How does protein structure optimise function?

It actually has something to do with strong electronic correlations and the Jahn-Teller effect. Furthermore, it illustrates chemistry is very local.

There is a nice review article, Molecular modelling for transition metal complexes: Dealing with d-electron effects by Deeth, Anastasi, Diedrich, and Randell. It discusses specific examples of the Jahn-Teller effect in d9 Cu(II) complexes. [Aside: this the same complex as in cuprate superconductors].

Type I blue copper proteins are involved in electron transfer in both plant and bacterial photo-synthesis. A key question concerns how the structure around the Cu active site might be optimised in order to facilitate electron transfer. It was proposed that the local environment was arranged to minimise the reorganisation energy in the Marcus-Hush electron transfer theory. However, using quantum chemistry calculations, Ryde, Olssen, Pierloot, and Roos, showed that this was not the case in 1996. In fact the local geometry was the same in vacuo as in the protein. Furthermore, the structure was dominated by the strong Cu-thiolate bond.

Hence, Deeth et al., state

We begin with the presumption that although protein molecules may be large and complex, they form simple M–L coordinate bonds – i.e. a Cu–N(imidazole) bond can be handled in exactly the same way irrespective of whether it corresponds to an isolated imidazole or to a histidine.

This allows them to develop force fields for molecular mechanics.
Chemistry is local. This is why localised approaches such as valence bond theory are often preferable to molecular orbital or DFT-based approaches which tend to delocalise electrons.