Pathological C-terminal phosphomimetic substitutions alter the mechanism of liquid-liquid phase separation of TDP-43 low complexity domain.

UNLABELLED: C-terminally phosphorylated TAR DNA-binding protein of 43 kDa (TDP-43) marks the proteinaceous inclusions that characterize a number of age-related neurodegenerative diseases, including amyotrophic lateral sclerosis, frontotemporal lobar degeneration and Alzheimer's disease. TDP-43 phosphorylation at S403/S404, and especially at S409/S410, is in fact accepted as a biomarker of proteinopathy. These residues are located within the low complexity domain (LCD), which also drives the protein's liquid-liquid phase separation (LLPS). The impact of phosphorylation at these LCD sites on phase separation of the protein is a topic of great interest, as these post-translational modifications and LLPS are both implicated in proteinopathies. Here, we employed a combination of experimental and simulation-based approaches to explore this question on a phosphomimetic model of the TDP-43 LCD. Our turbidity and fluorescence microscopy data show that Ser-to-Asp substitutions at residues S403, S404, S409 and S410 alter the LLPS behavior of TDP-43 LCD. In particular, in contrast to the unmodified protein, the phosphomimetic variants display a biphasic dependence on salt concentration. Through coarse-grained modeling, we find that this biphasic salt dependence is derived from an altered mechanism of phase separation, in which LLPS-driving short-range intermolecular hydrophobic interactions are modulated by long-range attractive electrostatic interactions. Overall, this in vitro and in silico study provides a physiochemical foundation for understanding the impact of pathologically-relevant C-terminal phosphorylation on the LLPS of the TDP-43 in a more complex cellular environment.

STATEMENT OF SIGNIFICANCE: Proteinaceous inclusions composed of phosphorylated, C-terminal TDP-43 fragments have long been recognized as hallmarks of several neurodegenerative diseases, in particular amyotrophic lateral sclerosis and frontotemporal dementia. A rapidly growing number of studies indicate that these proteinopathies may be closely related to liquid-liquid phased separation (LLPS) of TDP-43, but the impact of phosphorylation on TDP-43 LLPS remains largely unexplored. In this study we used a combination of experimental methods and coarse-grained simulations to ascertain, in mechanistic terms, how phosphorylation at pathologically-critical C-terminal sites impacts liquid-liquid phase separation of the low complexity domain of TDP-43. Our results broaden our understanding of the mechanisms driving pathogenic process in these neurodegenerative diseases.