CNV instability associated with DNA replication dynamics: evidence for replicative mechanisms in CNV mutagenesis

Abstract

Copy number variation (CNV) in the human genome is of vital importance to human health and evolution of our species. However, much of the molecular basis of CNV mutagenesis remains to be elucidated. Considering the DNA replication model of 'fork stalling and template switching' for CNV formation, we hypothesized that replication fork progression could be important for CNV mutagenesis. However, molecular assays of replication fork progression at the genome level are technically challenging. Instead, we conducted an estimation of DNA replication dynamics, as the statistic R, using the readily available data of replication timing. Small R-values can reflect 'slowed' replication, which could result from less fork initiation, reduced fork speed or fork barriers. We generated genome-wide profiles of R in the genomes of human, mouse and Drosophila. Intriguingly, the CNV breakpoints in all three genomes showed significantly biased distributions toward the genomic regions with small R-values, suggesting potential replication stress-induced CNV instability. Notably, among the human CNVs with distinct breakpoint junction characteristics, the homology-mediated and VNTR-mediated CNVs contribute the most to the correlation between CNV instability and the statistic R, consistent with the recent findings in the C. elegans and yeast genomes of repeat-induced DNA replication error and consequent CNV formation. The statistic R may reflect both replication stress and the effect of local genome architecture on fork progression. Our concordant observations suggest an important role for DNA replicative mechanisms in CNV mutagenesis and genome instability.

Figure 1.

Investigation of DNA replication dynamics in human lymphoblastoid cells based on their replication timing profile. (A) A DNA segment on the human chromosome 4 is exemplified for data processing. The red curve depicts the original data of replication timing obtained from a previous study (18). The blue curve depicts the differences in replication timing (|△Timing|) between two boundaries of each chromosomal window across the investigated segment. A large value of |△Timing| indicates that it takes a long time to accomplish DNA replication in the 100 kb segmented window. Therefore, large values of |△Timing| suggest reduced rates of DNA replication, and vice versa. The black square indicates the location of two 100 kb windows that are illustrated in Column B. (B) The mean value of replication timing data is calculated in each 20 kb bin at window boundary. Then △Timing is obtained to be the difference in mean value between two boundary bins for each chromosomal window. Our proposed statistic R = window size (in kb)/|△Timing|.

Figure 2.

Correlation between CNV instability and the statistic R is consistent across species. The chromosomal windows across the human, mouse and Drosophila genomes are, respectively, sorted into 20 proportions based on their R-values. A linear regression fitting was implemented for each plot. (A) The human germline CNVs identified using the CGH method (the correlation coefficient r² = 0.5206). (B) The human germline CNVs identified by the NGS methods (r² = 0.4402). The human DNA replication profile used in (A and B) is based on human lymphoblastoid cells. (C) Distribution of mouse germline CNV breakpoints in the genomic proportions with variable R-values for mouse ESC D3 line (r² = 0.3128). (D) Distribution of the de novo somatic CNV breakpoints in the genomic proportions with variable R-values for Drosophila S2 cell cultures (r² = 0.4197).

Figure 3.

The breakpoint distributions of the human CNVs mediated by different molecular mechanisms. The human germline CNVs obtained from the 1000 Genome Project (39) can be categorized into four groups based on their breakpoint junction characteristics. (A) Homology (r² = 0.2779); (B) VNTR (r² = 0.3875); (C) MEI (r² = 2 × 10⁻⁷); and (D) Non-homology (r² = 0.0301). The human DNA replication profile is based on human lymphoblastoid cells.