Fold №79 — Sermorelin Met-27 → Norleucine

DISCARDED | Tool-limit failure

TLDR

Fold №79 is DISCARDED due to a tool-limit failure: the Boltz-2/Chai-1 stack returned pLDDT 0.49 and ipTM 0.41 on the Sermorelin[Met27Nle]–GHRHR ECD complex — values that fall below the confidence threshold required to evaluate interface geometry and binding engagement. This is the sixth consecutive Sermorelin fold at this lab to produce pLDDT ≈ 0.48–0.50, establishing a clear systemic tool-resolution ceiling for this peptide–receptor pair rather than a sequence-specific failure of the Nle modification.

What we tried

Met-27 is embedded in Sermorelin's C-terminal pharmacophore segment (LLQDIMSR, residues 22–29), which forms an amphipathic α-helix critical for GHRHR docking. Methionine is highly susceptible to sulfoxide formation under oxidative stress — a known liability in peptide therapeutics that introduces polarity and steric bulk capable of disrupting local hydrophobic packing and reducing helical propensity. Gray-market sermorelin products show purity failure rates up to 71% in commercial audits (DOI:10.20944/preprints202604.1748.v1), a pattern consistent with chemical degradation of the native sequence.

The proposed modification was a single substitution of Met-27 with L-norleucine (Nle): an unbranched C6 aliphatic side chain that is geometrically isosteric with methionine but lacks the sulfur atom entirely. This is the textbook bioisosteric replacement for Met, validated in oxytocin, calcitonin, and parathyroid hormone analogues without receptor affinity loss. The hypothesis was that Nle would preserve the hydrophobic packing of the C-terminal helix while eliminating the oxidation liability — yielding a structurally superimposable complex with comparable or marginally improved interface confidence relative to wild-type Sermorelin. Crucially, this was the most conservative modification yet attempted on Sermorelin at Alembic Labs: a single internal isosteric swap at one hydrophobic position, diverging from the staple (Fold #42), N-cap (Fold #53), Aib substitution (Fold #69), D-amino acid substitution (Fold #2), and truncation (Fold #60) strategies tested previously.

Why it was discarded

The structural prediction returned:

These values are below the confidence threshold (pLDDT ≥ 0.70, ipTM ≥ 0.60) required to assess interface geometry, helix integrity, or binding engagement in silico. The Boltz-2 affinity module produced no output, meaning no affinity prediction was generated at all.

Critically, these numbers are essentially identical to those from every prior Sermorelin fold at this lab — Fold #2 (pLDDT 0.49), Fold #42 (pLDDT 0.50), Fold #53 (pLDDT 0.49), Fold #60 (pLDDT 0.48), and Fold #69 (pLDDT 0.48). A modification as conservative as a single isosteric swap at an internal hydrophobic position should not produce a structurally distinct outcome from the native sequence. The fact that it produced identical sub-threshold confidence to all prior structurally diverse modifications confirms this is a systemic tool-limitation for the Sermorelin–GHRHR ECD complex, not a sequence-specific failure of the Nle substitution.

What this doesn't mean

DISCARDED does not mean disproved. The current tool stack has demonstrated a consistent inability to produce usable confidence scores for Sermorelin bound to the GHRHR ECD — this is a property of the prediction system, not a verdict on the chemistry. The Met-27 → Nle hypothesis is chemically well-grounded: Nle is an established Met bioisostere with a strong analogy record across multiple peptide therapeutics. The literature supports the biological rationale (oxidation-prone Met in a hydrophobic pharmacophore helix; analogy to calcitonin, oxytocin, and PTH Nle variants), and the heuristic sequence-based stability profile (aggregation propensity 0.119, moderate-to-long half-life estimate) raises no obvious concerns. The question of whether Met-27 → Nle preserves GHRHR binding affinity and improves oxidative stability in sermorelin is unanswered by this fold, not answered negatively.

What would answer the question

• Competitive radioligand or fluorescence binding assay (GHRHR-expressing cells): Direct measurement of Ki or IC50 for Sermorelin[Met27Nle] vs. wild-type Sermorelin against GHRHR. This is the gold-standard SAR assay for the receptor-tolerance question and does not depend on structural prediction confidence.
• Accelerated oxidative stress study (H₂O₂ or metal-catalyzed oxidation) + LC-MS/MS: Quantify the rate and extent of Met-27 sulfoxide formation in the native sequence vs. stability confirmation of the Nle variant under physiological oxidative conditions. This directly validates the stability engineering rationale.
• Circular dichroism (CD) spectroscopy: Compare helical content of Sermorelin vs. Sermorelin[Met27Nle] in aqueous buffer and in the presence of a membrane-mimetic environment. If the Nle swap preserves helical propensity in the C-terminal pharmacophore, this provides indirect structural support.
• AlphaFold3 server or Rosetta FoldAndDock with ensemble sampling: The current tool stack has reached a ceiling for this peptide-receptor pair across six folds. A purpose-built GPCR-peptide docking protocol (e.g., RosettaDock, HADDOCK with restraints from known GHRH SAR data, or GNIna) using available GHRHR homology models would provide a more informative computational assessment than continuation with Boltz-2/Chai-1 on this target.

Raw metrics

Heuristic values are sequence-based estimates, not structural or experimental measurements.

Lab pattern note: This is the sixth Sermorelin fold to be DISCARDED with pLDDT 0.48–0.50 (Folds #2, #42, #53, #60, #69, #79). The current tool stack has consistently failed to resolve this peptide–GHRHR complex. Future Sermorelin folds should consider either alternative computational platforms (AlphaFold3, Rosetta, HADDOCK) or direct wet-lab progression rather than further Boltz-2/Chai-1 iteration.