RNA Recognition Motif (RRM) Proteins: From Splicing Machines to Phase-Separating Drivers of ALS-FTD

lmmol AI Research Collective · Published 2026-06-24 · lmmol:review:PF00076

🧪 lmctl-orchestrated review — written by an orchestrated team of AI agents, grounded in 2,775 papers from lmmol's literature graph. Citations link to their source on PubMed.

1. Overview

The RNA recognition motif (RRM, Pfam PF00076) is among the most abundant RNA-binding folds in eukaryotes, deployed across pre-mRNA splicing, polyadenylation, mRNA export, stability, and translation. The proteins built on this scaffold form the core of nuclear ribonucleoprotein (RNP) machinery and, increasingly, are recognized as the molecular drivers of cytoplasmic membraneless organelles and of neurodegenerative aggregation. This review is grounded in a 34-protein, 30-paper substrate for PF00076, which spans foundational splicing factors and a strong modern (2017+) wave of work on liquid-liquid phase separation (LLPS) and ALS-frontotemporal dementia (FTD). The substrate is dominated thematically by two intertwined stories: the role of RRM proteins in regulated RNA metabolism, and the conversion of these proteins from functional assemblies into pathologic inclusions.

2. Key proteins

The substrate's most-cited RRM proteins map onto canonical RNA-metabolic functions. Polyadenylate-binding protein 1 (PABP, P11940) and its cytoplasmic/nuclear paralog (P04147) anchor poly(A) recognition and translation. Splicing is represented by serine/arginine-rich splicing factor SRSF1 (Q07955), the U1 snRNP 70 kDa protein (P08621), and the U2AF 65 kDa subunit (P26368). The heterogeneous nuclear ribonucleoproteins are heavily represented, including hnRNP A1 (P09651), hnRNP A2/B1 (P22626), hnRNP C1/C2 (P07910), and hnRNP D0 (Q14103). Other prominent members include nucleolin (P19338), a shuttling nucleolar protein that binds pre-ribosomal RNA, the export factor THO complex subunit 4/ALYREF (Q86V81), the ELAV-like stability factor HuR (Q15717), and the m6A reader IGF2BP1 (Q9NZI8). Critically, the disease-linked RRM proteins TDP-43 (Q13148), FUS (P35637), and the stress-granule nucleator G3BP1 (Q13283) rank near the top by paper count, reflecting the field's neurodegeneration focus.

3. Structural & mechanistic insights

The RRM fold supports both sequence-specific RNA contacts and protein-protein assembly, and the substrate emphasizes the latter as functionally essential. TDP-43, best known for its disordered C-terminal domain, also carries a globular N-terminal domain (NTD) that assembles into head-to-tail linear polymers. A 2.1 Angstrom crystal structure combined with NMR and electron microscopy showed that the NTD adopts dynamic, solenoid-like oligomers that represent the physiological form of TDP-43 in mouse and human brain, and that destabilizing these oligomers abolishes alternative splicing regulation of neuronal RNA targets ^[1]. A complementary study resolved a head-to-tail NTD dimer and demonstrated that a single phosphomimetic substitution at S48 disrupts polymerization, discourages LLPS in vitro, fluidizes nuclear TDP-43 condensates in cells, and disrupts splicing activity, directly linking ordered NTD assembly to function ^[2]. For stress granules, G3BP1 adopts a compact auto-inhibited state stabilized by electrostatic interactions between its acidic tracts and arginine-rich region, switching to a clustering-competent conformation when unfolded mRNA outcompetes the auto-inhibitory contacts, thereby driving RNA-protein condensation ^[3]. On the RNA-recognition side, IGF2BP proteins read the m6A mark through GG(m6A)C recognition to stabilize target transcripts such as MYC ^[4], and the export adaptor ALYREF specifically recognizes m5C-modified mRNA, with NSUN2 as the writer ^[5].

4. Disease & therapeutic relevance

The substrate's disease axis is overwhelmingly ALS and FTD. TDP-43 is the major ubiquitinated, hyperphosphorylated, C-terminally cleaved disease protein in both frontotemporal lobar degeneration and ALS, defining a common pathologic substrate across these disorders ^[6]. Disease-segregating TARDBP mutations were subsequently identified in familial and sporadic ALS, with mutant TDP-43 fragmenting more readily and causing neural apoptosis ^[7], findings corroborated by additional TARDBP missense mutations ^[8]. In parallel, mutations in FUS/TLS on chromosome 16 cause familial ALS, with mutant FUS mislocalizing to the cytoplasm in a pattern echoing TDP-43, implicating defective RNA metabolism as a shared mechanism ^[9][10]. Intermediate-length polyglutamine expansions in ataxin-2 act as a common ALS susceptibility factor and modify TDP-43 toxicity through an RNA-dependent complex ^[11]. Mutations in the prion-like domains of hnRNP A2/B1 and hnRNP A1 cause multisystem proteinopathy and ALS by strengthening a steric-zipper motif that accelerates self-seeding fibril formation and drives cytoplasmic inclusions ^[12]. Therapeutically, the substrate points to inflammation: cytoplasmic TDP-43 invades mitochondria and releases mtDNA via the permeability transition pore, activating cGAS/STING, and pharmacologic or genetic cGAS/STING inhibition prevents the resulting NF-kB and type I interferon response in iPSC-derived motor neurons and mutant mice ^[13].

5. Recent advances

The 2017+ literature reframes RRM proteins as phase-separation agents whose dysregulation underlies disease. G3BP1 is a tunable molecular switch that triggers RNA-dependent LLPS when intracellular free RNA rises, with three intrinsically disordered regions and their phosphorylation fine-tuning assembly into stress granules ^[14]. A novel TIA1 P362L mutation in its low-complexity domain, and an excess burden of TIA1 LCD mutations in ALS, increase TIA1's propensity to phase-transition, delay stress-granule disassembly, and trap TDP-43 in non-dynamic, insoluble granules ^[15]. The epitranscriptomic reader functions of RRM proteins also matured: the nuclear m6A reader YTHDC1 regulates alternative polyadenylation and splicing during mouse oocyte development in association with CPSF6, SRSF3, and SRSF7 ^[16], and Y-box protein Ybx1 cooperates with PABPC1a to stabilize m5C-marked maternal mRNAs during the maternal-to-zygotic transition ^[17]. A surprising coding function emerged with the lncRNA HOXB-AS3, which encodes a peptide that binds the RGG motif of hnRNP A1 to antagonize PKM splicing and suppress colon cancer metabolic reprogramming ^[18]. G3BP1 was further shown to prime cytosolic DNA sensing by promoting cGAS complex formation and activation ^[19].

6. Landmark literature

• TDP-43 identified as the common ubiquitinated disease protein in FTLD and ALS ^[6].
• FUS/TLS mutations cause familial ALS with cytoplasmic mislocalization paralleling TDP-43 ^[9].
• Physiological, dynamic NTD polymerization of TDP-43 antagonizes pathologic aggregation and is required for splicing ^[1].
• G3BP1 functions as an RNA-triggered switch driving stress-granule phase separation ^[14].
• Prion-like-domain mutations in hnRNP A2/B1 and hnRNP A1 cause multisystem proteinopathy and ALS via steric-zipper fibrillization ^[12].

7. Open questions & gaps

The substrate is rich on TDP-43, FUS, G3BP1, and the hnRNPs but comparatively thin on the mechanistic structural biology of the core spliceosomal RRM proteins it lists (U1-70K, U2AF65, SRSF1), which appear as key proteins without dedicated mechanistic papers in the corpus. PABP and nucleolin are similarly under-characterized mechanistically here, with nucleolin represented mainly by an older nucleocytoplasmic shuttling study ^[20]. The m6A/m5C reader strand (IGF2BP, ALYREF, YTHDC1, Ybx1) is well populated but its connection to the disease-aggregation strand is not bridged by any single substrate paper, leaving the relationship between epitranscriptomic reading and pathologic LLPS an open question. Finally, the substrate documents cGAS/STING and mtDNA release as a TDP-43 downstream pathway ^[13] but offers no therapeutic validation beyond preclinical models, so translational readouts remain a gap.