abstract: amma annual medal lecture

Microsoft Word - MM2019_booklet.docxAbstract: AMMA Annual Medal Lecture
MM2019 Conference Programme December 5th – 8th 2019 Bintan Lagoon Resort Indonesia
General Information
1
Welcome! It is our great pleasure to welcome you to the MM2019 Conference on Bintan island. This is the latest in the series of “MM” conferences, the premier conferences of the Association of Molecular Modellers of Australasia (AMMA). This year will also feature the award lecture of the 2019 winner of the AMMA medal, Prof. Amanda Barnard.
This booklet contains the programme, maps for the conference venue and conference dinner locations, abstracts for all talks and posters as well as a participants list. We would like to take this opportunity to thank all speakers, presenters, sponsors who made this event possible. We are looking forward to four days of exciting science!
Please note that photographs may be taken during the event. These photographs may be used by AMM and the Bioinformatics Institute (BII) A*STAR for publicity and other related purposes.
Should you have any questions, suggestions, problems, comments or concerns please feel free to approach any of the conference organisers or helpers:
Local Organizing Committee: Peter J Bond & Chandra Verma, Bioinformatics Institute (BII) A*STAR ([email protected]) Organizing Committee: David Winkler, Monash, La Trobe, Nottingham universities, CSIRO Bret Church, Faculty of Medicine and Health, University of Sydney Ricardo Mancera, Curtin University Evelyne Deplazes, University of Technology Sydney & Curtin University Brian Smith, La Trobe University Helpers, BII (A*STAR): Roland G. Huber, Nguyen Thanh Binh, Aishwary Shivgan, Jan K. Marzinek, Sonia Nicolaou, Pietro Aronica, Ashar J. Malik, Alister Boags, Lorena Zuzic, Shruti Khare
General Information
General Information Practicalities
• Hotel check-in on the first day will be run concurrently with conference registration, in the Fairways Room.
• Hotel room check-out on the final day will be at 11am. Guests may leave luggage at the hotel reception.
• All delegates taking the 5:10pm ferry back to Singapore on the final day should gather at the hotel reception at 4pm for timely transfer to the ferry.
• The temperature in Bintan is warm (25-30°C) but we are just entering the monsoon season, so be prepared for unpredictable weather throughout the day.
• Because of the weather, the ferry ride could be rough, so you may wish to bring motion sickness tablets (but note that the journey time is short).
• The dress-code for the conference is casual. But please bear in mind that despite the warm weather, there may be strong air-conditioning inside the lecture halls.
• You may wish to bring: a good-quality insect repellent to protect against mosquitoes and sand flies; sun-safe and light-weight clothing, hat, sunglasses, sunscreen, etc.; beach-friendly / swimming gear; comfortable shoes / sandals.
Lectures • All lecture sessions, and the AMMA Annual General Meeting (AGM) on Friday
afternoon, will be held in the Fairways room. • Speakers should arrive 10 minutes before their session starts, to check the equipment
works and/or to upload presentations. Speakers may use their own laptop if they prefer, but should bring an adapter if required.
• Plenary Lectures (PLs), Keynote Lectures (KLs), Contributed Talks (CTs), and Short Talks (STs) will be allotted a total presentation time (including Q&A) of 50 minutes, 30 minutes, 20 minutes, and 10 minutes, respectively.
• Chairs will give PLs, KLs, CTs, and STs a warning at 10 minutes, 5 minutes, 5 minutes, and 2 minutes, respectively, before the end of their allotted time.
• The Manado Room will be available throughout the conference for delegates to use as a room for private work / break-out discussions.
Posters • Poster presenters should affix their posters in the Sulawesi room, on the first day.
Posters may be removed on the last day. • Poster viewing will be possible during afternoon tea breaks on Friday and Saturday,
and during the Friday evening poster session, in the Sulawesi room. • Posters will be judged for prizes during the Friday evening poster session.
Refreshments • Breakfast is served each day from 6.30am in the Fiesta restaurant. • Lunch will be served each day in the Fiesta restaurant, except on Saturday when a
special brunch will be served with live jazz in the Nelayan restaurant. • Dinner on Thursday and Saturday will be served at the Nelayan Upper Deck. • A drinks reception and conference BBQ dinner will take place on Friday on the beach. • Morning tea breaks will be held in the Fairways room, whereas afternoon tea breaks
with poster viewing will take place in the Sulawesi room.
General Information
4
AMMA Annual Medal Lecture: Amanda Barnard CSIRO Data61 "Digital but Different - Combining Computational Modeling and Machine Learning" Amanda S. Barnard, Ben Motevalli, Amanda J. Parker, J. Meli Fisher, Chris A. Feigl and George Opletal A fundamental aim of nanomaterials research is to identify features of materials that can be tuned to control how the nanomaterial performs under specific application conditions. The combination of computational chemistry, computational materials science with machine learning provides a powerful way of relating structural features with functional properties, but combining these fundamentally different scientific approaches is not as straightforward as it seems. Machine learning methods were developed for large data sets with small numbers of consistent features. Typically materials data sets are small, with high dimensionality and high variance in the feature space, and suffer from numerous destructive biases. None of the established data science or machine learning methods in widespread use today were devised with materials data sets in mind, but there are ways to overcome these issues and use them reliably. In this presentation we will discuss the impact of domain- specific constraints on data-driven materials design, and explore the differences between materials simulation and materials informatics that can be leveraged for greater impact.
Abstracts: Plenary Lectures
5
PL-1: Gerhard Hummer Max Planck Institute of Biophysics "What's wrong with diffusion?" Gerhard Hummer, Soeren von Buelow, Marc Siggel, Martin Voegele, Max Linke, Juergen Koefinger Molecular dynamics simulations of the diffusion of proteins and other macromolecules in dense solutions and in lipid membranes revealed unexpected complexities. In systems mimicking the interior of a living cell, with densely packed proteins, the translational and rotational diffusion of proteins slow down dramatically at high protein concentrations, and the Stokes-Einstein relation appears to break down. In membranes, the apparent diffusion coefficient appears to grow without bound as the box size is increased. We resolve these issues by showing, first, that transient clustering of proteins explains quantitatively the increase in the apparent Stokes radius and the rise in the viscosity. Second, we show that the divergence of membrane diffusion is the result of the unusual hydrodynamics under periodic boundary conditions. A hydrodynamic correction not only explains the divergence, but also produces proper diffusion coefficients and difficult-to-calculate membrane viscosities. In essence, membrane simulations provide a molecular realization of the Stokes Paradox. Overall, hydrodynamic theory explains why diffusion in concentrated solutions is hindered dramatically even by weak and transient protein-protein interactions and why diffusion in membranes appears to break down entirely. Accounting for hydrodynamics, we obtain diffusion coefficients that can be interpreted meaningfully and compared to experiment.
6
PL-2. Tim Clark FAU Erlangen-Nürnberg "Charge transport in materials for solar cells"
Very large scale semiempirical MO calculations provide wavefunctions for aggregates (up to 100,000 atoms) and periodic systems (up to 50,000 atoms in the repeat unit) relevant to organic and hybrid solar cells. Semiempirical configuration-interaction calculations (up to 5,000 atoms) are also used to characterize excited states involved in singlet fission. Charge transport itself can be simulated using propagation of the electron density in imaginary time.
PL-3. Lars Nordenskiöld Nanyang Technological University - Singapore "Multiscale Modelling of DNA and Nucleosome Phase Separation: From Atomistic to Mesoscale Level by Inverse Monte Carlo" Lars Nordenskiöld, Tiedong Sun, Alexander Mirzoev, Vishal Minhas, Nikolay Korlev and Alexander Lyubartsev
DNA condensation and phase separation is of utmost importance for DNA packing in vivo with important applications in chromosome organization, medicine, biotechnology and polymer physics. E.g., the presence of hexagonally ordered DNA is observed in virus capsids, sperm heads and in dinoflagellates. Rigorous modelling of these processes in all-atom MD simulations is presently difficult to achieve due to size and time scale limitations. We used a bottom-up hierarchical approach for systematic coarse-graining following the inverse Monte Carlo (IMC) approach to extract solvent mediated effective CG potentials for all interactions in the system from structural properties of the underlying system. Following validation, we modeled DNA and nucleosome phase separation at mesoscale level, induced by multivalent cations. Solvent-mediated effective potentials for a CG model of DNA were extracted from underlying all-atom MD simulations. Simulations of hundreds of 100-bp-long CG DNA oligonucleotides in the presence of explicit cobalt(III)-hexammine ions demonstrated aggregation to a liquid crystalline hexagonally ordered phase. Proceeding to further coarse- graining and extraction of effective potentials, we conducted modelling at mesoscale level. In agreement with electron microscopy observations, simulations of a 10 kbp-long DNA molecule showed phase separation to either a toroid or a fibre with distinct hexagonal DNA packing. The approach used here is based only on the underlying all-atom force field and uses no adjustable parameters and is being generalized to modelling chromatin condensation. We developed a CG model for the nucleosome core particle (NCP) and extracted the effective potentials on the basis of underlying all-atom MD simulations and simulated NCP phase separation in the presence of multivalent cations.
PL-4. Jim Warwicker University of Manchester "Molecular modelling for protein therapeutics: shape and charge and other old friends"
Jim Warwicker and Max Hebditch Developability is the somewhat unwieldy term describing the field of ensuring that protein therapeutics are able to be processed, stored and delivered as desired and at sufficiently high concentration. It incorporates many aspects of standard protein chemistry, for example limiting chemical modification and protein aggregation. Many groups are developing models in this area, with a focus on aggregation and other biophysical properties. Our own work will be discussed, based around tools available at the www.protein-sol.manchester.ac.uk site. These are amino acid sequence based and structure based methods, although in either case non-polar content and charge feature amongst properties that correlate with experiment. Alongside these perhaps obvious features, we have come across a few interesting sidelines, for example that lysine and arginine are quite different in regards to protein solubility, and that a simple surface tension parameterised hydrophobic effect can fail badly. Recent work describing machine learning models for the biophysical behaviour of antibodies will also be presented.
PL-5. Rebecca Wade Heidelberg Institute for Theoretical Studies (HITS) and Heidelberg University "Computational Approaches to Protein Dynamics and Binding Kinetics for Drug Discovery"
The dynamic nature of protein structures and the diversity of protein binding pocket dynamics provide challenges and opportunities for ligand design [1]. We have developed TRAPP, a toolbox of computational approaches to identify TRAnsient Pockets in Proteins for ligand design. I will present recent developments in TRAPP to identify pocket conformations with high druggability. Protein binding site flexibility is one of the factors that can affect drug- target binding kinetics. Growing evidence that the efficacy of a drug can be correlated to target binding kinetics has led to the development of many new methods aimed at computing rate constants for receptor-ligand binding processes [2], see also: kbbox.h-its.org. Here, I will describe our studies to explore the determinants of structure-kinetic relationships and to develop computationally efficient methods to estimate drug-target binding kinetic parameters. I will introduce our -random acceleration molecular dynamics simulation (RAMD) method to compute relative residence times [3] and discuss how machine learning analysis of RAMD trajectories [4] and the application of Comparative Binding Energy (COMBINE) Analysis [5] can be used to decipher the determinants of drug-target residence times. [1] Stank A, Kokh DB, Fuller JC, Wade RC. Protein binding pocket dynamics. Acc. Chem Res., 2016, 49:809-815. [2] Bruce NJ, Ganotra GK, Kokh DB, Sadiq SK, Wade RC. New approaches for computing ligand- receptor binding kinetics. Curr Opin Struct Biol. 2018, 49: 1-10. [3] Kokh DB, Amaral M,……Wade RC. Estimation of drug-target residence times by -random acceleration molecular dynamics simulations, J. Chem. Theory Comput. 2018, 14: 3859–3869. [4] Kokh DB, Kaufmann T, Kister B, Wade RC. Machine Learning Analysis of RAMD Trajectories to Decipher Molecular Determinants of Drug-Target Residence Times, Frontiers. Mol. Biosci. 2019, 6: 36. [5] Ganotra GK, Wade RC. Prediction of Drug–Target Binding Kinetics by Comparative Binding Energy Analysis. ACS Medicinal Chemistry Letters 2018, 9: 1134–1139.
Abstracts: Keynote Lectures
KL-1. David Winkler Monash, La Trobe, Nottingham Universities, CSIRO "Machine learning at the (nano)materials-biology interface"
David A. Winkler, R.M.T. Madiona, N. Welch, B.W. Muir, P.J. Pigram, Mikulskis, M. Alexander, T.C. Le, M. Penna, I. Yarovsky, S.R. Ghaemi, B. Delalat, A.L. Hook, S. Gronthos, N.H. Voelcker The past decade in particular has seen a spectacular rise in research into materials, with biomaterials in particular playing a dominant role in regenerative medicine. This has, in part, been driven by improvements in instrumentation and rapid rise in high throughput synthesis and characterization methods that now generate orders of magnitude more data than in previous eras. It is essential to extract as much information as possible on the molecular details of materials’ interaction with biological systems to use this for rational design of materials that generate desirable biological outcomes while minimizing undesirable side effects. We show how a concomitant rise in computational and algorithmic research methods is allowing complex, rich data sets to be efficiently analyzed and used to design new materials. New artificial intelligence and machine learning methods, coupled with improved ways of representing the molecular and physicochemical properties of materials to train such models and effective sparse feature selection methods, have addressed the critical need for effective modeling tools. As in many other areas of science and technology, deep learning methods are playing increasingly important roles in providing. Models that make robust, accurate predictions of biological properties for increasingly diverse classes of materials Here we illustrate how these powerful but accessible modeling methods have been used to model a variety of material-biology systems. We focus on the development of immune- instructive materials, materials that support growth and proliferation of stem cells, polymers that do not support attachment of microbial pathogens or proteins, and materials that minimize foreign body responses to implantable medical devices.
11
KL-2. Lee Hwee Kuan Bioinformatics Institute (BII) A*STAR "Solving partial differential equations using Neural Networks"
Hwee Kuan Lee, Kenta Shiina, Sojeong Park Computer simulations are used widely to perform predictions on physical or biophysical processes. Most of these computer simulations are essentially solving partial differential equations that models the physical process that we wish to predict. However numerical integration is very computationally intensive leading to long simulation times. We propose an alternative way to speed up the search for solutions of partial differential equations using neural networks. As examples, solving of several systems of partial differential equations will be presented.
Indian Institute of Science Education and Research (IISER) Pune "Integrative modeling of the 3D structure of intermediate filaments"
Neelesh Soni and M.S. Madhusudhan Intermediate filaments along with micro tubules and micro filaments form the cytoskeletal structure in cells. Unlike the other two components the molecular structure of the intermediate filament has not been determined at residue-level resolution. In this study, we determined the structure of the Keratin K5-K14 Intermediate filament using integrative modeling. Our model utilises data from experiments such as chemical cross linking, X-ray crystallography and SAXS to resolve the structure of large parts of the filamentous assembly. Our structure of the Keratin filaments helps identify the molecular mechanisms behind hundreds of point mutants that disrupt the filament and hence lead to disease. Our model also suggest how filament bundles are formed and these findings are consistent with experimental observations that were not used in model construction. Further, we have been able to suggest and experimentally validate new mutations that lead to filament disruption. We believe that this first 3D model of the full length intermediate filament would now help study other intermediate filaments. It would also be useful in designing therapeutic strategies against disease conditions that stem from filament instability.
13
KL-4. David Ascher Baker Institute and Bio21 Institute "Unraveling the molecular mechanisms behind mutations and their link to phenotypes"
The wealth of experimental, structural, genomic and proteomic information accumulated over the years has opened up opportunities to use advanced computational approaches to better understand how mutations will affect protein structure and function. These methods fall into several major classes based on whether they use protein sequence or structural information, and whether they rely on statistical, machine learning or dynamics approaches. Each approach comes with distinct advantages and limitations; however we have shown that they are often complementary, and combining distinct approaches can lead to significantly improved performance. One problem, especially with machine learning approaches, has been the unbalanced nature of the curated data- proteins have highly refined structures and functions, and alterations to it are more likely to be disruptive. Many of the developed methods therefore do not perform as well at identifying stabilising mutations. A number of strategies have been used, with varying success, to address this, including the use of reverse hypothetical mutations. By refining how the reverse mutations are implicated, we have developed more balanced and predictive tools. One of the few approaches that has been applied to rapidly characterise a wide spectrum of molecular consequences is the mCSM platform, which uses graph-based signatures to represent the wild-type environment of a residue in order to predict pathogenicity and the effects of a mutation on protein stability and affinity for protein partners, nucleic acids, metal ions and small molecules, including drugs and ligands. Analysis of their performance reveals they perform equally well on homology models. By integrating these approaches, we can understand the molecular consequences associated with mutations. This has been successfully used to guide protein engineering, more accurately identify diseases and predict patient outcomes, and guide the development of better drugs.
KL-5. Jane Allison University of Auckland "CherryPicker: Automated parameterisation of large biomolecules for molecular simulation"
Ivan Welsh, Jane Allison Molecular simulations allow investigation of the structure, dynamics and thermodynamics of molecules at an atomic level of detail, and as such, are becoming increasingly important across many areas of science. As the range of applications increases, so does the variety of molecules. Simulation of a new type of molecule requires generation of parameters that result in accurate representation of the behaviour of that molecule, and, in most cases, are compatible with existing parameter sets. While many automated parameterisation methods exist, they are in general not well suited to large and conformationally dynamic molecules. I will describe the CherryPicker method for automated assignment of parameters for large, novel biomolecules, and demonstrate its usage for peptides of varying degrees of complexity. CherryPicker uses a graph theoretic representation to facilitate matching of the target molecule to molecular fragments for which reliable parameters are available. It requires minimal user input and creates parameter files compatible with the widely-used GROMACS simulation software.
15
KL-6. Debra Bernhardt The University of Queensland "Modelling the growth and properties of two-dimensional materials"
Wenyu Wei, Saiyu Bu, Debra J. Searles and Qinghong Yuan The University of Queensland, Australia and East China Normal University, China The growth of crystalline two-dimensional materials can be controlled through changes in the catalyst, feedstock, temperature and pressure. In this talk we will discuss how these factors influence the structures of graphene, two-dimensional nitrogen-doped graphene and two- dimensional carbon nitrides. For example, the growth of graphene using chemical vapour deposition (CVD) can be catalyzed using copper-nickel alloy catalysts and the nucleation density and size of the graphene flakes has been to vary greatly with the copper:nickel ratio. Computational modelling has allowed us to explain the optimal ratio, with the computational results in good agreement with experiment (Y Liu, et al. Advanced Science, 5, 1700961). In another example we will discuss how experimental conditions can alter the concentration and type of nitrogen in nitrogen-doped graphene. Finally, we will show how the electronic properties of the materials can be tuned using different physical constraints on structure of layered two-dimensional materials.
16
KL-7. Jonathan Essex University of Southampton "How well do we model protein electrostatics?"
Fixed-charge atomistic force fields use Coulomb interactions to model electrostatics. However, because of their empirical, highly parameterised nature, the extent to which these force fields correctly model protein electrostatics is unclear. To address this question, we have compared conventional fixed-charge force fields, the AMOEBA polarisable force field, linear-scaling Density Functional Theory, and experimental fields derived using vibrational Stark effect spectroscopy, to study the electric fields in two protein systems. The first, the peptidylprolyl isomerase Cyclophilin A, catalyzes the cis/trans isomerization of the amide preceding proline residues in proteins. Its mechanism is believed to be electrostatically modulated, at least in part, by the conserved R55 residue, which has been proposed to provide a stabilizing electric field for the transition state. For the second, ketosteroid isomerase, very large electric fields have been reported in the binding site (-143 MV cm-1) with implications for the catalytic mechanism. We find in both systems that the ability of fixed charge force fields to calculate accurate electric fields is inferior to that of the AMOEBA force field. Given that AMOEBA is explicitly designed to model electrostatics through a combination of distributed multipoles and inducible dipoles, this is perhaps not surprising. The broader implications of this result for QM/MM and continuum electrostatic calculations, will be discussed. .
17
KL-8. Habibah A. Wahab Universiti Sains Malaysia "Drug resistant influenza strains, what happens before the drugs bind to the active sites?" The recurrence of influenza pandemics indicated that the influenza virus has continuously evolved creating a constant threat to the humanity. Although there are treatment options available, the increased number of reported drug resistant strains of the influenza virus demands a complete understanding of the mechanism of resistance. Many studies have been conducted to uncover the mechanism of oseltamivir resistances in H274Y NA. However, most of the reported studies on H274Y were only focused on drug-bound-system, while direct effects of the mutation towards NA itself, prior to drug binding, remains unclear. Therefore, molecular dynamics of NA in apo-form, followed by principal component analysis and interaction energy calculation, were performed to investigate the structural changes of NA binding site, as a result of H274Y mutation. Sliding-box docking suggested that the binding pathway of OTV was compromised due to this binding site disruption. This study also highlighted the differences of H274Y effects in N1-and N2-subtypes and the importance of functional group at position C6 of sialic acid mimicry. Rational design of a series of potential neuraminidase inhibitors are also highlighted.
Abstracts: Contributed Talks
19
CT-1. Sebastian Maurer-Stroh Bioinformatics Institute (BII) A*STAR "When Sequences meet Structures" Molecular modelling of protein structures is a wide field and I want to raise awareness of the many synergies sequence studies can contribute to inform and complement structural modelling studies and vice-versa. I will start with successful examples of remote homology detection supported by structural core motifs, new trends in industry projects with enzyme design, followed by examples of predicting peptide-protein interactions using evolutionary sequence conservation to guide docking and will end by how large-scale virus sequence analysis combined with structural modelling can provide powerful insights into viral fitness and drug resistance mutations.
20
CT-2. Ivo C. Martins Universidade de Lisboa "Combining wet lab experiments with in silico data to understand Zika, West Nile and Dengue virus capsid protein" Ana S. Martins, André F. Faustino, Nina Karguth, Nuno C. Santos, Ivo C. Martins Dengue, West Nile and Zika, closely related viruses of the Flaviviridae family, are an increasing global threat, due to the expansion of their mosquito vectors. Our work, dealing with peptides and protein-ligand interactions in general [1–20], is particulalrly focused on the capsid (C) protein of these viruses [12–20]. This is a crucial structural protein that mediates not only viral assembly, but also encapsidation, by interacting with host lipid systems, as shown by us, via bioinformatics and wet lab experimental approaches [12-20]. This also led to pep14-23, a drug candidate designed by us [12,15]. We recently investigated further the C protein [17– 20]. It forms a dimer with a disordered N-terminal region, an intermediate flexible fold section and a very stable conserved fold region [17,18]. Comparing and analyzing relevant mosquito- borne Flavivirus C protein sequences and their predicted structures shows alternative conformations enabled by the N-terminal region essential for its function [17–20]. Using dengue virus C protein as main model, we then correlated protein size, thermal stability and function with its structure/dynamics features [17]. The findings suggest that minor allosteric changes may modulate the C protein biological activity. Therefore, this knowledge contributes to future drug development strategies against Zika, dengue and closely related flaviviruses. REFERENCES 1. Peptides, 49 (2013). 11. Biotech, 6 (2016). 2. Reprod Biol Endocrinol, 11 (2013). 12. J Virol, 86 (2012). 3. Brain, 140 (2017). 13. Biochem J, 444 (2012). 4. EMBO J, 29 (2010). 14. Sci Rep, 5 (2015). 5. Nat Methods, 7 (2010). 15. ACS Chem Biol, 10 (2015). 6. EMBO J, 27 (2008). 16. Nanomed (NBM), 10 (2014). 7. Chem Soc Rev, 43 (2014). 17. Int J Mol Sci, 20, 3870 (2019). 8. Arch Biochem Biophys, 531 (2013). 18. Sci Rep, 9 1647 (2019). 9. Proc Natl Acad Sci, 102 (2005). 19. Front Microbiol, 9 (2018). 10. Nanomed, 13 (2018). 20. Front Cell Infect Microbiol, 9 (2019).
21
CT-3. Roland G. Huber Bioinformatics Institute (BII) A*STAR "Integrative Modelling of Viral Genome Structures: Data and Strategies" Roland G Huber, Wan Yue Dengue (DENV) and Zika (ZIKV) viruses are clinically important members of the Flaviviridae family with an 11kb positive strand RNA genome. While structures have been mapped primarily in the UTRs, much remains to be learnt about how the rest of the genome folds to enable function and regulation. Here, we performed structure and interaction mapping on four DENV serotypes and four ZIKV strains inside their virions and in infected cells. Comparative analysis of SHAPE reactivities across serotypes nominated potentially functional regions that are highly structured, show structure conservation, and low synonymous mutation rates. Interaction mapping by SPLASH further reveals new pair-wise interactions, in addition to the circularization sequence. Approximately 40% of pair-wise interactions form alternative structures, suggesting extensive structural heterogeneity. Analysis of shared interactions between serotypes revealed a conserved macro-organization whereby interactions can be preserved at their physical locations beyond sequence identities. Comparing genome structures in virions, released into solution, and in host cells, revealed that long-range interactions tend to be disrupted inside cells. Compensatory mutations further demonstrate the importance of one of these new interactions for virus fitness. Our findings provide a structural framework to examine DENV and ZIKV genome organization and serve as a resource for design of RNA therapeutics that target the RNA structures of viruses.
Australian National University "Modelling divalent cations: the curious case of PsaA"
Hugo MacDermott-Opeskin, Megan O'Mara Manganese homeostasis is crucial for the viability of the pathogenic bacteria, S. pneumoniae, protecting against oxidative stress and aiding cellular metabolism. The substrate binding protein, PsaA, controls the selective uptake of manganese via coupling to a membrane import complex. PsaA lacks a metal chelating cofactor and faces significant competition from other d block metal species. Competitive and irreversible binding by other d block metals has been identified as a mechanism for bacterial susceptibility to zinc and cadmium. In this work compare and benchmark three molecular dynamics divalent cation models (LJ 6-12, LJ 6-12- 4, multisite models) to probe how effectively these cation models reproducing the experimental data for metal ion binding, coordination and release at the PsaA cation binding site.1,2 We use free energy calculations to reveal mechanisms of Mn2+ and Zn2+ binding to PsaA. We show that Mn2+ is scavenged more effectively than competing metal ions from solution by mobile carboxylates located near the entrance to the binding site. We demonstrate that ligand coordination by water molecules is essential in controlling the reversibility of binding. Our work reveals that the coordination chemistry of PsaA is precisely controlled to provide selectivity and reversibility for Mn2+.
Australian National University "Molecular Insights into Inhibition of the neurotransmitter transporter GlyT2"
Katie A. Wilson and Megan L. O'Mara Chronic pain is a condition that effects 1 in 5 Australians with the prevalence of chronic pain expected to increase as the population ages. Unfortunately, the current treatments have low efficacy and unacceptable side effects. Therefore, new treatments for chronic pain must be explored to develop new drugs to safely manage chronic pain. The glycine transporter, GlyT2, is a specific neurotransmitter transporter involved in neuropathic pain and therefore is of interest for the therapeutic treatment of chronic pain. Interestingly, previous work has shown that an endogenous bioactive lipid, N-arachidonyl-glycine (NAGly), inhibits GlyT2 and successfully reduces chronic pain. Based on this work, NAGly has been used as a lead compound to develop novel, potent, selective, and metabolically stable compounds that are inhibitors of GlyT2. The developed inhibitors all contain a lipid structure with a hydrophilic amino acid based headgroup and a single hydrophobic lipid tail. Using a combination of atomistic and coarse grain molecular dynamics simulations the binding of the inhibitors to GlyT2 has been characterised. Based on these calculations we proposed a mechanism of inhibition that is mediated by cholesterol and involves inhibitor binding to a novel extracellular allosteric binding site. Furthermore, a structure-function relationship has been developed for a variety of GlyT2 lipid inhibitors that vary in the stereochemistry and chemical composition of the lipid headgroup, and the number of carbon atoms in the lipid tail.
University of Santiago de Compostela "From molecular dynamics simulations of cyclodextrin-based molecular machines to macroscopic functional materials"
Ángel Piñeiro, Pablo F. Garrido, Martín Calvelo, Rebeca García-Fandiño The behavior of native cyclodextrins (CDs) in aqueous solution and at the air/water interface is much more complex than expected just considering their apparently simple molecular structure. Consequently, the correct prediction of their different properties and skills such as their solubility, their ability to adsorb at interfaces or to encapsulate different molecules is not straightforward; not to mention their propensity to aggregate forming different patterns in the bulk and at the interface. It is not a surprise then that the behaviour of modified cyclodextrins, as well as that of the supramolecular complexes they form upon interacting with different types of molecules, is not trivial. The surface film resulting from the self- assembly of some of these complexes at the water/air interface exhibits a strong viscoelastic response to mechanical perturbations perfectly visible in macroscopic experiments. In the present work we will show how this behavior can be explained from the cooperative effect of 2:1 molecular complexes showing a piston-like movement of the guest molecule inside the nanocylinder formed by the two CD, coordinated with a chiral-driven directional oscillation of the primary hydroxyl groups of the cyclodextrins. Moreover, a surprisingly specific electric dipole moment vector field and packing of the water molecules around the complex indicate that those solvent molecules should be considered as a part of the molecular machine. Acknowledgments This work was supported by the Spanish Agencia Estatal de Investigación (AEI) and the ERDF (RTI2018-098795-A-I00). M.C. thanks to Xunta de Galicia for a predoctoral fellowship. P. F. G. thanks the Spanish Ministry of Economy and Competitiveness and the European Social Fund for his predoctoral research grant, reference BES-2016-076761. R.G.-F. thanks to Ministerio de Ciencia, Innovación y Universidades for a “Ramón y Cajal” contract (RYC-2016-20335).
La Trobe Institute For Molecular Science "Computational Studies of Insulin Structure and Function"
Nicholas A. Smith, Brian J. Smith Diabetes Mellitus, a disease which is principally caused by the functional dysregulation of the peptide hormone insulin, effects millions of people worldwide with little discrimination. The hormone which is produced in the pancreas and acts by binding to the insulin receptor expressed on cells throughout the body, plays an integral role in homeostatic metabolism, and the processing of glucose from meals. Although research of insulin and diabetes spans close to a hundred years, leading to numerous Nobel prizes, an extensive body of literature, and most recently an understanding of the atomic structures of the complex between the hormone and receptor, the prevalence of the disease continues to increase. The emergence of computational techniques, principally that of the simulation method molecular dynamics, seeks to help arrest this trend, allowing researchers to comprehend molecular interactions which occur on an atomic scale, seldom achievable by other experimental techniques. Exploiting this method, we focus here on conducting a comprehensive investigation into the activation of the insulin receptor by the native insulin peptide and insulin analogues. We examine through a combination of computational modelling, simulations, free energy investigations and thermodynamic analysis naturally occurring insulin analogues; from that of insulin Wakayama, an analogue containing a single amino acid mutation which causes significant loss of function, single-chain insulins which are refractory to thermal degradation, and cone snail venom insulins. We make numerous conclusions on the internal dynamics and functional interactions which underpin the molecular interaction between the hormone and the insulin receptor; with the perspective gained seeking to assist in the designing of modern therapeutics which have a more advantageous pharmacokinetic profile.
Bioinformatics Institute (BII) A*STAR "Structome: Exploring the structural neighbourhood of proteins"
Ashar J. Malik*[Bioinformatics Institute, Agency for Science, Technology and Research, Singapore], Jane R. Allison [School of Biological Sciences, University of Auckland, New Zealand] Evolutionary relationships are conventionally uncovered using protein sequences. Protein structure as opposed to the sequence can hold evolutionary signals over longer timescales and can therefore prove useful towards uncovering deep evolutionary relationships. To this end, Structome has been developed as a resource for researchers to quickly determine the structural neighbourhood of a query structure. The structural neighbourhood comprises protein structures within a certain user selected structural similarity cutoff. This resource, therefore allows the inspection of the neighbourhood of a query protein structure from which inferences can be made about the evolutionary relationships, through the use of phylogenetic networks. Domain annotation from SCOP and CATH databases are also provided, to allow users to validate their observations, along with sequence similarity. Covering ~70% of the proteins in RCSB PDB, Structome is a comprehensive tool for the analysis of the protein structure landscape.
CSIR-Institute of Genomics and Integrative Biology "Human LC3 and GABARAP subfamily members achieve functional specificity via specific structural modulations"
Nidhi Jatana, David B. Ascher, Douglas E.V. Pires, Rajesh S. Gokhale & Lipi Thukral Autophagy is a key cellular degradation mechanism in nearly all organisms. Core components of this machinery are the six human Atg8 orthologs that play a critical role in autophagosome initiation and expansion. Although LC3B labeling is perhaps the most well-established and representative marker for autophagy, the ambiguity regarding its evolutionary partners (LC3A, LC3C, GABARAP, GABRAPL1, and GATE16) and their binding specificities is striking. In part, the difficulty in constructing functional signatures arises due to remarkable structural similarity as all of them share ubiquitin fold. The question arises as to how specific sequence variation amongst the family members lead to different substrate recognition preferences. To understand this, we developed a computational pipeline to define structural determinants of human Atg8 family members that dictate functional diversity. We observed a clear evolutionary separation between Atg8 orthologs and defined novel sequence recognition motif for human LC3 and GABARAP subfamilies. By analyzing molecular dynamics (MD) trajectories of known protein structures, we observed that functional modules or microclusters reveal a pattern of intramolecular network, including distinct hydrogen bonding of key residues that may directly modulate their interaction preferences. In addition, multiple simulations were performed to characterize how these proteins interact with a common protein binding partner, PLEKHM1. This analysis showed remarkable differences in binding modes via intrinsic protein dynamics, with PLEKHM1-bound GABARAP complexes showing less fluctuations and higher number of contacts. We further demonstrate that distinct cancer- related mutations were likely to lead to significant structural changes. Our findings provide an extensive structural framework of diversity within human ATG8 proteins they may serve as an underlying mechanism of cross-talk and molecular control with other related signaling pathways.
University of Auckland "Phylogenetic inference from protein structure using a toroidal diffusion model"
Alex Popinga, Fabio Mendes, Jane Allison Typical models designed to infer phylogenetic trees, evolutionary relationships of the input data, rely upon aligned sequence data and either discard completely or only indirectly/partially take into account structural information. Not taking into account structural data can be problematic, particularly for deep-rooted trees, as protein structures may contain information about evolutionary trajectories that has been lost in their amino acid and corresponding gene sequences. We introduce two novel extensions of a recently published angular diffusion model for homology detection (Golden et al., 2017) in the Bayesian phylogenetic inference package BEAST2 (Bouckaert et al., 2019). The first model, SPITE (Structural Phylogenetic Inference using Toroidal Evolution), is the foundation for performing phylogenetic inference using protein structures described by their dihedral angles. MINOTOR (Molecular dynamics-Informed iNference On a TORus) is our proposed extension of SPITE that will incorporate simulations of molecular dynamics to inform the likelihood of observing sampled proteins as well as the probability of each ancestral structure.
Bioinformatics Institute (BII) & Experimental Drug Development Centre A*STAR "Planet of the Apps: Scientific Phone Apps & Mobile Devices"
Many significant biomedical discoveries of old were made in the private property of famous scientists e.g.Leeuwenhoek and Archimedes. Today, discoveries are made in brightly-lit, hi- tech, ergonomic buildings that house research institutes. While such development is advantageous in many aspects, the spatial restriction of research into well-organized structures may delay and limit the spontaneity necessary for discoveries. The smartphone and peripheral mobile devices have the potential to not only increase the productivity and mobility of biomedical research, but also restore some freedom from spatial constraints. One possible way this can occur is the development of a mobile biomedical lab that allows researchers to carry out core research processes ‘on-the-go’ without being spatially restrained within a building or availability of equipment. For this exciting prospect, we surveyed the Google and Apple app stores, discussing the limitations and the potential of this area. Based on the developments, it appears to be just a matter of time before the majority of biomedical labs processes and equipment become mobile, centred on the smartphone andperipheral devices.
Schrödinger, Inc., Bangalore, India "Molecular modelling - an industrial perspective"
Schrödinger INC is pioneer in developing applications for both materials and molecular design. There are more than 2000 research groups across the globe use Schrödinger simulation suites for their research. Understanding the molecular level structure-property relationship allows the researchers to effectively tackle many complicated problems and fine tune the properties of materials into the desired range. Schrodinger has designed several automated workflows to effectively predict the chemical and physical properties and stability of materials. This presentation will show few case studies from pharmaceutical formulations, polymers, OLED, catalysis and energy storage sectors of the industry.
CT-13. Raghu Rangaswamy
Vice President, Schrödinger, Inc., Bangalore, India "FEP+ the game changer for lead optimisation in drug design"
Over the past decade, computational drug design has undergone tremendous advances in the form of forcefield development, advancement in algorithms, availability of enormous amount of crystal structures, GPU and cloud computing etc. These advances in computational development and enormous amount of data has enabled the molecular simulations to identify the potential lead molecules much faster and accurate. This is really a boon for the pharma R & D a lot where they can reduce the cost and identify the lead molecules much faster.
The presentation would highlight on the developments in lead identification methods and on the lead optimisation methods. Today free energy perturbation calculation methods have matured to the point where we can calculate the binding affinity of a molecule with in few hours, which is a boon to a chemist to decide and go for synthesis. We have observed in our drug discovery collaborations that the use of free energy perturbation can lead to a measurable and significant acceleration of drug discovery projects in prospective applications.
The presentation would highlight few case studies from our drug discovery collaborations. How we did virtual screening and selected few lead molecules from database of millions of compounds. How the selected leads were expanded using computational methods to find potential analogs and how we did lead optimization to select very few successful lead candidates which are at par with experimental results.
Santiago de Compostela University "“Intelligent” Materials Targeting Bacterial And Cancer Cell Membranes: How Computational Tools Can Help?"
Rebeca Garcia-Fandino, Martin Calvelo, Gideon F. Tolufashe, Antonio Peón, Bárbara Claro, Alicia Muñoz, Daniel Conde, Ángel Piñeiro, Margarida Bastos, Juan R. Granja There is a need for the development of new antineoplastic and antimicrobial therapies, with higher selectivity, leading to fewer side effects than current ones. One strategy proposed is the use of bacterial or cancer cell membranes as a therapeutic target so that their basic properties are perturbed, altering the membrane potential and inhibiting the control functions on the signaling, communication or production bioenergy processes. Therapeutic peptides are a novel and promising approach for the development of both antimicrobial and anti-cancer agents that could specifically target bacteria or cancer cells with lower toxicity to normal tissues, which will offer new opportunities for cancer and infection treatments. Since lipid membranes are the major target of most of these therapeutic peptides, the development of drug resistance is less likely to occur since damaging cytotoxicity can take place within minutes of peptide introduction. Traditional design and optimization studies of peptides (or peptidomimetics) are known to be expensive and time-consuming. A detailed understanding of the molecular details of the membrane permeabilization process would allow the rational design of new molecules with the same mechanism of action, but with improved activity, selectivity, and bioavailability. Recent advances in computer power and methodology, including Molecular Dynamics simulations, have made possible to systematically explore events that take place into ranges where direct comparison and experimental testing are starting to be feasible, realizing the synergistic potential of a combined in-silico/in-vitro approach in the characterization of the membrane destabilization process by antitumoral or antimicrobial molecules. In this talk, some recent examples of our research related to the study of different peptides and peptidomimetics acting at the membrane level will be illustrated.
CT-15. Ricardo Mancera
Curtin University "How does sucrose change its mechanism of stabilization of lipid bilayers during desiccation? The roles of hydration and concentration"
Slawomir S. Stachura, Chris J. Malajczuk and Ricardo L. Mancera The interactions between sugars and membranes are thought to be responsible for the stabilisation of cells during desiccation and freezing, usually associated with a decrease of the main phase transition temperature of phospholipid bilayers. However, the underlying molecular mechanism is still not well understood. There are two opposing views on how this is achieved: the direct sugar-phospholipid interaction at the bilayer interface (water replacement hypothesis) and an entropy-driven phase transition where sugar molecules concentrate away from the lipid interface (hydration forces explanation). Various experiments support these two mechanisms but molecular dynamics (MD) simulations have overwhelmingly shown the occurrence of direct sugar-phospholipid interactions. We have conducted MD simulations of DOPC bilayers at different levels of hydration and in the presence of different sucrose contents as a representative system. Sucrose was found to behave in a manner that depends on both the sucrose contents and the level of hydration: at high sucrose concentration at low hydration it is best described by the hydration forces explanation model, whereas at low sucrose concentration it is consistent with the water replacement hypothesis. At low concentration, sucrose molecules are revealed to preferentially interact directly with the bilayer interface, while at high concentration they preferentially accumulate in the inter-bilayer solution. We have observed for the first time that the transition between the two modes of interaction is determined by the saturation of the lipid bilayer interface with sucrose molecules, which occurs more rapidly as the level of hydration decreases.
National Yang-Ming University "Planar and spherical - comparison of lipid dynamics in bilayers"
Meng-Han Lin and Wolfgang B. Fischer Spherical bilayers such as vesicles comprise an important container system as drug delivery vehicles as well as bio-reactors. Lipid composition and functionalization of the vesicles are a prerequisite for their area of application. Computational screening of the properties of vesicles of different composition constitutes a desirable tool to support the development of this system. Computational lipid bilayers and vesicles are built using zwitterionic (POPC, DOPC) and negatively charged (DOPS) lipids and mixtures thereof (5 % DOPS in either POPC or DOPC) as coarse-grained models (MARTINI) in molecular dynamics (MD) simulations. Vesicles (Ø 10 nm) are generated by placing each lipid on a sphere based on the area-per-lipid value leading to stable and precisely defined vesicles (tailor-made vesicles). The effect of restrained positive point charges (Ca-ions) external to bilayers and vesicles on the dynamics of the lipid molecules is investigated. The curves shape leads to a reduction of the diffusivity of the lipids. Independent of the shape of the bilayer and the type of mixture, the presence of the external charges reduces the diffusivity of the DOPS molecules the most. The reduced lipid diffusivity leads to lower DOPS density around the external charges in the vesicles than around the external charges in the planar lipid bilayer.
CT-17. Baris Demir
The University of Queensland "Cation effect on the electrolyte structure with an applied potential: a molecular level investigation using a constant potential method"
Baris Demir, Debra J. Searles Future energy storage devices require both high power and energy densities. In general, batteries have a high energy density whereas supercapacitors have a high power density. Developing devices that achieve both is therefore of significance. A striking difference between these two systems is the energy storage mechanism. Energy is stored on the electrodes of a battery through chemical reactions while it is stored via ion adsorption on the electrodes in supercapacitors. To advance the performance of supercapacitors, it is vital to understand the molecular-level formation of layers in the vicinity of electrodes, or the electric double layers (EDLs) as this is a key factor for their performance. In this work, we investigate the effect of the cation structure and dynamics on the layering in the electrolyte-to-electrode interface using molecular dynamics (MD) simulations. We implement a constant potential method to fully capture the dynamics of C2mimNTf2 and N4,1,1,1NTf2 ionic liquids (IL) at varying potential differences applied across the electrochemical cell. Two ILs are examined: C2mimNTf2 and N4,1,1,1NTf2, with graphene electrodes. A constant potential method has been applied to model electrochemical cells at the molecular level. After equilibration of the IL at 294 K in the presence of graphene electrode, a potential difference was applied. Our MD simulation results indicate that the applied potential created ion-dense layers at both electrode interfaces. This affected the charge density distribution in the EDL. Asymmetric charge density distribution profiles formed in the EDL on both electrode surfaces when the potential difference was turned on. The EDLs have substantially different compositions near both electrodes as a function of applied potential. We explore the charging/discharging behaviour of supercapacitors at various potential difference by varying the IL to better understand the dynamics of ions for the aim of tuning and designing new ILs.
University of Wollongong "Atomistic insights into photoprotein formation: Computational prediction of the properties of coelenterazine and oxygen binding in Obelin"
Thomas M. Griffiths, Aaron J. Oakley, Haibo Yu Bioluminescence in marine systems is dominated by the use of coelenterazine for light production. The bioluminescent reaction of coelenterazine is an enzyme catalysed oxidative decarboxylation: coelenterazine reacts with molecular oxygen to form carbon dioxide, coelenteramide and light. One such class is the Ca2+-regulated photoproteins. These proteins bind coelenterazine and oxygen, and trap 2-hydroperoxycoelenterazine, an intermediate along the reaction pathway. The reaction is halted until Ca2+ binding triggers the completion of the reaction. There are currently no reported experimental, atomistic descriptions of this ternary Michaelis complex. This study utilised computational techniques to develop an atomistic model of the Michaelis complex. Extensive molecular dynamics simulations were carried out to study the interactions between four tautomeric/protonation states of coelenterazine and wide-type and mutant obelin. Only minor differences in binding modes were observed across all systems. Interestingly, no basic residues were identified in the vicinity of the N7-nitrogen of coelenterazine. This observation was surprising considering that deprotonation at this position is a key mechanistic step in the proposed bioluminescent reaction. This work suggests that coelenterazine binds either as the O10H tautomer, or in the deprotonated form. Implicit ligand sampling simulations were used to identify potential O2 binding and migration pathways within obelin. A key oxygen binding site was identified close to the coelenterazine imidazopyrazinone core. The O2 binding free energy was observed to be dependent on the protonation state of coelenterazine. Taken together, the description of the obelin-coelenterazine-O2 complexes established in this study provides the basis for future computational studies of the bioluminescent mechanism.
Taylor's University "Benchmarking the performance of MM/PBSA in virtual screening enrichment with the GPCR-Bench dataset"
Mei Qian Yau, Abigail L. Emtage, Jason S.E. Loo Recent breakthroughs in G protein-coupled receptor (GPCR) crystallography and the subsequent increase in number of solved GPCR structures has allowed for the unprecedented opportunity to utilize their experimental structures for structure-based drug discovery applications. As virtual screening represents one of the primary computational methods used for the discovery of novel leads, the GPCR-Bench dataset was created to facilitate comparison among various virtual screening protocols. In this study we utilize 16 GPCR targets, consisting of 3026 actives and 182,927 decoys from the GPCR-Bench dataset, to benchmark the performance of Molecular Mechanics/Poisson Boltzmann Surface Area (MM/PBSA) in improving virtual screening enrichment obtained from docking using Glide. The top 10% of the docked database was rescored using MM/PBSA, using both the top-ranking pose and the top ten poses. We additionally performed the MM/PBSA rescoring following the Binding Estimation After Refinement protocol, which consists of a short 100ps MD simulation and energy minimization followed by rescoring. Our results indicate that MM/PBSA performance in improving enrichment factors was mixed. MM/PBSA rescoring resulted in an improvement in the EF1% for 5 targets and a decline for 11 targets, while the EF5% showed an improvement for 7 targets and a decline for 9 targets. While some targets such as ADRB1 and CRFR1 showed significant improvements when using MM/PBSA rescoring, other targets demonstrated significant decline. There was no clear benefit in rescoring the top ten binding poses compared to the top docked pose, as well as in refinement using short MD simulations. In both cases, although virtual screening performance varied between individual targets, we observed only small differences in enrichment factors on average. Overall, the performance of MM/PBSA rescoring in improving virtual screening enrichment obtained from docking of the GPCR-Bench dataset was found to be relatively modest.
CT-20. SrinivasarRaghavan Kannan
Bioinformatics Institute (BII) A*STAR "Macrocyclization of an all-D linear peptide improves target affinity and imparts cellular activity: A novel stitched α-helical peptide modality"
Srinivasaraghavan Kannan, Pietro G. A. Aronica, Simon Ng, Dawn Thean Gek Lian, Yuri Frosi, Sharon Chee, Jiang Shimin, Tsz Ying Yuen, Ahmad Sadruddin, Hung Yi Kristal Kaan, Arun Chandramohan, Jin Huei Wong, Yaw Sing Tan, Fernando J. Ferrer, Prakash Arumugam, Yi Han, Shiying Chen, Christopher J. Brown, Charles W. Johannes, Brian Henry, David P. Lane, Tomi K. Sawyer, Chandra S. Verma, Anthony W. Partridge Peptide-based inhibitors hold great potential for targeted modulation of intracellular protein- protein interactions (PPIs) by leveraging vast chemical space relative to primary structure via sequence diversity as well as conformationally through varying secondary and tertiary structures. However, the development of peptide therapeutics has been hindered because of their limited conformational stability, proteolytic sensitivity and cell permeability. Several contemporary peptide design strategies address these issues to varying degrees. Strategic macrocyclization through optimally placed chemical braces such as olefinic hydrocarbon crosslinks, commonly referred to as staples, may address these issues by i) restricting conformational freedom to improve target affinities, ii) improving proteolytic resistance, and iii) enhancing cell permeability. Conversely, molecules constructed entirely from D-amino acids are hyper-resistant to proteolytic cleavage, but generally lack conformational stability and membrane permeability. Since neither approach is a complete solution, we have combined these strategies to identify the first examples of all-D α-helical stapled and stitched peptides. As a template, we used a recently reported all D-linear peptide that is a potent inhibitor of the p53-Mdm2 interaction, but is devoid of cellular activity. To design both stapled and stitched all-D-peptide analogues, we used computational modelling to predict optimal staple placement. The resultant novel macrocyclic all D-peptide was determined to exhibit increased α-helicity, improved target binding, complete proteolytic stability and, most notably, cellular activity.
Abstracts: Short Talks
Indian Institute of Technology Madras "Molecular Basis of Differential Stability and Temperature Sensitivity of Zika versus Dengue Virus Envelopes"
Chinmai Pindi, Venkat R Chirasani, Mohd Homaidur Rahman, Mohd Ahsan, Prasanna D Revanasiddappa, and Sanjib Senapati* Rapid spread of zika virus (ZIKV) and its association with severe birth defects have raised worldwide concern. Recent studies have shown that ZIKV can survive in high fever, unlike dengue and other flaviviruses. In spite of recent cryo-EM structures that showed similar architecture of zika and dengue virus (DENV) envelopes, little is known that makes the former so unique. Here, we unravel the molecular basis of greater thermal stability of ZIKV envelope over DENV by employing all-atom molecular dynamics (MD) simulations. Our results show that ZIKV envelope retains its structural integrity while DENV envelope loosens up through the inter-raft interfaces. Protein structural network extracted from simulation data traced crucial residues involved in electrostatic and H-bonding interactions that make the ZIKV raft- raft interfaces robust. The residue-level information obtained here could pave way for designing small molecule inhibitors and specific antibodies to inhibit ZIKV E protein assembly and membrane fusion.
ST-2. Syeda Lubna
BITs Pilani "Evolution of NS1 protein from H1N1 influenza A virus across the Indian population from 2007 to 2019"
Debashree Bandyopadhyay, Suma Chinta, Prakruthi Burra, Kiranmayi Vedantham, Sibnath Ray, and Syeda Lubna. Background: Pandemic outbreak of influenza virus dated back to 2009, in India. Severity in viral infection varied across different Indian states in last ten years, indicating possible changes in viral genome. NS1 protein of H1N1 virus is a potent antagonist to host antiviral immune response system, thus contributing directly towards viral propagation and pathogenicity. Objective: This work documented comparative analyses of NS1 protein sequences and structures from Indian and global isolates and their possible consequences on NS1 interacting partners. Methods: Sequence analyses of NS1 protein were based on public and restricted databases. Crystal structure analyses and hydrophilicity calculations were based on PDB database. Antigenicity of the NS1 was predicted and compared across all the sequences, based on online tools. Functional changes in NS1 proteins were curated from literature, over last one hundred years. Results: Sequence analyses have identified changes in NS1 sequence positions, 25, 26, 48, 55 and 67, within the time span 2007 to 2017 from Indian isolates. Those residue positions were present on the antigenic segments of NS1 proteins from Indian isolates on and after 2015, only. Changes in hydrophilicity of four residues, 25, 26, 48 and 67 were noted based on NS1 crystal structures obtained from various isolates at different time. Conclusion: Overall study indicated shift of the antigenic hydrophilic surface of NS1 protein from Effector Domain (ED) to RNA Binding Domain (RBD) as the virus evolved from 2009 to 2017 in India. These changes in NS1 proteins lead to alteration of interacting cellular partners.
41
ST-3. Raphael Tze Chuen Lee Bioinformatics Institute (BII) A*STAR "FluSurver: current status and future development plans"
Raphael Tze Chuen LEE, Sebastian Maurer-Stroh FluSurver is an online tool that helps researchers analyze, identify and interpret the phenotypic consequences of mutations in influenza sequences sampled from routine surveillance efforts and new influenza outbreaks. It is widely used by National Influenza Centres, WHO Collaborating Centres and users of the EpiFlu database by the GISAID initiative. FluSurver has aided the discovery of new influenza strain variants with altered antiviral susceptibility, host specificity, glycosylation and shifts in antigenic properties. Currently, it uses curated mutational data from the literature to flag them out to users, highlights mutations on structural models, collates epidemiological data and processed them for geographic and temporal visualization. Going forward, it will allow users to compare their sequences with pre-built genome phylogenies, conduct predictions on antigenic drifts and egg adaptations, and allow users to visualize the local and global transmission chains of their viruses. We will also highlight some areas of our research where the use of structural models has assisted us in the analysis of mutations and better understand the biology of the virus.
Bioinformatics Institute (BII) A*STAR & The University of Manchester "Solvent mapping approach for uncovering cryptic pockets in membrane-bound proteins"
Lorena Zuzic, Jan K. Marzinek, Jim Warwicker, Peter J. Bond Flaviviruses are vector-borne human pathogens which include viruses such as dengue, Zika, yellow fever virus or the West Nile virus. Dengue virus is associated with 390 million clinical cases worldwide and the research into a safe and effective neutralizing antibody or a drug molecule is still ongoing, with a particular focus being placed on a viral envelope consisting of a protein and a membrane component. Molecular dynamics simulations with small organic probes added to the solvent are being used as a method to efficiently reveal cryptic pockets usually hidden from the water environment. We have designed a benzene-mapping method which relies on a modified force-field with added repulsive forces between the membrane and the benzene molecules, enabling effective cryptic pocket discovery even with a membrane present in the system. Repulsion parameters were optimized using a system containing a heterogeneous membrane. The method was tested on a dengue envelope and subsequently applied to six viral strains with a goal of finding a cryptic pocket conserved within the flaviviral family and with a potential to be a drug-binding site. Cryptic pocket which experimentally binds n-octyl-β-D-glucoside was consistently revealed in all simulations with benzene probes in the solvent. The addition of benzene also enhanced the flexibility and hydrophobic exposure of pockets within the domain III across all simulated systems, suggesting the drug-binding capacity of the envelope protein and the potential to modify or obstruct the interactions between the virus and the host receptors.
Indian Institute of Technology Madras "Electrostatically determined asymmetry of substrate binding in HIV-1 protease: A comprehensive MD simulation study"
Mohd Ahsan, Sanjib Senapati HIV-1 protease (HIVpr), an aspartyl protease is one of the key viral enzymes in the life cycle of HIV as it mediates the processing of gag and gag-pol polyproteins into protein products essential for viral maturation. Therapeutic inhibition of this protein affects the viral maturation and results in the formation of immature noninfectious virions. Hence, HIV protease is one of the prime targets to combat HIV infection. Currently, more than ten FDA approved drugs acting as competitive inhibitors are available in the market against HIV protease. However, studies have reported the emergence of resistance making these drugs ineffective. Drug resistance in protease causes imbalance in the molecular recognition process where the resistant enzyme no longer allows the effective binding of drug molecules but still can recognize and process the natural substrates. Hence, detailed understanding of substrate binding and recognition is required to combat drug resistance. Here, in this comprehensive MD simulation study we show that all the substrates under the study when bound to the active site displayed a asymmetrical binding with higher interaction energy towards unprotonated monomer, both drugs showed a symmetrical binding with similar interaction energy towards each monomer. This was primarily due to the preferential binding of backbone atoms (BB) of substrate residues with unprotonated monomer and such a behavior is not observed in the rigid drug molecules. Further, our electrostatic calculations revealed considerate difference in electrostatic interaction energy of substrates and drug molecules. Based on our results we propose that imparting of asymmetric binding pattern by improving electrostatic component and fitting in substrate envelope by vander waal component in drug molecules would make them less susceptible to resistant mutations.
Aarhus University "A generic protocol for constructing nanodiscs in silico"
Lisbeth Ravnkilde Kjølbye, Birgit Schiøtt A quantum leap in studies of membrane proteins took place upon the establishment of the Nanodisc (ND) technology in the late 1990s. In NDs engineered versions of the human apolipoprotein I (Apo-AI), referred to as membrane scaffold proteins (MSPs), self-assemble into discoidal phospholipid bilayers wrapped with an amphipathic helical belt surrounding the alkyl chains on the phospholipids. NDs can be used to obtain membrane proteins stable in solution, a main advantage compared to other available membrane mimics. The selection of the MSP has so far been governed by two opposing considerations: i) optimal nanodisc stability obtained with small sizes and ii) minimization of the finite size effects, obtainable with larger sizes. A faster and easier approach to select the optimal ND for the membrane protein in question, can be obtained by getting a baseline understanding of how the lipid properties change with size and MSP variant. In this work we present a generic protocol for constructing molecular models of NDs for molecular dynamics simulations. The protocol is written in python, making it fast and easy to modify. We validated and tested the protocol by simulating seven different NDs in various sizes and versions. The structural and lipid properties were analysed and shown to be in good agreement with previously reported studies.
The University of Auckland "Markov State Model of antibacterial peptides"
Aparajita Chakraborty, Paul Harris, Margaret Brimble, Bettina Keller, Jane Allison Modern medicine relies heavily on antibiotics for treating bacterial infections. However, bacterial resistance against many antibiotics has become a major issue, and this problem is expected to become much more serious in the not-too-distant future. This drives the continuous search for new synthetic and natural antibacterial agents. One promising source of these is antimicrobial peptides. Antimicrobial peptides are short peptides often produced by bacteria themselves as a defence mechanism while competing for food and resources in “bacterial wars”. [1] Our study focuses on linear structural analogues of battacin. To understand completely the functionality of these analogues we need to capture a detailed picture of both the thermodynamics and the kinetics of the system at atomistic level. Molecular dynamics simulations can provide this, but unfortunately, it is extremely challenging to reach biologically relevant timescales with molecular dynamics simulation, and hence even more challenging to obtain statistics necessary for accurate understanding of the characteristics of the system. Using Markov State Models (MSMs), however, we can overcome this issue. A Markov model represents a network of conformational states and a transition probability matrix describing the chances of movement from one state to another at a small time interval.[2] Since the states in an MSM are defined based on kinetic criteria rather than on geometric criteria, we can accurately identify the boundaries between free energy basins and model the complete free energy landscape of these peptides. References- 1. Cornforth, D.M.; Foster, K.R., Antibiotics and the art of bacterial war. Proceedings of the National Academy of Sciences 2015, 112 (35), 10827-10828. 2. J. H. Prinz et al., “Markov models of molecular kinetics: Generation and validation,” J. Chem. Phys.2011, vol. 134, no. 17.
Aarhus University "Allosteric Communication in the NMDA Receptor"
Nils A. Berglund, Jose Flores-Canales, Birgit Schiøtt The NMDA receptor (NMDAR) is a ligand-gated ion channel present in postsynaptic neurons. It is part of the glutamate receptor family and plays an essential function in synaptic plasticity and memory formation. Dysregulation of this receptor has been implicated in a range of conditions including Alzheimer’s, epilepsy, schizophrenia and depression. The NMDAR is therefore of great therapeutic interest, and modulation of the receptor has thus far shown promising results. One method used to gain additional insight into the intricacies of the NMDAR is molecular dynamics. This can provide atomistic scale insight into the interactions between protein and ligands, as well as any alterations in protein dynamics based on ligand binding, something that is of great interest in drug development. A significant challenge in simulations of the NMDAR is the size of the protein, as well as the number of loosely connected domains. Crystal structures have resolved the structure of the amino-terminal, ligand-binding and trans- membrane domains, together making up over 3000 residues, leading to systems over 500,000 atoms in size, resulting in sampling issues. The large size of the system makes these simulations expensive to run and the number of domains means long timescales are required to ensure inter-domain communication is achieved and the protein reaches an equilibrated state. For this reason alternative ways of gaining atomistic insight than expensive multi- microsecond simulations are required, in this work we used allosteric network analysis to compare the inter-domain communication pathways in the presence and absence of ligands, showing clear differences between the apo state and in the presence of glutamate and glycine.
ST-9. Lanie Ruiz-Perez
Curtin University "Permeation of short peptides across a stratum corneum model: application of a new, flexible enhanced sampling method"
Lanie Ruiz-Perez, Carlo Martinotti, Evelyne Deplazes and Ricardo L. Mancera In the context of dermal and transdermal drug delivery there is great interest in predicting which drug candidates show faster permeation rates across the stratum corneum (SC), the outer layer of the skin which acts as a barrier. The extracellular environment of the SC features a series of stacked lipid bilayers (LBs) in the gel phase. The ceramide-rich composition and high dehydration state of these LBs contribute to their remarkable resistance to permeation of exogenous compounds (1). The slow dynamics of LBs in the gel phase hampers the accuracy of permeability estimates from molecular dynamics simulations. Umbrella sampling (US), the most popular enhanced sampling method, suffers from insufficient sampling as the system is easily trapped in low-energy configurations of the permeant for up to hundreds of nanoseconds. Therefore, convergence in the calculation of free energies of permeation and hence permeability coefficients is usually poor except for very small molecules. We have developed a flexible implementation of replica exchange with solute scaling (REST2) (2) within GROMACS 4.6.7 that can be combined with US. Our implementation allows the independent scaling of electrostatic and van der Waals non-bonded interactions, such that they can be independently modulated between any pairs of components in the system. This US+REST implementation was used to obtain the potential of mean force (PMF) for the permeation of the dipeptide Ala-Tpr across a SC model. Our method yielded broader sampling of orientations and conformations of the dipeptide, confirming its ability to overcome local energy minima, especially in US windows where the dipeptide is embedded in the bilayer. Similarly, the PMF shows less pronounced global minima and maxima compared to that obtained by conventional US. These findings suggest significant enhancement of sampling which is validated against experimental permeability coefficients and free energy of partition (3).
Indian Institute of Technology- Gandhinagar, Ahmedabad University “New age antimicrobial peptides: Revealing mode of actions of multi functional AMPs using molecular dynamics simulation study”
Nirali Desai, Dr. Stephen Fox, Dr. Chandra Verma Currently, antimicrobial resistance developed by many infectious pathogens is a severe emerging problem. Antimicrobial peptides can be used as potential alternatives to conventional antibiotics because of their multi functionality and non-specificity in targeting pathogens. To understand different mechanisms of killing via bacterial membrane by AMPs in detail and to see differences in the mode of action of two peptides, magainin2 and pleurocidin with different modes of action we performed Molecular Dynamics simulations. Experimentally, magainin2 is known to form toroidal pores in the membrane whereas pleurocidin is known to interact with the intracellular targets. Molecular dynamic simulations were run for both peptides and for each orientation for 100-1000 ns using Gromacs and the charmm36m force field. Modelling a bacterial membrane (POPE:POPG in 3:1) solvated in the TIP3P water model and 0.15M NaCl ions. Magainin2 was found to significantly disrupt the membrane by forming toroidal pores, however pleurocidin also seemed to be form pores when forced in the membrane.
La Trobe University "Molecular Evolution of the Switch for Progesterone / Spironolactone from Mineralocorticoid Receptor Agonist to Antagonist"
Ruitao Jin, Sitong He, Peter Fuller, Brian Smith The mineralocorticoid receptor (MR) is highly conserved across vertebrate evolution. In terrestrial vertebrates the MR mediates sodium homeostasis by aldosterone and also acts as a receptor for cortisol. Although the MR is present in fish, they lack aldosterone. The MR binds progesterone and spironolactone as antagonists in human MR but as agonists in zebra fish MR. We have defined the molecular basis of these divergent responses using MR chimeras between the zebra fish and human MR coupled with reciprocal site-directed mutagenesis and molecular dynamic simulation on the structures of the MR ligand-binding domain. Substitution of a leucine by threonine in helix 8 of the ligand-binding domain of the zebra fish MR confers the antagonist response. This leucine is conserved across fish species whereas threonine (serine in rodents) is conserved in terrestrial vertebrate MR. MD identified an interaction of the leucine in helix 8 with a highly conserved leucine in helix 1 that stabilises the agonist conformation including the interaction between helices 3 and 5, an interaction which has previously been characterised. This switch in the MR coincides with the evolution of terrestrial vertebrates and of aldosterone synthesis. It was perhaps mandatory if the appearance of aldosterone as a specific mediator of the homeostatic salt retention was to be tolerated. The conformational changes also provide novel insights into the structural basis of agonism versus antagonism in steroid receptors, with potential implications for drug design in this important therapeutic target.
University of Cambridge "Mechanism of completion of peptidyltransferase centre assembly in eukaryotes"
Kargas V, Castro-Hartmann P, Escudero-Urquijo N, Dent K, Hilcenko C, Sailer C, Zisser G, Marques-Carvalho MJ, Pellegrino S, Wawiórka L, Freund SM, Wagstaff JL, Andreeva A, Faille A, Chen E, Stengel F, Bergler H, Warren AJ.
Eukaryotic ribosome biogenesis initiates in the nucleus and involves the concerted action of over 200 trans-acting assembly factors to produce functional 40S and 60S ribosomal subunits. Upon export to the cytoplasm, the remaining assembly factors on the pre-60S ribosomal particles are released and the last ribosomal proteins are integrated, sculpting the key functional site of the ribosome called peptidyltransferase centre (PTC). Despite the recent advances in structural biology, cytoplasmic ribosome biogenesis still remains elusive. Here, we set out to determine the mechanism of cytoplasmic 60S subunit maturation and completion of PTC assembly by using tandem affinity purification, immunoblotting, cryo electron microscopy (cryo-EM) and molecular modelling. Single particle cryo-EM analysis resulted in a series of pre-60S structures representing six distinct late cytoplasmic assembly states that could be ordered into a sequential maturation pathway. More specifically, recruitment of eL40 stabilises helix 89 to form the uL16 binding site. The loading of uL16 unhooks helix 38 from Nmd3 to adopt its mature conformation. In turn, partial retraction of the L1 stalk is coupled to a conformational switch in Nmd3 that allows the uL16 P-site loop to fully accommodate into the PTC where it competes with Nmd3 for an overlapping binding site. Our data reveal how the central functional site of the ribosome is shaped and suggest how the formation of translation-competent 60S subunits is disrupted in leukaemia- associated ribosomopathies.
ST-13. Malancha Karmakar
University of Melbourne "SUSPECT-PZA: An empirical tool to determine novel drug resistance in Pyrazinamide"
Malancha Karmakar, Justin T. Denholm and David B. Ascher Pyrazinamide, a first-line drug with sterilizing activity, plays an important role in tuberculosis treatment; however, its use is complicated by side-effects and challenges with reliable drug susceptibility testing. Resistance to pyrazinamide is largely driven by mutations in pyrazinamidase (pncA), responsible for drug activation, but genetic heterogeneity has hindered development of a molecular diagnostic test. Our objective was to use information from the proteins 3D structure to accurately identify resistance mutations in pncA. To achieve this, we curated 610 pncA non-synonymous single nucleotide mutations with associated high confidence experimental and clinical information on pyrazinamide susceptibility. The molecular consequences of these mutations were assessed using the mCSM platform, which provided insights into changes in protein stability, conformation, and interactions for each mutation. Using these structural and biophysical effects, we could correctly classify mutations as either susceptible or resistant with an accuracy of 80%. Our model was validated against a previously documented set of non-redundant clinically resistance mutations and the CRyPTIC dataset, achieving 79% and 81% accuracy respectively. We further validated our model using a novel set of previously unreported clinical mutations with experimental drug susceptibility testing from over 600 Victorian patients, and obtained 71% accuracy. Using the insights from this model, we also performed a real-time analysis on a Victorian tuberculosis patient, in which pyrazinamide treatment would not be effective and led to its discontinuation. This was the first use of structural information to guide clinical resistance detection. We have made this model freely available through a user-friendly web interface called SUSPECT-PZA. This will be a valuable resource to analyse any pncA missense mutation, providing structural insight to help guide patient treatment decisions and screening programs.
GSK and University of Strathclyde "Binding Pose Metadynamics: Exploring Ligand stability in Protein Crystal Structures"
Lucia Fusani, David Palmer, Don Somers and Ian Wall The prediction of the correct protein-ligand binding pose or poses is important in structure- based drug design and crucial for the evaluation of protein-ligand binding affinity. Most three- dimensional protein/ligand structures are obtained from single crystal X-ray crystallography experiments which result in a single static model of an ensemble of conformations. The Binding Pose Metadynamics [1] (BPMD) tool allows the study of ligand stability in full atomistic detail in a computationally efficient manner averaging over 10 × 10 ns metadynamics runs with the root-mean square deviation of the ligand heavy atoms as the collective variable. The basic principle of BPMD is that ligand poses which are unstable (average RMSD > 2 Å) under the bias of the metadynamics simulation are likely to be infrequently occupied in the energy landscape and make minimal contributions to the protein-ligand binding affinity. The robustness of the method is studied using crystal structures with ligands known to be incorrectly modelled as well as a wider data set of 63 crystal structures with ligand fit to electron density from the Twilight [2] database. Results show that BPMD can successfully discriminate between compounds whose binding pose is not supported by the electron density from those with well-defined electron density. We expect that this protocol will enable modelers to choose high-quality ligand protein crystal structures for the progression of structure-based drug design projects. 1. Clark, A.J., et al., Prediction of Protein-Ligand Binding Poses via a Combination of Induced Fit Docking and Metadynamics Simulations. J Chem Theory Comput, 2016. 12(6): p. 2990-8. 2. Weichenberger, C.X., E. Pozharski, and B. Rupp, Visualizing ligand molecules in Twilight electron density. Acta crystallographica. Section F, Structural biology and crystallization communications, 2013. 69(Pt 2): p. 195-200.
ST-15. Carlo Martinotti
Curtin University "Development of new enhanced sampling approaches for the prediction of the free energy of interaction of small molecules and peptides with cell membranes"
Carlo Martinotti, Evelyne Deplazes and Ricardo L. Mancera Understanding the interactions and binding affinities of small drug-like molecules with biological membranes is important in fields such as pharmacology, toxicology and rational drug design. Development of molecular simulation protocols that can provide an accurate prediction of the free energy of binding (ŒîGb) of small molecules to lipid membranes remains a challenging task in computational biophysics. Approaches such as umbrella sampling suffer from insufficient configurational sampling of the molecule, resulting in the system becoming trapped in local energy minima for very long times, preventing simulations from reaching convergence and/

abstract: amma annual medal lecture

Documents