Peptide stability and degradation
Peptides degrade by a small number of well characterized chemical routes, and each one is readable from the sequence before any material exists. Knowing which apply lets you design the storage, the buffer and the stability study around the real risks rather than testing everything.
Hydrolysis at the backbone
Aspartate is the reactive residue. Under mildly acidic conditions the Asp-Xaa bond hydrolyzes faster than any other backbone amide, with Asp-Pro the most labile pair by a wide margin. If a sequence contains Asp-Pro and the formulation sits below pH 5, backbone cleavage is the degradation route to expect.
Under neutral and alkaline conditions the dominant route through aspartate is different. The side chain attacks the following backbone nitrogen to form a cyclic succinimide, which then opens to a mixture of aspartate and isoaspartate. Isoaspartate adds a methylene to the backbone, which almost always destroys activity and is invisible by mass. Asp-Gly is the fastest motif, followed by Asp-Ser and Asp-Asn, and the rate climbs steeply above pH 7.
Deamidation
Asparagine deamidates to aspartate and isoaspartate through the same succinimide intermediate, adding one dalton. Asn-Gly is the fast motif and can have a half-life of days at physiological pH and temperature. Glutamine deamidates by the same chemistry roughly an order of magnitude more slowly.
The consequence is a charge change, so deamidation is detectable by ion exchange or by capillary isoelectric focusing long before it is obvious by reversed phase. Where an Asn-Gly sits in or near the active region, it is worth knowing whether the sequence can tolerate a conservative substitution.
Oxidation
Methionine oxidizes to the sulfoxide, adding sixteen daltons, driven by dissolved oxygen, trace metals and light. Cysteine oxidizes to disulfides and further to sulfinic and sulfonic acids. Tryptophan oxidizes through several products and is also the residue most sensitive to light.
Trace metal is usually the accelerant rather than oxygen alone, which is why a chelator at low concentration often does more than headspace control. Where methionine oxidation is the risk, the practical measures are a chelator, protection from light, and avoiding peroxide-containing excipients, some polysorbate lots included.
Disulfide behavior
A peptide with two cysteines has one possible disulfide. With four there are three regioisomers, and with six there are fifteen. All have identical mass, so mass spectrometry cannot distinguish them and a clean total ion chromatogram says nothing about connectivity.
Scrambling is base-catalyzed and thiol-catalyzed. Above pH 7, and in the presence of any free thiol including a trace of reduced peptide, the disulfides equilibrate toward a mixture. Keeping the formulation mildly acidic and free of reducing agent is the main defense, and confirming connectivity requires peptide mapping rather than intact mass.
Aggregation and surface loss
Hydrophobic and amphipathic peptides aggregate, and many form amyloid-like fibrils given time and an interface. Agitation, freeze-thaw cycling and air-liquid interfaces all nucleate it.
Surface adsorption is the failure that masquerades as low potency. A peptide at low micromolar in a plain polypropylene tube can lose a substantial fraction of its mass to the plastic, and the assay then reports a weak compound. Low-binding tubes, a carrier protein where the assay tolerates one, and avoiding dilute aqueous intermediate stocks all address it.
What this means for handling
Store lyophilized, cold, dry and dark. The dry state suppresses every hydrolytic route, and a peptide that survives years as a powder can degrade measurably in solution within a week.
Aliquot on reconstitution so that no vial is thawed twice, and record the actual date of reconstitution rather than the date on the certificate.
Choose the buffer pH from the sequence. Mildly acidic, around pH 4 to 5, suppresses deamidation, succinimide formation and disulfide scrambling, and is the right default unless an Asp-Pro bond or a solubility limit argues otherwise.