Structure-activity relationships of GLP-1 analogs

Let's trace the SAR evolution chronologically through the approved drugs: EXENATIDE (Byetta, 2005) — the lizard solution: Exendin-4, from Gila monster (Heloderma suspectum) saliva, shares ~53% sequence homology with GLP-1 but is naturally DPP-4 resistant (Gly² instead of Ala²) and NEP-resistant. Key structural differences from GLP-1: - Position 2: Gly (not Ala) — DPP-4 resistance - Position 3: Glu (not Glu in GLP-1 too, but different flanking residues) - C-terminal extension: 9-amino acid "Trp-cage" motif (PSSGA PPPS-NH₂) that stabilizes the α-helical structure - Multiple substitutions throughout that improve proteolytic stability > "The Trp-cage motif at the C-terminus of exendin-4 (residues 32-39) forms a compact tertiary structure that shields the C-terminal helix from proteolysis and contributes approximately 3-fold to receptor binding affinity through direct ECD contacts." > — Runge et al., *Journal of Biological Chemistry*, 2008; 283(17):11340–11347 Half-life: ~2.4 hours → requires BID injection. Better than 2 minutes, but not ideal.

LIRAGLUTIDE (Victoza, 2010) — the acylation breakthrough: Instead of finding a naturally resistant peptide, Novo Nordisk took native GLP-1 and engineered stability through lipid conjugation: Key modifications from GLP-1(7-37): - Position 34: Lys → Arg (prevents acylation at this site) - Position 26: Lys side chain acylated with C-16 palmitic acid via glutamic acid spacer These two changes + the C-16 acylation achieved: - Albumin binding → ~98% bound → renal clearance protection - Self-association → heptameric complexes at injection site → slow absorption - DPP-4 resistance → the albumin-bound form is sterically protected from DPP-4 > "Liraglutide self-associates to form heptameric complexes at the subcutaneous injection site concentration, with an in vitro dissociation half-life of approximately 4 hours, creating a depot effect that contributes to the 13-hour absorption half-life independent of albumin binding." > — Knudsen et al., *Journal of Medicinal Chemistry*, 2000; 43(9):1664–1669 Half-life: ~13 hours → once-daily injection. A 400-fold improvement over native GLP-1. The elegance: only TWO amino acid changes + a fatty acid. The peptide backbone is 97% native GLP-1 sequence.

SEMAGLUTIDE (Ozempic, 2017) — optimizing the acylation: Building on liraglutide, Novo Nordisk made three additional modifications: 1. Position 8: Ala → Aib (α-aminoisobutyric acid): The gem-dimethyl group sterically blocks DPP-4 access to the scissile bond. This provides DPP-4 resistance independent of albumin binding — critical for the free fraction. 2. Position 34: Lys → Arg (same as liraglutide) 3. Position 26: C-18 diacid via OEG-OEG linker (as discussed in the albumin binding thread) Why Aib at position 8 was the key innovation: > "Incorporation of Aib at position 8 of GLP-1 analogs increased DPP-4 resistance from t½ = 4.7 hours (Ala⁸) to t½ > 120 hours (Aib⁸) while maintaining 95% GLP-1R binding affinity, as the additional methyl group fills a small hydrophobic pocket near the peptide N-terminus that is not involved in receptor contacts." > — Lau et al., *Journal of Medicinal Chemistry*, 2015; 58(18):7370–7380 The combination of Aib⁸ (protease resistance) + C-18 diacid with OEG linker (albumin binding) produced: - Half-life: ~168 hours - ~13-fold improvement over liraglutide - ~5,000-fold over native GLP-1 Total changes from native GLP-1(7-37): Position 8 (Ala→Aib), position 34 (Lys→Arg), position 26 (Lys acylated). THREE modifications. That's it. The rest is native human GLP-1.

TIRZEPATIDE (Mounjaro, 2022) — the chimeric approach: Tirzepatide takes a completely different design philosophy. Instead of modifying GLP-1, it's based on the GIP(1-42) backbone with modifications to introduce GLP-1R activity: The peptide is a 39-amino acid chimera: - Positions 1-14: primarily GIP-derived (responsible for GIPR agonism) - Positions 15-39: hybrid GIP/GLP-1 sequence (provides GLP-1R activity) - Position 2: Aib (DPP-4 resistance, same principle as semaglutide) - Position 20: Lys, acylated with C-20 eicosandioic acid via OEG linker (albumin binding, longer than semaglutide's C-18) - C-terminus: amidated > "Tirzepatide was designed by grafting key GLP-1R-activating residues onto a GIP scaffold, creating a unimolecular dual agonist with full efficacy at GIPR (EC₅₀ 0.15 nM) and partial agonism at GLP-1R (EC₅₀ 0.52 nM, Emax ~79%)." > — Coskun et al., *Molecular Metabolism*, 2018; 18:3–14 Key SAR insights for the dual agonism: - Position 1 (Tyr): Common to GIP and GLP-1; essential for both receptors - Position 12 (Ile→Phe swap): GLP-1 has Phe¹², GIP has Ile. Tirzepatide uses Ile but compensates with other GLP-1R-contacting residues - Position 27: Critical for GLP-1R selectivity; tirzepatide's choice here provides the partial GLP-1R agonism The C-20 fatty diacid (vs. C-18 in semaglutide) provides slightly higher albumin affinity, contributing to the ~120-hour half-life.