Repository logo
 

Multiscale Modelling of Biomolecular Phase Behaviour


Loading...
Thumbnail Image

Type

Thesis

Change log

Authors

Garaizar, Adiran 

Abstract

Elucidating the physicochemical laws that govern the phase behaviour of biomolecules is a cardinal milestone to understand crucial open questions in biology ranging from the spatiotemporal organisation of the cytoplasm in the eukaryotic cell to the emergence of pathological cellular states. Understanding the governing principles of protein and nucleic acid phase behaviour implies deciphering how atomic-scale processes modulate molecular interactions, and how such interactions, in turn, dictate the large-scale thermodynamic behaviour of biomolecules. In this thesis, Molecular Dynamics and Statistical Mechanics have been used to understand the driving forces that modulate protein collective behaviour along such wide-ranging spatial scales.

Simulations with a minimal coarse-grained model showed how enriching the structural ensemble of proteins in open, expanded conformations can promote their propensity to undergo liquid-liquid phase separation. This is because more expanded conformations favour inter-protein interactions which increase the molecular connectivity inside protein droplets, favouring their stability.

In addition, I present a pioneering chemically specific coarse-grained force field for intrinsically disordered proteins that explicitly considers the effect of solvent in phase separation of proteins. I find our model in near-quantitative experimental agreement both with single-protein structural descriptors as well as with available thermodynamic data of phase separated-protein condensates such as water and protein content in FUS condensates. The model enables studying the structure of the solvent inside protein droplets and recovers experimentally observed salt-inhibited and salt-driven phase separation.

Next, I focus on delineating the role of spontaneous beta-sheet folding – prime drivers for neurodisease-linked phase transitions – in the kinetic and transport properties of condensates as a function of the abundance and strength of such transitions. Using atomistic umbrella sampling simulations I estimated the free energy of binding when 8-residue NUP-98 sequence replicas remain disordered as opposed to when they undergo folding into a beta-sheet, finding that folding can increase the binding free energy up to an order of magnitude. Transferring such information from the atomistic to a coarse-grained model, it was established how the timescale between diffusion loss and droplet coalescence determines the shape of phase-separated protein condensates.

Finally, I study how disorder-to-order transitions can give rise to single component Fused in Sacroma (FUS) multi-phase condensates. Atomistic free energy calculations were used to establish that such structural transitions can increase four-fold the interaction between FUS sequences which are prone to spontaneous beta-sheet folding. I transferred these findings to a sequence-dependent coarse-grained model which predicted that proteins that underwent folding contribute to lower the interfacial free energy of the condensates. Moreover, our dynamical minimal model informed with the findings from the atomistic and coarse-grained scales predicts single component FUS multi-phase condensates with an inhomogenous organisation where a liquid-core of disordered proteins is surrounded by a gel-like shell of the proteins which underwent disorder-to-order transitions.

Taken together, this work sheds light on the molecular driving forces of protein phase behaviour that are crucial to understand cell function both in health and disease and proposes a pioneering approach that advances the realism and explanatory power of coarse-grained biomolecular simulations.

Description

Date

2022-01-04

Advisors

Сollepardo-Guevara, Rosana
Rene Espinosa, Jorge

Keywords

Molecular Dynamics

Qualification

Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
Sponsorship
Engineering and Physical Sciences Research Council (2513376)
EPSRSC grant agreement EP/N509620/1

Collections