Detecting site-specific biochemical constraints through substitution mapping.

The neutral theory of molecular evolution states that most mutations are deleterious or neutral. It results that the evolutionary rate of a given position in an alignment is a function of the level of constraint acting on this position. Inferring evolutionary rates from a set of aligned sequences is hence a powerful method to detect functionally and/or structurally important positions in a protein. Some positions, however, may be constrained while having a high substitution rate, providing these substitutions do not affect the biochemical property under constraint. Here, I introduce a new evolutionary rate measure accounting for the evolution of specific biochemical properties (e.g., volume, polarity, and charge). I then present a new statistical method based on the comparison of two rate measures: a site is said to be constrained for property X if it shows an unexpectedly high conservation of X knowing its total evolutionary rate. Compared to single-rate methods, the two-rate method offers several advantages: it (i) allows assessment of the significance of the constraint, (ii) provides information on the type of constraint acting on each position, and (iii) detects positions that are not proposed by previous methods. I apply this method to a 200-sequence data set of triosephosphate isomerase and report significant cases of positions constrained for polarity, volume, or charge. The three-dimensional localization of these positions shows that they are of potential interest to the molecular evolutionist and to the biochemist.