Fig. 7: L1 repeats promote phase separation of HP1α.

a Metagene analysis showing ChIRP DNA-seq signals for L1 and Malat1 RNA, and ChIP-seq signals for HP1α and H3K9me3 in L1-rich compartments. ChIRP-seq and ChIP-seq reads densities were normalized to the input DNA. The shadow around each line represents standard error. b Genome browser view of sequencing tracks in two regions in mouse chr17: 18–60 Mb (left) and chr19: 3–12 Mb (right). The first five rows show the genomic density of B1 repeats, B1-related sequencing data (ChIRP-seq signals of Malat1, ChIP-seq signals of Pol II, and H3K27ac), and RNA-seq in mESCs. The lower tracks show the genomic density of L1 repeats (highlighted by beige shading) and L1-related sequencing data, including ChIRP-seq signals of L1 RNA, ChIP-seq signals of HP1α and H3K9me3. Refseq gene annotations are also included. c Boxplot showing the ChIRP-seq signal of L1 and Malat1 RNA in B1 and L1-rich compartments. Y-axis showing the fold change of raw ChIRP-seq signal to input DNA. P values were calculated with two-tailed Student’s t-test and are shown in each plot. d RT-qPCR analysis of relative enrichments of various transcripts in chromatin fractions of mESCs separated by sucrose gradient centrifugation. Enrichment of RNA in each fraction was normalized to the input nuclear extracts (top). Data are shown as means ± SD (n = two biological replicates). Western blot analysis confirms the effectiveness of chromatin fractionation (bottom). e Biotinylated L1 RNA pulls down recombinant HP1α protein. Reactions without addition of L1 RNA were served as mock control. f Representative images of droplet formation at different concentrations of HP1α protein and L1 RNA. Concentrations of HP1α and RNA are indicated at the top and left of the images, respectively. Scale bars, 10 μm. Data are representative of three independent experiments. g The model. First, the intrinsic self-assembly property of L1 and B1/Alu repeats provides numerous nucleation points to seed the formation of nuclear subdomains. Repetitive DNA sequences also serve as anchor sites for transcription machinery, regulatory proteins and RNAs. Second, the embedded structural information in DNA repeats may be translated by their RNA transcripts, together with interacting DNA- and/or RNA-binding proteins. Molecular crowding generated by interactions among DNA sequences, RNAs, and proteins in subnuclear domains seeded by individual clusters of L1 or B1/Alu, may drive the aggregation of compartments containing the same repeat type through a phase-separation mechanism, which consequently folds the genome. Third, the nuclear segregation of L1-rich compartments and B1/Alu-rich compartments may be further reinforced by attaching their DNA sequences to subnuclear structures such as nuclear speckles and the nucleolus, respectively, which serve as scaffolds to stabilize the nuclear architecture.