Developing New Viral Inhibitors:
The Role of Multiscale Dynamics

Introduction to Dengue

Dengue virus (DENV) and Zika virus (ZIKV) are members of the flavivirus genus and infect hundreds of millions of people a year. Currently there are no approved drugs against ZIKV or highly effective vaccines against all DENV serotypes. A detailed understanding of the structural and dynamical aspects of the various components of DENV and ZIKV virus is being pursued in Singapore in an interdisciplinary effort that engages 14 different experimental and computational groups.

A Flavivirus is like an onion. The outermost layer is made up of proteins called the envelope (E) protein. The next layer of the onion is a lipid bilayer, within which parts of the E proteins along with membrane (M) proteins are embedded. Finally, the inner core of the onion contains capsid (C) proteins in complex with a mass of RNA; the RNA contains the genetic information of the virus.

The virus infects cells of the immune system. The mechanisms behind this process of infection are partly understood. First the virus is encapsulated in small compartments of cells called endosomes. It is within these compartments that a lowering of the pH occurs that triggers a mechanical/structural response within the outermost layer of the virus. The E proteins, which were initially lying parallel to the surface of the viral sphere in the form of dimers, undergo structural changes and transition to a trimeric form (Figure 1). This structural rearrangement is accompanied by exposure of the hydrophobic tips of the E protein, referred to as the fusion peptide (FP) regions. This exposure subsequently initiates fusion of the viral and the immune cell membranes, leading to release of the viral genome into the cytoplasm and hence infection. The pattern of amino acids that make up FP is conserved across all flaviviruses (these include dengue virus, west nile virus, yellow fever virus, Japanese encephalitis, tick-borne encephalitis) and naturally is a potential target for antiviral drugs that aim to block viral fusion and hence infection.


Figure 1. Structure of the DENV. (A) The cross section of a coarse-grained model of an entire dengue virus envelope, recently developed at the Bioinformatics Institute, A*STAR. DENV is covered with 90 E protein dimers (blue and purple) which are embedded in a lipid bilayer (yellow and black); (B) Structure of a single dimeric E protein unit colored according to its domains: DI (red), DII (yellow), DIII (blue). The dimer undergoes pH triggered rearrangement forming (C) trimers. DII consists of the (D) hydrophobic fusion peptide (green) which initiates interaction with the immune cell resulting in infection.

The Project

We know that the function of macromolecules in biology is intimately linked to their structure, mediated through dynamical changes. However, these changes occur over multiple time scales: thermal fluctuations in the range of femtoseconds (fs), picoseconds (ps) and nanoseconds (ns), and global rearrangements, folding, and assembly processes such as antibody binding in the range of microseconds (μs) to milliseconds (ms) (Figure 2). Each of these motions can be captured by different experimental and computational techniques. Wouldn't it be nice to integrate them seamlessly: to have a rheostat that can be tuned to any time window to obtain a movie of the associated molecular motions? The current project brings together the various practitioners who extend the timescales of their recording devices towards this aim. This will result in a detailed movie of the workings of the dengue virus (as shown in Figure 2) and hopefully guide the development of novel therapeutics. The methodological basis for the movie will rely primarily upon the use of molecular dynamics (MD) simulation - this represents a "computational microscope" that enables us to zoom in on the jiggling of atoms and molecules.


Figure 2.Time scale of experimental and computational techniques (top) together with a summary of multiscale modelling / simulation approaches (bottom), currently applied in the Bioinformatics Institute, A*STAR.

In the Bioinformatics Institute (BII) A*STAR, multiscale MD simulation approaches1,2 allow our scientists to probe biologically and experimentally relevant time and length scales, by application/development of both atomically-resolved and simplified coarse-grained models (Figure 2). Coarse-graining is achieved by representing all-atom systems with a reduced number of degrees of freedom, i.e. groups of atoms are represented as beads (Figure 2) and validated against all-atom simulations and experimental data. This powerful tool is illustrated by a simulation model of the entire dengue virus envelope, recently developed in our group (Figure 1).3 On the other end of the spectrum, we recently used fully atomistic simulation methods to study the conformational dynamics of the E protein FP region, a major epitope for potential antibodies against dengue, and found results in excellent agreement with data from fluorescence spectroscopy experiments.4 Likewise, the thermodynamics of the interaction of FP with lipid bilayers are being investigated via parallel computational and experimental approaches, providing novel insights into the mechanisms of viral/host membrane fusion.5 In addition atomistic approaches were employed in order to understand the virus C protein dynamics 6 as well as its autoinhibitory role of the disordered N-Terminus mediating DENV C interaction with biological targets.7 At the same time, we looked at the reversibility of the virus "breathing" and expansion upon changes in temperature and presence of divalent cations. Our results showed that for DENV the intrinsic dynamics but not the specific morphologies are correlated to viral infectivity.8 More recently, we have extended these models to understand how recognition of specific epitopes on immature virus particles by host antibodies can exacerbate pathogenesis, yielding a molecular rationale for the so-called phenomenon of “antibody-dependent enhancement” that can result in the most serious forms of dengue infection.9 In parallel, we are investigating with workers at Duke-NUS how dengue epidemics result from mutations that modulate interactions of viral lipoprotein particles with the immune system10, and exploring the potential of novel synthetic peptidomimetic compounds11 developed with researchers at BTI (A*STAR) to inhibit viral dynamics and host interactions. At the same time computational approaches for solvent mapping are being developed to reveal new druggable cryptic pockets.12 We have also reported the genome sequences of ZIKV strains from two cases in Singapore and found through phylogenetic analyses that these strains form an earlier branch distinct from the recent large outbreak in the Americas.13

Summary

The current project combines experiments and molecular simulations for studying DENV (single proteins, complexes, whole virus), producing details of the virus at unparalleled spatial and temporal resolution. The emerging methodology will enable the application of new protocols we are developing at the Bioinformatics Institute (BII) A*STAR to probe the fluctuating surfaces of the virus in its various constituent states with a view to unravelling cryptic pockets/crevices that can subsequently be drugged with inhibitors of the viral motions and hence life cycle. Using dengue as an example and combining both experimental and computational results at different time and spatial scales will provide new insights into viral mechanisms and a platform for interrogating other complex biological systems, including related viruses such as ZIKV.

Principal investigators: Dr Peter J. Bond and Chandra Verma (Bioinformatics Institute, BII, A*STAR).

Funding notes: 2014-2018: the Ministry of Education in Singapore (MOE AcRF Tier 3 Grant Number MOE2012-T3-1-008). 2018-2021: CRP Grant Number NRF-CRP19-2017-03).

References

  1. Huber RG, Marzinek JK, Holdbrook DA, Bond PJ. "Multiscale molecular dynamics simulation approaches to the structure and dynamics of viruses", 2016, Progress in Biophysics & Molecular Biology, S0079-6107(16)30081-5.
  2. Sharma KK, Marzinek JK, Tantirimudalige SN, Bond PJ, Wohland T. "Single-molecule studies of flavivirus envelope dynamics: Experiment and computation", 2018, Progress in Biophysics & Molecular Biology, S0079-6107(16)30081-5.
  3. Marzinek JK, Holdbrook DA, Huber RG, Verma C, Bond PJ. "Pushing the Envelope: Dengue Viral Membrane Coaxed into Shape by Molecular Simulations", 2016, Structure, 24:1410-1420.
  4. Marzinek JK, Lakshminarayanan R, Goh E, Huber RG, Panzade S, Verma C, Bond PJ. "Characterizing the Conformational Landscape of Flavivirus Fusion Peptides via Simulation and Experiment", 2016, Scientific Reports, 5, 19160.
  5. Marzinek JK, Bag N, Huber RG, Holdbrook DA, Wohland T, Verma C, Bond PJ. "A Funneled Conformational Landscape Governs Flavivirus Fusion Peptide Interaction with Lipid Membranes", 2018, Journal of Chemical Theory Computation, 14(7):3920-3932.
  6. Boon PLS, Saw WG, Lim XX, Raghuvamsi PV, Huber RG, Marzinek JK, Holdbrook DA, Anand GS, Gruber G, Bond PJ. "Partial Intrinsic Disorder Governs the Dengue Capsid Protein Conformational Ensemble", 2018, ACS Chemical Biology, 13(6):1621-1630.
  7. Faustino AF, Guerra GM, Huber RG, Hollmann A, Domingues MM, Barbosa GM, Enguita FJ, Bond PJ, Castanho MARB, Da Poian AT, Almeida FCL, Santos NC, Martins IC. "Understanding Dengue Virus Capsid Protein Disordered N-Terminus and pep14-23-Based Inhibition", 2015, ACS Chemical
  8. Sharma K, Lim XX, Tantirmudalige SN, Gupta A, Marzinek JK, Holdbrook DA, Lim XYE, Bond PJ, Anand GS, Wohland TS. "Infectivity of dengue virus serotypes 1 and 2 is correlated to E protein intrinsic dynamics but not to envelope conformations", 2018, Structure, accepted.
  9. Wirawan M, Fibriansah G, Marzinek JK, Lim XX, Ng TS, Sim AYL, Zhang Q, Kostyuchenko VA, Shi J, Smith SA, Verma CS, Anand G, Crowe JE Jr, Bond PJ, Lok SM. "Mechanism of Enhanced Immature Dengue Virus Attachment to Endosomal Membrane Induced by prM Antibody" , 2018, Structure, S0969-2126(18)30370-8.
  10. Chan KW, Watanabe S, Jin JY, Pompon J, Teng D, Alonso S, Vijaykrishna D, Halstead SB, Marzinek JK, Bond PJ, Burla B, Torta F, Wenk MR, Ooi EE, Vasudevan SG. "Dengue disease severity caused by elevated levels of secreted NS1 arising from a single T164S substitution" , 2019 Science Translational Medicine, Accepted.
  11. Lim TC, Cai S, Huber RG, Bond PJ, Chia PXS, Khou SL, Gao S, Lee SS, Lee SGS. "Facile Saccharide-free Mimetics that Recapitulate Key Features of Glycosaminoglycan Sulfation Patterns", 2018, Chemical Science. 9:7940-7947.
  12. Tan YS, Spring DR, Abell C, Verma C "The Application of Ligand-Mapping Molecular Dynamics Simulations to the Rational Design of Peptidic Modulators of Protein-Protein Interactions", 2015, Journal of Chemical Theory Computation, 11:3199-3210.
  13. Maurer-Stroh S, Mak T, Ng Y, Phuah S, Huber RG, Marzinek JK, Holdbrook DA, Lee RT, Cui L, Lin RT "South-east Asian Zika virus strain linked to cluster of cases in Singapore" , 2016, Eurosurveillance, 21(38).

    Figure 3. On the cover:2 Dengue virus is responsible for millions of infections per year. Exposure to low pH within the endosomes triggers fusion with the viral envelope and release of the genome. We have performed computer simulations of the entire dengue envelope protein complex embedded within a vesicle lipid bilayer. The cross-section of the resultant model (shown on the cover) overlays closely with cryo-electron microscopy maps and reveals in unprecedented detail how envelope proteins globally organize and locally curve the viral lipid bilayer in preparation for membrane fusion.