Modeling and Simulation Proposes New Insights for SARS-CoV-2


In our previous blog, we discussed the use of predictive modeling tools to build initial atomic-level structures of potential drug targets (e.g., proteins) and to refine regions that were not experimentally determinable (See Video). These tools include the addition of hydrogen atoms and flexible loops that are sometimes not capable of being resolved experimentally. We examined this in the context of the cryo-electron microscopy (cryo-EM) of the SARS-CoV-2 spike (S) protein recently published in the journal Science (DOI: 10.1126/science.abb2507).

In this blog, we will detail how molecular modeling and simulation of refined structural models such as the SARS-CoV-2 S protein can assist in the generation of new hypotheses for the discovery and design of putative therapeutics to treat COVID-19.

Drug Binding Is Dependent on Structural Changes

In living systems, proteins naturally exist as dynamic entities. Their dynamics often predetermine their function. The physicist Richard Feynman once said:

“If we were to name the most powerful assumption of all, which leads one on and on in an attempt to understand life, it is that all things are made of atoms and that everything that living things do can be understood in terms of the jigglings and wigglings of atoms.” 1

The SARS-CoV-2 S protein is no exception to Feynman’s principle. Prior to entering human cells, the S protein binds a receptor referred to as angiotensin-converting enzyme 2 (ACE2).2 The receptor-binding domain (RBD) is part of the S protein that binds ACE2 (Figure 1A). The RBD can exist in at least two primary conformational states called the up (receptor-accessible) and down (receptor-inaccessible) states (Figure 1B). When the RBD is in the up state, the S protein is more “open” to facilitate the binding of ACE2.2 Studies have suggested that the down, receptor-inaccessible state, is more stable.2 This implies that putative therapeutics such as small organic molecules capable of binding RBD could stabilize the RBD in the down state and prevent the virus from interacting with ACE2; thus, stopping COVID-19 from infecting people.

Figure 1.
Figure 1. (A) Crystal structure of the receptor-binding domain (RBD) (blue) in complex with ACE2 (green). (B) Superimposition of the SARS-CoV-2 Spike protein in the up (red: PDB 6VYB) and down (blue: PDB 6VXX) states. (C) Homology model of the RBD (blue) with the modeled missing loops (red) and predicted hinge binding region.


The RBD of the S Protein Is Like a Hinge on a Door

A flexible linker connects the RBD to the remaining S protein. The flexibility of the linker enables RBD to transition from the down to the up state through a hinge-bending motion (Figure 1B). From the SARS-CoV-2 S protein, we have extracted the RBD with the associated flexible linker and adjacent domain (Figure 1C). We used the RBD structure from the S protein in the up state (PDB 6VYB), but this structure lacks three loops potentially important for binding ACE2 (Figure 1C). As a result, we had to build a homology model of the RBD with the flexible linker using the cryo-EM structure from which the truncation was originally made (PDB 6VYB) with an additional template. The additional template was the crystal structure of the RBD alone in complex with ACE2 (PDB 6M17) (Figure 1A).

The RBD structure contains loops including one greater than 20 amino acid residues, which are not present in the open state structure (PDB 6VYB). Two of these loops form the interaction with ACE2; thus, homology modeling is necessary to understand the interactions of the Spike protein. We can then assign hydrogen atoms in appropriate protonation states to mimic physiological conditions such as pH.

We could then perform a molecular dynamics (MD) simulation to simulate the conformational transition and/or predict possible binding sites where putative small molecules may bind to disrupt the interaction of the S protein with ACE2. When we predicted possible binding sites using BIOVIA Discovery Studio for our homology model of the RBD with the flexible linker, we identified a binding site located at the hinge of the flexible linker (Figure 1C). We labelled this region the hinge and note that it may be worthy of further investigation for the discovery of putative therapeutics. If a small molecule were to bind at the hinge region, it could conceivably lock the RBD in a down state and thus prevent ACE2 binding.

Further Investigations

Extensive investigations including computational predictions and biological experiments might further clarify the utility of the hinge region. Examples of computational predictions could include normal mode analysis (NMA) and/or MD simulation.3 For example, a long timescale, approximately hundreds of nanoseconds, of MD simulation could allow scientists to sample multiple conformations of the hinge region.

On the other hand, an NMA could provide a coarse and quick estimation of the conformational transitions.3 Specific conformations from the MD simulation and/or NMA are starting points for high-throughput virtual screening of potential small molecule databases. Scientists could dock and score each small molecule to all conformations. They could then rank all the resulting poses and submit the best hits for experimental validation. Studies have shown that this method of computational drug discovery, often referred to as ensemble-based virtual screening, improves the chances of identifying drug candidates.4 The method also reflects the reality that drug binding depends on structural changes in the protein, as noted above. We believe that the preliminary results identified here are interesting and worthy of further investigation.

Secondly, we would like to note that the refined S protein structural model could be usable as a target for immunotherapeutics.5 Scientists could design monoclonal antibodies that bind to the SARS CoV-2 S proteins based on prior knowledge of the ACE2 binding site. They could then perform in silico affinity-maturation studies to improve the binding specificity.6

As an active supporter of the scientific community that is collaborating today on COVID-19 solutions, BIOVIA Dassault Systèmes develops BIOVIA Discovery Studio. This proven life sciences modeling and simulation environment brings together over 30 years of peer-reviewed research and world-class in silico techniques. The software provides scientists with a complete toolset for use from target identification through lead optimization, including tools for biologics design and analysis, classical simulations, structure- and fragment-based design, virtual ligand screening, as well as ADME and toxicity prediction.

As part of Dassault Systèmes’ corporate social responsibility, BIOVIA is pleased to offer qualifying academic research groups involved in SARS-CoV-2 related studies a no-charge, six-month license to BIOVIA Discovery Studio to assist them in the search for rapid, safe and effective therapeutic drug candidates against the SARS-CoV-2 virus.  If you are an academic researcher in this field, please request a software license and download. This offer will run through June 30, 2020.




  1. Ruth Nussinov, A. S. (2015, October 27). PLOS Blogs. Retrieved from
  2. Wrapp, Daniel, et al. “Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation.” Science 367.6483 (2020): 1260-1263.
  3. Greener, Joe G., and Michael J. E. Sternberg. “Structure-based prediction of protein allostery.” Current Opinion in Structural Biology 50 (2018): 1-8.
  4. Ellingson, Sally R., et al. “Multi-conformer ensemble docking to difficult protein targets.” The Journal of Physical Chemistry B 119.3 (2015): 1026-1034.
  5. Yuan, Meng, et al. “A highly conserved cryptic epitope in the receptor-binding domains of SARS-CoV-2 and SARS-CoV.” Science (2020).
  6. Cannon, Daniel A., et al. “Experimentally guided computational antibody affinity maturation with de novo docking, modelling and rational design.” PLoS Computational Biology 15.5 (2019): e1006980.