ABSTRACT: Purpose: Chromodomain helicase DNA-binding (CHD) chromatin remodelers regulate transcription and DNA repair. Chd1-8 show upon germline disruption pronounced, often developmental lethal phenotypes. Here we show that contrary to Chd1-8 disruption, Chd9–/– animals are viable, fertile and display no developmental defects or disease predisposition. Germline deletion of Chd9 only moderately affects gene expression in tissues and derived cells, whereas acute depletion in human cancer cells elicits more robust changes suggesting that CHD9 is a highly context-dependent chromatin regulator that, surprisingly, is dispensable for mouse development. Methods: The role of CHD9 in gene expression regulation during murine development and lymphomagenesis, as well as in human cancer cell lines was explored by RNA-seq analysis. Embryonic (E15) brain and liver were isolated from littermates and snap-frozen. Mouse embryonic fibroblasts (MEFs) were isolated from the Chd9+/-x Chd9+/- cross at day 13.5 post coitum and cultured using standard methods. Three wild-type control and four Chd9 knockout derived fibroblast lines were used. Embryonic stem cells (ESC) were derived from the Chd9+/-x Chd9+/- cross as described before (Huijbers IJ, et al, 2015). Three wild-type control and three Chd9 knockout derived ESC lines were used. For all the experiments early passage MEFs and ESCs were used. Enlarged lymph nodes from three moribund Emu-Myc control, Chd9+/- and Chd9-/- mice were isolated and snap-frozen. Human SNB19 and K562 cells bearing doxycycline-inducible short hairpin RNAs (shRNA) targeting CHD9 were induced with 1ug/ml doxycycline for 48 hours, harvested, and pellet was snap-frozen. Total RNA was extracted, prepared using TruSeq protocol, and standard sample preparation protocols and RNA-seq was performed on a Hiseq2000 machine (Illumina) at the NKI Genomics Core Facility. Results: Sequencing data were aligned to mm9 (MEFs, ESCs), mm10 (E15 brain and liver; Emu-Myc lymphomas), hg37 (K562 and SNB19 cells) using TopHat (v2.1.0 using bowtie 1.1.0). and number of reads per gene were measured with HTSeq count (v0.5.3). Statistical analysis of the differential expression of genes were performed using DESeq2. Results were considered statistically significant at an adjusted p < 0.05. Normalized expression values were obtained by correcting for differences in sequencing depth between samples using DESeqs median-of-ratios approach. Analysis of Chd9 knockout MEFs revealed only downregulation of Chd9 and upregulation of the Hmgb2 and pseudogene Gm13237, (adjusted p–value 0.05, log2FC ±1). Conversely, in derived Chd9–/– ESCs maintained in serum–free 2i media with leukemia inhibitory factor (LIF), 63 upregulated and 212 downregulated genes were observed (adjusted p–value 0.05, log2FC ±1). RNA–seq of Chd9 knockout embryonic (E15) liver and brain revealed only a handful of differentially expressed genes (DEGs) (adjusted p–value 0.05, log2FC ±1). Both liver and brain show concordant upregulation of an uncharacterized pseudogene Gm4322. Despite high Chd9 expression in the developing brain, only marginal upregulation of the Hoxa (a7), Hoxb (b6, b7, b9) and Hoxc (c5, c6) homeotic gene clusters was observed. Analysis of lymphomas from Emu-Myc Chd9 wildtype, heterozygous and knockout mice revealed that in heterozygous tumours 20 genes were upregulated, and only Lrnn4 was downregulated. In the knockout tumours 28 genes were up– and 8 down–regulated. The presence of a shared six upregulated genes in both CHD9 heterozygous and knockout lymphomas suggests that they might be direct target genes. RNA-seq analysis of human cancer cell lines revealed profound gene expression changes 48 hours upon CHD9 depletion. In K562 cells, CHD9 knockdown resulted in 215 down- and 635 upregulated genes, whereas in SNB19 cells, 121 down- and 113 up–regulated genes were observed. Notably, a number of deregulated genes in CHD9-depleted K562 cells – such as PBX1, APLNR, and BIN1– are implicated in neuronal development and pathology. This suggests a more widespread role for CHD9 in regulation of neuronal-specific genes, which even shows up in the context of malignant hematopoietic K562 cells. We did not find any significant overlap in DEGs between both human cancer cell lines. Conclusions: Our study represents the first analysis of the role of CHD9 in gene expression regulation during murine development and lymphomagenesis, as well as in human cancer cells. The very modest effects on gene expression seen upon germline deletion of Chd9, as compared to the substantial change in gene expression upon acute depletion in two human cell lines, point to compensatory me chanisms mitigating the effects of germline deletion of Chd9. Our gene expression analysis failed to reveal any compensatory transcriptional deregulation of CHD enzymes or other chromatin regulators. Our results reveal that, despite ubiquitous expression, Chd9 is dispensable for embryonic development. Furthermore, knockout animals are viable, fertile, display no overt phenotypes and have a normal lifespan.
