Nonallelic transvection of multiple imprinted loci is organized by the H19 imprinting control region during germline development

Abstract

Recent observations highlight that the mammalian genome extensively communicates with itself via long-range chromatin interactions. The causal link between such chromatin cross-talk and epigenetic states is, however, poorly understood. We identify here a network of physically juxtaposed regions from the entire genome with the common denominator of being genomically imprinted. Moreover, CTCF-binding sites within the H19 imprinting control region (ICR) not only determine the physical proximity among imprinted domains, but also transvect allele-specific epigenetic states, identified by replication timing patterns, to interacting, nonallelic imprinted regions during germline development. We conclude that one locus can directly or indirectly pleiotropically influence epigenetic states of multiple regions on other chromosomes with which it interacts.

Figure 1.

4C analysis of a genome-wide chromosomal network impinging in the H19 ICR. (A) Reprogramming of interactions during in vitro differentiation of ES cells. The scatter plots compare intra- and interchromosomal (cis and trans) interactions between ES and EB cells (P = 2.9e-15 for the difference in cis/trans interactions in ES and EB cells). (B) Difference in the location of the H19 ICR in relation to its chromosomal territory in ES cells and EBs. Chromosome 7 territory is shown in green, and Igf2/H19 is shown in red. Bar, 1 μm. (C) Significant overrepresentation of imprinted genes (out of 107 total) within 500 kb flanking the interactors (FPR = 6.5e-03, 10⁴ randomizations). (D) Interactions at imprinted domains in ES cells and EBs identified at FDR = 0.05. Black bars represent the imprinted genes in the domain, while blue and red bars represent the location of interactors in ES and EB cells, respectively. Each domain is named as the first and last gene in the imprinted cluster shown. Magnified chromosomal regions are ∼3 Mb (Atp10a-Frat3), ∼2 Mb (H19-Dhrc7), and ∼1Mb (Dlk1-Dio3) in length. A detailed redundant list of interactors proximal to imprinted genes is given in Supplemental Table 4.

Figure 2.

Quantitative in-situ analyses of trans chromosomal interactions. (A) Examples of nuclear staining of colocalized imprinted genes. Igf2/H19 is shown in blue, interactors are shown in red/green, and DAPI is shown in gray. All Z planes are merged together in these images; single-plane images are shown in Supplemental Fig. 6. Bar, 1 μm. (B) Frequencies of close proximity among imprinted loci. The dashed line shows the average interaction frequency for noninteracting controls. (C) Chromosomal network of imprinted loci. Length, thickness, and color gradient of each edge indicate corresponding frequency of close proximity.

Figure 3.

Dynamics of cis and trans interactions. (A) Preferred sites of cis and trans interactions as revealed by chromosome painting experiments. Chromosome 7 territory is shown in blue, Igf2/H19 is shown in green, cis/trans interactors are shown in red, and DAPI is shown in gray. (B) Frequency of cis (Atp10a in green) and trans (Dlk1 and Copg2 in red) interactions. (C) Date (pairwise) versus party (simultaneous) interactions. Igf2/H19 is shown in blue, while other trans imprinted loci are shown in red (Dlk1, Copg2, and Slc22a2) or green (Ins1, Gatm, and Htr2a). Bar, 1 μm.

Figure 4.

Trans-regulation of epigenetic states by H19 ICR. (A) Frequencies of close proximity among imprinted domains in mouse spermatogonia (Spg), spermatocytes (Spc), and round spermatids (Spt). Cells were isolated from mice carrying a paternally (SD7 × 142*) or maternally (142* × SD7) inherited mutated H19 ICR allele defective in CTCF binding. The dotted line shows random control taken from noninteracting regions in our 4C screen. (B,C) Frequencies of close proximity in the neonatal liver of mouse crosses and ES cells (wild-type R1 and 142 R1 cells harboring a mutation in CTCF-binding sites within the maternally inherited H19 ICR allele; note that the 142 allele behaves the same as 142*, as illustrated in Supplemental Fig. 7). Dotted lines represents the noninteracting control. (D) Asynchronous replication timing pattern (percent SD) for spermatogonia cells from mouse crosses as above. Control is shown in open boxes and average values are shown in horizontal bars. (E) Replication timing pattern in PCNA-positive nuclei of neonatal liver. SS indicates two unreplicated/unseparated alleles, DD indicates two replicated and separated alleles, and SD represents one of each. (F) Asynchronous replication timing pattern in wild-type R1 and 142 R1 ES cells. More than 200 nuclei were counted for each instance in A–F. Genes analyzed are H19, Ins1, Copg2, Gatm, Htr2a, and Dlk1 in D, and H19, Ins1, Impact, Gatm, Htr2a, and Dlk1 in E and F.