Proteoglycans are abundant components of the cell surface and extracellular matrix that mediate critical interactions between cells and their environment. They play a variety of biological roles in normal tissues and in response to injury and disease. Proteoglycans regulate the distribution of extracellular signaling molecules such as morphogens and chemokines, and modulate signaling events at the cell surface that influence cell fate determination, proliferation, adhesion and motility. Secreted proteoglycans can also serve as structural constituents of the matrix, whereas cell surface proteoglycans may participate in endocytosis and vesicular trafficking, regulating the movement of molecules between intracellular and extracellular compartments.
Proteoglycans consist of a core protein carrying long unbranched disaccharide chains. Different types of chain are distinguished by the nature of the repeating disaccharides; heparan sulfate (HS) contains alternating N-acetyl glucosamine (GlcNAc) and glucuronic or iduronic acid residues. Heparan sulfate proteoglycans (HSPGs) carry one or more HS chains, and can be grouped into three distinct classes: the syndecans, which have a single transmembrane domain; the glypicans, which are linked to the outer plasma membrane by a glycosylphosphatidylinositol (GPI) anchor; and a varied group of secreted proteoglycans, including perlecan, agrin and collagen XVIII. HS chains are attached at specific serine residues on the core proteins through a tetrasaccharide linker. Following chain initiation, HS is synthesized by glycosyltransferases of the exostosin (EXT) family. Initially the chains contain only GlcNAc and glucuronic acid, but following chain elongation, additional modifications take place: GlcNAc N-deacetylase/N-sulfotransferase (NDST) converts N-acetyl groups in some regions to N-sulfates, C5 epimerase modifies nearby glucuronic acid residues to iduronic acid and specific sulfotransferases introduce additional sulfate groups (see Esko and Selleck, 2002; Kramer and Yost, 2003; Bulow and Hobert, 2006). These modification reactions do not go to completion, which leads to enormous structural variety in HS chains. This diversity allows HSPGs to bind to a wide range of proteins and influence many biological processes.
Genetic analysis has shown that both the core proteins and HS chains contribute to proteoglycan function, but they are not equally important in all contexts. Mutations causing global disruption of HS synthesis (for example, in the EXT genes) have severe developmental consequences, such as early embryonic lethality (Bulow and Hobert, 2006), as does disruption of the secreted HSPG perlecan (Iozzo, 2005). However, many core protein or HS modification mutants survive early development and show surprisingly mild or specific phenotypes (Bulow and Hobert, 2006). Analysis of these mutants reveals important roles for HSPGs in a variety of processes such as eye development, neuronal pathfinding and angiogenesis. Detailed genetic characterization also suggests that redundancy and compensatory mechanisms exist in vivo to safeguard the functions of these critical molecules (Merry et al., 2001; Grobe et al., 2002; Kamimura et al., 2006; Kreuger et al., 2006).
The classic view of HSPGs is that they serve as signaling co-receptors that bind via their HS chains to various ligands and promote interactions with cognate receptors. FGF was the first growth factor shown to depend on HS for interaction with its receptor (Rapraeger et al., 1991; Yayon et al., 1991; Pellegrini et al., 2000; Schlessinger et al., 2000). This general model has been extended to many other pathways, including Hedgehog, BMP and Wnt signaling, all of which are affected by loss of EXT function (Bornemann et al., 2004; Han et al., 2004; Takei et al., 2004). However, several lines of evidence suggest that signaling by some growth factors relies on contributions from the protein core in addition to the HS chains (Kramer and Yost, 2003). Mutant glypicans that lack HS chains retain some capacity to promote Wnt signaling (Ohkawara et al., 2003; Capurro et al., 2005; Kirkpatrick et al., 2006), demonstrating that the core protein is sufficient in some instances. Although HSPGs are generally thought to enhance the activity of receptors on the same cell, they have also been shown to promote VEGF signaling in trans on adjacent cells (Jakobsson et al., 2006). Glypican, syndecan and perlecan mutants all display phenotypes consistent with signaling defects, and the affected signaling pathways have been identified in some cases. Different glypicans specifically influence Wnt, BMP and/or Hedgehog signaling, syndecans are known to regulate Wnt and Slit signaling, and perlecan affects FGF and Hedgehog signaling (Jackson et al., 1997; Lin and Perrimon, 1999; Tsuda et al., 1999; Paine-Saunders et al., 2000; Capurro et al., 2005; Topczewski et al., 2001; Ohkawara et al., 2003; Desbordes and Sanson, 2003; Lum et al., 2003; Alexander et al., 2000; Johnson et al., 2004; Merz et al., 2003; Park et al., 2003; Lee and Chien, 2004; Tkachenko et al., 2005; Bulow and Hobert, 2006). These findings suggest that serving as a co-receptor for cell-cell signaling is a general property of many HSPGs.
Cell adhesion and invasion
In addition to promoting growth factor signaling, some HSPGs modulate integrin signaling to influence cell adhesion and motility. Syndecans, in particular, are thought to affect integrin responses to extracellular matrix cues through lateral interactions on the cell surface. Syndecan 1 (Sdc1) regulates cell attachment and spreading by stimulating αvβ3 or αvβ5 integrin signaling output in response to low levels of a matrix ligand (Beauvais and Rapraeger, 2003; McQuade et al., 2006). The mechanism of Sdc1 action is unknown, but these activities are mediated by its ectodomain, which may help to cluster the integrin or recruit it into an active signaling complex with the ligand. A different region of the Sdc1 ectodomain has been shown to inhibit invasion of B lymphoid ARH-77 cells into collagen gels (Liu et al., 1998; Langford et al., 2005), but again the mechanism is not understood. These different Sdc1 activities vary in their dependence on HS chains, which are not required to stimulate αvβ3 signaling but are needed in the other two cases (Couchman, 2003; Beauvais et al., 2004; McQuade et al., 2006).
Syndecan 4 (Sdc4), which is found with α5β1 integrin at focal adhesions, influences cell adhesion in a different way. In response to an extracellular ligand such as fibronectin, the Sdc4 cytoplasmic domain binds phosphatidylinositol 4,5-bisphosphate and directly activates PKCα, which in turn activates RhoA to promote focal adhesion assembly (Oh et al., 1997; Saoncella et al., 1999; Couchman, 2003; Lim et al., 2003; Dovas et al., 2006). Thus, Sdc4 and Sdc1 co-signaling activities depend on different domains of the HSPG core protein. Also, focal adhesion assembly requires both the integrin-binding and HS-binding domains of fibronectin.
HSPGs can also function as cell adhesion receptors themselves. For example, they mediate the initial interaction of circulating neutrophils with vascular endothelial cells during inflammation. Disrupting HS modification specifically in endothelial cells weakens the binding of neutrophil L-selectin to endothelial cells and alters neutrophil rolling velocity (Wang et al., 2005). Mutant animals with a targeted disruption of Ndst1 in endothelial cells show impaired neutrophil invasion in vivo, demonstrating that HSPGs are physiologically important for this process.
HSPGs also affect the stability and distribution of signaling molecules in the extracellular matrix. Mutations that block HS synthesis prevent accumulation of several different morphogens in mutant cells and also affect morphogen levels in adjacent wild-type cells (Bornemann et al., 2004; Han et al., 2004; Takei et al., 2004). Binding to HS restricts the range of some signals, such as BMP4 in the early Xenopus embryo (Ohkawara et al., 2002), VEGF-A in mouse vascular development (Ruhrberg et al., 2002) and Ihh during mouse chondrocyte differentiation (Koziel et al., 2004). In these cases, removing the HS-interacting domain from the ligand or reducing HS levels leads to increased ligand dispersal and broader signaling. In other contexts, HSPGs can have very different effects: the Drosophila glypican Dally-like (Dlp) limits Wingless accumulation near the morphogen source and promotes its spread across the wing disc (Kirkpatrick et al., 2004; Kreuger et al., 2004; Franch-Marro et al., 2005; Han et al., 2005; Marois et al., 2006). The mechanism of Dlp activity is not known in detail, but it depends on shedding of Dlp from the cell surface by the enzyme Notum, which cleaves Dlp at its GPI anchor (Kreuger et al., 2004). Since HSPGs influence both the stability of extracellular signals and their ability to spread through tissues, the net effects of HSPGs on a morphogen gradient depend on the balance between these activities in a particular context. Thus the mechanism of HSPG action in a new situation is difficult to predict a priori and must be determined experimentally.
HSPGs also regulate the tissue distribution of chemokines, small proinflammatory signaling proteins that form chemotactic gradients to recruit neutrophils to sites of injury. HSPGs are not required for chemokines to recruit leukocytes in vitro, but are essential for chemokine activity in vivo because they retain chemokines on endothelial cell surfaces, allowing gradient formation (Proudfoot et al., 2003; Wang et al., 2005). In addition, HSPGs play a role in transcytosis of chemokines from the injured tissue across the vascular endothelium to the lumenal surface of blood vessels (Middleton et al., 1997; Wang et al., 2005). Shedding of Sdc1 (with bound chemokine) from the cell surface by matrix metalloprotease cleavage increases chemokine migration into injured lung tissue, promoting inflammation (Li et al., 2002); however, shedding can also limit inflammation by reducing chemokine levels on endothelial cells (Marshall et al., 2003), highlighting the complexities of HSPG function in this process.
Although HSPGs are best known for their roles at the cell surface and in the extracellular matrix, evidence indicates they have additional functions in intracellular membrane trafficking. Mice lacking NDST2, one of the enzymes responsible for the initial steps of HS modification, have mast cell secretory granule defects and do not make heparin, which implicates HSPGs in secretion (Forsberg et al., 1999; Humphries et al., 1999). Cells with a defect in HS synthesis cannot secrete FGF-2 (Zehe et al., 2006). FGF-2 does not have a signal peptide and is not secreted by the conventional mechanism of translocation into the endoplasmic reticulum and transport through the Golgi (Nickel, 2005). The mechanism of FGF-2 export has not been elucidated, but it requires HS chains on the cell surface (Zehe et al., 2006).
HSPGs also participate in endocytosis. It has been known for some time that endocytosis of low density lipoprotein (LDL) by the LDL-receptor-related protein (LRP) requires HSPGs (Mahley and Ji, 1999), but recent work has shown that HSPGs independently mediate uptake of triglyceride-rich lipoproteins in the liver (MacArthur et al., 2007). Because LRP participates in Wingless/Wnt signaling (Pinson et al., 2000; Wehrli et al., 2000), and internalized Wingless can be depleted by heparinase treatment of Drosophila wing discs (Greco et al., 2001), HSPGs might participate in endocytosis of Wnts, but this has not been demonstrated directly. Sdc2 is internalized from the plasma membrane and recycled to the cell surface through its interaction with the adaptor protein syntenin, which provides a link to the Arf6/PIP2 recycling pathway (Zimmermann et al., 2005). Interestingly, other proteins that associate with syndecans, such as the FGF receptor and β1 integrin, accumulate in recycling endosomes when Arf6-mediated recycling is blocked, which suggests that syndecans may mediate the trafficking of a variety of proteins to and from the cell surface.
The various functions of HSPGs described above are probably interconnected, and the role of HSPGs in complex biological processes such as angiogenesis, muscle development or nervous system function may involve a combination of these activities. For example, the binding of growth factors to HSPGs influences both their extracellular distributions and their ability to signal through their receptors. Also, the localization of glypicans to lipid rafts within cell membranes (Sharom and Lehto, 2002) and the role of Sdc4 at focal adhesions indicate that HSPGs may participate in creating cell surface microdomains where the critical players in a particular process are assembled into functional complexes (Couchman, 2003). Further studies of HSPG function will reveal whether they can internalize cell surface proteins into endosomes and thus regulate the localization of signaling receptors as well as ligand distributions. In addition, identification of novel HSPG-interacting proteins [such as LAR, a phosphatase important for nervous system development (Johnson et al., 2006)] and careful analysis of mutant phenotypes will provide new insights into their varied roles in development. Another important focus of current and future research is how specific HS sequences contribute to HSPG function and whether some features of HS structure are less critical for activity in certain biological contexts.
- © The Company of Biologists Limited 2007