Introduction

Heparan sulfate (HS) is a linear, sulfated polysaccharide of the Glycosaminoglycan (GAG) family, which is ubiquitously present on cell surface and extracellular matrix (ECM) in most animal tissues [1–3]. Over the years, HS has been shown to participate to major cellular processes, through its ability to bind – and in some instance to modulate – a multitude of structural and signaling proteins. HS consists of repeating disaccharide units, composed of N-acetylglucosamine (GlcNAc) and glucuronic acid (GlcA) residues (Fig. 1). In specialized regions termed NS-domains, the structure of this disaccharide motif is further diversified by a series of enzymatically-catalyzed modifications: the amino sugar is N-sulfated (GlcNS), most GlcA have been epimerized into iduronic acid (IdoA) and saccharide units are variably substituted with O-sulfate groups, at the C-6 (and occasionally C-3) of glucosamine and at C-2 of uronic acid residues. This confers to HS a unique molecular organization, in which homogeneous, non or low sulfated regions (NAc domains) alternate with the hypervariable and highly sulfated NS-domains.

Fig. 1
figure 1

Structure of heparan sulfate. HS consists of repeating GlcA-GlcNAc disaccharide units. In regions termed NS domains (red boxes), this motif is extensively modified by a series of modifications including epimerization and sulfation reactions. The overall structure of the polysaccharide chain thus exhibits an alternation of NS-domains and non-modified regions (NAc-domains, black line), separated by intermediate NAc/NS transition zones (dashed boxes)

Full size image

Binding to HS mainly involves electrostatic interactions between positively charged residues on proteins and sulfate and carboxyle groups of the polysaccharide. HS biological properties are thus largely dictated by the structure and sulfation pattern of NS-domains, which comprise the binding sites for most HS ligands [4–6]. These structural features are determined by a complex and highly regulated biosynthesis pathway. HS biosynthesis is a multi-step process involving: (i) the formation of a tetrasaccharide linker that will serve as attachment site for HS chains on proteoglycans, (ii) the synthesis of a uniform polysaccharide precursor composed of GlcA-GlcNAc repeats, and (iii) the maturation of this precursor chain through the above mentioned series of structural modifications (Fig. 2a). The first of these reactions is the N-deacetylation/N-sulfation of glucosamine catalyzed by enzymes of the NDST family (4 isoforms). This is followed by the epimerization of GlcA into IdoA residues (by the C5 epimerase) and 2-O-sulfation of IdoA by the 2-O-sulfotransferase (2OST). Finally, sulfate groups at C6 and C3 of the glucosamine are added by the 6-O-sulfotransferase (6OST, 3 isoforms) and 3-O-sulfotransferase (3OST, 7 isoforms) families, respectively. HS structural features are thus finely tuned by the concerted action of these enzyme families and the differential expression of their multiple isoforms [7–10]. In addition, further regulation of HS 6-O-sulfation occurs post-synthetically, through the action of extracellular sulfatases of the Sulf family [11–13].

Fig. 2
figure 2

Biosynthesis of heparan sulfate. a Modification steps during HS biosynthesis. Elongating, uniform heparan chains successively undergo N-deacetylation/N-sulfation of glucosamine by the NDSTs, epimerization of GlcA into IdoA by the C5 epimerase, 2-O-sulfation of IdoA by the 2OST, 6-O-sulfation by the 6OSTs and finally 3-O-sulfation by the 3OSTs. b Substrate preferences of 6OSTs

Full size image

During the recent years, considerable progress has been made in the understanding of the tight relationships of HS structure and function and many studies have highlighted the major contribution of 6-O-sulfates to HS structure and activity. The aim of this review is thus to provide an overview and discuss the mechanisms regulating HS 6-O-sulfation pattern, as well as their consequences on the polysaccharide biological functions.

The 6-O-sulfation of HS: structural features and biological relevance

It is well established that HS/protein interactions involve saccharide motifs with defined sulfation patterns. The degree of structural specificity of these binding sites varies from one ligand to another, but can be very high for some proteins, depending on the nature of the sulfates involved. For instance, rare 3-O-sulfates have been associated with highly selective interaction, while abundant N-, 2-O-sulfates are part of most protein binding sites [5, 9]. Finally, 6-O-sulfation may occupy an intermediate status, with clear indications of distinctive features.

These differential contributions of sulfates to HS structural and functional diversity could be deduced from the mechanisms leading to NS-domain assembly during HS biosynthesis. As the first modification reaction, N-sulfation has been shown to be critical for determining NS-domain size and distribution, and 2-O-sulfation influences the Glc/IdoA epimerization ratio. These early modification steps may thus provide common structural and functional properties to the polysaccharide. In agreement with this, recent studies have demonstrated that NDSTs, C5-epimerase and 2OST show processive activities [14, 15], and earlier work on the biochemical characterization of HS reported that NS-domains were composed of relatively homogeneous blocks of contiguous GlcNS-IdoA,2S disaccharides [16, 17]. In contrast, the addition of 6-O- (and 3-O-) sulfates is much more heterogeneous. These late modifications may thus provide further diversity, and therefore potential specificity/selectivity, to the polysaccharide. In addition, it is worth noting that 6-O-sulfation is the only type of sulfation to be significantly found outside NS-domains, with almost half of the 6-sulfates present in the NAc/NS flanking regions. It is also the only modification step to be regulated through both biosynthetic and post-synthetic processes (see below). Finally, a wealth of studies has reported that expression and structure of HS are highly dynamic and vary dramatically amongst cell types and physiopathological conditions. This is particularly significant for HS 6-O-sulfation pattern, which has been shown to undergo significant changes during both embryonic development [18, 19] and aging [20, 21], as well as in pathologies, such as cancer [22, 23], amyloidosis [24, 25], chronic renal fibrosis [26], inflammation [27, 28] and diabetes [29, 30].

Functionally, HS 6-O-sulfation has been associated with major biological functions of the polysaccharide (Table 1). One of the first examples is that of the antithrombin III (AT-III) pentasaccharide binding motif, which characterization provided the molecular basis for the anticoagulant properties of heparin [31]. Critical structural elements of this highly specific saccharide motif are the well-known internal 3-O-sulfated glucosamine, but also a 6-O-sulfate located on the non-reducing end residue [32]. However, the selective contribution of 6-O-sulfates in HS activities has been most clearly exemplified in the FGF/HS interaction model [33–35]. FGFs represent a family of 23 heparin-binding growth factors. Amongst these, FGF-1 and FGF-2 are the archetypal members, which provided the first evidence that HS could serve as a growth factor co-receptor, which was necessary for promoting their activity [36, 37]. Binding sites to FGF-1 and FGF-2 share very similar structural features, both in term of size (5–6 sugar units) and saccharide composition, with a requirement for GlcNS and IdoA2S residues [38–40]. However, while 6-O-sulfates are essential for binding to FGF1 and enable optimal contact between sugar and protein [41–43], they are not involved in the interaction with FGF-2 [44, 45]. In contrast, induction of FGF-2 activity requires longer oligosaccharides (10–12 sugar units) and the presence of 6-O-sulfates [16, 46, 47]. These findings provided the first evidence of possible uncoupling between HS interactive properties and biological functions, and highlighted 6-O-sulfation as the critical determinant for discriminating these two activities. The rationale behind such mechanism is that saccharide extension and 6-O-sulfates provide an additional binding site for FGF receptor (FGFR) that induces formation of a FGF/FGFR/HS ternary complex able to trigger cell signaling. Interestingly, it has been recently reported that heparin oligosaccharides featuring a few (1–2) 6-O-sulfates located on the reducing end glucosamine residues exhibited full FGF-2 promoting activity [48]. The presence and also the specific positioning of 6-O-sulfates within saccharide sequences may therefore be critical for activation of FGF-2.

Table 1 Implication of 6-O-sulfates in HS/protein interactions
Full size table

Aside the well-documented examples of FGF-1 and FGF-2, HS 6-O-sulfation has been involved (to various extent) in the binding and/or activation of other FGFs, such as FGF-4, −7, −9, −10 and −18 [42, 43, 47, 49, 50], as well as other growth factors, including Platelet Derived Growth Factor (PDGF) [51], Hepatocyte Growth Factor (HGF) [52], le Glial cell-line Derived Neurotrophic Factor (GDNF) [53, 54], Heparin-binding Epidermal Growth Factor (HB-EGF) [55] and Vascular Endothelial Growth Factor (VEGF) [56–58].

HS 6-O-sulfation also participates in the interaction with other signaling proteins, such as chemokines and morphogens. Chemokines are small proteins implicated in many biological processes such as development, inflammation and immunosurveillance (ref. reviews, see [59, 60]). All chemokines bind to HS, which regulate their activity in different ways. HS sequesters chemokines and protects them from enzymatic cleavage, it increases their local concentration in the ECM to form chemokine gradients guiding migrating cells, and it induces their oligomerization to facilitate their interaction with receptors [61–65]. Contribution of 6-O-sulfates has been demonstrated for CXCL12 [66–69], CXCL8 [70, 71] and CXCL4 (Platelet factor 4) [72, 73]. Interestingly, a recent study showed that addition of a single 6-O-sulfate group at the non-reducing end of a chemically synthesized [IdoA,2S, GlCNS]6 heparin oligosaccharide switched its inhibitory properties from CXCL8 toward CXCL12, thereby highlighting the importance of 6-O-sulfate positioning for binding to these chemokines [74]. Finally, the crystal structure of CCL5 in complex with Heparin disaccharides revealed electrostatic interactions between the 6-O-sulfate of the sugar and the K45 residue of the chemokine [75].

Morphogens are signaling proteins that dictate cell fate and tissue development during embryogenesis. HS binds to members of the 3 major families of mammalian morphogens: Wnt/β-catenin, hedgehog (HH) and transforming growth factor beta (TGF-β)/bone morphogenic protein (BMP). As for growth factors and chemokines, interaction with HS regulates morphogen distribution and contributes to the formation of gradients [76, 77]. The 6-O-sulfation of HS has been reported to modulate activity of both TGFβ and Wnt in two opposite ways. Interaction with HS promotes TGFβ1 activity and 6-O-sulfate groups are important structural determinant for the binding [78, 79]. This is in agreement with a recent study showing that a decrease in HS 6-O-sulfation reduced the response of primary fibroblasts to TGFβ1 [80, 81]. In contrast, 6-O-sulfates may act as a negative regulator of Wnt signaling. A proposed mechanism is that Wnt binds with high affinity to 6-O-sulfated HS, which prevents access to its cell-surface receptor Frizzled (Fz). In support to this, it has been shown that enzymatic removal of 6-O-sulfates by the Sulfs (see below) reduced Wnt/HS binding affinity, thereby enabling interaction with Fz and induction of cell response [82]. However, it has also been reported in other studies that 6-O-desulfation could also have an inhibitory effect on Wnt, by facilitating its release and degradation [83]. Regulation of Wnt signaling may thus be dictated by a complex interplay between HS 6-O-sulfation status and Fz bioavailability.

6-O-sulfates have also been found to participate to the binding of HS to L-selectin [27, 69], endostatin [84] and axon guidance protein slit-2 [85]. It has also been involved in cell-surface attachment of hepatitis E virus [86] and in the promotion of neuregulin-1/erbB receptor interaction [87]. Finally, a number of studies have shown evidence of an implication in Alzheimer, as 6-O-sulfates take part in the binding to β-amyloid peptides [88] and in the modulation of amyloid precursor protein (APP) processing [89].

Determination of HS 6-O-sulfation status during biosynthesis

The 6-O-sulfation of HS is directed by the 6-O-sulfotransferase family (HS 6OSTs), which catalyze transfer of a sulfate group from adenosine 3′-phospho adenosine 5′-phosphosulphate (PAPS) donor to position 6 of glucosamine residues. HS6OST (termed thereafter 6OST1) was first cloned and characterized from Chinese-hamster ovary in 1995 [90]. On the basis of sequence homology, HS6OST2 and HS6OST3 isoforms were subsequently identified [91], as well as an alternatively spliced form of HS6ST2, HS6ST2-S, featuring a 40 amino acid deletion [92]. Orthologs are found in Xenopus, C. elegans and Drosophila (one isoform), while two isoforms have been described in chicken, four in zebrafish and three in human and mouse. In human, 6OST isoforms are encoded by three distinct genes located on chromosoms 2, X and 13, respectively [92]. Amino acid sequences of 6-OSTs are less conserved than other glucosaminyl-sulfotransferases. The 6OST1 displays 51 % and 57 % of sequence identity with 6OST2 and 6OST3 respectively, while 6OST2 and 6OST3 share 50 % of similarity [91]. However, the sequence located in the central region at the level of potential PAPS binding sites is highly conserved between the three isoforms [93].

6OSTs are type II transmembrane proteins that reside in the Golgi apparatus. They feature a N-terminal stem region, which is essential for controlling protein trafficking and localization, and for maintaining the enzyme in an active state [94]. Contrary to most other HS biosynthesis enzymes, 6OSTs can also be found in the extracellular environment [90]. The biological relevance and mechanisms underlying enzyme secretion still remain unclear, but for 6OST3, it has been shown to involve the cleavage of the short hydrophobic cytoplasmic domain by β-secretase [95]. Interestingly, inhibition of β-secretase resulted in 6OST3 accumulation in the Golgi apparatus and increased HS 6-O-sulfation. The balance of 6OST intracellular/extracellular distribution in active or inactive forms could therefore contribute to the tuning of HS 6-O-sulfation.

During HS biosynthesis, most steps of polysaccharide assembly and maturation are controlled by enzyme families. Such occurrence of multiple isoforms with specific activities, substrate preferences and tissue distribution is believed to be fundamental for determining the polysaccharide fine structure. Rather surprisingly, studies of 6OSTs did not reveal major differences in isoform substrate specificities. All isoforms can indistinctly add sulfate groups on both GlcNAc and GlcNS, regardless of the nature of the adjacent uronic acid (IdoA or GlcA) [96], and overexpression of any of the 6OSTs in HEK cells affected similarly HS structure [97]. However, studies using libraries of structurally-defined oligosaccharides revealed some subtle substrate preferences (Fig. 2b). 6OST1 preferentially acts on IdoA-GlcNS motifs and generate 6-O-sulfated sequences with lower 2-O-sulfate content. 6OST2 is more active towards 2-O-sulfate containing substrates [98]. Finally, 6OST3 displays an intermediate substrate specificity, between 6OST1 and 6OST2.

In contrast, 6OST isoforms can be distinguished by specific spatio-temporal distribution in tissues [91, 92, 99]. In adult mice, 6OST1 transcripts are mostly present in liver, with moderate expression detected in brain, heart kidney and lung. 6OST2 is found in the brain and spleen, while its alternatively spliced variant 6OST2-S is more present in the ovary, placenta and fetal kidney. Finally, 6OST3 transcripts are ubiquitously expressed. 6OSTs are also differentially expressed during organogenesis in a stage-dependent manner, suggesting critical roles during development. 6OST1 is found in tissues of epithelial and neuronal origins, while 6OST2, is localized in mesenchymal tissues. In contrast, 6OST3 expression is much more restricted and is limited to the later stages of development. Noteworthy, expression of HS-binding morphogens with distinct 6-O-sulfation requirements varies extensively during development. Differential expression of 6OST isoforms may therefore provide HS structures with defined binding properties, to finely regulate these morphogens’ activities and induce specific development processes.

Altogether, these data clearly indicate that 6OSTs tune HS 6-O-sulfation through overlapping yet complementary activities. Regulatory mechanisms involved remain far from being fully apprehended, but interesting data came from studies on animals, in which 6OSTs have been downregulated or knocked out. In this regard, Drosophila melanogaster and Caenorhabditis elegans provided valuable models, as these organisms present only one isoform of 6OST. In Drosophila, where structural properties of HS are similar to those present in vertebrates, knockdown of 6OST leads to high lethality and defects in the migration of the tracheal cells, in which 6OST is specifically expressed [100]. Interestingly, these phenotypes are similar to those observed for mutants of FGF signaling pathways, suggesting that these may due to impaired FGF signaling. In C. elegans, the unique 6OST form is only expressed in neuronal tissues. Consistently, 6OST knockout leads to defects in the ventral cord interneurons [101].

In mouse, the knockout of 6OST1 leads to high level of lethality during late embryonic stages. Surviving mice are fertile but show growth retards and abnormal morphological phenotypes, such as impaired ossification, reduction of the body weight, defects in placental vascularization, impaired lung morphology and erroneous axon navigation at the optic chiasm [102, 103]. In contrast, the knockout of 6OST2 did not cause significant phenotype abnormalities, whereas 6OST1/6OST2 double KO mice died at earlier stage than 6OST1 KO mice [47]. Analysis of HS composition from various organs of KO mice showed a reduction in 6-O-sulfation content. For 6OST1 KO mice, 6-O-sulfation reduction was moderate, with more pronounced effects in tissues naturally expressing high levels of this isoform. Transcriptional analysis showed no increase in 6OST2 and 6OST3, thereby indicating an absence of isoform compensatory mechanisms in these mice. Noteworthy, [IdoA,2S–GlcNS,6S] units were less affected than other 6-O-sulfate containing disaccharides. This suggests that 6OST1 may not be primarily involved in the 6-O-sulfation of heparin and highly sulfated HS. In contrast, analysis of HS from 6OST2 KO mouse embryonic fibroblasts (MEFs) showed a marked decrease in 6-O-sulfation, [IdoA,2S–GlcNS,6S] being the disaccharide unit the most affected. Finally, 6OST1/6OST2 double knockout resulted in an almost complete loss of 6-O-sulfation. Noteworthy, analysis of double KO MEF showed an increase in 2-O-sulfates, suggesting a possible compensatory effect between OSTs. In line with this, overexpression of any of the 6OSTs led to an increase in 6-O-sulfation, accompanied by a decrease in 2-O-sulfation [97]. Such compensation mechanism is still unclear and may extensively vary amongst tissues and species, but this clearly underlines the tight connections between the various enzymatic modification steps during HS biosynthesis.

Post-synthetic regulation of HS 6-O-sulfation status by the sulfs

In addition to the biosynthesis processes, HS 6-O-sulfation is further regulated through a unique post-synthetic mechanism involving extracellular enzymes of the Sulf family. Sulfs are endosulfatases that were first identified in Quails [104]. Since, Orthologs of these enzymes have been identified in mouse, rat, zebrafish and in human. In human, Sulfs exist as two isoforms: HSulf-1 and HSulf-2 that feature a common structural organization [105].

Sulfs are initially synthetized as pre-pro-proteins. Following removal of the signal sequence, the 125 KDa pro-proteins are then processed by furin-type proteases, yielding mature proteins composed of two subunits linked by one or more disulfide bonds: a 75 KDa N-terminal and the 50 KDa C-terminal regions [106, 107]. The N-terminal region (CAT) shares significant sequence homology with all members of the sulfatase family. It comprises the active site of the enzyme, including the strictly conserved the N-formylglycine residue (FGly) that is essential for arylsulfatase (sulfatase-generic) activity. In contrast, the C-terminal region contains a hydrophilic, highly basic domain (HD) that is a unique feature of the Sulfs, sharing no homology with any other known protein. Although not necessary for the arylsulfatase activity, the HD domain is responsible for the recognition and high affinity binding to HS substrates and is required for the enzyme endosulfatase activity (i.e. Sulf-specific ability to catalyze HS 6-O-desulfation) [106, 108]. Interestingly, HD domains from Sulf1 and Sulf2 isoforms show poor sequence homology, suggesting the existence of isoform-specific substrate preferences. Finally, the end of Sulf C-terminal region is homologous to Glucosamine-6-sulfatase (G6S), suggesting a role of this domain in the recognition of glucosamine motifs. Sulfs are N-glycosylated proteins, glycosylation accounting for ~20 % of the protein molecular weight. Although not investigated in human forms yet, a study on Quail Sulf-1 indicated that these glycosylations were necessary for appropriate cell-surface localization and enzymatic activity [109].

In contrast with other sulfatases, Sulfs are secreted proteins and exert their activity in the extracellular medium, on both cell-surface and ECM HS. Sulf-catalyzed 6-O-desulfation process is not fully understood yet, but recent data have provided further insights into the underlying mechanisms. Contrary to other sulfatases, Sulfs are endo-enzymes, which preferentially target internal HS highly sulfated NS-domains. HSulfs first bind with high affinity to the polysaccharide through their HD domains. Surface plasmon resonance analysis of Hsulf-1 HD domain [106] or full-length HSulf-2 (Vivès RR, unpublished data) interaction with heparin yielded high affinity Kds (in the nanomolar range) and showed the formation of very stable Enzyme-substrate complexes. Importantly, binding data could not be fitted to a simple 1:1 binding model, thereby suggesting the existence of a complex mode of interaction. Unique dynamic properties of HSulf/HS interactions were supported further by recent biophysical studies, which suggested the existence of multiple HS binding sites within HSulf-1 HD domain [110], and showed that this interaction was of the catch-bond type, which exhibits increased lifetime when subject to external forces [111]. Importantly, 6-O-sulfation was mandatory for high affinity HS/HD domain interaction [106, 110].

In a recent study, it has been shown that Sulf-driven desulfation always initiated at the non-reducing end of HS S-domains and proceeded towards the domain reducing end in a processive manner [48]. This implies that Sulf HD domain would primarily bind a saccharide motif downstream the S-domain non-reducing end to fit adequately the first 6-O-sulfated glucosamine in the active site. In addition to the HD domain, other epitopes present within the CAT domain may also participate to the binding and correct positioning of the polysaccharide in the active site (Vivès RR, unpublished data).

In absence of any detailed study, Sulf desulfation mechanism per se has been assumed to occur following the general arylsulfatase course [112]. This first involves hydroxylation of the active site FGly residue. The activated aldehyde hydrate then reacts and pulls out the 6-O-sulfate group of the glucosamine by a transesterification reaction. Finally, the FGly aldehyde group is regenerated by release of the sulfate group through an intramolecular rearrangement. After desulfation of the first glucosamine, highly dynamic, multiple binding sites within the HD domain may act as a sensor of following 6-O-sulfated residue and pull the enzyme along the polysaccharide [48]. Once the NS-domain reducing end is reached and in the absence of downstream sulfated residue, the affinity of the HD domain for the HS chain would drop and the enzyme would be released from the desulfated polysaccharide.

Although shared by both HSulf-1 and HSulf-2, such processive mechanism points out the HD domain as the key regulator of the enzyme activity and of isoform specificities. In agreement with this, HSulf-1 and HSulf-2 were able to catalyze full 6-O-desulfation of a heparin oligosaccharide in vitro, but with different rates [48]. HD domains may also contribute to differences in HSulf-1 and HSulf-2 substrate preferences. Saccharide structural requirements for Sulf substrate recognition are still poorly documented. Both enzymes essentially target [UA2S–GlcNS6S] trisulfated disaccharides units, although residual activity on [UA-GlcNS6S] disulfated disaccharides has also been reported [48, 73, 82, 113]. The absence of activity on GlcNAc-containing disulfated disaccharides suggests a requirement for N-sulfate groups. In addition, HSulfs seem to indistinctly accommodate both IdoA- of GlcA-containing disaccharides [73], while QSulf shows activity on [GlcA-GlcNS(6S)] but not on [IdoA-GlcNS(6S)] disaccharides [82, 114].

Because of these stringent substrate preferences ([UA2S–GlcNS6S] units usually account for less than 10 % of HS disaccharide content), consequences of Sulf activity on HS are structurally very limited. However, by specifically targeting HS functional NS domains, Sulfs have been shown to dramatically alter the polysaccharide binding properties, with very variable functional consequences (Fig. 3). Sulfs indeed inhibit a variety of heparin binding proteins, including growth factors such as FGF1 [48, 67], FGF2 [115–118], HGF [117, 119], HB-EGF [115, 120], amphiregulin [121] or TGFβ [79]. In many cases, downregulation is most likely the consequence of a structural alteration of cell-surface HS acting as coreceptors for these growth factors. Compromised binding may also have direct consequences for proteins such as chemokines, which activity relies on the formation of protein gradient stabilized through interactions with proteoglycans. Inhibition of HS interaction with CXCL12 and CCL21 by HSulf-2 has been demonstrated in vitro [67]. Although the physiological relevance of such regulation has not been investigated yet, effects on leukocyte recruitment/homing may be anticipated. However, Sulfs have also been shown to induce signaling pathways, as originally demonstrated for Wnt morphogen [104]. As explained above, Sulf mediated 6-O-desulfation weakens Wnt/HS interaction, allowing formation of a ternary HS/Wnt/Fz signaling complex [82]. Likewise, a similar mechanism has been reported for the regulation of Glial cell-derived neurotrophic factor (GDNF) [54, 122]. Sulfs have also been shown to indirectly promote BMP signaling, by modulating its inhibition by Noggin [114, 123]. Finally, Sulfs have demonstrated antagonist activities for some ligands. For instance, Sulf-1 has been shown to either inhibit or induce or inhibit Sonic Hedgehog (Shh) signaling in gastric cancer or during neuronal development, respectively [124, 125].

Fig. 3
figure 3

Regulation of Heparin-binding protein signaling by the Sulfs. At the cell surface, 6-O-desulfation mediated by the Sulfs can have different consequences on the signaling of HS binding proteins. It can downregulate FGFs (left side), by preventing formation of the FGF/HS/FGFR ternary complex that triggers intracellular signaling, and it can induce Wnt (right side), by lowering the affinity of the Wnt/HS interaction, thereby allowing subsequent binding to the Fz receptor. In contrast, alteration of ECM HS structure by the Sulfs may release sequestered proteins (including FGFs) to elicit their functions

Full size image

Thanks to these complex HS regulatory properties, Sulfs are involved in major physiological processes. Notably, Sulfs have been shown to play central roles during development, which could be in part deduced from the study of KO animals. In mouse and chick, Sulfs have been implicated in neuronal [54, 124, 126–128], skeletal and cartilage [129, 130] development, formation of the inner ear [131] and dentinogenesis [132]. In C. elegans, Sulf1 participates to the dorsal ventral patterning of the neural tube [133, 134]. Finally, in Zebrafish, alteration of BMP, FGF and CXCL12 signaling resulting from the loss of sulf1 led to poor differentiation of the somitic trunk muscle, loss of the horizontal myoseptum, reduction of pigmentation along the mediolateral stripe, and incorrect migration of the lateral line primordium [135].

In adults, Sulfs have been implicated in tissue and organ regeneration [136–138], cartilage homeostasis [123] as well as maintenance and stability of renal glomerular filtration barrier [139, 140]. Sulfs have also been very early associated with a number of pathologies and disease. The role of Sulfs in cancer has been thoroughly described elsewhere and will not be detailed here (for reviews, see [11–13, 141]). Briefly, Sulfs have been associated with large variety of cancers (leukemia, ovarian, liver, pancreas, lung, breast, brain, kidney, bladder, colon, gastric, and head and neck cancer) and have been involved in all major stages of the disease, including tumoral transformation [142], growth, invasion and metastatization [116, 143–147], as well as tumor cell sensitization/resistance to drugs [115, 148]. HSulf-2 in particular has been marked as a target of interest in cancer therapy, especially for tumors of poor prognosis, such as lung squamous cell carcinoma and lung adenocarcinoma [12, 142].

Increased expression of both Sulf-1 and Sulf-2 has also been reported in osteoarthritic and aging cartilage [149]. In osteoarthritis, abnormal chondrocyte activation and cartilage degradation may result from Sulf-catalyzed alteration of HS structure and binding properties, and subsequent effects on the signaling of many heparin-binding growth factors. More recently, expression of Sulf-1 has been found to be induced in degenerative intervertebral disc cells [150]. Although a direct link between the enzyme expression and the pathology remains to be demonstrated, this study emphasized further the role of Sulfs in cartilage homeostasis and disease. Another emerging but yet poorly explored area is the implication of the Sulfs during inflammation. As explained before, HS 6-O-sulfation is critical for the binding of many chemokines and Sulfs may therefore regulate these interactions, as demonstrated in vitro for CXCL12 and CCL21 [67]. HS 6-O-sulfation is also required for binding to L-selectin, which is implicated in the early events of leukocyte extravasation [27], and has been associated with heparin-induced leukocytosis, through disruption of Selectin- and CXCL12-mediated leukocyte trafficking [69]. This suggests that Sulfs could act as a modulator of leukocyte migration and adhesion to activated endothelium. In line with this, an in vivo study on renal allograft biopsies showed that Sulf-1 expression was repressed in inflammatory conditions [151]. In another study, Sulf-2 has been implicated in idiopathic pulmonary fibrosis, presumably, through a regulation of TGFβ1 signaling in type 2 alveolar epithelial cells [80]. TGF-β1 is a major actor during the pathogenesis of pulmonary fibrosis [152], which binding to HS requires 6-O-Sulfation and has been shown to induce both Sulf1 and Sulf2 expression in lung fibroblasts [79, 80] and renal epithelial cells [26]. Control of TGF-β1/Sulf expression and activities may thus occur through negative feedback loop system. Further study of such regulatory mechanism could provide novel insights into the role of Sulfs in other TGF-β1 involving processes, such as cell differentiation, chemoattraction, and the control of the balance between cell survival and apoptosis. Finally, Sulf-2 has been associated with Type-2 diabetes mellitus, the enzyme inhibiting very-low-density lipoprotein (VLDL) binding to HS hepatocytes and disrupting triglyceride clearance [29]. Interestingly, Sulf-2 was also found overexpressed in the serum of cirrhotic patients, suggesting potential use as serologic biomarker [153]. More recently, Sulf-1 genetic polymorphism has been associated with in vitro fertilization (IVF) failure [154, 155].

However, despite increasing interests, Sulfs are still poorly characterized enzymes and major issues remain to be clarified. Amongst them is the existence of possible differences between Sulf isoforms. In vitro, analysis of HSulf-1 and HSulf-2 activity did not show any isoform-specific activities or substrate preferences. In contrast, these enzymes exhibited either redundant, overlapping or even antagonist functions in vivo, according to the biological system considered. Studies of the role of Sulfs during development and in cancer have provided strong evidence of such discrepancies.

In mice, the single knockout of Sulf-1 or Sulf-2 was not lethal and did not cause severe abnormalities or histological defects, while Sulf-1/Sulf-2 double knockout led high neonatal mortality and to multiple phenotype anomalies [130, 156]. These data therefore pointed out major overlapping functions and/or compensation effects between the two isoforms during development. In agreement with this, analysis of HS from KO mice revealed that 6-O-sulfation content was significantly higher in Sulf double KO than in simple KO mice, thereby supporting some functional cooperativity of Sulf-1 and Sulf-2 isoforms [157]. Surprisingly, increased 6-O-sulfation was significantly detected in HS NAc/NS transition zones. This was therefore in contradiction with in vitro analyses, which reported that Sulfs exclusively targeted highly sulfated disaccharides that are normally present within NS-domains. Interestingly, a significant reduction in HS N- and 2-O-sulfate content, along with changes in 2OST and 6OST expression were also observed [158]. These data thus suggested the existence of interconnexions between Sulf activity and HS biosynthesis machinery in vivo.

However, these studies also provided indications of isoform specificities. Interestingly, increased expression of Sulf-1 could compensate the loss of Sulf-2, but Sulf-2 could not completely substitute for the lack of Sulf −1 [157]. Furthermore, comparative analysis of HS from single Sulf-1 and Sulf-2 KO mouse organs revealed sulfation differences [159]. Non-redundancy of Sulfs functions was also reported during mouse brain development, as the two isoforms differently contributed to neurite outgrowth of cerebellar and hippocampal neurons, synaptic plasticity, and motor activity [126]. Major differences between Sulf-1 and Sulf-2 activities have also been reported in Cancer. Although still debated, HSulf-1 has been frequently associated with anti-oncogenic activities, while HSulf-2 is generally associated with pro-oncogenic activities [11, 12]. In line with this, both HSulfs have been shown to downregulate pro-angiogenic growth factors in vitro, but HSulf-2 promotes tumour angiogenesis in vivo [160, 161]. The rationale behind such a discrepancy still remains poorly understood. However, one explanation could be that the two isoforms feature subtle substrate specificities and/or act on different HS subsets. For instance, targeting cell-surface HS that act as pro-angiogenic/pro-oncogenic growth factor coreceptors, or ECM HS involved in sequestration and storage of these growth factors, may have opposite effects (Fig. 3).

Concluding remarks

Over the past years, understanding of HS structure/function relationships has considerably evolved. HS has now emerged from a mere structural component of the extracellular matrix/cell surface interface, to a critical regulator of cellular signaling networks, most of these functions being the results of its unique interactive properties. Characterization of AT-III binding pentasaccharide three decades ago led the field to envision a concept of one saccharide sequence binding to one protein ligand for achieving one function, but accumulating evidence about the structural basis of HS/protein interactions tends now to suggest that this may prove right for a limited number of examples, involving notably rare 3-O-sulfated saccharide motifs [5, 9]. For the three major sulfate positions (ie. N-, 2-O-S and 6-O-sulfates), studies on HS biosynthesis enzyme KO organisms have clearly underlined the existence the existence of compensatory effects, and specificity for protein interactions may rather rely on saccharide motifs exhibiting appropriate spatial distribution of sulfates, than on linear sequences.

Nevertheless, the specific importance of 6-O-Sulfation pattern in HS structure/function relationships still rises intriguing questions. As explained above, 6-O-sulfates are critical contributors to HS structural diversity and are involved in major functions of the polysaccharide. While N- and 2-O-sulfates are ubiquitously required for protein binding, 6-O-sulfate contribution is much more restricted and has been associated with binding selectivity and regulatory mechanisms, as exemplified by the HS/FGF interaction system. Furthermore, HS 6-O-sulfation is the only modification being determined at two distinct stages: during HS biosynthesis and post-synthetically. However, the 6OST and Sulf families that are involved in these processes still remain poorly characterized enzymes. Mechanisms regulating the 6-O-sulfation status of HS may thus be the most complex and the least understood of all HS chain maturation processes.

For instance, the presence of significant amounts of 6-O-sulfates in the NAc/NS flanking regions is still puzzling. This suggests that 6OSTs somehow escape the stepwise nature of the HS biosynthesis process. However, the discrete substrate preferences identified to date for the various 6OST isoforms have failed to provide a rationale for such an atypical sulfate distribution. Recent studies on HS biosynthesis have led to the emergence of the GAGosome concept, which suggests that HS biosynthesis enzymes function collectively within supramolecular complexes [10]. Evidence of cooperativity and formation of complexes between EXT/NDST [162], and C5-epimerase/2OST [15] have already been reported. Although interaction of 6OST with 2OST and C5-epimerase has been reported in Drosophila S2 cells [163, 164], the question of the place of 6OSTs within the GAGosome remains to be addressed. Furthermore, 6OSTs remain the only family of HS biosynthesis enzymes, for which no crystal structure is available yet. Sulfs are also highly elusive enzymes, which distinguish from all other sulfatases both in term of structure and function. Study of the Sulfs remains extremely challenging, as they are difficult to express recombinantly and analysis of their enzyme activity is clearly hampered by the natural complexity of their HS substrates. Furthermore, increasing evidence points out their HD domain as the major determinant of the enzyme function. However, this domain is presumed to be poorly structured and thus extremely difficult to characterize. Unexplained discrepancies have been observed between their in vivo and in vitro activities, and between Sulf isoforms, which display either analogous or antagonist activities according to the physiopathological conditions. This could suggest the existence of isoform substrate preferences, yet to be defined, and possibly of non-catalytic mechanisms regulating the enzyme distribution, diffusion throughout tissues and access to specific HS subsets.

The 6-O-sulfation status may thus be defined by the complex interplay between these two enzymes families. This may most likely involve specific spatiotemporal expression patterns, as illustrated recently during prostate development [165]. However, competition and/or coordination of these enzyme activities may be far more intricate. One could even challenge the present paradigm restraining 6OST and Sulfs activities to the Golgian and extracellular compartments, respectively. It is indeed intriguing to note that 6OSTs are amongst the rare HS biosynthesis enzymes to be secreted in the extracellular compartment [90] and conversely, that Quail Sulf-1 has been shown to exhibit enzyme activity in the Golgi apparatus [82]. Although the physiological relevance of these activities is not clear yet, these may play an additional role in the regulation of HS 6-O-sulfation status. Undoubtedly, apprehending the structural and functional subtleties of HS 6-O-sulfation will require further understanding of the 6OST and Sulf enzyme activities.