www.elsevier.com/locate/matbio

Matrix Biology 23 (2004) 267–276

Endostatin/collagen XVIII—an inhibitor of angiogenesis—is expressed

in cartilage and fibrocartilage

Thomas Pufea,*, Wolf J. Petersenb, Nicolai Miosgec, Mary B. Goldringd,Rolf Mentleina, Deike J. Varogae, Bernhard N. Tillmanna

aDepartment of Anatomy, Christian-Albrechts-University, Olshausenstrasse 40, D-24098 Kiel, GermanybDepartment of Trauma, Hand and Reconstructive Surgery, University Hospital Munster, Germany

cDepartment of Anatomy Georg-August-University, Gottingen, GermanydBeth Israel Deaconess Medical Center, New England Baptist Bone and Joint Institute and Harvard Medical School,

Boston, MA 02115, USAeDepartment of Orthopaedic Surgery, University Hospital Schleswig-Holstein, Campus Kiel, Germany

Received 20 January 2004; received in revised form 8 June 2004; accepted 8 June 2004

Abstract

Aim of the study was to get a deeper insight in the mechanisms regulating avascularity of cartilaginious tissues. In the center of our interest

was the expression of the anti-angiogenic fragment of collagen XVIII and its potency to inhibit angiogenesis. We observed a strong endostatin/

collagen XVIII production in articular and fibrocartilage and an inhibitory potency concerning the VEGF-signalling pathway.

Introduction: Cartilaginous tissue is mainly avascular and shows a limited intrinsic capacity for healing. Aim of this study was to

investigate the expression of the antiangiogenic peptide endostatin/collagen XVIII in cartilage and fibrocartilage. Results: In fetal

epiphyseal cartilage of humans high endostatin/collagen XVIII levels could be detected by ELISA whereas significantly lower levels

were found in articular cartilage of adults. In the fibrocartilaginous tissue of the menisci, there was no significant difference in the

endostatin/collagen XVIII concentrations between samples of fetuses and adults. But in the menisci of adults, endostatin/collagen XVIII

concentrations were higher in the internal avascular two thirds of the meniscus whereas in the fetal menisci higher endostatin/collagen

XVIII concentrations were found in the external third. Endostatin/collagen XVIII immunostaining of rat articular cartilage shows that

endostatin/collagen XVIII downregulation starts soon after birth. In fetal cartilage and fibrocartilage of rats and humans, endostatin/

collagen XVIII could be immunostained in the extracellular matrix and in the pericellular matrix of endothelial cells, fibrochondrocytes

and chondrocytes. In adult cells, weak endostatin/collagen XVIII immunostaining was restricted to the pericellular matrix of

fibrochondrocytes and chondrocytes. The detection of endostatin/collagen XVIII could be verified by in situ hybridization. Chondrocytes

in vitro released measurable amounts of endostatin/collagen XVIII into culture supernatants. Stimulation of chondrocytes with EGF, as

an example of a growth factor, or dexamethasone had no influence on endostatin/collagen XVIII expression. Endostatin inhibited VEGF-

induced phosphorylation of MAPK in chondrocytes. Conclusions: The spatial and temporal expression of endostatin/collagen XVIII in

cartilaginous tissue and its potency regarding inactivation of VEGF signalling suggests that this antiangiogenic factor is important not

only for the development but also for the maintenance of avascular zones in cartilage and fibrocartilage. Experimental procedures: We

analyzed the spatial and temporal expression of endostatin/collagen XVIII—an endogenous angiogenesis inhibiting factor—in cartilage

and fibrocartilage of humans and rats by immunohistochemical and biochemical (ELISA) methods and by in situ hybridization. To

elucidate possible factors responsible for the induction or suppression of endostatin/collagen XVIII in cartilaginous tissues, chondrocytes

(cell line C28/I2) were exposed to EGF and dexamethason. To study the possible interaction of endostatin/collagen XVIII with

angiogenic factors, the immortalized human chondrocytes (C28/I2) have been incubated with VEGF and the phosphorylation of the

MAPK Erk 1/2 (extracellular-regulated kinases), a known signal transduction pathway for VEGF has been determined under the

influence of endostatin.

D 2004 Elsevier B.V./International Society of Matrix Biology. All rights reserved.

Keywords: Endostatin; Vascular endothelial growth factor (VEGF); Cartilage; Angiogenesis; Antiangiogenesis

* Corresponding author. Tel.: +49-431-880-3087; fax: +49-431-880-1557.

0945-053X/$ - see front matter D 2004 Elsevier B.V./International Society of Matrix Biology. All rights reserved.

doi:10.1016/j.matbio.2004.06.003

E-mail addresses: t.pufe@anat.uni-kiel.de (T. Pufe), wolfpetersen@hotmail.com (W.J. Petersen), nmiosge@gwdg.de (N. Miosge),

mgoldrin@bidmc.harvard.edu (M.B. Goldring), rment@anat.uni-kiel.de (R. Mentlein), d.varoga@anat.uni-kiel.de (D.J. Varoga), btillma@anat.uni-kiel.de

(B.N. Tillmann).

T. Pufe et al. / Matrix Biology 23 (2004) 267–276268

1. Introduction 1990), interferon-alpha (Angionello et al., 1995), thrombo-

Hyaline cartilage of adults is an avascular tissue with a

poor repair potential. In contrast, certain tunnels containing

blood vessels and known as cartilage canals are generally

found in the cartilaginous epiphyses of the fetal skeleton.

The significance of these vessels in cartilage nutrition and in

initiation of centers of ossification has still to be accurately

defined and the mechanism controlling the development and

regression of these structures are largely unknown (Brookes

and Revell, 1998).

In contrast to articular cartilage, these canals are lacking

during the fetal development of fibrocartilaginous structures

such as in the menisci of the knee (Petersen and Tillmann,

1995). Nevertheless, microvascular studies have shown that

in the late fetal stage and at the time of birth the complete

knee joint meniscus is vascularized. This vascular network

regresses soon after birth and in the adult the vascular supply

of the meniscus is restricted to a small rim close to the joint

capsule (Petersen and Tillmann, 1995). This specific vascu-

lar supply has clinical relevance regarding the healing of

meniscus tears. Lesions in the vascularized external part

have a good repair response, whereas lesions in the avascular

inner two thirds frequently fail to heal (Petersen and Till-

mann, 1995). Since the vascular status of cartilage and

fibrocartilage is closely related to its repair abilities, an

understanding of the mechanism regulating the tissue-spe-

cific vascularity might give further insights into the mecha-

nisms of healing and degeneration.

Angiogenesis—the formation of new blood vessels from

preexisting capillaries—is balanced controlled by a variety

of stimulating (angiogenic) and inhibiting (antiangiogenic)

peptides that act on invading endothelial and smooth muscle

cells (Mentlein and Held-Feindt, 2003). Endogenous inhi-

bition of angiogenesis is necessary for the development and

maturation of tissues which are largely avascular. This

might be caused either by expression of inhibitory factors

for vascular endothelial cells or by an intrinsic insufficiency

of cartilage cells to express stimulatory peptides (Kim et al.,

2000).

One of the most important angiogenetic factors is the

vascular endothelial growth factor (VEGF) (Ferrara, 1999).

VEGF has been identified in the zone of hypertrophic

cartilage in the growth plate (Gerber et al., 1999). VEGF

plays an integral role in endochondral ossification and for

the longitudinal growth of the skeleton. In adult articular

cartilage, VEGF is largely downregulated (Pufe et al.,

2001a) but in the disease state such as in osteoarthritis, this

angiogenic factor is reexpressed by chondrocytes. In the

knee joint menisci, high VEGF levels have been found

during the healing of meniscus lesions (Becker et al., in

press). These studies show that angiogenesis plays an

important role not only for physiological but also for

pathological events of the skeleton.

Several endogenous inhibitors of angiogenesis have been

identified. These include platelet factor 4 (Maione et al.,

spondin (Good et al., 1990), metastatin (Liu et al., 2001),

troponin-1 (Moses et al., 1999) or angiostatin (O’Reilly et

al., 1994), or endostatin (O’Reilly et al., 1997).

Endostatin is a 20-kDa antiangiogenic factor originally

identified from murine hemangioendothelioma cells and

seems to counteract many VEGF induced effects (O’Reilly

et al., 1997). Endostatin inhibits VEGF-induced endothelial

cell migration (Yamaguchi et al., 1999), VEGF-mediated

neovascularization (Takahashi et al., 2003) and VEGF-

induced vascular permeability (Takahashi et al., 2003).

Under physiologic conditions, endostatin is critical for the

regression of blood vessels in the vitreous of the eye (Fukai

et al., 2002). Despite some structural homologies of the

vitreous and cartilage, there are no information about endo-

statin/collagen XVIII expression in articular cartilage or the

menisci in the literature. However, in a recent study we

could identify endostatin/collagen XVIII in the fibrocarti-

lage of a ‘‘wrap around tendon’’ (retromalleolar region of

the posterior tibial tendon) (Pufe et al., 2003).

The aim of this study was to examine the spatial and

temporal expression of endostatin/collagen XVIII in carti-

lage and fibrocartilage, to elucidate possible factors respon-

sible for the induction or suppression of endostatin/collagen

XVIII in chondrocytes, and to study the possible interaction

of endostatin with a well-known signal transduction path-

way for VEGF (MAPK Erk 1/2 [extracellular-regulated

kinases]).

2. Results

2.1. Spatial and temporal expression of endostatin/collagen

XVIII in cartilage and fibrocartilage

Endostatin/collagen XVIII could be intensively immu-

nostained in fetal menisci and in epiphyseal cartilage (Fig. 1).

In the fetal meniscus, most cells had a fibroblastic

phenotype and type II collagen immunostaining was nega-

tive. Endostatin/collagen XVIII was present throughout the

basement membranes of vessels and in the pericellular

matrix of meniscus cells (Fig. 1). Sections that were immu-

nostained with antibodies against the angiogenic peptide

vascular endothelial growth factor (VEGF) showed that the

majority of cells in the fetal menisci express this peptide as

well. Factor VIII immunostaining reveled that the fetal

meniscus was nearly completely vascularized (not shown).

In the epiphysis of the distal femur-intensive endostatin/

collagen XVIII, immunostaining was observed not only in

the pericellular matrix of chondrocytes but also in the

extracellular matrix of the cartilage (Fig. 1a). Endostatin/

collagen XVIII immunostaining was negative in chondro-

cytes and in the extracellular matrix around cartilage

channels (Fig. 1a). The chondrocytes around the cartilage

channels showed a strong immunoreactivity for VEGF

(Fig. 1b). In the growth plate, endostatin/collagen XVIII

Fig. 1. Immunostaining of endostatin/collagen XVIII in human cartilage and menisci. Immunostaining of endostatin/collagen XVIII (a, c– f) and of VEGF (b)

in fetal epiphyseal (a) and adult (c) articular cartilage and in growth plate (e) and in fetal (b) and adult menisci (d). In adult cartilage (f), no immunoreactions

were detectable after adsorption of the endostatin/collagen XVIII antibody to recombinant endostatin. Nuclei in the sections were counterstained with Meyer’s

hemalum. Bar = 10 Am (c–e) or 100 Am (a, b, e); original magnification: 175-fol

Fig. 2. (A) Light microscopic in situ-hybridisation with the anti-sense probe

for endostatin of collagen type XVIII in human osteoarthritic cartilage. The

mRNA (asterisk) is found in the chondrocytes adjacent to the tidemark

(arrows). (B) The corresponding sense probe shows no staining, bar = 70 Am.

T. Pufe et al. / Matrix Biology 23 (2004) 267–276 269

could be immunostained in the proliferation zone but not

in the zone of hypertrophic cartilage (Fig. 1e). This zone

was strongly immunopositive for VEGF (not shown).

In the adult, immunolabeling for endostatin/collagen

XVIII was positive in the pericellular matrix of articular

chondrocytes (Fig. 1d) and meniscus fibrochondrocytes (not

shown). The extracellular matrix of cartilage and menisci

was endostatin/collagen XVIII negative.

2.2. Expression of collagen XVIII mRNA in human cartilage

Collagen-type XVIII mRNA could be detected in fetal,

normal (not shown) and osteoarthritic cartilage (Fig. 2A).

The mRNA (asterisk) is found in the chondrocytes adjacent

to the tidemark (arrows) (Fig. 2B). The corresponding sense

probe shows no staining.

2.3. Endostatin/collagen XVIII expression is age-related

in rat articular cartilage and menisci

To study endostatin/collagen XVIII concentrations dur-

ing childhood, a rat model was used. Endostatin/collagen

XVIII could be intensively immunostained in fetal menisci

and in epiphyseal cartilage (Fig. 3a and b). The expression

of endostatin/collagen XVIII decreased shortly after birth

(Fig. 3c,d,f,g). The immunostaining in the articular cartilage

of 1- and 2-day-old rats was markedly weaker than in the

fetal animal. In the articular cartilage and menisci of adult

animals, endostatin/collagen XVIII was only slightly detect-

able (Fig. 3h and i).

2.4. Endostatin/collagen XVIII concentrations in

cartilage and menisci

The endostatin/collagen XVIII concentrations in fetal and

adult menisci samples were homogenized and endostatin/

collagen XVIII was measured by a sensitive ELISA (Fig. 4).

The highest average endostatin/collagen XVIII concen-

trations were found in epiphyseal cartilage (mean: 964 pg/

d (a, b, e); 750-fold (c, d, e).

Fig. 3. (a– i) Immunostaining of endostatin/collagen XVIII in fetal (a and b), 1-day-old (c–e), 2-day-old (f, g) and 5-month-old rat joints (h, i). In articular

cartilage, immunoreactions were relatively high in fetal and less intense in normal adult cartilage and meniscus tissues. No immunoreaction was detectable after

adsorption of the endostatin/collagen XVIII antibody to recombinant endostatin (e). Nuclei in the sections were counterstained with Meyer’s hemalum.

Bar = 100 Am (a, c–h); Bar = 10 Am (b, i); original magnification: 45-fold (a, c); 90-fold (f, h); 175-fold (d, e, g); 350-fold (b); 560-fold (i).

T. Pufe et al. / Matrix Biology 23 (2004) 267–276270

mg fresh weight). In the articular cartilage of adults, endo-

statin/collagen XVIII concentrations were significantly low-

er ( p < 0.05), but average endostatin/collagen XVIII

concentrations of 228 pg/mg fresh weight could be still

measured. The lowest endostatin/collagen XVIII concentra-

tions were measured in cartilage samples from patients with

osteoarthritis (147 pg/mg fresh weight).

In the fetal menisci, the average endostatin/collagen

XVIII concentrations (mean: 569 pg/mg fresh weight) were

significantly lower than in fetal epiphyseal cartilage but still

higher than in adult articular cartilage. In the adult, there

was no significant change in the overall endostatin/collagen

XVIII concentrations of the menisci as compared to fetal

tissue. However in the adult, endostatin/collagen XVIII

concentrations increased significantly ( p < 0.05) in the in-

ternal two thirds (ITT) and decreased significantly in the

external third (ET) of the menisci ( p < 0.05) (Fig. 4).

These results could also be verified in Western blotting

experiments, further proving the specificity of the immuno-

reaction. One band of 20 kDa that corresponds to endostatin

protein was stained in menisci and in traction tendons

(Fig. 5).

2.5. Endostatin/collagen XVIII is expressed in cultured

chondrocytes (C28/I2) but is not affected by growth factors

To elucidate if endostatin/collagen XVIII is also ex-

pressed by chondrocytes under in vitro conditions, we cul-

tured immortalized chondrocytes and measured endostatin/

collagen XVIII concentrations by a sensitive ELISA.

Chondrocytes cultured under normal oxygen pressure

released measurable amounts of endostatin/collagen XVIII

into their culture supernatants. Application of EGF, as an

example of a growth factor, or dexamethasone, as example

Fig. 4. Endostatin/collagen XVIII concentration in menisci and articular cartilage. In menisci, fetal and adult endostatin/collagen XVIII concentrations are

comparably high, in particular the external third (EXT) in fetal menisci and in the avascular internal two thirds (INT) of adults. In articular cartilage, endostatin/

collagen XVIII concentrations are relatively high in fetal, but not in normal adult and osteoarthritic cartilage. MeansF standard deviations from n= 5

individuals each, fetal tissues were from 4- to 8-month-old fetuses.

T. Pufe et al. / Matrix Biology 23 (2004) 267–276 271

for a possible inhibiting steroid hormone, had no effect on

endostatin/collagen XVIII secretion (Fig. 6). The differences

of endostatin/collagen XVIII concentrations in the superna-

tant after stimulation with EGF and dexamethasone were to

the supernatant of unstimulated cells were statistically not

significant ( p>0.05).

In comparison to chondrocytes, cultured tenocytes re-

leased only low amounts of endostatin/collagen XVIII. The

difference of tenocytes to all three cultures of chondrocytes

was statistically significant ( p>0.05).

2.6. VEGF induces phosphorylation of MAPK in chondro-

cytes which can be inhibited by endostatin

To elucidate a potential paracrine effect of endostatin

on chondrocytes, we incubated immortalized human chon-

drocytes (C28/I2) with VEGF and determined the phos-

Fig. 5. Endostatin protein can be detected byWestern blot in traction tendons

(lanes 1) and in menisci (2–4) of the inner two thirds. Surgical samples were

boiled for 5 min in sample buffer for sodium dodecylsulphate-polyacryl-

amide gel electrophoresis (SDS-PAGE) in the presence of dithiothreitol,

proteins were separated by SDS-PAGE and transferred onto nitrocellulose

membranes that were stained with an endostatin/collagen XVIII antibody.

Chemiluminescence reaction detected one band of immunoreactive endo-

statin in traction tendons and menisci with molecular mass of 20 kDa. There

is no cross reactivity with collagen type XVIII (180 kDa).

phorylation of the MAPK Erk 1/2 (extracellular-regulated

kinases), a known signal transduction pathway for VEGF

under the influence of endostatin by using Western blot

technique. Indeed, stimulation with VEGF induced phos-

phorylation of Erk 1/2, in particular Erk 2 (Fig. 7).

Endostatin alone had no effect on basal levels whereas

co-incubation with endostatin blocked VEGF-induced Erk

1/2 phosphorylation. This experiment shows that endo-

statin and VEGF target chondrocytes, but have opposing

effects.

Fig. 6. Influence of growth factors and glucocorticoids (dexamethasone) on

endostatin/collagen XVIII secretion by cultured chondrocytes (C28/I2) and

tenocytes (control). Tenocytes and chondrocytes were incubated for 24

h under normoxic (20% O2) conditions without or with epidermal growth

factor (EGF, 10 ng/ml), dexamethasone (50 nM, added from a 1000-fold

concentrated stock solution in ethanol). Conditioned media were analysed

for endostatin/collagen XVIII concentration by ELISA. Values were related

DNA measured in the cell layers after the experiment (meanF standard

deviations from n= 3 cultures each). Basal endostatin/collagen XVIII

concentrations in the culture supernatants of chondrocytes were neither

increased nor decreased by stimulation with EGF or dexamethasone.

Tenocytes produced negligible amounts of endostatin/collagen XVIII.

Fig. 7. VEGF induces the phosphorylation of the MAPK Erk 1/2 which is

suppressed by co-incubation with endostatin. Human chondrocytes (C-28/

I2) were stimulated for 20 min with VEGF and endostatin (both at 10 ng/

ml) alone or with a combination, lysed and analysed for Tyr-204

phosphorylated Erk 1 (44 kDa) and Erk 2 (42 kDa) by Western blotting

using a specific antibody. Equal amounts of proteins were used in the

Western blot, as shown by incubating membranes with an antibody against

non-activated Erk2 (lower panel).

T. Pufe et al. / Matrix Biology 23 (2004) 267–276272

2.7. Endostatin decreases NO production in C-28/I2

chondrocytes

Since OA is associated with an increased synthesis of

NO, we measured the influence of endostatin on NO

production in the C-28/I2 cell line (Fig. 8). In fact, as

measured with the Griess reaction, endostatin reduces the

NO release into the culture medium after 24 h.

Fig. 8. Endostatin increases NO production of chondrocytes moderately.

C28/I2 cells were exposed to endostatin (10 ng/ml) for 24 h, and aliquots of

the conditioned medium assayed for nitrate/nitrite (NO) by the Griess

reaction; values are the mean and S.D. of n= 6 experiments; p= 0.01 versus

control.

3. Discussion

A mircosequence analysis of endostatin revealed its

identity to the C-terminal fragment of the NC1 domain

of collagen XVIII (O’Reilly et al., 1997). Collagen XVIII/

endostatin is expressed in the developing and postnatal

eye, along with collagens II and IX (Fukai et al., 2002).

Mice lacking collagen XVIII and its proteolytically derived

product endostatin show delayed regression of blood

vessels in the vitreous along the surface of the retina after

birth (Fukai et al., 2002). These results suggest that

collagen XVIII/endostatin is critical for normal blood

vessel formation of the eye and might be involved in the

development of other avascular tissues. In some aspects,

the structure of cartilage is similar to the vitreous, which

contains cartilage-specific forms of collagen II and IX

(Fukai et al., 2002).

This study has shown that endostatin/collagen XVIII an

endogenous inhibitor of angiogenesis is expressed in human

cartilage and fibrocartilage. In the articular cartilage of

adults, endostatin/collagen XVIII concentrations were sig-

nificantly higher than in the fetal epiphysis. Since material

of young human donors is rare, we studied the temporal

expression of endostatin/collagen XVIII during childhood in

a rat model. In the rat, downregulation of endostatin/

collagen XVIII occurred soon after birth. Since in this stage

the distal femur was already ossified with a well-developed

articular cartilage, this time point might correspond to the

end of the first decade in humans.

In contrast to adult articular cartilage, there was no

significant decrease in the endostatin/collagen XVIII con-

centrations of the menisci. From adult articular cartilage,

other antiangiogenic factors such as thrombospondin (Good

et al., 1990), metastatin (Liu et al., 2001), and troponin-1

(Moses et al., 1999) have been identified. The presence of

these molecules might explain why lower endostatin con-

centrations are necessary to maintain avascularity in artic-

ular cartilage compared to meniscus fibrocartilage. Another

explanation might be that the menisci are not completely

avascular. In the external part adjacent to the vascularized

joint capsule, there are some vessels in the meniscus tissue.

A strong physiologic inhibitory stimulus might be necessary

to prevent ingrowth of these vessels in the inner two thirds

of the menisci.

Avascularity is an important feature of adult articular

cartilage and fibrocartilage. The extracellular matrix (ECM)

of both tissues is functionally adapted to bear high com-

pressive and tensile forces. Ingrowing vascular elements

would impair the material properties of this tissue-specific

ECM, therefore a protection by angiogenesis inhibiting

molecules might be necessary to maintain the tissue homeo-

stasis of cartilage and fibrocartilage.

High endostatin/collagen XVIII levels in developing

cartilage and fibrocartilage might reflect the angiogenic

activity of fetal tissue where angiogenesis is balanced by

inhibiting and stimulatory peptides. This leads to the

question why angiogenesis inhibitors should be present

in tissues that are angiogenic. One possibility is that the

proteolytic activity that accompanies fetal growth, may

also mobilize circulating angiogenesis inhibitors from

precursor proteins that are not antiangiogenic them-

selves—a mechanism that has been postulated for tumor

angiogenesis (Mentlein and Held-Feindt, 2003). A second

possibility is that endostatin has a physiological function in

fetal development to inhibit vascular overgrowth which

T. Pufe et al. / Matrix Biology 23 (2004) 267–276 273

might be induced by high levels of angiogenic factors such

as VEGF. In both cartilage and fibrocartilage of fetuses,

we observed also strong immunostaining for VEGF. In

epiphyseal and growth plate cartilage, endostatin/collagen

XVIII and VEGF expression was spatially separated but

no such separation was found in the menisci. In the adult

articular cartilage and in the menisci, VEGF was down-

regulated but endostatin could still be measured.

The mechanism by which endostatin exerts its effects is

unknown, but several possibilities are emerging (Takahashi

et al., 2003). Three types of interactions have been sug-

gested by which endostatin may alter the behavior of

endothelial cells.

First, it may interact with specific endostatin receptors

located on endothelial cells, which would result in activation

of intracellular signalling pathways and altered gene expres-

sion. Cell surface glypicans are low-affinity endostatin

receptors (Karumanchi et al., 2001).

Second, endostatin may bind to extracellular proteins

and alter their ability to influence endothelial cell behavior.

For example, endostatin inhibits adhesion of endothelial

cells to collagen I via alpha(2)beta(1) integrin (Furumatsu

et al., 2002).

Third, endostatin may bind to other cell surface receptors

and thereby modulate their signaling. A recent study has

shown that endostatin blocks VEGF-mediated signalling via

direct interaction with the VEGF receptor 2 (KDR) (Kim et

al., 2002). Recently, we and others have identified the

VEGFR-2 on the surface of hypertrophic and osteoarthritic

chondrocytes (Petersen et al., 2002; Pufe et al., 2001a,b;

Pfander et al., 2001; Enomoto et al., 2003). In human

umbilical vein endothelial cells endostatin-inhibited VEGF

induced ERK and p38 MAPK activation which is a down-

stream event of the VEGFR-2/KDR signaling (Kim et al.,

2002). The results of the present study show that endostatin

interferes with the same signaling cascade in human chon-

drocytes. In tumor cells and endothelial cells, endostatin

inhibits activation of metalloproteinase (MMP) 1 and 2

(Kim et al., 2000). In cartilage and in C28/I-2 chondrocytes,

VEGF induces MMP-1 and -3 and -13 (Pufe et al., 2004a,b).

MMPs are known to be crucial for degrading extracellular

matrix components (Westermarck and Kahari, 1999). In

C28/I-2 chondrocytes, endostatin is able to reduce NO

release. This is remarkable concerning the inhibitory effect

of NO on matrix synthesis in articular cartilage (Studer et

al., 1999). Since endostatin/collagen XVIII is downregu-

lated in osteoarthritic cartilage, this molecule might play a

role for the pathogenesis of this degenerative disease. More

research is needed to elucidate possible effects of endostatin

on chondrocytes.

In situ hybridization and immunostaining experiments

using fetal and adult tissue samples demonstrated that

collagen XVIII, the precursor for endostatin, is ubiquitous-

ly located in basement membrane zones (Halfter et al.,

1998; Fukai et al., 2002; Miosge et al., 2003), its expres-

sion patterns almost identical to that of laminin. Interest-

ingly, typical integral components of basement membranes

such as laminin have been identified and immunolocalized

in cartilage (Durr et al., 1996) and in fibrocartilage (Hayes

et al., 2001).

Since formation of cartilaginous and fibrocartilaginous

tissue is a functional adaptation to compressive and shear-

ing forces (Pauwels, 1960; Altmann, 1964), it seemed

likely that the avascular nature of fibrocartilage may also

be influenced by mechanical stimuli. In a previous in vitro

study, we analyzed tendon fibroblast cultures which were

incubated with intermittent hydrostatic pressure to show

that mechanical factors are able to stimulate fibroblasts to

release factors that inhibit endothelial cell proliferation

(Pufe et al., 2003). Those experiments demonstrated clear-

ly that mechanical factors influence endothelial growth not

only in a direct way but also via humoral factors in a

paracrine manner. Increased endostatin/collagen XVIII

concentrations in the supernatant of fibroblasts cultured

under the influence of intermittent hydrostatic pressure

might explain this observation. In contrast to the influence

of mechanical factors, the expression of endostatin/colla-

gen XVIII is unaffected by inflammatory growth factors or

antiinflammatory glucocorticoids.

In conclusion, our study gives evidence that the differ-

ential spatial expression of endostatin expression plays an

important role in the organization of blood vessels in

cartilage. This peptide appears to be one of the antiangio-

genic factors regulating the vascular status of cartilage by

counteracting effects of VEGF. More research is needed to

elucidate the effects of endostatin on chondrocytes.

4. Experimental procedures

4.1. Human tissues

The knee joints of human fetuses (18, 23, 26, 32, and 34

weeks post gestation) were harvested during routine autop-

sy, the femora and the menisci were dissected, fixed in 4%

formalin and prepared for immunohistochemistry. The knee

of the contralateral side was obtained for the biochemical

analysis. The menisci and the epiphyseal cartilage were

immediately frozen in liquid nitrogen.

Articular cartilage and menisci from adults were obtained

from body donors (23–76 years). The material was divided;

one part was fixed in 4% formalin, the other part was frozen

in liquid nitrogen. Before freezing, the menisci were sharply

divided into two parts (the internal two thirds = avascular

zone, the external one third = vascular zone). The study was

approved by the ethical commission of the Medical Faculty

of the Christian-Albrechts-University of Kiel.

4.2. Rat model

To investigate endostatin/collagen XVIII expression in

cartilage during childhood, a rat model was used. Eighteen

T. Pufe et al. / Matrix Biology 23 (2004) 267–276274

fetal, postnatal (1 and 2 days old) and adult Wistar rats were

killed by CO2. Then the knee joints were harvested, fixed in

4% formaline and prepared for immunohistochemistry.

These procedures are in accordance with the National

Institute of Health guidelines for the use of laboratory

animals. A permission of the local government animal rights

protection authorities was not required.

4.3. Cell cultures

The immortalized human chondrocyte cell line, C-28/I2

(Goldring et al., 1994), was cultured in Dulbecco’s modified

Eagle’s medium (DMEM)/Ham’s F12 (1/1, v/v) containing

10% fetal calf serum (FCS) and passaged using trypsin-

EDTA solution at >95% confluency every 5 to 6 days.

These cells express type II collagen, aggrecan, and other

markers of the differentiated phenotype and have been used

to study the regulation of gene expression and signalling in

response to cytokines and other factors (Loeser et al., 2000;

Osaki et al., 2003; Parker et al., 2003; Tan et al., 2003; Grall

et al., 2003).

Culture and stimulation of rat tenocytes: Achilles tendons

were dissected from postnatal (2–5 days old) rats, cut into

small pieces of less than 0.5 mm, transferred into a small

volume (2 ml per 10 cm2 culture flasks; TPP, Switzerland)

of Dulbecco’s modified Eagle’s medium (DMEM) with

50% foetal calf serum (FCS), and left for 24 h at 37 jCwithout enzymatic digestion. Then 3 ml DMEM plus 50%

FCS were added, and tissue pieces left for another 24 h at 37

jC. During this time, tenocytes migrate out of the tissue and

adhere to the bottom of the culture dish. After removing the

medium with the remaining tissue pieces floating on top,

fresh DMEM plus 10% FCS was added, and the cells were

cultivated for 96 h at 37 jC. For subculture, cells were

detached by a short treatment (about 20 s) with trypsin

(0.1%; 1:250, Gibco, Paisley, UK). 106 cells were seeded

into fresh dishes, and cultivated for 24 h in DMEM plus

10% FCS.

4.4. Immunohistochemistry

For immunohistochemistry, the samples were embedded

in paraffin, sectioned in a longitudinal direction (5 Am),

irradiated at 750 W in a microwave oven with 3% hydrogen

peroxide in 0.01 M sodium citrate buffer, pH 6.0 (twice for

5 min), dewaxed, immunostained with anti-endostatin/col-

lagen XVIII (1:40 in Tris-buffered saline, 60 min; AB 1878

anti human endostatin/collagen XVIII polyclonal antibody,

Chemicon, USA or for rat 1:20 in Tris-buffered saline, code

number 1092, courtesy of Prof. Dr. Rupert Timpl), anti-

VEGF (1:40 in Tris-buffered saline, 60 min; sc7269 mouse

monoclonal IgG2a, Santa Cruz Biotechnology, CA, USA),

anti-collagen type II (1:10; CIIC1mouse monoclonal anti-

body; Developmental studies hybridoma bank) or for dem-

onstration of blood vessels by anti-factor VIII (1:200;

DAKO rabbit polyclonal antibody) followed by biotinylated

secondary antibodies and a peroxidase-labeled streptavi-

din–biotin staining technique; nuclei were counterstained

with hemalum.

4.5. In situ-hybridization

A 205-bp sac/hinc II fragment corresponding to the c-

terminal endostatin domain of collagen type XVIII was

used for hybridization (Sasaki et al., 1998). Non-radioac-

tively labeled sense and anti-sense RNAs were produced

in vitro with the digoxigenin (DIG) labeling kit (Roche

Diagnostics, Mannheim, Germany). Paraffin sections were

deparaffinized, rehydrated and pre-treated with proteinase

K, then incubated with the hybridization solution (50%

formamide, 5� SSC, 0.1 mg/Al yeast-tRNA) for 1 h at

37 jC. The RNA concentration for collagen type XVIII

was 50 ng digoxigenin labeled anti-sense probe in 50

Al hybridization solution per section. In control sections,

an equivalently labeled amount of sense-RNA was used

instead of anti-sense-RNA. Post-hybridization treatment

included washes with 1� SSC (2� 15 min at 50 jC),1� SSC (2� 15 min), 0.1� SSC (4� 15 min) both at

60 jC and finally PBS for 15 min at RT. Specimens

were subsequently incubated with a 1:60 dilution of anti-

DIG peroxidase-labeled antibody (Dakopats) in PBS for 1

h at RT.

4.6. Enzyme-linked immunosorbent assay (ELISA) and

Western blots

For ELISA, frozen tissue samples were crushed in an

achate mortar under liquid nitrogen, homogenized in 150

mM NaCl 20 mM Tris/HCl-buffer, pH 7.4; a soluble

fraction was obtained by centrifugation (48000� g, 60

min), and aliquots (100 Al) were analyzed by a sandwich

ELISA for endostatin/collagen XVIII (Chemicon, USA

Cyt274). Human recombinant endostatin (Chemicon,

USA) served as standard. For statistical analysis of the

ELISA results, the Dunnett’s test was used.

For Western blots, samples were reduced in the presence

of 10 mM dithiothreitol, proteins separated by sodium

dodecylsulphate-polyacrylamide gel electrophoresis (SDS-

PAGE; 10% gels), transferred onto nitrocellulose membranes

that were blocked and incubated with antibodies according to

standard techniques as described (Held-Feindt et al., 1999).

Signals were detected by chemoluminescence reaction

(ECL-Pus; Amersham-Pharmacia, Uppsala, Sweden).

4.7. Stimulation of chondrocytes

The immortalized human chondrocyte cell line, C-28/I2

(20,21) was used for in vitro examinations. 106 cells were

seeded into fresh dishes, and cultivated for 24 h in Dulbec-

co’s modified Eagle’s medium (DMEM/ Ham’s F12 (1:1))

plus 10% fetal calf serum (FCS). Then, the medium was

replaced by DMEM/ Ham’s F12 (1:1) (without FCS), and

T. Pufe et al. / Matrix Biology 23 (2004) 267–276 275

the cells exposed to different stimulators for 24 h. Condi-

tioned medium was withdrawn, and aliquots assayed for

endostatin/collagen XVIII content. The cells were washed

with phosphate-buffered saline, lysed, the DNA content was

measured fluorometrically with the CyQuant reagent (Mo-

lecular Probes, Eugene, OR, USA), and related to a standard

number of cells (counted with trypsinised cells (Pufe et al.,

2001b)).

Statistical significance was evaluated by the Dunnett’s

multiple comparisons test or the t-test.

4.8. Analysis of kinase phosphorylation

The C28/I2 chondrocytes were seeded in fresh dishes and

cultured for 48 h in DMEM/ HAM F12 (1:1) plus 10% FCS.

Then the medium was replaced with DMEM/ Ham’s F12

without FCS and the cells exposed to the stimulators for 1

h (VEGF and endostatin each 10ng/ml). For detection of

phosphorylated MAPK, cells were rinsed once in cold

phosphate-buffered saline (PBS), lysed with Triton-lysis-

buffer (50 mM Tris–HCl, pH 7.8, 100 mM NaCl, 2 mM

EDTA, 1% Triton X-100, 2 mM Na3VO4) and scraped off

by a rubber policeman. The lysates were mixed vigorously

(vortex mixer) and clarified in an Eppendorf centrifuge (15

min, 14,000� g, 4 jC). After protein determination from an

aliquot, samples with equal amounts were boiled in 50–200

Al SDS-PAGE sample buffer, separated by SDS-PAGE

(10%), transferred onto a polyvinylidene difluoride (PVDF)

membrane that was blocked with 5% BSA overnight or for

1 h. The blots were incubated in 1% BSA with anti-

phosphorylated extracellular-signal related kinases ERK 1/

2 (1:200, mouse monoclonal reacting with Tyr-204 phos-

phorylated ERK 1/2; Santa Cruz sc-7383, CA, USA),

followed (after washings) by horseradish peroxidase-labeled

anti-mouse or anti-rabbit IgG (1:30,000; DAKO, Glostrup,

Denmark) and visualized by enhanced chemiluminescence

(ECL system; Amersham). In some experiments, the blots

were stripped by washing with methanol (18 h 4 jC) and re-

probed with an antibody to ERK 2 (1:200; Santa Cruz sc-

1647).

4.9. Determination of nitrite/nitrate

Subconfluent C28/I2 chondrocytes (5� 105 cells) were

stimulated for 24 h in medium with 10% FCS, and nitrite /

nitrate oxidized from nitric oxide (NO) was measured

spectrophotometrically by the Griess reaction (Sigma-

Aldrich G-4410).

Acknowledgements

We wish to thank Inka Kronenbitter, Frank Lichte,

Marion Lorenzen, Miriam Lemmer, Sonja Seiter, Karin

Stengel and Regine Worm for their expert technical

assistance.

Part of this work was supported by a grant of the

Deutsche Forschungsgemeinschaft (DFG Pu 214/3-1 and Pe

873/2-1) and by a grant from the ‘‘Stiftung zur Forderung

der Medizinischen Forschung’’ of the Medical Faculty of

the University of Kiel. Dr. Goldring’s research is supported

by NIH grants AR-45378 and AG-22021.

References

Altmann, K., 1964. Zur kausalen Histiogenese des Knorpels. W. Roux’s

Theorie und experimentelle Wirklichkeit. Z. Anat. Entwickel. Ges. 37,

1–167.

Angionello, A.L., Sgadari, C., Taub, D.D., Liao, F., Farber, J.M., Miahesh-

wari, S., Kleinman, H.K., Reaman, G.H., Tosato, G., 1995. Human

recombinant interferone-inducible protein 10 is a potent inhibitor of

angiogenesis. J. Exp. Med. 182, 155–162.

Becker, R., Pufe, T., Kulow, S., Giessmann, N., Neumann, W., Mentlein,

R., Petersen, W., 2004. Expression of the vascular endothelial growth

factor (VEGF) during meniscus healing in a rabbit model. J. Bone Jt.

Surg., Br. (in press).

Brookes, M., Revell, W.J., 1998. Blood Supply of Bone. Springer, New

York.

Durr, J., Lammi, P., Goodman, S.L., Aigner, T., von der Mark, K., 1996.

Identification and immunolocalization of laminin in cartilage. Exp. Cell

Res. 222, 225–233.

Enomoto, H., Inoki, I., Komiya, K., Shiomi, T., Ikeda, E., Obata, K.,

Matsumoto, H., Toyama, Y., Okada, Y., 2003. Vascular endothelial

growth factor isoforms and their receptors are expressed in human

osteoarthritic cartilage. Am. J. Pathol. 162, 171–181.

Ferrara, N., 1999. Molecular and biological properties of vascular endo-

thelial growth factor. J. Mol. Med. 77, 527–543.

Fukai, N., Eklund, L., Marneros, A.G., Oh, S.P., Keene, D.R., Tamarkin,

L., Niemela, M., Ilves, M., Li, E., Pihlajaniemi, T., Olsen, B., 2002.

Lack of collagen XVII/endostatin results in eye abnormalities. EMBO J.

21, 1535–1544.

Furumatsu, T., Yamaguchi, N., Nishida, K., Kawai, A., Kunisada, T.,

Namba, M., Inoue, H., Ninomiya, Y., 2002. Endostatin inhibits adhe-

sion of endothelial cells to collagen I via alpha(2)beta(1) integrin, a

possible cause of prevention of chondrosarcoma growth. J. Biochem.

(Tokyo) 131, 619–626.

Gerber, H.P., Vu, T.H., Ryan, A.M., Kowalski, J., Werb, Z., Ferrara, N.,

1999. VEGF couples hypertrophic cartilage remodeling, ossification

and angiogenesis during endochondral bone formation. Nat. Med. 5,

623–628.

Goldring, M.B., Birkhead, J.R., Suen, L.-F., Yamin, R., Mizuno, S., Glo-

wacki, J., Arbiser, J.L., Apperley, J.F., 1994. Interleukin-1-h-modulated

gene expression in immortalized human chondrocytes. J. Clin. Invest.

94, 2307–2316.

Good, D.J., Polverini, P.J., Rastinejad, F., Le Beau, M.M., Lemons, R.S.,

Frazier, W.A., Bouck, N.P., 1990. A tumor suppressor inhibitor of

angiogenesis is immunologically and functionally indistinguishable

from a fragment of thrombospondin. Proc. Natl. Acad. Sci. U. S. A.

87, 6624–6628.

Grall, F., Gu, X., Tan, L., Cho, J.-Y., Inan, M.S., Pettit, A., Thamrongsak,

U., Choy, B.K., Manning, C., Akbarali, Y., Zerbini, L., Rudders, S.,

Goldring, S.R., Gravallese, E.M., Oettgen, P., Goldring, M.B., Lib-

ermann, T.A., 2003. Responses to the pro-inflammatory cytokines

interleukin-1 and tumor necrosis factor-a in cells derived from rheu-

matoid synovium and other joint tissues involve NF-nB-mediated

induction of the Ets transcription factor ESE-1. Arthritis Rheum.

48, 1249–1260.

Halfter, W., Dong, S., Schurer, B., Cole, G.J., 1998. Collagen XVIII is a

basement membrane heparan sulfate proteoglycan. J. Biol. Chem. 273,

25404–25412.

T. Pufe et al. / Matrix Biology 23 (2004) 267–276276

Hayes, A.J., Benjamin, M., Ralphs, J.R., 2001. Extracellular matrix in

development of the intervertebral disc. Matrix Biol. 20, 107–121.

Held-Feindt, J., Krisch, B., Mentlein, R., 1999. Molecular and functional

analysis of the somatostatin receptor subtype 2 (sst2) in human glioma

cells. Mol. Brain. Res. 64, 101–107.

Karumanchi, S.A., Jha, V., Ramchandran, R., Karihaloo, A., Tsiokas, L.,

Chan, B., Dhanabal, M., Hanai, J.I., Venkataraman, G., Shriver, Z.,

Keiser, N., Kalluri, R., Zeng, H., Mukhopadhyay, D., Chen, R.L., Land-

er, A.D., Hagihara, K., Yamaguchi, Y., Sasisekharan, R., Cantley, L.,

Sukhatme, V.P., 2001. Cell surface glypicans are low-affinity endostatin

receptors. Mol. Cell 7, 811–822.

Kim, Y.M., Jang, J.W., Lee, O.H., Yeon, J., Choi, E.Y., Kim, K.W., Lee,

S.T., Kwon, Y.G., 2000. Endostatin inhibits endothelial and tumor cel-

lular invasion by blocking the activation and catalytic activity of matrix

metalloproteinase. Cancer Res. 60, 5410–5413.

Kim, Y.M., Hwang, S., Kim, Y.M., Pyun, B.J., Kim, T.Y., Lee, S.T., Gho,

Y.S., Kwon, Y.G., 2002. Endostatin blocks vascular endothelial growth

factor-mediated signaling via direct interaction with KDR/Flk-1. J. Biol.

Chem. 277, 27872–27879.

Liu, N., Lapcevich, R.K., Underhill, C.B., Han, Z., Gao, F., Swart, G.,

Plum, S., Zhang, L., Green, S.J., 2001. Metastatin: a hyaluronan-bind-

ing complex from cartilage inhibits tumor growth. Cancer Res. 61,

1022–1028.

Loeser, R., Sadiev, S., Tan, L., Goldring, M.B., 2000. Integrin expression

by primary and immortalized human chondrocytes: evidence of a dif-

ferential role for a1h1 and a2h1 integrins mediating chondrocyte ad-

hesion to types II and VI collagen. Osteoarthr. Cartil. 8, 96–105.

Maione, T.E., Gray, G.S., Petro, J., Hunt, A.J., Donner, A.L., Bauer, S.I.,

Carson, H.F., Sharpe, R.J., 1990. Inhibition of angiogenesis by recom-

binant human platelet factor-4 and related peptides. Science 247, 77–79.

Mentlein, R., Held-Feindt, J., 2003. Angiogenesis factors in gliomas—a

new key to tumour therapy? Naturwissenschaften 90, 385–394.

Miosge, N., Simniok, T., Sprysch, P., Herken, R., 2003. The collagen type

XVIII endostatin domain is co-localized with perlecan in basement

membranes in vivo. J. Histochem. Cytochem. 51, 285–296.

Moses, M.A., Wiedeschain, D., Wu, I., Fernandez, C., Ghazizadh Lane,

W.S., Flynn, E., Sytkowski, A., Tao, T., Langer, R., 1999. Troponin I

is present in human cartilage and inhibits angiogenesis. Proc. Natl.

Acad. Sci. 96, 2645–2650.

O’Reilly, M.S., Holmgren, L., Shing, Y., Chen, C., Rosenthal, R.A., Moses,

M., Lane, W.S., Cao, Y., Sage, E.H., Folkman, J., 1994. Angiostatin: a

novel angiogenesis inhibitor that mediates the suppression of metastases

by a Lewis lung carcinoma. Cell 79, 315–328.

O’Reilly, M.S., Behm, T., Sing, Y., Fikai, N., Vasios, G., Lane, W.S.,

Flynn, E., Birkhead, J.R., Olsen, B.R., Folkman, J., 1997. Endostatin:

an endogenous inhibitor of angiogenesis and tumor growth. Cell 88,

277–285.

Osaki, M., Tan, L., Choy, B.K., Yoshida, Y., Auron, P.E., Cheah, K.S.E.,

Goldring, M.B., 2003. The TATA-containing core promoter of the type

II collagen gene (COL2A1) is the target of interferon-g-mediated inhi-

bition in human chondrocytes: requirement for Stat1a, Jak1, and Jak2.

Biochem. J. 369, 103–115.

Parker, W.L., Goldring, M.B., Philip, A., 2003. Endoglin is expressed on

human chondrocytes and forms a heteromeric complex with betaglycan

in a ligand and type II TGF-h receptor independent manner. J. Bone

Miner. Res. 18, 289–302.

Pauwels, F., 1960. Eine neue Theorie uber den Einfluß mechanischer Reize

auf die Differenzierung der Stutzgewebe. Zehnter Beitrag zur funktio-

nellen Anatomie und kausalen Morphologie des Stutzapparates. Z.

Anat. Entwicklungsgesch. 121, 478–515.

Petersen, W., Tillmann, B., 1995. Age-related blood and lymph supply of

the knee menisci. A cadaver study. Acta Orthop. Scand. 66, 308–312.

Petersen, W., Tsokos, M., Pufe, T., 2002. Expression of VEGF121 and

VEGF165 in hypertrophic chondrocytes of the human growth plate

and epiphyseal cartilage. J. Anat. 201, 153–157.

Pfander, D., Kortje, D., Zimmermann, R., Weseloh, G., Kirsch, T., Ges-

slein, M., Cramer, T., Swoboda, B., 2001. Vascular endothelial growth

factor in articular cartilage of healthy and osteoarthritic human knee

joints. Ann. Rheum. Dis. 60, 1070–1073.

Pufe, T., Petersen, W., Tillmann, B., Mentlein, R., 2001a. Expression of the

vascular endothelial growth factor in osteoarthritic cartilage. Arthritis

Rheum. 44, 1082–1088.

Pufe, T., Petersen, W., Tillmann, B., Mentlein, R., 2001b. The angiogenic

peptide vascular endothelial cell growth factor (VEGF) is expressed in

fetal and ruptured tendons. Virchows Arch. 439, 579–585.

Pufe, T., Petersen, W., Kurz, B., Tsokos, M., Tillmann, B., Mentlein,

R., 2003. Mechanical factors influence the expression of endosta-

tin—an inhibitor of angiogenesis—in tendons. J. Orthoptera Res. 21,

610–616.

Pufe, T., Lemke, A., Kurz, B., Petersen, W., Tillmann, B., Grodzinsky, A.J.,

Mentlein, R., 2004a. Mechanical overload induces VEGF in cartilage

discs via hypoxia-inducible factor. Am. J. Pathol. 164, 185–192.

Pufe, T., Harde, V., Petersen, W., Goldring, M.B.;, Tilmann, B., Mentlein,

R., 2004b. Vascular endothelial growth factor (VEGF) induces matrix

metalloproteinase expression in immortalized chondrocytes. J. Pathol.

202, 367–374.

Sasaki, T., Fukai, N., Mann, K., Gohring, W., Olsen, B.R., Timpl, R., 1998.

Structure, function and tissue forms of the C-terminal globular domain

of collagen XVIII containing the angiogenesis inhibitor endostatin.

EMBO J. 17, 4249–4256.

Studer, R., Jaffurs, D., Stefanovic-Racic, M., Robbins, P.D., Evans, C.H.,

1999. Nitric oxide in osteoarthritis. Osteoarthr. Cartil. 7, 377–379.

Takahashi, K., Saishin, Y., Saishin, Y., Silva, R.L., Oshima, Y., Oshima, S.,

Melia, M., Paszkiet, B., Zerby, D., Kadan, M.J., Liau, G., Kaleko, M.,

Connelly, S., Luo, T., Campochiaro, P.A., 2003. Intraocular expression

of endostatin reduces VEGF-induced retinal vascular permeability, neo-

vascularization, and retinal detachment. FASEB J. 17, 896–898.

Tan, L., Peng, H., Osaki, M., Choy, B.K., Auron, P.E., Sandell, L.J.,

Goldring, M.B., 2003. Egr-1 mediates transcriptional repression of

COL2A1 promoter activity by interleukin-1h. J. Biol. Chem. 278,

17688–17700.

Westermarck, J., Kahari, V.M., 1999. Regulation of matrix metalloprotei-

nase expression in tumor invasion. FASEB J. 13, 781–792.

Yamaguchi, N., Anaud-Apte, B., Lee, M., Sasaki, T., Fukai, N., Shapiro,

R., Que, I., Lowik, C., Timpl, R., Olsen, B.R., 1999. Endostatin inhibits

VEGF-induced endothelial cell migration and tumor growth indepen-

dently of zinc binding. EMBO J. 18, 4414–4423.