STRC h26 Phase 1c Contact Re-Cluster 2026-04-23

Verdict. GREEN — secondary contact cluster identified at stump 1077-1114 (25 residues, 21 disulfide geometries) entirely outside Phase 1 target 1579-1581. Native C1081 sits within Cb–Cb 6.87-7.09 Å of A1078/C1079/S1080 on the opposing chain — latent inter-chain disulfide site. Phase 1d Cys-engineering AF3 queued; h26 stays B-tier pending AF3 confirmation, but a green path is open.

Why Phase 1c

Phase 1 AF3 tested four mutations at the ARM sub-cluster 1579-1581 and all four destabilized the homodimer (per STRC Engineered Homodimer Phase 1 Results). The R-R repulsion hypothesis was falsified, but the dimer consensus analysis (ultramini_homodimer_consensus.json) already hinted that the 1579-1581 ARM was only part of a two-zone interface:

  • “stump” zone 1077-1131 — 13 strict-consensus residues, never mutagenized
  • “deep ARM” zone 1493-1590 — 16 strict-consensus residues (Phase 1 only touched 1579-1581 inside this zone)

Phase 1c asks: if we spatially cluster the full inter-chain contact set on the Ultra-Mini homodimer CIF, how many independent cluster-nodes are there, and are any of them (a) outside the Phase 1 target and (b) geometrically set up for Cys-engineering?

Method

  • engineered_homodimer_phase1c_contact_cluster.py
  • Input: /Users/egorlyfar/Sites/site-strc-egor-lol/public/models/job-ultramini-homodimer.cif (chain A + chain B, Ultra-Mini 1075-1775 × 2 at offset 1074 from local residue numbering)
  • Inter-chain heavy-atom contacts at 5.0 Å cutoff → residue-pair min distances
  • A-side Cα coords of all contact residues → DBSCAN (eps 6.5 Å, min_pts 3)
  • Per cluster: residue composition, C2-symmetry (does partner on B sit inside the same canonical range), disulfide-engineering Cb–Cb distance (4.5-7.5 Å window around native S-S geometry), mutation candidates ranked by steric / electrostatic / covalent logic

Results

Cluster topology

clusterzonen_resresidue rangeC2-symmetric pairsdisulfide candidatesPhase 1 overlap
1stump_1075_1200251077-111418 / 2521False
2deep_arm_1401_1775161529-159010 / 1611True (1579-1581 inside)
3stump (satellite)41128-11310 / 40False

Total A-side contact residues: 49. Total inter-chain residue pairs at 5 Å: 114. Noise: 4 residues.

Cluster 1 (stump 1077-1114) — primary GREEN target

  • Composition: 13 hydrophobic, 7 polar, 3 positive charged, 1 negative, 1 aromatic
  • Native C1079 and C1081 present — with C1081 forming inter-chain Cb–Cb contacts at 6.87-7.09 Å to A1077 / C1079 / S1080 on the opposite chain. This looks like a native cystine near-pair that could be fixed to an engineered inter-chain disulfide with a single S-to-C or A-to-C mutation on the A-side.
  • Top disulfide-engineering picks (Cb–Cb in S-S window):
mut site (chain A)partner (chain B)Cb–Cb (Å)rationale
A1078 A→CC1081 (native)~6.9A1078 Cb close to native C1081; single mutation locks one-sided cross-link
S1080 S→CC1081 (native)6.97Equivalent single-sided lock
C1079 nativeC1081 (native)7.09Already cystine-cystine — may form natively at oxidizing conditions
  • Steric disruption picks: A1078→W, L1082→W, L1089→W, A1096→W (all destabilizing; used only as negative controls in Phase 1d)
  • Electrostatic: R1088 has same-sign Arg-Arg contacts with K1104 and K1108 on opposite chain; R1088→D would relieve the electrostatic penalty and is the single best “dimer-stabilizing charge flip” candidate.

Cluster 2 (deep ARM 1529-1590) — Phase 1 target zone, residual candidates

  • Phase 1 tested R1581 and S1579 only; 13 contact residues in this cluster were never touched.
  • Disulfide candidates include V1549 → H1585 Cb–Cb 4.9 Å (tightest S-S geometry on the whole interface) and D1550 → D1587 Cb–Cb 6.9 Å (homotypic D-D; D→C on both sides gives a symmetric disulfide at the cost of losing a weak salt bridge that isn’t contributing because D-D is repulsive anyway).
  • Homotypic S1579→C at Cb-Cb 6.94 Å — gives a C2-symmetric Cys-Cys pair at the exact residue where Phase 1 mutagenized with the aromatic Y/F. The disulfide mutation was never tested.

Cluster 3 (1128-1131) — satellite

4 residues, no C2-symmetric partners, no disulfide geometry. Ignore.

Mutation candidate ranking — Phase 1d design

Six independent mutations at single residue, ranked by Nobel-path value:

rankmut (A / B, if symmetric)clustermechanismrisk
1A1078C / A1078C1 stumpinter-chain disulfide with native C1081 on both chains — symmetric, 2 native cystines stay intactmildly destabilizing if A1078 is in a packing context — check AF3 ipTM
2S1080C / S1080C1 stumpinter-chain disulfide with native C1081 on both chainspolar-to-sulfur swap, minimal steric perturbation
3S1579C / S1579C2 ARMC2-symmetric homotypic disulfide at tested positionreplaces Phase 1’s failed S1579W — re-opens ARM target with disulfide instead of aromatic
4R1088D (single chain)1 stumprelieve R-R electrostatic repulsion with K1104/K1108changes a charged residue; verify no loss of native salt bridge
5D1550C + D1587C (symmetric on chain A→B)2 ARMsymmetric DD→CC double mutant disulfide; loses repulsive salt bridge and gains covalent crosslinktwo-mutation change; higher complexity
6A1078W + A1078W (negative control)1 stumppredicted destabilizer (already listed as steric disruptor)negative control for Phase 1d

Verdict

phase_1c_green — one cluster outside 1579-1581 (stump, 25 residues) with 21 inter-chain disulfide Cb-Cb geometries in range. The existing native C1081 on both chains is a latent cross-chain S-S site that a single A1078C or S1080C mutation converts into a covalent inter-chain lock. Phase 1d AF3 queued.

Ranking delta

h26: B held (no AF3 run yet) — verdict green but AF3 dimerization confirmation is still required before any tier move. Promotion path is explicit: Phase 1d AF3 triple batch (A1078C + S1080C + S1579C all as homotypic symmetric double mutants, plus negative control A1078W) → if any passes homodimer gate (ipTM ≥ 0.50 AND homodimer contacts ≥ Phase 1 WT) then h26 → A-tier and rolls into h03 (drop-in Ultra-Mini mutant increases cochlear-avidity pre-clinical). If all four fail → C-tier.

Files:

Next-step tree

  1. Phase 1d AF3 (user-invoked; needs external AF3 server): 4 jobs at 3 seeds each =
    • A1078C/A1078C Ultra-Mini × 2 chains homodimer
    • S1080C/S1080C Ultra-Mini × 2 chains homodimer
    • S1579C/S1579C Ultra-Mini × 2 chains homodimer
    • A1078W/A1078W Ultra-Mini × 2 chains homodimer (negative control)
  2. Phase 1e disulfide geometry validation: if AF3 passes, extract Cα / Cβ / Sγ coords from AF3 output, confirm engineered S-S geometry (Cα–Cα 6-7 Å, Cβ–Cβ 4-6 Å, Sγ–Sγ < 2.5 Å with proper χ1 χ2 χ3 angles), and compute ΔΔG vs WT from structure-based models.
  3. If any mutant passes AF3 dimer gate → h26 A-tier → h03 drop-in: promote Phase 1d mutant as default Ultra-Mini variant in h03 pAAV clone → coIP uses engineered homodimer as positive control.
  4. If all fail → h26 C-tier, document disulfide engineering branch killed.

Connections