Parallel molecular evolution of deletions and nonsense mutations in bacteriophage T7.

Abstract

The extent to which identical evolutionary changes appear in closely related, independent lineages, a phenomenon known as parallel evolution, is a central question in evolutionary biology (Kimura 1983; Gould 1989). Inferring parallel evolution from extant lineages is instructive (e.g., Swanson et al. 1991; Doyle, Doyle, and Palmer 1995; Shyue et al. 1995), but may be misleading if rates are high enough to bias either the reconstruction of ancestral character states or reconstruction of the phylogeny itself (Felsenstein 1978, 1985). The surest way to detect parallel evolution is with experimental studies of microorganisms. Although previous studies of experimental evolution have documented parallel phenotypic changes, these changes have not generally been documented at the DNA sequence level (e.g. Dykhuizen 1990; Travisano et al. 1995). Phylogenies of bacteriophage T7 grown in the presence of a mutagen have proven to be an unusually flexible experimental system to evaluate phylogenetic methods (Hillis et al. 1992; Hillis, Huelsenbeck, and Cunningham 1994), and to study molecular evolution (Studier 1980; Bull et al. 1993). This system not only displays very rapid DNA sequence evolution, but allows the direct observation of ancestral states during the evolution of each lineage. Here we document a complex pattern of parallel evolution at the DNA sequence level. Our results suggest caution when reconstructing ancestral states of characters under directional selection, as well as caution against giving undue phylogenetic weight to insertion-deletion events. We serially propagated six bifurcating lineages of bacteriophage T7 in the presence of a mutagen according to an earlier protocol (fig. 1; Hillis et al. 1992). The lineages were bottlenecked to a single individual at each node shown in figure 1. This approach allowed the ancestral condition to be known with certainty at several points in the history of each lineage. Our analysis focused on DNA sequences from a region that, in wildtype phage, spans just over 2 kb and contains five genes 0.3, 0.4, 0.5, 0.6, and 0.7 (Dunn and Studier 1983: DNA alignments deposited in EMBL. Aligned 0.3 sequences as DS 27 143, aligned 0.7 sequences as DS 27 142). Although our wild-type ancestral stock contained the entire 0.3-0.7 region, every lineage evolved a 1.5-kb deletion that fused the 0.3 and 0.7 genes. Nine independent deletions were observed, but seven of them had breakpoints identical to the previously characterized