Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Fine Structure of Heparan Sulfate Proteoglycans

1.1 HSPG Biosynthesis

Heparan sulfate proteoglycan (HSPG) biosynthesis begins with translation of an HSPG core protein. There are three general configurations of HSPG core proteins: (1) a type 1 transmembrane (TM) protein, with a signal peptide and extracellular domain, a single-pass TM domain, and a cytoplasmic domain, such as members of the syndecan family; (2) a GPI-linked surface protein, with a signal peptide domain and an extracellular domain, such as members of the glypican family; or (3) extracellular proteins, such as agrin or perlecan. The HSPG core protein transitions toward the cell surface through the Golgi, where GAG chain synthesis is initiated at specific serine residues on the extracellular domain of the core protein. Initially, a four-sugar linker is covalently attached to a serine on the core protein, upon which alternating addition of N-acetylglucosamine and glucuronic acid residues creates an immature HS chain. Initial chain modification begins with conversion of select N-acetyl groups to N-sulfo groups by a family of N-deacetyl/N-sulfotransferases (NDST). Subsequently, C5-epimerase converts a subset of glucuronic acid residues to iduronic acid residues, and three families of O-sulfotransferases add sulfate groups to C2 of the uronic acids, to C6, and more rarely to C3 of the glucosamine residues to create mature HSPG chains (Esko and Selleck 2002).

Unlike the core proteins which are encoded by the genome, HS GAG chain initiation, polymerization, and modification appear to occur in a template-independent fashion in the Golgi. Currently, the mechanism by which mature HS chains are created is poorly understood. Open questions within the field include whether there are specific sub-compartments in the Golgi in which select modifications are made or whether there is an absolute requirement for some modifications to be made before others in vivo. A recent study suggests the biosynthetic pathway is more fluid than previously thought (Raman et al. 2011).

The collective actions of the entire HS chain biosynthetic process result in a mature HS chain with a heterogeneous mix of sulfated and non-sulfated residues (Lindahl et al. 1998; Esko and Lindahl 2001). The unique, sulfated domains created through the placement and spacing of sulfate and epimerization modifications create the HS “fine structure.” The specificity of any given HS chain to a particular ligand is thought to be dictated by the HS fine structure and defines the corresponding biological activities (Bernfield et al. 1992; Carey 1997; Rosenberg et al. 1997; Bernfield et al. 1999; Rapraeger 2001; Kramer and Yost 2003; Hacker et al. 2005). Fine structure detail, and ultimately the specificity of any particular HS chain, is defined during the last three steps in the HS chain maturation process, which constitutes the HS O-sulfation pathway.

1.2 HS O-Sulfation Pathway

The O-sulfation pathway represents the final three steps in the HS chain biosynthesis pathway—the addition of sulfates to the 2-O position of urinoic acid residues, catalyzed by the HS 2-O-sulfotransferase family (2-OST); addition of sulfates to the 6-O position, catalyzed by the 6-O-sulfotransferase (6-OST) family; and the 3-O position, catalyzed by the 3-O-sulfotransferase (3-OST) family, of glucosamine residues. A single 2-OST family member has been discovered in both invertebrate and vertebrate species (Cadwallader and Yost 2007), but recently a second 2-OST family member was discovered in zebrafish (Cadwalader et al. 2012). The number of 6-OST genes varies depending upon organismal complexity. Invertebrates, such as Caenorhabditis elegans and Drosophila, contain one 6-OST gene within their genome. In vertebrate species, the number of isoforms jumps to three in chicken, human, and mouse, with the addition of a splice-variant in the latter two species, and to four in zebrafish (Cadwallader and Yost 2006b). The existence of multiple 6-OST genes in the vertebrate species suggests the possibility of divergent regulation and function, as discussed below.

The 3-O-sulfation of glucosamine residues is the rarest HS chain modification, accounting for less than 0.5 % of the total sulfate modification (Colliec-Jouault et al. 1994; Shworak et al. 1994). However, a disproportionate number of 3-OST isozymes are needed to create this modification. The invertebrate model organisms contain one (C. elegans) or two (Drosophila) 3-OST family members. The number greatly expands in vertebrate species with seven family members in human and mouse and eight family members in zebrafish (Cadwallader and Yost 2006a). The rarity of the 3-O modification, as well as the extensive number of 3-OST family members, suggests that the 3-O modification acts as functional regulator of HS fine structure formation. A unique understanding of evolution, biochemistry, developmental biology, and gene regulation will be needed to understand the highly complex yet regulated HS fine structure.

1.3 Does Diversity Within the O-Sulfotransferase Families Contribute to the Glycocode?

There is a wealth of information focused on the biochemical importance of HSPGs and the O-sulfation pathway. Studies have looked at how these enzymes modify HS, how the modified HS interacts with various growth factors, and how these interactions are altered through changes in HS biosynthesis in cell culture. However, there is a lack of information about HS fine structure modifications with respect to development or disease. Most studies on HS in vivo function come from work in invertebrates, specifically C. elegans and Drosophila. While these models helped to define the roles of O-sulfation pathway members in an organism, the increased complexity of vertebrate organisms—the increase in growth factor isoforms and concordant increase in HS O-sulfation pathway members—provides a better model system for understanding both the complex functions and regulation of HSPGs in an in vivo system.

It is interesting to note that the final steps of the HS biosynthetic pathway represent the stages which contain the most isozymes. Why has there been an increase in the number of 6-OST and 3-OST isozymes within vertebrate species? One intriguing possibility is that the isozymes themselves have specificity. Each member of a given family has the same biochemical function. However, the glycocode hypothesis proposes that each family member selectively performs that biochemical function in a context-dependent mechanism. For example, every member of the 3-OST subgroup 2 family can transfer a sulfate to its substrate 3-carbon (of glucosamine residues) on a GAG chain, either in vitro or in vivo, but perhaps each member of the family has distinct preferences for 3-O-sulfation depending on the sequences of other GAG chain modifications on the substrate HS chain. These other modifications might be quite a distance from the primary modification site and would be challenging to assess biochemically.

Is the expansion of all growth factor families, i.e., FGF and Wnt family members, coordinated with the concordant increase in HS OST family members? Drosophila contains 3 FGF genes, while mice and humans contain 22 FGF genes (Bottcher and Niehrs 2005; Thisse and Thisse 2005). The actual number of zebrafish FGF genes has not been accurately determined, but recent evidence suggests approximately 27 (Itoh 2007; Itoh and Konishi 2007). The number of 6-OST and 3-OST isozymes in human is sevenfold greater than in Drosophila, comparable to the sevenfold increase in FGF family members from Drosophila to human. Has the HS field been led astray by studies focusing on the wrong FGF molecules? The study by Kaminura et al. which concluded a charge density role for HS used Drosophila as the model organism (Kamimura et al. 2006). The controversial studies looking at the FGF-FGFR-HS complexes have only examined members of the FGF1 superfamily, namely, FGF1 and FGF2 (Pellegrini et al. 2000; Plotnikov et al. 2000; Schlessinger et al. 2000). The FGF1 superfamily, FGF1 and FGF2, has the most widespread expression in both development and adult of the seven FGF superfamilies (Burgess and Maciag 1989; Baird and Klagsbrun 1991). Does this widespread expression mean a reduced stringency for specific HS fine structures? In contrast, FGF7 has a much more limited expression and correspondingly was shown to require a 3-OST modification (Ye et al. 2001).

The increased number of OST isoforms could act as a sorting mechanism, allowing for the “fine-tuning” necessary to direct growth factor signaling in vertebrate organism development. This rationale has led us to develop a model of HS fine structure function termed the “glycocode” hypothesis. The glycocode hypothesis suggests that the HS fine structure functions much like a zip code to regulate specific growth factor binding. Each growth factor and/or HS-binding ligand recognizes a specific sequence of modifications on HS chains in vivo. Some molecules can recognize more liberal HS fine structures, such as FGF1 and FGF2 not requiring 3-O-sulfation, whereas some ligands require more specific sequences, such as FGF7 and antithrombin requiring specific 3-O-sulfation. The amount of ligand binding and the location of signaling are controlled entirely by the HS fine structures present on a particular cell. The increase in 6-OST and 3-OST family members, along with diverse expression of each family, creates an advanced sorting mechanism to fine-tune and target growth factor signaling.

2 Roles of HSPG Sulfation in Development

While the HS field has primarily focused on the biochemical aspects of HS biosynthesis, recent studies have begun to examine the developmental and disease implications of HS fine structure modification. Early studies focused on the invertebrate species C. elegans and Drosophila, where the HS biosynthesis pathway is less complex. More recent studies in vertebrates, predominantly mice and zebrafish, have begun to examine how the HS fine structure affects specific organs and tissues.

2.1 HS-2-O-Sulfotransferase

The first step of the O-sulfation pathway, 2-O-sulfation of uronic acid residues, has been studied extensively in many species. Knockout of the 2-OST gene in C. elegans (hst-2) resulted in multiple phenotypic effects, including aberrant morphology, defects in egg laying, and neuronal defects (Bulow and Hobert 2004; Kinnunen et al. 2005). Surprisingly, knockout of HS2ST in Drosophila had only minor defects and showed only partial lethality when both maternal and zygotic HS2ST activity were eliminated (Kamimura et al. 2006). A second gene in Drosophila, pipe, shows very high homology to the vertebrate HS-2OST genes and was thought to encode a second 2-OST gene. However, work by Zhu et al. showed that while Pipe possesses sulfotransferase activity, the target substrate is not heparan sulfate (Zhu et al. 2005). Work in the chicken showed that HS2ST was expressed throughout the limb bud (Nogami et al. 2004). siRNA knockdown of HS2ST in the developing limb bud resulted in a truncation of the limb and reduced FGF expression throughout the limb (Kobayashi et al. 2007). HS 2-OST−/− mice die neonatally from bilateral renal agenesis (Bullock et al. 1998) and exhibit less penetrant eye and skeletal defects, as well as cleft palate and polydactyl (Bullock et al. 1998; McLaughlin et al. 2003). In contrast to these later-onset phenotypes, morpholino knockdown of maternally encoded HS 2-OST in zebrafish has early embryonic effects. 2-OST-deficient embryos have decreased cell adhesion and fail to initiate epiboly, one of the first steps in gastrulation. It is likely that 2-OST plays multiple roles, in multiple cell signaling pathways, to give this strong and pleiotropic phenotype in early development. Reduced cell adhesion in 2-OST is correlated with decreased b-catenin and E-cadherin at cell membranes, and can be rescued by downstream components of the canonical Wnt signaling intracellular pathway, but not by Wnt8 ligand, suggesting that 2-OST plays an integral role in Wnt signaling during early development (Cadwalader et al. 2012). Additional cell signaling pathways are currently being explored.

2.2 HS-6-O-Sulfotransferases

The 6-OST family, with multiple vertebrate family members, has provided insight into both the complexity and the spatial and temporal regulation of HS modification. Invertebrate species, with only one family member, show a range of phenotypes. Knockdown of 6-OST modification in C. elegans resulted in neural and cellular guidance defects (Bulow and Hobert 2004) and can suppress overexpression of the axon branching regulator kal-1(Bulow et al. 2002). In Drosophila, P-element excision of HS6ST resulted in lethality and defects in the tracheal system (Kamimura et al. 2001, 2006). In vertebrate species, in which multiple isozymes exist, targeted knockdown or knockout of individual isozymes showed distinct functions. In zebrafish, two independent phenotypes have been attributed to morpholino knockdown of 6-OST-2. 6-OST-2 is essential during somite specification and differentiation of muscle cells, but acts independently of the somite oscillatory mechanism (Bink et al. 2003). 6-OST-2 knockdown resulted in defects in caudal vein branching morphogenesis, while 6-OST-1a knockdown had no overt phenotype (Chen et al. 2005). To date, only one 6-OST family member has been knocked out in mice, 6-OST-1. Knockout of Hs6st-1 resulted in growth retardation and other developmental abnormalities, including skeletal and neural growth defects, with most embryos dying between E15.5 and perinatal stages (Pratt et al. 2006; Habuchi et al. 2007; Izvolsky et al. 2008).

2.3 HS-6-O-Endosulfatases

While most of the modifications of HS chains occur in the Golgi as the HSPG is transiting to the cell surface, it is important to note that cells have the ability to further regulate the complexity of the HS fine structure after the HSPGs have exited the Golgi. Two related HS-6-O-endosulfatases, commonly referred to as the Sulfs, have the unique ability to specifically remove glucosamine-6S residues from HS chains at the cell surface (Dhoot et al. 2001; Morimoto-Tomita et al. 2002). This radically changes the previous view that HS biosynthesis was complete prior to the HS chains moving to the cell surface. To date, the Sulfs have only been identified in vertebrate species. Cranial neural crest cell migration was altered following knockdown of XtSulf1 and XtSulf2 in Xenopus (Guiral et al. 2010). Knockout of Sulf1 or Sulf2 in the mouse had little effect on the mouse other than increased mortality (Holst et al. 2007; Lamanna et al. 2007; Lum et al. 2007). However, knockout of both Sulf1 and Sulf2 resulted in high rates of neonatal lethality with skeletal, renal, lung, and growth defects (Holst et al. 2007; Lamanna et al. 2007; Lum et al. 2007; Ratzka et al. 2008).

2.4 HS-3-O-Sulfotransferases

The importance of 3-O-sulfation in fine structure formation is not well understood. The 3-OST family can be divided into two subgroups based upon biochemical specificity and gene structure. It is thought that all bilaterians have at least one 3-OST from each subgroup (Cadwallader and Yost 2006a; Mizuguchi et al. 2009). One of the defining characteristics for subgroup 1 isozymes is the creation of the fine structure necessary for HS–antithrombin interaction. However, knockout of 3-OST-1 in the mouse failed to provide the supporting in vivo evidence for functions in antithrombin activity (Shworak et al. 2002; HajMohammadi et al. 2003). However, the 3-OST-1 knockout mouse did exhibit strain-specific lethality and intrauterine growth retardation (Shworak et al. 2002; HajMohammadi et al. 2003). Members of the 3-OST subgroup 2 family have been characterized for their ability to generate the HSV-1 binding site in vivo. RNAi-mediated disruption of HS3st-B, the Drosophila subgroup 2 homolog, results in a high rate of lethality along with neurological defects and a reduction in Notch signaling (Kamimura et al. 2004). Knockdown of HS3st-A, the subgroup 1 homolog, also exhibited a high rate of lethality (Kamimura et al. 2004). In zebrafish, knockdown of 3-OST-5, a subgroup 1 family member, or 3-OST-6, a subgroup 2 family member, causes left–right laterality defects. Strikingly, each gene functions through a separate mechanism that contributes to asymmetric fluid flow in Kupffer’s vesicle, the ciliated organ of asymmetry in zebrafish. 3-OST-5 is required in the FGF pathway, and knockdown diminishes cilia length via decrease in ciliogenic transcription factors FoxJ1a and Rfx2. In contrast, 3-OST-6 knockdown does not alter cilia length, but results in cilia immobility via diminished kinesin motor molecule (Kif3b) expression and loss of dynein arms within the normal-length cilia arm (Neugebauer et al. 2013). The observation that individual 3-OST isozymes that function within the same cells (ciliated cell lineage in Kupffer’s vesicle) have distinct functions in distinct cellular pathways suggests that each of these 3-OST family members creates distinct modified domains or “glycocodes” on cell surface proteoglycans. To date, this is some of the strongest functional evidence supporting the glycocode hypothesis.

3 Developmental Regulation of the Glycocode

Our working hypothesis is that a unique glycocode regulates specific components of numerous pathways, including the FGF, TGFβ, BMP, Wnt, and hedgehog cell signaling pathways. These codes are embedded in the fine structure or sequences of GAG modifications unique to each cell type. A crucial aspect of the glycocode model is that individual cells regulate the distinct GAG fine structures that they synthesize. While there is growing evidence that sulfation of HSPGs is important for development, one of the unmet questions in the field is whether specific fine structure sulfation patterns are generated and regulated in a cell-specific fashion. Analysis of HS from different mammalian tissues revealed the tissue-specific composition of samples, pointing to strict regulation of biosynthetic polymer modifications (Maccarana et al. 1996; Ledin et al. 2004). How this regulation occurs in vivo has been a topic of debate in the HS field. Does every cell in an organism express every OST family member? Are the OSTs regulated in a cellular and tissue specific manner? Are complexes of different OST family members selectively forming glycocodes during development?

3.1 Developmental Regulation of Distinct OST Family Members

While it was previously assumed that every cell expresses every OST gene, examining the developmental expression patterns of the various gene families from multiple organisms has refined our understanding of the mechanisms governing HS fine structure formation. It was previously thought that 2-OST expression would be ubiquitous given that only one family member exists in vertebrates (with a possible duplication in zebrafish). However, studies in mouse (Bullock et al. 1998), chicken (Nogami et al. 2004), and zebrafish (Cadwallader and Yost 2007) suggest that 2-OST expression is regulated in distinct tissues during development. RNA in situ studies looking at the transcripts encoding multiple isoforms of the 6-OST family in chicken (Nogami et al. 2004), Xenopus (Winterbottom and Pownall 2009), mouse (Habuchi et al. 2003; Sedita et al. 2004; Yabe et al. 2005), and zebrafish (Chen et al. 2005; Cadwallader and Yost 2006b) show each isozyme has specific spatial and temporal expression patterns. Not surprisingly, the Sulfs exhibit expression patterns coordinated with 6-OST expression in Xenopus (Winterbottom and Pownall 2009; Guiral et al. 2010) and mouse (Lum et al. 2007; Ratzka et al. 2008). Members of the 3-OST family also show a distinct localization patterns when examined in mouse (Yabe et al. 2005; Lawrence et al. 2007) and zebrafish (Cadwallader and Yost 2006a).

Summarizing these expression studies, most of the OST family members are ubiquitously expressed during early stages of embryogenesis. This ubiquitous expression makes sense if a large range of glycocodes are required for relatively pluripotent cells that are poised to receive multiple cell signals. In contrast, temporally and spatially dynamic subsets of OSTs and Sulfs are expressed during later development, suggesting the possibility that each isozyme plays a unique role in building distinct sequences within the glycocode during critical decision points in development. Thus, later in development, perhaps only a restricted set of glycocodes are available in a given cell type. This can be seen firsthand with respect to the 3-OST family and the neural system during development. Every member of the family is expressed in the neural system (Yabe et al. 2005; Cadwallader and Yost 2006a; Lawrence et al. 2007), but expression is restricted to specific brain areas or specific neurons (Hasegawa and Wang 2008). As development progresses, the glycocode is refined and restricted by the specific OST family members that are expressed in a given cell type during development. Future studies will need to address not only expression in a given organ or tissue, but which cells and cell types are being affected.

3.2 GAGosome

Another assumption in the HS field is that the HS polymerization functions independently and occurs prior to HS chain modification. This view may be propagated by the way in which the HS synthesis and modification pathway is presented as a sequential series of biochemical steps (Esko and Lindahl 2001). Recent evidence, including Ext2 regulation of NDST1 and Ext1 (Presto et al. 2008), NDST scaffold formation (Ledin et al. 2006; Raman et al. 2011), as well as HS isolated from various knockout mice (Merry et al. 2001; Li et al. 2003), has begun to question how mature HS chains are produced.

An interesting idea which challenges the stepwise HS chain modification pathway hypothesizes the HS chain modification enzymes expressed by a cell compose a large Golgi resident complex. This complex, most commonly termed the “GAGosome” (Esko and Selleck 2002), could completely synthesize and modify a given section of HS chain at one time. The review by Esko and Selleck provides evidence for the possibility of a complex. Recent findings have extended the evidence for the GAGosome. These findings include (a) the finding that NDST1 competes with Ext1 for binding to Ext2 (Presto et al. 2008), (b) the ability of members of the 3-OST subgroup1 family, which lack a transmembrane domain, to remain in the Golgi (Shworak et al. 1997; Cadwallader and Yost 2006a), (c) the compensation in sulfation when a member of the O-sulfation pathway is missing (Merry et al. 2001; Kamimura et al. 2006), and (d) growing evidence that multiple modifications are carried out in a coordinated and specific manner (Victor et al. 2009; Nguyen et al. 2011).

While no direct evidence yet exists for GAGosome formation, how might the GAGosome function within specific cells? We propose the possibility that distinct GAGosomes might contribute to distinct fine structural modifications of GAG chains within a cell. If each member of the 3-OST family has a slightly different specificity (Sect. 1.3), this might be played out in the context of the GAGosome. Furthermore, cells may contain more than one GAGosome type, which would enable cells to produce HS chains with distinct glycocodes, as suggested by distinct roles of different members of the 3-OST family in ciliated cells in zebrafish (Neugebauer et al. 2013).

3.3 Charge Density vs. Specificity

The exact nature of the interaction between HS and potential ligands has become a controversial topic within the HS community. The current dogma of the field is that the position and distribution of sulfates on HS chains are necessary for proper ligand binding (Esko and Lindahl 2001; Esko and Selleck 2002). The structural diversity, or specificity, of HS chains is a result of the strictly regulated but variable distribution of sulfate modified residues. This diversity is thought to enable selective interaction with proteins in a sequence and time-specific manner (Esko and Lindahl 2001). There are two major pieces of evidence suggesting that chain specificity is responsible for the HS–protein interactions. The first example is the antithrombin-binding sequence, which is responsible for the anticoagulant activity of heparin (Marcum et al. 1983; Bourin and Lindahl 1993; Princivalle et al. 2001). Each saccharide of the pentasaccharide structure (GlcNAc6S-GlcA-GlcNS3S-IdoA-GlcNS) is essential for high-affinity interaction with antithrombin. The second example often cited is the N-unsubstituted GlcN3S residue which mediates the specific binding of herpes simplex gD protein to cell surface HS during viral infection (Shukla et al. 1999).

Opposed to this is the idea that charge density, or the relative amount of negative sulfate charges in a given HS chain length, regulates how factors bind HS chains, and importantly, that the positions of the sulfates are much less critical (Kamimura et al. 2006; Kreuger et al. 2006; Gorsi and Stringer 2007). Therefore, it is thought that many HS–protein interactions may depend more on the overall charge of HS chains than on their specific fine structure. A study from Drosophila suggests that the overall charge density of a particular HS chain is more important than fine structure specificity (Kamimura et al. 2006). Mutations in Hs2st or Hs6st had little effect on FGF-mediated tracheal morphogenesis. Structural analysis of HS chains from these mutants revealed a loss of corresponding sulfation, but a compensatory increase in sulfation at other positions, which maintained the overall level of HS charge density. The Hs2st, Hs6st double mutant strongly disrupted tracheal morphogenesis. These findings suggest the overall sulfation level, and thus the overall charge level, maybe more important than strictly defined HS fine structures in some developmental contexts.

Several questions need to be addressed before this controversy can be resolved. Knockout of 2-OST in the mouse (Sect. 1.3) (Merry et al. 2001) resulted in major developmental defects, including renal agenesis (Bullock et al. 1998; McLaughlin et al. 2003), and an increase in 6-O-sulfation. Thus, a comparable level of charge density still gave a significant developmental phenotype. Knockout of 6-OST-1 in the mouse showed a minimal increase in 2-O-sulfation, but this result could be due to multiple isozymes being present. The phenotypes of both knockout mice suggest that fine structure specificity is important for the phenotypes in both mice, and maintaining overall charge density was not sufficient for normal development. Does the difference in phenotypes between Drosophila and the mouse knockouts suggest HS specificity differences exist between invertebrate and vertebrate species?

The examples highlighting HS specificity for ligands, namely, the antithrombin-binding site and the HSV-1 gD protein interaction, both contain a 3-O-sulfate modification. Does the 3-OST modification provide the HS fine structure with ligand binding specificity? As stated above, vertebrate species contain at least seven different isozymes to catalyze a modification which accounts for less than 0.5 % of the total HS sulfate modifications. Concomitantly, vertebrate species have seen an increase in growth factors proportional to the increase in 3-OST isozymes. Is the increase in 3-OST isozymes necessary to provide specificity to the increasing numbers of growth factors?

Our glycocode hypothesis proposes that inherent sulfation levels of HS, primarily due to the action of NDSTs, 2-OST, and 6-OSTs, create a low-affinity receptor and that actions of individual members of the large 3-OST family might confer a higher level of specificity. The generation of low-affinity receptors by NDSTs, 2-OST, and 6-OSTs would explain the differences seen in sulfation levels of various organs and tissues. The liver, which contains high inherent levels of HS sulfation, uses HS as a low Kd but abundant receptor for lipoproteins (MacArthur et al. 2007). Cells expressing low-sulfation HS, such as stem cell populations (Pickford et al. 2011), may be protected from hyperexcitation within their niche. High specificity, required by heparan sulfate-binding growth factors, is tightly regulated through the expression of specific isozymes and functions through the GAGosome. We speculate that in most cases, the specificity requires specific modifications, such as a 3-O-sulfation or post-synthesis modification provided by the Sulf enzymes. We think the glycocode hypothesis explains the differences in phenotypes seen in specific knockdown of the individual HS O-sulfation pathway enzymes.

4 Future Directions

The HS biosynthetic pathway is capable of producing 48 unique disaccharide units, yet only 23 units have been characterized in nature (Toyoda et al. 2000; Esko and Selleck 2002). Invertebrate and vertebrate species use the same 23 disaccharide units to make mature HS chains (Toyoda et al. 2000). The only difference between the various species is how these units are put together. Our glycocode hypothesis suggests that the specific placement of sulfates in the fine structure controls the specificity of interaction between HS and individual ligands. Cells have the ability to dynamically regulate expression of each enzyme and perhaps the ability to coordinate how the individual enzymes are packaged into different GAGosomes. While the overall level of sulfation serves as a low Kd binding platform, high specificity requires precise sequence specificity.

Much of the work previously done to identify and characterize specific members of the O-sulfation pathway has relied upon in vitro techniques and cell culture. These experiments and their conclusions have made significant contributions to the dogma in the field. Only recently has the field begun to understand the function of HS and specifically the O-sulfation pathway in vivo. Several examples are available where the in vitro and in vivo data simply do not match. Understanding the physiological roles of unique modifications of GAG chains on HSPGs will hopefully lead to a better understanding of complex choices made by cells in the context of their extracellular matrix.

Understanding the roles of the expanded number of 6-OST and 3-OST family members in vertebrates remains a priority for the field. The increased numbers of isozymes suggest that individual family members are highly regulated or essential for specific developmental processes. How does their in vivo expression and function relate with dynamic developmental choices made by cell lineages? Promising new technologies, including computer modeling (Spencer et al. 2010), synthetic organelle systems (Martin et al. 2009), and GAG array technologies (Wakao et al. 2008; Puvirajesinghe et al. 2012), show great potential moving forward. These technologies, along with others, will attempt to define the sequence specificity of numerous heparan sulfate-binding proteins. The key, however, will be in understanding how data derived from in silico applications reflects the fast-paced, dynamic interactions seen in vivo during development.