Poly(ADP-ribosyl)ation regulates insulator function and intrachromosomal interactions in Drosophila

Abstract

Insulators mediate inter- and intrachromosomal contacts to regulate enhancer-promoter interactions and establish chromosome domains. The mechanisms by which insulator activity can be regulated to orchestrate changes in the function and three-dimensional arrangement of the genome remain elusive. Here, we demonstrate that Drosophila insulator proteins are poly(ADP-ribosyl)ated and that mutation of the poly(ADP-ribose) polymerase (Parp) gene impairs their function. This modification is not essential for DNA occupancy of insulator DNA-binding proteins dCTCF and Su(Hw). However, poly(ADP-ribosyl)ation of K566 in CP190 promotes protein-protein interactions with other insulator proteins, association with the nuclear lamina, and insulator activity in vivo. Consistent with these findings, the nuclear clustering of CP190 complexes is disrupted in Parp mutant cells. Importantly, poly(ADP-ribosyl)ation facilitates intrachromosomal interactions between insulator sites measured by 4C. These data suggest that the role of insulators in organizing the three-dimensional architecture of the genome may be modulated by poly(ADP-ribosyl)ation.

Figure 1. Drosophila insulators undergo PARylation in vivo and in vitro

(A–C) CP190, Su(Hw) and Mod(mdg4)2.2 are PARylated in S2 cells. Cell lysates (Lys) were immunoprecipitated (IP) with either preimmune serum (IgG) or antibodies that recognize different insulator proteins. The various fractions were subjected to western analysis with PAR antibody followed by antibodies to different insulator proteins. Black arrow heads point to the possible PARylated form of Mod(mdg4)2.2 protein. (D) CP190 and dCTCF can be PARylated in vitro. GST, GST-tagged CP190 and GST-tagged dCTCF were PARylated in vitro using biotin-NAD+ as a substrate in the presence or absence of the inhibitor 3-aminobenzamide (3AB). In vitro products were western-blotted with Streptavidin-HRP, followed by dCTCF and CP190 antibodies. (E) The K566 residue within the putative PBZ domain is essential for PARylation of CP190. GST, GST-tagged CP190 and GST-tagged CP190:K566A proteins were PARylated in vitro and western-blotted with streptavidin-HRP and CP190 antibody. Asterisk indicates the location of CP190 in the gel. The sequence of the PBZ domain and the location of the K566A mutation are indicated at the bottom of the panel. (F) CP190:K566A protein is not PARylated in vivo. Lysates from S2 cells transfected with wild-type (Wt) or K566A mutant (KA) CP190-myc constructs were IP with myc antibody and probed with PAR and CP190 antibodies. See also Figure S1.

Figure 2. Mutation of the Parp gene affects gypsy and Fab-8 insulator activity

(A) The level of abdomen pigmentation is inversely correlated to insulator activity at the y² locus. Percentage of flies with different levels of pigmentation in y²; +/+, y²; Parp^CH1/+, y²; CP190^4-1/CP190^H312 and y²; CP190^4-1/CP190^H312-Parp^CH1 lines. Flies were examined for y² expression 1 day after eclosion. Chi-square test: p<0.0001 (+/+ and Parp^CH1/+) and p=0.04 (Parp^CH1/+ and CP190^4-1/CP190^H312-Parp^CH1). (B) The severity of the cut wing margin phenotype correlates with insulator activity at the ct⁶ locus. Percentage of flies with different levels of cut wing margin phenotypes in ct⁶; +/+, ct⁶; Parp^CH1/+, ct⁶; mod(mdg4)^T6 and ct⁶; mod(mdg4)^T6/mod(mdg4)^T6-Parp^CH1 lines. Chi-square test: p<0.0001 between mod(mdg4)^T6 and mod(mdg4)^T6/mod(mdg4)^T6-Parp^CH1. (C) Eyes of male flies of the genotype Fab8^60.39.2/+; +/+, Fab8^60.39.2/+; CP190^H312/+, Fab8^60.39.2/+; Parp^CH1/+, Fab8^60.39.2/+; CTCF^y+6/+ and Fab8^60.39.2/+; CP190^H312Parp^CH1/+. Eye color was examined 1 h after eclosion. (D) Amount of red-eye pigment extracted from the eyes of male files of the respective genotypes. Blue bars indicate the presence of the Fab8^60.39.2/+ transgene. Mean absorbance at OD 485 nm and standard deviation. At least 23 animals from each genotype were assayed. (E) Wild-type but not K566A transgene restores insulator activity at the ct⁶ locus in null CP190^H312/P11 flies. Top: The majority of the flies from transgenic strain Wt²⁵⁸ have a cut wing while most transgenic KA¹²² flies have wing margins that resemble those of hypomorphic CP190^4-1/H312 (hypo) flies. Chi-square test: p<0.001. Bottom: Western blot of lysate from five adult flies of different genotypes. See also Figure S1.

Figure 3. Interaction between insulator proteins and their association with the nuclear lamina are stabilized by PARylation

(A) PARylation stabilizes interactions between CP190 and dCTCF proteins. Lysate (Lys) from control (Ct) and 3AB treated cells was immunoprecipitated (IP) with preimmune serum (IgG) or CP190 antibody (CP190). These fractions were subjected to western blot analysis with CP190 and dCTCF antibodies. Quantification of the relative level of dCTCF protein that was pull-down by CP190 antibody and standard deviation from 6 independent experiments (**: p<0.005). (B) CP190:K566A protein interacts weakly with dCTCF in S2 cells. Lysates from cells transfected with wild-type (Wt) or K566A (KA) construct were IP with myc antibody and probed with CP190 and dCTCF antibodies. Quantification of the relative level of dCTCF protein that was pull-down by myc antibody and standard deviation from 3 independent experiments (**: p<0.008). (C) PARylation promotes the association of insulator proteins with the nuclear lamina. Nucleus, nuclear matrix, and soluble fractions isolated from control and 3AB treated cells were western-blotted with CP190, Su(Hw), Mod(mdg4)2.2, Lamin Dm0 and histone H3 antibodies. Mean band intensities quantified by ImageJ software and standard deviation from at least 5 independent experiments (***: CP190, p=0.0004, n=6; Su(Hw), p=0.00001, n=7 and Mod(mdg4)2.2, p=0.0006, n=5). (D) Formation of CP190 insulator bodies is impaired in Parp⁰³²⁵⁶ mutant larvae. Immunolocalization of CP190 (green) and Lamin Dm0 (red) in diploid nuclei from imaginal wing disc cells with DNA stained by DAPI (blue). Histogram depicting the distribution of the nuclear CP190 staining pattern in OR (wild type) and Parp⁰³²⁵⁶ mutant larvae. See also Figure S2.

Figure 4. A subset of insulator binding sites is regulated by PARylation

(A) Comparison of CP190 ChIP-seq peaks across an 840 kb region of Drosophila chromosome 3R in control (Ct) and 3AB treated S2 cells. Arrow indicates site at which CP190 binding is disrupted by 3AB treatment. (B) Graphs representing the number of Su(Hw), CP190, dCTCF and Mod(mdg4)2.2 binding sites that exhibit greater than 2-fold changes between Ct and 3AB samples (left). Right: Venn diagram of the overlap between different 3AB-downregulated insulator sites. * represents a site where binding of four insulator proteins was reduced by 3AB treatment. (C) Distribution of genomic CP190 binding sites (green) and sites with more than 3-fold reduction in the 3AB treated cells (blue) with respect to TSS. Each interval on the x-axis represents a 200 bp window. (D) Graph representing the distance between adjacent genomic (green), 3AB-downregulated (blue) and randomly-pick (black) CP190 sites. (E) Percentage of independent and aligned insulator binding sites affected by 3AB treatment. ‘Genome-wide’ refers to insulator binding sites that contain consensus motifs and are bound by different insulator proteins. ‘Ct > 3AB’ refers to 3AB-downregulated insulator binding sites. (F) Percentage of genome-wide and 3AB-downregulated insulator sites that are within 2 kb of TAD borders. See also Figure S3

Figure 5. Intra-chromosomal interactions between specific distant CP190 binding sites are regulated by PARylation

(A) Bait A represents an aligned insulator site with dCTCF and Su(Hw) consensus motifs. Left: ChIP signal of four insulator proteins in control cells (top). Relative ChIP-qPCR of CP190 and dCTCF at the bait and the standard deviation from 4 independent experiments (bottom). Right: Graphical depiction of the intra-chromosomal interactions between bait A and twelve CP190 binding sites in control cells. 3AB-responsive interactions are represented by a green line. Middle: ChIP signal of CP190 surrounding bait A in control (CT) and 3AB-treated cells. Graph of relative crosslinking frequency between bait A and the four affected sites in control and 3AB samples and standard deviation from 4 independent experiments. (B) Bait B represents an independent dCTCF site that is bound by CP190 protein, ChIP-qPCR validation and standard deviation from 4 independent experiments (left). Nineteen intra-chromosomal interactions between the bait B fragment and other CP190 binding sites were validated in control cells (middle); six of these interactions were reduced by 3AB treatment with error bar indicating the standard deviation from 4 independent experiments (right). (C) Bait C represents a class of CP190 binding sites devoid of DNA consensus motifs for BEAF-32, Su(Hw) and dCTCF proteins. Left: Validation of CP190 binding at bait C by ChIPqPCR with standard deviation from 7 independent experiments. Eighteen intra-chromosomal interactions between bait C and other CP190 binding sites were validated in control cells (middle), of which two interactions were downregulated by 3AB. Error bars indicate standard deviation from 4 independent experiments (right). *: p<0.05, **: p<0.02, ***: p<0.005. See also Figure S4 and S5

Figure 6. Model of how PARylation regulates insulator-mediated chromosome organization

PARylation at CP190 K566 promotes its interaction with dCTCF. PARylation modulates the binding of insulator proteins within TADs, which in turn affect the intra-chromosomal interactions between distant insulator sites and their association with the nuclear lamina.