Emerging Roles for Intermolecular RNA-RNA Interactions in RNP Assemblies

Abstract

Eukaryotic cells contain large assemblies of RNA and protein, referred to as ribonucleoprotein (RNP) granules, which include cytoplasmic P-bodies, stress granules, and neuronal and germinal granules, as well as nuclear paraspeckles, Cajal bodies, and RNA foci formed from repeat expansion RNAs. Recent evidence argues that intermolecular RNA-RNA interactions play a role in forming and determining the composition of certain RNP granules. We hypothesize that intermolecular RNA-RNA interactions are favored in cells yet are limited by RNA-binding proteins, helicases, and ribosomes, thereby allowing normal RNA function. An over-abundance of intermolecular RNA-RNA interactions may be toxic since perturbations that increase RNA-RNA interactions such as long repeat expansion RNAs, arginine-containing dipeptide repeat polypeptides, and sequestration or loss of abundant RNA-binding proteins can contribute to degenerative diseases.

Figure 1.. Protein-protein interactions promote RNP granule formation.

Proteins can interact in four ways that contribute to the multivalency of RNP granules. A) Classical stereospecific interactions between well-folded domains on proteins. B) Specific proteins bind short linear motifs (SLiMs), conserved sequences within intrinsically disordered regions (IDRs). Often, structure emerges from the IDR upon binding. C) IDRs can also interact specifically with other IDRs through interaction domains with key amino acid characteristics. For example, LARKs are short repeated sequences in IDRs that contain tyrosine and can interact weakly with other LARKs on neighboring proteins. D) IDRs can also provide promiscuous interactions, potentially through π-π or cation-π interactions, which enhance assembly once components are at high-local concentrations.

Figure 2.. RNA contributes to RNP granule formation.

A) RNA can serve as a scaffold for multivalent RNA-binding proteins. These proteins can then interact with each other as described in Figure 1. B) RNAs can interact non-specifically with each other through Watson-Crick base-pairing, non-canonical base-pairing, and helical stacking. C) Molecular crowding has a greater effect on the effective concentration of larger molecules, where the available solvent is much more reduced for larger molecules. This is visualized by the two panels; the left illustrates the accessible solvent to a 55 kDa protein (dark red area) and the right a 7.5 kb RNA (dark blue area). Darker coloring denotes accessible locations of the center of the protein or RNA, respectively. D) The association of larger molecules and complexes is favored by depletion attraction, which is a force only exerted in crowded conditions. The association of two larger molecules decreases the excluded volume and increases the entropy of the smaller macromolecules also in solution. Counterintuitively, the aggregation of larger complexes is entropically favored in crowded environments like the cell.

Figure 3.. RNA-RNA interactions can be promiscuous or specific.

A) Promiscuous interactions may be prevalent in granules containing a diverse set of RNAs in a high local concentration. Here, RNAs are predicted to assemble with a variety of interactions facilitated by the interaction capabilities of an RNA with any other RNA. An example of this may be the formation of stress granules, in which cells experience a large influx of free RNA following ribosomal run-off. B) RNA-RNA interactions contributing to assembly can also be specific. Two examples have been described in Drosophila development in which specific RNA-RNA homodimers of either bicoid or oskar RNAs are important for RNP granule assembly. C) Transcription sites may be a common location of RNP granule assembly, driven in part by the high-local concentration of RNA. As NEAT1 is transcribed, for example, a high local concentration of NEAT1 RNA is achieved. We hypothesize that newly transcribed NEAT1 RNA is capable of forming interactions with neighboring NEAT1 RNAs. As transcripts are released from transcription, a paraspeckle remains. Mature paraspeckles have a clear orientation with the middle of the RNA and certain proteins found in the center, and RNA ends oriented on the outside.

Figure 4.. A four-phase model of RNP granule assembly incorporates protein-protein and RNA-RNA interactions and has specific implications.

A) RNP granules can form through either RNA or protein dominated assembly pathways, or through combinations of these pathways. Increasing key protein-protein interactions can shift monomeric RNPs upwards into an RNP granule regime (black arrow). Increasing RNA-RNA interactions (yellow arrow) can stimulate assembly, but the depletion of assembly RBPs (red arrow) can prevent assembly, even in conditions where RNA-RNA interactions are increased. B) RNP granules can have different requirements for assembly. The relative contributions of RNA-RNA or protein-protein interactions is expected to vary from one type of RNP granule to the next. Protein-driven granules are expected to be highly influenced by the overexpression or deletion of key protein components. These granules are also expected to be regulated by post-translational modifications and chaperones. In contrast, granules primarily driven by RNA-RNA interactions are predicted to have a high local concentration of RNA, with an enrichment for long RNAs or RNAs with stable, specific interactions. In addition, enzymes acting on RNAs, such as helicases, would be expected to modulate the dynamics of RNA-based assemblies. Most granules will reside somewhere in the middle, with some characteristics properties from both sides of the spectrum.

Figure 5.. Concentrations and properties of granule components influence the formation and properties of granules themselves.

A) Effects of monovalent and multivalent RNA-bp on assemblies formed by RNA-RNA interactions. At low concentrations of protein, both monovalent and multivalent RNA-bp are recruited to RNA assemblies, but do not dramatically change the assembly (top left and right) (Van Treeck et al., 2018). In contrast, at high concentrations a monovalent RNA-bp can inhibit assembly by competing for RNA-RNA interactions (bottom left), while a multivalent RNA-bp can enhance assembly by providing additional cross-linking interactions between RNA molecules (bottom right) (Bounedjah et al., 2014). B) Effects of RNAs on self-assemblies of RNA-binding proteins. The addition of low concentrations of short or long RNAs should result in the RNA being effectively recruited to the assembly (top left and right). At high concentrations, both types of RNAs can also inhibit assembly by competing with the protein-protein interaction surface (bottom) (Lin et al., 2015; Schwartz et al., 2013; Maharana et al., 2018). However, RNA can also enhance the assembly of RNA-binding proteins (Banerjee et al., 2017; Molliex et al., 2015; Patel et al., 2015) by either providing a scaffold to increase the valency of RNA-protein complex interactions (Figure 2A), by triggering a conformational change in the protein that promotes assembly, which can occur with short RNAs (middle left), or through RNA-RNA interactions that promote assembly, which is favored for longer RNAs (middle right).

Figure 6.. A model for the modulation of interactions between RNPs.

We hypothesize RNPs may be more prone to associate when their components are in high local concentrations, they are composed of longer RNAs, or contain multivalent RNA-binding proteins or RNAs with increased interaction propensity with other RNAs (such as in repeat expansion diseases). Disassembly may be promoted by increased recruitment of monovalent RNA-binding proteins or short RNAs that limit associations between RNPs. Active remodelers, like chaperones, can also disassemble RNP granules (reviewed in Protter and Parker 2016). In addition, helicases and ribosomes can unwind intermolecular RNA-RNA interactions and block potential RNA-RNA and RNA-protein interactions, respectively.