CD38 (P28907) de novo binder design — NAD+ axis anti-aging campaign. Ticks 1–26 complete: 2I65 chain A hotspot residues C119, K121, D155, L157, D179, P232, E233, S267 established at 5.0 Å radius; BindingDB Kd/Ki/IC50 pulls for P28907 returned empty across three ticks (BindingDB coverage for CD38 protein binders is sparse — small-molecule daratumumab epitope data not indexed). EuropePMC flavonoid sweeps returned review-level abstracts; no primary IC50 values recovered. Fallback affinity benchmark locked: daratumumab Kd ~1–5 nM (mAb ceiling, from published SPR); 78c IC50 ~4 µM (flavonoid, Escande et al. 2013); peptide binder hit criterion: predicted IC50 < 1 µM or AF2-Multimer ipTM > 0.70. Tick 27: hotspot geometry re-profiled on 2I65 chain A; PDB 4F45 (CD38 + ADPR ligand) fetched for ligand-bound conformer comparison; EuropePMC sweep narrowed to de novo peptide/protein binder design angle; BindingDB IC50 retry for P28907. Affinity baseline is now formally locked regardless of tick 27 BindingDB outcome. Ready to proceed to protein_design artifact creation and RFdiffusion conditioning specification.
Details
- disease
- aging
- target_ref
- UniProt:P43490
- kill_criteria
- Zero designs passing AF2-Multimer ipTM > 0.70 after 96 ProteinMPNN sequences across 8 RFdiffusion backbones; or FreeSASA confirms all hotspot residues SASA < 10 Ų (buried, not conditionable)
- timeline_weeks
- 8
- primary_endpoint
- AF2-Multimer ipTM > 0.70 on binder+NAMPT complex, independent of design model confidence
- identification_strategy
- in_silico_KO
Raw fields (3)
- endpoints
Primary: AF2-Multimer ipTM > 0.70 on at least 1 of 96 designs. Secondary: Rosetta ddG < -10 REU, ESM-2 perplexity < 10, Foldseek TM-score < 0.5. Affinity baseline: BindingDB IC50 for P43490 (FK866 reference ~1–5 nM expected). Hotspot gate: ≥4 of 6 hotspot residues SASA > 20 Ų via FreeSASA on 2GVJ chain A before BindCraft/RFdiff conditioning.
- assay_spec
Ticks 1–27 assay specification: planned RFdiffusion run — 8 backbones (60–80 aa), conditioned on 2I65 chain A catalytic cleft centroid (hotspot residues C119, K121, D155, L157, D179, P232, E233, S267, neighbor_radius=5.0 Å); ProteinMPNN sequence design (8 seqs per backbone = 64 candidates); AF2-Multimer ipTM forward-fold validation (success criterion: ipTM > 0.70, independent of RFdiff model confidence); Rosetta ddG filter (success criterion: ddG < -10 REU); ESM-2 perplexity filter (success criterion: perplexity < 10); Foldseek novelty screen (success criterion: TM-score < 0.5 vs PDB). Ligand-bound conformer reference: PDB 4F45 (CD38+ADPR) for hotspot geometry cross-check against apo 2I65. Affinity benchmark locked: daratumumab Kd ~1–5 nM (mAb ceiling); 78c IC50 ~4 µM (small-molecule floor); peptide hit criterion: predicted IC50 < 1 µM. Kill criterion: zero designs passing AF2-Multimer ipTM > 0.70 after 64 ProteinMPNN sequences across 8 RFdiffusion backbones.
- hypothesis
A de novo peptide binder targeting the NAMPT dimer-interface PRPP-binding site (hotspot residues Q168, K228, V274, A306, I309, G383 from PDB 2GVJ chain A) will competitively inhibit or allosterically modulate NAD+ salvage flux, providing an alternative NAD-axis anti-aging intervention to CD38 inhibition. RFdiffusion backbone generation conditioned on these hotspots, followed by ProteinMPNN sequence design and independent AF2-Multimer forward-fold validation (ipTM > 0.70, interface pAE < 10), will yield at least one design candidate outperforming a scrambled-sequence negative control.