Concerted copy number variation balances ribosomal DNA dosage in human and mouse genomes

Abstract

Tandemly repeated ribosomal DNA (rDNA) arrays are among the most evolutionary dynamic loci of eukaryotic genomes. The loci code for essential cellular components, yet exhibit extensive copy number (CN) variation within and between species. CN might be partly determined by the requirement of dosage balance between the 5S and 45S rDNA arrays. The arrays are nonhomologous, physically unlinked in mammals, and encode functionally interdependent RNA components of the ribosome. Here we show that the 5S and 45S rDNA arrays exhibit concerted CN variation (cCNV). Despite 5S and 45S rDNA elements residing on different chromosomes and lacking sequence similarity, cCNV between these loci is strong, evolutionarily conserved in humans and mice, and manifested across individual genotypes in natural populations and pedigrees. Finally, we observe that bisphenol A induces rapid and parallel modulation of 5S and 45S rDNA CN. Our observations reveal a novel mode of genome variation, indicate that natural selection contributed to the evolution and conservation of cCNV, and support the hypothesis that 5S CN is partly determined by the requirement of dosage balance with the 45S rDNA array. We suggest that human disease variation might be traced to disrupted rDNA dosage balance in the genome.

Keywords: bisphenol A; concerted evolution; gene dosage balance; nucleolus; ribosome.

Fig. 1.

cCNV between 5S and 45S rDNA loci residing in different chromosomes. Genomic location of the 5S and 45S rDNA arrays in human (A) and mouse (B). Gray bars represent chromosomes, and black rectangles represent the approximate chromosomal location of the 5S and 45S rDNA loci. 5S (y axis) and 45S (x axis) rDNA CN is highly correlated in human (C) and mouse (D). Gray and black points represent CEU and YRI human populations, and laboratory/inbred and wild mouse isolates, respectively. Spearman correlations reflecting combined populations are reported. The 45S rDNA CN is the average CN of selected segments across the 18S, 5.8S, and 28S components. Correlations are consistent when stratified by population and sex (Table S1).

Fig. 2.

cCNV of rDNA loci is stronger between functionally linked components of the ribosome. (A) Schematic diagram of the ribosome. The large 60S subunit is shown in red, and the small 40S subunit is shown in blue. (B) The 5S, 5.8S, and 28S rRNA molecules assemble with the large subunit, and the 18S rRNA assembles with the small subunit of the ribosome. 5S rDNA CN is more strongly associated with rRNA molecules that comprise the large ribosomal subunit in human (C) and mouse (D). 5S rDNA CN is shown on the y axis, and 18S, 5.8S, and 28S rDNA CNs are depicted on the x axis. Spearman correlations of combined populations are reported.

Fig. 3.

cCNV between 5S and 45S loci is manifested in human pedigrees. The pedigree and association between the 5S (y axis) and 5.8S (x axis) is reported for each family (A–C). Estimates of CN were divided by the value of the individual with the smallest CN array within each family and component (relative CN). Generalized linear models indicate significant associations between 5S and 5.8S elements (P < 0.01; n = 29). The gray shaded area represents the 95% CI for the linear coefficients within each family.

Fig. 4.

Rapid environmental induction of cCNV between the 5S and 45S rDNA arrays. CN of rDNA loci in human cell culture after 24 h growth in control and BPA-containing media. CN of each rDNA component was estimated with qPCR (n = 3; biological replicates). CNs were normalized, with CN of each component set to 1 in the control media (relative CN). Black bars in each graph represent the SD. The CN of the 45S (A–C) and 5S (D) rDNA contracted upon treatment with BPA. Chromosomal location of 5S rDNA array and the TCEB3 and NBPF gene arrays (E). CN of TCEB3 and NBPF, which are CN variable tandemly repeated arrays located on chromosome 1, remained stable upon BPA treatment (F and G).