This is just a guess, but it's probably because the heat measurement is actually measuring the sum of many, many molecules reactions. It is a fact of mathematics, rather than natural science, that forces such phenomena to approximately follow bell curves (i.e., Gaussian functions). Basically, the central limit theorem states that if you have many instances of the same (random) variables and average them together, then the result will look like a bell curve. Here, the local thermodynamic change induced by a single individual protein molecule, as a function of the temperature, is presumably that random variable, since the thermodynamic behaviour of each protein is roughly similarly distributed. The randomness comes from the essentially chaotic random movements of the molecules.
See here or here for more.