DISTILLATION №30 — DISCARDED

MOTS-c i,i+4 Stapled Helix (Glu-5/Lys-9)

Verdict: DISCARDED — structural prediction uninformative pLDDT: 0.60 | ipTM: 0.23 | pTM: 0.57

Mechanism of Action (Background)

MOTS-c is a 16-amino acid mitochondrial-derived peptide (MDP) encoded within the 12S rRNA of the mitochondrial genome. Its primary established mechanism (Lee et al., 2015, PMID:25738459) is inhibition of the folate cycle and de novo purine biosynthesis, leading to accumulation of AICAR (5-aminoimidazole-4-carboxamide ribonucleotide). AICAR is a well-validated indirect activator of AMPK — the master cellular energy sensor — which drives glucose uptake, fatty acid oxidation, and metabolic homeostasis. The AICAR–AMPK axis is the consensus signaling route in the literature (PMID:36677050, PMID:36761202), placing MOTS-c's AMPK engagement primarily in the indirect category.

Beyond metabolic regulation, MOTS-c has demonstrated activity in gestational diabetes, bone metabolism, pulmonary fibrosis, ovarian cancer (where it competes for protein-protein interaction surfaces with LARS1/USP7), and cardiovascular conditions. Nuclear translocation under stress conditions has been documented, implicating multiple conformational modes and binding partners across cellular compartments. No experimental structure of MOTS-c — free or bound — has been published, and no residue-level SAR data exists.

Modification Hypothesis (What We Tested)

Fold #30 introduced an all-hydrocarbon i,i+4 staple between positions 5 and 9 of MOTS-c, replacing Met-5 and Tyr-9 with (S)-2-(4'-pentenyl)alanine (S5) non-natural residues followed by ruthenium-catalyzed ring-closing metathesis:

Native:  M  R  W  Q  E  M  G  Y  I  F  Y  P  R  K  L  R
         1  2  3  4  5  6  7  8  9  10 11 12 13 14 15 16
Stapled: M  R  W  Q  E [S5] G  Y  I [S5] F  Y  P  R  K  L  R

The design rationale was threefold:

1. Conformational pre-organization: Enforce a single helical turn spanning residues 5–9 to reduce the entropic cost of AMPK engagement, exploiting the well-established principle that intrinsically disordered peptides bind targets with lower affinity due to the conformational entropy penalty.
2. Oxidation resistance at Met-5: Replacing oxidation-prone Met-5 with a non-oxidizable S5 stapling residue addressed the same stability liability flagged in the Nle substitution of Fold #5 — a two-for-one modification.
3. Preservation of the binding surface: The staple was positioned on the solvent-exposed face (Met-5/Tyr-9), leaving the cationic C-terminal patch (R12/K13/R16, the AMPK-engaging surface explored computationally in Fold #19) and aromatic residues Trp-3/Phe-10 unperturbed.

This was the first application of hydrocarbon stapling to MOTS-c or any MDP in the literature, and the first stapled peptide in this lab's longevity series.

Why the Prediction Was Uninformative

Global structural confidence did not improve. pLDDT returned at 0.60 — marginally lower than the native MOTS-c baseline of ~0.62 observed across Folds #5, #19, and #25. For a single-run prediction, this difference is within noise. The predicted improvement to >0.65–0.70 in the stapled region did not materialize.

Inter-chain docking confidence collapsed. The ipTM of 0.23 is the most diagnostic metric here. For a peptide–protein complex prediction, ipTM values below ~0.40 indicate that the model cannot confidently place the peptide at the target interface. Boltz-2 essentially could not predict a meaningful bound geometry for the stapled MOTS-c against the AMPK alpha-2 catalytic subunit. The affinity module returned no values, consistent with this failure mode.

The predicted helical turn was not observed. Despite the staple spanning residues 5–9, the structural output did not show a clearly defined α-helical segment with elevated local pLDDT in that region. The C-terminal cationic patch (R12/K13/R16) and aromatic residues appeared structurally unperturbed — consistent with design intent — but absence of disruption at the retained surface is not evidence of functional improvement.

Possible technical contributors to the failure:

• Boltz-2 and similar structure predictors are trained primarily on natural amino acids. The S5 non-natural stapling residue (pentenyl side chain with a covalent hydrocarbon bridge) is chemically exotic and may not be modeled accurately, potentially explaining why the expected local helical stabilization was not predicted.
• The staple is represented as a sequence-level modification, but the covalent constraint geometry requires accurate modeling of the metathesis-derived bridge — a feature that current AlphaFold-family tools handle imperfectly at best.
• Single-run prediction without ensemble or multiple seeds cannot distinguish whether the low confidence reflects genuine disorder or prediction uncertainty from the non-natural residue representation.

What This Tells Us (Negative Results Are Data)

The structural ambiguity of MOTS-c is not resolved by i,i+4 hydrocarbon stapling at positions 5/9 — at least not in a way current predictors can detect. Across five MOTS-c folds in this lab:

The 0.62–0.63 pLDDT cluster is now a reproducible feature of MOTS-c predictions regardless of modification class (point substitution, lipidation, covalent constraint). This consistent ceiling suggests either: (a) the sequence intrinsically lacks helical propensity that current tools can amplify; (b) the AMPK alpha-2 complex geometry is not well-captured for this peptide class; or (c) both.

The negative result also raises a literature-grounded concern: if MOTS-c's primary AMPK mechanism is indirect (via AICAR accumulation through folate/purine pathway inhibition), then conformational optimization for direct AMPK binding may be chasing the wrong target entirely. Locking MOTS-c into a single helical conformation could actively impair its multi-compartment function (cytoplasmic metabolism, nuclear translocation, membrane interaction), consistent with the literature agent's assessment that conformational flexibility may be required for MOTS-c's broad activity profile.

The Tyr-9 → S5 substitution introduces a specific concern that deserves explicit note: Tyr-9 is an aromatic residue with a hydroxyl group that could contribute to hydrogen bonding at a binding interface. Replacing it with a hydrophobic pentenyl chain eliminates both features, a modification that could reduce biological activity independent of the helical constraint effect.

Alternative Hypotheses to Test

Given the consistent structural prediction ceiling for MOTS-c and the indirect-mechanism concerns from the literature:

1. Shorter, truncated analogs: Rather than constraining the full 16-mer, identify a minimal active fragment (e.g., the C-terminal cationic tail, residues 12–16, as a standalone peptide) and test whether a shorter sequence achieves better structural confidence. The conformational entropy problem scales with sequence length.

2. Extend the lipidation strategy from Fold #25: Myristoylation returned PROMISING at pLDDT 0.63 with a delivery-focused rationale. Testing a palmitoyl (C16) or stearoyl (C18) variant, or adding a PEG spacer between the fatty acid and Met-1, may improve membrane engagement metrics without disrupting the functional sequence. This parallels the palmitoylation work in Fold #27 on FOXO4-DRI.

3. Lactam stapling as an alternative constraint chemistry: The TB-500 lactam cyclization fold in the recent lab history demonstrated that ring-forming strategies can produce interpretable outputs. A Glu-5/Lys-9 or Asp-5/Lys-9 lactam bridge — avoiding the non-natural amino acid representation problem that may have confounded this fold — could be attempted as a chemically distinct conformational constraint on the same helix-spanning positions.

4. C-terminal modification building on Fold #19: K13R delivered a modest but consistent PROMISING signal. A double substitution (K13R + R12 methylation or R16 modification) targeting the cationic patch more aggressively, without touching the central region, avoids the conformational disruption risk entirely.

5. Abandon direct AMPK targeting; explore AICAR pathway mimicry: If MOTS-c acts primarily via AICAR, designing an analog optimized for folate cycle enzyme inhibition (direct binding to ATIC or MTHFD enzymes) rather than AMPK engagement may be mechanistically better aligned. This would require target switching and a new structural hypothesis.