In silico modelling and assessment of key functional amino acid residues on disease-related proteins: the case study of ceruloplasmin

Nowadays, bioinformatics has a key role in several scientific fields due to the necessity of managing big data, extracting meaningful information from the large omics dataset available, and reducing the increasingly high research costs. This aspect is particularly important in plasma-derived product (PDMP) production for rare or orphan diseases, characterized by expensive and time-consuming processes along with the employment of a resource (i.e., human plasma) that is limited, requires optimization, and ethical use. The usage of industrial waste fractions can partially address these limitations, supporting the maximization of the resource, the prioritization of experiments, and the cost reduction to identify novel drugs. Computational methods can be effectively employed in several steps of the process, and integrated with experimental approaches in developing new PDMPs or improving their therapeutical employment, through the production of recombinant forms. Here, a Ph.D. project in collaboration with Kedrion Biopharma S.p.A and as part of FSE-REACT-EU programme, PON “Ricerca e Innovazione” (“Dottorati su tematiche Green”) is presented. The Ph.D. project focused on identifying novel therapeutic candidates for rare diseases from plasma samples through industrial waste plasma fractions analysis and characterization, integrating bioinformatics and in vitro experimental methods, by exploiting a collaboration between the Departments of Computer Science and Pharmacy of the University of Pisa. In particular, several computational approaches have been employed to conduct a deeper analysis of waste fractions’ proteomic content and evaluate the effects of phosphorylation on ceruloplasmin (CP) structural stability and functionality to prioritize the production of recombinant CP. Specifically, in Chapter 1, plasma samples derived from Kedrion industrial waste fractions were analyzed, deepening a previous study that prioritized CP as a promising novel PDMP [58]. Starting from already published proteomics data, the analysis was focused on pathways and PPIs and supported the choice of CP for additional investigations. Indeed, in the subsequent chapters, computational approaches were employed to study CP structure and function. In particular, in Chapter 2, phosphomimetic/abrogative mutations were introduced in the protein’s sequence, the 3D structures were predicted, and molecular dynamics simulations (MDS) were conducted to evaluate the mutations’ effects on CP stability. Then, in Chapter 3, the mutated isoforms upon MDS were employed in docking studies to better understand phosphorylation impact on PPIs and CP resistance to proteolytic cleavage. Overall, how phosphorylation can affect the protein’s structure and function was evaluated, comprehensively considering changes in stability as well as in activity by analyzing CP PPIs with two different serine-proteases (i.e., trypsin and plasmin) and one of its main interaction partners, namely myeloperoxidase. The in silico studies suggested how phosphomimetic/abrogative mutations affected CP stability and functionality differently depending on the introduced amino acid and the involved positions. Overall, phosphomimetic isoforms had stronger effects on CP structural organization and were associated with a potentially higher resistance to proteolysis. Different bioinformatics approaches were successfully employed in deepening the knowledge about plasma industrial waste, considering the complexity of the biological matrix with an in-depth analysis level that would not have been accessible only using traditional methods. In addition to the studies related to CP and phosphorylation effects on the protein’s structural organization and function, an appendix is presented regarding a drug repurposing study focused on neuroinflammation. Considering the importance of bioinformatics in the drug repositioning fields, a systems biology approach was here applied through a previously validated computational pipeline, and the identified repurposed candidates were subsequently tested on two different in vitro models of human inflamed microglia.