FOXO4-DRI (D-retro-inverso FOXO4 peptide) demonstrated selective senescent-cell apoptosis in vivo (Baar et al. 2017, Nature). Two 2025 Nature Communications papers (Kohoutova et al., DOI 10.1038/s41467-025-59106-5; Bourgeois et al., DOI 10.1038/s41467-025-60844-9) clarified the structural plasticity and the disordered p53-TAD as the direct target. The 3L2C crystal structure of FOXO4-DBD provides a resolved hotspot surface. This plan proposes: (1) hotspot-conditioned RFdiffusion backbone generation targeting the N-terminal helical bundle of FOXO4-DBD (residues 99-105, 135-138) to design binders that mimic the p53-TAD engagement surface; (2) ProteinMPNN sequence design on top-ranked backbones; (3) AF2-Multimer forward-fold validation (independent of RFdiffusion) requiring pLDDT > 80 and ipTM > 0.75; (4) ESM-2 perplexity filter (mean log-likelihood > -1.5 per residue); (5) Foldseek novelty check against PDB to confirm scaffold novelty vs prior senolytics; (6) Rosetta FastRelax ddG as a final triage gate. Success criterion: at least one candidate with predicted Kd < 100 nM (benchmarked against FOXO4-DRI Ki ~ 200 nM), pLDDT > 82, ipTM > 0.78, and Foldseek TM-score < 0.5 vs nearest PDB hit.
Details
- disease
- cellular senescence, aging
- target_ref
- UniProt:P98177 / PDB:3L2C chain A
- primary_endpoint
- ≥8 chain A residues with relative SASA burial ≥ 40% in dimer vs monomer (confirms a tractable interface)
- identification_strategy
- in_silico_KO
Raw fields (3)
- assay_spec
Fetch PDB:2GVJ (NAMPT dimer), extract chain A hotspot shell residues via BioPython SASA (FreeSASA) + 5 Å contact map, output ranked hotspot JSON + interface contact heatmap PNG to gate BindCraft run; nampt_2gvj_hotspots.json — ranked list of chain A residues by relative SASA loss on dimer vs monomer (burial score), with Cα coordinates and neighbor shell, top 12 residues flagged as BindCraft hotspot seeds; nampt_2gvj_interface_contacts.png — chain A × chain B residue-residue contact heatmap (5 Å all-atom), hotspot residues highlighted, saved as artifact; {'kind': 'claim', 'text_template': 'NAMPT dimer interface buries ≥ N residues with >40% SASA reduction; top hotspot cluster spans residues R, S, T — these are the BindCraft target_pdb_hotspot inputs'}- hypothesis
The FOXO4 DNA-binding domain engages the disordered p53 transactivation domain through a defined helical interface (3L2C hotspot residues N99, S101, Y102, A103, L105, K135, K137, G138). Designing a compact, cell-penetrant binder that occludes this interface with higher affinity than FOXO4-DRI will selectively trigger apoptosis in senescent cells without affecting proliferating cells, improving the senolytic therapeutic index.
- kill_criteria
Abort or revise if: 2GVJ is a substrate-bound structure (PRPP + NMN); substrate atoms must be excluded from SASA calculation to avoid artefactual burial scores; If freesasa is not in the base env, fall back to BioPython SASA via ShrakeRupley — document which algorithm was used in the JSON metadata field