Functional Tuning of Intrinsically Disordered Regions in Human Proteins by Composition Bias

Intrinsically disordered regions (IDRs) in protein sequences are flexible, have low structural constraints and as a result have faster rates of evolution. This lack of evolutionary conservation greatly limits the use of sequence homology for the classification and functional assessment of IDRs, as opposed to globular domains. The study of IDRs requires other properties for their classification and functional prediction. While composition bias is not a necessary property of IDRs, compositionally biased regions (CBRs) have been noted as frequent part of IDRs. We hypothesized that to characterize IDRs, it could be helpful to study their overlap with particular types of CBRs. Here, we evaluate this overlap in the human proteome. A total of 2/3 of residues in IDRs overlap CBRs. Considering CBRs enriched in one type of amino acid, we can distinguish CBRs that tend to be fully included within long IDRs (R, H, N, D, P, G), from those that partially overlap shorter IDRs (S, E, K, T), and others that tend to overlap IDR terminals (Q, A). CBRs overlap more often IDRs in nuclear proteins and in proteins involved in liquid-liquid phase separation (LLPS). Study of protein interaction networks reveals the enrichment of CBRs in IDRs by tandem repetition of short linear motifs (rich in S or P), and the existence of E-rich polar regions that could support specific protein interactions with non-specific interactions. Our results open ways to pin down the function of IDRs from their partial compositional biases.

Automated summary

Intrinsically disordered regions (IDRs) in protein sequences have low structural constraints and faster evolutionary rates. Their lack of evolutionary conservation limits the use of sequence homology in the classification and functional assessment of IDRs. IDRs require other properties for their classification and functional prediction. While composition bias is not necessary for IDRs, compositionally biased regions (CBRs) are commonly found within IDRs. This study investigates the overlap between IDRs and specific types of CBRs in the human proteome. Results show that 2/3 of residues in IDRs overlap with CBRs. CBRs enriched in certain amino acids tend to be fully included within long IDRs, partially overlap shorter IDRs, or overlap IDR terminals. CBRs also overlap more frequently with IDRs in nuclear proteins and proteins involved in liquid-liquid phase separation. Analysis of protein interaction networks reveals the enrichment of CBRs in IDRs through tandem repetition of short linear motifs and the presence of E-rich polar regions that support specific protein interactions with non-specific interactions. These findings provide a new approach to infer the function of IDRs based on partial compositional biases.