To identify novel targets for acute myeloid leukemia (AML) therapy, we performed genome-wide CRISPR-Cas9 screening using AML cell lines, followed by a second screen in vivo. Here, we show that the mRNA decapping enzyme scavenger (DCPS) gene is essential for AML cell survival. The DCPS enzyme interacted with components of pre-mRNA metabolic pathways, including spliceosomes, as revealed by mass spectrometry. RG3039, a DCPS inhibitor originally developed to treat spinal muscular atrophy, exhibited anti-leukemic activity via inducing pre-mRNA mis-splicing. Humans harboring germline biallelic DCPS loss-of-function mutations do not exhibit aberrant hematologic phenotypes, indicating that DCPS is dispensable for human hematopoiesis. Our findings shed light on a pre-mRNA metabolic pathway and identify DCPS as a target for AML therapy.

Chromatin reorganization is necessary for pluripotent stem cells, including embryonic stem cells (ESCs), to acquire lineage potential. However, it remains unclear how ESCs maintain their characteristic chromatin state for appropriate gene expression upon differentiation. Here, we demonstrate that chromodomain helicase DNA-binding domain 2 (Chd2) is required to maintain the differentiation potential of mouse ESCs. Chd2-depleted ESCs showed suppressed expression of developmentally regulated genes upon differentiation and subsequent differentiation defects without affecting gene expression in the undifferentiated state. Furthermore, chromatin immunoprecipitation followed by sequencing revealed alterations in the nucleosome occupancy of the histone variant H3.3 for developmentally regulated genes in Chd2-depleted ESCs, which in turn led to elevated trimethylation of the histone H3 lysine 27. These results suggest that Chd2 is essential in preventing suppressive chromatin formation for developmentally regulated genes and determines subsequent effects on developmental processes in the undifferentiated state.

Biomedical research often involves conducting experiments on model organisms in the anticipation that the biology learnt will transfer to humans. Previous comparative studies of mouse and human tissues were limited by the use of bulk-cell material. Here we show that transfer learning—the branch of machine learning that concerns passing information from one domain to another—can be used to efficiently map bone marrow biology between species, using data obtained from single-cell RNA sequencing. We first trained a multiclass logistic regression model to recognize different cell types in mouse bone marrow achieving equivalent performance to more complex artificial neural networks. Furthermore, it was able to identify individual human bone marrow cells with 83% overall accuracy. However, some human cell types were not easily identified, indicating important differences in biology. When re-training the mouse classifier using data from human, less than 10 human cells of a given type were needed to accurately learn its representation. In some cases, human cell identities could be inferred directly from the mouse classifier via zero-shot learning. These results show how simple machine learning models can be used to reconstruct complex biology from limited data, with broad implications for biomedical research.

Chd5 is an essential factor for neuronal differentiation and spermatogenesis and is a known tumor suppressor. H3K27me3 and H3K4un are modifications recognized by Chd5; however, it remains unclear how Chd5 remodels chromatin structure. We completely disrupted the Chd5 locus using the CRISPR-Cas9 system to generate a 52 kbp long deletion and analyzed Chd5 function in mouse embryonic stem cells. Our findings show that Chd5 represses murine endogenous retrovirus-L (MuERV-L/MERVL), an endogenous retrovirus-derived retrotransposon, by regulating H3K27me3 and H3.1/H3.2 function.

Non-allelic histone variants are considered as epigenetic factors that regulate genomic DNA functions in eukaryotic chromosomes. In this study, we identified three new human histone H3 variants (named H3.6, H3.7, and H3.8), which were previously annotated as pseudogenes. H3.6 and H3.8 conserve the H3.3-specific amino acid residues, but H3.7 shares the specific amino acid residues with H3.1. We successfully reconstituted the nucleosome containing H3.6 in vitro and determined its crystal structure. In the H3.6 nucleosome, the H3.6-specific Val62 residue hydrophobically contacts the cognate H4 molecule, but its contact area is smaller than that of the corresponding H3.3 Ile62 residue. The thermal stability assay revealed that the H3.6 nucleosome is substantially unstable, as compared to the H3.3 nucleosome. Interestingly, mutational analysis demonstrated that the H3.6 Val62 residue is fully responsible for the H3.6 nucleosome instability, probably because of the weakened hydrophobic interaction with H4. We also reconstituted the nucleosome containing H3.8, but its thermal stability was quite low. In contrast, purified H3.7 failed to form nucleosomes in vitro. The identification and characterization of these novel human histone H3 variants provide important new insights into understanding the epigenetic regulation of the human genome.