Targeting Heparanase Through Multi-Modal Inhibitor Design and Ancestral Reconstruction

Abstract

Heparanase (HPSE) is the only known mammalian enzyme that catalyses the degradation of heparan sulfate (HS), a major component of the extracellular matrix and represents a validated therapeutic target in oncology and inflammatory disease. Despite its clinical significance, the development of selective inhibitors has been hindered by three factors. Firstly, glycosidase families have demonstrated a high degree of structural conservation, particularly within their catalytic sites. Secondly, HS mimetics have been shown to possess inherent off-target anticoagulant toxicities. Thirdly, there have been challenges in producing stable, native-like recombinant enzymes for mechanistic characterisation. This thesis addresses these challenges through four distinct but complementary research chapters. Firstly, we introduced a computational-to-experimental pipeline for allosteric peptide inhibitor discovery (see Chapter 2). We identified Peptide 5 as a blueprint for a next-generation inhibitor that functions through a reversible covalent bond at the non-catalytic Cys179 exosite on HPSE, inducing conformational changes that inactivate the enzyme's catalytic activity. In parallel, the characterisation of tyrosine-containing library variants demonstrated that mimicking HS electrostatics through post-synthetic tyrosine sulfation affects potency of binding to HPSE. These findings demonstrate that computational peptide binder design targeting HPSE requires the integration of reactive chemical features, specifically cysteine-targeting moieties or electrostatic mimicry, to achieve biochemical activity. Addressing the well-known limitation of recombinant protein production, the methodology of ancestral sequence reconstruction (ASR) was applied to generate a human-primate-node HPSE ancestor (see Chapter 3). Unlike the previously designed P6 HPSE variant, solubilised through artificial surface change modification, the ASR variant produces robust, chaperone-free solubility while preserving the native electrostatic topology with higher sequence similarity compared to wild-type (WT) HPSE. Biophysical characterisation confirms this structural integrity, revealing a canonical a-helical TIM-barrel fold with a robust melting point (Tm) of 61.49 oC. This establishes a structurally similar, physiologically relevant platform for HPSE structural biology and inhibitor discovery. In Chapter 4, we expanded the design scope to miniprotein scaffolds with the aim of developing novel miniprotein inhibitors of HPSE. To achieve this, we conducted a comparative analysis of two deep learning architectures, namely BindCraft and RFdiffusion, that were targeted towards heparin-binding domains (HBDs), which are critical for substrate recognition. One of the candidates neutralised the electropositive HBD surface potential, excluding extended HS chains through electrostatic remodelling rather than direct active-site obstruction. Finally, we employed fragment-based drug design (Chapter 5) to explore novel chemical space, identifying novel small-molecule scaffolds that inhibit HPSE in the micromolar range. Overall, this thesis advances the knowledge of HPSE inhibition by providing a validated solution to the challenges of recombinant expression and establishing a broad spectrum of inhibitory modalities, ranging from covalent allostery to electrostatic neutralisation, that offers a blueprint for next-generation HPSE therapeutics. Show more