HOXA9 promotes MYC-mediated leukemogenesis by maintaining gene expression for multiple anti-apoptotic pathways

HOXA9 is often highly expressed in leukemias. However, its precise roles in leukemogenesis remain elusive. Here, we show that HOXA9 maintains gene expression for multiple anti-apoptotic pathways to promote leukemogenesis. In MLL fusion-mediated leukemia, MLL fusion directly activates the expression of MYC and HOXA9. Combined expression of MYC and HOXA9 induced leukemia, whereas single gene transduction of either did not, indicating a synergy between MYC and HOXA9. HOXA9 sustained expression of the genes implicated in the hematopoietic precursor identity when expressed in hematopoietic precursors, but did not reactivate it once silenced. Among the HOXA9 target genes, BCL2 and SOX4 synergistically induced leukemia with MYC. Not only BCL2, but also SOX4 suppressed apoptosis, indicating that multiple anti-apoptotic pathways underlie cooperative leukemogenesis by HOXA9 and MYC. These results demonstrate that HOXA9 is a crucial transcriptional maintenance factor that promotes MYC-mediated leukemogenesis, potentially explaining why HOXA9 is highly expressed in many leukemias.


Introduction 1
Mutations of transcriptional regulators often cause aberrant gene regulation of 2 hematopoietic cells, which leads to leukemia. Structural alterations of the mixed 3 lineage leukemia gene (MLL also known as KMT2A) by chromosomal translocations 4 cause malignant leukemia that often associates with poor prognosis despite the current 5 intensive treatment regimens (1). MLL encodes a transcriptional regulator that 6 maintains segment-specific expression of homeobox (HOX) genes during 7 embryogenesis (2), which determines the positional identity within the body (3)(4)(5). 8 During hematopoiesis, MLL also maintains the expression of posterior HOXA genes 9 and MEIS1 (another homeobox gene), which promote the expansion of hematopoietic 10 stem cells and immature progenitors (6-10). The oncogenic MLL fusion protein 11 constitutively activates its target genes by constitutively recruiting transcription 12 initiation/elongation factors thereto (11-13). Consequently, HOXA9 and MEIS1 are 13 highly transcribed in MLL-rearranged leukemia (7). Forced expression of HOXA9 14 (but not MEIS1) immortalize hematopoietic progenitor cells (HPCs) ex vivo (14, 15). 15 Co-expression of HOXA9 with MEIS1 causes leukemia in mice which recapitulates 16 MLL-rearranged leukemia (15). Moreover, HOXA9 is highly expressed in many non-17 MLL-rearranged leukemias such as those with NPM1 mutation and NUP98 fusion and 18 is associated with poor prognosis (16). These findings highlight HOXA9 as a major 19 contributing factor in leukemogenesis. Nevertheless, the mechanism by which 20 HOXA9 promotes oncogenesis remains elusive. 21 HOXA9 is considered to function as transcription factor, which retains a 22 sequence-specific DNA binding ability. HOX proteins have an evolutionally 23 conserved homeodomain which possesses strong sequence preferences (17). HOXA9 24 associates with other homeodomain proteins such as PBX and MEIS family proteins 25 (14, 18). HOXA9 and those HOXA9 cofactors form a stable complex on a DNA 1 fragment harboring consensus sequences for each homeodomain protein (18,19), 2 suggesting that they form a complex of different combinations in a locus-specific 3 manner depending on the availability of the binding sites. Recently, it has been 4 reported that HOXA9 specifically associates with enhancer apparatuses (e.g. 5 MLL3/4) to regulate gene expression (20)(21)(22). However, the mechanisms by which 6 HOXA9 activate gene expression remain largely unclear. 7 In this study, we reveal the oncogenic roles for HOXA9 and its target gene 8 products in leukemogenesis and its unique mode of function as a transcriptional 9 maintenance factor which preserves an identity of a hematopoietic precursor. remained abundant in MLL fusion-expressing cells ( Figure 1B). Myc was highly 20 expressed in KSL, common myeloid progenitors (CMP), and granulocyte/macrophage 21 progenitors (GMP), which contain actively dividing populations (25), but was 22 completely suppressed at highly differentiated c-kit low /Mac1 high stages. In the two 23 MLL fusion-expressing cell lines, Myc was expressed at comparable levels to those in 24 the progenitor fractions (CMP and GMP; Figure 1B). Mxd1, a differentiation marker, 25 was highly expressed only in differentiated populations. These results indicate that 1 Myc, Hoxa9, Hoxa10, and Meis1 are intrinsically programmed to be silenced in 2 normal hematopoietic differentiation but are aberrantly maintained by MLL fusion 3 proteins. 4 To assess the oncogenic potential of MLL target genes, we performed 5 myeloid progenitor transformation assays, wherein HPCs were isolated from mice, 6 retrovirally transduced with each MLL target gene, and cultured in semi-solid 7 medium supplemented with cytokines promoting myeloid lineage differentiation 8 ( Figure 1C, D). HPCs transduced with MLL-ENL and MLL-AF10 produced a large 9 number of colonies in the third and fourth passages with high mRNA levels of Myc, 10 Hoxa9, Hoxa10, and Meis1 in colonies of the first passage, confirming their potent 11 transforming capacities. These cells were considered "immortalized" as they 12 proliferate indefinitely in this ex vivo culture (26,27 Transforming potential of MLL target genes. Clonogenic potential of the indicated 6 constructs was analyzed by myeloid progenitor transformation assays. Colony 7 forming unit per 10 4 cells (CFU) (Mean with SD, n = 3, biological replicates), relative 8 colony size (Mean with SD, n ≥ 100), and relative mRNA levels of indicated genes 9 (Mean with SD, n = 3, PCR replicates) were measured at the indicated time points. #: 10 Both endogenous murine transcripts and exogenous human transcripts were detected 11 by the qPCR primer set used. replicates) and mRNA level of Myc (Mean with SD, n = 3 PCR replicates) are shown. 16 Statistical analysis was performed using ordinary one-way ANOVA with the vector 17 control. ****P < 0.0001. 18 19 HOXA9 confers the identity of a hematopoietic precursor while MYC drives 20 anabolic pathways. 21 To identify the genes specifically regulated by HOXA9 but not by MYC, we 22 performed RNA-seq analysis of HOXA9-ICs and MYC-ICs which do not express 23 HOXA9 ( Figure 1C). Genes highly expressed in HOXA9-ICs but lowly expressed in 24 MYC-ICs (defined as "HOXA9 high signature") were associated with hematopoietic 25 identity/functions, whereas genes highly expressed in MYC-ICs and lowly expressed 26 in HOXA9-ICs (defined as "MYC high signature") were associated with anabolic 27 pathways (Figure 2A, B). MLL-AF10-ICs, which express endogenous Hoxa9 and 28 Myc at high levels ( Figure 1C it did not exhibit any leukemia-associated signs. HOXA9 did not induce leukemia 1 within 200 days either, suggesting that the HOXA9 high signature alone is also 2 insufficient to induce leukemia. Taken together, these results suggest that both high 3 MYC activity and HOXA9-mediated resistance to apoptosis are necessary for driving 4 leukemogenesis in vivo. MEIS1 are plotted separately by leukemia phenotype (ALL or AML) (B). C, 2 Transforming potential of various combinations of MLL target genes. CFU (Mean  3 with SD, n = 3, biological replicates) and relative colony size (Mean with SD, n ≥ 4 100) are shown as in Figure 1C. D, Morphologies of the colonies and transformed 5 cells. Bright field (left) and May-Grunwald-Giemsa staining (right) images are shown 6 with scale bars. E, In vivo leukemogenic potential of various oncogene combinations. 7 Kaplan-Meier curves of mice transplanted with HPCs transduced with the indicated 8 genes are shown as in Figure 3D. Bone marrow cells from moribund mice were 9 harvested and used for secondary transplantation. 10 11 HOXA9 functions as a transcription maintenance factor 12 Some HOXA9 high signature genes, namely Bcl2, Sox4, and Igf1, have been 13 implicated in leukemogenesis (31)(32)(33)(34), suggesting that they may be responsible for the 14 synergy between HOXA9 and MYC. Indeed, Bcl2, Sox4, and Igf1 were highly 15 Two genes were transduced into HPCs in a simultaneous manner. B, Gene expression 6 after inactivation of HOXA9. HOXA9-ER and MYC were doubly transduced into 7 HPCs and cultured in the presence of 4-OHT ex vivo. After 4-OHT withdrawal, RT-8 qPCR analysis was performed for the indicated genes (Mean, n=4, biological 9 replicates). Statistical analysis was performed using unpaired two-tailed Student's t-10 test. **P < 0.01, *P < 0.05. C. Gene expression of HPCs immortalized by step-wise 11 transduction of various transgenes. Relative mRNA levels of HOXA9 target genes in 12 myeloid progenitors transformed by various combinations of MLL target genes are 13 shown as in A. Two genes were transduced into HPCs in a stepwise manner. 14 15 BCL2 and SOX4 promote MYC-mediated leukemogenesis by alleviating 16 apoptosis 17 To identify the roles for BCL2 and SOX4 in leukemic transformation, we evaluated 1 the leukemogenic potential of combined expression of BCL2 or SOX4 with MYC. In 2 myeloid progenitor transformation assays, SOX4 by itself showed weak 3 immortalization capacity as previously reported (31) genes. CFU (Mean with SD, n = 3, biological replicates) is shown as in Figure 1C. B, 5 In vivo leukemogenic potential of various combinations of MYC and HOXA9 target 6 genes. Kaplan-Meier curves of mice transplanted with HPCs transduced with the 7 indicated genes are shown as in Figure 3D.  shown on the right. Statistical analysis was performed using the log-rank test and 10 Bonferroni correction with the vector control. D, Rescue of in vivo leukemogenic 11 potential by sgRNA-resistant transgenes. Before transduction of sgRNA, MLL-ENL-12 LCs were transduced with sgRNA-resistant BCL2 or SOX4. In vivo leukemogenesis 13 assay was performed as described in Figure 3D. E, A model illustrating HOXA9-14 mediated pathogenesis in MLL-rearranged leukemia. 15 16

Discussion 1
In this study, we found that HOXA9 regulates a variety of genes to maintain 2 hematopoietic precursor identity and its associated anti-apoptotic properties. In 3 leukemic transformation, MLL fusion proteins exploit both HOXA9 and MYC 4 downstream pathways. Accordingly, HOXA9 and MYC synergistically induce 5 leukemia in mouse models. Thus, we propose that MLL fusion proteins employ two 6 arms to promote oncogenesis: MYC-mediated proliferation and HOXA9-mediated 7 resistance to differentiation/apoptosis. 8 It is widely accepted that two types of mutations need to occur before 9 leukemia onset; class I mutations that confer proliferative advantages and class II 10 Therapeutic efficacy of BCL2 inhibitor has been reported in AML including 1 MLL leukemia (43). Because there is a correlation between the expression levels of 2 HOX proteins and the sensitivity to BCL2 inhibitor in AML patient samples, the 3 aberrant expression of HOXA9 is the potential mechanisms for BCL2-dependence of 4 MLL leukemia (44, 45). Oncogenic MYC expression often leads to apoptosis, which 5 need to be alleviated by additional genetic events for leukemic cell survival (28). 6 Indeed, co-expression of HOXA9, BCL2 or SOX4 promoted MYC-mediated 7 leukemogenesis, while MYC alone was insufficient to induce leukemia in vivo. 8 Although the downstream mechanisms could not be addressed, SOX4 also exhibited 9 anti-apoptotic effects on MYC-expressing cells. Importantly, there was a partial effect 10 of single gene knockout of Bcl2 and Sox4 on leukemia initiation and maintenance. 11 This indicates that multiple anti-apoptotic pathways are exploited by MLL fusion 12 proteins and that blocking a single anti-apoptotic pathway may be insufficient to 13 completely abrogate leukemic potential. Thus, simultaneously blocking multiple anti-14 apoptotic pathways may be required for efficient molecularly targeted therapy of 15

HOXA9-expressing leukemia. 16
Our results also provide an insight into the mode of function of HOXA9. HOX 17 genes are known to express in a position-specific manner, conferring a positional 18 identity to a cell. For example, similarly functional fibroblasts derived from different 19 parts of a body express different HOX genes (5). Thus, it is unlikely that HOXA9 20 functions as a major upstream factor which determines tissue-specific gene expression 21 by turning a silenced chromatin into transcriptionally active chromatin. HOX proteins 22 likely play a supportive role to maintain gene expression which was activated by other 23 transcriptional regulators. Our observation that HOXA9 cannot reactivate gene 24 expression once silenced fits to this hypothesis. Accordingly, HOXA9 maintains a 25 subset of genes related to hematopoietic identity when expressed in hematopoietic 1 precursors. Recently, it has been reported that HOXA9 may recruit enhancer 2 apparatuses, as it colocalizes with active enhancer mark (i.e. acetylated histone H3 3 lysine 27). We speculate that HOXA9 may support the maintenance of an active 4 enhancer, but unlikely establishes it on a silenced chromatin. Further functional 5 analysis of HOXA9 is required to understand how HOXA9 regulates gene expression. 6 In summary, our results describe the oncogenic roles for HOXA9 as 7 transcriptional maintenance factor for multiple anti-apoptotic genes, which are 8 necessary to promote MYC-mediated leukemogenesis. In case of MLL-rearranged 9 leukemia, MLL fusion proteins directly activate both MYC and HOXA9, while 10 HOXA9 maintains expression of MYC, BCL2, and SOX4, achieving high MYC 11 activity and anti-apoptotic properties simultaneously ( Figure 7E). Thus, MLL-12 rearranged leukemia cells acquired highly proliferative potentials and survival 13 advantages at the same time, using HOXA9 as a key mediator.  Table 1). The MSCV-7 neo MLL-ENL, and MLL-AF10 vectors have been previously described (23).  Table 1 The myeloid progenitor transformation assay was carried out as previously described 19 (26,27). Bone marrow cells were harvested from the femurs and tibiae of 5-week-old 20 female C57BL/6J mice. c-Kit + cells were enriched using magnetic beads conjugated 21 with an anti-c-Kit antibody (Miltenyi Biotec), transduced with a recombinant 22 retrovirus by spinoculation, and then plated (4 × 10 4 cells/ sample) in a 23 methylcellulose medium (Iscove's modified Dulbecco's medium, 20% FBS, 1.6% 24 methylcellulose, and 100 µM β-mercaptoethanol) containing murine stem cell factor 25 (mSCF), interleukin 3 (mIL-3), and granulocyte-macrophage colony-stimulating 1 factor (mGM-CSF; 10 ng/mL each). During the first culture passage, G418 (1mg/mL) 2 or puromycin (1μg/mL) was added to the culture medium to select for transduced 3 cells. Hoxa9 expression was quantified by qRT-PCR after the first passage. Cells 4 were then re-plated once every 4-6 days with fresh medium; the number of plated 5 cells for the second, third, and fourth passages was 4 × 10 4 , 2 × 10 4 , and 1 × 10 4 6 cells/well, respectively. CFUs were quantified per 10 4 plated cells at each passage. 7 8 In vivo leukemogenesis assay 9 In vivo leukemogenesis assays were carried out as previously described (26,52). c-10 Kit + cells (2 ´ 10 5 ) prepared from the femurs and tibiae of 5-week-old female 11 C57BL/6J mouse were transduced with retrovirus by spinoculation and intravenously 12 transplanted into sublethally irradiated (5-6 Gy) C57BL/6J mice. For secondary 13 leukemia, leukemia cells (2 ´ 10 5 ) cultured ex vivo for more than three passages were 14 transplanted. As for knockouts of Bcl2 and Sox4, mice heterozygous for Bcl2 or Sox4 15 were crossed, and c-Kit + cells were isolated from fetal livers at E14-15 (for Bcl2) or 16 E13 (for Sox4). The next day, cells were transduced with MLL-AF10 or 17 HOXA9/MEIS1 and transplanted intravenously into sublethally irradiated (2.5 Gy) 18 SCID mice [2 × 10 5 (for Bcl2) or 1 × 10 5 cells/mouse (for Sox4)]. 19 20

qRT-PCR 21
Total RNA was isolated using the RNeasy Mini Kit (Qiagen) and reverse-transcribed 22 using the Superscript III First Strand cDNA Synthesis System (Thermo Fisher 23 Scientific) with oligo (dT) primers. Gene expression was analyzed by qPCR using 24 TaqMan probes (Thermo Fisher Scientific). Relative expression levels were 25 normalized to those of GAPDH/Gapdh or TBP/Tbp and determined using a standard 1 curve and the relative quantification method, according to manufacturer's instructions 2 (Thermo Fisher Scientific). Commercially available PCR probes used are listed in 3 Supplemental Table 1.  4 5

ChIP-qPCR and ChIP-seq 6
The eluted material obtained by fanChIP was extracted by phenol/chloroform/isoamyl 7 alcohol. DNA was precipitated with glycogen, dissolved in TE buffer, and analyzed 8 by qPCR (ChIP-qPCR) or deep sequencing (ChIP-seq). The qPCR probe/primer 9 sequences are listed in Supplemental Table 4. Deep sequencing was performed using 10 the TruSeq ChIP Sample Prep Kit (Illumina) and HiSeq2500 (Illumina) at the core 11 facility of Hiroshima University and described in our previous publication (23). 12

RNA-seq 14
Total RNA was prepared using the RNeasy Kit (Qiagen) and analyzed using a 15 Bioanalyzer (Agilent Technologies). Deep sequencing was performed using a 16 SureSelect Strand Specific RNA Library Prep Kit (Agilent Technologies) and 17 HiSeq2500 (Illumina) with 51-bp single-end reads at the core facility of Hiroshima 18 University. Sequenced reads were mapped to the mouse genome assembly mm9 using 19 TopHat 2.0.14 (53) and read counts were normalized with Cufflinks 2. 2.1 2010 (54).  Statistical analysis was performed using GraphPad Prism 7 software. Data are 1 presented as the mean with standard deviation (SD). Comparisons between two 2 groups were analyzed by unpaired two-tailed Student's t-test, while multiple 3 comparisons were performed by ordinary one-way ANOVA followed by Dunnett's 4 test or two-way ANOVA. Mice transplantation experiments were analyzed by the log-5 rank test and Bonferroni correction was applied for multiple comparisons. P values < 6 0.05 were considered statistically significant. n.s.: P>0.05, *: P ≤ 0.05, **: P ≤ 0.01, 7 ***: P ≤ 0.001, and ****: P ≤ 0.0001. 8 9

Study approval 10
All animal experimental protocols were approved by the National Cancer Center 11 (Tokyo Japan) Institutional Animal Care and Use Committee. 12

Acknowledgments 1
We thank Yuzo Sato, Makiko Okuda, Megumi Nakamura, Etsuko Kanai, Aya 2 Nakayama, Boban Stanojevic, and Ayako Yokoyama for technical assistance. We 3 thank Drs. Yoshihide Tsujimoto and Hans Clevers for providing us the knockout 4 mouse lines of Bcl2 and Sox4, respectively. We also thank all members of the Shonai   LCs were harvested from moribund mice during the transplantation assay as described 5 in Figure 4E. Relative mRNA levels of the indicated genes (Mean with SD, n=3, PCR 6 replicates) are shown along with those of the respective ICs.