Attaching a fatty acid to a peptide changes almost everything a chemist measures without changing the sequence at all. The backbone is untouched. The residues are untouched. And yet the molecule now carries a hydrophobic tail long enough to bury itself in a protein pocket, in a micelle, or on the wall of a pipette tip. Lipidation is therefore best understood as two subjects that happen to share a reaction: what the tail is for, and what the tail does to every method used to characterize the material afterward.

The tail is a binding element, not ballast

The design logic behind fatty-acid acylation is that serum albumin already transports fatty acids, and a peptide carrying a fatty acid can borrow that transport. The association is reversible and noncovalent: the peptide is not permanently attached to anything, it simply spends a large fraction of its time bound. A 2026 Perspective on fatty-acid-conjugated antimicrobial peptides describes lipidation as enhancing membrane interactions and proteolytic stability, often through binding to serum albumin, and examines common lipidation approaches, including N-terminal acylation and side-chain modification [1].. The scope there is antimicrobial sequences, but the chemistry it catalogs is general.

That framing matters because one tail drives several behaviors at once, and they compete. A lipidated peptide can bind albumin, and it can also bind copies of itself. Work on a series of lipidated GLP-1 analogs leveraged NMR and surface plasmon resonance to elucidate the oligomerization propensity and the albumin interaction of each analog, the authors stating that lipidation affects a peptide's tendency to oligomerize and its interaction with plasma proteins alike [2]. Any account of a lipidated peptide that discusses only albumin is describing half the molecule's behavior.

What albumin's fatty-acid sites look like

The structural picture came from crystallography. Human serum albumin was solved in complex with five molecules of myristate at 2.5 Å resolution; the fatty acids sit in long hydrophobic pockets capped by polar side chains, many of them basic, and those pockets are distributed asymmetrically through the protein despite its symmetrical repeating domain architecture [3]. The asymmetry is the useful part. The sites are not equivalent, so a designed acyl group is not choosing among interchangeable slots.

Which of those sites bind most tightly was addressed by combining crystallography with NMR. Titrating albumin with carboxyl-labeled fatty acid produced three chemical shifts associated with high-affinity pockets, and albumin mutants that disrupt coordination of the bound carboxyl allowed two of them to be assigned: sites four and five, both within domain III, were shown to represent two of the three high-affinity long-chain fatty-acid sites [4].

The map is not closed. A later study using a fluorophore-bearing C12 fatty acid as the probe reported a single high-affinity site with 1:1 stoichiometry, located by X-ray crystallography in subdomain IIA, and used competition with warfarin and ibuprofen to argue that this site is distinct from the classical Sudlow drug-binding sites [5]. Whether that is best described as a new site or as a re-description of known density depends partly on which ligand is doing the reporting. The site inventory should be treated as an active question rather than settled numbering.

Do acylated peptides actually use them?

This is the question that separates a design rationale from a demonstrated mechanism. One study addressed it directly, using methyl-labeled oleic or palmitic acid as NMR probes together with albumin mutants engineered to report on fatty-acid binding. Two-dimensional NMR signals corresponding to the high-affinity sites were assigned, and competitive displacement experiments with a set of acylated peptides identified a primary fatty-acid site in albumin that is used in acylated peptide binding. The authors describe this as an important first step toward a structural basis for the interaction, not as a finished model [6].

Affinity is also not one number. Isothermal titration calorimetry on an acylated growth hormone derivative showed high-affinity binding of one molecule in the approximately 100–1000 nM range and lower-affinity binding of one to two further molecules in the approximately 1000–10000 nM range. The high-affinity site was placed in domain III by size-exclusion chromatography and calorimetry, while a crystal structure with an engineered albumin variant carrying four residue replacements located a low-affinity site at fatty-acid site 6 in domain II [7]. A lipidated molecule can therefore occupy more than one site with affinities an order of magnitude apart, which is not what a single reported dissociation constant conveys.

Where the acyl group is installed

The two approaches already named differ in what they offer. In practice the side chain is most often a lysine ε-amino group, because that is the nucleophile a standard Fmoc building block already supplies. The N-terminus is attractive because there is exactly one of it. A lysine side chain is attractive because it can be placed anywhere in the sequence, which turns position into a design variable.

Position is not a free choice. Three lipidated derivatives of the peptide hormone PYY(3-36) were prepared at different sequence positions — one within the β-turn region and two within the α-helical region, referenced to the published solution NMR structure. They gave distinctly different self-assembled nanostructures from one another and from the unmodified parent, across a range that included micelles, nanotapes, and fibrillar hydrogels depending on concentration, pH, and temperature [8]. Same peptide, same lipid, different attachment point, different physical behavior.

Mechanically, the acylation is usually performed while the peptide is still on the resin. That requires a side-chain protecting group on the target lysine that can be removed selectively, leaving the rest of the protection intact. The 4-methyltrityl (Mtt) group is the workhorse: Nα-Fmoc-Nε-Mtt lysine is commercially available and extensively used for exactly this purpose, and it is conventionally removed with 1–3% trifluoroacetic acid in dichloromethane — conditions that a systematic comparison notes must be adjusted for each specific application to avoid premature loss of Boc groups or cleavage from the linker. The milder alternatives identified in that work, 30% hexafluoroisopropanol or 30% perfluoro-tert-butanol in dichloromethane, removed Mtt while leaving Boc and tert-butyl protection unaffected even after four hours [9]. Other orthogonal handles exist that come off under conditions unrelated to acid, which is the point of having more than one.

Linker chemistry

The acyl chain need not connect straight to the amine. A spacer can be inserted between the two, and that choice is a real design parameter rather than a formality. A series of insulins acylated at position B29 with fatty acids through glutamic acid spacers was examined precisely to deduce the structural requirements for the self-association behavior the authors wanted, using size-exclusion chromatography with light scattering and circular dichroism; their conclusion was that the behavior was critically dependent on the acylation moiety [10]. The spacer is part of that moiety.

What a spacer does separates into a few effects. It moves the hydrophobic tail away from the peptide surface so the tail can reach into a pocket without dragging the peptide against the protein. It adds its own polarity, offsetting some of the hydrophobicity the tail introduced. And it changes conformational freedom. Whether longer is better is genuinely unresolved. A study of octanoyl lipopeptides bearing glycine spacers of one to three residues opens by stating that a clear consensus on how linker length influences physicochemical properties and biological activities remains elusive, then shows that linker length changed both the physicochemical properties and the self-assembly behavior of the series: the shortest-linker compound formed micelles the authors describe as exceptionally stable, while the longer-linker compounds formed micelles above their critical aggregation concentrations that disassembled on dilution [11].

The bond formed matters as much as the length. An amide between an acyl group and an amine is the default because it survives the conditions peptides are normally handled in. Ester linkages to serine, threonine, or tyrosine hydroxyls do not, which turns them from a design option into a side-reaction problem. Thioether linkages made by Michael addition of a cysteine to a maleimide are common across bioconjugation but are subject to retro-Michael exchange, and stabilizing that adduct remains an area of active method development.

Acylation is less selective than the reagent labels suggest

N-hydroxysuccinimide esters are sold as amine-selective reagents, and they are — up to a point. One study reports that even a small molar excess of NHS ester may lead to acylation of hydroxyl-containing amino acids as a side reaction, and evaluated a remedy on two peptides and one protein that were acetylated at a high excess of NHS-acetate: the large molar excess acylated all the primary amines completely, and the hydroxyl-containing amino acids reacted as well. Incubating those solutions in a boiling water bath then hydrolyzed the acetyl-ester bonds while the acetyl-amide bonds remained stable, which the authors offer as an alternative to treatment with hydroxylamine under alkaline conditions [12].

The most striking demonstration is older. Gonadotropin-releasing hormone and two analogs were reacted with sulfonated NHS esters of biotin. The parent peptide contains no reactive amino group and one analog contained only one, yet reversed-phase HPLC showed multiple derivatives for each. Hydroxylamine treatment and mass spectrometry established that substantial O-biotinylation had occurred at Ser4, with a minor contribution at Tyr5. Unrelated peptides reacted under identical conditions showed no significant O-acylation of seryl residues, and running the reaction under denaturing conditions largely abolished it — evidence the authors read as an intrinsically elevated, conformation-dependent reactivity of that particular serine hydroxyl [13]. The general lesson for lipidation in solution is that a peptide with no free amine other than the intended one can still return more than one acylated product.

What the tail does to the analysis

Chromatography

A C16 tail is a large hydrophobic increment on a reversed-phase column. Retention increases, often enough that a gradient developed for the parent peptide is no longer fit for purpose, and peak shape can degrade when the analyte exchanges between free monomer and aggregate on the timescale of the separation. The practical consequence is that purity by area percent becomes harder to interpret. Area percent is a ratio whose denominator is the total integrated area of the peaks that eluted and were detected. Which way material that fails to elute pushes that ratio depends on what the missing material is. Strongly retained impurity that stays on the column is absent from the denominator while the main peak is unchanged, so the reported purity rises. Aggregated analyte that never elutes as a resolved peak is missing from the numerator and the denominator alike, which pushes the reported purity down instead. Either way the figure describes only the material that eluted and was detected, not the material that was injected.

Mass arithmetic

The mass shift from acylation is the acid minus water, because an amide bond forms with loss of water. Palmitic acid is C16H32O2; acylating an amine with it adds a net C16H30O to the peptide's formula, because the acid gives up its hydroxyl and the nitrogen gives up a hydrogen. That increment is 238.23 Da monoisotopic and 238.41 Da average. A γ-glutamyl spacer contributes a glutamyl residue, C5H7NO3, at 129.04 Da monoisotopic. A peptide of monoisotopic mass M acylated on a lysine ε-amine with palmitic acid through a single γ-Glu spacer therefore appears at M + 129.04 + 238.23 = M + 367.27. Two checks follow from that.

  • M + 238.23 — acylated directly, without the spacer increment. Consistent with a construct in which the spacer coupling failed, or with a regiochemical variant.
  • M + 605.50 — the intended spacer-plus-palmitoyl increment carrying a second palmitoyl, 367.27 + 238.23. This is what over-acylation of a second nucleophile looks like on a spectrum of the spacered construct, and it is the species the selectivity problems described above generate.

Note that neither check distinguishes positional isomers. A peptide acylated on the intended lysine and the same peptide acylated on a different lysine have identical masses and are separated, if at all, by chromatography rather than by the mass spectrometer.

Self-assembly, and why concentration stops being simple

Lipidated peptides are surfactants. All three lipidated PYY(3-36) derivatives described above self-assembled above a critical aggregation concentration determined by pyrene fluorescence, formed nanostructures characterized by small-angle X-ray scattering and cryogenic electron microscopy, and gelled at higher concentrations in certain pH windows where the unmodified peptide did not [8]. The state of the material therefore depends on the concentration at which it is measured. A UV determination made above the aggregation threshold and a scattering measurement made below it are not describing the same species.

Two further consequences are standard practice rather than findings from any one paper. Amphiphilic molecules adsorb to glass and plastic, so apparent concentration can fall during handling in a way that tracks surface area rather than elapsed time. And size-exclusion chromatography of a self-associating lipopeptide reports an apparent size that depends on running concentration and buffer, not a molecular weight.

What is not known

The inventory and relative affinities of albumin's fatty-acid sites are still being revised, and different probe molecules give different answers [4, 5]. The structural basis for how an acylated peptide — as opposed to a free fatty acid — occupies those sites is described by the people who studied it directly as a first step [6]. Whether a longer linker improves or degrades a given property has no consensus answer [11]. And the relationship between a lipopeptide's self-assembly behavior and its other properties is characterized, rather than predicted, in most of the work cited here. A lipid tail is a reliable way to change a peptide's physics. It is not yet a predictive one.