Pre-registered confirmatory analysis of two mechanistic hypotheses testing whether AC1 modulation is sufficient to explain the full dose-response trajectory and cross-pool heterogeneity across n=155 pooled substrates (SC-3 n=45 + Part 1 synthetic n=20 + Part 2 real-world n=90). Pre-registration DOI: 10.5281/zenodo.20564791. H-SC7-Primary (AC1 θ-sweep binary monotonicity, co-verdict): NOT CONFIRMED. Criterion A (pooled n=155): 4/7 adjacent pairs pass — inversions at θ=0.70→0.80, 0.80→0.90, 0.90→0.95, consistent with SC-3 composition effect anticipated in Gate-0 v10 Cycle 3 M-1. Criterion B (Part 2 n=90): 7/7 PERFECT MONOTONE. H-SC7-Secondary (Cochran's Q three-tier verdict): PARTIAL. Q reduction: 73.2% (28.22→7.56, ≥50% ✅). Q p=0.0228 (<0.05 ❌). Part 1 synthetic sub-pool ρ=−0.321 is the dominant observed contributor to residual heterogeneity. Key finding: AC1 modulation of the MSE↔z relationship is substrate-class-conditional. Part 2 real-world: AC1↑→ρ↓ (Suppressor, 7/7 perfect). SC-3 legacy: AC1↑→ρ↑ (Reverser). Part 1 synthetic: AC1↑→ρ fixed at −0.321 (Bounded). 5-AI adversarial review pipeline (Gate-0 v10): UNCONDITIONAL ACCEPT (Cycle 3).Hardware: AMD Ryzen 9 9800X3D + RTX 4080 SUPER, Windows 11, C:\PHILIA. Additional simulations: 0.