Differential lineage-specific amplification of transposable elements is responsible for genome size variation in Gossypium

Abstract

The DNA content of eukaryotic nuclei (C-value) varies approximately 200,000-fold, but there is only a approximately 20-fold variation in the number of protein-coding genes. Hence, most C-value variation is ascribed to the repetitive fraction, although little is known about the evolutionary dynamics of the specific components that lead to genome size variation. To understand the modes and mechanisms that underlie variation in genome composition, we generated sequence data from whole genome shotgun (WGS) libraries for three representative diploid (n = 13) members of Gossypium that vary in genome size from 880 to 2460 Mb (1C) and from a phylogenetic outgroup, Gossypioides kirkii, with an estimated genome size of 588 Mb. Copy number estimates including all dispersed repetitive sequences indicate that 40%-65% of each genome is composed of transposable elements. Inspection of individual sequence types revealed differential, lineage-specific expansion of various families of transposable elements among the different plant lineages. Copia-like retrotransposable element sequences have differentially accumulated in the Gossypium species with the smallest genome, G. raimondii, while gypsy-like sequences have proliferated in the lineages with larger genomes. Phylogenetic analyses demonstrated a pattern of lineage-specific amplification of particular subfamilies of retrotransposons within each species studied. One particular group of gypsy-like retrotransposon sequences, Gorge3 (Gossypium retrotransposable gypsy-like element), appears to have undergone a massive proliferation in two plant lineages, accounting for a major fraction of genome-size change. Like maize, Gossypium has undergone a threefold increase in genome size due to the accumulation of LTR retrotransposons over the 5-10 Myr since its origin.

Figure 1.

Evolutionary relationships among diploid members of Gossypium. Gossypium is a monophyletic genus composed of ∼50 species that are widely distributed throughout many tropical and subtropical regions. Diploid species have a haploid complement of 13 chromosomes. Gossypium is divided into eight genome groups based on cytogenetic data and level of fertility in interspecific hybrids (Endrizzi et al. 1985). Multiple molecular data sets support the phylogenetic relationships indicated, including the outgroup relationship of Gossypioides kirkii (Wendel and Albert 1992; Seelanan et al. 1997; Small et al. 1998, 1999). Despite conservation of chromosome number among the diploids, genome size varies threefold, from an average of 885 Mb in the New World D-genome species to an average of 2576 Mb in the Australian K-genome species (Hendrix and Stewart 2005).

Figure 2.

Copy number estimates for repetitive sequences in Gossypium. Copy numbers for repetitive sequences recovered in the WGS libraries were estimated as described (see Methods). The majority of repetitive sequences are LTR retrotransposons, particularly in the larger-genome species. In both the A and K genomes, massive amplification of Gorge3 gypsy-like sequences has occurred, contributing predominantly to genome size expansion in these two lineages. In the smallest Gossypium genome, G. raimondii (D genome), and copia-like sequences have proliferated and are primarily responsible for genome size expansion in this lineage. Class II sequences were less abundant and appear to contribute little to genome size evolution in the genus. Tandem repeats are approximately evenly distributed among all four species, with pXP1–80 sequences slightly elevated in G. exiguum (K genome).

Figure 3.

Neighbor-joining analysis of Gossypium gypsy-like Gorge1, 2, and 3 reverse transcriptase sequences. Unrooted Neighbor-joining analysis of 373 Gossypium, 24 Arabidopsis, and 36 Brassica gypsy reverse transcriptase sequences provides support for the three distinct classes of gypsy-like sequences in Gossypium. Gorge1 is similar to Arabidopsis gypsy sequence athila, Gorge2 is similar to maize cinful1, and Gorge3 is similar to del1–46 from Lilium henryi and dea1 from Ananas comosus. Bootstrap values for the deeper nodes are shown.