From Proteins to Organisms

Introduction One of the most intriguing problems in molecular biology is the origin of the vast population diversity of protein families. Following the assumption that the protein families are populated at random, one would expect a multinomial distribution of the family populations.5 However, it has been discovered6-9 that distribution of the family populations is by far nonexponential, but has a long tail, which signifies that some specific mechanisms govern populations of protein families. To explain such diversity, there emerged two views of convergent and divergent evolution (Fig. 1). In the convergent evolution scenario, it is postulated that the present population distribution of protein fold families is the result of convergent processes in the course of evolution which were selectively populating folds. In the simplest scenario, it is presumed that evolution has reached equilibrium in the protein structural space.** Due to the underlying physical nature of evolutionary processes, i.e., the physical nature of amino acid interactions that underlie the properties of specific folds, the expected equilibrium distribution of family population follows the Boltzman distribution. Thus, more “designable” folds, that can be encoded by many sequences, have a higher representation in genomes. This assumption, called the “designability principle”, is based on phenomenological considerations13 and on observations drawn from exhaustive enumeration of all sequences in simplified twoand three-dimensional lattice protein models. In the course of evolution more designable folds become more populated than less designable folds, which results in the uneven distribution of observed populations of protein families. There have been several arguments based on various observations favoring convergent evolution. Teichmann et al proposed that structural similarities arise solely due to physical interactions that favor particular packing and chain topologies. Functional pressure was proposed to be the paladin of protein structural convergence. One of the most striking example is that of the Ser/His/Asp catalytic triad,10,21 which is found in a number of folds that have no significant sequence similarity. Antifreeze proteins (AFP) provide a crucial defense for