How do peptides degrade? Degradation pathways, warning signs and stability testing

A peptide stops being the molecule on its label in two ways. Its covalent structure can change: an amide hydrolyses, a sulfur picks up oxygen, a bond breaks or a residue flips its stereochemistry. Or its molecules can stick together into aggregates and fibrils while each chain stays chemically intact. Both kinds of change can happen in solution and, more slowly, in a lyophilized powder.

This guide goes pathway by pathway: which residues are at risk, which conditions drive each reaction, how each degradant shows up in HPLC and mass spectrometry, and how forced degradation studies under ICH Q1A(R2) and Q1B are used to map them. For storage temperatures and retest periods, see our guide to lyophilised peptide stability and storage.

How do peptides degrade? Chemical versus physical instability

The standard way to organise peptide and protein degradation comes from Manning, Patel and Borchardt’s 1989 review. Chemical instability “can include proteolysis, deamidation, oxidation, racemization, and beta-elimination”, while physical instability “refers to processes such as aggregation, precipitation, denaturation, and adsorption to surfaces” [1]. Their 2010 update added a section on the interrelationship between chemical and physical instability [2].

Regulators now expect the same thinking for every synthetic peptide. EMA’s guideline, in effect from 1 June 2026, lists the pathways that “can contribute to the formation of degradation products of synthetic peptides”: oxidation, hydrolysis, isomerisation, deamidation, cyclic imide (aspartimide) formation, diketopiperazine and pyroglutamic acid formation, β-elimination, condensation and dehydropeptide formation, disulfide cleavage or exchange, and acetylation [3]. It asks for these to be discussed “taking into account the amino acids composition and sequence” [3]. In other words, you can predict much of a peptide’s weak points from its sequence before running a single stability sample.

Peptide degradation pathways at a glance

Pathway	Residues or motifs at risk	Conditions that drive it	How it is detected	How it is slowed	Ref
Deamidation	Asn, fastest when followed by Gly; also Gln	Neutral to basic pH, temperature, buffer catalysis	LC-MS (about +1 Da); isoAsp and Asp products resolved by HPLC	pH control, cold storage, low moisture	[4], [5], [6], [7]
Isomerisation and Asp-X cleavage	Asp, notably Asp-Gly	Isomerisation above pH 4; cleavage in strong acid	HPLC (isoAsp has the same mass as Asp); LC-MS for cleavage fragments	pH control, cold storage	[8], [4]
Oxidation	Met, Cys, His, Trp, Tyr	Peroxides, transition metals, light, oxygen	LC-MS (+16 Da per oxygen); peptide mapping to locate the site	Chelators for metal-catalysed oxidation, scavengers, light protection, inert atmosphere	[9], [10], [3]
Disulfide scrambling and β-elimination	Free Cys; disulfide bonds	Neutral to basic pH	Non-reduced versus reduced LC-MS, peptide mapping; dimers by SEC	Acidic to neutral pH; thiol control	[11]
Backbone hydrolysis	Asp-Gly, Asp-X, Asn-Ser, Asn-Pro, Asn-Leu bonds	Strong acid; neutral to alkaline pH for Asn-linked cleavage	LC-MS: fragments whose masses sum to the parent plus water	pH control, low moisture	[8], [5], [4]
Diketopiperazine formation	N-terminal dipeptide with Pro in position 2	Free (unprotonated) N-terminal amine, base catalysis	LC-MS: loss of the N-terminal dipeptide	Lower pH (protonated N-terminus reacts more slowly)	[12]
Pyroglutamate formation	N-terminal Gln or Glu	pH 4 and 8 faster than pH 6.2 (Glu)	LC-MS (loss of water from Glu, ammonia from Gln)	pH control near the slowest point	[13], [14]
Racemisation	Any residue; succinimide intermediates racemise fast	Succinimide formation in storage; Fmoc deprotection during synthesis	Chiral GC after hydrolysis in deuterated acid; specific HPLC methods	Synthesis process control; the same pH controls that limit succinimide formation	[4], [15], [3]
Aggregation and fibrillation	Hydrophobic and amyloid-prone sequences	Concentration, pH, ionic strength, agitation, interfaces, lyophilisation, impurities	SEC for high-molecular-weight species; thioflavin T; turbidity	Formulation, low agitation, controlled handling	[16], [17], [3]

The mass shifts in the table are the elemental change of each reaction (an NH2 group replaced by OH, one oxygen atom added, water or ammonia lost) rounded to the nearest dalton.

Deamidation: why Asn-Gly is the hot spot

Deamidation turns an asparagine (or, more slowly, a glutamine) side-chain amide into a carboxylic acid. It adds a negative charge and, at neutral pH, usually goes through a cyclic intermediate rather than direct hydrolysis.

The classic measurement is Geiger and Clarke’s 1987 study of the hexapeptide Val-Tyr-Pro-Asn-Gly-Ala. At 37 °C and pH 7.4, it deamidated “with a half-life of only 1.4 days” to an aspartyl succinimide, which then hydrolysed (half-time 2.3 hours) and racemised (half-time 19.5 hours) [4]. The end product was a mixture of L- and D-aspartyl and isoaspartyl peptides. Replacing the glycine with a bulky leucine or proline slowed degradation 33–50-fold [4]. This was a model-peptide study in buffer, but it is why Asn-Gly is flagged in any sequence review.

Tyler-Cross and Schirch extended the picture with pentapeptides. At neutral and alkaline pH, the residue on the amino side of Asn had little effect, while the residue on the carboxyl side mattered a lot: larger and more branched side chains cut the rate by “as much as 70-fold compared to glycine” [5]. Products formed in about a 3:1 isoAsp:Asp ratio at neutral and alkaline pH, whereas in acid only the Asp peptide formed [5]. Patel and Borchardt saw the same switch in mechanism: from pH 5 to 12 deamidation ran “exclusively via a cyclic imide intermediate”, with buffer catalysis between pH 7 and 11, while at acidic pH the amide hydrolysed directly [6].

Deamidation is not a niche problem. Robinson and Robinson calculated deamidation rates for 1,371 asparaginyl residues across 126 human proteins and concluded that it is biologically relevant “in a remarkably large percentage of human proteins” [7].

Detection. Deamidation adds roughly 1 Da, so it is visible in LC-MS only with adequate resolution, and the isoAsp and Asp products have identical masses. They are told apart by retention time, which is why a related-substances HPLC method has to be shown to separate them.

Aspartate isomerisation and Asp-X cleavage

Aspartic acid residues form the same succinimide as deamidating asparagine, only more slowly: in the Geiger and Clarke model, swapping Asn for Asp gave “a 34-fold decrease in the rate of succinimide formation” [4]. Once formed, the ring opens to either normal aspartate or isoaspartate, a β-linked residue that inserts an extra backbone carbon.

pH decides what happens to an Asp residue. Oliyai and Borchardt studied the hexapeptide Val-Tyr-Pro-Asp-Gly-Ala [8]:

• At pH 0.3–3.0, the main reaction was acid-catalysed hydrolysis of the Asp-Gly bond, cutting the peptide into a tetrapeptide and a dipeptide.
• At pH 4.0 and 5.0, isomerisation to isoAsp and Asp-X cleavage ran side by side.
• Above pH 6.0, isomerisation to the iso-Asp hexapeptide was “the only observed product”.

The ionised side chain was more reactive than the unionised form in both pathways [8]. In synthesis, the same chemistry appears as aspartimide formation, which EMA lists separately as “cyclic imide formation (aspartamide)” [3].

Oxidation of methionine, cysteine and tryptophan

“Methionine, cysteine, histidine, tryptophan, and tyrosine are the amino acid residues most susceptible to oxidation,” wrote Li, Schöneich and Borchardt in their review of oxidation mechanisms [9]. Oxidation can be induced by contaminating oxidants, catalysed by transition metal ions or induced by light, and it is influenced by pH, temperature and buffer composition [9]. Two mechanisms behave differently. Oxidation by oxidants already in the system is non-site-specific and can be inhibited with antioxidants or radical scavengers; metal-catalysed oxidation is site-specific, and there “the addition of antioxidants may accelerate the oxidation reaction”, so chelating agents are the better tool [9].

Which residue oxidises depends on the oxidant. Ji and colleagues used the 34-residue parathyroid hormone fragment PTH(1-34) as a model and found that hydrogen peroxide and t-butyl hydroperoxide “primarily oxidized the two Met residues”, while a radical generator (AAPH) and hydrogen peroxide with Fe(II) oxidised both Met and Trp; histidine oxidised when copper replaced iron [10]. Free methionine protected the Met residues against peroxide, mannitol and EDTA were effective against peroxide plus Fe(II), and free tryptophan protected Trp against AAPH [10]. The authors chose those stresses because they mimic real sources: peroxides from degraded polysorbate, residues of aseptic agents and metals from stainless steel [10].

EMA singles out “oxidation of Cys and Met residues” as a pathway to discuss [3]. Methionine sulfoxide adds one oxygen (+16 Da), so it is easy to see in LC-MS; locating which methionine was hit needs MS/MS or peptide mapping, the approach Ji and colleagues used [10].

Disulfide scrambling and β-elimination

Peptides with cysteine pairs, such as linaclotide, a 14-residue guanylate cyclase-C agonist cyclised by three disulfide bridges, carry an extra risk: the disulfides can break and re-form in the wrong pairing. Trivedi, Laurence and Siahaan’s review notes that “most of the disulfide bond degradation” in peptides and proteins is observed “under neutral and basic conditions” [11].

They describe three routes [11]:

1. Hydroxide attack on sulfur, opening the disulfide to a sulfenic acid and a thiolate. The thiolate can attack another disulfide, giving dimers, trimers and higher oligomers.
2. β-elimination, in which the α-proton of cysteine is abstracted to give dehydroalanine and a persulfide. Re-addition can scramble stereochemistry at the cysteine α-carbon, and the dehydroalanine can react with a thiolate to form a thioether or with a lysine or N-terminal amine to form a secondary amine. The thioether route has been seen in an IgG1 antibody and in cyclic disulfide peptides.
3. α-elimination, giving a thiolate and a thioaldehyde.

A free cysteine “may cause unwanted intramolecular disulfide scrambling, and covalent oligomerization via intermolecular disulfide formation” [11]. Scrambled isomers have the same mass as the native peptide, so they are found by comparing retention times and by mapping which cysteines are linked. EMA asks for characterisation of “disulfide bridges and cyclic structures” where relevant [3].

Hydrolysis of the peptide backbone

Peptide bonds are kinetically stable, but a few motifs are weak points. Under strong acid, the Asp-Gly bond in Oliyai and Borchardt’s model peptide was the one that hydrolysed [8]. At neutral and alkaline pH, asparagine-linked cleavage takes over. In Geiger and Clarke’s study, peptide cleavage products appeared when the Asn was followed by leucine or proline [4]. Tyler-Cross and Schirch found that every peptide with serine after asparagine also cleaved in neutral and alkaline solution, at “about 10% of the rate of the deamidation pathway” [5].

Hydrolysis also happens in dry powders. When D’Hondt and colleagues heated solid buserelin to test whether it could survive hot-melt extrusion, the degradation hypotheses the authors drew from mass spectrometry of the degradants included direct backbone hydrolysis, β-elimination at serine and isomerisation [18]. That was a deliberate forced-degradation study, not a storage study, but it shows the reactions a stability-indicating method has to capture.

Cleavage products are usually the easiest degradants to assign: two fragments whose masses add up to the parent plus one water molecule.

Diketopiperazine and pyroglutamate formation

Both reactions attack the N-terminus.

Diketopiperazine (DKP). The free N-terminal amine attacks the carbonyl of the second peptide bond, and the first two residues leave as a cyclic dipeptide. Goolcharran and Borchardt measured this with Phe-Pro model compounds and found the rate depended on the ionisation of the N-terminal amine, “with the unprotonated reactant being more reactive than the protonated form”, a kinetic pKa of 6.1, general base catalysis by phosphate between pH 5 and 8, and a significant effect of the residue preceding proline [12]. D’Hondt and colleagues list DKP among the typical degradation-related impurities of peptide medicines [15]. In LC-MS it shows up as a truncated peptide missing its first two residues.

Pyroglutamate. An N-terminal glutamine or glutamic acid can cyclise onto its own side chain. “Cyclization of N-terminal glutamine to pyroglutamate is a common modification” of recombinant antibodies, usually identified by LC-MS [14]. For N-terminal glutamic acid, Chelius and colleagues found minimal pyroglutamate formation at pH 6.2, more at pH 4 and pH 8, and a half-life for N-terminal glutamate of about 9 months at pH 4.1 and 45 °C [13]. Their work was on antibodies, so the numbers transfer to peptides only as a guide; the pH dependence is the useful lesson.

Racemisation

Racemisation converts an L-residue to its D-form. It changes shape and potentially activity without changing mass, which makes it one of the harder degradants to see.

During storage, the fastest route goes through the succinimide: in the Geiger and Clarke peptide, the succinimide racemised with a half-time of 19.5 hours, and “most of” the observed racemisation “could be accounted for” by it [4]. During synthesis, Fmoc deprotection “can cause racemization of amino acid residues and thus diastereomeric impurities” [15].

Detection needs dedicated methods. Chiral gas chromatography after acid hydrolysis is often used, and because hydrolysis itself causes some epimerisation, it is carried out in deuterated hydrochloric acid so that residues racemised during the test can be distinguished by mass spectrometry from those racemised before it [3]. EMA also notes that “control of diastereomers may require the development of specific methods” [3].

Aggregation and fibrillation

Physical instability means molecules associating while staying chemically intact. Zapadka and colleagues review the drivers: sequence, concentration, pH, net charge, excipients, chemical degradation, surfaces and interfaces and impurities, plus physical parameters including “pressure, temperature, agitation and lyophilization” [16]. Aggregates can be amorphous or “highly structured fibrillar species” [16].

Fibrillation follows nucleation kinetics. In Nielsen and colleagues’ insulin study, followed by thioflavin T fluorescence, fibril formation had a lag phase for nucleation followed by growth; added seeds “eliminated the lag phase”, higher concentration shortened the lag, and acidic pH gave shorter lag times and faster growth than neutral pH [17]. In their proposed model, the air-water interface stimulates formation of the partially folded intermediates that nucleate fibrils [17]. Glucagon shows how varied the outcome can be: it “can form several types of fibrils” through distinct nucleation-dependent pathways, each able to self-propagate by seeding [19]. Both are in vitro findings on specific hormones, and in both, existing fibrils seeded further fibril growth.

For synthetic peptides, EMA expects high-molecular-weight impurities such as aggregates and oligomers to be controlled, for example by SEC-LC, and suggests thioflavin T and intrinsic tryptophan fluorescence to investigate “whether a peptide can form fibrillary aggregates” [3].

Does lyophilization stop peptide degradation?

It slows it. Lai and Topp’s review of solid-state stability lists deamidation, peptide bond cleavage, oxidation, the Maillard reaction, β-elimination and dimerisation or aggregation as reactions that still occur in dried peptides and proteins [20]. The factors that set the rate are temperature, moisture content, excipients and the physical state of the formulation, amorphous or crystalline [20]. Lyophilisation is also on Zapadka’s list of physical stresses that can promote aggregation [16].

That is why EMA expects water content to be “part of the stability protocols” for hygroscopic peptide powders [3], and why the usual storage conditions are refrigerated (5 °C ± 3 °C) or frozen (−20 °C ± 5 °C) [3]. The formulation and cycle choices that control residual moisture are covered in our peptide lyophilisation guide.

Signs of peptide degradation: what changes in the data

Most degradation is invisible to the eye. The reliable signs are analytical:

What you see	Likely cause	Method
Main HPLC peak falls; new peaks before or after it	Any chemical pathway	Related-substances RP-HPLC
New peak about 1 Da heavier than the parent	Deamidation	LC-MS
New peak 16 Da heavier (or 32 Da)	Oxidation (one or two oxygens)	LC-MS; MS/MS or peptide mapping to locate it
Same mass, different retention time	Isomerisation (isoAsp), racemisation, disulfide isomer	HPLC with a method shown to resolve them; chiral analysis
Two smaller species whose masses sum to the parent plus water	Backbone hydrolysis	LC-MS
Species missing the first two residues	Diketopiperazine formation	LC-MS
Species about 17 or 18 Da lighter at the N-terminus	Pyroglutamate from Gln or Glu	LC-MS
Species at twice the mass or higher	Dimers and oligomers	SEC; LC-MS
Rising thioflavin T fluorescence	Fibrils	ThT assay

Visible changes, when they occur, mostly signal physical instability: precipitation is one of the physical degradation routes Manning and colleagues describe [1], and fibrillation can be followed in solution because fibrils bind thioflavin T [17]. A clear solution is not evidence of a stable peptide. Our guide to peptide purity testing explains how the related-substances method is built, and peptide mass spectrometry covers how degradants are identified.

Forced degradation under ICH Q1A(R2) and Q1B

Forced degradation, or stress testing, deliberately breaks the molecule to learn how it breaks. ICH Q1A(R2) says stress testing “can help identify the likely degradation products”, establish degradation pathways and “validate the stability indicating power of the analytical procedures used” [21]. It is likely to be done on a single batch and should include [21]:

• temperature in 10 °C increments (for example 50 °C, 60 °C) above the accelerated condition;
• humidity of 75% RH or greater where appropriate;
• oxidation;
• photolysis;
• hydrolysis “across a wide range of pH values” in solution or suspension.

ICH Q1B splits photostability into two parts: forced degradation testing, whose design “is left to the applicant’s discretion”, and confirmatory testing [22]. Confirmatory samples receive an overall illumination of not less than 1.2 million lux hours and integrated near-ultraviolet energy of not less than 200 watt hours per square metre [22]. Degradants that appear only under forcing conditions and are shown not to form in confirmatory studies “need not be further examined” [22].

How hard to push? Blessy and colleagues note that degradation “between 5% and 20% has been accepted as reasonable for validation of chromatographic assays”, and note that some scientists consider about 10% optimal for small molecules [23]. Their table of commonly used conditions includes 0.1 M hydrochloric acid and 0.1 M sodium hydroxide at 40 °C and 60 °C, 3% hydrogen peroxide at 25 °C and 60 °C, a radical initiator, light at one and three times the ICH exposure, and dry and humid heat at 60 °C and 80 °C, each sampled at 1, 3 and 5 days [23]. Those are small-molecule defaults; for peptides and proteins, Hawe and colleagues point out that standard forced-degradation procedures “are largely unavailable, except for photostability”, and add freeze-thawing and mechanical stress to the list [24].

For synthetic peptides, EMA says forced degradation studies “are foreseen to evaluate both, the degradation of the peptide and the ability of the analytical procedures to detect the degradation” [3]. That second purpose links stress testing to method validation: a related-substances method is only stability-indicating if the degradants from the table above are separated and measured. Bakshi and Singh’s review of stability-indicating assay methods lists separation of all degradation products and establishment of mass balance among the critical issues [25]. Our guide to HPLC method validation under ICH Q2(R2) covers that step.

How to slow peptide degradation in the laboratory

Each pathway above has a lever, and most of them are the same few controls:

• pH. Choose a pH that suits the sequence: acid favours Asp-X cleavage and direct Asn hydrolysis [8], [6]; neutral to basic pH favours succinimide formation, DKP formation and disulfide degradation [5], [12], [11].
• Temperature. Deamidation showed marked temperature dependence in model peptides [6], and refrigerated or frozen storage is EMA’s usual default for peptides [3].
• Water. Keep powders dry and let cold containers reach room temperature before opening, since solid-state rates depend on moisture content [20].
• Oxygen, metals and light. Protect from light, consider desiccant or an inert atmosphere for hygroscopic powders [3], and remember that for metal-catalysed oxidation chelators are the better tool, because antioxidants may accelerate it [9].
• Handling. Avoid unnecessary agitation and repeated freeze-thaw cycles, both recognised protein stresses [24], [16].

How Winhycare Pharma supplies research peptides

Winhycare Pharma supplies research peptides from Hyderabad, India, as research-use-only material, at purity of ≥99% as stated in our product brochure. Semaglutide, tirzepatide, retatrutide, cagrilintide and BPC-157 are also supplied as API to licensed manufacturers; see peptide APIs. For sequences outside the catalogue, including peptides with sensitive residues, we offer custom peptide synthesis. Ask for the batch Certificate of Analysis with your quote: request research peptides.

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