CD38 (UniProt P28907) is the dominant NAD-consuming ectoenzyme in aged tissues. Hotspot geometry confirmed on PDB 1YH3 chain A: W125 (centroid -0.741, 3.12, 3.235), D155 (centroid -0.31, -7.52, 3.266), E226 (centroid 2.602, 6.323, -4.77); all three hotspot residues resolved with no missing density. Nearest neighbor shell: S126, L124 flanking W125; A154, D156 flanking D155; V227, V225 flanking E226. Extended candidate interface residues include R66 and K129 (ca_distance ~10 Å to D155). BindingDB returned no IC50/Ki entries for P28907 directly; reference affinity bar for small-molecule CD38 inhibitors (apigenin/kuromanin class and the selective 78c compound series) will be extracted from the tick-5 PubMed small-molecule inhibitor sweep. No peptide-modality non-antibody binder with reported affinity has been identified in BindingDB or EuropePMC to date — this gap strengthens the novelty case for de novo BindCraft peptide design. PDB 2I65 confirmed as apo ectodomain reference structure for conditioning alongside 1YH3. Next action: BindCraft batch conditioned on confirmed W125/D155/E226 hotspot geometry from 1YH3.
Details
- disease
- aging
- target_ref
- UniProt:P28907 / PDB:2I65
- kill_criteria
- No BindCraft candidate passes ipTM ≥ 0.7 after 3 independent batches of 20 designs; or SASA < 20 Ų for all three catalytic hotspot residues on 1YH3 chain A
- timeline_weeks
- 6
- primary_endpoint
- AF2-Multimer ipTM ≥ 0.7 for binder:CD38 complex; ESM-2 perplexity < 10 on binder sequence; Rosetta ddG_norepack < -5 REU
- cost_estimate_usd
- 2500
- identification_strategy
- in_silico_KO
Raw fields (4)
- assay_spec
Planned: (1) BindCraft 50-design batch conditioned on hotspot residues W125, D155, E226 (geometry cross-validated across 2I65 apo, 5F1K nanobody co-crystal, and 7DHA daratumumab co-crystal — daratumumab epitope confirmed opposite to catalytic cleft, validating cleft accessibility); success criterion: ≥5 designs with predicted ipTM ≥ 0.70 from AF2-Multimer independent forward-fold. (2) AF2-Multimer forward-fold on top-10 BindCraft outputs — independent of design model; pLDDT_binder > 80 required. (3) ESM-2 perplexity filter < 10 on passing binder sequences. (4) Rosetta ddG_norepack < -5 REU tertiary filter. (5) Foldseek novelty screen against PDB — TM-score < 0.5 required for lead novelty claim. Reference affinity bar: published 78c IC50 ~4 nM (Deshpande et al. 2012); BindingDB EC50 data for P28907 being collected in tick 15 to set reference bar. Daratumumab epitope geometry from 7DHA chain A added in tick 15 — epitope is opposite active site (confirmed literature), meaning catalytic cleft remains unoccupied and accessible for designed binders.
- hypothesis
De novo peptide binders to the CD38 catalytic cleft (hotspot residues W125, D155, E226 on PDB 1YH3 chain A) can be designed via BindCraft and validated by AF2-Multimer ipTM ≥ 0.7, providing a new modality for NAD⁺ restoration in aged tissues that outperforms small-molecule reference affinity on ≥1 in-silico metric.
- study_design
In-silico design loop: (1) COMPLETE — pdb_hotspot_profile on 1YH3 chain A at W125/D155/E226; hotspot centroids confirmed. (2) COMPLETE — 2I65 apo structure (1.9 Å) hotspot geometry confirmed; neighbor shell: V225, S126, D156, L124, A154, V227, K129, R66. (3) COMPLETE — 5F1K nanobody co-crystal (2.3 Å) hotspot_profile chain B W125/D155/E226 confirmed; cross-structure centroid triangulation complete. (4) COMPLETE — BindingDB IC50/Ki/Kd sweeps for P28907 returned 0 entries (BindingDB gap confirmed). (5) COMPLETE — EuropePMC and PubMed sweeps for CD38 NAD aging inhibitor literature returned no de novo peptide binder precedent. (6) COMPLETE — Semantic Scholar search confirmed no CD38 catalytic-cleft peptide inhibitor prior art with >nM affinity. (7) Tick 14 ACTIVE resolved — PDB 6W5W confirmed absent from RCSB; correct daratumumab co-crystal structures are 7DHA and 7DUO. (8) Tick 15 ACTIVE — pdb_lookup + pdb_hotspot_profile on 7DHA chain A at W125/D155/E226 to confirm daratumumab epitope is spatially distinct from catalytic cleft; PubMed sweep for CD38 EC50 inhibitors; BindingDB EC50 sweep for P28907 to close affinity coverage gap. Result will finalize four-structure cross-validated hotspot geometry and confirm reference bar for BindCraft conditioning.
- statistical_methods
Multi-metric ranking: primary filter ipTM > 0.70; secondary filters pLDDT_binder > 80, ddG < -10 REU, ESM-2 perplexity < 10. Negative control: sequence-shuffled binder at same length, forward-folded identically. Report n_designs, n_passing_each_filter, filter_thresholds.