266 Btochimtca et Btoplo'stca Aeta.

1051 I 1990l 266 275

I~lsevicr

BBAMCR 12642

Production of collagens, collagenase and collagenase inhibitor

during the dedifferentiation of articular chondrocytes
by serial subcultures
V6ronique Lefebvre, Chantal Peeters-Joris and Gilbert Vaes
Laboratoire de Chimie Physiologique (Connective Tissue Group), Uni~ersit~ de Louvain and lnternattonal Institute of Cellular
and Molecular Pathology, Brussels (Belgium)

(Received 26 June 1989)

Key words: Chondrocyte dedifferentiation; Collagen synthesis: Collagenase; lnterleukin l; Tissue inhibitor of metalloproteinase:
(Rabbit cell)

Rabbit articular chondrocytes were cultured in monolayer and the progressive loss of their differentiated phenotype was
monitored from passage to passage. The cell densities achieved in confluent cultures decreased abruptly between the
primoculture and the second or third subculture, and more slowly thereafter, reflecting parallel morphological changes.
The synthesis of collagen (but not that of other proteins) decreased sharply, and a smaller proportion of collagen was
incorporated into the matrix. Cells in primocuiture synthesized mainly the cartilage-specific collagens, types II and XI,
which were mostly deposited in the matrix, but no type I nor |II collagen. With increasing passages, the synthesis of
type II collagen decreased progressively while that of types I and III collagens increased, the latter being almost
completely released in the culture medium. Simultaneously, the production of type XI collagen was apparently switched
to that of type V. Fully differentiated confluent chondrocytes in primoculture produced the collagenase inhibitor TIMP
(tissue inhibitor of metalioproteinases) but no detectable procoilagenase; their production of procollagenase was,
however, induced by interleukin 1. The production of TIMP increased from passage to passage. A spontaneous
production of procollagenase was only occasionally observed in confluent cultures of dedifferentiated chondrocytes.
However, interleukin 1 induced an always higher production of procollagenase from dedifferentiated chondrocytes than
from cells in primocuiture.

Introduction phenotype modulation (often called dedifferentiation)

Although normally responsible for synthesizing the
matrix of articular cartilage, chondrocytes can become
activated to produce enzymes which degrade the col-
lagens and proteoglycans of this tissue (reviewed in Ref.
1). This activation is thought to play a critical role in
the joint destruction that occurs in arthritis. Therefore,
unravelling the conditions leading to or accompanying
such activation is of importance.
In osteoarthritic cartilage, an increased production of
collagenase has been reported [2] as well as the ap-
pearance of type I collagen, a sign of chondrocyte
Abbreviations: APMA, 4-aminophenylmercuric acetate; DMEM,
Dulbecco's modified Eagle's medium; FCS, foetal calf serum; IL-1,
interleukin 1; TIMP, tissue inhibitor of metalloproteinases.
Correspondence: G. Vaes, Laboratoire de Chimie Physiologique, UCL
75.39, Avenue Hippocrate 75, B-1200 Bruxelles, Belgium.
towards a fibroblastic phenotype [3,4]. Contrary to
non-activated chondrocytes [5], fibroblasts are a par-
ticularly good source of collagenase and related tissue
metalloproteinases, such as proteoglycanase (stromely-
sin) [6]. This suggests that the production of collagenase
by chondrocytes could be linked to their dedifferentia-
tion.
To investigate this hypothesis, we cultured rabbit
articular chondrocytes in monolayer for several pas-
sages, allowing the progressive loss of their differenti-
ated phenotype [7]. This was monitored by fluoro-
graphic visualization of the collagen type chains pro-
duced by the cells after their electrophoretic separation,
and by quantitative evaluation of the collagen and
non-collagenous proteins produced. In parallel, both the
spontaneous and IL-l-induced [8] chondrocyte produc-
tions of procollagenase were evaluated. Because of its
potentially important regulatory role on the activity of
collagenase [9], the production of the collagenase inhibi-
tor T I M P was also recorded. New assay procedures

0167-4889/90/$03.50 ,~5 1990 Elsevier Science Publishers B.V. (Biomedical Division)


were used, established [10] to improve the recovery of
TIMP and the validity of the enzymatic assay of procol-
lagenase in crude culture media which contain both
procollagenase and TIMP. Our study provides an over-
view of both quantitative and qualitative changes occur-
ring in collagen production during the chondrocyte
dedifferentiation. It shows that these changes are
accompanied by enhanced production of TIMP and by
increased IL-l-induced production of procollagenase.
Materials and Methods
Materials
Recombinant human IL-la (specific activity 108
U/mg) was a gift from Dr. P.T. Lomedico (Hoffmann-
LaRoche, Nutley, N J, U.S.A.). Mouse bone-conditioned
culture medium containing latent tissue procollagenase
was kindly donated by J.M. Delaiss6 from our labora-
tory (see Ref. 10). DAPI (4',6-diamidino-2-phenylin-
dole) was from Serva Feinbiochemica (Heidelberg,
F.R.G.); L-[-2,3-3H]proline and 14C-methylated pro-
teins, from Amersham Belgium (Brussels); 4-amino-
propionitrile fumarate, from Janssen Chimica (Beerse,
Belgium); calf thymus DNA, N-ethylmaleimide and
purified bacterial collagenase (type VII), from Sigma
Chemical, MO, U.S.A.); pepsin, from Boehringer,
Mannheim (F.R.G.); z-ascorbic acid and ammonium
hydroxide, from Merck (Darmstadt, F.R.G.); X-ray
films, from Fuji Photo Film Company (Japan). Other
special chemicals, culture plates and culture media were
from suppliers previously mentioned [10].
Articular chondrocyte cultures
Cartilage pieces from rabbit humeri, femurs, patellae
and tibiae were freed from adhering other tissues by
successive digestions at 37 °C with hyaluronidase (0.05%
(w/v) for 10 min) and trypsin (0.25% (w/v) for 30 min)
in phosphate-buffered saline. The pieces were then
digested overnight with crude bacterial collagenase
(0.15% (w/v) in DMEM supplemented with 10%, v/v,
FCS). The dissociated chondrocytes were washed and
cultured following procedures described for synovial
fibroblasts [11], except that the trypsin solution used to
disperse the cells at each passage was supplemented
with 0.05% (w/v) bacterial collagenase. The chondro-
cytes were plated at (0.8-1.0)- 105 cells per cm2 culture
dish and cultured in DMEM supplemented with 10%
FCS; the medium was renewed once or twice a week.
Subcultures were done at 7-day intervals and at a split
ratio of 1 : 2 in order to allow the progressive dediffer-
entiation of the chondrocytes [7].
Cell densities were determined at the end of each
experiment by assaying the DNA content of the cul-
tures with the DAPI fluorimetric method [12]. Calf
thymus DNA (50 pg/ml) was used as a standard and it
was assumed that 1 0 6 cells contain 6 #g of DNA.
267
Synthesis of collagen and non-collagenous proteins
The proteins synthesized by confluent chondrocytes
at successive passages were labelled for 24 h with L-[2,3-
H3]proline (15 /~Ci/ml) in fresh DMEM supplemented
with 10% FCS, 4-aminopropionitrile fumarate (100
/xg/ml) and ascorbic acid (100 /~g/ml). The culture
medium was then removed and centrifuged to sediment
the floating cells. The matrix and cell layer was dis-
solved in 0.25 M NH4OH and combined with the
floating cells.
The production of collagen and non-collagenous pro-
teins was quantified according to Peterkofsky et al. [13].
After extensive dialysis against water, the samples were
incubated for 90 min at 37°C in a 50 mM Tris-HC1
buffer (pH 7.5) containing 150 mM NaCI, 5 mM CaC12,
0.2 mg of NaN3/ml, 0.05% (v/v) Triton X-100 and 3
mM N-ethyl-maleimide, and with or without 20 /zg of
purified bacterial collagenase/ml. Undigested proteins
were precipitated by 10% (w/v) cold trichloroacetic
acid in the presence of 1 mg of bovine serum
albumin/ml. After centrifugation (35000 g-min) the
pellets were dissolved in 0.2 M NaOH, and the amount
of radioactivity was determined in the pellets and in the
supernatants. Collagenase sensitive and unsensitive pro-
teins were considered as collagen and non-collagenous
proteins respectively.
The types of collagen were determined from samples
treated with pepsin (0.2 mg/ml in 0.5 M acetic acid for
3 h at 15°C, followed by extensive dialysis against
water). To facilitate the identification of the collagens
or of their subtypes, aliquots of the samples were fur-
ther digested either by purified bacterial collagenase (as
described above), or by tissue collagenase (5 U / m l for 4
h at 25°C, in the same buffer as that used for the
bacterial collagenase digestion). Types V [14] and XI
collagens [15] are indeed resistant to the action of tissue
collagenase while types I, II and III [16], as well as type
I trimer [17], are cleaved into characteristic 3/4-1/4
fragments. Procollagenase-containing mouse bone-con-
ditioned culture medium was used as a source of tissue
collagenase, after the activation of procollagenase by
trypsin [18]. Collagenase-digested and non-digested
aliquots were concentrated 25-fold by dialysis and
lyophilization, resuspended in electrophoresis buffer
with or without dithiothreitol, and denatured at 100 °C
for 2 min. Sodium dodecyl sulfate-polyacrylamide gel
electrophoreses [19] were performed on 5-8% (w/v)
polyacrylamide linear gradient gels. Fluorographs [20]
were exposed for 2 or 8 days at - 8 0 ° C . Type I
guinea-pig skin collagen, partially digested by animal
collagenase, and 14C-methylated proteins were used re-
spectively as collagen and protein standards. Individual
collagen bands resolved from the samples were quanti-
tated by scanning densitometry of the fluorographs.
268
Production of collagenase and collagenase inhibitor
(TIMP)
Collagenase and TIMP were evaluated as described
[10] in conditioned culture media obtained (unless
otherwise indicated) at cell confluence, in DMEM with
or without 10% FCS, 10% NU serum or IL-1, as speci-
fied in the figure legends. Briefly, free TIMP was as-
sayed by measuring the inhibition of the trypsin-
activated collagenase present in a crude skin fibroblast-
conditioned medium. This assay underestimates the total
amount of TIMP present because it does not consider
the proportion of T I M P that binds to other metal-
loproteinases present in the fibroblast medium. There-
fore, all parallel assays were done using the same
fibroblast medium. TIMP bound to metalloproteinases
was measured after destruction of the enzymes and
dissociation of the complexes at pH 2 and 100°C for 30
min. Prior to the activation of procollagenase by either
trypsin or APMA, the free TIMP present in the condi-
tioned media was neutralized whenever needed (see the
figure legends) by the binding of other tissue metal-
loproteinases (i.e., the proteoglycanase and gelatinase
present in the culture medium of rabbit bone marrow-
derived macrophages [21]). Collagenase was then as-
sayed at 2 5 ° C using soluble [3H]acetylated guinea-pig
skin collagen as a substrate. The procollagenase values
presented have been corrected for the stimulation ( × 1.5)
exerted on collagenase activity by the macrophage-con-
ditioned medium [10]. 1 U of collagenase degrades 1 ~g
of collagen per min at 25 ° C and 1 U of TIMP inhibits
2 U of collagenase by 50%.
Cultures of chondrocytes on collagen/proteoglycan-coated
plates
The capability of chondrocytes, at various passages,
to degrade type I or type II collagen and cartilage
proteoglycans was assayed by culturing the cells on the
surface of [14C]collagen/[3H]proteoglycan films [11].
Type II collagen was prepared from calf articular carti-
lage [16] and 14C-acetylated [22].
Results
Morphological aspects and cell densities of confluent
cultures
Observed at low magnification (not shown),
chondrocytes in primoculture appear as small polygonal
cells surrounded by an abundant matrix. In some areas,
the cells pile up in small multilayer 'cartilaginous nod-
ules'. This morphology changes progressively with in-
creasing passages. The abundance of extracellular ma-
trix decreases and confluent cells appear larger and
flattened, although still polygonal in shape, and they no
longer pile up. In agreement with this evolution, the cell
densities achieved in confluent cultures decrease from
passage to passage. An abrupt diminution, approx. 50%
mE
o
20,
1.6
1.2
6
2 5
08
~o 0.4
0.0 5
1 I I I I I
0 2 4 6 8 10
passage number
Fig. l. Evolution of the cell densities achieved by confluent chondro-
cytes during their dedifferentiation by serial monolayer subcultures.
Six different chondrocyte cell lines (referred to as lines l to VI in Figs.
2, 3, 5 and 6) were cultured on plastic for 6 to 10 passages to allow'
their progressive dedifferentiation. The cellular densities achieved
after 9 to 10 days of culture at various passages were evaluated by
measuring the DNA content of the cultures; the results, expressed in
10 5 cells/cm 2, are the means of evaluations done on 1 to 6 of the
different cell lines, as indicated by the corresponding figures on the
graph; vertical bars correspond to + 1 S.D.
loss, is observed between the primoculture and either
the second or third culture passage, depending upon the
experiment. This is followed by a slower decrease there-
after (Fig. 1).
Relative production of collagen and non-collagenous pro-
teins
The chondrocyte synthesis of collagen and non-col-
lagenous proteins was characterized during the progres-
sive dedifferentiation of several cell lines by serial
monolayer subcultures. The results of a representative
experiment, obtained with cell line I (see Figs. 1, 3, 5
and 6), are illustrated in Fig. 2 for the relative produc-
tion of collagen and non-collagenous proteins, and in
Fig. 3, for the collagen types produced.
Collagen is actively synthesized in primoculture and
first subculture where it is mainly incorporated in the
matrix and cell layer. A sharp reduction of collagen
synthesis occurs in subsequent passages, along with a
decreased incorporation into the matrix and cell layers
and a slightly increased release into the media (Fig. 2A
and B). Inversely, the production of non-collagenous
proteins increases steadily from the primoculture to the
10th passage. Most of these proteins are located in the
matrix and cell layers but with increasing passages the
amounts released into the media increase more, relative
to those incorporated in the matrix and cell layers (Fig.
2A and B). Consequently, the ratio of newly synthesized
collagen to newly synthesized total proteins decreases
sharply during the first two passages and more slowly
thereafter (Fig. 2C). The ratio of newly synthesized
collagen incorporated into the matrix and cell layers
relative to the total of newly synthesized collagen de-
creases progressively with increasing cell passages (Fig.
2D).
 269

A B

0.8 0.4 _

0.6 0.3

0, 02
0.2 0.1 0
o.o [- -~- o.o
I I l I I Ic I I l I I Dl

~6 - 80 ~ .

..j
.~
Z
+ 8
--, OOio
40
0
0 ~ 4 20
0 0

0 2 4 6 8 10 0 2 4 6 8 10
passage number passage number

Fig. 2. Production of collagen and non-collagenous proteins by dedifferentiating chondrocytes in monolayer culture. The chondrocyte cell line I
(see Figs. 1, 3, 5 and 6) was characterized for its synthesis of collagen (m) and non-collagenous proteins (©) during its progressive dedifferentiation
by serial monolayer subcultures. A and B present the accumulation of these newly synthesized proteins in either the matrix and cell layer (A) or the
culture medium (B) as a function of the cell passage number. Results are expressed in cpm incorporated into protein per cell during the labelling
period with [3H]proline (see Materials and Methods). Each point is the mean of three determinations; vertical bars correspond to _+1 S.D. (shown
only if its magnitude exceeds the size of the symbol used for the experimental points). C shows the progressive reduction in the percent ratio of
newly synthesized collagen to newly synthesized total proteins with increasing passage number. D shows the parallel reduction in the percentage of
newly synthesized total collagen deposited in the matrix and cell layer, The former ratio only (C) was calculated on cpm values corrected according
to Peterkofsky et al. [13] in order to include in the collagen determinations the collagenase-resistant propeptides of procollagens otherwise included
in the non-collagenous proteins and to take account of the higher proline and hydroxyproline content of collagen relative to non-collagenous
proteins. (COLL, collagen; NCP, non-collagenous proteins).

Collagen types produced chains [lower band, migrating together with the al(II)
The matrix and cell layers as well as the culture chains] of type XI collagen. This is another product of
media of the chondrocyte cell line presented above (Fig. differentiated chondrocytes existing predominantly un-
2) were monitored from the primoculture to the seventh der a heterotrimer [al(XI).a2(XI).a3(XI)] form [23].
subculture to determine the types of collagen synthe-
sized (Fig. 3) and their relative proportions (Table I). TABLE I
In primoculture, the most abundant neosynthesized
Relative proportions of collagen types synthesized by chondrocytes during
collagen found in the medium (Fig. 3A) and, in consid- their progressive dedifferentiation in monolayer culture
erably larger amounts, in the matrix and cell layer (Fig.
The table was established by compiling the data obtained from
3B), has a migration similar to that of the al(I) chain densitometry of fluorographs presented in Fig. 3. Type III collagen
standard. No a2(I) chains are detected. This indicates was considered to correspond to band A of the culture medium, since
the absence of type I collagen with a chain composition the contribution of types I and II collagens to this band, visualized
of [al(I)]2.a2(I), but the presence of s o m e [al(I)] 3 can- after dithiothreitol reduction (not shown), was found to be minimal.
All other collagen types were evaluated from both the medium and
not be excluded. Thus, the major collagen synthesized
the cell and matrix compartment: type I collagen [(al)2.a2 ] was
appears to be type II, a typical product of differentiated estimated equal to 3-fold the values of bands F; type XI, to 3-fold the
chondrocytes, consisting of three al(II) chains which values of bands D; type V, to 1.5-fold the difference between bands C
are transformed by a crude preparation of tissue col- and D; type II (and type 1 trimer), to the difference between bands E
lagenase (Fig. 3C and D) into fragments of the size of and the values estimated for the participation of types I, V and XI
collagens to these bands.
the 3 / 4 ~1 and 3 / 4 a2 standards or smaller (discussed
below). This material is followed by two bands, visible Cell Relative proportions (%) of collagen type
almost exclusively in the matrix layer, which migrate passage
I II III V XI
between the 3 / 4 fl(I) and the al(I) standards. These number
t w o b a n d s p e r s i s t a f t e r d i g e s t i o n w i t h c r u d e t i s s u e col-
0 0 59.2 0 10.9 29.9
l a g e n a s e as d o e s a m i n o r c o m p o n e n t o f t h e b a n d 1 0 62.1 0.5 11.2 26.2
m i g r a t i n g a t t h e a l ( I ) level. T h e s e t i s s u e c o l l a g e n a s e - r e - 2 19.1 41.8 7.3 19.8 12.0
sistant b a n d s a p p e a r to be c o l l a g e n o u s by their c o m - 3 23.2 36.0 9.7 31.1 0
plete removal by bacterial collagenase (not shown). 5 48.9 10.0 5.1 36.0 0
7 57.0 0.7 9.2 33.1 0
T h e y c o n t a i n m a i n l y t h e a l ( u p p e r ) , c~2 ( c e n t r e ) a n d a 3
 270

A 0 1 2 3 5 7 gs B '~ o- 0 1 2 3 5 7 gs

~ ~ A
%r
34/"

200 ,5

C
D
~1 E
I )2 F
4111 G
34(~ 92.5 341~2
H
~iiiiiiiiiiii!i~~ -

69

46

cs 1 2 3 5 gs
C r~ cs 0* 0 1 2 3 5 gs
I'
~,,~' I"

/; 200

(t2
34,h 92.5
34(t2

99

Fig. 3. Analysis of the collagen types synthesized by chondrocytes during their dedifferentiation in monolayer culture. The chondrocyte cell line l
(see Figs. 1, 2, 5 and 6) was characterized at several passages for the collagen types it produced. The newly synthesized proteins labelled with
[3H]proline and accumulated in the culture medium (A and C) or in the cell and matrix layer (B and D) were submitted to a limited pepsin
digestion following which aliquots were treated with crude tissue collagenase. Fluorographs of the denatured proteins were prepared after their
separation by electrophoresis. Identical sample volumes were deposited in each well without correction for cell density variation with increasing celt
passages. Type I collagen partially digested by animal collagenase was used as a collagen standard (well cs): "~, [~, M and c~2 symbols correspond
respectively to trimers, dimers and monomers of the typical M(I) and ~2(l)chains: their 3 / 4 equivalents are the 3 / 4 fragments generated by animal
collagenase. ]4C-Methylated proteins were used as globular protein standards (gs); their M r- 10 3 are indicated on the right. The pepsin-resistant
(A D) and animal collagenase-resistant (C and D) collagens and peptides are fractionated in the wells 0 *, 0. 1, 2, 3, 5 and 7, indicating cell
passage number. Fluorographs of 0 * wells were exposed for 2 days and those of all other samples for 8 days. They display different bands (bands
A M) characteristic of the following peptides or collagen chains: A='((I,II,II1); B=/~(I,II); C = M ( V , X I ) ; D = ~ 2 ( X I ) ; E = c d ( l . [ I ) , c~2(V),
c~3(XI); F = c~2(I); G, H and I = respectively 108, 83 and 63 kDa peptides: J = ~2(g), c~3(XI); K = 3 / 4 M(I,II,III), [1/4 cd(lll)]3; L = 3 / 4 ~2(ll
and fragments from M(II); M = fragments from al(II). The M, of the peptides migrating in the G. ft and 1 bands were estimated by comparison
with the globular protein standards. Band H disappeared after reduction by dithiothreitol (not shown).

The presence of some newly synthesized cd(V) and
c~2(V) chains, comigrating respectively with the M(XI)
and the ~3(XI) chains is possible, because the ratio of
M(XI) to ~2(XI) or c~3(XI) to ~2(XI) chains is greater
than 1:1. The absence of reducible "},-components in
the gels indicates that type III collagen is not synthe-
sized by the cells in primoculture. Identical collagen
bands were observed in gels run under reducing condi-
tions (not shown).
With increasing cell passages, the pattern of collagen
types synthesized changes progressively. Chondrocytes
in the first passage still produce types II and Xl (possi-
bly also type V) collagens but no type I. They have
initiated synthesis of type Ili collagen, visible almost
exclusively in the media as y-components transformed
by dithiothreitol reduction (not shown) into al compo-
nents comigrating with the M(I) chains. With further
passages, the synthesis of type III collagen increases
 271

progressively as well as that of type I collagen, demon-
strated by the presence of ct2(I) chains in both matrix
layers and media. Conversely, the synthesis of type II
collagen decreases progressively. Types I and III col-
lagens are digested by a crude mouse bone collagenase
preparation but, in contrast to type II collagen, they
produce only typical 3 / 4 ct fragments. Interestingly, the
3 / 4 etl fragments of (pepsinized) type II collagen are
further degraded into smaller pieces by either the col-
lagenase or, more likely, by contaminating proteinases
such as proteoglycanase (stromelysin) or gelatinase,
known to be present in the crude collagenase prepara-
tion [24]. The disappearance of these smaller pieces
after the fifth passage provides evidence for the virtual
switching off of type II collagen synthesis after this
passage. A progressive reduction in type XI collagen
synthesis occurs in parallel, shown by the disappearance
of the a2(XI) chains after the fifth passage as well. The
simultaneous disappearance of al and ct3 chains of type
XI collagen is likely but presumably masked by a
progressive increase in the synthesis of type V collagen.
Several unidentified peptides, resistant to bacterial col-
lagenase (not shown) as well as to pepsin, are produced
by chondrocytes after the second passage in parallel
with their dedifferentiation (Fig. 3, bands G, H and I).
It should be noted that, from one experiment to
another, the dedifferentiation of chondrocytes may be
initiated at slightly different passages. We frequently
observed chondrocyte populations which were half-
dedifferentiated in the first passage and we occasionally
saw chondrocyte cultures that remained essentially dif-
ferentiated in the second passage.
Production of active collagenase and collagenase inhibitor
Free active collagenase was not detectable in the
medium from non-stimulated chondrocytes (not shown).
The fact that no active collagenase (nor proteo-
glycanase) was produced by these cultures, was ascer-
tained by culturing chondrocytes from different pas-
sages on 14C-labelled type II or type I collagen/
[3H]proteoglycan-coated plates. Regardless of passage
number, the cells did not degrade either substrate (Fig.
4A), although they all did after stimulation with IL-1
(shown for first-passage cells in Fig. 4B). In a unique
case (not shown), we observed degradation of type II
collagen by non-stimulated chondrocytes in the 9th
passage, but not by chondrocytes in the 2nd passage,
when the cells were cultured in the same experiment on
type II collagen-coated plates.
The absence of active collagenase in the medium can
be related to the presence of the inhibitor TIMP pro-
duced by the chondrocytes [10]. This production was
always higher in cultures with serum, either FCS or NU
serum, than in its absence (compare cell lines I and III
to the others in Fig. 5). Regardless of the culture media,
the production of TIMP increased steadily from passage
to passage (Fig. 5I-VI). The inhibitor was almost com-
pletely in the free form as its amount did not increase

~ ~oo 100
"6 (~ 80 80
60
== 8 ,o 40
~ 2o 20
,,, 0 0
"r
I I I l l I I I I I l l I I
o
z lOO - -
- 100
a
Lu z 80 80
GO uJ
< (-9 60 60
"'
rr" ~0 40 40
in"" o 20
~lt~ .
~ ~-- , ~- . 20
s t.lr-- ~lr I_lr
m m
0
~0 I I I I I I l I I I l I l l
0 2 4 6 8 10 12 0 2 4 6 8 10 12

TIME IN CULTURE ( days )

Fig. 4. Degradation of collagen and proteoglycans by chondrocytes at different passages and after IL-1 stimulation. 50000 chondrocytes were
cultured for the time indicated in 2 ml of serum-free medium [11] on []4C] (type II) collagen/[3H]proteoglycan - (experiment A) or [14C] (type I)
collagen/[3H]proteoglycan-coated plates (experiment B). Substrate degradations were evaluated by the release of soluble 14C- or 3H-labelled
materials from the plates and expressed as a percentage of the total amount of substrate initially present in the wells. Each point is the mean of
three cultures; vertical bars correspond to ___S.D. and are shown only if their magnitude exceeds the size of the symbol used for the experimental
points. Arrows show the maximal release of 14C-soluble products achieved by trypsin (50 t.tg/well). In experiment A, chondrocytes were cultured
without stimulant at passages 1 (o), 2 (zx), 4 (tz) and 8 (=); in experiment B, first-passage chondrocytes were stimulated with 0 (m), 0.3 (~), 3 (O) or 30
(A) U of IL-1/ml.
 272

12 - II 1.2
10 10 _~
8 08 ~
Go
~ 6 0.6 ~
4 0.4 E
2 - 0.2
- 0.0
I I I I I I . I I I I I I

5 - III 10 - 1.0
4 ,0 - 0.8
GO 3 - 06 '-°o
• .0
2 -04
E E
- 02
- 00
I I 1 I 1 I . I I I l I I

1,0 - IJI -05

/
0.8 -- 04

~°o 0.6 -- 0.3 Go

0.4 --02
E E
D 0.2 .-01
z3
0.0 - 0.0
I I I I I I I I I I I I
0 2 4 6 8 10 0 2 4 6 8 10
passage number passage number

Fig. 5. Production of procollagenase and TIMP by chondrocytes during their dedifferentiation in monolayer culture. The 6 chondrocyte cell lines
(Fig. 1) were considered independently for their procollagenase and collagenase inhibitor production (I VI). In 1, conditioned media were produced
from the 2nd to the 9th day of culture in DMEM with 10% FCS; before assays, they were freezed and thawed several times in order to denature
their a2-macroglobulin content. In II to VI, conditioned media were produced at or near confluence for 2 days in DMEM without (full lines) or
with (dotted lines) 10% NU serum (a serum devoid of active collagenase inhibitors). Free TIMP was measured in all media without other treatment
(o) or after optimal prometalloproteinase activation ( o ) with APMA alone (IV to VI) or APMA and trypsin (I). In II to VI, procollagenase (~) was
measured after neutralization of the free inhibitor present by binding to active tissue metalloproteinases added to the media prior to activation of
the proenzyme with trypsin. Results are expressed as a function of cell passage number, in I, in U / m l , as the cell numbers were increasing during
conditioned media production, and in II to VI, in U / m l per 106 cells, as the latter cell lines were studied at or near confluence.

10 III 5

/
8 4

GO 6 3 Go

4 2
E
2
0
1 I I I 1 1 I 1 I I I I

5 -- ~ I0 _ ? IJl 12 5
4 1O0

/
Go 3 0.75 G o
2 - 0.50
E
3 1 - 0.25 D
- 0.00
- I I I I I I t 1 I I I I I
0 2 4 6 8 10 0 2 4 6 8 10
passage number passage number

Fig. 6. Production of procollagenase by chondrocytes stimulated with IL-1 during their dedifferentiation in monolayer culture. In parallel with the
experiments reported for non-stimulated cells in Fig. 5, the chondrocyte cell lines I1, II1, IV and VI, cultured at or near confluence in DMEM, were
stimulated for 2 days with 100 U of lL-1/ml. Procollagenase was assayed in the conditioned media as done for the corresponding non-stimulated
cells. Results are expressed as a function of cell passage number in U / m l per 106 cells.

by heating the media at acid pH (not shown). The
absence of bound TIMP suggests again that only
minimal or no active collagenase ( a n d / o r other tissue
metalloproteinases) was formed in the cultures.
Spontaneous production of procollagenase and response to
IL-1
Treatment of media with APMA a n d / o r trypsin
(which activates the proenzymes of collagenase and
other tissue metalloproteinases under conditions where
it does not inactivate TIMP [10]) did not elicit measura-
ble collagenase activity in either differentiated primo-
cultures or dedifferentiated cultures from later passages.
The inactivation of free TIMP by complex formation
with an exogenous metalloproteinase, before activation
of procollagenase by trypsin, did not elicit collagenase
activity in media from differentiated primocultures
analyzed at or near confluence (Fig. 5II-VI). With
increasing passage numbers and cell dedifferentiation,
collagenase became detectable, after this treatment, in
only two (lines II and III) out of the five cell lines tested
at or near confluence. However, preliminary evidence
(not shown) suggests that low levels of collagenase can
be produced transiently by differentiated cells, and in
relatively larger quantities by dedifferentiated cells, be-
fore approaching confluence.
When collagenase was detected (as for cell lines II
and III of Fig. 5), its activity remained lower than that
of the free TIMP present in the same medium. In the
four other cell lines (Fig. 5I, IV, V and VI), treatment of
the media with APMA, followed or not by trypsin,
decreased the amount of free TIMP measurable in late
passage cultures but not in primocultures.
Production of procollagenase could be induced by
IL-1 in all chondrocyte cultures; it was optimal using an
IL-1 concentration of 100 U / m l regardless of the cell
passage number (not shown). However, the procol-
lagenase response of differentiated chondrocytes in
primoculture was always lower than that of dedifferenti-
ated cells in later passage cultures (Fig. 6). The
chondrocyte production of TIMP was unaffected by
IL-1 [25] and usually lower (in U / m l ) than that of
procollagenase.
Discussion
Serial subcultures allowed us to establish a parallel
between the loss of the differentiated phenotype of
rabbit articular chondrocytes, as expressed by their
production of collagens, and the modulation of their
production of collagenase inhibitor and procollagenase.
The differentiated phenotype expressed by chondro-
cytes in primoculture (Table I) was characterized by
abundant synthesis of types II and XI collagens [26].
Apparently some type V collagen was also produced, in
agreement with other reports [27,28]. Pepsin-resistant
273
fragments from type IX collagen [29] were not identifia-
ble in our fluorographs, unless sometimes after long
exposure (not shown). This absence was apparently not
due to the technical procedure used (unpublished data).
It is therefore likely that our chondrocytes did not
synthesize significant amounts of type IX collagen, at
least during the labelling period with [3H]proline. Dif-
ferentiated chondrocytes simultaneously produced other,
non-collagenous proteins among which the collagenase
inhibitor TIMP, but no or very low levels of procol-
lagenase. Thus, their metabolic behaviour is oriented
towards the synthesis and preservation of the cartilage
matrix collagen.
As expected [26,30,31], the subsequent monolayer
cultures of chondrocytes caused their progressive dedif-
ferentiation. This was manifested by changes in the
collagen types synthesized, as described previously by
others [4,7,26,27,30,31]. Furthermore, it involved a re-
duction in the ratio of newly synthesized collagen to
newly synthesized non-collagenous proteins. This reduc-
tion is explained by the progressive disappearance of
the abundant synthesis of types II and XI collagens and
its replacement by only low productions of types I, III
and V collagens. Moreover, the ratio of newly synthe-
sized collagen incorporated into the matrix to that
released into the culture medium decreased in parallel.
This decrease manifested principally the shifts from
types II and XI collagens, mainly deposited in the
matrix and cell layers, to types I and V collagens, more
soluble in the culture medium, and to type III collagen,
almost exclusively present in the medium. Nevertheless,
the proportions,of type II plus type I collagens and of
type XI plus type V collagens remained constant at all
passages, the former representing about 60% of the total
collagen produced and the latter about 30-40%. The
proportion of type III collagen never exceeded 10% of
the total collagen synthesized.
Changes in production of collagenase inhibitor and
procollagenase and shifts in collagen synthesis were
evaluated in the same chondrocyte cultures while they
progressively dedifferentiated throughout subcultures.
Articular chondrocytes have been previously shown to
produce TIMP [10,32,33] but they produce only very
low or undetectable levels of collagenase, unless
activated by cytokines (most notably, IL-1) or other
agents [1,5,8,25,33,34]. However, in most of these stud-
ies, the phenotype of the chondrocytes was not con-
trolled, although the shift from a differentiated to a
modulated phenotype was sometimes suspected [34].
Also, problems arising in enzymatic assays of col-
lagenase or TIMP [10], due to the possible simultaneous
presence of (pro)collagenase and inhibitor in the culture
medium, were generally not discussed.
Using modified assay procedures established to
improve the recovery of both TIMP and procollagenase
in medium where they co-exist [10], we have shown here
274
that fully differentiated, non-proliferating chondrocytes
in primoculture secrete some T I M P but no detectable
procollagenase. Interestingly, preliminary evidence sug-
gests that low levels of procollagenase are produced by
proliferating chondrocytes before they approach con-
fluence, but the significance of this observation remains
to be established. The production of procollagenase by
differentiated cells can, however, be induced by IL-1. In
parallel with their dedifferentiation, the chondrocytes
progressively increase their production of TIMP. Dedif-
ferentiated chondrocytes also respond to IL-1 by greater
production of procollagenase than differentiated cells,
suggesting an increased sensitivity to the cytokine. A
'spontaneous' (i.e., not induced by exogenous IL-1)
induction of procollagenase production was observed in
some cell lines after the first or second subculture and
this production increased with further dedifferentiation.
This could result from an autocrine stimulation of the
cells by IL-1, since this cytokine can be produced by
chondrocytes [35]. Indirect evidence was obtained sug-
gesting that, even in the absence of a detectable 'spon-
taneous' induction of procollagenase, the synthesis of
other latent tissue metalloproteinases can occur in
dedifferentiated chondrocyte cultures. Treatment of
their medium with A P M A followed (or not) by trypsin,
under conditions where these agents do not destroy
T I M P but activate the zymogens of tissue metallopro-
teinases [10], caused a reduction in the amount of free
TIMP. This effect, which was not observed with medium
from differentiated primocultures, is likely due to the
activation of some tissue metalloproteinases (such as
proteoglycanase or gelatinase) that upon activation form
irreversible complexes with free T I M P [10]. Due to the
excess of free T I M P present, it was, however, not possi-
ble to assay these enzymes in the medium after the
activation of their latent proenzymes, because the exog-
enous metalloproteinase preparation used to neutralize
T I M P prior to the assay of procollagenase is itself active
on proteoglycans and gelatin [21].
Taking evidences altogether, it appears that dedif-
ferentiated chondrocytes are more likely to switch to-
wards a 'catabolic' behaviour characterized by enhanced
production of procollagenase in response to IL-1, possi-
bly even by a spontaneous induction of zymogen pro-
duction. This could be relevant to the pathological
events occurring in osteoarthritic cartilage, where foci of
type I collagen-producing dedifferentiated chondrocytes
coexist [3,4] with collagenase production and matrix
degradation [2,36]. To degrade the cartilage matrix con-
stituents, these proenzymes need to be activated and the
level of active enzymes must be higher than that of the
surrounding TIMP. We did not find active collagenase
in the procollagenase-containing culture fluids from
non-stimulated chondrocytes, nor did we find evidence
for significant amount of T I M P bound to collagenase or
to other metalloproteinases. This indicates that the
zymogens were not significantly activated by the
chondrocytes in culture and that they did not 'auto-
activate' in the media (a fact confirmed in nonreported
experiments). Moreover, in non-stimulated subcultures.
the level of free collagenase inhibitor recovered was
always higher than that of procollagenase when the
latter was found. It is thus not surprising that, even
after full activation of the proenzyme, no collagenase
activity could be detected in the media unless we had
neutralized the inhibitor before the activation of procol-
lagenase. It is interesting that non-stimulated chondro-
cytes from various passages, cultured on [14C]collagen-
coated plates, did not degrade collagen, as would be
expected if T I M P was in excess over the potential
collagenase activity produced by the cells. Thus, the
balance between production of procollagenase, its rate
of activation into collagenase (a problem not investi-
gated in the present study) and the production of TIMP,
is obviously important for maintaining the collagen
matrix surrounding cells. We observed here that non-
stimulated dedifferentiated chondrocytes can sometimes
increase their production of procollagenase, although
they usually keep the balance in favour of preserving
the surrounding matrix by maintaining the production
of inhibitor at a higher level than that of procol-
lagenase. It was only through IL-I stimulation that
more proenzymes than T I M P were produced [25] and
that active metalloproteinases appeared and directly
degraded the [14C]collagen and [3H]proteoglycans
coated on the culture plates.
Acknowledgements
This work was supported by the Fund for Medical
Scientific Research (Belgium) and by the Belgian State-
Prime Minister's Office Science Policy Programming
(interuniversity attraction poles, grant No. 7bis and
concerted actions, grant No, 88/93-122). V.L. was a
Research Fellow of the Institut pour l'Encouragement
de la Recherche Scientifique dans l'Industrie et l'Agri-
culture ' I R S I A ' . We gratefully acknowledge the expert
technical assistance of F. Chenut and B. Dieudonn6 and
the skillful secretarial work of Y. Marchand. Our thanks
also go to Dr. K. Willard-Gallo for checking the
manuscript.
References
1 Jasin, H.E. (1983) in Advances in Inflammation Research (Weiss-
man, G., ed.), Vol. 5, pp. 87-106, Raven Press, New York.
2 Pelletier, J.P.. Martel-Pelletier, J., Howell, D.S., Ghandur-
Mnaymneh, L., Enis, J.E. and Woessner, J.F. (1983) Arthritis
Rheum. 26, 63 68.
3 Nimni, M. and Deshmukh, K. (1973) Science 181, 751 752.
4 Gay, S., Miiller, P.K.~ Lemmen. C., Remberger, K., Matzen, K.
and KiJhn, K. (1976) Klin. Wschr. 54, 969 976.

5 Deshmukh-Phadke, K., Lawrence, M. and Nanda, S. (1978) Bio-
chem. Biophys. Res. Commun. 85, 490-496.
6 Laub, R., Huybrechts-Godin, G., Peeters-Joris, Ch. and Vaes, G.
(1982) Biochim. Biophys. Acta 721,425-433.
7 Benya, P.D., PadiUa, S.R. and Nimni, M.E. (1977) Biochemistry
16, 865-872.
8 Gowen, M., Wood, D.D., Ihrie, E.J., Meats, J.E. and Russell,
R.G.G. (1984) Biochim. Biophys. Acta 797, 186-193.
9 Cawston, T.E., Murphy, G.E., Mercer, E., Galloway, W.A., Hazle-
man, B.L. and Reynolds, J.J. (1983) Biochem. J. 211, 313-318.
10 Lefebvre, V. and Vaes, G. (1989) Biochim. Biophys. Acta 992,
355-361.
11 Peeters-Joris, Ch., Emonds-Alt, X. and Vaes, G. (1981) Biochem.
J. 196, 95-104.
12 Brunk, C.F., Jones, K.C. and James, T.W. (1979) Anal. Biochem.
92, 497-500.
13 Peterkofsky, B., Chojkier, M. and Bateman, J. (1982) in lmmuno-
chemistry of the Extracellular Matrix (Furthmayr, H., ed.), Vol. II,
pp. 19-47, CRC Press, Boca Raton.
14 Liotta, L.A., Abe, S., Robey, P.G. and Martin, G.R. (1979) Proc.
Natl. Acad. Sci. USA 76, 2268-2272.
15 Eyre, D.R., Wu, J.J. and Woolley, D.E. (1984) Biochem. Biophys.
Res. Commun. 118, 724-729.
16 Miller, E.J., Harris, E.D, Jr., Chung, E., Finch, J.E,, Jr., Mc-
Croskey, P.A. and Butler, W.T. (1976) Biochemistry 15, 787-791.
17 Narayanan, A.S., Meyers, D.F., Page, R.C. and Welgus, H.G.
(1984) Collagen Rel. Res. 4, 289-296.
18 Vaes, G. (1972) Biochem. J. 126, 275-289.
19 Laemmli, U.K. (1970) Nature 227, 680-685.
20 Bonner, W.M. and Laskey, R.A. (1974) Eur. J. Biochem. 46,
83-88.
21 Hauser, P. and Vaes, G. (1978) Biochem. J. 172, 275-284.
275
22 Cawston, T.E. and Barrett, A.J. (1979) Anal. Biochem. 99, 340-345.
23 Morris, N.P. and Biichinger, H.P. (1987) J. Biol. Chem. 262,
11345-11350.
24 Vaes, G., Eeckhout, Y., Lenaers-Claeys, G., Fran~jois-Gillet, Ch.
and Druetz, J.E. (1978) Biochem. J. 172, 261-274.
25 Lefebvre, V., Peeters-Joris, Ch. and Vaes, G. (1990) Biochim.
Biophys. Acta, in press.
26 Mayne, R. and vonder Mark, K. (1983) in Cartilage (Hall, B.K.,
ed.), Vol. I, pp. 181-214, Academic Press, New York.
27 Benya, P.D., Padilla, S.R. and Nimni, M.E. (1978) Cell 15,
1313-1321.
28 Eyre, D.R., Wu, J.J. and Apone, S. (1987) J. Rheumatol. 14,
25-27.
29 van der Rest, M. and Mayne, R. (1987) in Structure and Function
of Collagen Types (Mayne, R. and Burgeson, R.E., eds.), pp.
195-221, Academic Press, New York.
30 Benya, P.D. and Brown, P.D. (1986) in Articular Cartilage Bio-
chemistry (Kuettner, K., Schleyerbarch, R. and Hascall, V.C.,
eds.), pp. 219-233, Raven Press, New York.
31 von der Mark, K. (1986) Rheumatology 10, 272-315.
32 Morales, T.I., Kuettner, K.E., Howell, D.S. and Woessner, J.F.
(1983) Biochim. Biophys. Acta 760, 221-229.
33 McGuire-Goldring, M.K.B., Murphy, G., Gowen, M., Meats, J.E.,
Ebsworth, N.M., Poll, C., Reynolds, J.J. and Russell, R.G.G.
(1983) Biochim. Biophys. Acta 763, 129-139.
34 Malemud, C.J., Norby, D.P., Sapolsky, A.I., Matsuta, K., Howell,
D.S. and Moskowitz, R.W. (1981) Biochim. Biophys. Acta 657,
517-529.
35 Ollivierre, F., Gubler, U., Towle, C.A., Laurencin, C. and Tread-
well, B.V. (1986) Biochem. Biophys. Res. Commun. 141,904-911.
36 Howell, D.S. (1986) Amer. J. Med. 80, 24-28.