An Introduction to Proteoglycans and Their Localization

Abstract

Proteoglycans comprise a core protein to which one or more glycosaminoglycan chains are covalently attached. Although a small number of proteins have the capacity to be glycanated and become proteoglycans, it is now realized that these macromolecules have a range of functions, dependent on type and in vivo location, and have important roles in invertebrate and vertebrate development, maintenance, and tissue repair. Many biologically potent small proteins can bind glycosaminoglycan chains as a key part of their function in the extracellular matrix, at the cell surface, and also in some intracellular locations. Therefore, the participation of proteoglycans in disease is receiving increased attention. In this short review, proteoglycan structure, function, and localizations are summarized, with reference to accompanying reviews in this issue as well as other recent literature. Included are some remarks on proteoglycan and glycosaminoglycan localization techniques, with reference to the special physicochemical properties of these complex molecules.

Proteoglycans are present in all members of the Bilateria and therefore have a long evolutionary history. Moreover, they are widespread components of most tissues and organs and can sometimes be a dominant component (e.g., in the vertebrate cartilage matrix). Studies of proteoglycans initially focused on this tissue, since the major component, now known as aggrecan (Aspberg 2012), is abundant and readily purified. However, it is now clear that virtually all extracellular matrices contain proteoglycans, and they are also present in the pericellular matrix of many cell types. Moreover, they are not restricted to extracellular locations, since some are specialized components of membrane-bound secretory granules—for example, heparin and highly sulfated forms of chondroitin are present in the granules of mast cells, other leukocytes, and endodermal and endothelial cells (see Rönnberg et al. 2012). Several reports also suggest that proteoglycans can become localized in the nucleus, although its functional significance remains unknown.

Given that proteoglycans are ubiquitous in multicellular animals and occur in many tissues in many forms, it is unsurprising that they are ascribed a variety of functions. Some are structural, such as aggrecan and its relatives, others control enzymatic activity, and still others serve as cell surface receptors. In development and tissue repair, proteoglycans may have key roles in controlling gradients and availability of potent growth factors, chemokines, cytokines, and morphogens. The roles are many but incompletely understood. However, interest is increasing in these molecules. Although their structural complexity is considerable, as explained here and in other reviews in this volume, their importance in homeostasis, development, repair, and disease, notably cancer, is apparent. Proteoglycan variation in form and function reaches its peak in mammals where the sophistication of the immune system and variety of extracellular matrices that comprise tissues across this group are underlying factors.

The accompanying series of short reviews on proteoglycans cover a range of topics relevant to mammalian biology and disease. They follow from the Nordic Proteoglycan Workshop held at the Copenhagen Biocenter, University of Copenhagen, on April 16-17, 2012. The workshop was funded, in large part, by the Alfred Benzon Foundation.

Proteoglycans—What Are They?

By definition, proteoglycans consist of a core protein to which one or more glycosaminoglycan chains are covalently attached. Glycosaminoglycans (GAGs) themselves are unbranched, often long polysaccharides with a repeating disaccharide structure. The number of glycosaminoglycan types is rather few, and their structures are shown in Figure 1. Correspondingly, the number of proteins that become substituted with GAGs is also small, and many of these fall into major families. For example, the large hyaluronan-binding (aggregating) proteoglycans, the lecticans (Yamaguchi 2000), consist of four members, aggrecan, versican, neurocan, and brevican. Aggrecan is the major proteoglycan of cartilage, whereas versican is enriched in large blood vessels. Neurocan and brevican are concentrated particularly in the specialized extracellular matrices of the brain. These have chondroitin sulfate as their major GAG, although aggrecan also possesses keratan sulfate. Versican is subject to alternate mRNA splicing that can give rise to several isoforms, one of which contains no GAGs (Lemire et al. 1999). A quite different family of proteoglycans is the syndecans, which are transmembrane receptors capable of signaling independently or in combination with other receptors, such as fibroblast growth factor receptors or the cell adhesion receptors, integrins (Couchman, 2003; 2010; Morgan et al. 2007; Murakami and Simons 2008). Table 1 summarizes some of the major groups of proteoglycans. It is likely that all mammalian extracellular matrices and all cell surfaces, with the exception of erythrocytes, possess one and, frequently, multiple proteoglycans.

Figure 1.

Composition of the glycosaminoglycans. Heparan sulfate (HS), chondroitin sulfate (CS), and dermatan sulfate (DS) all have a common stem oligosaccharide consisting of xylose and two galactose residues followed by a single glucuronic acid. Heparan sulfate (and the related highly sulfated heparin) then consists of a repeating disaccharide of N-acetylglucosamine and glucuronic acid. Subsequently, some of the N-acetylglucosamine is deacetylated and N-sulfated (NS). This usually occurs in blocks along the chain. These sulfated domains then undergo uronic acid epimerization, converting glucuronic acid to iduronic acid, some of which is then 2-O-sulfated (2S). Further modifications are 6-O and (rarely) 3-O sulfation of the hexosamine residues. In heparan sulfate, these modifications do not go to completion but culminate in domains of high sulfation (often referred to as NS domains) interspersed in regions of low or no sulfation (NA domains). Zones of intermediate sulfation can occur at the boundaries between these two extremes (NS/NA). Chondroitin and dermatan sulfates are similar: both consist of a repeating disaccharide of N-acetylgalactosamine and glucuronic acid. However, dermatan sulfate, by definition, has some of the uronic acid epimerized to iduronic acid by epimerase enzymes. Some of this iduronic acid can be sulfated at the 2-O position (2S). In the example shown here, both the CS and DS chains are sulfated at the 4-O position on the N-acetylgalactosamine (4S). However, chondroitin can be unsulfated or 6-O sulfated, and commonly a chain can contain more than one type of sulfation. Oversulfated forms of CS are known; for example, chondroitin sulfate E contains N-acetylgalactosamine sulfated at both the 4- and 6-O positions. Oversulfated forms of DS are also known. Hyaluronan (HA) comprises N-acetylglucosamine and glucuronic acid in a repeating disaccharide and may form chains of Mr = 1 × 10⁶ or more. It is synthesized at the cell surface and is therefore not modified by, for example, sulfation. Keratan sulfate (KS) consists of repeating disaccharides of galactose and N-acetylglucosamine, both of which can be 6-O sulfated.

Table 1.

Characteristics of Some Human Proteoglycans

Proteoglycan	Core Protein Size, kDa	Type of GAG Chains	Human Chromosome Localization	Tissue Location
Glypicans
Glypican 1	56	HS	Chromosome: 2 location: 2q35-q37	GPI-anchored cell surface
Glypican 2	59	HS	Chromosome: 7 location: 7q22.1	GPI-anchored cell surface
Glypican 3	59	HS	Chromosome: X location: Xq26.1	GPI-anchored cell surface
Glypican 4	58	HS	Chromosome: X location: Xq26.1	GPI-anchored cell surface
Glypican 5	59	HS	Chromosome: 13 location: 13q32	GPI-anchored cell surface
Glypican 6	58	HS	Chromosome: 13 location: 13q32	GPI-anchored cell surface
Syndecans
Syndecan-1	33	HS, CS/DS	Chromosome: 2 location: 2p24.1	Transmembrane, extracellular
Syndecan-2	23	HS	Chromosome: 8 location: 8q22-23	Transmembrane, extracellular
Syndecan-3	43	HS, CS/DS	Chromosome: 1 location: 1pter-p22.3	Transmembrane, extracellular
Syndecan-4	22	HS	Chromosome: 20 location: 20q12	Transmembrane, extracellular
Lecticans
Aggrecan	208–220	CS/KS	Chromosome: 15 location: 15q26.1	Extracellular
Versican (0/1/2/3 isoforms)	373/265/180/72	CS	Chromosome: 5 location: 5q14.3	Extracellular
Neurocan	145	CS	Chromosome: 19 location: 19p12	Extracellular
Brevican	96	CS	Chromosome: 1 location: 1q31	Extracellular
SLRPs
Decorin	36	CS/DS	Chromosome: 12 location: 12q21.33	Extracellular
Biglycan	38	DS/CS	Chromosome: X location: Xq28	Extracellular
Fibromodulin	42	KS	Chromosome: 1 location: 1q32	Extracellular, intracellular
Lumican	38	KS	Chromosome: 12 location: 12q21.3-q22	Extracellular
Keratocan	37	KS	Chromosome: 12 location: 12q22	Extracellular
Mimecan	25	KS	Chromosome: 9 location: 9q22	Extracellular
Others
Thrombomodulin	58	CS	Chromosome: 20 location: 20p11.2	Transmembrane
CD44 (19 isoforms)	37–81	CS/DS	Chromosome: 11 location: 11p13	Transmembrane, extracellular, intracellular
NG2/CSPG4	251	CS	Chromosome: 15 location: 15q24.2	Transmembrane
Invariant chain	31	CS	Chromosome: 5 location: 5q32	Cell surface, intracellular
Neuroglycan-C	120–150	CS	Chromosome: 3 location: 3p21.3	Transmembrane
Type XVIII collagen	180–200	HS	Chromosome: 21 location: 21q22.3	Extracellular
Perlecan	400–450	HS	Chromosome: 2 location: 1p36.1-p34	Extracellular
Agrin	212	HS	Chromosome: 1 location: 1p36.33	Transmembrane, extracellular
Betaglycan	110	HS/CS	Chromosome: 1 location: 1p33-p32	Transmembrane
SV2	20	KS	Chromosome: 1 location: 1q21.2	Transmembrane
Serglycin	10–19	HS/CS	Chromosome: 10 location: 10q22.1	Intracellular
Endocan	50	DS	Chromosome: 5 location: 5q11.2	Circulating extracellular
Neuropilin-1	130	HS/CS	Chromosome: 10 location: 10p12	Transmembrane
Type IX collagen	270	CS	Chromosome: 6 location: 6q12-q14	Extracellular
Testican 1	48	HS/CS	Chromosome: 5 location: 5q31.2	Extracellular
Testican 2	45	HS/CS	Chromosome: 10 location: 10pter-q25.3	Extracellular

CS, chondroitin sulfate; DS, dermatan sulfate; GAG, glycosaminoglycan; GPI, glycosylphosphatidylinositol; HS, heparan sulfate; KS, keratan sulfate; SLRP, small leucine-rich family of proteoglycans.

A few words need to be said about hyaluronan, or hyaluronic acid (HA), as it is also known. This GAG has a deceptively simple composition, yet takes part in a number of essential processes in vertebrate development, repair, and disease. It is absent from invertebrates. However, hyaluronan is not synthesized as part of a proteoglycan (i.e., it is not attached to a core protein). It is synthesized at the cell surface, and three mammalian HA synthases are known (Wang A et al. 2011). It is therefore released to the extracellular environment without passage through the Golgi. Commensurate with this unique synthetic pathway, HA is not sulfated or otherwise modified at the time of synthesis. It can, however, be modified subsequently by covalent linkage to inter-α-inhibitor (IαI; Zhang et al. 2012) family members, a group of serine protease inhibitors. The cross-linkage is catalyzed by tumor necrosis factor (TNF)–stimulated gene 6 (TSG-6; Sangaard et al. 2008), and the complex is anti-inflammatory. Moreover, the HA-IαI heavy chain complex is a major component of the cumulus matrix and is essential for ovulation and fertilization in the mouse (Zhuo et al. 2001). Hyaluronan can also interact noncovalently with a number of matrix components. It may also be bound at the cell surface by receptors such as CD44. This may be important in enhancing signaling through cell surface growth factor receptors (Bourgignon et al. 2007; Kim Y et al. 2008; Misra et al. 2008). CD44 null mice have a mild phenotype, yet disease models show a role in limiting inflammatory responses (Wang Q et al. 2002; Ponta et al. 2003; Puré and Assoian 2009).

It can be appreciated that the complexity of GAG structures (e.g., heparan, chondroitin, and dermatan sulfate) can only be achieved by organized synthesis and modifications that are controlled by the cell. A large number of transferases and modifying enzymes together have the role of regulating the fine structural properties of the GAGs. The following reviews by Kreuger and Kjellén (2012), Multhaupt and Couchman (2012), and Malmström et al. (2012) highlight some of the key issues. Although GAG synthesis (with the exception of HA) takes place in the Golgi apparatus, and all of the enzymes involved are known from many genomes, including human, mouse, and genetic models such as zebrafish, Drosophila, and Caenorhabditis elegans, the regulation of synthesis remains, for the most part, unknown. Immunocytochemistry is playing a part in tracking where the synthases and transferases are localized in the Golgi, as part of unraveling this complex synthetic pathway (Multhaupt and Couchman 2012). Heparan sulfate is perhaps the most complex polysaccharide known to date, in terms of modifications by sulfation and epimerization, so it is a challenge to understand how cells regulate its synthesis.

Proteoglycans—Where They Are and How to Localize Them

Immunohistochemistry and immunocytochemistry have been important techniques in discovering the distributions of the proteoglycans. Many core protein antibodies are available, although some are not as specific and high affinity as would be desirable. Caution is needed with a number of commercial antibodies generally (Couchman 2009; Saper 2009). It should be pointed out that Western blotting is often part of antibody characterization, but for proteoglycans, this can be challenging because the GAGs may be a dominant physicochemical feature of the whole macromolecule. Some GAGs may be >20 kD in size, and when present in multiple copies on each core protein, the proteoglycan may appear very heterodisperse by SDS-PAGE. Moreover, purifying proteoglycans usually requires some specific methods (Iozzo 2001) that take advantage of GAG properties, such as anion exchange chromatography or density gradient centrifugation. One way to reduce the heterodisperse nature of the proteoglycan is to remove the GAGs by specific bacterial enzymes, such as chondroitinase ABC or heparitinase. These enzymes can remove most of the corresponding GAG chain, apart from a small carbohydrate “stub” (Fig. 2). The core proteins may then resolve as discrete polypeptides. In a similar way, immunohistochemistry for core proteins can also be more complex because of the GAGs. Fixatives, particularly those that precipitate, can cause the GAGs to collapse and mask the core protein. However, it is quite possible to use the same GAG-degrading enzymes to treat paraffin sections, for example, after the antigen retrieval steps. This will remove the GAGs and, in some cases, render the core proteins more accessible to specific antibodies. The GAG-degrading enzymes often require specific buffers that are suggested by the suppliers but are also in the literature (Iozzo 2001).

Figure 2.

Action of heparinases and chondroitinases on their respective glycosaminoglycans. These bacterial enzymes are eliminases and so, on cleavage of the polysaccharides, create an unsaturated uronic acid residue. One of these units will be present at the end of the carbohydrate “stub” remaining on each core protein. In addition, disaccharides or larger oligosaccharides will be released, each bearing a similar terminal unsaturated uronic acid. (A) There are three bacterial heparinases, I, II, and III, each of which has particular substrate specificity in terms of the heparan sulfate (HS) structure it will cleave. Heparinase III, for example, will cleave at unsulfated regions, and since the region proximal to the core protein is usually of low sulfation, this enzyme will pare back the polysaccharide close to the core protein. As a result of selectivity in heparan sulfate substrates that the three enzymes will cleave, oligosaccharides composed of varying numbers of disaccharide units will be released. To maximize heparan sulfate disaccharide cleavage products, it is common to use all three heparinases synchronously. For simplicity, the infrequent 3-O-sulfate modification of glucosamine is not shown. (B) Chondroitinase ABC will cleave chondroitin 4-sulfate (CS-A), dermatan sulfate (CS-B), and chondroitin 6-sulfate (CS-C), whereas chondroitinase ACII, for example, will not cleave dermatan sulfate. Chondroitin 4-sulfate is shown as a substrate in the model with symbols, but chondroitin may be unsulfated, 6-O sulfated, or a combination of these forms. However, in each case, a product with terminal unsaturated uronic acid will result from enzyme cleavage. R—H or SO₃; R′—COCH₃ or SO₃.

The GAGs themselves are not very antigenic, although a number of antibodies, some commercial, are available to them. A series of single-chain antibodies with preference for particular sulfation motifs in heparan sulfate have been reported, for example (van Kuppevelt et al. 1998; Thompson et al. 2009). Some antibodies are also available against chondroitin and dermatan sulfates, as well as keratan sulfate (Caterson 2012). Hyaluronan provides a special challenge, however, as it is difficult to prepare specific antibodies. A review on this area was published recently (de la Motte and Drazba 2011). The most commonly used technique for this GAG is to use a fragment of the aggrecan core protein, which binds HA with high affinity. A recent study of human skin shows this technique (Siiskonen et al. 2011). A small number of monoclonal antibodies have particular use for proteoglycans. The bacterial enzymes, such as chondroitinases and heparinases, are eliminases, rather than hydrolases. Therefore, they leave a terminal unsaturated uronic acid after digestion (Fig. 2). This in turn is quite antigenic, being a nonnative carbohydrate structure. Some antibodies against chondroitin and heparan sulfate have been characterized with this specific activity (Couchman et al. 1984; David et al. 1992). To be used in immunocytochemistry or immunohistochemistry, however, the cells or tissue sections must first be treated with the requisite bacterial enzyme. Moreover, it should be borne in mind that the antibody will not have specificity for the core protein, only the remaining carbohydrate “stub.” This can be useful, for example, in locating all heparan sulfate proteoglycans in a tissue or on a Western blot. A recent study of cutaneous disease provides an example of this technique (Kim JS and Werth 2011). Such antibodies can be used alongside more traditional, core protein-specific antibodies.

Proteoglycan Distributions

Proteoglycans were first identified as extracellular matrix components, and older histology textbooks referred to ground substance, an all-embracing term that included the location of these soluble macromolecules. As with collagens and glycoproteins, the spectrum of proteoglycans that occur in any particular type of extracellular matrix may be distinct. However, it seems that all extracellular matrices contain proteoglycans. Proteoglycans vary greatly in form and function. The small leucine-rich proteoglycan, decorin, is abundant in dense connective tissues where it associates with, and can regulate the formation of, interstitial collagen fibers (Rada et al. 1993; Reed and Iozzo 2002). Decorin bears a single dermatan or chondroitin sulfate chain. There is a family of related proteoglycans of the small leucine-rich type (Table 1). At the opposite extreme, aggrecan is most abundant in cartilage and can be substituted with around 100 chains (Aspberg 2012). Cartilage also contains type IX collagen that can possess a single chondroitin sulfate chain (Brewton et al. 1991). Basement membranes contain several proteoglycans, three of which usually are substituted with heparan sulfate. They are perlecan, type XVIII collagen, and agrin. In terms of sequence and structure, the core proteins of these proteoglycans are completely unrelated. Nevertheless, basement membrane heparan sulfate may be an important determinant of, for example, growth factor sequestration, protection from proteases, and presentation in the correct configuration to cognate receptors (Saksela et al. 1988; Friedl et al. 1997; Iozzo et al. 2009). Perlecan is not basement membrane restricted but occurs in several extracellular matrices, including cartilage.

On all but erythrocytes, one or more cell surface proteoglycans are present. Most are transmembrane, such as the HA receptor CD44 and the syndecans, but others, notably the six mammalian glypicans, are anchored by a glycosylphosphatidylinositol moiety to the outer membrane leaflet. It is common that glypican and syndecans coexist on cell surfaces, although each may be enriched in different compartments. It is possible that glypicans, by virtue of their anchor, localize to raft structures. In polarizing epithelia (Dick et al. 2012), it is known that glypicans can concentrate in the apical membrane domain (Gallet et al. 2008). Both the syndecans (Eriksson and Spillman 2012) and glypicans have heparan sulfate GAGs predominantly, although syndecans-1 and -3 may also have chondroitin or dermatan sulfate in addition. The reason that these two proteoglycans may be hybrids is not entirely clear. Some cell surface proteoglycans may not be glycanated all the time. CD44, for example, is often not glycanated on leukocytes but in epithelia may bear heparan sulfate chains (Brown et al. 1991). Frequently, in fibroblasts, it is substituted with a chondroitin or dermatan sulfate chain (Malmström et al. 2012). This receptor is known for its highly complex alternate splicing at the mRNA level, and it is a particular splice variant that carries the sequence that includes the heparan sulfate acceptor site. Other cell surface proteoglycans such as neuropilin-1, a receptor for vascular endothelial growth factor (VEGF) and semaphorins, and betaglycan, a type III transforming growth factor (TGF)–β receptor, may on some occasions possess a heparan sulfate chain but on other occasions a chondroitin sulfate chain. It is known that the properties of neuropilin-1 may change as a result of this alternate glycanation (Shintani et al. 2006). The transmembrane signaling proteoglycans have been reviewed in depth recently (Couchman 2010). Many link indirectly to the actin cytoskeleton and also to submembranous protein networks represented by PDZ domain proteins (Subbaiah et al. 2011).

It has been known for some time that mucosal and connective tissue mast cells contain granules with the proteoglycan serglycin (Rönnberg et al. 2012). This proteoglycan gets its name from the serine-glycine rich motif that dominates its sequence. The serine residues are sites for heparin or chondroitin sulfate substitution. Commercial heparin for a long time has been prepared from gut that is rich in mucosal mast cells. Serglycin is, however, not restricted to mast cells but is found more widely, including in vascular endothelial cells, where it may regulate secretion of the chemokine GROα/CXCL1 (Meen et al. 2011).

A number of proteoglycans, including syndecans and glypicans, have been reported from nuclear locations. For many years, the existence of nuclear proteoglycans was controversial, but reports are increasing, based mainly on immunocytochemistry. It has yet to be ascertained why this occurs and what function these highly anionic molecules may have in the nucleus (Richardson et al. 2001; Chen and Sanderson 2009). It is also not clear if whole proteoglycans are transported to nuclei or whether there is partial cleavage to liberate specific domains, presumably including some of the GAGs, which are then translocated. Certainly, many proteoglycans are sensitive to proteolytic cleavage. Syndecans in particular are highly susceptible to matrix metalloproteinases and other proteinases (Manon-Jensen et al. 2010), and this may be an important part of their physiology with relevance to tissue repair and disease. Glypicans can be released from the cell surface by the GPI-phospholipase Notum (Traister et al. 2008). In turn, this can lead to suppression of signaling through Wnts, for example.

Many proteoglycans are turned over and degraded through well-studied lysosomal pathways. Indeed, some important and devastating diseases in humans are known that arise from the lack of a specific enzyme in GAG hydrolysis. These rare genetic storage diseases are the subject of much current research with some new potential therapeutic approaches being considered (Muenzer 2011; Valayannopoulos and Wijburg 2011). In addition, HA has been described as a more general cytoplasmic component (Hascall et al. 2004). The pathways that lead to this localization of HA are incompletely understood, and more recent research suggests that this accumulation may be a consequence of disturbed glucose metabolism (Wang A and Hascall 2009; Wang A et al. 2011).

Proteoglycan Functions in Development, Tissue Architecture, and Disease

Knockouts of Genes Involved in Heparan Sulfate Synthesis and Observations on Stem Cells

A very large number of potent growth factors, morphogens, and cytokines have heparin-binding and heparan sulfate-binding properties. Therefore, it is not surprising that there is an extensive literature detailing many roles for heparan sulfate proteoglycans (HSPGs) in invertebrate and vertebrate development. As might be expected, there are some remarkably conserved roles for these proteoglycans. Moreover, genetic ablation of EXT1 or EXT2, which encode the components of a heterodimeric protein complex that is the major heparosan polymerase, leads to embryonic lethality. This is true of the mouse, and the equivalent experiments in Drosophila yield the same results, where, remarkably, human EXT1 can rescue flies where the equivalent gene, tout velu, is deleted (Dasgupta et al. 2007). Therefore, heparan sulfate is essential for multicellular life. A key property of heparan sulfate on the cell surface is that it can concentrate ligands for signaling through high-affinity receptors. A well-studied example is fibroblast growth factor (FGF; Rapraeger et al. 1991). Many members of this large family bind heparan sulfate, and crystallographic studies reveal how the FGF/HS complex interacts with the FGF receptor (FGFR) kinases (Pellegrini et al. 2000; Schlessinger et al. 2000). Heparan sulfate can also control morphogen gradients (Nahmad and Lander 2011) necessary for tissue differentiation events. Examples include wingless (Wnt in vertebrates) and hedgehog.

Much recent attention has been focused on embryonic stem cells and their potential for the treatment of a range of diseases. Stem cells are relatively rich in proteoglycans, and heparan sulfate has been shown to be essential for differentiation (Smith et al. 2011; Tamm et al. 2012). Embryonic stem cells derived from Ext1^–/– mice, for example, are incapable of differentiation but can be rescued by exogenous heparin. These experiments underline the essential nature of heparan sulfate for mammalian development, probably because essential growth and differentiation factors require GAG interactions at the cell surface. This area is reviewed by Tamm et al. (2012). However, although heparan sulfate may be essential, no single knockout of a syndecan or glypican is embryonic lethal in the mouse, suggesting that in the embryonic stem cell context, there is redundancy among the core proteins that bear heparan sulfate.

Specific deletions of transferases that are involved in heparan sulfate synthesis have varying results. Deletion of NDST1 in the mouse is embryonic lethal (Ringvall et al. 2000). This is perhaps unsurprising because N-sulfation is an early event in HS synthesis, so that a failure in this step leads to a global undersulfation of the GAG. However, deletion of the more selectively expressed NDST2 leads to defective mast cells, since heparin synthesis is compromised, but is not embryonic lethal (Forsberg et al. 1999). Deletions of the 5′ epimerase or 2-O-sulfotransferase have similar outcomes, most notably renal agenesis, which leads to perinatal lethality (Bullock et al. 1998; Li et al. 2003). A later step in the synthesis of HS is 6-O-sulfation. However, the binding of many growth factors depends on this modification. As a result, 6OST1 deletion is also embryonic lethal in the mouse and in Drosophila (Habuchi and Kimata 2010).

Proteoglycan Core Protein Genes—Knockouts and Genetic Diseases of Man

In contrast to ablation of enzymes involved in heparan sulfate synthesis, knockout of single syndecan core protein genes in the mouse has mild outcomes. Loss of syndecan-1, -3, or -4 all leads to phenotypes, but mice survive in every case and are fertile. For example, syndecan-1 and -4 mice have slowed tissue repair in injury models (Bernfield et al. 1999; Echtermeyer et al. 2001), whereas syndecan-3 null mice have neural defects that are quite mild (Hienola et al. 2006). Syndecan-2 null mice have not been reported. This leads to the obvious conclusion that there is redundancy among the syndecans. Despite this, syndecans have specific signaling functions that do not overlap. Syndecan-4, for example, binds inositol phospholipid and protein kinase Cα, which then becomes activated and signals downstream to the actin cytoskeleton (Couchman 2003, 2010). It seems likely that a common function of syndecans has yet to be elucidated that might explain the apparent redundancy that genetic experiments suggest. In invertebrates, there is a single syndecan gene, and evidence from C. elegans suggests that it has a role in migration of specific neurons through development (Rhiner et al. 2005). As yet, however, the biochemical basis for its action is not understood.

There are six mammalian glypicans, and a rare human disease, Simpson-Golabi-Behmel syndrome, is associated with loss of function in glypican-3. This causes overgrowth, both in human patients and in the mouse knockout (Filmus and Capurro 2008). Patients are also tumor prone and are particularly at risk of Wilms tumors. Research now suggests that glypican-3 has a role in inhibiting Hedgehog signaling through competition with its receptor, patched (Capurro et al. 2008). More recently, there has been much interest in the connection between high levels of glypican-3 expression and the development of hepatocellular carcinoma (Allegretta and Filmus 2011). This proteoglycan is now a target for cancer therapy. Human mutations in glypican-6 that are functionally null have been characterized in autosomal recessive omodysplasia, with pronounced limb shortening and craniofacial dysmorphism (Campos-Xavier et al. 2009). A genome-wide association study, on the other hand, shows that acquired nephrotic syndrome can be associated with specific single-nucleotide polymorphisms (SNPs) in the glypican-5 gene (Okamoto et al. 2011). In these cases, the suspicion is once again that there are defects in heparin-binding growth factor metabolism, for example, of the FGF family.

In contrast to syndecans, deletion of the perlecan gene, which encodes a major basement membrane HSPG, has catastrophic effects. In the mouse, around 40% mice die at around day 10 of embryonic development, with the remainder exhibiting profound defects, not in basement membranes as might have been predicted but in cartilage and cranial development. Perlecan is, therefore, essential for skeletal development. In a similar vein, patients with dyssegmental dysplasia of the Silverman-Handmaker type show gross skeletal malformations. They have functional null mutations in the perlecan gene. Less severe phenotypes are shown by patients with Schwartz-Jampel syndrome, where truncated forms of perlecan are expressed (Arikawa-Hirasawa et al. 2001, 2002). The disease is marked by myotonic myopathy and chondrodysplasia. Along with structural roles, it may be that the heparan sulfate chains of perlecan have a role in binding skeletal growth factors and morphogens such as FGF18 and Indian Hedgehog (Koziel et al. 2004; Chuang et al. 2010). The lecticans get their name from a C-terminal globular domain that has lectin-like properties (Aspberg 2012). A recent report from Stattin et al. (2010) reveals that mutations in this domain of the large chondroitin sulfate proteoglycan (CSPG) aggrecan, seen in dominant familial osteochondritis dissecans, can have structural effects on the cartilage, with loss of extracellular matrix from the joint surface.

Proteoglycans and Acquired Diseases

Given that proteoglycans occur in every tissue and on or in many varied cell types, it is not surprising that some receive attention through their potential roles in pathogenesis of acquired diseases. Moreover, GAGs can bind many potent mitogens that are thought to be relevant to proliferative diseases such as cancer. Many studies have been directed, for example, into the roles of syndecans in tumor progression. Syndecan-1 has been the focus of a majority of this work (Teng et al. 2012), and in some cases, such as breast cancer, its expression, particularly where it is present not only in epithelial tissue but also in stroma, is correlated with poor prognosis (Lendorf et al. 2011; Yoneda et al. 2012). Much work also strongly suggests that the synthesis of syndecan-1 in myeloma is congruent with disease progression (Khotskaya et al. 2009), to the extent that a current research focus is whether inhibition of syndecan-1 expression may lead to a novel therapeutic approach to this devastating disease.

Heparan sulfate-modifying enzymes are also under scrutiny with respect to tumor progression. In the human genome, there is a single heparanase that can cleave heparan sulfate chains into oligosaccharides that retain considerable biological activity (Fux et al. 2009). This can affect the distribution of heparin-binding ligands such as growth factors, with downstream influences on gene expression and motility. It is upregulated in a number of tumor types and is associated with poor prognosis in many cases (Arvatz et al. 2011; Sanderson and Iozzo 2012). In addition, there are two sulfatases that can edit heparan sulfate after it is synthesized, probably on the cell surface. These enzymes selectively remove sulfate from the 6-O-position on heparan sulfate but not chondroitin sulfate. Many potent polypeptides use 6-O-sulfates as part of their binding site(s) on heparan sulfate, so their removal can liberate the proteins and increase their bioavailability. In this regard, Sulf2, in particular, is thought to be important for the progression of non–small cell lung cancer, as well as pancreatic and hepatocellular carcinoma (Rosen and Lemjabbar-Alaoui 2010).

Heparanase has also recently been shown to be essential for development of diabetic nephropathy in mice (Gil et al. 2012). Diabetes is a fast-growing problem in the developed world, and proteoglycans have long been a focus of research into pathological changes to extracellular matrices in diabetic tissue. Organ failure in long-term diabetics is in part due to pathological changes in matrix composition, and those changes occurring in the diabetic kidney are reviewed elsewhere (Kolset et al. 2012). Originally, it was thought that basement membrane thickening and loss of heparan sulfate were key to filtration defects and proteinuria, but as Kolset et al. (2012) describe, the situation appears to be more complex and diabetic nephropathy may encompass multiple changes to glomerular and tubular proteoglycans.

Aggrecan is one of four members of the lectican family of CSPGs. It is abundant in cartilage, where is contributes to essential structural properties, such as the ability to resist compressive forces. In part, this is due to the high negative charge of the GAG chains that attract counter-ions and water. Therefore, the loss of aggrecan in arthritis is an important part of disease etiology. Metalloproteinases are the major suspects in aggrecan cleavage and loss, and recent work has focused on the ADAMTS group of proteinases (Majumdar et al. 2007; Stanton et al. 2011; Troeberg and Nagase 2012).

Versican, a large chondroitin proteoglycan of the lectican family, is a major vascular wall component that can also be the target of ADAMTS proteinases. This has implications for pathological angiogenesis (Fu et al. 2011). Versican, therefore, has structural roles. However, an emerging theme is that many such structural proteoglycans have, in addition, signaling roles, particularly with respect to the immune system. Versican has been identified as upregulated in many tumors, including, lung, gastric, ovarian, breast, and melanoma (Kim S et al. 2009; Ricciardelli et al. 2009; Du et al. 2010; Ween et al. 2011, Hernández et al. 2011). In 2009, it was identified as a macrophage activator, through cell surface TLR2:TLR6 complexes that in turn induced TNF-α secretion by myeloid cells, with downstream promotion of lung cancer metastasis (Kim S et al. 2009). Versican may act at other levels also. In common with other family members, versican core protein has N-terminal (G1) and C-terminal (G3) globular domains (discussed for aggrecan in Aspberg 2012). These globular domains may interact with many extracellular matrix components, and G1 interactions with HA are a defining character of the lectican class. However, it has been reported that the versican G3 domain may interact, directly or indirectly, with the epidermal growth factor receptor (Hernández et al. 2011). Such signaling may promote mammary tumor and melanoma progression, for example (Du et al. 2010; Hernández et al. 2011).

Whether the other two members of the lectican family, neurocan and brevican, have similar functions through their globular domains is not yet clear. However, these two hyaluronan-binding proteoglycans are believed to have important roles in the central nervous system. A recent genome-wide association study reported significant association of a SNP in the 3′ untranslated region of the neurocan gene with bipolar disorder (Cichon et al. 2011). Brevican is a component of the perineuronal net and is believed to have roles in synaptic plasticity, glioma invasion, postlesion plasticity, and, perhaps, Alzheimer disease (Frishknecht and Seidenbecher 2012).

Several members of the small leucine-rich family of proteoglycans (SLRPs) are known to have roles in regulating collagen fibrillogenesis. The decorin null mouse, for example, has abnormal interstitial collagen fibers in the dermis (Danielson et al. 1997). These small proteoglycans have the ability to bind to distinct regions on the collagen and through this regulate the assembly of fibrils into fibers. However, quite independent of this activity, it is also known that decorin can bind and inhibit TGF-β (Border et al. 1992). TGF-β has many roles, both in development but also in response to tissue injury. Of particular significance is that TGF-β has been implicated in the pathogenesis of fibrosis. Signaling from TGF-β receptors leads fairly directly to transcriptional regulation of, for example, type I collagen genes (Penttinen et al. 1988; Biernacka et al. 2011). Therefore, the decorin/TGF-β axis has received much interest from a clinical perspective. A recent review on decorin and liver fibrosis covers this aspect in depth (Baghy et al. 2012). Recent work, summarized in this volume (Nastase et al. 2012), points to yet another vital function for members of the SLRP family, notably biglycan and decorin, and that is regulation of the innate immune system (Moreth et al. 2010; Merline et al. 2011).

Hyaluronan is now also considered a regulator of immune function (Jiang et al. 2011), interacting with toll-like receptors, as can biglycan. This GAG is a widespread extracellular matrix component, is particularly abundant in developing tissues, and may contribute to a more open, hydrated, extracellular matrix network that allows cell migration. It is also an essential component of cartilage, so that HA is present in extracellular matrices of widely differing physical properties. In the mammalian eye, hyaluronan is a major component of the vitreous as well as being present in the corneal epithelium, conjunctiva, tears, and trabecular meshwork. HA-related products are now in use and under development for conditions such as vitreous replacement, dry eye, and retinal surgery (Rah 2011). In addition, since HA can promote cell migration, it is also now under investigation with respect to tumor progression (Afratis et al. 2012), and in some cases, hyaluronan-rich matrices may impede drug delivery or access to cell surface receptors, for example, ErbB2 (Jacobetz et al. 2012; Váradi et al. 2012). It is has also been shown that expression of the HAS2 HA synthase is related to that of TIMPs, tissue inhibitors of matrix metalloproteinases (Bernert et al. 2011). The discovery some years ago that HA could be a cytoplasmic component has triggered further work to understand its origins and functions (Wang A et al. 2011). It now seems that it may be associated with hyperglycemia and autophagy (Wang A and Hascall 2009). Connections to diabetes are therefore possible, and since HA consists of N-acetylglucosamine and glucuronic acid, it suggests that cells may respond to elevated glucose levels by elevating synthesis of this GAG, which requires little energy consumption (Tammi et al. 2011).

Conclusions

A select group of proteins can become proteoglycans through the addition of GAG chains. Although some of these can be assigned into families, such as the syndecans, glypicans, SLRPs, and lecticans, others are single examples (e.g., CD44 and perlecan). The array of proteoglycans is wider in vertebrates than in invertebrates, but nevertheless there are some remarkable examples of conserved function through evolution. These are most obvious for the syndecans and glypicans. The lecticans and SLRPs are vertebrate inventions and relate to the increased complexity and variety of extracellular matrices in vertebrates and their more sophisticated immune systems. Considered in their entirety, however, it is clear that proteoglycans perform important functions in many and varied aspects of development, tissue repair, and also some common and rare diseases in humans. Much remains to be learned regarding the regulation of GAG synthesis, physiological roles, and potential importance as targets for disease treatment. The reviews that follow address some of these key issues.