Peptide Half-Life Extension: How Minutes Become Weeks

Peptide science

Peptide half-life extension is the reason a hormone that survives about two minutes in blood can be sold as a once-weekly injection. Native GLP-1 has a plasma half-life of roughly 1.5 to 2 minutes. Semaglutide shares 94% of its sequence and has an elimination half-life of about one week. That gap is not a single clever trick — it is four separate engineering strategies, each aimed at a different clearance mechanism, and each one carries a cost that shows up somewhere else in the molecule.

Why peptide half-life extension is needed at all

Three things remove a circulating peptide, and they work at very different speeds.

Proteases. The FDA label for liraglutide states the problem plainly: native GLP-1(7-37) has a half-life of 1.5 to 2 minutes “due to degradation by the ubiquitous endogenous enzymes, dipeptidyl peptidase IV (DPP-IV) and neutral endopeptidases (NEP).” Exopeptidases chew inward from the ends; endopeptidases cut internally. Both are everywhere, including in blood.

Renal filtration. Small proteins are simply filtered out at the glomerulus. The size threshold is genuinely contested in the literature — a widely cited 2015 review puts it near 70 kDa, while a 2026 review states an empirically established range of roughly 36 to 44 kDa and elsewhere uses “below 50 kDa.” What both agree on is that mass is not the only variable: negatively charged polypeptides are repelled by the negatively charged glomerular basement membrane and filtered less readily, while cationic polypeptides are removed even faster.

The combination. Exenatide illustrates it: its label describes elimination “by glomerular filtration with subsequent proteolytic degradation,” a mean apparent clearance of 9.1 L/hour and a terminal half-life of 2.4 hours. Filtered first, then destroyed. Every strategy below attacks one of these routes — or sidesteps all three by changing how fast the drug enters the blood in the first place.

Strategy 1: outrun the renal filter

The bluntest approach is to make the molecule too large — or too negatively charged — to pass the glomerulus. Polyethylene glycol was the original tool here, and its successors are unstructured polypeptide tails that do the same job out of ordinary amino acids.

XTEN is a designed random-coil polypeptide built from a deliberately narrow alphabet. Hydrophobic residues were omitted to prevent aggregation, asparagine and glutamine to protect long-term storage stability, the positively charged residues to minimise nonspecific membrane binding, and cysteine to avoid disulfide heterogeneity. What is left is roughly 8% alanine, 12% glutamate, 18% glycine, 17% proline, 28% serine and 17% threonine. The strong negative charge produces electrostatic repulsion at the glomerular basement membrane on top of the expanded hydrodynamic volume. Fused to exendin, XTEN took the half-life in monkeys from about 30 minutes to 60 hours; a growth-hormone-XTEN fusion reached a 131-hour half-life in humans.

PAS polymers use only proline, alanine and serine — threonine was excluded for its tendency to promote beta-sheet formation, and glycine was left out too. PASylated leptin went from 26 minutes to 19.6 hours. PASylated erythropoietin gained a 15.6-fold extension in plasma half-life despite reduced in vitro potency, which is the recurring pattern in this whole field.

ELP tails are repeats of Val-Pro-Gly-X-Gly that reversibly phase-separate above a transition temperature. An ELP of 636 residues fused to GLP-1 (PB1023, about 70 kDa) reached a half-life near 36 hours and entered phase 2.

Strategy 2: hitch a ride on FcRn

The neonatal Fc receptor is a salvage pathway. Immunoglobulin G and human serum albumin are constantly taken into cells by pinocytosis; as the vesicle acidifies to about pH 6.0, both bind FcRn and are shuttled back to the cell surface, where the neutral pH of blood (about 7.4) releases them again. The affinity of FcRn for albumin drops roughly 200-fold across that pH transition, which is exactly what makes the switch work.

The payoff is large. That recycling loop confers a nominal 14- to 21-day half-life on human IgG1, IgG2 and IgG4, and about 19 days on albumin. IgG and albumin bind FcRn at different epitopes, so they do not compete — and roughly 700 albumin molecules are recycled for every IgG. There are three ways to exploit it.

Fuse to an Fc. Dulaglutide is two identical disulfide-linked chains, each a DPP-4-resistant GLP-1 analogue joined to a modified human IgG4 Fc by a small peptide linker, about 63 kDa total. Its half-life is approximately 5 days, with a time to peak concentration of 24 to 72 hours. Two design details are worth noting: IgG4 was chosen rather than IgG1 specifically to minimise effector function, since killing cells that express the GLP-1 receptor would be a catastrophic side effect; and the length and structure of the linker turned out to be critical, because without it the GLP-1 agonist activity was minimal. The first Fc fusion protein was a CD4-Fc built in 1989; etanercept, approved in 1998, was the first to reach the market.

Fuse to albumin itself. Albiglutide was a GLP-1(7-37)-albumin fusion of about 73 kDa with a half-life near 5 days. It was authorised in the EU on 21 March 2014 and the marketing authorisation was withdrawn on 29 October 2018 — at the manufacturer’s own request, for commercial reasons, not safety. That distinction matters and is routinely garbled online.

Borrow albumin without fusing to it. Attach a fatty acid and the peptide binds circulating albumin reversibly. Liraglutide is 97% homologous to native GLP-1: lysine 34 was swapped for arginine so that a C16 palmitic acid with a glutamic acid spacer could be attached to the remaining lysine at position 26. Its label credits three mechanisms at once — self-association that delays absorption, plasma protein binding, and stability against both DPP-4 and neprilysin — for a 13-hour half-life. Semaglutide pushes the same idea further: a hydrophilic spacer and a C18 fatty di-acid at position 26, a change at position 8 for DPP-4 stability, and a minor change at position 34 to ensure only one di-acid attaches. Its label names albumin binding as the main protraction mechanism, “which results in decreased renal clearance and protection from metabolic degradation,” giving an elimination half-life of about one week and detectable drug for roughly five weeks after the last dose.

Position matters as much as chemistry. Moving the acyl group toward the N-terminus (position 8) took the EC50 from about 61 pM to 1,260 pM. Attaching two fatty acids pushed it to 7,000 and 16,700 pM. The gamma-glutamate linker in liraglutide exists partly to compensate for the acidic group consumed by the amide linkage — the linker is not decoration, it is damage control.

One correction worth making: insulin detemir and degludec are usually described as albumin binders, but that is not their dominant mechanism. Detemir was the first clinically approved protein modified by a fatty acid, and its protraction comes primarily from subcutaneous deposition after injection, with only a small increase in intravenous half-life attributable to albumin binding. Same chemistry, different physics.

Strategy 3: resist the proteases

DPP-4 cleaves after the second residue. Native GLP-1 has alanine there and is destroyed in minutes. Exendin-4, the lizard peptide behind exenatide, naturally has glycine in the same position and survives roughly 30 minutes after intravenous dosing. That single-residue difference is the cleanest natural experiment in the field.

Copying it is not free. Introducing glycine at position 8 of GLP-1 reduces DPP-4 lability but also significantly reduces receptor affinity — in one matched pair, potency fell about five-fold from that substitution alone. Albiglutide carried the same glycine 8 swap and its receptor affinity was about 20 nM against 0.02 nM for exenatide. The only alanine substitution that delivered both DPP-4 stability and high receptor affinity was aminoisobutyric acid (Aib), which is what semaglutide uses at position 8.

Liraglutide makes the opposite point. Its N-terminal His-Ala is unchanged, and it is still partially protected from DPP-4 — probably through reversible albumin binding or direct steric hindrance. Protease resistance does not have to come from a point mutation.

Beyond substitutions, the classic backbone tools are terminal caps, D-amino acids and ring closure. Leuprolide is the textbook example: a nonapeptide with a pyroglutamate cap at the N-terminus, D-leucine at position 6 and an N-ethylamide replacing the C-terminal glycinamide. If you want the chemistry behind building these sequences, see our guide to solid-phase peptide synthesis, and for the enzymes doing the cutting, what DPP-4 actually does.

Strategy 4: slow the input instead of the clearance

This is the one most often mislabelled. Exenatide extended-release incorporates the peptide into 50:50 poly(D,L-lactide-co-glycolide) microspheres — 37.2 mg of polymer per dose — from which exenatide is released over approximately 10 weeks. The molecule still has a 2.4-hour terminal half-life. Nothing about it was made more stable.

Lupron Depot does the same for leuprolide, with polylactic or lactide-glycolide microspheres formulated for one, three, four or six months. Insulin glargine achieves it chemically instead: a di-arginine extension on the C-terminus of the B chain shifts the isoelectric point so the drug precipitates in subcutaneous tissue and redissolves slowly.

The pharmacology term for what you observe is flip-flop kinetics — the apparent half-life is absorption-rate-limited, not elimination-rate-limited. It works, but it has its own signature cost: in one exenatide ER study, 24 of 31 subjects (77%) developed at least one injection-site nodule, which the label attributes to the known properties of the microspheres.

The scatter above is the most useful corrective in the whole topic. Pentameric IgM weighs 970 kDa and lasts about 6 days. Thyroglobulin weighs 660 kDa and lasts under 3 days. IgG1 weighs 146 kDa and lasts about 20 days. Three protein classes — IgG, albumin and transferrin — persist far longer than their size alone would predict, and the reason is receptor biology, not mass.

What PEGylation taught the field

PEG works — it is highly flexible, uncharged, and generates a larger hydrodynamic radius than an equivalently sized protein. It is also, in the words of the same review, non-biodegradable. In pegvisomant, the PEG is roughly half the mass of the finished drug: a 21,998 Da protein carrying predominantly four to six PEG chains of about 5,000 Da each, for a total near 42,000 to 52,000 Da and a half-life of about 6 days.

The regulatory record is blunt about the consequences. A 2012 European Medicines Agency safety document notes that repeated parenteral dosing of PEGylated proteins has been associated with cellular vacuolation in macrophages, histiocytes and renal tubular cells, and that vacuolation of choroid plexus ependymal cells has been seen with PEG moieties of at least 40 kDa in six-week-plus monkey studies. Immunohistochemistry confirmed the vacuoles contain PEG, and in one 52-week study the ependymal vacuolation persisted through a 26-week recovery period.

The other lesson is immunological. PEG was long treated as inert and non-antigenic; it is not. Anti-PEG antibodies developed in 42% of patients in the pegloticase programme, anti-drug antibodies overall in 92%, and infusion reactions ran 53% in high-titre patients against 6% in the rest. The label carries a dedicated drug-interaction subsection warning about other PEGylated products. Background prevalence in people who have never been treated has climbed across the literature from 0.2% in 1984 blood donors to about 27% and then 42% in later surveys — plausibly because PEG is now ubiquitous in consumer products. Our note on anti-drug antibodies covers why this matters for any repeat-dosed biologic.

The modern approach stacks all of it

The clearest current example is a factor VIII product approved in the US in 2023 whose established name literally contains its engineering: Fc-VWF-XTEN fusion protein. Its label describes three mechanisms in three consecutive paragraphs. The D-prime-D3 domain of von Willebrand factor is appended so the molecule stops depending on endogenous VWF, which was the ceiling on factor VIII half-life. The IgG1 Fc region engages FcRn and delays lysosomal degradation. And two XTEN polypeptides alter the hydrodynamic radius, reducing rates of clearance and degradation. The result is a 3- to 4-fold longer half-life than standard and extended half-life factor VIII products.

That is what a mature half-life extension strategy looks like in 2026: not one trick, but three clearance routes closed at once. For the underlying kinetics — what a half-life actually measures and why it is not a shelf-life — see our primer on peptide half-life, and for the notation you will meet on spec sheets, what DAC means on a peptide.

Frequently asked questions

Does a longer half-life make a peptide more potent?

No — and the two are often traded against each other. Every measurement in the acylation series above shows receptor potency falling as albumin affinity rises, because only unbound drug can reach the receptor. Persistence and potency are separate properties, and a long-acting analogue is frequently a weaker agonist that simply stays around longer.

Why do fusion proteins not inherit their partner’s half-life?

Albumin lasts about 19 days but albiglutide lasted about 5. IgG lasts 14 to 21 days but most approved Fc fusions land at roughly 4 to 5 days. The peptide “head” of the fusion is still an accessible protease target even when the tail is being faithfully recycled — the weakest link sets the rate. There are exceptions at the long end, so this is a tendency rather than a hard ceiling.

Is a depot injection the same thing as half-life extension?

No. Molecular engineering changes how fast the body removes the drug; a depot changes how fast the drug arrives. A microsphere formulation can give once-weekly or six-monthly dosing while the peptide inside still has a two-hour half-life. Both produce long dosing intervals, but they behave differently on missed doses, in renal impairment and at the injection site.

Has PEGylation been abandoned?

No. Several PEGylated products remain approved and in use, and PEG is still described in the literature as a mature technology. What changed is that it is no longer treated as biologically inert: anti-PEG antibodies, tissue accumulation and heterogeneous conjugation are now designed around rather than assumed away, which is much of why unstructured polypeptide tails were developed.

Informational only — not medical advice · 21+