Expansion of human centromeric arrays in cells undergoing break-induced replication

Abstract

Human centromeres are located within α-satellite arrays and evolve rapidly, which can lead to individual variation in array lengths. Proposed mechanisms for such alterations in lengths are unequal cross-over between sister chromatids, gene conversion, and break-induced replication. However, the underlying molecular mechanisms responsible for the massive, complex, and homogeneous organization of centromeric arrays have not been experimentally validated. Here, we use droplet digital PCR assays to demonstrate that centromeric arrays can expand and contract within ~20 somatic cell divisions of a cell line. We find that the frequency of array variation among single-cell-derived subclones ranges from a minimum of ~7% to a maximum of ~100%. Further clonal evolution revealed that centromere expansion is favored over contraction. We find that the homologous recombination protein RAD52 and the helicase PIF1 are required for extensive array change, suggesting that centromere sequence evolution can occur via break-induced replication.

Figure 1.. Organization of human centromeres and copy number quantification using ddPCR-based assays.

(A) Schematic of the human centromere. Higher-order repeats (HOR, grey box) comprise tandemly oriented 170–171 bp monomers (colored boxes). Specific HOR copy number (CN) can be quantified based on sequence identities between HORs and polymorphisms present in monomers. (B) Schematic of single-copy assay workflow. Each HOR (red box) or single-copy reference gene on the same chromosome (dark blue box) in a subclone is isolated by restriction enzyme digestion, partitioned into >18,000 droplets, and simultaneously amplified using HOR-specific or single-gene primers (black arrows) in separate reactions. The droplets that contain targets (green peaks) are counted by signal amplitude and the CN is calculated. The HOR CN per array is determined by normalization with single-gene copies (e.g. HOR copies/single-gene copies). (C) Histogram showing HOR CNs of D11Z1, D18Z1, and D6Z1 in the CHM13 cell line either measured by the single-copy assay or the 5S assay. Values represent mean ± SD of three independent measurements. For the 5S assay, the CNs of the HOR and 5S were measured and the HOR CN per 5S CN were determined. Next, the 5S and a single gene located on the same chromosome were measured to calculate the 5S CN per chromosome. Finally, this number is multiplied by the HOR CN per 5S CN to calculate the HOR CN per chromosome.

Figure 2.. Centromere arrays can expand and contract within ~20 somatic cell divisions.

(A) Experimental scheme to quantify HOR CN within somatic cell divisions using single-copy ddPCR-based assay (image created with BioRender.com). (B–E) Box-whisker plots showing the D11Z1, D18Z1, D6Z1, and DXZ1 CNs in U2OS Group1 subclones. In these and subsequent box-whisker plots, each dot indicates a single PCR reaction, which is normalized by the mean of the parental cell (PC) HOR CN (dotted line). Colors indicate technical replicates. Asterisks indicate degree of significance in CN changes between parental cells and subclone pairs determined by Tukey’s HSD test (n=8, Tukey’s HSD, P<0.05). (F) Box-whisker plot showing the D11Z1 CN in K562 Group1 subclones. (n=16, Tukey’s HSD, P>0.05). (G) Cartoon summary of CN changes in U2OS Group1 subclones.

Figure 3.. Expansion of centromere arrays is favored over contraction

(A) Schematic of single cell isolation and D11Z1 CN changes over time. Relative magnitudes of D11Z1 CN changes are indicated by colors. (B–E) Box-whisker plots showing D11Z1 CNs in SC3 Group2 (n=13, Tukey’s HSD, P<0.05), SC4 Group2 (n=12, Tukey’s HSD, P<0.05), SC3.3 Group3 (n=15, Tukey’s HSD, P<0.05), and SC4.4 Group3 (n=12, Tukey’s HSD, P<0.05) subclones. Individual subclones are identified as follows: parental cell name followed by a period and subclone number (e.g. SC3.3).

Figure 4.. RAD52 and PIF1 are required for D11Z1 CN changes

(A) Overview of the break-induced replication (BIR) model. (B) Expression level of RAD52 (top) and PIF1 (bottom) detected by western blot analysis. Arrows indicate WT bands present in the non-target control. Asterisks indicate KD. The band above the asterisks or arrows are non-specific bands. (C–E) Box-whisker plot showing the D11Z1 CN in either NTC (n=16, Tukey’s HSD, P<0.05), RAD52^KO C1 (n=20, Tukey’s HSD, P>0.05), or PIF1 ^KD C2 clones (n=16, Tukey’s HSD, P>0.05).