When comparative information leads us astray: the receptor-binding region of colicin E9

In an effort to develop derivatives of the Escherichia coli antimicrobial protein colicin E9 that exhibit novel interactions with a target cell, we mutagenized a 10-amino acid region located at the C terminus of the colicin receptor-binding domain. We subsequently selected for those colicin molecules that retain the antimicrobial phenotype and found that, despite a mutagenic strategy that alters every amino acid in the targeted domain, more than 70% of the engineered colicins retained antimicrobial activity. This result is all the more surprising given the extensive phylogenetic conservation of this receptor-binding domain, which originally suggested the operation of strong selective constraints on the amino acid sequence of this region. This apparent contradiction between our experimental results and the comparative data is resolved by exploring the fitness consequences of the experimentally induced amino acid substitutions. In 17 of 52 cases we examined, the fitness of cells harboring the functional engineered colicins was lower than that of our control line (containing wild-type colicin E9), and in 33 of 52 cases, equal to it. Paradoxically, two of the engineered colicins appear to confer a higher fitness to the producer cell lines. While the mechanism linking changes in the amino acid sequence of the colicin receptor-binding domain and the growth rate of the cells remains unclear, these results illustrate the surprising versatility of the colicin/receptor interaction and underscore the importance of distinguishing molecular function from organismal fitness.