Investigating the Contribution of Disordered Domains to the Biological Activity of RNA-binding Proteins

Abstract

Many proteins contain disordered domains under physiological conditions. These disordered regions may be functional, although under pathological conditions they may lead to protein aggregation and degradation, as observed in proteins related to neurodegenerative diseases. In my thesis study, I aimed to understand how the primary sequence of these proteins encodes for the diverse ensemble of conformations rather than a stable folded state. I focused on the role of disordered domains in the activity of RNA-binding proteins involved in post-transcriptional regulation, but may lead to pathogenesis in many diseases. The human TIS11 proteins bind to AU-rich elements in the 30 UTR of mRNAs through a CCCH-type tandem zinc finger (TZF) domain. Mutations in these proteins have been linked to cancer. A member of this protein family, Tristetraprolin (TTP), is partially unfolded in the C-terminal zinc finger in the apo state, but folds upon RNA binding. The homolog protein TIS11d is folded in both free and bound states. Previous studies have shown that the extent of structure of the TZF domain in the apo state does not affect the affinity to target RNA in vitro, however it modulates the activity of the protein in cell. To understand which interactions determine the zinc affinity of the C-terminal zinc fingers of TTP and TIS11d, I investigated the stability of their TZF domains using homology modeling and molecular dynamics (MD) simulations. I found that, in the C-terminal zinc finger of TIS11d, a hydrogen bond is necessary to allow for [pi-[pi] stacking between the side chains of a conserved phenylalanine and the zinc-coordinating histidine. Using mutagenesis and nuclear magnetic resonance (NMR) spectroscopy, I demonstrated that the lack of this hydrogen bond is responsible for the reduced zinc affinity, and thus lack of structure, of the C-terminal zinc finger in TTP. These results suggest that the CCCH-type TZF domain in different proteins have evolved to differentiate their function through a disorder-to-order transition. In Caenorhabditis elegans several RNA-binding proteins contain a TZF domain homologous to the RNA-binding domain of TIS11 proteins, but have different RNA-binding specificity. I characterized the structure and the dynamics of the C. elegans protein MEX-5 using NMR spectroscopy and MD simulations. I found that MEX-5, like its mammalian counterpart TTP, contains a zinc finger that is partially unfolded in the free state but that folds upon RNA-binding. To assess if the disorder-to-order transition upon RNA-binding contributes to MEX-5 function, I designed a variant MEX-5 where both zinc fingers are stably folded in the absence of RNA. I characterized the RNA-binding activity of this variant MEX-5 and I found that the binding affnity and specificity are unchanged compared to the wild type protein. Together with Ryder's lab, we used CRISPR-hr to introduce this variant into the endogenous C. elegans mex-5 locus. Homozygotes animals are sterile, form massive uterine tumors within a few days of reaching adulthood, and often die by bursting. These results show that the unfolded state of MEX-5 is critical to its function in vivo by a mechanism distinct from its RNA-binding activity. To further investigate how the equilibrium between structural order and disorder affects the function of a protein in the cell, I focused on the human protein TDP-43, a major component of the cellular proteinaceous aggregates found in amyotrophic lateral sclerosis and other neurodegenerative diseases. Previous studies have shown, both in vitro and in vivo, that the second RNA recognition motif (RRM2) of TDP-43 domain contains peptide regions that are particularly prone to fibril formation. In addition, RRM2 has been shown to populate, to a small degree, one or more partially folded states under native conditions. To determine if the partially folded states of TDP-43 RRM2 contribute to the formation of aggregates observed in the human diseases, I characterized the structures of these states using MD simulations including enhanced sampling methods and restraints from experimental chemical shifts. I found that in these states the protein exposes to the solvent aggregation-prone regions that are instead buried in the protein core in the native state. These results suggest a role in fibrogenesis for the transient partially folded states of TDP-43 RRM2.