Small-scale expression screening

For a new antigen or a difficult construct, the fastest way to find out what expresses is to try several versions at small scale in parallel. Twenty-four to ninety-six conditions in deep-well plates or small shake flasks, harvested after a few days, tells you more in a week than reasoning about signal peptides ever will.

What to vary

Construct. Domain boundaries first: for an ectodomain, two or three truncations often differ by an order of magnitude in yield. Then tag position and identity, signal peptide, and whether to include a linker or a fusion partner.

Host. HEK and CHO where relevant, and for small stable domains such as VHHs, a bacterial periplasmic construct is worth including because it is so cheap.

Conditions. Temperature shift after transfection, harvest day, and for bacteria, induction temperature and time.

Vary one axis at a time only if you have the wells to spare. For a first pass, a sparse matrix that covers the plausible combinations is more informative than a clean one-factor screen.

Reading the screen

Titer alone is a trap. A construct that expresses well and comes out aggregated is worse than a construct with half the yield and good monomer content. Screen with at least two readouts: quantity, typically by a quick affinity-based titer or by a small-scale purification, and quality, typically an analytical size exclusion run or a non-reducing gel.

For an antigen, add a functional check: does it still bind a reference antibody or its natural ligand? A truncation that expresses beautifully and lost its fold is the usual reason a screen looks successful and the antigen is useless.

What small scale predicts

Relative ranking of constructs transfers reasonably well to larger scale. Absolute titer does not; deep-well cultures have different mixing and gas transfer than a shake flask, and both differ from a bioreactor.

Purification behavior transfers only partly. A construct that binds the resin well at small scale can still be difficult at larger loads, and aggregation often appears at the concentrations only larger scale reaches.

So the screen picks the construct. The first real run still has to be characterized properly.

Practical notes

• Normalize everything to culture volume and to viable cell density so conditions are comparable.
• Keep the same detection method across the screen; switching between a titer assay and a gel densitometry mid-screen makes the comparison meaningless.
• Save small aliquots of every supernatant. Re-running a question on the same material later is worth far more than the freezer space.
• Record transfection details; the most common reason two identical constructs differ is the transfection, not the construct.

The screen is cheap insurance against the alternative, which is scaling a construct that was never going to work and finding out at the purification step.