Widespread gene conversion in centromere cores

Abstract

Centromeres are the most dynamic regions of the genome, yet they are typified by little or no crossing over, making it difficult to explain the origin of this diversity. To address this question, we developed a novel CENH3 ChIP display method that maps kinetochore footprints over transposon-rich areas of centromere cores. A high level of polymorphism made it possible to map a total of 238 within-centromere markers using maize recombinant inbred lines. Over half of the markers were shown to interact directly with kinetochores (CENH3) by chromatin immunoprecipitation. Although classical crossing over is fully suppressed across CENH3 domains, two gene conversion events (i.e., non-crossover marker exchanges) were identified in a mapping population. A population genetic analysis of 53 diverse inbreds suggests that historical gene conversion is widespread in maize centromeres, occurring at a rate >1x10(-5)/marker/generation. We conclude that gene conversion accelerates centromere evolution by facilitating sequence exchange among chromosomes.

Figure 1. Correspondence between CRM2 marker number and CRM2 FISH intensity.

Metaphase chromosomes from a B73/Mo17 hybrid line (from a single cell). CRM2 LTR and telomeres are shown in green, CentC and the knob 180 bp repeat are shown in red, and chromosomes are shown in blue. The lower panel shows CRM2 FISH signal (in white), and beneath each centromeric region is the total number of CRM2 TD markers recovered from that centromere.

Figure 2. ChIP display.

The image shows CRM2 elements labeled with P³³ on a polyacrylamide gel. The left panel shows results from chromatin immunoprecipitation with controls: pos, B73 nuclei used for the ChIP experiment; sup, supernatant that did not bind to CENH3 antibodies; C, CENH3-bound markers; neg, no antibody control (shows non-specific binding to the sepharose beads used for precipitation). The right panel shows an annotated comparison between sup and C lanes. The chromosomal locations of the bands precipitated are indicated. The dashes next to the S lane denote non-precipitated bands.

Figure 3. The B73_8_ACC165 gene conversion event.

This figure illustrates marker gain as primary data; see also Figure S2 for a visualization of how the data are interpreted. Panels show gel images acquired using fluorescent (FAM) labeling and capillary electrophoresis (images produced by GeneMarker software). IBM10 contains all Mo17 markers from centromere 8 as well as the centromere 8 B73_8_ACC165 marker (B73 markers are labeled in blue and Mo17 markers are labeled in red). IBM11 and IBM12 contain normal Mo17 and B73 centromeres, respectively. Only a subset of the (total 30) markers for centromere 8 is shown; see Figure S2 for the complete list.

Figure 4. CRM2 marker data from a set of diverse inbreds.

Panels A and B together represent the entire data set. Columns show the 53 inbreds scored, while rows show the presence (black) or absence (white) of 75 CRM2 TD markers for the indicated centromeres. The columns containing B73 and Mo17 reference data are highlighted in grey. For centromere 2, only sequence-confirmed data are shown, whereas all other data were interpreted with a presumed false positive rate of 1.8%.

Figure 5. Linkage disequilibrium in centromere 2.

(A) Pairwise LD plotted against distance, fit to a decay function , using a value of ρ = 8.81×10⁻⁷. Inset shows the decay over the first 5 kb (in black) and the same function fit using the genome-wide median of ρ (in grey) . (B) Heatmap of pairwise LD. Lighter colors show higher LD. The black box demarcates three markers that show high LD and flank the only cluster of CentC repeats on this centromere (the 180 kb region between positions 89.88 and 90.06 Mb on the physical map).

Figure 6. Haplotype estimation of gene conversion.

Shown is the expected number of haplotypes observed under varying levels of gene conversion (γ) from coalescent simulations of a maize domestication bottleneck. The solid line indicates the mean number of haplotypes, and the shaded region encloses the empirical 95% confidence intervals. Horizontal dotted lines represent the number of haplotypes observed from the centromeres indicated (m is the number of markers in that centromere). The most probable gene conversion rates occur where the dotted lines intersect with the solid lines. The last panel shows the outcome if all centromere 8 data are considered together (from both B73 and Mo17, such that m = 19).