Supplementary Figure 2: A summary of variations in noncoding sequences.

(a) The origin regions and repeat structure of the Drosophila noncoding region. The central panel shows type I (brown) and type II (gray) repeats of a typical D. melanogaster mitochondrial genome (for example, mt:ND2^del1 + mt:CoI^T300I). Individual repeats for each type are highly conserved¹. The ATP6[1] genome lacks two entire type I repeats (B1 and A/C) and two entire type II repeats (B2 and C)², whereas the D. yakuba genome lacks the majority (>75%) of the repeated sequences³. Nucleotide sequences near the origins of replication for the heavy (O_H) and light (O_L) chains for the mt:ND2^del1 + mt:CoI^T300I, ATP6[1] and D. yakuba genomes show conservation with potentially significant polymorphisms. The direction of replication is indicated by an arrow. The 5′ ends of mtDNA were mapped to define the sites at which mtDNA synthesis began in the D. yakuba genome: the nucleotides in the green boxes are the sites where the ends were mapped for O_H and O_L (ref. 4). (b) Rapid divergence of the noncoding region suggests positive selection. The divergence of the noncoding region is so fast and includes so many deletions and insertions that it is difficult to make a meaningful determination of the number of changes separating the genomes of different species. Recently reported complete genome sequences for 13 mitochondrial haplotypes from diverse wild strains of D. melanogaster⁵ allowed us to compare more recently diverged genomes. Even among these, the number of changes makes some comparisons difficult. We selected four pairs of strains where each pair represents two especially closely related genomes and each pair represents a different branch of the tree of relatedness for the 13 sequenced genomes⁵. The tree shown here (left) gives the relatedness of the four pairs of sequences we analyzed. For comparison, the relatedness to D. yakuba, the reference sequence for D. melanogaster (NC 024511), and the sequence we obtained for the temperature-sensitive genome used in this study (derived from a laboratory stock of w1118) are also indicated. Because the target size for synonymous mutation in protein-coding sequence is about one-third of the total, the ~12 kb of protein-coding sequence for the mitochondrial genome provides about 4,000 potential sites for synonymous changes, an amount roughly equal to the total number of sites in the non-coding region. As synonymous changes are in general considered neutral, the number of synonymous changes in the protein-coding regions should reflect the neutral mutation rate, and if the number of changes in the noncoding region is substantially higher positive selection could be invoked. The distribution of sequence differences distinguishing the members of each pair is shown on the right. (c) Southern blot analysis showing length polymorphism of the noncoding region of mtDNA from several Drosophila species. Total DNA was isolated from flies homoplasmic for various mitochondrial genotypes and then digested with XbaI and HindIII. The Southern blot was probed with a DIG-labeled DNA fragment recognizing mt21–400 (pink).