We tested the hypothesis that RBPs drive the formation of RNP granules in cells by a process known as phase separation. We further tested the model that aberrant phase transitions of these RBPs from a liquid to a solid-like state may be closely tied to the pathogenesis associated with diseases such as ALS.
First, we attempted to rebuild bona fide RNP granules in the test tube. We focused on stress granules that form when stressed cells shut down translation and release mRNA molecules from polysomes. Previous work had shown that stress granule formation is dependent on the RNA-binding protein G3BP. We showed that under non-stressed conditions, the intrinsically disordered acidic tracts and positively charged arginine-rich regions of G3BP are sequestered by electrostatic intramolecular interactions to stabilize a compact, auto-inhibited state. Upon release from polysomes, unfolded mRNA molecules outcompete the auto-inhibitory interaction, engendering a conformational transition that facilitates assembly of G3BP through protein-RNA interactions into networked RNA/protein condensates. We used mutagenesis to identify a sequence-encoded molecular grammar underlying the driving forces of phase separation of FUS and related prion-like proteins and we tested aspects of this grammar in cells. We found that phase separation is primarily governed by multivalent interactions among tyrosine residues from prion-like domains and arginine residues from RNA-binding domains, which are modulated by negatively charged residues. Glycine residues enhanced the fluidity, whereas glutamine and serine residues promoted hardening. Based on this insight, we developed a model showing that the measured saturation concentrations of phase separation are inversely proportional to the product of the numbers of arginine and tyrosine residues. These results provide us with an improved molecular understanding of disease-related mutations in RBPs and suggest that it will soon be possible to predict phase separation properties based on amino acid sequences.We have also been able to understand the environmental triggers that lead to stress granule assembly, e.g. during heat shock. Heat shock induces the disassembly of the translation initiation complex eIF4F, where eIF4G and eIF4E assemble into translationally arrested mRNA ribonucleoprotein particles (mRNPs) and heat shock stress granules, whereas eIF4A promotes HS translation. We could show that a conformational rearrangement of the thermo-sensing eIF4A-binding domain of eIF4G dissociates eIF4A and promotes the assembly with mRNA into mRNPs, which recruit additional translation factors, including Pab1p and eIF4E, to form multi-component condensates. This allowed us for the first time to propose a model for the assembly of heat-induced multicomponent stress granules. In another study we focused on the role of mRNA in regulating the material properties of RNP granules (work being prepared for submission). We could show that RNA-binding proteins such as G3BP1 are highly dynamic in reconstituted RNP granules, while mRNA is not. We demonstrated that the RNA dynamics are determined by trans RNA-RNA interactions that influence the overall properties and stability of reconstituted SGs. We then performed a biochemical characterization of the RNA helicase and stress granule component DDX3X. We revealed that DDX3X is capable of attenuating RNA-RNA interactions inside reconstituted SGs, resulting in more dynamic destabilized SG. We conclude that helicases such as DDX3X regulate the stability of SGs through modulating RNA-RNA interactions to control the translatability of mRNAs.We further demonstrated that the phase behavior of RBPs in vitro critically depends on the RNA concentration: low RNA/protein ratios promote phase separation into liquid condensates, whereas high ratios prevent condensate formation. Moreover, reduction of RNA levels in cells causes excessive phase separation and the formation of pathological solid-like structures. This means that prion-like RBPs such as TDP43 and FUS are normally soluble in the nucleus but form solid pathological aggregates when mislocalized to the cytoplasm. Based on these findings, we proposed that the nucleus is a buffered system in which high RNA concentrations keep RBPs soluble. These new insights constituted a critical advance because they suggested pathways for how diseases such as ALS and FTD could arise.
We have also been able to reveal how aggregates of the ALS-linked nuclear protein TDP-43 arise in cytosolic stress granules, and we identified the triggers of cytosolic TDP-43 aggregation. We found that TDP-43 aggregation requires a double event: up-concentration in stress granules beyond a threshold, and simultaneous oxidative stress. These two events collectively induce intra-condensate demixing, giving rise to a dynamic TDP-43 enriched phase within stress granules, which subsequently transitions into pathological aggregates.
We obtained further evidence that the first line of defense against aberrant protein misfolding and aggregation inside RNP granules is formed by molecular chaperones. We could show that diverse chaperone systems are recruited to RNP granules to maintain their liquid-like properties and prevent aggregation of disease-causing proteins such as TDP-43 and FUS.
In summary, we have shown that aberrant phase transitions within RNP granules lie at the heart of age-related diseases such as ALS and that the cellular chaperone machinery specifically targets and controls these processes.
Dissemination of the results has been in multiple journal publications. The results were also presented at various conferences and seminars.
