The intrinsically disordered transcriptional activation domain of CIITA is functionally tuneable by single substitutions: An exception or a new paradigm?

Abstract

During protein evolution, some amino acid substitutions modulate protein function (“tuneability”). In most proteins, the tuneable range is wide and can be sampled by a set of protein variants that each contains multiple amino acid substitutions. In other proteins, the full tuneable range can be accessed by a set of variants that each contains a single substitution. Indeed, in some globular proteins, the full tuneable range can be accessed by the set of site-saturating substitutions at an individual “rheostat” position. However, in proteins with intrinsically disordered regions (IDRs), most functional studies – which would also detect tuneability – used multiple substitutions or small deletions. In disordered transcriptional activation domains (ADs), studies with multiple substitutions led to the "acidic exposure" model, which does not anticipate the existence of rheostat positions. In the few studies that did assess effects of single substitutions on AD function, results were mixed: The ADs of two full-length transcription factors didnotshow tuneability, whereas a fragment of a third AD was tuneable by single substitutions. Here, we tested tuneability in the AD of full-length human class II transactivator (CIITA). Sequence analyses and experiments showed that CIITA’s AD is an IDR. Functional assays of singly-substituted AD variants showed that CIITA’s function was highly tuneable, with outcomes not predicted by the acidic exposure model. Four tested positions showed rheostat behaviour for transcriptional activation. Thus, tuneability of different IDRs can vary widely. Future studies are needed to illuminate the biophysical features that govern whether an IDR is tuneable by single substitutions.

Rheostatic contributions to protein stability can obscure a position's functional role

“…Third, of the ZmPYK positions tested, none were rheostat positions; instead, the substitution variants at many positions were evenly split between those with WT-like affinity and those that were inactive (Page et al, 2022). This novel substitution phenotype (named "WT/inactive split") was not previously observed in any of our prior studies investigating substitutions in nine natural and eight engineered proteins (Fenton et al, 2020(Fenton et al, , 2022Hodges et al, 2018;Martin et al, 2020;Meinhardt et al, 2013;Ruggiero et al, 2021Ruggiero et al, , 2022Sousa et al, 2016;Sreenivasan et al, 2024;Swint-Kruse et al, 2021Swint-Kruse & Fenton, 2024;Wu et al, 2019).…”

“…However, in experiments, none of the predicted ZmPYK positions showed rheostatic outcomes for K app-PEP (Page et al, 2022). Furthermore, the substitution phenotypes for nine positions did not align with any of the previously-observed phenotypes (rheostat, neutral, or toggle) (Campitelli et al, 2021;Fenton et al, 2020Fenton et al, , 2022Hodges et al, 2018;Martin et al, 2020;Meinhardt et al, 2013;Miller et al, 2017;Ruggiero et al, 2021Ruggiero et al, , 2022Sreenivasan et al, 2024;Swint-Kruse, 2016;Swint-Kruse et al, 2021Swint-Kruse & Fenton, 2024;Wu et al, 2019). Instead, substitutions at these positions were almost evenly split between wild-type-like and inactive substitution variants (the "WT/inactive split" phenotype).…”

“…Several unexpected outcomes were observed in the ZmPYK dataset: First, a high fraction (47%) of the substituted proteins were inactive, which was much higher than in other proteins assessed for rheostat positions ($10%) (Fenton et al, 2020(Fenton et al, , 2022Hodges et al, 2018;Martin et al, 2020;Meinhardt et al, 2012Meinhardt et al, , 2013Ruggiero et al, 2021Ruggiero et al, , 2022Sousa et al, 2016;Sreenivasan et al, 2024;Swint-Kruse et al, 2021;Wu et al, 2019). Second, of the active ZmPYK substitution variants, values for K app-PEP spanned <10-fold whereas this parameter spanned >600-fold for hLPYK substitution variants (Page et al, 2022;Swint-Kruse et al, 2023).…”

“…One complicating factor is that a set of nonconserved positions can contain both “neutral” and “rheostat” positions (Miller et al, 2019; Swint‐Kruse et al, 2021). Substitutions at neutral positions have no effect upon protein function (Hodges et al, 2018; Martin et al, 2020), whereas substitutions at a rheostat position modulate, but seldom abolish, protein function (Fenton et al, 2020; Hodges et al, 2018; Meinhardt et al, 2013; Sreenivasan et al, 2024; Swint‐Kruse et al, 2021). Indeed, the defining characteristic of a rheostat position is that a series of different substitutions sample a wide functional range.…”

“…Indeed, the defining characteristic of a rheostat position is that a series of different substitutions sample a wide functional range. Such “functional” rheostat positions are common: They have been identified in proteins from bacteria to humans and in many classes of proteins (globular soluble, integral membrane, and intrinsically disordered) (Campitelli et al, 2021; Fenton et al, 2020, 2022; Hodges et al, 2018; Meinhardt et al, 2013; Ruggiero et al, 2021, 2022; Sreenivasan et al, 2024; Swint‐Kruse et al, 2021, 2023; Wu et al, 2019); rheostat positions can also be prevalent within a protein, with at least two model proteins containing >30% rheostat positions (Meinhardt et al, 2013; Sousa et al, 2016; Swint‐Kruse et al, 2021, 2023). Thus, our understanding of which protein features discriminate between rheostat and neutral positions, as well as the underlying biophysics that govern rheostat positions, must be improved in order to advance such far‐ranging fields as protein engineering and natural variant interpretation.…”