Adaptive evolution of centromere proteins in plants and animals

Abstract

Background: Centromeres represent the last frontiers of plant and animal genomics. Although they perform a conserved function in chromosome segregation, centromeres are typically composed of repetitive satellite sequences that are rapidly evolving. The nucleosomes of centromeres are characterized by a special H3-like histone (CenH3), which evolves rapidly and adaptively in Drosophila and Arabidopsis. Most plant, animal and fungal centromeres also bind a large protein, centromere protein C (CENP-C), that is characterized by a single 24 amino-acid motif (CENPC motif).

Results: Whereas we find no evidence that mammalian CenH3 (CENP-A) has been evolving adaptively, mammalian CENP-C proteins contain adaptively evolving regions that overlap with regions of DNA-binding activity. In plants we find that CENP-C proteins have complex duplicated regions, with conserved amino and carboxyl termini that are dissimilar in sequence to their counterparts in animals and fungi. Comparisons of Cenpc genes from Arabidopsis species and from grasses revealed multiple regions that are under positive selection, including duplicated exons in some grasses. In contrast to plants and animals, yeast CENP-C (Mif2p) is under negative selection.

Conclusions: CENP-Cs in all plant and animal lineages examined have regions that are rapidly and adaptively evolving. To explain these remarkable evolutionary features for a single-copy gene that is needed at every mitosis, we propose that CENP-Cs, like some CenH3s, suppress meiotic drive of centromeres during female meiosis. This process can account for the rapid evolution and the complexity of centromeric DNA in plants and animals as compared to fungi.

Figure 1

The rat CENP-A protein. (a) Alignment of predicted CENP-A proteins of mammals. Relative to other mammalian CENP-As, rat CENP-A has a 25 amino-acid insertion that arises from a duplication of the amino terminus, shown as over-lined regions. The boundary between the tail and the histone-fold domains (HFD) is indicated below the alignment, along with the position of Loop 1. (b) Alignment of duplicated regions of the rat Cenpa gene (rat1 and rat2) with Cenpa genes of mouse and Chinese hamster. The region that became duplicated in rat extends from upstream of the start codon to codon 22 in mouse and hamster, and is bounded by a conserved dodecamer repeat. The encoded amino acids are shown above (rat1) or below (rat2) the duplicated sequence.

Figure 2

Sliding-window analysis of K_a/K_s for selected pairs of Cenpc genes. Each point represents the value of K_s, K_a, or K_a/K_s for a 99 nucleotide (33 codon) window plotted against the codon position of the midpoint of the window. K_a/K_s is not defined where K_s = 0. The aligned coding sequence is represented at the top of each graph, with the CENPC motif represented by a filled rectangle; exons are also indicated for the plant sequences. Regions of statistically significant positive selection (black bars) and negative selection (gray bars) are marked. (a) Rat and mouse. The interrupted gray bar indicates that p = 0.06 for this region. (b) Arabidopsis thaliana and Arabidopsis arenosa. (c) Maize (CenpcA) and Sorghum bicolor. (d) Wheat and barley, exons 9p-14.

Figure 3

Comparisons of CENP-C proteins in animals, yeast and plants. The CENPC motif and conserved regions found at the termini of CENP-C proteins are indicated. For pairwise comparisons of protein-coding sequences, regions of positive and negative selection between the species compared are shown. (a) Alignment of animal and fungal CENP-Cs. Mammalian CENP-Cs align throughout their lengths, as do the two Saccharomyces Mif2p proteins, but others align only at conserved regions. Portions of the human CENP-C protein implicated in centromere-targeting (purple bars) and DNA-binding (black bars) are shown at the top. The scale bar at the top marks the length of human CENP-C in amino acids. (b) Alignment of plant CENP-Cs. Within angiosperm families, proteins align throughout their lengths. Between families, weak conservation is found at the amino terminus and strong conservation at the carboxyl terminus. (c) Logos representation of an alignment of the CENPC motif from human; mouse; cow; chicken; Caenorhabditis elegans; budding yeast; Schizosaccharomyces pombe; Physcomitrella patens; maize CenpcA; rice; A. thaliana; black cottonwood, soybean, and tomato.

Figure 4

Alignment of conserved regions of angiosperm CENP-C predicted proteins. (a) Short regions of conservation are encoded in the first six exons of Cenpc genes from five families. The dipeptide SQ (underlined) is relatively frequent in exon 5. (b) Multiple alignment reveals strong conservation in the carboxyl termini of encoded proteins from six families. The CENPC motif is indicated. At, A. thaliana; Mt, barrel medic; Os, rice; Zm, maize CENP-CA; St, potato; SLe, tomato; Bv, beet; Pbt, black cottonwood.

Figure 5

Gene models of selected plant Cenpc genes. Exon/intron structure is conserved across families from exon 1 through the beginning of exon 6, and for the final two exons and introns. Exon sizes are given to the nearest codon where genomic sequence is available to confirm predicted exons. Duplicated exons are indicated by gray shading.

Figure 6

CENP-C exon repeats in the grasses. (a) Alignments of copies of the duplicated exons 9, 10, 11, and 12 from the grass species in this study, excluding pseudogenes, are shown in Logos format. (b) A neighbor-joining phylogram (with gaps excluded) of the exon pairs 9-10 and 11-12 in grass species. A parsimony tree gave essentially the same topology. Dots indicate the locations of inferred duplication events in the tree. Presumed pseudogenes are marked with ψ. (c) Schematic representation of exon duplication events leading to various Cenpc gene structures, and examples of grass species with these structures. Pairs of arrows indicate duplication events; lines terminating in a filled circle indicate loss of an exon pair in derivatives.