Les Houches-TSRC Workshop on Protein Dynamics

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) macrodomain within the nonstructural protein 3 counteracts host-mediated antiviral adenosine diphosphate-ribosylation signaling. This enzyme is a promising antiviral target because catalytic mutations render viruses nonpathogenic. We conducted a massive crystallographic screening and computational docking effort, identifying new chemical matter primarily targeting the active site of the macrodomain. X-ray data collection to ultra-high resolution and at physiological temperature enabled assessment of the conformational heterogeneity around the active site. Neutron diffraction data is guiding hydrogen placement to improve docking calculations. Several hits have promising activity in solution and provide starting points for development of potent SARS-CoV-2 macrodomain inhibitors. The role of entropy in modulating binding affinity will also be discussed.

Machine learning and high-performance simulation are helping us to make significant progress in all fields of science, and protein biophysics is no exception. In this talk, I will give an update of the state of the art in machine learning for protein biophysics and specifically focus on rare event sampling and detection, learning coarse-grained molecular dynamics force fields from all-atom simulations, and generating protein structures and thermodynamics via generative deep learning.

The essential role of protein dynamics for enzyme catalysis has become more generally accepted. Since evolution is driven by organismal fitness hence the function of proteins, we are asking the question of how enzymatic efficiency has evolved. I will describe how combining the advantages of each technique, NMR, x-ray crystallography inclusion ensemble refinement, MD simulations and Ancestral Sequence Reconstruction (ASR) lead to a thorough description of the energy landscape dictating function.

First, I will address the evolution of enzyme catalysis in response to one of the most fundamental evolutionary drivers, temperature. Using ASR, we answer the question of how enzymes coped with an inherent drop in catalytic speed caused as the earth cooled down over 3.5 billion years. Tracing the evolution of enzyme activity and stability from the hot-start towards modern hyperthermophilic, mesophilic and psychrophilic organisms illustrates active pressure versus passive drift in evolution on a molecular level. Second, I will share a novel approach to visualize the structures of transition-state ensembles (TSEs), that has been stymied due to their fleeting nature despite their crucial role in dictating the speed of biological processes. We determined the transition-state ensemble in the enzyme adenylate kinase by a synergistic approach between experimental high-pressure NMR relaxation during catalysis and molecular dynamics simulations. Third, in “forward evolution” experiments, we discovered how directed evolution reshapes energy landscapes in enzymes to boost catalysis by nine orders of magnitude relative to the best computationally designed biocatalysts. The underlying molecular mechanisms for directed evolution, despite its success, had been illusive, and the general principles discovered here (dynamic properties) open the door for large improvements in rational enzyme design. Finally, visions (and success) for putting protein dynamics at the heart of drug design are discussed.

FRET spectroscopy and imaging can provide state-specific information on the structure and dynamics of complex dynamic biomolecular assemblies under ambient conditions with nanosecond time resolution and single-molecule sensitivity. To overcome the sparsity of FRET experiments, we developed procedures to combine these with computer simulations to map biomolecular dynamics and to resolve quantitative integrative structure models at a precision and accuracy better than 3 Å. The integrative structure models are deposited in the new protein data bank, PDB-dev [1-3]. Moreover, we combined super-resolution microscopy via stimulated emission depletion (STED) and Multi-parameter Fluorescence Image Spectroscopy (MFIS) [4] to reach molecular resolution with sub-nanometer precision in molecular imaging of biomolecules and their complexes. While STED-MFIS captures the spatial and temporal information of the cellular context with a resolution below 10 nm, the concurrent measurement of Förster resonance energy transfer (FRET) between an excited donor and acceptor provides a zoom with Ångström precision. Thus, integrative super-resolution FRET image spectroscopy exploits these synergies to reach molecular resolution.

I will introduce the concepts of our novel optical tools and demonstrate recent applications: (1) Detection of a so far hidden functionally important conformational state in the enzyme T4 Lysozyme [5], (2) Resolving the conformational transitions of Guanylate binding proteins (GBPs) during GTP-controlled phase transition to exert their function as part of the innate immune system of mammalian cells [6]. (3) Mapping the dynamic exchange network in chromatin fibers by studying a 12-mer nucleosome array [7].

[1] Kalinin et al.; Nat. Methods 9, 1218-1225 (2012).

[2] Dimura et al.; Curr. Opin. Struct. Biol. 40, 163–185 (2016).

[3] Dimura et. al. Nat Commun. 11, e5394 (2020).

[4] Weidtkamp-Peters et al.; Photochem. Photobiol. Sci. 8, 470-480 (2009).

[5] Sanabria et. al. Nat Commun. 11, e1231 (2020).

[6] Kravets et. al.; eLife 5, e11479 (2016).

[7] Kilic et al.; Nat. Commun. 9, 235 (2018).

Classical molecular dynamics (MD) simulations based on atomic models play an increasingly important role in a wide range of applications in physics, biology and chemistry. The approach consists of constructing detailed atomic models of the macromolecular system and, having described the microscopic forces with a potential function, using Newton's classical equation, F=MA, to literally "simulate" the dynamical motions of all the atoms as a function of time. The calculated trajectory, though an approximation to the real world, provides detailed information about the time course of the atomic motions, which is impossible to access experimentally. While great progress has been made, producing genuine knowledge about biological systems using MD simulations remains enormously challenging. Among the most difficult problems is the characterization of large conformational transitions occurring over long time scales. Issues of force field accuracy, the neglect of induced polarization in particular, are also a constant concern. A powerful paradigm for mapping the conformational landscape of biomolecular systems is to combine free energy methods, transition pathway techniques and stochastic Markov State Model based massively distributed simulations. These concepts will be illustrated with a few recent computational studies of Src tyrosine kinases, K+ channels, and the P-type ion pumps.