FOXO4-DRI (a D-amino-acid retro-inverso peptide spanning FOXO4 residues ~150-180 in the CR3/transactivation domain) blocks FOXO4 from sequestering p53 in the nucleus of senescent cells, releasing p53 to drive apoptosis selectively in those cells. The original Baar et al. 2017 (Cell) proof-of-concept showed clearance of p53-FOXO4 nuclear foci and reversal of several aging phenotypes in mice. Key limitations: (1) the original peptide has modest affinity and likely requires high doses; (2) RI peptides have improved proteolytic stability but still require delivery optimization; (3) no crystal structure of the FOXO4 CR3 – p53 TAD complex exists, requiring homology or AF3 modelling to define the binding interface. This plan will: (a) use available structural data on p53 TAD interactions (PDB 1H9T: p53 TAD–MDM2 as geometry reference; PDB 2LVM: FOXO4 FKH domain) plus AF3/AlphaFold-Multimer modelling of the FOXO4 CR3 – p53 TAD complex to map the interface hotspots; (b) run BindCraft conditioned on those hotspots to generate candidate binder scaffolds targeting p53 TAD at the FOXO4-contact surface; (c) sequence-design candidates with ProteinMPNN; (d) validate independently with AF2-Multimer (ipTM, pAE, pLDDT) and score with ESM-2 log-likelihood; (e) benchmark against the original FOXO4-DRI peptide sequence on all metrics; (f) triage top candidates by Foldseek novelty search against PDB. Success criteria: at least one candidate with AF2-Multimer ipTM ≥ 0.75, pLDDT > 80 on designed chain, ESM-2 pseudo-log-likelihood improvement ≥ 0.5 nats/residue vs. FOXO4-DRI, and no Foldseek hit < 0.5 TM-score to existing PDB structures.
Details
- disease
- aging, cellular senescence
- target_ref
- UniProt:P04637 (TP53 TAD, residues 1-67) × UniProt:P98177 (FOXO4 CR3)
- primary_endpoint
- ≥6 hotspot residues identified with ΔSASA > 20 Ų each at NAD+ analog interface
- identification_strategy
- in_silico_KO
Raw fields (3)
- assay_spec
PDB 2I65 chain extraction → SASA-based hotspot mapping (FreeSASA) → contact residue enumeration at NAD+ binding pocket → BindCraft input JSON generation for de novo competitive binder campaign against CD38 catalytic cleft; PyMOL-style hotspot map PNG: CD38 chain A surface colored by ΔSASA burial, hotspot residues labeled, NAD+ analog shown as sticks — planned; hotspot_residues.json: ranked list of CD38 residues by ΔSASA and contact frequency with NAD+ analog, with backbone coordinates for RFdiffusion/BindCraft conditioning — planned; bindcraft_cd38_config.json: BindCraft campaign input specifying target PDB path, hotspot residue indices, binder length range 60–90 aa, number of trajectories 50 — planned; {'kind': 'claim', 'text_template': 'CD38 catalytic cleft hotspot set comprises residues {list} with mean ΔSASA burial {X} Ų per residue relative to apo surface, defining the conditioning epitope for de novo binder design — planned'}- hypothesis
A de novo retro-inverso or stapled peptide mimicking the FOXO4 CR3 domain can outperform FOXO4-DRI on affinity for the p53 transactivation domain (TAD) in senescent cells while retaining proteolytic stability, and can be validated by AF2-Multimer ipTM ≥ 0.75 and ESM-2 log-likelihood improvement over the FOXO4-DRI reference sequence.
- kill_criteria
Abort or revise if: 2I65 may have crystal contacts near catalytic site — verify chain A is the biological unit before computing SASA; NAD+ analog in 2I65 is arabino-NAD not canonical NAD+; contact geometry is close but not identical — note in hotspot annotation; BindCraft config schema may have version drift; pin to version used in research_plan:5e4168c3 if specified there