phenotypic stability of articular chondrocytes in vitro

8/10/2019 Phenotypic Stability of Articular Chondrocytes in Vitro

1/9

Phenotypic Stability of Articular Chondrocytes In Vitro: The

Effects of Culture Models, Bone Morphogenetic Protein 2,and Serum Supplementation

MATTHEW C. STEWART,* KATHRYN M. SAUNDERS,

NANCY BURTON-WURSTER, and JAMES N. MACLEOD

ABSTRACT

Numerous in vitro culture models have been developed for the investigation of chondrocyte and cartilage

biology. In this study, we investigated the stability of the chondrocytic phenotype in monolayer, aggregate,

pellet, and explant culture models and assessed the effects of recombinant human bone morphogenetic protein 2

(rhBMP-2) and serum supplementation on the phenotype in each model. Phenotypic effects were assessed by

analyses of procollagen type II, aggrecan, (VC) fibronectin, and procollagen type I messenger RNA

expression. In monolayer cultures, we noted a characteristic loss of procollagen type II and induction of

procollagen type I expression. The aggregate and pellet culture models supported matrix protein gene

expression profiles more reflective of in vivo levels. In explant cultures, expression of matrix protein genes was

consistently depressed. Treatment with rhBMP-2 significantly increased the expression of procollagen type II

and aggrecan in monolayer cultures; however, other models showed comparatively little response. Similarly,

serum supplementation significantly down-regulated procollagen type II and aggrecan expression in monolayer

cultures but had less effect on gene expression in the other models. Serum supplementation increased

procollagen type I expression in monolayer and aggregate cultures. These results suggest that the influence of

exogenous BMP-2 and serum on expression of chondrocyte-specific matrix protein genes is influenced byaspects of substrate attachments, cellular morphology, and/or cytoskeletal organization. Finally, the analyses of

fibronectin expression suggest that V and C region alternative splicing in chondrocytes is linked to the

establishment of a three-dimensional multicellular complex. (J Bone Miner Res 2000;15:166174)

Key words: articular chondrocyte, cartilage, BMP-2, matrix protein, phenotype

INTRODUCTION

ARTICULAR CARTILAGEis an avascular, aneural tissue thatcovers the articulating surfaces of bones. It distributesload and minimizes friction associated with joint movement.

Articular cartilage is a relatively acellular tissue. Chondro-

cytes comprise only 12% of the biomass but are respon-

sible for the synthesis and maintenance of the extracellular

matrix, which constitutes the bulk of the tissue.(1) The

load-bearing properties of articular cartilage result from the

unique components and organization of the extracellular

matrix. The high water content of cartilage (7080% of total

weight) is maintained by electrostatic interactions betweenwater molecules and the sulfated glycosaminoglycan moi-

eties of aggrecan complexes.(2) This proteoglycan gel,

anchored to hyaluronan polymers, provides the compressive

strength of the tissue. The arcuate arrangement of collagen

type II fibrils constrains the proteoglycan gel and contributes

to shear force resistance at the articular surface.(3) The

extracellular matrix is also active in regulating chondrocyte*Current affiliation: Department of Orthopaedics, Case Western

Reserve University, Cleveland, Ohio, U.S.A.

James A. Baker Institute for Animal Health, College of Veterinary Medicine, Cornell University, Ithaca, New York, U.S.A.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 15, Number 1, 2000

2000 American Society for Bone and Mineral Research

166


2/9

activity, through transmission of biomechanical stimuli via

integrin-mediated (46) and other receptor-mediated(7,8) signal-

ing pathways and by influencing the distribution and bioavail-

ability of growth factors such as fibroblast growth factor,

transforming growth factor 1, and bone morphogenetic

proteins.(9,10)

Fibronectin is an extracellular matrix protein that is

ubiquitously expressed in connective tissues. Fibronectin isencoded by a single gene, but substantial protein heterogene-

ity is introduced by alternative splicing of the primary

transcript. A cartilage-specific splice variant of fibronectin

that lacks the nucleotides encoding both the V and C regions

of the protein has been described.(11) In mammals,

this (VC) fibronectin isoform constitutes 3080% of

total fibronectin in articular cartilage, with the percen-

tage increasing with age.(12) The relative expression of

this isoform drops significantly during chondrocyte isolation

and remains low in monolayer culture.(11,12) Unlike the

collagen and proteoglycan components of the matrix, the

functional roles of (VC) fibronectin have not been

defined.Many in vitro culture models have been developed for the

investigation of chondrocyte and cartilage biology, includ-

ing explant models, several forms of three-dimensional

culture systems, and monolayer cultures.(13) There is also

considerable variation in aspects of media composition and

supplementation that affect the expression of the chondro-

cytic phenotype. Chondrocytes grown in monolayer culture

undergo a characteristic process of dedifferentiation, marked

by a loss of collagen type II and aggrecan core protein

expression and the induction of collagen type I expres-

sion.(14,15) This phenomenon is influenced to some extent by

seeding density(16) and is accelerated by growth in medium

supplemented with serum and by passage.(14) Conversely,

members of the bone morphogenetic protein (BMP) familyprevent dedifferentiation of monolayer chondrocyte cul-

tures.(1720) Growth of chondrocytes under conditions that

support a rounded morphology also facilitates maintenance

of the differentiated chondrocytic phenotype(2123); however,

the phenotypic effects of serum supplementation and BMP-2

treatment in nonadherent or suspension cultures have been

less well characterized.

The present study was conducted to investigate three

related issues: the phenotypic stability of articular chondro-

cytes in differing in vitro models, the phenotypic effects of

recombinant human bone morphogenetic protein 2 (rh-

BMP-2) and serum supplementation in these models, and

the relationship between fibronectin V and C region RNAsplicing and the expression of recognized markers of the

chondrocytic phenotype. Equine articular chondrocytes were

maintained as monolayers, nonadherent aggregates, pellets,

or explants. The three-dimensional models used in these

experiments support a rounded cellular morphology but

differ in aspects of cell-to-cell and cell-to-matrix interac-

tions. Messenger RNA (mRNA) levels of procollagen type

II, aggrecan core protein, the (VC) fibronectin splice

variant, and procollagen type I were assayed to monitor the

effects of the culture models on the chondrocytic phenotype.

MATERIALS AND METHODS

Materials

Dulbeccos modified Eagle medium (DMEM), Hams F12

medium, Opti-MEM, Geys balanced salt solution (GBSS),

Hanks balanced salt solution (HBSS), fetal bovine serum

(FBS), penicillin/streptomycin and amphotericin B were

purchased from GIBCO BRL/Life Technologies (GrandIsland, NY, U.S.A.). Cell culture flasks and Ultra Low

Attachment, hydrogel-coated plates were purchased from

Corning Inc. (Corning, NY, U.S.A.). L-Ascorbic acid phos-

phate was purchased from Wako Pure Chemicals (Rich-

mond, VA, U.S.A.). [32P]Deoxycytidine triphosphate

([32P]dCTP) and [32P]uridine triphosphate ([32P]UTP) were

purchased from Amersham (Arlington Heights, IL, U.S.A.).

Collagenase type CLS1 from Clostridium histolyticum was

purchased from Worthington Biochemicals (Freehold, NJ,

U.S.A.). Erythrosin B and dimethyl sulfoxide were pur-

chased from Sigma Chemical Company (St. Louis, MO,

U.S.A.).

Isolation of chondrocytes

The protocols for isolation and cryopreservation of equine

articular chondrocytes were adapted from the techniques

described by Nixon et al.(24) Articular cartilage shavings

were collected from the limb joints of six skeletally imma-

ture horses ranging from 1 week to 14 months of age. The

shavings were stored on ice in GBSS that contained 200 U of

penicillin/200 g of streptomycin (2% v/v) and 2.5 g of

amphotericin B/ml. Partial thickness sections were collected

to avoid including chondrocytes from the underlying epiphy-

seal ossification center. Procollagen type X mRNA, a marker

of the hypertrophic chondrocyte phenotype, was undetect-

able in subsequent Northern blot analyses (data not shown).

The articular cartilage pieces were cut into approximately1-mm-thick slices and either used directly for explant

cultures or digested overnight in 0.1% collagenase type

CLS1 (1:1) in DMEM/Hams F12, 10% FBS, 2% penicillin/

streptomycin, and 2.5 g/ml amphotericin B at 37C.

Isolated chondrocytes were recovered by filtration through

40-m nylon mesh, counted using a hematocytometer, and

assessed for viability by erythrosin B exclusion. Chondro-

cytes that were not used immediately were resuspended at a

concentration of 1.5 107 cells in 1.5 ml of DMEM, 10%

FBS, and 10% dimethyl sulfoxide and then stored in liquid

nitrogen. The viability of cryopreserved cells was reassessed

by erythrosin B exclusion after thawing; viability was

routinely90%.

Culture conditions

A schematic diagram of the establishment sequence for

the in vitro models is shown in Fig. 1. Each experiment was

carried out with cells isolated from an individual donor.

Chondrocytes from each donor were cultured in at least two

of the in vitro models assessed in this study. Cartilage used

in explant cultures was diced into thin sections, as described

earlier (1025 mg wet weight), and maintained in six-well

plates. Each well contained 300350 mg of cartilage in 7 ml

167IN VITRO CHONDROCYTIC PHENOTYPE STABILITY


3/9

of medium. Chondrocyte pellets were established in micro-

centrifuge tubes by suspension of 2 105 cells in 500 l of

medium. The tubes were centrifuged at 300 relative

centrifugal force (rcf) for 5 minutes in a benchtop centrifuge.

The tops of the tubes were perforated with an 18-gauge

needle after centrifugation to permit gaseous exchange.

After 72 h, corresponding to the first time the medium was

changed, the pellets were gently aspirated from the microcen-

trifuge tubes and transferred to six-well hydrogel-coated

plates. Groups of 1520 pellets/well were maintained in 6 ml

of medium. Nonadherent aggregate cultures also were

maintained in hydrogel-coated six-well plates. Three million

cells were placed in each well in 5 ml of medium. During the

first 72 h of culture, the floating cells formed clearly visibleaggregates. At each medium change, spent medium was

aspirated and centrifuged at 300 rcf for 5 minutes. Any

aspirated cells were returned to the appropriate wells, along

with fresh medium. Monolayer cultures were established by

seeding 5 106 cells in T25 flasks at an initial seeding

density of 2 105 cells/cm2. The monolayers were 70%

confluent 24 h after seeding and had reached confluence by

72 h.

In preliminary experiments that confirmed changes in

gene expression in monolayer culture, dedifferentiation of

equine articular chondrocytes was evident within 710 days

and was accelerated by culture in serum-containing medium,

consistent with data from other studies.(14,15) Subsequent

experiments were conducted within this time frame. Articu-lar chondrocytes or cartilage explants were cultured for the

first 72 h under serum-free conditions in Opti-MEM, a

defined culture medium that contains insulin, transferrin,

and selenous acid and is specifically designed for use under

low-serum conditions. This initial period allowed attach-

ment and expansion of monolayer cultures to confluence, the

formation of the nonadherent aggregates, and the consolida-

tion of pellets after centrifugation.

The culture medium was changed after 72 h (Time 0), at

which point cultures were left in Opti-MEM, were treated

with 100 ng/ml rhBMP-2 (a generous gift from the Genetics

Institute, Cambridge, MA, U.S.A.), or were transferred to

DMEM/Hams F12 medium (1:1) that contained 10% FBS.

DMEM/Hams F12 medium was selected for use in these

experiments because it is commonly used for serum-

supplemented chondrocyte cultures. Samples were collected

for analyses on days 1, 4, and 7 after commencement of

treatments. All media were supplemented with 50 g/mlL-ascorbic acid phosphate, 1% (v/v) penicillin/streptomycin,

and 1.0 g/ml amphotericin B and were replaced every 48 h

after the initial medium change. The cultures were main-

tained at 37C, 95% air/5% CO2in a humidified incubator.

Histology

Histomorphological characteristics of monolayer cultures

were assessed directly by phase-light microscopy. Pellets

and aggregates were collected on day 7; fixed in 4%

paraformaldehyde in phosphate-buffered saline (PBS), pH

7.4, for 23 h at 4C; and suspended in low-melting-

temperature agar to facilitate sectioning. After fixation, the

samples were dehydrated in serial ethanol solutions andembedded in paraffin. Sections 5 m thick were stained with

hematoxylin and eosin to enhance cellular detail or with

Alcian blue/periodic acidSchiff (PAS)(25) to identify the

extracellular matrix.

RNA isolation and Northern blot analyses

On days 1, 4, and 7 of culture, samples were collected

from control and rhBMP-2 treated cultures and snap frozen

in liquid nitrogen. Additional experiments were conducted

that were sampled only on day 7, to increase sample size and

statistical power of analyses at this time point. Samples from

serum-supplemented cultures were collected only on day 7.

Total RNA was isolated from cartilage samples and fromexplants using a protocol described previously.(11) Total

RNA was isolated from pellets by using a commercial kit

(RNeasy; Qiagen, Inc., Chatsworth, CA, U.S.A.) operated

according to the manufacturers instructions. Total RNA was

isolated from aggregate and monolayer chondrocyte cultures

by using a commercially available single-step phenol/

guanidine thiocyanate protocol (Tri Reagent; Molecular

Research Center Inc., Cincinnati, OH, U.S.A.), based on the

method of Chomczynski and Sacchi.(26)

Northern analyses of total RNA samples were carried out

according to standard protocols.(27) Sample loading and

transfer efficiency of each blot were normalized by densito-

metric comparison of ethidium bromidestained 28S and

18S ribosomal bands on the nylon membranes immediatelyafter transfer, using Scion Image version 1.60 gel documen-

tation software (Scion Corporation, Frederick, MD, U.S.A.).

Radiolabeled probes were prepared from gel-purified cDNA

insert preps by using [32P]dCTP and random hexanucleotide

primers (Promega, Madison, WI, U.S.A.) and purified with

Sephadex G-50 spin columns (Boehringer-Mannheim, India-

napolis, IN, U.S.A.). Prehybridization, hybridization, and

wash conditions followed protocols recommended by the

manufacturer of the nylon membranes (MSI, Westboro, MA,

U.S.A.). After exposure to radiographic film, the Northern

FIG. 1. Establishment of the in vitro chondrocyte culture

models. Articular cartilage shavings were harvested from the

limb joints of 6 skeletally immature horses, 1 week to 14

months old. Cartilage was either cultured directly as ex-

plants or digested with collagenase to isolate chondrocytes

for culture as pellets, aggregates, or monolayers as described

in Materials and Methods. Chondrocytes isolated from each

animal were cultured in at least two of the models described.

168 STEWART ET AL.


4/9

blot data were quantified by using the Fujix bio-imaging

analyzer system (BAS1000 MacBAS; Fuji Medical Systems

USA, Stamford, CT, U.S.A.).

Ribonuclease protection assays

Relative amounts of alternatively spliced mRNAs that

encoded different isoforms of fibronectin were determinedby ribonuclease protection assays (RPAs) as described

previously.(12) An equine-specific version of the I-10/I-11

cDNA fragment used as a template for riboprobe synthesis

to measure the cartilage-specific (VC) fibronectin iso-

form was generated by polymerase chain reaction (PCR)

using 5-TCCTCTAGAGCCAGTGCTTAGGGTTTG (sense)

and 5-TCCTCTAGAAGACACG-TGCAGCTCATC (anti-

sense) primers that included Xba 1 linkers. The amplified

fragment was then cloned into pGEM-3Zf() (Promega).

Antisense riboprobe was synthesized in the presence of

[32P]UTP by in vitro transcription using SP6 RNA polymer-

ase (MAXISCRIPT; Ambion, Austin, TX, U.S.A.). The RPA

was performed by using a commercial kit (RPA II; Ambion)

and following the manufacturers recommended protocol.Protected fragments were separated on a 6% polyacrylamide

gel that contained 8 M urea and were quantitated directly

from the dried gel using the Fujix bio-imaging analyzer.

Resulting numerical data were then corrected for sizes of the

protected fragments to determine the ratio of (VC)

fibronectin mRNA to other fibronectin splice variants.

cDNA probes

A 3000-bp equine procollagen type II cDNA(28) was

generously provided by Dr. Dean Richardson (The Univer-

sity of Pennsylvania). A 1950-bp Xho 1 fragment of this

cDNA was used for probe synthesis. Equine-specific cDNA

probes for aggrecan core protein (306 bp) and fibronectin(840 bp) were generated by reverse-transcription (RT) PCR,

as described previously.(29) A 516-bp equine type I (2)

procollagen cDNA fragment was generated by RT-PCR

using total bone RNA and 5-TCGCCCTGGAGAGCCT

(sense) and 5-GGACCTCGGCTTCCAATAGG (antisense)

primers. The amplified region corresponds to bases 1454

1969 and 13931908 in the published canine(30) (accession

no. AF035120) and human(31) (accession no. Y00724)

sequences, respectively.

Statistical analyses

Variation in articular cartilage matrix gene expression was

present between animals. As reported in previous stud-ies,(18,28,32) chondrocytes isolated from younger animals

exhibited greater biosynthetic activity than cells isolated

from the older donors in the study. To account for this

individual variation, experimental data were normalized

against representative tissue values obtained from articular

cartilage samples collected from each donor and analyzed in

parallel. Thus, the in vitro data were expressed as a ratio of

the measured in vivo expression. Procollagen type I mRNA

values were normalized against total RNA isolated from a

single equine ligament sample. Means and standard devia-

tions of the normalized data were computed from replicate

experiments in each model. Mean normalized expression of

procollagen type II, aggrecan, and total and (VC) fibro-

nectin were compared with in vivo expression (value of

1.00) in each model under serum-free conditions by using a

one-sample Studentst-test. Mean procollagen type I expres-

sion was analyzed in the same manner by comparison with

0.00, because type I procollagen mRNA is undetectable inarticular cartilage by Northern blot analysis. The effects of

rhBMP-2 and serum supplementation were analyzed by

comparing mean treated and control values at each time

point. Fibronectin data from matched control and rhBMP-2

treated samples were pooled for culture model analysis,

because treatment with rhBMP-2 had no detectable effect on

total fibronectin expression nor on relative expression of the

(VC) fibronectin isoform in any model. The effects of

treatment with rhBMP-2 or with FBS were analyzed by the

Wilcoxon rank-sum test to account for differences in sample

sizes and variances; p values 0.05 were considered to be

significant.

RESULTS

Histomorphology

After reaching confluence, chondrocytes in monolayer

cultures exhibited a polygonal morphology, and developed a

characteristic cobblestone appearance. The cell layer

became increasingly refractile with time in culture, indica-

tive of pericellular matrix production.(15) Histologic analysis

of aggregates showed rounded cells surrounded by extracel-

lular matrix (Fig. 2A) that stained strongly with Alcian

blue/PAS. A zone of flattened cells, distinctly different from

the rounded cells in the center of the structures, covered the

surface of the aggregates (Fig. 2B). Alcian blue staining of

the matrix immediately below this peripheral zone of cellswas less than in the center of the aggregates. Pellets were

more cellular than aggregates with less intercellular matrix

(Fig. 2C). Flattened cells were evident on the surface of

pellets, similar to those seen with aggregates but limited to

the most peripheral layer (Fig. 2D).

Effects of in vitro models on matrix protein

gene expression

Representative Northern analyses of procollagen type II,

aggrecan, and procollagen type I expression under serum-

free conditions from single experiments are shown in Fig. 3

along with representative tissue blots. The mean steady-state

mRNA levels in each model normalized against in vivoexpression are shown in Fig. 4.

Procollagen type II

In monolayer cultures, procollagen type II mRNA levels

were 30% of in vivo levels throughout the experiments,

significantly below in vivo expression at all three time

points. Procollagen type II expression in aggregate cultures

was stable throughout the time course of the experiments at

50% of tissue levels. Expression in pellets was low



5/9

initially but approached in vivo levels by day 4. Procollagen

type II mRNA levels in explants were substantially lower

than tissue levels at all time points; however, the differences

at days 1 and 4 were not statistically significant because ofthe small sample size.

Aggrecan

In vitro patterns of aggrecan mRNA expression were

similar to those of procollagen type II. Steady-state mRNA

levels of aggrecan in monolayer cultures were lower than

tissue levels initially but approached in vivo levels by day 7.

In contrast, expression in aggregate and pellet cultures

equaled or exceeded tissue levels throughout the culture

period, with a more than 2-fold increase in expression in

pellets at day 4. As with procollagen type II, aggrecan

expression in explants was reduced, averaging 50% of tissue

levels by day 7.

Procollagen type I

Induction of procollagen type I expression was evident by

day 1 in monolayer cultures and peaked on day 4. Low levels

of procollagen type I expression also were detected in some

aggregate and pellet experiments. No procollagen type I

mRNA was detected in explant cultures, despite the substan-

tial reduction in the expression of chondrocyte-specific

markers.

(VC) fibronectin

Steady-state levels of (VC) fibronectin transcripts in

articular cartilage from animals used in this study ranged

from 26% to 65% of total fibronectin mRNA, increasing

with age. The relative expression of (VC) fibronectin in

each culture model, normalized against tissue levels, is

shown in Fig. 5A. Relative levels of the (VC) isoform in

monolayer cultures were significantly lower than in vivo

values throughout the culture period, consistent with previ-

ous results.(11,12) In contrast, relative amounts increased with

time in aggregate and pellet cultures and were not signifi-

cantly different from tissue levels by day 4 in these models.

In explants, the relative expression of (VC) transcriptsexceeded tissue levels at all three time points.

Changes in relative expression of (VC) fibronectin did

not follow the same patterns as that of procollagen type II

and aggrecan. However, changes in isoform ratios were

caused in part by changes in total fibronectin expression

(Fig. 5B). Total fibronectin expression, primarily C iso-

forms, was markedly increased in the early stages of

monolayer (8-fold), aggregate (8-fold), and pellet (4-fold)

cultures. In aggregate and pellet models, the re-establish-

ment of relative (VC) isoform expression consistent with

FIG. 2. Histomorphological appearance of equine articu-

lar chondrocytes cultured as aggregates or pellets in serum-

free Opti-MEM for 10 days. This interval included a 72-h

consolidation period. Aggregates and pellets were fixed in

4% paraformaldehyde in PBS, pH 7.4, and processed for

histologic analysis. Sections (5 m thick) were stained with

Alcian blue/PAS. (A) Central regions of aggregates con-

tained rounded cells separated by matrix. (B) Peripherally, a

zone of more flattened cells was evident (arrowheads). (C)

Central regions of pellets were substantially more cellular

than aggregates. (D) A peripheral layer of flattened cells also

was evident (arrowheads) on the surface of pellets. (Magni-

fication100.)

FIG. 3. Representative Northern analyses of matrix gene

expression in equine articular chondrocytes. Chondrocytes

were cultured as monolayers (M), aggregates (A), pellets

(P), or explants (E) in serum-free Opti-MEM. After 72 h,

cultures remained under control conditions or were treated

with 100 ng/ml rhBMP-2. Total RNA was isolated at days 1,

4, and 7. Total RNA from representative cartilage tissue (T)

was included in each analysis. Variability in signal intensity

in the tissue samples is reflective of individual variation in in

vivo expression, differing probe specific activities, and

lengths of autoradiographic exposure in each experiment.

170 STEWART ET AL.


6/9

in vivo values was associated with a return of total fibronec-

tin expression toward tissue levels. This was not the case in

monolayer cultures. In explants, the increase in relative

(VC) levels was associated with a significant reduction in

total fibronectin expression.

Effects of rhBMP-2 on matrix protein gene expression

Representative Northern analyses of matrix gene expres-

sion in cultures treated with 100 ng/ml rhBMP-2 for 1, 4,

and 7 days are shown alongside serum-free control culturesin Fig. 3. Quantitative comparisons of control and rhBMP-2

treated cultures are given in Fig. 4. The addition of rhBMP-2

to serum-free monolayer cultures increased the expression

of procollagen type II and aggrecan mRNAs approximately

2- and 3-fold, respectively. The increase in matrix protein

synthesis was associated with increased refractility of mono-

layer cultures treated with rhBMP-2 (data not shown).

Aggregate, pellet, and explant cultures were comparably less

responsive to rhBMP-2 supplementation. In aggregate cul-

tures, both procollagen type II and aggrecan expression

FIG. 4. Effects of culture models, rhBMP-2, and FBS on

matrix protein gene expression in equine articular chondro-

cytes. Chondrocytes were cultured as monolayers (days 1

and 4,n 4 replicates; day 7,n 6), aggregates (days 1 and

4,n 4; day 7,n 5), pellets (days 1 and 4,n 3; day 7,

n 5), or explants (days 1 and 4, n 2; day 7, n 3) in

serum-free Opti-MEM without (Control) or with 100 ng/mlof rhBMP-2, or in DMEM/Hams F12 medium (1:1) that

contained 10% FBS. Steady-state mRNA levels of procolla-

gen type II and aggrecan from culture samples were

measured by Northern blot analyses (as described in Materi-

als and Methods) and normalized against in vivo levels

obtained from samples of articular cartilage collected from

each animal. Procollagen type I expression was normalized

against values obtained from equine ligament. Means and

standard deviations were calculated from the results of

replicate experiments. Single-sample Students t-test was

used to compare expression under serum-free conditions

with cartilage expression, which was assigned a value of

1.00 for collagen type II and aggrecan and 0.00 for type I

collagen. a, significantly different from tissue values. (p 0.05). b, significant effects of rhBMP-2 and FBS supplemen-

tation on matrix protein gene expression compared with

expression in serum-free control samples at the same time

points (p 0.05). Bars SD.

FIG. 5. Expression of (VC) fibronectin in equine

articular chondrocytes. Chondrocytes were cultured as mono-

layers (n 8 replicates), aggregates (n 6), pellets (n 6),

or explants (n 6) in serum-free Opti-MEM (Control) or in

DMEM/Hams F12 medium (1:1) that contained 10% FBS.

Total RNA was collected from serum-free cultures on days

1, 4, and 7; samples were collected from serum-supple-

mented cultures only on day 7. (A) Relative levels of

(VC) fibronectin were determined by RPAs. (B) Total

steady-state fibronectin expression was determined by North-

ern analyses. RPA and Northern data were normalized

against levels obtained from representative cartilage samples.

a, experimental samples that differ significantly from tissue

expression, assigned a value of 1.00. (p 0.05). b, signifi-

cant effects of FBS supplementation at day 7 compared

with expression under serum-free conditions (p 0.05).

Bars SD.



7/9

increased by 50%. The responses in pellet and explant

cultures were not statistically significant at any time point.

Treatment with rhBMP-2 did not significantly affect

expression of procollagen type I mRNA, nor was there any

detectable effect on the pattern of total fibronectin expres-

sion or of the relative expression of the (VC) splice

variant in any culture model (Fig. 3).

Effects of FBS on matrix protein gene expression

The effects of serum supplementation on matrix protein

gene expression were assessed after 7 days (Fig. 4). A

profound down-regulation of procollagen type II and aggre-

can expression occurred under monolayer conditions. In

contrast, significant FBS-induced changes in the expression

of these genes were not observed in the three-dimensional

models. Serum supplementation did increase procollagen

type I expression in aggregates, as was observed in monolay-

ers. The relative expression of the (VC) fibronectin

isoform in aggregates was significantly reduced by serum

supplementation (Fig. 5A); however, total fibronectin expres-

sion was increased in both monolayer and aggregate cul-tures. Serum supplementation did not alter the down-

regulation of fibronectin expression in explants.

DISCUSSION

In vitro culture systems are, in all instances, simplified

models of an in vivo milieu. Such models allow experimen-

tal manipulation of specific factors under highly controlled

conditions, circumstances often unattainable in an in vivo

context. However, the value of in vitro data is predicated on

the model being representative of and consistent with the in

vivo situation under investigation. All four in vitro models

evaluated in this study excluded biomechanical forcesexperienced by articular chondrocytes in vivo. With the

exception of explants, the models also utilized cells isolated

from cartilage matrix, although resynthesis of extracellular

matrix components has been documented in these sys-

tems.(28,33,34)

The documented alterations in cellular morphology and

matrix gene expression that occur in monolayer culture,

particularly when grown in the presence of serum, are

important parameters to consider when this model is used to

address issues of chondrocyte biology. Significant changes

in expression of all four matrix genes were noted in this

study. The proliferative capacity of chondrocytes is substan-

tially higher under monolayer conditions compared with

nonadherent systems. This capacity has been exploited forpurposes of population expansion, followed by re-establish-

ment of a chondrocytic phenotype by transfer to a three-

dimensional culture system.(21,22)

Both the aggregate and pellet culture models supported

the chondrocytic phenotype throughout the time course of

these experiments and were relatively resistant to FBS-

induced phenotypic changes. The differences between the

patterns and levels of matrix gene expression in these

models during the early stages of culture may reflect the

methods of their establishment. Aggregates developed spon-

taneously through cell-to-cell interactions during the initial

stages of culture, as previously documented.(35) In contrast,

pellets were effectively established as ultrahigh-density

cultures, initially lacking pericellular and intercellular ma-

trix domains. The difference in cellularity remains apparent

at day 7 of culture in the histologic sections shown in Fig. 2.

Relating these data to those of other studies,(5,6,36) the

establishment of a pericellular matrix and intercellularseparation may be requisite for stable expression of the

articular chondrocyte phenotype, in agreement with the

chondron model proposed by Poole and co-workers.(3739)

Expression of type II procollagen, aggrecan, and fibronec-

tin was markedly reduced in explant cultures, consistent

with previous findings.(40) The dimensions and weights of

the explants used in these experiments were comparable to

those in other published studies.(4043) The total RNA/g of

explant tissue did not change appreciably during the course

of the experiments, suggesting that cell viability in the

explants was not compromised. Down-regulation of matrix

protein gene expression in explants may represent a physi-

ological response, following removal of biomechanical

stimuli and the impetus for matrix turnover. Reducedchondrocyte biosynthetic activity has been observed in vivo

after joint immobilization or protection from weight-bearing

activity.(44,45) Although no significant changes were detected

in explants in response to rhBMP-2 or serum, explants in

parallel experiments increased fibronectin expression by 8-

to 10-fold after treatment with 5.0 ng/ml of transforming

growth factor 1 (TGF-1; data not shown), indicating that

the cells remained responsive to specific exogenous stimuli.

These findings are of direct relevance to investigations of

matrix protein synthesis in explant models, because these

data indicate that basal levels of expression are significantly

lower than in vivo.

Variable induction of procollagen type I expression in the

aggregate and pellet models was observed despite retentionof chondrocyte marker expression and levels of expression

comparable to those in vivo. Inclusion of collagen type

Iexpressing cells in the initial isolation procedure was

unlikely, because equine articular cartilage is a well-defined

and homogeneous tissue for collection purposes. Procolla-

gen type I mRNA was undetectable by Northern analyses of

samples of cartilage and freshly isolated chondrocytes.

Furthermore, few cell types are able to survive under

nonadherent conditions used for establishing aggregate

cultures. It has been suggested that the in vitro expression of

collagen type I by chondrocytes is indicative of transdiffer-

entiation of terminally differentiated chondrocytes into an

osteoblastic lineage.(4648) This explanation may be rele-

vant to chondrocytes isolated from cartilage undergoingendochondral ossification but does not seem to be applicable

to articular chondrocytes, which do not undergo hypertro-

phic nor osteoblastic differentiation under normal condi-

tions.

A peripheral zone of spindle-shaped cells was apparent in

histologic analyses of aggregates and pellets (Fig. 2B and

D). These flattened cells have been noted in other three-

dimensional chondrocyte culture systems(23,35) and have

been distinguished from underlying, collagen type II

expressing cells on the basis of cell adhesion molecule

172 STEWART ET AL.


8/9

(N-CAM) expression.(36) These surface cells are analogous

in some respects to collagen type Iexpressing perichondrial

cells that cover the surfaces of cartilage anlaga in vivo. (49)

Their development in cultures initiated with articular chon-

drocytes implies a considerable degree of plasticity in what is

generally considered to be a fully differentiated phenotype.

Treatment of monolayer cultures with rhBMP-2 signifi-

cantly increased the expression of procollagen type II andaggrecan core protein, consistent with the results of other

studies.(20) In contrast to its effect under monolayer condi-

tions, rhBMP-2 caused relatively little biosynthetic response

in the aggregate, pellet, and explant cultures. This difference

may reflect limitations in the ability of rhBMP-2 to diffuse

into and through the extracellular matrices of aggregates,

pellets, and explants; however, consistent induction of

procollagen type X in response to rhBMP-2 has been noted

in growth plate chondrocytes cultured under analogous

culture conditions.(50) The administration of rhBMP-2 to

monolayer cultures may compensate for the loss of endog-

enous BMP-2 or other biosynthetic and/or morphogenic

stimuli that are sustained in three-dimensional models.Relative expression of (VC) fibronectin drops quickly

when articular chondrocytes are enzymatically isolated from

cartilage and grown as monolayer cultures(11,12) (Fig. 5).

This phenomenon is also evident in the early stages of

aggregate and pellet cultures. The current data indicate that

two processes are occurring simultaneously. First, total

fibronectin expression is rapidly up-regulated. Second, alter-

native splicing patterns of fibronectin transcripts shift,

decreasing relative levels of the (VC) isoform. In abso-

lute terms, the amount of the (VC) splice variant is

increased by the processes of chondrocyte isolation and

culture but substantially less so than the increase in C

isoforms.

As was seen in monolayer cultures, relative expression of

(VC) fibronectin is lost when chondrocytes are seeded

and remain isolated in a three-dimensional alginate bead

matrix.(12) By contrast, in the aggregate and pellet culture

models that allow at least transient intercellular contact,

multicellular organization, and the development of a three-

dimensional matrix, the relative expression of the (VC)

transcript returns to in vivo levels by day 4 of culture. These

data with equine articular chondrocytes suggest that the

expression of specific V and C region fibronectin splice

variants in cartilage is linked to the establishment of a three-

dimensional, multicellular structureand may reflect the establish-

ment of a structurally mature cartilaginous tissue in vitro.

ACKNOWLEDGMENTS

This study was supported by National Institutes of Health

Grant AR44340 and funding from the Arthritis Foundation.

The authors thank Marlene Crissey for assistance with

preparation of samples for histologic analyses and the

Genetics Institute for providing the rhBMP-2 used in these

experiments.

REFERENCES

1. Buckwalter JA, Mankin HJ 1997 Articular cartilage Part I:Tissue design and chondrocyte:matrix interactions. J Bone JtSurg (Am)79:600611.

2. Hardingham TE, Fosang AJ 1992 Proteoglycans: Many forms,many functions. FASEB J6:861870.

3. Eyre DR, Wu J-J 1995 Collagen structure and cartilage matrix

integrity. J Rheumatol Suppl43:8285.4. Arner EC, Tortorella MD 1995 Signal transduction throughchondrocyte integrin receptors induces matrix metalloprotein-ase synthesis and synergizes with interleukin-1. ArthritisRheum38:13041314.

5. Enomoto M, Leboy PS, Menko S, Boettiger D 1993 1Integrins mediate chondrocyte interaction with type I collagen,type II collagen, and fibronectin. Exp Cell Res 205:276285.

6. Enomoto-Iwamoto M, Iwamoto M, Nakashima K, Mukudai Y,Boettiger D, Pacifici M, Kurisu K, Suzuki F 1997 Involvementof 51 integrin in matrix interactions and proliferation ofchondrocytes. J Bone Miner Res 12:11241132.

7. Ishida O, Tanaka Y, Morimoto I, Takigawa M, Eto S 1997Chondrocytes are regulated by cellular adhesion through CD44and hyaluronic acid pathway. J Bone Min Res 12:16571663.

8. Schlessinger J 1997 Direct binding and activation of receptortyrosine kinases by collagen. Cell91:869872.

9. Reddi AH 1994 Cartilage morphogenesis: Role of bone andcartilage morphogenetic proteins, homeobox genes and extra-cellular matrix. Matrix Biol 14:599606.

10. Spivak-Kroizman T, Lemmon MA, Divic I, Ladbury JE,Pinchasi D, Huang J, Jaye M, Crumly G, Schlessinger J, Lax I1994 Heparin-induced oligomerization of FGF molecules isresponsible for FGF receptor dimerization, activation, and cellproliferation. Cell79:10151024.

11. MacLeod JN, Burton-Wurster N, Gu DN, Lust G 1996Fibronectin mRNA splice variant in articular cartilage lacksbases encoding the V, III-15, and I-10 protein segments. J BiolChem271:1895418960.

12. Burton-Wurster N, Borden C, Lust G, MacLeod JN 1998Expression of the (VC) fibronectin isoform is tightly linkedto the presence of a cartilaginous matrix. Matrix Biol17:193203.

13. Adolphe M, Benya P 1992 Different types of cultured chondro-

cytes: The in vitro approach to the study of biologicalregulation. In: Adolphe M (ed.) Biological Regulation of theChondrocytes. CRC Press, Ann Arbor, Michigan, U.S.A., pp.105139.

14. Hering T, Kollar J, Huynh TD, Varelas JB, Sandell LJ 1994Modulation of extracellular matrix gene expression in bovinehigh-density chondrocyte cultures by ascorbic acid and enzy-matic resuspension. Arch Biochem Biophys314:9098.

15. Lefebvre V, Garofalo S, Zhou G, Metsaranta M, Vuorio E, DeCrombrugghe B 1994 Characterization of primary cultures ofchondrocytes from type II collagen/-galactosidase transgenicmice. Matrix Biol14:329335.

16. Ronziere M-C, Farjanel J, Freyria A-M, Hartmann DJ, HerbageD 1997 Analysis of types I, II, III, IX and XI collagenssynthesized by fetal bovine chondrocytes in high-densityculture. Osteoarthritis Cartilage5:205214.

17. Hiraki Y, Inoue H, Shigeno C, Sanma Y, Bentz H, Rosen DM,Asada A, Suzuki F 1991 Bone morphogenetic proteins (BMP-2and BMP-3) promote growth and expression of the differenti-ated phenotype of rabbit chondrocytes and osteoblasticMC3T3-E1 cells in vitro. J Bone Miner Res 12:13731385.

18. Luyten FP, Chen P, Paralkar V, Reddi AH 1994 Recombinanthuman bone morphogenetic protein-4, transforming growthfactor-beta 1, and activin A enhance the cartilage phenotype ofarticular chondrocytes in vitro. Exp Cell Res210:224229.

19. Luyten FP, Yu YM, Yanagishita M, Vukicevic S, HammondsRG, Reddi AH 1992 Natural bovine osteogenin and recombi-nant human bone morphogenetic protein-2B are equipotent inthe maintenance of proteoglycans in bovine articular cartilageexplant cultures. J Biol Chem267:36913695.



9/9

20. Sailor LZ, Wang JH, Morris EA, Hewick RH 1996 BoneMorphogenetic Protein-2 (BMP-2) maintains the phenotype ofarticular chondrocytes in long term monolayer culture. JOrthop Res14:937945.

21. Binette F, McQuaid DP, Haudenschild DR, Yaeger PC, Mc-Pherson JM, Tubo R 1998 Expression of a stable articularcartilage phenotype without evidence of hypertrophy by adulthuman articular chondrocytes in vitro. J Orthop Res 16:207216.

22. Bonaventure J, Kadhom N, Cohen-Solal L, Ng KH, Bour-guignon J, Lasselin C, Freisinger P 1994 Reexpression ofcartilage-specific genes by dedifferentiated human articularchondrocytes cultured in alginate beads. Exp Cell Res 212:97104.

23. Hauselmann HJ, Fernandes RJ, Mok SS, Schnid TM, BlockJA, Aydelotte MB, Kuettner KE, Thonar EJ-MA 1994 Pheno-typic stability of bovine articular chondrocytes after long-termculture in alginate beads. J Cell Science107:1727.

24. Nixon AJ, Lust G, Vernier-Singer M 1992 Isolation, propaga-tion, and cryopreservation of equine articular chondrocytes.Am J Vet Res53:23642370.

25. Preece A 1972 A Manual for Histologic Technicians, 3rd Ed.Little, Brown, and Co, Boston, MA, U.S.A.

26. Chomczynski P, Sacchi N 1987 Single-step method of RNAisolation by acid guanidinium thiocyanate-phenol-chloroformextraction. Anal Biochem162:156159.

27. Sambrook J, Fritsch E F, Maniatis T 1989 Molecular Cloning:A Laboratory Manual, 2nd Ed. Cold Spring Harbor Laboratory,Cold Spring Harbor, NY, U.S.A.

28. Richardson DW, Dodge GR 1997 Cloning of equine type IIprocollagen and the modulation of its expression in culturedequine articular chondrocytes. Matrix Biol16:5964.

29. MacLeod JN, Fubini SL, Gu DN, Tetrault JW, Todhunter RJ1998 Effect of synovitis and corticosteroids on transcription ofcartilage matrix proteins. Am J Vet Res59:10211026.

30. Campbell BG, Wooten JA, MacLeod JN, Minor RR 1998Sequence of canine COL1A2 cDNA: Nucleotide substitutionsaffecting the cyanogen bromide peptide map of the alpha 2(I)chain.Arch Biochem Biophys357:6775.

31. Kuivaniemi H, Tromp G, Chu ML, Prockop DJ 1988 Structureof a full-length cDNA clone for the prepro alpha 2(I) chain ofhuman type I procollagen: Comparison with the chicken geneconfirms unusual patterns of gene conservation. Biochem J

252:633640.32. Seghatoleslami MR, Lichtler AC, Upholt WB, Kosher RA,

Clark SH, Mack K, Rowe DW 1994 Differential regulation ofCOL2A1 expression in developing and mature chondrocytes.Matrix Biol14:753764.

33. Fedewa MM, Oegema TR Jr, Schwartz MH, MacLeod A,Lewis JL 1998 Chondrocytes in culture produce a mechani-cally functional tissue. J Orthop Res 16:227236.

34. Ballock RT, Reddi AH 1994 Thyroxine is the serum factor thatregulates morphogenesis of columnar cartilage from isolatedchondrocytes in chemically defined media. J Cell Biol 126:13111318.

35. Reginato AM, Iozzo RV, Jimenez SA 1994 Formation ofnodular structures resembling mature articular cartilage inlong-term primary cultures of human fetal epiphyseal chondro-cytes on a hydrogel substrate. Arthritis Rheum37:13381349.

36. Tavella S, Raffo P, Tacchetti C, Cancedda R, Castagnola P 1994

N-CAM andN-Cadherin expression during in vitro chondrogen-esis. Exp Cell Res215:354362.

37. Chang J, Poole CA 1997 Confocal analysis of the molecularheterogeneity in the pericellular microenvironment produced

by adult canine chondrocytes cultured in agarose gels. Histo-chem J29:515528.

38. Lee GM, Poole CA, Kelley SS, Chang J, Caterson B 1997Isolated chondrons: A viable alternative for studies of chondro-cyte metabolism in vitro. Osteoarthritis Cartilage 5:261274.

39. Poole CA 1997 Articular cartilage chondrons: Form, functionand failure. J Anat191:113.

40. Hascall VC, Handley CJ, McQuillan DJ, Hascall GK, Robin-son HC, Lowther DA 1983 The effect of serum on biosynthesis

of proteoglycans by bovine articular cartilage in culture. ArchBiochem Biophys224:206223.

41. Burton-Wurster N, Lust G 1990 Fibronectin and proteoglycansynthesis in long term cultures of cartilage explants in HamsF12 supplemented with insulin and calcium: Effects of additionof TGF-. Arch Biochem Biophys283:2733.

42. Plaas AHK, Sandy JD 1993 A cartilage explant system forstudies on aggrecan structure, biosynthesis and catabolism indiscrete zones of the mammalian growth plate. Matrix 13:135147.

43. Rayan V, Hardingham T 1994 The recovery of articularcartilage in explant culture from interleukin-1: Effects onproteoglycan synthesis and degradation. Matrix Biol14:263271.

44. Jortikka MO, Inkinen RI, Tammi MI, Parkkinen JJ, Haapala J,Kiviranti I, Helminen HJ, Lammi MJ 1997 Immobilizationcauses long lasting matrix changes both in the immobilized andcontralateral joint cartilage. Ann Rheum Dis 56:255261.

45. Muller FJ, Setton LA, Manicourt DH, Mow VC, Howell DS,Pita JC 1994 Centrifugal and biochemical comparison ofproteoglycan aggregates from articular cartilage in experimen-tal joint disuse and joint instability. J Orthop Res 12:498508.

46. Bianco P, Cancedda FD, Riminucci M, Cancedda R 1998 Boneformation via cartilage models: The borderline chondrocyte.Matrix Biol17:185192.

47. Gerstenfeld LC, Shapiro FD 1996 Expression of bone-specificgenes by hypertrophic chondrocytes: Implications of the com-plex functions of the hypertrophic chondrocyte during endo-chondral bone development. J Cell Biochem62:19.

48. Scammell BE, Roach HI 1996 A new role for the chondrocytein fracture repair: Endochondral ossification includes directbone formation by former chondrocytes. J Bone Miner Res11:737745.

49. Lee K, Deeds JD, Segre GV 1995 Expression of parathyroidhormone-related peptide and its receptor messenger ribo-nucleic acids during fetal development in rats. Endocrinology136:453463.

50. Stewart MC, Ballock RT, MacLeod JN 1998 Thyroid hormoneand BMP-2 induce hypertrophic differentiation and up-regulatep21cip1/waf1 expression in growth plate but not articular chondro-cytes. Trans Orthop Res Soc23:58 (abstract).

Address reprint requests to:

Matthew Stewart

Department of Orthopaedics

Case Western Reserve University

11100 Euclid Ave

Cleveland, OH 44106

Received in original form March 5, 1999; in revised form May 28,1999; accepted July 7, 1999.

174 STEWART ET AL.

phenotypic stability of articular chondrocytes in vitro

Documents