SEARCH

SEARCH BY CITATION

Keywords:

  • skeletal development;
  • integrins;
  • integrin signaling;
  • cartilage;
  • synovial joint;
  • articular cartilage;
  • bone;
  • tendon;
  • mesenchymal stem cell;
  • knockout mice

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

Integrins are cell surface receptors that connect extracellular matrix (ECM) components to the actin cytoskeleton and transmit chemical and mechanical signals into the cells through adhesion complexes. Integrin-activated downstream pathways have been implicated in the regulation of various cellular functions, including proliferation, survival, migration, and differentiation. Integrin-based attachment to the matrix plays a central role in development, tissue morphogenesis, adult tissue homeostasis, remodeling and repair, and disturbance of the ECM-integrin-cytoskeleton signaling axis often results in diseases and tissue dysfunction. Increasing amount of in vitro and in vivo evidences suggest that integrins are pivotal for proper development, function, and regeneration of skeletal tissues. In this paper, we will summarize and discuss the role of integrins in skeletogenesis and their influence on the physiology and pathophysiology of cartilage, bone, and tendon. Birth Defects Research (Part C) 102:13–36, 2014. © 2014 Wiley Periodicals, Inc.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

The skeletal system of vertebrates originates from the mesoderm via a complex morphogenetic process and consists of bone, cartilage, joints, ligaments, and tendons which, together with the musculature, provide support for the body, protect vital organs, and allow movement. Skeletal tissues are adopted for specific mechanical demands through distinct organization of their hierarchical composite material, consisting of a network of collagen fibrils embedded into a heterogeneous mixture of extracellular matrix (ECM) molecules. In bone, type I collagen fibrils representing about 90% of the organic matrix and the inorganic hydroxyapatite crystals act together to give the tissue its hardness and fracture toughness. Tendons and ligaments are mainly composed of parallelly oriented collagen I fibrils and a smaller fraction of elastin embedded in a proteoglycan–water matrix endowing the tissue with high tensile force and resilience, meanwhile preventing damage and disconnection of the fibers under stress conditions. In the hyaline cartilage, the meshwork of heterotypic collagen II/IX/XI fibrils supplies cartilage with its tensile strength, whereas the hydrated glycosaminoglycan chains of proteoglycans generate an osmotic swelling pressure that resists compressive forces. The composition of skeletal matrices is reflected in the synthetic activities of their resident cell types, osteoblasts, tenocytes/ligamentocytes, and chondrocytes, and is maintained with the help of carefully balanced anabolic and catabolic processes. Structural and compositional changes induced by aging, environmental, or genetic factors result in skeletal diseases that typically involve various degrees of motion deficits with tremendous socioeconomic impact. Injuries are another challenge of the skeletal system and bone–tendon–cartilage correlating with the blood and nerve supply presents a continuum of good-to-bad-to-non-healer tissues. Skeletal differentiation, tissue function, and repair critically depend on the continuous interactions between cells and the ECM. Integrins are the major class of adhesion and signaling receptors on the surface of skeletogenic cells, and their role for numerous aspects of skeletal tissue properties is increasingly recognized. In this paper, we overview integrin expression and function in cartilage, bone, and tendon, and summarize the genetically engineered mouse strains carrying mutations in genes encoding integrins or integrin-interacting proteins that significantly advanced the knowledge of the skeletal research field (Table 1).

Table 1. Mouse Models with Gene Modifications for Integrins and Integrin-Interacting Molecules
GeneModelPhenotypeReferences
Cartilage   
Itgb1Col2a1creLethal chondrodysplasia, chondrocyte shape change, proliferation and cytokinesis defects, abnormal ECMAszodi et al. (2003)
Itgb1Prx1creDisorganization of the AC, chondrocyte proliferation defect, abnormal ECMRaducanu et al. (2009)
Itga1NullAccelerated development of knee OAZemmyo et al. (2003)
Itga10NullMild dwarfism, moderate chondrocyte shape change and proliferation defectBengtsson et al. (2005)
Adam15NullAccelerated development of knee OABöhm et al. (2005)
FlnbNullStunted growth, delayed hypertrophy, increased apoptosis, diminished phosphorylation of β1 integrinLu et al. (2007)
IlkCol2a1creLethal chondrodysplasia, proliferation defect, mild chondrocyte shape changeGrashoff et al. (2003); Terpstra et al. (2003)
MIA/CD-RAPNullMild collagen network abnormalitiesMoser et al. (2002)
  AC disorganization, enhanced repair in injury OA modelSchmid et al. (2002)
Bone   
Itgb1TransgenicImpaired intramembranous bone formationZimmermann (2000)
  Decreased skeletal mass and bone strengthGlobus et al. (2005)
Itgb1Col1a1creMild hystomorphometric defectsBentmann et al. (2010)
  Increased stiffness and strength during unloadingPhilips et al. (2008)
Itgb3NullOsteopetrosis, dysfunctional osteoclastsMcHugh et al. (2000)
  Accelerated early phase healing of tibial fracturesHu et al. (2010)
Itga1NullNormal skeleton, impaired fracture healingEkholm et al. (2002)
Itgb5NullIncreased bone loss in ovariectomized miceLane et al. (2005)
Itga2NullProtection against age-associated bone deteriorationStange et al. (2013)
Itga5InducedEnhanced bone formation and repairSrouji et al. (2012)
FakCol1a1creNormal skeletal development, delayed bone healingKim et al. (2007)
IcapNullImpaired osteoblast differentiation and mineralizationBouvard et al. (2007)
IlkCol1a1creTransient increase in bone volumeEl-Hoss et al. (2014)
Kindlin-3NullSevere osteopetrosis, defect in osteoclast activationSchmidt et al. (2011)
Tendon   
Itga7NullMyotendinous junction defectsMayer et al. (1997)
Itga11NullDisorganized periodontal ligaments, tooth eruption defectPopova et al. (2007)
IlkHsaCreMyotendinous junction defectsWang et al. (2008)
Talin-1HsaCreMyotendinous junction defectsConti et al. (2008)

Integrins

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

Integrins are heterodimeric transmembrane proteins consisting of α and β subunits (Legate et al., 2009). There are 18 α and 8 β subunits, which can combine into at least 24 different heterodimers and bind ECM components or interact with counter-receptors on adjacent cells. Integrin receptors usually bind to various ligands and, conversely, many ECM or cell surface adhesive molecules are able to interact with multiple integrin heterodimers. Traditionally, integrins can be classified into four main categories, based on the molecular properties of integrin–ligand interactions (Humphries et al., 2006; Campbell and Humphries, 2011). RGD-binding integrins, α5, α8, αv, and αIIb in pairs with β1 or β3, interact with the arginine–glycine–aspartate tripeptide motif found in many ECM molecules, such as fibronectin, tenascin, and vitronectin. Eight integrin heterodimers, including the four members of the β2 subfamily, bind to an RGD-related LDV or LDV-homologous site. Four α integrins (α1, α2, α10, and α11), which couple with the β1 subunit, are distinct collagen/laminin receptors carrying an “inserted” or “αI domain”. The I-domain, which is present in five additional α subunits, contains a metal ion-dependent adhesive site known to be crucial for divalent cations-dependent ligand binding. Finally, integrins α3β1, α6β1, α7β1, and α6β4 are highly selective laminin-binding heterodimers without αI-domain.

Integrin occupancy and clustering by the substrate supports cell adhesion and is essential for embryonic development, tissue remodeling and repair, host defense, and hemostasis. Integrins connect the ECM to the intracellular actin cytoskeleton, providing the mechanical basis for anchorage, cell shape determination, force transmission, and migration. Furthermore, integrins transmit chemical and biomechanical signals into the cells (outside-in signaling) through adhesion complexes, which are multiprotein platforms around the integrin cytoplasmic tails and are composed of numerous enzymes, adaptor, and actin-binding molecules. The diverse signaling mechanisms triggered by ligand-binding includes enzyme phosphorylation (e.g., focal adhesion kinase-FAK; mitogen-activated protein-MAP kinases), calcium influx, and activation of Rho GTPases (e.g., RhoA, Rac1, and Cdc42), which all together influence nearly all aspects of cell physiology (Legate et al., 2009). To effectively sense changes in the cellular microenvironment, integrin–ligand-binding and subsequent signal transduction are dynamically regulated via the control of integrin activation involving a series of events referred as inside–out signaling. Thus, integrins are unique transmembrane receptors which utilize a bi-directional signaling machinery to fulfill demands of cell differentiation and tissue organization. Moreover, integrins interact and share common signaling pathways with growth factor receptors to modulate a variety of cellular functions. At least 12 integrin receptors contain the β1 subunits, thus forming the biggest integrin subfamily. β1 integrins are crucial for development, since ablation of the β1 integrin gene in mice results in peri-implantation lethality owing to inner cell mass failure (Fassler and Meyer, 1995; Bouvard et al., 2001).

In the recent decade, our increasing understanding of the central role of integrins in governing various processes important for cell behavior has directed basic research and the pharmaceutical industry to develop strategies targeting integrin signaling to treat cancer, ischemic stroke, inflammatory and autoimmune diseases, and other pathological conditions (Cantor et al., 2008). For instance, natalizumab, a humanized monoclonal antibody against α4 integrin was approved by FDA in 2006 for the treatment of relapsing multiple sclerosis. In skeletal conditions, especially in osteoarthritis and osteoporosis, the beneficial role of the modulation of integrin adhesion/signaling is less understood and it may involve both activation and inactivation of particular integrin subunits.

Cartilage Development and Function

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

A major part of the mammalian skeleton is laid down by a multi-step process called endochondral bone formation (Aszodi et al., 2000; Kronenberg, 2003). During embryogenesis, first cartilaginous templates of the future bones are formed by the condensation of skeletogenic mesenchymal stem cells (MSCs). In precartilaginous condensations, under the control of TGFβ and the canonical Wnt/β-catenin signaling, skeletogenic cells stop to proliferate and express adhesive molecules such as N-cadherin, N-CAM, and tenascin-C, which support their aggregation. The resulting molds are surrounded by a mesenchymal sheet (perichondrium) and consist of immature, proliferative chondrocytes, which synthesize a cartilage-specific ECM composed of type II collagen fibrils, proteoglycans (mainly aggrecan), and non-collagenous glycoproteins. Chondrocyte early differentiation is tightly controlled by the transcription factors Sox trio (Sox5/Sox6/Sox9) and growth factor signaling (BMPs, FGFs, and TGFβ). In the next step, fully differentiated chondrocytes produce a vast amount of cartilage-specific matrix, then centrally located cells stop to proliferate, mature, and differentiate into hypertrophic chondrocytes, which express collagen type X. The terminal differentiation of hypertrophic chondrocytes is accompanied by ECM mineralization, invasion of blood vessels, chondroclasts, and osteoblast precursors leading, to the replacement of the cartilage scaffold by bone. The program of proliferation, maturation, and remodeling takes place within a specialized structure, called growth plate. Growth plate chondrocytes form horizontal zones reflecting their differentiation stage (resting, proliferative, maturation, and hypertrophic), whereas the flattened proliferative cells are arranged into vertical columns. Proliferation, matrix production, and hypertrophy of growth plate chondrocytes are the major factors, which are essential to achieve longitudinal elongation of endochondral bones until the end of puberty (Hunziker, 1994). Abnormalities of cartilage and bone development result in genetic, congenital conditions, such as dysostoses and skeletal dysplasias. Dysostoses are skeletal malformations occurring in a subset of skeletal elements (e.g., fingers), whereas the more than 200 skeletal dysplasias (e.g., chondrodystrophies, osteodystrophies, dwarfisms) lead to abnormal size and shape of the skeleton and disproportion of the long bones, spine and head (Warman et al., 2011).

Chondrogenesis, Integrin Expression, and In Vitro Studies

Immunohistochemical studies on fetal limb cartilages have revealed that chondrocytes, depending on their location, differentially express integrin receptors for cartilage matrix ligands such as α1β1, α2β1, and α10β1 (for collagen II), α5β1, αvβ3, and αvβ5 (for fibronectin), and α6β1 (for laminin). In human knee, α5β1, α6β1, and αvβ5 integrins were found in the growth plate, epiphyseal, and the forming articular cartilage (AC), while the α1β1 and α2β1 were absent in growth plate chondrocytes (Salter et al., 1995). The distribution of three collagen-binding integrins (α1β1, α2β1, and α10β1) was analyzed in detail in the developing mouse limb (Camper et al., 2001). The integrin subunit α10 was found to be co-localized with collagen II in the forming cartilaginous molds as early as embryonic day 11.5 (E11.5). Integrin α10β1 remained the dominating integrin in all cartilage regions until birth, while α1 was seen only in the articular surface and the hypertrophic zone and α2β1 was undetectable in any regions. At 4 weeks after birth, α1β1 was also weakly detectable in growth plate chondrocytes. These experiments suggested that different cell–matrix interactions mediated by integrins are involved in cartilage differentiation and chondrocyte function, and pointed towards a critical role of the β1 subunit.

It is a well-known phenomenon that isolated primary chondrocytes rapidly de-differentiate in plastic monolayer culture by changing their shape (from round to fibroblastic), actin cytoskeleton organization (from cortical to stress fibers), and gene expression repertoire (e.g., from collagen II and aggrecan to collagen I and fibronectin). A sensitive marker for chondrocyte phenotypic changes in culture is the relative expression of the α10 and α11 integrin subunits (Gouttenoire et al., 2010). Extended culturing of chondrocytes and/or TGFβ1 treatment leads to de-differentiation accompanied by a decrease in the chondrocyte-specific α10 integrin level and an increased expression of the fibroblast-specific α11 integrin. In contrast, BMP-2 administration to the culture medium stabilizes the chondrogenic phenotype and high levels of α10 expression (Gouttenoire et al., 2010). An earlier study has indicated that monolayer-cultured chondrocytes maintain their shape and differentiated phenotype for a longer time when they interact with collagen type II via β1 integrins, suggesting that collagen-binding integrins are essential for chondrocyte stabilization (Shakibaei et al., 1997). Indeed, it was later found that in suspension culture, the formation of chondrocyte aggregates with a pericellular matrix environment similar to the native cartilage tissue requires α10β1 integrin-collagen II interaction (Gigout et al., 2008).

The functional roles of integrins in cartilage development were first studied in micromass culture systems. High density culture of skeletogenic MSCs cells isolated from limb buds recapitulates several steps of developmental chondrogenesis in vitro, including condensation and early chondrogenic differentiation. Exogenous administration of β1 integrin-blocking antibodies during micromass culture, but not antibodies against the α1 and α5 chains, was shown to prevent the formation of cartilaginous nodules (Shakibaei, 1998; Bang et al., 2000). In chicken, both fibronectin and β1 integrin-mediated tyrosine phosphorylation of FAK was required for precartilage condensation and subsequent chondrogenic differentiation of wing bud cells (Bang et al., 2000). Further studies revealed that during precartilage condensation, matrix metalloprotease-2 (MMP-2) activity regulated by microRNA-488 negatively influences cell migration by disrupting matrix adhesion sites and down-regulating FAK-β1 integrin interaction (Jin et al., 2007). Interestingly, FAK-deficient mouse fibroblasts effectively form cartilage nodules, implying that FAK signaling actually suppresses chondrogenesis by regulating the shift from cell–ECM to cell–cell interactions as a prerequisite of early chondrogenic differentiation (Pala et al., 2008). These data indicate that focal adhesion complexes containing fibronectin, β1 integrins, and FAK might be important for the transition of mesenchymal precursors to chondrocytes in vitro.

In chicken sternal organ culture, blocking β1 integrin by antibodies led to reduced growth, increased apoptosis, and abnormal organization of the actin cytoskeleton suggesting a pivotal role of integrins in chondrocyte behavior (Hirsch et al., 1997). Studies with isolated cells demonstrated that proliferation of rabbit growth plate chondrocytes requires α5β1 and fibronectin (Enomoto-Iwamoto et al., 1997), whereas survival of chicken chondrocytes in suspension culture is mediated via β1 integrin–collagen interactions (Cao et al., 1999). Other experiments highlighted the importance of integrins in hypertophic chondrocyte differentiation: signaling via α1β1 and α5β1 integrins was required for transglutaminase-induced hypertrophy of cultured AC chondrocytes (Johnson et al., 2008) and a β1 integrin-blocking antibody impaired collagen X deposition (a marker for hypertrophic chondrocytes) in sternal organ culture (Hirsch et al., 1997). In vitro evidence for the link between β5 integrins and apoptosis of terminally differentiated hypertrophic chondrocytes has been also reported (Wang and Kirsch, 2006). Overexpression of annexin V, a collagen II receptor and cytoplasmic-signaling protein, which co-expresses with β5 integrin in hypertrophic chondrocytes and binds to the integrin cytoplasmic tail, increased the death of cultured growth plate chondrocytes. In contrast, siRNA-mediated silencing of annexin V increased chondrocyte viability.

Studies with Knockout Mice

The in vitro experiments suggested a critical role for integrins, especially β1 integrins, in cartilage differentiation, development, and function. Constitutive inactivation of the genes encoding the β1 (Itgb1) and α5 (Itga5) subunits results in early embryonic lethality before chondrogenesis (Bouvard et al., 2001), therefore these null mutants cannot be used to draw conclusions about their biological function in chondrocytes. To overcome the problem of lethality and investigate the in vivo roles of β1 integrins during endochondral bone formation, we have conditionally inactivated the floxed β1 integrin gene (β1fl/fl) using the loxP-Cre system. By utilizing a transgene driving the expression of the Cre recombinase under the control of the mouse collagen II promoter (Col2a1cre, Sakai et al., 2001), we generated mice lacking β1 integrin in chondocytes (β1fl/fl-Col2a1cre, Aszodi et al., 2003). To establish a loss-of-function mouse line before chondrogenesis (Raducanu et al., 2009), we used the Prx1cre transgene, which deletes in limb bud mesenchymal precursors for chondrocytes, osteoblasts, tenocytes, and synoviocytes (Logan et al., 2002). The major difference between the two strains is that the lack of β1 integrins in the entire cartilaginous skeleton (β1fl/fl-Col2a1cre) results in perinatal lethal chondrodysplasia because respiratory distress, while β1fl/fl-Prx1cre mice having a limb-restricted phenotype survive after birth allowing the investigation of the consequence of β1-deficiency in the adult appendicular skeleton. Importantly, the lack of β1 integrin in the condensing mesenchyme does not affect chondrogenesis and the formation of cartilaginous templates in β1fl/fl-Prx1cre mice (Raducanu et al., 200 and unpublished results), demonstrating that β1 integrins-mediated cell–matrix interactions and signaling processes are either not required for early stages of endochondral bone formation (aggregation and early chondrogenic differentiation) per se, or are efficiently compensated by members of other β subfamilies. Despite the difference in viability, the embryonic skeletal phenotype is essentially the same in both strains, which was reported in details in the β1fl/fl-Col2a1cre mice (Aszodi et al., 2003; Raducanu et al., 2009).

The most dramatic phenotype of mice lacking β1 integrins on chondrocytes were observed in the growth plate (Aszodi et al., 2003). The normal columnar growth plate is a polarized structure characterized by specifically shaped and oriented cells in the proliferative zone. Proliferative chondrocytes display a strongly flattened, anisotropic geometry and align along the medio-lateral (ML) axis of the growth plate, thus perpendicular to the proximo-distal (PD) direction of the longitudinal growth. The mitotic figures in these elongated chondrocytes lie also perpendicular to the PD axis and cell divisions occur parallel with the columns (Doods 1930). The two mediolaterally oriented, semi-circular daughter cells subsequently undergo gradual flattening and extensive rotational or gliding movements around each other to arrange into the longitudinal column. Polarized cell migration and ML intercalation is characteristic for convergent extension, an essential morphogenetic process resulting in tissue narrowing along the ML axis and lengthening along the PD axis. This mechanism was completely disrupted in β1 integrin-deficient growth plates, resulting in shortened and broadened long bones. The mutant chondrocytes in the proliferative zone were rounded up, failed to rotate, and showed random orientation of the mitotic spindle and the cleavage plain (Aszodi et al., 2003 and unpublished). Although an increasing body of evidence suggests that β1 integrins are actively involved in spindle and cell division positioning in various tissues, including the skin (Lechler and Fuchs, 2005) and mammary epithelium (Taddei et al., 2008), the mechanisms by which β1 integrins govern mitotic spindle orientation and gliding movements in cartilage are still not fully understood. In contrast to other cell types, normal proliferative chondrocytes do not round up during mitosis, but remain flattened throughout the cell cycle. For such elongated cells, Hertwig's laws can be applied, postulating that (1) spindles are formed in the longest axis of the cells and (2) the division plane is always perpendicular to the long axis (reviewed in Shah, 2010). In this model, simple geometric constraints provide the necessary cue for orientation of the mitotic spindle and direction of cell division. β1-null chondrocytes do not adhere to collagen II and laminin, and only partially to fibronectin (Aszodi et al., 2003), prompting the speculation that such significant loss of cell anchorage to the cartilage ECM causes the shape change (rounding) of chondrocytes in the proliferative zone of the mutant growth plate. Consequently, the isotropic β1-deficient chondrocytes lack geometric and adhesive cues which guide oriented cell division and rotational movements resulting in disorganized growth plate.

In addition to the crucial role of β1 integrins for spindle and division positioning, β1fl/fl-Col2a1cre mice exhibited further cellular and ECM defects (Aszodi et al., 2003). Mutant chondrocytes were frequently bi-nucleated and displayed actin abnormalities, demonstrating that disrupted cell–ECM interactions may impair adhesion-generated pulling forces and actomyosin ring constriction at the cleavage furrow, which are required to complete cytokinesis. β1-null chondrocytes also showed abnormal G1/S transition caused by reduced cyclin D level, increased nuclear translocation of Stat1/Stat5a, and elevated expression of the cell cycle inhibitors, p16 and p21. Furthermore, an increased apoptosis rate of growth plate chondrocytes was observed. These abnormalities were accompanied with slightly reduced activation of FAK and Erk1/Erk2, suggesting that altered integrin signaling is likely contributing to the proliferation and survival defects. Indeed, cartilage-specific deletion of the gene encoding integrin-linked kinase (Ilk), an effector molecule interacting with the cytoplasmic domain of β1/β3 integrins, leads to similar but milder growth plate abnormalities, including reduced proliferation, disorganization of the cytoskeleton, and chondrocyte shape change (Grashoff et al., 2003; Terpstra et al., 2003). Finally, the β1fl/fl-Col2a1cre cartilage has abnormal ultrastructure characterized by thick and disordered collagen fibrils, indicating that β1-integrins on chondrocytes are important for proper assembly of the collagenous fibrillar network. In fibroblast, a preformed fibronectin matrix is essential for the formation of the collagen fibrillar network, and the collagen-binding α2β1 and α11β1 integrins strongly enhance this process (Velling et al., 2002). We have shown that mice lacking fibronectin in cartilage have normal growth plate structure (Aszodi et al., 2003), predicting that one or more collagen-binding β1 integrins directly modulate the polymerization of collagen II and/or incorporation of the fibrils into the meshwork. In accordance, β1 integrins were shown to play a role in the arrangement of the ECM around chondrocytes by anchoring and bending type II collagen fibrils (Lee and Loeser, 1999). The absence of β1 integrins on the cell surface and the ECM changes had profound effects on the diffusion and biomechanical properties of the cartilage (Bougault et al., 2013). Rheological measurements on intact costal cartilage isolated from 17.5-day-old embryos revealed increased compressive stiffness of β1fl/fl-Col2a1cre samples, likely because high cellular rigidity induced by uncoupling the cytoskeleton–integrin interaction. In contrast, the diffusivity of macromolecular fluorescent solutes was greatly reduced in close vicinity of the cell membrane, indicating that the lack of β1 integrin–ECM interactions influences the organization and/or composition of the pericellular matrix compartment.

In contrast to the severe cartilage abnormalities of the β1fl/fl-Col2a1cre and β1fl/fl-Prx1cre mice, most mutant strains lacking other integrin subunits expressed on chondrocytes, such as α1, α2, α6, αv, β3, and β5, do not show skeletal defects (Bouvard et al., 2001). The only exception is the α10 integrin-deficient mice, which partially recapitulate the β1-null phenotype. Constitutive deletion of the gene encoding α10 (Itga10) results in a mild, non-lethal chondrodysplasia characterized by mild chondrocyte shape change and disorganization of the growth plate (Bengtsson et al., 2005). Furthermore, the lack of α10 on chondrocytes impairs proliferation via the modulation of the Stat1/Stat5a/p16 pathway, and leads to reduced collagen fibrillar density in the cartilage matrix, as seen in the β1 null growth plate. These findings imply that although α10β1 integrin is the most critical collagen-binding integrin, which modulates chondrocyte function, the absence of multiple β1 integrins is necessary for the most severe cartilage defects, indicating partial compensation among β1 integrin-containing heterodimers. It is worth mentioning that double knockout mice lacking both α2β1 and α11β1 integrin develop dwarfism, which is, however, caused by decreased production of insulin-like growth factor-1 in the liver and does not result from growth plate dysfunction (Blumbach et al., 2012).

To date, mutations in genes encoding integrin subunits have not been directly associated with inherited human chondrodysplasias (Warman et al., 2010). However, very recently a truncating mutation in exon 16 of the canine ITGA10 gene was identified in two breeds of Nordic hunting dogs with disproportionate short-limbed dwarfism (Kyostila et al., 2013). The phenotype of the diseased growth plate is similar to that of the α10-deficient cartilage mouse and caused by the lack of α10 protein in the affected dogs. As the canine and human ITGA10 genes are highly homologous, these findings suggest that a similar, naturally occurring loss-of-function mutation in human ITG10 is a potential candidate for chondrodysplasias with so far unidentified genetic background. Mutations in the gene coding for filamin B (Flnb), an intracellular scaffolding protein that interacts with actin and β1 integrin, result in a range of human skeletal disorders (Krakow and Rimoin, 2010) including boomerang dysplasia and spondylocarpotarsal syndrome. A mouse model lacking Flnb shows reduced phosphorylation of β1 integrin at Ser785, which is required for cell adhesion and develops cartilage ECM and chondrocyte abnormalities seen upon blocking β1 integrin (Lu et al., 2007). Thus, disruption of the ECM-β1 integrin-filamin B axis may contribute to skeletal dysplasias both in mouse and human.

SYNOVIAL JOINT AND AC

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

Diarthroidal joints unite different type of tissues to ensure load transmission within skeletal structures and coordinate their movements. The opposing bones are covered with AC and are separated by a joint cavity enclosed in a capsule linking the skeletal elements. The capsule is lined by a synovial membrane secreting synovial fluids, which lubricates the AC surface and provides nutrition to the components of joints.

Synovial joints develop through two main processes called joint specification and joint morphogenesis (reviewed in Pacifici et al., 2005). In the step of joint specification, skeletogenic MSCs expressing the Sox trio are directed to the articular fate under the control of TGF-beta receptor 2 (Spagnoli et al., 2007), canonical Wnt (e.g., Wnt4 and Wnt9a/14) (Hartmann and Tabin, 2001; Guo et al., 2004), and growth differentiation factor-5 (Gdf5) (Settle et al., 2003) signaling cascades, resulting in the formation of presumptive joint regions called interzones. Subsequently, a cavity forms in the interzone between the articulating skeletal elements driven by a not fully understood mechanism involving high-level synthesis of hyaluronan (Matsumoto et al., 2009), specific ECM composition (Pacifici et al., 1993; Choocheep et al., 2010), and skeletal movement (Kahn et al., 2009). During joint morphogenesis, specific articular cells differentiate and form the joint tissues. The continued expression of the Sox trio is pivotal for articular chondrocyte differentiation, while it seems that Wnt/beta-catenin signaling prevents chondrocyte differentiation at sites not destined for AC, such as the synovial lining and ligaments.

Integrins and Joint Morphogenesis

The function of integrin-mediated cell–cell or cell–matrix interactions in joint determination and morphogenesis, although anticipated (Pacifici et al., 2005), is poorly investigated and understood. Almost a decade ago, Garciadiego-Cázares et al. (2004) reported that α5 integrin is downregulated in the interzones of developing mouse fingers, suggesting a role for α5β integrin in patterning of the hands. Indeed, blocking integrins in E14.5 mouse forelimb organ culture by microinjecting anti-β1 and anti-α5 monoclonal antibodies or RGD peptides into the wrist region induced ectopic joint formation at the distal part of the radius and the ulna. A narrow gap region expressing interzonal markers, such as Gdf5 and Wnt9a/14, but not expressing the prehypertrophic marker Indian hedgehog (Ihh), had formed in between the proliferative and hypertrophic zones. In contrast, forced expression of human cDNAs coding for human α5 and β1 integrin subunits in embryonic chick legs had resulted in joint fusions of the phalanges, accompanied by the differentiation of prehypertrophic chondrocytes expressing Ihh. Consequently, the authors proposed that α5β1 integrin-RGD ligand (fibronectin) interactions are pivotal in cell fate determination during the development of the appendicular skeleton by dictating either the differentiation of interzonal cells or prehypertrophic chondrocytes on the basis of α5β1 expression level in chondrocytes. Although this hypothesis is clearly attractive, genetic experiments with conditional knockout mice did not support a mechanistic role for α5β1-mediated matrix attachments in joint morphogenesis. To date, no in vivo experiments have been performed to investigate the role of α5 integrins in skeletogenesis. However, neither β1fl/flCol2cre mice nor β1fl/flPrx1cre mice display ectopic joint formation, and in both models, Ihh is expressed at the normal level in the prehypertrophic zone (Aszodi et al., 2003; Raducanu et al., 2009). The lack of any obvious phenotype in joint formation in the absence of β1 integrins on skeletogenic MSCs, perichondrial cells, and chondrocytes implies that antibody perturbation might trigger non-specific and/or non-physiological responses (e.g., apoptosis) not seen upon genetic ablation.

Integrins and the AC

AC is a permanent, specialized type of hyaline cartilage that functions as a load bearing tissue and can resist shearing forces and friction. The AC, similar to the growth plate, is characterized by a high degree of anisotropy and is divided into four zones with distinct organization of the matrix and cells: superficial (or tangential), intermediate (or transitional), deep (or radial), and calcified (Poole et al., 2001). The superficial zone is the thin, outermost layer of the AC, which contains elongated cells and densely packed collagen fibrils arranged parallel to the surface. This zone has low aggrecan content but it is enriched in small proteoglycans, such as decorin and biglycan (Poole et al., 1996). The intermediate zone contains individual spherical chondrocytes and a relatively disorganized collagen network, whereas in the deep zone the cells tend to organize into columns with radially oriented collagen fibril bundles between them. The pericellular compartment of these zones has high collagen VI, decorin, and perlecan content, while aggrecan is more abundant in the territorial and interterritorial regions. The calcified zone is separated from the upper zones by the tidemark and its main function is to anchor the AC to the subchondral bone. The hypertrophic chondrocyte marker collagen X is expressed in this zone (Gannon et al., 1991; Hoyland et al., 1991; von der Mark et al., 1992), but unlike in the growth plate, the calcified matrix normally resists vascular invasion and resorption.

AC chondrocytes have a similar set of integrin heterodimers as growth plate chondrocytes with the most prominent expression of α1β1, α2β1, α10β1 (for collagens); α5β1, αvβ3, αvβ5 (for fibronectin), and α6β1 (for laminin) (Durr et al., 1993; Woods et al., 1994; Salter et al., 1995; Camper et al., 2001). In vitro studies revealed that human AC chondrocytes primarily use β1 integrins, including α5β1 and αvβ5 integrins, for adhesion to the cartilage matrix (Kurtis et al., 2003), whereas α5β1-fibronectin interaction was shown to promote survival of cultured AC chondrocytes (Pulai et al., 2002). Therefore it is not surprising that, analogous to the growth plate, β1 integrins are critical regulators of chondrocyte phenotype in the AC as well. β1fl/flPrx1cre mice display severe abnormalities of AC chondrocytes including rounding of the superficial zone cells, loss of columnar arrangement in the deep zone, reduced proliferation, increased cell death and bi-nucleation, and disrupted actin cytoskeleton (Raducanu et al., 2009). These observations extend the view that in vivo β1 integrins are pivotal for maintaining the normal shape, organization, proliferation, and survival of chondrocytes, independent of their location in the cartilaginous skeleton. Furthermore, β1 integrin-deficiency of AC chondrocytes leads to prominent ECM defects, including disorganized collagen II network in the interterritorial compartment and reduced deposition of collagen VI in the pericellular compartment, suggesting that β1 integrins are important organizers of the collagenous matrices. Importantly, the ablation of β1 integrins differentially affects chondrocyte maturation in the growth plate and the AC. While in β1fl/flCol2cre and β1fl/flPrx1cre mice the hypertrophic differentiation of growth plate and epiphyseal chondrocytes is delayed, the AC in the β1fl/flPrx1cre mice exhibits reduced ratio of the non-calcified and calcified zones and contains collagen X expressing chondrocytes above the tidemark, suggesting accelerated hypertrophic differentiation (Aszodi et al., 2003; Raducanu et al., 2009). This may imply that β1 integrins have a spatially restricted, tissue context- dependent role in chondrocyte maturation.

Interestingly, while no single α subunit-deficient mice have been reported with structural defects of the AC, the secreted melanoma inhibiting activity/cartilage-derived retionic acid sensitive protein (MIA/CD-RAP)-deficient mice partially recapitulate the AC abnormalities seen in the β1fl/flPrx1cre mice. MIA/CD-RAP is expressed in chondrocytes (Dietz and Sandell, 1996; Bosserhoff et al., 1997) and its absence in mice leads to mild ultrastructural abnormalities of the collagen fibrils in the developing growth plate cartilage (Moser et al., 2002). In 3 month old mice, MIA/CD-RAP-deficient AC shows disturbed columnar arrangement of the chondrocytes and increased height of the calcified zone (Schubert et al., 2010). Based on in vitro studies with cultured chondrocytes, it was proposed that MIA/CD-RAP physically binds to and inhibits α5β1 integrin, leading to reduced ERK signaling, stabilization of the chondrocyte phenotype, and delay of hypertrophy. This finding is consistent with a recent report demonstrating correlation between ERK activation and hypertrophic differentiation of human AC chondrocytes (Prasadam et al., 2010). Furthermore, an antibody-activating α5β1 integrin/ERK signaling (Forsyth et al., 2002) was shown to abolish the negative effect of MIA/CD-RAP on the expression of the osteogenic marker osteocalcin in cultured AC chondrocytes (Schubert et al., 2010).

Taken together, in vitro and loss-of-function genetic studies in mice strongly suggest that integrins are required for proper differentiation and spatial organization of AC chondrocytes. Since the ablation of both β1 integrin and the α5β1 integrin-inhibitory MIA/CD-RAP results in a similar phenotype, the exact nature of integrin-mediated signaling pathways on articular chondrocyte hypertrophy remains to be elucidated.

Integrins and Osteoarthritis

AC is highly prone to injury and pathological degeneration. Once damaged, the AC is unable to initiate an efficient regenerative process, owing to the loss of mitotic activity of adult AC chondrocytes, decreased responsiveness to anabolic growth factors, and synthesis of less functional matrix molecules. Osteoarthritis (OA) is a common degenerative disorder of the synovial joints that not only affects the AC but also involves changes in subchondral bone, synovium, ligaments, and muscle. OA is a complex disorder influenced by both environmental and genetic factors. Although the precise etiology is largely unknown, it is generally accepted that uncontrolled metabolism of skeletal tissues is critical for the pathophysiology of this disabling condition. In primary OA, the normal turnover of ECM molecules produced by chondrocytes is disturbed, and the equilibrium between anabolic and catabolic processes is shifted towards proteolytic activities, which eventually results in progressive, age-dependent AC destruction (Goldring and Goldring, 2007). In addition, congenital abnormalities of the joints, as a typical result of impaired development of endochondral bones, are often associated with early onset degeneration and loss of the AC (secondary OA). Physiological ECM remodeling of the AC occurs in a spatially and temporally controlled fashion, and involves the activities of both proteinases and proteinase inhibitors, which are tightly regulated at multiple levels. The regulatory mechanisms are only partially understood and include, in addition to growth factor and cytokine signaling pathways, chemical and mechanical signals modulated by matrix–matrix and cell–matrix interaction (Knudson and Loeser, 2002). There is a growing consensus that changes of ECM composition, e.g., by proteolytic degradation of matrix constituents; or alterations of the biomechanical microenvironment of chondrocytes caused by chronic stress or injury, significantly increase the risk of OA through the perturbation of integrin-mediated signaling.

The strongest evidence for the participation of integrins in OA has emerged from experiments studying integrin expression and ECM fragments-induced signaling processes. In osteoarthritic monkey AC, increased immunostaining of α1, α3, and α5β1 integrins was reported, compared with normal cartilage (Loeser et al., 1995). In human OA samples, the expression of β1 integrin was inversely correlated with the severity of the lesions, suggesting that β1 integrin-mediated-ECM adhesion is decreased with OA progression (Lapadula et al., 1998). In contrast, osteoarthritic human femoral head cartilage expressed the integrin subunits β2, α2, and α4, which were not observed in normal AC (Ostergaard et al., 1998). Subsequently, numerous reports have shown that fibronectin (FN) and collagen II degradation fragments are present in synovial fluid and cartilage of OA patients as a result of increased catabolic activity, and these fragments can further mediate joint destruction through the activation of integrin signaling (reviewed in Yasuda, 2006, Sofat, 2009). Treatments of cultured rabbit synovial fibroblasts with the central, 120 kDa cell-binding FN fragment (FN-f) carrying the RGD sequence or ligation of α5β1 integrin with a monoclonal antibody were shown to stimulate MMP-1 and MMP-3 expression (Werb et al., 1989). Similarly, the central FN-f and α5β1/α2β1 activating antibodies were shown to upregulate MMP-13 synthesis in cultured human AC chondrocytes (Forsyth et al., 2002; Loeser et al., 2003). Mechanistically, the stimulation of α5β1 integrin with ligation antibodies or FN-f activates protein kinase C (PKC)-delta, which in turn phosphorylates the proline-rich tyrosine kinase-2 (PYK2), resulting in increased MAPK (Erk-1/2, p38, and JNK) signaling and increased phosphorylation of c-jun and NF-kappa B (Loeser et al., 2003; Pulai et al., 2005; Ding et al., 2008). In addition to central FN-f, the amino-terminal 29 kDa and the gelatin-binding 50 kDa FN-fs are also chondrolytic and likely interfere with the native FN/α5β1 signaling pathway (Homandberg and Hui, 1994; Homandberg et al., 2002a, b). The physiological relevance of Fn-fs in cartilage destruction was evidenced by studies demonstrating that injection of Fn-fs into knee joint dramatically increases proteoglycan loss of rabbit AC (Homandberg et al., 1993). The exact mechanism by which ECM degradation fragments accelerate cartilage catabolism is unclear. The most popular explanation supposes that FN and collagen degradation products disturb or even disrupt normal integrin clustering induced by native FN or collagen II, which in turn alters physiological outside-in signaling and cartilage metabolisms (Peters et al., 2002). However, beside degradation fragments, collagen itself is able to induce MMP expression upon binding to α1β1 integrin. Collagen I is upregulated in OA cartilage (Adam and Deyl, 1983; Miosge et al., 2004) and in the chondrogenic cell line MC615, and it stimulates the expression of MMP-13 via the activation of ERK and JNK MAP kinases. In the presence of α1 or β1 blocking antibodies the induction of MMP-13 expression was inhibited (Ronziere et al., 2005). Furthermore, a recent report has suggested that collagen II fibrils modified by the lipid peroxidation end-product 4-hydroxynonenal, which is upregulated in synovial fluid of OA patients, induce cell death and catabolic and inflammatory responses of human AC chondrocytes, presumably through perturbation of α1β1 signaling (El-Bikai et al., 2010).

In AC, physiological mechanical loading leads to anabolic changes that help to maintain tissue integrity, while non-physiological mechanical stimuli (e.g., overloading and injury) are associated with cartilage damage and predispose to OA. Chondrocytes that are exposed to mechanical forces transmit the signals via various mechanotransducers, including stretch activated ion channels, receptor tyrosine kinases, the hyaluronan receptor CD44, and integrins (Millward-Sadler and Salter, 2004; Bader et al., 2011; Vincent, 2013). Under physiological loading, integrins (e.g., α5β1) are activated on the surface of isolated chondrocytes, leading to recruitment of adaptor proteins in focal adhesion complexes and to lateral interactions with growth factor receptors and ion channels (Millward-Sadler et al., 2004). Induction of the integrin signaling pathway may lead to the secretion of interleukin-4 (IL-4), which via autocrine/paracrine signaling events, blocks catabolic processes and enhances GAG and proteoglycan synthesis (Millward-Sadler et al., 1999; Holledge et al., 2008). The application of dynamic cyclic compression on human chondrocytes in the presence of TGFβ3 also enhanced proliferation and proteoglycan synthesis, which was reversed by adding the α5β1 ligand GRGDSP to the system (Chowdhury et al., 2004). Mechanotransduction via α5β1 might be regulated by the α5 integrin-associated protein CD47, since a function-blocking anti-CD47 antibody inhibited membrane hyperpolarization, tyrosine phosphorylation, and the elevation of aggrecan mRNA expression induced by mechanical stimulation (Orazizadeh et al., 2008). In contrast to this protective role, non-physiological mechanical loading disrupts the actin cytoskeleton via α5β1-mediated signaling that activates the MAP kinases and NF-kappa B. These, in turn, result in elevation of nitric oxide, prostaglandin E2, reactive oxygen species, proteases, and cytokines, which mediate AC degradation (Bader et al., 2011).

Although these expression studies and in vitro experiments highlight the importance of integrin signaling in OA and suggest that inhibition of integrin pathways might be a potential approach to ameliorate ECM and ECM fragment-mediated AC destruction, the in vivo role of integrins in OA pathogenesis has just begun to be elucidated. Most mutant lines lacking a particular α subunit develop an apparently normal AC without reported abnormalities. The collagen-binding α1β1 integrin normally is expressed at low levels in the deep zone of healthy mouse AC but it is significantly upregulated in the middle and upper zones of early osteoarthritic cartilage, suggesting that this integrin heterodimer supports cartilage remodeling (Zemmyo et al., 2003). Knee joints of α1-null mice display precocious proteoglycan loss, cartilage erosion associated with increased MMP-2 and MMP-3 expression, and synovial hyperplasia (Zemmyo et al., 2003). Accelerated development of knee OA was also observed in mice lacking the membrane-anchored disintegrin and metalloproteinase ADAM15 (Bohm et al., 2005). ADAM15 carries a RGD motif and has multiple pericellular functions, including the modulation of outside-in signaling in chondrocyte-matrix interactions (Bohm et al., 2009), which in turn enhances chondrocyte adhesion to collagen II and promotes AC homeostasis. Interestingly, β1fl/fl-Prx1cre+ mice, despite the severe chondrodysplasia and efficient deletion of β1 integrins in AC chondrocytes, showed no significant differences in erosion of the knee AC, in the expression of matrix-degrading proteases, or in the exposure of aggrecan and collagen II cleavage neoepitopes, compared with controls (Raducanu et al., 2009). In contrast to the previous in vitro experiments, no evidence was found for disturbed activation of MAP kinases in mutant AC and in hip explants upon stimulation with a 40 kDA cell-binding FN fragment. These results suggest that either β1 integrins are not required for ECM–ECM fragment induced MAPK activation per se, or, and this is the most likely scenario, the lack of β1 integrins is efficiently compensated via signaling through β3/β5 integrins.

AC Repair

AC has a poor intrinsic capacity to heal owing to limited synthetic and proliferative potential of AC chondrocytes; low efficiency of AC resident chondrogenic progenitors to induce rapid and successful regeneration; and the lack of vascular system to deliver MSCs into the damaged tissue. Current orthopedic therapeutic strategies include marrow stimulation-based, osteochondral transfer, and cell-based techniques (Richter, 2009), with various advantages and disadvantages of each. Subchondral bone drilling stimulates the recruitment, proliferation, and chondrogenic differentiation of bone marrow-derived MSCs, but it frequently produces fibrocartilage, which has a different biochemical composition and insufficient biomechanical properties, compared to native hyaline cartilage. Autologous chondrocyte implantation (ACI) and the more recent matrix assisted ACI (MACI) have provided encouraging clinical results to heal cartilage defects; however, these techniques are hampered by the low proliferation rate of isolated chondrocytes in culture and their high de-differentiation potential to fibroblasts during the expansion period. MSCs isolated from various sources (e.g., bone marrow, synovial membrane, adipose tissue) have sufficient proliferative activity in vitro but require special culture conditions to differentiate into chondrocytes. Inducing and/or stabilizing the chondrogenic phenotype is a crucial requirement of all regenerative therapies, and the importance of integrins in this process has been increasingly recognized.

MSCs differentiation into different lineages (chondrogenic, osteogenic, adipogenic) in vitro can be modulated via numerous factors, including growth factors, cell–matrix interactions, 3D microenvironment, cytoskeletal tension, or mechanical stimulation (McBeath et al., 2004; Engler et al., 2006; Woods et al., 2007; Kelly and Jacobs, 2010). Collagen II, for example, enhances chondrogenesis of both bone marrow-derived MSCs (BMSC) and adipose-derived MSCs (ASC) in hydrogels (Bosnakovski et al., 2006; Lu et al., 2010) by inducing rounded cell shape through β1 integrin-mediated RhoA/Rock signaling (Lu et al., 2010). Co-culturing BMSCs and collagen II fibrils in 3D led to remodeling of the fibers, differentiation of spherical chondrocytes, and synthesis of cartilaginous markers on β1 and α2 integrin-dependent manner (Chang et al., 2007). In contrast, interaction with RGD peptides inhibits chondrogenesis of BMSCs by promoting spreading through αvβ3 integrin and actin cytoskeletal reorganization (Connelly et al., 2007, 2008). Stem cell commitment can be further regulated via ECM remodeling. The membrane-bound MT1-MMP (MMP-14), a pericellular collagenase, modulates cell shape, which in turn activates a signaling pathway involving β1 integrin, Rho GTPase activity, and the YAP/TAZ transcriptional coactivators (Tang et al., 2013). Using MT1-MMP-deficient MSCs and mice lacking MT1-MMP in skeletogenic MSCs (Mmp14fl/fl/Dermo1-cre), it was shown that the absence of this protease directs MSC fate towards chondrogenesis and adipogenesis at the expense of osteogenesis both in vitro and in vivo. Stiffness of the hydrogels and external mechanical loading (e.g., cyclic compression and hydrostatic pressure) also influence MSC chondrogenesis (Huang et al., 2004; Pelaez et al., 2009; Meyer et al., 2011), which involves integrin-mediated cell-matrix interactions (Steward et al., 2012, 2013). Recently, an α10 integrin expressing subpopulation of monolayer expanded human BMSCs with high chondrogenic potential in pellet culture was identified, suggesting that α10β1 can be a unique biomarker for quality assurance of MSCs used for cartilage repair (Varas et al., 2007). Furthermore, α10 expression was significantly downregulated during extensive monolayer culture of BMSCs, which was reversible by FGF-2 treatment.

Further in vitro studies have demonstrated the prospect of the modulation of integrin signaling in AC repair. The activation of FAK is increased upon exogenous stimulation of alginate beads with collagen II, whereas FAK knockdown reduces chondrocyte re-differentiation in the same culture system (Kim and Lee, 2009). As discussed earlier, MIA/CD-RAP induces chondrogenic differentiation via abolishing α5 integrin activity and inhibiting the downstream ERK signaling; and stimulates cartilage synthesis of chondrocytes obtained from patients suffering from OA (Tscheudschilsuren et al., 2006; Schubert et al., 2010). Importantly, MIA/CDRAP-deficient mice exhibit increased regeneration potential of the AC in a surgically induced OA model because enhanced proliferation of the chondrocytes (Schmid et al., 2010). In contrast, deficiency of tenascin C, a ligand of some β1 and β3 integrins, delays the repair of full thickness cartilage defects in mice (Okamura et al., 2010). Finally, a recent report has shown that magnesium enhances the adhesion and hyaline cartilage differentiation of human synovial MSCs at the sites of osteochondral plugs in the rabbit knee through activation of β1 integrins (Shimaya et al., 2010).

Bone Formation and Function

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

Intramembranous ossification is the second process by which bones can form during development of the mammalian skeletal system. In contrast to endochondral bone formation where the growth plate cartilage is replaced by bone, intramembranous bones arise directly from mesenchymal precursors without an intermediate cartilaginous phase. Skeletogenic MSCs at the sites of future bones first aggregate into dense nodules within a connective tissue sheet and develop capillaries or differentiate into osteoblasts establishing an ossification center. The newly formed osteoblasts begin to secrete a collagen I-rich ECM termed osteoid, which will be subsequently mineralized and organized into compact bone, while the embedded osteoblast became osteocytes. Flat bones of the skull, the lateral part of the clavicles, and the bone collar around the growth plate originate by intramembranous ossification covered by dense connective tissues from outside (periosteum) and inside (endosteum). Intramembranous ossification is also important during bone healing, where the endosteal and periosteal surfaces near the fracture gap are filled and remodeled by this process (Isefuku et al., 2004). Both endochondral and intramembranous bones undergoes continuous remodeling. Osteoclasts derived from the hematopoietic lineage resorb the bone tissue and new osteoid forms by osteoblasts that subsequently become mineralized. Imbalance in this remodeling mechanisms leads to many bone diseases, including osteopetrosis (abnormally dense bone) and osteoporosis (less dense bone). Intramembranous bone growth can occur through bone formation within the periosteum or at the sutures. Sutures are mesenchymal stripes which connect cranial bones and contain osteoprogenitors that differentiate into osteoblasts at the osteogenic front (Opperman, 2000).

During embryonic development the differentiation of progenitor cells into osteoblasts or chondrocytes is controlled by the Wnt/beta-catenin signaling. The common progenitors express both Sox9 and Runx2, being respectively the transcription factors labeling the commitment to the chondrogenic and osteogenic lineages. Normally, the canonical Wnt signaling inhibits Sox9 and promotes Runx2 expression, driving bone formation through the process of intramembranous ossification at the expenses of chondrogenesis (Day et al., 2005). Osteoblast differentiation and maturation, bone mineralization, and expression of late markers such as osteocalcin (Bgb) and bone sialoprotein (Bsp) are under the control of various transcription factors and secreted molecules (reviewed in (Karsenty et al., 2009; Lefebvre and Bhattaram, 2010). For example, the transcription factor osterix (Osc) acts downstream of Runx2, induces the expression of Col1a1 and Bgb, and is required for osteoblastic differentiation of Runx2-expressing progenitors. In contrast, ATF4 (activating transcription factor-4) is not needed for osteoblast differentiation but is important for synthesis of collagen I by regulating amino acid import of the cells. In addition, orchestration of various signaling pathways, including FGFs, BMPs, Wnts, or Ihh, is necessary for the precise control of osteoblastogenesis and bone formation. Osteoclasts are derived from myeloid stem cells through a series of events, including commitment, differentiation, multinucleation, and maturation under the control of multiple regulations, including macrophage colony-stimulating factor (M-CSF), tumor necrosis factor (TNF) family cytokines, and receptor activator of nuclear factor (NF)-κB ligand (RANKL) (reviewed in (Yavropoulou and Yovos, 2008).

Integrin Expression and In Vitro Studies

Numerous studies have been performed to characterize the integrin expression profile in bone tissue, osteoblasts, osteoclasts, and their precursor cells. Osteoblasts express a broad range of integrin receptors for collagen, fibronectin, and vitronectin, including α1β1, α2β1, α3β1, α5β1, and αvβ3 (reviewed in (Shekaran and Garcia, 2011). Cell culture experiments have shown that integrin-ECM interactions are critical for osteoblast differentiation and survival (Moursi et al., 1997; Globus et al., 1998; Marie, 2013) and play pivotal roles in bone mechanotransduction (Katsumi et al., 2004; Thompson et al., 2012). In osteoclasts, antibodies or antagonists against αvβ3 demonstrated the essential role of this integrin for attachment, migration, and bone resorption (Duong et al., 2000; Horton, 2001). Osteoclasts resorb bone by reorganizing their cytoskeleton and forming an actin ring, the sealing zone, and releasing proteases at defined membrane convolutions called ruffled borders. It was recently showed that ligation of αvβ3 on vitronectin-coated coverslips induces these resorptive structures, thereby αvβ3 ligands and not the mineralized bone itself is required for osteoclast activation (Fuller et al., 2010).

Another set of experiments aimed to identify key integrins that drive osteogenesis. The integrin subunits α1, α2, α3, α5, α6, αV, β1, β3, and β4 have been reported to be expressed in human MSCs (reviewed in (Docheva et al., 2007). In particular, the α1β1, α2β1, α5β1, α6β1, αVβ3, and αVβ5 heterodimers co-express with the STRO-1 antigen on the cell surface of human stromal precursors (Gronthos et al., 2001). Our research group has recently shown that the collagen-binding α2β1 and α11β1 integrins are essential for BMSC adhesion, migration, and survival, and that the expression level of α2β1 is significantly reduced in BMSCs from osteoporotic patients (Popov et al., 2011). α2 integrin activates the FAK-ERK-ROCK and PI3K/Akt signaling pathways (Shih et al., 2011) and its knockdown in BMSCs severely impairs (Popov et al., 2011), while its overexpression promotes osteogenic differentiation (Hu et al., 2013). Integrin α5 was also found to be upregulated during dexamethasone-induced osteoblast differentiation of human BMSC (Hamidouche et al., 2009). Overexpression of ITGA5 or stimulation of α5β1 activation in MSCs induced the expression of osteoblast markers and enhanced osteogenesis both in vitro and in vivo through the activation of FAK-ERK and PI3K signaling pathways (Fromigue et al., 2012). Plating BMSCs on laminin 5-coated surface upregulates α3β1 integrin and increases phosphorylation of the osteogenic transcription factor Runx2 (Klees et al., 2005). Finally, it has been shown that cellular migration of hMSC is enhanced by bone sialoprotein, a ligand of αVβ3, by bridging of MMP-2 to the integrin, thus facilitating the disruption of matrix barriers (Karadag and Fisher, 2006).

Transgenic and Knockout Models

Although being the most highly expressed subunit on osteoblasts and despite the well-documented importance of β1 integrin in osteoblast differentiation and function in vitro, genetic studies with the manipulated Itgb1 gene in mice resulted in relatively mild bone phenotype. Transgenic mice overexpressing a function blocking dominant-negative form of the β1 integrin in mature osteoblasts and osteocytes under the control of the osteocalcin promoter display impaired intramembranous bone formation. These mice are characterized by thinner cortical and flat bones, and the bone that formed appeared disorganized, because changes of osteoblast shape and activity and matrix organization (Zimmerman et al., 2000). Follow-up studies showed that disruption of β1 integrin signaling in mature osteoblasts reduces skeletal mass and bone strength of the growing animals (Globus et al., 2005), but it has no major effect on the development of cancellous osteopenia (reduced bone mineral density) induced by mechanical unloading (Iwaniec et al., 2005). The hypothesis that integrin signaling might play a role in adjusting bone structure and strengths to mechanical loading was further challenged in conditional β1 integrin knockout mice (Phillips et al., 2008). The floxed β1 integrin was deleted in osteoblasts using Cre driven by the 2.3 kb promoter of the gene coding for the mouse α1 chain of collagen I (Col1a1), leading to depletion of β1 integrin from cortical osteocytes of the Colα1(I)-Cre+/β1fl/fl mutant mice. Although wild type and mutant mice were indistinguishable under normal mechanical loading, short-duration disuse by hindlimb unloading led to changes in cortical geometry and increased bone stiffness and strength in mutants compared to controls. These results demonstrated that β1 integrins in cortical bones limit periosteal apposition and marrow expansion during disuse, and provided in vivo evidence for a role of β1 integrin in mediating mechanotransduction in osteocytes. A Colα1(I)-Cre+/β1fl/fl mutant strain, generated independently in another laboratory, exhibits mildly disturbed bone histomorphometric parameters, such as decreased apposition rate and mineralization lag time (Bentmann et al., 2010). Interestingly, mice with cre-mediated deletion of Fak in osteoblasts did not show skeletal defects, suggesting that integrin signaling through FAK has no major impact on osteoblasts function in vivo (Kim et al., 2007). This surprising finding could be explained with the compensatory up-regulation of the related non-receptor type kinase Pyk2 at the focal adhesions of FAK-deficient osteoblasts or with redundant pathways controlling bone homeostasis. In accordance, only a mild and transient increase in trabecular bone volume was reported in mice with osteoblast-specific deletion of ILK (El-Hoss et al., 2014).

The role of the collagen-binding α2β1 integrin for age-related control of bone metabolism has been recently discovered (Stange et al., 2013). α2-null mice show protection against age-associated deterioration of bone structural and biomechanical properties, owing to impaired sensing of the collagenous matrix. Collagen I expression is upregulated in the mutant osteoblasts, leading to increased quantity of collagen in the bone matrix and stiffer bone in aged animals. These observations establish α2β1 integrin as a negative modulator of collagen turnover in the bone tissue.

The critical role of integrin signaling in osteogenesis has been demonstrated in mice with constitutive ablation of the integrin cytoplasmic domain-associated protein 1 (ICAP-1). ICAP-1 interacts specifically with the cytoplasmic domain of β1 integrin and negatively regulates integrin activation and function (Bouvard et al., 2006, 2007). ICAP-1-deficiency results in impaired osteoblast differentiation both in vivo and in vitro (Bouvard et al., 2007; Millon-Fremillon et al., 2008). Icap-1 null mice display growth retardation, defective mineralization of the calvarial bones, and delayed closure of the calvarial sutures (Bouvard et al., 2007). In the absence of ICAP-1, β1 activity is increased in osteoprogenitors which results in their impaired aggregation in vitro and defective condensation of mesenchymal cells at the osteogenic front of the sutures in vivo. Independent from the condensation phenotype, mutant osteoblasts also show reduced proliferation rate. More recently, it has been shown that the loss of ICAP-1 affects fibronectin assembly, collagen I deposition, and osteoblast mineralization (Brunner et al., 2011). Mechanistically, ICAP-1-β1 integrin interaction modulates the dynamics of fibrillar adhesion, which are structures where fibronectin self-assembles into fibrils and promotes mineralization. Collectively, the data implicate that the β1 integrin-ICAP-1 axis is a critical regulator of early as well as late osteoblast functions.

The common consensus that αvβ3 integrin is the major and functionally the most important adhesive protein on osteoclasts was clearly supported by the phenotype of mice harboring null mutation for the β3 integrin gene. β3 null mice developed progressive osteopetrosis (bone overgrowth) because dysfunctional osteoclasts (McHugh et al., 2000). Despite the increased bone mass, osteoclast differentiation was not affected, but the numerous mutant osteoclasts had greatly reduced resorptive capacity. In the absence of matrix-derived, αvβ3-mediated signals, the osteoclasts failed to polarize and reorganize the actin cytoskeleton required for adhesion, migration, and bone resorption (McHugh et al., 2000; Faccio et al., 2003). Integrin ligation leads to phosphorylation of the β3 cytoplasmic domain-associated tyrosine kinase c-Src, which in turn “switch on” a signaling complex, including Syk kinase, the adaptor ITAM proteins, the guanine nucleotide exchange factor Vav3, and Rho GTPases (Zou and Teitelbaum, 2010). Deletion of any of these molecules impairs osteoclast function, resulting in compromised bone resorption and osteopetrosis (Soriano et al., 1991; Mocsai et al., 2004; Faccio et al., 2005). The residual resorption activity of β3-deficient osteoclasts may indicate compensation by other integrins, e.g., β1 integrins, which also regulate osteoclast function in vivo (Helfrich et al., 1996; Rao et al., 2006). Patients with Glanzmann thrombasthenia (GT) lack β3 integrin and display bleeding disorder because platelet dysfunction, but do not present osteopetrosis. It has been shown that GT osteoclasts lack αvβ3 but upregulate α2β1 which partially compensates and allow bone resorption to proceed (Horton et al., 2003). The importance of different β integrin classes in osteoclast biology has been recently demonstrated in mice lacking kindlin-3 (Schmidt et al., 2011). Kindlin-3, an intracellular FERM (four-point-one, ezrin, radixin, moesin) domain-containing protein, is expressed in hematopoietic cells and required for integrin activation (Moser et al., 2009). Kindlin-3 null mice develop severe osteopetrosis, owing to defects in osteoclast adhesion, spreading, and failure in forming podosomes and sealing zones. Ablating β1, β2, and αv integrins individually, in pairs or simultaneously, and comparing their phenotypes to β3 integrin mutants revealed that all these integrin classes are needed for normal podosome formation and osteoclast function. Kindlin-3 regulates the activation of these integrins in osteoclasts, thereby orchestrating bone homeostasis. It is important to note that in contrast to kindlin-3 and β3-deficiency, the loss of β1/β2/αv integrins at the same time prevents the formation of multinucleated osteoclasts, pointing to an overlapping role of these integrin classes in osteoclastogenesis (Schmidt et al., 2011). Interestingly, the deletion of αvβ5 integrin, which is oppositely regulated with αvβ3 during osteoclastogenesis, accelerates osteoclast maturation, leading to increased bone loss in ovariectomized mice, indicating that αvβ5 is actually an inhibitor of osteoclast formation and reduces bone loss induced by estrogen-deficiency (Lane et al., 2005).

Bone Healing

Broken bones can repair through a physiological process which produces new bone and restores the original physiological and mechanical properties of the bone (McKibbin, 1978). Immediately after fracture, inflammatory cells are recruited to the site of the injury and contribute to the replacement of the necrotic bone, while growth factors and cytokines induce mitogenic and osteogenic effects on the osteoprogenitor cells. In the reparative phase, stabilized fractures may heal by intramembranous ossification which requires reestablishment of the blood supply, osteoclastic resorption, and osteoblastic deposition of the new bone. When absolute immobilization is not possible, first a cartilaginous callus tissue is formed that is then replaced by bone in the process of endochondral ossification. Practically, most fractures are repaired through the combination of the two ossification processes.

While a plethora of in vitro evidence suggests that integrins are pivotal for the repair of injured bones (reviewed in Marie, 2013), in vivo models assessing integrin function during fracture healing are scarce. The α1 integrin knockout mice, which develop normal skeleton, display impaired healing of diaphyseal tibial fractures, because significantly less callus and defective cartilage tissue formation (Ekholm et al., 2002). The results suggested that α1β1 regulates callus growth through the control of skeletal progenitor proliferation and cartilage matrix production. In contrast, the early phase of fracture healing was accelerated upon the loss of integrin β3 (Hu et al., 2010). β3-null mice had increased bleeding time, significantly more new bone and cartilage tissues at seven days of post-fracturing, while they exhibited no apparent changes at later stages compared with control animals. The cause of these effects is unclear, but the extended bleeding might have delivered more cytokines to the fracture sites, which boosted the initial stages of the healing. A recent study has addressed the role of α5 integrin in bone repair. Lentiviral-mediated overexpression of Itga5 in human BMSC greatly enhanced bone formation and repair in critical-size cranial and long bone defects in athymic nude mice (Srouji et al., 2012), presumably through activation of the FAK-ERK signaling pathway (Hamidouche et al., 2009). Corroborating with this finding, healing of focal tibial cortical bone defects in mice lacking FAK in osteoblasts is delayed (Kim et al., 2007). Furthermore, a defective bone matrix was deposited perturbing osteoclast attachment and bone remodeling that eventually led to the development of exuberant bony calluses in the mutant mice at the late stage of the repair process. Taken together, the results of these studies support the idea that interfering with integrin activation and/or signaling during fracture healing is a promising target for the development of novel therapies for non- or difficult-to-heal bone injuries.

Tendons and Ligaments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

The functional integrity of the musculoskeletal system depends entirely on the appropriate connections between muscles and skeletal elements. Thus, it is of great importance that the development of cartilage, tendon, and muscle occurs in a precise and well-coordinated manner. So far, our knowledge on the ontogeny of tendons and ligaments lags far behind the other mesodermal tissues, which is mostly because the lack of gene markers that specifically label the tendon and ligament cell lineages and can discriminate the discreet steps of their cell specification and differentiation. Major breakthroughs in the field came upon the identification of the beta helix-loop-helix (bHLH) transcription factor scleraxis as an unique and early marker of tendon and ligament progenitors during embryonic development (Cserjesi et al., 1995; Schweitzer et al., 2001), and from studies in mice harboring genetic mutations leading to tendon and ligament phenotypes (reviewed in Tozer and Duprez, 2005; Liu et al., 2011). Analysis of scleraxis-deficient mice revealed severely disturbed force-transmitting tendons, and disorganized tendon matrix and cellular alignment. However, it seems that scleraxis is a crucial but not the ultimate transcription factor in tendon and ligament development, because in the knockout mice, ligaments and short-range anchoring tendons were not affected (Murchison et al., 2007). This suggests that in the development of these tissues, there are additional transcription and regulatory factors still to be uncovered and studied. Nevertheless, some novel insights into the onset of tendon and ligament development and some of the molecules involved in this process have been provided.

The vertebrate musculoskeletal system originates from segmental blocks of mesoderm, lying along either side of the neural tube and notochord. In response to signals from the surrounding tissues, somites are patterned into distinct compartments giving rise to different cell lineages such as muscle, cartilage, and tendons. Based on the expression of scleraxis, Brent et al. (2003) have identified a novel somite compartment, the syndetome, whose location was mapped at the dorsolateral edge of the early sclerotome. The authors discovered that tendon progenitor cells originate from the developing sclerotome, but require signals, such as FGF8, from the dermomyotome. While the myotome positively regulates tendon development, the sclerotome represses scleraxis via the cartilage-dependent expression of Pax1, Sox5, and Sox6 (Brent et al., 2003). The mechanism by which this repression is mediated, as well as the precise molecular cascade activated by scleraxis resulting into the specification of the tendon progenitor lineage, is currently unknown.

The development of the limb tendons occurs in a different way compared to the tendons from axial skeleton. In the limb, tendons, similarly to cartilage, arise in situ and contrary to the axial tendons, do not require signaling from the myotome (Kardon, 1998). The exact mechanism of limb tendon formation is still unclear. Schweitzer et al. (2001) proposed that the ectoderm plays a stimulatory role in the onset of tendon progenitors, since removal of the dorsal limb ectoderm blocked scleraxis expression in the limb mesenchyme, while surgical excision of the developing muscle did not. Furthermore, analysis of MyoD/Myf5 double mutant mice revealed that scleraxis-positive tendon progenitors are still present in the muscle-less limbs, further confirming that in tendon, limb development muscle-derived signals are dispensable, however they might be required for tendon differentiation and growth (Schweitzer et al., 2001). Certain members of the TGFβ/BMP superfamily can also regulate limb tendon development (Kuo et al., 2008). For example, in the absence of TGFβ signals, as in the TGFβ2/TGFβ3 double-mutant mice and the type II TGFβ receptor null mice, most of the tendons and ligaments are lost, even though the induction of scleraxis-expressing tendon progenitors is not affected (Pryce et al., 2009). Other stimulatory factors are BMP-12, −13, and −14, which are known to induce neotendon and ligament formation when delivered at ectopic sites (Wolfman et al., 1997).

Taken together, tendon and ligament development is coordinated by many factors derived endogenously and from neighboring compartments, however, the exact mechanisms of tendon progenitor specification and differentiation still remain largely unknown. One open question is whether descendants of the tendon progenitors indentified during embryonic development remain existent in adult tendons. Bi et al. (2007) have identified a stem cell niche within human and mouse tendons containing residing tendon stem/progenitor cells (TSPC). These cells exhibit the classical stem cell features self-renewal, clonogenicity, and multipotential; express MSC surface antigens and tendon marker genes such as scleraxis and tenomodulin; and furthermore, have the potential to generate tendon- and enthesis-like tissues in vivo. Further studies have also demonstrated the existence of a TSPC population within human supraspinatus and Achilles tendons, respectively (Tempfer et al., 2009; Kohler et al., 2013). The discovery of TSPC had a major impact in the field, since TSPC might be involved in tendon tissue homeostasis and repair, or they can be practically used in tissue engineering strategies for injured tendons. Still, there is a remaining need to clarify if embryonic tendon progenitors and TSPC are identical cell populations, as well as to generate a solid data for TSPC existence, location and functions in vivo. Improving our knowledge on the above questions can provide novel fundamental understanding, not only on the development of tendon and ligament tissues but also on their sustainability and repair.

Integrin Expression and In Vitro Studies

The integrin expression profile of tendons and ligaments has been investigated over the years; however, most of the experimental data was obtained from cells derived from animals (mice, chickens, dogs, rabbits, and rats), and only a small number of publications have demonstrated integrin expression in human tendons and ligaments (Table 2). Altogether, these studies identified that the α subunits 1-7, 9-11, and V, and the β subunits 1, 3, 5, and 8, are expressed by tendon and ligament cells. The expression of α1, α2, α5, αV, β1, β3, and β5 receptors was reported more often than the other integrin subunits, and was found by using different experimental techniques in various tendon and ligament tissues and cells in few species. This suggests that their expression is widespread in tendons and ligaments. Furthermore, the protein expression of these common receptors in tendons has been recently confirmed by proteomic analyses (Smith et al., 2012). The rest of the integrin subunits seem to be rather specific, for example integrins α3, α6, and α7 were detected mainly in myotendinous junctions (Bao et al., 1993; Conti et al., 2008) and flexor tendons (Harwood et al., 1998; Smith et al., 2012), while α9 and α10 were found up to now only in the cells from anterior cruciate ligament (Brune et al., 2007).

Table 2. Integrin Expression in Tendons and Ligaments
Integrin subunitExpressionSpeciesMethodsReferences
  1. ACL, anterior cruciate ligament; MTJ, myotendinous junction; PCR, polymerase chain reaction; WB, western blot; FACS, flow cytometry; IHC, immunohistochemistry; IF, immunofluorescence; AA, adhesion assay.

α1Achilles and flexor tendons, MTJ TSPC, aponeurotic and ACL cellsMouse, human, chickPCR, WB, FACS proteomicsBao et al. (1993); Quaglino et al. (1997); Brune et al. (2007); Smith et al. (2012); Bell et al. (2013); Kohler et al. (2013)
α2Achilles and flexor tendons Aponeurotic cells, TSPCMouse, humanPCR, WB, FACS, proteomicsQuaglino et al. (1997); Smith et al. (2012); Bell et al. (2013); Kohler et al. (2013)
α3MTJChickIHCBao et al. (1993)
α4Flexor tendonRatIHCJorgensen et al. (2005)
α5Achilles and flexor tendons, MTJ Aponeurotic and ACL cells, TSPC, patellar and ACL fibroblasts, flexor tendon cellsMouse, rabbit, dog, chick, humanPCR, WB, FACS, IF proteomicsBao et al. (1993); Quaglino et al. (1997); Harwood et al. (1998); Harwood et al. (1999); Nawrotzki et al. (2003); Brune et al. (2007); Tetsunaga et al. (2009); Smith et al. (2012); Bell et al. (2013); Kohler et al. (2013)
α6Flexor tendon, MTJMouse, dog, chickWB, PCR, IHC, proteomicsBao et al. (1993); Harwood et al. (1998); Smith et al. (2012)
α7Flexor tendon, MTJMouse, chickWB, IF, proteomicsBao et al. (1993); Conti et al. (2008); Smith et al. (2012)
α8not reported   
α9ACL cellsHumanPCRBrune et al. (2007)
α10ACL cellsHumanPCRBrune et al. (2007)
α11all tendons and ligaments, periodontal ligament embryonic fibroblasts, TSPC, ACL cellsMouse, humanIF, PCRBrune et al. (2007); Popova et al. (2007); Sun et al. (2010); Kohler et al. (2013)
αVAchilles and flexor tendons, MTJ TSPC, ACL cells, flexor tendon cells, ACL and patellar tendon fibroblastsMouse, chick, humanPCR, WB, AA, proteomicsBao et al. (1993); Harwood et al. (1998, 1999); Brune et al. (2007); Mori et al. (2007); Tetsunaga et al. (2009); Smith et al. (2012); Bell et al. (2013); Kohler et al. (2013)
β1Achilles, flexor and patellar tendons, MTJ Aponeurotic and ACL cells, TSPC, ACL fibroblasts, tendon-like MSC, tenocytesMouse, human, rat, dog, chickWB, FACS, IF, PCR, proteomicsBao et al. (1993); Quaglino et al. (1997); Harwood et al. (1998, 1999); Yamamoto et al. (2004); Hannafin et al. (2006); Yao et al. (2006); Brune et al. (2007); Conti et al. (2008); Tetsunaga et al. (2009); Kapacee et al. (2010); Smith et al. (2012); Bell et al. (2013); Kohler et al. (2013)
β2not reported   
β3Achilles tendon TSPC, ACL cells, flexor tendon cells, ACL fibroblastsMouse, dog, humanPCR, IFHarwood et al. (1999); Hannafin et al. (2006); Brune et al. (2007); Tetsunaga et al. (2009); Bell et al. (2013); Kohler et al. (2013)
β4not reported   
β5Achilles tendon Tendon-like MSC, TSPC, ACL cellsMouse, human, dogPCRBrune et al. (2007); Kapacee et al. (2010); Bell et al. (2013); Kohler et al. (2013)
β6not reported   
β7not reported   
β8Tendon-like MSCHumanPCRKapacee et al., (2010)

Several in vitro studies have examined how integrin expression levels change in response to selected growth factors or mechanical stimulation. It was reported that a range of FGF-2 or PDGF-BB concentrations positively influences the expression of the fibronectin receptors α5β1 and αvβ3, which are known to be important during tissue repair, in flexor tendon cells (Harwood et al., 1999). Moreover, the addition of these two growth factors strongly stimulated the outgrowth and proliferation of the tendon cells. Analyses on the effect of FGF-2, BMP-2, GDF-5, and -7 on integrin expression and cell migration of anterior cruciate ligament and medial collateral ligament fibroblasts demonstrated that they strongly modulate integrin α2 levels, as well as α2-dependent cell adhesion and migration (Date et al., 2010). Interestingly, the stimulation with these growth factors altered in a different manner the location of the α2 receptor in both tendon fibroblasts. In the presence of FGF-2 and GDF-5, α2 was enriched in perisomatic filopodia, while when GDF-7 was added, α2 shifted to the filopodial anlagen. Taken together, there is now some evidence that the presence of growth factors can change integrin expression and localization in tendon cells. However, in-depth studies on the crosstalk of growth factors and integrin signaling, as well as integrin-mediated cellular responses of tendon and ligament cells, are still missing.

Recently, we investigated the expression of integrins in TSPC derived from human Achilles tendons of two groups of patients—young and healthy, and with age-related tendon degeneration (Kohler et al., 2013). As a previous study has suggested that distorted TSPC functions can be linked to the progression of tendon pathologies (Bi et al., 2007), we compared the above two types of TSPC, to identify novel molecular factors and mechanisms involved in tendon aging and degeneration. Gene ontology and literature annotation analysis of our microarray findings suggested an intriguing transcriptomal shift in the expression of genes related to stem cell division and maintenance, cell-matrix and cell-cell interactions, actin cytoskeleton, and cell migration. Detailed analysis of mRNA levels of the collagen I-binding (α1β1, α2β1, and α11β1) and fibronectin-binding (α5β1, αvβ3, and αvβ5) integrins revealed that with aging, TSPC downregulate collagen I and elevates fibronectin-binding integrin expression. In addition to the observed dysregulation of the integrin expression during TSPC aging, we found that the knockdown of α2 and α11 integrins in young TSPC impairs their typical ability to remodel collagen I, which was comparable to the result obtained in TSPC from aged/degenerated tendons (manuscript in preparation). This data suggests that changes in integrin-mediated signaling might be involved in the progression of certain tendon degenerative pathologies, which are characterized with compromised matrix quality and abnormal cell–matrix interactions. In conclusion, our studies demonstrated an important role of collagen I-binding integrins for MSCs, and TSPC behavior (Popov et al., 2011; Kohler et al., 2013). Follow up investigations should aim at dissecting the exact mechanisms of how integrins regulate TSPC functions in normal and diseased tendons.

The effect of different mechanical loading protocols on integrin expression levels in tendons and ligaments was the subject of few publications. The expression of integrin α5, β1, and β3 in tendon fibroblasts significantly increased upon stimulation with 5% mechanical stretch (Hannafin et al., 2006). In contrast, no differences were observed in the integrin α1, α5, and β1 levels between control and strained tendon monolayer cultures when stimulated with 2.5% cyclic strain (Henshaw et al., 2006). These results suggest that tendon cells might be responsive to a specific magnitude of mechanical stimuli and only then elevate integrin receptor levels. Interestingly, it has been shown that mechanical stimulation could also influence the location of integrin receptors in tendon fibroblasts. In the first hour of stretching, β1 focal adhesions aligned parallel to the long axes of the cells and were heavily enriched at the polar sides (Kaneko et al., 2009). Hence, this article suggested that there might be a particular time course in which the deposition of the integrin receptors is first re-organized and later stabilized upon mechanical stress. Tetsunaga et al. (2009) found that there is a regional difference in the integrin expression after mechanical stretching. The authors stimulated and compared cells derived from different areas of anterior cruciate ligaments with 7% uniaxial cyclic stretch and found that mid-substance cells increase the expression of integrin α5 and αV, while junction cells elevate only integrin αV, and α5 levels remain unchanged. In summary, the above studies have generated new intriguing data; however, the mechanisms of integrin mechanotransduction in tendon cells remain very unclear and we believe that because of the acknowledged significance of mechanical stress for tendon function and repair, the elucidation of these mechanisms should be a key goal in tendon and ligament research.

Knockout Models

Only few of the established integrin knockout mouse models have been analyzed for tissue and cellular alterations in tendons and ligaments. The major reason is lack of obvious and severe tendon and ligament phenotype, which in turn speaks of powerful compensatory mechanisms between the integrin receptors in these tissues.

Integrins α1 and α2 are widely expressed in mice, whereas the distribution pattern of integrin α11 is restricted to areas of highly organized collagen I, including tendons and ligaments. This restricted expression pattern of α11 suggests a strict transcriptional control. Furthermore, the expression data showed that α11 integrin is expressed in a manner comparative with scleraxis, leading to the notion that this tendon-specific transcription factor might directly regulate α11 expression (Tiger et al., 2001). Integrin α1 and α2 knockout mice have mild phenotypes without tendon pathology. These mild phenotypes and additional studies with cells derived from the mice concluded redundancy and compensation between these receptors (Gardner et al., 1996; Holtkotter et al., 2002; Mercurio, 2002). In contrast, mice that lack integrin α11 are significantly smaller in size than littermate controls, and show abnormal and disorganized periodontal ligaments, leading to failure in the incisor tooth eruption. In addition, the ligaments in α11 knockouts have increased thickness because increased collagen content, as judged by sirius red staining (Popova et al., 2007). The apparent phenotype in periodontal ligaments and not in other ligaments or tendons is because α11 is the sole collagen receptor in periodontal ligamenotocytes. Mice with combined deficiency of α2 and α11 have been recently generated and the overall phenotype has been analyzed; however, so far, it is has not been investigated if there are changes occurring in tendon and ligament tissue and cellular organization (Blumbach et al., 2012). The double mutants displayed skeletal dwarfism and smaller internal organs, which suggests that the growth of tendons and ligaments is also affected. Future investigation of α2- and α11-null tendons and ligaments will be necessary to understand how tendon cells operate in the absence of these two collagen receptors and what are the compensatory mechanisms that take place. Studies with embryonic fibroblasts, ablated from α1, α2, and α11, displayed severe defects in vitro, characterized by greatly reduced spreading and adhesion on collagen I, reduced proliferation, and ability to retract collagen I lattices (Gullberg and Lundgren-Akerlund, 2002; Zhang et al., 2006; Popova et al., 2007). Therefore, it will be of great interest to perform studies with TSPC derived from α1, α2, and α11 single and double mutant mice, and to compare these to their human counterparts with applied integrin knockdown. Despite the expected compensatory effects when single collagen receptor is lost and the mild in vivo phenotypes, understanding the exact role of collagen-binding integrins in TSPC might have a great impact on tendon regenerative strategies where TSPC can be combined with various collagen scaffolds or decellularized tendon matrixes.

Another mouse model with tendon-related phenotype is the knockout of the laminin receptor α7β1. More accurately, these mice display severe myotendinous junction defects, since α7β1 expression is enriched. Ultrastructural analysis revealed that the organization and ECM deposition in α7-deficient myotendinous junctions are disrupted, which was accompanied with the development of muscular dystrophy (Mayer et al., 1997). Furthermore, when α7 knockout mice were challenged by running exercises, extensive muscle damage was induced in areas of high myotendinous junctions and high mechanical force near the Achilles tendon (Boppart et al., 2008). Interestingly, mice ablated for talin or ILK in skeletal muscle, also exhibit destabilized myotendinous junctions, suggesting that proper integrin-triggered signaling is indispensable for proper attachments of muscle to tendon (Conti et al., 2008; Wang et al., 2008). Still, in the above models the analyses were exclusively focused on the alterations in the myotendinous junctions and the muscle structure and function. Therefore, further studies are required to determine if, consecutively to the not normally operating myotendionos junctions, tendon tissue organization, biomechanics, and performance are also affected upon loss of integrin α7 and its downstream signaling.

Taken together, the lack or the available very mild phenotypic changes in tendons and ligaments in mice with single integrin knockout is because of extensive functional compensation by other integrin receptors. Hence, investigation of double knockout mouse models combined with proper in vitro studies will be necessary to understand the importance of integrins in tendon and ligament tissues and cells. Such models can be further used in challenging experiments mimicking certain tendon pathology, for example surgically induced tendon or ligament injury, and provide novel insights on the role of integrins during progression of tendon disease and healing.

Tendon and Ligament Repair

Only very few studies have addressed the involvement of integrins in healing of tendons and ligaments. Schreck et al. (1995) investigated the expression of the integrin subunits β1, α5, α6, and αv in wounded rabbit ligaments, and found a striking increase of β1, α5, and αV levels in cells within the repair site of the medial collateral ligament but unchanged baseline levels in the anterior cruciate ligament. The α6 subunit was expressed exclusively in vascular structures in the healing ligaments. The authors suggested that the known lower healing capacity of the anterior cruciate ligaments might be related to the failure of ligamentocytes to elevate integrin expression subsequently to injury. In a similar study on early healing of a canine intrasynovial flexor tendon, it was shown that the expression of β1 and α5 was increased at ten days post-injury, while the expression of αv and α6 peaked at 2 weeks (Harwood et al., 1998). These results correlate with previous findings on the augmented fibronectin deposition and angiogenesis in the early phases after tendon rupture, suggesting an important role of fibronectin–integrin adhesion events during tendon repair (Amiel et al., 1991; Gelberman et al., 1991). It seems that the fibronectin-binding α5β1 receptor might be also involved during tissue remodeling after ligament immobilization, since it was reported that in stress-deprived ligaments the expression of α5β1 is markedly increased (AbiEzzi et al., 1995). Another study investigated the effect of fatigue on the expression of integrin subunits α11 and β1 in rat patellar tendons and found a differential response, as a downregulation of α11 and an upregulation of β1 was observed. The exact role of these integrins in the tendon cells upon tissue damage because of tendon overload remains unclear.

Taken together, the above studies have delivered new information on the expression of certain integrin subunits in injured, stress-deprived, or fatigue-loaded tendons and ligaments. However, it is important to raise the notion that up to now there is a significant lack of mechanistic knowledge on the role of integrins during tendon and ligament healing. Possible reasons are: (1) the precise cellular and molecular events behind the steps of tendon healing are still poorly understood; (2) there is a deficiency of gene markers that can allow to label or follow different cell subpopulations in tendons and ligaments; (3) because the very dense nature of the collageneous matrix in these tissues, it is difficult to identify focal adhesions and to study changes in their organization and distribution in vivo. Therefore, future studies have to overcome the above issues impeding a deeper understanding of integrin importance in the repair process of tendons and ligaments.

Concluding Remarks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References

Over the years, a vast body of work has addressed and in turn unveiled the importance of integrins in skeletal development, function, and repair. This body of gathered data is powerful enough to provide basis for strategies that aim at affecting or modulating the functions of integrins and consequently the functions of the skeletal cells in their unique niches, thus supporting the processes of tissue repair. Still there are areas needing further exploration, for example, dissecting redundant and unique roles of integrin subunits, integrin functions in tendon and ligament tissues, or coupling of existing knockout strains with disease-mimicking models, which may result in deciphering of molecular pathomechanisms, as well as in translation of basic biological findings into clinical applications.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Integrins
  5. Cartilage Development and Function
  6. SYNOVIAL JOINT AND AC
  7. Bone Formation and Function
  8. Tendons and Ligaments
  9. Concluding Remarks
  10. References