Going down the folding funnel, proteins may sample a wide variety of conformations, some being outright detrimental to the organism. Yet, the vast majority of polypeptide molecules avoid such pitfalls. Not only do they reach the native minimum of the energy landscape; they do so via blazingly fast, biased, routes. This specificity and speed is remarkable, as the surrounding solution is filled to the brim with other molecules that could potentially interact with the protein and in doing so stabilise non-native, potentially toxic, conformations. How such incidents are avoided while maintaining native structure and function is not understood. 

This doctoral thesis argues that protein structure and function can be separated in the folding code of natural protein sequences by use of multiple partly uncoupled factors that act in a concerted fashion. More specifically, we demonstrate that: i) Evolutionarily conserved functional and regulatory elements can be excised from a present day protein, leaving behind an independently folded protein scaffold. This suggests that the dichotomy between functional and structural elements can be preserved during the course of protein evolution. ii) The ubiquitous charges on soluble protein surfaces are not required for protein folding in biologically relevant timescales, but are critical to intermolecular interaction. Monomer folding can be driven by hydrophobicity and hydrogen bonding alone, while functional and structural intermolecular interaction depends on the relative positions of charges that are not required for the native bias inherent to the folding mechanism. It is possible that such uncoupling reduces the probability of evolutionary clashes between fold and function. Without such a balancing mechanism, functional evolution might pull the carpet from under the feet of structural integrity, and vice versa. These findings have implications for both de novo protein design and the molecular mechanisms behind diseases caused by protein misfolding.