Antigen quality control
A discovery campaign inherits the quality of its antigen. Selections run against it, assays measure against it, and every ranking is relative to it. A misfolded, aggregated or partly degraded antigen produces binders to the wrong thing, and the problem usually surfaces months later when the panel fails on cells.
Check one: is it the right molecule
Intact mass confirms sequence, tags, and whether the signal peptide cleaved where you expected. It also catches the construct that carried a mutation past plasmid sequencing. This takes one run and removes an entire class of downstream confusion.
Check two: is it monomeric
Analytical size exclusion, ideally with light scattering, tells you the oligomeric state and the aggregate content. This matters more for antigens than for antibodies, for two reasons. An aggregated antigen presents repeated epitopes, which creates avidity and makes weak binders look strong. And an antigen that is dimeric in the tube but monomeric on the cell selects for binders to an interface that does not exist in vivo.
Some antigens are legitimately oligomeric. Know which state yours should be in, and measure whether it is there.
Check three: is it folded
Mass and size say nothing about conformation. The practical test is binding to a reference antibody or to the natural ligand, with a measurable potency, not a yes or no. A conformational reference binder is the single most useful reagent to have for an antigen, and it is worth sourcing before the campaign starts.
Thermal unfolding is a useful supporting measurement: a correctly folded domain gives a cooperative transition, and a preparation that unfolds gradually across a wide range is usually a mixture.
Check four: is it modified the way it should be
Glycosylation, disulfide status and any post-translational modification that the epitope depends on. An antigen expressed in bacteria has none of the human modifications, which is fine for some targets and disqualifying for others. Phospho-epitopes need enzymatic or synthetic installation and verification.
Lot to lot
Treat antigen as a lot-controlled reagent. Record expression host, construct, tag, purification date, concentration, buffer, monomer content and reference binder potency for every lot, and re-run the reference binder when a new lot enters service. A quiet change in antigen lot between screening and confirmation is a standard explanation for a panel that stops reproducing.
For biotinylated antigen, add degree of labeling and the labeling method to that record, because random chemical labeling changes which epitopes are available.
When the antigen cannot be made well
Some targets are simply hard: multi-pass membrane proteins, complexes that need partners, antigens that are only correctly folded on a cell. In those cases the workable options are cell-based selection and assay, virus-like particles or nanodiscs, or a stabilized engineered construct whose relationship to the native protein you have verified. What does not work is pretending that a poorly behaved recombinant fragment is the target, and then explaining the results later.