òòò½Íø

We build computational models that contribute to our understanding of the mechanisms of disease and host immune responses, with a particular focus on infective pathogens. Our main method of choice is molecular simulation, using multiscale approaches that allow us to access a broad range of spatio-temporal scales, from atomically-detailed systems to highly simplified coarse-grained resolutions. We interact closely with our numerous experimental collaborators, and develop new ways to integrate diverse structural, biophysical, and biochemical data into models that enable us to work towards novel therapeutic strategies.

FLAVIVIRUSES: TOWARDS VACCINES & INHIBITORS

Flaviviruses are enveloped viruses which encompass numerous human pathogens of escalating global health concern that are predominantly transmitted by mosquitoes and ticks. Traditionally, we have focused on dengue virus, which represents a significant burden to human health and the economy in Singapore. We have developed “virtual flavivirus” models that have been used to elucidate key viral dynamics associated with infection and antibody recognition [1]. As part of a collaborative team across Duke-NUS and A*STAR funded by an NRF CRP grant, we are pursuing mRNA-based vaccine development to target all four dengue serotypes. In parallel, we are exploring other vaccine technologies, including engineering of dengue virus-like particles (VLPs) [Patent No. 10202303148T; 6th November 2023] supported by a YIRG grant awarded to Jan K. Marzinek (Figure 1A). Nevertheless, dengue exhibits a phenomenon known as antibody-dependent enhancement which can exacerbate infections and complicates traditional vaccine-based approaches. As an alternative approach, the pH-dependent conformational changes of the envelope protein required for fusion are attractive points for inhibition. To this end, we have developed a computational platform to probe membrane proteins for hidden druggable pockets, which has been used to uncover a novel cryptic site in flaviviruses associated with pH-sensing and is now being pursued for potential antivirals [2].

Figure 1. Viral therapeutics. (A) Predicting the maturation process of dengue VLPs, towards improved secretion properties for use in vaccines. (B) SARS-CoV-2 spike protein modelling for (left) mapping receptor binding domain (RBD) epitopes of novel antibodies, and (right) predicting allosteric motions induced by full-length IgG antibody.

CORONAVIRUSES: THERAPEUTIC STRATEGIES AND FUTURE PREPAREDNESS

We are leveraging the methods we have developed for flaviviruses to other enveloped viruses including SARS‑CoV‑2. Supported by an ID HTPO Seed Fund grant, we have characterized the structure and dynamics of the viral surface spike protein for a wide range of SARS-CoV-2 variants. We computationally mapped its glycan coating as part of a global collaboration [3], showing that constructs derived from different labs are highly similar and closely resemble their viral counterpart, helping to rationalize the success of spike-based vaccines. A multidisciplinary approach has also revealed that the spike protein plays previously undefined pathogenic roles in serious COVID-19 infections [4, 5] including augmenting hyperinflammation by acting as a conduit to over-activate host immune pathways. Our findings highlight potential therapeutic strategies for future viral infections that cause over-amplified immune reactions and sepsis. Furthermore, we have discovered the importance of long-range allostery [1] in governing spike function [6]. We showed with collaborators at NUS that antibodies may exploit such allostery to tune neutralizing efficacy, while also exhibiting synergistic binding behavior which may offer potential routes to antiviral “cocktail” development (Figure 1B). Finally, we have leveraged our platform for probing membrane protein pockets [2] to uncover potentially druggable sites on the spike protein [8]; looking towards future epidemic preparedness, we are aiming to develop fusion-inhibitory therapeutics by targeting conserved sites across different coronaviruses, while also bearing in mind the potential for such viruses to evolve alternative mechanisms of infection [9].

TACKLING ANTIMICROBIAL RESISTANCE

Antimicrobial resistance (AMR) is a leading cause of death worldwide, and there is an urgent need to discover new ways to battle AMR infections. For example, as part of a collaboration with P&G, we are establishing computational approaches to predict the modes of action of disinfectants against viral capsids and are working with collaborators at the Universities of Southampton and Oxford to better understand the impact of hand sanitizer ingredients upon pathogen envelopes [10]. A multiscale simulation approach can provide an improved understanding of bacterial membrane structure and function [11-14] and contribute towards the development of novel strategies to battle AMR [15]. For example, in collaboration with NTU, we have helped to clarify the mechanisms of action of drugs that inhibit the essential F-ATP synthase enzyme in mycobacteria, which cause diseases such as tuberculosis, and are leveraging this to target other AMR species [16-18]. We are also interested in developing strategies to target the excessive activation of inflammatory pathways that can occur during systemic infections by AMR bacteria. In a long-running collaboration with researchers at Lund University, a cyclic molecule has been developed based on the evolutionarily conserved innate structural fold of endogenous thrombin-derived C-terminal peptides (TCPs) which targets both bacterial toxins and host immune receptors [19] (Figure 2A). We have also helped to show how such TCPs may be incorporated into sutures to reduce the incidence of surgical site infections [20] (Figure 2B).

Figure 2.: Exploiting host defence peptides for therapeutics. (A) A hydrophobic staple was added to TCP to develop a dual-action drug that both neutralizes bacterial lipopolysaccharide (LPS) and inhibits the host CD14/TLR4 pathway to block inflammatory reactions. (B) Multiscale models of Vicryl suture polymers were used to predict the stable incorporation of TCPs, a strategy which was then demonstrated to successfully prevent surgical site infections.

OTHER KEY DEVELOPMENTS

We have a broad interest in other challenges within biophysics [21], bionanotechnology [22], and biomedicine [23]. For example, as part of an NMRC OF-IRG grant with Duke-NUS and NTU, we are focused on overcoming drug resistance in cancer due to ABCB1 transporter overexpression, and recently reported a novel cyclic tryptophan-based molecule which inhibits ABCB1 with an efficacy better than classical inhibitors [23]. We also contribute important molecular simulation methodological developments, such as new approaches to calculate protein glass transition temperatures which correlate with their biochemical activities and may be modulated by bioprotective compounds [24], and coarse-grained forcefields for describing the glycosaminoglycans (GAGs) in proteoglycans which play vital roles in numerous physiological and pathological processes, particularly governing cellular communication and cell surface attachment [25] (Figure 3).

Figure 3. Multiscale modelling of GAGs. Physiologically realistic mammalian cell surfaces may be modelled by simulating GAGs (red) on membrane-bound proteoglycans and receptors.