Human imprinted chromosomal regions are historical hot-spots of recombination

Abstract

Human recombination rates vary along the chromosomes as well as between the two sexes. There is growing evidence that epigenetic factors may have an important influence on recombination rates, as well as on crossover position. Using both public database analysis and wet-bench approaches, we revisited the relationship between increased rates of meiotic recombination and genome imprinting. We constructed metric linkage disequilibrium (LD) maps for all human chromosomal regions known to contain one or more imprinted genes. We show that imprinted regions contain significantly more LD units (LDU) and have significantly more haplotype blocks of smaller sizes than flanking nonimprinted regions. There is also an excess of hot-spots of recombination at imprinted regions, and this is likely to do with the presence of imprinted genes, per se. These findings indicate that imprinted chromosomal regions are historical "hot-spots" of recombination. We also demonstrate, by direct segregation analysis at the 11p15.5 imprinted region, that there is remarkable agreement between sites of meiotic recombination and steps in LD maps. Although the increase in LDU/Megabase at imprinted regions is not associated with any significant enrichment for any particular sequence class, major sequence determinants of recombination rates seem to differ between imprinted and control regions. Interestingly, fine-mapping of recombination events within the most male meiosis-specific recombination hot-spot of Chromosome 11p15.5 indicates that many events may occur within or directly adjacent to regions that are differentially methylated in somatic cells. Taken together, these findings support the involvement of a combination of specific DNA sequences and epigenetic factors as major determinants of hot-spots of recombination at imprinted chromosomal regions.

Figure 1. LD Analysis at Human Chromosomal Regions Containing Imprinted Genes

(A) Comparison of LDU values at imprinted versus control regions. Each LDU/Mb value obtained for a given imprinted bin was plotted against the LDU/Mb value of the corresponding control region. Note that most of the imprinted regions reach higher LDU values compared with their corresponding control regions (dotted line with slope 1 corresponds to virtual positions in cases with equal LDU/Mb values at imprinted and control regions). (B) The number of haplotype blocks/Mb is higher at imprinted genes compared with their corresponding control regions (dotted line indicates equal values). (C) The mean sizes of haplotype blocks are significantly smaller at imprinted regions versus flanking control regions (dotted line indicates equal sizes). (D) There is a significant excess in the number of hot-spots of recombination at imprinted versus control regions (see text), and the total length of the hot-spots appears greater in imprinted regions than in control regions. Each value corresponds to the total length of all hot-spots of recombination (in base pairs [bp]) for a given imprinted or control region.

Figure 2. Distribution of Recombination Events at Human 11p15.5 Imprinted Cluster

Positions of markers used for mapping recombinants in this region are indicated in Mb from the telomeric end (Tel) of the short arm. Imprinted genes are shown on the left side of the figure. Arrows correspond to direction and parental-specific origin of transcription: blue are paternally transcribed genes, red are maternally expressed genes, and black are genes with biallelic expression or unknown imprinting status. The two known germline imprints at this locus are shown by colored oval shapes on the left side of the figure: the blue oval corresponds to the paternally methylated IGF2/H19 DMR and the red oval corresponds to the maternally methylated KCNQ1OT1 DMR. Each vertical bar on the right side of the figure corresponds to a meiotic recombination event, delimitated by the nearest informative markers: Labeled in blue are crossovers in paternal meiosis, and labeled in red are recombinations in maternal meiosis. An asterisk (*) represents an unidentified polymorphism found at MUC5B locus, and double asterisks (**) indicate an unidentified TaqI polymorphism found at TH locus (genotypes available through CEPH database—see Materials and Methods).

Figure 3. LD Analysis at Human 11p15.5 Imprinted Cluster

(A) Population-specific metric LD map for about 1-Mb region containing imprinted genes at human 11p15.5 chromosome. Positions along the chromosome are shown in bp on the x-axis. Straight lines are representing the genome-wide slopes (LDU/Mb) corresponding to each population, as extrapolated from De La Vega et al. [16]. Note that LD extends less far in the region containing imprinted genes compared with a region of similar length from the rest of genome, in agreement with the interpretation of higher recombination in these areas (i.e., breakdown of LD has been converted to implied recombination rate and rendered graphically as red rectangular “hot spots”). Location of CpG islands in the region are depicted as shown in MapView; dark blue represents CpG islands larger than 500 bp, and light blue represents CpG islands over 200 bp. For both categories, G + C content is higher than 50% and the observed CpG/expected CpG content is higher than 0.6. The two black arrows correspond to the regions containing the primary germline imprints at H19/IGF2 DMR (left arrow) and KCNQ1OT1 DMR (right arrow), respectively. Both are located at regions exhibiting steps of LD and recombination hot-spots and are zoomed-in in (B) and (C). The red open arrows correspond to smaller steps, which are variable between populations and do not correspond with any hot-spot of recombination. (B) The metric LD map for the region containing H19/IGF2 DMR using data from the four populations (HapMap) and the set of CEPH individuals analyzed in this study. The three horizontal bars correspond to recombinants mapped at this region, one in maternal meiosis (red) and two in paternal meioses (blue). The blue oval shape corresponds to the H19/IGF2 DMR. (C) The metric LD map for the region containing KCNQ1OT1 DMR using data from the four populations (HapMap). Two recombinants (horizontal blue bars) were mapped at this region in paternal meioses. The red oval shape corresponds to the KCNQ1OT1 DMR.

Figure 4. Pairwise LD Analysis and Pairwise FGT at the H19/IGF2 DMR

(A) Pairwise LD test between ten SNPs covering a 31-kb region containing H19/IGF2 DMR shows a major breakdown of LD which corresponds to the LDU step shown in Figure 3B. Intensity of LD is coded in colors as shown. (B) Pairwise FGT between the same ten SNPs. A “1” indicates recombination between that pair of loci (all four gametes) and “0” indicates only three types of gametes (recombination between the two loci is uncertain). Considering that a historical recombination would break the haplotype inside of which it appeared, at least eight haplotype blocks could be identified. The ten markers used for both analyses in Figure 4A and 4B are the same as depicted in Figure 3B (CEPH)