Modeling and simulation of intrinsically disordered proteins

Abstract: Proteins are large, complex molecules that play many critical roles in the body and are required for the structure, function, and regulation of the body’s tissues and organs. In very simple terms, a protein can be defined as a linear chain of subunits called amino acid residues. The individual amino acid residues are sequentially bonded together by peptide bonds. There are 20 standard amino acids in nature and their specific order of appearance in a protein chain is thought to determine its structure and function. For a very long time it was assumed that the structure of a protein and its function were mutually inclusive, and exceptions to the norm were often either “swept under the rug” or branded as mere curiosities of little relevance. Around the turn of the millennium, mounting evidence of structural disorder in a considerable amount of otherwise perfectly functional proteins lead to a change in paradigm. It is now known that structural disorder is not only abundant in all species, as it is also an advantage for proteins involved in functions which benefit from structural malleability.Despite the considerable, recent interest in the study of intrinsically disordered proteins (likely due to their implication in a number of human diseases), the lack of a well-defined structure represents a substantial obstacle to their structural characterization by classic, high-resolution experimental methods. Some lower resolution methods can provide information about the average shape and size of the collection of structures that a disordered protein can attain in solution. However, computational methods are generally necessary to aid in interpreting and complementing the information that can be obtained from experimental data. One of such methods is the computer simulation of proteins, where a computer model of the protein is run for a certain period of time, in order to observe and register the most relevant spatial arrangements which the protein may adopt and freely convert between. It is from this collection of arrangements that interesting structural and thermodynamic properties can be calculated, making computer simulations very powerful tools.The papers included in this thesis deal with the development, validation and application of computer simulation models for flexible and disordered proteins, both in solution and at interfaces. In Paper II it was found that a simple physical model can be used to mimic the properties of flexible proteins, helping to understand how and why these proteins adsorb to surfaces under certain conditions. In Paper III, the same simple model shown that two disordered proteins from different sources (saliva and milk) have very similar properties in solution and when adsorbed to surfaces. Thus, it was hypothesized that it may be possible to use one of them as a substitute for the other under a pharmaceutical context. Paper I was the catalyst for a series of studies (Papers IV – VI) involving more detailed protein models. Among other things, this study provided an indication that the atomistic models used until then, for the simulation of proteins with well-defined structures, may not be applicable to their disordered counterparts. This was later confirmed in Paper IV, by evaluating several such models against experimental evidence. A similar evaluation was conducted for two new independent approaches developed with disordered proteins in mind. The results (presented in Papers IV and V) were shown to be in excellent agreement with each other and with experiment, which represents a considerable step forward in the search for accurate and predictive models for the simulation of disordered proteins. Finally, in Paper VI, one of the new atomistic models was used to perform the structural characterization of a disordered peptide conjugated to a small molecule, which has been shown to possess promising therapeutical applications. The value of computer simulations is well illustrated in this study, as the insight obtainable from experiment is limited and it is only through the analysis of the simulations that a possible link between the average conjugate structure and its increased antifungal activity was established.