Matrix Biology 32 (2013) 443–451

Contents lists available at ScienceDirect

Matrix Biology

j ourna l homepage: www.e lsev ie r .com/ locate /matb io

Chondrocalcin is internalized by chondrocytes and triggers cartilagedestruction via an interleukin-1β-dependent pathway

Claudie Bantsimba-Malanda a,1,2, Justine Cottet a,2, Patrick Netter a, Dominique Dumas a,b, Didier Mainard a,Jacques Magdalou a, Jean-Baptiste Vincourt a,c,⁎a Molecular, Cellular, Therapeutic Engineering & Glycosyl Transferases (MolCelTeG), UMR 7365 CNRS-Université de Lorraine, Biopôle du Campus Biologie-Santé, Faculté de Médecine, 9,Avenue de la Forêt de Haye, BP 184, 54505 Vandoeuvre-lès-Nancy, Franceb Plate-forme d'Imagerie et de Biophysique Cellulaire et Tissulaire, FR 3209, Biopôle du Campus Biologie-Santé, Faculté de Médecine, 9, Avenue de la Forêt de Haye, BP 184,54505 Vandoeuvre-lès-Nancy, Francec Proteomics platform FR3209, Biopôle du Campus Biologie-Santé, Faculté de Médecine, 9, Avenue de la Forêt de Haye, BP 184, 54505 Vandoeuvre-lès-Nancy, France

Abbreviations: HRP, horseradish peroxidase; BSA,phenylarsine oxide; PFA, paraformaldehyde; IL-1β, Interleureceptor I; MMP13, Matrix metalloprotease-13; COL, collag⁎ Corresponding author at: UMR 7365 CNRS-UL, Biop

Faculté deMédecine, 9, Avenue de la Forêt deHaye, BP 184France. Tel.: +33 3 83 68 54 12; fax: +33 3 83 68 5409.

E-mail address: Jean-Baptiste.Vincourt@univ-lorrain1 Present address: INSERM U606, Université Paris

Centre Viggo Petersen, 2, rue Ambroise Paré, 75010 Par2 Both authors contributed equally.

0945-053X/$ – see front matter © 2013 Elsevier B.V. Alhttp://dx.doi.org/10.1016/j.matbio.2013.06.002

a b s t r a c t

a r t i c l e i n f o

Article history:Received 27 May 2013Received in revised form 28 June 2013Accepted 30 June 2013

Keywords:OsteoarthritisCollagenChondrocytesMetalloproteaseEndocytosisMMP13

Chondrocalcin is among the most highly synthesized polypeptides in cartilage. This protein is released fromits parent molecule, type II pro-collagen, after secretion by chondrocytes. A participation of extracellular,isolated chondrocalcin in mineralization was proposed more than 25 years ago, but never demonstrated.Here, exogenous chondrocalcin was found to trigger MMP13 secretion and cartilage destruction ex vivo inhuman cartilage explants and did so by modulating the expression of interleukin-1β in primary chondrocytecultures in vitro. Chondrocalcin was found internalized by chondrocytes. Uptake was found mediated by asingle 18-mer peptide of chondrocalcin, which does not exhibit homology to any known cell-penetratingpeptide. The isolated peptide, when artificially linked as a tetramer, inhibited gene expression regulationby chondrocalcin, suggesting a functional link between uptake and gene expression regulation. At the sametime, the tetrameric peptide potentiated chondrocalcin uptake by chondrocytes, suggesting a cooperative mecha-nismof entry. The corresponding peptide from type I pro-collagen supported identical cell-penetration, suggestingthat this property may be conserved among C-propeptides of fibrillar pro-collagens. Structural modeling localizedthis peptide to the tips of procollagen C-propeptide trimers. Our findings shed light on unexpected function andmechanism of action of these highly expressed proteins from vertebrates.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Chondrocalcin was first discovered as a calcium-binding proteinlocalizing very preferentially to calcifying zones of growing cartilageandwas named on the basis of its proposed participation in calcification(Poole et al., 1984). Although obvious correlations between de novocalcification and chondrocalcin amounts in growth plate chondrocytecultures were evidenced (Hinek et al., 1987; Hinek and Poole, 1988;Kujawa et al., 1989), a demonstration of the calcifying function ofchondrocalcin has been lacking since then. Chondrocalcin corresponds

bovine serum albumin; PAO,kin-1beta; IL1-RI, interleukin-1en; GAG, glycosaminoglycan.ôle du Campus Biologie-Santé,, 54505 Vandoeuvre-lès-Nancy,

e.fr (J.-B. Vincourt).Diderot, Hôpital Lariboisière,is, France.

l rights reserved.

to the C-terminal propeptide of type II procollagen (PC2CP), which, aspart of the procollagen, plays a major structural role in the initiation ofthe triple helical assembly of type II procollagen chains in the endoplas-mic reticulum of chondrocytes. Indeed, the assembly of procollageninitiates with the disulfide bound homotrimerization of its C-terminalpropeptide, which then propagates towards the N-terminal side.The C-terminal propeptide requires to be cleaved from its parent mole-cule, however, prior to collagen fibrillogenesis, whereby it becomeschondrocalcin. Again as part of type II procollagen, chondrocalcin isone of the most abundantly synthesized polypeptides in cartilage.Indeed, it was found as a major component of immature cartilage inbovine (Niyibizi et al., 1987). It has been extensively proposed as acandidate marker of cartilage metabolism in general and of articulardiseases in particular (Garnero, 2006; Conrozier et al., 2008). Dependingon pathological contexts and species, its concentration was found toreach several micrograms per milliliter in synovial fluids (de Grauw etal., 2009) and serum (Hosogane et al., 2012). Itwas never demonstratedas a useful marker to predict long term evolution of the articulation,however.

In a previous study, we found that chondrocalcin accumulated inbenign chondrogenic tumors at levels higher than in adult articular

444 C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

cartilage and we initiated its functional characterization (Vincourt etal., 2010). Chondrocalcin was found to induce the expression ofMatrix metalloprotease 13 (MMP13) in chondrocytes, suggesting that,coming back to the articular context, it may favor cartilage breakdown.Here, we investigated ex vivo this property of chondrocalcin and in vitroits mechanisms of action.

2. Results

2.1. Administration of chondrocalcin to human articular cartilage explantcultures triggers cartilage degradation

We have previously observed the in vitro induction of MMP13 byrecombinant chondrocalcin in primary chondrocyte cultures (Vincourtet al., 2010). In order to verify this property of chondrocalcin on carti-lage, we have performed ex vivo treatment of human cartilage explantswith chondrocalcin for 48 h and investigated the downstream conse-quences over cartilage metabolism (Fig. 1). Chondrocalcin significantlyincreased the release of MMP13 into the culture medium (Fig. 1A)when administered at 50 μg/ml, but not 5 μg/ml, in agreement withour previous findings in chondrocyte cultures (Vincourt et al., 2010).Concomitantly, chondrocalcin induced degradation of type II collagen(COL2, Fig. 1B). Chondrocalcin also increased GAG release (Fig. 1C),and at the same time, decreased sulphur incorporation into cartilageproteoglycans (Fig. 1D). Therefore, the increased GAG release in theculture medium (Fig. 1C) accounted for increased extracellular matrixdegradation. Altogether, these data indicated that chondrocalcin in-duced MMP13 expression and cartilage degradation.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

none BSA(50) (5) (50)

rele

ased

MM

P13

(n

g/g

cart

ilage

/ 24

hour

s)

protein added (µg/ml)

A

0

0.5

1

1.5

2

2.5

rele

ased

gly

cosa

min

ogly

cans

(µg

/ mg

cart

ilage

)

none BSA(50)

protein added (µg/ml)

C

***

**

Chondrocalcin

(5) (50)Chondrocalcin

Fig. 1. Chondrocalcin induces cartilage degradation and alters anabolism in cultured cartilage(A, B, C) increased degradation of cartilage components upon chondrocalcin treatment. CultuII collagen proteolytic cleavage product) ELISA or (C) glycosaminoglycan assay. (D) Decreasulphate incorporation. * indicates p b 0.1, ** indicates p b 0.05 and *** indicates p b 0.02 v

2.2. Dose-dependent down-regulation of anabolism and up-regulation ofosteoarthritis mediators at the gene level by chondrocalcin

In order to understand by which mechanisms chondrocalcin alterscartilage metabolism, we investigated its dose-dependent effects overgene expression in cultured chondrocytes (Fig. 2A–B). Chondrocalcininduced an over 10-fold increase of MMP13 mRNA levels (Fig. 2A), asfound earlier (Vincourt et al., 2010) and an even higher induction ofinterleukin-1β (IL-1β)mRNA. The induction ofMMP13 releasewas clear-ly observable bywestern blot analysis of culture supernatants, even at thelowest doses tested (Fig. 2B). Examination of chondrocalcin levels withinthe cell fraction revealed that it did associate to cells in a dose-dependentmanner (Fig. 2B). Of note, association of chondrocalcin to cells wasobserved even in the absence of exogenously added chondrocalcin, indi-cating that basal chondrocalcin released by chondrocytes behaved likeexogenous chondrocalcin in this respect.

2.3. Chondrocalcin acts via interleukin-1β signaling

Considering that IL-1β very potently stimulatesMMP13 (Borden etal., 1996), we hypothesized that chondrocalcin might affect cartilagemetabolism via IL-1β-dependent mechanisms. Chondrocytes werepre-treated with blocking anti-interleukin-1 receptor 1 (anti-IL1-RI)antibodies and then stimulated with chondrocalcin before geneexpression and protein analysis (Fig. 2C & D). MMP13 induction bychondrocalcin was almost abolished by the anti-IL1-RI, both at themRNA (Fig. 2C) and protein levels (Fig. 2D). It is worth noticing thatIL-1β gene expression induction by chondrocalcin was not inhibited

0

5

10

15

20

25

none BSA(50)

C2C

(ty

pe II

col

lage

n cl

eava

ge)

(ng/

mg

cart

ilage

/ 24

hour

s)

protein added (µg/ml)

0

5

10

15

20

25

30

35

S35

inco

rpor

atio

n (1

03 cp

m/ m

g ca

rtila

ge)

none BSA(50)

protein added (µg/ml)

B

D

***

*

(5) (50)Chondrocalcin

(5) (50)Chondrocalcin

explants. Explants were cultured in 96-well plates and treated with indicated proteins.re media were collected and processed for (A) anti-MMP13 ELISA, (B) anti-C2C (a typesed glycosaminoglycan synthesis upon chondrocalcin treatment as determined by 35Sersus controls (n = 5).

A

1

10

100

1000

IL-1βMMP13

mR

NA

nor

mal

ized

to c

ontr

ol

[chondrocalcin] (µg/ml)

B

GAPDH

chondrocalcin

MMP13

Cells

medium

0 0.5

2 5 20 50

chondrocalcin added[µg/ml]

C

0

5

10

15

20

25

30

anti-IL-1R chondrocalcin anti-IL-1R +chondrocalcin

MM

P13

mR

NA

rel

. to

cont

rol

D

GAPDH

chondrocalcin

MMP13

chondrocalcin

Cells

Medium

anti-IL1R- - + +- + - +

**

0

10

20

30

40

50

60

70

IL-1

B m

RN

A r

el. t

o co

ntro

l

MMP13

IL-1β

0.1 1 10 100

Fig. 2. Gene regulation in chondrocalcin-treated primary chondrocytes and its inhibition by anti-IL1-RI antibody. Primary chondrocytes were treated in 24-well plates with indi-cated concentrations of chondrocalcin for 24 h and processed for mRNA extraction or protein extraction of the culture supernatant and layer, respectively. (A) Gene expressionanalysis as measured by real-time PCR and expressed as relative to GAPDH and normalized to control cells. (B) Western blot analysis of the expression of MMP13 and incorporationof chondrocalcin. (C & D) Chondrocytes were pre-treated with 2 μg/ml anti-interleukin-1 receptor 1 blocking polyclonal antibody for 20 min and then stimulated with 50 μg/mlchondrocalcin for 24 h. (C) Gene expression analysis as measured by real-time PCR and expressed as relative to GAPDH and normalized to control cells; ** indicates p b 0.05.(D) Western blot analysis of the release of MMP13 and appearance of chondrocalcin in the cellular fraction.

445C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

by the anti-IL1-RI (Fig. 2D), nor was the appearance of chondrocalcinwithin the cellular fraction (Fig. 2D), indicating that chondrocalcindid not signal directly through the IL1-RI, but rather stimulated secre-tion of IL-1β, which in turn signaled through its receptor.

2.4. Chondrocalcin is specifically internalized by chondrocytes

In order to investigate the molecular basis of the appearanceof chondrocalcin to the cellular fraction, cells were seeded on cover-slips, grown as for chondrocalcin stimulation and the localization ofchondrocalcin was investigated by immunofluorescence microscopyusing an antibody directed against chondrocalcin, after or withoutcell permeabilization (Fig. 3A–D). Without permeabilization, in eithernon-treated (not shown), control-treated cells (treated with BSA), orchondrocalcin treated cells, the staining for chondrocalcin remainedat the background level (Fig. 3A & C). When permeabilized, control-treated chondrocytes exhibited a heterogeneous, extended, perinuclearstaining for chondrocalcin (Fig. 3B). As explained in SupplementalFig. S1, this staining corresponded to unprocessed type II procollagen.Chondrocalcin-treated chondrocytes also exhibited this staining(Fig. 3D), but additionally demonstrated punctuate, cell-associatedstaining. This finding indicated that exogenously added chondrocalcincould be immuno-detected in treated chondrocytes only if cells werepermeabilized. It did not localize to the nucleus (Fig. 3, SupplementalFigs. S3 & S4), the Golgi apparatus or the endoplasmic reticulum (datanot shown).

In order to verify the specificity of this phenomenon, we investigatedthe association of biotinylated chondrocalcin or biotinylated BSA tochondrocytes (Fig. 3E & F), using fluorescently tagged streptavidincoupled to fluorescence microscopy. BSA treated cells (Fig. 3E) did notexhibit significant staining compared to untreated cells (not shown),while chondrocalcin-treated cells (Fig. 3F) did exhibit a punctuate

staining as found in the immunofluorescence experiment (Fig. 3D). Fur-thermore, performing this experiment in chondrocyte culture mediumcontaining 10% FCS resulted in similar results, indicating that majorserum proteins did not compete for chondrocalcin association tochondrocytes. Finally, this experiment was also analyzed by confocalmicroscopy (Supplemental Fig. S2), revealing that the punctuate stain-ing for chondrocalcin was at least mostly intracellular. Altogether,these experiments demonstrated that chondrocalcin was specificallyinternalized by chondrocytes.

2.5. The internalization of chondrocalcin is inhibited by phenylarsine oxide

A direct ELISA-style quantification method was developed to mea-sure levels of internalized, biotinylated chondrocalcin. This assay allowedus to demonstrate that the incorporation of chondrocalcin was depen-dent on its extracellular concentration and did not saturate in the con-centration range achievable in our experiments (Fig. 4A). The half-lifeof internalized chondrocalcin was also evaluated and was found closeto 3 h (Fig. 4B), suggesting its rapid degradation. In an attempt to identifyby which type of mechanism chondrocalcin was internalized, we testedthe effects of various treatments over its entry upon a 4 hour exposure(Fig. 4C). Earlier studies fromus and others suggested that chondrocalcinmight interact with β1 integrins (Davies et al., 1997; Vincourt et al.,2010). Here, the addition of an anti-β1 integrin blocking antibody didnot significantly reduce chondrocalcin incorporation, indicating that β1integrins did not serve as an essential receptor for chondrocalcin entryinto chondrocytes. Phenylarsine oxide (PAO), a widely used general in-hibitor of endocytosis (Ivanov, 2008), did fully abrogate chondrocalcinuptake at 1 μM but not at 50 nM (Fig. 4C). It is worth mentioning thatPAO is often used at concentrations above 1 μM to inhibit endocytosiswithout major cellular damages on the short term (Sun et al., 2009;Kiyoshima et al., 2011). In contrast, Cytochalasin D, the inhibitor of

BSA chondrocalcin

no p

erm

eabi

lizat

ion

perm

eabi

lizat

ion

biotinylatedBSA

biotinylatedchondrocalcin

A

B

C

D

pro-collagen II

Chondrocalcin

FE

Fig. 3. Exogenously added chondrocalcin is internalized by chondrocytes. (A–D) anti-chondrocalcin immuno-staining antibody (red) of chondrocytes treated with either BSA orchondrocalcin. The heterogenous, intense, peri-nuclear staining corresponds to endogenous pro-collagen type II. (E & F) streptavidin staining (red) of chondrocytes treated with eitherbiotinylated BSA or biotinylated chondrocalcin (20 μg/ml each) for 4 h before fixation and permeabilization. Green: phalloidin counterstaining. Bars: 10 μm.

446 C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

actin polymerization, which has also been found to potently inhibitendocytosis (Gottlieb et al., 1993), did not very significantly affectchondrocalcin incorporation, even at doses which profoundly affectedcell morphology and survival on the long run. The relatively specificinhibition of chondrocalcin incorporation by PAO (as compared toCytochalasinD) suggested the involvement of particular cell constituentsin its cellular uptake.

2.6. A single, 18-mer peptide derived from the chondrocalcin sequencemimics its cellular uptake

In a previous study, we generated 19 peptides encompassing thewhole sequence of chondrocalcin (Vincourt et al., 2010). Here, thesepeptides were biotinylated and exposed at 100 μg/ml to chondrocytes

cultured on coverslips for 6 h. Chondrocytes were then fixed andlabeled with fluorescent streptavidin to screen peptides which labeledthe cells. Staining was observed with one peptide only, termed “D4”(Fig. 5A) in reference to its annotation in our previous study (Vincourtet al., 2010). The staining was punctuate and very comparable to thatobserved with chondrocalcin (not shown, but evidenced farther). In-deed, when co-treating chondrocytes with biotinylated peptide D4and recombinant chondrocalcin, both molecules largely co-localized(Fig. 5B–D), demonstrating that D4 was sufficient for the cellulartargeting of chondrocalcin. Of note, this peptide promoted poor cellattachment in our previous study (Vincourt et al., 2010) and is one ofthe least conserved regions between C-propeptides of fibrillar collagens(see the alignment in Fig. 5A and that in (Vincourt et al., 2010)). Despitethis poor conservation, the corresponding peptide in the C-propeptide

0

10

20

30

40

[biotinylated protein] (µg/ml)

A

(ng

biot

inyl

ated

pro

tein

/wel

l)

0

10

20

30

40

time (hours)

B

0

5

10

15

20

25

30

cont

rol

PAO, 1 µ

M

PAO, 50

nM

CD, 1 µ

M

PFA, 4%

(ng

Cho

ndro

calc

in/w

ell)

anti-

β1 in

tegr

in

(ng

Cho

ndro

calc

in/2

0 00

0 ce

lls)

C

ND ND* *

0 5 10 15 20 250 10 20 30 40 50 60

Fig. 4. Quantification of intracellular biotinylated chondrocalcin under various conditions. Quantification was performed by direct ELISA using dilutions of directly coated,biotinylated chondrocalcin as standards. (A) Dose-dependent incorporation of chondrocalcin (full line) or BSA (dotted line). (B) time-course of intra-cellular chondrocalcin degra-dation after a 4-hour extracellular pulse. (C) Effects of various treatments of chondrocytes over chondrocalcin incorporation during a 4 hour exposure. Chondrocyte fixation withparaformaldehyde prior to exposure to chondrocalcin was used as a positive control of incorporation inhibition. The blocking anti-β1 integrin antibody was MAB1959 and inhibitedMatrilin-3 signaling in parallel experiments. Bars: SD. * indicates p b 0.01 versus control (n = 5). ND indicates amounts lower than the standards. CD, Cytochalasin D; PAO,phenylarsine oxide; PFA, paraformaldehyde.

447C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

of procollagen Iα1 (shown in Fig. 5A) localized similarly (not shownbutvery alike Fig. 5B–D), suggesting that key residues for cellular uptakeare among the 9 amino acids conserved between both peptides(highlighted by the red line in Fig. 5A). Recently, the crystal structureof the C-terminal propeptide of type III procollagen, which is a veryclose homolog of chondrocalcin, was solved (Bourhis et al., 2012).

D

A

B

Fig. 5. A single peptide from the chondrocalcin sequence is internalized by chondrocytespeptide D4 to its counterparts in propeptides of other fibrillar procollagens. Amino acids idin at least 3 sequences are bowed in gray. The red line indicates amino acids identical betwco-localizes to exogenously added chondrocalcin. Chondrocytes were treated with 20 μg/mimmunostaining; (C) streptavidin staining; (D) overlay with DAPI staining. Bars: 10 μm. (Ewere colored blue, red and white, respectively and their residue positions 94–111, correspo

Examination of the localization of the peptide corresponding to D4within type III procollagen C-propeptide (Fig. 5E) revealed that thisregion of the protein is very exposed not only to the outside of the tri-mer but also at the tips of each monomer. Therefore, the interventionof peptide D4 in interactions to potential uptake receptors is relativelycredible.

C

D4PC1CP D4

E

in a pattern similar to full size chondrocalcin. (A) Amino acid sequence alignment ofentical in at least 3 sequences are bowed in black. Amino acids functionally conservedeen peptide D4 and its counterpart in procollagen Iα1. (B–D) biotinylated peptide D4l chondrocalcin and 100 μg/ml biotinylated peptide D4 for 4 h. (B) anti-chondrocalcin) Peptide D4 localizes to the tip of C-propeptide monomers. The three peptide chainsnding to peptide D4, were colored green.

448 C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

2.7. Multimerized peptide D4 potentiates cellular uptake but counteractssignaling of chondrocalcin

Since D4 internalized chondrocytes in a manner similar tochondrocalcin, its ability to trigger MMP13 gene expression was in-vestigated. However, none was evidenced at peptide concentrationsranging from 1 to 100 μg/ml (that is, up to a 50-fold molar excesscompared to highest chondrocalcin concentrations, shown farther).Therefore, cellular internalization via the chondrocalcin route isnot sufficient to modulate gene expression. Competitions betweenchondrocalcin and D4 were performed, but even at a 100-fold molarexcess, peptide D4 was not able to significantly alter chondrocalcinentry or signaling. Capacities of biotinylated chondrocalcin versusbiotinylated peptide D4 to undergo uptake were then compared usingthe ELISA-style incorporation assay mentioned above. Chondrocalcinwas significantly internalized at concentrations about 100-fold lowerthan peptide D4. While peptide D4 remains presumably under amonomeric state in solution, chondrocalcin is a disulfide-stabilizedhomotrimer, as earlier described (Niyibizi et al., 1987) and verifiedin the course of its purification (Fig. 6A & B). Therefore, a singlechondrocalcin trimer possesses 3 physically linked copies of peptideD4 (Figs. 5E & 6B). We hypothesized that artificially combiningseveral copies of peptide D4 may profoundly increase their efficiencyof incorporation. Since peptide D4 was biotinylated, its syntheticmultimerization was performed by co-incubation with fluorescently orHRP-labeled streptavidin at equivalent monomer concentrations. Thismethod has been used successfully in other contexts (Ramachandiranet al., 2007). Since streptavidin arranges as a tetramer and eachmonomerbinds biotin with inequivalent affinity, this simple procedure results in

A

250

150

100

80

50

37

25

+ DTT-DTT

10

Chondrocalcin ~ 100 kDa

streptavidin / D4 complex ~ 60 kDa

(D4S4)

B

GAPDH

Chondrocalcin

Cells

medium

ChondrocacinStreptavidin

D4D4S4

substance added

+ + +

+ ++

+ +

D

ChonSt

E

MW(kDa)

MMP13

Fig. 6. Excess tetrameric peptide D4 favors chondrocalcin uptake but inhibits its signaling.chondrocalcin. 4 μg purified chondrocalcin was loaded under non-reducing or reducing cokDa, kiloDaltons. (B) Schematic representations of the organizations of chondrocalcin andthe locations of presumed disulfide bounds. Pacmen represent streptavidin. (C) Direct ELline, full squares), isolated peptide D4 (full line, empty squares) or tetrameric peptide D4molecules. (D) Western blot analysis of the effect of tetrameric peptide D4 over chondroctetrameric peptide D4 over MMP13 gene expression regulation by chondrocalcin. Levels ar* indicates p b 0.05 (n = 3). D4S4: streptavidin-mediated tetramer of peptide D4.

the very preferential formation of a complex containing 4 copies ofpeptide D4 (D4S4, for streptavidin-mediated tetramer of D4, Fig. 6B).Fluorescently labeled D4S4 was used for functional experiments andincorporation visualization while HRP-labeled D4S4 was used toquantify its incorporation into cells. The dose-dependent incorporationof chondrocalcin, D4 and D4S4 (respectively, Fig. 6C) into cells demon-strated that D4S4 was internalized significantly at monomer molarconcentrations approximately 20-fold lower than D4 and only 5-foldhigher than chondrocalcin, confirming that the multimerization of D4favored its cellular incorporation. The capacity of D4S4 to competechondrocalcin at a 100:1 ratio in both terms of cellular uptake and signal-ing was then investigated (Fig. 6D & E). Surprisingly, D4S4 very signifi-cantly (estimated to 5-fold by densitometry) increased chondrocalcinuptake (Fig. 6D). At the same time, excess D4S4 abolished the inductionof MMP13 expression by chondrocalcin, both at the protein and mRNAlevels (Fig. 6D & E). Therefore, the D4 sequence in chondrocalcin doesplay an important signaling function, which is distinct from its internal-ization capacity as the two are oppositely affected by D4S4.

3. Discussion

Since the first description of the cell penetration by HIV Tat-1 proteindue to its 13 amino acid long transduction domain, several dozens ofsuch cell penetrating proteins and corresponding peptides have beendescribed (reviewed in (Milletti, 2012)). They differ by their functions,structures and mechanisms of cellular uptake. In many cases it remainsunclear whether they penetrate cells via active endocytosis or not.Here, exogenous chondrocalcin was found to accumulate into cells(Figs. 3 & 4) in a granular pattern which resembles that observed for

C

0 10 102 103 104 105

inco

rpor

ated

pm

ol /

wel

l

concentration added (nM)

0

0.2

0.4

0.6

0.8

1

02468

1012141618

drocalcinreptavidin

D4D4S4

+ + +

+ ++

+ +

MM

P13

/GA

PD

Hre

l. to

con

trol

*

(A) SDS-PAGE analysis of the trimeric, disulfide bound-stabilized state of recombinantnditions onto SDS-PAGE and detected by Coomassie staining. MW, molecular weight;tetrameric peptide D4. Thick gray bars represent peptide D4. Thin gray bars indicateISA-based quantification of the dose-dependent incorporation of chondrocalcin (full(dotted line, empty triangles). Indicated concentrations are those of the monomeric

alcin uptake and MMP13 protein secretion. (E) Real-time PCR analysis of the effect ofe normalized to those of GAPDH and expressed as relative to control cells (untreated).

449C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

several other penetrating proteins. Chondrocalcin incorporation wasresistant to cytoskeleton disassembly, but sensitive to PAO, suggestinga possible involvement of endocytosis (Fig. 4). Chondrocalcin did notlocalize to clathrin- or caveolin vesicles, nor did it co-localize to any ofthe receptors that we suspected, or to transferrin, which is known toundergo endocytosis (Supplemental Fig. S3). Indeed, at concentrationswhich inhibited chondrocalcin entry, PAO also promoted extracellularmatrix catabolism and cell death after a 24 hour exposure (notshown), corroborating earlier findings in other cell types (Charoensuket al., 2009). Therefore, the specificity of PAO towards endocytosis isquestionable and we could not use PAO to investigate the functionalrelationship between cellular uptake and gene expression regulationby chondrocalcin. Peptide D4 and its counterpart from the C-terminalpropeptide of procollagen Iα1 were internalized in a pattern similarto that of chondrocalcin (Fig. 5). Indeed, the corresponding peptide inthe C-propeptide of type III procollagen localizes to the tips of theC-propeptide trimer (Fig. 5E), favoring molecular interactions withpotential receptors. Chondrocalcin induced cartilage catabolism via in-duction of IL-1β and downstream signaling (Figs. 1–2). It promotedMMP13 release (Fig. 2) at concentrations as low as 500 ng/ml, whichwere reported in synovial fluid depending on the pathological condition(de Grauw et al., 2009). Assuming that chondrocalcin was internalizedand signaled via a single receptor, we anticipated that elevated concen-trations of isolated D4 could titrate the receptor and compete forchondrocalcin entry and signaling. Indeed, multimerized D4 (D4S4) didprofoundly affect the induction of MMP13 expression by chondrocalcin(Fig. 6D & E), indicating that the signaling of chondrocalcin depends ona functional interaction mediated by this peptide. On the other hand,excess D4S4, but not isolated D4 potentiated chondrocalcin uptake(Fig. 6D). This finding profoundly disfavors the hypothesis of a specific,saturable receptor mediating chondrocalcin entry and suggests thatmultiple D4 regions arrange to cooperatively promote internalization.

We found that chondrocalcin could be internalized by cells but didnot evidence that the entry was required for expression regulation.Both events, however, imply the D4 region of chondrocalcin, which isrestricted to 18 amino acids; it makes it quite likely that the two arefunctionally linked. Several penetrating proteins translocate to thenucleus and act as transcriptional regulators (reviewed in (Milletti,2012)). An early study suggested that the C-propeptide of procollagenI was targeted to the nucleus of fibroblasts (Wu et al., 1991). Anotherstudy demonstrated that C-propeptides of procollagens I to IIIinteractedwith Ciz, a transcriptional factor found able to trigger nuclearlocalization of recombinant procollagens (Hayata et al., 2008). Geneexpression regulation by chondrocalcin exerts at the mRNA level inour experiments (Figs. 2 & 6). This would be compatible with the ideathat chondrocalcin translocates and acts in the nucleus. However,chondrocalcin localization to the nucleus was never found convincingor even likely in our hands (Figs. 3 & 5 and Supplemental Figs. S4 &S5). We failed to observe its colocalization to Ciz, even outside thenucleus (Supplemental Fig. S4). Fragments of Fibronectin also induceMMP13 via IL-1β (Saito et al., 1999) and are believed to do so viaintegrin signaling (Forsyth et al., 2002) and downstream ERK (Gembaet al., 2002) and AKT (Yasuda, 2011) activation. We and others foundin the past that chondrocalcin induces chondrocyte adhesion in a β1integrin-dependent manner (Davies et al., 1997; Vincourt et al., 2010).However, we did not observe localization of chondrocalcin to β1integrins (Supplemental Fig. S4) and anti-β1 integrin blocking antibod-ies did not inhibit chondrocalcin uptake (Fig. 4). Furthermore, we didnot observe activation of ERK or AKT upon chondrocyte exposure tochondrocalcin (data not shown). Therefore, the pathways used bychondrocalcin appear very distinct from those used by Fibronectinfragments.

As part of type II procollagen, chondrocalcin serves as a structuralscaffold for triple-helix formation. It is secreted by chondrocytes andthen cleaved from its parent molecule to allow collagen fibrillation.More than 25 years ago, it was proposed to play a mineralizing

function on its own within the extracellular matrix (Poole et al.,1984). We have investigated the regulation of the mineralizing pro-cess in the ATDC5 cell line by exogenously added chondrocalcin, butdid not evidence any significant changes from untreated controls(data not shown), possibly because environmental situations inwhich chondrocalcin participates in this process are subtle. However,we found that extracellular chondrocalcin acts in again another way:it is able to shuttle back to the inside of chondrocytes and modulatecartilage catabolism, after what it appears to be relatively quicklydegraded (Fig. 3). Gene expression regulation by the C-propeptideof collagen Iα1 resembles that by chondrocalcin (Vincourt et al.,2010) and its region corresponding to peptide D4 was internalizedby cells in a similar way. Therefore, cellular uptake and the inductionof extracellular matrix catabolism by chondrocalcin may be generalfunctions of C-propeptides of other fibrillar collagens as well.

4. Experimental procedures

4.1. Tissue procurement

Human articular cartilage was procured under general anesthesiaduring total knee replacements. This study was approved by our localResearch Institution review board (registration number UF 9757-CPRC2004-Cellules souches et chondrogénèse), and patients gave writteninformed consent in accordance with the usual ethical regulations incollaboration with our local bone bank.

4.2. Protein and peptide engineering

Recombinant human chondrocalcin was produced from human cellcultures and purified as explained in Vincourt et al. (2010). Bovineserum albumin (BSA, fraction V) was purchased from Euromedex.Proteins and peptides were biotinylated using Sulfo-NHS biotinylationkit (PIERCE) according to manufacturer's instructions. Proteins weredialyzed twice against culture medium to remove unbound biotin, butpeptideswere not, due to their lowmolecularmasses. Instead, for chon-drocyte phenotype experiments, D4 andD6peptideswere purchased asN-terminally biotinylated, purified peptides. D4S4 and D6S4 were gen-erated by mixing peptides and Alexa-555 coupled streptavidin at a 1:1monomer ratio for 3 h, and dialyzed twice against culture medium.

4.3. Ex vivo stimulation of cartilage and analysis

Chondrocalcin was reported at elevated levels upon osteoarthritis(Garnero, 2006; Conrozier et al., 2008). Therefore, we anticipated thatosteoarthritic cartilage might respond less efficiently to exogenouschondrocalcin. For this reason, cartilage for ex vivo stimulation wasobtained from patients undergoing prosthesis hip replacement dueto femoral head fracture consecutive to osteonecrosis without majormacroscopic cartilage degradation, rather than osteoarthritic carti-lage. Cartilage was cut into pieces of the largest surfaces and thicknesspossible and then sub-divided using 4-mm punchers. Punches from asame initial cartilage piece were dispatched into separate experimentalgroups to diminish the participation of local cartilage condition to thevariations of observed criteria. Explants were cultured in serum freeDMEM/F12 for one week. Explants were then stimulated for 48 h withchondrocalcin. Culture media were collected and processed for GAG,COL2 and MMP13 content measurement. Total GAGs were measuredusing the dimethylmethylene blue assay (Farndale et al., 1982) withchondroitin sulphate as a standard. COL2 proteolytic degradation andMMP13 release were measured using the C2C (Ibex) and MMP13 (R&DSystems) ELISA kits, respectively. For GAG synthesis measurement,explants were pulsed with 0.5 μCi 35S sulphate for the last 6 h, rinsed 8times in fresh culture medium over 2 h, weighed for total contentnormalization, digested with 2 mg/ml collagenase A overnight andtheir radioactive content was measured by scintillography.

450 C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

4.4. Monoclonal antibody generation

Two peptide sequences (NH2-DQAAGGLRQHDAEVDA-NH2 andNH2-NPANVPKKNWWS-NH2) were used for mouse immunization,performed at Eurogentec. Clones were selected based on direct ELISA,western blot and immunofluorescence experiments using either re-combinant chondrocalcin or pro-COL2 as targets. Clone 11D3 wasprocessed for hybridome construction based on its good specificityand sensitivity in all respects. This antibody was found directed againstthe first peptide as characterized by ELISA. Therefore, it does not exhibitany detectable reactivity against peptide D4. It was used in all experi-ments aiming at detecting non-biotinylated chondrocalcin.

4.5. Cell culture, gene and protein expression analysis

Articular chondrocytes were prepared and cultured strictly asdescribed in Vincourt et al. (2012). Gene expression was investigatedby real time PCR and protein expression by western blotting strictlyas in Vincourt et al. (2012). Rabbit anti-GAPDH and Goat polyclonalblocking anti-human interleukin-1 receptor (IL1-RI) were from R&DSystems.

4.6. Immunofluorescence microscopy

When needed, chondrocytes were grown as described above buton glass coverslips in 24 well plates. Coverslips were rinsed in culturemedium and fixed for 20 min in 4% paraformaldehyde (PFA) preparedin culture medium. Cells were permeabilized for 20 min in PBS contain-ing 0.2% (w/v) Triton-X100. Non-specific binding was blocked in 5%BSA in PBS. 11D3 antibody was added at 1 μg/ml overnight at 4 °Cin blocking buffer. Alexa-coupled donkey anti-mouse or streptavidin(Invitrogen) were diluted 200-fold and coverslips were mounted inDAPI containing Vectashield. Images were captured on a DMI3000Bmicroscope (Leica) with a 40× oil immersion objective.

4.7. Chondrocalcin, peptide or D4S4 incorporation

Chondrocytes were stimulated in 96-well culture plates withbiotinylated proteins or peptides, or with horseradish peroxidase(HRP)-labeled D4S4. At the indicated time-points, cells were fixed andpermeabilized with Triton X-100. For BSA, chondrocalcin or peptidequantification, HRP-coupled streptavidin was added and detectedwith ELISA detection kit (both from R&D Systems, used as specified).For D4S4 quantification, ELISA detection was performed without addi-tion of HRP-coupled streptavidin. Standards were coated in separateplates overnight, rinsed three times and processed like samples.Measures fitted a logarithmic regression curve within the range ofreported concentrations. Details are shown in Supplemental Fig. S3.

4.8. Structural mapping of peptide D4 in the C-propeptide of typeIII procollagen

The crystal structure of the C-propeptide of type III procollagen asdetermined in Bourhis et al. (2012) was obtained from the molecularmodeling database from NCBI under reference 102834. The PDB filewas manipulated using RasMol software (Sayle and Milner-White,1995) to visualize the polypeptide chains in blue, red and white,respectively, and their residue positions 94 to 111, corresponding tothe D4 peptide, in green.

4.9. Statistical analysis

All gene expression and western blot experiments were analyzedin experimental triplicates (n = 3). ELISA-based measurements andGAG determination were analyzed in experimental pentaplicates(n = 5). Each experiment was performed at least three times from

independent patients and provided conceptually similar results.p-Values were calculated when required by the two tailed Studentt-test.

Acknowledgments

This work was supported by Association pour la Recherche contre leCancer, Ligue Régionale contre le Cancer, Fondation pour la RechercheMédicale, and Région Lorraine.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.matbio.2013.06.002.

References

Borden, P., Solymar, D., Sucharczuk, A., Lindman, B., Cannon, P., Heller, R.A., 1996. Cytokinecontrol of interstitial collagenase and collagenase-3 gene expression in humanchondrocytes. J. Biol. Chem. 271, 23577–23581.

Bourhis, J.M., Mariano, N., Zhao, Y., Harlos, K., Exposito, J.Y., Jones, E.Y., Moali, C.,Aghajari, N., Hulmes, D.J., 2012. Structural basis of fibrillar collagen trimerizationand related genetic disorders. Nat. Struct. Mol. Biol. 19, 1031–1036.

Charoensuk, V., Gati, W.P., Weinfeld, M., Le, X.C., 2009. Differential cytotoxic effects ofarsenic compounds in human acute promyelocytic leukemia cells. Toxicol. Appl.Pharmacol. 239, 64–70.

Conrozier, T., Poole, A.R., Ferrand, F., Mathieu, P., Vincent, F., Piperno, M., Verret, C.,Ionescu, M., Vignon, E., 2008. Serum concentrations of type II collagen biomarkers(C2C, C1, 2C and CPII) suggest different pathophysiologies in patients with hiposteoarthritis. Clin. Exp. Rheumatol. 26, 430–435.

Davies, D., Tuckwell, D.S., Calderwood, D.A., Weston, S.A., Takigawa, M., Humphries, M.J.,1997. Molecular characterisation of integrin-procollagen C-propeptide interactions.Eur. J. Biochem. 246, 274–282.

de Grauw, J.C., van de Lest, C.H., van Weeren, P.R., 2009. Inflammatory mediators andcartilage biomarkers in synovialfluid after a single inflammatory insult: a longitudinalexperimental study. Arthritis Res. Ther. 11, R35.

Farndale, R.W., Sayers, C.A., Barrett, A.J., 1982. A direct spectrophotometric microassay forsulfated glycosaminoglycans in cartilage cultures. Connect Tissue Res. 9, 247–248.

Forsyth, C.B., Pulai, J., Loeser, R.F., 2002. Fibronectin fragments and blocking antibodiesto alpha2beta1 and alpha5beta1 integrins stimulatemitogen-activated protein kinasesignaling and increase collagenase 3 (matrix metalloproteinase 13) production byhuman articular chondrocytes. Arthritis and rheumatism 46, 2368–2376.

Garnero, P., 2006. Use of biochemical markers to study and follow patients withosteoarthritis. Curr. Rheumatol. Rep. 8, 37–44.

Gemba, T., Valbracht, J., Alsalameh, S., Lotz, M., 2002. Focal adhesion kinase andmitogen-activated protein kinases are involved in chondrocyte activation by the29-kDa amino-terminal fibronectin fragment. J. Biol. Chem. 277, 907–911.

Gottlieb, T.A., Ivanov, I.E., Adesnik, M., Sabatini, D.D., 1993. Actin microfilaments play acritical role in endocytosis at the apical but not the basolateral surface of polarizedepithelial cells. J. Cell Biol. 120, 695–710.

Hayata, T., Nakamoto, T., Ezura, Y., Noda, M., 2008. Ciz, a transcription factor with anucleocytoplasmic shuttling activity, interacts with C-propeptides of type I collagen.Biochem. Biophys. Res. Commun. 368, 205–210.

Hinek, A., Poole, A.R., 1988. The influence of vitamin D metabolites on the calcificationof cartilage matrix and the C-propeptide of type II collagen (chondrocalcin). J. BoneMiner. Res. 3, 421–429.

Hinek, A., Reiner, A., Poole, A.R., 1987. The calcification of cartilage matrix in chondrocyteculture: studies of the C-propeptide of type II collagen (chondrocalcin). J. Cell Biol.104, 1435–1441.

Hosogane, N., Watanabe, K., Tsuji, T., Miyamoto, T., Ishii, K., Niki, Y., Nakamura, M.,Toyama, Y., Chiba, K., Matsumoto, M., 2012. Serum cartilage metabolites as biomarkersof degenerative lumbar scoliosis. J. Orthop. Res. 30, 1249–1253.

Ivanov, A.I., 2008. Pharmacological inhibition of endocytic pathways: is it specificenough to be useful? Methods Mol. Biol. 440, 15–33.

Kiyoshima, D., Kawakami, K., Hayakawa, K., Tatsumi, H., Sokabe, M., 2011. Force- andCa(2)(+)-dependent internalization of integrins in cultured endothelial cells.J. Cell Sci. 124, 3859–3870.

Kujawa, M.J., Weitzhandler, M., Poole, A.R., Rosenberg, L., Caplan, A.I., 1989. Associationof the C-propeptide of type II collagen with mineralization of embryonic chick longbone and sternal development. Connect Tissue Res. 23, 179–199.

Milletti, F., 2012. Cell-penetrating peptides: classes, origin, and current landscape. DrugDiscov. Today 17, 850–860.

Niyibizi, C., Wu, J.J., Eyre, D.R., 1987. The carboxypropeptide trimer of type II collagen isa prominent component of immature cartilages and intervertebral-disc tissue.Biochim. Biophys. Acta 916, 493–499.

Poole, A.R., Pidoux, I., Reiner, A., Choi, H., Rosenberg, L.C., 1984. The association of anewly discovered protein, called chondrocalcin, with cartilage calcification. ActaBiol. Hung. 35, 143–149.

Ramachandiran, V., Grigoriev, V., Lan, L., Ravkov, E., Mertens, S.A., Altman, J.D., 2007. Arobust method for production of MHC tetramers with small molecule fluorophores.J. Immunol. Methods 319, 13–20.

451C. Bantsimba-Malanda et al. / Matrix Biology 32 (2013) 443–451

Saito, S., Yamaji, N., Yasunaga, K., Saito, T., Matsumoto, S., Katoh, M., Kobayashi, S., Masuho,Y., 1999. The fibronectin extra domain A activates matrix metalloproteinase gene ex-pression by an interleukin-1-dependent mechanism. J. Biol. Chem. 274, 30756–30763.

Sayle, R.A., Milner-White, E.J., 1995. RasMol: biomolecular graphics for all. TrendsBiochem. Sci. 20, 374.

Sun, X.M., Canda-Sanchez, A., Manjeri, G.R., Cohen, G.M., Pinkoski, M.J., 2009. Phenylarsineoxide interferes with the death inducing signaling complex and inhibits tumornecrosis factor-related apoptosis-inducing ligand (TRAIL) induced apoptosis.Exp. cell Res. 315, 2453–2462.

Vincourt, J.B., Etienne, S., Cottet, J., Delaunay, C., Malanda, C.B., Lionneton, F., Sirveaux, F.,Netter, P., Plenat, F., Mainard, D., Vignaud, J.M., Magdalou, J., 2010. C-propeptides of

procollagens I alpha 1 and II that differentially accumulate in enchondromas versuschondrosarcomas regulate tumor cell survival and migration. Cancer Res. 70,4739–4748.

Vincourt, J.B., Etienne, S., Grossin, L., Cottet, J., Bantsimba-Malanda, C., Netter, P., Mainard, D.,Libante, V., Gillet, P., Magdalou, J., 2012. Matrilin-3 switches from anti- to pro-anabolicupon integration to the extracellular matrix. Matrix Biol. 31, 290–298.

Wu, C.H., Walton, C.M., Wu, G.Y., 1991. Propeptide-mediated regulation of procollagensynthesis in IMR-90 human lung fibroblast cell cultures. Evidence for transcriptionalcontrol. J. Biol. Chem. 266, 2983–2987.

Yasuda, T., 2011. Activation of Akt leading to NF-kappaB up-regulation in chondrocytesstimulated with fibronectin fragment. Biomed. Res. 32, 209–215.