Interplay Between BALL and CREB Binding Protein Maintains H3K27 Acetylation on Active Genes in Drosophila

CREB binding protein (CBP) is a multifunctional transcriptional co-activator that interacts with a variety of transcription factors and acts as a histone acetyltransferase. In Drosophila, CBP mediated acetylation of histone H3 lysine 27 (H3K27ac) is a known hallmark of gene activation regulated by trithorax group proteins (trxG). Recently, we have shown that a histone kinase Ballchen (BALL) substantially co-localizes with H3K27ac at trxG target loci and is required to maintain gene activation in Drosophila. Here, we report a previously unknown interaction between BALL and CBP, which positively regulates H3K27ac. Analysis of genome-wide binding profile of BALL and CBP reveals major overlap and their co-localization at actively transcribed genes. We show that BALL biochemically interacts with CBP and depletion of BALL results in drastic reduction in H3K27ac. Together, these results demonstrate a previously unknown synergy between BALL and CBP and reveals a potentially new pathway required to maintain gene activation during development.


INTRODUCTION
In metazoans, covalent modifications of histones act in a combinatorial manner to regulate developmental gene expression (Allis and Jenuwein, 2016;Zhao and Shilatifard, 2019). Among these modifications, histone H3 lysine 27 acetylation (H3K27ac) is considered a hallmark of trithorax group (trxG) mediated gene activation that antagonizes repression by Polycomb group (PcG) proteins (Tie et al., 2009). Although the anti-silencing effect of H3K27ac and its catalyzing enzyme CREB binding protein (CBP) is widely documented in flies and mammals, how this CBP mediated H3K27ac is maintained on actively transcribed genes remains elusive (Tie et al., 2009;Pasini et al., 2010). In Drosophila, presence of only one homolog of CBP known as nejire, makes it suitable to understand its function. Importantly, chromatin immunoprecipitation sequencing (ChIP-seq) data from previously published reports revealed the presence of CBP on a majority of actively transcribed genes in Drosophila S2 cells (Philip et al., 2015). Our lab recently showed that Drosophila Ballchen (BALL), a known serine-threonine kinase that mediates histone H2AT119 phosphorylation (Aihara et al., 2004), associates with trxG target genes enriched with H3K27ac (Khan et al., 2021). In this report, we have further analyzed the link between BALL, H3K27ac and CBP. We have discovered that BALL not only binds to chromatin enriched with H3K27ac but also shares more than 77% genomic binding sites with CBP. Both BALL and CBP primarily associate with transcription start sites (TSS) and regions up to 1 Kb upstream of TSS. Four of top five DNA motifs in chromatin that CBP binds are also bound by BALL. Importantly, BALL biochemically interacts with CBP and positively contributes to maintenance of H3K27ac by CBP since depletion of BALL leads to diminished H3K27ac.

RESULTS AND DISCUSSION
Recently we reported that BALL exhibits trxG like behavior by contributing to the maintenance of gene activation in flies (Khan et al., 2021). Since BALL was shown to co-localize with Trithorax (TRX) at genes enriched with H3K27ac, we investigated if BALL binds to chromatin together with CBP. To this end, we analyzed ChIP-seq data of BALL (Khan et al., 2021) and CBP (Philip et al., 2015) from Drosophila S2 cells. Comparison of BALL and CBP binding to chromatin revealed their co-occupancy at 4577 genes ( Figure 1A and Supplementary Table 1). Further analysis revealed that both BALL and CBP shared binding sites in genomic regions up to 1 Kb upstream of the transcription start sites (TSS), however, CBP is also found at the regions that are further upstream of TSS (Figures 1B,C). This finding is in line with the reported binding of CBP in promoter as well as enhancer regions (Philip et al., 2015;Boija et al., 2017). Analysis of the genomic distribution of CBP and BALL in terms of their binding at promoter regions, gene body, introns and intergenic regions also revealed similar chromatin binding patterns ( Figure 1D). Of 6,553 overlapping binding sites for both BALL and CBP, 5,865 were mainly clustered on promoter regions (Supplementary Figure 1 and Supplementary Table 2). Importantly, promoters of different genes have variable number of peaks, i.e., some promoters have multiple binding sites while others have a single binding site. Binding profile of BALL closely mirrored that of CBP and H3K27ac at a majority of genomic regions. Additionally, most of these shared loci were also expressed in S2 cells, as highlighted by analysis of RNA-seq data (Gan et al., 2010; Figure 1E). Of 4577 genes where both BALL and CBP are present, 3114 (68%) were indeed found to be actively transcribed in S2 cells ( Figure 1F and Supplementary Table 1). Comparison between top scoring DNA motifs present within chromatin regions where both BALL and CBP preferentially bind revealed a distinct similarity ( Figure 1G). Collectively, our data highlights that BALL binding pattern substantially overlaps the binding of CBP across the whole genome. These findings prompted us to investigate whether BALL and CBP biochemically interact with each other. To this end, co-immunoprecipitation (co-IP) experiments were performed using S2 cells that stably expressed FLAG-tagged BALL. Immunoprecipitation of BALL-FLAG resulted in precipitation of endogenous CBP as compared to empty vector control ( Figure 1H). To confirm this interaction, a reciprocal co-IP was performed using anti-CBP antibody, which resulted in immunoprecipitation of BALL-FLAG ( Figure 1I) as compared to mock IP. These results illustrate a hitherto unknown interaction between BALL and CBP, which may explain their co-occupancy at actively transcribed regions.
Next, we investigated if depletion of BALL affects CBP mediated H3K27ac levels globally. Western blot analysis of cells where ball was knocked down revealed drastic reduction in H3K27ac levels as compared to control cells treated with dsRNA against LacZ (Figure 2A). The global levels of histone H3 remained unchanged which were used as normalization control to quantify reduction in levels of H3K27ac (Supplementary Figure 2). To validate this finding in vivo, we generated homozygous ball 2 mitotic clones using the flp/FRT system. Immunostaining of imaginal discs with ball 2 mutant clones showed a drastic reduction in H3K27ac (Figures 2B-E and Supplementary Figure 3). Notably, not all homozygous ball 2 mitotic clones show diminished H3K27ac which reiterates similar effects observed in trx mutant clones on expression of actively transcribed genes. It could be explained by the complex interplay of different cell signaling pathways in imaginal discs (Klymenko and Müller, 2004).
Together, these results demonstrate that BALL positively contributes to CBP mediated H3K27ac but the molecular nature of this interaction between BALL and CBP remains unclear. Whether they are part of a single multi-protein complex and if they directly interact with each other or it involves an indirect interaction remains elusive. Since BALL is a histone kinase, it is imperative to investigate whether the impact that BALL exerts on H3K27ac is through the phosphorylation of neighboring H3 residues or due to the direct phosphorylation of CBP or its interacting proteins. Interestingly, VRK1 (Vaccinia related kinase 1), the mammalian homolog of BALL, was shown to phosphorylate CREB at serine 133 (Kang et al., 2008). This phosphorylation is known to enhance CREB interaction with CBP, leading to increased transcription of CREB target genes (Radhakrishnan et al., 1998). The VRK1, CBP and CREB nexus may explain the regulation of CBP mediated gene activation by BALL in flies, which warrants further investigation.

Chromatin Immunoprecipitation Sequencing Analysis
Chromatin immunoprecipitation sequencing (ChIP-seq) analysis was done as described previously (Khan et al., 2021). Briefly, the sequencing datasets were uploaded to the Galaxy web public server (Jalili et al., 2020). Using Bowtie version 2, the ChIP-seq data was mapped to Drosophila genome (dm6) (Langmead et al., 2009). Peak calling was performed using MACS version 2 with FDR (q-value) of 0.05 as cut-off and Genes co-occupied by BALL and CBP display active transcription as inferred by RNA-seq analysis of S2 cells (bottom panel). (F) Venn diagram depicting CBP and BALL enriched genes overlaid with actively transcribed genes in S2 cells. Majority of the genes that were co-bound by CBP and BALL were found to be transcriptionally active. (G) Four of the top five short DNA binding motifs within CBP associated chromatin, generated from MEME-ChIP database, show-striking similarity with the similar motifs within BALL bound chromatin. Each motif is presented with its respective reverse complementary motif and E-value. (H) Immunoblots of endogenous CBP and FLAG-tagged BALL after co-immunoprecipitation (IP) using anti-FLAG antibody in Drosophila S2 cells containing inducible BALL-FLAG. As compared to IP from empty vector (EV) control cells, noticeable enrichment of CBP can be seen in IP using anti-FLAG antibody, which also resulted in enrichment of FLAG-BALL. (I) Immunoblots of endogenous CBP and FLAG-tagged BALL after co-immunoprecipitation (IP) using anti-CBP antibody in Drosophila S2 cells. BALL-FLAG is enriched after IP with anti-CBP antibody as compared to mock IP used as negative control. Enrichment of CBP validates successful IP.
read extension of 200 bp (Zhang et al., 2008). List of annotated peaks was extracted from peak file using ChIPseeker (Yu et al., 2015). Comparative heat maps of BALL and CBP were generated using deepTools, (bamCompare, computeMatrix, and plotHeatmap) (Ramírez et al., 2016). To find out overlapping peaks between BALL and CBP, "bedtools Intersect intervals" FIGURE 2 | Depletion of BALL affects H3K27ac levels. (A) Western blot analysis of whole cells lysates from Drosophila S2 cells treated with dsRNA against ball exhibit reduced H3K27ac. Cells treated with dsRNA against LacZ served as control while total levels of histone H3 were used as a normalization control in immunoblot. Analysis of the relative intensities of H3K27ac signal, normalized against total H3 levels using ImageJ, showed more than 37% reduction in H3K27ac upon ball depletion. Experiment was performed thrice and replicates are shown in Supplementary Figure 2. (B-E) Haltere imaginal discs containing mitotic clones of ball 2 were stained with anti-H3K27ac (D). Mitotic clones, marked by the absence of GFP, showed drastic reduction of H3K27ac. Imaginal discs showed uniform DAPI staining (B). Mitotic clones are encircled and highlighted with arrows. The ex vivo knock down of ball was performed in triplicate while the mitotic clone experiment was performed in duplicate.
was utilized in galaxy database. MEME-ChIP -motif discovery, enrichment analysis and clustering on large nucleotide datasets (Galaxy Version 4.11.2 + galaxy1) was utilized to generate DREME (Discriminative Regular Expression Motif Elicitation) output of BALL and CBP ChIP-seq data for de novo motif discovery (Jalili et al., 2020). ChIP-seq data of CBP, BALL, and H3K27ac were taken from GSE64464 (Philip et al., 2015), GSE165685 (Khan et al., 2021) and GSE81795 (Rickels et al., 2016), respectively. RNA-seq data of Drosophila S2 cells (GSM480160) (Gan et al., 2010) was mapped using TopHat Gapped-read mapper for RNA-seq data (Galaxy Version 2.1.1). Exon read count was calculated using htseq-count to generate list of active genes.

Generation of BALL Stable Cell Line
To generate stable cell line with inducible expression of FLAG tagged BALL, w 1118 embryos were used to prepare cDNA and amplify ball CDS. The ball CDS was first cloned in pENTR-D-TOPO entry vector (Thermo Fisher Scientific) followed by sub-cloning in pMTWHF destination vector, a gift from Paro lab (Kockmann et al., 2013), by setting up LR Clonase reaction following manufacturer's protocol (Thermo Fisher Scientific). The resulting pMT-ball-FLAG vector map is given in the Supplementary Figure 4. Drosophila S2 cells were transfected with pMT-ball-FLAG using Effectene transfection reagent (Qiagen) and a stable cell line was generated by following the manufacturer's instructions (Qiagen).

Co-immunoprecipitation
For co-immunoprecipitation, Drosophila S2 cells expressing FLAG-tagged BALL were washed twice with PBS after harvesting them by centrifugation at 3000rpm for 5 min. Cells were suspended in 1 mL lysis buffer containing 140 mM NaCl, 20 mM Tris (7.4 pH), 1 mM EDTA, 0.5% NP40, 10% glycerol, 0.2 mM Na 2 VO 4 , pepstatin 0.5 µg/mL, leupeptin 0.5 µg/mL, aprotinin 0.5 µg/mL and PMSF 1 mM followed by incubation on ice for 30 min. The lysate was centrifuged at 14,000rpm for 15 min and supernatant was added to 40 µL of M2-FLAG agarose beads (Sigma-Aldrich) which were pre-washed with lysis buffer. After 1 h of incubation, beads were washed thrice with lysis buffer, re-suspended in 30 µL Laemmli Loading buffer and heated at 95 • C for 5 min. IP samples were loaded on Novex precast Tris-acetate 3-8% gradient gels (Thermo Fisher Scientific). For CBP IP, 1:1 Protein A and G Dyna beads (Thermo Fisher Scientific) mixture was incubated with anti-CBP antibody overnight. Cell lysates were prepared, and incubated for 1 h at 4 • C with the beads-antibody complex. After washing thrice with lysis buffer, samples were proceeded for western blotting as described above.

Ex vivo Knock Down
PCR amplification of the templates for dsRNAs preparation was done using T7 promoter (TAATACGACTCACTATAGGGAGA) tailed oligonucleotides as primers. Primers used for ball template are ATAGTTCACCACCCAGCCAG and ATCCTGGTCCGCTTTCTTTT for the DRSC amplicon ID DRSC26607 (Ramadan et al., 2007). Primers used for LacZ template are GGAAGATCAGGATATGTGG and CTTCATCAGCAGGATATCC. In vitro transcription of the templates was performed using MEGAscript TM T7 Transcription Kit according to the manufacturer's instructions (Ambion). For setting up the knock down experiment, D. Mel-2 cells were treated with 10 µg/mL of respective dsRNA for 4 days and total cell lysates were prepared in Laemmli loading buffer before heating at 95 • C for 5 min.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

AUTHOR CONTRIBUTIONS
AS and MHFK performed the research. MHFK, AS, ZU, HA, and MT wrote the manuscript. All authors contributed to the article and approved the submitted version.