The chromatin remodeler CHD8 governs hematopoietic stem/progenitor survival by regulating ATM-mediated P53 protein stability

Hematopoiesis is a tightly regulated process that constantly generates a large number of mature blood cells through a cascade differentiation of hematopoietic stem/progenitor cells (HSPCs). The longevity and integrity of the stem cell pool governs the lifelong homeostasis of the hematopoietic system.¹  The adult HSPCs are maintained in the bone marrow (BM) and are capable of dramatic expansion and contraction to ensure proper homeostatic replacement of blood cells upon stress signals.^2,3  At the cellular level, well-maintained chromatin organization is critical to HSPC health and function, as defective chromatin structure can be susceptible to physiological stress resulting in DNA damage and genomic instability. Mice lacking components and regulators of the chromosomes and genomic stability display severe hematopoietic phenotypes and HSPC defects.^4-10 

Adenosine triphosphate (ATP)-dependent chromatin remodeling enzymes play an essential role in stem cell biology.¹¹  These enzymes use the energy from ATP hydrolysis to regulate DNA accessibility, which enables transcription regulators to bind, to activate or repress gene expression.¹²  Among the evolutionarily conserved SWI-like ATP-dependent chromatin remodeling complexes, the CHD remodeling enzymes are defined by the presence of 2 chromodomains: a helicase/ATPase domain and a DNA-binding domain.^11,13  Genetic, biochemical, and structural studies demonstrate that CHD proteins are important regulators of transcription and are indispensable during developmental processes.^14,15  Several CHD proteins, including CHD3, CHD4, and CHD7, have been found to be involved in hematopoiesis.^16-20 

CHD8 was first identified as a negative regulator of the Wnt/β-catenin signaling pathway and is one of the most frequently mutated genes involved in the autism spectrum disorders.^21-23  Recent studies have focused on its role in neurogenesis and brain development.^24,25  CHD8 appears to be essential for B acute lymphoblastic leukemia cell survival and correlates with Brd4-NSD3 across the genome in acute myeloid leukemia cells.^26,27  In the current study, we conditionally deleted Chd8 in hematopoietic cells using the Mx1-Cre driver and found CHD8 essential for the maintenance of normal hematopoiesis, different from its role in most other tissues. Loss of CHD8 led to severe pancytopenia and BM failure because of its essential role in regulating HSPC survival. Although CHD8 is not a DNA damage response and repair cue, it binds directly to P53 and ATM to restrict their function, and loss of CHD8 causes a change in chromatin integrity resulting in ATM activation and consequent P53 activation, leading to cell apoptosis. We implicate the chromatin remodeling factor CHD8 as an essential controller of HSPC survival and hematopoiesis.

Chd8^Flox mice and their genotyping were reported previously,²⁴  and P53^Flox mice were purchased from The Jackson Laboratory (#008462). To induce hematopoietic cell–specific deletion, Chd8^Flox and P53^Flox mice were crossed with Mx1-Cre mice. All animal experiments were performed in accordance with the protocols from the Institutional Animal Care and Use Committee at Cincinnati Children’s Hospital Medical Center. For Cre-mediated recombination, 6- to 8-week-old control and Cre-expressing animals were injected with polyinosinic/polycytidylic acid (pI/pC) (15 mg/kg; GE Healthcare) 4 times, and BM cells were harvested for analysis 6 days, 10 days, and 8 weeks after the last injection. pI/pC-treated Chd8^F/F mice were considered wild type (WT), and pI/pC-treated Chd8^F/F;Mx1-Cre mice that lost Chd8 alleles were considered to be Chd8^−/−.

For direct transplantation, 2 × 10⁶ fresh, isolated BM cells from Chd8^F/F and Chd8^−/− mice (CD45.2) were transplanted into lethally irradiated wild-type mice (CD45.1). For competitive transplantation, 1 × 10⁶ fresh, isolated BM cells from donors (CD45.2) were mixed 1:1 with WT (CD45.1) cells and transplanted into lethally irradiated WT recipients (CD45.1). BM chimerism was monitored by fluorescence-activated cell sorting in peripheral blood (PB) at 4 weeks, and then the mice were injected 4 times with pI/pC (GE Healthcare) at a dose of 15 mg/kg. Chimerism was assessed every 4 weeks for 16 weeks by cell sorting. For a secondary transplant, 2 × 10⁶ BM cells from primary recipient mice were transplanted into lethally irradiated WT recipients (CD45.1). The mice were euthanized 16 weeks after transplantation and used for flow cytometry.

Low-density BM cells were harvested from the Chd8^F/F and Chd8^−/− mice using Histopaque-1083 and then cKit⁺ cells were further enriched with CD117 microbeads (Miltenyi Biotec). The number of cells was determined, and the cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM; Thermo Fisher Scientific) supplemented with a final concentration of 10% fetal bovine serum (FBS), 100 ng/mL stem cell factor (PeproTech), 100 ng/mL troponin (PeproTech), and 100 ng/mL granulocyte colony?stem cell factor (G-CSF; PeproTech), with or without specific inhibitors or stress treatment. The final concentration of chemicals was 10 μM MG132 (#S2619; Selleck Chemicals), 10 μg/mL cycloheximide (#01810; MilliporeSigma), 1 μM and 10 μM KU-55933 (#SML1109; MilliporeSigma), and 5 μM etoposide (#33419-42-0; Cayman Chemical). The dose of irradiation was 3 Gy. After culture, the cells were further harvested for analysis.

The ubiquitination assay was performed according to a published protocol.²⁸  The low-density BMs were cultured in Iscove’s modified Dulbecco’s medium for 4 hours with 10 M MG132 and harvested by centrifuge. The cells were lysed in cell lysis buffer (2% sodium dodecyl sulfate, 150 mM NaCl, and 10 mM Tris-HCl [pH 8.0]) with inhibitors, and the protein was harvested by centrifuge. Protein concentration was determined with Quick Start Bradford (Bio-Rad) and similar amounts of proteins across samples were used for P53 immunoprecipitation. The elution was further blotted with a ubiquitin antibody to detect ubiquitinated P53.

Statistical analyses were performed with GraphPad Prism software. All data are shown as means ± standard error of the mean. The 2-tailed Student t test was used to determine statistical significance.

Chd8 is found expressed in all subpopulations of HSPCs by reverse transcription-polymerase chain reaction (RT-PCR) (supplemental Figure 1A, available on the Blood Web site). To determine whether Chd8 plays a role in hematopoiesis, we crossed Chd8^Flox mice with an Mx1-Cre driver and conditionally deleted Chd8 by pI/pC induction. After pI/pC injection, Chd8^F/F mice were considered to be WT and Chd8^F/F;Mx1-Cre mice were considered to be Chd8^−/− (Figure 1A). In Chd8^−/− mice, Chd8 messenger RNA (mRNA) in whole-BM cells and CHD8 protein in the sorted hematopoietic progenitors were found to be efficiently deleted 6 days after pI/pC administration (Figure 1B; supplemental Figure 1B). PB cell analyses showed that Chd8 depletion led to pancytopenia (Figure 1C). Hematoxylin and eosin staining of femur sections showed that Chd8^−/− mice had BM failure, with the total BM cellularity remarkably reduced compared with the WT control (Figure 1D-E). Consistently, we observed a decreased number of mature hematopoietic cells in the BM of Chd8^−/− mice (Figure 1F). As the likely result of severe anemia, Chd8^−/− mice showed deficient survival (supplemental Figure 1C). The spleen weight was also reduced in Chd8^−/− mice, but the white pulp structure in the spleen appeared normal (supplemental Figure 1D). These data indicate that CHD8 is critical for hematopoiesis and an acute loss of CHD8 results in rapid BM failure.

To examine how the rapid BM failure in Chd8^−/− mice occurs, we used flow cytometry to analyze hematopoietic progenitors. We found that CHD8-deficient mice had drastically fewer Lineage⁻cKit⁺Sca1⁺ (LSK) and Lineage⁻cKit⁺ (LK) cell populations (Figure 2A-B). Moreover, LSK cells combined with CD150 and CD48 analysis showed that SLAM HSCs (LSK CD150⁺CD48⁻) and multipotent progenitors MPP2 (LSK CD150⁺CD48⁺) and MPP3 (LSK CD150⁻CD48⁺) were profoundly depleted after CHD8 loss (Figure 2D-E). The common myeloid progenitors (CMPs, LK CD34⁺CD16/32⁻), granulocyte-macrophage progenitors (GMPs; LK CD34⁺CD16/32⁺), and megakaryocyte-erythroid progenitors (MEPs; LK CD34⁻CD16/32⁻) were nearly absent from Chd8^−/− BM (supplemental Figure 2A-B). Consistent with the loss of HSPCs in Chd8^−/−, we detected little colony-forming activity after Chd8 depletion (Figure 2C; supplemental Figure 2C). These results indicate that CHD8 is essential for the maintenance of HSPCs.

To determine the blood cell–intrinsic effects of CHD8 depletion, we harvested Chd8^F/F and Chd8^−/− BM cells 6 days after pI/pC injection and performed BM transplantation. Lethally irradiated recipients receiving CHD8-deficient BM cells died within 3 weeks (supplemental Figure 2D), manifesting a failure in HSPC engraftment and BM repopulation in the absence of CHD8. This is further evidenced by the observation that no donor CD45.2⁺Chd8^−/− cells were detected in 1:1 WT vs Chd8^−/−-competitive BM transplantations (supplemental Figure 2E-F). When a competitive BM transplantation was performed with an equal number of BM cells from CD45.2⁺Chd8^Flox;Mx1-Cre mice, Chd8^Flox littermates, and CD45.1 WT BM cells, to fully engraft in the lethally irradiated recipients, followed by pI/pC injection (Figure 2F), it is clear that CD45.2⁺ CHD8-deficient cells were rapidly outcompeted by CD45.1 WT cells over a 16-week period (Figure 2G). Analyses of the mature blood cells from these mice showed a dramatic loss of CHD8-deficient donor cells compared with Chd8^F/F cells (Figure 2H). In addition, the LK, LSK, and MPP subpopulations of the CD45.2⁺, CHD8-deficient cells were drastically reduced (Figure 2I). In the secondary and tertiary transplants, all Chd8^−/− cells were outcompeted by the WT competitors (Figure 2J-K). Moreover, we performed reciprocal transplantation by transplanting CD45.1 WT cells into lethally irradiated Chd8^Flox or Chd8^Flox;Mx1-Cre mice, followed by pI/pC injection. No detectable phenotypes from PB or BM cells were observed (supplemental Figure 2G-I). These results show that the hematopoietic failure caused by CHD8 loss is related to the intrinsic defects of CHD8-deficient HSPCs.

In mice, CHD8 haploinsufficiency could present autisticlike phenotypes. However, CHD8 haploinsufficiency in hematopoietic cells did not yield detectable phenotypes in the PB (supplemental Figure 3A). Transplantation or competitive transplantation of Chd8^+/− BM cells into WT recipients resulted in no significant changes in hematopoiesis compared with the Chd8^F/+ control (supplemental Figure B-G), indicating that CHD8 heterozygosity is insufficient to cause blood defects in mice.

To examine the underlying mechanism of HSPC loss in Chd8^−/− mice, we performed annexin V staining of the BM cells. There were significantly elevated annexin V⁺ SLAM HSC and LK cells after deletion of CHD8, which accounted for up to 80% of dying cells (Figure 3A-C). Western blot of cleaved caspase 3 in sorted LK cells showed dramatically increased apoptosis in Chd8^−/− cells compared with WT cells (Figure 3D). In addition, we observed increased apoptosis in mature Mac1⁺Gr1⁺ cells, but not in B220⁺ cells (supplemental Figure 4A-B). These results show that CHD8 is essential for the survival of HSPCs.

Next, we examined the proliferation of the Chd8^−/− HSPCs by Ki67 staining. CHD8 depletion led to a loss of quiescence in LSK cells, as there are more Chd8^−/− cells in the G₁ and S-G₂-M phases compared with WT control (Figure 3E-F). Ki67 staining of SLAM HSCs and MPPs further showed that CHD8 is necessary to maintain HSC and MPP quiescence (Figure 3G-H; supplemental Figure 4C), which is probably caused by a compensatory response to the Chd8^−/− caused by BM failure and pancytopenia.

To rule out a possible effect of pI/pC on HSPC survival, we examined the mice at a later time points after pI/pC injections. At 10 days, the hematopoietic phenotypes were similar to those observed at 6 days after pI/pC (supplemental Figure D-F), and the Chd8^−/− HSPCs still showed dramatic apoptosis (supplemental Figure 4G). The pancytopenia of Chd8^−/− mice extend to 8 weeks after pI/pC injection as the cell death effect in LK cells (supplemental Figure 4H-I). The surviving HSPCs showed a differentiation block and cell cycle progression block (supplemental Figure 4J-K). In addition, in vitro knockdown or Cre transduction assays found a significant increase of apoptosis and/or reduction in colony-forming activity in Chd8-deficient LSKs (supplemental Figure 5A-E). Administration of an IFNAR1 mAb to the cell culture did not restore the colony-forming unit–defects in the Chd8^−/− BM cells (supplemental Figure 5F).

CHD8 is an ATP-dependent chromatin remodeling enzyme that has roles in regulating chromatin accessibility. To determine whether CHD8 regulates chromatin accessibility in HSPCs, we performed assay for transposase-accessible chromatin sequencing (ATAC-Seq) with sorted LSK cells from Chd8^F/F and Chd8^−/− mice. We found a globally increased genomic accessibility after CHD8 loss (Figure 4A-B). We also performed CHD8 CUT&RUN (cleavage under targets & release using nuclease) assay in WT LSK cells and found that most of the peaks were in the gene body regions (supplemental Figure 6A-B). Chd8^−/− LSK cells showed increased chromatin accessibility at CHD8-binding sites compared with the Chd8^F/F cells (supplemental Figure 6C-D). Whereas previous work has found that CHD8 could recruit H3K4 histone methyltransferase to orchestrate oligodendrocyte lineage progression,²⁴  we did not detect any significant change in H3K4me3 or H3K27me3 modification in CHD8-depleted LSK cells in a CUT&RUN assay (supplemental Figure 6E-F). Consistent with these observations, we did not observe any changes at the protein level in the H3K4me3, H3K27Ac, and H3K27me3 chromatin markers caused by Chd8 depletion (Figure 4C; supplemental Figure 7A).

To elucidate the mechanism driving the HSPC loss, we performed RNA-seq in LSK cells of WT and Chd8^−/− mice. Triplicate samples from each genotype are well correlated (Figure 4D): more than 1000 genes are differentially expressed between WT and Chd8^−/− cells (Figure 4E). A decreased HSC signature and increased interferon α response pathway in Chd8^−/− mice were detected, but Wnt signaling appeared normal (Figure 4F; supplemental Figure 6G). Among the differential pathways seen in a Kyoto Encyclopedia Genes and Genomes (KEGG) analysis, P53 signaling was highly enriched and significantly increased in the Chd8^−/− HSPCs (Figure 4F; supplemental Figure 6G). The heat map of P53 target genes shows a clear pattern of the p53 signatures (Figure 4G). Further reverse transcription polymerase chain reaction (RT-PCR) analysis of LSK cells and SLAM HSCs found highly elevated P53 target genes, including P21 and Noxa but no change in P53 mRNA level (Figure 4H).

CHD8 could interact with P53 and suppress its transactivation during embryonic development.²⁹  To examine whether CHD8 plays a similar role in hematopoietic cells, we performed CHD8 immunoprecipitation in HPC7 cells and found that CHD8 formed a complex with P53 (Figure 5A). The CUT&RUN analysis of primary LSK cells shows that CHD8 binds directly to the gene body of Cdkn1a and a promotor region of Ccng1 (Figure 5B), suggesting that CHD8 plays a role in repressing P53 transactivation of its targets in HSPCs.

Interestingly, we found that the P53 protein level in CHD8-deficient LSK cells increased massively (Figure 5C; supplemental Figure 7B). Deletion of 1 allele of the P53 gene was not sufficient to restore the elevated P53 protein level in the Chd8^−/− cells (Figure 5D; supplemental Figure 7C). Alternatively, the excessive P53 protein in Chd8^−/− HSPCs could have been due to increased protein stability. To examine this possibility, we treated BM progenitors with MG132, a proteasome inhibitor and found that protease inhibition could significantly stabilize P53 in WT progenitors, but with a reduced effect on Chd8^−/− cells (Figure 5E; supplemental Figure 7D). In addition, cycloheximide (CHX) blockage of P53 translation showed that P53 protein degradation in Chd8^−/− progenitors was significantly slower (Figure 5F; supplemental Figure 7E). Consistently, the ubiquitinated P53 was also significantly reduced in Chd8^−/− low-density BM cells (Figure 5G; supplemental Figure 7F). We also examined the effect of CHD8 deletion in CD45−CD31⁺ BM endothelial cells and liver cells and found that CHD8 loss did not affect cell survival, the P53 protein level, or genomic stability of these cells from the same Mx1-Cre;Chd8^F/F mice used for the hematopoiesis studies (supplemental Figure 7G-L). Together, these results demonstrate that CHD8 loss leads to increased P53 protein stability and elevated P53 in HSPCs.

To determine the role of P53 in HSPCs in the Chd8-controlled hematopoiesis, we crossed P53^Flox mice to Chd8^Flox;Mx1-Cre mice and generated P53;Chd8 compounding deletion by pI/pC injection. CHD8 and P53 protein can be efficiently deleted in these mice (Figure 6A). Whereas deletion of 1 allele of P53 did not rescue the apoptosis of HSPCs (data not shown), deletion of both alleles of P53 readily restored cell counts of Mac1⁺Gr1⁺ and B220⁺ cells because of CHD8 loss (supplemental Figure 8B). In addition, total BM cellularity in the CHD8-depleted mice was significantly restored (Figure 6B; supplemental Figure 8A).

Next, we analyzed hematopoietic progenitors of the mice by flow cytometry. The numbers of both LK and LSK cells were restored in Chd8^−/−P53^−/− BM (Figure 6C), whereas the total cell counts of these populations showed an overcompensation compared with WT in Chd8^−/−P53^−/− mice (Figure 6D). The counts of committed progenitors and HSCs were also restored by P53 deletion (supplemental Figure 8D-F), even though the percentages of CMP and GMP in LK cells remained relatively lower. Consistent with these genotypic observations, the colony-forming unit activity of the in Chd8^−/− BM cells also recovered in the Chd8^−/−P53^−/− mice (supplemental Figure 8C). Functionally, the survival defects of SLAM HSC and LK cells were fully rescued by P53 deletion (Figure 6E). Competitive BM transplantations showed that the deletion of P53 in CD45.2⁺Chd8^Flox;Mx1-Cre mice enabled an effective engraftment of the CHD8-depleted HSPCs with the chimerism in PB and progenitors restored to that of WT (Figure 6F-G). Moreover, RNA-seq analysis of Chd8^−/−P53^−/− LSK cells showed that the P53 deletion mostly restored global gene expression in the Chd8^−/− cells (Figure 6H; supplemental Figure 8G). However, there remain ∽200 differentially expressed genes between WT and Chd8^−/−P53^−/− LSK cells, which may contribute to the overcompensation of some progenitors (supplemental Figure 9A-B). Thus, P53 regulated cell survival plays a central role in CHD8-mediated HSPC function and hematopoiesis.

Posttranslational regulation of P53 is important for p53 protein stability. Phosphorylation and acetylation of P53 could interrupt its binding to MDM2 and suppress P53 degradation. In the Chd8^−/− cells, we observed a significant increase of p-P53 at serine 15, whereas acetyl P53 K379 remained unchanged (Figure 7A; supplemental Figure 9C). In parallel, we saw an increased p-ATM, p-CHK2, and γH2A.X, but not p-ATR or reactive oxygen species (Figure 7A, supplemental Figure 10A-C), indicating an activated ATM signaling pathway in Chd8^−/− cells. Moreover, the increased p-ATM and DNA damage marker γH2A.X in Chd8^−/− cells was not rescued by compounding P53 deletion (Figure 7B; supplemental Figure 9D), suggesting that CHD8 suppresses genomic instability independent of p53. Consistent with increased γH2A.X expression, we observed increased γH2A.X foci in Chd8^−/− progenitors (Figure 7C). During chemostress, this phenomenon was exaggerated, producing a more intense γH2A.X immunofluorescence in Chd8^−/− progenitors (Figure 7C). To determine if CHD8 is involved in DNA damage response, we irradiated WT and CHD8-null cells at a 3-Gy dose and harvested cells after various times of recovery. The DNA damage response in CHD8 deleted cells, as indicated by p-P53 and γH2A.X expression, appears to be similar to that of WT cells in a 6-hour recovery period (Figure 7D). In addition, we found that the recruitment of BRCA1 to γH2A.X foci (Figure 7E), along with several other DNA damage repair factors, were not affected in Chd8^−/− progenitors (supplemental Figure 10D-F). The γH2A.X level was also repaired in Chd8^−/−P53^−/− progenitors (supplemental Figure 10G). These data show that, although CHD8 is required maintain genomic integrity, it is not involved in DNA damage repair.

To examine whether activated ATM contributes to the increased P53 protein level, we administered an ATM inhibitor, KU55933, to BM progenitors. KU55933 significantly decreases p-P53 and P53 levels dose dependently in the Chd8^−/− cells (Figure 7F; supplemental Figure 9E), suggesting that the elevated ATM activity in CHD8-cKO cells is responsible for P53 elevation. To better understand how ATM is involved, we performed a coimmunoprecipitation test with HPC7 cells and found that CHD8 and ATM directly interact with each other in either a CHD8 immunoprecipitation or an ATM immunoprecipitation (Figure 7G). Moreover, the activated p-ATM appeared to lose interaction with CHD8 upon irradiation (Figure 7H). A knockdown of CHD8 in these cells mimicked the DNA damage response effects with increased pP53 and γH2A.X expression as in Chd8^−/− HSPCs (supplemental Figure 10H). Thus, CHD8 forms a complex with both ATM and p53 and maintains genomic stability by integrating the ATM-p53 signaling.

CHD8 is an ATP-dependent chromatin remodeling enzyme that is essential for brain development, and its mutations represent a subclass of autism in humans,^22,24  but its role in hematopoiesis remains unknown. In this study, we found that loss of CHD8 led to dramatic loss of HSPCs, because of increased apoptosis driven by P53 activation, and loss of p53 rescued the survival defects in Chd8^−/− HSPCs. Moreover, our findings revealed that CHD8 is essential for maintaining the genomic integrity of HSPCs by forming a complex with ATM and p53 and regulating ATM-p53 mediated signaling and cell survival. We demonstrated that CHD8 is essential for normal hematopoiesis.

Aside from neurodevelopmental defects, autism is associated with anemia and other blood disorders.^30,31  To date, no information is available for the underlying mechanism of autism-associated blood abnormalities beyond disease-related food intake/iron deficiencies.³²  In a subpopulation of patients with autism, heterozygous Chd8 mutations, mostly loss of function changes, are considered functionally significant. Although we observed severe BM failure in Chd8^−/− mice, we did not detect significant hematopoietic phenotypes in mice with a 1-allele deletion of Chd8, perhaps because of mouse vs human differences or involvement of other compounding genetic factors in autism. Nonetheless, our finding of an essential function of CHD8 in hematopoiesis raises a possible mechanism of the blood cell intrinsic role of Chd8 in autism beyond defective food intake and iron deficiencies.

HSPC survival is a critical function for the maintenance of the stem/progenitor cell pool, and in this study, CHD8 was found to be essential for this process. One of the well-known functions of CHD8 is as a negative regulator of P53.²⁹  Indeed, our results showed overactivation of P53 signaling in Chd8^−/− mice. Consistent with previous findings, we observed a direct binding between P53 and CHD8 in HSPCs and found that CHD8 was recruited to P53 target promotors, such as Cdkn1a and Ccng1. More interestingly, we observed dramatic accumulation of P53 protein in Chd8^−/− HSPCs, without a detectable mRNA change. Further analysis showed that CHD8 also interacted with ATM, and loss of CHD8 led to activation of the ATM signaling pathway, which resulted in increased P53 phosphorylation and decreased P53 ubiquitination. The role of CHD8 in P53 protein stabilization has not been seen in other cell types, including U2OS osteosarcoma and endothelial cells. We propose a model in which, during normal hematopoiesis, CHD8 maintains the HSPC cell survival by integrating ATM and p53 to restrict P53 protein stability and regulate P53 transactivation (Figure 7I). Consistent with what we observed in HSPCs, CHD8 contributes to B-ALL cell survival,²⁷  suggesting an essential role of CHD8 in maintaining hematopoietic cell survival. CHD8 is also a negative regulator of the Wnt-β catenin signaling pathway in multiple cell types other than blood lineages,^33,34  but we did not observe significant changes of this pathway in HSPCs, suggesting that CHD8 involvement in restricting Wnt pathway does not apply to HSPCs.

Chromatin structure and dynamics are key factors in the maintenance of genomic integrity,³⁵  and the CHD family chromatin remodeling enzymes are emerging players.³⁶  Unlike the function of CHD8 in chromatin opening in oligodendrocyte progenitors,²⁴  we detected globally increased chromatin accessibility after CHD8 loss in HSPCs by ATAC-seq, suggesting that CHD8 is important for chromatin compaction rather than opening. The relaxation of chromatin may be causal to the increased sensitivity to a DNA damage response in Chd8^−/− HSPCs. Multiple CHD proteins have been found to act in regulating DNA damage response and repair.^37-41  In our study, however, unlike other CHD proteins, the DNA damage repair response appears normal in Chd8^−/− HSPCs as indicated by decreased γH2A.X after irradiation and the timely recovery of DNA damaged foci in both Chd8^−/− and Chd8^−/−P53^−/− HSPCs. This suggests that CHD8 is mainly required for chromatin integrity but is not involved in DNA damage repair per se. Our and other studies have established that the biological role and the p53 stabilization/ATM sequestration mechanism of CHD8 is specific to HSPCs as CHD8 loss in BM endothelial cells, and liver cells, and other lineages do not yield similar effects, as in blood cells.^42-44 

Overall, our study discovered an HSPC-specific role of CHD8 in regulating hematopoiesis by maintaining chromatin integrity and genomic stability through a CHD8-ATM-p53 complex.

The authors thank James Johnson for offering technical assistance; the Research Flow Cytometry Core, the Comprehensive Mouse and Cancer Core, Biomedical Informatics Core, DNA Core, and the Pathology Core at Cincinnati Children’s Hospital Medical Center for their services and technical support.

This work was partially supported by National Institutes of Health, National Cancer Institute grants R01CA234038 and R01CA204895.

Contribution: Z.T., Q.R.L., and Y.Z. designed the study; Z.T. performed the experiments, interpreted the data, and generated the figures; C.W., A.K.D., M.H., C.Z., and M.X. performed the experiments; Z.T. and Y.Z. wrote the manuscript; and Y.Z. provided funding support.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Yi Zheng, Cincinnati Children‘s Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229; e-mail: yi.zheng@cchmc.org.