Regulation of chondrocyte differentiation by the actin cytoskeleton and adhesive interactions


  • Anita Woods,

    1. CIHR Group in Skeletal Development and Remodeling, Department of Physiology and Pharmacology, Schulich School of Medicine and Dentistry, The University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Guoyan Wang,

    1. CIHR Group in Skeletal Development and Remodeling, Department of Physiology and Pharmacology, Schulich School of Medicine and Dentistry, The University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Frank Beier

    Corresponding author
    1. CIHR Group in Skeletal Development and Remodeling, Department of Physiology and Pharmacology, Schulich School of Medicine and Dentistry, The University of Western Ontario, London, Ontario, Canada
    • Department of Physiology and Pharmacology, University of Western Ontario, London, Ontario, Canada, N6A 5C1.
    Search for more papers by this author


Chondrocyte differentiation is a multi-step process characterized by successive changes in cell morphology and gene expression. In addition to tight regulation by numerous soluble factors, these processes are controlled by adhesive events. During the early phase of the chondrocyte life cycle, cell–cell adhesion through molecules such as N-cadherin and neural cell adhesion molecule (N-CAM) is required for differentiation of mesenchymal precursor cells to chondrocytes. At later stages, for example in growth plate chondrocytes, adhesion signaling from extracellular matrix (ECM) proteins through integrins and other ECM receptors such as the discoidin domain receptor (DDR) 2 (a collagen receptor) and Annexin V is necessary for normal chondrocyte proliferation and hypertrophy. Cell–matrix interactions are also important for chondrogenesis, for example through the activity of CD44, a receptor for Hyaluronan and collagens. The roles of several signaling molecules involved in adhesive signaling, such as integrin-linked kinase (ILK) and Rho GTPases, during chondrocyte differentiation are beginning to be understood, and the actin cytoskeleton has been identified as a common target of these adhesive pathways. Complete elucidation of the pathways connecting adhesion receptors to downstream effectors and the mechanisms integrating adhesion signaling with growth factor- and hormone-induced pathways is required for a better understanding of physiological and pathological skeletal development. J. Cell. Physiol. 213: 1–8, 2007. © 2007 Wiley-Liss, Inc.

Chondrocytes are the cellular component of cartilage, responsible for generating and maintaining the cartilaginous extracellular environment. One essential function of cartilage is to provide the intermediate template for endochondral bone formation. The initial step of cartilage template formation requires the recruitment of mesenchymal cells to future sites of skeletal development, the formation of cellular condensations, and the differentiation of mesenchymal cells to the chondrogenic lineage, a process termed chondrogenesis. Chondrocytes that continue to differentiate to hypertrophy form the cartilage intermediate that is replaced by bone. The chondrocytes remaining on either end of the mineralized bone are organized into growth plates and are responsible for longitudinal growth of long bones. Chondrocyte differentiation is marked by distinct changes in cellular morphology, correlated to specific changes in gene expression, but the molecular mechanisms connecting these processes were largely unknown. However, studies over the last years have shown that adhesion both to other cells and to the extracellular matrix (ECM) control chondrocyte differentiation through the regulation of specific intracellular signaling pathways and actin dynamics.


The development of a cartilage template is essential for proper formation of endochondral bones. The intermediary cartilage anlagen, which result from chondrogenesis, provide the template on which bone is laid down (Erlebacher et al., 1995). The control of chondrogenesis is best understood in the context of the long bones of the limbs (Fig. 1). In the developing limbs, cells originating from the lateral plate mesoderm are stimulated to aggregate (‘condense’) without an increase in proliferation, creating increased cell density and cell–cell interactions (Cohn and Tickle, 1996). These interactions are most likely involved in propagating signal transduction events. In mouse development, this is a mid embryonic event, occurring at 10.5–12.5 days post-coitum (dpc) (Karsenty, 1999). Culture systems have been developed that promote chondrogenesis. The three-dimensional micromass system is an excellent culture model providing sufficiently high cell density that is required for chondrogenesis as well as a method to reduce the interactions of cells with cell culture plastic (Ahrens et al., 1977; DeLise et al., 2000). Instead, cells undergo cell–cell interactions and morphological changes that resemble in vivo chondrogenesis.

Figure 1.

Chondrogenesis. The process of chondrogenesis begins with the stimulation of mesenchymal cells to condense. Cells within these condensations then differentiate into chondrocytes and start to proliferate. Collectively, these processes determine the size and location of future skeletal elements.

Mesenchymal cells within the condensations express cell adhesion molecules such as N-cadherin (DeLise and Tuan, 2002) and neural cell adhesion molecule (N-CAM) (Hall and Miyake, 2000). The level of chondrogenic differentiation is directly correlated to the density of the initial condensation (Ahrens et al., 1977). Cells undergoing chondrogenesis acquire a distinct spherical cell morphology and initiate expression of the transcription factors Sox9 (Bi et al., 1999), Sox5, and Sox6, which regulate the genes encoding the ECM molecules Collagen II and Aggrecan (Lefebvre et al., 1997; Sekiya et al., 2000; Smits et al., 2001). The ECM produced by differentiated chondrocytes is essential for the formation of the future bone and maintains and regulates the chondrocyte phenotype. Collagen II provides tensile strength to the cartilaginous matrix and is important in the establishment of temporal and spatial organization with other matrix components such as the main proteoglycan, Aggrecan (Doege, 1999). Aggrecan is heavily modified by sulfated glycosaminoglycans (GAGs), attracts numerous water molecules and forms large aggregates in cartilage (Hascall and Heinegard, 1974; Krueger et al., 1990). Aggrecan and other proteoglycans provide a cushioning role of the matrix, but also act to immobilize and store growth factors and thereby function as molecular organizers of the ECM and cartilage in general (Rouslahti and Yamaguchi, 1991). The importance of these pathways for skeletal development is shown by several human diseases caused by mutations in the genes encoding Sox9, Collagen II, and Aggrecan, or regulators of GAG sulfation (Eyre et al., 1986; Spranger et al., 1994; Vertel et al., 1994; Wagner et al., 1994; Chan et al., 1995; Hastbacka et al., 1996; Bi et al., 2001; Rossi and Superti-Furga, 2001; Gleghorn et al., 2005).

Chondrocytes mature towards two different fates after commitment to the chondrogenic lineage; they will remain as chondrocytes (persistent cartilage) or differentiate to hypertrophic chondrocytes (transient cartilage). The chondrocytes that stop differentiating form the persistent cartilage located on the articular surfaces in joints. These chondrocytes do not differentiate further or mineralize their ECM (Trippel et al., 1980). Chondrocytes that continue to differentiate to the hypertrophic state contribute to the formation of the growth plate (Beier, 2005).

Endochondral Ossification and Growth Plate Development

Chondrocytes within the cartilage anlagen proliferate, and the most central chondrocytes differentiate to hypertrophy. Hypertrophic chondrocytes increase in cell size, downregulate the expression of Collagen II and instead initiate expression of Collagen X (Reichenberger et al., 1991). Hypertrophic chondrocytes mineralize their surrounding matrix and undergo apoptosis (Shapiro et al., 2005). Vascular tissue is stimulated to invade this region, allowing the entrance of osteoclast and osteoblast precursors, which remodel the remaining hypertrophic matrix and lay down bone tissue (Gerber et al., 1999). This process of replacement of hypertrophic cartilage matrix in the cartilage center is termed primary ossification. The cartilage segments that remain on either side of the primary ossified region are termed the growth plates and are responsible for the longitudinal growth of long bones (van der Eerden et al., 2003).

The chondrocytes of the growth plate organize into distinctive regions of resting/reserve, proliferative, and hypertrophic chondrocytes (Fig. 2). Chondrocytes of each region display distinctive behavior, cell shape, and functions. The resting/reserve zone is a region of rounded chondrocytes, which are spaced further apart and express high levels of Collagen II and Aggrecan (Abad et al., 2002). The chondrocytes of the proliferative region undergo strictly regulated, unidirectional proliferation, resulting in highly organized, columnar structures (van der Eerden et al., 2003). Chondrocytes produce ECM predominantly in the vertical direction, generating thickened longitudinal septae that contribute to the generation of the columns (Hunziker, 1994). After cells exit the cell cycle, they differentiate to hypertrophy (Beier, 2005). Besides the hallmarks of enlarged cell shape and Collagen X expression, hypertrophic chondrocytes express the transcription factor Runx2 (Takeda et al., 2001), the matrix molecule bone sialoprotein (BSP) (Chen et al., 1991) and secreted factors such as vascular endothelial growth factor (VEGF) and matrix metalloproteases 9 and 13 (MMP9, MMP13). The metalloproteases function to degrade the ECM during the remodeling process to allow for hypertrophic enlargement, proper vascularization and ossification (Mwale et al., 2002; Tchetina et al., 2003). The remaining ECM provides the template for osteoblasts to adhere and deposit the matrix creating the bone (Karsenty and Wagner, 2002; Stickens et al., 2004).

Figure 2.

Endochondral Ossification. Endochondral ossification requires the formation of a transient cartilage template. The process of chondrogenesis results in formation of the initial cartilage template. Chondrocytes in the most central region of the template differentiate to the terminal stage of the hypertrophic chondrocyte. Chondrocytes located between the resting/reserve zone and the hypertrophic zone proliferate in an unidirectional manner, resulting in characteristic columns. Hypertrophic chondrocytes mineralize the matrix surrounding them, undergo apoptosis, and hypertrophic cartilage is invaded by blood vessels. The mineralized cartilage is degraded by osteoclasts and replaced with bone tissue through the activity of osteoblasts. The regions on either side of the bone tissue are termed the growth plates and responsible for longitudinal growth.

The processes of chondrocyte proliferation and differentiation to hypertrophy must be strictly regulated in order to generate bone of normal length (Hunziker, 1994; Ballock and O'Keefe, 2003). The volume increase of hypertrophic chondrocytes and the number of proliferative cycles a chondrocyte undergoes control longitudinal growth of the long bones (Hunziker, 1994; Wilsman et al., 1996a,b). Disruptions of this intricate balance commonly result in shortened and/or misshaped bones.

It is widely accepted that distinct cell shape is indicative of the chondrocyte phenotype. Therefore, the study of molecular mechanisms that control cell shape (such as regulation of the cytoskeleton, cell–cell interactions and cell–matrix signaling) provides an intriguing approach to study the regulation of chondrocyte differentiation.

Contribution of the Cytoskeleton

Evidence for a role of the cytoskeleton in chondrocyte differentiation has accumulated over the last 20 years (Daniels and Solursh, 1991). Although there is likely a role for the other components (microtubules, intermediate filaments) of the cytoskeleton in chondrocyte differentiation (Farquharson et al., 1999; Capin-Gutierrez et al., 2004; Blain et al., 2006), most work has focused on the contributions of the actin cytoskeleton to chondrogenesis and hypertrophic differentiation.

It is well known that maintenance of chondrocytes in monolayer culture results in loss of the differentiated phenotype. Chondrocytes plated in a monolayer culture tend to change morphology to flattened cells and to cease production of Collagen II and GAGs (von der Mark et al., 1977; Grundmann et al., 1980). It was discovered that the addition of dihydrochalasin B or cytochalasin D, inhibitors of actin polymerization, stimulated rounding of de-differentiated chondrocytes (Zanetti and Solursh, 1984; Benya et al., 1988; Brown and Benya, 1988; Loty et al., 1995). The rounded cells display a return to chondrogenic gene expression, as exhibited by increases in Collagen II and GAG production (Benya et al., 1988; Brown and Benya, 1988; Benya and Padilla, 1993). However, the mechanisms responsible for this reversion had not been determined. In addition, murine embryonic stem cells can be stimulated to differentiate into chondrocytes by treatment with cytochalasin D as noted by expression of the chondrogenic genes Sox9, Collagen II, and Aggrecan (Zhang et al., 2006).

Recent data suggest that actin dynamics not only control chondrogenesis, but also growth plate physiology. Adservin, an actin-binding protein, is upregulated during hypertrophy, and overexpression of this molecule results in rearrangement of the actin cytoskeleton, increase in cell volume, and upregulation of hypertrophic markers (Nurminsky et al., 2007). Importantly, several human chondrodysplasias have been linked to mutations affecting the actin cytoskeleton. Mutations in the genes encoding the actin modifying proteins Filamin A and B have been linked to a multitude of human skeletal disorders (Table 1) such as frontometaphyseal dysplasia, spondylocarpotarsal syndrome, Larsen's syndrome, and atelosteogenesis type I and III, further suggesting that the actin cytoskeleton contributes to proper cartilage and bone development and growth plate function (Robertson et al., 2003; Feng and Walsh, 2004; Krakow et al., 2004; Farrington-Rock et al., 2006; Robertson et al., 2006). Our preliminary microarray analyses also suggest that cytochalasin D treatment of primary chondrocytes in monolayer culture results in enhanced expression of hypertrophy-specific genes (Woods et al., unpublished). Finally, the human disease Hereditary Multiple Exotoses is characterized by stellate-shaped chondrocytes with large aggregates of actin bundles (Bernard et al., 2000). The increased levels of microfilaments are proposed to increase cell adhesiveness and to interfere with binding of signals to receptors, contributing to the phenotype of inappropriate chondrocyte proliferation (Duke et al., 2002).

Table 1. Gene mutations affecting chondrocyte adhesion signaling
GenePhenotype/diseaseExperimental model/mutationReferences
α1 IntegrinNo growth plate abnormalities, osteoarthritisKnock-outZemmyo et al. (2003)
α10 IntegrinGrowth retardation, increased apoptosis, abnormal cell shapeKnock-outBengtsson et al. (2005)
β1 IntegrinDisorganized growth plate, unorganized proliferative columns, reduced proliferation, abnormal cell shape, decreased adhesionTissue-specific knock-outAszodi et al. (2003)
DDR2Chondrodysplasia, reduced proliferation of growth plate chondrocytesKnock-outLabrador et al. (2001)
Filamin AFrontometaphyseal dysplasia, X-linked, abnormally shaped tubular bonesMissense mutations and small deletions in Human diseaseRobertson et al. (2006)
Filamin BAtelogenesis I, III; autosomal dominant, disharmonious skeletal maturation, joint dislocationsSpondylocarpotarsal dysplasia; autosomal recessive, disrupted vertebral segementationHuman DiseaseMissense mutations, point mutations in the actin binding domainHuman diseasemissense mutations in CH2 domain, premature stop codon mutationsKrakow et al. (2005); Farrington-Rock et al. (2006)
ILKChondrodysplasia, reduced chondrocyte proliferation, abnormal cell shape, decreased adhesionTissue-specific knock-out

Grashoff et al. (2003

); Terpstra et al. (2003)

N-cadherinNo phenotypeKnock-outLuo et al. (2005)
Rac1Short bones, disorganized hypocellular growth plates, decreased adhesion and proliferationTissue-specific knock-outWang et al., submitted

Because of these established roles of the actin cytoskeleton, investigations into the upstream regulators of cytoskeletal dynamics in chondrocyte physiology were initiated by us and others. Rho GTPases are molecular switches in the cell that activate downstream effectors when bound to GTP and are inactivated by the hydrolysis to GDP. Rho GTPase activity is highly regulated by interaction with other proteins and post-translational modifications (Nobes and Hall, 1994; Kwon et al., 2000; Ridley, 2001; Sawada et al., 2001; Ellerbroek et al., 2004). Of the more than 20 Rho GTPases family members, the three most intensively studied molecules are RhoA, Rac1, and Cdc42. RhoA, Rac1, and Cdc42 control the formation of stress fibers, filopodia, and lamellipodia, respectively (Nobes and Hall, 1995; Amano et al., 1997; Wittman and Waterman-Storer, 2001). Rho GTPases regulate the actin cytoskeleton via multiple kinases (such as ROCK, LIM kinases, and PAK), which in turn control the activity of actin modifying proteins such as Cofilin, Profilin, IQGAP, and Phosphatidylinositol (PIP2) (Machesky and Insall, 1998; Miki et al., 1998; Maekawa et al., 1999; Rohatgi et al., 1999; Tolias et al., 2000; Noritake et al., 2004; DesMarais et al., 2005). Additional downstream targets of Rho GTPases include WAVE and WASP proteins and the Arp2/3 complex that control nucleation of actin polymerization. Rho GTPases thus control organization, polymerization, and depolymerization of actin through multiple pathways.

RhoA signaling through its main effector ROCK, and possibly other mediators, stimulates stress fiber formation and promotes elongated cell shape. We showed that RhoA/ROCK signaling results in the inhibition of chondrogenesis by inhibiting expression of the essential transcription factor Sox9 (Woods et al., 2005; Woods and Beier, 2006). Similarly, chondrocytes isolated from articular cartilage de-differentiate in monolayer culture, but inhibition of ROCK signaling leads to superinduction of Sox9 expression and a rescue of the chondrocyte phenotype (Tew and Hardingham, 2006). In addition, our data show that RhoA/ROCK signaling also regulates expression of Sox5 and Sox6 as well as phosphorylation of Sox9 (Woods and Beier, 2006). Interestingly, the effects of this pathway on Sox gene expression are dependent on the culture model, highlighting the importance of three-dimensional context in the maturation of chondrocytes.

RhoA/ROCK signaling also inhibits the differentiation of chondrocytes to hypertrophy (Wang et al., 2004). In addition, RhoA signaling promotes proliferation of chondrocytes, likely because of increased expression of Cyclin D1, an essential regulator of chondrocyte proliferation (Beier et al., 2001; Beier, 2005). Furthermore, inhibition of RhoA/ROCK signaling in organ culture models results in impaired growth plate function (Woods et al., unpublished work). Overall, these data demonstrate that RhoA signaling inhibits multiple stages of chondrocyte differentiation and promotes chondrocyte proliferation.

In contrast, we demonstrated recently that Rac1 promotes chondrogenesis through the stimulation of N-cadherin expression and Sox gene expression, and Cdc42 promotes chondrogenesis through the regulation of Sox9 expression (Woods et al., submitted). Rac1 and Cdc42 signaling also promote the differentiation of chondrocytes to hypertrophy, likely exerting their effects through p38 MAP kinases (Wang and Beier, 2005). In addition, both Rac1 and Cdc42 signaling inhibit proliferation of chondrocytes when overexpressed in cell culture. Furthermore, cartilage-specific ablation of the Rac1 gene results in shortened long bones and disorganized, hypocellular growth plates (Wang et al., 2007). Chondrocytes isolated from Rac1-deficient growth plates display reduced adhesion to Collagen II and Fibronectin, abnormal cell shape and actin organization, demonstrating disrupted cell-ECM interactions. Thus, each of the three prototype Rho GTPases regulates multiple stages of chondrocyte differentiation in a unique manner.

Cell–Cell Interactions

In addition to signaling by growth factors, adhesion of cells to each other provides important information and determines cellular responses (Takeichi, 1988; Leckband and Prakasam, 2006). Cellular adhesion molecules, such as cadherins, immunoglobins, and selectins, mediate cell–cell interactions (Juliano, 2002). These adhesion molecules are typically transmembrane proteins that link the extracellular and intracellular environments via scaffolding, effector, and adaptor proteins that often connect receptors to the actin cytoskeleton (Aberle et al., 1996; Brown and Sacks, 2006).

As outlined above, chondrogenesis relies heavily on cell–cell interactions. In chondrocyte biology, the most intensively studied adhesion molecules are N-cadherin and N-CAM, an immuoglobin family protein. N-cadherin, a member of the classical cadherin family (Takeichi et al., 1990; Gumbiner, 1996), binds in a homophilic, calcium-dependent manner (Takeichi, 1991). Intracellularly, cadherins interact with catenins that function to link cadherins to the actin cytoskeleton (Aberle et al., 1996), which is a requirement for functional adherens junctions. N-cadherin is highly expressed in limb development during the stages of cellular condensation (Oberlender and Tuan, 1994). Inhibition of N-cadherin by a neutralizing antibody in cellular condensations results in the inability of mesenchymal cells to condense in vitro and in vivo and therefore inhibits subsequent chondrogenesis (Oberlender and Tuan, 1994). Surprisingly, a knock-out model of N-cadherin does not demonstrate a skeletal phenotype; however, it is suggested that there may be molecular compensation by a closely related cadherin family member, cadherin-11, which is also expressed during chondrogenesis (Luo et al., 2005). It has been determined in other cell types that cadherin function is regulated by RhoA and Rac1 (Braga et al., 1997; Takaishi et al., 1997). On the other hand, Rho GTPases act downstream of cadherins, providing a link between cadherin signaling and actin organization as well as other cellular responses (Charrasse et al., 2002; Nelson et al., 2004). We recently provided evidence for another level of interaction between cadherins and Rho GTPases by showing that Rac1, but not RhoA or Cdc42, stimulates expression of N-cadherin during chondrogenesis (Woods et al., submitted). Overall, the relation between cadherins and Rho GTPases is therefore complex, reciprocal, and cell type-specific.

N-CAM is a glycoprotein that binds in a homophilic or heterophilic fashion to other cell surface adhesion molecules or to the matrix (Cunningham et al., 1987; Edelman et al., 1987). During prechondrogenic condensations, cells increase expression of N-CAM (Tavella et al., 1994). N-CAM is suggested to function in initiating and stabilizing cellular condensations but may not directly contribute to the differentiation to chondrocytes (Fang and Hall, 1999).

Cell–Matrix Interactions

As chondrocytes mature, they secrete vast amounts of matrix, and the reliance on signaling switches from cell–cell to cell–matrix interactions (Goggs et al., 2003). Adhesion receptors such as integrins bind a wide range of extracellular components such as ECM molecules and other cell surface proteins (ffrench-Constant and Colognato, 2004). Integrin attachment stimulates the formation of focal adhesion complexes, an intracellular protein complex that transduces signals from the ECM to intracellular effectors such as the cytoskeleton (Parsons et al., 1994; Clancy et al., 1997). It should be noted that integrin signaling has been studied extensively in articular chondrocytes and has been implicated in cartilage homeostasis and arthritis (reviewed in Loeser, 2002; Millward-Sadler and Salter, 2004). However, this review focuses on developmental aspects of ECM signaling in chondrocytes.

Twenty-four unique integrin dimers form in vivo from the 18α a subunits and 8β subunits found in mammalian cells (Hynes, 2002). The composition of the ECM regulates which integrin heterodimers are utilized and expressed in a given cell type (Stupack and Cheresh, 2002). Integrins can regulate cell shape by directing regions of ligand binding (van der Flier and Sonnenberg, 2001). In addition, intracellular cell signals can regulate the affinity of the integrin heterodimers for extracellular ligands (Arnaout et al., 2005). Chondrocytes express a subset of integrin subunits including Fibronectin receptors (α5β1, αnβ3, αnβ5), a Laminin receptor (α6β1), and Collagen receptors (α1β1, α2β1, α10β1) (Shakibaei et al., 1995; Loeser, 2000; Loeser, 2002; Egerbacher and Haeusler, 2003).

The β1 chain is therefore a component of most chondrocyte integrins. Thus, it comes as no surprise that cartilage-specific inactivation of the β1 integrin gene results in a severe cartilage phenotype (Aszodi et al., 2003). Knock-out mice develop chondrodysplasia, characterized by distorted growth plates demonstrating unorganized proliferative columns and abnormal cell shape due to loss of adhesion to Collagen II. Chondrocytes from these mice display abnormal cell shape, reduced proliferation, and deregulated expression of cell cycle proteins including D-type cyclins and cyclin-dependent kinase inhibitors.

In vitro experiments also suggest that loss of β1 integrin and αnβ5 integrins promotes apoptosis in growth plate chondrocytes (Wang and Kirsch, 2006) and that blocking antibodies against β1, α2, or α3 integrins repress hypertrophic differentiation and decrease chondrocyte survival (Hirsch et al., 1997). Growth plate chondrocytes also express α5β1 integrin, with the highest expression in proliferative and hypertrophic chondrocytes. Blocking α5β1 interactions of growth plate chondrocytes results in reduced chondrocyte proliferation (Enomoto-Iwamoto et al., 1997). In addition, α5β1 integrin has been determined to be involved in the formation of joints in the appendicular skeleton. Missexpression of α5β1 leads to the fusion of joints and results in premature formation of pre-hypertrophic chondrocytes (Garciadiego-Cazares et al., 2004).

Knock-out of the α10 integrin gene also resulted in growth plate dysfunction and growth retardation (Bengtsson et al., 2005). The loss of this collagen-specific integrin causes abnormal cell shape of chondrocytes and increased rates of apoptosis. In contrast, although integrin α1 has been shown to be expressed primarily in hypertrophic chondrocytes, knock-out mice for α1 integrin exhibit osteoarthritis but no growth plate abnormalities (Zemmyo et al., 2003).

As aforementioned, integrin signaling stimulates the formation of signaling complexes made up of multiple proteins, which regulate multiple cellular processes such as organization of the cytoskeleton (Schwartz et al., 1995). Included within these signaling complexes are scaffolding proteins such as Talin, Paxillin and alpha-Actinin, and kinases including FAK and integrin-linked kinase (ILK) (Lo, 2006). Mice lacking ILK in cartilage display chondrodysplasia as well as abnormal chondrocyte cell shape, defective cell adhesion, and reduced chondrocyte proliferation (Grashoff et al., 2003; Terpstra et al., 2003). This phenotype resembles that of mice with a cartilage-specific disruption of the b1 integrin (Aszodi et al., 2003) or Rac1 (Wang et al., submitted) genes, suggesting that all three genes act in a common pathway.

Integrins are not the only ECM receptors in cartilage. The discoidin domain receptors (DDR) are members of a subfamily of tyrosine kinase receptors whose ligands are collagens (Shrivastava et al., 1997; Vogel et al., 1997; Leitinger and Kwan, 2006). Mice lacking DDR2 exhibit dwarfism due to decreased proliferation of growth plate chondrocytes (Labrador et al., 2001). Increased expression of DDR2 occurs in articular chondrocytes in mice with osteoarthritis, suggesting a role for this receptor in the pathogenesis of this disease (Xu et al., 2005). The related family member, DDR1 is also a receptor for collagens, but is not expressed in chondrocytes, nor does the knock-out result in a skeletal phenotype (Gross et al., 2004). Annexin V is another receptor for collagen (von der Mark and Mollenhauer, 1997), specifically for a fragment of Collagen II (the N-telopeptide) in articular chondrocytes (Lucic et al., 2003). Treatment of chondrocytes with collagen fragments leads to decreased chondrocyte matrix synthesis, suggesting a physiological role for Annexin V in chondrocytes (Jennings et al., 2000). It is also suggested that Annexin V is involved in regulating apoptosis and matrix mineralization of growth plate chondrocytes (Kirsch, 2005; Wang and Kirsch, 2006).

In addition to the cell–cell adhesions discussed above, cell–matrix interactions are also involved in chondrogenesis. CD44 is a cell surface glycoprotein that acts as a receptor for collagens and Hyaluronan. During chondrogenesis, CD44 expression increases, as well as expression of some hyaluronidases, suggesting a role of Hyaluronan signaling and turnover in chondrogenesis (Nicoll et al., 2002). Furthermore, it is suggested that CD44 establishes and organizes the pericellular matrix in chondrocytes, and disruption of this function may alter chondrocyte survival (Knudson, 2003).

Fibronectin and Laminins are important to embryogenesis and regulate chondrogenesis (Tavella et al., 1997; Hashimoto et al., 2005). A functional splice variant of Fibronectin is highly expressed just prior to condensations and is downregulated just after condensation (Gehris et al., 1997). Fibronectin expression is regulated by transforming growth factor-b (TGF-β), and together these molecules control the initiation of condensations (Roark and Greer, 1994). Fibronectin also acts to activate the expression of N-CAM (Chimal-Monroy and Diaz de L, 1999). Mesenchymal condensations also require interactions with the ECM molecule Tenascin (Gehris et al., 1997; Hall and Miyake, 2000) for chondrogenic differentiation. Tenascins stimulate signaling through FAK and Paxillins which has been demonstrated to promote the differentiation to fully differentiated chondrocytes (Bang et al., 2000; Vinall et al., 2002). Conversely, tension force inhibits chondrogenesis and stimulates expression of both β1 integrin and FAK. Recent studies suggest that the inhibition of chondrogenesis by tension is mediated by FAK and integrins β1, α2, and α5 (Takahashi et al., 2003; Onodera et al., 2005). Similarly, plating of mesenchymal stem cells on RGD-containing peptides inhibits chondrogenesis which is mediated by the integrins β1 and αvβ3 (Connelly et al., 2007). In contrast, blockade of integrin β1 function was shown to result in inhibition of chondrogenesis and chondrocyte matrix production as well (Shakibaei, 1998). Thus, integrins (in particular β1 integrin) have been shown to both promote and inhibit chondrogenesis, dependent on the context. In vivo studies will therefore be required to obtain a complete understanding of integrin function in early chondrogenesis.


Regulation of cell shape and signaling from cell–cell and cell–ECM interactions are essential to the maturation of chondrocytes. Determination of signaling pathways and targets downstream of these events will aid in the better understanding of musculoskeletal pathologies, the development of novel strategies for cartilage replacement and new approaches for regeneration, prevention, and treatment of cartilage disorders. In addition, elucidation of the mechanisms integrating signals from adhesion receptors with those from growth factor and hormone receptors will contribute greatly to a better understanding of physiological and pathological endochondral ossification. Sophisticated approaches to gene mutagenesis in mice in conjunction with advanced genomic, proteomic, and imaging tools will provide a strong platform for rapid progress in these areas over the next years.