Heparan Sulfate in Skeletal Development, Growth, and Pathology: The Case of Hereditary Multiple Exostoses

Abstract

Heparan sulfate (HS) is an essential component of cell surface and matrix-associated proteoglycans (HSPGs). Due to their sulfation patterns, the HS chains interact with numerous signaling proteins and regulate their distribution and activity on target cells. Many of these proteins, including bone morphogenetic protein family members, are expressed in the growth plate of developing skeletal elements, and several skeletal phenotypes are caused by mutations in HS-synthesizing and modifying enzymes. The disease we discuss here is Hereditary Multiple Exostoses (HME), a disorder caused by mutations in HS synthesizing enzymes EXT1 and EXT2, leading to HS deficiency. The exostoses are benign cartilaginous-bony outgrowths, form next to growth plates, can cause growth retardation and deformities, chronic pain and impaired motion, and progress to malignancy in 2-5% of patients. We describe recent advancements on HME pathogenesis and exostosis formation deriving from studies that have determined distribution, activities and roles of signaling proteins in wild type and HS-deficient cells and tissues. Aberrant distribution of signaling factors combined with aberrant responsiveness of target cells to those same factors appear to be a major culprit in exostosis formation. Insights from these studies suggest plausible and cogent ideas about how HME could be treated in the future.

Introduction

Heparan sulfate (HS) is a versatile and essential component of cell surface and matrix-associated proteoglycans (HSPGs) (Bernfield et al., 1999). Due to their specific chemistry and highly negative charge, the HS chains can bind to a number of proteins, including growth factors, signaling proteins, integral membrane receptors, chemokines, and extracellular matrix proteins (Hacker et al., 2005; Bishop et al., 2007). Studies have indicated that in particular, the HS chains endow the proteoglycans with the key ability to regulate the distribution and availability of the growth and signaling proteins and their respective interactions, function and bioactivity on target cells (Bernfield et al., 1999; Lin, 2004; Umulis et al., 2009). Because many of these proteins, including members of the hedgehog, bone morphogenetic protein (BMP), fibroblast growth factor (FGF) and Wnt families, are expressed in the growth plate (Kronenberg, 2003), it is evident that HS influences many important processes in skeletogenesis and skeletal growth and morphogenesis. Mice deficient in Ext1, an essential HS polymerizing enzyme, show altered patterns of Indian hedgehog (Ihh) diffusion, increasing the range of signaling and resulting in significant changes in growth plate morphology (Koziel et al., 2004; Hilton et al., 2005). N-sulfotransferase 1 (Ndst1) is another critical enzyme in HS assembly, establishing tissue-specific sulfation patterns by replacing acetyl groups with sulfate modifications. Loss of Ndst1 also causes severe changes in Hedgehog distribution and growth plate function (Yasuda et al., 2010). The significance of HS chains and HSPGs in skeletogenesis is reiterated by the fact that there are a number of skeletal and craniofacial phenotypes related to genetic mutations in HS-synthesizing and modifying enzymes and in HSPG expression (Bishop et al., 2007). A case in point is Hereditary Multiple Exostoses (HME), a pediatric autosomal-dominant disorder during which cartilage outgrowths called exostoses form next to the growth plate of skeletal elements such as long bones, ribs and pelvis and protrude into the adjacent perichondrium and neighboring tissues (Porter and Simpson, 1999; Hecht et al., 2005). In turn, the exostoses can cause skeletal deformities, chronic pain and early onset osteoarthritis, among a variety of other pathological events (Dormans, 2005; Jones, 2011). HME –also called Multiple Osteochondroma (MO) or Multiple Hereditary Exostoses (MHE) - is estimated to affect 1 in 50,000 children, and is almost 100% penetrant. The majority of cases of HME are caused by loss-of-function mutations in EXT1 and EXT2, which encode for Golgi-associated glycosyltransferases. After a number of other enzymes create a linkage tetrasaccharide on a serine residue of the core protein, Ext1 and Ext2 are collectively responsible for polymerization of the HS chain (Esko and Selleck, 2002). The number and type of EXT mutations are many and lead to varying degrees of HS deficiency (Jennes et al., 2009). What is not fully understood is whether and how the HS deficiency leads to exostosis formation, whether it can account for all the various symptoms and complications of the syndrome, and what could be done therapeutically to treat or reverse it. Several recent studies reported by this and other groups have aimed to decipher the possible roles of HS deficiency on exostosis formation and HME pathogenesis in general, and have also provided clues on the genesis of other complications of the disease. In so doing, these studies have contributed to further understand also the normal roles of HS and HSPGs in skeletal development and growth. This review article summarizes the studies and provides a critical assessment of current findings and underlying hypotheses.

HME Pathology and Population Studies

HME presents with predominantly orthopedic manifestations. The exostoses are the most evident trait of the syndrome from which it takes its name, and are composed of a growth plate-like cartilaginous cap overlaying a bony base (Fig. 1) (Jones, 2011). Because of their location, size, number and interactions, the exostoses can cause: compression of nerves, blood vessels, and tendons with consequent pain and impairment of motion; skeletal deformities and growth retardation by interfering with normal growth plate function; and early onset osteoarthritis (Fig 2, A-B). Patients with HME often show slightly shortened stature, bowing and shortening of forearm elements, changes in angulations of the knee and fingers, and limb-length inequalities. Previous hypotheses suggested that exostoses growing along the bones of the forearm affected their shape and introduced a “steal phenomenon” that caused bone shortening. However, recent experiments show no change in overall bone volume and no correlation between forearm shortening and the presence of an osteochondroma. This work indicates that HS loss may be important for directing the ratio between peripheral and longitudinal growth (Jones et al., 2013). Exostoses can also sometimes occur in craniofacial elements such as the mandible (Ruiz and Lara, 2012). Chronic pain and restrictions in activity are common consequences of the disease. A recent national cohort study in the Netherlands found a significantly lower outcome in physical functioning and resultant role limitations, social functioning, vitality, pain, and general health perception in patients with HME compared with three separate reference groups (Goud et al., 2012). Unfortunately, as these symptoms and difficulties progress, patients must undergo surgery, with 70% of patients undergoing an operation by the time they reach 18 years of age. In children, surgery can be dangerous as it can cause irreversible damage to the adjacent growth plate. Although no new exostoses form after puberty when the growth plates close, existing exostoses can continue to grow and cause further pain and complications, leading to a surgical rate of 67% in adults (Goud et al., 2012).

FIGURE 1.

Histological comparison of a typical human growth plate and an exostosis. (A) Longitudinal section through a normal growth plate shows the stratified organization of the chondrocytes and their distinct morphologies in each zone. Note that the chondro-osseous border is located at the bottom of the picture. (B) The exostosis contains similar diverse populations of chondrocytes but absence of clear organization. Note that the chondro-osseous border is located on the left reflecting the 90° orientation of the exostosis relative to the longitudinal axis of the adjacent growth plate. The cartilaginous portions stains strongly with Safranin-O, while the adjacent bone is stained with fast green. (B, Inset) A macroscopic view of exostosis tissue removed from a patient. Stereotypic exostoses are continuous with normal medullary and cortical bone and are covered by perichondrial tissue. Bar for A and B, 75 μm; bar for inset, 2 mm.

FIGURE 2.

Images from patients with Hereditary Multiple Exostoses. (A) Frontal and (B) lateral X-ray images of the forearm of a 14 year-old HME patient reveal the presence of a large exostosis at the distal end of the radius (arrowheads). (C) X-ray image and (D) CT scan reveal the presence of a chondrosarcoma lesion near the scapula in a 17 year-old patient (arrows). The humerus also contains exostoses (arrowheads). (E-F) These intraoperative CT scans demonstrate the presence of an exostosis lesion within the spinal canal at the T12 level (arrowhead).

One potentially serious complication of HME is the transformation of benign exostoses to malignant chondrosarcomas, a life-threatening progression of this syndrome that occurs in about 2 to 5% of patients (Fig. 2, C-D). For patients whose exostoses undergo malignant degeneration, the mean age at time of diagnoses- 35 years old- is younger than that for general HME patients (Bjornsson et al., 1998). Another serious complication of HME is the formation of exostoses on the surface of the vertebrae, which can compress the spinal cord or nerve roots (Fig. 2, E-F). Spinal cord compression due to exostoses has been seen to manifest as motor or sensory deficits including gait disturbance, weakness or numbness, amplified reflex responses and spasticity, and incontinence (Zaijun et al., 2011; Bari et al., 2012). Vertebral exostoses can also impinge on the esophagus, impairing normal swallowing (Perrone, 1967). Additionally, exostoses on both vertebral as well as costal surfaces can interfere with lung function and can cause spontaneous hemothorax, pneumothorax, and pericardial effusion, typically leading to immediate surgical intervention (Assefa et al., 2011).

It is interesting that, although HS production in HME patients is likely to decrease in all tissues, the only truly apparent and diagnostic phenotype are the exostoses themselves. However, as indicated above, HME patients can suffer from a variety of less obvious problems that can include wound healing delay, learning disabilities, dental problems and others (Hosalkar et al., 2007; Wiweger et al., 2012). This clinical and biomedical complexity, though still not fully understood and clear, certainly relates well with the fact that HSPGs regulate numerous, if not most, physiologic processes in the growing and adult organism. Thus, a generalized deficiency in HS could, and should, have broad and widespread consequences the severity of which would reflect the specific importance and roles that HS and HSPG have in different tissues and organs and distinct biological contexts and processes (Bishop et al., 2007).

HS Synthesis and HME Genetics

Heparan sulfate (HS) constitutes the glycosaminoglycan moiety of such cell surface and matrix proteoglycans as syndecans, glypicans and perlecan (Bernfield et al., 1999). The HS chains are composed of repeating d-glucuronic acid (GlcA) and N-acetyl-d-glucosamine (GlcNAc) residues that are assembled into linear polysaccharides in a biosynthetic process including initiation, elongation, and modification steps. Synthesis is initiated in the endoplasmic reticulum by the addition of a xylose to a serine residue in the proteoglycans core protein by xylosyltransferase. The linker tetrasaccharide is completed in the Golgi apparatus by the addition of two galactose residues and a glucuronic acid, carried out by galactyltransferases I and II and glucuronosyltransferase I, respectively. HS elongation is then performed in a stepwise manner by the Golgi-associated heterodimer complexes of EXT1 and EXT2 that are ubiquitously expressed type-II transmembrane glycoproteins (Esko and Selleck, 2002). Both proteins and expression of both alleles are required to produce and maintain normal physiologic HS levels and homeostasis. During assembly, the nascent chains undergo extensive modifications that include N-deacetylation/N-sulfation, epimerization, and O-sulfation, requiring a number of enzymatic reactions and providing a range of HS structural variability between and within tissues that confer structural and protein-binding specificity (Bulow and Hobert, 2006).

EXT1 and EXT2, originally considered to be tumor-suppressor genes, are part of a family of EXT proteins that also includes three EXTL (exostosin-like) members. The latter proteins have roles in HS synthesis as well, including addition of the first carbohydrate residue to the tetrasaccharide linkage (Kitagawa et al., 1999; Kim et al., 2001; Busse et al., 2007; Okada et al., 2010). Mutations in either EXT1 or EXT2 are responsible for about 90% of HME cases, with the majority (~65%) seen in EXT1, the protein believed to have enzymatic activity within the EXT1/EXT2 complex (EXT2 plays a structural role and may also have chaperone functions) (McCormick et al., 1998). To date, over 650 unique mutations have been found in these two genes, most of which are nonsense, frame shift, or splice-site mutations (Jennes et al., 2009; Ciaverella et al., 2012). These inactivating mutations result in premature termination of EXT proteins, causing premature degradation and nearly complete loss of function (Wuyts and Van Hul, 2000). Very recently, a family of patients negative for EXT1 or EXT2 mutations was shown to have intronic rearrangement within the first intron of EXT1, suggesting a possible mechanism for HME that would not be detected with conventional diagnostic techniques (Waaijer et al., 2013).

Several studies have been carried out with the goal of establishing a relationship between genotype and phenotype, but no firm relationship has been obtained yet. Interestingly, there is a wide spectrum of variation in disease presentation within families and between individuals with the same mutation, suggesting additional genetic, hormonal, or environmental influences in phenotype specification and severity. However, some general correlations have been determined in a number of HME population studies. Due to its enzymatic function, EXT1 mutations are correlated to more severe presentations of the disease (Francannet et al., 2001; Porter et al., 2004). Additionally, male patients typically show a more severe clinical presentation, which is hypothesized to be caused by a later growth plate closure, allowing more time for exostosis formation. Accordingly, patients with a greater number of exostoses (>20) usually have more disabilities and deformities (Alvarez et al., 2006; Pedrini et al., 2011).

Animal Models of HME

The development of skeletal elements in the skull, trunk and limbs initiates with the formation of ecto-mesenchymal and mesenchymal cell condensations at prescribed times and locations. Several condensations located in the skull region undergo intramembranous ossification and produce skeletal elements such as the calvaria and jaw. The remaining and more numerous condensations undergo endochondral ossification during which the condensed mesenchymal cells differentiate into chondrocytes and become organized into growth plates closely surrounded by perichondrial tissues. The growth plate chondrocytes proliferate, undergo hypertrophy and are replaced by endochondral bone and marrow, thus sustaining skeletal growth until the end of puberty and producing definitive skeletal elements throughout the body that include ribs, vertebrae and long bones (Kronenberg, 2003). As pointed out above, the growth chondrocytes express both cell surface and matrix-associated HSPGs including syndecan-3, glypicans-5 and perlecan that are required for function, including regulation of growth factor distribution and action and relationships of growth plate with surrounding tissues and most importantly perichondrium (Arikawa-Hirasawa et al., 1999; Viviano et al., 2005; Habuchi et al., 2007; Yasuda et al., 2010). As pointed out above, the exostoses exhibit an intriguing growth plate-like organization in which their main axis of elongation is at about 90° angle compared to that of the adjacent native growth plate. Understanding the role of HS and HSPGs in the growth plate is thus critical for elucidating the processes by which exostoses can form and grow next to, but never within, the growth plates and protrude into perichondrium and surrounding tissues.

A number of zebrafish models have been developed to assess changes in HSPGs and their affect on cartilage development. Dackel (dak/ext2) mutants lack HS chains and showed a severe cartilage phenotype, with disorganized cells that failed to flatten and intercalate into stacks and lost expression of collagen10a1. However, these cells are able to express markers of both early chondrocytes as well as perichondrium, suggesting that they are capable of forming components of an exostosis. Interestingly, pinscher mutants (pic/slc35b2) exhibit an even more severe phenotype, with thickened perichondrium and reduced matrix deposition. Pic mutants are unable to transport sulphate donors into the Golgi and produce sulphate-less GAGs (including keratin sulfate and chondroitin sulfate). This demonstrates the necessity of GAG chains in maintaining proper chondrocyte and perichondrial cell phenotype and morphology (Wiweger et al., 2011). Transplant experiments show that most dak mutant cells can be rescued by surrounding WT cells, taking on a proper flat and intercalated phenotype. However, these cells occasionally grow out from the edge of developing cartilage, oriented perpendicular to the WT cells; this growth mimics developing exostoses in humans that consistently form along the side of the growth plate, extending into the perichondrium. This model suggests that the location of mutant cells within a cartilage element may dictate their responsiveness to changes in growth factor distribution or ability to contact neighboring WT cells (Clement et al., 2008).

Over the last decade or so, several mammalian models of HME have also been developed to uncover and understand exostosis pathogenesis as well as the roles of HS in normal mammalian skeletal development (Table 1). Ext1-null mice are embryonic lethal at E8.5, indicating the essential importance of HSPGs for sustaining and regulating critical developmental stages (Lin et al., 2000). Indeed, as early as day E10.5, Ext1 expression becomes conspicuous in the developing limb buds (Stickens et al., 2000). Interestingly, mice heterozygous-null for Ext1- originally created as a model of HME- were found to be largely normal and to lack an obvious skeletal phenotype (Hilton et al., 2005). However, upon closer examination their growth plates exhibited subtle changes including an increase in collagen II expression, a decrease in collagen X expression, and increased BrdU incorporation in the proliferative zone, collectively suggesting altered chondrocyte proliferation and maturation patterns. Because the overall length of limb bones in adult mutant mice were similar to those in wild-type mice, it was proposed that the above effects of Ext1 and HS deficiency would be compensated over time (Hilton et al., 2005). Mice lacking Ext2 were also found to be embryonic lethal and the embryos actually failed to undergo gastrulation, likely as a result of a disruption of several signaling pathways critical to mesoderm development and formation of extra-embryonic structures (Stickens et al., 2005). Heterozygous null Ext2^+/− mice showed growth plate disturbances similar to those in their Ext1^+/− counterparts, with a disorganized proliferative zone and changes in Ihh expression domains. Of significant interest was the fact that a small percentage of the Ext2^+/− mice did develop small exostosis-like outgrowths along the ribs near the costochondral junction, supporting the idea that a partial loss of Ext function and HS production may be sufficient for exostosis formation (Stickens et al., 2005).

Table 1.

Summary of current models of EXT deficiency in mice.

Genotype	Skeletal phenotype	Detection of exostoses*	Reference
Ext1+/−	Minor alterations in growth plate chondrocytes	None	Hilton, 2005
Ext2+/−	Minor alterations in growth plate chondrocytes	Small exostoses in ribs of <20% of mice	Stickens, 2005
Ext1+/−;Ext2+/−	Bowed forearms, shortened HS chains	Stereotypic exostoses in long bones in 50% of mice; exostoses display growth plate-like characteristics	Zak, 2011
Ext1Gt/Gt	Shortened skeletal elements, delayed chondrocyte differentiation, prenatal lethal at E15	None	Koziel, 2004
Col2-rtTA-Cre;Ext1e2neofl/e2neofl	Shortened limbs	Aggressive epiphyseal osteochondromas in all mice	Jones, 2009 Jones, 2012
Col2-CreERT;Ext1fl/fl (stochastic model)	Bowed forearms, scoliosis, short stature	Overgrowth of growth plate cartilage to form cartilage-capped bony protrusions	Matsumoto, 2010a
Prx1Cre;Ext1fl/fl	Severe skeletal abnormalities, joint fusion, shortened limbs, missing carpals/metacarpals, postnatal lethal	None	Matsumoto, 2010b
GDF5Cre;Ext1fl/fl	Shortened skeletal elements, joint fusion in digits, postnatal lethal	Small exostoses in all mice	Mundy, 2010 Huegel, 2013
Col2CreER;β-cateninfl/fl	Changes in chondrocyte proliferation and differentiation	Lateral outgrowth of the growth plate to form chondroma-like masses	Cantley, 2013

^*Note that original nomenclature has been replicated in this table. The term exostosis is considered synonymous with a number of other terms including osteochondroma and chondroma.

Intriguingly, however, the HS levels observed in surgical retrieval specimens of human exostosis cartilage were found to be very low and apparently lower than the 50% levels presumably caused by a single heterozygous EXT mutation (Hecht et al., 2005). Loss-of-heterozygosity could explain such low HS levels, but LOH has actually been documented in only a few cases (Bovee et al., 1999) though it may have been underestimated originally because of the mixed nature of cells within exostoses (Jones, 2011). Thus, it is possible that secondary mutations in other HS-related genes, action of HS degrading enzymes such as heparanase (Trebicz-Geffen et al., 2008), or other mechanisms may play roles in reducing HS levels further. To test these ideas, we recently created and analyzed double heterozygous Ext1^+/−;Ext2^+/− mice (double hets) (Zak et al., 2011). Because a full complement of Ext proteins is needed for normal levels of HS synthesis, it was predicted that these compound mice would have a significant decrease in HS levels (about 25%) compared to single heterozygous mutant (about 50%) and wild types (100%). Indeed, we found that in addition to developing rib exostosis-like outgrowths as seen in single Ext2^+/−heterozygous mice, almost half of the compound Ext1^+/−;Ext2^+/− mutants exhibited stereotypic exostoses next to the growth plates of long bones and deformations in their pelvis. The exostoses displayed a growth plate-like arrangement of chondrocyte zones with typical expression patterns of zone markers such as Sox9, Col2, Indian hedgehog (Ihh) and Col10. The HS levels in growth plates and exostoses of these mice were significantly reduced compared to those in wild-type littermates as indicated by immunostaining, and the HS chains that were produced by cultured compound mutant chondrocytes were shorter than those produced by WT chondrocytes. We also found evidence of changes in chondrocyte response to growth factors important for normal skeletal development (Zak et al., 2011). These phenotypes strongly support the idea that significant, but not necessarily complete, loss of Ext expression and HS production is sufficient for formation of multiple stereotypic exostoses. Thus, it appears that exostosis initiation and frequency are inversely related to overall Ext expression and function and increase almost linearly in single het vs double hets vs WT. This finding may explain the broad range of phenotype severities seen in HME patients in whom EXT protein function and HS production may vary depending on the nature of the EXT mutation and other concurrent genetic modulations.

Other mouse models of HME have provided additional information about exostosis pathogenesis and the role of HS in normal growth plate development. A hypomorphic, gene-trapped Ext1^Gt/Gt mouse line expressing a truncated form of Ext1 displayed shortened skeletal elements and fused vertebrae at E15.5 (Koziel et al., 2004). These changes were accompanied by increased distribution and signaling of Ihh within the growth plate and delayed hypertrophic differentiation, reiterating the idea that Ext expression and HS production are needed to regulate action and distribution of signaling proteins within the growth plate. Postnatal, chimeric and conditional inactivation of Ext1 in chondrocytes in mouse growth plate cartilage was found to result in the formation of numerous exostoses throughout the appendicular skeleton, indicating that a full Ext loss leads to aggressive exostosis formation (Jones et al., 2010). Conditional deletion of Ext1 from limb mesenchyme utilizing Prx1-Cre transgenic mice caused severe limb skeletal defects including joint fusion and variable numbers of developing digits. These effects were at least partially attributed to broader and non-physiologic distribution of bone morphogenetic proteins (BMPs) and BMP signaling activity on targets (Matsumoto et al., 2010).

HS and Signaling Proteins

The above studies clearly indicate that HS and HSPGs are critical to determine and supervise the distribution of signaling proteins, their range of action, and the effects exerted on their targets. They also make it plain that HS deficiencies can have profound repercussions on those mechanisms and could directly contribute to pathologic changes. Interactions between growth factors and HS have been thoroughly studied for many years in a multitude of organ systems, cell types, and in vivo scenarios. One major overall insight stemming from all this work is that however, those interactions and relationships are contextually dependent and can specifically vary depending on the developmental stage of the tissue or organ, the types of cells involved, the presence of other systemic and local factors, etc. (Rider and Mulloy, 2010). One well-characterized HS interaction is the one with members of the fibroblast growth factor (FGF) family. In this relationship, HS is most often required for effective signal transduction as it acts as a FGF co-receptor (Hacker et al., 2005). Early stage Ext2 null embryos do not in fact respond to FGF signaling, conceivably an explanation for their early embryonic lethal phenotype (Shimokawa et al., 2011). Fibroblasts isolated from Ext1^Gt/Gt mice show decreased amounts of cell surface HS as expected and, in addition, exhibit a reduced signaling response to FGF2 and consequent decrease in proliferation after growth factor stimulation (Osterholm et al., 2009).

Hedgehog signaling is another important regulator of axial and limb skeletal development. Since primary cilia are largely responsible for this signaling pathway, there is much evidence that they are also essential for skeletal development (Huangfu and Anderson, 2005). Conditional deletion of the primary cilium component Kif3a in chondrocytes resulted in both limb and cranial skeletal abnormalities, including exostosis-like cartilaginous masses forming near the growth plate. We showed that unexpectedly, the mutant Kif3a-null growth plates displayed a drastic reduction in HSPG expression (and in particular syndecan-3 and perlecan) and a concomitant broader distribution of Ihh within the growth plate and all along the adjacent perichondrium. This was associated, and probably directly caused, ectopic hedgehog signaling all along the perichondrium, ectopic chondrogenesis and then local formation of ectopic exostosis-like cartilaginous masses, suggesting a role for hedgehog proteins and defective Kif3a-related mechanisms in exostosis formation as well (Koyama et al., 2007). Primary cilia are also at least partially responsible for organizing chondrocytes into their growth-plate specific columnar structures. Substantiating the role of the primary cilia in HME, this polarity has been found to be lost in human osteochondroma cells, suggesting significant changes in cell adhesion and motility and related cell-matrix and cell-cell communication mechanisms (de Andrea et al., 2010).

Bone morphogenetic proteins (BMPs), a subfamily of the TGFβ superfamily of secreted proteins, also regulate a number of stages in skeletal development. Depletion of HS from the surface of C2C12 cells enhanced BMP2 bioactivity while inhibiting its internalization (Jiao et al., 2007). Disruption of HS chains with exogenous heparinase also increased pSmad1/5/8 signaling in human mesenchymal stem cells (Manton et al., 2007). Several members of both the BMP and FGF families are expressed in growth plate and/or perichondrium and have been shown to be part of interactive loops regulating Ihh and PTHrP expression and overall growth plate activities (Zou et al., 1997; Pathi et al., 1999). Misexpression of Ihh causes changes in the expression of pro-chondrogenic BMPs as well as their anti-chondrogenic antagonists Noggin and Chordin, altering the differentiation of cells in the growth plate as well as the bordering perichondrium (Pathi et al., 1999).

During mesenchymal condensation in early skeletal development, Wnt ligands induce an accumulation of β-catenin in the cytoplasm. In turn, β-catenin translocates to the nucleus and binds to transcription factors, controlling downstream gene transcription to determine an osteogenic lineage. Ablation of β-catenin initiates cartilage formation while transgenic overexpression of Wnt signaling promotes osteoblast differentiation (Day et al., 2005). Wnt ligands also bind tightly to HSPGs at the cell surface, which act to maintain the solubility of hydrophobic Wnt and stabilize its activity (Fuerer et al., 2010; Kikuchi et al., 2011). The Wnt/β-catenin signaling pathway also plays important roles in cartilage maintenance and growth plate function (Yuasa et al., 2009). Interestingly, postnatal β-catenin ablation in cartilage causes exostosis-like cartilage masses to form off of the growth plate as well as in the periosteum. Additionally, cartilage tissue collected from osteochondromas of HME patients showed little to no β-catenin positive cells, potentially extending the lifespan of exostosis chondrocytes (Cantley et al., 2012). In sum, the above and related studies have provided clear evidence for essential roles that HS-dependent mechanisms have to orchestrate and coordinate the different functions and processes within the growth plate as they relate to the function and action of signaling proteins and factors (Minina et al., 2002).

Signaling Proteins and Exostosis Initiation

The above studies hint to the possibility that a decrease in Ext expression/HS levels and a concomitant increase in signaling protein distribution/action may initiate ectopic chondrogenesis and exostosis formation. To test these possibilities more directly, we recently resorted to in vitro studies in which we isolated mouse embryo limb mesenchymal cells and grew them in chondrogenic conditions in micromass culture (Huegel et al., 2013). Cultures were maintained in absence or presence of Surfen, a small molecule heparan sulfate antagonist (Schuksz et al., 2008). Indeed, Surfen treatment caused a dose-dependent increase in the number of alcian blue-positive cartilaginous nodules (Fig 3, A-C). Interestingly, the nodules in Surfen-treated cultures had lost their typical round, individual morphology and fused with one another, indicating that the nodules were unable to maintain their border and circumscribed perimeter normally occupied by flat perichondrium-like cells (Ahrens et al., 1979). Isolated RNA from the micromass cultures was subjected to qPCR. Surfen treatment clearly caused a significant increase in expression of characteristic chondrogenic genes including aggrecan, collagen type II, Runx2 and Sox9, confirming that loss of HS function stimulates chondrogenic differentiation of progenitor cells (Fig 3D). The same effects were observed when Ext1^fl/fl limb bud cells in micromass cultures were treated with a Cre-expressing adenovirus, thus reducing Ext expression and HS production; the effects and responses were abrogated and reduced by addition of the BMP antagonist Noggin (not shown) (see (Paine-Saunders et al., 2002)).

FIGURE 3.

Pharmacological interference with HS stimulates chondrogenesis. (A-C) Mouse embryo limb bud mesenchymal cells in micromass cultures were treated with vehicle (control) or 5 or 10 μM Surfen for 5 to 7 days. Note that the treated cultures display a much larger number of alcian blue-positive cartilaginous nodules and that several of them are fused into each other. (D) qPCR analysis shows that expression of indicated cartilage marker genes is up-regulated by Surfen in a dose-dependent manner in the micromass cultures. Vertical bars in each histogram indicate standard deviations (Huegel et al., 2013).

As discussed previously, HS mediates the interaction of chondrogenic factors with target cells, limiting their availability, distribution and action (Bernfield et al., 1992; Lin, 2004). Thus, we reasoned that Surfen could have stimulated chondrogenesis by increasing availability and action by signaling proteins. For proof-of-principle, we chose BMP signaling since it has strong pro-chondrogenic activity (Weston et al., 2000). Indeed, BMP signaling -as indicated by pSmad1/5/8 phosphorylation levels- was greatly enhanced by Surfen treatment of limb mesenchymal micromass cultures and its increase preceded the increase in cartilage nodule formation. These responses were abrogated by addition of the BMP antagonist Noggin. Reporter plasmid assays confirmed that cells treated with Surfen were more sensitive to both endogenous and exogenous BMP2 (Huegel et al., 2013). To verify these observation in vivo, we conditionally deleted Ext1 in perichondrium flanking the upper portions of the growth plate in Ext1^fl/fl;Gdf5Cre mouse embryos. We found that many mutant perichondrial cells became positive for pSmad1/5/8 signaling in the nucleus, while most cells in controls were negative. Limb explants treated with Surfen showed similar changes in signaling (Fig 4, C-D). This ectopic BMP signaling was followed by formation of ectopic cartilage within perichondrium (Fig 4, A-B). Collectively, these studies suggest that Ext1 and HS are critical regulators of the perichondrium phenotype -allowing it to act as an anti-chondrogenic border around the growth plate- and are also essential to restrain and contain cartilage growth. In an Ext1- and HS-deficient environment, BMP signaling would be enhanced and mis-regulated, leading to abnormal behavior and growth of chondrocytes and enhanced chondrogenic response of perichondrium.

FIGURE 4.

Heparan sulfate inhibition leads to ectopic BMP signaling and cartilage formation. (A-B) Histological images of Safranin-O/fast green-stained epiphyses showing that ectopic cartilage was forming along the flanking perichondrium in Surfen-treated explants (outlined in D), while the chondro-perichondrial border was continuous and intact in controls. (C-D) Immunostaining images showing that phosphorylated Smad1/5/8 staining is apparent in the perichondrium flanking the epiphysis of Surfen treated explants (arrowheads in B), while there is background staining in control long bones (Huegel et al., 2013).

Other Modifiers of HS

Another interesting aspect of the biology of HS chains and HSPGs is that the chains can be modified extracellularly by the action of enzymes such as heparanase (HPSE) and sulfatases (Ai et al., 2007; Fux et al., 2009). Indeed, increased expression of HPSE has been described in exostosis tissue (Trebicz-Geffen et al., 2008; Yang et al., 2010). Responsible for cleaving HS chains into small fragments, the endoglucuronidase HPSE has pro-proliferative activity and is implicated in a range of cancers by assisting in the structural remodeling of the extracellular matrix during cellular invasion and release of growth factors (Ilan et al., 2006). This invasive characteristic is also seen in tooth development when dental follicular cells penetrate through the epithelial root sheath to differentiate into cementoblasts, which coordinates with up-regulated HPSE expression (Hirata and Nakamura, 2006). Additionally, increased levels of HPSE are present at the chondro-osseous junction in developing bones, suggesting that HPSE plays a role in late chondrocyte differentiation during endochondral ossification (Brown et al., 2008). It also suggests that regulated degradation of HS chains could promote factor release and signaling during other key points of the bone formation process.

Sulfatases modify extracellular HS chains by selectively cleaving 6-O-sulfate groups, altering the structural pattern and heterogeneity and impacting interaction with signaling molecules (Ai et al., 2007). The idea that the HS sulfation pattern can be fluid after synthesis provides another level of cell or tissue-specific regulation of HSPG function. During Xenopus embryogenesis, Sulf1 enzyme expression is highly regulated, and acts negatively on both BMP and FGF signaling, allowing for the formation of morphogen gradients vital to axis patterning, somitogenesis, and other key developmental processes (Freeman et al., 2008). More recently, articular cartilage in Sulf1^−/− mice was shown to display a spontaneous osteoarthritis phenotype with decreased BMP and upregulated FGF signaling (Otsuki et al., 2010). Evidently, Sulf expression is also necessary for maintaining articular cartilage homeostasis by regulating signaling between chondrocytes.

Concluding Remarks and Future Directions

The above studies have provided major new insights into the interplays amongst HS biology, growth plate function and exostosis formation as well as the complexities and subtleties of these mechanisms. It appears plausible and likely therefore that exostosis formation is the ultimate outcome of changes in HS-dependent signaling pathways, including BMP and Ihh pathways, converging to create pro-chondrogenic responses and proliferative environments along the border of growth plates. Exostosis formation would thus result from increased distribution of these factors, increased responsiveness of the cells to these factors, and reduced capacity of growth plate cartilage and/or perichondrium to remain distinct and phenotypically stable.

Other aspects of exostosis biology remain to be clarified. For instance, exostoses have been shown to often consist of a mix of HS-expressing and HS-null cells (Jones, 2011), suggesting that the cell population within each exostosis may be varied and phenotypically diverse and may have varying developmental origins. With regard to the latter, different hypotheses have been put forward over the years regarding the origin of exostosis forming cells. One hypothesis is that the cells represent borderline growth plate chondrocytes that would misbehave as a result of Ext/HS deficiency and lack of Ext tumor suppressor function, thus behaving as benign tumor cells and solely responsible for exostosis formation (Fig 5A). A second hypothesis is that the cells would originate in perichondrium and would be progenitors in nature (Fig 5B). The cells would lose their fibrogenic/progenitor phenotype and become reassigned to the chondrogenic lineage, and would undergo de novo chondrogenesis and give rise to the exostoses. The third possibility is that the cells would originate in the groove of Ranvier, a specialized region of perichondrium near the epiphysis which is rich in progenitor cells and may contain a specialized stem cell niche (Fig 5C). Data in favor of, and against, these various hypotheses exist in the literature, thus requiring further work and more refined tools to be clarified and defined in a conclusive manner. As pointed out above, it may be that these hypotheses are not mutually exclusive and that exostosis formation could co-involve growth plate and perichondrial/groove cells. Their common denominators would be: enhanced responsiveness to growth factors; greater availability of those factors to act; increased enzymatic degradation of existing HS chains by heparanase; and inclusion of a mixed cell population (Fig 5D). Based on our studies, we believe the contribution of perichondrial/groove cells may be preponderant.

FIGURE 5.

Schematic summarizes current hypotheses regarding the origin of exostosis-forming cells. Those cells are currently thought to be: (A) growth plate chondrocytes themselves (blue); (B) perichondrial cells (red); or (C) cells originating in the groove of Ranvier (purple). Data in favor and against each of these theses have been reported in recent years. Our own work indicates that perichondrium, including the groove of Ranvier, could represent a source of these cells. It is also possible that the exostosis-founding cells could reside in growth plate or perichondrium at the onset of exostosis formation, and would subsequently recruit cells from surrounding sites to sustain and boost the outgrowth process. Thus, exostosis formation could involve more than one source of cells (D). Other factors may also play a role in initiation and growth of ectopic cartilage, including increased range and responsiveness of growth factor signaling as well as upregulated heparanase.

An equally critical issue to be addressed in the future is whether and how exostosis formation could be prevented or even reversed therapeutically. As discussed above, symptomatic exostoses are removed surgically at present, but surgery is dangerous, can have complications and cannot be used to remove each exostosis since the number of exostoses is usually very high in each patient. Hence, biological solutions are needed to aid surgery. Given that the exostoses likely represent a de novo chondrogenic process involving growth plate chondrocytes, perichondrial cells or both, it is plausible to assume that anti-chondrogenic tools could be effective to block exostosis formation. Powerful anti-chondrogenic mechanisms include the retinoid pathway, Wnt signaling, and BMP antagonists such as Noggin and Gremlin. We have recently used acute activation of retinoid signaling via nuclear retinoic acid receptor-selective synthetic agonists to prevent and block heterotopic ossification, an ectopic endochondral processes triggered by trauma or gain-of-function activin receptor 1a mutations (Shimono et al., 2011). It is conceivable that such therapy may work in HME as well, and we will be testing this thesis in one of our HME mouse models in the near future. Another possibility is that microRNAs could be used to treat HME as it is being currently tested in other cancer fields. A recent study has shown that the miR transcriptome in human exostosis tissue differs from that of normal human cartilage pointing to multiple putative therapeutic targets (Zuntini et al., 2010). Last but not least, the identification of Surfen as an HS antagonist shows that HS can be affected and modulated by pharmacological means. Thus, it is possible that HS agonists could be identified and would increase HS bioactivity, reduce turnover or increase EXT expression. Whatever their mechanisms of action, the drugs would then be tested for ability to increase or restore HS function/levels in HME mouse models and eventually patients, reduce exostosis formation and ameliorate other symptoms. Our genetic studies show that exostosis formation does not require a complete loss of Ext function to occur. It may be that pharmacological prevention of exostosis formation may require a significant, but not a major and complete, increase in HS function and levels as well.

• Heparan sulfate proteoglycans and a number of growth factors they control are expressed and active in the growth plate and surrounding perichondrium

• Congenital mutations in HS-synthesizing and modifying enzymes and HSPG expression cause severe skeletal and craniofacial phenotypes

• Recent developments in understanding Hereditary Multiple Exostoses (HME) suggest that aberrant growth factor signaling plays a major role in exostosis initiation and growth