Predicting disease in tubulin: investigating isotypes, evaluating computational methods and exploring pathogenic mechanisms
From assisting division to molecular transport, cells depend on microtubules for several vital functions. Their fundamental nature is reflected in the extensive links between disease and mutations in their building blocks: heterodimers of tubulin-α and -β proteins. The first reported tubulin mutations were solely associated with neurodevelopment. Since then, disorders including bleeding disorders and infertility have also been attributed to tubulins, with several isotypes in the tubulin family now implicated in disease. However, the mechanisms governing how these mutations give rise to disease remain unclear. This thesis integrates computational and experimental strategies to address this question. First, I focused on comparing isotypes using both approaches. Bioinformatic tools allowed me to evaluate how tubulin isotypes differ in tissue expression, sequence divergence and evolutionary constraint. I then complemented this by utilising cell biology techniques to investigate the capability of tubulin isotypes to be incorporated into microtubules. Through this work, I could also interrogate strategies for tagging tubulin isotypes in vivo. Next, I systematically analysed mutations across the tubulin family to understand how they can result in disease with the view to improve predictions of their phenotypic effects. Variant effect predictors (VEPs) and structural analyses showed an inadequate ability to distinguish between pathogenic mutations and putatively benign variants. Moreover, their performance was markedly worse in genes linked to diseases outside of neurodevelopment. I then highlighted the mutations that had the poorest mutations so I could characterise them experimentally. Mutant tubulin incorporated into microtubules successfully, indicating that their pathogenic phenotype results from dominant-negative effects. Finally, I interrogated mutations in TUBβ4B associated with primary ciliary dyskinesia (PCD), a disease newly ascribed to tubulin. As these mutations varied in structural location and predicted effect, they also provided an excellent case study for identifying differences in pathogenic mechanisms. Indeed, experimental characterisation in vitro showed that mutations occurring close to the heterodimeric binding interface and predicted more likely to be pathogenic (P259S and P259L) disrupted heterodimer formation. In contrast, a mutation predicted more poorly (P358S) close to the microtubule lumen formed a stable heterodimer, suggesting it may exert dominant-negative effects. In vivo experiments confirmed these differences in terms of microtubule incorporation. Overall, this thesis highlights tubulin mutations as a blind spot for pathogenicity prediction, and this is potentially attributed to their ability to cause disease through dominant-negative effects.