Ancient polyploidization predating divergence of the cereals, and its consequences for comparative genomics

Abstract

Integration of structural genomic data from a largely assembled rice genome sequence, with phylogenetic analysis of sequence samples for many other taxa, suggests that a polyploidization event occurred approximately 70 million years ago, before the divergence of the major cereals from one another but after the divergence of the Poales from the Liliales and Zingiberales. Ancient polyploidization and subsequent "diploidization" (loss) of many duplicated gene copies has thus shaped the genomes of all Poaceae cereal, forage, and biomass crops. The Poaceae appear to have evolved as separate lineages for approximately 50 million years, or two-thirds of the time since the duplication event. Chromosomes that are predicted to be homoeologs resulting from this ancient duplication event account for a disproportionate share of incongruent loci found by comparison of the rice sequence to a detailed sorghum sequence-tagged site-based genetic map. Differential gene loss during diploidization may have contributed many of these incongruities. Such predicted homoeologs also account for a disproportionate share of duplicated sorghum loci, further supporting the hypothesis that the polyploidization event was common to sorghum and rice. Comparative gene orders along paleo-homoeologous chromosomal segments provide a means to make phylogenetic inferences about chromosome structural rearrangements that differentiate among the grasses. Superimposition of the timing of major duplication events on taxonomic relationships leads to improved understanding of comparative gene orders, enhancing the value of data from botanical models for crop improvement and for further exploration of genomic biodiversity. Additional ancient duplication events probably remain to be discovered in other angiosperm lineages.

Fig. 1.

Arrangement of duplicated protein-encoding genes in Oryza. Both X and Y axes represent 45,174 genes in their chromosomal order (based on a 7/23/2003 assembly from www.tigr.org). The single best-matching gene pairs (identified as described in Materials and Methods) are plotted and color-coded to indicate same (red) or opposite (green) transcriptional orientations. The lower left area shows only the subset of best-matching gene pairs for which synonymous substitution rates range from 0.75 to 0.95. Duplicated regions are highlighted in blue along the y axis. Red dots indicate locations of centromeres.

Fig. 2.

Patterns of colinearity between sorghum and rice. The x axis represents 45,174 rice genes in their chromosomal order, and the y axis represents 2,509 loci in their recombinational arrangement along a sorghum STS-based genetic map (1). Rice chromosomes (1-12) and sorghum linkage groups (A-J) are arranged consecutively, and labeled at top and left, respectively. Red circles represent inferred locations of sorghum centromeres (1). Each dot represents a best match (≤1e-06) between a sorghum STS and a rice gene. The total number of probes mapping to each intersection of sorghum and rice chromosomes is shown at upper left in each cell. Horizontal subdivisions of sorghum linkage groups B, C, and F delineate locations at which sorghum-rice syntenic relationships change, consistent with subdivisions of the counts of incongruent loci shown in Table 2.

Fig. 3.

An early phylogenetic tree of genomic duplications for the angiosperms. By integrating data described herein and elsewhere (7), ancient duplications in the monocots and dicots, respectively, are superimposed on a partial angiosperm phylogenetic tree that also represents well established recent duplications in several lineages. Open circles indicate possible chromosomal duplication or polyploid formation events. Question marks indicate (i) the need for additional data to support tentative indications of a polyploidization event in the Gossypium lineage (J. Rong, J.E.B., and A.H.P., unpublished data) and (ii) uncertainty about the dating of the γ event (7). Gene tree analyses (see text and Table 1) support largely “one-to-one” correspondence of the rice chromosomes to those of other diploid cereals, but suggest the need for “one-to-two (or more)” comparisons to more distant lineages. More recent duplications and/or polyploid formation within many lineages further complicate comparative genomics. Branch lengths along the y axis approximate divergence times cited (7) or Ks data reported herein, converted to MY (millions of years) by using the average of current molecular clocks (18, 19).