Two 2025 NMR/crystal papers (Kohoutova et al. Nat Commun 2025; Bourgeois et al. Nat Commun 2025) have resolved the FOXO4-DBD:p53-TAD interface at atomic resolution and confirmed the disordered p53 TAD as the direct binding target of FOXO4-DRI. The 3L2C crystal structure (1.87 Å, FOXO4-DBD:DNA) provides the scaffold geometry for hotspot-conditioned binder design. Campaign steps: (1) Retrieve updated FOXO4-DBD:p53-TAD co-complex structure from 2025 papers or AlphaFold3; (2) Run BindCraft conditioned on FOXO4-DBD hotspot patch (W97, G98, N99, Q100, S101, Y102, A103, L105, K135) to generate 50-100 de novo peptide backbones; (3) Filter by ESM-2 perplexity (< 10) and AF2-Multimer ipTM (> 0.65); (4) Rosetta FastRelax ddG shortlist to top 10; (5) Foldseek novelty screen vs PDB; (6) Forward-fold top 5 with AF3-Multimer independent of design model. Success criterion: at least one candidate with AF3-Multimer ipTM > 0.70 and predicted Kd < 100 nM (BindingDB p53-TAD reference anchored), structurally novel (Foldseek TM-score < 0.5 vs PDB), ESM-2 perplexity < 8. Selectivity guard: AF3-Multimer run against p53-DBD and BCL-2 must show ipTM < 0.5 for final lead. Wet-lab route: gene synthesis → E. coli expression or SPPS for stapled peptide → ITC/SPR for Kd → HCT116 p21-high senescent cell viability assay.
Details
- disease
- aging / cellular senescence
- target_ref
- PDB:3L2C / UniProt:P98177 (FOXO4) × UniProt:P04637 (TP53)
- primary_endpoint
- ≥1 design passes AF2-Multimer ipTM > 0.6 AND pLDDT > 75 on the binder chain
- identification_strategy
- in_silico_KO
Raw fields (3)
- assay_spec
Hotspot-anchored RFdiffusion backbone generation targeting CD38 NAD+-binding pocket (PDB: 2I65, chain A residues W125, D202, E226, W189), followed by ProteinMPNN sequence design (T=0.1, 8 seqs/backbone), ESM-2 perplexity filter (ppl < 10), and AF2-Multimer forward-fold validation (ipTM > 0.6, pLDDT > 75); 10 RFdiffusion backbone PDBs targeting CD38 NAD+ pocket — planned; 80 ProteinMPNN candidate sequences (8 per backbone) in FASTA — planned; ESM-2 perplexity scores per sequence; filtered set with ppl < 10 — planned; AF2-Multimer predicted complexes (top 5 ESM-filtered candidates vs CD38 chain A) with ipTM and pLDDT per model — planned; Lead binder candidate: sequence, ipTM, pLDDT, predicted binding epitope overlap with W125/D202/E226 — planned
- hypothesis
A de novo peptide or stapled-helix mimetic targeting the FOXO4 DNA-binding domain surface (hotspot residues W97, G98, N99, Q100, S101, Y102, A103, L105, K135; 3L2C chain A geometry confirmed) will competitively displace p53-TAD binding and selectively kill senescent cells, outperforming FOXO4-DRI on predicted binding affinity (target Kd < 100 nM by AF3-Multimer ipTM > 0.7) while preserving selectivity versus p53-DBD interactions.
- kill_criteria
Abort or revise if: 2I65 is an apo structure; induced-fit effects upon binder docking may not be captured by rigid RFdiffusion conditioning — flag for MD follow-up; Active-site binders risk cross-reactivity with CD157 (BST1), a paralog sharing the ADP-ribosyl cyclase fold — Foldseek novelty check required post-generation; AF2-Multimer may underpredict binding for short helical peptides vs globular binders — ipTM threshold is conservative at 0.6