The Effects of ERK1=2 Inhibitor on the Chondrogenesisof Bone Marrow– and Adipose Tissue–Derived

Multipotent Mesenchymal Stromal Cells

Hye-Joung Kim, M.S., and Gun-Il Im, M.D.

This study tested the hypothesis that mitogen-activated protein kinase inhibitors suppress hypertrophy andenhance chondrogenesis during chondrogenesis of multipotent mesenchymal stromal cells (MSCs). The effects ofPD98059 (an extracellular signal–regulated kinase-1=2 inhibitor) and SB203580 (a p38 inhibitor) were tested onbone marrow–derived MSCs (BMMSCs) and adipose tissue–derived MSCs (ATMSCs). In vitro pellet cultures werecarried out using 2.5�105 MSCs in a chondrogenic medium containing 5 ng=mL of transforming growth factor-b2

(TGF-b2) for BMMSCs, and 5 ng=mL of TGF-b2 and 100 ng=mL of bone morphogenetic protein-7 (BMP-7) forATMSCs. From the 14th day of culture, the pellets were additionally treated with PD98059 or SB203580. After 14more days of in vitro culture, pellets were harvested for analysis. PD98059 increased DNA content and glycos-aminoglycan amount in BMMSCs and ATMSCs, whereas SB203580 had little effect. Collagen type I (COL1A1)mRNA decreased to almost a quarter in BMMSCs treated with PD98059. The mRNA levels of collagen type II(COL2A1) and SRY (sex determining region Y)-box 9 (SOX-9) increased several fold in both cells after PD98059treatment, whereas SB203580 had only a slight effect. The gene expression of collagen type X (COL10A1) and runt-related transcription factor 2 (Runx-2) decreased by half after PD98059 treatment in BMMSCs, and decreasedfurther in ATMSCs. SB203580 elevated COL10A1 and Runx-2 gene expression in both cell types. Safranin-Ostaining and immunohistochemistry generally mirrored findings from real-time PCR except for diminished ex-pression of type I collagen in ATMSCS, and more pronounced decrease in type X collagen and Runx-2 in BMMSCsafter PD98059 treatment. Our study demonstrated that PD98059 suppressed hypertrophy and promoted chon-drogenesis of MSCs, and provides a ground for using them in cartilage tissue engineering.

Introduction

Multipotent mesenchymal stromal cells (MSCs)from adults are capable of self-renewal and differen-

tiation into several cell types of connective tissue,1,2 and havebeen investigated as potential alternatives to chondrocytesfor cartilage repair.3–5 Bone marrow provides the most uni-versal source of MSCs, but other tissues such as periosteum,muscle, synovial membrane, and adipose tissue also possessMSCs. Of these, adipose tissue offers unique advantages interms of accessibility and abundance.6

In cartilage tissue engineering from MSCs, hypertrophicchanges and inadequate differentiation pose formidablechallenges.7 Hypertrophic change is a tricky problem becauseit represents the usual fate of chondrocytes in the skeletalformation during developmental process. Most of naturalchondrogenesis in humans occurs predominantly with cellu-

lar hypertrophy, which is the traditional representation ofchondrocyte maturation.8 Nevertheless, it is still important tosuppress hypertrophic changes in the tissue engineering ofarticular cartilage, if not for chondrogenesis in the broadsense, to produce a neocartilage bearing the characteristics ofhyaline articular cartilage. Although MSCs may differentiateinto any lineage of musculoskeletal tissue, individual cellshave different propensity in becoming chondrocytes.9 Chon-drogenic growth factors promote chondrogenesis, but do notcompletely block the differentiation into other cell types.

Growth factors that have been used to induce chon-drogenesis of MSCs, including transforming growth factor-b(TGF-b) and bone morphogenetic protein (BMP), act up-stream of signal transduction pathways; however, moleculesthat affect intracellular events downstream of these path-ways may enhance chondrogenesis of MSCs, possibly in amore efficient way. While SMAD, a pathway whose name

Department of Orthopaedics, Dongguk University Ilsan Hospital, Goyang, Republic of Korea.

TISSUE ENGINEERING: Part AVolume 16, Number 3, 2010ª Mary Ann Liebert, Inc.DOI: 10.1089=ten.tea.2009.0070

851

comes from the combination of the Drosophila gene ‘‘mothersagainst decapentaplegic’’ (Mad) and the Caenorhabditis elegansgene Sma, is the canonical signaling pathway for TGF-band BMP. TGF-b and BMP also act via mitogen-activatedprotein kinase (MAPK) pathways.10 MAPKs play a key rolein a variety of cellular responses, including proliferation,differentiation, and apoptosis.10 The major MAPK subtypesinclude extracellular signal–regulated kinase-1 (ERK-1) andERK-2, C-Jun N-terminal kinase, and p38 MAPK.10 In-dividual MAPs are activated by distinct upstream kinasesthat respond to a wide variety of extracellular stimuli.11

MAPK subtypes regulate cellular functions either coopera-tively or antagonistically in many cases.

While p38 has been suggested to have an obligatory roleduring chondrogenesis, the role of ERK appears to be morecomplex and may depend on the intensity or duration of anupstream activating signal or on its localization withincells.11 In chondrogenic ATDC5 cells from mouse teratocar-cinoma, inhibition of the ERK1=2 or p38 MAPK pathwaysrepressed TGF-b-induced aggrecan gene expression.12 Con-versely, the inhibition of ERK1=2 enhanced chondrogenesisup to 1.7-fold, while the inhibition of p38 reduced chon-drogenesis by 30% in micromass cultures of chick embryomesenchyme.13 In hMSCs, ERK1=2 inhibition by PD98059increased sulfated glycosaminoglycan (GAG) levels two-fold.14 Conversely, Chang et al. recently reported that inhi-bition of p-ERK with PD98059 blocked the expression ofgrowth-arrest-specific protein 7 (Gas7), a protein required forthe chondrogenic differentiation of hMSCs.15

ERK1=2 and p38 kinases also have controversial roles interms of controlling hypertrophy. ERK1=2 and its upstreamkinase are required for the normal expression of type Xcollagen and p21 in chondrocytes,16,17 and the absence ofc-RAF (an MAP3K for ERK1=2) severely delays endochondralossification.18 On the other hand, p38 suppression results inthe accumulation of cartilage matrix protein, and inhibits theexpression of type X collagen in the chick embryo, whichsuggests that p38 signaling is required for the transition fromprehypertrophic to fully hypertrophic chondrocytes.19

Since many of the molecular events of embryogenesis arerecapitulated during chondrogenesis of MSCs in three-dimensional in vitro cultures, it is possible that modulationsof these pathways could be used to inhibit hypertrophy andenhance chondrogenesis. To date, it has not been thoroughlyinvestigated in MSCs isolated from human adults. In thisstudy we tested the hypothesis that MAPK inhibitors sup-press unwanted hypertrophic changes and enhance chon-drogenesis during chondrogenesis of hMSCs. Specifically,we examined the effects of PD98059 (an ERK1=2 inhibitor)20

and SB203580 (a p38 inhibitor)21 on chondrogenesis of bonemarrow–derived MSCs (BMMSCs) and adipose tissue–derived MSCs (ATMSCs).

Materials and Methods

Procurement of samples, cell isolation, and cultivation

The bone marrow samples used to isolate MSCs (i.e.,BMMSCs) were obtained from three donors (mean age, 50years; range, 37–64 years) who underwent total hip re-placement for osteoarthritis. ATMSCs were isolated from li-poaspirates generated during elective liposuction procedureson three nonsmoking=nondiabetic donors (mean age,

37 years; range, 30–42 years). Body mass index was 19.6� 1.7in BMMSC donors, and 24.0� 1.4 in ATMSC donors. Thisstudy was approved by the institutional review board of ouruniversity, and informed consent was obtained from all do-nors involved. BMMSCs and ATMSCs were isolated andexpanded as described previously.22–24 Briefly, bone marrowsamples (10 mL) were mixed with 0.3 mL of heparin to pre-vent coagulation, and then this mixture was diluted with20 mL of phosphate buffered saline (PBS). The cells werefractionated on a Lymphoprep� density gradient (Axis-Shield, Oslo, Norway) by centrifuging at 600 g for 10 min.The interface mononuclear cells were isolated, these cellswere washed with PBS, and then erythrocyte (RBC) lysisbuffer (0.154 M NH4Cl, 10 mM KHCO3, and 0.1 mM EDTA)was added to destroy the RBC contaminants. The cells werewashed twice by centrifugation (600 g) in PBS. One millioncells were cultured in F-12=DMEM medium (Gibco BRL,Green Island, NY) containing 10% fetal calf serum (GibcoBRL) and 1% antibiotics (penicillin 100 U=mL, streptomycin0.1 mg=mL, and amphotericin B 0.25 mg=mL; Gibco BRL) at378C in a humidified atmosphere with 5% CO2.

The ATMSCs were isolated from lipoaspirate. Briefly, thelipoaspirates were washed three times with PBS and thematrix was digested with 1.5 mg=mL collagenase, and thenfiltered through a 250-mm nylon mesh. The erythrocytes wereremoved using erythrocyte lysis buffer. The remaining cellswere placed in culture flasks and maintained under condi-tions identical to those used for the BMMSCs.

Induction of in vitro chondrogenesis

To induce chondrogenesis, in vitro pellet cultures wereprepared using 2.5�105 cells at passage 3 in DMEM=F-12(Gibco BRL) supplemented with 1% ITS (Gibco BRL), 10�7 Mdexamethasone, 50mM ascorbate-2-phosphate, 50mM L-proline, 1 mM sodium pyruvate, and respective growth fac-tors. Growth factor–starting concentrations were determinedbased on our previously published concentrations: 5 ng=mLof TGF-b2 (R&D Systems, Minneapolis, MN) for BMMSCs,and 5 ng=mL of TGF-b2 and 100 ng=mL of BMP-7 (R&D Sys-tems) for ATMSCs.23–25 For pellet cultures, 1 mL aliquots ofcell suspension (2.5�105 cells) at passage 3 were placed in15 mL polypropylene centrifuge tubes, and spun in a bench-top centrifuge at 500 g for 5 min. The tubes were then placed inan incubator at 378C in a humidified 95% air=5% CO2 atmo-sphere for up to 28 days. Tube caps were loosened to allow airexchange. The medium was changed every third day. Fromthe 14th day of culture, subsets of pellets were treated withPD98059 [0, 1, 10mM] (Sigma-Aldrich, St. Louis, MO) orSB203580 [0, 1, 10 mM] (Sigma-Aldrich), as referred fromprevious studies that investigated chondrogenic differentia-tion of MSCs.13,26 Cells were then cultured for an additional14 days in vitro, and pellets were harvested for analysis.

DNA quantification

Cell pellets were digested overnight in papain buffer at608C, and the DNA content was determined using the DNAbinding fluorochrome Hoechst 33258. Briefly, 10mL lysatesamples were transferred to the wells of a 96-well polysty-rene plate containing 90mL of TNE buffer (10 mM Tris, 2 MNaCl, at pH 7.4, and 1 mM EDTA). Hoechst 33258 (Sigma-Aldrich) solution (100 mL and 1 mg=mL) was then added to a

852 KIM AND IM

concentration of 20 mg=mL in TNE buffer. Plates were thenread using a Fluostar optima fluorescence plate reader (BMGLabtech, Offenburg, Germany) at an emission wavelength of460 nm and an excitation wavelength of 355 nm. The DNAcontent was determined using a standard curve drawn usingserial dilutions of calf thymus DNA.

Analysis of the GAG content

Pellets obtained as described above were digested in papainbuffer at 608C overnight. Samples were then transferred to1.5 mL microcentrifuge tubes. For 1,9-dimethylmethyleneblue assays, 50mL aliquots of each sample were added 50mLof an appropriate buffer. GAG levels were determined usingBlyscan kits (Biocolor, Carrickfergus, Northern Ireland) ac-cording to the manufacturer’s instruction. GAG levels areexpressed as micrograms of GAG per microgram of DNA.

Reverse transcription and real-time PCR

RNA was isolated using the standard guanidine iso-thiocyanate Tri-Reagent� (Sigma-Aldrich) protocol. IsolatedRNA samples (approximately 0.2 mg per each pellet) wereconverted to cDNA using reverse transcriptase (SuperScriptIII�; Invitrogen, Carlsbad, CA) and oligo (dT) primers. AllPCRs were performed on the LightCycler 480 System�

(Roche Diagnostics, Mannheim, Germany) in reaction vol-umes of 20 mL containing 5 mL cDNA, 0.5 mL of 10 mM senseand 0.5 mL of 10 mM antisense primer, 10mL LightCycler 480SYBR Green I Master mix (Roche Diagnostics), and 4mL ofRNAse-free water. The expression of the following geneswas examined: collagen type I (COL1A1), collagen type II(COL2A1), collagen type X (COL10A1), SRY (determiningregion Y)-box 9 (SOX-9), and runt-related transcriptionfactor-2 (Runx-2); glyceraldehyde-3-phosphate dehydroge-nase (GAPDH) was used as a housekeeping gene. Theprimers and reaction conditions used for amplification arelisted in Table 1. After polymerase activation (958C for10 min), amplification was performed over 45 cycles (10 sdenaturation at 958C, 10 s annealing at 658C, and 10 s ex-tension at 728C). Melting curve analysis was performed im-mediately after amplification using the following conditions:

5 s at 958C (hold time after reaching temperature), 1 min at658C, and 1 s at 978C. Temperature change rates were 208=s,except for the final step, for which a temperature change rateof 0.18=s was used. The peak melting temperatures obtainedwere considered to be those of the specific amplified prod-ucts. To guarantee the reliability of the results obtained, allsamples were processed in triplicate. Each assay was per-formed using positive and negative controls. The thresholdcycle (Ct) value of each gene was measured for each reversetranscript (RT) sample. The Ct value of GAPDH was used asan endogenous reference for normalization purposes (Userbulletin no. 2; Applied Biosystems–Roche Molecular System,Alameda, CA). The values thus obtained were normalizedversus the control, and are expressed as fold changes.

Histological analysis

After 28 days in culture, pellets were fixed in a 4%paraformaldehyde solution for 3 h, dehydrated with 100%ethanol, washed with xylene, and embedded in paraffin.Four-micrometer-thick sections were cut from the paraffinblocks and placed on glass slides. Safranin-O staining forproteoglycan and immunohistochemistry for collagen type I,type II, and type X and Runx-2 was then performed. ForSafranin-O staining, sections were deparaffinized with xy-lene and ethanol, treated with aqueous Safranin-O (0.1%) for30 min, and then washed with distilled water. For immu-nohistochemistry, Dakocytomation LSAB2 System HRPkits� (Dako, Hamburg, Germany) were used. Briefly, sec-tions were deparaffinized, treated with Pepsin Soluble� (FineLife Science, Seoul, Korea) for 15 min, washed in a 1�washbuffer (Dako), and incubated for 5 min with a peroxidase-blocking solution. Primary antibodies in goat serum(mouse anti-chicken collagen type II monoclonal antibody[200mg=mL; Chemicon, Temecula, CA] diluted at 1:100,rabbit anti-human collagen type I antibody [50mg=mL;AbD Serotec, Oxford, United Kingdom] diluted at 1:500,mouse monoclonal anti-human collagen type X antibody[1.0 mg=mL; Sigma-Aldrich] diluted at 1:1000, and mousemonoclonal anti-human Runx-2 antibody [1.0 mg=mL; Ab-cam, Cambridge, United Kingdom] diluted at 1:500) werereacted with sections overnight at 48C. After three washes in

Table 1. Primers Used for Real-Time Polymerase Chain Reaction

Gene Sequence Accession no. Amplicon (bp) Reference

Collagen type I 50-CCGCCGCTTCACCTACAGC-30 NM_000088 83 2850-TTTTGTATTCAATCACTGTCTTGCC-30

Collagen type II 50-CCGAATAGCAGGTTCACGTACA-30 NM_001844 79 2850-CGATAACAGTCTTGCCCCACTT-30

SOX-9 50-CACACAGCTCACTCGACCTTG-30 Z46629 76 2850-TTCGGTTATTTTTAGGATCATCTCG-30

Collagen type X 50-AAAGGCCCACTACCCAACAC-30 NM_000493 18250-CTTCCGTAGCCTGGTTTTCC-30

Runx-2 50-CCCAGCCACCTTTACCTACA-30 NM_001024630 12850-TATGGAGTGCTGCTGGTCTG-30

GAPDH 50-ATGGGGAAGGTGAAGGTCG-30 NM_002046 70 2850-TAAAAGCAGCCCTGGTGACC-30

Amplification was performed over 45 cycles (10 s denaturation at 958C, 10 s annealing at 658C, and 10 s extension at 728C) for all primers.GAPDH, glyceraldehyde-3-phosphate dehydrogenase; SOX-9, SRY (sex determining region Y)-box 9; Runx-2, runt-related transcriptionfactor-2.

ERK INHIBITOR PROMOTES CHONDROGENESIS OF MSCS 853

1� wash buffer, sections were incubated with horseradishperoxidase–labeled anti-mouse goat secondary antibodies(Dako) for 30 min, extensively washed, and reacted withsubstrate buffer and diaminobenzidine chromogen (Dako;50:1) for 10 min, and mounted. Normal human articularcartilage with subchondral bone from the femoral head wasused as a staining control. Three pellets were prepared fromcell isolates from bone marrow or lipoaspirates from eachdonor in this study. Sections were made from all pelletsgenerated in this study. Semiquantitative assessment of his-tological grading was performed based on the scale of in-tensity, which was precisely defined as very high (4), high(3), moderate (2), faint (1), or no reaction (0). The score wasobtained from each sample after the agreement of three mi-croscopists who were blinded to the samples.

Statistical analysis

Descriptive statistics was used to determine group meansand standard deviations. Statistical analysis was performedusing analysis of variance (ANOVA) followed by Bonfer-roni’s correction for multiple comparisons. Significance wasset at p< 0.05.

Results

Assessment of cell number by DNA content,and of proteoglycan synthesis by GAG analysis

PD98059 dose dependently increased the DNA content inBMMSCs ( p¼ 0.029) and ATMSCs ( p¼ 0.027) up to 2.5-fold(at 10 mM, a high dose), although basal ATMSC cell numberswere much lower. SB203580 also elevated the DNA contentin BMMSCs up to 1.6-fold. SB203580 did not show consistenteffect on ATMSCs (Fig. 1A).

PD98059 [10mM] doubled the GAG content compared to theuntreated controls [0mM] in BMMSCs ( p¼ 0.001) and increasedthem by 70% compared to the untreated controls [0mM] inATMSCs ( p¼ 0.019); however, no pronounced dose–responserelationship was observed. SB203580 did not significantlychange GAG levels in either BMMSCs or ATMSCs (Fig. 1B).

Gene expression of COL1A1, SOX-9, COL2A1,COL10A1, Runx-2 by quantitative reversetranscription (qRT)–PCR

The expression of COL1A1 dose dependently decreased byalmost 75% after PD98059 [10mM] treatment in BMMSCs

FIG. 1. DNA (A) and glycosaminogly-can (GAG) (B) amounts normalizedversus DNA in pellets after 28 days ofculture. Treatment with PD90509 orSB203580 started from the 14th day. Thebars represent means� SEM (n¼ 3:*p< 0.05, **p< 0.01, and ***p< 0.001).NS, not significant; SEM, standard errorof the mean; CM, MSCs treated withtheir respective chondrogenic medium;BMMSCs, bone marrow–derived mes-enchymal stromal cells; ATMSCs, adi-pose tissue–derived mesenchymalstromal cells.

854 KIM AND IM

( p¼ 0.007); however, it remained unchanged in ATMSCs.SB203580 affected COL1A1 expression in an irregular man-ner, but elevated it at 1 mM in BMMSCs and at 10mM inATMSCs (Fig. 2A).

The mRNA levels of COL2A1 were increased by three- andsixfold by 1 and 10mM of PD98059 in BMMSCs ( p¼ 0.004and p¼ 0.0001, respectively), and by three- to fourfold byPD98059 in ATMSCs ( p¼ 0.009 at 1 mM and p¼ 0.002 at10 mM). SB203580 induced COL2A1 gene expression dosedependently in both BMMSCs and ATMSCs, but at lowerlevels than PD98059 (Fig. 2B).

The mRNA level of SOX-9, the master gene of chon-drogenesis,27 increased more than threefold after PD98059treatment [1mM] ( p¼ 0.006) in BMMSCs, but this was notincreased further by a higher concentration ( p¼ 0.002 at10 mM). Dose dependence was observed with ATMSCs ( p¼0.023 at 10mM). SB203580 increased SOX-9 mRNA levels 1.4-fold [10mM] in BMMSCs and 1.9-fold [10mM] in ATMSCs.These increases were not statistically significant (Fig. 2C).

COL10A1, a marker of hypertrophic chondrocytes,28

decreased by a half by PD98059 [10 mM] in BMMSCs al-though the difference was not statistically significant, and in

ATMSCs, the decrease was less consistent. On the otherhand, SB203580 elevated COL10A1 mRNA levels in both celltypes, but by more than threefold in BMMSCs (Fig. 2D).

The mRNA levels of Runx-2, the master transcriptionfactor for osteogenesis and another marker of hypertrophicchange,29 also decreased 50% after PD98059 treatment[10mM] in BMMSCs, and decreased further in ATMSCs. Incontrast, SB203580 elevated Runx-2 mRNA levels in both celltypes (Fig. 2E).

Histological findings

Histological findings of Safranin-O and immunohisto-chemistry for type I, II, and X collagen and Runx-2 generallymirrored the findings of GAG assays and qRT–PCR exceptfor diminished expression of type I collagen in ATMSCS, andmore pronounced decrease in type X collagen and Runx-2after PD98059 treatment in BMMSCs. Safranin-O stainingshowed increased metachromatic staining after PD98059treatment in both cell types, whereas SB203580 did not havenotable effects. Type I collagen expression substantially de-clined by PD98059 in both BMMSCs and ATMSCs, but was

FIG. 2. Reverse transcription and real-time PCR. The mRNA expression of collagen type I (COL1A1) (A), collagen type II(COL2A1) (B), SOX-9 (C), collagen type X (COL10A1) (D), and runt-related transcription factor-2 (Runx-2) (E) was nor-malized versus glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Values are expressed as folds of the control. The barsrepresent means� SEM (n¼ 3: *p< 0.05, **p< 0.01, and ***p< 0.001). CM, MSCs treated with their respective chondrogenicmedium.

ERK INHIBITOR PROMOTES CHONDROGENESIS OF MSCS 855

Table 2. Scores from Histological Gradings (n¼ 3)

BMMSCs, mean (SD) ATMSCs, mean (SD)

Antigen Control 10 mM PD98059 10 mM SB203580 Control 10mM PD98059 10 mM SB203580

Safranin-O 2 (0) 4 (0) 2 (0) 2 (0) 3.7 (0.6) 2.7 (0.6)COL I 2.7 (0.6) 1 (0) 2.3 (0.6) 2.7 (0.6) 1 (0) 2.7 (0.6)COL II 1 (0) 2.7 (0.6) 1.7 (0.6) 0.7 (0.6) 2.3 (0.6) 0.7 (0.6)COL X 1.7 (0.6) 0.7 (0.6) 3 (0) 1.3 (0.6) 0.3 (0.6) 3 (0)Runx-2 1.3 (0.6) 0.3 (0.6) 2 (0) 1 (0) 0.7 (0.6) 1 (0)

BMMSCs, bone marrow–derived mesenchymal stromal cells; ATMSCs, adipose tissue–derived mesenchymal stromal cells; SD, standarddeviation; COL I, collagen type I; COL II, collagen type II; COL X, collagen type X; Runx-2, runt-related transcription factor.

FIG. 3. Safranin-O staining (�400) and immunohistochemistry (�400) for type I collagen (COL I), type II collagen (COL II),type X collagen (COL X), and Runx-2 of cultured pellets of BMMSCs (A) and ATMSCs (B). Undifferentiated BMMSCs andATMSCs (cultured without growth factor for 3 days) are shown as negative controls. Positive controls were articular cartilagefrom human femoral head for Safranin-O and type II collagen, trabecular bone from human femoral head for type I collagenand Runx-2, and rat epiphyseal plate for type X collagen (C). CM, MSCs treated with their respective chondrogenic medium.Color images available online at www.liebertonline.com=ten.

856 KIM AND IM

not affected by SB203580 treatment in both cell types. Type IIcollagen expression moderately increased in BMMSCs andATMSCs by PD98059, while SB203580 also caused an in-crease in type II collagen in BMMSCs but had no observableeffect in ATMSCs. Type X collagen expression was reducedby PD98059 in both BMMSCs and ATMSCs, but was ele-vated by SB203580 in both cell types. Runx-2 expressionbecame very faint after PD98059 treatment, but slightly in-creased by SB203580 in BMMSCs. Runx-2 had very weakexpression in ATMSCs, and was not influenced by PD98059or SB203580 (Table 2; Fig. 3).

Discussion

Chondrogenesis of MSCs has not provided a straight-forward solution for cartilage tissue engineering.7,30 De-spite the known pathways of chondrogenesis, generationof engineered cartilage tissues with MSCs has not out-

performed those of chondrocytes.30 To achieve chondro-genesis of acceptable quality from MSCs, our understandingof the characterization and differentiation of MSCs must beexpanded. The knowledge thus obtained might then betterenable us to produce articular cartilage–like tissues fromMSCs.

In the present study, we used two substances that inhibitMAPK pathways—PD98059, which indirectly inhibitsERK1=2 by acting on its upstream kinase (MEK, i.e., mitogen-activated protein kinase=ERK kinase), and SB203580, whichinhibits p38 MAPK.10 Anticipating that hypertrophy wouldbegin after initial chondrogenic differentiation, we startedusing MAPK inhibitors from day 14. Our results showed thatPD98059, when administered during the latter period ofchondrogenic in vitro culture, suppressed hypertrophy asdemonstrated by the decrease of type X collagen and Runx-2at both mRNA level and protein level, and promoted chon-drogenesis as shown by accumulation of GAG and increase of

FIG. 3. (Continued).(Figure continued ?)

ERK INHIBITOR PROMOTES CHONDROGENESIS OF MSCS 857

type II collagen. Conversely, SB203580 rather enhanced hy-pertrophy, and did not notably promote chondrogenesis.

Our current findings suggest that BMMSCs were moreresponsive to ERK and p38 inhibitors than ATMSCs. Thissuggests that other mechanisms that involve MAPK path-ways also are accountable for the lower chondrogenicpotential of ATMSCs in addition to lower levels of TGF-breceptors.31 Of the genes tested in BMMSCs in the presentstudy, type II collagen was found to be dose dependentlyrelated to PD98059, whereas SOX-9 was not. Of interest,SOX-9 was found to be dose dependently related to PD98059in ATMSCs. PD98059 decreased type I collagen RNA levelsin BMMSCs, but not in ATMSCs. These results also under-score the different natures of BMMSCs and ATMSCs.

The strength of this study is that it involves an investigationof ATMSCs, which have been identified as a potential cell

source for tissue engineering purposes.6 While the reportedlyinferior chondrogenic potential of ATMSCs has dampenedinitial enthusiasm, the present study demonstrates that anERK1=2 inhibitor induces a relatively strong chondrogenicresponse in ATMSCs and BMMSCs. Our findings thus lend afurther support for using ATMSCs for cartilage tissueengineering. Future studies will be focused on the mechanismof action of these molecules in in vitro–cultured MSC modelsbecause in this background the differentiation process is cer-tain to proceed in a different manner from that observed indeveloping embryo. Future studies should also include ap-plication of the results to in vivo implantation and other in vitrochondrogenesis models utilizing different natural and syn-thetic scaffolds.

It is of note that relatively high concentrations of PD98059tested in this study failed to completely abrogate hypertrophyas demonstrated by the extent of decrease in COL10A1 andRunx-2 gene expression, and definitely diminished but per-sistent type X collagen protein expression. These findingssuggest that this intrinsic characteristic of MSCs may not becompletely overcome using this drug. The starting point ofPD98059 treatment also deserves attention because we plan-ned the chondrogenesis of MSCs on the assumption thathypertrophy constitutes a latter part of chondrogenic differ-entiation. There are reports that type X collagen appeared asearly as type II collagen, and the authors of these studiesquestioned the validity of type X collagen as a marker of hy-pertrophy in association with stem cell differentiation.32,33

Our previous studies also showed the simultaneous increaseof type X collagen along with type II collagen in the chon-drogenesis of MSCs, demonstrating the characteristics ofMSCs inherently different from chondrocytes.25,34,35 Still,the reduction of type X collagen by ERK inhibitors inthis study should not be depreciated because it represents animprovement in the quality of induced neocartilage. Theuse of an artificial compound that inhibits a specific cellularsignal pathway to reproduce a natural process of histogenesisis a new approach. Therefore, the long-term effect of thecompound on the cell survival and metabolic pathwaysshould be further investigated before application for tissueengineering.

Lastly, it should also be mentioned that the results ob-tained in artificial models, including the pellet culture systemused in this study, may not be exactly reproduced in naturalcartilage histogenesis, considering continuous cell–matrixinteractions and positional and molecular feedback infor-mation cells receive in the cartilage development.

In conclusion, the present study suggests that PD98059, anERK 1=2 inhibitor, suppresses hypertrophy and promoteschondrogenesis of MSCs. These pathways may be furthermanipulated and tailored for establishing a model of carti-lage tissue engineering.

Acknowledgment

This work was supported by a grant from the KoreanResearch Foundation (KRF-2006-311-E00359).

Disclosure Statement

No competing financial interests exist.

FIG. 3. (Continued).

858 KIM AND IM

References

1. Caplan, A.I. Mesenchymal stem cells. J Orthop Res 9, 641,1991.

2. Pittenger, M.F., Mackay, A.M., Beck, S.C., Jaiswal, R.K.,Douglas, R., Mosca, J.D., Moorman, M.A., Simonetti, D.W.,Craig, S., and Marshak, D.R. Multilineage potential of adulthuman mesenchymal stem cells. Science 284, 143, 1999.

3. Tuan, R.S., Boland, G., and Tuli, R. Adult mesenchymal stemcells and cell-based tissue engineering. Arthritis Res Ther 5,

235, 2003.4. Chen, F.H., Rousche, K.T., and Tuan, R.S. Technology in-

sight: adult stem cells in cartilage regeneration and tissueengineering. Nat Clin Pract Rheumatol 2, 373, 2006.

5. Caplan, A.I. Review: mesenchymal stem cells: cell-basedreconstructive therapy in orthopedics. Tissue Eng 11, 1198,2005.

6. Zuk, P.A., Zhu, M., Mizuno, H., Huang, H., Futrell, J.W.,Katz, A.J., Benhaim, A.J., Lorenz, H.P., and Hedrick, M.H.Multilineage cells from human adipose tissue: implicationsfor cell-based therapies. Tissue Eng 7, 211, 2001.

7. Pelttari, K., Winter, A., Steck, E., Goetzke, K., Hennig, T.,Ochs, B.G., Aigner, T., and Richter, W. Premature inductionof hypertrophy during in vitro chondrogenesis of humanmesenchymal stem cells correlates with calcification andvascular invasion after ectopic transplantation in SCID mice.Arthritis Rheum 54, 3254, 2006.

8. Bohme, K., Winterhalter, K.H., and Bruckner, P. Terminaldifferentiation of chondrocytes in culture is a spontaneousprocess and is arrested by transforming growth factor-beta 2and basic fibroblast growth factor in synergy. Exp Cell Res216, 191, 1995.

9. Muraglia, A., Cancedda, R., and Quarto, R. Clonal mesen-chymal progenitors from human bone marrow differentiatein vitro according to a hierarchical model. J Cell Sci 113, 1161,2000.

10. Stanton, L.A., Underhill, T.M., and Beier, F. MAP kinases inchondrocyte differentiation. Dev Biol 263, 165, 2003.

11. Hagemann, C., and Blank, J.L. The ups and downs of MEKkinase interactions. Cell Signal 13, 863, 2001.

12. Watanabe, H., de Caestecker, M.P., and Yamada, Y. Tran-scriptional cross-talk between Smad, ERK1=2, and p38 mi-togen-activated protein kinase pathways regulatestransforming growth factor-beta-induced aggrecan geneexpression in chondrogenic ATDC5 cells. J Biol Chem 276,

14466, 2001.13. Oh, C.D., Chang, S.H., Yoon, Y.M., Lee, S.J., Lee, Y.S., Kang,

S.S., and Chun, J.S. Opposing role of mitogen-activatedprotein kinase subtypes, ERK-1=2 and p38, in the regulationof chondrogenesis of mesenchymes. J Biol Chem 275, 5613,2000.

14. Hwang, N.S., Varghese, S., Puleo, C., Zhang, Z., and Elis-seeff, J. Morphogenetic signals from chondrocytes promotechondrogenic and osteogenic differentiation of mesenchy-mal stem cells. J Cell Physiol 212, 281, 2007.

15. Chang, Y., Ueng, S.W., Lin-Chao, S., and Chao, C.C. In-volvement of Gas7 along the ERK1=2 MAP kinase and SOX9pathway in chondrogenesis of human marrow-derivedmesenchymal stem cells. Osteoarthritis Cartilage 16, 1403,2008.

16. Beier, F., Taylor, A.C., and LuValle, P. Raf signaling stimu-lates and represses the human collagen X promoter throughdistinguishable elements. J Cell Biochem 72, 549, 1999.

17. Beier, F., Taylor, A.C., and LuValle, P. The Raf-1=MEK=ERK pathway regulates the expression of thep21(Cip1=Waf1) gene in chondrocytes. J Biol Chem 274,

30273, 1999.18. Wojnowski, L., Stancato, L.F., Zimmer, A.M., Hahn, H.,

Beck, T.W., Larner, A.C., Rapp, U.R., and Zimmer, A. Craf-1protein kinase is essential for mouse development. MechDev 76, 141, 1998.

19. Zhen, X., Wei, L., Wu, Q., Zhang, Y., and Chen, Q. Mitogen-activated protein kinase p38 mediates regulation of chon-drocyte differentiation by parathyroid hormone. J Biol Chem276, 4879, 2001.

20. Pang, L., Sawada, T., Decker, S.J., and Saltiel, A.R. Inhibitionof MAP kinase kinase blocks the differentiation of PC-12cells induced by nerve growth factor. J Biol Chem 270, 13585,1995.

21. Cuenda, A., Rouse, J., Doza, Y.N., Meier, R., Cohen, P.,Gallagher, T.F., Young, P.R., and Lee, J.C. SB 203580 is aspecific inhibitor of a MAP kinase homologue which isstimulated by cellular stresses and interleukin-1. FEBS Lett364, 229, 1995.

22. Barry, F., Boynton, R.E., Liu, B., and Murphy, J.M. Chon-drogenic differentiation of mesenchymal stem cells frombone marrow; differentiation-dependent gene expression ofmatrix components. Exp Cell Res 268, 189, 2001.

23. Im, G.I., Shin, Y.W., and Lee, K.B. Do adipose tissue-derivedmesenchymal stem cells have the same osteogenic andchondrogenic potential as bone marrow-derived cells? Os-teoarthritis Cartilage 13, 845, 2005.

24. Im, G.I., Jung, N.H., and Tae, S.K. Chondrogenic differen-tiation of mesenchymal stem cells isolated from patients inlate adulthood: the optimal conditions of growth factors.Tissue Eng 12, 527, 2006.

25. Kim, H.J., and Im, G.I. Combination of transforming growthfactor-beta-2 and bone morphogenetic protein 7 enhanceschondrogenesis from adipose tissue-derived mesenchymalstem cells. Tissue Eng Part A 15, 1543, 2009.

26. Chang, S.H., Oh, C.D., Yang, M.S., Kang, S.S, Lee, Y.S.,Sonn, J.K., and Chun, J.S. Protein kinase C regulates chon-drogenesis of mesenchymes via mitogen-activated proteinkinase signaling. J Biol Chem 273, 19213, 1998.

27. Lefebvre, V., Huang, W., Harley, V.R., Goodfellow, P.N.,and de Crombrugghe, B. SOX-9 is a potent activator of thechondrocyte-specific enhancer of the pro-alpha1(II) collagengene. Mol Cell Biol 17, 2336, 1997.

28. Castagnola, P., Moro, G., Descalzi-Cancedda, F., and Can-cedda, R. Type X collagen synthesis during in vitro devel-opment of chick embryo tibial chondrocytes. J Cell Biol 102,

2310, 1986.29. Enomoto, H., Enomoto-Iwamoto, M., Iwamoto, M., No-

mura, S., Himeno, M., Kitamura, Y., Kishimoto, T., andKomori, T. Cbfa1 is a positive regulatory factor in chon-drocyte maturation. J Biol Chem 275, 8695, 2000.

30. De Bari, C., Dell’Accio, F., Tylzanowski, P., and Luyten, F.P.Multipotent mesenchymal stem cells from adult human sy-novial membrane. Arthritis Rheum 44, 1928, 2001.

31. Hennig, T., Lorenz, H., Thiel, A., Goetzke, K., Dickhut, A.,Geiger F., and Richter, W. Reduced chondrogenic potentialof adipose tissue derived stromal cells correlates with analtered TGFbeta receptor and BMP profile and is overcomeby BMP-6. J Cell Physiol 211, 682, 2007.

32. Mwale, F., Stachura, D., Roughley, P., and Antoniou, J.Limitations of using aggrecan and type X collagen as

ERK INHIBITOR PROMOTES CHONDROGENESIS OF MSCS 859

markers of chondrogenesis in mesenchymal stem cell dif-ferentiation. J Orthop Res 24, 1791, 2006.

33. Shen, B., Wei, A., Tao, H., Diwan, A.D., and Ma, D.D. BMP-2enhances TGF-beta3-mediated chondrogenic differentiationofhumanbonemarrowmultipotentmesenchymalstromalcellsin alginate bead culture. Tissue Eng Part A 15, 1311, 2009.

34. Kim, H.J., and Im, G.I. Chondrogenic differentiation of adi-pose tissue-derived mesenchymal stem cells: greater dosesof growth factor are necessary. J Orthop Res 27, 612, 2009.

35. Kim, H.J., Kim, Y.J., and Im, G.I. Is continuous treatmentwith transforming growth factor-beta necessary to inducechondrogenic differentiation in mesenchymal stem cells?Cells Tissues Organs 190, 1, 2009.

Address correspondence to:Gun-Il Im, M.D.

Department of OrthopaedicsDongguk University Ilsan Hospital

814 Siksa-DongGoyang 411-773

Republic of Korea

E-mail: gunil@duih.org

Received: January 31, 2009Accepted: October 5, 2009

Online Publication Date: November 9, 2009

860 KIM AND IM