Pericellular Matrilins Regulate Activation of Chondrocytes by Cyclic Load-Induced Matrix Deformation

Authors

  • Katsuaki Kanbe,

    1. Department of Orthopaedic Surgery, Tokyo Women's Medical University/Medical Center East, Tokyo, Japan
    Search for more papers by this author
  • Xu Yang,

    1. Cell and Molecular Biology Laboratory, Department of Orthopaedics, Brown Medical School/Rhode Island Hospital, Providence, Rhode Island, USA
    Search for more papers by this author
  • Lei Wei,

    1. Cell and Molecular Biology Laboratory, Department of Orthopaedics, Brown Medical School/Rhode Island Hospital, Providence, Rhode Island, USA
    Search for more papers by this author
  • Changqi Sun,

    1. Cell and Molecular Biology Laboratory, Department of Orthopaedics, Brown Medical School/Rhode Island Hospital, Providence, Rhode Island, USA
    Search for more papers by this author
  • Qian Chen PhD

    Corresponding author
    1. Cell and Molecular Biology Laboratory, Department of Orthopaedics, Brown Medical School/Rhode Island Hospital, Providence, Rhode Island, USA
    • Department of Orthopaedics, Brown Medical School/Rhode Island Hospital, 1 Hoppin Street, Suite 402, Providence, RI 02903, USA
    Search for more papers by this author

  • The authors state that they have no conflicts of interest.

Abstract

Pericellular matrix is at the ideal location to be involved in transmitting mechanical signals from the microenvironment to a cell. We found that changes of the content of matrilins that link various pericellular molecules surrounding chondrocytes affect mechanical stimulation of chondrocyte proliferation and gene expression. Thus, pericellular matrilins may play a role in chondrocyte mechanotransduction.

Introduction: Chondrocytes reside in a capsule of pericellular matrix (chondron), which has been hypothesized to play a critical role in transducing mechanical signals to the cell. In this study, we test the hypothesis that the levels of matrilin (MATN)-1 and -3, major components of the chondrocyte pericellular matrix network, regulate activation of chondrocyte proliferation and differentiation by cyclic load–induced matrix deformation.

Materials and Methods: Functional matrilins were decreased by expressing a dominant negative mini-MATN in primary chondrocytes or by using MATN1-null chondrocytes. The abundance of matrilins was also increased by expressing a wildtype MATN1 or MATN3 in chondrocytes. Chondrocytes were cultured in a 3D sponge subjected to cyclic deformation at 1 Hz. Chondrocyte gene expression was quantified by real-time RT-PCR and by Western blot analysis. Matrilin pericellular matrix assembly was examined by immunocytochemistry.

Results: Elimination of functional matrilins from pericellular matrix abrogated mechanical activation of Indian hedgehog signaling and abolished mechanical stimulation of chondrocyte proliferation and differentiation. Excessive or reduced matrilin content decreased mechanical response of chondrocytes.

Conclusions: Normal content of matrilins is essential to optimal activation of chondrocytes by mechanical signals. Our data suggest that the sensitivity of chondrocytes to the changes in the microenvironment can be adjusted by altering the content of matrilins in pericellular matrix. This finding supports a critical role of pericellular matrix in chondrocyte mechano-transduction and has important implications in cartilage tissue engineering and mechanical adaptation.

INTRODUCTION

It is known that mechanical signals regulate cartilage homeostasis not only during fracture healing and joint repair in adults but also during endochondral bone formation in development.(1,2) Cartilage matrix deformation, as a result of mechanical loading, regulates chondrocyte proliferation, maturation, and gene expression.(3) Using a 3D culture system in which primary chondrocytes were cultured in collagen sponges subjected to intermittent 5% deformation, we showed that mechanical signals significantly stimulated chondrocyte proliferation and maturation.(3) Mechanical activation of Indian hedgehog (Ihh), an essential regulator of chondrocyte proliferation and differentiation,(4) is required for mechanical stimulation of chondrocyte proliferation.(5) Cyclic loading induces chondrocyte expression of Ihh and its receptor patched (Ptc), a target gene of Ihh signal,(6) which in turn, mediates mechanical loading stimulated cell proliferation.

Chondrocytes are completely surrounded by extracellular matrix in cartilage. They are encapsulated by their immediate pericellular matrix network, which together with the residing chondrocyte, forms a functional unit termed a chondron.(7,8) The complete chondron is further embedded in a matrix high in type II collagen and hyaluronan/aggrecan/link protein complex that provide the tissue with its mechanical properties. Thus, pericellular matrix is situated at the ideal location to transmit mechanical load–induced matrix deformation signals to the cell.

Matrilins (MATNs), oligomeric matrix proteins that form filamentous networks, are major components of pericellular matrix.(9,10) As the prototype of the matrilin family, MATN1 interacts with the chondrocyte membrane through the α1β1 integrin(11) and with other pericellular matrix components such as collagen VI.(12) In addition, matrilins interact with major extracellular matrix molecules in cartilage such as collagen type II and IX,(13) proteoglycans including aggrecan and small leucine-rich proteoglycans,(14) and cartilage oligomeric matrix protein (COMP).(15) The matrilin family consists of four members, all of which contain the adhesive von Willebrand factor A (vWF A) domains and form filamentous matrix networks.(16) Whereas MATN2 and 4 are expressed in many tissues such as bone and lung, MATN1 and 3 are expressed mainly in cartilage.(10) MATN1 and 3 form various oligomeric forms, including homo-oligomers (MATN-1)3, (MATN-3)4, and heterotetramer (MATN-1)2 (MATN-3)2.(17,18) Mutations in the gene encoding MATN3 are associated with autosomal dominant multiple epiphyseal dysplasia (MED), an osteochondrodysplasia characterized by delayed and irregular ossification of the epiphyses and early-onset osteoarthritis.(19) In addition, the abundance of matrilins change during aging and under pathological conditions.(20) However, it is not known how the changes in matrilin content affect chondrocyte properties.

The assembly of matrilin filaments consists of two steps, which are governed by various domains in a matrilin molecule. They include the vWF A domain, epithelial growth factor (EGF)-like domain, and a coiled-coil domain at the –COOH terminus. In the first step, a matrilin monomer assembles into an oligomer through the coiled-coil domain(21) and is stabilized by intermolecular covalent disulfide bonds from a pair of cysteines within the domain.(17,22) Thus, the coiled coil domain is responsible for MATN3 being able to form homo- and hetero-oligomers. In the second step, matrilin oligomers form filaments through the adhesive A domains. The metal-ion–dependent adhesion site (MIDAS) within the A domain is responsible for establishing adhesion of matrilins in matrix.(9) Deletion of either A domain of MATN1 completely abolishes its ability to form a filamentous network.(9) Furthermore, overexpression of a mini-matrilin1 that lacks the A domain in chondrocytes achieves a dominant negative effect on endogenous filamentous network formation by sequestering the wildtype (WT) matrilins.(9)

To test the hypothesis that matrilins are involved in transducing mechanical signals, we altered the content of matrilins in pericellular matrix. Functional matrilins were decreased by expressing a dominant negative mini-MATN3 in primary chondrocytes or by using MATN1-null chondrocytes. The abundance of matrilins was also increased by expressing a WT MATN1 or MATN3 in chondrocytes. Transfected chondrocytes were cultured in a 3D sponge subjected to cyclic deformation at 1 Hz. We determined the effect of alteration of matrilin content on the mechanical responsiveness of chondrocytes including activation of the Ihh signaling pathway, stimulation of cell proliferation and maturation, and changes in gene expression.

MATERIALS AND METHODS

Cloning of a full-length chicken MATN3 cDNA

A full length of matrilin-3 cDNA (MATN3) was cloned from total RNA isolated from chick sternal cartilage by RT-PCR using primers, 5′-ATGCGGCGGGCGCTCGGAACGCTGG-3′ (sequence from 1–25) and 5′-TCAGACAACCTGCTGTCTGTCTTGATAGGCTTGCA-3′ (sequence from 1335–1369), designed according to a DNA sequence from GenBank (accession no. AJ000055). Total RNA was isolated from 17-day embryonic chick sterna using RNeasy kit (Qiagen). RT-PCR of matrilin-3 mRNA was performed using Superscript II RTase (GIBCO BRL, Rockville, MD, USA) following the manufacturer's protocol. The nucleotide sequence of the full-length 1358-base pair matrilin-3 cDNA was determined by DNA sequencing and was found to be identical to that from GenBank (accession no. AJ000055), except the difference of eight nucleotides as follows: T631C, C704G, T755C, T851C, C920T, T1038C, T1172C, and G1188A. Among these, changes of three amino acids occurred: Ile203 to Thr203, Tyr338 to His338, and Val388 to Ile388. MATN3 and a mini-MATN3 derived from MATN3 were cloned into a retroviral vector as described previously.(23) In addition, a full-length matrilin 1 cDNA (MATN1) that had been cloned into RCAS (Replication Competent Adenovirus with Slice acceptor) previously(23,23) was used in this study (Fig. 1B). A V5 tag (GKPIPNPLLGLDST) was attached to the –COOH terminus of miniMATN3, whereas a myc-tag (EQKLISEEDL) was attached to the –COOH terminus of MATN1 and MATN3, respectively. The RCAS vector without any cDNA insert (RCAS) was used as a control for possible effects of retroviral infection. Transfection without cDNA (Mock) was also performed for control of transfection procedure. Construction of a mini-MATN3 was described previously.(17)

Figure Figure 1.

(A) Constructs of matrilin-3 and matrilin-1 cDNA. Full-length mat-3 consists of a signal peptide, A1 domain, four EGF repeats, and a coiled-coil domain. Full-length mat-1 consists of a signal peptide, two A domains (A1 and A2) separated by an EGF domain and a coiled-coil. Mini-MATN3 is derived from the full-length mat-3 with A domain deleted and c terminus tagged with a V5 motif. Antibodies 1H1 and D2 are used to detect matrilin-1, whereas antibody V5 is used to detect mini-matrilin-3. The antigenic sites of these antibodies are indicated. (B) Mechanical signals increase the mRNA level of matrilin-1 but not that of matrilin-3. Chondrocytes were incubated in 3D culture at 37°C for 2 days, with or without cyclic load as indicated. Total RNA was extracted from the cells, which were collected after digestion of sponges. The mRNA levels of MATN1 and MATN3 were determined by real-time quantitative RT-PCR (mean ± SD, n = 3, the mean of the transcript level at the 2-day incubation under loaded conditions was set to 10 units. *p < 0.01, compared with its counterpart from nonload).

Primary culture of chondrocytes

Primary cultures of chick embryonic chondrocytes were established as follows. Sterna from 17-day embryonic chicks were subjected to enzymatic treatment with 0.1% trypsin (Sigma, St Louis, MO, USA), 0.3% collagenase (Worthington, Freehold, NJ, USA), and 0.1% type 1 testicular hyaluronidase (Sigma; dissociation medium). After an incubation of 30 minutes at 37°C, the dissociation medium was replaced with fresh medium and incubated at 37°C for an additional 1 h. Chondrocytes were resuspended in Ham's F-12 medium containing 10% FBS (Life Technologies, Grand Island, NY, USA) and 0.01% testicular hyaluronidase. After culturing overnight, the medium was replaced with fresh medium without hyaluronidase. Medium was changed every other day.

Transfection of matrilin transgenes

Primary cell cultures that produced retrovirus that harbor transgenes were established as described.(23) Briefly, confluent fibroblasts were split 1:5 and incubated overnight. Six micrograms of retroviral cDNA constructs was transfected into monolayer fibroblasts using the calcium phosphate method. Cultures were incubated for 1 week and assessed for viral infection by immunostaining using an antibody against viral gag protein within the retroviral cDNA constructs.(9) For preparation of recombinant virus, the medium of infected fibroblasts was replaced by a thin layer of fresh medium. After 24 h, the culture supernatant was collected, and debris was removed by centrifuging at 3750 rpm for 5 minutes. The supernatant that contained the viral particles was stored frozen at −80°C. Infection of primary cultures of chicken embryo chondrocytes (CECs) or CEFs (chicken embryo fibroblasts) with recombinant virus was achieved as follows. Confluent cultures were split 1:5 and plated overnight. Cells were infected with a thin layer of filtered culture supernatant that contained the virus. After 1 h of incubation, medium was added to the normal level (15 ml/100-mm dish). The infected culture was incubated for 1 week to allow the infection to spread before further characterization. For production of recombinant MATN1, conditioned medium was collected after CECs were transfected with MATN1 for 2 days.

Immunofluorescent cytochemistry and Western blot analysis

For immunofluorescence staining, cultured cells were fixed at room temperature with 4% paraformaldehyde for 20 minutes. Slides were washed with PBS and incubated with primary antibodies. After washing with PBS, affinitypurified fluorescein- or rhodamine-conjugated donkey anti-mouse or donkey anti-rabbit antibodies (Jackson ImmunoResearch, West Grove, PA, USA) were applied with Hoechst nuclear dye (0.5 mg/ml). Slides were washed and mounted in 95% glycerol in PBS. Single or multiple exposure photography was performed with a microscope from Nikon (Melville, NY, USA).

Western blot analysis was performed with both the conditioned medium and cell extracts. Twenty milliliters of conditioned medium from transfected cultures was collected from each plate. After removal of the medium, cells were washed with PBS for 5 minutes three times and with 0.1 M glycine in PBS for 20 minutes. After another three washes with PBS, the monolayers were extracted with 1 ml extraction buffer (4 M urea, 50 mM Tris, pH 8.5, and 0.1 mM phenylmethylsulphonylfluoride [PMSF]). Cells were scraped off the dishes and passed through 21-gauge needles to shear DNA. After incubation for 30 minutes on ice, the supernatant (soluble fraction) was used for electrophoretic analysis. For nonreducing condition, cell extracts or medium were mixed with standard 2× SDS gel-loading buffer. For reducing conditions, the loading buffer contained 5% β-mercaptoethanol and 0.05 M DTT. Protein concentration in each sample was determined by BCA protein assay (Pierce, Rockford, IL, USA). Five micrograms of extracted protein was boiled for 10 minutes before being loaded into each lane of a 10% SDS-PAGE gel. After electrophoresis, proteins were transferred onto an Immobilon-polyvinylidene difluoride membrane (Millipore, Bedford, MA, USA) in 25 mM Tris, 192 mM glycine, and 15% methanol. The membranes were blocked in 2% BSA fraction V (Sigma) in PBS for 30 minutes and probed with antibodies. Horseradish peroxidase (HRP)-conjugated goat anti-rabbit, goat anti-mouse, or rabbit anti-goat immunoglobulin G (heavy and light chain; Bio-Rad, Melville, NY, USA) was diluted 1:3000 and used as a secondary antibody, respectively. Visualization of immunoreactive proteins was achieved by using the enhanced chemiluminescense (ECL) Western blotting detection reagents (Amersham, Arlington Heights, IL, USA) and exposing the membrane to Kodak (Rochester, NY, USA) X-Omat AR film. The exposed film was quantified using Discovery Series Quantity One software from Protein Databases (Huntington, NY, USA).

Matn1 was detected by mAb 1H1 or pAb D2, as described previously.(23) Myc tag was recognized by mAb 9E10, which was against an epitope in human c-myc. These antibodies were generated in our laboratory.(23) V5 tag was detected by a monoclonal antibody against V5 (Invitrogen). We also used mAb 5E1 to detect Ihh (Developmental Hybridoma Bank) as described previously.(5) We used mAb II-II6B3 to detect type II collagen (Developmental Hybridoma Bank) as described previously.(23)

Mechanical stimulation of chondrocytes

Chondrocytes were cultured in 3D collagen sponges as described previously.(3) Briefly, 100 μl of cell suspension (1 × 106/ml) was applied to 2 × 2 × 0.25-cm Gelform sponges (Upjohn, Kalamazoo, MI, USA) presoaked with Hanks' balanced salt solution (HBSS). After overnight incubation, >80% of cells attached to collagen scaffoldings in the sponge. Sponges were stretched with an intermittent pattern (5% elongation, 60 stretches/minute, 15 minutes/h) for 2 days with a Bio-stretch device (ICCT). The extent of matrix deformation generated by such device was found to be comparable with that experienced in vivo.(3)

Determination of cell growth

After mechanical stimulation, collagen sponges with cultured cells were washed three times in HBSS before digestion with 0.03% (wt/vol) collagenase in HBSS for 15 minutes at 37°C. Cells were collected by centrifuging at 1000 rpm for 7 minutes and resuspended in HBSS. Cell number was determined by counting with a hemacytometer (American Optical, Buffalo, NY, USA) under a microscope. The viability of cells was determined by trypan blue exclusion assay.

Real-time quantitative PCR

Total RNA from chondrocytes cultured in collagen sponges was extracted with RNeasy mini kit (Qiagen), after sponges were washed thoroughly with HBSS, cut into small pieces, and digested in 0.03% (wt/vol) collagenase in HBSS at 37°C for 10 minutes. Chondrocytes were collected by centrifugation for RNA preparation. One microgram total RNA was used for each reverse transcriptase reaction in a reaction buffer containing 1 μl oligo(dT) and 1 μl 10 mM dNTP Mix (Bio-Rad). Real-time quantitative PCR amplification was performed using QuantiTect SYBR Green PCR kit (QIAGEN) with DNA Engine Opticon 2 Continuous Fluorescence Detection System (MJ Research, Waltham, MA, USA). The same amount of RNA was used for both mechanically loaded and nonloaded samples, and the levels of mRNA were adjusted against that of 18S rRNA. The probes and primers used to detect chicken MATN1, MATN3, and α1(X) mRNA were described by Zhang and Chen,(17) and those to detect Ihh and Ptc were described by Wu et al.(5)

The forward and reverse primers to detect mouse type X collagen (ColX) mRNA were 5′-CCAGGTGTCCCAGGATTCCC-3′ (sequence from 2063–2082) and 5′-CAAGCGGCATCCCAGAAAGC-3′(sequence from 2293–2312), respectively (GenBank accession no.: NM_009925). ColX mRNA levels were normalized to housekeeping gene 18S RNA levels. The forward and reverse primers for 18S were 5′-CGGCTACCACATCCAAGGAA-3′ (sequence from 386–406) and 5′-GCTGGAATTACCGCGGCT-3′(sequence from 553–570), respectively (GenBank accession no.: AY040696). Calculation of mRNA values was performed as previously described. The 18S RNA was amplified at the same time and used as an internal control. The cycle threshold (Ct) values for 18S RNA and that of samples were measured and calculated by computer software (PE ABI). Relative transcript levels were calculated as x = 2−ΔΔCt, in which ΔΔCt = ΔE − ΔC, and ΔE = Ctexp − Ct18s; ΔC = Ctctl − Ct18s. Data were presented as mean ± SE for three samples and analyzed using two-way ANOVA. The level of Col X in WT unloaded cells was designated as 1. Statistical significance was taken at p < 0.05 or 0.01.

Isolation and culture of newborn mouse rib chondrocytes in 3D collagen system

The ventral parts of the rib cages were dissected from 7-day-old WT and matrilin-1 knockout (KO) mice(24) and washed with cold sterile PBS. Rib cage cartilage was incubated in collagenase D solution (3 mg/ml) in the CO2 incubator at 37°C for 1.5 h to digest the attached soft tissue, followed by washing with PBS and incubation in fresh collagenase D for 6 h. The suspension was transferred to a new tube and centrifuge to pellet the cells. After washing, chondrocytes were plated in 3D collagen system at the density of a million of cells per sponge as previously described.(5) After incubation overnight, the sponges were mechanically loaded to induce 5% elongation at 60 cycles per minute by a computer controlled Bio-Stretcher (ICCT Technologies) for 48 h. Control sponges were not cyclically loaded. Total RNA was extracted from cells in sponges with RNeasy mini kits (QIAGEN).

Statistical analysis

Two-tailed t-tests were used to compare mRNA levels from mechanically loaded chondrocytes in the sponge to those in the sponge under nonload conditions. mRNA levels in chondrocytes under different conditions were analyzed by one-way ANOVA with Dunnett multiple comparison posthoc test. For these calculations, p < 0.05 was considered to be statistically significant unless specifically noted.

RESULTS

MATN1 and MATN3

To determine the function of matrilins, we cloned a full-length MATN3 cDNA from developing chicken sternal cartilage and constructed a mini-MATN3 that lacked the adhesive vWF A domain (Fig. 1A). In addition, a full-length MATN1 cDNA that we cloned previously(23) was also used in this study. We first determined whether expression of endogenous MATN1 and -3 in primary embryonic chicken chondrocytes was responsive to mechanical regulation. The mRNA levels of MATN1 and -3 in chondrocytes cultured in 3D collagen sponges in response to mechanical deformation were quantified by real-time RT-PCR. The MATN1 mRNA level was increased 122% by cyclic load; however, the mRNA level of MATN3 was not significantly increased in response to mechanical load (Fig. 1B). This suggested that, whereas the expression of MATN1 in chondrocytes was sensitive to mechanical load induced matrix deformation, the expression of MATN3 was not.

Expression of matrilins

We transfected chondrocytes with different matrilin cDNA constructs. To examine matrilin expression levels in these cells, we quantified MATN1 and -3 mRNA levels by real-time RT-PCR. Transfection of MATN1 cDNA resulted in >100-fold increase of MATN1 mRNA in chondrocytes under both load and nonload conditions (Fig. 2A, MATN1). Transfection of MATN3 or Mini-MATN3 cDNA resulted in >800-fold increase of MATN3 mRNA levels under both load and nonload conditions (Fig. 2B, Mini-MATN3 and MATN3, the MATN3 primer pair for real-time PCR, could detect both MATN3 and Mini-MATN3 mRNA). Thus, transfected matrilin cDNA was successfully expressed by chondrocytes. In response to cyclic loading, WT chondrocytes increased MATN1 mRNA levels significantly (Fig. 2A, Mock and RCAS), but did not increase MATN3 mRNA levels (Fig. 2B, Mock and RCAS). This is consistent with the conclusion drawn from Fig. 1A using nontransfected chondrocytes. Expression of Mini-MATN3 diminished mechanical stimulation of MATN1 mRNA levels (Fig. 2A, Mini-MATN3). In contrast, overexpression of MATN1 had no effect on MATN3 expression in either the unloaded or loaded cells (Fig. 2B, MATN3).

Figure Figure 2.

Expression of matrilins in chondrocyte cultures. Transfected chondrocytes were incubated in 3D culture at 37°C for 2 days, with or without load as indicated. Total RNA was extracted from the cells, which were collected after digestion of sponges. The mRNA levels of (A) MATN1 and (B) MATN3 were determined by real-time quantitative RT-PCR (mean ± SD, n = 3, the mean of the transcript level in Mock under nonload conditions was set to 1 unit; *p < 0.01, compared with its counterpart from nonload; #p < 0.01, compared with RCAS control; |P%p < 0.01, compared with Mini-MATN3 in LOAD group). Note the primers for MATN3 detect both MATN3 and Mini-MATN3.

Secretion of matrilins

To determine whether transfected matrilins were secreted by chondrocytes successfully, we performed Western blot analysis using conditioned medium collected from cell culture after transfection. The antibody against the V5 tag detected Mini-MATN3 recombinant protein specifically (Fig. 1A). It detected four products containing Mini-MATN3 in the medium under nonreducing conditions (Fig. 3A, V5). Based on their molecular mass (Fig. 3B), these products were predicted to be putative Mini-MATN3 monomer, Mini-MATN3 homo-oligomers, and Mini-MATN3 hetero-oligomers with endogenous MATN1 in chondrocytes (Fig. 3A, V5). Under reducing conditions, all of these products were shifted to the 34-kDa Mini-MATN3 monomer with only a trace amount of the 68-kDa oligomer. This showed that the Mini-MATN3 recombinant protein was secreted into medium successfully. Western blot analysis using antibody D2 against MATN1 showed the presence of 54-kDa MATN1 monomers in conditioned medium from all groups under reducing conditions (Fig. 3A, D2). Therefore, expression of Mini-MATN3 did not interfere with secretion of endogenous MATN1. Under nonreducing conditions, MATN1 oligomers at 200 kDa were detected in the medium in all groups except Mini-MATN3–expressing cells, which secreted a 176-kDa hetero-oligomeric protein containing both Mini-MATN3 and MATN1 (Fig. 3A, D2).

Figure Figure 3.

Secretion of matrilins in chondrocyte cultures. (A) Western blot analysis of conditioned medium from cultures of transfected chondrocytes with a mAb against the V5 tag in recombinant Mini-MATN3 (V5, top) or with a pAb against MATN1 (D2, bottom). Chondrocytes were transfected with no DNA (Mock), empty vector (RCAS), matrilin-1 transgene (MATN1), mini-MATN3 (Mini-MATN3), or WT matrilin-3 (MATN3), as indicated. (B) Calculated molecular weight of matrilin-1 and mini-matrilin-3 homo- and hetero-oligomers.

Assembly of pericellular matrilins

To determine whether the change of secretion from 200-kDa endogenous MATN1 oligomers to 176-kDa hetero-oligomer containing Mini-MATN3 by chondrocytes affected pericellular matrix assembly, we performed immunocytochemistry analysis of primary chondrocytes with a monoclonal antibody (1H1) against MATN1. Extensive pericellular filamentous matrilin capsules were detected around WT chondrocytes (Fig. 4A) and around chondrocytes transfected with MATN1 (Fig. 4C) or MATN3 (Fig. 4D). They resulted in an integrated matrix network connecting a group of chondrocytes (Figs. 4A, 4C, and 4D). This matrix network contains both type II collagen–associated and type II collagen–free matrilin filaments (Fig. 4E). However, pericellular matrilin capsules were missing from Mini-MATN3–expressing chondrocytes, although intracellular matrilins could still be seen (Fig. 4B). This suggested that pericellular matrilin network was not assembled around Mini-MATN3–expressing chondrocytes, which resulted in cells that were not connected by matrilin networks (Fig. 4B).

Figure Figure 4.

(A–D) Micrographs of immunofluorescent analysis of transfected chondrocyte cultures with rhodamine coupled mAb against MATN1 (1H1). Chondrocyte nuclei were stained by DAPI (blue), and MATN1 was visualized by rhodamine (red). Chondrocytes were transfected with (A) empty vector RCAS, (B) mini-MATN3, (C) MATN1, or (D) MATN3. Note the lack of MATN1 pericellular matrix in mini-MATN3 transfected cell culture (B). (E) Micrograph of double immunofluorescent analysis of RCAS transfected chondrocyte cultures with rhodamine coupled mAb against type II collagen (II-II6B3) and fluorescen coupled pAb against MATN1 (D2). Type II collagen–associated MATN1 filaments were visualized by the yellow color, whereas type II collagen free MATN1 filaments were visualized by the green color. Bar = 10 μm.

Mechanical activation of Ihh signaling

It has been previously reported that the Ihh signaling pathway was activated by mechanical load and played an important role in mechanotransduction of chondrocytes.(5) To determine whether mechanical induction of Ihh was dependent on the abundance of matrilins, we quantified the Ihh mRNA levels in chondrocytes in response to mechanical signals. Ihh mRNA levels in WT chondrocytes (mock or viral vector RCAS transfected) were increased 50-fold by mechanical stimulation (Fig. 5A, Mock and RCAS, compare Nonload to Load). MATN1- and -3–overexpressing cells increased Ihh mRNA by 30-fold in response to mechanical signals (Fig. 5A, MATN1 and MATN3), which was significantly lower than the increase in WT chondrocytes. However, mechanical stimulation of Ihh mRNA was abolished in mini-MATN3–expressing cells (Fig. 5A, Mini-MATN3, compare Nonload to Load).

Figure Figure 5.

Mechanical stimulation of Ihh signaling pathway. Transfected chondrocytes were incubated in 3D culture at 37°C for 2 days, with or without mechanical load as indicated. Total RNA or protein was extracted from the cells, which were collected after digestion of sponges. (A) The mRNA level of Ihh was determined by real-time quantitative RT-PCR (mean ± SD, n = 3, the mean of the transcript level in Mock under nonload conditions was set to 1 unit; *p < 0.01, compared with its counterpart from nonload; #p < 0.01, compared with RCAS control). (B) The protein level of Ihh in chondrocyte lysate was quantified by Western blots (mean ± SD, n = 3, the mean of the protein level in Mock under nonload conditions was set to 1 unit; *p < 0.01, compared with its counterpart from nonload; #p < 0.01, compared with RCAS control; one representative Western blot was shown; equal amount of total protein was loaded in each lane). (C) The mRNA levels of Ptc, which are induced by Ihh signal, were determined by real-time quantitative RT-PCR (mean ± SD, n = 3, the mean of the transcript level in Mock under nonstretched conditions was set to 1 unit; *p < 0.01, compared with its counterpart from nonload; #p < 0.01, compared with RCAS control).

To determine whether the Ihh protein levels were increased in response to mechanical signals, we performed Western blot analysis. Basal levels of Ihh protein were observed in all groups of chondrocytes under nonload conditions (Fig. 5B). In response to cyclic load, WT chondrocytes (RCAS) increased Ihh protein levels by 15-fold, whereas MATN1- and MATN3-overexpressing chondrocytes increased Ihh protein levels by only 6-fold. No difference of Ihh protein levels was observed between load and nonload conditions in Mini-MATN3–expressing chondrocytes (Fig. 5B).

To determine whether the Ihh signaling pathway was altered in matrilin-expressing chondrocytes, we quantified the mRNA levels of Ptc, a receptor of Ihh that was induced by hh signals. The mRNA levels of Ptc were increased 2.5-fold in mechanically stimulated WT chondrocytes (Fig. 5C, Mock and RCAS). This stimulation was abolished in mini-MATN3–expressing cells (Fig. 5C, mini-MATN3) and reduced in MATN1- and -3–overexpressing cells (Fig. 5C, MATN1 and -3).

Mechanical stimulation of chondrocyte proliferation

It has been shown previously that mechanical activation of Ihh pathway mediates mechanical stimulation of chondrocyte proliferation and differentiation.(5) To determine the effect of matrilins on mechanical stimulation of chondrocyte proliferation, the cell number in a 3D sponge was quantified during the course of cyclic loading (Fig. 6A). After 2 days of cyclic loading, the number of WT chondrocytes (RCAS) increased 144% under mechanical load conditions, whereas the number of MATN1- and MATN3-overexpressing chondrocytes increased only 40.7% and 22.3%, respectively. After 4 days of loading, the number of WT chondrocytes increased 97.7% under load conditions, whereas MATN1- and MATN3-overexpressing cells increased 72.5% and 60.6%, respectively, under mechanical loading. After 6 days of loading, the number of WT chondrocytes increased 48.8% under load conditions, whereas MATN1- and MATN3-overexpressing chondrocytes increased only 21.9% and 17.2%, respectively (Fig. 6A). There was no significant difference of cell number in Mini-MATN3–expressing chondrocytes between load and nonload conditions throughout the entire course (days 2, 4, and 6; Fig. 6A).

Figure Figure 6.

Mechanical stimulation of chondrocyte proliferation and differentiation. (A) The number of chondrocytes in a collagen sponge in 3D culture under nonload and load conditions for 2, 4, and 6 days (mean ± SD, n = 3; *p < 0.01, compared with its counterpart from nonload; #p < 0.01, compared with RCAS control). There is no statistical difference between load and nonload for all the mini-MATN3 samples, p = 0.12 (D2), p = 0.07 (D4), and p = 0.05 (D6). (B) mRNA levels of Col X. Transfected chondrocytes were incubated in 3D culture at 37°C for 2 days, with or without load as indicated. Total RNA was extracted from the cells, which were collected after digestion of sponges. The mRNA levels of type X collagen were determined by real-time quantitative RT-PCR (mean ± SD, n = 3, the mean of the transcript level in Mock under nonload conditions was set to 1 unit; *p < 0.01, compared with its counterpart from nonload; #p < 0.01, compared with RCAS control). (C) Percentage of type X collagen mRNA levels increase in chondrocytes in response to mechanical loading. Chondrocytes were transfected with MATN1, Mini-MATN3, or no DNA (Mock), followed by incubation in 3D culture at 37°C. Before applying mechanical load, the following media were added to the Mini-MATN3–transfected chondrocyte cultures: control medium (recombinant MATN1–), conditioned medium containing recombinant MATN1 (5 μl/ml, recombinant MATN1+), or conditioned medium containing high concentration of recombinant MATN1 (40 μl/ml, recombinant MATN1+++). After incubation under mechanical load for 2 days, total RNA was extracted from the cells, and the mRNA levels of type X collagen were determined by real-time quantitative RT-PCR (mean ± SD, n = 3).

Mechanical stimulation of chondrocyte hypertrophy

Next we determined whether the abundance of functional matrilins also affected mechanical stimulation of chondrocyte differentiation. We quantified the mRNA levels of type X collagen, a chondrocyte hypertrophy marker. Cyclic load greatly increased the mRNA level of α1(X) in WT chondrocytes (Fig. 6B, Mock and RCAS). The levels of α1(X) mRNA were also increased in MATN1- and MATN3-overexpressing chondrocytes, but the extent of the increase was much less than that of the WT chondrocytes (Fig. 6B, MATN1 and MATN3). Mini-MATN3–expressing chondrocytes exhibited a decrease of α1(X) mRNA levels in response to mechanical load (Fig. 6B, Mini-MATN3).

To determine whether MATN1 could rescue the inhibitory effect of Mini-MATN3 on mechanical stimulation of α1(X) mRNA levels, we produced recombinant MATN1 from transfecting chicken skin fibroblasts with MATN1-RCAS cDNA. Adding conditioned medium containing recombinant MATN1 into Mini-MATN3–transfected chondrocyte culture partially rescued the inhibitory effect of Mini-MATN3 on mechanical stimulation of Col X mRNA expression (Fig. 6C). However, adding large amounts of MATN1 (8 times of the “rescuing” amount) diminished the rescuing effect of MATN1 (Fig. 6C).

Mechanical response of MATN1-null chondrocytes

To determine whether the reduction but not the depletion of functional matrilins affects mechanical response of chondrocytes, we cultured chondrocytes from MATN1 KO mice under mechanical loading. MATN1 KO chondrocytes did not synthesize MATN1 but produced normal amount of other matrilins.(24) In contrast to the significant stimulation of cell proliferation as exhibited by WT chondrocytes in response to mechanical loading, MATN1 KO chondrocytes exhibited a diminished response (Fig. 7A). Real-time RT-PCR analysis showed that the mRNA levels of collagen type X were increased >2.5-fold in response to cyclic loading in WT chondrocytes, but increased only 0.5-fold in MATN1 null chondrocytes (Fig. 7B). Therefore, reduction of normal matrilin content diminished mechanical response of chondrocytes.

Figure Figure 7.

Reduced mechanosensitivity in MATN1 KO chondrocytes. (A) Percentage of cell number increase in response to mechanical loading in MATN1 WT and MATN1 KO chondrocytes. Primary mouse rib chondrocytes were seeded in a collagen sponge (105 cells/sponge) and cultured for 4 or 6 days in the absence or presence of cyclic loading at 1 Hz. The number of chondrocytes was counted after digestion of collagen sponges with collagenases (mean ± SD, n = 4). (B) The mRNA levels of type X collagen. Primary mouse rib chondrocytes were cultured in 3D collagen system for 2 days in the absence or presence of cyclic loading at 1 Hz. Real-time PCR analysis of total mRNA was performed. The asterisks indicate a statistically significant difference between the nonload control and the load (*p < 0.05; **p < 0.01; n = 3).

DISCUSSION

Mechanical responsiveness is a hallmark of skeletal cells, which is necessary for morphogenesis and remodeling of skeleton in response to mechanical load.(25,26) It has been shown that mechanical signals regulate chondrocyte proliferation, maturation, and gene expression.(1,3) Mechanotransduction is essential even for cartilage under non–weight-bearing conditions, because elimination of mechanical stimulation in ovo by paralysis of muscles attached to the bone results in significant inhibition of chondrocyte proliferation in the growth plate. However, the identity of extracellular matrix molecules participating in transmitting mechanical signals to chondrocytes remains elusive. The pericellular matrix, a region of tissue that surrounds chondrocytes in cartilage, has been hypothesized to influence the mechanical environment of the chondrocyte in cartilage, and therefore may play a role in modulating cellular response to micromechanical factors.(27) The pericellular matrix has its unique biochemical composition. Type VI collagen is preferentially localized in the pericellular microenvironment of articular cartilage and increases during osteoarthritis.(28) It forms a “cargo-net like” organization around each chondrocyte in vitro.(29) In addition, type IX collagen, type II collagen, and aggrecan have been identified as components of pericellular matrix.(30–32) However, it is not known how these various molecules, which include both collagens and proteoglycans, form a pericellular matrix network.

Matrilins including MATN1 and -3 are among the most upregulated extracellular matrix proteins during chondrogenesis.(33,34) In human cartilage, MATN1 has been shown to distribute in a ring-like structure around chondrocytes.(35,36) In primary cell culture, MATN1 and MATN3 form pericellular filamentous networks surrounding chondrocytes.(9,37) Importantly, matrilins have been shown to interact with all the major components of pericellular matrix, including collagen types II, VI, and IX and aggrecan.(12,15) Thus, matrilins may serve as a “connector” of various matrix molecules to form an integrated pericellular matrix network. Furthermore, MATN1 has been shown to bind chondrocytes through integrins α1β1.(11) Such molecular properties make matrilins a highly attractive candidate to transmit mechanical load–induced matrix deformation signals to a chondrocyte through pericellular matrix.

To test this hypothesis, we altered the content of matrilins in pericellular matrix and quantified the effect of such alteration on mechanical responsiveness of chondrocytes in terms of cell proliferation, differentiation, and gene expression. Previously, the role of pericellular matrix molecules was studied by correlating the presence of a particular matrix molecule to its function after enzymatic or mechanical isolation of chondrons.(8,32) Alteration of the content of pericellular matrix molecules such as type VI collagen was also achieved by treatment of chondrocytes with cytokines including IL-1.(30) However, such studies were correlative in nature, because the contents of multiple matrix molecules and matrix degradation enzymes may be altered simultaneously by such treatment. In addition, very few studies were performed under mechanically active environment; thus, the role of pericellular matrix in modulating cellular response to micromechanical factors was, in large part, implied.

In this study, we systematically altered the content of matrilins with the following four approaches: (1) functional matrilins were eliminated by overexpression of a dominant negative matrilin, (2) matrilin-1 content was eliminated by using MATN1 null chondrocytes, (3) matrilin-1 content was increased by overexpressing MATN1, and (4) matrilin-3 content was increased by overexpressing MATN3.

In the first approach, we took advantage of our previous observation that a mini-matrilin that lacks the N-terminal adhesive A domain but retains the C-terminal coiled-coil domain disrupts pericellular matrilin filamentous network formation by sequestering endogenous WT matrilins.(23) Such mini-matrilin has the advantage to disrupt filament formation of all members of matrilin family, because the coiled-coil domain from one matrilin is able to associate with that from another member of matrilin family.(22) Indeed, our data have shown that mini-MATN3 expressing chondrocytes do not form pericellular MATN1 networks (Fig. 4), despite producing normal amount of MATN1 mRNA (Fig. 2) and secreting normal amount of MATN1 protein (Fig. 3). Chondrocytes that lack pericellular matrilin networks do not respond to mechanical signals by activating the Ihh signaling pathway (Fig. 5), a key mediator of mechanical regulation of cartilage growth.(5) They do not increase the rate of proliferation or the expression level of chondrocyte hypertrophic marker collagen X (Fig. 6) in response to mechanical signals. It is worth noting that, under nonload conditions, expression of mini-MATN3 in chondrocytes does not alter the expression levels of Ihh or its receptor Ptc 1 (Fig. 5), the proliferation rate of chondrocytes (Fig. 6A), or the expression level of collagen X (Fig. 6B). The regulatory effects of pericellular matrilin become apparent only under mechanical load conditions. Thus, we conclude that pericellular matrilin filaments are required for mechanical stimulation of chondrocyte proliferation and differentiation, although they do not affect the rate of chondrocyte proliferation and differentiation in the absence of external mechanical load.

In the second approach, we eliminated the content of matrilin-1 by using chondrocytes from MATN1 KO mice. MATN1 KO mice exhibit relatively normal skeletal development, normal content of other types of matrilins, and normal expression of the hypertrophic marker Col X.(24,38) This could be because of the redundancy of different matrilin forms in developing cartilage.(39) Interestingly, MATN1-null chondrocytes exhibit a diminished response to mechanical signals (Fig. 7). This suggests that MATN1 plays a role in mechanical regulation of chondrocytes. However, this function of MATN1 becomes apparent only in a mechanically active environment. Based on these data, we predict that MATN1 KO mice may develop skeletal abnormalities in vivo only under mechanically challenging conditions. Such predictions remain to be tested. On the other hand, unlike Mini-MATN3–expressing cells, MATN1-null chondrocytes still respond positively to mechanical load, although at a reduced level. This suggests that other matrilin members may compensate at least part of the function of matrilin-1.

We have shown that recombinant MATN1 partially rescued the inhibition of mechanically stimulated type X collagen levels in Mini-MATN3–expressing cells (Fig. 6C). However, MATN1 does not restore the full extent of mechanical stimulation as experienced by normal chondrocytes. This supports the conclusion that other pericellular matrilins may also be involved in chondrocyte mechanotransduction.

In the third and fourth approaches, we increased the content of MATN1 and MATN3, respectively. Unexpectedly, overexpression of either matrilin leads to a reduced mechanical sensitivity in comparison with WT chondrocytes (Figs. 5 and 6). It is not known how increased matrilin content would result in a reduction of cellular response to mechanical signals. Elevation of matrilin content increases the molecular ratio of matrilins to their matrix ligands including integrin and collagens. This increase may saturate the matrix binding sites of those ligands, thereby altering the balance of their interactions. MATN1 and MATN3 may contribute at a different extent to this process because they contain different numbers of the matrix-binding domains. Alternatively, increase of matrilin content may affect mechanical properties of the pericellular matrix. Previous studies suggest that the mechanical properties of the pericellular matrix, relative to those of the chondrocyte and extracellular matrix, may significantly influence the stress-strain, physicochemical, and fluid-flow environments of the cell.(27)

In this study, we showed that mechanical loading increases the content of matrilin-1 (Fig. 1). However, the increase of matrilin-1 content may reduce the mechanical responsiveness of chondrocytes (Figs. 5 and 6). Therefore, the increase of matrilin content, as a result of mechanical loading, may diminish further mechanical stimulation of chondrocytes. This suggests that the threshold of chondrocyte mechanical sensitivity may be decreased by mechanical loading through modification of pericellular matrix. This may form a negative feedback loop for a mechanically stimulated chondrocyte to achieve adaptation.

The content of matrilins in cartilage change greatly during aging and in osteoarthritic pathogenesis. Whereas both MATN1 and MATN3 are expressed in developing growth plate cartilage, they are diminished in adult articular cartilage.(40,41) During osteoarthritis pathogenesis, expression of both MATN1 and MATN3 is upregulated in OA cartilage.(42) In addition, the amount of matrilin-1 protein is gradually increased in permanent cartilage such as trachea during aging.(43) Intriguingly, it has shown that the mechanical properties of pericellular matrix is significantly altered in OA cartilage.(44) Our data raised a possibility that the changes of matrilin content may contribute to the alteration of mechanical sensitivity of cartilage during aging and in osteoarthritic pathogenesis through modifying pericellular matrix.

Acknowledgements

This study was supported by grants from the NIH to QC and LW and a grant from the Arthritis Foundation to QC.

Ancillary