Distributions of molecular conformations and interactions revealed by EPR spectroscopy - methodology and application to hnRNPA1

Proteins are long chains of peptide-bond linked amino acids, which have the fascinating ability to fold into a huge variety of three-dimensional shapes. To describe peptide sequences that have particularly large conformational flexibility, the umbrella terms intrinsically disordered domains (IDDs), resp. proteins (IDPs) have emerged. Many different combinations of amino acid composition (both polar, or hydrophobic) and sequence have been identified as IDDs, with the unifying feature that the domains are enriched in only a few types of amino acids, thus also referred to as low complexity domains (LCDs). The functions and interactions have recently been shown to be important for example in cellular processes, such as stress response, which has initiated a new era in structural biology. Building on the well-established structural characterisation of folded domains, the aim of biophysical characterisation extends also to understanding the dynamics and intrinsic conformational flexibility of proteins. A multitude of biophysical and biochemical methods have been developed for the former aspect, which include X-ray crystallography, nuclear magnetic resonance (NMR), single molecule fluorescence techniques, and more recently, high resolution cryo-electron microscopy to elucidate biomolecular structures. Structural characterisation of proteins containing IDDs, however, is notoriously difficult precisely due to the large conformational space and faster inter-changing of conformations. An additional experimental complication is that disordered protein domains often have a strong tendency to interact with other biomolecules, or with themselves, making it virtually impossible to study isolated molecules. In this thesis we present methods based on electron paramagnetic resonance (EPR) spectroscopy to characterise biomolecular structures and interactions in highly disordered systems, where a large distribution of states is encountered. As a biologically relevant application we demonstrate how well-established pulsed dipolar spectroscopy (PDS) methods can be applied for the structural characterisation of the partially disordered human protein 'heterogeneous nuclear ribonucleoprotein A1' (hnRNPA1). Approximately the last carboxy-terminal third of hnRNPA1 is an LCD, enriched in glycine and arginine residues. It is known, for example from solution state NMR, that the LCD of hnRNPA1 is an IDD, and exhibits fast dynamics at ambient temperature. Furthermore, the IDD of hnRNPA1 is known to mediate a process known as liquid-liquid phase separation (LLPS) both in vivo (under conditions of stress), and in vitro. We use site-directed spin labelling and a combination of continuous wave and pulsed EPR spectroscopy to address the structural characterisation of hnRNPA1 in vitro. We identify protein-protein interactions by pulsed double electron-electron resonance (DEER) experiments with singly spin-labelled hnRNPA1, and we determine buffer conditions in which we are able to study predominantly monomeric hnRNPA1. These stabilised conditions consequently allowed us to perform pair-wise distance measurements between spin labelling sites in the folded domains of hnRNPA1 and sites in the IDD by DEER experiments on doubly spin-labelled hnRNPA1. The probabilistic interpretation of the DEER experiments in terms of distance distributions enabled us to use a large set of distance restraints for the generation of an ensemble model of full length hnRNPA1. This ensemble model can be validated against additional, independent biophysical experiments. Following the characterisation of the free state of the protein we present experimental results for the characterisation of samples of full length hnRNPA1 after inducing LLPS. They are of particular interest in combination with RNA-binding experiments, since it is known that RNA binding and LLPS are tightly linked processes for many protein involved in RNA processing, which is also what we found for hnRNPA1. In the experiments using RNA and hnRNPA1 we encountered situations where it is a great benefit to be able to detect multiple pair-wise distance in a spin-labelled sample. PDS with spectroscopically orthogonal spin labels has been developed as a powerful extension of conventional distance measurements between like spin labels, and offers promising opportunities towards this goal. In the methodological sections of this thesis we describe method development towards the application of spectroscopically orthogonal spin labels for the routine characterisation of biomolecules and biomolecular complexes. We discuss existing approaches with PDS and spectroscopially orthogonal spin labels, in particular the DEER experiment between Gd(III)-based spin labels and nitroxides. The RIDME technique is an emerging complementary method to the currently predominantly used DEER technique in may situations. In particular, we describe a detailed analysis of the performance of RIDME for distance measurements between nitroxide and Cu(II)-based spin labels using experiments with molecular rulers. A broadband microwave pulse version of the RIDME experiment was introduced in this context. Broadband coherent excitation of the nitroxide spin allowed both one-step orientation averaging, and EPR-correlated RIDME experiments. Building on the Cu(II)-nitroxide experiments, we performed experiments with molecular rulers for nitroxide high-spin pairs (in particular with Gd(III) and Mn(II) metal ions). Additional considerations must be made when using the RIDME experiments with high spin metal ion-based spin labels for distance analysis, due to contributions of harmonic overtones of the dipolar coupling frequency