The Difference in Structural States between Canonical Proteins and Their Isoforms Established by Proteome-Wide Bioinformatics Analysis

Abstract

Alternative splicing is an important means of generating the protein diversity necessary for cellular functions. Hence, there is a growing interest in assessing the structural and functional impact of alternative protein isoforms. Typically, experimental studies are used to determine the structures of the canonical proteins ignoring the other isoforms. Therefore, there is still a large gap between abundant sequence information and meager structural data on these isoforms. During the last decade, significant progress has been achieved in the development of bioinformatics tools for structural and functional annotations of proteins. Moreover, the appearance of the AlphaFold program opened up the possibility to model a large number of high-confidence structures of the isoforms. In this study, using state-of-the-art tools, we performed in silico analysis of 58 eukaryotic proteomes. The evaluated structural states included structured domains, intrinsically disordered regions, aggregation-prone regions, and tandem repeats. Among other things, we found that the isoforms have fewer signal peptides, transmembrane regions, or tandem repeat regions in comparison with their canonical counterparts. This could change protein function and/or cellular localization. The AlphaFold modeling demonstrated that frequently isoforms, having differences with the canonical sequences, still can fold in similar structures though with significant structural rearrangements which can lead to changes of their functions. Based on the modeling, we suggested classification of the structural differences between canonical proteins and isoforms. Altogether, we can conclude that a majority of isoforms, similarly to the canonical proteins are under selective pressure for the functional roles.

Per Aspera ad Chaos: Vladimir Uversky’s Odyssey through the Strange World of Intrinsically Disordered Proteins

“…In addition to conventional predictors, which often make use of compositional bias analysis, one increasingly finds artificial intelligence-based programs that enable accurate and rapid analysis of entire proteomes. Four articles can be found in this section of SI [ 23 , 24 , 25 , 26 ].…”

Re-evaluating the impact of alternative RNA splicing on proteomic diversity

“…For a given gene, the most expressed transcript is usually defined as coding for the canonical protein. This canonical status is determined based on the transcript expression across different tissues of an organism, the conservation of its exon combination with other species, and/or the existence of a functional role for the protein ( Osmanli et al, 2022 ). Compared to their canonical proteins, isoform proteins can lose or gain certain domains and, therefore, can differ in their functional properties by the alteration of localization signals, sequences for post-translational modifications, or interaction with other proteins ( Kriventseva et al, 2003 ; Stamm et al, 2005 ; Leoni et al, 2011 ; Light and Elofsson 2013 ).…”

Predicting the Structural Impact of Human Alternative Splicing

“…In contrast, the impact of alternative splicing on protein structure has been studied much less. Previous studies have shown that alternative splicing can induce unstable protein conformations 10 , change protein localization 4 , alter transmembrane domains 11 , and create variations in repeat regions 12,13 . However, these studies have been relatively constrained in scope 13 , and a thorough and systematic investigation of how splicing affects structure is imperative.…”