The biosynthesis of GAGs, and HS in particular, is not a template-driven process, meaning that the extent of modification reactions that the glycan chains may undergo, and the consequent HS final structural motifs, are not directly encoded in the genome. The current general view is that HS biosynthesis is regulated by the availability and expression levels of the enzymes involved in this pathway, and by the binding specificities of these enzymes towards the sequences that are formed in HS growing chains (19). Hence, the structural features resulting from each stage of HS biosynthesis determine subsequent modification reactions, contributing to the “step-by-step” fine tuning of HS overall structure. However, not all of the HS substrates are equally modified, thus giving rise to a great HS structural variability.

The majority of the enzymes that intervene and control HS biosynthesis are located in the Golgi apparatus and are classified as type II transmembrane proteins (15). A HS biosynthesis model has proposed that these enzymes are in tight interaction with each other, forming a supramolecular complex called “GAGosome”, which ensures quick and concerted reactions towards the formation of the GAG chains (20). However, the regulatory mechanisms underlying the activity of this possible enzymatic machinery complex and its impact in HS high structural variability are not yet fully disclosed.

HS biosynthesis is initiated by the formation of a universal tetrasaccharide linker covalently attached to the core protein of all GAG bearing proteoglycans, including HSPGs. Assembly of this tetrasaccharide linker starts with the transfer of a xylose (Xyl) residue to specific protein serine amino acids by two O-Xylosyltransferases (XYLT1 and XYLT2) ( Figure 1 ) (22). This is followed by the sequential transfer of two galactose (Gal) residues, successively added by the Galactosyltransferase-I/β4-Galactosyltransferase 7 (β4Gal-T7) and the Galactosyltransferase-II/β3-Galactosyltransferase 6 (β3Gal-T6), and lastly by the addition of one GlcA residue by the Glucuronyltransferase-I (GlcAT-I), to form the linker GlcAβ1-3Gal-β1-3Gal-β1-4Xyl-β1-O-Ser. During this assembly, transient phosphorylation of the Xyl residue by the kinase FAM20B and dephosphorylation by the 2-Phosphoxylose phosphatase (XYLP) also take place, which enhance the activity of β3Gal-T6, further promoting the maturation of the linker (23, 24). The polymerization of HS chains is then initiated by the transfer of a GlcNAc residue to the tetrasaccharide linker, which is regulated by Exostosin Like 2 (EXTL2) (25) and EXTL3 glycosyltransferases (26), and is followed by further chain elongation promoted by a hetero-oligomeric complex formed by EXT1 and EXT2, which catalyses the alternate transfer of GlcNAc and GlcA residues ( Figure 1 ) (27). Polymerized HS chains then undergo extensive processing and modification reactions that give rise to fully mature HS chains, which are the main focus of this review.

Although unique and distinctive biologically active HS structural motifs have not been disclosed for most protein-HS interactions, distinct roles of the most common types of sulfation (N-, 2-O- and 6-O-sulfation) have been reported for protein binding affinities and biological activities. Basic Fibroblast Growth Factor (FGF2 or bFGF) was the first growth factor reported to depend upon cell surface HS as a co-receptor required for the formation of a biologically active FGFR-FGF2-HS ternary complex (66, 67). FGFs constitute a large family of mitogens involved in the regulation of cellular migration, proliferation, differentiation, and survival (68). FGF2-HS interaction, in particular, is important not only for receptor dimerization and activation, but also for evasion of degradation by FGF2 molecules (69). Maccarana M. et al. showed that IdoA2S and GlcNS were important units for FGF2 binding, whereas the presence of GlcN6S residues was irrelevant for HS-FGF2 interaction (70). It was later reported that 6-O-sulfated residues were required to bridge FGF2 and FGFR, promoting the formation of the full ternary complexes and receptor activation (71–73).

Interestingly, a specific trisaccharide that included both IdoA2S and GlcNS6S residues (IdoA2S–GlcNS6S–IdoA2S) was invariably detected in FGF1 binding sites in HS chains (74). It was demonstrated that the presence of this specific motif repeated in HS polysaccharides contributed to higher FGF1 binding affinity, when compared with HS polysaccharides with increased overall sulfation, and that 6-O-sulfation was important for FGF1-HS interaction (74). Overall, HS structural requisites for binding of FGF1 and FGF2 illustrate the consensus on the general role of N- and 2-O-sulfation in ligand binding and the specificity provided by 6-O-sulfation.

Another example illustrating the biological impact of 6-O-sulfation in HS binding specificities relates to Vascular Endothelial derived-Growth Factors (VEGF) driven-signalling. VEGFs are key molecules in vasculogenesis and angiogenesis, regulating these events during homeostasis and pathology progression (75). In a recent report, it was revealed that HS chains present on the transmembrane HSPG syndecan-2 formed a ternary complex with VEGFA₁₆₅ and its target receptor VEGFR2. VEGFA₁₆₅ is an important growth factor involved in retinal angiogenesis, and the formation of this complex was shown to enhance significantly VEGFR2 signalling in endothelial cells, when compared to the binary complex VEGFA-VEGFR2 (76). GAG disaccharide analyses revealed high 6-O-sulfation in HS of syndecan-2 (76), a structural feature that had been previously shown to be specifically related with the increased VEGFA₁₆₅ binding affinity to HS chains (77).

HSPGs also modulate the activity of morphogens, which are key elements in the cellular events that occur during early development, by providing a scaffold for stabilizing extracellular gradients of these signalling molecules, or acting directly as cell surface co-receptors for receptors activation. Noteworthy, signalling pathways of major HS-binding morphogens including Hedgehog, Bone Morphogenic Protein (BMP) and Wnt, were shown to be impaired by aberrant HS biosynthesis, resulting from the loss of EXT gene expression and function (78, 79). HS bioactive domains for Wnt and for a member of the BMP family, BMP-2, have also been previously disclosed. Wnt interaction with HS and subsequent activation were associated mainly with the polysaccharide 6-O-sulfation content, and its structural remodelling by the Sulfs (80, 81). The most critical components of BMP-2 HS-binding domain were revealed to be GlcNS residues, while 6-O- and 2-O-sulfation were also important, but to a lesser extent (82).

The most selective interactions take place in regions of HS chains that contain rare modifications, such as GlcNH₃ and 3-O-sulfated GlcN residues (83). The first identified and most studied example of a HS specific saccharide sequence required for protein high specificity binding and activity relates to antithrombin, an important inhibitor of blood coagulation, whose inhibitory activity is potentiated upon binding to heparin (84). Since the late 70s, many studies have been dedicated to disclose a specific binding epitope within heparin responsible for antithrombin interactions and activity, and led to the identification of an antithrombin binding pentasaccharide GlcNAc6S-GlcA-GlcNS3S6S-IdoA2S-GlcNS6S (85, 86). Within this sequence, the GlcNS3S residue is highlighted as a fundamental distinctive structural feature as it is an otherwise rare unit within heparin/HS chains (85, 87).

The binding of Herpes simplex virus type 1 glycoprotein D to cell surface HS, that functions as a critical receptor for the virus entry in host cells, was shown to also depend on specific sites within HS chains that include 3-O-sulfation (either IdoA2S-GlcNH₂3S6S or IdoA2S-GlcNH₂3S) (88). In addition to its role in antithrombin activity and cellular viral entry, this less frequent type of HS modification has been associated with a few highly selective interactions with other types of bioactive molecules with impact in cell physiology, including the growth factor FGF7 (89), cyclophilin B (90) and the cell surface receptor Neuropilin-1 (91). Lastly, a recent report showed that 3-O-sulfation significantly enhances tau binding to cell surface HS, and subsequent tau cellular uptake, ultimately contributing to the prion-like spread of tau pathology in Alzheimer’s Disease (92).

Regarding NDST isozymes that act in the initial step of HS modification, namely N-deacetylation and N-sulfation of GlcN, loss of NDST4 gene expression in colorectal cancer has been associated with higher pathological stages and patient’s poor prognosis (95). Additionally, Fuster M. M. et al. studied the impact of another NDST isoform, NDST1, on tumour angiogenesis, and showed that endothelial cells isolated from Ndst1 KO mice synthesized structurally modified HS chains that impaired angiogenesis-related signalling pathways, leading to decreased vascularization of lung tumours and consequent reduced tumour growth (96). Altered expression of GLCE, the epimerase controlling HS GlcA/IdoA ratio, was also reported in cancer. Studies indicated significant decreased expression of GLCE in breast tumours (97), even at premalignant stages of the disease, and in lung cancer cells (99). Furthermore, anti-proliferative effects were reported on both breast (98) and small-cell lung (99) cancer cells after induced re-expression of GLCE, highlighting this gene as a potential tumour-suppressor gene.

The sulfotransferases that catalyse the more common HS O-sulfation reactions, 2OST1 and 6OSTs, have also been implicated in cancer pathology. Gene expression datasets from the Oncomine database were explored and revealed that 2OST1 expression is upregulated in prostate carcinoma and functional assays showed that it correlates with cell proliferation and invasion capabilities, known to enhance cancer cells metastatic behaviour (100). In the same line, but in a different model, it was demonstrated that downregulation of 2OST1, and concomitant change in HS structural patterns, accompanied the granulocytic differentiation of SKM-1 leukaemia cells and were associated with cells’ growth inhibition and less aggressive phenotypes (101). In breast cancer, 2OST1 expression was shown to promote the acquisition of cancer stem cell-like properties, by activating stemness-associated signalling pathways (102). Additionally, and contradicting the previously reported 2OST1 “tumour promoter gene”-like features, upregulation of 2OST1 in breast cancer cell lines was shown to promote cell-cell and cell-ECM adhesion and to inhibit cell migration and invasion capabilities (103). These effects were associated with increased HS 2-O-sulfation and subsequent altered growth factor binding affinities, decreased expression and activation of the Epidermal Growth Factor Receptor (EGFR), which impaired different signalling pathways that modulate cancer cell invasiveness, and increased expression of E-cadherin (103).

Regarding 6OSTs, Waaijer C. et al. reported upregulated expression of 6OST1 and 6OST2 during chondrosarcoma progression, correlated with increasing tumour histological grade, higher levels of 6OST3 in most of the analysed cartilage tumours, and higher levels of 6-O-sulfated HS disaccharides in high-grade chondrosarcoma cell lines (104). Similarly, the isoforms 6OST2 and 6OST3 were also found to be overexpressed in colorectal (105) and breast (106) cancer, respectively. 6OST3 expression in particular was shown to influence breast tumour cell growth, invasion and migration (106). HS 6-O-sulfation content was also shown to be augmented in the ovarian endothelium of patients with advanced stage ovarian cancer (115). In agreement with this work, it was later disclosed the impact of HS 6-O-sulfation content, majorly determined by 6OST isozymes, on ovarian cancer angiogenesis (116). In this work, Cole C. L. et al. showed that HS 6-O-sulfation stimulates the heparin-binding EGF-like growth factor (HB-EGF)-dependent activation of cell surface receptor EGFR, which leads to increased expression of angiogenic cytokines (interleukin 6, interleukin 8 and FGF2) by ovarian tumour cells (116). HS post-biosynthesis editing by 6-O-endosulfatases represents an additional regulatory mechanism frequently altered in different tumour models, further emphasizing the role of polysaccharide altered 6-O-sulfation in cancer (16, 117).

Lastly, aberrant expression of 3OSTs, which catalyse the rarest HS modification, was also reported in various human cancers, suggesting an important role in cell malignant transformation and tumour cells’ functional features. Miyamoto K. et al. described hypermethylation of 5’ region of the 3OST2 gene and consequent loss of its expression in human primary breast, colon, lung and pancreatic cancers (107). Later, Kumar A. V. et al. have taken this premise and further investigated the possible functions of 3OST2 in tumorigenesis (109). These authors reported that re-expression of this gene in a highly invasive breast cancer cell line augmented the activation of Mitogen-Activated Protein Kinase (MAPK) and Wnt/β-catenin signalling pathways, which promoted breast cancer cell invasiveness, motility and chemoresistance (109). Additionally, 3OST2 expression in breast cancer cells was shown to be related with acquisition of a cancer stem cell-like phenotype (102). Hypermethylation and silencing of 3OST2 was also reported in non-small cell lung cancer, and it was associated with poor overall patient survival. Hwang J. A. et al. induced exogenous expression of this gene in two lung cancer cell lines and observed reduced cells’ migration and invasion capabilities, as well as their lower proliferation rate, supporting that 3OST2 hypermethylation might promote lung tumorigenesis (108). Regarding different isoforms, 3OST3A was revealed to be epigenetically repressed in breast cancer cell lines representative of distinct molecular subgroups, and it was shown to induce opposite effects, either oncogenic or tumour-suppressive, depending on tumour cell phenotypes (110). Moreover, 3OST3B1 expression was associated to acute myeloid leukaemia progression, by inducing expression and shedding of proangiogenic factors (111), and, more recently, it was found to be upregulated on non-small cell lung cancer and to regulate epithelial-to-mesenchymal transition (112). Interestingly, the telomerase protein TRF2, whose expression is upregulated in tumour cells and associated to suppression of immune response, was shown to regulate 3OST4 expression and to prevent natural killer (NK) cell recruitment and promote cancer immune escape (118, 119).

Together with changes in HS modifying enzymes, altered HS sulfation levels and structural conformations have also been studied and described in different types of cancer, in an attempt to profile specific sulfation patterns as distinctive traits in cancer pathologies that can potentially serve as predictive biomarkers.

GAG disaccharide profiling of cancer patients’ tissue samples has revealed decreased HS 6-O-sulfation and total O-undersulfation in Hepatocellular cancer (120) and overall decreased HS sulfation, both N and O, in renal cell carcinoma, mainly due to lower levels of mono 6-O-sulfated disaccharides, and all disulfated disaccharides (121). HS sulfation was also shown to vary in breast cancer but in a cell dependent manner. As such, it was shown that MCF7 breast cancer cells, expressing oestrogen and progesterone receptors, produce HS with reduced content in 2-O-sulfated disaccharides and enriched in 6-O-sulfation. In contrast, HS isolated from MDA-MB-231 and HCC38 cells, triple-negative for oestrogen and progesterone receptors, as well as human epidermal growth factor receptor-2 (HER2), were more 2-O-sulfated and less 6-O-sulfated, both compared with a non-tumorigenic human mammary gland epithelial cell line (122).

On the other hand, data regarding 3-O-sulfation levels in tumours is very scarce. This is mainly due to the shortage of commercially available 3-O-sulfated disaccharide standards that can be used for comparison in HS disaccharide structural analyses, as well as the low susceptibility of 3-O-sulfated disaccharides to the currently used Heparinases (I, II and III) for exhaustive HS depolymerisation, which is a prerequisite for this type of analyses.

Despite the known role of HS sulfotransferases in determining HS conformations and interacting partners, which influences cell signalling networks, the targeting of these enzymes has been considerably less explored for innovative therapeutic strategies. Some reports have focused on the use of small molecules to manipulate the binding and/or activity of HS sulfotransferases over the last years ( Figure 3A ). In one of these projects, Byrne D. P. et al. screened several small-molecules from a Public Kinase Inhibitor Set library, and validated cell permeable compounds as probes with inhibitory activity against 2OST, namely three polyanionic compounds - suramin, aurintricarboxylic acid and rottlerin - and one oxindole RAF kinase inhibitor - GW407323A (123). In addition, a new approach for quicker, reproducible and cheaper detection of substrate sulfation was employed. This sulfation detection strategy is based on microfluidics and differential scanning fluorimetry, and in the future it could be applied for the discovery of new inhibitory compounds against other sulfotransferases, like 6OSTs and 3OSTs (123).

More recently, it was reported a methodology for transient and reversible inhibition of HS biosynthesis also involving a small cell-permeable compound, the tetra-acetylated N-azidoacetylgalactosamine (Ac₄GalNAz). Maciej-Hulme M. et al. have shown that the treatment of Chinese hamster ovary cells with Ac₄GalNAz induces early termination of HS elongation, reducing HS levels and length, potentially by interfering with the activity of NDSTs (124). In a first instance these types of tools will be extremely relevant to further study HS and HS-sulfation dependent cell mechanisms and impact in cancer pathology, and in the long-term these could potentially be applied in cancer research and therapy.

HS mimetics are synthetic and homogenous molecules, that can either be saccharide or non-saccharide based, with great potential as anti-cancer agents. These molecules function by competing with HS native chains and blocking HS-protein interactions ( Figure 3B ), and a few of them either have been or are currently under clinical trials, namely CX-01 (ODSH), Roneparstat (SST0001), Necuparanib (M402), Muparfostat (PI-88) and Pixatimod (PG545) (125). A great variety of these HS mimetics inhibit either Heparanase and/or Sulfs activity and directly target HS-growth factor interactions, and their specific features and functions have been recently reviewed in (126). Furthermore, it has been reported that these mimetics may also inhibit the activation of RTKs. For example, Roneparstat was shown to inhibit FGF, Insulin Growth Factor (IGF), EGF and Platelet-Derived Growth Factor (PDGF) receptors expressed in sarcoma cells (127).

In addition to these widely studied HS mimetics, several different HS-like compounds have been developed and/or discovered and studied specifically due to their ability to interfere with HS mediated signalling cascades, that are yet to be further explored in clinical studies.

Sutton A. et al. revealed the anti-tumour effects of two synthetic sulfated polymers HS mimetics, OTR4120 and OTR4131, in hepatoma cells. These compounds interfere with a HS-mediated signalling pathway regulated by the CC-chemokine Regulated upon Activation, Normal T Cell Expressed and Presumably Secreted (RANTES), also referred to as CCL5, which was shown to induce cell migration and invasion in a HS-dependent manner (128). The G2.2, another synthetic sulfated non-saccharide HS mimetic, has also been evaluated in regard to its anti-tumour activity and it was shown to selectively target and reduce colorectal cancer stem cells, by inducing the activation of p38 MAPK and impairing cell self-renewal, ultimately reducing tumour growth (129). In a different study, Shanthamurthy C. D. et al. synthesized a highly sulfated IdoA-based oligosaccharide as a novel HS mimetic to target chemokines and modulate its activity in cancer progression, and reported its high binding affinity towards several homeostatic and inflammatory chemokines. Additionally, they showed that this HS mimetic, potentially by binding to CCL2, inhibits breast cancer CCL2-mediated cell proliferation and CCR2/CCL2- mediated cell migration, and it reduces cell invasiveness (130). More recently, Jain P. et al. have resorted to a library of HS tetrasaccharide ligands with varying sulfation patterns and high-throughput array binding assays to further address HS binding specificities, and validated two HS analogues, HT-2,6S-NAc and HT-6S-NAc, as potential ligands to target VEGF₁₆₅-mediated cellular events, as these were shown to not only bind with high affinity to the mentioned growth factor but also to inhibit endothelial cells’ VEGF₁₆₅-induced proliferation, migration and tube formation, which are known to be relevant tumour related features (131).

Another promising tool for cancer therapy pertains to the use of synthetic xylosides. These molecules comprise a xylose residue linked to an aglycone group and serve as primers for protein-free GAG biosynthesis in the Golgi. Synthetic xylosides compete with PGs’ core protein for biosynthetic enzymes, impairing the elongation and modification of GAG chains attached to PGs, while on the other hand the xyloside primed GAG chains secreted into the extracellular environment also compete with endogenous PG-linked GAGs for binding of different ligands ( Figure 3C ) (132).

The use of synthetic xylosides with HS priming activity was shown to effectively inhibit several tumour-related cellular events. Raman K. et al. revealed significant reduction of the invasive capabilities of glioma cells treated with click-xylosides (133). It has also been reported the anti-angiogenic efficacy of two different fluoro-xylosides that by inhibiting HS synthesis in endothelial cells, ultimately inhibit tumour cells’ angiogenesis (134), which is in agreement with the previously mentioned role of HSPGs as important binding partners and co-receptors of pro-angiogenic factors, including VEGF and FGF. More recently, novel xyloside-derived compounds with potential to be used in therapy for glioblastoma were identified due to their ability of impairing endogenous GAG biosynthesis by binding to XYLT1 and β4GAL T7 active sites, and consequently decreasing glioblastoma cell viability (135). In a different study, Mani K. et al. demonstrated that the use of the synthetic xyloside 2(6-hydroxynaphthyl)-β-D-xylopyranoside inhibited tumour cell growth in vitro, in human lung and hepatocellular carcinoma, and SV40-transformed mouse embryonic fibroblasts 3T3 cells, as well as in vivo, in human bladder carcinoma cells (136). However, unlike the previously mentioned studies, they hypothesised that this anti-proliferative effect, observed at low xyloside doses, was due to the accumulation of antiproliferative products resulting from degradation of the xyloside-primed HS chains inside the cells, rather than competition with PG core protein or PG-linked GAGs for HS biosynthesis or ligand-binding, respectively (136).

HSPGs expressed at cell glycocalyx also function as internalizing receptors for extracellular vesicles (EVs). EVs are cell-derived vesicles that mediate intercellular communication and display crucial roles in a multitude of physiological and pathophysiological processes. In recent years, it has been suggested that EVs released by tumour cells play an important role in the establishment of premetastatic niches and in metastasis. Christianson H. C. et al. have shown that treatment of glioma cells with xylosides inhibited EV cellular uptake and consequently it reduced cell migratory capabilities (137).

Treatment of cancer cells that feature aberrant expression of HS sulfotransferases with xylosides could potentially lessen the synthesis of endogenous PGs modified with HS chains with altered sulfation patterns, diminishing the effects of PG-linked GAG oversulfation on different cell signalling pathways involved in tumorigenesis. Moreover, a higher impact over HS sulfotransferase activities could be achieved by manipulating the structure of the aglycone portion of the synthetic xylosides (138), which could be employed to further increase enzyme affinity towards these synthetic molecules. This type of xyloside dependent HS structural changes were also reported by Chen Y. et al. when comparing the GAG priming efficiency of two synthetic xylosides, 2-naphthyl-β-D-xylopyranoside and its derivative 2-(6-((3-aminopropyl)oxy)-naphthyl)-β-D-xylopyranoside, particularly in terms of sulfation (139).

In addition to the previously mentioned strategies, a different line of investigation has been dedicated to the development and study of molecules that impair HS-ligand interactions by directly targeting HS and HSPGs, namely anti-HSPG antibodies ( Figure 3D ).

Gao W. et al. generated a human monoclonal HS-specific antibody, HS20, targeting HS chains found on glypican-3 (140), a cell surface HSPG known to be highly expressed in hepatocellular carcinoma and associated with patient poor prognosis (141). Treatment of hepatocellular carcinoma cells with HS20 was revealed to block the activation of the HGF/Met pathway, and consequently to inhibit HGF-induced cell migration, motility, and 3D-spheroid formation, as well as in vivo liver tumour growth (140).

Syndecan-1 is another cell-surface HSPG that has been revealed to be a potential immunotherapeutic target for cancer therapy, more specifically for multiple myeloma. Jiang H. et al. developed modified NK cells expressing a syndecan-1-specific chimeric antigen receptor (CAR), that showed enhanced in vitro and in vivo cytotoxicity against syndecan-1 positive multiple myeloma cells, presenting a new possible approach for efficient and specific cancer immunotherapy (142). More recently, it was also developed a new monoclonal anti-syndecan-1 antibody, VIS832, to use in multiple myeloma therapy, and it was shown to induce potent NK cell-mediated antibody-dependent cellular cytotoxicity and macrophage-mediated antibody-dependent cellular phagocytosis against myeloma cells either sensitive or resistant to current therapies (143). In this study it was further demonstrated the efficacy of the VIS832 treatment in a murine model of disseminated human multiple myeloma, both as monotherapy and combined with a proteasome inhibitor (bortezomib) used in therapy (143).

In addition, anti-HSPG antibodies also hold great potential for safer and highly specific drug delivery in cancer immunotherapy. Indatuximab ravtansine (BT062) is an example of such therapeutic approach, as an anti-syndecan-1 monoclonal antibody (nBT062) conjugated with a highly cytotoxic maytansinoid derivative (DM4). This antibody-drug conjugate displayed in vitro and in vivo anti-tumour activity against multiple myeloma cells expressing syndecan-1, both as monotherapy (144) and in combination with other clinically approved anti-myeloma drugs (145), and more recently it has been under phase I and phase I/IIa clinical trials as monotherapy for relapsed and/or refractory multiple myeloma (146). In a different study, Bosse K. R. et al. have identified glypican-2 as a potential immunotherapeutic target for neuroblastoma due to its significantly higher expression in high-risk neuroblastomas, which was also associated with patient worse overall survival, and developed a highly cytotoxic antibody-drug conjugate that specifically targets glypican-2-expressing neuroblastoma cells (147).