Tissue-specific regulation of chromatin insulator function

Abstract

Chromatin insulators organize the genome into distinct transcriptional domains and contribute to cell type-specific chromatin organization. However, factors regulating tissue-specific insulator function have not yet been discovered. Here we identify the RNA recognition motif-containing protein Shep as a direct interactor of two individual components of the gypsy insulator complex in Drosophila. Mutation of shep improves gypsy-dependent enhancer blocking, indicating a role as a negative regulator of insulator activity. Unlike ubiquitously expressed core gypsy insulator proteins, Shep is highly expressed in the central nervous system (CNS) with lower expression in other tissues. We developed a novel, quantitative tissue-specific barrier assay to demonstrate that Shep functions as a negative regulator of insulator activity in the CNS but not in muscle tissue. Additionally, mutation of shep alters insulator complex nuclear localization in the CNS but has no effect in other tissues. Consistent with negative regulatory activity, ChIP-seq analysis of Shep in a CNS-derived cell line indicates substantial genome-wide colocalization with a single gypsy insulator component but limited overlap with intact insulator complexes. Taken together, these data reveal a novel, tissue-specific mode of regulation of a chromatin insulator.

Figure 1. Shep associates directly with gypsy insulator complexes.

(A) Diagram of Shep protein isoforms. RRMs (blue) and alternative amino acid stretches (not to scale, orange) are shown. Regions of Shep utilized for antibody production or contained in the yeast two-hybrid clone, which corresponds to exons present in isoform E, are indicated. (B) Coomassie staining of recombinant GST fusion proteins used for binding reactions in (C). Protein marker is run in lane 1. (C) Interaction of purified, soluble His-Mod(mdg4)2.2 (lane 1, 4.5% input) with immobilized GST (lane 2), GST-Su(Hw) (lane 3) or GST-Shep isoforms (lanes 4–6). Binding of His-Mod(mdg4)2.2 to GST-fusion proteins was detected by Western blotting. (D) Coomassie staining of recombinant GST fusion proteins used for binding reactions in (E). (E) Interaction of purified, soluble His-Su(Hw) (lane 1, 6.3% input) with immobilized GST (lane 2), GST-Mod(mdg4)2.2 (lane 3) or GST-Shep isoforms (lanes 4–6). Binding of His-Su(Hw) to GST-fusion proteins was detected by Western blotting.

Figure 2. Coimmunoprecipitation of gypsy insulator proteins with Shep isoforms.

(A) Identification of Shep isoforms in vivo. Western blotting for Shep from larval extracts that are wildtype (lane 1), expressing Act5C::Gal4 driving single copy UAS-shep dsRNA (lane 2), expressing Act5C::Gal4 driving single copy UAS-shep C and E (lane 3), or containing a P-element insertion that disrupts the coding region of isoform A (lane 4). Pep is shown as a loading control. (B) Coimmunoprecipitation of gypsy insulator proteins with Shep. Embryo nuclear extracts (lane 1) were immunoprecipitated (IP) with either Pre-Immune (Pre Im; lanes 2 and 4) or α-Shep (lanes 3 and 5) serum. Shep, Mod(mdg4)2.2, Su(Hw), and CP190 were detected in nuclear extracts (Nuc Ext), supernatants (Sup) (lanes 2–3) and IPs (lanes 4–5) by Western blotting. Approximately 0.02% CP190, 0.02% Su(Hw), and 0.1% Mod(mdg4)2.2 of total were recovered in the IP.

Figure 3. Identification of Shep loss-of-function alleles.

(A) Diagram of lesions in the shep locus. P-element insertion sites are denoted below the gene model, and genomic deficiencies are indicated above the gene model. Hatched lines indicate that deletions extend beyond the shep locus. See Table 1 for P-element details. (B) Western blotting of larval extracts of mod(mdg4) ⁺ and homozygous shep P-element insertion larval extracts for Shep, Mod(mdg4)2.2, Su(Hw), and CP190 in the mod(mdg4) ⁺ background. Lane numbers of gel are indicated. (C) Western blotting for CP190, Su(Hw), Pep, and Shep in larval extracts of mod(mdg4) ⁺, mod(mdg4)^u1, and heterozygous or transheterozygous shep deficiencies in the mod(mdg4)^u1 background.

Figure 4. Loss-of-function shep alleles disrupt gypsy insulator activity at ct⁶.

(A) Effects of shep mutations on the ct⁶ phenotype. All flies are homozygous for mod(mdg4)^u1. At the shep locus, flies are wildtype (shep⁺), harbor a heterozygous deficiency, or contain a homozygous P-element insertion as indicated. Percent of population scored on a scale of 0–4 is indicated for each genotype. 0, no notching; 1, slight notching in one wing; 2, slight notching in both wings; 3, pronounced notching in hinge distal wing margin; 4, severe notching in both hinge proximal and distal margins. Asterisks denote P-element insertions showing extensive synthetic lethal interaction with mod(mdg4)^u1 for which rare escapers were scored (49≤n≤180 for all genotypes). (B) Hemizygous alleles of shep affect ct⁶. Phenotypes of ct⁶ of shep^BG00836 and shep^d05714 mutations transheterozygous with Df(3L)Exel6104. All flies are homozygous for mod(mdg4)^u1. Flies were scored in parallel with those in (A) (85≤n≤180). (C) Male abdominal pigmentation due to y² expression is unchanged in mod(mdg4)^u1 compared to shep^BG00836, mod(mdg4)^u1 flies.

Figure 5. Shep negatively regulates gypsy activity in the CNS.

(A) Confocal imaging of Shep distribution in stage 14 wildtype Oregon R embryo by indirect immunofluorescence using guinea pig α-Shep (green) and mouse α-Elav (red) antibodies detected by α-guinea pig Alexa-488 and α-mouse Alexa-594 secondary antibodies. DAPI staining (blue) is also shown in the merged image. A, anterior; P, posterior; D, dorsal; V, ventral. (B) Western blotting of anterior third instar larval extracts (lane 1), brains (lane 2), eye discs (lane 3), leg discs (lane 4), wing discs (lane 5), and salivary glands (lane 6) for Shep, Su(Hw), Mod(mdg4)2.2, Pep, and Lamin. (C) Epifluorescence imaging of insulator body localization by indirect immunofluorescence using rabbit α-CP190 and α-rabbit Alexa-594 in whole mount brain, leg imaginal disc, or eye imaginal disc tissues in wild type; mod(mdg4)^u1; or shep^BG00836, mod(mdg4)^u1 larvae. White dotted lines outline one example nucleus in each image. (D) Western blotting of larval extracts for Shep, Su(Hw), CP190, Mod(mdg4)2.2 and Pep in wildtype (lane 1), non-insulated (lanes 2–5), and insulated (lanes 6–9) luciferase lines. Act5c::Gal4 was used to drive single copy UAS-su(Hw) dsRNA (lanes 3 and 7), UAS-shep dsRNA (lanes 4 and 8) or Shep overexpression (UAS-shep, lanes 5 and 9). (E–G) Relative luciferase units were quantified in individual larvae expressing Act5C::Gal4 (E), l(3)31-1::Gal4 (F) Mef2::Gal4 (G), dsRNA hairpin, and/or UAS-shep as indicated. Luciferase values across the population are plotted as box and whisker plots where boxes represent upper and lower quartiles proximal to the median, and whiskers represent the range excluding outliers. Populations were compared by 1-way ANOVA, and pairwise p values were calculated by Tukey HSD post hoc tests. Outliers falling outside a normal distribution are shown (dots) but were not used to calculate p values. For each genotype, n≥12 larvae. For (F), non-insulated control vs. non-insulated shep RNAi, p = 0.18; for (G), insulated control vs. insulated shep RNAi, p = 0.99.

Figure 6. Comparison of Su(Hw), Mod(mdg4)2.2, and Shep ChIP–seq profiles in BG3 cells.

(A) Screenshot of Su(Hw), Mod(mdg4)2.2, and Shep ChIP-seq signals at the dnt neuronal-expressed locus. The large gap in ChIP signal corresponds to a highly repetitive region to which sequence reads could not be aligned with high confidence. (B) Screenshot of the caps neuronal-expressed locus. (C) Classification of Su(Hw), Mod(mdg4)2.2, and Shep ChIP-seq peaks in BG3 cells. Number of sites and percentage of total in parentheses corresponding to TSS, transcription start site; CDS, coding sequence; 5′ UTR, 5′ untranslated region; 3′ UTR, 3′ untranslated region. See methods for classification hierarchy of overlapping categories. (D) Heat map of log₂ enrichment scores for pairwise comparisons of binding sites for Su(Hw), Mod(mdg4)2.2, Shep, and additional data sets. Color scale corresponding to enrichment value is indicated (right). Positive values indicate significant enrichment while negative values indicate significant negative correlation of enrichment. Self-self comparisons are indicated in grey, and pairwise comparisons that are not statistically significant (p>0.001) are indicated in white. Numbers along top of each column indicate the total number of features in each data set, and the number of sites overlapping with Shep are indicated in parentheses. Data from Richter (2011) were derived from larval brains and imaginal discs; all other datasets are derived from BG3 cells. Data from modENCODE are indicated by an asterisk. Full heat map with hierarchical clustering is shown in Figure S4. (E) Binary heat map of Su(Hw), Mod(mdg4)2.2, and Shep binding sites in BG3 cells ordered by supervised hierarchical clustering. Each row represents a single genomic location, and a mark in a column represents the presence of a particular factor.