DISTILLATION №22 — PROMISING

Humanin S14C: Disulfide-Bridged Cyclization to Pre-Organize the BAX-Binding Helix

Peptide: Humanin (modified) — MAPRGFSCLLLLTCEIDLPVKRRA
Class: LONGEVITY
Target: BAX (UniProt Q07812)
Modification: Ser-14 → Cys, forming an intramolecular disulfide with native Cys-8 (i,i+6 register)
Verdict: PROMISING
pLDDT: 0.558 | pTM: 0.470 | ipTM: 0.357

Mechanism of Action

Humanin (HN) is a 24-amino acid mitochondrial-derived peptide that directly binds the pro-apoptotic BCL-2 family protein BAX and prevents its activation cascade. Under apoptotic stress, BAX undergoes conformational activation, exposes its N-terminus and BH3 domain, translocates from the cytosol to the mitochondrial outer membrane, oligomerizes, and drives MOMP — releasing cytochrome c and committing the cell to caspase-dependent death. HN intercepts this cascade upstream: Guo et al. (2003, PMID:12732850) demonstrated by co-immunoprecipitation and functional assay that HN physically binds BAX, blocks its conformational change, and prevents mitochondrial translocation. This is a direct, physical anti-apoptotic mechanism — not a downstream signaling effect.

Morris et al. (2019, PMID:31690630) extended this picture using CD spectroscopy, fluorescence, and negative-stain EM, revealing that HN and BAX co-form fibers with BAX undergoing secondary and tertiary structural rearrangements upon engagement. Critically, HN mutations that alter anti-apoptotic potency also alter fiber morphology — directly linking HN's own conformation to its functional output. Luciano et al. (2005, PMID:15661735) demonstrated HN also inhibits BimEL and other BCL-2 family pro-apoptotic members, but the BAX axis is the most mechanistically resolved and is the focus of this distillation.

Modification Rationale

Native HN is largely disordered in solution. Its anti-apoptotic function requires the central amphipathic segment — spanning approximately residues 7–17, encompassing the LLLL hydrophobic tract and the EIDL acidic region — to adopt an α-helical conformation competent to engage the BAX hydrophobic groove or fiber interface. This conformational disorder imposes an entropic cost upon binding: HN must fold-upon-binding, reducing the effective affinity relative to a pre-organized helix.

The S14C modification exploits the presence of native Cys-8 to introduce a disulfide bridge with minimal mutational footprint — only a single substitution is required. An i,i+6 disulfide spans approximately two turns of an α-helix (Cα–Cα ~9–10 Å, compatible with SS geometry) and is well-established in the constrained peptide literature as an effective helix-nucleating constraint. Ser-14 was selected because it sits on the solvent-facing, polar face of the helix in the predicted helical model, away from the hydrophobic Leu-rich face hypothesized to contact BAX — minimizing disruption to the binding interface while enforcing backbone geometry.

This modification also opens a new experimental avenue for the lab: DISTILLATION №22 is the first cyclization-class modification in the LONGEVITY series, complementing the substitution strategies explored in Fold #19 (MOTS-c K13R) and Fold #17 (SS-31 Nal), and providing a structural pre-organization hypothesis not previously tested on any peptide in the recent rotation.

Predicted Properties — Where the Signal is Moderate

The structural prediction places S14C-Humanin near BAX with a partially helical central segment and a Cys8–Cys14 register consistent with the intended disulfide geometry. This is the predicted structure's most encouraging signal: the folding algorithm independently arrives at a helical register that would accommodate the proposed SS bond, without being explicitly told to enforce it. However, current AF/Chai tools do not model disulfide bonds as covalent constraints — the SS bond is inferred from geometry, not enforced, which means the actual constrained peptide could behave differently.

The low ipTM (0.357) is the primary limiting factor: the docking interface between peptide and BAX is not confidently resolved. The hydrophobic Leu face is partially exposed but not unambiguously positioned in the BAX hydrophobic groove. The PROMISING verdict reflects genuine structural signal in the peptide itself, not a confirmed binding event.

What Would Strengthen This Signal

Computational next steps:

1. Ensemble prediction with explicit SS constraint: Run the S14C variant with explicit disulfide bond modeling in tools that support covalent constraint definition (e.g., RosettaFold with constraint files, or Schrödinger's CovDock). A single unconstrained prediction cannot capture the true effect of the covalent loop on backbone geometry.

2. MD simulation of constrained vs. unconstrained HN: Molecular dynamics with the disulfide bond explicitly enforced (GROMACS/AMBER with SS bonding) would quantify the change in helical content and conformational entropy of residues 8–14, directly testing the hypothesis that the bridge pre-organizes the helix.

3. Chai-1 / Boltz-2 affinity module re-run: This fold lacked Chai-1 agreement scores and Boltz-2 affinity values. A second-model agreement run would substantially increase or decrease confidence in the predicted pose.

4. Comparison fold — lactam analog: Run the lactam-bridged equivalent (Lys-8, Glu-14 → lactam, or vice versa) as a non-reducible helix staple at the same i,i+6 position. If the lactam variant shows higher ipTM and similar helix geometry, it would both validate the pre-organization concept and sidestep the redox limitation (see below).

5. Wild-type Humanin baseline fold: A direct structural comparison fold of native HN against BAX would establish a pLDDT/ipTM baseline, allowing the S14C modification to be assessed as a delta rather than in isolation.

Critical biological question to resolve computationally:

The S14G substitution (HNG) is the most potent known HN analog, and Gly at position 14 specifically increases local flexibility — the opposite of disulfide constraint. A direct structural comparison fold of HNG (S14G) vs. S14C-Humanin against BAX, under identical prediction conditions, would test whether the predictor favors the flexible or rigid variant for BAX engagement. If HNG scores higher ipTM than S14C-HN, the flexibility hypothesis would be supported and the disulfide strategy would warrant reconsidering. This comparison fold is the single most informative next experiment available without wet lab resources.

Wet lab validation priorities (if resources allow):

1. Oxidative folding and HPLC verification of disulfide bond formation in the purified S14C peptide.
2. CD spectroscopy comparing S14C-HN vs. native HN helical content in buffer — directly testing the pre-organization hypothesis.
3. BAX pull-down or SPR binding assay under non-reducing conditions (to maintain disulfide) vs. reducing conditions (to demonstrate disulfide-dependence of any affinity gain).
4. Cell-based apoptosis assay (staurosporine or H₂O₂ challenge) in reducing intracellular context — the critical test of whether disulfide-mediated pre-organization survives to the relevant cellular compartment.

Unresolved Complications

Two issues carry enough weight to be flagged prominently:

1. The HNG paradox. Position 14 is the defining residue of HNG, the most potent HN analog — but HNG's Gly substitution increases flexibility, not rigidity. If enhanced potency at position 14 is mechanistically linked to local conformational freedom (enabling induced-fit engagement or fiber-competent rearrangement), the disulfide constraint is working against the grain of the SAR data at this exact position. This is not a reason to discard the hypothesis — it may be that HNG's potency derives from reduced steric clash rather than increased flexibility — but it is a genuine contradicting signal that the next experiments should be designed to resolve.

2. Redox context. BAX inhibition is required in the cytosol — a reducing environment maintained by millimolar glutathione. A disulfide-bridged peptide delivered extracellularly would need to reach the cytosol with its SS bond intact, which is not guaranteed. If disulfide reduction occurs before intracellular delivery, the modification provides no advantage over native HN. This is a fundamental translational constraint. Non-reducible staples (hydrocarbon, lactam) at the same i,i+6 position would be the rational next iteration if wet lab data confirms pre-organization benefit.

In silico only. All properties are computational predictions, not experimental measurements. This report does not constitute medical advice. Heuristic peptide properties (aggregation, stability, BBB, half-life) are sequence-based estimates, not validated assay results. Disulfide bond geometry is inferred from predicted Cα positions; the SS bond is not explicitly modeled by the prediction tools used.