Contingency and chance erase necessity in the experimental evolution of ancestral proteins

To understand why evolution produced the biological systems that exist today, we must know how important chance, contingency, and necessity were during history. Previous observations suggest that each of these modes of causality affects evolution in various settings, but their relative roles and interactions are not well characterized because they have never been systematically assessed in a single system or on a timescale relevant to evolutionary history. To this end, we reconstructed ancestral B-cell-lymphoma-2-family proteins and developed a continuous evolution method to select for defined protein-protein interaction specificities. By repeatedly evolving a series of ancestral proteins to acquire specificities that occurred during history, we show that contingency steadily overwhelms chance and erases necessity as the primary cause of sequence variation in proteins over long phylogenetic timescales. As a result, evolutionary trajectories launched from distant starting points are essentially unpredictable, even under strong and identical selection pressures. Genetic dissection of the outcomes shows that chance arises because numerous sets of mutations can alter specificity at any point in time, while contingency arises because historical substitutions change these sets. Patterns of variation in extant protein sequences are therefore largely the idiosyncratic product of a particular course of unpredictable historical events.