Heparan sulfates in skeletal muscle development and physiology


  • Guido J. Jenniskens,

    1. Department of Biochemistry 194, University Medical Center, NCMLS, Nijmegen, The Netherlands
    2. Biological Engineering Division, Massachusetts Institute of Technology, Cambridge, Massachusetts
    Search for more papers by this author
  • Jacques H. Veerkamp,

    1. Department of Biochemistry 194, University Medical Center, NCMLS, Nijmegen, The Netherlands
    Search for more papers by this author
  • Toin H. van Kuppevelt

    Corresponding author
    1. Department of Biochemistry 194, University Medical Center, NCMLS, Nijmegen, The Netherlands
    • Department of Matrix Biochemistry 194, UMC, NCMLS, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands.
    Search for more papers by this author


Recent years have seen an emerging interest in the composition of the skeletal muscle extracellular matrix (ECM) and in the developmental and physiological roles of its constituents. Many cell surface-associated and ECM-embedded molecules occur in highly organized spatiotemporal patterns, suggesting important roles in the development and functioning of skeletal muscle. Glycans are historically underrepresented in the study of skeletal muscle ECM, even though studies from up to 30 years ago have demonstrated specific carbohydrates and glycoproteins to be concentrated in neuromuscular junctions (NMJs). Changes in glycan profile and distribution during myogenesis and synaptogenesis hint at an active involvement of glycoconjugates in muscle development. A modest amount of literature involves glycoconjugates in muscle ion housekeeping, but a recent surge of evidence indicates that glycosylation defects are causal for many congenital (neuro)muscular disorders, rendering glycosylation essential for skeletal muscle integrity. In this review, we focus on a single class of ECM-resident glycans and their emerging roles in muscle development, physiology, and pathology: heparan sulfate proteoglycans (HSPGs), notably their heparan sulfate (HS) moiety. J. Cell. Physiol. 206: 283–294, 2006. © 2005 Wiley-Liss, Inc.


2-OST; HS2ST, heparan sulfate 2-O-sulfotransferase; 3-OST, heparan sulfate 3-O-sulfotransferase; 6-OST; HS6ST, heparan sulfate 6-O-sulfotransferase; AChE, acetylcholine esterase; AChR, acetylcholine receptor; BL, basal lamina; CHO, Chinese hamster ovary; CS, chondroitin sulfate; DHPR, dihydropyridine receptor; DS, dermatan sulfate; ECM, extracellular matrix; FGF, fibroblast growth factor; GAG, glycosaminoglycan; Gal, D-galactose; GalNAc, N-acetyl-D-galactosamine; GlcA, D-glucuronic acid; GlcNAc, N-acetyl-D-glucosamine; HA, hyaluronic acid; HS, heparan sulfate; HSPG, heparan sulfate proteoglycan; IduA, L-iduronic acid; InsP3R, inositol trisphosphate receptor; KS, keratan sulfate; MuSK, muscle-specific kinase; NDST, N-deacetylase/N-sulfotransferase; NMJ, neuromuscular junction; RyR, ryanodine receptor; sBL, synaptic basal lamina; SR, sarcoplasmic reticulum; T-tubule, transverse tubule; Xyl, D-xylose.

The extracellular connective tissue of muscle (the interstitium) is organized into three discrete but mutually continuous sheets: the epimysium, the perimysium, and the endomysium. The epimysium encircles the whole muscle. Blood vessels and nerves are located within the perimysium, where they divide into a rich network of capillaries and nerve branches extending into the myofiber-ensheating endomysium, which also contains fibroblasts, mast cells, and macrophages. The basal lamina (BL) is a 50–100 nm thick specialized sub-structure of the ECM into which the extracellular domains of cell surface proteins may extend, and which completely envelops individual muscle fibers.


The BL is a macromolecular network, which provides a structural and regulatory environment for cells. It is composed of several collagenous glycoproteins, noncollagenous glycoproteins, and proteoglycans (Table 1; Hall and Sanes, 1993). Whereas collagens are major structural components of BL, the adhesion of cells to collagenous substrates is mediated by noncollagenous glycoproteins. Proteoglycans are major structural and regulatory components of both the BL and the cell surface.

Table 1. Components of the basal lamina and cell surface of muscle cells and their location
  • Due to limitations of space, only a single reference has been cited for each entry. Abbreviations: BL, basal lamina; Cell S, cell surface; c, common to all BL; c/s, common but concentrated (or different constitution) in synaptic BL; e, extrasynaptic BL; n, neural BL; s, synaptic BL

  • a

    Heparan sulfate proteoglycan.

Acetylcholinesterases, BLMcMahan et al. (1978)
Acetylcholine receptors, Cell SSalpeter and Loring (1985)
Carbohydrates (Lectin specific)s, Cell SSanes and Cheny (1982)
Chondroitin sulfate proteoglycanc, BLCarrino and Caplan (1982)
Collagen IV (α1 and α2)e, BLSanes et al. (1990)
Collagen IV (α3, α4, and α5)s, BLMiner and Sanes (1994)
Collagen Ve, BLSanes and Cheny (1982)
Collagen XIIIc, BLSund et al. (2001)
Collagen XVIIIac, BLHalfter et al. (1998)
Decorinc, BLAndrade and Brandan (1991)
Dermatan sulfate proteoglycanc, BLEggen et al. (1994)
Dystrophin–glycoprotein complexc/s, Cell SGrady et al. (2000)
ErbB2, 3, and 4s, Cell SZhu et al. (1995)
Fibronectinc, BLSanes and Cheny (1982)
Glypicansac, Cell SBrandan et al. (1996)
Heparin-binding growth associated molecules, BLSzabat and Rauvala (1996)
Heparan sulfate epitopesc/s, BLJenniskens et al. (2000, 2002)
Heparan sulfate proteoglycans, BLAnderson and Fambrough (1983)
Integrin-α3β1s, Cell SCohen et al. (2000)
Integrin-α7(A/ B)β1s, Cell Svon der Mark et al. (1991)
Laminin-2 (merosin)e,n, BLPatton et al. (1997)
Laminin-3 (s-laminin)s, BLHunter et al. (1989)
Laminins-4, -9, and -11s, BLPatton et al. (1997)
Laminin-8n, BLPatton et al. (1997)
M-cadherinc/s, Cell SCifuentes-Diaz et al. (1996)
Muscle-specific kinase (MuSK)s, Cell SValenzuela et al. (1995)
N-acetylgalactosaminyl-terminated carbohydrates, BLScott et al. (1988)
N-acetylgalactosaminyl-terminated transferases, Cell SScott et al. (1990)
Neural agrinas, BLMcMahan (1990)
Neural cell adhesion molecule (N-CAM)s, Cell SRieger et al. (1985)
Neuregulin-β3s, BLRimer et al. (1998)
Neuregulin-αs, Cell SRimer et al. (1998)
Nexin Is, BLFestoff et al. (1991)
Nidogenc, BLLerner and Torchia (1986)
Perlecanac, BLIozzo (1994)
S-nidogens, BLChiu and Ko (1994)
Syndecan-2as, Cell SHsueh et al. (1998)
Syndecansac, Cell SBernfield et al. (1992)
Voltage gated sodium channelsc, Cell SLupa and Caldwell (1991)
Tenascin-Cc, BLMatsumoto et al. (1994)
Tenascin-Xc, BLMatsumoto et al. (1994)
Tenascin-Yc, BLHagios et al. (1999)

Collagen and laminin self-assemble to form supra-molecular networks, in which cells are firmly anchored through interactions between sarcolemma-resident receptors and BL-resident ligands. Prominent examples are the binding of collagen IV and laminins to integrins, laminin interactions with the dystrophin–glycoprotein complex, and the binding of laminins to cell surface proteoglycans (Fig. 1; Timpl and Brown, 1995).

Figure 1.

Complex interactions between the cell surface and the basal lamina. a: At the neuromuscular junction, the nerve terminal of a motor neuron lies in a shallow gutter formed by the postsynaptic (muscle) membrane. The BL of the muscle and the nerve are continuous, though different in composition. The postsynaptic membrane contains clusters of AChRs on top of the junctional folds (adapted from Hall and Sanes, 1993 (with permission)). b: In extrasynaptic areas the dystrophin glycoprotein complex (composed of dystrophin, α-dystroglycan (α), β-dystroglycan (β), sarcospan, sarcoglycans, and MuSK (M)) interacts with laminin(-2 or -8), which in turn interact with, e.g., nidogen and collagens. c: In synaptic areas the dystrophin glycoprotein complex contains utrophin instead of dystrophin and connects with AChRs through rapsyn (R). Agrin (A), a HSPG, interacts with α-dystroglycan, MuSK, and laminin(-3, -4, -9, or -11), which in turn interact with e.g. nidogen, collagens, and the HSPG perlecan, to which AChE is bound (adapted from Jacobson et al., 2001 (with permission)).

The BL shows structural and functional specializations at sites of interaction with other cells. The neuromuscular nerve terminal and the facing muscle sarcolemma are separated by a 50 nm wide synaptic cleft, filled with the synaptic BL (sBL; Engel, 1994). The sBL is continuous with the BL of the myofiber and the Schwann cell, traverses the synaptic cleft, and extends into the junctional folds. Although morphologically indistinguishable from the extrasynaptic BL, the sBL is highly specialized biochemically. Whereas it only encompasses 0.1% of the myofiber surface, several protein isoforms are concentrated in or specific for the sBL. These include neural agrin (McMahan, 1990), collagen IV isoforms (Sanes et al., 1990), laminin isoforms (Patton et al., 1997), Acetylcholine esterase (AChE; McMahan et al., 1978), and HSPGs (Anderson and Fambrough, 1983). Some collagen IV isoforms are totally absent from the sBL, whereas fibronectin and other collagen and laminin isoforms are shared (Sanes and Chiu, 1983). Studies with lectins or monoclonal antibodies have previously identified several synapse-specific carbohydrates (Sanes and Cheney, 1982; Iglesias et al., 1992). Using phage display-derived antibodies we recently showed that specific HS epitopes are differentially distributed between synaptic and extrasynaptic BL (Jenniskens et al., 2000).

Transverse tubules (T-tubules) are tubular invaginations of the sarcolemma, which penetrate transversely towards the center of the myofiber. The inside of T-tubules is continuous with the extracellular environment and contains ECM (Davis and Carlson, 1994). Satellite cells are undifferentiated multipotent myogenic precursor cells that mediate the postnatal growth and regeneration of muscle. Satellite cells are located between the BL and the sarcolemma of myofibers, but have a separate BL with a discrete composition.

The BL plays an active role in many developmental and physiological processes, by creating spatiotemporal microenvironments of BL-components contributed by the muscle, the innervating nerve, and satellite cells. Focal sequestering of BL-bound biologically active proteins, e.g., growth factors, may be causal for processes such as myogenesis, synaptogenesis, and regeneration. Pre- and postsynaptic specializations re-occur at the site of the original sBL during NMJ regeneration, even in absence of the muscle or nerve cell, respectively (Sanes, 1986), whereas the structural integrity is lost when the BL is degraded (Salpeter et al., 1992).


Collagens are a protein family that distinguishes itself from other glycoproteins by a triple-helical segment. Several collagen isoforms are demonstrated in skeletal muscle ECM, of which collagen XVIII bears HS chains (Halfter et al., 1998) and types IX and XII are part-time proteoglycans, sometimes having GAG side chains. Whereas collagens I and III are major components of the epimysium and the perimysium, collagen V is present in lesser amounts, and isoforms of collagen IV are differentially distributed with regard to the sBL (Miner and Sanes, 1994). AChE contains a collagenous subunit that mediates its BL-anchoring and its synaptic localization, through perlecan and heparin (Brandan et al., 1985).

Down regulation of certain collagen isoforms may be causal for myopathies, as shown for collagen VI and XIII (Lamandé et al., 1998; Kvist et al., 2001).


Laminins are composed of three disulfide-linked subunits: α, β, and γ. Whereas laminin-2 (merosin) is located in the extrasynaptic neural BL, isoforms-3 (s-laminin), -4, -9, and -11 are specific for the sBL (Patton et al., 1997). During myogenesis and synaptogenesis as well as after denervation, the expression of laminin isoforms is tightly regulated (Sanes et al., 1990), thus implicating them in myofiber formation.

Abnormal expression of laminin isoforms induces several muscular dystrophies and congenital myopathies (Tome et al., 1994; Merlini et al., 1999), and aberrant distribution can be used to diagnose congenital muscular dystrophies in mouse and man (Herrmann et al., 1996; Li et al., 1997; Ringelmann et al., 1999).


The hexameric Tenascins are primarily involved in cell adhesion in NMJ regions. Whereas tenascin-C expression is modulated during muscle healing, tenascin-X, and tenascin-Y are involved in myogenesis (Burch et al., 1995; Hagios et al., 1999).

Tenascin-C expression is correlated with muscular dystrophy and with inflammatory and congenital myopathies (Ringelmann et al., 1999), and its deficiency results in abnormal NMJ formation and defective re-innervation (Cifuentes-Diaz et al., 1998).


Many ECM proteins carry covalently attached oligosaccharides or glycans. Glycans are defined as polyhydroxyaldehydes or polyhydroxyketones, depending on their monosaccharide composition. Common eukaryotic classes of glycans are: N-glycans, which are linked to a polypeptide through an asparagine residue; O-glycans, which are typically linked to a serine or threonine residue; a glycophospholipid anchor is a glycan bridge between phosphatidylinositol and a phosphoethanolamine in amide linkage to the carboxy-terminus of a protein; proteoglycans have one or more attached glycosaminoglycan (GAG) chains; and glycolipids, oligosaccharides attached to the lipid moiety ceramide (Varki et al., 1999).

Recently, a surge of reports linked mutations in genes that affect glycan metabolism or protein glycosylation to neuromuscular diseases (considering the scope of this review, we refer the reader to some excellent reviews on this subject: Martin and Freeze, 2003; Martin-Rendon and Blake, 2003; Muntoni et al., 2004).


Proteoglycans are complex macromolecules consisting of a protein core to which one or more GAG moieties are covalently attached. GAGs are linear polymers of repeating disaccharides that consist of a hexosamine and a hexuronic acid. GAGs differ depending on their basic disaccharide unit (Table 2): Hyaluronic acid (HA) is the simplest GAG, consisting of N-acetyl-D-glucosamine (GlcNAc) and glucuronic acid (GlcA) linked by β1-3 and β1-4 bonds, of up to 25,000 disaccharides. HA is the only GAG that is not linked to a protein core.

Table 2. Various types of glycosaminoglycans
GAGDisaccharide composition# UnitsPossible modifications
  1. Gal, D-galactose; GalNAc, N-acetyl-D-galactosamine; GlcA, glucuronic acid: GlcNAc, N-acetyl-D-glucosamine; IduA, L-iduronic acid.

CSGlcA(β1-3)GalNAc(β1-4)2504- or 6-O-sulfation
DSGlcA(β1-3)GalNAc(β1-4)2502-, 4-, and 6-O-sulfation
 IduA(α1-3)GalNAc(β1-4)2502-, 4-, and 6-O-sulfation
HS/HepGlcA(β1-4)GlcNAc(α1-4)150N-deacetylation/ N-sulfation (<50%/>70%), C5-epimerization, 2-, 3-, and 6-O-sulfation
 IduA(α1-4)GlcNAc(α1-4)150N-deacetylation/ N-sulfation (<50%/>70%), C5-epimerization, 2-, 3-, and 6-O-sulfation

There are two galactosaminoglycans: Chondroitin sulfates (CS) have a backbone of up to 250 β1-3 and β1-4 linked N-acetyl-D-galactosamine (GalNAc) and GlcA disaccharides. Depending on the sulfation positions of GalNAc residues, CS is referred to as chondroitin-4-sulfate or chondroitin-6-sulfate. Dermatan sulfates (DS) have a backbone similar to CS, in which individual D-GlcA residues are epimerized to L-iduronic acid (IduA), which may be 2-O-sulfated. By convention, the β1-3 glycoside bond designation changes to α1-3 upon epimerization.

The glucosaminoglycans consist of three types: Keratan sulfates (KS) consist of alternating GlcNAc and galactose (Gal), linked by β1-3 and β1-4 bonds, of up to 80 disaccharides. Sulfation can occur on the 6-position of either sugar residues. Heparan sulfates (HS) have a α1-4 and β1-4 linked GlcNAc-GlcA backbone, approximately 150 disaccharides in length. Certain regions in the HS backbone are extensively modified by N-deacetylation and N-sulfation, C-5 epimerization, 2-O-sulfation, 3-O-sulfation, and/ or 6-O-sulfation, resulting in highly sulfated domains alternating with poorly sulfated domains. Heparin is structurally similar to HS, but has a significantly higher degree of N-sulfation. The term ‘heparin’ is reserved for the GAG component of the proteoglycan serglycin, which is synthesized by mast cells; all structurally related GAGs are called ‘heparan sulfate’ (Kjellén and Lindahl, 1991).

CS, DS, and HS are the most common GAGs in skeletal muscle. The CS/DS proteoglycans biglycan and decorin, a large DS proteoglycan, and a CS proteoglycan are present throughout the perimysium and endomysium (Carrino and Caplan, 1982), whereas a specific CS proteoglycan is located within the T-tubular network (Davis and Carlson, 1994).


Heparan sulfate proteoglycans (HSPGs) are the major PGs in the BL and on the cellular surface of skeletal muscle (Sanes, 1986). One or more HS moieties are attached to variety of core proteins (Fig. 2), both defining the activities of HSPGs. Binding of various ECM components or growth factors can be effectuated by the protein core, the HS sequence, or both. BL-resident proteoglycans include collagen XVIII (Halfter et al., 1998), perlecan (Iozzo, 1994), and agrin (McMahan, 1990). Cell surface proteoglycans include the families of syndecans (Bernfield et al., 1992) and glypicans (Campos et al., 1993).

Figure 2.

Schematic representation of the HSPGs present in muscle. A total of thirteen genes encode agrin, perlecan, collagen type XVIII, six isoforms of glypican, and four isoforms of syndecan core proteins. Splice variants have been reported for agrin, perlecan, and type XVIII collagen. The bottom part provides the keys to the protein modules and HS chain attachment sites (Modified from Iozzo, 2001 (with permission)).


The 220–250 kDa agrin core protein contains two GAG attachment sites (Tsen et al., 1995). Although HS makes up almost half its mass, agrin can also contain O-linked glycans, and has sites for N-linked glycosylation. Neural agrin is a key factor in the clustering and expression of AChRs in the postsynaptic membrane (Fallon et al., 1985; Hoch et al., 1994). Agrin also mediates the clustering of AChE (Wallace et al., 1985; Nitkin et al., 1987) and voltage-gated sodium channels (vgNa+Chs; Lupa and Caldwell, 1991). Alternatively spliced agrin isoforms (Ferns et al., 1992; Ruegg et al., 1992) are developmentally regulated (Hoch et al., 1993) and specifically distributed, especially with respect to the NMJ (Hagiwara and Fallon, 2001). Certain domains in neural agrin isoforms induce postsynaptic differentiation (Reist et al., 1992) and control the stability of these specializations (Denzer et al., 1997). Muscle-derived agrin orchestrates AChR aggregation and their attachment to cytoskeletal proteins (Wallace, 1995).

Agrin interacts with several BL and cell surface proteins, including collagen type IV, laminin isoforms, tenascin, and the dystrophin–glycoprotein complex (Campanelli et al., 1994; Gee et al., 1994). Clustering of AChR by agrin occurs through the MuSK transmembrane receptor tyrosine kinase complex and rapsyn (Willmann and Fuhrer, 2002). Subsequently, AChR clusters are stabilized by tyrosine kinases and by attachment to the dystrophin–glycoprotein complex (Apel et al., 1995). The importance of these interactions is shown in mice knock out for agrin, MuSK, or rapsyn, all of which have severely disrupted NMJs (DeChiara et al., 1996; Gautam et al., 1996, 1999).

Agrin can also interact with carbohydrates (Martin and Sanes, 1995; Campanelli et al., 1996; Gesemann et al., 1996), which may affect its activity: Heparin, HS (Hirano and Kidokoro, 1989; Wallace, 1990), and sialic acid (Grow and Gordon, 2000) inhibit agrin-induced AChR clustering in vitro. Moreover, muscle cells deficient in HS biosynthesis have severely reduced AChR clustering and agrin responsiveness (Gordon et al., 1993).


Perlecan was the first BL-HSPG identified (Hassell et al., 1980). Its 270–467 kDa core protein carries three HS chains at its N-terminal domain that contribute ∼50% of the total mass (Costell et al., 1997). Perlecan plays important roles in the assembly and integrity of virtually all BLs, through its interactions with BL-resident proteins such as type IV collagen, fibronectin, and laminins (Costell et al., 1999). In the sBL, perlecan mediates the anchoring of AChE via its collagen tail (Torres and Inestrosa, 1983; Brandan et al., 1985), as was illustrated in NMJs of perlecan-null mice, which lack AChE (Arikawa-Hirasawa et al., 2002). During myogenesis, perlecan expression is down regulated (Larraín et al., 1997a). Various splice variants have been detected in human, mouse, and in Caenorhabditis elegans (Iozzo, 1994).

Collagen XVIII

Collagen XVIII is the only collagen type shown to be a HSPG. The 180 kDa protein contains ten triple-helical domains and a C-terminal endostatin domain, which interacts with several BL proteins including laminin-1 and perlecan (Sasaki et al., 1998). Three isoforms of collagen XVIII are described in mice and two in human, all of which are expressed in a tissue specific manner (Rehn and Pihlajaniemi, 1995).


Glypicans are anchored in the extracellular face of the plasma membrane via glycosylphosphatidylinositol. The globular core proteins carry HS chains proximal to the plasma membrane. Six mammalian isoforms are reported, and one ortholog (dally) has been described in Drosophila melanogaster. Although specific roles in myogenesis are unknown, glypicans can regulate the activity of growth and survival factors (Filmus and Selleck, 2001). Changes in glypican isoform expression are reported during development (Brandan et al., 1996).


Syndecans are transmembrane proteins with three HS chains attached to the N-terminus, distal from the plasma membrane. Some syndecans also contain CS chains, but biological activity depends on the HS moieties (Langford et al., 1998). The four family members (syndecan 1–4) differ significantly in their extracellular domains, except for the location and number of HS attachment sites (Bernfield et al., 1992). Syndecan isoforms are differentially expressed during skeletal muscle development (Larraín et al., 1997b), and are instrumental in orchestrating cell morphology, adhesion, and migration (Woods et al., 1998), the inhibition of myoblast differentiation (Larraín et al., 1998), satellite cell maintenance and muscle regeneration (Cornelison et al., 2001), and the acceleration of myogenesis (Fuentealba et al., 1999).


The proposed pathway of HS biosynthesis consists of a series of reactions catalyzed by various enzymes, each of which is responsible for a single step (Fig. 3; Lindahl et al., 1998).

Figure 3.

Biosynthesis of heparosan and modification to heparin/ HS. a: The biosynthesis of heparosan occurs on specific serine residues in the core protein by the sequential action of the respective glycosyltransferases to add xylose, two galactose molecules, and a glucuronic acid, followed by the addition of the first N-acetylglucosamine residue by GlcNAc transferase I, to form the GAG linker tetrasaccharide (GlcA(β1,3)Gal(β1,3)Gal(β1,4)Xylβ1-O-). b: HS co-polymerase, which has both GlcA transferase and GlcNAc transferase activity, synthesizes the repeating disaccharides ([-4GlcNAc(α1-4)GlcAβ1-]n) by alternative addition of β-GlcA and α-GlcNAc molecules. c: The bare HS-backbone can subsequently be modified by a series of enzymes. N-deacetylation and N-sulfation is performed by glucosaminyl N-deacetylase/N-sulfotransferase, C5-epimerization by D-glucuronyl C5-epimerase, 2-O-sulfation by heparan sulfate 2-sulfotransferase, 6-O-sulfation by 6-O-sulfotransferase, and 3-O-sulfation by heparan sulfate 3-O-sulfotransferase. Depicted are the consensus pentasaccharide sequence for Antithrombin III binding (GlcNAc6S-GlcA-GlcNS3S6S-IdoA2S-GlcNS6S) and the preferred sulfation pattern for FGF-2/ FGFR ternary complex formation (n = ∼150).

Biosynthesis of the HS backbone (heparosan)

First, a linker tetrasaccharide is attached to the serine residue of a consensus proteoglycan linkage region (Ser-Gly-X-Gly (X represents any amino acid) or Ser-Gly-Asp; Bourdon et al., 1987; Costell et al., 1997; Dolan et al., 1997)). The proximal occurrence of Ser-Gly repeats as well as clusters of acidic amino acids favors glycosylation (Zhang et al., 1995). The linker tetrasaccharide (GlcA(β1-3)Gal(β1-3)Gal(β1-4)Xylβ1-O-Ser) is formed through the stepwise addition of each monosaccharide residue by the respective glycosyltransferase, followed by a GlcNAc by GlcNAc transferase I. Next, the repeating disaccharides ([-4)GlcA(β1-4)GlcNAc(α1-]n) are added by HS co-polymerase, an enzyme with both α-GlcNAc transferase and β-GlcA transferase activity.

Modification of heparosan to HS

The heparosan backbone is subsequently modified by a series of enzymes that convert it into an active HS molecule with functional domains, containing unique sulfation patterns.

  • 1The first modifications of the polysaccharide chain are catalyzed by the bifunctional enzyme glucosaminyl N-deacetylase/N-sulfotransferase (NDST). NDST N-deacetylates and N-sulfates selected GlcNAc residues. Four NDST isoforms are cloned (Hashimoto et al., 1992; Eriksson et al., 1994; Orellana et al., 1994; Aikawa and Esko, 1999). NDST-1 and NDST-2 have a wide tissue distribution (Humphries et al., 1997, 1998; Kusche-Gullberg et al., 1998; Aikawa et al., 2001), whereas NDST-3 and NDST-4 are restrictively expressed. Disruption (Fan et al., 2000; Ringvall et al., 2000) or decreased expression (Aikawa and Esko, 1999) of NDST-1 significantly lowers N-sulfation in HS and is incompatible with life. NDST-2-deficient mice show no HS defects, but are unable to synthesize fully sulfated heparin (Forsberg et al., 1999; Humphries et al., 1999).
  • 2C-5 epimerization of D-GlcA to L-IduA, performed by the enzyme D-glucuronyl C5-epimerase, is the second HS modification (Crawford et al., 2001; Li et al., 2001). No isoforms are known, and disruption of the D-glucuronyl C5-epimerase gene results in neonatal lethality (Li et al., 2003).
  • 32-O-sulfation of uronic acid residues (IduA or GlcA) is conducted by heparan sulfate 2-O-sulfotransferase (2-OST; HS2ST), as shown by studies of the purified (Kobayashi et al., 1996) and cloned (Kobayashi et al., 1997) enzyme, as well as in a CHO (Chinese Hamster Ovary) cell mutant, pgsF-17, defective in 2-O-sulfation (Bai and Esko, 1996). One 2-OST gene has been identified in mammals, which disruption leads to renal agenesis and neonatal lethality (Bullock et al., 1998).
  • 43-O-sulfation of glucosamine residues is catalyzed by heparan sulfate 3-O-sulfotransferase (3-OST), as concluded from experiments with purified enzyme (Liu et al., 1996) and cloned 3-OST-1 (Schworak et al., 1997). Four additionally cloned isoforms (3-OST-2, -3A, -3B, -4, and -5; Schworak et al., 1999; Xia et al., 2002) are differentially expressed in various tissues.
  • 56-O-sulfation of glucosamine residues is performed by heparan sulfate 6-O-sulfotransferase (6-OST; HS6ST), an enzyme that has been purified (Habuchi et al., 1995) and cloned (Habuchi et al., 1998), and of which four isoforms are reported (6-OST-1; -2; -2S; and -3; Habuchi et al., 2000, 2003). These isoforms are tissue-specifically expressed and have different substrate specificities, suggesting important roles in the synthesis of tissue-specific HS epitopes.

The modifications in its backbone make HS a very information-dense biopolymer, to which proteins may bind in a sequence-specific way.


Unique HS sequences are generated through variations in tissue-expression and substrate specificity of HS-biosynthetic enzymes. Inter- and intra-organ distribution of HS epitopes varies in organs at different stages of development (David et al., 1992; van den Born et al., 1992; van Kuppevelt et al., 1998). Mutations in HS-biosynthetic enzymes in Drosophila, Caenorhabditis, vertebrates, and in cell lines demonstrated critical roles for HS in developmental processes and signaling pathways. Knock out mice for various core proteins or enzymes involved in HS biosynthesis have significantly enhanced our understanding of the physiological importance of HS (Forsberg and Kjellén, 2001). Its affinity for a large number of proteins (Table 3), has implicated HS in many cellular processes such as cell proliferation, cell differentiation, angiogenesis, metastasis, tissue repair and (re)modeling, cell adhesion, microbial invasion, and viral infection.

Table 3. Various groups of glycosaminoglycan-binding proteins
  1. Modified from Gunay and Linhardt, 1999

 Lipolytic enzymes
 Carbohydrate hydrolases, eliminases, transferases
 Proteases and esterases
 Nucleases, polymerases, topoisomerases
 Other enzymes, oxidases, synthetases, dismutases
Enzyme inhibitors
 Serine protease inhibitors (serpins)
 Low and very low density lipoproteins
Growth factors
 Fibroblast growth factors
 Epidermal growth factors
 Hepatocyte growth factors
 Platelet-derived growth factors
 Transforming growth factors
 Vascular endothelial growth factors
 CXC chemokines
 CC chemokines
Extracellular matrix proteins
Receptor proteins
 Steroid receptors
 Growth factor receptors
 Channel proteins
Viral coat proteins
Nuclear proteins
 Transcription factors
Other proteins
 Prion proteins
 Amyloid proteins
 Ion channels

HS in development

Unique sulfation patterns on the HS moiety of HSPGs are considered key regulators of developmental signaling. HS sequence requirements are essential for activation of individual members of the fibroblast growth factor (FGF) family (Habuchi et al., 1992; Turnbull et al., 1992; Guimond et al., 1993). This has become the best-studied system in growth factor modulation where HS mediates a ternary complex between FGFs and their receptors, thus facilitating FGF-mediated muscle differentiation (Rapraeger et al., 1991; Yayon et al., 1991). HS is also involved in the binding of other growth factors and associated co-factors, such as hepatocyte growth factor (Lyon et al., 1994), insulin-like growth factor binding proteins (Arai et al., 1994), platelet-derived growth factor (Feyzi et al., 1997), and transforming growth factor-beta (McCaffrey et al., 1992). In Drosophila development, Wnt and Hedgehog (Hh) create morphogenic gradients that tightly regulate developmental gene expression. HSPGs and HS are involved in Wnt and Hh distribution and signaling activities, rendering HS a central modulator in tissue formation (Lin and Perrimon, 2003).

Several models have been proposed to explain how HS modulates development (Lin, 2004). By facilitating ligand–receptor interactions, HS may initiate or enhance cell signaling following the formation of di- or oligomeric ternary ligand–HS–receptor complexes. Morphogenic gradients may be generated by HS, thus regulating the movement of secreted morphogen molecules through restricted diffusion. HS may retain and stabilize ligands in the ECM by limiting diffusion or by increasing their half-life. Finally, by reducing the dimensionality of ligand diffusion from three to two dimensions, HS may create focal concentrations that exceed the activation-threshold for a given signaling pathway.

Only recently, Dally-like glypican was shown to provide a context-dependent feedback mechanism that regulates Wingless morphogen activity (Kirkpatrick et al., 2004; Kreuger et al., 2004). As new regulatory mechanisms are being discovered, it becomes apparent that HS affects developmental signaling pathways in previously unanticipated ways.

HS in myogenesis

Prior to the attribution of specific functions to individual proteoglycans, reports recognized the involvement of ECM components in skeletal muscle development. Several studies established the developmental regulation of proteoglycan expression and HS biosynthesis during myogenesis in vitro (Noonan et al., 1986; Larraín et al., 1997a,b; Jenniskens et al., 2000) and in vivo (Velleman et al., 1999; Cifuentes-Diaz et al., 2000; Cornelison et al., 2001; Olguin and Brandan, 2001; Jenniskens et al., 2002). Specific illustrations include the C. elegans ortholog of perlecan (Unc-52), which is responsible for muscle integrity (Rogalski et al., 1995), the requirement of 6-O-sulfation for muscle development in zebrafish (Bink et al., 2003), and impairment of myofiber and axonal guidance upon mutations in the Drosophila ortholog of syndecan (sdc; Steigemann et al., 2004).

Satellite cells are activated upon muscle injury resulting from mechanical trauma, direct injury, or disease (Bischoff, 1994). As a result, severe changes occur in HSPG expression, notably the transient upregulation of satellite cell marker syndecan-3 (Casar et al., 2004). Knocking out syndecan-4 renders satellite cells unable to reconstitute damaged muscle (Cornelison et al., 2004). Changes in HSPG levels are linked to satellite cell growth factor responsiveness in mdx mice (Crisona et al., 1998) and turkey (McFarland et al., 2003). The addition of heparin, HS, or HS mimetics stimulates in vitro satellite cell proliferation and differentiation, and muscle regeneration (Olwin and Rapraeger, 1992; Meddahi et al., 2002; Papy-Garcia et al., 2002). These HS-like polymers change the natural occurrence of HS species on satellite cells, potentially affecting their myogenic activity (Barbosa et al., 2005).

Current models for the myogenic activity of HS evolve around the dynamic spatiotemporal expression of proteoglycans and HS epitopes. This concert of extracellular glyco-epitopes provides a microenvironment for satellite cell proliferation, differentiation, fusion, and the guidance of myofibers and neural axons through various developmental signaling pathways.

HS in synaptogenesis

Parallel to the recognition of HS involvement in myogenesis, roles emerged in neural differentiation (Nurcombe et al., 1993) and in the genesis of specialized cellular structures like NMJs (Martin, 2002; Yamaguchi, 2002). NMJ-resident HSPGs were identified as structural components of the sBL (Anderson and Fambrough, 1983; Chiu and Sanes, 1984) and implicated in postsynaptic differentiation and in the stabilization of AChR and AChE clusters. In vitro studies confirmed that HS and heparin can block AChR accumulation (Hirano and Kidokoro, 1989; Wallace, 1990) and myotube innervation (Mars et al., 2000). Axonal path finding towards and through muscle primordia during development and regeneration is very accurate and guided by interstitial molecules (Letourneau et al., 1994). Glypican and syndecan are involved in the regulation of Slit-Robo signaling, a ligand–receptor complex responsible for axonal guidance and skeletal muscle development (Ronca et al., 2001; Steigemann et al., 2004). Interestingly, CS and HS act as inhibitor respectively activator of axonal development through the semaphorin family of transmembrane proteins (Qu et al., 2002; Kantor et al., 2004), and HS sulfation patterning affects axonal guidance in C. elegans through sax-3/Robo and kal-3/Anosmin-1 systems (Bülow and Hobert, 2004). We previously described the tightly regulated spatiotemporal distribution of HS epitopes in mammalian NMJ, which may underlie similar mechanisms (Jenniskens et al., 2000, 2002).

Analogous to current ideas on their roles in myogenesis, extracellular guidance clues provided by HS epitopes may modulate various signaling pathways, thus facilitating axonal growth, guidance, and the genesis of synaptic specializations. Furthermore, HS-cues in the sBL may direct regeneratory processes to re-innervate denervated muscle fibers at sites of original NMJs, besides their structural role in healthy muscle.

HS in myopathies

An increasing number of reports link myopathies to inherited or artificially induced defects in HSPGs of skeletal muscle interstitium (Table 4). In mdx mice, an animal model for Duchenne muscular dystrophy, various ECM molecules are over-depositioned (Quirico-Santos et al., 1995), whereas perlecan, syndecan-3, and glypican-1 are upregulated in human Duchenne muscular dystrophy (Alvarez et al., 2002).

Table 4. Inherited or artificially induced defects in cell surface and BL components and the resulting (congenital) interstitial myopathies
AffectedCauseaInterstitial myopathyReferences
  • Due to limitations of space, only a single reference has been cited for each entry.

  • a

    AI, artificially induced; I, inherited.

  • b

    Heparan sulfate proteoglycan.

  • c

    Perlecan ortholog in C. elegans.

AgrinbAICongenital muscular dystrophyMoll et al. (2001)
 AIDefective neuromuscular synaptogenesisGautam et al. (1999)
Collagen VIIBethlem myopathyLamandé et al. (1998)
Collagen XIIIAIProgressive myopathyKvist et al. (2001)
Glypican-3bISimpson-Golabi-Behmel syndromePilia et al. (1996)
Laminin-2IMerosin-negative congenital muscular dystrophyTome et al. (1994)
Laminin-α2ICongenital muscular dystrophyHerrmann et al. (1996)
Laminin-α5AIMerosin-negative congenital muscular dystrophyRingelmann et al. (1999)
Laminin-β1ILimb-girdle muscular dystrophyLi et al. (1997)
Laminin-β1IBethlem myopathyMerlini et al. (1999)
MuSKAIDefective neuromuscular synaptogenesisGautam et al. (1999)
PerlecanbISchwartz-Jampel syndromeNicole et al. (2000)
 ISilverman-Handmaker dyssegmental dysplasiaArikawa-Hirasawa et al. (2001)
 AIAChE clustering defectsArikawa-Hirasawa et al. (2002)
RapsynAIDefective neuromuscular synaptogenesisGautam et al. (1999)
SarcoglycansILimb-girdle muscular dystrophySewry et al. (1996)
Syndecan-3bAIMuscular dystrophyCornelison et al. (2004)
Tenascin-CAINeuromuscular junction defectsCifuentes-Diaz et al. (1998)
Unc-52cAIBody wall muscle integrity and paralysisRogalski et al. (1995)

Individually affected HSPGs include perlecan, which gene is mutated in chondrodystrophic myotonia (Schwartz-Jampel syndrome; Nicole et al., 2000) and dyssegmental dysplasia (Silverman-Handmaker type; Arikawa-Hirasawa et al., 2001). In Schwartz-Jampel syndrome, significantly reduced numbers of full-length perlecan, and various truncated forms are described. Partial impairment of perlecan function causes myotonic myopathy, consistent with its AChE clustering activity at the NMJ (Arikawa-Hirasawa et al., 2002). Moreover, mutations in C. elegans Unc-52 are lethal or lead to paralysis (Rogalski et al., 1995). Agrin knock out induces aberrant synaptogenesis in mice (Gautam et al., 1996, 1999), and its artificial expression rescues congenital muscular dystrophy (Moll et al., 2001). The Simpson-Golabi-Behmel syndrome is attributed to glypican-3 defects (Pilia et al., 1996; Cano-Gauci et al., 1999), whereas knocking out syndecan-3 results in the development of muscular dystrophy (Cornelison et al., 2004).

Although the picture is far from complete, various HSPGs are essential for skeletal muscle physiology and compromising their expression is causal to severe pathologies. The functional loss of cell signaling modulators and structural interstitium elements severely interrupts normal physiology, and regeneration potential upon damage, wear, and natural tissue turn over. Future research may identify pathological roles of other HSPG isoforms, and differentiate between the roles of the core protein and the HS moiety in the pathologic processes.

HS in excitation–contraction coupling

Little is known about the role(s) of HS in skeletal muscle ion housekeeping, let alone about its mechanism of action. The effect of extracellular heparin on mammalian muscle is controversial: the ryanodine receptor (RyR) is activated by heparin in vitro (Ritov et al., 1985; Bezprozvanny et al., 1993), and the heparin antagonist protamine reversibly blocks the RyR (Koulen and Ehrlich, 2000). Paradoxically, the structurally similar inositol trisphosphate receptor (InsP3R) is inhibited by heparin (Gosh et al., 1988; Kobayashi et al., 1988), whereas in frog muscle heparin cannot activate RyR nor inhibit InsP3R (Pape et al., 1988; Rojas and Jamovich, 1990). Heparin also affects the kinetic properties of the dihydropyridine receptor (DHPR; Knaus et al., 1990; Martinez et al., 1996): intracellular application blocks excitation–contraction coupling in toad (but not in rat) muscle fibers, an effect likely caused by desensitizing the DHPR voltage sensor (Lamb et al., 1994). Ion buffering by the negatively charged biopolymers has been proposed as a mechanism for ion channel modulation by HS or heparin (Gosh et al., 1988; Bezprozvanny et al., 1993). Direct evidence for the involvement of HS in muscle calcium kinetics was obtained in myoblast cultures in which HS epitopes were physiologically knocked out (Jenniskens et al., 2003a) and in primary myoblast cultures from mice carrying targeted disruptions in the NDST-1 gene (Jenniskens et al., 2003b).

The modulating effect of HS on ion housekeeping can be envisioned to depend on sequence-specific molecular interactions with various ion channels. Alternatively, HS may alter local ion concentrations in the direct vicinity of these transmembrane proteins through the sequestering of cations (notably calcium), thereby regulating ion channel activity.


In this review, we summarized the current knowledge on HS in skeletal muscle ECM, its biosynthesis, and emerging roles in muscle development, physiology, and pathology. Whereas the involvement of HS in myogenesis and synaptogenesis is well established, new roles for this class of polysaccharides are emerging in muscle regeneration, interstitial myopathies, and muscular ion housekeeping.

We have only just begun to grasp the tremendous complexity of the processes by which HS mediates a wealth of biological roles. Expression profiling of the entire glycome (all glycosylated proteins and the enzymes involved in glycan biosynthesis and depolymerization) may shed light on the occurrence and involvement of glycans, notably HS, in developmental, physiological, and pathological processes. Concomitant sequencing of the HS epitopes that bind myogenic factors and characterization of these protein–HS interactions will greatly improve our understanding of the roles that HS plays in underlying signaling pathways. Currently, we are unable to routinely sequence biologically relevant HS sequences or make statements on their specific occurrence on individual HSPG (sub)species. This is indicative for the fundamental nature of some of the major challenges that the field of glycomics has yet to overcome.

The sheer complexity of HS biosynthesis and its various biological activities require a very creative combinatorial employment of genetic, cell biological, biochemical, and analytical approaches to elucidate the mechanism by which this information-dense biopolymer exerts its functions. In conclusion, the exploration of the intriguing field of skeletal muscle glycobiology has only just kicked off, but holds substantial challenges for years to come.