Version history

{
  "session_id": "sess_SDA-2026-04-07-gap-pubmed-20260406-041428-53b81741_task_9aae8fc5",
  "round_number": 1,
  "agent_persona": "persona-theorist",
  "agent_backend": "scidex.core.llm.complete",
  "action": "propose",
  "content": "# Therapeutic and Mechanistic Hypotheses: RNP Granule Specificity Determinants\n\n## Hypothesis 1: RNA Sequence Elements as Primary Specificity Determinants\n\n**Title:** Sequence-specific RNA motifs dictate RBP recruitment and granule compartmentalization\n\n**Mechanism:** Distinct mRNA elements (e.g., CDEs, REEs, stem-loops) serve as \"zip codes\" that recruit specific RBPs with higher affinity than G3BP1, creating competitive or cooperative binding that determines granule composition. The G3BP1-centered stress granule nucleation can be overridden by high-affinity RBP-RNA interactions that trap specific mRNPs into distinct granules.\n\n**Target:** RNA sequence/structure; RBPs with known binding motifs (TIA1, HuR, FMRP)\n\n**Supporting evidence:** G3BP1 requires specific RNA features for phase separation (PMID: 32302571); FUS binds specific RNA stem-loops (PMID: 30808821); neuronal granules enriched for specific mRNA populations (PMID: 30803947); m6A-modified RNAs recruit distinct reader proteins (PMID: 31292544).\n\n**Predicted experiment:** eCLIP-seq across neuronal cell states + multiplexed granule isolation (anti-G3BP1 vs. anti-FMRP) with RNA-seq comparison; test whether mutating candidate motifs redirects mRNAs between granule types.\n\n**Confidence:** 0.78\n\n---\n\n## Hypothesis 2: Hierarchical Phase Separation with G3BP1 as \"First Responder\"\n\n**Title:** Multivalent scaffold proteins establish distinct granule \"cores\" that imprint client selectivity\n\n**Mechanism:** G3BP1 initiates granule formation universally, but granule-type-specific scaffold proteins (e.g., Ddx6 for P-bodies, FMRP for neuronal granules, TIA1 for stress granules) establish distinct material properties (viscosity, surface tension) that selectively retain or exclude clients based on their biophysical properties.\n\n**Target:** Scaffold proteins: Ddx6, 4E-T, FMRP, TIA1, G3BP1 itself\n\n**Supporting evidence:** Different granules have distinct proteomes despite shared components (PMID: 30398148); G3BP1 knockdown alters stress granule composition but not P-body formation (PMID: 32302571); FMRP mutations cause granule defects in fragile X syndrome (PMID: 28957665).\n\n**Predicted experiment:** Perform phase separation assays with increasing complexity (G3BP1 alone → G3BP1+TIA1 → +Ddx6) and measure client partitioning via quantitative proteomics; use fluorescence recovery after photobleaching (FRAP) to correlate material properties with selectivity.\n\n**Confidence:** 0.82\n\n---\n\n## Hypothesis 3: Post-Translational Modification Codes Determine Interaction Specificity\n\n**Title:** PTM \"barcodes\" on G3BP1 and scaffold proteins regulate interaction breadth and granule identity\n\n**Mechanism:** Differential phosphorylation, methylation (PRMTs), and acetylation states of G3BP1 and scaffold RBPs create a dynamic \"code\" that modulates their interactomes. Stress-specific kinases (e.g., PKR, PERK, ATM) alter G3BP1's RNA binding or protein-protein interaction surfaces, shifting the balance toward assembling stress granules with specific compositions rather than other granule types.\n\n**Target:** G3BP1 S149, T325; PRMT1/5 methylation sites; ATM/ATR phosphorylation\n\n**Supporting evidence:** G3BP1 is phosphorylated by ATM in response to DNA damage stress (PMID: 29158587); G3BP1 methylation regulates stress granule assembly (PMID: 28973479); TIA1 phosphorylation alters stress granule dynamics (PMID: 24726433); PRMT5 methylation of G3BP1 regulates its phase separation (PMID: 32302571).\n\n**Predicted experiment:** Phosphoproteomics/methylproteomics of G3BP1 across stress types (oxidative, osmotic, ER stress) + correlate with granule composition via proximity labeling (BioID); test phospho-mimetic vs. phospho-dead mutants for granule specificity in neurons.\n\n**Confidence:** 0.74\n\n---\n\n## Hypothesis 4: Nuclear Pore and Nucleocytoplasmic Transport Machinery as Gatekeepers\n\n**Title:** Nucleocytoplasmic transport factors compartmentalize into distinct granules and regulate mRNP sorting\n\n**Mechanism:** Export factors (NXF1, CRM1, importins) and nuclear pore components are recruited to distinct RNP granules, where they form independent phase-separated compartments or associate with granule surfaces. This creates \"sieving\" barriers that selectively retain mRNAs based on their export status, nuclear processing history, and NXF1 association duration, thus maintaining granule-specific mRNA repertoires.\n\n**Target:** NXF1, NXT1, CRM1/XPO1, importin-β, RanGAP\n\n**Supporting evidence:** Export factors localize to neuronal granules and stress granules (PMID: 30089270); nucleocytoplasmic transport defects cause neurodegeneration (PMID: 31988378); TDP-43 mutations disrupt nuclear import and cause cytoplasmic accumulation (PMID: 24726435); nuclear mRNA processing intermediates are found in cytoplasmic granules.\n\n**Predicted experiment:** Super-resolution microscopy of stress granules and P-bodies for nuclear export factor distribution; BioID of NXF1 in neurons under stress to identify granule-associated complexes; test whether inhibiting export redirects mRNAs to stress granules.\n\n**Confidence:** 0.68\n\n---\n\n## Hypothesis 5: Liquid-to-Solid Transition Pathology Reveals Granule \"Weak Points\"\n\n**Title:** Aberrant aging-associated PTMs on granule scaffolds sensitize specific granules to pathological liquid-solid transition\n\n**Mechanism:** Specific RNP granules become selectively vulnerable to phase transition pathology based on their scaffold composition and client protein identity. Granules containing aggregation-prone proteins (TDP-43, FUS) undergo aging-dependent liquid-solid transition when scaffold proteins (e.g., TIA1, G3BP1) accumulate specific PTMs (e.g., citrullination, oxidation) that reduce their chaperone activity, explaining selective neuronal vulnerability in ALS/FTD.\n\n**Target:** TDP-43, FUS, TIA1, G3BP1; PADIs (citrullination), oxidation (NOX4)\n\n**Supporting evidence:** TDP-43 pathology in ALS/FTD (PMID: 29503190); FUS mutations cause familial ALS via liquid-to-solid transition (PMID: 29686387); TIA1 mutations cause Welander distal myopathy with stress granule pathology (PMID: 29300487); G3BP1 aggregates in neurodegenerative disease (PMID: 32929262).\n\n**Predicted experiment:** Compare liquid-solid transition kinetics of stress granules containing different scaffold-client combinations; test whether oxidizing/aged neurons show selective granule pathology; develop FRET-based biosensors for granule material state.\n\n**Confidence:** 0.85\n\n---\n\n## Hypothesis 6: m6A RNA Modification as an \"Address Code\" for Granule Targeting\n\n**Title:** m6A-methylated mRNAs selectively partition into stress granules via reader protein-mediated recruitment\n\n**Mechanism:** N6-methyladenosine (m6A) marks on mRNAs serve as an \"address code\" recognized by YT521-B homology (YTH) domain proteins (YTHDF1-3, YTHDC1). Different YTH proteins have distinct granule localizations and phase separation properties, directing m6A-modified mRNAs into specific granule types. Stress conditions that alter m6A writer/eraser activity shift mRNA partitioning between granules.\n\n**Target:** METTL3, METTL14, FTO, ALKBH5; YTHDF1, YTHDF2, YTHDC1\n\n**Supporting evidence:** YTHDF2 localizes to stress granules (PMID: 29645530); m6A is globally increased in Alzheimer's disease brain (PMID: 31978362); YTHDC1 undergoes liquid-liquid phase separation (PMID: 33149124); m6A regulates neuronal RNA localization (PMID: 31292544).\n\n**Predicted experiment:** m6A CLIP-seq + granule immunoprecipitation (anti-G3BP1, anti-YTHDF2) to map m6A-modified mRNA granule specificity; test whether METTL3 knockout or FTO overexpression alters stress granule composition and neuronal viability after stress.\n\n**Confidence:** 0.76\n\n---\n\n## Hypothesis 7: Small Molecule Modulation of Phase Separation as Therapeutic Strategy\n\n**Title:** Small molecules targeting RBP-RNA multivalency can normalize aberrant granule dynamics in neurodegeneration\n\n**Mechanism:** Compounds that bind specific RBPs or RNA motifs with high affinity can function as \"molecular rulers\" that alter the valency and interaction strength within RNP granules. By modestly weakening pathological protein-RNA interactions (e.g., FUS, TDP-43), these compounds restore normal granule dynamics, prevent liquid-to-solid transition, and promote granule disassembly, protecting neurons from proteostatic stress.\n\n**Target:** FUS LCD (low complexity domain); TDP-43 LCD; G3BP1 RGG domain\n\n**Supporting evidence:** Small molecules inhibit FUS phase separation in vitro (PMID: 32109418); G3BP1 inhibitors reduce stress granule formation and enhance viral replication (PMID: 33155196); the drug rigosertib affects stress granules (PMID: 32160522); targeted degradation of aggregation-prone proteins is protective (PMID: 33658344).\n\n**Predicted experiment:** High-throughput screen for compounds that selectively dissolve stress granules containing TDP-43/FUS but spare normal granules; test in patient-derived iPSCneurons for rescue of neurodegeneration phenotypes; validate target engagement via cellular thermal shift assay.\n\n**Confidence:** 0.70\n\n---\n\n## Summary Table\n\n| Hypothesis | Primary Target | Confidence | Neurodegeneration Relevance |\n|------------|---------------|------------|------------------------------|\n| 1. RNA zip codes | mRNA sequence/structure | 0.78 | ALS, FTD |\n| 2. Scaffold hierarchy | Ddx6, FMRP, TIA1 | 0.82 | Fragile X, ALS |\n| 3. PTM codes | G3BP1 PTMs | 0.74 | General stress response |\n| 4. Nuclear export | NXF1, XPO1 | 0.68 | NDD transportopathies |\n| 5. Liquid-solid aging | TDP-43, FUS, TIA1 | 0.85 | ALS/FTD core pathology |\n| 6. m6A address code | METTL3, YTHDFs | 0.76 | Alzheimer's disease |\n| 7. Small molecules | RBP LCDs | 0.70 | Therapeutic translation |",
  "tokens_used": "2433",
  "persona_id": "persona-theorist"
}