Hotspots of mutation and breakage in dog and human chromosomes

Abstract

Sequencing of the dog genome allows an investigation of the location-dependent evolutionary processes that occurred since the common ancestor of primates and carnivores, approximately 95 million years ago. We investigated variations in G+C nucleotide fraction and synonymous nucleotide substitution rates (Ks) across dog and human genomes. Our results show that dog genes located either in subtelomeric and pericentromeric regions, or in short synteny blocks, possess significantly elevated G+C fraction and Ks values. Human subtelomeric, but not pericentromeric, genes also exhibit these elevations. We then examined 1.048 Gb of human sequence that is likely not to have been located near a primate telomere at any time since the common ancestor of dog and human. We observed that regions of highest G+C or Ks ("hotspots"; median sizes of 0.5 or 1.3 Mb, respectively) within this sequence were preferentially segregated to dog subtelomeres and pericentromeres during the rearrangements that eventually gave rise to the extant canine karyotype. Our data cannot be accounted for solely on the basis of gradually elevating G+C fractions in subtelomeric regions as a consequence of biased gene conversion. Rather, we propose that high G+C sequences are found preferentially within dog subtelomeres as a direct consequence of chromosomal fission occurring more frequently within regions elevated in G+C.

Figure 1.

Variations in dog and human K_s, and different G+C fractions, as functions of distance (in base pairs) along dog Chromosome 1. These quantities are shown as median values for 10 gene overlapping windows (see Methods). (A) The variation in K_s values (in black) of dog and human orthologs along this chromosome. K_s value hotspots are indicated in green above A. (B) The syntenic locations (see Methods) of human telomeres (in red) on dog Chromosome 1. Short synteny blocks (<4 Mb) are indicated below in light blue. (C) Variations in GC53 or GC4D fractions (as percentages). %GC53 and %GC4D values of dog genes are shown in light blue and red, respectively, whereas %GC53 and %GC4D values of their human orthologs are shown in green and dark blue, respectively. %GC4D hotspots, exceeding the 80th centile for the whole chromosome (see Methods), are marked above C in red. (D) The differences in %GC53 (ΔGC53, dark blue), and in %GC4D (ΔGC4D, green), between dog and human orthologs. Above D, ΔGC4D hotspots (i.e., dog regions elevated in GC4D, with respect to human orthologs; see Methods) are indicated in red, whereas ΔGC4D cold spots (i.e., dog regions suppressed in GC4D, with respect to human orthologs; see Methods) are shown in blue.

Figure 2.

Variations in dog and human K_s, and different G+C fractions, as functions of distance (in base pairs) along human Chromosome 1. These quantities are shown as median values for 10 gene overlapping windows (see Methods). (A) The variation in K_s values (in black) of human and dog orthologs along this chromosome. K_s value hotspots are indicated in green above A. (B) The syntenic locations (see Methods) of dog telomeres (in red) and dog centromeres (in pink) on human Chromosome 1. Short synteny blocks (<4 Mb) are indicated below in light blue. (C) Variations in GC53 or GC4D fractions (as percentages). %GC53 and %GC4D values of human genes are shown in green and dark blue, respectively, whereas %GC53 and %GC4D values of their dog orthologs are shown in light blue and red, respectively. %GC4D hotspots, exceeding the 80th centile for the whole chromosome (see Methods), are marked above C in dark blue. (D) The differences in %GC53 (ΔGC53, green), and in %GC4D (ΔGC4D, dark blue), between dog and human orthologs. Above D, ΔGC4D hotspots (i.e., dog regions elevated in GC4D, with respect to human orthologs; see Methods) are indicated in red, whereas ΔGC4D cold spots (i.e., human regions elevated in GC4D, with respect to dog orthologs; see Methods) are shown in blue.

Figure 3.

Variations in median K_s and GC4D percentage of dog and human single orthologs (set D2) with respect to log distance to a dog or human telomere or centromere. Values were obtained from nonoverlapping bins of 200-250 sequential genes. K_s was calculated between dog-human 1:1 orthologs (see Methods). D is the distance in bases. All plots have been fitted with a first-order exponential, y = A + B exp(X/C), except those plotted with respect to distance from the human centromere, which are fitted with straight lines.

Figure 4.

The “fragile breakage” model for the evolution of the canine karyotype. Schematic models of chromosome breakage within hotspots for metacentric (A) and acrocentric (B) chromosomes, showing how initially interstitial hotspots may be redistributed to subtelomeric and pericentromeric locations. (A) A metacentric Chromosome A breaks at the hotspot (shown in red), thereby forming Chromosome B and fragment C that lacks the original centromere. The hotspot fragment now resides within the subtelomeric regions of B and C. Further breakage within the subtelomeric hotspot of B may give rise to Chromosome D and, through neocentromerization, the microchromosome E. Fragment C may acquire a centromere adjacent to its hotspot. The hotspot, once interstitial in Chromosome A, is now distributed across two subtelomeric, and one pericentromeric, regions. (B) Acrocentric Chromosome I fissures at a hotspot (shown in red), enabling a pericentric inversion that relocates the centromere medially, giving rise to the metacentric Chromosome J. The hotspot fragments now reside in the proximal subtelomeric, and in the quartal pericentromeric, regions of J. A subsequent break at the centromere gives rise to acrocentric Chromosomes K and L. The interstitial hotspot is again redistributed to the subtelomeric and pericentromeric regions.

Figure 5.

Density of dog/human orthologs' K_s values versus distance of the dog ortholog to a dog/human synteny breakpoint. The density color key is shown to the right of the plot. Increasing K_s and gene G+C content at fourfold degenerate positions (data not shown) are each significantly correlated with decreasing distance to a human/dog synteny breakpoint.