Angiogenesis has emerged as a novel target for anti-cancer therapies through randomized clinical trials that tested the benefit of adding vascular endothelial growth factor (VEGF) inhibitors to conventional cytotoxic therapies. However, despite improvements in the progression-free survival, the benefit in overall survival is modest. Tumour angiogenesis is regulated by a number of angiogenic cytokines. Thus innate or acquired resistance to VEGF inhibitors can be caused, at least in part, through expression of other angiogenic cytokines, including fibroblast growth factor 2 (FGF2), interleukin 8 (IL-8) and stromal-cell-derived factor 1α (SDF-1α), which make tumours insensitive to VEGF signalling pathway inhibition. The majority of angiogenic cytokines, including VEGF-A, FGF2, IL-8 and SDF-1α, manifest an obligate dependence on heparan sulfate (HS) for their biological activity. This mandatory requirement of angiogenic cytokines for HS identifies HS as a potential target for novel anti-angiogenic therapy. Targeting multiple angiogenic cytokines with HS mimetics may represent an opportunity to inhibit tumour angiogenesis more efficiently. Our published studies and unpublished work have demonstrated the feasibility of generating synthetic HS fragments of defined structure with biological activity against a number of angiogenic cytokines.
Angiogenesis, the formation of new blood vessels, has been validated as a target in the oncology clinic in numerous Phase III clinical trials that have compared conventional therapy with the same regimen supplemented with a vascular endothelial growth factor (VEGF) pathway inhibitor; the prototypic class of anti-angiogenic agents . Phase III clinical trials in colorectal , ovarian , renal , neuroendocrine  and breast cancer  have shown that the addition of such drugs improves the progression free interval, the time from the start of the study until disease progression and, in some cases, overall survival.
The clinical benefit from anti-angiogenic agents, although initially very promising in the laboratory, has been somewhat less impressive in the clinic, most probably because of heterogeneity in tumours and therefore in the resultant vasculature. Evidence to support this has arisen from clinical studies of the VEGF receptor tyrosine kinase inhibitor, cediranib, when used to treat patients with glioma . During periods of drug-holiday or progressive disease, the plasma concentration of fibroblast growth factor 2 (FGF2) increased, implying that this cytokine-mediated escape from the control is afforded by VEGF inhibition. Additional evidence for heterogeneity arises from studies in colorectal  and ovarian cancer , which have shown that drugs that target the angiopoietin–Tie2 axis also improve disease control following previous exposure to VEGF inhibitors.
Redundancy in the regulation of angiogenesis is not surprising but it highlights the need for new strategies for treatment in which multiple angiogenic pathways can be targeted simultaneously. The mandatory role that heparan sulfate (HS) plays in regulating angiogenic growth factor activity offers such a therapeutic opportunity.
Heparan sulfate (HS) is a linear glycosaminoglycan that consists of alternate N-substituted glucosamine residues and a hexuronic acid residue that is either L-iduronate or D-glucuronate. These chains of saccharides, which are covalently linked to core proteins that are embedded in the cell membrane or extracellular matrix to form HS proteoglycans (HSPGs), are variably sulfated to generate domains of high and low anionic charge density that engage cytokines largely through ionic interactions with protein ligands. The sulfated domains are generated by the post-translational modification of core proteins that occurs by the enzymatically-mediated sequential modification of the heparan backbone through N-sulfation, epimerization to generate iduronate and then specific 2-O-, 6-O- or N-sulfation of hexuronic and glucosamine residues .
The HS–cytokine axis
Many important angiogenic cytokines are critically dependent on HS for the assembly of signalling complexes and thereby their biological activity. Examples include the FGFs, most VEGFs, interleukin (IL)-6, IL-8, hepatocyte growth factor (HGF) and stromal-cell-derived factor 1α (SDF-1α) . The significance of angiogenic growth factor binding to HS has been demonstrated by mutating HS-binding sites in the HGF isoform neurokinin 1 (NK1) and VEGF. The impact was to disrupt HS-cytokine complex formation, converting NK1 and VEGF into potent competitive antagonists of signalling and inhibitors of angiogenesis . Taken in conjunction with older data that demonstrated the importance of HS in FGF2 activity [12,13], these data highlight the pivotal role that HS plays in regulating the activity of angiogenic growth factors.
The polysaccharide backbone structure of HS has been known for several decades and early studies involving HS-anti-thrombin III structure–function relationships described a specific pentasaccharide that engages the protein  leading to the assumption that there would be specific structure–function relationships with other protein ligands. In the 1980s and 1990s, much subsequent work focused on the identification of specific binding sequences for protein ligands, but these proved elusive. Nevertheless, for some proteins critical structural properties were discovered. For instance, in the case of FGF2, the interaction between HS and FGF2 was reported to be dependent on N-sulfate moieties in glucosamine and 2-O-sulfate groups in iduronate ; the content of the latter also being an important determinant of cytokine binding. However, a critical observation at the time was that the HS–FGF2 interaction was not the key determinant of biological activity. Rather, this required the presence of 6-O-sulfate moieties on N-sulfated glucosamine, so that interactions could occur between HS and FGF receptor 1 (FGFR1), leading to formation of the trimolecular signal-transducing complex . On the other hand, FGF1-induced activity was dependent on fine-tuning of 6-O-sulfation levels, but the degree of 6-O-sulfation that was required for activation of FGF1 and FGF2 was different, highlighting differential 6-O-sulfation requirements for FGF1 and FGF2 .
The precise pattern of HS sulfation that determines the strength and biological significance of HS interactions with other growth factors and chemokines is less clearly defined. In particular, there are clear differences between oligosaccharide structures that support growth factor activity in HS-denuded cells or inhibit it in the context of wild-type HS. VEGF165 binds to oligosaccharides containing at least eight residues  but requires quite long species of saccharides effectively providing two sulfated domains  to support growth factor activation . On the other hand, we demonstrated that oligosaccharides composed of 12 monosaccharides that were homogeneously sulfated at the 2-O-position of iduronate and the N- and 6-O-positions in glucosamine were poor inhibitors of VEGF165-induced phenotypes, whereas homogeneous sulfation at the 2-O- and N-positions inhibited VEGF165-associated biological functions . Binding experiments between HS and SDF-1α, showed that cytokine interacts with heparin through 2-O- and N-sulfate residues. Removing 6-O-sulfate from heparin also affected the interaction between SDF-1α and heparin, although to a lesser degree . The significance of 6-O-sulfation for interaction between HS and IL-8 was demonstrated in genetically modified mice that lacked 2-O-sulfation, which resulted in elevated N- and 6-O-sulfate levels and enhanced neutrophil recruitment in models of acute inflammation . As with VEGF165, available evidence points to the requirement for two sulfated domains in HS for engagement with IL-8 . Finally, further evidence to support the importance of 6-O-sulfation arises from the existence of unique 6-O-endosulfatases that remove this critical moiety from extracellular HS . Together, these data show that 6-O-sulfation levels play an important role in regulating biological activity of FGF2, VEGF165, SDF-1 and IL-8, all of which have been implicated in the regulation of ovarian cancer angiogenesis.
Together, these biological relationships identified a potential strategy for the development of anti-angiogenic oligosaccharides. In parallel, they also have important implications for our patients who are treated with fractionated low-molecular-mass heparin, a highly sulfated analogue of HS. Since these clinical preparations contain mixtures of differentially sulfated short and long molecules, they may also have the potential to support signalling complexes consisting of HS-cytokine-receptor complexes although of course, differentially sulfated HS fragments of different lengths may be required to inhibit specific angiogenic cytokines.
HS in cancer
HS has been implicated in several malignancy-associated phenotypes . We showed previously that transformation, in vitro, is associated with increased 6-O-sulfation, particularly in the sulfated domains of HS  and that this was associated with increased activity through the FGF2 signalling axis ; a cytokine that increased proliferation and migration in vitro . Knowing that HS was an important mediator of growth factor activity, we sought to explain its role in human ovarian cancer. Having shown that the HSPG syndecan 1 was of prognostic significance in the disease and that N-syndecan was aberrantly expressed in the vasculature , we used a receptor construct to show that ovarian cancer vasculature expressed HS that had the capacity to assemble the FGF2–FGFR1 signalling complex, identifying vascular HS as a potential target for anti-angiogenic therapy in the disease . In keeping with these data, in situ studies demonstrated that the mRNA for the 6-O-sulfotransferase 1 was present in the ovarian tumour vasculature but that 6-O-endosulfatase 1 was barely detectable in the tumour vasculature . Indeed, reduced expression of 6-O-sulfotransferases 1 and 2 in endothelial  or tumour  cells suppressed angiogenesis, which in the case of ovarian cancer cells was critically dependent on a heparin-binding epidermal growth factor (HB-EGF)-mediated effect.
Previous crystallography data had suggested that there was a critical requirement for HS oligosaccharides of sufficient length to support the formation of trimolecular signalling complexes involving FGF2 and FGFR [34,35]. We therefore sought to define the optimum length of heparin oligosaccharides that would inhibit angiogenesis in vitro and in vivo. Using models of FGF2-induced angiogenesis that were based on cytokine-impregnated sponges or cell lines engineered to overexpress FGF2, we demonstrated that heparin octa- and deca-saccharides inhibited angiogenesis in vivo . Additional studies of chemically de-6-O-sulfated heparin also demonstrated anti-angiogenic activity in vivo. Nevertheless, despite clear reductions in angiogenesis in vivo, our studies did not show tumour growth restraint, suggesting that combination therapy with cytotoxic agents or other anti-angiogenic drugs are needed to control the process.
Other studies have confirmed the importance of HS sulfation in models of tumour growth in vivo with associated angiogenesis . Lung xenograft tumours generated in mice that were unable to sulfate N-glucosamine showed reduced tumour growth due to impaired angiogenesis . Decreased 6-O-sulfation in human endothelial cells significantly affected neo-angiogenesis in Matrigel plugs implanted into mice . Thus these studies provide compelling data to support the development of drugs that inhibit ovarian cancer endothelial HS activity.
Chemical synthesis and evaluation of HS oligosaccharides
The above experiments  were completed through size-exclusion chromatographic separation of heparin oligosaccharides from commercially available low-molecular-mass heparin. The generation of sufficient material for experimental purposes was laborious and we therefore elucidated one of the first synthetic schemes that allowed the generation of multi-gram amounts of site-specifically sulfated HS oligosaccharides. Having developed the first inexpensive synthetic route to the generation of L-iduronate , we employed this to provide scalable access to disaccharides, tetrasaccharides and by further polymerization into oligosaccharides containing up to 12 saccharide residues [39–41]. Through the use of differential protecting groups, we then generated species of length-defined oligosaccharides that were homogeneously 6-O-sulfated, in the first instance. Comparing species that contained one 2-O-sulfate per disaccharide with others that bore 2-O-sulfated iduronate and N-sulfated glucosamine; the data showed that the latter species were a more effective inhibitor of endothelial-cells proliferation and migration than the former species .
We also examined HS oligosaccharides containing three sulfate moieties per disaccharide (2-O-, 6-O- and N-sulfate)  but showed that this family of oligosaccharides did not demonstrate superior inhibitory potential against FGF2-induced endothelial proliferation or migration in vitro. Thus these data corroborated the importance of 6-O-sulfate in HS, discussed above. We therefore investigated this relationship in vivo using these homogeneous species. The results in several models demonstrated that 2-O- and N-sulfated HS 12-mers were able to inhibit vascular FGFR signalling and were associated with statistically significant reductions in tumour microvascular density in vivo (Figure 1). However, tumour growth continued in these models, suggesting that enhanced anti-angiogenic chemical modifications are required or that combination treatment with cytotoxic or other anti-angiogenic agents is needed. Having established the 2-O- and N-sulfated 12-mer as the optimum backbone for further modification, we examined the effect of adding a single 6-O-sulfate moiety at the reducing end of the molecule. This imparted some specificity to the biological effects of the resultant oligosaccharide, which differentially affected chemokine activity. Finally, through radiolabelling of the molecules, we established their intra-tumoral pharmacokinetics and demonstrated that sufficient concentrations were achieved in tumours in vivo (Figure 2).
De-6S dodecasaccharide inhibits ovarian tumour angiogenesis
Intra-tumoral concentrations of 2-O- and N-sulfated 12-mer
The position of polysaccharide anti-angiogenic agents
Our data have shown that HS is critical for the biological activity of several angiogenic cytokines and that 6-O-sulfate moieties are of particular biological importance. Because of this, we developed the complete chemical synthesis of site-specifically sulfated heparin oligosaccharides, which when evaluated in vivo, confirmed the findings of our earlier study , which showed that such oligosaccharides can be delivered subcutaneously and safely in vivo. Systemically administered HS oligosaccharides were well-tolerated and were associated with reduced signalling and micro-vessel density in vivo, but tumour growth was maintained.
So where do these studies leave HS oligosaccharides as anti-angiogenic agents? We know that the molecules are safe, tolerable and achieve adequate concentrations in vivo. First and importantly, we have been able to omit the anti-thrombin III-binding sequence that characterizes heparin from our synthetic oligosaccharides, thereby removing the principal dose-limiting toxicity of heparin anti-coagulation. Thus pre-clinically, there was no maximum tolerated dose of the oligosaccharides. Secondly, other synthetic oligosaccharides (e.g. PI-88 ) have been associated with immune thrombocytopenia in the clinic, presumably because the oligosaccharide is recognized as foreign. We are making HS oligosaccharides that are essentially chemically pure wild-type structures and which are therefore poorly immunogenic.
On the other hand, both sets of our in vivo studies have shown that although we can inhibit angiogenic cytokine signalling and microvessel density, tumour growth is not affected by oligosaccharide administration. To some extent that is not surprising since experimental models of tumour growth in vivo are dominated by the hypoxic production of VEGF, including HS-independent VEGF121. Human cancer is not as reliant on VEGF, explaining the disparity between experimental and clinical observations of VEGF inhibitors. Nevertheless, although we can anticipate that synthetic oligosaccharides can be safely administered subcutaneously in saline to patients with advanced cancer, it is possible that the true benefit of this modality of treatment will only emerge once medicinal chemistry has been applied to the oligosaccharides or when these oligosaccharides are administered with other treatments such as cytotoxic chemotherapy or VEGF pathway inhibitors.
Angiogenesis and Vascular Remodelling: New Perspectives: A Biochemical Society Focused Meeting held at University of Chester, U.K., 14–16 July 2014. Organized and Edited by Roy Bicknell (Birmingham University Medical School, U.K.), Michael Cross (University of Liverpool, U.K.), Stuart Egginton (University of Leeds, U.K.), Victoria Heath (University of Birmingham, U.K.) and Ian Zachary (University College London, U.K.).
This work was funded by Cancer Research UK [grant number C2075/A9106] and the Medical Research Council [grant numbers G0601746 and G902173].