Substitutions at non-conserved, rheostat positions in intrinsically disordered regions of CIITA modulate its function.

Abstract

Personalized medicine requires understanding which exome changes cause medically relevant phenotypes. To predict exome phenotypes, existing algorithms rely on textbook rules for amino acid substitutions that were derived from decades of studies biased towards conserved positions. Recent studies targeting non-conserved positions found a special class of “rheostat” positions. Substitutions at these positions exhibited a range of functional outcomes that were not predicted by existing rules. Understanding such rheostat positions is critical to improve predictions. To that end, we tested rheostat positions in globular proteins and found substitutions altered protein dynamics. Here, we extended our studies to the extremely dynamic and non-conserved intrinsically disordered (ID) proteins using the Class II transactivator (CIITA) as our model. CIITA is the master transcriptional activator of the MHC class II genes. Sequence analyses and experiments show that the N-terminal transcriptional activation domain of CIITA has positions with high and low probabilities for ID. This region has putative binding for interaction with multivalent partners. We hypothesized that rheostats would be in positions with low ID probabilities. We measured the effects of 82 single substitutions across 7 positions on (i) expression, (ii) transcriptional activation, and (iii) overall biological output. We observed modest effects on expression, identified strong rheostats for transcriptional activation and overall biological output. Surprisingly, rheostats were found in both regions of low and high ID probabilities. Furthermore, substitutions at just 7 of the 1130 positions sampled nearly the entire range of possible transcription activation levels. These results show that this multi-valent, highly non-conserved, ID protein has a substitution sensitivity that is similar to globular soluble and integral membrane proteins. Finally, retrospective inspection of alanine scans of other transcription activation domains suggest that strong rheostat positions exist in other many other IDPs.