Interactions of signaling proteins, growth factors and other proteins with heparan sulfate: Mechanisms and mysteries

Abstract

Heparan sulfate (HS) is a component of cell surface and matrix-associated proteoglycans (HSPGs) that collectively, play crucial roles in many physiologic processes including cell differentiation, organ morphogenesis and cancer. A key function of HS is to bind and interact with signaling proteins, growth factors, plasma proteins, immune-modulators and other factors. In so doing, the HS chains and HSPGs are able to regulate protein distribution, bio-availability and action on target cells and can also serve as cell surface co-receptors, facilitating ligand-receptor interactions. These proteins contain an HS/heparin-binding domain (HBD) that mediates their association and contacts with HS. HBDs are highly diverse in sequence and predicted structure, contain clusters of basic amino acids (Lys, Arg) and possess an overall net positive charge, most often within a consensus Cardin-Weintraub (CW) motif. Interestingly, other domains and residues are now known to influence protein-HS interactions, as well as interactions with other glycosaminoglycans, such as chondroitin sulfate. In this review we provide a description and analysis of HBDs in proteins including amphiregulin, fibroblast growth factor family members, heparanase, sclerostin and hedgehog protein family members. We discuss HBD structural and functional features and important roles carried out by other protein domains, and also provide novel conformational insights into the diversity of CW motifs present in Sonic, Indian and Desert hedgehogs. Finally, we review progress in understanding the pathogenesis of a rare pediatric skeletal disorder, Hereditary Multiple Exostoses (HME), characterized by HS deficiency and cartilage tumor formation. Advances in understanding protein-HS interactions will have broad implications for basic biology and translational medicine as well as for the development of HS-based therapeutics.

Introduction

The heparan sulfate proteoglycans (HSPGs) constitute a large and important family of cell surface and extracellular matrix (ECM)-associated macromolecules. The HSPGs display distinct patterns of expression and regulate a variety of physiologic roles including cell differentiation, cell-cell interactions, tissue morphogenesis and organ function; when dys-regulated, they can also have roles in pathologies such as cancer or skeletal dysplasias (reviewed in ^1,2,3,4). Each HSPG consists of a core protein to which one or more HS chains are covalently attached via hydroxyl groups on serine residues. The HS chains are polymerized sequentially, and the process initiates with the initial attachment of xylose to the serine residue followed by 2 galactose residues and glucuronic acid to form the linkage region. Polymerization continues with the sequential addition of glucuronic acid (GlcA) and N-acetyl- glucosamine (GlcNAc) to produce repetitive disaccharide units producing chains with an average size of 20–25 kDa (^2,3). Individual saccharides along the HS chains are modified via epimerases and also by specific sulfotransferases. The latter catalyze the sulfation of carbohydrate carbons at position 2, 3 or 6 around the sugar rings, eliciting exceedingly complex sulfation patterns referred to as the “sulfation code” that have major biological significance (²). In toto, there are over 25 enzymes involved in HS chain polymerization and modification (reviewed by ^{2, 3}). Additional complexity and subtleties are produced by selective removal of some of the sulfate groups by Sulf1 and Sulf2 sulfatases, eliciting segments with low/minimal sulfation along the HS chain intercalated by high sulfation segments (⁵). In addition, HS chains can be selectively removed from the cell surface or the ECM by the action of heparanase, the single entity in the human genome with the ability to do so (⁶). The family of mammalian HSPGs includes 4 syndecans whose core proteins span the cell surface bilayer and 6 glypicans whose core proteins are bound to the cell surface via a GPI anchor (Table 1) (^2,3). It includes also a number of extracellular HSPGs such as perlecan, betaglycan and collagen XVIII and serglycan which resides in the secretory granules of mast cells (^{7, 8}; see Table 1). As indicated above, HSPGs display selective patterns of expression in different tissues and organs and at different stages of development, adding to their functional complexity but also introducing significant specificity to their biological action and function (³).

Table 1.

Heparan Sulfate Proteoglycans Identified in Mammalian Cells^*

HSPG**	Location	Reference***
Agrin	Transmembrane	57
Betaglycan	Transmembrane	58
Syndecans (4)	Transmembrane	9,51,59
Neuropilin-1	Transmembrane	60
Glypicans (6)	Membrane, GPI anchored	61
Serglycin	Intracellular	62
Collagen XVIII	Extracellular	57,63
Perlecan	Extracellular	64
Testican (2)	Extracellular	65

^*Adapted from ³.

^**Numbers in parenthesis- number of known gene family members.

^***Published Reviews

Because of their sulfation, the HS chains bear multiple anionic charges. One of the key functions of HSPGs stemming from this unique feature is their ability to interact with numerous proteins bearing a reciprocally charged HS-binding domain (HBD). The HS-binding proteins include plasma proteins, extracellular matrix components, cell surface proteins, and members of the major growth factor and signaling protein families including Wnt, hedgehog, bone morphogenetic protein, fibroblast growth factor and vascular endothelial growth factor families (^9,10,11). The protein-HS interactions are very important and serve multiple functions including: modulating protein function and distribution; limiting protein range of action on target cells; stabilizing proteins and protect them from degradation; setting up morphogen protein gradients during development and growth; and presenting specific proteins to their cognate receptors for activation of signaling (^{2, 3}). The nature and general traits of the HBD have been studied in many proteins and some overall features have become apparent (^12,13). In general the HBDs contain basic residues (Arg and/or Lys), have an average pI > 11, and contain hydropathic amino acids spacing the basic residues apparently important for accommodating GAG chains in the binding pocket, but exhibit variable length and diverse amino acid sequence (Table 2). In general, the HBDs bind HS and heparin with high affinity when measured in standard biochemical assays, but can also bind chondroitin sulfate and hyaluronic acid, usually with lower affinity.

Table 2.

Primary Sequences of HS/Heparin Binding Protein Domains

Name*	Accession**	Position	Sequence	Reference***
AAMP	AAA68889	14–25	RRLRRMESESES	66
AREG	AAI46968	126–144	RKKKGGKNGKNRRNRKKKN	16
Antistatin	AAB29420	93–120	PNGLKRDKLGCEYCECRPKRKLIPRLS	14
AT III	BAC21176	64–71	RRVWELSK	12
AT III	XP_005245255	105–118	FFFAKLNCRLYRKA	67
AT III	AAA51794	156–177	AKLNCRLYRKANKSSKLVSANR	13
AT III	AAB40025	319–332	KPEKSLAKVEKELT	68
Apo B100	CAA28420	3168–3182	LSVKAQYKKNKHRHSI	69
Apo B100	1211338A	3352–3371	YKLEGTTRLTRKRGLKLATA	69
Apo E	1NFN_A	144–151	LRKRLLRD	70
Apo E	AAB59518	229–236	GERLRARM	70
Dhh	O43323	32–38	RRRYARK (by homology)
FGF 1	CAA41788	34–54	GLKKNGSCKRGPRTHYGQKAI	71
FGF 2	AAK52309	84–101	LKRTGQYKLGSKTGPGQK	13
FGF 2	NP_001997	261–280	KRTGQYKLGSKTGPGQKAIL	24,22
FGF 4	NP_001998	181–200	LSKNGKTKKGNRVSPTMKVT	72
FGF 7	C46289	71–91	LNQKGIPVRGKKTKKEQKTAH	24
FGF 8	NP_001193318	78–98	FTRKGRPRKGSKTRQHQREVH	24
FGF 10	AAM46926	142–162	LNGKGAPRRGQKTRRKNTSAH	24
FGF 18	NP_003853	153–173	FTKKGRPRKGPKTRENQQDVH	24
FBN	CAC86916	31–66	R31-G66	73
FBN	AAI17177	1847–1865	YEKPGSPPREVVPRPRPGV	74
FBN	AAI17177	1887–1901	KNNQKSEPLIGRKKT	75
Glia Nexin	NP_033281	82–105	RYNVNGVGKVLKKINKAIVSKKNK	13
HGF	ACX45438	67–99	A67-F99	76
HB-EGF	NP_001936	93–113	KRKKKGKGLGKKRDPCLRKYK	77
Hep Cofac 2	NP_032249	181–198	FRKLTHRLFRRNFGYTLR	12
Heparinase	NP_001159970	158–171	KKFKNSTYSRSSVDC	17
IFN-γ	NP_000610	147–166	AKTGKRKRSQMLFRGRRASQ	30
IGFBP-3	CAA46087	242–259	KKGFYKKKQCRPSKGRKR	78
IGFBP4	NP_034647	202–216	RNGNFHPKQCHPALDQ	78
IGFBP-5	NP_000590	221–238	RKGFYKRKQCKPSRGRKR	79
IGFBP-6	NP_002169	192–209	HRGFYRKRQCRSSQGQRR	78
IL10	CAG46790	116–138	LKTLRLRLRRCHRFLPCENKSKA (putative)	80
Ihh	Q14623	37–43	RRRPPRK (by homology)
Laminin	XP_006240058	662–681	RYVVLPRPVCFEKGMNYTVR	81
Laminin	EAW83423	247–263	RIQNLLKITNLRIKFVK	82
Laminin	EDL38338	3030–3051	KQNCLSSRASFRGCVRNLRLSR	82
LPL	AAC61679	163–181	RKNRCNNLGYEINKVRAKR	83
NCAM	AAH29119	150–167	IWKHKGRDVILKKDVRFI	84
PDGF-A	EAW87161	198–215	GRPRESGKKRKRKRLKPT	85
PRG4	XP_009438019	94–107	RSPKPPNKKKTKKV	86
Prot C Inhib	AAB60386	283–302	SEKTLRKWLKMFKKRQLELY	87
Sclerostin	NP_079513	70–172	F70-R170	31
Serglycin	NP_035287	26–40	YPARRARYQWVRCKP	15
Shh	NP_000184	32–38	KRRHPKK	39,40
EC-SOD	AAA66000	223–240	REHSERKKRRRESECKAA	88
Sulf1	NP_001121678	417–735	K417-K735	47
TFPI	NP_006278	240–272	G240-F272	89
TGF-β1	AAQ18642	14–32	DFRKDLGWKWIHEPKGYHA	90
TSP	AAA61178	41–50	RKGSGRRLVK	91
TSP	AAA61178	95–101	RQMKKTR	91
VEGFA	NP_001165100	137–185	A137-R185	26
VTN	AAH05046	366–380	AKKQRFRHRNRKGYR	92
vWF	CCQ25771	1328–1350	YIGLKDRKRPSELRRIASQVKYA	93
XO	XP_003262893	781–795	LGVPANRIVVRVKRM	94
XDH	AAA75287	1106–1122	KKKNPSGSWEDWVTAAY	13

^*Abbreviations: AAMP, angio-associated, migratory cell protein, AREG, amphiregulin, ATIII, Antithrombin III, FBN, fibronectin, HGF, hepatocyte growth factor, Hep Cof, Heparin cofactor II, LPL, Lipoprotein Lipase, TFP1, thrombospondin, VTN, vitronectin.

^**Genbank accession number.

^***Reference

Nature and primary structure of the HBDs

In pioneering work, Cardin and Weintraub set out to identify the domain(s) of proteins responsible for interaction with HS and other glycosaminoglycans and focused on four proteins: apolipoprotein B, apolipoprotein E, vitronectin, and platelet factor-4 (see ¹²). This comparative analysis allowed them to identify two binding motifs -XBBXBX and XBBBXXBX- in which B represents a basic residue and X represents any other residue. Analyses of additional proteins have since confirmed these findings and identified analogous motifs frequently containing clusters of basic amino acids -XBX, XBBX and XBBBX)- again separated by non-charged residues (¹⁴). The sequences first identified by Cardin and Weintraub have entered the vernacular and are referred to as the consensus Cardin-Weintraub (CW) motif. Indeed, Verrecchio and coworkers utilized consensus sequence templates to design peptides with high affinities for heparin and endothelial cell proteoglycans, and found that peptides with the highest affinity were tandem repeats of the sequence XBBBXXBX ¹⁵.

Amphiregulin (AREG) is an HS-binding growth factor that associates with and activates the epidermal growth factor (EGF) receptor tyrosine kinase. The mitogenic activity of AREG on cultured cells is blocked by addition of exogenous heparin and prior treatment of the cells with heparitinase (¹⁶). A peptide comprising residues 26–44 of fully processed AREG (Table 2), and comprising the putative HBD, blocks AREG binding to immobilized heparin, indicating that much of the HS-binding activity of AREG resides in that domain ¹⁶. Using solid phase binding assays, we have indeed found that the peptide does bind to heparin and HS with high affinity, but binds also to CS or HA albeit with lower affinity (Fig. 1). We carried out similar binding assays with human heparanase fragments. Not surprisingly, this enzyme also contains a HBD targeting it to its natural substrate. The HBD in human heparanase spans amino acids Lys158-Asp171 and can by itself block binding of full length enzyme to heparin (¹⁷). We synthesized the Lys158-Asp171 peptide and tested its binding capacities using solid phase assays with different GAGs as above. As one would expect, we did observe very high affinity binding to heparin and HS, but there was appreciable binding to CS and hyaluronic acid (HA) (not shown). These and similar experiments indicate that the HBD has a primary role in protein-HS interaction but other regions of the proteins are likely to be involved in order to enhance specificity of interaction and action.

Figure 1.

Differential binding of AREG peptide to different GAGs. The indicated type of GAG was immobilized on 96 well plates and the binding of Flag- tagged AREG (DYKDDDDKGG RKKKGGKNGKNRRNRKKKN; AREG sequence underlined) peptide was determined using an anti-Flag Ab and secondary antibody-HPR conjugate. Results included are from a representative experiment and were highly reproducible.

Some of the most detailed analyses of protein interactions with HS/heparin have been carried out with members of the FGF growth factor family. The family comprises 22 proteins with essential functions in cell growth, morphogenesis, tissue repair and angiogenesis (¹⁸). The FGFs interact with cognate cell surface tyrosine kinase receptors FGFR1, FGFR2, FGFR3 and FGFR4 and have an obligatory requirement for HS to exert their biological activity (Reviewed in^18,19, ²⁰). The most efficient signaling structure for FGF2 was recently shown to consist of a symmetrical complex [FGF2]₂-[HS block]₂-[FGFR]₂ (²¹). Interestingly, a comparison of the HBDs in FGF-1, FGF-2 and FGF-7 with those in other FGFs shows that the HBDs are quite diverse in their primary sequences (²²). Heparin/HS binding was found to be mediated in part by a “glycine box” with the consensus sequence XBXGXXBBG in which the location of the basic residues varies from FGF to FGF (^23,24). Using a library of HS octasaccharides with defined 2-O and 6-O sulfation patterns, Ashikari-Hada et al. were able to define and classify the interaction of FGFs and other growth factors with specific patterns along the HS chains. It was determined that growth factor-HS interactions can be categorized into five groups: Group 1 includes FGFs with affinity for 2-O-sulfated octasaccharides (ex., FGF-2); Group 2 factors have affinity for 6-O-octasaccharides (ex., FGF-10); Group 3 factors have affinity for both 2-O-sulfated and 6-O-sulfated octasaccharides but prefer the 2-O-sulfated ones (FGF-18 and HGF); Group 4 requires both 2-O-sulfate and 6-O-sulfate within the octasaccharides for binding (FGF-4, FGF-7); and Group 5 includes FGFs and other proteins with weak binding to any octasaccharide tested (FGF-8, BMP6 and VEGF) (²⁴). These results indicate that FGF binding is at least partially mediated by the glycine box sequence and differences in the primary sequence of this domain and sulfation patterns can both influence protein-GAG recognition in different FGFs and other proteins.

Secondary structure analyses of HBDs

In addition to amino acid sequence analysis, nuclear magnetic resonance (NMR) has been utilized to gain further structural insights into protein-HS interactions using approaches such as short and defined peptides. These studies have also yielded information about amino acid residues involved in binding that are distant from the primary HBD as the following examples demonstrate. Vascular endothelial growth factor (VEGF) is an important regulator of angiogenesis and its association and action on its cognate high affinity receptors involve HS and HSPGs (²⁵). The HBD of VEGF comprises residues 116–165 at the C-terminus whose solution structure has been solved by NMR. The structure consists of a disordered N-terminal region followed by 2 disulfide-bonded subdomains containing two short stranded antiparallel β sheets followed by an α-helical region at its C-terminus. Most of the basic residues mediating HS/heparin binding turned out to reside on the N and C-terminal domains and are thus brought together along one face (²⁶). Interestingly, cell surface-associated Syndecan 1 is frequently up-regulated in multiple myeloma²⁷. The HS chains on Syndecan 1 are believed to play an important role in VEGF signaling and are actually implicated in the pathogenesis of multiple myeloma (²⁸).

Mammalian heparanase (an endo-β-glucuronidase) is the only enzyme encoded in the genome that specifically cleaves HS at intra-chain sites. Its ability to recognize -and act upon HS and heparin- appears to reside in a specific HBD at position Lys158–Asp171 (31). NMR analysis of Lys158–Asp171 peptide mixed with a synthetic oligosaccharide mimicking heparin showed chemical shifts of a sub-domain from Lys158 to Asn162, suggesting that this sub-domain is critical for HS/heparin binding.

Similar experimental approaches were used to study IFNγ and its interactions with HS. This factor is an important immunomodulator and as such, needs to be strictly controlled within the extracellular milieu (²⁹). Its interaction with HS/heparin involves a HBD residing at the C-terminus and spanning amino acids Arg 124 to Gln 143. NMR has revealed that two subdomains within this peptide, D1 (Lys 125- Arg 131) and D2 (Arg 137- Arg 140) exhibit the most pronounced chemical shifts following HS/heparin binding (³⁰).

Sclerostin is a secreted cysteine-knot HS-binding protein that negatively regulates Wnt signaling and has critical roles in the regulation of bone formation and homeostasis (³¹). The interaction of Sclerostin with HS/heparin involves 3 loops within a structured HBD core region spanning residues 52-147: loop 1 spans Arg-57 to Val-80; loop 2 spans Gly-86 to Arg-109; and loop 3 spans Ile-111 to Ser-140. The loops are stabilized and cross-linked by a network of disulfide bonds, and are arranged together to form a linear and positively charged area covering one entire side of the protein. Following HS/heparin binding, chemical shifts were observed in amide nitrogens within loops 2 and 3 (³¹).

The above examples highlight the fact that domains and subdomains involved in HS/heparin binding in a given protein are structured to produce a congruent and often linear surface displaying all the negatively charged sugars, likely augmenting binding effectiveness, specificity and strength.

More complex interactions

Recent studies on certain signaling factors have shown that interactions with HS are more complex both structurally and functionally than previously realized and can actually involve concurrent interactions of other protein domains with other GAG types. A prominent example of this phenomenon can be found in members of the hedgehog family that include Sonic (Shh), Indian (Ihh) and Desert (Dhh) hedgehogs (^{11, 32,33, 34}). The Hh proteins are very potent signaling factors and have critical functions in embryonic development as well as tissue function and homeostasis and skeletal and non-skeletal pathologies including cancer (^{4,35, 36, 37}). Interactions with HS and HSPGs have been shown to regulate the distribution and action of Hh proteins on target cells and tissues and also their ability to form morphogen gradients within/amongst tissues during embryonic development (³⁸). The Shh signaling pathway in particular is frequently up-regulated in pancreatic cancer and interestingly, the HS/heparin binding activity of Shh is required for its action on the proliferation and metastatic spread of pancreatic ductal adenocarcinoma cells (PDAC) (see ³⁹).

The HBD resides within the amino-terminal and biologically active portion of Hh proteins and contains a CW motifs with the consensus BBBXXBB (Table 3; Reviewed in ^{11, 35, 40}). Site- specific mutagenesis has shown that the Lys residues at position 37 and 38 within the CW are important for Shh function, including its ability to induce osteoblast differentiation in C3H/10T1/2 cells (³⁹). Somewhat unexpectedly, these studies have also shown that a Lys residue at position 178 and thus far away from the CW is equally important for Shh function. A recent important crystallographic study on murine Shh combining site-directed mutagenesis, biological analysis and protein-GAG and protein-protein interactions has confirmed that the above 3 residues are involved in HS/heparin binding and function, but has also uncovered new aspects of Hh biology (⁴⁰). The study has revealed that additional basic amino acids at position K88, R124, R154 and R156 (corresponding to K87, R123, R153, R155 in human SHH) influence Shh interactions with HS, and this region was termed the “Hh core GAG-binding site” (⁴⁰). The CW at the N-terminus and the core site were shown to both be involved in Shh interactions with HS, but also mediated interactions of Shh with CS. Because the Hh proteins oligomerize to diffuse, to form morphogen gradients and to interact with cells, the authors examined these aspects of Hh biology and found that HS favors oligomerization much more than CS. Equally interesting was the finding that the GAG binding site partially overlaps with the domain of Shh interactions with other protein partners including cell surface receptor Patched and interference hedgehog proteins, suggesting unique mechanisms of Hh signaling modulation by HS, CS and other interacting proteins.

Table 3.

Sequence of CW motif and flanking amino acids in Sonic, Indian and Desert Hedgehogs.

Hh*	Organism	Sequence**, ***	Accession****
SHH	Human	GPGRGFGKRRHPKKLTPLAY	Q15465
SHH	Mouse	GPGRGFGKRRHPKKLTPLAY	NP_033196
SHH	Sheep	GPGRGFGKRRNPKKLTPLAY	XP_004008418
SHH	Aardvark	GPGRGFGKRRHPKKLTPLAY	XP_007951664
SHH	Python	GPGRGFGKRRHPKKLTPLAY	XP_007433256
SHH	Chicken	GPGRGIGKRRHPKKLTPLAY	AAA72428
IHH	Human	GPGRVVGSRRRPPRKLVPLAY	Q14623
IHH	Mouse	GPGRVVGSRRRPPRKLVPLAY	AAH46984
IHH	Aardvark	GPGRVVGSRRRPPRKLVPLAY	XP_007957594
IHH	Python	GPGRVVGSRRRPPRKLIPLAY	XP_007419884
IHH	Chicken	GPGRVVGSRRRPPRKLIPLAY	NP_990288
DHH	Human	GPGRGPVGRRRYARKQLVPLL	NP_066382
DHH	Mouse	GPGRGPVGRRRYVRKQLVPLL	EDL04156
DHH	Aardvark	GPGRGPVGRRRYVRKQLVPLL	XP_007936046
DHH	Dolphin	GPGRGPVGRRRYVRKQLVPLL	XP_004326854
DHH	Python	GPGRGPVGRKQSSRKSLAPLQ	XP_007437346

^*Hedgehog protein

^**Sequence of CW motif (Bold type) and flanking amino acids; sequence variations underlined.

^***CW consensus sequence BBBXXBB.

^****GenBank accession number.

We have uncovered additional interesting aspects of Hh biology by examining further and comparing the sequences of the CW motifs of Shh, Ihh and Dhh and surrounding sequences. As summarized in Table 3, uniformly spaced Gly residues are present on the N-terminal side of the CW in all three Hh proteins. It is well established that glycines provide conformational flexibility to polypeptide chains (⁴¹) and likely increase plasticity that may enable the CWs to adjust and optimize their interactions with GAG chains. Intriguingly, a highly conserved and invariant proline is present in the 5th position of Shh’s CW domain, while two prolines occupy positions 4 and 5 in Ihh’s CW and none are present in Dhh’s CW. Proline residues induce bends or kinks in polypeptide chains (⁴²) and presumably, are expected to have a major impact on overall protein conformation. Thus, we used a secondary structure prediction program in the PSIPRED Protein Sequence Analysis Workbench (Link: bioinf.cs.ucl.ac.uk/psipred) and scanned the PDB using the I-TASSER server (^43,44,45) to predict and analyze the CW conformation in Shh, Ihh and Dhh. The data obtained point to substantial conformational differences (Figure 2). Specifically, the CW of Shh is predicted to have a random coil-like conformation with a central kink generated by its single proline, while the CW of Ihh would have a similar configuration but with a more pronounced and complex kink generated by its 2 prolines. Because Dhh’s CW lacks prolines, its configuration is predicted to be a continuous and uniform helix as one would expect (Figure 2). Though these predictions need to be verified by more stringent methods, they raise the interesting possibility that the CWs of the 3 Hh proteins may be intrinsically different, could differentially affect the interactions of each Hh member with HS chains, different sites with distinct sulfation patterns and/or different HSPGs, and could in turn influence protein distribution, metabolism, overall conformation and perhaps even function (see ³⁴). In view of the identification of the “Hh core GAG-binding site” (⁴⁰), we also wonder whether the different CWs could have repercussions on that sites function as well in each Hh member.

Figure 2.

Structure of CW motifs in Shh, Ihh and Dhh. Secondary structure predications were carried out using the I-TASSER server (⁴⁵) and resulting structures were visualized using PyMol. The peptides are oriented with the N-terminus on the left; side chains of basic residues within the CW motifs are in blue while the side groups of proline are in red. Note that CWs of Shh and Ihh have a largely random coil configuration with central kinks due to the proline residue(s) while the CW of Dhh exhibits a more helical configuration. The sequences examined are presented below each structure; the CW motif is in bold type and flanking residues are in plain typeset.

Related and very interesting insights have been gained in a recent study that analyzed the secreted sulfatase Sulf1. As pointed out above, this enzyme is involved in remodeling the 6O-sulfation state of cell surface HSPGs and must thus possess the ability to recognize and act selectively upon its natural substrate (^5,46). However, full-length Sulf 1 was found to bind not only to HS/heparin, but also to CS and dermatan sulfate (DS) in affinity chromatography assays. Further structural and biochemical analyses using different Sulf1 deletion mutants showed that the interactions of Sulf1 with HS encompass a very large protein region called the hydrophilic domain (HD), spanning 319 amino acids Lys417-Lys735 (⁴⁷). This large region includes a consensus HBD and when assayed in solution and solid phase binding assays, was found to bind HS with high affinity and specificity but not to CS, DS or 6-O-desulfated HS (⁴⁸). HS/heparin binding was found to be mediated by 2 sites located in the inner and C-terminal regions of HD (⁴⁸). The authors concluded that the substrate specificity of Sulf1 is mediated by the HD, involves at least two separate HS-binding sites, and clearly depends on presence of 6-O sulfation, that is its own substrate.

HS- and HSPG-associated pathologies

Given the important roles that HS and HSPGs play in normal cell and tissue physiology, it is not surprising that genetic and metabolic defects in these macromolecules are linked to a number of pathological changes and disorders, in addition to those pointed out above. For instance, loss-of-function mutations in Glypican-3 cause the human X-linked disorder Simpson-Golabi-Behmel syndrome that is characterized by embryonic and postnatal overgrowth, cardiac malformations and predisposition to certain tumors (^49,50). Abnormally high shedding of syndecan-2 from the cell surface has been linked to progression and malignancy of various human tumors (^51,52). For sometime now, we have been studying the pediatric skeletal disorder Hereditary Multiple Exostoses (HME). This disease is characterized by benign cartilaginous outgrowths –termed exostoses- that form next to, but never within, the growth plates in developing and growing skeletal elements including long bones, ribs, vertebrae and pelvis (^4,53). The majority of HME cases are linked to heterozygous loss-of-function mutations in EXT1 or EXT2 and as a result, the patients have a systemic deficiency in HS levels amounting to about half of that seen in healthy people (⁵⁴). It has remained unclear and controversial as to whether and how the HS deficiency causes exostosis formation, why the exostoses form near but not inside the growth plates, whether the exostoses are formed by aberrant function of growth plate chondrocytes themselves or involve perichondrial progenitors, and what growth factors trigger and sustain exostosis outgrowth (^4,53). In studies we and others carried out previously, we found that mutant mice lacking one allele of Ext1 or Ext2 did not reproduce the human HME skeletal phenotype, while double heterozygous Ext1^+/−; Ext2^+/− mice or conditionally ablated Ext1^−/− mice did (^{55, 56}). The data strongly indicated that a partial decrease in HS levels such as that elicited by a simple heterozygous EXT mutation is insufficient for exostosis formation and HME progression, but a deeper decrease is required. The mechanisms for the latter are still unclear, but several possibilities have been suggested including loss of heterozygosity or a second hit to an unrelated gene (⁴). In order to identify growth factors involved in exostosis formation, we monitored BMP signaling -a key chondrogenic pathway- in Ext1 conditionally-ablated mice (⁴). Specifically, we ablated both alleles of Ext1 along the perichondrium of growing long bones in mice and monitored BMP signaling and exostosis formation over time. We found that BMP signaling levels revealed by phosphorylated SMAD immunohistochemistry were very low in perichondrium in control mice, but were significantly higher in the mutant mice. This was followed by initiation and growth of exostoses at later time points, indicating that ectopic BMP signaling precedes exostosis formation and may be a major driver in their outgrowth (⁴). In good agreement, studies by others have shown that the distribution and signaling of Ihh also become abnormally wide and off-target within the growth plate of Ext1 hypomorphic mice that produce about 10–20% of normal HS levels, leading to abnormal growth plate organization, chondrocyte function and skeletal growth retardation (³⁸). Together, the above studies have indicated that a major role of HS and HSPGs in the growth plate is in fact to regulate growth factor distribution and restrict and delimit signaling action of growth factors on target cells. Thus, a wider and uncontrolled distribution of growth factors, resulting from severe decrease of Ext gene expression and HS levels, would derange normal growth plate and/or perichondrial function and cause pathologies including exostosis formation.

Discussion and Prospective

The ability of many signaling proteins, growth factors and other proteins to interact with HS and HSPGs has long been known, but the present review and analysis of current literature show that while there is been considerable progress, significant gaps in our understanding remain. There is little doubt that because of its primary structure and concentration of positively-charged amino acid residues, the CW motif plays a very important role in establishing and favoring HS-protein interactions and may even initiate them. However, the specific sequence, organization and predicted configuration of the HBDs vary considerably from protein to protein (see Table 2), and the significance and consequences of this variability remain incompletely understood. The diversity of these domains does suggest however, that they may have evolved to have multiple and subtle functions, to selectively interact with defined regions present in HS and possibly other GAGs, and to contribute to define and regulate the roles and activities of the interacting proteins. Since different HSPGs are expressed in different tissues and at different developmental stages and because the sulfation patterns along their HS chains differ significantly as well (^2,3,4), the diversity of HBDs could thus introduce elements of specificity and selectivity of HS-protein interactions and permit differential retention, accumulation and/or activity of given signaling and growth factor proteins on cells. This feature could be particularly important to insure that distinct factors, present within the same tissue and organ, would be able to coordinately interact with their specific targets. We provide above clear example of these paradigms regarding Ihh, a powerful signaling protein that becomes abnormally and broadly distributed and active at ectopic sites when HS and/or HSPGs are deficient or deficiently expressed (^4,38).

As we describe above, it has become clear that in addition to CW motifs, other protein domains and regions are involved in regulating protein-GAG interactions and in particular those with HS and CS. The recent models proposed for Sulf1 and Shh are particularly attractive and revealing. In the case of Sulf1, portions of the protein at the N- and C-termini would provide surfaces for interactions with CS-rich proteoglycans within the matrix, serving as a support and guiding system to allow the more centrally-located catalytic hydrophilic domain (HD) to act on the 6-O-sulfate substrate groups on neighboring HS chains and carry out sequential desulfation (⁴⁸). In the case of Shh, the CW motif at the N-terminus would have a major role in binding to HS and signaling function, but the centrally-located Hh core GAG-binding site would be equally important for interactions with HS and CS and overall hedgehog protein distribution, structural features and function (^11,40). As summarized above, interactions of Shh with HS would favor oligomerization and thus function and signaling, whereas interactions with CS would provide a lower degree of oligomerization, possibly promoting storage of Shh for subsequent function. Our own analysis of the CW in Shh indicates that its configuration is quite different from the CW of Ihh and Dhh, suggesting that proteins as powerful as hedgehogs may have acquired additional structural features during evolution, introducing additional elements of regulation. It is thus clear that the interactions of proteins with GAG chains and their respective proteoglycans introduce several levels of regulation and as pointed out in recent reviews, can in fact fine-tune cell and tissue physiology and developmental processes and create pathologies when abnormal (^2,3,4). All the above findings, insights and speculations are attractive, intriguing and interesting, but much remains to be learned about their basis, true relevance and roles and implications. Advances and new insights in these complex research fields hopefully in the near future will have broad repercussions and significance for basic biology and translational medicine as well as for the conception and creation of HS-based therapeutics.

Abbreviations

AREG

Amphiregulin

CS

chondroitin sulfate

CW

Cardin-Weintraub motif

ECM

extracellular matrix

FGF

fibroblast growth factor

G-box

glycine box

GAG

glycosaminoglycan

GlcA

glucuronic acid

GlcNAc

N-Acetyl-D-glucosamine

HA

hyaluronic acid

HBD

heparin/heparan binding domain

HME

Hereditary Multiple Exostoses

HS

heparan sulfate

HSPGs

heparan sulfate proteoglycans