-tinase ABC digested cartilage explants, chondrocytes werelabelled with CMFDA and monitored using fluorescencemicroscopy. In agreement with results from the Boydenchamber experiments (Fig. 2), CS chains on the aggrecancore protein inhibit the migration of isolated chondrocytesinto the cut edge of young articular cartilage explants[Figs. 5(B) and 6(B)]. Minimal migration of CMFDA-labelledchondrocytes, isolated from both young [Fig. 5(A)] and ma-ture [Fig. 6(A)] articular cartilages, was observed in the con-trol explants over the initial 7-day culture period. In explantscultured for longer than 14 days, a small amount of migrationinto the superficial and transitional zones of the control ex-plants was observed where concentrations of proteoglycansare at their lowest [Figs. 5(A) and 6(A)]. In comparison, therewas a marked migration of young [Fig. 5(B)] and mature[Fig. 6(B)] CMFDA-labelled chondrocytes into the chondroiti-nase ABC digested explants. Migration was only observed inthe zone of digestion, although CMFDA-labelled chondro-cytes were also observed at the cut edge. The majority ofthe chondrocytes that had migrated into the tissue were

observed within the superficial zone, consistent with thegreater degree of digestion within this region. In comparisonto the superficial and transitional zones, active migration ofchondrocytes into the deeper zones was substantially less[Figs. 5(B) and 6(B)].

Grids were used to quantify the number of CMFDA-labelled chondrocytes that had migrated into the superfi-cial, transitional and deep zones of the cut articularcartilage explants. Increasing numbers of labelled chon-drocytes isolated from both young [Fig. 5(C)] and mature[Fig. 6(C)] articular cartilages were demonstrated to mi-grate into the chondroitinase ABC digested explantswith time in culture [Fig. 5(C)]. This trend appeared toplateau after 14 days with the chondrocytes isolatedfrom the young articular cartilage [Fig. 5(C)]. In contrast,the number of chondrocytes isolated from the mature car-tilage that had migrated into the digested explants contin-ued to increase to day 28 [Fig. 6(C)]. Statistical analysisconfirmed significantly higher numbers of CMFDA-labelledmature chondrocytes in the chondroitinase ABC digestedexplants at days 7, 14 and 28 when compared to controls(P< 0.005, 0.5 and 0.001, respectively). Initially fewerCMFDA-labelled chondrocytes isolated from the maturearticular cartilage were seen to migrate into the cut ex-plants when compared to chondrocytes isolated fromthe young cartilage. At later time points the numbers oflabelled chondrocytes within the digested explants werecomparable between the age groups.

Fig. 3. Immunofluorescence photomicrographs, using monoclonal antibodies 1-B-5 and 2-B-6, demonstrating the presence of chondroitinaseABC generated CS-stubs originating from the exposed cut edge (white dotted line). CS-stub formation is present in the superficial (AeD),transitional (EeH) and deep (IeL) zones of the articular cartilage explant, with increasing depth from the cut edge with time in culture (upto 24 h). Chondroitinase ABC generated CS-stubs are labelled with FITC (green) and nuclei counterstained with propidium iodide (red).

Bar¼ 50 mm.

Fig. 4. Fluorescence photomicrographs illustrating ethidium homodimer (dead) labelled chondrocytes within full depth articular cartilage (A) 8 hchondroitinase ABC treated and (B) PBS treated explants. The cut edge is depicted by a white dotted line. SZ¼ superficial zone and

DZ¼ deep zone. Bar¼ 50 mm.

859Osteoarthritis and Cartilage Vol. 16, No. 8

Day 1 Day 7 Day 14Day 5 Day 21 Day 28

Superficial-

Transitional

Zone

Transitional

Zone

Deep Zone

B

Day 1 Day 7 Day 14Day 5 Day 21 Day 28

Superficial-

Transitional

Zone

Transitional

Zone

Deep Zone

A

0

10

20

30

40

50

60

70

Contro

l

Day 1

Treated

Day 1 Con

trol

Day 5

Treated

Day 5

Contro

l

Day 7

Treated

Day 7

Contro

l

Day 14 Trea

ted

Day 14 Con

trol

Day 21 Trea

ted

Day 21 Con

trol

Day 28 Trea

ted

Day 28

Treatment

Averag

e N

um

ber o

f M

ig

rated

Ch

on

dro

cytes / 0.1m

m2

C

*

P=0.073

P=0.109

ig. 5. Immunofluorescence photomicrographs demonstrating the migration of CMFDA-labelled young articular chondrocytes into (A) PBSontrol and (B) chondroitinase ABC digested explants over 28 days in culture. The antibodies 2-B-6 and 1-B-5 and a secondary Texased� conjugated antibody have been used to highlight the chondroitinase ABC generated CS-stubs (scale bar¼ 50 mm). The cut edge isepicted by a white dotted line. (C) Quantification of migrating CMFDA-labelled chondrocytes (green) visualised in (A) and (B). Data are

FcRd

presented as mean� S.E.M. (n¼ 3; *P< 0.05).

Day 1 Day 7 Day 14Day 5 Day 21 Day 28

Superficial-

Transitional

Zone

Transitional

Zone

Deep Zone

A

Day 1 Day 7 Day 14Day 5 Day 21 Day 28

Superficial-

Transitional

Zone

Transitional

Zone

Deep Zone

B

0

5

10

15

20

25

30

35

40

45

50

ControlDay 1

TreatedDay 1

ControlDay 5

TreatedDay 5

ControlDay 7

TreatedDay 7

ControlDay 14

TreatedDay 14

ControlDay 21

TreatedDay 21

ControlDay 28

TreatedDay 28

Treatment

Averag

e N

um

ber o

f M

ig

rated

Ch

on

dro

cytes / 0.1m

m2

C

* ***

P=0.072

P=0.064

**

Fig. 6. Immunofluorescence photomicrographs demonstrating the migration of CMFDA-labelled mature articular chondrocytes into (A) PBScontrol and (B) chondroitinase ABC digested explants over 28 days in culture. The antibodies 2-B-6 and 1-B-5 and a secondary TexasRed� conjugated antibody have been used to highlight the chondroitinase ABC generated CS-stubs (scale bar¼ 50 mm). The cut edge isdepicted by a white dotted line. (C) Quantification of migrating CMFDA-labelled chondrocytes (green) visualised in (A) and (B). Data are

presented as mean� S.E.M. (n¼ 3; *P< 0.05, **P< 0.01, ***P< 0.001).

862 L. C. Davies et al.: CS impedes migration in a sub-population of chondrocytes

MIGRATING ARTICULAR CHONDROCYTES ADOPT THE

MORPHOLOGY OF RESIDENT ARTICULAR CHONDROCYTES

The morphology of the CMFDA-labelled chondrocytesthat had migrated into the different zones of the cut cartilageexplants was assessed using fluorescence microscopy(Fig. 7). By day 21, the CMFDA-labelled chondrocytesthat had migrated into the explant tissue demonstrated themorphology of the resident cells of the zones that theyhad migrated into and were occupying lacunae (Fig. 7). Inparticular, labelled chondrocytes within the superficialzone were flattened and arranged in rows parallel to the ar-ticular surface of the explant [Fig. 7(A) and (B)]. Within thetransitional zone migrated chondrocytes demonstrateda more spherical morphology and were randomly distrib-uted within the extracellular matrix [Fig. 7(C) and (D)]. Thelacunae were clearly visible within both the transitionaland the deep zone. Labelled chondrocytes within thedeep zone were spherical and occupied lacunae; thesecells were arranged in columns perpendicular to the

Fig. 7. High magnification fluorescence photomicrographs of migrated CMarticular cartilages into the superficial (AeB), transitional (CeD) and deepexplants. Arrowheads indicate migrated chondrocytes displaying characte

articular surface and aligned with the resident chondrocyteswithin the explants [Fig. 7(E) and (F)].

Discussion

We have demonstrated for the first time that a sub-population of primary chondrocytes isolated from both youngand skeletally mature articular cartilages has the ability to mi-grate into native articular cartilage that has been pre-treatedwith chondroitinase ABC. Although the use of chondroitinaseABC in enhancing migration of synoviocytes in a partial-thickness defect in an in vivo rabbit model has been previ-ously described21, we report, for the first time, the ability ofmigrating chondrocytes to adopt the cellular morphologyand inherent characteristics of the resident tissue that hasbeen subject to the chondroitinase ABC digestion.

Using both a Boyden chamber model of cell migrationand an in vitro full thickness articular cartilage explantmodel we have shown that CS on aggrecan impedes the

FDA-labelled chondrocytes (green) isolated from young and mature(EeF) zones of the chondroitinase ABC treated articular cartilage

ristic morphology of cells within their respective zones. Bar¼ 10 mm.

863Osteoarthritis and Cartilage Vol. 16, No. 8

migration of articular chondrocytes and that prolonged di-gestion of cartilage with chondroitinase ABC promotes themigration of isolated chondrocytes into the digested matrix.Our data, using the Boyden chamber assay, suggest thatthere is a sub-population of cells within cartilage that hassignificantly greater migratory potential than the bulk of cellsas only 8% of the chondrocytes seeded onto the uppermostside of the chamber membranes, migrated to the lower side.This was true for both age groups of chondrocytes. Thismay relate to the identification of a chondroprogenitor cellpopulation within the superficial zone of articular cartilage31.At present there are no clear markers for this progenitor cellalthough their colony forming potential could be determined.The potential of this progenitor cell for cartilage repair hasyet to be fully assessed32.

The anti-adhesive properties of proteoglycans have previ-ously been demonstrated15 and specifically this is thought tobe due to the presence of their sGAG chains12e14,16,17. ThesGAG chain component has been shown directly to have ananti-migratory effect on fibroblasts14, endothelial cells19,nerve cells20 and synoviocytes21. Using the Boyden cham-ber assay we have demonstrated that a chondrocyte sub-population can migrate on aggrecan coated membranesand that removal of the CS glycosaminoglycan (GAG) chainsenhanced their migratory capacity. The concentration ofGAGs in articular cartilage is approximately 50 mg/ml33

which exhibits both anti-adhesive and anti-migratory proper-ties. However, the concentration of aggrecan used in this as-say was only 50 mg/ml which is too low a concentration toprevent adhesion and migration of the chondrocytes thuslimiting the effect of chondroitinase ABC digestion. There-fore, we used a cartilage explant model to investigate therole of the GAG chains in inhibiting chondrocyte migration.

Initially we had to ascertain appropriate chondroitinaseABC digestion conditions to denude the endogenous aggre-can of the CS GAGs. We utilised antibodies that specificallyrecognise the GAG ‘‘stubs’’ remaining after chondroitinaseABC digestion to determine the extent of digestion; a celldeath assay was also used to determine the effect of thisdigestion on the viability of the explants after digestion. As ex-pected, digestion of the surface zone, where the proteogly-can concentration is relatively low, was readily achieved butmore extensive digestion was necessary to obtain significantremoval of the CS chains from the deep zone. An incubationperiod of 8 h appeared to be optimal for digestion through allzones of the tissue. Although digestion was more limited inthe transitional and deep zones in comparison to that of thesuperficial zone of the explant allowing an incubation periodof 24 h resulted in complete digestion of the CS chains withinthe superficial zone with a blanket of chondroitinase ABCgenerated CS-stubs being seen. Complete digestion insuch a manner could ultimately result in permanent damageto the cartilage and the ability of the cartilage to withstandcompressive forces would be diminished.

Cell death induced by chondroitinase ABC digestion ofthe explants was not observed. Additionally, limited celldeath is seen at the cut edge caused by excision of the tis-sue with a scalpel blade in accordance with previous stud-ies34,35. As expected the zone of cell death within the deepzone was largest, a factor that can be attributed to celldeath induced during removal of the explant from thebone for experimentation. No further cell death was seendeeper into the tissue than that demonstrated in the PBStreated control explants that can be correlated with diges-tion of the tissue with chondroitinase ABC.

This optimised in vitro model was subsequently used tovisualise the migration of CMFDA-labelled chondrocytes

into articular cartilage explants. Negligible migration ofCMFDA-labelled chondrocytes from both young and maturearticular cartilages was visualised in control explants overthe first 7 days in culture confirming the anti-adhesive/anti-migratory nature of native articular cartilage6,15. At thelater time points a small amount of migration was observedin the superficialetransitional zones of the control explantswhere concentrations of proteoglycans are at their lowest33.This was expected as this is a static model of migration. Aspreviously demonstrated within a number of models, immo-bilisation of a synovial joint leads to loss of proteoglycansfrom the articular cartilage, allowing the migration of isolatedchondrocytes into the matrix36e42.

In comparison, labelled chondrocytes from both agegroups migrated into chondroitinase ABC digested explants,with increasing numbers of labelled cells observed within thedigested matrix with time in culture. This migration only ex-tended as far as the zone of digestion, demonstrating the in-hibitory effect of the CS GAG chains. Quantitative analysisshows that the total number of chondrocytes migrating com-pared to undigested controls is statistically significant inmany cases, especially with the mature chondrocytes, butclear trends are also observed with the cells derived fromyoung cartilage. In most cases the variability is quite highand the number of total cells is quite low, therefore n¼ 3 isinsufficient to achieve statistical significance. Analysis byzone showed, as expected, that most cells migrated intothe superficial zone but this again only reached significancewith the mature cells after 14 and 28 days. Interestingly, atthe early time points a sub-population of young cells clearlymigrate more rapidly than the cells derived from mature car-tilage. However, in both cases the number of cells plateausat a similar level again suggesting that there is a limited pop-ulation of chondrocytes capable of adhering to and migratinginto the explants consistent with the Boyden chamber data.Whether the removal of GAGs abolishes the inhibitory activ-ity of the proteoglycans or unmasks other pro-adhesive/pro-migratory matrix macromolecules, such as fibronectin andtype II collagen is at present unknown.

Interestingly by days 21 and 28, fluorescently labelled mi-grated chondrocytes were demonstrating the correct cellularmorphology for the zones into which they had migrated. La-belled chondrocytes within the superficial zone were typicallyflattened and arranged in rows parallel to the articular sur-face. Cells within the transitional and deep zones weremore spherical with a random distribution in the transitionalzone, and in columns perpendicular to the articular surfacewithin the deep zone. Migrated chondrocytes can clearly beseen to occupy lacunae within transitional and deep zones.

This is the first study to demonstrate that chondrocytesmigrating into an established articular cartilage matrix canadapt to the morphology of the resident chondrocytes al-though a recent study of neocartilage grafts produced in cul-ture also adopted a zonal organisation similar to nativecartilage43. In our current study, the morphology of the la-belled chondrocytes is likely to have been influenced bythe organisation of the matrix of the cartilage explant.Clearly, removal of GAGs by chondroitinase ABC appearsnot to have significantly affected this matrix organisationsuggesting that once the proteoglycans have been replacedthe cartilage will retain its normal biomechanical properties.

Effective tissue integration of cartilage grafts with endoge-nous cartilage requires adequate and appropriate matrixsynthesis that is assembled across the repair interface. Wehave previously demonstrated that a combination of IGF-1and TGFb1 enhances cartilage matrix biosynthesis in skele-tally mature articular chondrocytes (Davies et al., submitted).

864 L. C. Davies et al.: CS impedes migration in a sub-population of chondrocytes

Using this in vitro model we have clearly demonstrated thatremoval of CS GAG chains at the edge of a cartilage defectpromotes chondrocyte migration, without adversely affectingcell viability or morphology, and therefore is likely to beadvantageous in promoting a more effective repair. A combi-nation of these two strategies has the potential to greatlyreduce the incidence of poor cartilage graft integration, amajor problem in current cartilage repair strategies.

Conflict of interest

The authors have no conflict of interest.

Acknowledgements

The authors would like to acknowledge funding from theEngineering and Physical Sciences Research Council andCardiff University (LCD), Arthritis Research Campaign(EJB, VCD and BC) and Biotechnology and BiologicalSciences Research Council (BC).

References

1. Brittberg M, Lindahl A, Nilsson A, Ohlsson C, Isaksson O, Peterson L.Treatment of deep cartilage defects in the knee with autologous chon-drocyte transplantation. N Engl J Med 1994;331:889e95.

2. Gratz KR, Wong VW, Chen AC, Fortier LA, Nixon AJ, Sah RL. Biome-chanical assessment of tissue retrieved after in vivo cartilage defectrepair: tensile modulus of repair tissue and integration with host carti-lage. J Biomech 2006;39:138e46.

3. van de Breevaart Bravenboer J, In der Maur CD, Bos PK, Feenstra L,Verhaar JA, Weinans H, et al. Improved cartilage integration and inter-facial strength after enzymatic treatment in a cartilage transplantationmodel. Arthritis Res Ther 2004;6:R469e76.

4. Ghadially FN, Thomas I, Oryschak AF, Lalonde JM. Long-term results ofsuperficial defects in articular cartilage: a scanning electron-micro-scope study. J Pathol 1977;121:213e7.

5. Shapiro F, Koide S, Glimcher MJ. Cell origin and differentiation in the re-pair of full-thickness defects of articular cartilage. J Bone Joint SurgAm 1993;75:532e53.

6. Chang C, Lauffenburger DA, Morales TI. Motile chondrocytes from new-born calf: migration properties and synthesis of collagen II. Osteoar-thritis Cartilage 2003;11:603e12.

7. Shimizu M, Minakuchi K, Kaji S, Koga J. Chondrocyte migration to fibro-nectin, type I collagen, and type II collagen. Cell Struct Funct 1997;22:309e15.

8. Maniwa S, Ochi M, Motomura T, Nishikori T, Chen J, Naora H. Effects ofhyaluronic acid and basic fibroblast growth factor on motility of chon-drocytes and synovial cells in culture. Acta Orthop Scand 2001;72:299e303.

9. Morales TI. Chondrocyte moves: clever strategies? OsteoarthritisCartilage 2007;15:861e71.

10. Kuroda Y. Studies on cartilage cells in vitro. I. Morphology and growth ofcartilage cells in monolayer cultures. Exp Cell Res 1964;35:326e36.

11. Kuettner KE, Memoli VA, Pauli BU, Wrobel NC, Thonar EJ, Daniel JC.Synthesis of cartilage matrix by mammalian chondrocytes in vitro. II.Maintenance of collagen and proteoglycan phenotype. J Cell Biol1982;93:751e7.

12. Culp LA, Murray BA, Rollins BJ. Fibronectin and proteoglycans as deter-minants of cell-substratum adhesion. J Supramol Struct 1979;11:401e27.

13. Lewandowska K, Choi HU, Rosenberg LC, Zardi L, Culp LA. Fibronectin-mediated adhesion of fibroblasts: inhibition by dermatan sulfate pro-teoglycan and evidence for a cryptic glycosaminoglycan-bindingdomain. J Cell Biol 1987;105:1443e54.

14. Merle B, Malaval L, Lawler J, Delmas P, Clezardin P. Decorin inhibitscell attachment to thrombospondin-1 by binding to a KKTR-dependentcell adhesive site present within the N-terminal domain of thrombo-spondin-1. J Cell Biochem 1997;67:75e83.

15. Winnemoller M, Schmidt G, Kresse H. Influence of decorin on fibroblastadhesion to fibronectin. Eur J Cell Biol 1991;54:10e7.

16. Winnemoller M, Schon P, Vischer P, Kresse H. Interactions betweenthrombospondin and the small proteoglycan decorin: interferencewith cell attachment. Eur J Cell Biol 1992;59:47e55.

17. Yamagata M, Suzuki S, Akiyama SK, Yamada KM, Kimata K. Regula-tion of cell-substrate adhesion by proteoglycans immobilized on extra-cellular substrates. J Biol Chem 1989;264:8012e8.

18. Perris R, Perissinotto D, Pettway Z, Bronner-Fraser M, Morgelin M,Kimata K. Inhibitory effects of PG-H/aggrecan and PG-M/versicanon avian neural crest cell migration. FASEB J 1996;10:293e301.

19. Johnson WE, Caterson B, Eisenstein SM, Roberts S. Human interverte-bral disc aggrecan inhibits endothelial cell adhesion and cell migrationin vitro. Spine 2005;30:1139e47.

20. Johnson WE, Sivan S, Wright KT, Eisenstein SM, Maroudas A,Roberts S. Human intervertebral disc cells promote nerve growthover substrata of human intervertebral disc aggrecan. Spine 2006;31:1187e93.

21. Hunziker EB, Rosenberg LC. Repair of partial-thickness defects in artic-ular cartilage: cell recruitment from the synovial membrane. J BoneJoint Surg Am 1996;78:721e33.

22. Jeffrey JE, Thomson LA, Aspden RM. Matrix loss and synthesis follow-ing a single impact load on articular cartilage in vitro. Biochim BiophysActa 1997;1334:223e32.

23. Walker EA, Verner A, Flannery CR, Archer CW. Cellular responsesof embryonic hyaline cartilage to experimental wounding in vitro.J Orthop Res 2000;18:25e34.

24. Lee MC, Sung KL, Kurtis MS, Akeson WH, Sah RL. Adhesive force ofchondrocytes to cartilage: effects of chondroitinase ABC. Clin OrthopRelat Res 2000;286e94.

25. Hunziker EB, Kapfinger E. Removal of proteoglycans from the surface ofdefects in articular cartilage transiently enhances coverage by repaircells. J Bone Joint Surg Br 1998;80:144e50.

26. Vaughan-Thomas A, Young RD, Phillips AC, Duance VC. Characteriza-tion of type XI collageneglycosaminoglycan interactions. J Biol Chem2001;276:5303e9.

27. Chua KH, Aminuddin BS, Fuzina NH, Ruszymah BH. Insulinetransferrineselenium prevent human chondrocyte dedifferentiationand promote the formation of high quality tissue engineered humanhyaline cartilage. Eur Cell Mater 2005;9:58e67 (Discussion 67).

28. Little CB, Flannery CR, Hughes CE, Mort JS, Roughley PJ, Dent C, et al.Aggrecanase versus matrix metalloproteinases in the catabolism ofthe interglobular domain of aggrecan in vitro. Biochem J 1999;344(Pt 1):61e8.

29. Caterson B, Christner JE, Baker JR, Couchman JR. Production andcharacterization of monoclonal antibodies directed against connectivetissue proteoglycans. Fed Proc 1985;44:386e93.

30. Couchman JR, Caterson B, Christner JE, Baker JR. Mapping by mono-clonal antibody detection of glycosaminoglycans in connective tis-sues. Nature 1984;307:650e2.

31. Dowthwaite GP, Bishop JC, Redman SN, Khan IM, Rooney P,Evans DJ, et al. The surface of articular cartilage contains a progenitorcell population. J Cell Sci 2004;117:889e97.

32. Archer CW, Redman S, Khan I, Bishop J, Richardson K. Enhancing tissueintegration in cartilage repair procedures. J Anat 2006;209:481e93.

33. Freeman ME. Adult Articular Cartilage. Pitman Medical Publishing CoLtd 1979.

34. Redman S, Thomson BM, Archer CW. The cellular responses of articu-lar cartilage to sharp and blunt trauma. Osteoarthritis Cartilage 2004;12:106e16.

35. Tew SR, Kwan AP, Hann A, Thomson BM, Archer CW. The reactions ofarticular cartilage to experimental wounding: role of apoptosis. Arthri-tis Rheum 2000;43:215e25.

36. Tammi M, Saamanen AM, Jauhiainen A, Malminen O, Kiviranta I,Helminen H. Proteoglycan alterations in rabbit knee articular cartilagefollowing physical exercise and immobilization. Connect Tissue Res1983;11:45e55.

37. Saamanen AM, Tammi M, Jurvelin J, Kiviranta I, Helminen HJ. Proteo-glycan alterations following immobilization and remobilization in thearticular cartilage of young canine knee (stifle) joint. J Orthop Res1990;8:863e73.

38. Palmoski M, Perricone E, Brandt KD. Development and reversal of aproteoglycan aggregation defect in normal canine knee cartilage afterimmobilization. Arthritis Rheum 1979;22:508e17.

39. Palmoski MJ, Colyer RA, Brandt KD. Joint motion in the absence of nor-mal loading does not maintain normal articular cartilage. ArthritisRheum 1980;23:325e34.

40. 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;30:801e9.

41. Haapala J, Arokoski JP, Ronkko S, Agren U, Kosma VM,Lohmander LS, et al. Decline after immobilisation and recovery afterremobilisation of synovial fluid IL1, TIMP, and chondroitin sulphatelevels in young beagle dogs. Ann Rheum Dis 2001;60:55e60.

42. Caterson B, Lowther DA. Changes in the metabolism of the proteogly-cans from sheep articular cartilage in response to mechanical stress.Biochim Biophys Acta 1978;540:412e22.

43. Hayes AJ, Hall A, Brown L, Tubo R, Caterson B. Macromolecular orga-nization and in vitro growth characteristics of scaffold-free neocarti-lage grafts. J Histochem Cytochem 2007;55:853e66.