Steven L. McKnight

Protein domains of low sequence complexity are unable to fold into stable, three-dimensional structures. In test tube studies, these unusual polypeptide regions can self-associate in a manner causing phase separation from aqueous solution. This form of protein:protein interaction has been implicated in numerous examples of dynamic morphological organization within eukaryotic cells. In several cases, the basis for low complexity domain (LCD) self-association and phase separation has been traced to the formation of labile cross-β structures. The primary energetic force favoring formation of these transient and reversible structures is enabled by polypeptide backbone interactions. Short, contiguous networks of peptide backbone amino groups and carbonyl oxygens are zippered together intermolecularly by hydrogen bonding as described by Linus Pauling seven decades ago. Here, we describe a simple, molecular biological method useful for the identification of localized, self-associating regions within larger protein domains of low sequence complexity.

Protein domains of low sequence complexity are unable to fold into stable, three-dimensional structures. In test tube studies, these unusual polypeptide regions can self-associate in a manner causing phase separation from aqueous solution. This form of protein:protein interaction has been implicated in numerous examples of dynamic morphological organization within eukaryotic cells. In several cases, the basis for low complexity domain (LCD) self-association and phase separation has been traced to the formation of labile cross-β structures. The primary energetic force favoring formation of these transient and reversible structures is enabled by polypeptide backbone interactions. Short, contiguous networks of peptide backbone amino groups and carbonyl oxygens are zippered together intermolecularly by hydrogen bonding as described by Linus Pauling seven decades ago. Here we describe a simple, molecular biological method useful for the identification of localized, self-associating regions within larger protein domains of low sequence complexity. Significance: This study describes a molecular biological method for analyzing protein domains of low sequence complexity in search of segments that mediate self-association and consequent phase separation both in vitro and in vivo . Small regions allowing for self-association correspond to sequences that specify the formation of labile cross-β structural order. When juxtaposed to the C-terminus of GFP, cross-β prone regions suppress fluorescence. A tiled scan of overlapping fragments of the low complexity domain (LCD) of the TDP-43 RNA binding protein pinpointed an evolutionarily conserved sequence of twenty amino acids essential for self-association, phase separation and the formation of nuclear speckles. The screening method described herein should be useful for the analysis of any LCD believed to function via homotypic self-association.

This perspective begins with a speculative consideration of the properties of the earliest proteins to appear during evolution. What did these primitive proteins look like, and how were they of benefit to early forms of life? I proceed to hypothesize that primitive proteins have been preserved through evolution and now serve diverse functions important to the dynamics of cell morphology and biological regulation. The primitive nature of these modern proteins is easy to spot. They are composed of a limited subset of the 20 amino acids used by traditionally evolved proteins and thus are of low sequence complexity. This chemical simplicity limits protein domains of low sequence complexity to forming only a crude and labile type of protein structure currently hidden from the computational powers of machine learning. I conclude by hypothesizing that this structural weakness represents the underlying virtue of proteins that, at least for the moment, constitute the dark matter of the proteome.

An evolutionarily conserved region of the TDP-43 low-complexity domain (LCD) twenty residues in length can adopt either an α-helical or β-strand conformation. When in the latter conformation, TDP-43 self-associates via the formation of a labile, cross-β structure. Self-association can be monitored via the formation of phase-separated protein droplets. Exposure of droplets to hydrogen peroxide leads to oxidation of conserved methionine residues distributed throughout the LCD. Oxidation disassembles the cross-β structure, thus eliminating both self-association and phase separation. Here, we demonstrate that this process reciprocally enables formation of α-helical structure in precisely the same region formerly functioning to facilitate β-strand-mediated self-association. We further observe that the α-helical conformation allows interaction with a lipid-like detergent and that exposure to lipids enhances the β-to-α conformational switch. We hypothesize that regulation of this oxidative switch will prove to be important to the control of localized translation within vertebrate cells. The experimental observations reported herein were heavily reliant on studies of 1,6-hexanediol, a chemical agent that selectively dissolves labile structures formed via the self-association of protein domains of low sequence complexity. This aliphatic alcohol is shown to exert its dissociative activity primarily via hydrogen-bonding interactions with carbonyl oxygen atoms of the polypeptide backbone. Such observations underscore the central importance of backbone-mediated protein:protein interactions that facilitate the self-association and phase separation of LCDs.

Summary: An evolutionarily conserved region of the TDP-43 low complexity domain twenty residues in length can adopt either an α-helical or β-strand conformation. When in the latter conformation, TDP-43 self-associates via the formation of a labile, cross-β structure. Self-association can be monitored via the formation of phase separated protein droplets. Exposure of droplets to hydrogen peroxide leads to oxidation of conserved methionine residues distributed throughout the low complexity domain. Oxidation disassembles the cross-β structure, thus eliminating both self-association and phase separation. Here we demonstrate that this process reciprocally enables formation of α-helical structure in precisely the same region formerly functioning to facilitate β-strand mediated self-association. We further observe that the α-helical conformation allows interaction with a lipid-like detergent, and that exposure to lipids enhances the β-to-α conformational switch. We hypothesize that regulation of this oxidative switch will prove to be important to the control of localized translation within vertebrate cells. The experimental observations reported herein were heavily reliant on studies of 1,6-hexanediol, a chemical agent that selectively dissolves labile structures formed via the self-association of protein domains of low sequence complexity. This aliphatic alcohol is shown to exert its dissociative activity primarily via hydrogen bonding interactions with carbonyl oxygen atoms of the polypeptide backbone. Such observations underscore the central importance of backbone-mediated protein:protein interactions that facilitate the self-association and phase separation of low complexity domains. Significance Statement: The TDP-43 protein is a constituent of RNA granules involved in regulated translation. TDP-43 contains a C-terminal domain of 150 amino acids of low sequence complexity conspicuously decorated with ten methionine residues. An evolutionarily conserved region (ECR) of 20 residues within this domain can adopt either of two forms of labile secondary structure. Under normal conditions wherein methionine residues are reduced, the ECR forms a labile cross-β structure that enables RNA granule condensation. Upon methionine oxidation, the ECR undergoes a conformational switch to become an α-helix incompatible with self-association and granule integrity. Oxidation of the TDP-43 low complexity domain is hypothesized to occur proximal to mitochondria, thus facilitating dissolution of RNA granules and activation of localized translation.

Father of molecular embryology.

Protein domains of low sequence complexity do not fold into stable, three-dimensional structures. Nevertheless, proteins with these sequences assist in many aspects of cell organization, including assembly of nuclear and cytoplasmic structures not surrounded by membranes. The dynamic nature of these cellular assemblies is caused by the ability of low-complexity domains (LCDs) to transiently self-associate through labile, cross-β structures. Mechanistic studies useful for the study of LCD self-association have evolved over the past decade in the form of simple assays of phase separation. Here, we have used such assays to demonstrate that the interactions responsible for LCD self-association can be dictated by labile protein structures poised close to equilibrium between the folded and unfolded states. Furthermore, missense mutations causing Charcot-Marie-Tooth disease, frontotemporal dementia, and Alzheimer’s disease manifest their pathophysiology in vitro and in cultured cell systems by enhancing the stability of otherwise labile molecular structures formed upon LCD self-association.

TAR binding protein 43 (TDP-43) is an RNA binding protein that assists in the maturation, export and sub-cellular localization of mRNA. The carboxyl terminal 153 residues of TDP-43 are of low sequence complexity and allow for self-association of the protein in a manner leading to its phase separation from an aqueous environment. These interactions assist TDP-43 in forming cytoplasmic RNA granules involved in the transport of mRNA for localized translation. Self-association of the TDP-43 low complexity (LC) domain is facilitated by a region of twenty five residues that are of extreme evolutionary conservation. The molecular basis for self-adherence of the protein through this region has been illuminated by a combination of structural and biochemical studies, allowing definition of a morphologically specific cross-β structure predicted to be weakly assembled by main chain hydrogen bonds. In this study we have investigated the importance of individual, Pauling hydrogen bonds hypothesized to facilitate self-adherence of the TDP-43 LC domain.

Mutations causative of neurological and neurodegenerative disease can occur in coding regions that specify protein domains of low sequence complexity. These autosomal dominant mutations can be idiosyncratic in their recurrent appearance at the same amino acid. Here we report studies of recurrent mutations in proline residues located within low complexity (LC) domains associated with the neurofilament light chain protein, the microtubule-associated tau protein, and the heterogeneous nuclear RNPA2 protein. All such mutations manifest their effects by directing formation of variant proteins endowed with the addition of a single, main chain hydrogen bond specified by the variant amino acid replacing proline. Here we show that methylation of the peptide backbone nitrogen atom associated with these variant amino acids eliminates the aberrant hydrogen bond and restores normal protein function.