Organised Chaos: defining different degrees of intrinsic disorder by molecular dynamics methods

Intrinsically disordered proteins (IDPs) and regions (IDRs) are defined as proteins, or proteins’ regions, that lack a stable 3D structure. The nature of the IDP(R)s amino acid residues and their sequence prevents these systems from folding. IDPs and IDRs are ubiquitous in biology, playing many different roles that range from mechanical, see for example elastin, to scaffolding the structure of macromolecular assemblies, such as the XPA in the Nucleotide Excision Repair (NER) pathway, to regulatory and signalling functions, such as p53 and short linear motifs (SLiMs)-containing regions. Trying to understand the relationships between structure and functions in these highly flexible and dynamic systems is rather challenging, because their intrinsic flexibility makes them extremely difficult to characterize experimentally through structural biology methods, such as X-ray crystallography, NMR, or Cryo-EM. Indeed, the actual classification of protein domains as “intrinsically disordered” derives from the inability of determining 3D atomic coordinates from experimental sources. According to the dogma underlying the whole field of structural biology, all the different functions that IDP(R)s are key player in should be linked to their structure, however this structure is undecipherable within the experimental timescale. The overarching design of my thesis work was to understand these structure-to-function relationships in specific IDP(R)s by molecular simulation techniques. My main goal was to understand how different levels of residual secondary structure propensity in specific IDP(R)s systems of importance in health and disease can explain their macroscopic function and more specifically, how structural disorder can regulate molecular recognition at the atomistic level of details. Molecular simulation techniques have now come to age and are well suited to play a starring role in structural biology, not only as a support for experimental methods, but also as equal partners and/or as the primary research tool for discovery. In this thesis I show how different molecular simulations techniques can be successfully used for discovery in the structure and function of IDP(R)s a) on their own, b) as an equal partners to experiments, c) as tools to understand experimental data and d) for the ad hoc design of new experiments. More specifically, we found that in many systems, namely in XPA and in the p53 and ECSIT C-terminal domains, molecular recognition can be explained in terms of a distinct residual secondary structure propensity, which results in an enrichment of the disordered conformational ensemble with secondary structure motifs that may act as molecular recognition features (MoRFs), or nucleation sites. According to this hypothesis, specific MoRFs are recognized by receptors and fold upon binding, within a scheme that lies between the two extremes of “conformational selection” and “induced fit”. As shown for XPA, the structure of these MoRFs obtained through computing can inform the design of macrocyclic systems that can be used to inhibit specific receptors or pathways in the absence of any other structural information, in view of the development of new diagnostic and/or therapeutic strategies. In this work I also show cases where the propensity to form MoRFs, also termed pre-structuring, is not required. More specifically, in the case of the very short SLiM LxCxEcontaining peptides, which can reach low nM binding affinities for the Retinoblastoma (Rb) protein, we have not detected any residual secondary structure propensity (or MoRF) in the unbound peptides. This particular study was conducted in partnership with experiments, through extensive sampling simulations for the bound and unbound peptides, in the presence of counterions or no counterions, and in the case of phosphorylation and in the absence of phosphorylation. Interestingly, in agreement with experiment, our simulations show that phosphorylation increases the level of polyproline II (PPII) structure in the unbound peptides, which has been also recently underlined as an important in signalling pathways. In summary, I believe that this research work on a few examples of structurally disordered systems contributes to shed some light into intrinsic disorder in biomolecular recognition and how intrinsic disorder should not be classed as one grey area, but instead viewed as incorporating many different shades of conformational degrees and diversity that modulate binding affinity through their structure, via enthalpy, and relative stability, via entropy.