Forced degradation
Forced degradation applies stresses harder than the molecule will normally see, to reveal its degradation routes quickly. It answers two questions: which liabilities are real rather than theoretical, and do the analytical methods detect the products when they appear.
The standard stresses
Thermal. Days at an elevated temperature. Accelerates most chemical routes and promotes aggregation.
Low pH. Mimics the Protein A elution and viral inactivation holds, and drives aspartate isomerization.
High pH. Drives asparagine deamidation hard, which is the quickest way to find out whether a predicted NG motif is exposed.
Oxidative. A peroxide challenge at low percentage for a few hours oxidizes exposed methionine and, more slowly, tryptophan. Metal-catalyzed oxidation is a more biologically relevant variant.
Photostability. Light exposure under defined conditions, which finds tryptophan oxidation and photo-induced fragmentation.
Mechanical. Agitation, freeze-thaw cycles, and pumping, which produce aggregate and particles by interfacial stress rather than by chemistry.
Reading the result
The aim is to reach partial degradation, not destruction. Conditions that leave something like five to twenty percent of the molecule modified give you products to measure and a molecule still worth measuring. A sample pushed to complete degradation tells you only that the stress was too hard.
For each stress, the readouts are size, by size exclusion and by a gel or capillary method; charge, by an ion exchange or isoelectric focusing method; potency, by the binding or functional assay; and chemistry, by peptide mapping, which localizes the modification to a residue.
Pairing potency loss with a site-level modification is the point of the exercise. A deamidation site that appears under stress and correlates with a drop in binding is a liability worth engineering out. One that appears with no functional consequence can be monitored instead.
Where it feeds back into design
A panel run through a short forced degradation screen ranks differently from the same panel ranked on affinity, and the difference is informative early, while sequences can still be changed. It is also the fastest way to decide whether a computed liability is real: the model flags an NG motif, the high-pH stress either produces deamidation at that site or does not, and the question is settled in a week.
What it does not tell you
Accelerated conditions do not predict shelf life quantitatively. They reveal routes; real-time stability at the intended storage condition establishes the shelf life. Programs that substitute one for the other end up either over-engineering a molecule against a route that would never have mattered, or missing a slow route that only appears at the real temperature.
Used as a diagnostic rather than a prediction, it is one of the highest-value experiments available at small scale.