Chromosomal rearrangement inferred from comparisons of 12 Drosophila genomes

Abstract

The availability of 12 complete genomes of various species of genus Drosophila provides a unique opportunity to analyze genome-scale chromosomal rearrangements among a group of closely related species. This article reports on the comparison of gene order between these 12 species and on the fixed rearrangement events that disrupt gene order. Three major themes are addressed: the conservation of syntenic blocks across species, the disruption of syntenic blocks (via chromosomal inversion events) and its relationship to the phylogenetic distribution of these species, and the rate of rearrangement events over evolutionary time. Comparison of syntenic blocks across this large genomic data set confirms that genetic elements are largely (95%) localized to the same Muller element across genus Drosophila species and paracentric inversions serve as the dominant mechanism for shuffling the order of genes along a chromosome. Gene-order scrambling between species is in accordance with the estimated evolutionary distances between them and we find it to approximate a linear process over time (linear to exponential with alternate divergence time estimates). We find the distribution of synteny segment sizes to be biased by a large number of small segments with comparatively fewer large segments. Our results provide estimated chromosomal evolution rates across this set of species on the basis of whole-genome synteny analysis, which are found to be higher than those previously reported. Identification of conserved syntenic blocks across these genomes suggests a large number of conserved blocks with varying levels of embryonic expression correlation in Drosophila melanogaster. On the other hand, an analysis of the disruption of syntenic blocks between species allowed the identification of fixed inversion breakpoints and estimates of breakpoint reuse and lineage-specific breakpoint event segregation.

Figure 1.—

Linkage chain analysis for reciprocal breakpoints generated by a single inversion event. Gene order is shown for two species where D. melanogaster has the standard gene order and D. ananassae has the rearranged gene order. The box indicates the segment that was inverted. The reusage statistic r (Sankoff and Trinh 2005) of 1.0 estimated from this linkage chain indicates no reusage of these breakpoints.

Figure 2.—

Linkage chain with five breakpoints that was generated by four inversion events. Gene order is shown for two species where D. melanogaster has the standard gene order and D. ananassae has the rearranged gene order. The reusage statistic (Sankoff and Trinh 2005) r of 1.6 estimated from this linkage chain indicates that 60% or three breakpoints were used more than once.

Figure 3.—

Plot of genome rearrangement for the six Muller elements among eight species of Drosophila. Each vertical line represents a single gene and the lines that connect the genes between the species help to determine the movement of the genes within and between chromosomal arms among the different species. The centromeres are indicated with solid half and whole circles. Each Muller element is shown in shades of a single color: A, red; B, green; C, blue; D, yellow; E, magenta; F, orange. Blocks of genes in D. grimshawi are arbitrarily colored within each Muller element. The blocks of genes in D. grimshawi do not correspond to syntenic blocks, but are presented as a visual heuristic to help observe shuffling of genes among the eight species. Some rearrangements might not be visible due to the compressed scale (on Muller F, for example).

Figure 4.—

Distribution of synteny blocks across various species based on the D. melanogaster euchromatic gene order. Species are shown in increasing evolutionary distance from D. melanogaster. The numbers to the left denote the approximate time range since a group of species shared most recent common ancestry with D. melanogaster (Tamura et al. 2004). The graph for each species shows the size of derived synteny blocks (in number of genes) on the horizontal axis (nonlinear scale showing size buckets: 2, 5, 10–100 interval 10, 150, 200–1000 interval 100, 1500, 2000) and the number of such blocks (log scale) on the vertical axis. These distributions show greater fragmentation of the genome compared to D. melanogaster with increasing evolutionary divergence. Additionally, species equidistant from D. melanogaster might show different degrees of fragmentation (as seen in D. yakuba vs. D. erecta, where D. yakuba exhibits greater fragmentation). These distributions are based on genome assembly scaffolds that were anchored to chromosome arms, where synteny blocks were allowed to span across scaffold breaks wherever possible. Species: D_ sec, D. sechellia; D_sim, D. simulans; D_yak, D. yakuba; D_ere, D. erecta; D_ana, D. ananassae; D_pse, D. pseudoobscura; D_per, D. persimilis; D_wil, D. willistoni; D_vir, D. virilis; D_moj, D. mojavensis; D_gri, D. grimshawi.

Figure 5.—

Paracentric inversions correlated with phylogenetic distance: Muller element B dot plots based on D. melanogaster (left column) and D. virilis (right column) gene orders. Species are shown in increasing evolutionary distance from D. melanogaster from top to bottom (reverse for D. virilis). Evolutionarily distant species show greater scrambling of gene order due to independent paracentric inversions across various lineages. For example, D. simulans shows greater arm-level synteny with D. melanogaster than D. virilis. Similarly, D. mojavensis exhibits the opposite behavior with respect to the reference species. Compressed scale is used to depict the chromosome arm.

Figure 6.—

Linkage chain distribution for two pairwise comparisons, D. melanogaster vs. D. ananassae and D. melanogaster vs. D. virilis. The numbers of chains are summed over Muller A–E for both comparisons.

Figure 7.—

Distribution of inferred multispecies conserved synteny blocks. The frequency of variously sized (number of genes) conserved blocks across nine species (excluding D. sechellia, D. simulans, and D. persimilis—see materials and methods) is shown (median = 3 genes, mode = 2 genes). The vertical axis (log scale) shows the number of blocks. Over 60% of the blocks have ≥3 genes. There are one or more large blocks of size ≥20 on each of the Muller elements, except Muller A (where the largest block has 15 genes). The largest block of 31 genes is found on Muller E.

Figure 8.—

Rearrangement phylogeny and estimated number of fixed chromosomal rearrangement breaks based on the NGP method (Bhutkar et al. 2007a), using ancestral adjacencies derived from D. melanogaster gene annotation. Each inferred break corresponds to a gene adjacency that existed at the immediate ancestor and was disrupted in that lineage. Paracentric inversions are assumed to be the dominant mechanism resulting in disruption of ancestral adjacencies and in shuffling the order of genes along a chromosome. This analysis includes macro-inversions as well as micro-inversions that shuffle the order and mutual transcriptional orientation between genes. Total counts of fixed breaks from the genus Drosophila root to each extant species (leaf node) are shown to the right of the figure. An. gambiae was used as the outgroup species to resolve ambiguities at the Drosophila genus root, wherever possible. Rearrangement counts for the subgenus Drosophila were found to be lower than those for the subgenus Sophophora. The low-coverage mosaic assembly for D. simulans was excluded from this analysis. See discussion for notes regarding the placement of D. willistoni.

Figure 9.—

Phylogeny of seven Drosophila species with the branch lengths based on inversion distances from pairwise linkage chain analysis. Each set of colored bars indicates the inversion rate for the five Muller elements for each branch. The triangles pointing up indicate that the number of inversions on this branch was greater than expected while the triangles pointing down indicate that the number of inversions was greater than expected. The expected values were determined on the basis of a chi-square test of homogeneity. The inversion rates were scaled to the second-codon position rate of change per 250 sites.

Figure 10.—

Synteny disruption over evolutionary time. (a) Synteny disruption based on divergence estimates from Tamura et al. (2004). The horizontal axis shows total independent evolutionary time between D. melanogaster and a given species (which is twice the time since their last common ancestor). The vertical axis shows the number of inferred synteny blocks with respect to the D. melanogaster gene order. An exponential model fits this data set well. A linear model (with comparable R² value) is also shown but it does not contain the data points as well as the exponential model. (b) Synteny disruption based on divergence estimates from Russo et al. (1995). A linear model fits this data set best. (c) An example of the distribution of synteny block lengths (in kilobases) between D. melanogaster and D. virilis. The vertical axis (log scale) shows the frequency of variously sized syntenic segments (in kilobases on the horizontal axis; log scale). A power law fits this distribution well, suggesting a bias toward a large number of small segments with fewer well-conserved larger segments. See text for a discussion of possible chromosomal breakage models.

Figure 11.—

Distribution of the average size (in number of genes) of synteny blocks across the Muller elements (A–E) of various species based on the D. melanogaster euchromatic gene order. The vertical axis shows the average size of the blocks (log scale). Outside the melanogaster subgroup Muller A shows a trend of being more fragmented than the other arms. Species: D. sec, D. sechellia; D. sim, D. simulans; D. yak, D. yakuba; D. ere, D. erecta; D. ana, D. ananassae; D. pse, D. pseudoobscura; D. per, D. persimilis; D. wil, D. willistoni; D. vir, D. virilis; D. moj, D. mojavensis; D. gri, D. grimshawi.

Figure 12.—

Embryonic gene expression correlation within some of the larger multispecies conserved blocks for each Muller element. Red squares indicate significant positive correlation (P < 0.05) and blue squares indicate significant negative correlation. Block identifiers refer to their tags in the supplemental material. Although some blocks exhibit high levels of positive correlation (like Muller E block 47, which contains the Osiris gene cluster; see discussion), others show different patterns. These plots raise the possibility of islands and networks of positive gene correlation that might select for these genes to be conserved in their order or proximity to each other (see discussion for details). For example, blocks 457 and 413 on Muller E have a major chunk each of positively correlated genes devoid of any negative correlation in these chunks. Further, they show positive correlation with some genes outside these chunks, which might hold the whole block together. Blocks 155 (Muller A) and 280 (Muller B) show patterns where some negatively correlated genes might be trapped within blocks (for example bgm in block 280). We find areas of overlap between our set of cross-species conserved blocks and blocks determined by Spellman and Rubin (2002) in a study based on expression analysis in D. melanogaster. Areas of overlap are shown boxed (in dashed lines). See text for an analysis of the overlap and discussion related to cross-species conservation and expression profiles.