FOLD №29 — Tesamorelin Ala-2 → Aib: DPP-IV Resistance & Helix Nucleation

Verdict: DISCARDED | Class: PERFORMANCE | Target: GHRHR (Q02643)

Mechanism of action (background)

Tesamorelin is a 44-residue synthetic analogue of human growth hormone-releasing hormone (GHRH) that stimulates the pituitary GHRHR to drive endogenous GH secretion. Its key engineering feature is a trans-3-hexenoyl acyl group on the N-terminal Tyr-1, which blocks the canonical DPP-IV cleavage site at the Tyr1-Ala2 amide bond — the same bond that causes rapid degradation of native GHRH (t½ ~7 min in plasma) and early analogues like Sermorelin. This modification produced a clinically viable agent with a half-life of ~26–38 minutes, enabling once-daily subcutaneous dosing and FDA approval in 2010 for HIV-associated lipodystrophy. The receptor-binding pharmacophore involves an amphipathic N-terminal α-helix (residues 1–13) that docks into the GHRHR extracellular domain and transmembrane interface, with residues 1–5 specifically required for receptor activation.

Modification hypothesis (what we tested)

This distillation asked: does substituting Ala-2 with Aib (α-aminoisobutyric acid / 2-aminoisobutyric acid; a Cα,α-disubstituted non-proteogenic residue) add a second, mechanistically independent layer of DPP-IV resistance while simultaneously nucleating the N-terminal α-helix?

The hypothesis rests on two orthogonal mechanisms:

1. Steric occlusion of DPP-IV: Aib's quaternary α-carbon presents a geometry incompatible with DPP-IV's active site, which requires a small penultimate residue (typically Ala or Pro). This is distinct from the acyl cap's mechanism, which blocks the amide nitrogen via steric and electronic effects — making the two modifications potentially additive, not redundant.

2. Helix nucleation: Aib is one of the strongest helix-inducing residues in peptide chemistry, strongly biasing φ/ψ toward canonical α-helical values (~−60°/−45°). Placing it at position 2 could pre-organize the N-terminal receptor-binding segment into its bioactive conformation, potentially improving both receptor affinity and on-rate.

This approach is stereochemically conservative: Aib retains L-configuration (avoiding the helix-disrupting chirality inversion that discarded Fold №2, Sermorelin D-Ala-2) and preserves the non-polar methyl side-chain character of the native Ala (adding only a second methyl group to Cα). It is also positionally targeted: unlike Fold №13 (Tesamorelin Gln-8 → Aib, DISCARDED), which tested Aib at a mid-helix position, this fold places Aib directly at the proteolytic hotspot.

Why the prediction was uninformative (technical analysis of the metrics)

All three primary confidence metrics fall below interpretable thresholds. The pLDDT of 0.467 means that the majority of backbone positions — including the N-terminal helix that is the structural centrepiece of the hypothesis — are not reliably placed by the model. The ipTM of 0.307 indicates that the docked interface geometry cannot be trusted; any inferred Tyr1/Aib2 N-cap arrangement or receptor contact pattern would be speculative artefact rather than meaningful prediction. The absence of Boltz-2 affinity output removes the one metric that could have provided partial signal even under low structural confidence.

This is the second consecutive Aib distillation on the Tesamorelin scaffold to fail on structural confidence grounds. Fold №13 (Gln-8 → Aib, pLDDT 0.49) exhibited nearly identical failure characteristics. The pattern strongly suggests a systematic tool limitation rather than a coincidental per-fold failure:

• The Tesamorelin scaffold is 44 residues with a non-standard N-terminal acyl modification (trans-3-hexenoyl on Tyr-1) that current AlphaFold-derivative models were not trained to handle as a covalent chemical entity
• Aib has restricted, non-standard Ramachandran geometry that general protein folding networks trained primarily on canonical amino acid sequences may not accurately sample
• The combination of these two non-standard features in a single long peptide appears to overwhelm the model's confidence calibration

By contrast, Fold №4 (Ipamorelin N-Me-Aib at position 1, REFINED, pLDDT 0.80) demonstrates that N-terminal non-canonical amino acids are not universally problematic — the failure mode here appears scaffold-specific, tied to Tesamorelin's length and existing chemical complexity.

What this tells us (negative results are data)

Despite the structural prediction failure, this fold is not without value:

1. The hypothesis remains scientifically unrefuted. A DISCARDED verdict here reflects a confidence failure of the prediction tool, not a demonstration that the Aib-2 modification is biologically inactive or receptor-disruptive. The mechanistic logic — orthogonal DPP-IV resistance plus helix nucleation — is intact and is not contradicted by any experimental evidence in the literature.

2. The Tesamorelin + Aib combination is a systematic blind spot for current tools. Two independent Aib placements on this scaffold (positions 2 and 8) have now produced structurally uninformative outputs. This is actionable information for lab tool strategy: attempting a third Aib variant on Tesamorelin using the same pipeline is unlikely to yield interpretable results without a different modelling approach.

3. The acyl cap may create a systematic modelling challenge. The trans-3-hexenoyl modification on Tyr-1 is a covalent chemical group that standard folding pipelines represent imperfectly or not at all. Any modification adjacent to this cap (i.e., at position 2) is particularly likely to produce low-confidence outputs because the model cannot accurately anchor the N-terminal geometry.

4. Chirality inversion (D-Ala, Fold №2) and Aib substitution at position 2 have now both failed to yield structural signal on closely related GHRH scaffolds (Sermorelin and Tesamorelin respectively). This is consistent with the interpretation that position 2 modifications — however chemically rational — are particularly difficult to model at the N-terminus of these acyl-capped peptides.

Alternative hypotheses to test (avoid the failure mode)

Near-term: Change the modelling approach before testing more variants

• Molecular dynamics with Aib-parameterized force fields (e.g., AMBER ff19SB + CMAP Aib parameters, or CHARMM36m with explicit Aib patch): These tools are purpose-built for non-standard residues and do not rely on learned folding confidence scores.
• Ensemble prediction / multiple seeds: If continuing with Boltz-2 or Chai-1, running 5–10 seeds and selecting the consensus structure would provide a more robust signal than single-run pLDDT.
• Shorter constructs: Testing an Aib-2 peptide spanning only residues 1–29 (the minimal pharmacophore), without the disordered C-terminal tail that may contribute to low global pLDDT, could improve structural confidence.

Experimental pathway (bypass the structural prediction bottleneck)

• The literature and mechanistic rationale support direct synthesis and testing of Aib-2 Tesamorelin. Solid-phase peptide synthesis with Fmoc-Aib is routine, and the modification cost is low.
• DPP-IV plasma stability assay (LC-MS/MS monitoring of Tyr1-Aib2 bond cleavage vs. native Tesamorelin and vs. Tesamorelin with acyl cap only): directly tests the additive resistance hypothesis.
• GHRHR binding assay (HTRF or radioligand displacement): tests whether the Aib at position 2 preserves or disrupts receptor affinity.
• GH secretion in primary pituitary cell culture: functional readout of agonism.

Alternative modifications on Tesamorelin targeting PK without position-2 Aib

• C-terminal amidation: Protects against carboxypeptidases; predictors handle this natively and prior folds on related scaffolds have yielded usable confidence scores.
• Lipidation at Lys-12 or Lys-21 (fatty acid conjugation for albumin binding, analogous to semaglutide strategy): extends half-life via a different mechanism entirely, not dependent on DPP-IV resistance at position 2.
• Stapled helix (i, i+4 or i, i+7 hydrocarbon staple) across the N-terminal helix: rigidifies the binding conformation and improves protease resistance without requiring non-canonical α-carbon substitution at position 2.

All results are in silico predictions only. No wet-lab validation has been performed. This is not medical advice.