CD38 is the dominant NAD-consuming ectoenzyme in aged mammalian tissues, with expression rising 2–3-fold with age. NAD⁺ decline is a hallmark of aging linked to mitochondrial dysfunction, senescence, and metabolic deterioration. Existing CD38 inhibitors are small molecules (daratumumab, isatuximab are mAbs targeting the extracellular domain; 78c is a small-molecule; apigenin is a weak flavonoid inhibitor). A designed protein binder targeting the catalytic cleft could achieve sub-nM affinity with tunable selectivity against the related BST-1/CD157. Design campaign: (1) Hotspot conditioning from PDB 2I65 chain A (C119, K121, W125, F143, E146, D155, L157, Q226 — geometric profile already computed); (2) BindCraft one-shot binder generation conditioned on these hotspots; (3) ProteinMPNN sequence diversification on top RFdiff backbones; (4) AF2-Multimer forward-fold validation (independent of design model); (5) Boltz-1 complex prediction for cross-model consensus; (6) ESM-2 log-likelihood filtering; (7) Foldseek novelty check against PDB/AlphaFold DB; (8) PyRosetta interface ddG and shape complementarity scoring. Success criteria: ipTM > 0.75, pLDDT > 80 on binder chain, AF2 RMSD vs design < 2 Å, Rosetta interface dG < −10 REU, Foldseek TM-score < 0.5 vs nearest PDB hit (novelty), ESM-2 pseudo-perplexity < 10.
Details
- disease
- aging, nad+ decline, age-related metabolic dysfunction
- target_ref
- UniProt:P28907 | PDB:2I65
- identification_strategy
- in_silico_KO
Raw fields (4)
- assay_spec
Tick 28 CONFIRMED: pdb_hotspot_profile on 2I65 returned all 8/8 hotspot residues (C119, K121, W125, F143, E146, D155, L157, Q226) with zero missing. Centroids confirmed. 2PGJ benchmark reference confirmed at 1.71 Å. Tick 29: BindingDB IC50 query returned not_found (0 results); EuropePMC query returned 255 hits but no direct protein/peptide binder papers in top-8. UniProt P28907 canonical sequence (300 aa) confirmed for BindCraft FASTA input. Tick 30: BindCraft recipe artifact created (c340aeb5-fd2c-4d01-9cec-bfaf579a5edf). BindingDB Ki fallback query returned not_found. PubMed protein-inhibitor survey returned 0 results. Tick 31: BindingDB Kd query dispatched; PubMed and EuropePMC affinity surveys dispatched. Reference affinity ceiling: daratumumab Kd ~1 nM (anti-CD38 IgG, clinical). Planned success criterion: ipTM >= 0.75 + AF2 pLDDT >= 85 as functional proxies for sub-100 nM competitive binders.
- hypothesis
A de novo protein binder targeting the CD38 catalytic cleft (hotspot residues C119, K121, W125, F143, E146, D155, L157, Q226 from PDB 2I65, 1.9 Å resolution) will competitively inhibit CD38 NADase activity, reduce NAD⁺ consumption in aged tissues, and outperform small-molecule inhibitors (e.g., 78c, isatuximab) on selectivity and tissue penetrance metrics assessable in silico.
- kill_criteria
Abort if 2I65 hotspot profile returns >3 missing residues from the 8-residue catalytic set — indicates numbering mismatch requiring manual re-alignment before design proceeds. Also abort if 2PGJ and 2I65 hotspot centroids diverge >3 Å RMS — implies conformational switch too large for single-conformation binder design without ensemble docking.
- primary_endpoint
Tick 31 deliverable: affinity floor established from literature and/or BindingDB. Reference ceiling: daratumumab anti-CD38 IgG Kd ~1 nM (prior knowledge). Design success criteria formally set: ipTM >= 0.75 AND AF2-Multimer pLDDT >= 85 as in-silico proxies for sub-100 nM functional binders. BindCraft recipe artifact (c340aeb5-fd2c-4d01-9cec-bfaf579a5edf) linked to research plan. Tick 32 goals: (1) create BindCraft execution notebook with full pipeline specification — hotspot conditioning on 2I65 residues C119,K121,W125,F143,E146,D155,L157,Q226; binder length 60–90 aa; 100 designs; AF2-Multimer-v3 independent validation; ipTM >= 0.75 and pLDDT >= 85 thresholds; ESM-2 perplexity <= 10; Rosetta ddG <= -10 kcal/mol; Foldseek novelty check. (2) Confirm notebook kernel availability before marking ready-to-execute.