Thermal shift assays
A thermal shift assay heats a protein slowly and watches it unfold. The midpoint of that transition, the melting temperature, is a fast and cheap measure of conformational stability, and it is one of the first numbers worth having on a new panel.
The two common methods
Dye-based differential scanning fluorimetry adds a dye that fluoresces when it binds exposed hydrophobic surface. As the protein unfolds, the signal rises. It runs on a standard qPCR instrument, needs little material, and is the cheapest option. The dye can interact with hydrophobic proteins and with detergents, so some molecules are not compatible.
Intrinsic fluorescence, often called nanoDSF, follows tryptophan and tyrosine fluorescence as their environment changes, with no dye at all. It works with nearly any protein, uses very little material, and many instruments simultaneously measure light scattering, which gives the aggregation onset temperature alongside the melting temperature.
Differential scanning calorimetry measures heat directly and remains the reference method, at the cost of far more material and time.
Two different numbers
The melting temperature describes when the fold comes apart. The aggregation onset describes when the molecules start sticking together. They are related and they are not the same, and for formulation the aggregation onset is often the more useful of the two, because a protein that unfolds reversibly at a high temperature can still aggregate early.
For a multi-domain molecule, expect more than one transition. An IgG typically shows separate transitions for its domains, and knowing which domain melted first is more informative than a single averaged number.
Running a fair comparison
- One buffer. Stability is a property of the protein in its buffer. A panel in mismatched buffers produces a ranking of buffers.
- One concentration. Aggregation onset in particular depends on concentration.
- One ramp rate. Unfolding is kinetically influenced, so the measured midpoint shifts with ramp rate.
- Replicates, because the midpoints of similar candidates often differ by less than the run-to-run spread.
- A reference molecule in every run to catch instrument and reagent drift.
What it predicts, and what it does not
Thermal stability correlates with shelf life and with manufacturability, loosely. It is a good screen for finding the candidates that will be difficult, and a poor substitute for a real stability study at storage temperature, because the degradation routes that matter over months are often chemical rather than thermal, and because agitation and interfaces drive aggregation in ways a thermal ramp does not capture.
A useful way to use it: rank a panel, drop the clear outliers, then run a short real-time and accelerated stability study on the survivors and let that decide.