Recursive Paleohexaploidization Shaped the Durian Genome

Abstract

The durian (Durio zibethinus) genome has recently become available, and analysis of this genome reveals two paleopolyploidization events previously inferred as shared with cotton (Gossypium spp.). Here, we reanalyzed the durian genome in comparison with other well-characterized genomes. We found that durian and cotton were actually affected by different polyploidization events: hexaploidization in durian ∼19-21 million years ago (mya) and decaploidization in cotton ∼13-14 mya. Previous interpretations of shared polyploidization events may have resulted from the elevated evolutionary rates in cotton genes due to the decaploidization and insufficient consideration of the complexity of plant genomes. The decaploidization elevated evolutionary rates of cotton genes by ∼64% compared to durian and explained a previous ∼4-fold over dating of the event. In contrast, the hexaploidization in durian did not prominently elevate gene evolutionary rates, likely due to its long generation time. Moreover, divergent evolutionary rates probably explain 98.4% of reconstructed phylogenetic trees of homologous genes being incongruent with expected topology. The findings provide further insight into the roles played by polypoidization in the evolution of genomes and genes, and they suggest revisiting existing reconstructed phylogenetic trees.

Figure 1.

Original and corrected synonymous nucleotide substitutions (Ks) among colinear genes. Continuous lines are used to show Ks distribution in a genome, and dashed lines are among genomes. A, Distributions fitted by using original Ks values; B, Inferred means; C, Distributions fitted by using corrected Ks values; D, Inferred evolutionary dates.

Figure 2.

Species phylogeny and polyploidization occurrence models using haploid reference genomes. For example, in Subfig. C, the triplicated cacao paralogs, C1, C2, C3, would each have one grape ortholog and two durian orthologs. Nonorthology homologs between genomes were defined as outparalogs. Between cacao (or grape) and durian, this results in a orthology ratio 1:2 and outparalogy ratio 1:4. A, Species phylogeny; B, Assuming no durian-specific event; C, Assuming a durian-specific tetraploidization; D, Assuming a durian-specific paleohexaploidization; E, If cotton is added to the phylogeny; F, Assuming a durian-specific paleohexaploidization and a Gossypium-specific paleodecaploidization.

Figure 3.

Examples of homologous gene dotplots among durian, cotton, grape, and cacao. Durian scaffold numbers and grape, cacao, and cotton chromosome numbers were shown on the tops and sides of plots, segment regions showed in megabyte (Mbp). Best-hit genes are shown in red dots, secondary hits with blue dots, and others in gray. Arrows show complement correspondence produced by chromosome breakages during evolution. A, Grape vs. durian; B, Cacao vs. durian; C, Cotton vs. durian.

Figure 4.

Alignment of durian, cotton, cacao, and grape genomes. The alignment was constructed by using inferred colinear genes among genomes with the grape genome as reference. The grape chromosomes form the innermost circle, and their paralogous genes in colinearity are linked curves. A grape chromosome region has 1, 5, and 3 orthologous regions in cacao, cotton, and durian genomes, respectively. The homologous regions form the other circles, displayed by short lines to show colinear genes. The short lines in grape chromosomes were colored as the 7 core-eudicot-common ancestral chromosomes inferred previously (Jaillon et al., 2007). The short lines forming chromosomes in other genomes were colored as to their respective source chromosome numbers. The color scheme is shown at the bottom. D, durian; G, cotton; V, grape .

Figure 5.

Alignment of homologous regions from durian, cotton, cacao, and grape genomes. A slice of genome-wide alignment shown in Fig. 4 is shown here in detail. This displays alignment in homologous local regions from considered genomes.

Figure 6.

Types of reconstructed phylogeny of homologous genes. Each reconstructed tree has one grape gene as the outgroup and its cacao ortholog. The reconstructed trees are divided into four types based on the location of the cacao gene or on genes with which plant species it grouped on the phylogenetic tree. A, The tree is of expected phylogeny; B, The cacao gene is grouped with durian homologs; C, The cacao genes are grouped with cotton genes; D, The cacao gene is grouped with but not outgroup to cotton and durian genes.