Thermal proteome profiling (TPP, also called CETSA at single-protein scale) heats intact cells or lysate to a series of temperatures, pellets the proteins that unfold and aggregate by centrifugation, and quantifies whatever stays in the soluble supernatant by mass spectrometry. Each protein's soluble fraction follows a sigmoidal melt curve:
f(T) = 1 / (1 + exp((T - Tm) / k))
where Tm is the temperature at which half the population has denatured and k sets the transition sharpness. A drug that binds and thermodynamically stabilises its target protein raises that protein's Tm — the classic melt-shift assay. The size of the shift follows a saturating (Langmuir) binding isotherm in ligand concentration [C]:
ΔTm(C) = ΔTm_max · C / (Kd + C)
An off-target / control protein that the drug does not bind shows no concentration-dependent shift — exactly what distinguishes a real target from background noise in a real TPP screen.
- Temperature slider / Run Melt Ramp — heat-challenges both populations; each of the 30 protein copies per group carries its own randomly offset melting threshold (population heterogeneity), so the group denatures gradually rather than all at once, matching a real sigmoidal curve.
- Concentration / Kd — set how much ligand is present and how tightly it binds; together they set the target's apparent ΔTm via the isotherm above. The control population ignores both sliders.
- Aggregation is irreversible — once a copy fully denatures it sinks into the pellet and stays there even if you cool back down, just as precipitated protein does not redissolve on ice.
Real-world relevance: this is the assay used to identify direct drug targets and off-target liabilities in intact cells without any prior knowledge of the mechanism — e.g. panobinostat and staurosporine target deconvolution, and large-scale kinase inhibitor profiling.