EPC1/2 regulate hematopoietic stem and progenitor cell proliferation by modulating H3 acetylation and DLST

Summary Enhancers of polycomb 1 (EPC1) and 2 (EPC2) are involved in multiple biological processes as components of histone acetyltransferases/deacetylase complexes and transcriptional cofactors, and their dysfunction was associated with developmental defects and diseases. However, it remains unknown how their dysfunction induces hematopoietic stem and progenitor cell (HSPC) defects. Here, we show that depletion of EPC1/2 significantly reduced the number of hematopoietic stem and progenitor cells (HSPCs) in the aorta-gonad mesonephros and caudal hematopoietic tissue regions by impairing HSPC proliferation, and consistently downregulated the expression of HSPC genes in K562 cells. This study demonstrates the functions of EPC1/2 in regulating histone H3 acetylation, and in regulating DLST (dihydrolipoamide S-succinyltransferase) via H3 acetylation and cooperating with transcription factors serum response factor and FOXR2 together, and in the subsequent HSPC emergence and proliferation. Our results demonstrate the essential roles of EPC1/2 in regulating H3 acetylation, and DLST as a linkage between EPC1 and EPC2 with mitochondria metabolism, in HSPC emergence and proliferation.


INTRODUCTION
5][6] Acetylation of histones and non-histones is involved in normal and malignant hematopoiesis by modulating chromatin status and adjusting the function of non-histone protein. 7,8Histone acetyltransferases (HATs) mainly include multiple families: MYST (MYST1 (KAT8), Tip60, MOZ (KAT6A), MORF, HBO1 (KAT7), MOF), p300/CBP, GNAT (PCAF, GCN5 (KAT2A), ELP3 (KAT9)), and nuclear (or steroid) receptor coactivator (NCOA1(KAT13A), NCOA2, NCOA3, and CLOCK), 8,9 etc. Loss of CREB binding protein (CBP) function has been reported to cause defects in primitive hematopoiesis 10 and disrupt HSPC development in mice. 11,12Lack of MOZ (KAT6A) HAT activity has been shown to cause defects in HSPCs 13 and accelerate HSPC apoptosis. 146][17] As transcriptional co-factors, EPC interact with RET finger protein (RFP) to repress gene transcription, and recruit homeodomain-only protein (HOP) or serum response factor (SRF) to induce skeletal muscle differentiation, 18,19 and regulate transcriptional activity by interacting with E2F transcription factor 1 (E2F1). 17EPC1 and EPC2 are also related to acute myeloid leukemia (AML), which can directly or indirectly suppress the accumulation of MYC and apoptosis of AML cells, thereby maintaining oncogenic potential. 20Based on morpholino-mediated gene knockdown, a large-scale reverse genetic screen of 425 human chromatin factors in zebrafish found that EPC2 (as an epigenetic factor) is required for hematopoietic development, especially primitive erythropoiesis. 21However, how hematopoietic cells develop in zebrafish with deficiency of gene EPC1 or EPC2 and the related underlying mechanisms still remain unknown.
The purpose of this study was to identify the effects, potential mediators, and underlying mechanisms of EPC1 and EPC2 in hematopoietic development by a zebrafish genetic model with deletion of the genes epc1a or epc2, and human K562 cells with knockdown of functional EPC1 or EPC2.We demonstrated that the functional deficiency of EPC1 or EPC2 led to impaired proliferation and reduced levels of acetylated H3 proteins in HSPCs.Importantly, we identified DLST and acetylated H3 proteins as downstream targets of EPC1 and EPC2 to positively regulate HSPC emergence and proliferation.

Epc1a and epc2 are indispensable for embryonic HSPC emergence
It has been reported that knockdown of epc2 induce primitive hematopoiesis defects in zebrafish, 21 while the hematopoiesis, especially the development of HSPCs, in embryos with either epc1a or epc2 deficiency is still unknown.In this study, we found that zebrafish epc1a and epc2 were ubiquitously and predominantly expressed in the brain and eyes, with expression in the intermediate cell mass (ICM) and AGM (Figure S1A), suggesting their potential involvement in zebrafish hematopoiesis.Epc1a À/À (with an 11-bp deletion in exon 1), epc2 À/À (with a 4-bp deletion in exon 1), and epc1a À/À epc2 À/À mutants constructed in this study all exhibited mild developmental delay, with no obviously morphological defects observed during embryogenesis (Figures S1B and S1C).Compared with the control, the deletion of either epc1a or epc2 induced significantly downregulated expression of the target gene in embryos (p < 0.01), respectively (Figure S1D), which was also verified by western blot (WB) (Figure S1E).Meanwhile, epc1b (homolog of epc1a and epc2) expression was significantly increased (p < 0.01) in both epc1a À/À and epc2 À/À mutants (Figure S1D).
To explore whether epc1a and epc2 deficiency affect the development of HSPCs in zebrafish, we detected the expression of HSPC markers runx1/cmyb in the mutants firstly.Whole-mount in situ hybridization (WISH) results displayed that compared with the control, epc1a À/À and epc2 À/À mutants showed reduced expression (p < 0.001) of runx1 and cmyb in the AGM at 33 hpf (Figure 1A) and in the CHT region at 72 hpf Figure 1.Loss of epc1a and epc2 perturbed the development of hematopoietic stem and progenitor cells (HSPCs) in zebrafish (A and B) Expression of HSPC genes cmyb and runx1 in the AGM (A1-A6) and CHT regions (B1-B6) in the control, epc1a À/À , and epc2 À/À embryos at 33 hpf and 72 hpf, and quantification of WISH data (A7 and A8, B7 and B8), with red arrowheads indicating positive signals.
In this study, in runx1 + cells sorted from Tg (runx1: GFP) embryos at 33 hpf, 23 runx1/cmyb exhibited abundant expression (p < 0.001), in contrast to obvious low expression of the neural gene olig2 and the muscle gene myoD (p < 0.001) (Figure 1D), suggesting the HSPC identity of runx1 + cells.The abundant expression of epc2 and approximate expression of epc1a were observed in runx1 + cells (Figure 1D), further supporting their potential involvement in HSPC development.Depletion of either epc1a or epc2 markedly decreased (p < 0.001) the percentage of runx1 + cells relative to the control group (Figure 1E).However, in Tg (epc1a À/À ; runx1: GFP), epc1a exhibited significantly reduced expression (p < 0.001) while epc2 increased in runx1 + cells and vice versa, and epc1b expression was markedly upregulated in epc1a-or epc2-deficient runx1 + cells (Figure 1F).Meanwhile, the ectopic expression of epc1 mRNA not only restored the decrease of runx1/cmyb in the AGM region at 33 hpf in epc1a À/À embryos, but also partially restored the decrease of runx1/cmyb in epc2 À/À embryos, and vice versa (Figure S4), suggesting there might be no off-target effects occurred in each of mutants and that epc1a and epc2 may function redundantly during HSPC development.
Histone H3 acetylation and dlst are key regulators for epc1a and epc2 modulation of HSPC emergence and proliferation As epigenetic factors, EPC1 and EPC2 have been found to play important roles in the histone acetylation/deacetylation process, [26][27][28] and involve in zebrafish primitive erythropoiesis via NuA4 or HDAC-NuRD complexes. 21In this study, acetylation-related Gene Ontology (GO) terms, such as internal protein amino acid acetylation, histone acetyltransferase complex, and histone H4 acetylation, were seen to be significantly enriched in epc1a À/À mutants versus the control (Figure S7A; Table S13).The levels of H3K9Ac, H3K27Ac, H3K56Ac (Figure 3A; Figure S7B), and H4K5/K8/K12/K16Ac (Figure S7C) were clearly downregulated (p < 0.01), while H3K4Me1 protein level showed no change (Figure S7D) and H3K27Me3 increased (Figure S7E), with loss of functional epc1a or epc2 relative to the control group.The fluorescence intensity of H3K9Ac, H3K27Ac, and H3K56Ac in runx1 + HSPC cells showed a marked decrease (p < 0.001), and the number of runx1 + cells was decreased significantly in the mutants versus the control (Figures 3B, S7F, and S7G).Meanwhile, depletion of epc1a or epc2 in zebrafish caused significant reduction (p < 0.05) of H3K27Ac (as a marker of active gene promoters and enhancers) 29 enrichment on promoters of cmyb (Figure 3C1) and runx1 (Figure 3C2).

Roles of EPC1 and EPC2 are conserved in human hematopoietic cells
In a previous study, EPC1 and EPC2 were shown to play critical roles in sustaining AML stem cell potential. 20Here, we inactivated EPC1 and EPC2 in K562 cells (K562 line is derived from a chronic myeloid leukemia/erythroleukemia patient, 37,38 and commonly used as an in vitro model to study hematopoietic development) 23,39 By lentivirus-mediated shRNA knockdown 40 in constructing shEPC1 or shEPC2 functional deficiency K562 cells, effective targets were screened by enrichment score based on qRT-PCR of the transcript level of hEPC1 or hEPC2.Compared with the control, ShEPC1-2 targeting EPC1 (abbreviated as shEPC1 in the following) and ShEPC2-3 targeting EPC2 (abbreviated as shEPC2) exhibited stable and obvious knockdown activities (p < 0.001) (Figures S13A and S13B), and K562 cells with stable expression of ShEPC1-2 (shEPC1) and of ShEPC2-3 (shEPC2), respectively, were selected and maintained as stable knockdown cell lines in the following tests.Both EPC1 and EPC2 were seen to be significantly downregulated (p < 0.001) in ShEPC1/2 (both knockdown of EPC1 and EPC1 via both ShEPC1-2 and ShEPC2-3 in K562 cells) stable cell lines (Figure S13C).Cell proliferation in K562 cells was significantly weakened (p < 0.001) by knockdown of either EPC1 or EPC2 versus the control (Figure S13D), consistent with the tendency in epc1a or epc2 mutant.Meanwhile, EPC1 or EPC2 knockdown induced a marked increase (p < 0.001) in cell apoptosis (Figure S13E).These data indicated that human erythroleukemia cells may depend on EPC1 and EPC2 for proliferation and anti-apoptosis defense.
The expressions of most hematopoietic genes were clearly reduced (p < 0.001) upon either EPC1 or EPC2 knockdown relative to the control (Figures S13F and S13G).The transcription profiles were further analyzed by RNA-seq upon knockdown of either hEPC1, hEPC2, or hEPC1/2.ShEPC1, ShEPC2, and ShEPC1/2 cells were seen to be downregulated in the expression of genes related to fatty acid metabolic process, transition metal ion homeostasis, and acyl-CoA biosynthetic process (Figure 5A; Table S13), and ShEPC1, ShEPC2, and ShEPC1/2 cells showed remarkable enrichment (p < 0.001) in GO terms related to hematopoiesis and acetylation (Figure 5B; Figure S13H; Tables S14 and S15), which was further confirmed by RT-qPCR results (Figures S13I and S13K; Table S16).
In the absence of either EPC1 or EPC2 or both, K562 cells exhibited remarkable reduction (p < 0.001) in H3K9Ac (Figure 5C1; Figure S14A1), H3K27Ac (Figure 5C2; Figure S14A2), and H3K56Ac (Figure 5C3; Figure S14A3), with remarkable enrichment (p < 0.001) in GO terms related to acyl-CoA biosynthetic process (Figure 5A) and histone acetylation, histone H2A acetylation, etc. (Figure 5B), while H3K27Me3 showed increased level (Figure S14B).Meanwhile, the decrease of acetylated histones and hematopoiesis-related genes could be partially restored by treatment with VPA (Figures 5D and 5E; Figure S14C and S14D), which was in line with the results in zebrafish (Figures S7H and Figures 3D-3F).Furthermore, knockdown of EPC1 or EPC2 or both caused obvious downregulation (p < 0.001) of DLST protein in K562 cells versus the control (Figure 5F) and obvious downregulation in binding enrichment of H3K27Ac on promoter of gene DLST (Figure 5G), and the decreased DLST expression could also be rescued by VPA treatment (Figure 5H).Overall, our findings supported the conserved roles of EPC1 and EPC2 in hematopoietic development via regulation of histone H3 acetylation and DLST, and DLST expression was modulated by histone H3 acetylation, which might be specific in hematopoiesis.

EPC1 and EPC2 co-operate with histone acetylation factors KATs and BRPF1 in modulating DLST
Previous studies have shown the involvement of EPC1 and EPC2 in HAT complexes, 26,27,41 and the aforementioned results showed that VPA could rescue the expression of DLST, which acted as a downstream gene of EPC1 and EPC2 in runx1 + HSPCs and K562 cells, suggesting that the transcription of DLST might be modulated via histone acetylation.In this study, we found that acetylation-related genes, such as ep300a, kat8, ncoa2, and gfi1a/b, were highly expressed in runx1 + HSPCs (Figure S15B).Meanwhile, epc1a-or epc2-deficient HSPCs and embryos both exhibited marked changes in the expressions of a series of the histone acetylation genes, such as kat2a, kat6a, kat7b, ncoa1, kat9, and brpf1 (p < 0.001) compared with their expression in the control (Figure 6A; Figure S15A).We speculated that the upregulated expressions of acetylation-related genes in EPC1/2 complexes in HSPCs might prefer to maintain the complex functions with either EPC1 or EPC2 functional deficiency, as genetic compensation response (GCR) reported recently. 42,43Thus, using DLST promoter activities and protein levels of .Continued (I) Double staining of runx1 + and BrdU in the CHT region in the control, epc1a À/À , and epc2 À/À embryos and the corresponding groups injected with dlst mRNA at 72 hpf (I1-I24), and quantification of runx1 + cells (I25) and runx1 + BrdU + cells (I26), with white arrowheads indicating double-positive cells.I4, I8, I12, I16, I20, and I24 show the magnified images for I3, I7, I11, I15, I19, and I23, respectively.Each experiment was repeated three times, and a representative result is shown.N changed /N total in the right bottom corner of each panel indicates embryos with changed expression/total tested embryos, and n in calculation panels indicates the number of total tested embryos in each group.All embryos are shown in lateral view, anterior to the left, and dorsal to the up.Scale bars, 100 mm.Data are presented as mean G SD (n R 3).t test, *p < 0.05, **p < 0.01, ***p < 0.001, NS, not significant.iScience Article H3K27Ac as indicators, we tested the interactions of EPC1/2 with the candidate histone acetylation-related regulators (Figure 6A, red arrowhead).The transcriptional activities of the DLST promoter were seen to be significantly activated (p < 0.01) through collaboration of histone acetylation writers KAT2A (GCN5), KAT6A (MYST3/MOZ), KAT12 (GTF3C4) or KAT13A (NCOAI), and histone acetylation reader BRPF1 (Figure 6B; Figure S15C), with either EPC1 or EPC2 (Table S17), coupled with increased levels of acetylated proteins in the co-transfected cells (Figure 6C; Figure S15D).Taken together, EPC1 and EPC2 may modulate histone acetylation by collaborating with histone acetylation regulators KATs and BRPF1, then to regulate DLST expression, and subsequently regulate HSPC development.

EPC1 and EPC2 transcriptionally regulate DLST in cooperation with SRF or FOXR2
5][46][47] The potential protein interaction between EPC and SRF or FOXR2 (forkhead box R2) was found by retrieving GeneCards database (https://www.genecards.org/).In this study, the transcriptional levels and protein levels of SRF and FOXR2 were significantly decreased in EPC1-or EPC2deficient runx1 + cells (Figure S16A) and K562 cells (Figures S16B and S16C).The binding sites of SRF on hDLST (human DLST) and zdlst (zebrafish dlst) promoter region has been reported and referenced in many studies, [48][49][50] and the FOXR2 binding sequence were predicted by consulting JASPAR database (http://jaspar.genereg.net/)(Table S18; Figure S17). 51To verify the predicted results, we performed dual-luciferase reporter experiments and observed that EPC1 and EPC2 significantly enhanced (p < 0.001) the DLST transcriptional activities by cotransfection with SRF or FOXR2 (Figure 7A).Co-immunoprecipitation assays further confirmed the interaction of EPC1 and EPC2 with SRF or FOXR2 (Figures 7B and 7C).Mutational analysis of the DLST promoter region showed that deletion of the predicted SRF binding site (CGTAAAAAGG) significantly decreased (p < 0.001) the transcriptional activity of DLST promoter when EPC1 or EPC2 was co-transfected with SRF, but deletion of the predicted FOXR2 binding sites in this study had no influence on the its transcriptional activity when co-transfecting EPC1 or EPC2 with FOXR2 (Figures 7D and 7E; Figure S18A).Furthermore, ChIP-qPCR assay indicated that knockdown of EPC1 and EPC2 led to significant decrease (p < 0.001) in EPC1/EPC2 and SRF/FOXR2 enrichments at the predicted SRF binding site and FOXR2 binding sites of DLST promoter in K562 cells (Figures 7F; Figures S18B and S18C).However, in zebrafish knock out embryos, their enrichments at the predicted SRF binding site were reduced (Figures S18C1 and S18C2), but remained unchanged at the predicted FOXR2 binding sites (Figures S18C3-S18C6).
In addition, we tried to delete the SRF or FOXR2 binding sites at zebrafish dlst promoter (Table S19) via the CRISPR-Cas9 system, and tested emergence of HSPCs in the F 0 mutants.We effectively got a line of F 0 mutants carrying an 8 bp insertion as shown in Figure S19A, which led to an increasing number of possible SRF motif (CArG [48][49][50] ), and the F 0 mutants exhibited increased cmyb expression at 33 hpf () and runx1 + BrdU + cells (Figure S19C) compared with the WT control.Meanwhile, ectopic expression of srf and foxr2 mRNA could partially recover the decrease of cmyb at 33 hpf (Figure S20A) and runx1 + BrdU + cells at 72 hpf (Figures S20B-S20C) in epc1a À/À and epc2 À/À mutants.
Collectively, these results suggested that EPC1 and EPC2 may recruit SRF or FOXR2 to transcriptionally activate DLST expression, and the predicted SRF/FOXR2 binding site might be the potential target of EPC1/EPC2 and SRF/FOXR2 complex on DLST promoter specifically in HSPCs.

DISCUSSION
EPC1 and EPC2 are indispensable for HSPC emergence and proliferation EPC1 and EPC2 have been reported to play essential roles in leukemic hematopoiesis 20 and involve in primitive erythropoiesis in zebrafish, 21 but how the deficiency of EPC1 and EPC2 induces hematopoiesis developmental defects is unknown.Here, we show that EPC1 and EPC2 are indispensable for HSPC emergence and proliferation via regulation of histone H3 acetylation and modulation of DLST expression.Apart from demonstrating EPC1 and EPC2 in histone H3 acetyltransferases complexes to regulating HSPC emergence and proliferation, this study unveils an important linkage of EPC1/2 with a TCA enzyme DLST then with HSPCs, and unveils the HSPC-specific role of DLST, downstream of EPC1/2 to link the epigenetic factors with mitochondrial metabolism, regulates HSPC emergence and proliferation.(B) GO analysis of DEGs involved in acetylation and hematopoiesis in control and ShEPC1 K562 cells (B1), and in control, ShEPC1, ShEPC2, and ShEPC1/2 K562 cells (B2), respectively, based on RNA-seq data.(C) Western blotting analysis of the protein levels of H3K9Ac (C1), H3K27Ac (C2) and H3K56Ac (C3) in control, ShEPC1, ShEPC2, and ShEPC1/2 cells, with actin and PCNA as the internal controls.(D) Immunofluorescence analysis of the control, ShEPC1, ShEPC2, and ShEPC1/2 K562 cells and corresponding groups treated with VPA.(D1) using the anti-H3K27Ac antibody, and quantification of fluorescence intensity (D2).(E) qRT-PCR expression analysis of RUNX1, CMYB, GATA1, and GATA2 in control, ShEPC1, ShEPC2, and ShEPC1/2 cells as well as the corresponding groups treated with VPA.(F) Western blotting analysis of the protein level of DLST in control, ShEPC1, ShEPC2, and ShEPC1/2 cells, with actin as the internal controls (F1), and quantification results (F2).(G) DLST expression in control, ShEPC1, ShEPC2, and ShEPC1/2 cells as well as corresponding groups treated with VPA.(H) ChIP-qPCR analysis of H3K27Ac occupancy at DLST promoters in the control, ShEPC1, ShEPC2, and ShEPC1/2 cells.The occupancy was presented as the percentage of input.Each experiment was repeated three times, and a representative result is shown.Scale bars, 5 mm, Data are presented as mean G SD (n R 3).t test, *p < 0.05, **p < 0.01, ***p < 0.001, NS, not significant.
A series of observations in this study support the essential roles of EPC1 and EPC2 in HSPC development.First, the expressions of the HSPC markers runx1/cmyb are markedly decreased in epc1a-or epc2-deficient embryos and runx1 + HSPCs.Second, depletion of either epc1a and epc2 results in impaired proliferation, malformed spindles, and impaired cell cycle in runx1 + HSPCs.Third, knockdown of either EPC1 or EPC2 or both restrains cell proliferation in K562 cells.These observations are consistent with the reports that EPC1 or EPC2 are involved in cell cycle progression 26,52 and cell proliferation 53 in yeast and mammalian cells, and unveil the essential roles of EPC1 and EPC2 in HSPC proliferation.Knockdown of EPC1 or EPC2 induced cell apoptosis in K562 cells rather than in embryonic HSPCs, not only agreeing with the reports that knockdown of EPC1 and/or EPC2 induced apoptosis in AML cells, 20 but also suggesting that roles of EPC1 and EPC2 in apoptosis are different between in vivo and in vitro cells.Collectively, our data shed light on the evolutionarily conserved roles of EPC1 and EPC2 in hematopoietic development.
Additionally, HSPC defects were more severe in epc1a À/À epc2 À/À embryos, and the ectopic expression of epc2 mRNA partially restores the decrease of runx1/cmyb in epc1a À/À embryos and vice versa, suggesting epc1a and epc2 may play redundant roles in HSPC development.Some studies have confirmed that organisms can upregulate related family genes to response to mutations by genetic compensation mechanism. 42,43In this study, the transcriptional adaptations between the three genes (epc1a, epc1b, and epc2) in runx1 + HSPCs and whole embryos are also observed.
EPC1 and EPC2 regulate HSPC emergence and proliferation by modulating histone H3 acetylation and epigenetically regulating DLST EPC1 and EPC2, the components of the EP400 and Tip60-EP400 complexes, have been reported to activate gene expression by stimulating H3.3 deposition into promoters and enhancers in U2OS human osteosarcoma cells. 54Here, we show that EPC1 and EPC2 could positively modulate the levels of histone acetylation in embryonic HSPCs and K562 cells, especially H3 histones, such as H3K9Ac, H3K27Ac, and H3K56Ac.Histone deacetylase inhibitors VPA could efficiently curb these inhibitory effects on the levels of acetylated histones and expressions of HSPC genes (CMYB and RUNX1) in either EPC1-or EPC2-deficient K562 cells and embryonic HSPCs, confirming that EPC1 and EPC2 contribute to H3 histone acetylation and the subsequent development of hematopoietic cells.Epl1 (yeast homologous gene of EPC1) depletion has been shown to induce the global loss of acetylated histone H4 and H2A in yeast. 26In this study, enriched GO terms of histone H4 acetylation and decreased H4K5/K8/K12/K16Ac level were observed in epc1a À/À or epc2 À/À mutant, and numerous genes regulated histone acetylation, such as ep300/crbbpa, ncoa1, clocka, and kat2b, which are capable of acetylating histone H2A/B or H4, 55 had a remarkable decrease in epc1a-and epc2-deficient HSPCs.Moreover, epc1a-and epc2-deficient embryonic cells show increased H3K27Me3 protein and normal H3K4Me1, but with significant reduction in the levels of H3K9Ac, H3K27Ac, and H3K56Ac which also exhibit significantly reduced expression in EPC1-and EPC2-deficient K562 cells, further supporting that histone H3 acetylation and methylation are regulated in a seesaw manner in cells, 56,57 and demonstrating the important roles of both EPC1 and EPC2 in histone H3 acetylation in HSPCs apart from their roles in H2 or H4 acetylation.
Here, we show that (1) functional deficiency of EPC1 and EPC2 induces the downregulation of DLST specifically in zebrafish embryonic HSPCs and K562 cells; (2) dlst morphants and mutants also show reduced HSPC emergence and proliferation; and (3) ectopic expression of dlst could effectively recover impaired HSPC proliferation in epc1a À/À and epc2 À/À mutants.These suggest the specific role of DLST as a downstream target of EPC1 and EPC2 in regulating HSPC proliferation, which may widen the role of DLST besides its role in MYC-mediated leukemogenesis 32 and neuroblastoma, 33 and also widen the roles of EPC1 and EPC2 in regulating HSPC development via DLST related mitochondria metabolism.Meanwhile, we show that VPA could effectively recover the reduced DLST expression in embryonic HSPCs and mammalian K562 cells induced by the functional deficiency of either EPC1 or EPC2, suggesting EPC1 and EPC2 may regulate DLST via histone acetylation machinery to cooperate with chromatin factor complexes.
Moreover, we demonstrate that writers in histone acetylation, such as GNAT family members, KAT2A and KAT9, MYST family members, KAT6A and KAT7, NCOA1 family member KAT13A, other HAT family member KAT12A and reader BRPF1, could significantly elevate the levels of acetylated H3 proteins and the transcriptional activities of DLST by co-transfection with EPC1 or EPC2.These findings not only support that EPC1 and EPC2 could positively regulate DLST via histone acetylation machinery in HSPCs specifically, but also suggest that the aforementioned proteins might be the potential co-components in EPC1 or EPC2 histone acetylation machinery.Meanwhile, these results are supported by the previous studies about the functions of KAT2A, 58,59 ELP3, 60 KAT6A, 61 KAT7, 62 KAT8, 63,64 NCOA1, 65 and BRPF1 66,67 in hematopoiesis.However, it is still unknown regarding the specific containing EPC1 and EPC2 histone acetylation complex in HSPCs, especially the specific writers of histone acetylation and HSPC factors for EPC1 and EPC2 histone acetylation machinery in HSPC proliferation.
Moreover, we show that deficiency of epc1a and epc2 induces reduced dlst specifically in definitive HSPCs while increased dlst in the whole embryos and trunk muscle, with increased meox1 in the whole epc1a À/À embryos.Meox1 is well-known in regulating somite-derived Figure 6.EPC1 and EPC2 cooperate with histone acetylation regulators (A) Expression of genes in histone acetylation machinery in runx1 + cells at 33 hpf, with red arrows indicating the genes verified by luciferase reporter assays in this study.(B) The transcriptional activities of the DLST promoter through co-transfection of EPC1 and EPC2 with hKAT7, hNCOA1 (B1), hKAT2A (B2), hKAT6A (B3), hKAT8, hKAT9 and hKAT12 (B4), and hBRPF1 (B5) in 293T cells, respectively.(C) Western blotting analysis of acetylated histone H3K27Ac in cells co-transfected of EPC1 or EPC2 with hNCOA1, hKAT7 (C1 and C2), hKAT2A, hKAT6A (C3 and C4), hKAT8 (C5 and C6), hKAT9 (C7 and C8), hKAT12 (C9 and C10), and hBRPF1 (C11 and C12), respectively, with PCNA as the internal controls.Data are presented as mean G SD (n R 3).t test, *p < 0.05, **p < 0.01, ***p < 0.001, NS, not significant.endothelial cells and the subsequent hematopoietic stem cell induction via restricting endotome which contributes to dorsal aorta colonization. 68We speculate that increased meox1 might expand other somite cell types at the expense of definitive HSPC contributor endotome, resulting in the increased dlst in somites while reduced dlst in definitive HSPCs.Similarly, knockdown of EPC1 or EPC2 results in downregulation of DLST protein level and transcriptional level in K562 cells (as an in vitro model of hematopoietic development 23,39 ), further confirming the function of DLST specifically in the development of hematopoietic cells, and EPC1/EPC2 positively regulate DLST expression specifically in hematopoiesis.
EPC1 and EPC2 bind with SRF or FOXR2 to regulate DLST transcriptionally EPC1 and EPC2 have been reported to interact with various complexes 69,70 or bind with FOXO (forkhead box protein O) to trigger the differentiation of multipotent hematopoietic progenitors in Drosophila. 46EPC1 has been shown to regulate neointima formation after arterial balloon injury by interacting with Myocd (myocardin) and SRF. 71Here, we show that EPC1 and EPC2 could significantly enhance DLST transcriptional activity by co-transfection with SRF or FOXR2, and the direct interactions are observed between EPC1/EPC2 and SRF or FOXR2, suggesting the potential role of transcriptional complex of EPC1/EPC2 with either SRF or FOXR2 in DLST expression and the subsequent HSPC proliferation.It has been reported that as DNA-binding transcriptional factors, SRF is involved in the development of megakaryocytes, 72 neutrophils, 73 and HSPCs, 74 and FOXR2 is associated with tumorigenesis. 75,76In this study, we identify the binding sites of SRF or FOXR2 at DLST promoter by dual-luciferase assays and ChIP-qPCR, and find that EPC1/EPC2 contact with SRF to effectively recognize the predict SRF binding sequence CGTAAAAAGG in DLST promoter to active its expression.Additionally, F 0 mutants carrying an 8 bp insertion with an increasing number of possible SRF motif (CArG) on gene dlst promoter exhibit increased cmyb expression and runx1 + BrdU + cells at 33 hpf, further demonstrating that Epc1/2 might cooperate with SRF to regulate Dlst expression and the subsequent HSPC emergence.Meanwhile, EPC1/EPC2 contact with FOXR2 to recognize the predict FOXR2 binding sequence in DLST promoter effectively in HSPCs while not in the whole embryos and in 293T cells, suggesting the difference and complex of EPC1/EPC2 and FOXR2 interaction in regulating DLST.
Based on the aforementioned findings, we proposed a model for EPC1 and EPC2 regulation of HSPC proliferation through direct modulation of histone acetylation machinery by interacting with KAT2A/KAT6A/KAT12/KAT13A, and regulation of DLST by recruiting SRF/FOXR2 and modulating histone acetylation machinery, with DLST as a downstream target of EPC1 and EPC2 to facilitate HSPC emergence and proliferation.Numerous studies have proved that metabolic pathways, such as fatty acid oxidation, 77,78 glycolysis, 79,80 oxidative phosphorylation (OXPHOS), [81][82][83] and glutamine metabolism, 34,84,85 play key roles in HSPC commitment and erythroid differentiation. 86The mitochondrial citrate metabolism (TCA cycle), has also been reported to regulate hematopoietic development through OXPHOS 36 and reactive oxygen species (ROS). 87DLST, as one of a-KGDH enzymatic complex components, its inactivation could lead to significant alterations in TCA cycle and ROS in triple-negative breast cancer, 88 and suppressed NADH production and impaired OXPHOS in high-risk neuroblastoma. 33Hence, we speculate the downregulation of DLST resulted from loss of EPC1 and EPC2 may cause HSPC defect by disturbing OXPHOS and ROS generation controlled by TCA cycle.In addition, acetyl-CoA, as one of metabolite in TCA cycle, play important roles in cell proliferation, 89 cell cycle progression, 90 and cell differentiation 91 via regulation of histone acetylation.In this study, GO terms related to acetyl-CoA were remarked enrichment in ShEPC1, ShEPC2, and ShEPC1/2 cells based on RNA-seq data, with the decreased expression of DLST.Therefore, in this study, the positive feedback loop between DLST and histone acetylation may be formed to regulate HSPC development, and DLST might function to be a pivotal linkage between EPC1/EPC2 with mitochondrion metabolism in HSPC development.
In this study, an 18 bp insertion in exon 8 in dlst gene induces almost no Dlst protein could be detected in dlst À/À mutants, and we predict that non-frameshifting (NFS, an insertion or deletion of a multiple of three nucleotides) might occur in the mutants.NFS in coding regions has been reported to affect protein function, and cause a series of diseases and developmental defects in human, 89,[92][93][94][95] in mice, 94 in rabbit, 96 in horses, 97 in quail, 98 and so on.In this study, the NFS insertion of 18 bp fragment in the exon 8 of gene dlst in the mutants resulted in alteration of DLST protein structure predicted by SWISSMODEL, 99 which may change its protein enzyme activity and stability, and then affect TCA cycle.
In this study, loss of EPC1 and EPC2 are unveiled to impair the development of hematopoietic cells, which are conserved from fish to mammalian.Meanwhile, EPC1 and EPC2 are unveiled to function in histone H3 acetylation apart from their roles in H2 or H4 acetylation in HSPCs via co-operating with histone acetylation writers KAT2A, KAT9, KAT6A, KAT7, KAT13A, KAT12A and reader BRPF1, but the HSPC-specific histone acetylation writer and reader in EPC1 and EPC2 acetyltransferase complexes still wait for discovery.Moreover, EPC1 and EPC2 are unveiled to positively regulate DLST expression specifically in HSPCs, and DLST links EPC1 and EPC2 with mitochondrial metabolism in HSPC proliferation.

Limitations of the study
In this study, we show that EPC1 and EPC2 regulate hematopoietic development by histone H3 acetylation machinery and DLST.However, there still are some unanswered questions regarding the HSPC-specific histone acetylation writers or readers in EPC1 and EPC2 acetyltransferase complexes, and the different molecular mechanisms between EPC1 and EPC2 in regulating hematopoietic development, these questions need to be addressed in future studies.

RNA interference
Four ShRNAs specific to human EPC1 and three ShRNAs specific to human EPC2 were synthesized and cloned into the pLKO.1-purovector using the primer sequences shown in Table S4, and were used for screen effective ShRNAs in knocking down expression of EPC1 or EPC2 in K562 cells in this study.Lentiviruses for different ShEPC1, ShEPC2, or ShGFP controls were generated by co-transfecting the packaging plasmid pCMV-dR8.91,envelope plasmid VSV-G/pMD2G and hairpin-pLKO.1 into 293T cells.K562 cells were infected with the lentivirus and then screened by puromycin (2-5 mg/mL) for at least 48 h.The efficiency of EPC1 or EPC2 knockdown was verified by qRT-PCR, with the primers for qRT-PCR listed in Table S8 and the ShGFP cells as the controls. 40Based on the data of knockdown efficiency of different ShEPC1 or ShEPC2, shEPC1-2 (abbreviated as shEPC1), shEPC2-3 (abbreviated as shEPC2) and shEPC1/2 (knockdown of both EPC1 and EPC2 with shEPC1-2 and shEPC2-3) were screened and further maintained as stable knock-down cell lines and were used for the next tests in this study.

Drug treatment
Zebrafish embryos from the control, epc1a À/À and epc2 À/À mutants were treated with 0.2 mM valproic acid (VPA) (Cat# V298968, aladdin, China) beginning at 11 hpf.Next, the treated embryos were collected at 33 hpf or 72 hpf.Meanwhile, the cultured cells from control, ShEPC1, ShEPC2 and ShEPC1/2 cells were incubated with the similar deacetylase inhibitors VPA at 37 C for 16-18 h, followed by collecting the cell samples for the following experiments.In this study, the zebrafish embryos or larvae were collected at the indicated stages for experiments as shown in Schema 1.

Morpholino (MO) and mRNA injection
The morpholinos of epc1a, epc2 and dlst were purchased from Gene Tools LLC (Philomath, Oregon, USA) and dissolved in ddH 2 O at 3 mM concentration (stock solution).The sequences of MOs are listed in Table S5.The dlst MO sequence has been reported previously. 21The fulllength epc1a, epc2 and dlst were amplified with the specific primers shown in Table S6, and synthesized using the Ambion MAXIscript T7 Kit (Cat# AM1344, Invitrogen, USA) as instructed by the manufacturer.In all experiments, the MOs and mRNAs were injected into one-cell stage embryos, with the MO doses of epc1a, epc2 and dlst at 0.6 mM, 0.9 mM and 0.6 mM, and the mRNA concentrations of epc1a, epc2, dlst, srf and foxr2 at 150 ng/mL, 150 ng/mL, 200 ng/mL, 200 ng/mL and 200 ng/mL, respectively.

RNA-sequencing (RNA-Seq) and analysis
In this study, fifty embryos of control and epc1a À/À mutants at 28 hpf, and K562 cells named control, ShEPC1 (EPC1 knockdown), ShEPC2 (EPC2 knockdown), and ShEPC1/2 (both EPC1 and EPC2 knockdown), were collected separately and used for mRNA extraction.The cDNA library construction and the next RNA sequencing (RNA-Seq) was performed and run on a BGISEQ-500 and Illumina NovaSeq 5000 platforms, respectively.Sequenced reads were 150 bp long with paired-ends.After filtering the raw reads by SOAPnuke soft (v1.5.2), 105 the sequencing data were mapped to zebrafish reference genome (GRCz11) using the HISAT (v2.0.4) and Bowtie2 (v2.2.5) tool. 25,106,107RNA abundance was expressed as the number of FPKM (fragments per kilobase of transcript per million).DEGs were defined using the DEGSeq method following the criteria: |fold-change| R 2 and adjusted p value<= 0.001. 108Enriched Kyoto Encyclopedia of Genes and Genomes pathway analysis was conducted for each sample using KOBAS v.2.0 based on the lists of DEGs.GO analysis was conducted using the lists of DEGs by GOseq Release 2.12.Hierarchical clustering was performed by TIGR Multi Experiment Viewer to generate different heatmaps.The sequencing data were deposited in GSA database (https://ngdc.cncb.ac.cn/gsa/), 109,110 and accession numbers are listed in the key resources table.

One step cell-direct qRT-PCR
One Step Cell-Direct qRT-PCR followed the protocol of the manufacturer.In this study, Tg (runx1: GFP) (as control group), Tg (epc1a À/À ; runx1: GFP), and Tg (epc2 À/À ; runx1: GFP) embryos were collected and homogenized in PBS supplemented with 5% FBS, and then centrifuged at 1000 3 g for 10 min at 4 C. Next, the precipitate was resuspended in PBS and subsequently filtered through a 0.2 mm filter to make single-cell suspension.The GFP-positive cells (5000-10000 sorted cells/sample) were sorted into the lysis solution provided by the
were washed with PBS and analyzed by WB analysis.The Amersham Imager 600 analyzer was used for photographing the blots.ImageJ software was used for quantifying the protein levels based on the band density obtained in the WB analysis.

Statistical analysis
The sample size was larger than 10 embryos (n > 10) for WISH and immunofluorescence, 50 embryos for RNA and protein extraction, and 100-150 embryos for FACS in each group.Two to three biological replicates were performed for each assay.Percentage analysis of the results among different groups was performed using hypergeometric distribution in R-console software. 118The signals of WISH, WB and immunofluorescence images were quantified using ImageJ software (NIH, Bethesda, Maryland), and the calculated data were analyzed using t test by GraphPad Prism 8.0 software.Each dot represents the signal level of a representative image in an individual embryo in each group.The qPCR data were analyzed by one-way analysis of variance (ANOVA) and post hoc Tukey's test in the IBM Statistic Package for Social Science Statistics for Windows, Version 22 (Released 2013; IBM Corp., USA) software.Each dot represents one repeat.
The number of HSPCs and HSPC proliferation were quantified from the confocal images of the AGM (470 mm 3 470 mm) and CHT (470 mm 3 470 mm) with z-stacks spanning the entire trunk thickness, and the number of flk1 + runx1 + and runx1 + BrdU + cells were manually counted using ImageJ software.At least ten randomly selected units were analyzed for each control group and experimental group.Also, the fluorescence intensity of acetylated histone protein in a single runx1 + cell or K562 cell was also quantified from the confocal images (22.15 mm 3 22.15 mm) with z-stacks spanning the entire cell thickness, and the fluorescence intensity of histone protein was measured using ImageJ.The statistical data of the signal area and fluorescence level in different samples were analyzed using t-test by GraphPad Prism 8 software, with each dot representing the signal level in an individual embryo or one cell in each group.The statistical analysis of the luciferase reporter assay results was performed using GraphPad Prism 8 software (unpaired t test).Statistically significant differences among groups are indicated by p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***).A criterion of R1.5-fold change (1.5-fold increase/decrease, p value %0.05) was used to define obvious/marked changes when comparing differences between groups in this study.

Figure 4
Figure 4. Continued (I) Double staining of runx1 + and BrdU in the CHT region in the control, epc1a À/À , and epc2 À/À embryos and the corresponding groups injected with dlst mRNA at 72 hpf (I1-I24), and quantification of runx1 + cells (I25) and runx1 + BrdU + cells (I26), with white arrowheads indicating double-positive cells.I4, I8, I12, I16, I20, and I24 show the magnified images for I3, I7, I11, I15, I19, and I23, respectively.Each experiment was repeated three times, and a representative result is shown.N changed /N total in the right bottom corner of each panel indicates embryos with changed expression/total tested embryos, and n in calculation panels indicates the number of total tested embryos in each group.All embryos are shown in lateral view, anterior to the left, and dorsal to the up.Scale bars, 100 mm.Data are presented as mean G SD (n R 3).t test, *p < 0.05, **p < 0.01, ***p < 0.001, NS, not significant.

Figure 7 .
Figure 7. EPC1 and EPC2 transcriptionally regulate DLST expression by interacting with SRF and FOXR2 (A) The transcriptional activities of the DLST promoter through co-transfection of EPC1 and EPC2 with hSRF or hFOXR2 in 293T cells, respectively.(B and C) CoIP analysis of the interaction between hEPC1, hEPC2, hSRF, and hFOXR2.IP, immunoprecipitated; IB, immunoblot; WCL, whole cell lysate.(D) Schematic illustration of the predicted binding sites of SRF and FOXR2 in DLST promoter region (D1), and DLST promoter (+13-1864 bp) mutated with SRF and FOXR2 binding sites named control, FOXR2 site1 mut, FOXR2 site2 mut, and SRF site1 mut, respectively (D2), the mutated regions are marked with arrows.(E) Luciferase reporter assays of DLST promoter with mutated SRF binding site.(F) ChIP-qPCR analysis of EPC1 and EPC2 occupancies at predicted SRF binding site and FOXR2 binding site in DLST promoter in the control, ShEPC1, ShEPC2, and ShEPC1/2 cells.The occupancy was presented as the percentage of input.Each experiment was repeated three times, and a representative result is shown.Data are presented as mean G SD (n R 3).t test, *p < 0.05, **p < 0.01, ***p < 0.001, NS, not significant.

TABLE d
B Zebrafish husbandry and lines B Cell lines d METHOD DETAILS B Generation of zebrafish mutants using CRISPR/Cas9 B RNA interference B Drug treatment B Morpholino (MO) and mRNA injection