Session A08: Physics of Proteins I: Structure & Dynamics of Proteins

Modeling Protein Dynamics from Brownian Motions to Transcription Regulation

Protein factors and enzymes largely employ Brownian motions for movements along DNA. How to achieve biological regulations out of thermal motions are of high interest as we model the proteins from atomic to coarse-grained (CG) and from molecular dynamics (MD) to stochastic simulations. Notably, we have revealed recently at atomic-scale molecular diffusion of a small transcription factor (TF) domain protein with 1-bp spontaneous stepping along DNA. Extensive samplings and kinetic model construction further show that hydrogen bonding dynamics for synchronization at the protein-DNA interface are rate limiting and likely serve for sequence information detection. The protein 1D diffusional search along DNA can be trapped upon the protein reorientation on specific DNA sequences. CG-MD to stochastic dynamics modeling then suggest that the protein stepping size can vary from one to several bps depending on protein-DNA interactions or protein responses to DNA sequence patterns. The level of stochasticity in the protein diffusion can nevertheless be regulated e.g., by varying solution ionic concentrations. The corresponding protein diffusional free energy profiling along DNA can be constructed and the ruggedness of the profiling can be tuned.  We additionally show that a dimeric TF protein coordinates its two DNA binding domains/regions to achieve ‘inchworm’ stepping on the DNA, and lowering the solution ionic concentration may trigger the dimer dissociation into monomers on DNA. By demonstrating similar Brownian motions for RNA polymerase translocation on DNA, we emphasize that the thermal fluctuations can be nicely employed by such an enzyme to respond sensitively to identities or energetic biases of incoming nucleotide substrates for “selective ratcheting” or transcription fidelity control.

    

Binding mechanism of different variants of SARS-CoV(-2) virus with hACE2 receptor protein

We have investigated the binding mechanism of different variants of SARS-CoV(-2) virus with hACE2 receptor protein. Understanding binding mechanism of different variants of SARS-CoV-2 virus with its receptor is essential for designing effective drug/vaccine for this virus. Classical molecular dynamics (MD) simulation has been carried out for the purpose. In our study, three important binding regions between different variants of SARS-CoV(-2) and hACE2 have been identified. Further relative strengths of binding of different variants of SARS-CoV(-2) virus with hACE2 have been estimated via study of hydrogen bonding, salt bridges, electrostatics and van der Waals interactions. The average number of hydrogen bonds in wild type, delta, mu and omicron variants are estimated to be 7, 7, 5 and 5 respectively. We have also estimated binding free energy for wild type, delta, mu and omicron variants with hACE2 by MM/GBSA method and the values are - 44.15 ± 8.07 kcal/mol, - 44.93 ± 7.69 kcal/mol, - 37.31 ± 7.98 kcal/mol and -36.20 ± 7.61 kcal/mol respectively. Moreover estimation of free energy by Umbrella sampling for wild type and delta variants shows the higher free energy of binding for the delta variants than the wild type.   

Geometric Signatures of Switching Behavior in Mechanobiology

The proteins involved in cell's mechanobiological processes have evolved specialized and surprising responses to applied forces. One example is catch-slip bonding, where a protein-protein bond switches from increasing strength to decreasing strength under increasing force. Another example is force-induced pathway switching, where a multi-pathway biochemical transformation switches from one pathway to another under applied force. These force-activated switching behaviors are important in various biological functions, from cell adhesion and mechanosensing, to molecular motors, proofreading, and antigen discrimination. We develop a theoretical framework that unifies these switching behaviors and identifies the signatures of a system's free energy landscape that generate specific switches. Remarkably, we find that almost every 2-dimensional bond will show catch-slip behavior under an appropriate pulling force—no specialized mechanisms are required. We use this framework to identify the signatures of switching in established catch bond models and we propose course-grained free energy landscapes for P-selectin, integrin, and actin/myosin catch bonds based on experimental data. Our framework suggests design principles for engineering novel bond behaviors and provides clues how sophisticated bonding mechanisms may have evolved from simple bonds.   

Protein dielectrophoresis in an exact nanotrap

Dielectrophoresis is the process by which a gradient of the squared electric field imparts motion to a particle. Interest in the phenomenon of dielectrophoresis has gained significant attention in recent years due to its potential in the sorting, manipulation, and trapping of solutes such as proteins in aqueous solutions. For many decades protein dielectropheresis was considered impossible, as the predicted magnitude of the force arising from experimentally accessible field strengths could not out-compete thermal kicks. This conclusion was drawn from the mainstay Clausius-Mossotti result for the susceptibility of the dielectrophoretic force. However, the CM result is inappropriate for proteins. By accounting for the repulsive core of the protein and the polarization of the solvent by its permanent dipole, a corrected susceptibility may be obtained. Here we propose an explicit geometry within which the correction may be put to the test. Trapping distances are explored as a function of the applied field strength and pore size.   

Dynamic Allosteric Modulation of G-Protein-Coupled Receptors

G-protein-coupled receptors (GPCRs) are the largest superfamily of human membrane proteins and represent primary targets of ~1/3 of currently marketed drugs. Allosteric modulators have emerged as more selective drug candidates compared with orthosteric ligands. However, many X-ray and cryo-EM structures of GPCRs resolved so far exhibit negligible differences upon binding of allosteric modulators. Mechanism of dynamic allosteric modulation of GPCRs remains unclear. In this talk, I will present our recent studies in allosteric modulation of adenosine and muscarinic GPCRs by combining cutting-edge cryo-electron microscopy, Gaussian accelerated molecular dynamics (GaMD) simulations and pharmacology experiments^1,2. In addition, we have applied the GaMD, Deep Learning and free energy prOfiling Workflow (GLOW) to map dynamic changes in free energy landscapes of GPCRs upon binding of allosteric modulators. Available high-resolution experimental structures of allosteric modulator-bound class A and B GPCRs were collected and a number of additional computational models were generated by changing target receptors of the modulators to different subtypes. All-atom GaMD simulations were performed for a total of 66 µs on 44 GPCR systems in the presence/absence the modulator. Comprehensive Deep Learning and free energy calculations of the GaMD simulations have provided important mechanistic insights into dynamic allostery of GPCRs and modulator selectivity, which shall greatly facilitate rational design of selective GPCR allosteric drugs.

    

Analysis of Zika virus’ electrostatic features

The Zika (ZIKV) virus is one of the most dangerous flavivirus diseases, which is usually transmitted by the bite of infected mosquitoes.  ZIKV can cause several disorders in adults which are mostly mild. However, it is of higher concern in pregnant women since it can develop deformations in the neonates’ brains and other organs. The study of Envelope (E) protein is fundamental since is the primary focus of the immune response against Zika. Therefore, the study of its structural characterization and the conformation changes during the infection is important to find potential cures or vaccines. By using several computational methods, we calculated electrostatic properties that allowed us to understand the interactions among multiple E proteins on the Zika virus capsid. The findings in our study may help the future drug design for the Zika virus.

    

Markovian Modeling of Coupled Global and Internal Dynamics in Freely Diffusing Molecules

We present progress on a Langevin formalism for coarse-graining of macromolecules that allows for viscous coupling between global diffusion and internal dynamics as well as an inertial transition between underdamped and overdamped internal motions. Obtaining accurate friction coefficients from observed trajectories is crucial to understanding the significance of both effects, and so the friction must be parameterized in a manner consistent with both types of motion. Meanwhile the intramolecular potential must be consistent with the global symmetry and average properties of the rotational coupling. We use a generalized Einstein Relation for interacting particles with memory to retrieve the long-time Markovian friction from lagged correlations between phase space variables. We discuss numerical instability at long times in systems with free global diffusion and bound internal degrees of freedom, due to saturated internal diffusion, and show a numerically stable route to the friction. We also demonstrate how a harmonic internal potential can be fit to fluctuations in the internal degrees of freedom in a manner consistent with the rotational coupling. We demonstrate accurate reproduction of time correlation functions from a toy simulation of a partially overdamped Markovian trimer. Additionally, our body-fixed formalism allows us to demonstrate that the dynamical features and long-time tail exponents of the internal motion can be modified by coupling with the global diffusion.   

Deciphering active site strain

Strain at the active site of proteins has been invoked to understand processes ranging from enzyme catalysis to energy storage in transient intermediate states during the first steps of vision. We are developing Raman Optical Activity (ROA), based on differences between Raman scatter using left- versus right-circularly polarized light, as an approach to quantitatively determine dihedral angle distortions at the active site of chromophoric proteins. Previously, we showed that pre-resonance conditions offer a unique spectral window to obtain structurally informative spectroscopic data, and reported that the ROA amplitude of selected vibrational modes can function as spectroscopic rulers for specific active site dihedral angles. In this work, quantum chemical calculations form an essential bridge connecting available crystallographic data with vibrational spectra. Currently, we are developing tools that can determine if high-resolution features of chromophoric active sites reported in crystal structures are viable structural models based on their ability to reproduce observed Raman data.   

Amide-I Vibrational Entanglement for a 2-State Crossing Point

Vibrational Manifolds of the Amide I band in both crystalline analogs and protein films show a curious sharp isosbestic point in absorbance as a function of temperature, indicating that teh amide-I decolazied system can exsist in 2 discrete states. Narrow frequency 4-wave mixing experiments at the isosbestic point indicate extremely long dephasing times, suggesting that when the system is excited precisely at the isosbestic point it exists as a quantum entangled superposition of 2 discrete states of extreme isolation from the environment.   

Computational modelling of protein structure and stability using molecular dynamics simulations with enhanced sampling

Protein structure and stability are frequently studied in vitro, while proteins naturally occur in crowded cellular environments that include other proteins, DNA, and numerous organelles. As a result, we lack a quantitative understanding of protein structure and stability in cells. In this work, we carry out extensive molecular dynamics simulations with enhanced sampling methods based on potential energy rescaling and Hamiltonian replica exchange to determine the accuracy of measurements of the melting temperature T_m. The ability to accurately measure the melting temperature will allow us to determine changes in protein structure and stability caused by different cellular environments. We show that we can accurately determine the change in melting temperature ΔT_m in vitro between wildtype peptides and those with point mutations, e.g. chignolin and cln025 and wwdomain and several single-point mutants. We also illustrate the usage of a small number of FRET experimental restraints in MD simulations to pin down the complete structure of the peptides. These techniques will be applied to proteins in the presence of large crowding molecules in future studies.