Targeting miR-126 in inv(16) acute myeloid leukemia inhibits leukemia development and leukemia stem cell maintenance

Abstract

Acute myeloid leukemia (AML) harboring inv(16)(p13q22) expresses high levels of miR-126. Here we show that the CBFB-MYH11 (CM) fusion gene upregulates miR-126 expression through aberrant miR-126 transcription and perturbed miR-126 biogenesis via the HDAC8/RAN-XPO5-RCC1 axis. Aberrant miR-126 upregulation promotes survival of leukemia-initiating progenitors and is critical for initiating and maintaining CM-driven AML. We show that miR-126 enhances MYC activity through the SPRED1/PLK2-ERK-MYC axis. Notably, genetic deletion of miR-126 significantly reduces AML rate and extends survival in CM knock-in mice. Therapeutic depletion of miR-126 with an anti-miR-126 (miRisten) inhibits AML cell survival, reduces leukemia burden and leukemia stem cell (LSC) activity in inv(16) AML murine and xenograft models. The combination of miRisten with chemotherapy further enhances the anti-leukemia and anti-LSC activity. Overall, this study provides molecular insights for the mechanism and impact of miR-126 dysregulation in leukemogenesis and highlights the potential of miR-126 depletion as a therapeutic approach for inv(16) AML.

Fig. 1. Cbfb-MYH11 (CM) upregulates miR-126 expression in HSPC.

a Relative levels of miR-126 in CD34⁺ and CD34⁻ cells from healthy (HL; black circle; n = 6; HL CD34⁺ vs. HL CD34⁻ p = 0.0036) donors and inv(16) AML patients (red square; n = 10; inv(16) CD34⁺ vs. HL CD34⁺ p = 0.024; inv(16) CD34⁻ vs. HL CD34⁻ p = 0.0053; inv(16) CD34⁺ vs. inv(16) CD34⁻ p = 0.0002). b Relative levels of EGFL7 in CD34⁺ and CD34⁻ cells from HL donors (black circle; n = 6; HL CD34⁺ vs. HL CD34⁻ p = 0.016) and inv(16) AML patients (red square; n = 10; inv(16) CD34⁺ vs. HL CD34⁺ p = 0.006; inv(16) CD34⁺ vs. inv(16) CD34⁻ p < 0.0001). c Counts per million (CPM) reads for miR-126-3p in PB based on miRNA-seq of control (Ctrl; black line; n = 7) and CM (red line; n = 6) mice over time until moribund with leukemia (two-way ANOVA analysis showed CM vs. Ctrl p < 0.0001). d Normalized reads per kilobase per million of transcript (RPKM) for Egfl7 in PB based on RNA-seq of Ctrl (black line; n = 7) and CM (red line; n = 6) mice over time (two-way ANOVA analysis showed Ctrl vs. CM p = 0.0003). e Relative levels of miR-126 in CM (red square) preleukemic (6 weeks after induction; left; n = 6) or leukemic (right; n = 6) vs. Ctrl (black circle; n = 6) HSPC populations, including LSK (preleukemic p = 0.0074; leukemic p = 0.049), GMP (preleukemic p = 0.0064; leukemic p = 0.043), Pre-Meg/E (preleukemic p = 0.00015; leukemic p = 0.0012), Pre-GM (preleukemic p < 0.0001; leukemic p = 0.0002), and EP (leukemic p < 0.0001). f Western blot for CBFβ and CM using anti-CBFβ in 32D cells transduced with CBFB vs. CM (left). Relative levels of primary pri-miR-126, precursor pre-miR-126, mature miR-126, and Egfl7 in 32D cells expressing CM (red dots/bars) vs. CBFβ (black dots, gray bars). g Western blot analysis for CM in 32D-CM non-silencing (NS) control vs. CM shRNA (A3, D4) cells (left). Relative levels of mature miR-126, pri-miR-126, pre-miR-126 and Egfl7 expression in NS (red), CM shRNA-A3 (blue) or CM shRNA-D4 (gray) cells (right). h Relative Gata2 expression in BM samples of control (black/gray; n = 5) vs. CM leukemia (red; n = 4; p = 0.0002) mice. i Relative level of Gata2 in leukemia BM cells transduced with NS control (red), Gata2 shRNA #2 (dark blue), Gata2 shRNA #4 (light blue). j Relative expression of miR-126 in leukemia BM cells transduced with NS control (red), Gata2 shRNA #2 (dark blue), Gata2 shRNA #4 (light blue). k EGFL7 promoter-firefly luciferase reporter activity normalized with renilla luciferase as internal control in 293T cells co-transfected with or without GATA2, RUNX1, CBFβ or CM expression vectors as indicated (gray: none; green: GATA2; light blue: RUNX1 + CBFβ; dark blue: RUNX1 + CBFβ + GATA2; orange: RUNX1 + CM; Red: RUNX1 + CM + GATA2). Each dot represents results from an individual sample and data are presented as the mean ± SEM in a, b, e, h; each line represents the trajectory of an individual mouse in c, d. Representative data of at least three independent experiments are shown in f, g, i, j and data are presented as the mean ± SD. Q-PCR data in a, b, e, f for pri-miR-126, pre-miR-126, Egfl7, and Gata2 were normalized to Beta-2-microglobulin (B2M) and miR-126 data were normalized to RNU44 (human) or snoRNA234 (mouse). The statistical significance for all comparisons shown was determined using two-tailed Student’s T tests (*p < 0.05; **p < 0.01; ***p < 0.001).

Fig. 2. CM upregulates miR-126 biogenesis through HDAC8/RAN-XPO5 axis.

a Western blot analysis of HDAC8 in 32D-CM cells with NS or Hdac8 shRNA (#1, #2) (left). Relative expression of pre-miR-126 and mature miR-126 in 32D-CM cells with NS (red) vs. Hdac8 shRNA#1 (light blue), Hdac8 shRNA#2 (dark blue) (right). b Relative expression in 32D-CM cells treated with vehicle (red) vs. HDAC8i (22d 2.5 μM or 5 μM; purple) for pre-miR-126 (left; 2.5 μM p < 0.0001; 5 μM p < 0.0001) and miR-126 (right; 2.5 μM p = 0.012; 5 μM p = 0.0055). c Representative image of IF co-staining of RAN (green) and HDAC8 (red) in 32D cells (left; scale bar 10 μm). IP with anti-IgG control or anti-HDAC8 followed by immunoblotting (IB) with anti-RAN antibodies (right). Representative results of two independent experiments with similar results are shown. d IP with anti-RAN and IB with antibodies for Ac-Lys, RCC1, XPO5, or RAN in 32D-CM cells not-treated (NT) or treated with DMSO vehicle or 22d (2 µM) (left). The levels of RAN-GTP determined by RAN activation assay (right). The densitometry of RAN-GTP levels measured from three assays are shown on the bottom. e IP with anti-RAN and IB with antibodies for Ac-Lys, RCC1, XPO5, or RAN (left), and RAN activation assay (right) in 32D-CM cells not-treated (NT) or transduced with shCtrl control or shCM for 24 h. f IP with anti-RAN and IB with antibodies for Ac-Lys, RCC1, XPO5, or RAN (left), and RAN activation assay (right) of BM cells isolated from Ctrl or CM mice. g Schematic model of the mechanism by which CM regulates miR-126/EGFL7 transcription as well as miR-126 biogenesis through enhancing HDAC8 activity which in turn promotes RAN-XPO5 mediated transportation. Pre-miR-126 was normalized to B2m and mature miR-126 was normalized to snoRNA234. Data are presented as the mean ± SD and statistical significance shown was determined using two-tailed Student’s T tests (*p < 0.05; **p < 0.01; ***p < 0.001).

Fig. 3. High miR-126 promotes survival of HSPCs and leukemia-initiation induced by CM.

a Relative Cbfb-MYH11 expression in control (Ctrl; black), CM (red; CM vs. Ctrl p = 0.0031), and CM/miR-126^Δ/Δ (blue; CM/miR-126^Δ/Δ vs. Ctrl p = 0.0006) preleukemic BM cells. b Relative expression levels of miR-126 in control (black), CM (red; CM vs. Ctrl p = 0.0062), and CM/miR-126^Δ/Δ (blue; CM vs. CM/miR-126^Δ/Δ p < 0.0001, CM/miR-126^Δ/Δ vs. Ctrl p < 0.0001) preleukemic BM cells. c Frequency of LSK (CM vs. Ctrl p < 0.0001; CM/miR-126^Δ/Δ vs. Ctrl p = 0.0009; CM vs. CM/miR-126^Δ/Δ p < 0.0001), myeloid/erythroid progenitor populations including Pre-GM (CM vs. Ctrl p < 0.0001; CM/miR-126^Δ/Δ vs. Ctrl p = 0.0046; CM vs. CM/miR-126^Δ/Δ p = 0.0012), GMP (CM vs. Ctrl p = 0.0002; CM/miR-126^Δ/Δ vs. Ctrl p = 0.0043), Pre-Meg/E (CM vs. Ctrl p = 0.0001; CM/miR-126^Δ/Δ vs. Ctrl p = 0.0028; CM vs. CM/miR-126^Δ/Δ p = 0.012), MP (CM vs. Ctrl p = 0.0001; CM/miR-126^Δ/Δ vs. Ctrl p = 0.017; CM vs. CM/miR-126^Δ/Δ p = 0.0079) in Ctrl (black), CM (red) and CM/miR-126^Δ/Δ (blue) preleukemic BM. Flow cytometry gating strategies for HSPC populations are shown in Supplementary Fig. 3A. d Frequency of LT-HSC (CM vs. Ctrl p < 0.0001; CM/miR-126^Δ/Δ vs. Ctrl p = 0.008), ST-HSC, and MPP (CM vs. Ctrl p < 0.0001; CM/miR-126^Δ/Δ vs. Ctrl p = 0.0009; CM vs. CM/miR-126^Δ/Δ p < 0.0001) subsets in control (black), CM (red), and CM/miR-126^Δ/Δ (blue) preleukemic BM. e Kaplan–Meier survival curve of induced CM (red line; n = 26; median survival 110 days) or CM/miR-126^Δ/Δ mice (blue line; n = 26; median survival 354 days) monitored up to 1 year. The statistical significance was determined using Log-rank (Mantel–Cox) test (p < 0.0001). f The hierarchical clustering heatmap is composed of log2-based RPKM values of the top 100 differential expression genes (≥1.5-fold change; p < 0.01) identified by RNA-seq analysis of CM/miR-126^Δ/Δ LSK (n = 5) vs. CM LSK (n = 5). The color scale is originated from the “redgreen” palette in Cluster 3.0 software, while the color intensity is related to the range of log2-based RPKM values (−3 to 3). The quasi-likelihood (QL) F-test (one-sided) with likelihood ratio test is carried out to determine the differentially expressed genes (DEG). The “Benjamini–Hochberg” approach is implemented for calculating adjusted p values in multiple comparisons. g Normalized enrichment score (NES) plot for top 10 most positively and negatively enriched Hallmark Signature pathways identified by GSEA. h Enrichment plots for most significantly enriched pathways (ordered by p value). The enrichment score (ES) is calculated based on the pre-ranked gene list obtained by “−log10(p value for DEG) × sign(log2(fold change))”. Nominal p value (one-sided) is estimated from background ES (derived from the random permutation of genes) by using the positive or negative portion of the distribution for the sign of ES. The adjusted significance of ES is calculated by false discovery rate (FDR) for multiple hypothesis testing. The detailed results for top 13 most enriched pathways are shown in Supplementary Table 2. Q-PCR data for Cbfb-MYH11 were normalized with Hprt and miR-126 was normalized to SnoRNA234 in a, b. Each dot in a–c represents result from an individual mouse and data are presented as the mean ± SEM; statistical significance was determined using two-tailed Student’s T-tests (*p < 0.05; **p < 0.01; ***p < 0.001).

Fig. 4. Expression of miR-126 enhances MYC activity through downregulating SPRED1, PLK2, and activating the ERK-MYC axis in CM-AML.

a Relative expression levels Spred1 (CM vs. Ctrl p = 0.015; CM/miR-126^Δ/Δ vs. Ctrl p = 0.0022; CM vs. CM/miR-126^Δ/Δ p = 0.0035) and Plk2 (CM vs. Ctrl p = 0.0002; CM vs. CM/miR-126^Δ/Δ p = 0.0042) normalized to level of B2m in Ctrl (black), CM (red) and CM/miR-126^Δ/Δ (blue) preleukemic LSK. b Representative immunostaining of p-MYC (S62; magenta) and p-ERK (cyan) in control and CM AML LSK (left; scale bar 10 μm); western blot of SPRED1, PLK2, p-ERK, p-MYC (S62), and MYC in control and CM AML lineage-negative (Lin⁻) BM cells (right). c Western blot of SPRED1, PLK2, p-ERK, p-MYC (S62) and MYC in CM vs. CM/miR-126^Δ/Δ Lin^- BM cells. d Relative expression in CM-AML cells expressing shCtrl (black) vs. shmiR-126 (red) for miR-126 (p = 0.0018) and Spred1 (p = 0.0028) normalized to snoRNA234 and B2m, respectively (left); western blot of SPRED1, p-ERK, p-MYC (S62), and MYC (right) 2 days after transduction with shCtrl or shmiR-126 (right). e Relative expression levels of E2f1 (p = 0.0072), Cdk4 (p = 0.0136), Npm1 (p = 0.0041) normalized to level of B2m in shCtrl (black) or shmiR-126 (red) expressing CM-AML cells. Each dot represents data from an individual mouse and presented as the mean ± SEM in a; data in d, e are presented as the mean ± SD and statistical significance for all comparisons shown was determined using two-tailed T tests (*p < 0.05; **p < 0.01; ***p < 0.001); representative results of two independent experiments with similar results are shown in b, c.

Fig. 5. Knockdown of miR-126 promotes apoptosis and reduces quiescence of primitive inv(16) AML cells.

a Relative levels of miR-126 in inv(16) AML CD34⁺ cells transduced with shCtrl (black) or shmiR-126 (red; p = 0.006), as assessed by qPCR and normalized with internal levels of RNU44. b Frequency (%) of G₀, G₁ or S/G₂/M phases of cell cycle in shCtrl (black) or shmiR-126 (red) transduced inv(16) AML CD34⁺ (G₀ p = 0.0073; G₁ p = 0.046; S/G₂/M p = 0.0094), CD34⁺CD38⁻ (G₀ p = 0.0069; G₁ p = 0.0184; S/G₂/M p = 0.0138), and CD34⁺CD38⁺ (G₀ p = 0.0072; G₁ p = 0.0015; S/G₂/M p = 0.0259) populations. c Frequency (%) of apoptotic cells defined by Annexin V⁺ in shCtrl (black) vs. shmiR-126 (red) transduced inv(16) AML CD34⁺ (p = 0.0057), CD34⁻ (p = 0.0261), CD34⁺CD38⁺ (p = 0.0027), and CD34⁺CD38⁻ (p = 0.0331) populations. d Representative FACS plots showing gating strategy and frequency of Annexin V/DAPI staining in CD34⁺ and CD34⁻ fractions of inv(16) AML samples transduced with shCtrl or shmiR-126. e Western blot of p-MYC (S62), MYC, BCL2, BAX in CD34⁺ cells from inv(16) AML patients (AML1298, 987, 1069) 2 days after transduction with shCtrl or shmiR-126. f Western blot of PARP and Cleaved Caspase 3 in CD34⁺ cells from inv(16) AML patients (AML1298, 987, 1069) 2 days after transduction of shCtrl or shmiR-126. g Relative levels of BCL2, BAX, BAK1 in CD34⁺ cells from inv(16) AML patients (AML#1298, 1069, 987) transduced with shCtrl (black) or shmiR-126 (red), as assessed by qPCR and normalized using internal levels of B2M. Data are presented as the mean ± SD. Two-way ANOVA showed p < 0.0001 for BCL2; p < 0.0001 for BAX, and p = 0.006 for BAK1. In a–c, each dot represents result from an individual patient and data are presented as the mean ± SEM; statistical significance shown was determined using two-tailed Student’s T tests (*p < 0.05; **p < 0.01; ***p < 0.001). Representative of two independent experiments with similar results are shown in e, f.

Fig. 6. Targeting miR-126 by miRisten increases AML cell apoptosis and delays CM-AML progression.

a Schematic of experimental design. AML cells isolated from CM induced leukemia mice were treated with miRisten or SCR control ex vivo and assessed for uptake, miR-126 expression, cell cycle, and apoptosis at various time points. b The frequency of uptake for miRisten (CpG-anti-miR-126^cy3; red) or vehicle (gray) in bulk or LSK cells from CM leukemia BM. c Relative levels of miR-126 in CM AML cells after treatment with SCR control (black) or miRisten (200 nM, 500 nM, 1 μM; pink/red; p < 0.0001 for all doses) as determined by qPCR and normalized to SnoRNA234 expression. d Frequency (%) of G₀, G₁, or S/G₂/M phases of cell cycle in CM AML cells after treatment with SCR control (black) or miRisten (red; G₀ p = 0.031; G₁ p = 0.0017). e Frequency (%) of apoptotic cells defined by Annexin V⁺ in CM AML cells after SCR (black) or miRisten (pink/red; p < 0.0001 for all doses) treatment. f Schematic of treatment regimen. CM AML cells (1 × 10⁶ cells/mouse) were transplanted into WT mice to generate a cohort of AML-bearing mice, which were randomly divided into two groups and treated with SCR control (20 mg/kg/dose, i.v., daily; n = 7) or miRisten (20 mg/kg/dose, i.v., daily; n = 7) for 21 days. g The frequency of uptake for miRisten (CpG-anti-miR-126^cy3; red) or vehicle (gray) in c-Kit⁺ AML blasts in BM or PB 1 h after intravenous (i.v.) treatment with miRisten 20 mg/kg. h Kaplan–Meier survival curve of CM leukemic mice treated with SCR control (black line; n = 7; median survival 78 days) or miRisten (red line; n = 7; median survival 154 days). Dotted line with gray shade indicates treatment window; statistical difference for survival curves was determined using Log-rank (Mantel–Cox) test (p = 0.0013). Each dot in c, d, e represents result from an individual mouse and data are presented as the mean ± SEM; statistics were calculated using two-tailed Student’s T tests (*p < 0.05; **p < 0.01; ***p < 0.001).

Fig. 7. Administration of miRisten effectively inhibits AML propagation and leukemia-initiating activity in inv(16) AML PDX model.

a Schematic of experimental design. Inv(16) AML PDX was established by directly injecting T cell-depleted primary AML cells (1 × 10⁶ cells/mouse) i.f. into irradiated NSGS and subsequently expanded into larger cohorts by secondary transplantation via i.v. injection. After 8 weeks, PDX mice were treated with SCR control (20 mg/kg/dose, i.v., daily) or miRisten (20 mg/kg/dose, i.v., daily) for 3 weeks and followed by assessment of human cell engraftment in BM and spleen. b Representative FACS plots showing gating strategy and frequency of hCD45⁺ cells in BM of mice treated with SCR control (top) or miRisten (bottom) for 3 weeks. c Representative FACS plots showing gating and frequency of hCD34⁺/CD45⁺ cells in BM of mice treated with SCR (top) or miRisten (bottom) for 3 weeks. d Frequency of hCD45⁺ AML cells in SCR (black) vs. miRisten (red) treated mice (n = 9/group) BM (p = 0.0222) or spleen (SP). e Frequency of hCD34⁺/CD45⁺ AML cells in SCR (black) vs. miRisten (red) treated mice (n = 9/group) BM (p = 0.01) or SP (p = 0.0385). f The Kaplan–Meier survival curve of mice treated with SCR (black line; n = 6; median survival 117 days) or miRisten (red line; n = 6; median survival 136 days; p = 0.0161) for 3 weeks starting at 8 weeks after transplant. Dotted line with gray shade indicates treatment window. g The Kaplan–Meier survival curve of second transplant recipients treated with SCR (black line; n = 5; median survival 137 days) or miRisten (red line; n = 5; median survival 199 days; p = 0.0127). The statistical significance for f, g was determined using log-rank (Mantel–Cox) test. Each dot in d, e represents result from an individual mouse and data are presented as the mean ± SEM; statistical significance was determined using two-tailed Student’s T-tests (*p < 0.05; **p < 0.01; ***p < 0.001).

Fig. 8. Anti-miR-126 therapy by miRisten effectively inhibits CM-AML burden and abrogates LSC activity.

a Schematic of experimental design. CM/tdTomato⁺ AML cells (1 × 10⁶) were transplanted into cohorts of WTwild-type syngenic mice. After 8 weeks, mice were treated with SCR control or miRisten (20 mg/kg/dose, daily for 3 weeks), or a “5 + 3” chemotherapy regimen consisted of Ara-C (50 mg/kg/day, 5 days) plus DNR (1.5 mg/kg/day, 3 days), or miRisten and “5 + 3” chemotherapy combined. AML engraftment was analyzed 7 days after the last dose of treatment and BM cells were transplanted into second recipients, which were analyzed for engraftment at 7 weeks or monitored for leukemia onset and survival. b Representative images of spleens from recipients of SCR control (Ctrl), miRisten, A + D, or miRisten/A + D group. c Spleen weight from treated recipients of SCR Ctrl (black; n = 10), miRisten (red; n = 8; miRisten vs. Ctrl p < 0.0001), A + D (blue; n = 8; A + D vs. Ctrl p < 0.0001), or miRisten/A + D (purple; n = 8; miRisten/A + D vs. Ctrl p < 0.0001; miRisten/A + D vs. miRisten p = 0.0004; miRisten/A + D vs. A + D p = 0.0337). d Representative FACS plots showing gating strategy and frequency of dTomato⁺ AML cells in the BM of SCR control, miRisten, A + D, or miRisten/A + D group. e Frequency of dTomato⁺ AML cells in the PB of mice treated with SCR control (black; n = 10), miRisten (red; n = 8; miRisten vs. Ctrl p = 0.0324), A + D (blue; n = 8; A + D vs. Ctrl p = 0.0012; A + D vs. miRisten p = 0.0047), or miRisten/A + D (purple; n = 8; miRisten/A + D vs. Ctrl p = 0.0018; miRisten/A + D vs. miRisten p = 0.014). f Frequency of dTomato⁺ AML cells in the SP of mice treated with SCR control (black; n = 10), miRisten (red; n = 8; miRisten vs. Ctrl p = 0.0103), A + D (blue; n = 8; A + D vs. Ctrl p < 0.0001), or miRisten/A + D (purple; n = 8; miRisten/A + D vs. Ctrl p < 0.0001; miRisten/A + D vs. miRisten p = 0.0004; miRisten/A + D vs. A + D p = 0.0071). g Frequency of dTomato⁺ AML cells in the BM of mice treated with SCR control (black; n = 10), miRisten (red; n = 8; miRisten vs. Ctrl p = 0.0038), A + D (blue; n = 8; A + D vs. Ctrl p < 0.0001), or miRisten/A + D (purple; n = 8; miRisten/A + D vs. Ctrl p < 0.0001; miRisten/A + D vs. miRisten p < 0.0001; miRisten/A + D vs. A + D p = 0.0291). h White blood cell (WBC) counts of second transplant recipients of SCR control (black; n = 10), miRisten (red; n = 10; miRisten vs. Ctrl p = 0.0225), A + D (blue; n = 10; A + D vs. Ctrl p = 0.0005), or miRisten/A + D (purple; n = 11; miRisten/A + D vs. Ctrl p = 0.0002) treated mice at 7 weeks. i Frequency of dTomato⁺ AML cells in the PB of second transplant recipients treated with SCR Ctrl (black; n = 10), miRisten (red; n = 10; miRisten vs. Ctrl p = 0.0358), A + D (blue; n = 10; A + D vs. Ctrl p = 0.0098), or miRisten/A + D (purple; n = 11; miRisten/A + D vs. Ctrl p < 0.0001; miRisten/A + D vs. A + D p = 0.0049) at 7 weeks. j The Kaplan–Meier survival curve of second transplant recipients treated with SCR Ctrl (black line; n = 10; median survival 84.5 days), miRisten (red line; n = 10; median survival 95.5 days; miRisten vs. Ctrl p = 0.0004), A + D (blue line; n = 10; median survival 84 days), or miRisten/A + D (purple line; n = 11; median survival 102 days; miRisten/A + D vs. Ctrl p = 0.0002; miRisten/A + D vs. miRisten p = 0.0386). The statistical significance was determined using log-rank (Mantel–Cox) test. Each dot in c, e, f, g, h, i represents data from an individual mouse and presented as the mean ± SEM; statistics for all comparisons shown were determined using two-tailed T-tests (*p < 0.05; **p < 0.01; ***p < 0.001).