The COVID-19 pandemic, caused by SARS-CoV-2, has had a profound global impact. The viral main protease (Mpro) is essential for replication and transcription, making it a promising target for therapeutic intervention. To examine how ligand binding affects Mpro function, I applied elastic network models to analyze its conformational flexibility and gain insight into its underlying dynamics.
[Picture]
Representation of protein structure as a coarsegrained elastic network (M. Kim, Jang, and Jeong 2006).
Elastic Network Models (ENMs) represent proteins as particles connected by springs based on a reference structure. Coarse-grained ENMs typically use one particle per amino acid (α-carbon), with connections determined by a distance cutoff (usually within 7 to 10 Å). There are two ways in which ENMs are commonly represented and analyzed: Gaussian Network Model and Anisotropic Network Model.
When only uniform (isotropic) fluctuations around a reference structure are considered, system behavior can be captured by the magnitude of each particle’s displacement without regard to direction. Under this assumption, the conformational space of a system containing N particles is reduced to N dimensions. Network models based on this framework, known as Gaussian Network Models (GNMs), apply normal mode analysis to estimate global fluctuation patterns across a molecule. In GNMs, molecular dynamics are described by an N × N matrix that reflects particle connectivity and effective stiffness.
When directional information is included, the conformational space expands to 3N dimensions, as each particle’s motion along the x, y, and z axes is taken into account. These models, referred to as Anisotropic Network Models (ANMs), incorporate both the orientation of inter-particle springs and changes in their length, with force directions aligned along spring vectors. ANMs are also analyzed using normal mode analysis to characterize molecular motions.
The Gaussian Network Model (GNM) examines protein flexibility by measuring the magnitude of atomic fluctuations, giving an overall view of how the molecule moves. In contrast, the Anisotropic Network Model (ANM) considers both the magnitude and direction of movements, providing a more detailed picture of protein dynamics.
[Picture]
Anisotropic Network Model use an elastic mass-and-spring network to represent biological macromolecule.
Proteins exhibit various types of molecular movements. In this work, we focused on predicting hinge movements, which involve the rotation of protein domains around a hinge region, typically a loop or linker.
[Picture]
The arrow represents the hinge axis.
Hinges usually consist of a few residues that undergo significant conformational changes, while most of the rotating domains remain largely unchanged. In contrast, shear movements involve sliding motions of protein parts relative to each other. Elastic network models can identify hinge regions—flexible areas around which domains rotate—and rigid regions that remain stable. Hinge rotations are critical for molecular binding, activation, or deactivation, making them structurally important for coordinated protein motions.
In my research, I identified key residues and hinge points in SARS-CoV-2 Mpro that are critical for its inhibition and highlight potential allosteric sites. Comparison with experimental and computational literature confirmed that several of these residues are already known to play roles in allosteric regulation, underscoring the relevance of our findings. Additionally, we discovered novel residues that may be important for inhibiting Mpro function. Analysis of the enzyme’s intrinsic dynamics further revealed key slowest modes governing its conformational changes. Together, these results provide valuable insights into Mpro’s structural and dynamic features, which could guide the design of targeted therapeutics.