Peptide modifications for stability and half-life: lipidation, PEGylation, Aib, cyclisation and fusion

A peptide that works in a receptor assay often fails in an animal, for two reasons that have nothing to do with its binding affinity. Proteases cut it, and the kidney filters it out. Native glucagon-like peptide-1 has a half-life of about 1.5 minutes after intravenous dosing in humans and 1.5 hours after subcutaneous dosing [2], and generally “proteins and peptides smaller than approximately 70 kDa are more likely to be eliminated by kidney filtration than are larger proteins” [1].

Peptide engineering is the set of chemical answers to those two problems. This guide covers each one with worked examples from molecules in this library: fatty acid acylation and albumin binding, PEGylation, non-natural residues such as Aib and D-amino acids, cyclisation and stapling, terminal capping, and protein fusion. It explains what each modification does, what it costs to make, and what the published evidence shows.

Two clearance problems, two families of fix

Problem	Mechanism	Modifications that address it
Proteolysis	Specific peptidases, for example dipeptidyl peptidase-4 cleaving the N-terminal dipeptide of incretins [9]	Aib and other non-natural residues, D-amino acids, N-methylation, cyclisation, terminal capping, backbone modification [16] [3]
Renal filtration	Glomerular filtration of molecules below roughly 70 kDa, influenced by size, shape, hydrodynamic radius and charge [1]	Fatty acid acylation with albumin binding, PEGylation, fusion to Fc or albumin [15] [13] [1]

The two are not independent. Albumin binding addresses both at once: the stated principle behind fatty-acid derivatisation of GLP-1 was “to prepare analogs that could bind to albumin in a reversible manner and, thereby, protect the peptide from both DPP-IV degradation and renal filtration” [2].

Lipidation: borrowing albumin’s half-life

Lipidation attaches a fatty acid or fatty di-acid to the peptide, typically on a lysine side chain. The lipid then binds serum albumin, and the peptide inherits albumin’s slow clearance. A 2019 review describes lipidation as “a clinically-proven post-translational modification” that improves half-life, reduces immunogenicity and can enable intracellular uptake and transport across epithelia [15].

Why albumin

Human serum albumin is abundant and long-lived: about 10 to 15 g is produced by the liver daily, giving a plasma concentration of 35 to 50 g/L, roughly 0.6 mM, and its own half-life runs to several weeks [2]. That longevity comes from pH-dependent binding to the neonatal Fc receptor, the same recycling mechanism that protects antibody Fc domains [2]. Crystallography has mapped seven binding sites for the fatty acids C10, C12, C14, C16 and C18 on albumin, so its capacity to carry lipidated cargo is large [2]. The half-life extension achieved by lipidation “is obtained by an increased binding to albumin, which is the most abundant protein within the extracellular fluid” [3].

The structure-activity rules

The relationship between lipid and activity was worked out explicitly for GLP-1. A 2000 study derivatised the 30-residue hormone with fatty acids and reported that all compounds with fatty acids of 12 carbons or longer were strongly protracted compared with GLP-1, that linear fatty acids up to 16 carbons could be attached almost anywhere in the C-terminal part without considerable loss of potency, that the longer the fatty acid the more potency was lost, and that attaching two fatty acid substituents led to a considerable loss of potency [6]. It also found that modifying the N-terminus for metabolic stability at the same time interfered with fatty acid derivatisation and cost potency [6].

The next generation optimised the linker rather than the lipid alone. The semaglutide discovery paper states the aim as designing “a once weekly GLP-1 analogue by increasing albumin affinity and secure full stability against metabolic degradation”, and identifies “the fatty acid moiety and the linking chemistry to GLP-1” as the key features [4]. The development account is specific: linker studies using palmitate showed that introducing OEG linkers gave both high receptor potency and albumin affinity, and a systematic test of C12 to C20 fatty acids found that “a C18 di-acid together with a gammaGlu-2xOEG linker resulted in the highest albumin affinity combined with GLP-1R potency” [2].

Worked example: semaglutide

Feature	Detail	Source
Backbone changes	Aib at position 8, Arg at position 34, versus human GLP-1	[4]
Acylation site	Lysine 26	[4] [5]
Side chain	C18 fatty di-acid through a hydrophilic spacer	[5]
Linker identified in design	gamma-glutamate plus two OEG units	[2]
Sequence homology to human GLP-1	94%	[5]
Albumin binding	More than 99% bound to plasma albumin	[5]
Human elimination half-life	About 1 week	[5]
Receptor affinity cost	GLP-1 receptor affinity 0.38 plus or minus 0.06 nM, a threefold decrease versus liraglutide	[4]
Elimination route	Proteolytic cleavage of the backbone plus sequential beta-oxidation of the fatty acid side chain	[5]

The label states the mechanism plainly: “the main protraction mechanism of semaglutide is albumin binding, facilitated by modification of position 26 lysine with a hydrophilic spacer and a C18 fatty di-acid”, and “a minor modification was made in position 34 to ensure the attachment of only one fatty di-acid” [5]. That position 34 change had a manufacturing pay-off as well as a chemical one: replacing the only other lysine with arginine enabled “a semi-recombinant process, where the peptide backbone was produced recombinantly and the fatty acid attached afterwards by a simple chemical reaction” [2], and the label records that the peptide backbone “is produced by yeast fermentation” [5]. Molecular formula C187H291N45O59 and average molecular weight 4113.58 g/mol [5], matching the figures in our semaglutide overview.

Note the trade recorded in the table: the design deliberately accepted a threefold loss of receptor affinity against liraglutide in exchange for higher albumin affinity [4]. Lipidation is not free potency.

Worked example: tirzepatide

Tirzepatide shows the same toolkit applied to a GIP-based sequence. Its label states that it “is based on the GIP sequence and contains aminoisobutyric acid (Aib) in positions 2 and 13, a C-terminal amide, and Lys residue at position 20 that is attached to 1,20-eicosanedioic acid via a linker”, with molecular weight 4813.53 Da and empirical formula C225H348N48O68 [7]. It “contains a C20 fatty diacid that enables albumin binding and prolongs the half-life”, is 99% bound to plasma albumin, and has an elimination half-life of approximately 5 days [7]. Metabolism runs by proteolytic cleavage of the backbone, beta-oxidation of the C20 di-acid and amide hydrolysis [7]. The discovery paper describes it as “a fatty acid modified peptide with dual GIP and GLP-1 receptor agonist activity designed for once-weekly subcutaneous administration” [8].

Four modifications are stacked in one molecule: two Aib substitutions against proteolysis, a C-terminal amide, a di-acid for albumin binding, and a linker between the two. This is the current state of the art, and it is the reason the molecule is made by a hybrid route rather than linear synthesis.

Making lipidated peptides

Lipidation is usually done while the peptide is still on the resin. The lysine intended for acylation is protected with an orthogonal group so it can be unmasked selectively: acid-labile Mmt or Mtt, removed with 1% trifluoroacetic acid in dichloromethane; base-labile ivDde or Dde, removed with 2% hydrazine in dimethylformamide; or Alloc, removed with catalytic palladium under inert conditions [3]. Once the side chain is free, “a selective removal of the side-chain protecting group enables specific amide-bond formation by the reaction of an activated carboxylic group of the fatty acid with the N-epsilon-group of the lysine” [3]. In one published synthesis of lipidated human pancreatic polypeptide, the lysines were protected with Dde during automated synthesis and the palmitic acid introduced afterwards [3].

The practical point for anyone costing a synthesis: these orthogonal building blocks and the extra deprotection, acylation and wash steps all sit on top of the standard cycle, and fatty acid and PEG introduction is “generally performed manually” [3]. Fmoc chemistry is the usual choice for modified peptides because many modifications are not stable to the anhydrous hydrogen fluoride that Boc chemistry needs for final cleavage [21].

PEGylation

PEGylation attaches polyethylene glycol chains of defined average molecular weight. The mechanism is physical rather than biological: “by increasing the molecular mass of proteins and peptides and shielding them from proteolytic enzymes, pegylation improves pharmacokinetics” [13]. It also increases hydrodynamic radius and so decreases the rate of glomerular filtration [1]. PEG is amphiphilic, non-toxic, weakly immunogenic and highly soluble, and the PEGylated peptide travels inside a large water shell, because every ethylene oxide unit binds about three water molecules [3].

The chemistry is straightforward in principle. PEG polymers are sold with a methoxy cap at one end and an activated group, usually an N-hydroxysuccinimidyl ester, at the other, and are available in a range of sizes and chain architectures [3]. Small PEGs can be coupled on resin after selective side-chain deprotection; “the introduction of larger PEG units has to be carried out in solution” after cleavage, when only the intended amino groups are available [3]. Conventional PEGylation targets lysine or cysteine residues, which is why site-specific methods, including genetic code expansion to install a uniquely reactive non-canonical amino acid, have been developed to avoid heterogeneous multi-PEGylation [16].

Three limitations are documented:

• Polydispersity and persistence. PEG is polydisperse and lacks biodegradability, which is why alternatives such as polysialylation, HESylation and PASylation have been developed [3].
• Anti-PEG immunity. Antibodies that specifically bind PEG can cause accelerated blood clearance of PEGylated therapeutics. In animals the response is typically robust but short-lived and predominantly IgM; in humans the presence of anti-PEG antibodies “has been correlated with reduced efficacy of PEGylated therapeutics in clinical trials”, and such antibodies occur in some healthy untreated individuals [14].
• Renal findings. PEG has been associated with vacuolisation of renal cortical tubular epithelium cells, which the fusion-protein review notes brings its safety “at least somewhat into question” [1].

There is also a regulatory consequence for manufacturing. EMA treats peptide fragments carrying modifications such as PEG chains, or that have been cyclised, as intermediates rather than as starting materials, which means they fall inside the GMP boundary [22].

Backbone modifications: Aib, D-amino acids, N-methylation

The cheapest way to defeat a protease is to make its substrate unrecognisable at the scissile bond. For incretin peptides that site is position 2, where dipeptidyl peptidase-4 cuts.

Aib

Aminoisobutyric acid is alanine with a second methyl group on the alpha carbon, which blocks the enzyme sterically without adding bulk elsewhere. The GLP-1 work established how narrow the window is: an early study showed DPP-IV lability could be reduced by alternative residues at position 8, but glycine at that position also significantly reduced receptor affinity, and “the only Ala substitution that resulted in both DPP-IV stability and high GLP-1R affinity was the introduction of aminoisobutyric acid (Aib) at position 8” [2]. Semaglutide’s label records exactly this: it “is modified in position 8 to provide stabilization against degradation by the enzyme dipeptidyl-peptidase 4 (DPP-4)” [5]. Tirzepatide carries Aib twice, at positions 2 and 13 [7].

Worked example: teduglutide, a single-residue solution

Teduglutide is the minimal case and a useful control. Its label describes it as “a 33 amino acid glucagon-like peptide-2 (GLP-2) analog” manufactured “using a strain of Escherichia coli modified by recombinant DNA technology”, with a molecular weight of 3752 Daltons [10]. The single change is a glycine for alanine swap at position 2. The 1997 study behind it showed that GLP-2(1-33) was degraded to GLP-2(3-33) by human placental DPP-IV or rat serum but not by serum from DPP-IV-deficient rats, and that the synthetic analogue r[Gly2]GLP-2 “was resistant to cleavage by both DPP-IV and rat serum in vitro” and produced a statistically significant increase in small bowel mass in wild-type rats [9].

What teduglutide does not have is any half-life extension beyond protease resistance. Its label reports a mean terminal half-life of approximately 2 hours in healthy subjects and 1.3 hours in subjects with short bowel syndrome [10]. That contrast is the clearest illustration available of the difference between blocking proteolysis and slowing renal clearance: one residue buys protease resistance, and nothing more. The molecular formula in our teduglutide overview is C164H252N44O55S with a molecular weight of 3752.13 g/mol, consistent with the label.

D-amino acids

Proteases are stereospecific, so a D-residue at a cleavage site resists them. Backbone modification strategies include “the substitution of L-amino acids by D-amino acids, insertion of methyl-amino acids, and the incorporation of beta-amino acids and peptoids” [16]. The catch is activity: “the insertion of D-amino acid can usually help to extend the plasma half-life of the peptide, but peptides with D-amino acid modification rarely exhibit effective biological activity” [16]. D-proline has a narrower, more reliable use, forcing a turn to stabilise beta-sheet structure [16].

D-residues also appear as unwanted impurities rather than design elements. Epimerisation during activation and coupling produces diastereomers with nearly identical physical properties, which is why racemisation control is a central concern in synthesis; cysteine and histidine are the most at-risk residues under base-mediated activation and microwave heating [21].

Cyclisation, stapling and terminal capping

Cyclisation

Cyclisation can be head-to-tail, backbone-to-side-chain or side-chain-to-side-chain [16]. It “can increase proteolytic stability and cell-permeability, and allows mimicking and stabilization of the peptide secondary structure”; a single linear sequence cannot hold a loop or turn, but cyclisation pre-organises the intramolecular interactions that make one [16]. It is also used to stabilise helices and beta-sheets [16].

This is a mature class, not an experimental one. A 2017 review states that “over 40 cyclic peptide drugs are currently in clinical use and around one new cyclic peptide drug enters the market every year on average”, with the great majority derived from natural products such as antimicrobials and human hormones, and newer entries coming from rational design and in vitro evolution [17]. Disulfide-rich cyclic peptides are a sub-class with their own chemistry: linaclotide, in this library, is described in its label as “a 14-amino acid peptide” whose chemical name ends “cyclic (1-6), (2-10), (5-13)-tris (disulfide)”, and that constrained, protease-resistant architecture is consistent with the label statement that it “is minimally absorbed with negligible systemic availability following oral administration” [19].

Modern resins carry linkers designed for on-resin cyclisation, which is what makes these molecules practical to make at scale [16].

Stapling

Stapling locks an alpha-helix with a covalent cross-link, “using non-natural electrophilic amino acids to replace residues at the i and i+4 or i and i+7 position” [16]. The founding report, in 2004, used hydrocarbon stapling on BH3 domain peptides from the BCL-2 family. The stapled peptides, named stabilised alpha-helices of BCL-2 domains, “proved to be helical, protease-resistant, and cell-permeable molecules that bound with increased affinity to multidomain BCL-2 member pockets”; one of them activated the apoptotic pathway to kill leukaemia cells and inhibited the growth of human leukaemia xenografts in vivo [18]. The evidence level there is cell culture plus a mouse xenograft model.

N- and C-terminal capping

The simplest intervention of all is to block the ends. Peptide stability “can also be optimized by blocking their respective termini through N-terminal acylation and C-terminal amidation” [3]. C-terminal amidation also has a structural rationale: an acid or an amide are the two naturally occurring C-terminal functionalities, and the resin linker is what decides which one the synthesis delivers [3]. Tirzepatide carries a C-terminal amide [7].

Albumin binding by covalent conjugation

Reversible albumin binding through a fatty acid is one approach. The other is to bond to albumin covalently, in the bloodstream, after administration. CJC-1295 is the worked example in this library.

The method places a reactive maleimide on the peptide so that it forms a covalent bond with the free thiol on cysteine 34 of serum albumin. Three maleimido derivatives of human GRF(1-29) were synthesised, conjugated to human serum albumin ex vivo, and all three showed enhanced in vitro stability against dipeptidyl peptidase-IV and were bioactive in a growth hormone secretion assay in cultured rat anterior pituitary cells [11]. The best compound, CJC-1295, is described as “a tetrasubstituted form of hGRF(1-29) with an added N-epsilon-3-maleimidopropionamide derivative of lysine at the C terminus”; it gave a fourfold increase in growth hormone area under the curve over two hours in rats compared with hGRF(1-29), was present in plasma beyond 72 hours, and appeared on Western blot as an immunoreactive band at the size of serum albumin from 15 minutes after administration [11].

In humans, two randomised, placebo-controlled, double-blind ascending-dose trials in healthy adults aged 21 to 61 estimated the half-life of CJC-1295 at 5.8 to 8.1 days; after multiple doses, mean IGF-I levels remained above baseline for up to 28 days, and no serious adverse reactions were reported [12]. CJC-1295 is not approved by any major regulator, and clinical development was discontinued; our CJC-1295 overview covers the distinction between this albumin-binding molecule and the version sold without the albumin-binding lysine.

The same albumin-as-carrier idea has been pursued with peptide and protein ligands rather than chemistry. A cyclic peptide with the core sequence DICLPRWGCLW binds albumin with 1:1 stoichiometry at a unique epitope and, fused to Fab fragments, increased half-life 37-fold in mice and 26-fold in rabbits, though with species-specific differences in binding affinity [2].

Fusion proteins, briefly

Fusion moves the problem to a protein that the body already recycles. Human IgG1, IgG2 and IgG4 and human serum albumin all bind the neonatal Fc receptor in a pH-dependent manner after pinocytosis, and that recycling “confers a nominal 14- to 21-day half-life on human IgG1, IgG2, and IgG4, and a ~19-day half-life on HSA” [1]. Immunoglobulins that do not bind the receptor, such as IgA, IgM, IgD and IgE, have no extended half-life, and IgG3, whose receptor-binding domain carries an altered residue, falls to about 5 to 7.5 days [1].

Applied to incretins, the effect is large. An approved GLP-1 Fc fusion “contains a DPP-4-resistant version of GLP-1 (V8-GLP-1) fused to the Fc of a human IgG4 (F234A/L235A) via a linker” [1], and an approved GLP-1 fusion to human serum albumin improved the half-life of pharmacologically active GLP-1 “from 1-2 min for native GLP-1 to 4-7 days, which allows for once weekly dosing” [1]. Across the field, “most of the Fc fusion proteins generated to date have half-life values in humans of about 4-5 days” [1]. Immunogenicity has been modest in the examples reviewed: for one albumin fusion product, about 5% of patients generated non-neutralising low-level anti-drug antibodies not correlated with clearance or loss of activity [1].

Fusion is a biologics strategy rather than a peptide-chemistry one, with the manufacturing and regulatory consequences that implies, so it sits at the edge of this topic rather than inside it.

Choosing a modification

Goal	First-choice modification	Evidence to check	Cost in synthesis
Block a known protease site	Aib, D-residue or N-methylation at the cleavage site	In vitro stability against the specific enzyme, plus retained receptor affinity [2] [9]	One non-standard building block
Extend half-life to days or a week	Fatty di-acid with an optimised linker, for albumin binding	Albumin binding percentage and the receptor affinity you gave up [4] [5]	Orthogonal lysine protection, selective deprotection, manual acylation [3]
Reduce renal clearance of a small peptide	PEGylation	Anti-PEG antibody risk and PEG size distribution [14] [3]	Site-specific chemistry; large PEGs attached in solution [3]
Lock a bioactive conformation	Cyclisation or stapling	Helicity, protease resistance, cell permeability [16] [18]	On-resin cyclisation linker, non-natural cross-linking residues
Stabilise the ends cheaply	N-terminal acetylation, C-terminal amidation	Comparative stability against the unmodified peptide [3]	Resin choice and one capping step
Reach weeks of exposure	Fc or albumin fusion	Receptor-mediated recycling, immunogenicity [1]	Expression system rather than synthesis

A closing caution from the field. More than 80 peptide drugs have reached the market since insulin, across diabetes, cancer, osteoporosis, multiple sclerosis, HIV infection and chronic pain, built on exactly these techniques [20]. But every modification is a trade. Semaglutide’s threefold loss of receptor affinity against liraglutide [4], the potency cost of long or doubled fatty acids [6], and the tendency of D-amino acid substitution to abolish activity [16] are all documented. Modification data are only interpretable when potency and pharmacokinetics are reported together.

How Winhycare Pharma handles modified peptides

Winhycare Pharma supplies research peptides as research-use-only material, at purity of at least 99% as stated in our product brochure, with QC documentation, including acylated and cyclised sequences such as semaglutide, tirzepatide, CJC-1295 and teduglutide. Custom peptide synthesis covers lipidation, PEGylation, cyclisation and non-natural residues for research sequences. Semaglutide, tirzepatide, retatrutide, cagrilintide and BPC-157 are also supplied as API to licensed pharmaceutical manufacturers through our peptide API hub; US-DMF and CTD Module 3 support is available for semaglutide only. To discuss a modified sequence, use the custom synthesis enquiry form.

For laboratory research use only. Not for human or veterinary use.