HHEX promotes myeloid transformation in cooperation with mutant ASXL1

Additional sex combs-like 1 (ASXL1), an epigenetic modulator, is frequently mutated in myeloid malignancies including acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), myeloproliferative neoplasms, and chronic myelomonocytic leukemia. ASXL1 mutations are associated with poor prognosis in various myeloid malignancies.^1-3 ASXL1 mutations in myeloid neoplasms typically occur near the last exon where the majority of the mutations are frameshift or nonsense mutations. As a result, C-terminally truncated forms of mutant ASXL1 (ASXL1-MT) are produced.⁴  Wild-type ASXL1 interacts with various epigenetic regulators, such as polycomb repressive complex 2 and BRCA1-associated protein 1 (BAP1).^5,6  ASXL1-MT also interacts with these epigenetic components and induces aberrant histone modifications in a dominant-negative or gain-of-function manner.^7-10  It has also been reported that ASXL1-MT, but not wild-type ASXL1, interacts with BRD4 to activate genes related to myeloid transformation.¹¹  Thus, these epigenetic changes induced by ASXL1-MT contribute to the development of myeloid neoplasms.

We recently generated ASXL1-MT conditional knockin (ASXL1-MT-KI) mice.¹²  ASXL1-MT-KI mice showed mild anemia, differentiation block of erythrocytes, myeloid skewing, and dysplasia. These features are indicative of early stage MDS. However, ASXL1-MT-KI mice never developed myeloid leukemia in 2 years, suggesting that additional elements are required for myeloid transformation of ASXL1-mutated cells. Our previous study on retrovirus-mediated insertional mutagenesis showed the susceptibility of ASXL1-MT-KI bone marrow (BM) cells to transform into myeloid leukemia.¹²  In the previous study, we identified the hematopoietically expressed homeobox (HHEX) gene as one of the common retrovirus integration sites.¹² 

HHEX is a homeodomain-containing gene, which was primarily isolated from hematopoietic tissue.¹³ HHEX encodes a highly conserved homeobox protein, HHEX protein, which contains a proline-rich homeodomain and is considered as a repressive transcriptional factor.^14,15  Loss of HHEX displays embryonic lethality because of impairment of vasculogenesis and heart tissue.^16,17  In fetal hematopoiesis, HHEX deletion results in reduced production of CD45⁺ hematopoietic cells from ES cells.¹⁸  In adult hematopoiesis, previous studies have shown that HHEX is highly expressed in hematopoietic stem-progenitor cells (HSPCs) and leukemia cells.¹⁹  In this context, overexpression of HHEX represses CDKN2A and blocks differentiation of promyelocytes.^19,20  Moreover, HHEX is essential for B lymphopoiesis but dispensable for hematopoiesis at a steady state.^21-23  In stress hematopoiesis, HHEX depletion leads to defects of T lymphopoiesis.²³  In contrast, HHEX expression negatively regulates regulatory T cells by inhibiting Foxp3 expression.²⁴  However, it remains unknown whether HHEX contributes to myeloid transformation in ASXL1-mutated cells, and how HHEX promotes leukemogenesis remains to be elucidated.

In this study, we showed functional cooperation between ASXL1-MT and HHEX in myeloid leukemogenesis either by themselves or in combination with FLT3-ITD and RUNX1-ETO. We also identified MYB and ETV5 as directly regulated genes by HHEX and demonstrated the pivotal roles of MYB and ETV5 in ASXL1-mutated myeloid leukemia.

Gene expression and correlation analyses of HHEX, ETV5, and MYB were generated using RNA sequencing datasets of the Beat AML study,²⁵  which were kindly provided by the Oregon Health & Science University to the Cleveland Clinic Foundation. Data from The Cancer Genome Atlas (TCGA) were also analyzed.

All animal experiments were approved by the Animal Care Committee at the Institute of Medical Science, the University of Tokyo (approval numbers: K15-32, PA13-19, PA17-75, PA16-31, PA18-46). All experiments using human cord blood (CB) cells were approved by the Ethics Committee at the Institute of Medical Science, the University of Tokyo (approval number: 27-34-1225). Details for other methods are provided in the supplemental Methods, available on the Blood Web site.

To investigate the role of ASXL1-MT and HHEX in myeloid transformation, we first performed colony-forming assay using c-Kit⁺ BM cells derived from control mice or ASXL1-MT-KI mice. Forced expression of HHEX enhanced colony-replating activity in cells derived from ASXL1-MT-KI mice, whereas it showed only a modest effect in normal myeloid progenitor cells (Figure 1A). The morphology of the colonies derived from cells expressing ASXL1-MT and HHEX is more compact than that derived from other genotypes (supplemental Figure 1A). In liquid culture, the murine cells expressing ASXL1-MT and HHEX also enhanced proliferation and blocked myeloid differentiation (Figure 1B-C; supplemental Figure 1B). We further evaluated the effect of ASXL1-MT and HHEX on proliferation using human CD34⁺ CB cells. Expression of ASXL1-MT and HHEX synergistically enhanced proliferation and colony-forming unit–granulocyte/macrophage (CFU-GM)–replating activity, whereas it impaired colony-forming unit erythroid (CFU-E) and burst-forming unit-erythroid (BFU-E) formation (Figure 1D-F; supplemental Figure 1C-E). To characterize the cell cycle profiles in ASXL1-MT– and HHEX-expressing CB cells, we assessed cell cycle and apoptosis status. Expression of ASXL1-MT and HHEX increased the proportion of S/G2/M-phase cells and decreased the Annexin V⁺ fraction (Figure 1G-H). Thus, expression of ASXL1-MT and HHEX induced increased proliferation of murine and human HSPCs via promoting cell cycle progression and inhibiting apoptosis.

Next, we assessed the effect of ASXL1-MT and HHEX on myeloid differentiation. Expression of HHEX reduced expression of differentiation markers including CD66b and CD16b (Figure 1I). In addition, expression of either ASXL1-MT or HHEX reduced expression of CD14 and CD117, but retained expression of CD34 in human CD34⁺ CB cells. We found that ASXL1-MT cooperated with HHEX in further inhibiting differentiation (Figure 1I-J). We also observed that ASXL1-MT and HHEX jointly inhibited calcitriol-induced monocytic differentiation block in a human leukemia cell line, HL-60 (supplemental Figure 1F-G). These data indicated that ASXL1-MT and HHEX cooperatively enhanced proliferation and impaired differentiation of murine and human myeloid progenitor cells in vitro.

To evaluate impacts of ASXL1-MT and HHEX on hematopoiesis in vivo, we transduced a control vector or HHEX into c-Kit⁺ BM cells derived from a control or ASXL1-MT-KI mice (Figure 2A). We then transplanted them into sublethally irradiated recipient mice. Five of 12 mice transplanted with cells expressing ASXL1-MT and HHEX developed AML, whereas those transplanted with cells expressing HHEX or ASXL1-MT alone never developed AML (Figure 2B). The AML mice induced by ASXL1-MT and HHEX displayed hepatosplenomegaly (Figure 2C; supplemental Figure 2A). The AML cells displayed blastic morphology and the following immunophenotypes: c-Kit⁺, CD11b^low, B220⁻, and CD3⁻ (Figure 2D-E; supplemental Figure 2B). The surviving mice transplanted with cells expressing ASXL1-MT and HHEX displayed mild leukocytopenia, anemia, macrocytosis, and thrombocytosis (Figure 2F). Unlike mice that developed AML, the chimerism of the ASXL1-MT– and HHEX-expressing cells in the whole BM was not high (Figure 2G). In addition, expression of either ASXL1-MT or HHEX showed myeloid skewing both in the peripheral blood (PB) and BM, and expression of ASXL1-MT and HHEX increased the frequency of CD11b⁺Gr-1⁻ cells (Figure 2H-I). Consisting with in vitro experiments, expression of either ASXL1-MT or HHEX exhibited the immature morphology and the increase of granulocyte-monocyte progenitor (GMP) and Lineage⁻ Sca-1⁻ c-Kit⁺ fractions, and expression of ASXL1-MT and HHEX showed further differentiation block (Figure 2J-K; supplemental Figure 2C-D). These data indicated that expression of ASXL1-MT and HHEX cooperatively inhibited myeloid differentiation of HSPCs in vivo, but it was insufficient to complete myeloid transformation.

Next, we examined possible cooperation between ASXL1-MT and HHEX in myeloid leukemogenesis in the presence of other mutations. Clinical data showed that ASXL1 mutations and RUNX1-ETO (also called AML1-ETO) fusion frequently cooccur in AML patients.²⁶  We previously showed that CB cells expressing RUNX1-ETO can grow over 6 months in culture retaining primitive CD34⁺ cells, and recapitulate many features of human RUNX1-ETO leukemia.^27,28  Therefore, we transduced a control vector or ASXL1-MT (coexpressing green fluorescent protein [GFP]) and another control vector or HHEX (coexpressing nerve growth factor receptor [NGFR]) into RUNX1-ETO–expressing CB cells and then traced the frequency of GFP⁺/NGFR⁺ double-positive fraction by fluorescence-activated cell sorting (FACS; supplemental Figure 3A). Cells expressing ASXL1-MT or HHEX moderately enhanced and cells coexpressing ASXL1-MT and HHEX dramatically enhanced proliferation of RUNX1-ETO–expressing CB cells (Figure 3A; supplemental Figure 3B-C). Expression of ASXL1-MT and HHEX increased the proportion of S/G2/M-phase cells and decreased the Annexin V⁺ fraction (Figure 3B-C). In addition, most of the ASXL1-MT– and HHEX-transduced RUNX1-ETO–expressing cells retained an immature phenotype on day 30 (Figure 3D). Thus, ASXL1-MT and HHEX enhanced proliferation of RUNX1-ETO–expressing CB cells in vitro.

Next, we investigated the synergistic effects of ASXL1-MT and HHEX on leukemogenesis in vivo using RUNX1-ETO9a, which is a shorter isoform of RUNX1-ETO and has a stronger leukemogenic activity.²⁹  We retrovirally transduced RUNX1-ETO9a (coexpressing Thy1.1) together with a control vector or HHEX (coexpressing NGFR) into c-kit⁺ fetal liver cells derived from Vav-Cre^+/− or Vav-Cre^−/− ASXL1-MT-KI, after which we transplanted them into sublethally irradiated recipient mice (Figure 3E). Consistent with our previous report,⁵  cells expressing ASXL1-MT showed susceptibility to myeloid leukemia driven by RUNX1-ETO9a (Figure 3F). Expression of HHEX further enhanced leukemogenesis induced by RUNX1-ETO9a and ASXL1-MT (Figure 3F-G; supplemental Figure 4A-B). Expression of ASXL1-MT and HHEX increased the frequency of c-Kit⁺ fraction in RUNX1-ETO9a AML cells (Figure 3H; supplemental Figure 4C). Moreover, we investigated the collaborating roles of ASXL1-MT and HHEX in another leukemia model, named FLT3-ITD leukemia. c-Kit⁺ BM cells derived from control or ASXL1-MT-KI mice were transduced with FLT3-ITD (coexpressing Kusabira Orange [KuO]) together with a control vector or HHEX followed by transplantation into sublethally irradiated mice (supplemental Figure 5A). In line with a previous report,³⁰  HHEX expression enhanced the leukemogenicity of FLT3-ITD (Figure 3I). Expression of ASXL1-MT and HHEX more rapidly induced the development of myeloid leukemia driven by FLT3-ITD (Figure 3I; supplemental Figure 5B-C). HHEX expression increased the frequency of CD11b⁺Gr-1⁻ cells and combined expression of ASXL1-MT and HHEX increased c-Kit⁺ fraction in FLT3-ITD AML cells (Figure 3J; supplemental Figure 5D). These data indicated that ASXL1-MT and HHEX cooperatively promote leukemogenesis induced by RUNX1-ETO and FLT3-ITD.

We next assessed whether endogenous HHEX promotes myeloid leukemogenesis of leukemia cells expressing ASXL1-MT. We previously generated 2 distinct myeloid leukemia cells, cSAM cells (cells with combined expression of SETBP1 and ASXL1 mutations) and cRAM cells (cells with combined expression of RUNX1 and ASXL1 mutations).^12,31,32  We first transduced Cas9 to cSAM cells and cRAM cells (hereafter referred to as cSAM/Cas9 cells and cRAM/Cas9 cells), and then either a nontargeting control vector (sgNT) or Hhex-targeting sgRNAs (sgHhex #1, sgHhex #2) into cSAM/Cas9 cells and cRAM/Cas9 cells, followed by performing colony-forming assay and mouse transplantation (Figure 4A-B). We observed that Hhex depleted cSAM/Cas9 cells, and cRAM/Cas9 cells reduced cell cycle progression and promoted differentiation and apoptosis (Figure 4C-D; supplemental Figure 6A-E). Importantly, depletion of Hhex both in cSAM/Cas9 cells and cRAM/Cas9 cells dramatically reduced colony-forming activity and leukemogenicity (Figure 4E-F).

Next, we assessed the role of HHEX in human leukemia cell lines with or without ASXL1 mutations. We here used 4 ASXL1-mutated cell lines (MEG-01, Kasumi-1, TS9;22, and NOMO-1) and 4 ASXL1-wildtype cells lines (THP1, U937, HL-60, and SKNO-1), respectively. We confirmed that CRISPR-Cas9-mediated depletion of ASXL1 inhibited growth of ASXL1-mutated cell lines, whereas ASXL1-wildtype cells showed only minor reduction of growth by ASXL1 depletion (supplemental Figure 7A-E). HHEX depletion attenuated the growth of all these ASXL1-mutated cell lines (Figure 4G-H). In contrast, the growth of ASXL1-wildtype leukemia cells was relatively insensitive to HHEX depletion (supplemental Figure 6F-G). We also confirmed that HHEX depletion promoted differentiation and apoptosis in MEG-01 and Kasumi-1 cells (Figure 4I-K; supplemental Figure 6H). These data indicated that endogenous HHEX promoted the survival of leukemia cells expressing ASXL1-MT via inhibiting differentiation and apoptosis.

To explore whether HHEX depletion affects CD34⁺ CB cells, we performed CRISPR-Cas9-mediated depletion of HHEX in human CD34⁺ CB cells by electroporation (supplemental Figure 8A). An sgRNA targeting CD45 was used as a control (supplemental Figure 8B). Both sgRNAs targeting HHEX (sgHHEX #1, sgHHEX #2) efficiently depleted HHEX protein (supplemental Figure 8C). We observed no significant differences for CD34 expression (supplemental Figure 8D), colony-forming activity (supplemental Figure 8E-F), myeloid cell expansion (supplemental Figure 8G), apoptosis (supplemental Figure 8H), cell cycle status (supplemental Figure 8I), and differentiation (supplemental Figure 8J-K) between control and HHEX-depleted CB cells. Thus, HHEX depletion has little impact on the proliferation or differentiation of CD34⁺ human CB cells.

To investigate how HHEX expression contributes to myeloid transformation, we performed RNA-seq analysis using RUNX1-ETO–expressing human CB cells (supplemental Figure 3A). We transduced a control vector or ASXL1-MT (coexpressing GFP) and a control vector or HHEX (coexpressing NGFR) into RUNX1-ETO–expressing CB cells. The cells expressing both GFP and NGFR were sorted by FACS. Total RNA was then extracted from the cells and subjected to RNA-seq analysis. We identified 1424 upregulated and 1433 downregulated genes in the HHEX-expressed cells (Figure 5A). By analyzing the enrichment of Gene Ontology (GO) biological process terms in these differentially expressed genes, we found that the upregulated genes are significantly involved in the process of cell proliferation and cell cycle (Figure 5B). In contrast, the downregulated genes are related to the process of apoptosis and differentiation (Figure 5B). Gene set enrichment analysis (GSEA)^33,34  revealed that genes related to leukemia stem cells were more enriched in cells expressing both ASXL1-MT and HHEX compared with control cells (Figure 5C), which was consistent with our observations in RUNX1-ETO9a and FLT3-ITD leukemia models shown in Figure 3F,I. On the contrary, genes related to monocytic differentiation were less enriched in cells expressing both ASXL1-MT and HHEX than control cells (Figure 5C). This also is consistent with our in vitro results that ASXL1-MT and HHEX inhibited monocytic differentiation shown in Figure 1I and supplemental Figure 1F-G.

To investigate the regulatory effect of HHEX bindings, we next performed chromatin immunoprecipitation (ChIP) against an anti-HA antibody using 293T cells transfected with HA-tagged HHEX followed by next-generation sequencing (ChIP-seq). We found that HHEX bound mainly to the intergenic region and promoter region (Figure 5D). In addition, HHEX binds to AT-rich loci, which resembles specific homeobox factors including Hoxb5 and Mnx1 (Figure 5E; supplemental Figure 9A-B). From the ChIP-seq analysis, we identified that HHEX bound to the promoter regions of 5202 genes (Figure 5F). The 5202 genes included 25% of the differentially expressed genes; 301 upregulated genes and 400 downregulated genes (Figure 5F). These results imply that HHEX regulates target gene expression by binding at their promoters.

To elucidate underlying molecular mechanisms by which the combination of ASXL1-MT and HHEX promotes myeloid leukemogenesis, we investigated target genes directly regulated by ASXL1-MT and HHEX. As ASXL1-MT induced gene upregulation via reducing H2AK119ub modifications,⁵  we focused on the upregulated genes by HHEX binding to promoter loci (Figure 5F). Among 301 upregulated genes, we found that MYB and ETV5 are upregulated by ectopic expression of HHEX in murine c-Kit⁺ cells and HL-60 cells (Figure 6A; supplemental Figure 10A). Meanwhile, Hhex depletion decreased expression of MYB and ETV5 in ASXL1-mutated AML, cSAM cells, cRAM cells, and MEG-01 cells, but not in ASXL1-wildtype AML THP1 cells (Figure 6B; supplemental Figure 10B-D). To further evaluate ASXL1-MT/BAP1 complex, we analyzed our previous RNA-seq data and found that ASXL1-MT/BAP1 complex increased MYB/ETV5 expression, but not HHEX expression (supplemental Figure 10E). In fact, depletion of ASXL1-MT increased the global level of H2AK119ub and reduced the expression of MYB and ETV5 in MEG-01 cells (supplemental Figure 10F-G). Consistent with mRNA expression data shown in Figure 6A, HHEX expression resulted in increased expression of MYB and ETV5 proteins in c-Kit⁺ BM cells (supplemental Figure 10H). On the contrary, HHEX depletion led to decreased expression of MYB and ETV5 proteins in cSAM/Cas9 cells (supplemental Figure 10I). Thus, HHEX upregulated MYB and ETV5 by binding promoter loci, which would be maintained by ASXL1-MT expression.

To explore the clinical implications of HHEX, MYB, and ETV5 expression in leukemia cells, we analyzed mRNA expression in human primary AML cells. HHEX, MYB, and ETV5 are highly expressed in primary AML patients regardless of the presence or absence of ASXL1 mutations (Beat AML)²⁵ ; this indicates that high expression of HHEX, MYB, and ETV5 could contribute to the survival of AML cells (supplemental Figure 10J). Next, we investigated the correlation between HHEX and MYB/ETV5 mRNA expression in primary AML samples in the Beat AML cohort and found that HHEX and MYB/ETV5 mRNA were positively correlated (Figure 6C). These positive correlations were also confirmed in another cohort from TCGA³⁵  (supplemental Figure 10K).

Next, we assessed whether the upregulation of Myb and Etv5 are directly regulated by ASXL1-MT and HHEX using ChIP-seq assay. ChIP-seq using 293T cells revealed that HHEX directly bound to Myb and Etv5 loci including the promoter region (Figure 6D). To validate this result in hematopoietic cells as well, we performed ChIP-qPCR assay using HL-60 cells expressing HA-tagged HHEX. HHEX bound to the Myb and Etv5 promoter region 2 kb upstream of the transcriptional start site (Figure 6E). Additionally, the level of H2AK119ub at Myb and Etv5 at the promoter loci was decreased (Figure 6F). Importantly, we found that expression of ASXL1-MT increased the binding of endogenous HHEX toward the Myb and Etv5 promoter loci (Figure 6G). HHEX expression did not change the global level of H2AK119ub (supplemental Figure 11A). In addition, we did not detect physical interaction between ASXL1-MT and HHEX (supplemental Figure 11B). To assess whether binding of HHEX alters the activity of the promoter at MYB and ETV5, we performed a reporter assay. Binding of HHEX increased the activity of promoter at MYB and ETV5, which was enhanced by the expression of ASXL1-MT and BAP1 (Figure 6H; supplemental Figure 11C). Thus, ASXL1-MT and HHEX directly upregulated the expression of MYB and ETV5 via activating promoter activity at these genes.

To assess the roles of Myb and Etv5 in ASXL1-mutated leukemia cells, we depleted Myb and Etv5 using the CRISPR-Cas9 system in cSAM/Cas9 cells and cRAM/Cas9 cells. We first generated 2 independent sgRNAs each targeting murine Myb or Etv5. These sgRNAs efficiently depleted Myb or Etv5 protein both in cSAM/Cas9 cells and cRAM/Cas9 cells (Figure 7A,D). Depletion of Myb increased apoptosis and decreased colony-forming activity in both cSAM/Cas9 and cRAM/Cas9 cells (Figure 7B-C). Etv5 depletion promoted differentiation in cSAM/Cas9 and cRAM/Cas9 cells, thereby reducing their colony-forming activity (Figure 7E-F). These growth-inhibitory effects of MYB or ETV5 depletion were also observed in human ASXL1-mutated leukemia cell lines (supplemental Figure 12A-B). Thus, endogenous MYB and ETV5 play important roles in promoting the proliferation of ASXL1-mutated leukemia cells.

Finally, we investigated whether MYB and ETV5 are essential targets of HHEX in myeloid leukemogenesis. To this end, cSAM/Cas9 cells were first transduced with control vector, MYB, or ETV5 (coexpressing KuO). Then, KuO⁺ cells were then sorted and transduced with sgNT or Hhex-targeting sgRNAs followed by a colony-forming assay (Figure 7G-H). Intriguingly, ectopic expression of MYB or ETV5 ameliorated the reduction of colony-forming activity of cSAM/Cas9 cells induced by Hhex depletion (Figure 7I). Altogether, these data indicated that the upregulation of MYB and ETV5 are critical for leukemogenesis driven by ASXL1-MT and HHEX.

In the previous study, we primarily identified HHEX as a candidate gene whose expression collaborates with ASXL1-MT in cellular transformation.¹²  In this study, we confirmed the functional collaboration between ASXL1-MT and HHEX. The combination of ASXL1-MT and HHEX induced AML with long latencies in about half of the transplanted mice, and it also promoted leukemogenesis induced by FLT3-ITD and RUNX1-ETO. Expression of HHEX promoted the myeloid transformation of ASXL1-mutated cells by inhibiting their differentiation and apoptosis and activating cell cycle progression, whereas differentiation and apoptosis/cell cycle arrest cannot be completely separated. Additionally, as shown in supplemental Figure 10J, we found that primary AML patients harboring ASXL1 mutations have higher HHEX expression compared with healthy controls. These data indicated that ASXL1-mutated preleukemic clones that have high expression of HHEX may be biologically selected and eventually transformed into leukemia cells, possibly with acquisition of additional mutations.

Moreover, we demonstrated that endogenous HHEX is required for the maintenance of ASXL1-mutated leukemia. In contrast, depletion of HHEX showed marginal growth inhibition in ASXL1-wildtype leukemia used in this study except for SKNO-1 cells (harboring a RUNX1-ETO fusion). Given that HHEX expression is generally high in AML, it is possible that HHEX could collaborate with various mutations other than ASXL1-MT, including RUNX1-ETO. We also found that HHEX-depleted CD34⁺ CB cells showed normal myeloid proliferation and differentiation, which is consistent with the previous reports that HHEX is dispensable for adult myelopoiesis in mice.¹⁹  Thus, HHEX could be a potential therapeutic target in AML, in particular ASXL1-mutated AML. In addition, the results presented in this study demonstrate that ASXL1-MT promotes the proliferation of the human ASXL1-mutated cell lines, although to formally prove this, gene-editing mediated repair of ASXL1 mutations will need to be demonstrated in a future study.

HHEX is one of the homeobox factors and is highly expressed in various subtypes of myeloid leukemia.¹⁹  However, it is not fully understood how HHEX contributes to myeloid leukemogenesis. Using transcriptome analyses, we demonstrated that HHEX activated genes related to the proliferation of myeloid cells. Moreover, coexpression of ASXL1-MT and HHEX induced the upregulation of genes involved in leukemia stem cells. In line with these observations, coexpression of HHEX and ASXL1-MT enriched immature fractions and accelerated the development of myeloid leukemia driven by RUNX1-ETO and FLT3-ITD. In RUNX1-ETO AML, ASXL1 mutations are frequently found as well as ASXL2 mutations.²⁶  Although FLT3-ITD mutations are not prevalent in ASXL1-mutated AML, a recent study showed that ASXL1 mutations are found in 31 of 328 AML (9.5%). The clinical significance of ASXL1 mutations in FLT3-ITD AML are warranted in a future study.

In mammalian cells, the binding sites of HHEX were not clearly defined. Our ChIP-seq analysis demonstrated that HHEX exclusively binds to AT-rich regions in mammalian cells. The motif where HHEX binds showed high similarities to other homeobox transcriptional factors, such as homeobox B5 (HOXB5) and motor neuron and pancreas homeobox 1 (MNX1). HOXB5 efficiently marks long-term hematopoietic stem cells,³⁶  and MNX1 is aberrantly expressed in t(7; 12) AML and blocks differentiation of HSPCs.³⁷  According to previous reports, HHEX has been considered a repressive transcriptional factor that silences genes such as PDX1, CDX1b, and CDKN2A.^22,38  However, our data suggest that binding of HHEX at promoter loci is involved in gene activation as well.

A recent report showed that the ASXL1/BAP1 complex is involved in gene activation by removing the H2AK119ub modification.³⁹  ASXL1-MT, but not wild-type ASXL1, induces aberrant gene activation by enhancing the catalytic activity of BAP1.^5,9  Along with these findings, expression of ASXL1-MT/BAP1 complex significantly upregulated the expression of MYB and ETV5, although the upregulation of MYB in c-Kit⁺ cells is minor. On the contrary, depletion of ASXL1-MT in MEG-01 cells significantly reduced the expression of MYB, indicating that ASXL1-MT is important for maintaining MYB expression in leukemia cells. Consistent with the clinical data that HHEX expressions are not different between ASXL1-MT and ASXL1-WT AML, our data indicated that HHEX is not the target of ASXL1-MT/BAP1 complex. However, our data also suggested that ASXL1-MT enhanced binding of HHEX to the promoter loci of MYB and ETV5 by reducing H2AK119ub. Interestingly, we also revealed that ASXL1-MT/BAP1 complex enhances the promoter activity in the reporter assay, which might be a result of the unknown mechanisms independent of histone modifications (eg, recruiting other transcriptional factors). This hypothesis should be verified in the future study.

MYB is known to enhance self-renewal activity in leukemia stem cells.⁴⁰  Recently, several studies have developed MYB-targeting drugs to treat myeloid leukemia.^41,42  Our study showed that MYB depletion in ASXL1-mutated leukemia cells strongly induced apoptosis. ETV5 is known to contribute to proliferation and differentiation block in B-ALL.^43,44  However, the role of ETV5 in myeloid leukemogenesis remained elusive. In this study, we demonstrated that the depletion of ETV5 promoted myeloid differentiation of ASXL1-mutated leukemia cells.

Collectively, we demonstrated that HHEX promotes leukemogenesis and survival of leukemia cells expressing ASXL1-MT via the upregulation of MYB and ETV5. Our study suggests that inhibition of the HHEX-MYB/ETV5 axis would be a promising therapeutic approach for myeloid malignancies harboring ASXL1 mutations.

The authors thank Akiho Tsuchiya and Shiori Shikata for expert technical assistance. The authors are grateful to Naoki Goshima for providing plasmids regarding HHEX constructs. The authors are grateful to Yuichi J. Machida for providing plasmids regarding BAP1 constructs. MSCV-puro-N3FLAG-MYB-FL was a gift from Christopher Vakoc (Addgene plasmid # 105596). pLX_TRC311 ETV5 was a gift from William Hahn (Addgene plasmid # 74984). The authors are grateful to Hiroshi Kimura for the kind gift of the antihistone H3 antibody. The authors thank the FACS Core and the Mouse Core at the Institute of Medical Science, The University of Tokyo. The authors thank the Oregon Health & Science University for providing sequencing data from the Beat AML study, and The Cancer Genome Atlas for providing public gene expression data of AML.

This work was supported by a grant from IMSUT International Joint Project (T.K.), a grant from The Japanese Society of Hematology Research Grant (T.K.), a grant from Princess Takamatsu Cancer Research Fund (S.G.), a grant-in-aid from the Japan Society for the Promotion of Science for a postdoctoral fellowship (S.A.; 19J01570), and a grant from the Deutsche Forschungsgemeinschaft (H.J.B.).

Contribution: R.T. performed most of the experiments, analyzed and interpreted the data, and wrote the manuscript; S.A. conceived the project, designed and performed the experiments, analyzed and interpreted the data, and wrote the manuscript; S.-J.P. and K.N. analyzed ChIP-seq data and edited the manuscript; A.Y., A.K., and T.I. performed ChIP-seq experiments; H.J.B., Y.T., and S. Yamazaki performed gene editing of cord blood cells; V.V., C.H., and J.P.M. analyzed data from the patients’ samples and edited the manuscript; A.M., S. Yamasaki, and T.S. performed RNA-seq experiments; Y.H., T.Y., M.T., and T. Fukushima assisted with experiments; T. Fukuyama, D.I., and H.H. advised on data interpretation; and S.G. and T.K. conceived and supervised the project, interpreted the data, and participated in writing the manuscript.

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

Correspondence: Shuhei Asada, Tokyo Women’s Medical University, 8-1 Kawada-cho, Shinjuku-ku, Tokyo 1628666, Japan; e-mail: shiasada-tky@umin.ac.jp; and Toshio Kitamura, The Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 1088639, Japan; e-mail: kitamura@ims.u-tokyo.ac.jp.