Drosophila Histone Demethylase KDM4A Has Enzymatic and Non-enzymatic Roles in Controlling Heterochromatin Integrity

Abstract

Eukaryotic genomes are broadly divided between gene-rich euchromatin and the highly repetitive heterochromatin domain, which is enriched for proteins critical for genome stability and transcriptional silencing. This study shows that Drosophila KDM4A (dKDM4A), previously characterized as a euchromatic histone H3 K36 demethylase and transcriptional regulator, predominantly localizes to heterochromatin and regulates heterochromatin position-effect variegation (PEV), organization of repetitive DNAs, and DNA repair. We demonstrate that dKDM4A demethylase activity is dispensable for PEV. In contrast, dKDM4A enzymatic activity is required to relocate heterochromatic double-strand breaks outside the domain, as well as for organismal survival when DNA repair is compromised. Finally, DNA damage triggers dKDM4A-dependent changes in the levels of H3K56me3, suggesting that dKDM4A demethylates this heterochromatic mark to facilitate repair. We conclude that dKDM4A, in addition to its previously characterized role in euchromatin, utilizes both enzymatic and structural mechanisms to regulate heterochromatin organization and functions.

Keywords: DNA repair; Drosophila; H3K36me3; H3K56me3; HP1a; dKDM4A; heterochromatin; histone demethylase; position-effect variegation; γH2Av.

Figure 1. dKDM4A is enriched in HC

(A) Live imaging of S2 cells stably expressing GFP-tagged dKDM4A and mCherry-tagged HP1a. (B) Time-lapse imaging of BG3 cells transiently expressing GFP-tagged HP1a and mCherry-tagged dKDM4A from G2 to G1. Dispersion of HP1a marks progression of cells through mitosis. (C) Time-lapse imaging of BG3 cells transiently expressing GFP-tagged dKDM4A or HP1a, and mCherry-tagged PCNA shows continued HC enrichment of dKDM4A (above) and HP1a (below) during late S phase, as marked by PCNA replication foci in HC. (D) Live imaging of Kc cells stably expressing GFP-HP1a and mCherry-dKDM4A after 5-day RNAi of bw (control), dKDM4A, Su(var)3-9, or HP1a. Pearson coefficient of correlation (p) is shown in merged images of bw (n=25), dKDM4A (n=22), Su(var)3-9 (n=27), and HP1a RNAi (n=23). (E) Western blot analysis of S2 cell extracts after 5-day RNAi of y (control), dKDM4A, or HP1a. (F) Live imaging of S2 cells transiently expressing GFP-tagged HP1a and mCherry-tagged dKDM4A mutated for the PxVxL motif required for HP1a binding. Scale bars = 5 μm. See also Figure S1.

Figure 2. Position effect variegation is dependent on dKDM4A levels

(A) Representative images of wildtype and mutant dKDM4A fly abdomens exhibiting silencing effects on a y⁺ reporter gene inserted in different HC regions of chromosomes 2, 3, and 4. Images represent the median of at least 10 individuals. (B) Fiji-based quantitation of pigmentation on the last two abdominal segments is shown for wildtype and dKDM4A mutant flies for each reporter gene strain. Error bars representing SD and p-values are shown. (C) Representative FISH images of S2 cells showing AACAC tandem repeats after GFP (control) or dKDM4A RNAi. Scale bars = 5 μm. Number of AACAC repeat foci is quantitated in (D), and a similar experiment in Kc cells using LNA probes for AATAACATAG is quantitated in (E). Error bars representing SD and p-values are shown. See also Figure S2.

Figure 3. dKDM4A does not significantly contribute to transcriptional regulation of HC elements

(A) Representative Western blot analysis of dKDM4A and H3K36me3 levels after 5-days RNAi depletion of dKDM4A, compared to GFP RNAi (control). Fiji-based quantitation of dKDM4A and H3K36me3 levels after normalization to GFP RNAi levels and the H3 loading control is shown below. (B) Number of genes and repetitive elements showing a significant change in RNA levels after 5-day dKDM4A or HP1a RNAi in S2 cells, compared to a GFP RNAi control, p<0.05. Number of HC genes affected by either dKDM4A or HP1a RNAi is shown in parentheses. “overlap” row indicates the number of genes and repetive elements co-regulated by dKDM4A and HP1a. (C) Graph of repetitive element transcript levels increased by HP1a RNAi (blue) over GFP RNAi is contrasted with their levels after dKDM4A RNAi (orange). (D) HP1a RNAi-induced decreases in repetitive element transcript levels is contrasted with dKDM4A RNAi effects. See also Figure S3 and Table S1, S2.

Figure 4. dKDM4A demethylation of H3K36me3 is not a significant contributor to HP1a domain structure

(A) Representative IF images of Kc cells stained for H3K36me3 and HP1a (top) and H3K36me3 and H3K9me2 (bottom) after 5-days of y (control) or dKDM4A RNAi. Scale bars = 5 μm. (B) Co-localization analysis of H3K36me3 and H3K9me2 signals after y or dKDM4A RNAi. Correlation co-efficient, p, is shown, based on 23 cells. Histograms showing frequency distributions of H3K36me3 (left) and H3K9me3 (right) ChIP-Seq peak enrichment values mapping to EC (C) and HC (D) after GFP control (blue) or dKDM4A (orange) RNAi. Data shown are cumulative peaks from 2 independent experiments, including p-values and D-statistics from two-sample Kolmogorov-Smirnov test. Plots showing H3K36me3 (E) and H3K9me3 (F) enrichment values for repetitive elements between GFP and dKDM4A RNAi-treated cells (mean of 2 experiments). The standard deviation of mean enrichment values (dotted lines) is used as a conservative measure of significantly increased or decreased association with H3K36me3 or H3K9me3 after dKDM4A depletion. In (F), a trendline drawn through the mean enrichment values of transposon sequences and a two-tailed t-test highlight H3K9me3 enrichment after dKDM4A RNAi.

Figure 5. dKDM4A enhances variegation independent of its catalytic activity

(A) Representative images of y⁺ PEV suppression between dKDM4A mutant flies expressing a single copy of wildtype or catalytically inactive dKDM4A transgene, and wildtype and dKDM4A mutant flies bearing no transgenes. Fiji-based quantitation of pigmentation levels of the two posterior segments is shown on the right, with Student’s t-test. (B) Representative images of y⁺ PEV suppression in wildtype flies with or without a single copy of wildtype or catalytically inactive dKDM4A transgene. Corresponding quantitation of pigmentation levels with Student’s t-test is shown. Error bars represent SD. See also Figure S4.

Figure 6. dKDM4A is required for efficient repair of heterochromatic DNA damage

(A) Graph comparing γH2Av foci frequency at DAPI-bright region between Kc cells after 5-day RNAi of bw (control), dKDM4A or HP1a, and multiple timepoints after 5 Gy of IR. Error bars represent SE from 2 independent experiments. (B) Western blot analysis of Kc cells stably expressing GFP-tagged wildtype or catalytically-inactive dKDM4A transgenes with alternative codons after 5-day RNAi depletion of endogenous dKDM4A. (C) Representative IF images of the same Kc cells showing retention of γH2Av foci in the heterochromatic DAPI-bright region (left, dotted line) and in HP1a-enriched HC (right, dotted line) 60 minutes after 5 Gy of IR. Scale bars = 5 μm. Quantitation of γH2Av foci frequency for each HC domain is shown to the right of each image, with Student’s t-test. (D) Western blot showing H3K56me3 and H2B levels in larval extracts from dKDM4A mutant flies or wildtype flies. (E) Representative Western blots of Kc nuclear extracts after 5 days of GFP (control) or dKDM4A RNAi, and up to 60 minutes after 5 Gy of IR, showing H3K56me3, H3K9me3, dKDM4A, and Lamin levels. (F) Quantitation of H3K56me3 and H3K9me3 levels in GFP or dKDM4A RNAi-treated cells post-IR, after normalization to Lamin levels. Error bars indicate SE based on 4 independent experiments, and p-value is based on Wilcox rank test. See also Figure S5.