EXPERIMENTAL CELL RESEARCH 238, 265–272 (1998)ARTICLE NO. EX973858

In Vitro Chondrogenesis of Bone Marrow-DerivedMesenchymal Progenitor Cells

Brian Johnstone,1 Thomas M. Hering,* Arnold I. Caplan,† Victor M. Goldberg, and Jung U. Yoo

Skeletal Research Center and Department of Orthopaedics, *Department of Medicine, and †Department of Biology,Case Western Reserve University, 2080 Adelbert Road, Cleveland, Ohio 44106

cartilage when implanted in vivo [1–4]. A culture sys-A culture system that facilitates the chondrogenic tem has been developed in which these cells will un-

differentiation of rabbit bone marrow-derived mesen- dergo osteogenic differentiation in vitro [5]. However,chymal progenitor cells has been developed. Cells ob- attempts to develop in vitro conditions in which mesen-tained in bone marrow aspirates were first isolated by chymal progenitor cells isolated from postnatal mam-monolayer culture and then transferred into tubes and malian bone marrow will progress down the chondro-allowed to form three-dimensional aggregates in a genic lineage have been less successful. There are stud-chemically defined medium. The inclusion of 1007 M ies reporting in vitro chondrogenesis using postnataldexamethasone in the medium induced chondrogenic mammalian cells [6, 7], but none have demonstrateddifferentiation of cells within the aggregate as evi- histologically identifiable cartilage formation, althoughdenced by the appearance of toluidine blue metachro- type II collagen production has been detected, sug-masia and the immunohistochemical detection of type gesting at least prechondroid tissue production.II collagen as early as 7 days after beginning three- We have developed a culture system that facilitatesdimensional culture. After 21 days, the matrix of the the chondrogenic differentiation of postnatal mamma-entire aggregate contained type II collagen. By 14 days

lian marrow mesenchymal progenitor cells. This systemof culture, there was also evidence for type X collagenis an adaptation of the ‘‘pellet’’ culture system that waspresent in the matrix and the cells morphologicallyoriginally described as a method for preventing the phe-resembled hypertrophic chondrocytes. However, chon-notypic modulation of chondrocytes in vitro [8, 9]. Moredrogenic differentiation was achieved in only approxi-recently, the system has been used in studies of the ter-mately 25% of the marrow cell preparations used. Inminal differentiation of growth-plate chondrocytes [10,contrast, with the addition of transforming growth fac-11]. This culture system allows cell–cell interactionstor-b1 (TGF-b1), chondrogenesis was induced in allanalogous to those that occur in precartilage condensa-marrow cell preparations, with or without the pres-tion during embryonic development [12]. However, thisence of 1007 M dexamethasone. The induction of chon-

drogenesis was accompanied by an increase in the al- cell configuration is not sufficient for the induction ofkaline phosphatase activity of the aggregated cells. chondrogenesis: the chondrogenic differentiation of theThe results of RT-PCR experiments indicated that marrow-derived progenitor cells required the use of aboth type IIA and IIB collagen mRNAs were detected defined medium to which were added certain bioactiveby 7 days postaggregation as was mRNA for type X factors, including dexamethasone and TGF-b1. Thiscollagen. Conversely, the expression of the type I colla- study describes the development of the system, and thegen mRNA was detected in the preaggregate cells but consequent production of hypertrophic chondrocytes bywas no longer detectable at 7 days after aggregation. the differentiation of bone marrow-derived mesenchymalThese results provide histological, immunohistochem- progenitor cells. This system provides a means for study-ical, and molecular evidence for the in vitro chondro- ing the process of chondrogenesis, including those factorsgenic differentiation of adult mammalian progenitor that regulate the progression of cells through the entirecells derived from bone marrow. q 1998 Academic Press chondrogenic lineage.

METHODSINTRODUCTION

Cells isolated from postnatal mammalian bone mar- Cell harvest and colony formation. Rabbit bone marrow was har-vested from either the iliac crests or tibias of 30 5-month-old Newrow have the potential for differentiation into the spe-Zealand White rabbits. The marrow was harvested from the proximalcialized cells of mesenchymal tissues such as bone andanterior tibial metaphysis or from the posterior superior iliac spinevia small skin incisions. An 18-gauge needle was used to penetrate

1 To whom correspondence and reprint requests should be ad- the cortex of the bone and 7–8 ml of marrow was aspirated into asyringe containing 3000 units of heparin. Dulbecco’s modified Eagle’sdressed. Fax: (216) 368-1332. E-mail: bxj9@po.cwru.edu.

265 0014-4827/98 $25.00Copyright q 1998 by Academic Press

All rights of reproduction in any form reserved.

AID ECR 3858 / 6i2d$$$201 12-23-97 19:26:41 ecal

266 JOHNSTONE ET AL.

medium (DMEM) with 10% fetal bovine serum (FBS) was added to for EF-1a [17] to control for differences in RNA loading. Blots wererinsed and then washed three times at 247C with 500 ml of 0.1Xthe aspirate and the number of nucleated cells was determined. The

cells were plated out at 20 1 106/100-mm dish and grown for 14 days SSPE, 0.1% SDS, and twice at 547C (a1(II) probe) or 657C (a2(I)probe) with 500 ml of 0.1X SSPE, 0.1% SDS, and exposed to X-rayat 377C, 5% CO2 with medium changes every 4 days.film. The a2(I) probe was a cloned 702-bp rabbit-specific RT-PCRAggregate culture. On day 14, adherent colonies of cells were tryp-product generated from rabbit cartilage RNA in our laboratory. Up-sinized, counted, and 2 1 105 cell aliquots were spun down at 500gper and lower primers, specific for sequence in exons 49 and 52 ofin 15-ml polypropylene conical tubes. The FBS containing mediumthe human HUMC1A2 gene [18] were: 5*-GGT GGT TAT GAC TTTwas then replaced with a defined medium, consisting of DMEM withGGT TAC-3 * and 5*-CAG GCG TGA TGG CTT ATT TGT-3 *, respec-ITS/ Premix (Collaborative Biomedical Products: insulin (6.25 mg/tively. The a1(II) probe was a 3.8-kb genomic fragment containingml), transferrin (6.25 mg/ml), selenous acid (6.25 mg/ml), and linoleicexons 45–54 of the human COL2A1 gene [19, 20] previously shownacid (5.35 mg/ml), with bovine serum albumin (1.25 mg/ml)). Pyruvateto hybridize to rabbit cartilage type II collagen RNA on a Northern(1 mM) and ascorbate 2-phosphate (37.5 mg/ml) were also added. Ag-blot (Hering, unpublished result).gregates were cultured with or without dexamethasone (1007 M), TGF-

cDNA synthesis and PCR. RNA from cultured mesenchymal pro-b1 (0.5 to 10 ng/ml, recombinant human, R&D Systems), or a combina-genitor cells and aggregate cultures (350 ng each) and rabbit articulartion of these agents. For some experiments, the 10% FBS containingcartilage (5 mg) was used for oligo d(T)-primed cDNA synthesis usingmedium was not replaced. The pelleted cells were incubated at 377C,MoMLV-H reverse transcriptase [21]. cDNA synthesized from equiv-5% CO2. Within 24 h of incubation, the cells formed an essentiallyalent amounts (50–100 ng) of preaggregate mesenchymal progenitorspherical aggregate that did not adhere to the walls of the tube. Me-cell or aggregate culture RNA, or 125 ng of cartilage RNA, was useddium changes were carried out at 2- to 3-day intervals and aggregatesas template for PCR amplification per 25-ml reaction volume usingwere harvested at time points up to 21 days.primer pairs designed using sequence obtained from the GenBankAlkaline phosphatase activity. Medium was removed from thedatabase: human type II collagen a1(II) chain [22, 23], human typeaggregate cultures and they were rinsed in Tyrodes solution priorI collagen a2(I) chain [24], and human type X collagen a1(X) [25].to incubation with 200 ml of 5 mM p-nitrophenylphosphate in 50 mMPrimer sets were as follows: collagen a1(II): 5*-CTG CTC GTC GCCTris, 150 mM NaCl, pH 9.0, for 30 min at room temperature. TheGCT GTC CTT-3 * and 5*-AAG GGT CCC AGG TTC TCC ATC-3 *,resulting colorimetric reaction was quantified by determining thecollagen a2(I); 5*-GGT GGT TAT GAC TTT GGT TAC-3 * and 5*-CAGabsorbance of the substrate solution at 405 nm.GCG TGA TGG CTT ATT TGT-3 *, collagen a1(X); 5 *-GCC CAA GAGHistology and immunohistochemistry. For histological and im-GTG CCC CTG GAA TAC-3* and 5 *-CCT GAG AAA GAG GAG TGGmunohistochemical analyses, the aggregates were frozen in OCT andACA TAC-3 *. Calculated optimal annealing temperatures (OLIGO5-mm sections were cut. For histological evaluation, sections werePrimer Analysis Software, National Biosciences Inc., Plymouth, MN)stained with toluidine blue. For immunohistochemistry, the frozenwere used for each primer pair, and samples were withdrawn forsections were fixed briefly for 10 min in methanol after a brief immer-analysis in agarose gels following 30 and 40 cycles of amplification.sion in distilled water to remove the OCT. Blocking of nonspecificExpected product sizes were as follows: collagen a1(II), 432 bp (IIAantibody binding sites was done by incubating the slides in 5% bovineform) and 225 bp (IIB form); collagen a2(I), 702 bp; collagen a1(X),serum albumin (BSA) in phosphate-buffered saline (PBS) for 1 h.703 bp. PCR products were analyzed by electrophoresis on a 1%The sections were then incubated with primary antibody for 30 min,agarose gel containing ethidium bromide [21]. Total PCR reactiondiluted in 0.5% BSA in PBS. Two antibodies with epitopes in typeproducts were ligated into a TA cloning vector (pCRII, Invitrogen)II collagen were used: a polyclonal antibody (affinity purified anti-and transformed into Escherichia coli. A number of plasmids repre-type II, Rockland, Inc.) and a monoclonal antibody (C4F6, kindlysenting different cloned PCR products were purified and inserts se-provided by Clinton Chichester, URI). In addition, an antibody toquenced by the DNA Sequencing Core Facility of the Northeasterntype X collagen was used (kindly provided by Gary Gibson, HenryOhio Multipurpose Arthritis Center using a Pharmacia BiotechFord Hospital, Detroit, MI). To facilitate antibody access to the colla-ALFexpress Automated DNA Sequencer. Rabbit-specific PCR prod-gens, the sections were predigested with chondroitinase ABC (0.1 U/ucts representing rabbit collagen a1(II) alternatively spliced formsml in 0.1 M Tris-acetate, Seikagaku). Reactivity was detected withusing these primers are 435 bp (IIA form) and 225 bp (IIB form).fluorescence microscopy after incubation for 30 min with an FITC-This partial rabbit IIA sequence has been deposited to the GenBanklinked secondary antibody (either anti-rabbit Ig or anti-mouse Ig,data base under Accession No. AF027122.Cappel) diluted in 0.5% BSA in PBS.

Southern blot analysis. Southern blot analysis using labeled oli-RNA isolation. For two separate marrow cell preparations, RNAgonucleotide probes internal to amplifying primers was performedwas prepared from the preaggregate mesenchymal progenitor cellas per standard protocols [21] to determine the relative abundance ofcultures, and subsequent aggregate cultures from the same prepara-the collagen IIA and IIB forms in the RT-PCR amplification reaction.tions, with a modification of the method of Chomczynski and SacciFollowing electrophoresis, samples were transferred from agarose[13]. Cells were lysed in culture dishes (1 ml/10 cm2) with TRIZOLgels to nitrocellulose membranes and were sequentially probed usingreagent (Life Technologies Inc., Grand Island, NY). For each time-5*-end 32P-labeled oligonucleotides. We designed an oligonucleotidepoint chosen, 20 pellets were pooled and homogenized using a Dounceprobe (probe AB) spanning exons 5 and 6 that would hybridize to bothhomogenizer in TRIZOL reagent, and RNA was prepared as per kitthe IIA and IIB splice variants and a probe that would specificallyinstructions. RNA was isolated directly from 5-week-old rabbit artic-hybridize to the IIA splice variant (probe A), recognizing exon 2. Theular cartilage using the method of Nemeth et al. [14]. RNA wassequences of these oligonucleotides were as follows: probe AB, 5*-quantified by comparing ethidium bromide fluorescence with a stan-TTC ACC TGC AGG TCC CTG AGG-3*; probe A, 5*-ACA CAG ATCdard series of RNA dilutions [15].CGG CAG GGC TCC-3 *. Following hybridization and washing, mem-

Northern hybridization. RNA samples (15 mg) were electropho- branes were exposed to Kodak BioMax film for varying periods ofresed through 1% agarose gels containing formaldehyde [15] and time at 0707C with an intensifying screen.were transferred overnight to a nitrocellulose filter [16]. Followingtransfer, filters were dried and baked over 2 h at 807C. Filters were

RESULTSprehybridized at 427C in a solution containing 50% formamide, 51SSPE, 21 Denhardt’s reagent, 100 mg/ml salmon sperm DNA, and

Aggregate Formation0.1% SDS [15] for several hours. Northern blots were sequentiallyhybridized at 427C in the same solution with 32P-labeled cDNA probes

During primary culture, adherent colonies of cellsfor the a2(I) chain of type I collagen, and the a1(II) chain of type IIcollagen prepared by random primer labeling, and with a cDNA probe formed. Differences in cell morphology were observed

AID ECR 3858 / 6i2d$$$202 12-23-97 19:26:41 ecal

267IN VITRO CHONDROGENESIS

between colonies: most cells were fibroblast-like, but the aggregates (data not shown). At 1 ng/ml TGF-b1,chondrogenic differentiation of the central aggregatesome were more flattened. After 14 days in culture the

cells were trypsinized to release them from the substra- region had not occurred by 21 days of culture, althoughdifferentiation was observed in the outer third of thetum and used for the aggregate culture studies, after

centrifugation into pellets as described above. In the aggregate; this was even more restricted in the aggre-gates incubated with 0.5 ng/ml. In order to better defineserum-free defined medium, condensation of the cells

into a single aggregate was seen within 16 h of incuba- the extent of the chondrogenic differentiation of themarrow-derived cells, we also immunostained repre-tion after centrifugation. However, cells incubated in

DMEM with 10% FBS did not form a clearly identifi- sentative sections of aggregates with an anti-type Xantibody. Type X collagen was detectable throughoutable aggregate at any time after centrifugation. There-

fore, only the aggregates from the defined medium in- the pellet by day 21 (Fig. 2).cubations, with or without additives (dexamethasone,

Alkaline Phosphatase ActivityTGF-b1), were available for analysis.

This was measured for the six preparations to whichEffect of Dexamethasone on the Cell AggregatesTGF-b1 was added, with or without dexamethasone.The alkaline phosphatase activity in the dexametha-Aggregate cultures were set up for marrow cell cul-

tures from 24 rabbits. For each rabbit cell preparation, sone-treated aggregates was low at day 7 and did notincrease over the time in culture (Fig. 3). This corre-aggregates were incubated with or without dexametha-

sone in the defined medium. Duplicate aggregates were lates with the histological and immunohistochemicalfindings that dexamethasone alone was not sufficientharvested at 7, 14, and 21 days after centrifugation

and frozen in OCT prior to analysis. In 6 of the 24 to induce chondrogenesis in the preparations used forthis assay. In contrast, the inclusion of TGF-b1 in themarrow cell preparations, aggregates incubated with

dexamethasone had metachromatic staining with tolu- aggregate cultures caused an increase in the measuredalkaline phosphatase activity. This increase correlatedidine blue that was characteristic of cartilage matrix

(Fig. 1). Within the metachromatic staining matrix with the chondrogenic differentiation of the cells.there were cells in lacunae with the appearance of hy-

Extracellular Matrix Gene Expression duringpertrophic chondrocytes. This was seen as early as dayAggregate Chondrogenesis7, and clearly identifiable by day 14. By day 21, the

morphology of some aggregates was entirely cartilagi- Northern hybridization of the total cellular RNAnous. This change in morphology and staining pattern from preaggregate cells was done with matrix moleculewas not observed in any aggregate incubated without probes for types I and II collagen (Fig. 4). The probedexamethasone. for collagen a1(I) hybridized to bands of the expectedThe metachromatic staining pattern of aggregated size (approximately 5 kb) in the preaggregate cell RNA,cells incubated in the presence of dexamethasone sug- as well as in the rabbit fibroblast control RNA. No hy-gested that a cartilaginous matrix had been synthe- bridization to a1(II) mRNA (which would be approxi-sized. To confirm this, immunohistochemistry was car- mately 5 kb) was detectable upon subsequent rehybrid-ried out with an antibody to type II collagen. Positive ization with the collagen a1(II) probe. RT-PCR analysisimmunostaining was observed only in regions of aggre- of aggregate cultures was performed using primersgates that had metachromatic staining, which by day spanning collagen a1(II) exons 1–7 (Fig. 5). Only the21 comprised the whole aggregate. Aggregates incu- collagen IIB (225 bp) splice variant was detectible fol-bated without dexamethasone had no detectable stain- lowing 30 cycles of amplification using RNA purifieding for type II collagen. from rabbit articular cartilage as a template. Amplifi-cation using RNA from 7-day aggregate cultures re-Effects of TGF-b1 on Aggregate Chondrogenesisvealed PCR products representing both the IIB and theIIA (432 bp) splice variants. The identity of these PCRSix rabbit cell preparations were used for testing the

effects of TGF-b1 on chondrogenesis. Aggregates were products was confirmed by sequence analysis. South-ern blot analysis using labeled oligonucleotide probesincubated with TGF-b1 (10 ng/ml), dexamethasone, or

both. In this series of aggregates, no chondrogenesis was also performed on the same RT-PCR amplificationproducts. Two bands were visualized when the blot waswas observed in those incubated with dexamethasone

alone (Fig. 2). However, TGF-b1, either alone or in com- probed with an oligonucleotide designed to hybridizewith exons 5 and 6 (Fig. 5B). For confirmation of speci-bination with dexamethasone, induced chondrogenesis

in the aggregated cells. The aggregates formed in the ficity, a probe designed to hybridize with exon 2, thealternatively spliced exon included in the IIA form, waspresence of both dexamethasone and TGF-b1 appeared

larger than those incubated with TGF-b1 alone. In sep- used and only the slower migrating band was detected.Further PCR experiments were done to assess thearate experiments, it was determined that lowering the

TGF-b1 concentration decreased the chondrogenesis in changes in the expression of types I and X collagen.

AID ECR 3858 / 6i2d$$$203 12-23-97 19:26:41 ecal

268 JOHNSTONE ET AL.

FIG. 1. Rabbit bone marrow-derived cells cultured as aggregates. (A, B, C, D) Immunostaining with type II collagen antibody. (E, F)Toluidine blue staining. (A) 7, (B) 14, and (C, E) 21 days with 1007 M dexamethasone; (D, F) 21 days without dexamethasone.

Samples were withdrawn following 40 cycles of amplifi- to cells from the synovium or subchondral bone beingincluded in the harvested tissue. A collagen a1(X) PCRcation. Collagen type I a2(I) chain mRNA appeared

more abundant in preaggregate cell RNA than in day 7 product was also amplified from cartilage RNA, andfrom day 7 aggregate cells, but no product was detectibleaggregate cells (Fig. 6A). This result correlates with high

steady state levels of a1(I) chain mRNA detected by in preaggregate cells (Fig. 6B). The detection of type Xin the articular cartilage sample was probably due toNorthern blot analysis of similarly cultured cells (Fig.

4). It was also detected in a rabbit articular cartilage the inclusion of hypertrophic cells in the harvest sincethe tissue was taken from 5-week-old rabbits that stillsample included in the analysis. The presence of type I

mRNA in the articular cartilage sample could be due possess an articular–epiphyseal cartilage complex [26].

AID ECR 3858 / 6i2d$$$203 12-23-97 19:26:41 ecal

269IN VITRO CHONDROGENESIS

FIG. 2. Rabbit bone marrow-derived cells cultured as aggregates in defined medium for 21 days: (A, C) with 1007 M dexamethasone;(B, D, E) with 1007 M dexamethasone and 10 ng/ml TGF-b1. (A, B) Toluidine blue staining; (C, D) immunostaining with anti-type IIantibody; (E) immunostaining with anti-type X collagen antibody.

DISCUSSION bone marrow can be reimplanted in vivo and undergoosteochondral differentiation [1–4]. Furthermore, invitro genesis of osseous tissue, in the form of bone nod-Mesenchymal progenitor cells with chondrogenic po-ules, is a well-recognized assay for the osteogenic poten-tential are present in many tissues of the body. Thosetial of these cells [28]. However, the formation of carti-of the bone marrow are of particular interest because oflage in vitro from postnatal mammalian bone marrowtheir ease of harvest and the potential for the use ofhas not been well demonstrated. Successful in vitrothese cells to facilitate cartilage repair [27]. There are

numerous studies demonstrating that cells isolated from chondrogenesis has been demonstrated with avian and

AID ECR 3858 / 6i2d$$$203 12-23-97 19:26:41 ecal

270 JOHNSTONE ET AL.

Dexamethasone is not a specific chondrogenic differ-entiation factor, as demonstrated by its ability to in-duce multiple end-phenotypes when added to culturedfetal rat calvarial cells with differentiation potential[32, 34]. In fact, impairment of chondrogenesis in mu-rine neonatal condylar cartilage has been observed inthe presence of dexamethasone [35], although it in-duced chondrogenesis in organoid cultures of murineembryonic cells [33]. The addition of dexamethasonefacilitated chondrogenic differentiation in only 25% ofthe rabbit marrow cell aggregate preparations. Thereasons for this are unclear, but it may be related tothe number of cells with chondrogenic capacity thatare within a given marrow aspirate and subsequentlyadhere to culture plates. The results of several studiesindicate that there appears to be a minimum numberFIG. 3. Alkaline phosphatase activity of the pellets as assessedof cells required before chondrogenesis can occur [36–by measuring the absorbance value for the postincubation solution43]. Such a variation between aggregates is possibleat 405 nm. (Dex) dexamethasone.since we are not using clonally isolated populations ofcells, and there is cellular heterogeneity in the marrow-derived monolayer cultures used for these experiments.However, how can we then explain the effect of addingembryonic mammalian cells and cell lines [6, 29–33],

but there has only been one report of induction of chon- TGF-b1, which induced chondrogenesis in all prepara-tions? Perhaps, this cytokine initially increases the pro-drogenesis from postnatal mammalian cells [7]. In the

present study, we describe the chondrogenic differentia- liferation of cells with chondrogenic potential to a pointwhere the critical number is reached and the differenti-tion of mesenchymal progenitor cells from postnatal

mammalian bone marrow. The presence of a metachro- ation proceeds. Alternatively, TGF-b1 may provide thecellular stimulus for differentiation, regardless of thematic-staining matrix, the chondrocytic appearance of

the cells, and the detection of type II collagen mRNAand protein signify that the tissue generated by thesemarrow-derived cells is cartilage. Furthermore, the cellsdifferentiate into their terminal phenotype, the hyper-trophic chondrocyte, as indicated by the detection of typeX collagen mRNA and protein and the concomitant risein alkaline phosphatase activity.

The induction of the chondrogenesis of these cellsrequired particular culture conditions. The cells weremaintained in a format resembling that of a precarti-lage condensation [12]. In a recent paper, Noble et al.(7) described experiments where the addition of dex-tran sulfate to porcine bone marrow cells grown to con-fluence on tissue culture plates caused retraction of thecells into nodular structures in which type II collagenwas immunolocalized after 6 days. Thus, as found inour studies, chondrogenesis was induced after the cellsformed precartilage condensation-like structures. Ofinterest is the fact that Noble et al. could achieve initia-tion of chondrogenesis in serum-containing medium,whereas induction of chondrogenesis does not occur inrabbit bone marrow-derived cell aggregates incubatedin medium containing fetal bovine serum. Likewise, ifthe serum is removed and the aggregates are incubatedin a defined medium without dexamethasone or TGF-

FIG. 4. Northern hybridization of preaggregate rabbit bone-mar-b1, no chondrogenesis occurs. This implies that therow-derived cell RNA with matrix molecule probes. Total cellularaddition of dextran sulfate may have effects on the RNA from the preaggregated cells (lane 1) and from cultured rabbit

porcine marrow cells other than simply creating a con- dermal fibroblasts (lane 2) hybridized with probes for (A) collagena1(I), (B) collagen a1(II).densation-like structure.

AID ECR 3858 / 6i2d$$$203 12-23-97 19:26:41 ecal

271IN VITRO CHONDROGENESIS

shown to inhibit growth plate chondrocyte hypertrophy[50]. In aggregate cultures of growth plate chondro-cytes, TGF-b1 stimulated cartilage-specific proteogly-can production, but when added at later stages, inhib-ited the appearance of type X collagen [10, 11]. How-ever, in our system, the cells progress through thechondrogenic lineage to hypertrophy, with type X colla-gen detected as early as 7 days. This occurred in thepresence of dexamethasone and TGF-b1. The con-

FIG. 5. (A) RT-PCR analysis to determine collagen type IIA and trasting effects of these factors on mesenchymal pro-IIB splice variants in rabbit cartilage (lane 1) and day 7 pellet culture genitor cell aggregate cultures compared with chick(lane 2) RNA. Lane 3 is a 1-kb ladder. (B) Southern blot analysis of

embryonic and rat growth plate cells may be due tothe RT-PCR amplification reaction products produced from day 7species or cell-type differences or to the dissimilar cul-aggregate culture extracted RNA, using either an oligonucleotide

probe that hybridizes to both collagen type IIA and IIB (lane 1) or ture conditions used.a probe specific for IIA (lane 2). When chondrogenesis was achieved, the morphology

of the aggregate changed from the appearance of a mes-enchymal cell condensation to that of cellular cartilage,such as is seen in embryonic limb formation. Further-number of cells present with chondrogenic capacity. In

the embryo, epithelial–mesenchymal interactions are more, at day 7 postaggregation, the presence of typeIIA collagen mRNA was detected by RT-PCR. Type IIAcrucial for differentiation. In the culture system, the

addition of the TGF-b1 provides the appropriate exter- collagen is the splice variant of type II collagen thathas been found in prechondrocytes and immature chon-nal signal, that when coupled with the endogenous fac-

tors communicated between cells within the condensa- drocytes [51, 52]. This form has an extra exon (exon 2,coding for 69 amino acids) spliced into it. Type II colla-tion, facilitates chondrogenic differentiation. In those

preparations where dexamethasone addition was suf- gen without this exon (designated IIB) is the form asso-ciated with maturing chondrocytes and those found inficient to induce differentiation, endogenous TGF-b1 or

other inductive cytokines may have been present in postnatal cartilage. These observations lead to the sug-gestion that the aggregate culture system is a modelhigh enough concentrations that the dexamethasone-

induced changes were enough to facilitate chondrogen- of embryonic chondrogenic differentiation and that theprocess is a recapitulation of embryonic events. Thisesis. TGF-b1, -b2, and -b3 have been detected in the

developing skeleton including the areas where the hypothesis is currently being explored.primitive mesenchymal tissue are undergoing conden-sation and cartilage formation [44–46]. All rabbit mar-row-derived pluripotential cell aggregates treated withTGF-b1 progressed on to form histologically identifi-able cartilage, thus supporting the idea that TGF-b1plays an important role in chondrogenesis.

The appearance of the type X collagen in the aggre-gates as shown by immunohistochemistry as well asPCR demonstrate that these cells will terminally differ-entiate into hypertrophic chondrocytes. The appear-ance of type X collagen is a rapid phenomenon oc-curring soon after the appearance of the type II colla-gen. This rapid appearance of type X collagen soon afterthe chondrocytic differentiation of mesenchymal cellshas been shown in vitro in avian chondrogenesis stud-ies [47]. When the aggregated cells underwent chondro-genesis, a concomitant elevation in the alkaline phos-phatase level was detected. This rise in the alkalinephosphatase activity is consistent with the differentia-tion of these cells into hypertrophic chondrocytes [48].However, this result contrasts those of other in vitrostudies of the effects of dexamethasone and TGF-b1

FIG. 6. RT-PCR analysis of rabbit articular cartilage (lane 1),on terminal differentiation. Although not necessary forpreaggregate mesenchymal cells (lane 2) and day 7 aggregate cellschick embryonic cell chondrogenesis in vitro, dexa- (lane 3) to determine the expression of mRNA for (A) collagen type

methasone supports cell viability but delays the ap- I and (B) collagen type X. The 123-bp ladders are also shown. Arrow-heads indicate the positions of the expected products.pearance of type X collagen [49]. TGF-b1 has also been

AID ECR 3858 / 6i2d$$$203 12-23-97 19:26:41 ecal

272 JOHNSTONE ET AL.

We thank John Kollar and Amad Awadallah for expert technical T., Solomon, E., Grant, M. E., and Boot-Handford, R. P. (1991)Biochem. J. 280, 617–623.assistance and R. Tracy Ballock, M.D., for advice on the culture

system. This study was supported in part by N.I.H. Grants AR-44390 26. Van Sickle, D. C., and Kincaid, S. A. (1983) Proc. Am. Assoc.(B.J.) and AR-37726 (V.G.) and AR-20618 (Northeastern Ohio Multi- Vet. Anat. 120, 14.purpose Arthritis Center). 27. Wakitani, S., Goto, T., Pineda, S. J., Young, R. G., Mansour,

J. M., Goldberg, V. M., and Caplan, A. I. (1994) J. Bone Jt. Surg.76, 579–592.REFERENCES

28. Beresford, J. N., Graves, S. E., and Smoothy, C. A. (1993) Am.1. Friedenstein, A. J. (1976) Int. Rev. Cytol. 47, 327–355. J. Med. Gen. 45, 163–178.2. Ashton, B. A., Allen, T. D., Howlett, C. R., Eaglesom, C. C., Hat- 29. Caplan, A. I., and Stoolmiller, A. C. (1973) Proc. Natl. Acad.

tori, A., and Owen, M. (1980) Clin. Orthop. 151, 294–307. Sci. USA 70, 1713–1717.3. Ashhurst, D. E., Ashton, B. A., and Owen, M. E. (1990) J. Or- 30. Ahrens, P. B., Solursh, M., and Reiters, R. (1977) Dev. Biol. 60,

thop. Res. 8, 741–749. 69–82.4. Goshima, J., Goldberg, V. M., and Caplan, A. I. (1991) Clin. 31. Asahina, I., Sampath, T. K., Nishimura, I., and Hauschka, P. V.

Orthop. Rel. Res. 262, 298–311. (1993) J. Cell Biol. 123, 921–933.5. Maniatopoulos, C., Sodek, J., and Melcher, A. H. (1988) Cell 32. Grigoriadis, A. E., Heersche, J. N. M., and Aubin, J. E. (1990)

Tiss. Res. 254, 317–330. Dev. Biol. 142, 313–318.6. Denker, A. E., Nicoll, S. B., and Tuan, R. S. (1995) Differentia- 33. Zimmermann, B., and Cristea, R. (1993) Anat. Embryol. 187,

tion 59, 25–34. 67–73.7. Noble, B. S., Dean, V., Loveridge, N., and Thomson, B. M. (1995) 34. Shalhoub, V., Conlon, D., Tassinari, M., Quinn, C., Partridge,

Bone 17, 375–382. N., Stein, G. S., and Lian, J. B. (1992) J. Cell. Biochem. 50,425–440.8. Holtzer, H., Abbott, J., Lash, J., and Holtzer, S. (1960) Proc.

Natl. Acad. Sci. USA 46, 1533–1542. 35. Silbermann, M., von der Mark, K., Maor, G., and van Menxel,M. (1987) Bone Mineral 2, 87–106.9. Manning, W. K., and Bonner, W. M., Jr. (1967) Arth. Rheum.

10, 235–239. 36. Steinberg, M. S. (1963) Science 141, 401–408.10. Kato, Y., Iwamoto, M., Koike, T., Suzuki, F., and Takano, Y. 37. Umansky, R. (1966) Dev. Biol. 13, 31–56.

(1988) Proc. Natl. Acad. Sci. USA 85, 9552–9556. 38. Caplan, A. I. (1970) Exp. Cell Res. 62, 341–355.11. Ballock, R. T., and Reddi, A. H. (1994) J. Cell Biol. 126, 1311– 39. Ede, D. A., and Flint, O. P. (1972) J. Embryol. Exp. Morphol.

1318. 27, 245–260.12. Fell, H. B. (1925) J. Morphol. Physiol. 40, 417–459. 40. Flickinger, R. A. (1974) Dev. Biol. 41, 202–208.13. Chomczynski, P., and Sacchi, N. (1987) Anal. Biochem. 162, 41. Hall, B. K. (1978) ‘‘Developmental and Cellular Skeletal Biol-

156–159. ogy,’’ Academic Press, New York.14. Nemeth, G. G., Heydemann, A., and Bolander, M. E. (1989) 42. Cottrill, C. P., Archer, C. W., and Wolpert, L. (1987) Dev. Biol.

Anal. Biochem. 183, 301–394. 122, 503–515.15. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) ‘‘Molecular 43. Solursh, M. (1990) Semin. Dev. Biol. 1, 45–53.

Cloning,’’ Cold Spring Harbor Laboratory Press, Cold Spring44. Heine, U. I., Munoz, E. F., Flanders, K. C., Ellingsworth, L. R.,Harbor.

Lam, H. Y. P., Thompson, N. L., Roberts, A. B., and Sporn, M. B.16. Thomas, P. S. (1980) Proc. Natl. Acad. Sci. USA 77, 5201–5205. (1987) J. Cell. Biol. 105, 2861–2876.17. Hering, T. M., Kollar, J., Huynh, T. D., Varelas, J. B., and San- 45. Pelton, R. W., Dickinson, M. E., Moses, H. L., and Hogan,

dell, L. J. (1994) Arch. Biochem. Biophys. 314, 90–98. B. L. M. (1990) Development 110, 609–620.18. de Wet, W. J., Bernard, M., Bensen-Chanda, V., Chu, M.-L., 46. Gatherer, D., TenDijke, P., Baird, D. T., and Akhurst, R. J.

Dickson, L., Weil, D., and Ramirez, F. (1987) J. Biol. Chem. (1990) Development 110, 445–460.262, 16032–16036.

47. Adams, S. L., Pallante, K. M., Niu, Z., Leboy, P. S., Golden,19. Sandell, L. J. (1984) J. Cell Biol. 99, 405a. E. B., and Pacifici, M. (1991) Exp. Cell Res. 193, 190–197.20. Goldring, M. B., Fukio, K., Birkhead, J. R., Dudek, E., and San- 48. Wuthier, R. E., and Register, T. C. (1995) in ‘‘The Chemistry

dell, L. J. (1994) J. Cell. Biochem. 54, 85–99. and Biology of Mineralized Tissue’’ (W. T. Butler, Ed.), pp. 113–21. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seid- 124, Ebsco, Birmingham.

man, J. G., Smith, J. A., and Struhl, D. (1991) ‘‘Current Proto- 49. Quarto, R., Campanile, G., Cancedda, R., and Dozin, B. (1992)cols in Molecular Biology,’’ Wiley, New York. J. Cell. Biol. 119, 989–995.

22. Baldwin, C. T., Reginato, A. M., Smith, C., Jimenez, S. A., and 50. Ballock, R. T., Heydemann, A., Wakefield, L. M., Flanders,Prockop, D. J. (1989) Biochem. J. 262, 521–528. K. C., Roberts, A. B., and Sporn, M. B. (1993) Dev. Biol. 158,

23. Ryan, M. C., Sieraski, M., and Sandell, L. J. (1990) Genomics 414–429.8, 41–48. 51. Ryan, M. C., and Sandell, L. J. (1990) J. Biol. Chem. 265,

24. de Wet, W. J., Chu, M.-L., and Prockop, D. J. (1983) J. Biol. 10334–10339.Chem. 258, 14385–14389. 52. Sandell, L. J., Morris, N., Robbins, J. R., and Goldring, M. B.

(1991) J. Cell Biol. 114, 1307–1309.25. Thomas, J. T., Cresswell, C. J., Rash, B., Nicolai, H., Jones,

Received April 24, 1997Revised version received October 3, 1997

AID ECR 3858 / 6i2d$$$204 12-23-97 19:26:41 ecal