Pathological LSD1 mutations repress enhancer-mediated gene regulatory networks in early differentiation

Abstract

Lysine-specific demethylase 1 (LSD1/KDM1A) removes methylation of histone and non-histone substrates, and recruits a repressive chromatin complex. De novo LSD1 mutations impairing protein function lead to a rare neurodevelopmental disorder, but the molecular mechanisms of the pathology are unclear. Using patient-derived fibroblasts, reprogrammed pluripotent stem cells, and differentiated cells, we found over 4000 differentially expressed genes and 68 transcription factors (TFs) whose motif accessibilities changed upon LSD1 mutation. An enhancer-mediated gene regulatory network approach identified impaired transcriptional repressor activity in fibroblast and stem cells, leading to erroneous activation of their target genes. Furthermore, our analysis revealed overall decreases in TF target genes specifically during early lineage differentiation of LSD1 mutant stem cells, likely caused by increased activity of repressive co-factors of LSD1 - histone deacetylases (HDACs). Consistently, HDAC inhibitor restored changes in gene expression including downregulation phenotype. Our findings provide insights into pathogenesis of LSD1 mutations and targets for further therapeutic studies.

GRaNIE and GRaNPA: Inference and evaluation of enhancer-mediated gene regulatory networks applied to study macrophages

“…Notably, in all eGRNs, only about 20-30% of the peaks are linked to their closest gene TSS (Fig. 1G), an observation that is consistent with previous observations in pulmonary arterial endothelial cells (7) and iPSC derived cardiomyocytes (31).…”

“…Notably, in all networks, only about 20-30% of the enhancers are linked to their closest gene TSS (Fig. 1F), an observation that is consistent with previous observations in pulmonary arterial endothelial cells [7] and iPSC derived cardiomyocytes [29]. Stratifying the enhancer-gene links based on their distance to the gene TSS revealed a steep drop in number of connections for distances greater than 10kb (likely reflecting extended promoter regions) with a slower decline in numbers of significant connections for distances greater than 10kb (Fig.…”

GRaNIE and GRaNPA: inference and evaluation of enhancer‐mediated gene regulatory networks

“…We observed a slightly larger number of TFs classified as activators than as repressors (1.5–2‐fold), yet activators were connected with more peaks resulting in over 10‐fold more peaks being linked to an activator than to a repressor (Appendix Fig S5 A–D; Table 1 ). Notably, in all eGRNs, only about 20–30% of the peaks are linked to their closest gene TSS (Fig 1G ), an observation that is consistent with previous observations in pulmonary arterial endothelial cells (Reyes‐Palomares et al , 2020 ) and iPSC‐derived cardiomyocytes (preprint: Bunina et al , 2021 ).…”

“…Enrichment of ChIP-seq peaks overlapping a GRaNIE-inferred TF-bound peak (same TF) are shown for different TF-peak FDRs in the naive macrophage eGRN. Statistical significance was determined using Fishers Exact test; test set: all TF-peak pairs where the peak contains the motif for the respective TF (n = 25,205, 39,408, 78,971, 109,228, 142,548, 147,226 Notably, in all eGRNs, only about 20-30% of the peaks are linked to their closest gene TSS (Fig 1G), an observation that is consistent with previous observations in pulmonary arterial endothelial cells (Reyes-Palomares et al, 2020) and iPSC-derived cardiomyocytes (preprint: Bunina et al, 2021). On average, a gene is linked to 4.4 (naive), 2.9 (infected) and 5.9 (primed) peaks, of which 1.8 (naive), 1.5 (infected) and 5.7 (primed) are TF-bound and thus part of a GRaNIE eGRN (Fig 1H,Appendix Fig S7).…”