Neutral evolution of the nonbinding region of the anthocyanin regulatory gene Ipmyb1 in Ipomoea

Abstract

Plant transcription factors often contain domains that evolve very rapidly. Although it has been suggested that this rapid evolution may contribute substantially to phenotypic differentiation among species, this suggestion has seldom been tested explicitly. We tested the validity of this hypothesis by examining the rapidly evolving non-DNA-binding region of an R2R3-myb transcription factor that regulates anthocyanin expression in flowers of the genus Ipomoea. We first provide evidence that the W locus in Ipomoea purpurea, which determines whether flowers will be pigmented or white, corresponds to a myb gene segregating in southeastern U.S. populations for one functional allele and one nonfunctional allele. While the binding domain exhibits substantial selective constraint, the nonbinding region evolves at an average K(a)/K(s) ratio of 0.74. This elevated rate of evolution is due to relaxed constraint rather than to increased levels of positive selection. Despite this relaxed constraint, however, approximately 20-25% of the codons, randomly distributed throughout the nonbinding region, are highly constrained, with the remainder evolving neutrally, indicating that the entire region performs important function(s). Our results provide little indication that rapid evolution in this regulatory gene is driven by natural selection or that it is responsible for floral-color differences among Ipomoea species.

Figure 1.—

Domains of Ipmyb1 from Ipomoea purpurea. The shaded region is the R2R3 binding domain. Solid vertical lines are intron positions (i1 and i2). The open region is the variable domain. Arrows (labels) indicate positions (identity) of PCR primers. Triangles indicate deletions in ipmyb1. The asterisk indicates the position of the premature stop codon in ipmyb1.

Figure 2.—

Phylogenetic relationship of the 13 species examined. Branch lengths convey no information. Numbers associated with each branch represent estimated d_S (outside parentheses) and d_N (inside parentheses) from model M1 (neutral model) for the variable region of Ipmyb1.

Figure 3.—

Amino acid alignment of variable-region sequences. Amino acid abbreviations in boldface type in the I. purpurea reference sequence indicate highly conserved sites. Dots indicate amino acids that do not differ from the I. purpurea reference sequence. Dashes indicate insertions/deletions.

Figure 4.—

Amino acid alignment of R2R3 binding-domain sequences. Dots indicate amino acids that do not differ from the I. purpurea reference sequence.

Figure 5.—

Estimated probability density distribution of ω from model M7. The distribution function is a β-distribution with parameters corresponding to the maximum-likelihood fit to the data. (A) Binding domain. (B) Variable domain.

Figure 6.—

Frequency histograms of mean change in amino acid properties derived from simulations under the assumption that nonsynonymous mutations are neutral. For each property, 1000 simulations were performed. Arrows indicate mean of observed amino acid substitutions. Bars above histograms indicate the 1st, 5th, 95th, and 99th percentiles of the cumulative frequency distribution.