Domino regulates the expression of a large set of target genes

One of the well-established functions of the Tip60 complex especially in stem cells is regulation of gene expression. Moreover, Dom and Tip60 have been shown to regulate a large set of target genes in Drosophila (Ellis et al., 2015; Fazzio et al., 2008b; Lorbeck et al., 2011). H4K8 acetylation is reduced in dom deficient NBs and functional analyses have demonstrated the importance of H2Av in NB maintenance. This indicates that Dom and the Tip60 complex regulate gene expression via histone modification and incorporation in larval NBs to regulate self-renewal, polarity and division. To gain a better understanding of key regulators of NB maintenance I investigated the genes regulated by Dom in larval neural cells.

Since Dom is ubiquitously expressed, its target genes might strongly vary depending on the cellular context. I was exclusively interested in target genes in cells of the larval nervous system. Dom null mutants are early larval lethal, thus larval dom null mutant brains are not available for analysis. Homozyogus mutant cell clones can be induced but are not frequent and very small, consequently do not provide enough material for gene expression studies. I decided to use the previously validated dom-RNAi line v7787 for conducting a transcriptome-wide analysis (Figure 53).

In short, the neural driver line insc::Gal4 was used to mark neural cells with CD8-GFP and for knockdown of dom by RNAi (for details see 2.2.6). For a wild type control the same Gal4 driver was crossed to w¹¹¹⁸. Larval brains were dissected and committed for

fluorescence activated cell sorting (FACS) based on GFP expression. Previous studies have used similar setups to sort NBs based on their larger size (Berger et al., 2012). However, due to the NB size decrease upon dom knockdown this was not possible. Thus, neural cells were sorted. Figure 54 shows the conditions and gates set for sorting neural, CD8-GFP positive cells. For each sort, a GFP-negative sample of brain cells was used to determine the gate for the GFP-expressing cell population (R2). Furthermore, cells were sorted by absence of propidium iodide staining (PI) to retain only living cells. For each replicate between 1.4 and 2.2 million GFP^high PI^low cells were sorted. FACS was conducted together with Christoph Göttlinger, University of Cologne, Institute for Genetics.

Figure 53: Experimental setup for the transcriptome-wide analysis

Stepwise procedure for the transcriptome analysis included rearing of larvae of the indicated phenotypes at 25 °C to exclude activation of heat shock genes. Larval brains were dissected and single cell suspensions were sorted by FACS, based on GFP expression and absence of propidium iodide (PI) staining to obtain live neural cells (together with Christoph Göttlinger). After RNA extraction, library preparation and RNA-sequencing was done at the Cologne Center for Genomics. Subsequent bioinformatic analysis was kindly conducted by Dr. Manu Tiwari.

Figure 54: FACS conditions for sorting of live neural larval cells

FACS conditions for sorting of neural cells for subsequent transcriptome were based on low PI for live cells (R1) and positive GFP fluorescence (R2). A GFP-negative sample was used to discriminate between GFP^low and GFP^high conditions (Christoph Göttlinger).

A small fraction of the sorted cells was immunostained to confirm that the desired cell population was obtained. FAC-sorted cells were GFP positive and expressed correct neural markers (Figure 55).

Figure 55: insc::CD8-GFP L3 brain cells sorted by GFP express neural markers.

GFP-positive cells were sorted by FACS from L3 brains expressing CD8-GFP with the neural driver insc::Gal4.

For the wild type control, the driver was crossed to w¹¹¹⁸ (A). For dom depletion the dom-RNAi line v7787 was used (B). Immunostaining of GFP-sorted cells confirmed that nearly all sorted cells expressed GFP (A, B). Most cells expressed high levels of Pros (arrows, A’ and B’). Fewer bigger cells were positive for Mira (arrowheads), reflecting the smaller amount of NBs versus cells undergoing neurogenesis.

I extracted RNA from FACS-sorted neural cells and forwarded it to the Cologne Center for Genomics for further downstream processing and RNA-sequencing. Subsequent bioinformatic analysis to identify differentially expressed genes was done by Dr. Manu Tiwari (University of Cologne, Anatomy I, Molecular Cell Biology). Principal component analysis indicated that similar samples cluster together, showing similar variances. No outliers were identified (Figure 56).

Figure 56: Principal Component Analysis

Principal Component Analysis (PCA) analysis was used to visualize sample variances. Samples clustering together indicate similar variances, suggesting that they represent comparable gene sets. Importantly, the triplicates of the wild type and the triplicates of the dom knockdown are present in two different clusters.

Having confirmed that all replicates pass the quality controls, differentially expressed genes in the dom knockdown neural cells were identified by following the pipeline described in the methods section (2.2.6.5). The MA plot, an indicator of reproducibility between experimental samples, exhibited good fit across the samples (Figure 57). In total, 3326 genes were found to be differentially expressed at FDR  0.05.

Figure 57: MA plot showing differential gene expression

Visualization of all differentially expressed genes in a mean average (MA) plot by DEseq2 analysis. Dots represent genes. Red dots mark differentially expressed genes (FDR ≤ 0.05). Genes with negative log fold change are downregulated upon dom-RNAi.

Importantly, dom is present in the downregulated genes (log2FC = 0.576; FDR = 7.88E-7).

The rather slight downregulation of dom might be due to the low efficiency of the RNAi and lower induction of the Gal4-UAS system at 25 °C. Alternatively, it is possible that cells with high dom knockdown are underrepresented, as loss of dom functions results in a loss of the affected cells (Ruhf et al., 2001).

Interestingly, p53 (log2FC = 0.652; FDR = 1.9E-4) and its downstream target dacapo (log2FC = 0.645; FDR = 5.9E-5) were upregulated upon dom knockdown, indicating that dom is usually required for their repression. In addition, also Myc was upregulated upon dom knockdown (log2FC = 0.474; FDR = 9.5E-3). Importantly, if Dom can induce Myc expression, this can explain the partial rescue of myc knockdown by DomB overexpression (Figure 49).

3.6.1. Domino target genes regulate neuroblast fate

To identify major pathways regulated by dom I chose significantly regulated (FDR  0.05) target genes with -0.75  log2FC  +0.75 (1355 genes) and performed Database for Annotation, Visualization and Integrated Discovery (DAVID) functional annotation of gene ontology (GO)-terms. Table 4 lists clusters of GO-terms that were significantly enriched in the dom target genes (enrichtment score  1.5).

Table 4: Database for Annotation, Visualization and Integrated Discovery (DAVID) functional annotation:

GO-term analysis of Domino target genes

I performed DAVID functional annotation to identify enriched GO-term clusters in dom differentially regulated genes. Clusters represented in red color predominantly comprise upregulated genes, while those in green comprise predominantly downregulated genes. Grey clusters contain roughly equal portions of down- and upregulated genes.

Sensory Perception GOTERM_BP_FAT 8.8E-5 37

Annotation Cluster 2, Enrichment Score 3.58

Homeobox INTERPRO 2.2E-7 26

Transcription Factor Activity GOTERM_MF_FAT 9.2E-6 56

Annotation Cluster 3, Enrichment Score 2.84

Neurotransmitter Binding GOTERM_MF_FAT 8.8E-4 15

Annotation Cluster 4, Enrichment Score 2.34

Heme Binding GOTERM_MF_FAT 2.6E-5 27

Microsome GOTERM_CC_FAT 2.3E-3 15

Annotation Cluster 5, Enrichment Score 2.32

Sensory Perception of Chemical Stimulus GOTERM_BP_FAT 1.2E-3 26 Annotation Cluster 6, Enrichment Score 1.85

Gastrulation GOTERM_BP_FAT 9.0E-4 14

Mesoderm Formation GOTERM_BP_FAT 1.3E-3 8

Annotation Cluster 7, Enrichment Score 1.79

Cell Fate Determination GOTERM_BP_FAT 5.6E-3 18

Ganglion Mother Cell Fate Determination GOTERM_BP_FAT 1.3E-2 4

NB Fate Determination GOTERM_BP_FAT 2.3E-2 6

NB Fate Commitment GOTERM_BP_FAT 3.2E-2 6

NB Differentiation GOTERM_BP_FAT 3.7E-2 6

Annotation Cluster 8, Enrichment Score 1.75

Plasma Membrane Part GOTERM_CC_FAT 1.4E-2 40

I found that many highly enriched clusters (clusters 1, 3 and 5), representing mostly upregulated genes, contained GO-terms implicated in function of neurons. This substantiates the previously presented results pointing towards cells undergoing differentiation upon dom knockdown. Two clusters (cluster 4 and 8) consist of GO-terms containing genes connected to plasma membrane or the ER, which produces the plasma membrane. The misregulation of genes in these clusters could explain the elevated CD8-GFP fluorescence upon dom knockdown, as CD8-CD8-GFP is a membrane tethered marker.

Importantly, cluster 7 contains GO-terms implicated in regulation of NB and offspring cell behavior, which supports the premise that dom is required for the expression of genes modifying NB division and fate.