Copy number variation and evolution in humans and chimpanzees

Abstract

Copy number variants (CNVs) underlie many aspects of human phenotypic diversity and provide the raw material for gene duplication and gene family expansion. However, our understanding of their evolutionary significance remains limited. We performed comparative genomic hybridization on a single human microarray platform to identify CNVs among the genomes of 30 humans and 30 chimpanzees as well as fixed copy number differences between species. We found that human and chimpanzee CNVs occur in orthologous genomic regions far more often than expected by chance and are strongly associated with the presence of highly homologous intrachromosomal segmental duplications. By adapting population genetic analyses for use with copy number data, we identified functional categories of genes that have likely evolved under purifying or positive selection for copy number changes. In particular, duplications and deletions of genes with inflammatory response and cell proliferation functions may have been fixed by positive selection and involved in the adaptive phenotypic differentiation of humans and chimpanzees.

Figure 1.

Whole-genome comparison of human and chimpanzee copy number variation. For each autosome, relative intensity log₂ ratios are superimposed for all 30 chimpanzees compared with Clint (top) and all 30 humans compared with NA10851 (middle). Log₂ ratios for the interspecies comparison of Clint vs. NA10851 are shown at the bottom. Clones in nonvariable regions are depicted as gray circles (log₂ ratios close to 0). Clones reporting copy number change with the CNVfinder algorithm are shown in orange, blue, and green/black for chimpanzees, humans, and the interspecies comparison, respectively. Fixed human–chimpanzee CNDs are indicated by the larger black circles on the interspecies profiles. Large gaps in clone coverage correspond to centromeric regions.

Figure 2.

PCR-based validation of a deletion CNV in chimpanzee. (A) The WGTP clones Chr17tp-10E4 and Chr17tp-7B11 (human chromosome 17q21.32) report a chimpanzee-specific deletion CNV. Based on an alignment of the chimpanzee and human genomes (Karolchik et al. 2003) for this region, the chimpanzee reference sequence (donor: Clint) has a gap of ∼68 kb including the first exon of the B4GALNT2 gene. In the between-species aCGH experiment, we observed a relative loss for these two clones in Clint compared with the human reference individual. (B) Bivariate clustering of Chr17tp-10E4 and Chr17tp-7B11 log₂ ratios. The inferred cluster class of Clint, the chimpanzee reference individual (i.e., log₂ ratio close to 0 for each clone) corresponds to the lowest copy number state among chimpanzees. (C) Results of a PCR-based genotyping assay using a 1.2% agarose gel with ethidium bromide staining. PCR primer positions are depicted in A. Note that primer combination A+C amplifies only when the intervening sequence is deleted. The 31 chimpanzees (including Clint) are in sample numerical order. One human individual is included as positive control. (D) PCR-based copy number genotype estimates and Chr17tp-10E4 log₂ ratio clusters are 100% concordant.

Figure 3.

Validation of CNV loci by fluorescence in situ hybridization. (A) An ∼2-Mb region encompassing the EGFL11 gene (human chromosome region 6q12) is copy-number variable in chimpanzees. The relative intensity log₂ ratios for clones throughout this region form four discrete clusters. In addition, there is a relatively large log₂ ratio difference between the human and chimpanzee reference individuals (ref., reference). Interphase FISH experiments with two labeled BAC probes (RP11-307F22 in green; RP11-13B22 in red) confirmed the presence of an extremely large, tandem, multi-allelic CNV in chimpanzees. The human reference individual NA10851 was found to have two diploid copies (one copy on each chromosome). The chimpanzee log₂ ratio clusters correspond to four (Pt421; 1 + 3), five (Pt238; 2 + 3), six (reference chimpanzee Clint; 3 + 3), and presumably seven copies of this genomic region per diploid cell. (B) The WGTP clones Chr1tp-8H4 and Chr1-32k-1C17 (human chromosome 1q23.3) report CNVs in both humans and chimpanzees. This region includes the FCGR2 and FCGR3 genes, and HSPA6. Fiber FISH experiments with two labeled fosmid probes (G248P87067E2 in green; G248P85846B10 in red) validated the WGTP results and determined absolute copy numbers of the FCGR3 genes, which have been associated with susceptibility to systemic autoimmune diseases (Fanciulli et al. 2007). The human reference individual NA10851 has five diploid FCGR3 copies (3 + 2), while the other humans apparently have either four (e.g., NA19108; 2 + 2) or three (e.g., NA19257; 2 + 1) copies per diploid cell. Chimpanzees have either four (Pt205 and Pt296; 2 + 2) or three (Pt19; 2 + 1) diploid copies of this genomic region.

Figure 4.

Genome architecture of the CCL3/CCL3L locus in humans and chimpanzees. (A) Log₂ ratio distributions of the WGTP clones Chr17tp-10G9 and Chr17tp-6G12 from the human 17q12 locus for humans (top) and chimpanzees (bottom; ref., reference). (B) Sequence annotation of this locus based on the human reference genome (hg18), with locations of the WGTP clones and fosmids used in this study. We compared two human samples and Clint to NA10851 by aCGH with a human oligonucleotide platform covering the 17q12 locus with a median spacing of 50 bp. The high-resolution profiles are concordant with the WGTP results displayed in A: NA18916 shows an increase in CCL3L1 copy number together with a decrease in TBC1D3 copy number compared with NA10851, while NA19108 shows relative copy number gains for both genes. In contrast, no difference in CCL3L1 copy number can be detected between Clint and NA10851. Instead, a high-fold relative copy number loss of the TBC1D3 gene is identified in Clint, thus explaining the CND loss detected with the WGTP platform. (C) Absolute CCL3L1 copy number measurement by fiber-FISH, using probes containing CCL3 (green), CCL3L1 (red), and a DNA segment between these two genes (in white). NA10851 carries a single copy of CCL3L1 per chromosome, while NA18916 carries one copy of CCL3L1 on one chromosome, but two copies on the other chromosome. For Pt205 and Pt296, no DNA fiber shows green and red signals together, suggesting that CCL3 and CCL3L1 are not adjacent genes in chimpanzee. In addition, we observed only single red signals for these two chimpanzee individuals, with no evidence of tandem CCL3L1 duplication. (D) Interphase-FISH with probes containing CCL3 (green) and CCL3L1 (red). In NA10851 cells, the signals corresponding to CCL3 and CCL3L1 cannot be discriminated spatially. In contrast, gaps between the red and green signals can be observed in chimpanzee nuclei, confirming major structural differences in the architecture of the CCL3L1 locus between human and chimpanzee. These results are concordant with the published sequence of chimpanzee chromosome 17 (panTro2 assembly), where a single copy of CCL3L1 is present at 19.41 Mb and CCL3 is mapped at 21.07 Mb.

Figure 5.

Colocalization of CNVs and SDs in the human and chimpanzee genomes. (A) Observed and expected proportion of human-specific CNVs (i.e., human CNVs that do not overlap any chimpanzee CNVs), chimpanzee-specific CNVs, and CNVs that were observed in orthologous regions in both species that overlap human-only, chimpanzee-only, and shared SDs. A given CNV may intersect more than one type of SD. Expected values are based on 10,000 randomized permutations. (B) Ratio of observed (obs) versus expected (exp) number of common CNVs (observed in two or more individuals) overlapping SDs in the human genome, binned by nucleotide sequence similarity between intrachromosomal SD paralogs. For CNVs overlapping multiple SDs, the SD with maximum (similarity x length) is reported. Expected values are based on 1000 randomized permutations. *P < 0.01; **P < 0.001; ***P < 1 × 10⁻⁸.