Amino acid preferences of small proteins. Implications for protein stability and evolution.

The dependence of amino acid frequency on sequence length has been examined for the 20 natural amino acids using a set of 2275 protein sequences with little sequence identity. As expected, the frequency of cysteine increases dramatically for sequences shorter than 100 amino acids with a length-dependence that corresponds to an average of two Cys per sequence independent of length. Surprisingly dramatic changes were also observed for the frequencies of arginine, lysine, aspartic acid, and glutamic acid: Arg and Lys frequencies increase for short sequences whereas Asp and Glu frequencies decrease. These changes do not appear to be due to an over-abundance of DNA- and membrane-binding proteins in the database and may, therefore, be related to protein stability. Possible stabilizing mechanisms include increased hydrogen bonding by Arg and increased hydrophobic stabilization due to the amphiphilic character of Arg and Lys. These observations suggest that amino acid composition played an important role in the evolution of small proteins.