I want to add some structural detail about the albumin interaction itself, because it's more sophisticated than simply "fatty acid sticks in a hydrophobic pocket." Human serum albumin has 7 fatty acid binding sites (FA1-FA7), identified crystallographically by Bhattacharya et al. The C-18 diacid of semaglutide primarily occupies FA site 4 (within subdomain IIIA, overlapping with Sudlow site II), based on competition binding and mutagenesis studies. Key features of this interaction: Hydrophobic contacts: The C-18 alkyl chain buries ~250 Ų of hydrophobic surface area in the FA4 tunnel. Each additional CH₂ unit (going from C-16 to C-18) adds ~25 Ų and contributes ~0.8 kcal/mol binding energy. Ionic interactions: The terminal carboxylate of the diacid forms salt bridges with Arg410 and Tyr411 of albumin. The second carboxylate (α to the carbonyl linking to the OEG spacer) interacts with Lys414. These ionic contacts are absent in the monoacid (liraglutide), contributing to the affinity differential. > "Crystallographic analysis of albumin-fatty acid complexes revealed that the C-18 octadecanedioic acid occupied FA site 4 with a Kd of 1.8 μM, compared to 7.2 μM for the C-16 palmitic acid, with the additional binding energy primarily derived from two CH₂ hydrophobic contacts and a second ionic interaction." > — Petitpas et al., *Journal of Molecular Biology*, 2001; 314(5):955–960 Species differences: Albumin binding affinity varies significantly across species, which complicates preclinical-to-clinical PK translation. Rat albumin binds semaglutide's fatty acid ~3-fold weaker than human, leading to shorter half-life in rat PK studies.

Great structural detail. To complete the picture on why semaglutide specifically uses the C-18 DIACID rather than longer chains or alternative lipid modifications: Why not C-20 or longer? - C-20 fatty acids showed improved albumin binding in vitro (~1.5-fold over C-18) - BUT: increased aggregation propensity. Longer acyl chains promote self-association of the peptide-lipid conjugate at high concentrations, causing injection site reactions and formulation instability - Also: diminishing returns on half-life extension. Beyond ~99.5% albumin binding, increasing affinity barely changes free fraction Why diacid, not monoacid? - Improved aqueous solubility (extra carboxylate is ionized at pH 7.4) - Better albumin binding (as detailed above) - Reduced CYP450 metabolism of the fatty acid chain (ω-oxidation blocked by carboxylate) > "The structure-activity relationship for albumin-binding lipid modifications of GLP-1 analogs followed a bell-shaped curve: optimal half-life extension was achieved with C-18 to C-20 diacids, while C-22+ analogs showed reduced solubility, increased aggregation, and paradoxically shorter in vivo half-lives due to rapid clearance of aggregates." > — Madsen et al., *Journal of Medicinal Chemistry*, 2007; 50(24):6126–6132 The Novo Nordisk team screened over 100 lipid-linker combinations to arrive at the final semaglutide structure. It's a beautiful example of multiparameter optimization in medicinal chemistry.

To close this thread with clinical PK implications: Steady-state considerations: With a t½ of ~168 hours and weekly dosing, semaglutide reaches steady state at approximately 4-5 weeks (4-5 half-lives). The accumulation ratio (AUCss/AUCsingle dose) is approximately 3-fold. This means the steady-state trough concentration is ~3x higher than the trough after a single dose. Clinically, this explains why: 1. Full efficacy isn't seen until 4-5 weeks at each dose level 2. Side effects (especially nausea) may emerge during the accumulation phase even at a stable dose 3. After discontinuation, pharmacological effects persist for 5-7 weeks (5 half-lives for washout) The albumin reservoir effect on drug interactions: Because >99% of semaglutide is albumin-bound, displacement interactions are theoretically possible. However, since semaglutide binds to fatty acid sites (not Sudlow site I where most drugs bind), clinically significant displacement is unlikely. Warfarin, ibuprofen, and other common albumin-binding drugs don't compete. > "No clinically significant pharmacokinetic interactions were observed between semaglutide and warfarin, digoxin, metformin, or atorvastatin in dedicated drug-drug interaction studies, consistent with the absence of binding site competition on human serum albumin." > — Kapitza et al., *Clinical Pharmacokinetics*, 2018; 57:733–745 The gastric emptying effect on oral drug absorption is a different story entirely — but that's for another thread.