A physical map for the Amborella trichopoda genome sheds light on the evolution of angiosperm genome structure

Abstract

Background: Recent phylogenetic analyses have identified Amborella trichopoda, an understory tree species endemic to the forests of New Caledonia, as sister to a clade including all other known flowering plant species. The Amborella genome is a unique reference for understanding the evolution of angiosperm genomes because it can serve as an outgroup to root comparative analyses. A physical map, BAC end sequences and sample shotgun sequences provide a first view of the 870 Mbp Amborella genome.

Results: Analysis of Amborella BAC ends sequenced from each contig suggests that the density of long terminal repeat retrotransposons is negatively correlated with that of protein coding genes. Syntenic, presumably ancestral, gene blocks were identified in comparisons of the Amborella BAC contigs and the sequenced Arabidopsis thaliana, Populus trichocarpa, Vitis vinifera and Oryza sativa genomes. Parsimony mapping of the loss of synteny corroborates previous analyses suggesting that the rate of structural change has been more rapid on lineages leading to Arabidopsis and Oryza compared with lineages leading to Populus and Vitis. The gamma paleohexiploidy event identified in the Arabidopsis, Populus and Vitis genomes is shown to have occurred after the divergence of all other known angiosperms from the lineage leading to Amborella.

Conclusions: When placed in the context of a physical map, BAC end sequences representing just 5.4% of the Amborella genome have facilitated reconstruction of gene blocks that existed in the last common ancestor of all flowering plants. The Amborella genome is an invaluable reference for inferences concerning the ancestral angiosperm and subsequent genome evolution.

Figure 1

Maximum likelihood trees for reverse transcriptase genes classified as Copia-type and Gypsy-type LTR and LINE elements. (a) Copia-type; (b) Gypsy-type LTRs; (c) Gypsy-type LINEs. The maximum likelihood trees show rate heterogeneity and no recent expansive radiations (that is, short terminal branches). Reverse transcriptase sequences were mined from BAC end sequence set.

Figure 2

K-mer analyses of Sanger shotgun sequences reveal low frequencies of short repeats in the Amborella genome relative to the sorghum and maize genomes.

Figure 3

Variation in rates of structural evolution evident in parsimony mapping of losses of synteny with 29 gene blocks inferred for the last common ancestor of all extant flowering plant lineages.

Figure 4

Hybridization of three BAC clones in the minimum tiling paths for contigs 1003 and 431 to mitotic squashes (2n = 26) verifies the FPC assemblies. (a-e) Results for contig 1003; (f-j) results for contig 431. Panels (a) and (f) show all three BAC-FISH probes merged; (e,j) DAPI staining; (b,c,d) show each of three BACs (red, green, white) for contig 1003; (g,h,i) show each of three BACs (red, green, white) for contig 431.

Figure 5

LASTZ dot plots comparing BAC contig 1003 syntenic regions in the grape and rice genomes. (a) Grape genome; (b) rice genome.

Figure 6

LASTZ dot plots comparing BAC contig 431 syntenic regions in the grape and rice genomes. (a) Grape genome; (b) rice genome.

Figure 7

Gene trees for auxin-independent growth promoter (AXI1), ceramidase and plant uncoupling mitochondrial protein 1 (PUMP1) gene families. (a) Auxin-independent growth promoter (AXI1); (b) ceramidase; (c) plant uncoupling mitochondrial protein 1 [PUMP1] gene families. The gene trees show divergence of genes on Amborella contig 431 diverging from lineages leading to Vitis γ homeologs mapping to syntenic blocks on chromosomes 6, 8 and 13 (shown in red). Genes sampled from major angiosperm lineages are highlighted.