Targeting miR‐193a‐AML1‐ETO‐β‐catenin axis by melatonin suppresses the self‐renewal of leukaemia stem cells in leukaemia with t (8;21) translocation

Abstract AML1‐ETO, the most common fusion oncoprotein by t (8;21) in acute myeloid leukaemia (AML), enhances hematopoietic self‐renewal and leukemogenesis. However, currently no specific therapies have been reported for t (8;21) AML patients as AML1‐ETO is still intractable as a pharmacological target. For this purpose, leukaemia cells and AML1‐ETO‐induced murine leukaemia model were used to investigate the degradation of AML1‐ETO by melatonin (MLT), synthesized and secreted by the pineal gland. MLT remarkedly decreased AML1‐ETO protein in leukemic cells. Meanwhile, MLT induced apoptosis, decreased proliferation and reduced colony formation. Furthermore, MLT reduced the expansion of human leukemic cells and extended the overall survival in U937T‐AML1‐ETO‐xenografted NSG mice. Most importantly, MLT reduced the infiltration of leukaemia blasts, decreased the frequency of leukaemia stem cells (LSCs) and prolonged the overall survival in AML1‐ETO‐induced murine leukaemia. Mechanistically, MLT increased the expression of miR‐193a, which inhibited AML1‐ETO expression via targeting its putative binding sites. Furthermore, MLT decreased the expression of β‐catenin, which is required for the self‐renewal of LSC and is the downstream of AML1‐ETO. Thus, MLT presents anti‐self‐renewal of LSC through miR‐193a‐AML1‐ETO‐β‐catenin axis. In conclusion, MLT might be a potential treatment for t (8;21) leukaemia by targeting AML1‐ETO oncoprotein.

creates an in-frame fusion gene between the conserved runt homology domain from AML1 to almost the entire ETO gene. 1 The AML1 encodes a subunit of the core-binding factor heterodimer, which mediates in transcriptional regulation during hematopoiesis. ETO represses transcription through recruiting a nuclear receptor corepressor, histone deacetylase complex and the mSin3 corepressor. 3 Thus, AML1-ETO is believed to block myeloid differentiation via partially inhibiting the transcription of AML1-driven genes involved in cell differentiation. Multiple studies indicate that AML1-ETO alone is not sufficient to induce AML in a murine model and thus additional genetic events are required for the onset of AML. 4 AML1-ETO rapidly induces murine leukaemia in cooperation with Wilm's tumour-1 (WT1), 5 suggesting that AML1-ETO is considered as the important onset oncogene for the leukemogenesis of AML.
Acute myeloid leukaemia is initiated from a small subset of leukaemia stem cell (LSC). Drug-resistant LSC, which is not completely eradicated by current standard therapies, causes most relapse of AML patients after chemotherapy. 6 LSC presents several important characteristics, including increased self-renewal, uncontrolled proliferation and dysregulated differentiation, distinguishing it from normal hematopoietic stem and progenitor cell (HSPC). These characteristics are caused by the expression of some of the leukaemia oncogenes, such as MLL-AF9 7 and AML1-ETO. 8 AML1-ETO enhances the self-renewal capacity of HSPC through COX-β-catenin signalling. 9 Therefore, eliminating AML1-ETO protein might inhibit the self-renewal capacity in AML patients with t (8;21). Our previous study has shown that honokiol, a natural phenolic compound isolated from the plant Magnolia officinalis, rapidly degrades AML1-ETO protein via increasing ubiquitin conjugase UbcH8 expression. 10 However, whether agents degrading AML1-ETO protein present anti-self-renewal of LSC is not yet determined.
Melatonin (MLT; N-acetyl-5-methoxytryptamine) is an ancient molecule with a wide range of physiological functions. 11 In mammals, the pineal gland synthesizes and secretes MLT into the blood circulation. However, extrapineal tissues, including bone marrow and skin, might also contribute to MLT levels. 12 The concentration of MLT may reach up to 0.5 nmol/L in mammalian blood at night. At this physiological concentration, MLT affects the biological clock, regulates the immune system, and presents antioxidant actions. 13 MLT influences the physiological function mainly through interaction with two wellcharacterized G protein-coupled seven-transmembrane-domain receptors, MT1 and MT2, which inhibit adenylate cyclase. 14 MLT at higher concentration (mM) presents anti-cancer activity through inhibiting proliferation and inducing apoptosis in multiple types of cancer including leukaemia. 15 Additionally, MLT enhances the antileukaemia activity of puromycin in HL-60 cells. 16 Fusion genes, such as AML1-ETO and PML-RARα, play important roles in leukemogenesis. However, whether MLT presents anti-leukaemia ability through degrading fusion proteins, such as AML1-ETO, is not determined. Therefore, we determined whether MLT can degrade AML1-ETO oncoprotein and found that MLT degraded AML1-ETO protein and suppressed the self-renewal of LSCs in leukaemia with positive AML1-ETO.
In this report, we found that MLT substantially reduces AML1-ETO protein in leukaemia cell lines, primary AML blasts, and AML1-ETO-induced murine leukaemia blasts. MLT induces apoptosis, decreases proliferation and reduces colony formation in leukaemia cells, but not in normal murine and human HSPC cells. Furthermore, MLT prolongs the overall survival in AML1-ETO-induced murine leukaemia. In addition, MLT decreases the self-renewal of LSC via inhibiting the expression of β-catenin. Mechanistically, MLT decreases AML1-ETO expression through up-regulation of miR-193a. Thus, MLT presents anti-self-renewal of LSC through miR-193a-AML1-ETO-β-catenin axis in AML cells carrying AML1-ETO. and were then incubated with streptavidin particles and placed in the magnetic field to remove the lineage-committed cells from progenitor cells. The lineage-negative (Lin − ) fraction of cells was resuspended in Iscove modified Dulbecco medium for colony forming count. Normal human CD34 + cells were isolated and enriched from umbilical cord blood by positive selection (Stemcell Technologies).

| Construction of plasmids
To construct the plasmid expressing miR-193a in mammalian cells, the paired primers were based on the primary sequence of pre-miR-193a and its flanking regions. Then, the PCR product was cloned into lentiviral expression vector pLVX-IRES-ZsGreen1 (pLVX-GFP, Clontech, Palo Alto, CA). The coding sequence of β-catenin was constructed in pLVX-GFP to produce plasmid expressing human β-catenin. MSCV-GFP-IRES-AML1-ETO was kindly provided by Prof. Ying Lu (Department of Pathophysiology, Shanghai Jiao Tong University School of Medicine). Whole CDS of WT1 was directly synthesized (Genewiz, Suzhou, China) and then inserted into retrovirus vector pMSCV-puro (Clontech). 21 All plasmids were confirmed by sequencing. Leukemic cells were seeded in 6-well plates at a density of 2.0 × 10 5 /mL per well. Special miR-193a inhibitor Agomir (100 pmol per well, GenePharma, Shanghai, China) or negative control was transiently transfected into leukemic cells by Hiperfect transfection reagent (Qiagen). Cells were collected and proteins were extracted for western blot after transfection for 48 hours.

| Other procedures
For details on reagents, apoptosis detection, virus production and cell transduction, western blot, luciferase reporter and mutagenesis assays, and other experiments sees supplemental materials and methods.

| Statistical analysis
All the results were expressed as Mean ± SD where applicable. The significance of the difference between groups was determined by Student's t test. A P < 0.05 was considered statistically significant.
All statistical analyses were performed with SPSS software (SPSS 22.0, Chicago, IL).

| Anti-leukaemia activity by MLT
To determine whether MLT has potential anti-leukaemia activity  or without 1 mmol/L MLT and colonies were counted. ** and ## P < 0.01 vs untreated cells, respectively. (D), Cell growth was measured by CCK-8 in two primary AML blasts with positive AML1-ETO, which were treated with or without 0.5, 1, 2 mmol/L MLT for 24 h. (E), Apoptosis was measured by Annexin V/PI staining in two primary AML blasts bearing AML1-ETO treated with or without 1 mmol/L MLT for 24 h. ** and ## P < 0.01 vs untreated cells, respectively. (F), CD34 + cells (1 × 10 3 ) were isolated from two primary AML blasts bearing AML1-ETO and were plated on methylcellulose medium incubated with or without 1 mmol/L MLT for colony counting. ** and ## P < 0.01 vs untreated cells, respectively. (G), Apoptosis was measured by Annexin V/PI staining in two CD34 + cells, which were isolated from two umbilical cord blood, treated with or without 1 mmol/L MLT for 24 h. (H), CD34 + cells (1 × 10 3 ) were isolated from two cord blood and were plated on methylcellulose medium treated with or without 1 mmol/L MLT for colony counting. (I), Lineage − cells (Lin − ) were isolated from BM mononuclear cells of two wild-type C57 mice. Lin − cells (2 × 10 3 ) were plated on methylcellulose medium for colony formation treated with or without 1 mmol/L MLT, followed by colony counting. (J), CD11b and CD14 were detected in Kasumi-1 cells treated with or without 1 mmol/L MLT for three days. Shown is the representative plot (Left) and the summary of CD11b + and CD14 + cells (Right). ** and ## P < 0.01 vs untreated cells, respectively. (K), Wright-Giemsa staining for Kasumi-1 cells treated with or without 1 mmol/L MLT for three days. Scale bars represent 10 μmol/L block MLT-induced degradation of AML1-ETO protein ( Figure 3C).

| MLT-induced inhibition of AML1-ETO is independent of caspase-3 activation, ubiquitinproteasome system and MT1/2 receptor
To further determine whether MLT enhances the degradation rate, the ubiquitinated proteins were measured in leukaemia cells treated with or without MLT for 12, 24 and 48 hours. MLT did not substantially enhance the ubiquitinated proteins ( Figure 3D).
As documented, 30 AML1-ETO interacts with heat shock protein 90 (HSP90), an important molecular chaperone that plays a key role in the conformational maturation and stabilization of signalling proteins.
Furthermore, 17-allylamino-geldanamycin (17-AAG), a HSP90 antagonist, induces the degradation of AML1-ETO. 30 To further exclude the possibility that MLT degraded the AML1-ETO protein through degrading HSP90, the expression of HSP90 was measured in Kasumi-1 and U937T cells treated with MLT for 24 and 48 hours. However, MLT failed to affect the protein expression of HSP90 ( Figure 3G).

| MLT degrades AML1-ETO through upregulation of miR-193a
Our data indicated that MLT-induced degradation of AML1-ETO is independent of activated caspase-3, ubiquitin-proteasome system  Figure 4D). However, the decreased luciferase activity was almost abolished by the mutation of putative miR-193a-binding sites ( Figure 4D).
To further confirm that miR-193a plays an important role in MLTinduced degradation of AML1-ETO, a specific inhibitor of miR-193a was transfected in MLT-treated cells and AML1-ETO was measured.

| Effects of MLT in Kasumi-1 and U937T-GFP xenograft mouse model
To explore whether MLT could reduce the tumourigenicity in vivo, Kasumi-1 cells were injected subcutaneously into the right flanks of mice to produce xenograft mouse model followed by MLT treatment or not. MLT treatment substantially reduced average tumour volume by 45.7% compared with vehicle mice ( Figure 5A,B).
Also, MLT treatment led to a 41.2% decrease in average tumour weight ( Figure 5C). Because MLT decreases the expression of AML1-ETO in vitro, we then determined whether MLT reduced the AML1-ETO expression in vivo. As expected, the protein levels of AML1-ETO were significantly decreased in the tumours from MLT-treated mice compared with vehicle mice ( Figure 5D). These results indicate that MLT inhibits the tumour growth of Kasumi-1

cells in vivo.
To further explore the anti-leukaemia activity, U937 cells were transduced with MSCV-GFP-IRES-AML1-ETO to produce U937T-GFP cells, which continuously expresses both GFP ( Figure S2A) and AML1-ETO ( Figure S2B). U937T-GFP cells were transplanted into NSG mice and then the mice were treated with or without MLT ( Figure 5E). GFP + cells were detected to reflect the infiltration of human leukaemia blasts. MLT decreased leukaemia cells in NSG mice by about 60% ( Figure 5F). Furthermore, MLT substantially weakened the expansion of engrafted leukemic cells in  Figure 5G) and prolonged the overall survival ( Figure 5H).  Figure 6E). Finally, the protein expression of AML1-ETO was measured in MLT-treated mice and vehicle mice. As expected, MLT decreased the protein expression of AML1-ETO ( Figure 6F).

| MLT decreases the long-term self-renewal of LSC in AML1-ETO-induced murine leukaemia
To assess the impact of MLT on long-term self-renewal of LSC with AML1-ETO, we performed serial mouse BMT assays. In the secondary mouse BMT assay, vehicle or MLT-treated BM blasts isolated from primary leukaemia mice were transplanted to lethally irradiated wild-type recipient mice. Vehicle and MLT-treated mice developed leukaemia much faster than the corresponding primary BMT groups, respectively ( Figure 7A vs Figure 6E To directly compare the effects of MLT treatment on the frequency of LSC, we conducted limiting dilution assays with mouse BM leukemic cells collected from tertiary BMT recipients as donor cells. Transplantation results showed that LSC frequency was decreased in MLT-treated mice vs vehicle mice (1/11287 vs 1/2924, P < 0.05, Figure 7C). Finally, to investigate the functional role of MLT on LSC function, we evaluated the colony formation ability of MLT-treated mice and vehicle mice. GFP + cells from MLT-treated mice formed colonies at lower frequency than those from vehicle mice ( Figure 7D).

| MLT inhibits self-renewal by reducing β-catenin activity
As documented, 9 β-catenin is essential for the hematopoietic selfrenewal of leukaemia cells with AML1-ETO. To investigate whether MLT reduces β-catenin activity, the protein expression of β-catenin was detected in Kasumi-1 and U937T cells treated with MLT. MLT decreased the protein expression of β-catenin in Kasumi-1 and U937T cells ( Figure S4A). To further investigate the role of β-catenin on the anti-self-renewal by MLT, Kasumi-1 and U937T cells were transduced with LVX-NC or LVX-GFP-β-catenin. Western blot indicated successful overexpression of β-catenin ( Figure S4B). Importantly, overexpression of β-catenin partially prevented MLT-induced inhibition of colony formation ( Figure S4C,D).
β-catenin was decreased as a consequence of the degradation of AML1-ETO. 9 To determine whether MLT directly regulates the expression of β-catenin independent of degradation of AML1-ETO in  Figure   S5A). Therefore, we speculate that MLT decreases the expression of β-catenin mainly through the degradation of AML1-ETO.

| D ISCUSS I ON
AML1-ETO is an important factor for the leukemogenesis through enhancing self-renewal of LSC in AML patients with t (8;21).
Inhibiting AML1-ETO may provide a potential therapy for these AML patients. Here, we report that MLT strongly decreases endogenous and exogenous expressions of AML1-ETO protein in leukaemia cell lines and primary AML blasts. MLT decreases AML1-ETO through up-regulation of miR-193a, which inhibits AML1-ETO expression via binding its CDS and 3'UTR sites. In addition, MLT decreases the expression of β-catenin, which has been reported as downstream of AML1-ETO, 9 suggesting that MLT presents anti-self-renewal of LSC through miR-193a-AML1-ETO-β-catenin axis in AML1-ETO-expressing cells ( Figure S4E). Therefore, MLT might provide the lead natural compound for the development of AML1-ETO-targeted agents.
miRNAs have recently been found to play an important role in biological regulations such as apoptosis, proliferation, and differentiation in haematological cells by modulating the expression of oncogenes or tumour suppressors. 21,33 We identified miR-193a as a candidate miRNA regulated by MLT. AML1-ETO triggers the silencing of miR-193a and in contrast, miR-193a decreases AML1-ETO expression via targeting its binding sites, suggesting that a feedback circuitry is formed in AML cells bearing AML1-ETO. 32  MLT has been reported to present anti-leukaemia activity through different mechanisms. 15,34,35 However, whether MLT inhibits self-renewal of LSC is not determined. The most important finding of this study is that MLT suppresses colony formation, decreases frequency of LSC, and extends overall survival in AML1-ETO-induced murine leukaemia. AML1-ETO enhances hematopoietic self-renewal and leukemogenesis through β-catenin signaling pathway. 9 Our results indicate that MLT decreases the expression of β-catenin and overexpression of β-catenin partially prevents MLT-induced anti-self-renewal, suggesting that MLT presents anti-self-renewal activity through AML1-ETO-β-catenin axis. Consistent with our results, MLT has been reported to promote the anticancer effect of paclitaxcel in brain cancer stem cells. 36 Most importantly, MLT has little effect on the colony formation in normal human and murine HSPC. Thus, MLT is considered as a low-toxicity protective agent against leukaemia cells. For example, MLT ameliorates the toxicity induced by therapeutic drugs 37 and X-ray radiation. 34 Therefore, MLT should be a potential anti-LSC agent in AML with t (8;21) translocation without severe side-effects.
Targeting AML1-ETO has achieved potential beneficial effector for AML patients with AML1-ETO. For example, oridonin shows significant anti-leukaemia ability in vitro and in vivo through inducing the degradation of AML1-ETO oncoprotein. 23 Depsipeptide, a histone deacetylase inhibitor, interrupts the association of AML1-ETO with HSP90 and induces the proteasomal degradation of AML1-ETO. 30 Furthermore, our results also demonstrate that honokiol induces the degradation of AML1-ETO via increasing the UbcH8, an E2-conjugase. 10 However, because AML1-ETO alone is essential but not sufficient for the leukemic transformation, additional gene mutations, such as c-Kit, are required for the leukaemia transformation in murine model. Thus, combination treatments targeting RUNX1/ ETO and the frequent mutated proteins have achieved pre-clinical efficiency. For example, oridonin and homoharringtonine exert synergistic effects against t (8; 21) leukaemia through targeting AML1-ETO and c-Kit mutation. 38 However, the five-year overall survival in AML patients with AML1-ETO is still less than 30%. Recently, more studies focus the target geness which are regulated by AML1-ETO and are required for leukemogenesis. For example, COX2 9 and CCND2 39 are two target genes by AML1-ETO. Targeting COX2 and CCND2 by approved drugs inhibits the self-renewal of LSCs and demonstrates preclinical therapeutic effects. Thus, specific AML1-ETO-targeting strategies will finally improve the overall survival in AML patients with AML1-ETO.
As documented, AML1-ETO protein can be degraded or cleaved in leukaemia cells, indicating that AML1-ETO protein is degraded through the UbcH8-mediated ubiquitin-proteasome system. In addition, E3-ligase SIAH-1 is mediated in the degradation of AML1-ETO protein. 27 Thus, AML1-ETO protein can be degraded by ubiquitin proteasome system including UbcH8 and SIAH-1. However, our results indicate that MG132, a prototypical proteasome inhibitor, fails to block MLT-induced degradation of AML1-ETO. Moreover, MLT does not increase the expression of UbcH8 and SIAH-1 (data not shown). Therefore, MLT-induced degradation of AML1-ETO protein might be independent of the ubiquitin-proteasome system.
Although inhibition of MT1/2 by a competitive antagonist luzindole did not prevent the degradation of AML1-ETO by MLT, the utilized concentration of luzindole (5 µmol/L) is considerably lower than the utilized concentration of MLT (1 mmol/L). Therefore, luzindole might be outnumbered by MLT. Furthermore, in addition to MT1/2, the quinone reductase enzyme family (MT3) is the natural receptor by MLT. 46 It is possible that MLT exerts its function via binding to its natural receptor MT3. In conclusion, we do not exclude the possibility that MLT degrades AML1-ETO protein through binding to its natural receptor. Several articles demonstrate that MLT reduces the side-effects of clinical drugs. For example, MLT alleviates the side-effects of chemotherapy in breast cancer 47 and metabolic side-effects of olanzapine in schizophrenia. 48 Furthermore, MLT as a dietary supplement has been used to improve the night time sleep for many years. No obvious side-effects of MLT as a dietary supplement for people are reported. Most importantly, our results demonstrate that MLT has little effects on the colony formation in normal human and murine HSPCs. However, whether long-term MLT treatment results in drug dependence and endocrine dyscrasia should be investigated.
The main shortfall of this study is that MLT decreases AML1-ETO in high concentration (mM). This high concentration is not easily achieved under normal physiological conditions. Thus, nanoparticlebased drug delivery should be used to maintain release, decrease the dose of MLT, and enhance treatment efficiency. 49 In addition, although MLT presents anti-LSC activity in leukaemia cells carrying AML1-ETO, whether MLT inhibits self-renewal of LSC in leukaemia without AML1-ETO is still unknown. Therefore, MLL-AF9-induced murine leukaemia is required for the further study. 50 Elimination of the initiating event in leukemogenesis has proved to be an extremely effective therapeutic strategy, such as Gleevec for BCR-ABL in chronical myeloid leukaemia 51 and alltrans retinoic acid for AML with PML-RARα. 52