Chromosomal aberrations and multiple genome-wide association studies (GWAS) have established a major hematopoietic quantitative trait locus in chromosome 6q23.3. The locus comprises an active enhancer region, in which some of the associated SNPs alter transcription factor binding. We now identify miR-3662 as a new functional driver contributing to the associated phenotypes. The GWAS SNPs are strongly associated with higher miR-3662 expression. Genome editing of rs66650371, a three-base-pair deletion, suggests a functional link between the SNP genotype and the abundance of miR-3662. Increasing miR-3662′s abundance increases colony formation in hematopoietic progenitor cells, particularly the erythroid lineage. In contrast, miR-3662 is not expressed in acute myeloid leukemia cells, and its overexpression has potent antileukemic effects in vitro and in vivo. Mechanistically, miR-3662 directly targets NF-κB–mediated transcription. Thus, miR-3662 is a new player of the hematopoietic 6q23.3 locus.

Significance: The characterization of miR-3662 has identified a new actor in the prominent hematopoietic quantitative trait locus in chromosome 6q23.3. The mechanistic insights into miR-3662′s function may reveal novel or only partially known pathways for normal and malignant hematopoietic cell proliferation. Cancer Discov; 6(9); 1036–51. ©2016 AACR.

This article is highlighted in the In This Issue feature, p. 932

Numerous elements located throughout the genome are involved in the complex regulation of hematopoietic development (1), including one specific genomic region that has been repeatedly highlighted as a master regulator of hematopoiesis: the chromosome 6q23.3 quantitative trait locus (QTL). Early studies showed associations between chromosomal abnormalities such as translocation breakpoints and deletions, on the one hand, and hematologic malignancies, on the other hand (2, 3). Thus, the concept that chromosome 6q23 harbors a locus or loci that regulate different aspects of hematopoiesis was already established when large-scale genome-wide association studies (GWAS) became available. These associations were confirmed and expanded in numerous independent GWASs to comprise phenotypes such as mean corpuscular volume (4–9), mean corpuscular hemoglobin (4–7, 9), erythrocyte count (4–7, 10, 11), monocyte count (9–10), platelet count (11–13), and fetal hemoglobin level (refs. 14–16; see Table 1 for details). The locus contains three protein coding genes, HBS1-like translational GTPase (HBS1L), v-myb avian myeloblastosis viral oncogene homolog (MYB), and Abelson helper integration site 1 (AHI1).

Table 1.

SNPs identified in GWASs and located in the 6q23.3 locus

SNPChromosomal positionTrait associated with SNP
rs7775698/(rs66650371) chr6:135418635 MCH, MCV (4) 
  RBC count, MCH (5) 
  MCH, MCV, MCHC, RBC count, platelet count (6) 
  MCH, MCV (10) 
rs7776054 chr6:135418916 MCH (7) 
rs9399137 chr6:135419018 Platelet count, platelet volume (11) 
  MCV, MCH, RBC count (4) 
  HBA2 level (14) 
  Platelet count (12) 
  RBC count, platelet count, monocyte count (10) 
  Fetal hemoglobin level (9) 
rs9373124 chr6:135423209 MCHC (7) 
rs4895441 chr6:135426573 MCHC (4) 
  WBC count (6) 
  MCV (7) 
rs9402686 chr6:135427817 MCV (8) 
rs9494145 chr6:135432552 Platelet volume (13) 
  MCV (5) 
rs9483788 chr6:135435501 RBC count, Hct (7) 
rs6569992 chr6:135452152 RBC count, MCV, MCH (5) 
SNPChromosomal positionTrait associated with SNP
rs7775698/(rs66650371) chr6:135418635 MCH, MCV (4) 
  RBC count, MCH (5) 
  MCH, MCV, MCHC, RBC count, platelet count (6) 
  MCH, MCV (10) 
rs7776054 chr6:135418916 MCH (7) 
rs9399137 chr6:135419018 Platelet count, platelet volume (11) 
  MCV, MCH, RBC count (4) 
  HBA2 level (14) 
  Platelet count (12) 
  RBC count, platelet count, monocyte count (10) 
  Fetal hemoglobin level (9) 
rs9373124 chr6:135423209 MCHC (7) 
rs4895441 chr6:135426573 MCHC (4) 
  WBC count (6) 
  MCV (7) 
rs9402686 chr6:135427817 MCV (8) 
rs9494145 chr6:135432552 Platelet volume (13) 
  MCV (5) 
rs9483788 chr6:135435501 RBC count, Hct (7) 
rs6569992 chr6:135452152 RBC count, MCV, MCH (5) 

NOTE: Summary of SNPs identified in independent GWASs to be associated with the respective traits (National Human Genome Research Institute Catalog of Published GWASs). The SNPs are all clustered in a region of only ∼40 kB: mean corpuscular hemoglobin (MCH), mean corpuscular volume (MCV), red blood cell (RBC), mean corpuscular hemoglobin concentration (MCHC), hemoglobin A2 (HBA2), white blood cell (WBC), hematocrit (Hct). The cited GWAS SNP rs7775698 shares its genomic location with a three-base-pair deletion, rs66650371. Of note, in our cohort we almost exclusively detected the deletion of rs66650371 and not rs7775698 (that is contained in the deletion) and therefore refer to this locus as rs66650371.

The region comprising the GWAS SNPs has been extensively studied. Functional dissection revealed strong erythroid-specific histone acetylation (17), RNA polymerase 2 and transcription factor binding (17, 18), indicating the presence of a regulatory element in the GWAS region. Indeed, some of the SNPs were shown to alter transcription factor binding sites (19). Furthermore, the region directly interacts with the promoter of the transcription factor MYB, with which it forms dynamic long-range chromatin looping, thereby regulating its expression (18, 19). Changes to the architecture of the locus lead to changes in the associated phenotypes (19, 20).

The proto-oncogene MYB has been shown to crucially influence hematopoietic development, including lineage commitment and differentiation (21–23). MYB especially affects the developmental process of the T-cell lineage (24–28) and megakaryopoiesis (29). However, the complexity of the phenotypes associated with the locus suggests the presence of hitherto unrecognized elements (30).

Enhancing our understanding of the biology of the 6q23.3 locus might reveal new mechanisms in the regulation of not only hematopoietic development, but possibly also leukemogenesis. As blocked differentiation of hematopoietic progenitor (HP) cells is a main feature of acute leukemias, factors regulating hematopoiesis under normal conditions can also play a role in leukemia.

miR-3662 Abundance Is Associated with the Risk Alleles of Several GWAS SNPs in the 6q23.3 Locus

Upon reviewing existing data on the 6q23.3 locus, we noticed a recently annotated microRNA located in intron 13 of the HBS1L gene: miR-3662 (31). We hypothesized that miR-3662 may provide a clue to the hitherto not fully explained associations of the 6q23.3 locus with the various hematologic parameters and may be causally involved in hematopoietic development.

To determine if any association exists between the QTL-defining SNPs within the 6q23.3 region and miR-3662 abundance, we genotyped four tag-SNPs in the locus in a cohort of 200 individuals without known hematologic diseases. Due to the high linkage disequilibrium, these four SNPs can account for the genotypes of all 10 reported hematopoietic GWAS SNPs (Fig. 1A; Table 1; Supplementary Table S1). We also determined miR-3662 abundance in these individuals. Of note, rs7775698 shares its genomic location with a three-base-pair deletion (rs66650371), with which it is in complete linkage disequilibrium. In our cohort, we almost exclusively detected the deletion of rs66650371 and not the cited GWAS SNP rs7775698, and therefore refer to this position as rs66650371. Strikingly, higher abundance of miR-3662 was associated with homozygosity for the risk alleles of all GWAS SNPs, represented by rs66650371, rs9402686, rs6569992, and rs9483788 (Fig. 1A). Meticulous dissection of the region previously identified rs66650371 as a binding site for erythropoiesis-related transcription factors (15, 19), especially all components of the activating LDB1 complex (GATA1, LDB1, TAL1, and ETO2; ref. 19), and as a major contributor to modulate the expression of MYB. In line with those functional data, query of the ENCODE transcription factor chip data hosted by the UCSC Genome Browser suggested only rs66650371 and rs9483788 exhibit transcription factor binding (Supplementary Fig. S1). For additional confirmation, we tested the alleles of both SNPs for differences in their binding potentials in electrophoretic mobility shift assays (EMSA; Fig. 1C; Supplementary Fig. S2). Only the deletion of rs66650371 strongly increased the binding affinity (Fig. 1C). Thus, rs66650371 may be a promising candidate to directly affect the abundance not only of MYB but also of miR-3662.

Figure 1.

The association of miR-3662 expression with tag-SNPs of the 6q23.3 locus and the effect of miR-3662 on growth and colony formation of HP cells. A, schematic depiction of the 6q23.3 locus (not drawn to scale). miR-3662 is located in intron 13 of HBS1L. In total, 10 SNPs identified in independent GWASs that are associated with various hematologic features are located in a ∼40-kB region between HBS1L and MYB (see Table 1). GWAS SNP rs7775698 shares its genomic location with a three-base-pair deletion, rs66650371. B, box plots depicting miR-3662 abundance with respect to the genotypes of four tag-SNPs in healthy controls (n = 200). The genotypes of the SNPs are indicated below the boxes (homozygous for ancestral allele, heterozygous, homozygous for variant allele). Due to the high linkage disequilibrium, the genotypes of the four SNPs are representative for the genotypes of 10 GWAS SNPs (see Supplementary Table S1). In our cohort, we almost exclusively detected the deletion of rs66650371 and not the cited GWAS SNP rs7775698 (that is contained in the deletion) and therefore refer to this locus as rs66650371. Box plots with median; *, P < 0.05; **, P < 0.005. P values were determined using the Kruskal–Wallace test and pairwise Wilcoxon test with adjusted P values. WT, wild-type. C, EMSA of rs66650731 to validate differences in the binding affinities of the ancestral and the variant alleles. Oligos: 20-bp sequence containing either the ancestral or the variant allele; NE, nuclear extracts of KG1a cells. The experiment was repeated twice, confirming the results. D, miR-3662′s abundance in HP cells and Kasumi1 cells with or without the three-base-pair deletion rs66650371. CRISPR/Cas9 technology was used to create HP cells and Kasumi1 cells either heterozygous or homozygous for the rs66650371 three-base-pair deletion. Individual clones were selected, and miR-3662′s abundance was determined by qPCR in RNA of successfully engineered cells. Depicted are two independent experiments of different clones (rs66650371+/+ vs. rs66650371+/−). E, qPCR showing the relative miR-3662 abundance at different stages of hematopoietic differentiation of HP cells. The abundance of miR-3662 is depicted relative to the abundance detected on day +4 of differentiation (set to 1). F, abundance of miR-3662 during the lineage-specific differentiation of HP cells. MiR-3662 expression was quantified every 4 days using qPCR. The experiment was repeated twice, confirming the results. EPO, erythropoietin. G, growth curves showing the effects of forced miR-3662 expression on HP cells of two healthy donors (HP cells #1 and #2) compared to scramble control. Cell growth was monitored over 12 days. **, P < 0.005 when comparing the final cell counts of miR-3662 versus scramble control. P values were determined using two-tailed Student t tests (equal variance). The experiment was repeated twice, confirming the results.H, colony counts showing the effects of forced miR-3662 expression on the colony formation of HP cells. Left, erythroid colonies (colony-forming units, erythroid [CFU-E] and burst-forming units–erythroid [BFU-E]) in red. Granulocyte/macrophage colonies (colony-forming units–granulocytes/macrophages [CFU-GM]) in gray. Middle/right, absolute colony counts of granulocyte/macrophage colonies and erythrocyte colonies comparing scramble- versus miR-3662–infected HP cells. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). I, representative colony assay images after forced expression or knockdown of miR-3662. n = 3 biological replicates.

Figure 1.

The association of miR-3662 expression with tag-SNPs of the 6q23.3 locus and the effect of miR-3662 on growth and colony formation of HP cells. A, schematic depiction of the 6q23.3 locus (not drawn to scale). miR-3662 is located in intron 13 of HBS1L. In total, 10 SNPs identified in independent GWASs that are associated with various hematologic features are located in a ∼40-kB region between HBS1L and MYB (see Table 1). GWAS SNP rs7775698 shares its genomic location with a three-base-pair deletion, rs66650371. B, box plots depicting miR-3662 abundance with respect to the genotypes of four tag-SNPs in healthy controls (n = 200). The genotypes of the SNPs are indicated below the boxes (homozygous for ancestral allele, heterozygous, homozygous for variant allele). Due to the high linkage disequilibrium, the genotypes of the four SNPs are representative for the genotypes of 10 GWAS SNPs (see Supplementary Table S1). In our cohort, we almost exclusively detected the deletion of rs66650371 and not the cited GWAS SNP rs7775698 (that is contained in the deletion) and therefore refer to this locus as rs66650371. Box plots with median; *, P < 0.05; **, P < 0.005. P values were determined using the Kruskal–Wallace test and pairwise Wilcoxon test with adjusted P values. WT, wild-type. C, EMSA of rs66650731 to validate differences in the binding affinities of the ancestral and the variant alleles. Oligos: 20-bp sequence containing either the ancestral or the variant allele; NE, nuclear extracts of KG1a cells. The experiment was repeated twice, confirming the results. D, miR-3662′s abundance in HP cells and Kasumi1 cells with or without the three-base-pair deletion rs66650371. CRISPR/Cas9 technology was used to create HP cells and Kasumi1 cells either heterozygous or homozygous for the rs66650371 three-base-pair deletion. Individual clones were selected, and miR-3662′s abundance was determined by qPCR in RNA of successfully engineered cells. Depicted are two independent experiments of different clones (rs66650371+/+ vs. rs66650371+/−). E, qPCR showing the relative miR-3662 abundance at different stages of hematopoietic differentiation of HP cells. The abundance of miR-3662 is depicted relative to the abundance detected on day +4 of differentiation (set to 1). F, abundance of miR-3662 during the lineage-specific differentiation of HP cells. MiR-3662 expression was quantified every 4 days using qPCR. The experiment was repeated twice, confirming the results. EPO, erythropoietin. G, growth curves showing the effects of forced miR-3662 expression on HP cells of two healthy donors (HP cells #1 and #2) compared to scramble control. Cell growth was monitored over 12 days. **, P < 0.005 when comparing the final cell counts of miR-3662 versus scramble control. P values were determined using two-tailed Student t tests (equal variance). The experiment was repeated twice, confirming the results.H, colony counts showing the effects of forced miR-3662 expression on the colony formation of HP cells. Left, erythroid colonies (colony-forming units, erythroid [CFU-E] and burst-forming units–erythroid [BFU-E]) in red. Granulocyte/macrophage colonies (colony-forming units–granulocytes/macrophages [CFU-GM]) in gray. Middle/right, absolute colony counts of granulocyte/macrophage colonies and erythrocyte colonies comparing scramble- versus miR-3662–infected HP cells. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). I, representative colony assay images after forced expression or knockdown of miR-3662. n = 3 biological replicates.

Close modal

Utilizing HP cells of a healthy donor with heterozygosity for rs66650371 and the Kasumi1 cell line, which is homozygous for the rs66650371 deletion, we performed genome editing using lentiviral CRISPR/Cas9 technology. Deletion of the three base pairs in the heterozygous donor HP cells increased miR-3662′s abundance, whereas reinsertion of the three base pairs significantly lowered the abundance of miR-3662 in Kasumi1 cells (Fig. 1D), suggesting a direct functional link between the deletion and transcription of miR-3662.

Because several hematopoietic phenotypes are associated with the presence of the risk allele of rs66650371, and we demonstrated that the genotypes of rs66650371 modulate miR-3662 expression, we now set out to explore whether miR-3662 itself affects hematopoiesis and thus may contribute to the phenotypes of the 6q23.3 locus.

miR-3662 Accelerates Growth and Colony Formation of HP Cells

Monitoring miR-3662′s abundance over the course of hematopoietic differentiation showed an increased expression starting at day +12 (Fig. 1E; Supplementary Fig. S3). Lineage-specific quantification showed that miR-3662 expression increases in all three lineages, but is especially pronounced during erythroid differentiation (Fig. 1F). To test whether manipulation of miR-3662′s abundance influences the proliferation rate and/or course of differentiation of HP cells, we stably infected HP cells of two nonleukemic donors with either miR-3662 or scramble control using a lentiviral expression system. Cells infected with miR-3662 had higher absolute cell counts (Fig. 1G) and increased colony formation (Fig. 1G) compared to those expressing scramble control. The effects of miR-3662 were especially pronounced in the erythroid lineage, as miR-3662 colony-forming plates had a greater proportion of erythroid colonies when compared to scramble control (Fig. 1H).

We repeated the colony formation experiment to validate our findings using HP cells of a third nonleukemic donor. In addition to scramble control and miR-3662–infected cells, we added an siRNA-mediated knockdown experiment of miR-3662 starting at day +12 (which corresponds to the time when endogenous miR-3662 increases; see Fig. 1E). We again observed an increase in clonogenic potential, predominantly of the erythroid lineage, caused by forced expression of miR-3662 (Fig. 1I). In contrast, knockdown of miR-3662 led to reduced clonogenic potential (Fig. 1I).

Reintroduction of miR-3662 Reduces the Growth and Survival of Leukemic Cells

As the abundance of miR-3662 increased during development and had prodifferential effects on HP cells during normal hematopoiesis, we wondered whether miR-3662 may also play a role in malignant hematopoiesis and explored its function in acute myeloid leukemia (AML). We first used the existing sequencing data of small RNAs from The Cancer Genome Atlas (TCGA) AML TCGA cohort to gain insights into the expression of miR-3662 in patients with AML (32). As AML blasts are undifferentiated, the abundance of miR-3662 was, expectedly, extremely low in the entire AML patient set, with 58% of the patients having no detectable miR-3662 expression (Fig. 2A). To test whether miR-3662 abundance may be associated with specific blast phenotypes, we next analyzed the miR-3662 expression levels with respect to the French–American–British (FAB) classification system of acute myeloid leukemias (FAB M0-M7), which is based on the originating cell type and the degree of maturity (33). Curiously, the two patients with AML FAB M6, corresponding to acute erythroid leukemia, both had a comparatively high miR-3662 abundance (Fig. 2A). We performed additional quantitative PCRs (qPCR) in a set of four patients with M6 AML and four patients with non-M6 AML (OSU; see Methods for details). The results further supported this interesting observation (Fig. 2B), indicating that miR-3662 abundance in AML is extremely low, with the possible exception of erythroid leukemia.

Figure 2.

Distribution of miR-3662 abundance in AML and effects of forced expression on the growth and survival of AML cells in vitro and in vivo.A, miR-3662′s abundance in the patients with AML of the AML TCGA cohort. The small RNA-sequencing data from the AML TCGA were analyzed to quantify the abundance of miR-3662. The patients are grouped by their reported blast morphology according to the FAB classification (x-axis). Normalized read counts are depicted on the y-axis. B, comparison of the relative abundance of miR-3662 in patients with AML (OSU) with AML M6 versus non-M6 AML as determined by qPCR, using four patients in each group. Two separate cDNA syntheses were performed for each patient, and qPCRs were run in duplicate. The expression of miR-3662 was normalized to RNU6. The depicted relative abundance of miR-3662 with respect to the median abundance of the group of patients with non-M6 AML was estimated and compared between two groups by applying the Wilcoxon rank sum test. *, P < 0.05. C, chemiluminescent TiterGlo assay showing the effects of forced miR-3662 expression on the viability of AML cell lines. n = 3 biological replicates, displayed as mean ± SD. *, P ≤ 0.05. P values were determined using two-tailed Student t tests (unequal variance). D, flow cytometry after Annexin V staining showing the effects of forced miR-3662 expression on the apoptosis rates of AML cell lines. The red boxes indicate apoptotic cells and cells currently undergoing apoptosis. The percentage of both cell populations combined is shown enlarged. The experiment was repeated twice, confirming the results. E, methylcellulose-based colony-forming assays showing the effects of forced miR-3662 expression on the clonogenic potential of MV4-11 cells. n = 3 biological replicates, displayed as mean ± SD. *, P < 0.05. P values were determined using two-tailed Student t tests (equal variance). The lower panel depicts representative images of the colony-forming assays. F, chemiluminescent caspase-3/7 assays (top) and TiterGlo assays (bottom) showing the effects of forced miR-3662 expression on the apoptosis rates and viability of primary AML patient blasts. n = 3 biological replicates, displayed as mean ± SD. *, P < 0.05. P values were determined using two-tailed Student t tests (equal variance). G, flank tumor xenograft model of female athymic nude mice injected with MV4-11 cells stably expressing miR-3662 or scramble (n = 9 mice/group). Top, macroscopic images of tumors with the corresponding weight (in mg). Bottom, macroscopic images of the mice at day +42 after flank tumor injection. H, in vivo leukemogenesis model of female NSG mice injected with splenocytes stably expressing miR-3662 or scramble (n = 8 mice/group).**, P < 0.005. P values were determined using two-tailed Student t tests (unequal variance).

Figure 2.

Distribution of miR-3662 abundance in AML and effects of forced expression on the growth and survival of AML cells in vitro and in vivo.A, miR-3662′s abundance in the patients with AML of the AML TCGA cohort. The small RNA-sequencing data from the AML TCGA were analyzed to quantify the abundance of miR-3662. The patients are grouped by their reported blast morphology according to the FAB classification (x-axis). Normalized read counts are depicted on the y-axis. B, comparison of the relative abundance of miR-3662 in patients with AML (OSU) with AML M6 versus non-M6 AML as determined by qPCR, using four patients in each group. Two separate cDNA syntheses were performed for each patient, and qPCRs were run in duplicate. The expression of miR-3662 was normalized to RNU6. The depicted relative abundance of miR-3662 with respect to the median abundance of the group of patients with non-M6 AML was estimated and compared between two groups by applying the Wilcoxon rank sum test. *, P < 0.05. C, chemiluminescent TiterGlo assay showing the effects of forced miR-3662 expression on the viability of AML cell lines. n = 3 biological replicates, displayed as mean ± SD. *, P ≤ 0.05. P values were determined using two-tailed Student t tests (unequal variance). D, flow cytometry after Annexin V staining showing the effects of forced miR-3662 expression on the apoptosis rates of AML cell lines. The red boxes indicate apoptotic cells and cells currently undergoing apoptosis. The percentage of both cell populations combined is shown enlarged. The experiment was repeated twice, confirming the results. E, methylcellulose-based colony-forming assays showing the effects of forced miR-3662 expression on the clonogenic potential of MV4-11 cells. n = 3 biological replicates, displayed as mean ± SD. *, P < 0.05. P values were determined using two-tailed Student t tests (equal variance). The lower panel depicts representative images of the colony-forming assays. F, chemiluminescent caspase-3/7 assays (top) and TiterGlo assays (bottom) showing the effects of forced miR-3662 expression on the apoptosis rates and viability of primary AML patient blasts. n = 3 biological replicates, displayed as mean ± SD. *, P < 0.05. P values were determined using two-tailed Student t tests (equal variance). G, flank tumor xenograft model of female athymic nude mice injected with MV4-11 cells stably expressing miR-3662 or scramble (n = 9 mice/group). Top, macroscopic images of tumors with the corresponding weight (in mg). Bottom, macroscopic images of the mice at day +42 after flank tumor injection. H, in vivo leukemogenesis model of female NSG mice injected with splenocytes stably expressing miR-3662 or scramble (n = 8 mice/group).**, P < 0.005. P values were determined using two-tailed Student t tests (unequal variance).

Close modal

Next, we aimed to elucidate whether the introduction of miR-3662 into myeloid cells could affect cell growth and survival. We stably infected three AML cell lines (KG1a, MV4-11, and Kasumi1) with miR-3662 or scramble control and assessed cell proliferation, survival, and colony formation. In contrast to the effects observed in nonleukemic HP cells, increasing the abundance of miR-3662 in AML cell lines led to a reduction of cell proliferation (Fig. 2C), an increase in cell death (Fig. 2D), and a reduction in clonogenic potential (Fig. 2E).

Next, we infected primary leukemic blasts from 10 patients with AML (patients 1–10; Supplementary Tables S2 and S3) with either miR-3662 or scramble control. As determined by both caspase-3/7 chemiluminescent assays (patients 1–4) and TiterGlo viability assays (patients 2, 3, and 5), miR-3662infected cells showed increased cell death compared to scramble control (Fig. 2F).

miR-3662 Inhibits Tumor Growth and Leukemogenesis In Vivo

To test the effects of miR-3662 on leukemogenesis in vivo, we used an MV4-11 cell flank tumor model to compare the tumor growth of stably expressing miR-3662 or scramble controls in female athymic nude mice (n = 9 mice/group). Tumors with increased abundance of miR-3662 showed a significant reduction in volume and weight, and also a lower engraftment rate (Fig. 2G). In two of the five mice expressing miR-3662, there was no detectable tumor after 20 days (Fig. 2G). The remaining 4 mice/group were observed until day +42. None of the miR-3662 injected mice had a noteworthy tumor growth, thereby further supporting the antitumor effect of miR-3662 in our xenograft model (Fig. 2G).

To more fully investigate the potential of miR-3662 to modulate leukemogenesis in vivo, we stably overexpressed miR-3662 in murine-adapted MV4-11 cells (harvested splenocytes), which we propagated in nonobese diabetic severe combined immunodeficient (NOD/SCID) γ knockout (NSG) mice. miR-3662– and scramble-expressing cells were injected into female NSG mice (n = 8 mice/group; refs. 34, 35). Strikingly, all scramble-infected mice died of their leukemia before the first miR-3662 mouse. Specifically, overexpression of miR-3662 delayed the onset of disease and prolonged the median survival of the mice by 7 days (miR-3662 vs. scramble, 47.5 days vs. 40.5 days; Fig. 2H), thus significantly reducing the aggressiveness of the disease. Macroscopic pictures of the spleens of scramble- and miR-3662–expressing mice (showing leukemia-associated splenomegaly) and histologic images of the spleens stained for human CD45 are shown in Supplementary Fig. S4.

miR-3662 Is Involved in Multiple Developmental and Cancer-Associated Pathways

To identify the most important direct target genes of miR-3662, we next utilized a comprehensive targeted RNA sequencing approach with a panel of 361 genes (TruSeq RNA platform; Illumina) on miR-3662– versus scramble-infected leukemic blasts from two additional patients with AML (patients 11 and 12; Supplementary Table S2). Pathway analysis of the differentially expressed genes suggested cell death and survival (n = 72/76 molecules; Supplementary Table S4), cell cycle (n = 40/76 molecules), and cellular growth and proliferation (n = 62/76 molecules) to be the top-scoring molecular and cellular functions affected by miR-3662 expression.

A total of 44 genes showed concordant ≥20% decreases in mRNA abundance in the miR-3662–infected cells compared to scramble (Fig. 3A). Six out of the 44 downregulated genes have predicted miR-3662 binding sites and therefore qualified as potential direct miR-3662 target genes (Fig. 3A, indicated by stars). We tested the predicted miR-3662 binding sites in the 3′-UTRs of four of the six genes with the highest scores for their reactivity to forced miR-3662 expression in luciferase assays with and without mutation of the binding site. Two genes, the inhibitor of kappa light polypeptide gene enhancer in B-cells, kinase beta (IKBKB) and SMAD family member 2 (SMAD2), showed reactivity to miR-3662, with IKBKB exhibiting a stronger rescue after mutation of the predicted binding site (Fig. 3B). We then tested the 3′-UTRs of the genes for other miRs with predicted binding potential to assess putative competitive binding. Although 115 miRs were predicted to have higher binding potential to the SMAD2 3′-UTR compared to miR-3662, it was the highest-ranked miRNA to bind to IKBKB (Table 2). Thus, although the miR-3662–associated phenotype is likely caused by the effects of several target genes, IKBKB became a promising candidate as a main effector of the observed cellular effects of miR-3662 during hematopoiesis.

Figure 3.

Involvement of miR-3662 in multiple developmental and cancer-associated pathways. A, heat map of gene expression changes in the primary blasts of 2 patients with AML after stable introduction of miR-3662 or scramble control. Depicted in red are genes with a concordant ≥20% increase in mRNA abundance in both experiments. These genes are potential indirect targets of miR-3662. Depicted in green are genes with a concordant ≥20% decrease in mRNA abundance in both experiments. These genes are potential direct targets of miR-3662. A star indicates the presence of a predicted miR-3662 binding site in the 3′-UTR of the respective gene. B, luciferase reporter assays showing the effects of miR-3662 on four 3′-UTR constructs and binding site–mutated controls of predicted miR-3662 target genes. n = 3 biological replicates, displayed as mean ± SD. *, P < 0.05; **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). C, proposed downstream effects of miR-3662. D, confocal microscopy images depicting the cellular localization of phospho-p65 in blasts from patients with AML and AML cell lines. White scale bars, 20 μm. The experiment was repeated twice, confirming the results.

Figure 3.

Involvement of miR-3662 in multiple developmental and cancer-associated pathways. A, heat map of gene expression changes in the primary blasts of 2 patients with AML after stable introduction of miR-3662 or scramble control. Depicted in red are genes with a concordant ≥20% increase in mRNA abundance in both experiments. These genes are potential indirect targets of miR-3662. Depicted in green are genes with a concordant ≥20% decrease in mRNA abundance in both experiments. These genes are potential direct targets of miR-3662. A star indicates the presence of a predicted miR-3662 binding site in the 3′-UTR of the respective gene. B, luciferase reporter assays showing the effects of miR-3662 on four 3′-UTR constructs and binding site–mutated controls of predicted miR-3662 target genes. n = 3 biological replicates, displayed as mean ± SD. *, P < 0.05; **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). C, proposed downstream effects of miR-3662. D, confocal microscopy images depicting the cellular localization of phospho-p65 in blasts from patients with AML and AML cell lines. White scale bars, 20 μm. The experiment was repeated twice, confirming the results.

Close modal
Table 2.

miRs predicted to bind to the IKBKB 3′-UTR

Target rankTarget scoremiRNA name
89 miR-3662 
88 miR-1825 
85 miR-3120-5p 
84 miR-148a-3p 
83 miR-148b-3p 
Target rankTarget scoremiRNA name
89 miR-3662 
88 miR-1825 
85 miR-3120-5p 
84 miR-148a-3p 
83 miR-148b-3p 

NOTE: miR-3662 has the highest target score to regulate IKBKB (determined by the miRDB algorithm). The target score represents the likelihood of all known miRs to bind to the 3′-UTR of a tested gene.

IKBKB encodes the protein inhibitor of the kappa light polypeptide gene enhancer in B-cells, kinase beta (IKKβ), which phosphorylates nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha (IκBα), a direct inhibitor of nuclear factor κB (NF-κB). IKKβ-mediated phosphorylation of IκBα dissolves the IκBα/NFκB complex, allowing the freed NF-κB transcription factor to localize to the nucleus and activate its target genes (pathway graphically demonstrated in Fig. 3C; ref. 36). Thus, downregulation of IKBKB by miR-3662 would lead to stabilization of the IκBα/NF-κB complex and reduce the transcriptional activity of NF-κB. The NF-κB pathway belongs to the most crucial effectors during hematopoietic development (37) and is activated in many patients with AML (38). Confocal microscopy to visualize the expression of phospho-p65 in AML patient samples and cell lines depicts the baseline activation status of NF-κB in our specimens (Fig. 3D).

miR-3662 Directly Targets IKBKB, Thereby Inhibiting the Function of NF-jB

To see whether IKBKB is a direct downstream target of miR-3662, we determined the effect of forced miR-3662 expression on IKBKB mRNA levels using miR-3662–infected HP cells from healthy donors and leukemic blasts from patients with AML (patients 1–7; Table 2). Increasing miR-3662 abundance lowered IKBKB mRNA expression in both the HP cells and AML patient blasts (as determined by qPCR; Fig. 4A). This was validated at the protein level (Fig. 4B). Expectedly, the miR-3662–mediated decrease in IKBKB levels led to a decrease in phosphorylated IκBα (pIκBα; Fig. 4A).

Figure 4.

Identification of IKBKB as a direct target of miR-3662. A, effect of forced expression of miR-3662 on IKBKB abundance in HP cells and primary AML patient blasts. Left, relative IKBKB abundance determined by qPCR, displayed as mean ± SD. *, P < 0.05. P values were determined using two-tailed Student t tests (equal variance). Right, Western blots depicting IKBKB abundance and phosphorylation of IKBα (pIKBα). The experiment was repeated twice, confirming the results. B, effects of miR-3662 on the transcriptional activation potential of NF-κB. 3xkB-Luc shows the effects of miR-3662 versus scramble control on a luciferase construct containing three NF-κB binding motifs. 3xkBmut-Luc shows the effects of miR-3662 versus scramble control on a luciferase construct containing three mutated forms of the NF-κB binding motifs. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). C, effects of miR-3662 on the nuclear localization of NF-κB in various cell types visualized by confocal microscopy. DAPI (blue) shows the cell nuclei and NF-κB p65 (red) shows NF-κB. MERGE depicts the combined images of DAPI/NF-κB. White scale bars, 20 μm. D, effects of miR-3662 on the nuclear localization of NF-κB in KG1a cells visualized by Western blotting after subcellular fractionation. C, cytoplasmic fraction of KG1a lysates; N, nuclear fraction of KG1a lysates. The experiment was repeated twice, confirming the results. E, IKBKB-mediated rescue of miR-3662′s effects. Left, total colony counts after coinfection with miR-3662 in HP cells stably expressing IKBKB or scramble control. Middle and right, chemiluminescent TiterGlo assay showing the effects of forced miR-3662 expression on the viability of AML cell lines stably expressing IKBKB or scramble control. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). F, effects of miR-3662 on the nuclear localization of NF-κB in KG1a cells stably expressing IKBKB, visualized by confocal microscopy. DAPI (blue) shows the cell nuclei, and NF-κB p65 (red) shows NF-κB. MERGE depicts the combined images of DAPI/NF-κB. The white scale bars indicate a length of 20 μm. The experiment was repeated twice, confirming the results.

Figure 4.

Identification of IKBKB as a direct target of miR-3662. A, effect of forced expression of miR-3662 on IKBKB abundance in HP cells and primary AML patient blasts. Left, relative IKBKB abundance determined by qPCR, displayed as mean ± SD. *, P < 0.05. P values were determined using two-tailed Student t tests (equal variance). Right, Western blots depicting IKBKB abundance and phosphorylation of IKBα (pIKBα). The experiment was repeated twice, confirming the results. B, effects of miR-3662 on the transcriptional activation potential of NF-κB. 3xkB-Luc shows the effects of miR-3662 versus scramble control on a luciferase construct containing three NF-κB binding motifs. 3xkBmut-Luc shows the effects of miR-3662 versus scramble control on a luciferase construct containing three mutated forms of the NF-κB binding motifs. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). C, effects of miR-3662 on the nuclear localization of NF-κB in various cell types visualized by confocal microscopy. DAPI (blue) shows the cell nuclei and NF-κB p65 (red) shows NF-κB. MERGE depicts the combined images of DAPI/NF-κB. White scale bars, 20 μm. D, effects of miR-3662 on the nuclear localization of NF-κB in KG1a cells visualized by Western blotting after subcellular fractionation. C, cytoplasmic fraction of KG1a lysates; N, nuclear fraction of KG1a lysates. The experiment was repeated twice, confirming the results. E, IKBKB-mediated rescue of miR-3662′s effects. Left, total colony counts after coinfection with miR-3662 in HP cells stably expressing IKBKB or scramble control. Middle and right, chemiluminescent TiterGlo assay showing the effects of forced miR-3662 expression on the viability of AML cell lines stably expressing IKBKB or scramble control. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests (equal variance). F, effects of miR-3662 on the nuclear localization of NF-κB in KG1a cells stably expressing IKBKB, visualized by confocal microscopy. DAPI (blue) shows the cell nuclei, and NF-κB p65 (red) shows NF-κB. MERGE depicts the combined images of DAPI/NF-κB. The white scale bars indicate a length of 20 μm. The experiment was repeated twice, confirming the results.

Close modal

As our results indicate that miR-3662 has the ability to directly reduce IKBKB abundance, we next tested whether this reduction leads to changes in NF-κB's transactivation potential. We stably transfected miR-3662– or scramble-expressing 293 cells with a luciferase reporter gene fused to three tandem repeats of the κB site from the MHC class I enhancer (3xκB-Luc) or mutated κB sites (3xκBmut-Luc). Luciferase activity driven by the 3xκB-Luc was lower in miR-3662–expressing cells compared to scramble control, indicating a lower NF-κB activation potential in cells with higher miR-3662 abundance (Fig. 4B). In contrast, luciferase activity remained unaltered in the 3xκBmut-Luc cells (Fig. 4B). To assess possible changes in the cellular localization of NF-κB, we performed confocal microscopy on empty vector– and miR-3662–infected HP cells, leukemic blasts from AML patient 2, and the three AML cell lines. Although NF-κB could easily be detected in the nucleus and cytoplasm in the empty vector–transfected cells, far less nuclear NF-κB was detectable in the miR-3662–transfected cells (Fig. 4C). This result was confirmed in cellular fractionation blotting for NF-κB in the extracts of KG1a cells (Fig. 4D).

Taken together, our results indicate that miR-3662 inhibits IKKβ-mediated phosphorylation of the IκB proteins, thus reducing NF-κB's presence in the nucleus and its transactivation potential.

However, as the observed effects and associated phenotypes of miRs are usually the result of the modulation of several target genes (often in the same or related pathways) we aimed to test if some of the observed effects can be rescued by reintroduction of IKBKB. Coexpression of IKBKB in HP cells during colony formation and proliferation assays in KG1a and MV4-11 cells indeed reverted some but not all of the effects of miR-3662 on cell differentiation and growth (Fig. 4E). Reintroduction into KG1a and MV4-11 cells completely changed the cellular localization of NF-κB (Fig. 4F).

miR-3662 Has Its Own Regulatory Element and Can Be Activated by C/EBPα (p30) and GATA1

Next, we studied the upstream regulation of miR-3662. Using online prediction programs, we identified a possible transcription start site (TSS) located 1304–1354 bp upstream of the stemloop of miR-3662 (score: 1.0, TSS-3662; ref. 39).

To test the activation potential of the transcription start site, we performed luciferase reporter assays. The addition of TSS-3662 to a promoterless luciferase vector increased the luciferase signal 38-fold above baseline (Fig. 5A). Furthermore, chromatin immunoprecipitation (ChIP) assays of TSS-3662 performed on MV4-11 cell lysates showed an enrichment of RNA polymerase II (POL II), trimethylated lysine 4 of histone H3 (H3K4), and total histone H3, thus further supporting that TSS-3662 is an active transcription start site (Fig. 5B).

Figure 5.

Characterization of the upstream regulation of miR-3662. A, luciferase reporter assay showing the activating potential of the predicted transcription start site for miR-3662 (TSS-3662). n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests. B, chromatin immunoprecipitations of RNA polymerase II (POLII), trimethylated lysine 4 of histone H3 (H3K4) and total histone H3 (H3) in MV4-11 cells. Ribosomal protein L30 (RPL30) was used as a positive control. Alpha Satellite repeats were used as a negative control. n = 3 biological replicates, displayed as mean ± SD. C, luciferase reporter assay of TSS-3662 with cotransfection of C/EBPα, C/EBPβ, GATA1, and MZF1. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests. D, EMSA of TSS-3662 to test for binding of GATA1 and C/EBPα. Oligos: 20-bp sequence of TSS-3662 with predicted binding sequence; NE, nuclear extracts of KG1a cells; AB, respective antibodies used. E, effects of C/EBPα and GATA1 on the abundance of miR-3662 in KG1a cells. n = 3 biological replicates. *, P < 0.05. P values were determined using one-tailed Student t tests. F, graphical summary of the upstream regulation and downstream effects of miR-3662. Top, depiction of the 6q23.3 locus. miR-3662 is located in intron 13 of HBS1L and can be directly regulated by GATA1 and C/EBPα binding to its transcription start site. Homozygosity for the 3-bp deletion rs66650371 further enhances miR-3662′s abundance by creating several transcription factor binding sites. Bottom left, miR-3662 directly inhibits IKBKB, thereby preventing NF-KB's translocation into the nucleus. Bottom right, miR-3662′s effects on hematopoiesis and leukemogenesis.

Figure 5.

Characterization of the upstream regulation of miR-3662. A, luciferase reporter assay showing the activating potential of the predicted transcription start site for miR-3662 (TSS-3662). n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests. B, chromatin immunoprecipitations of RNA polymerase II (POLII), trimethylated lysine 4 of histone H3 (H3K4) and total histone H3 (H3) in MV4-11 cells. Ribosomal protein L30 (RPL30) was used as a positive control. Alpha Satellite repeats were used as a negative control. n = 3 biological replicates, displayed as mean ± SD. C, luciferase reporter assay of TSS-3662 with cotransfection of C/EBPα, C/EBPβ, GATA1, and MZF1. n = 3 biological replicates, displayed as mean ± SD. **, P < 0.005. P values were determined using two-tailed Student t tests. D, EMSA of TSS-3662 to test for binding of GATA1 and C/EBPα. Oligos: 20-bp sequence of TSS-3662 with predicted binding sequence; NE, nuclear extracts of KG1a cells; AB, respective antibodies used. E, effects of C/EBPα and GATA1 on the abundance of miR-3662 in KG1a cells. n = 3 biological replicates. *, P < 0.05. P values were determined using one-tailed Student t tests. F, graphical summary of the upstream regulation and downstream effects of miR-3662. Top, depiction of the 6q23.3 locus. miR-3662 is located in intron 13 of HBS1L and can be directly regulated by GATA1 and C/EBPα binding to its transcription start site. Homozygosity for the 3-bp deletion rs66650371 further enhances miR-3662′s abundance by creating several transcription factor binding sites. Bottom left, miR-3662 directly inhibits IKBKB, thereby preventing NF-KB's translocation into the nucleus. Bottom right, miR-3662′s effects on hematopoiesis and leukemogenesis.

Close modal

Analysis of predicted transcription factor binding sites in a 150-bp region surrounding TSS-3662 (TFsearch) revealed four transcription factors with a score of >85%: CCAAT/enhancer binding protein (C/EBP), C/EBP alpha (C/EBPα), C/EBP beta (C/EBPβ), GATA binding protein 1 (GATA1), and myeloid zinc finger 1 (MZF1). All of these transcription factors are known actors in hematopoiesis and AML.

When testing the activating potential of the transcription factors, only cotransfection with C/EBPα (p30) and GATA1 led to an increase in luciferase activity, indicating a possible physical interaction of C/EBPα (p30) and GATA1 with TSS-3662 (Fig. 5C). EMSAs validated the binding of both C/EBPα (p30) and GATA1 to TSS-3662, thereby further supporting the activating potential of both transcription factors for miR-3662 (Fig. 5D).

Finally, forced expression of C/EBPα (p30) or GATA1 led to an increase of miR-3662 abundance in KG1a cells (Fig. 5E). Overall, our results implicate C/EBPα (p30) and GATA1 as upstream regulators of miR-3662, which may additionally modulate miR-3662 expression in hematopoietic differentiation.

In summary, we characterize the complex upstream regulation and downstream effects of miR-3662, a player in hematopoiesis, which is located in the 6q23.3 locus (Fig. 5F).

In the past decade, numerous GWASs based on SNP arrays have illuminated our understanding of genomic space and enabled us to connect many human traits back to their underlying genomic changes (40, 41). However, often the association of an identified locus with the observed phenotype cannot be fully explained by the action of the genes located in the genomic region (41). Although early association studies mainly focused on the coding genes in identified regions, long-range enhancers and noncoding RNAs are now also considered as potential causative mechanisms (18, 19, 42, 43).

In an effort to identify new, crucial, and possibly targetable molecules involved in hematopoiesis and leukemogenesis, we revisited one of the most important genomic loci involved in hematopoietic differentiation and lineage commitment: the HBS1L–MYB region on chromosome 6q23.3. Strikingly, at least 15 independent GWASs identified one or more SNPs in the 6q23.3 region as a major candidate associated with hematologic traits (Table 1).

We now characterize the role of a newly annotated miRNA within intron 13 of the HBS1L gene: miR-3662.

Two partly overlapping stories evolved during our studies. The first is the identification of miR-3662 as a player in the 6q23.3 locus in nonleukemic individuals. Although GWAS studies identified rs66650371 to associate with higher hemoglobin levels and erythrocyte counts, our genome editing showed that rs66650371 directly affects the abundance of miR-3662. Closing the circle, our functional studies imply that miR-3662 increases erythroid differentiation. Thus, we formally conclude that miR-3662 causally contributes to the phenotypes associated with rs66650371. Mechanistically, we demonstrate that miR-3662 is regulated by the binding of GATA1 and C/EBPα to its own promoter, and that miR-3662′s abundance can be further increased in individuals who are homozygous for the deletion of rs66650371, as it creates additional transcription factor binding sites.

The second story is the antiproliferative effect of forced expression of miR-3662 in leukemia cells, which evolved out of the hypothesis that factors affecting normal hematopoiesis may also play a role in leukemia. We found that miR-3662′s abundance is low in both normal and malignant undifferentiated hematopoietic cells. While miR-3662 expression increases during normal hematopoiesis, and possibly contributes to successful hematopoietic development, AML blasts are caught in a differentiation block. Upregulation of miR-3662 never occurs. A curious exception is the higher expression of miR-3662 in erythroid leukemia, which is characterized by a more mature “erythroid” phenotype of the blasts.

Our overexpression and knockdown experiments revealed contrasting effects of miR-3662 on both normal and malignant hematopoiesis. Although the differential effects on cell growth and survival of nonleukemic and leukemic cells at first seem surprising, they may be explained by the strong prodifferentiation effects of miR-3662. This proposed mechanism is in line with previous reports of antiproliferative and/or proapoptotic effects of agents or genes that induce differentiation in leukemic blasts (42).

The identification of the classic NF-κB pathway as a potential downstream effector of miR-3662 is in accord with the well-known expression profile of NF-κB (p65) signaling during hematopoietic differentiation (36) and in AML (37, 38). The expression profile and activity of NF-κB is highly dependent on the cell type and the stage of differentiation (36, 43). Although the activity of NF-κB generally increases during the differentiation process, it shows a decrease during the course of erythroid differentiation (44). This matches with the identified expression profile of miR-3662, which increases during hematopoietic differentiation. Thus, although our results also suggest additional miR-3662 target genes, in both HP and AML cells, IKBKB is likely a major mediator of the miR-3662–associated phenotypes.

Our results go a long way toward clarifying an additional mechanism involved in the 6q23.3 locus. Furthermore, the mechanistic insights into miR-3662′s function may open up novel or only partially known pathways for normal and malignant proliferation. Our findings emphasize the role of noncoding genes and regulatory elements associated with phenotypic features.

Healthy Controls, AML Patient Samples, and CD34+ HP Cells

DNAs and RNAs extracted from the peripheral blood of 200 control individuals without known hematologic diseases were obtained from the Human Cancer Genetics Tissue Bank of the Ohio State University (OSU). Primary AML blasts from 12 patients with newly diagnosed AML were obtained from the OSU Leukemia Tissue Bank for functional studies. Additionally, RNAs from 8 patients with AML (4 patients each with AML FAB M6/AML FAB non-M6) were provided by the OSU Leukemia Tissue Bank for the determination of miR-3662 expression. All patients gave written consent according to the Declaration of Helsinki to use their tissue for studies, according to OSU institutional review board guidelines. Cytogenetic and molecular information about the patients can be found in Supplementary Tables S2 and S3. The endogenous abundance of miR-3662 and IKBKB in the primary blasts and cell lines used is shown in Supplementary Fig. S5. The successful overexpression of miR-3662 was validated using qPCR (Supplementary Fig. S6).

CD34+ HP cells from nonleukemic donors were obtained from the Tissue Bank of Cincinnati Children's Hospital.

Animal Models

MV4-11 cells were stably infected with miR-3662 or scramble control. Successful expression of miR-3662 was validated using qPCR. Per mouse, a total of 3.5 million cells in 75 μL RPMI were mixed with 75 μL Matrigel and injected subcutaneously into the left flank of female athymic nude mice 9 weeks of age. Tumor measurements were performed every third day starting at day 7 after injection. All athymic nude mice used for the experiment were bred within the Targeted Validation Shared Resource (TVSR) of The Ohio State University, and the xenograft experiments were performed within this shared resource and approved by the OSU Institutional Animal Care and Use Committee (IACUC) board.

To assess leukemogenesis in NSG mice, 6-week-old female mice (The Jackson Laboratory) were intravenously injected through the tail vein with 3.0 × 105 murine-adapted MV4-11 cells. Depending on the number of cells injected, these mice developed an aggressive leukemia 3 to 6 weeks after injection and died within 2 to 5 days after development of initial disease symptoms (34, 35). Overexpression of miR-3662 was validated using qPCR. Engraftment was validated by flow cytometric determination of CD45 in mice without external signs of disease 17 days after cell injection. Organ samples were harvested from the mice and processed for microscopy. These studies were performed in accordance with the OSU institutional guidelines for animal care and under protocols approved by the OSU IACUC (protocol number 2013A00000067).

Tissue Culture

KG1a (Lot#58683026) and Kasumi1 cells (Lot#60770973) were obtained from the American Type Culture Collection (ATCC) in January 2015, and MV4-11 cells (Lot#12; 2-7-14) were obtained from DSMZ on February 24, 2015, for studying miR-3662 in AML. All cell lines were authenticated by ATCC using a COI assay for interspecies determination, and short tandem repeat analysis for intraspecies determination of the unique DNA profiles. The cells were cultured in RPMI medium supplemented with 10% FBS and 1% antibiotic–antimycotic (Gibco). The patient cells were cultured in StemSpan media with added StemSpan100 cytokine cocktail (Stemcell Technologies). Cells from the 293TN cell line were obtained from SBI for luciferase assays and virus production. The cells were cultured in DMEM supplemented with 10% FBS, l-glutamine (200 mmol/L), and antibiotic–antimycotic agents (Life Technologies/Gibco) and grown at 37°C with 5% CO2. CD34+ HP cells from healthy donors were cultured in StemSpan media with added StemSpan100 cytokine cocktail (Stemcell Technologies). In certain experiments, M-CSF, G-CSF, and erythropoietin (EPO; all Stemcell Technologies) were added for lineage differentiation.

Genome Editing

To directly test the effects of rs66650371, a three-base-pair deletion, on the abundance of miR-3662, we performed genome editing on the Kasumi1 cell line, which is homozygous for rs66650371 and HP cells of a healthy donor heterozygous for rs66650371. Kasumi1 cells and HP cells were infected with lentiviral Cas9 nuclease (GeneCopoeia). After 72 hours, eGFP-positive cells were selected using the ARIAIII cell sorter. Next, single-stranded templates were introduced into the stably Cas9-expressing cells using transferrin-coated nanoparticles (ssOGN sequence: ACATCAGGATTAAAT TCACTCTGGACAGCAGATGTTATATCAAAATTACAAAATGTTATCAGGGCGGTTC; IDT). After 24 hours, the cells were infected with lentiviral rs66650371 sgRNA (mCherry, GeneCopoeia). The cells were semi-depleted after 48 hours and DNA was extracted for screening using the T7 endonuclease I assay kit (GeneCopoeia). Next, the cells were seeded on methylcellulose to ensure single-cell origin for each clone (50 cells/plate, 6 plates total). After 7 days, colonies were picked and transferred to a 96-well plate for further growth. At a final cell count of ∼100,000 cells/well, each well was semi-depleted and used for DNA extraction and genotyping. Cells from the well with successful genotyping were used for RNA extraction and determination of miR-3662 expression using qPCR.

RNA Extraction

RNA was extracted from cells using TRIzol Reagent (Life Technologies).

cDNA Synthesis and Gene/miRNA Abundance Analyses

RNA was harvested and reverse transcribed to cDNA using either the TaqMan MicroRNA Reverse Transcription Kit (Life Technologies Corporation/Applied Biosystems) or the Superscript III First-Strand cDNA Synthesis Kit (Life Technologies Corporation/Invitrogen). Both kits were used according to the manufacturer's instructions. The abundances of miR-3662 and its target genes were determined by qPCR.

Methylcellulose-Based Colony Forming Unit Assays

We added 100 μL of cells infected with either miR-3662 or scramble control to a 3-mL vial of pre-aliquoted MethoCult GF M3434 (StemCell Technologies) with recombinant cytokines and EPO. A total of 1.2 mL of the mixture, in duplicate, was used for the assay.

Forced Abundance of miR-3662, miR-3662 Knockdown, and Forced Abundance of IKBKB

miR-3662′s stemloop with 200-bp flanking sequence was cloned into an HIV-based lentiviral dual promoter vector (pCDH-CMV-MCS-EF1-copGFP+Puro cDNA; System Biosciences) for stable expression.

Primer sequences: miR-3662-clonF, gcgtGCTAGCgtgttttattttgtctgtatc; miR-3662-clonR, gcgtGGATCCtagaggtgaggtcctatatg. A custom-made antagomiR-3662 was purchased for targeted knockdown of miR-3662 (System Biosciences). A lentiviral scramble control miRNA was used as a negative control, according to the manufacturer's instructions (miRZiP000, System Biosciences). For stable overexpression of IKBKB, the ORF expression clone was purchased and handled according to the manufacturer's instructions (GeneCopoeia). The lentiviral construct (45 μg) was transfected into HEK-293TN cells, using 45 μL pPACKH1 and 55 μL PureFection (System Biosciences). The supernatant containing the pseudoviral particles was collected after 48 and 72 hours, and the virus was precipitated at 4°C overnight using 5X PEG-IT virus precipitation solution (System Biosciences). PBS (200 μL) and 25 μmol/L Hepes Buffer were used for resuspension of the pelleted virus. We infected 200,000 cells in triplicate with 20 IU of virus using 5 μL Transdux Infection Reagent (System Biosciences). RNA was harvested after 24 and 48 hours. For infection of primary patient blasts, 600,000 cells/mL were infected in triplicate with 20 IU virus, using 5μL Transdux Infection Reagent (System Biosciences). RNA was harvested after 48 and 72 hours. For overexpression studies in primary AML patient blasts, an increase in miR-3662 abundance was validated by qPCR in 7 out of 10 patients (patients 1–7). Five samples with both successful overexpression of miR-3662 and cell viability of ≥60% were selected for functional studies (patients 1–5).

The endogenous expression levels of AML cell lines (compared to nonleukemic HP cells during differentiation, total bone marrow from nonleukemic donors, and differentiated cells from the peripheral blood) as well as the expression levels achieved by the lentiviral expression construct are shown in Supplementary Fig. S3.

TiterGlo and Caspase-3/7 Assays

Cell viability and apoptosis changes in cell lines and primary patient blasts infected with lentiviral miR-3662 or scramble control were analyzed using the chemiluminescent Capase-3/7 and TiterGlo assays (both Promega) 72 hours after infection (primary cells) or puromycin selection (cell lines, 2.5 mg/mL) using 20,000 cells in duplicate of three biological replicates according to the manufacturer's instructions.

miR-3662 Promoter Analysis

Luciferase reporter constructs (∼50-bp genomic sequence) containing the predicted transcription start site for miR-3662 were cloned into the multiple cloning site of the promoterless luciferase reporter vector (pGL4.11, Promega) using the KpnI and SacI restriction sites. Primer sequences: TSS-3662-clonF, gcgtGGTACcaacatatcaagctcatag; TSS-3662-clonR, gcgtGCTAGCtacactgttccacatatga;TSS-3662-GATA1-mut, gcgtGGTACCaacagggcaagctcat; TSS-3662-CEBPA-mutR, gcgtGCTAGCtacactgttccacatat. Cells from the 293 cell line were transfected in triplicate with 250-ng luciferase reporter construct, 100-ng control construct (Renilla, pGL4.74, Promega), and 50 ng of the different expression constructs or empty pIRES2-EGFP vector control. Transfected cells were incubated for 24 hours at 37°C with 5% CO2 in Opti-MEM II medium containing the Lipofectamine/plasmid combination. Protein lysates were assessed for firefly luciferase and Renilla luciferase activities according to the recommendations detailed in the Dual-Luciferase Reporter Assay System (Promega). For further analysis, relative expression was normalized using cotransfected Renilla luciferase.

Confocal Staining and Microscopy

Confocal staining was performed 24 hours after transfection by standard procedures using the following antibodies: p65 (#8242S, Cell Signaling Technology; diluted 1:30 with normal goat serum), Phospho-p65 (sc-101752, Santa Cruz Biotechnology; diluted 1:30 with normal goat serum), and Alexa Fluor 647 goat anti-mouse and Alexa Fluor 488 donkey anti-rabbit (BD Biosciences; diluted 1:200 with normal goat serum). Confocal micrographs were taken using the FV1000 Confocal Laser Scanning Microscope (Olympus) with a UPLFLN 40× Oil, N.A. 1.3 lens.

TruSeq-Targeted RNA Analysis

A customized Add-on panel composed of 361 genes using the backbone of the Illumina apoptosis panel and the Illumina stem cell panel was designed using DesignStudio (Illumina) for the TruSeq Targeted RNA expression analysis. As potential targets to test, we included the genes on Illumina's preselected apoptosis and stem-cell panels and added all genes with a predicted target score for miR-3662 in the 3′-UTR of ≥99% as determined by miRDB (n = 57), and genes with a score of ≥85% that have known roles in hematopoiesis (n = 23). RNA from the primary blasts of 2 patients with AML (miR-3662 vs. scramble control) was extracted 72 hours after infection. Library preparations using 100-ng total RNA and the MiSeq run were performed according to the manufacturer's instructions. MiSeqReporter software was used to estimate target hits for each transcript after aligning reads against references specified by Targeted Oligo Pool, using banded Smith–Waterman alignment. The raw count data were then normalized using the R library DESeq (V 1.14.0), built based on negative binomial distribution, with variance and mean linked by local regression (20). The percentage of relative changes of mRNA expression of miR-3662 compared to scramble was estimated.

Chromatin Immunoprecipitation

ChIPs to determine histone modifications and Pol II enrichment were performed using the SimpleChIP Enzymatic Chromatin IP Kit (#9003, Cell Signaling Technology) according to the manufacturer's instructions, using qPCR-based analysis. Primer sequences used were TSS-3662F AACATATCAAGCTCATAG and TSS-3662R TACACTGTTCCACATATGA. Primers for RPL30 (positive control, #7014P) and alpha satellite repeats (negative control, #4486S) were purchased from Cell Signaling Technology. Antibodies used were Histone H3 (D2B12) XP (R) (#4620S), Tri-Methyl-Histone H3 (K4, C42D8; #9751S), Rpbl CTD (4H8; = Pol II, #2629S), and normal IgG rabbit AB (#2729, all Cell Signaling Technology).

EMSA

Nuclear proteins were extracted from KG1a cells using the Nuclear Extract Kit (Active Motif) according to the manufacturer's instructions. For EMSA, the Thermo Scientific LightShift Chemiluminescent EMSA Kit (Pierce/Thermo Fisher Scientific) was used according to the manufacturer's instructions. Antibodies used were GATA1 (M-20, sc-1234, Santa Cruz Biotechnology) and C/EBPα (14AA, sc-61, Santa Cruz Biotechnology). Oligo sequences (5′-biotinylation, manufactured at IDT): EMSA CEBPA F, 5′Biosg/ACATTACAAAAAGAG; EMSA CEBPA R, 5′Biosg/CTCTTTTTGTAATGT; EMSA_rs66650371 del in F, 5′Biosg/CAGATGTTACTATATCAAAA; EMSA_rs66650371 del in R, 5′Biosg/TTTTGATATAGTAACATCTG; EMSA_rs66650371 del out R, 5′Biosg/CAGATGTTATATCAAAA; EMSA_rs9483788T F, 5′Biosg/CTACTAAATATAGGATTTGT; EMSA_rs9483788T R, 5′Biosg/ACAAATCCTATATTTAGTAG; EMSA_rs9483788C F, 5′Biosg/CTACTAAATACAGGATTTGT; EMSA_rs9483788C R, 5′Biosg/ACAAATCCTGTATTTAGTAG.

Western Blots

Western blotting was performed according to standard procedures. Antibodies used were Actin (I-19) sc-1616, IKKβ (G-8) sc-271782, and pIkBα (Ser 32/36) sc-101713 (all Santa Cruz Biotechnology; primary antibodies diluted 1:10,000 with milk; secondary antibodies diluted 1:20,000 with milk).

Statistical Methods

Data are represented as mean ± SD of at least 3 independent experiments unless otherwise indicated and analyzed by the two-tailed or one-tailed Student t test. The means and SD were calculated and displayed in bar graphs as the height and the corresponding error bar, respectively. A P value of <0.05 was considered statistically significant.

No potential conflicts of interest were disclosed.

Conception and design: S.E. Maharry, D.C. Guttridge, A. de la Chapelle, A.-K. Eisfeld

Development of methodology: S.E. Maharry, X. Huang, D.C. Guttridge, A.-K. Eisfeld

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.E. Maharry, C.J. Walker, S. Mehta, M. Patel, M.A. Bainazar, M.A. Lankenau, K.W. Hoag, P. Ranganathan, R. Garzon

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.E. Maharry, C.J. Walker, S. Liyanarachchi, K.W. Hoag, R. Garzon, J.S. Blachly, D.C. Guttridge, A.-K. Eisfeld

Writing, review, and/or revision of the manuscript: S.E. Maharry, C.J. Walker, J.S. Blachly, C.D. Bloomfield, A. de la Chapelle, A.-K. Eisfeld

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.E. Maharry, M. Patel, J.S. Blachly

Study supervision: C.D. Bloomfield, A. de la Chapelle, A.-K. Eisfeld

Other (aided in accumulation of data through experimentation): S. Mehta

We thank Jan Lockman for technical support, Dr. David Lucas and Donna Bucci of the OSU Leukemia Tissue Bank at The Ohio State University Comprehensive Cancer Center, Columbus, OH, for sample processing and storage services, and The Ohio State University Comprehensive Cancer Center's Nucleic Acid and Microarray Shared Resources for technical support. We also thank the Target Validation Shared Resource for their support with the xenograft experiments, and Xiaomei Meng for her help with the NOD/SCID in vivo model.

This work was supported in part by the NCI (grants CA114725, CA140158, CA33601, CA16058, and CA095512), the Coleman Leukemia Research Foundation (A.-K. Eisfeld), and the Pelotonia Fellowship Program (A.-K. Eisfeld, S.E. Maharry, and C.J. Walker).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Attar
EC
,
Scadden
DT
. 
Regulation of hematopoietic stem cell growth
.
Leukemia
2004
;
18
:
1760
8
.
2.
Barletta
C
,
Pelicci
P-G
,
Kenyon
LC
,
Smith
SD
,
Dalla-Favera
R
. 
Relationship between the c-myb locus and the 6q- chromosomal aberration in leukemias and lymphomas
.
Science
1987
;
235
:
1064
7
.
3.
Harper
ME
,
Franchini
G
,
Love
J
,
Simon
MI
,
Gallo
RC
,
Wong-Staal
F
. 
Chromosomal sublocalization of human c-myb and c-fes cellular onc genes
.
Nature
1983
;
304
:
169
71
.
4.
Li
J
,
Glessner
JT
,
Zhang
H
,
Hou
C
,
Wei
Z
,
Bradfield
JP
, et al
GWAS of blood cell traits identifies novel associated loci and epistatic interactions in Caucasian and African-American children
.
Hum Mol Genet
2013
;
22
:
1457
64
.
5.
Kullo
IJ
,
Ding
K
,
Jouni
H
,
Smith
CY
,
Chute
CG
. 
A genome-wide association study of red blood cell traits using the electronic medical record
.
PloS ONE
2010
;
5
:
e1301
.
6.
Kamatani
Y
,
Matsuda
K
,
Okada
Y
,
Kubo
M
,
Hosono
N
,
Daigo
Y
, et al
Genome-wide association study of hematological and biochemical traits in a Japanese population
.
Nat Genet
2010
;
42
:
210
5
.
7.
Ganesh
SK
,
Zakai
NA
,
van Rooij
FJ
,
Soranzo
N
,
Smith
AV
,
Nalls
MA
, et al
Multiple loci influence erythrocyte phenotypes in the CHARGE Consortium
.
Nat Genet
2009
;
41
:
1191
8
.
8.
Soranzo
N
,
Spector
TD
,
Mangino
M
,
Kühnel
B
,
Rendon
A
,
Teumer
A
, et al
A genome-wide meta-analysis identifies 22 loci associated with eight hematological parameters in the HaemGen consortium
.
Nat Genet
2009
;
41
:
1182
90
.
9.
Menzel
S
,
Jiang
J
,
Silver
N
,
Gallagher
J
,
Cunningham
J
,
Surdulescu
G
, et al
The HBS1L-MYB intergenic region on chromosome 6q23.3 influences erythrocyte, platelet, and monocyte counts in humans
.
Blood
2007
;
110
:
3624
6
.
10.
Ferreira
MA
,
Hottenga
JJ
,
Warrington
NM
,
Medland
SE
,
Willemsen
G
,
Lawrence
RW
, et al
Sequence variants in three loci influence monocyte counts and erythrocyte volume
.
Am J Hum Genet
2009
;
85
:
745
9
.
11.
Shameer
K
,
Denny
JC
,
Ding
K
,
Jouni
H
,
Crosslin
DR
,
de And rade
M
, et al
A genome- and phenome-wide association study to identify genetic variants influencing platelet count and volume and their pleiotropic effects
.
Hum Genet
2014
;
133
:
95
109
.
12.
Gieger
C
,
Radhakrishnan
A
,
Cvejic
A
,
Tang
W
,
Porcu
E
,
Pistis
G
, et al
New gene functions in megakaryopoiesis and platelet formation
.
Nature
2011
;
480
:
201
8
.
13.
Qayyum
R
,
Snively
BM
,
Ziv
E
,
Nalls
MA
,
Liu
Y
,
Tang
W
, et al
A meta-analysis and genome-wide association study of platelet count and mean platelet volume in African Americans
.
PloS Genet
2012
;
3
:
e1002491
.
14.
Menzel
S
,
Garner
C
,
Rooks
H
,
Spector
TD
,
Thein
SL
. 
HbA2 levels in normal adults are influenced by two distinct genetic mechanisms
.
Br J Haematol
2013
;
160
:
101
5
.
15.
Farrell
JJ
,
Sherva
RM
,
Chen
ZY
,
Luo
HY
,
Chu
BF
,
Ha
SY
, et al
A 3-bp deletion in the HBS1L-MYB intergenic region on chromosome 6q23.3 is associated with HbF expression
.
Blood
2011
;
18
:
4935
45
.
16.
Thein
SL
,
Menzel
S
,
Peng
X
,
Best
S
,
Jiang
J
,
Close
J
, et al
Intergenic variants of HBS1L-MYB are responsible for a major quantitative trait locus on chromosome 6q23.3 influencing fetal hemoglobin levels in adults
.
Proc Natl Acad Sci U S A
2007
;
104
:
11346
51
.
17.
Wahlberg
K
,
Jiang
J
,
Rooks
H
,
Jawaid
K
,
Matsuda
F
,
Yamaguchi
M
, et al
The HBS1L-MYB intergenic interval associated with elevated HbF levels shows characteristics of a distal regulatory region in erythroid cells
.
Blood
2009
;
114
:
1254
62
.
18.
Stadhouders
R
,
Thongjuea
S
,
Andrieu-Soler
C
,
Palstra
RJ
,
Bryne
JC
,
van den Heuvel
A
, et al
Dynamic long-range chromatin interactions control Myb proto-oncogene transcription during erythroid development
.
EMBO J
2012
;
31
:
986
99
.
19.
Stadhouders
R
,
Aktuna
S
,
Thongjuea
S
,
Aghanjanirefah
A
,
Pourfarzad
F
,
van Ijcken
W
, et al
HBS1L-MYB intergenic variants modulate fetal hemoglobin via long-range MYB enhancers
.
J Clin Invest
2014
;
124
:
1699
710
.
20.
Suzuki
M
,
Yamazaki
H
,
Mukai
HY
,
Motohashi
H
,
Shi
L
,
Tanabe
O
, et al
Disruption of the Hbs1l-Myb locus causes hereditary persistence of fetal hemoglobin in a mouse model
.
Mol Cell Biol
2013
;
33
:
1687
95
.
21.
Emambokus
N
,
Vegiopoulos
A
,
Harman
B
,
Jenkinson
E
,
Anderson
G
,
Frampton
J
. 
Progression through key stages of haematopoiesis is dependent on distinct threshold levels of c-Myb
.
EMBO J
2003
;
22
:
4478
88
.
22.
Sandberg
ML
,
Sutton
SE
,
Pletcher
MT
,
Tarantino
LM
,
Hogenesch
JB
,
Cooke
MP
. 
c-Myb and p300 regulate hematopoietic stem cell proliferation and differentiation
.
Dev Cell
2005
;
8
:
153
66
.
23.
Sakamoto
H
,
Dai
G
,
Tsujino
K
,
Hashimoto
K
,
Huang
H
,
Fujimoto
T
. 
Proper levels of c-Myb are discretely defined at distinct steps of hematopoietic cell development
.
Blood
2006
;
108
:
896
903
.
24.
Badiani
P
,
Corbella
P
,
Kioussis
D
,
Marvel
J
,
Weston
K
. 
Dominant interfering alleles define a role for c-Myb in T-cell development
.
Genes Dev
1994
;
8
:
770
82
.
25.
Allen
RD
 III
,
Bender
TP
,
Siu
G
. 
c-Myb is essential for early T cell development
.
Genes Dev
1999
;
13
:
1073
8
.
26.
Pearson
R
,
Weston
K
. 
c-Myb regulates the proliferation of immature thymocytes following β-selection
.
EMBO J
2000
;
19
:
6112
20
.
27.
Bender
TP
,
Kremer
CS
,
Kraus
M
,
Buch
T
,
Rajewsky
K
. 
Critical functions for c-Myb at three checkpoints during thymocyte development
.
Nat Immunol
2004
;
5
:
721
9
.
28.
Lieu
YK
,
Kumar
A
,
Pajerowski
AG
,
Rogers
TJ
,
Reddy
EP
. 
Requirement of c-myb in T cell development and in mature T cell function
.
Proc Natl Acad Sci U S A
2004
;
101
:
14853
8
.
29.
Carpinelli
MR
,
Hilton
DJ
,
Metcalf
D
,
Antonchuk
JL
,
Hyland
CD
,
Mifsud
SL
, et al
Suppressor screen in Mpl -/- mice: c-Myb mutation causes supraphysiological production of platelets in the absence of thrombopoietin signaling
.
Proc Natl Acad Sci U S A
2004
;
101
:
6553
8
.
30.
Clappier
E
,
Cuccuini
W
,
Kalota
A
,
Crinquette
A
,
Cayuela
JM
,
Dik
WA
, et al
The c-MYB locus is involved in chromosomal translocation and genomic duplications in human T-cell acute leukemia (T-ALL), the translocation defining a new T-ALL subtype in very young children
.
Blood
2007
;
110
:
1251
61
.
31.
Hansen
TB
,
Bramsen
JB
,
Kjems
J
. 
Re-inspection of small RNA sequence datasets reveals several novel human miRNA genes
.
PLoS One
2010
;
5
:
e10961
.
32.
The Cancer Genome Atlas Research Network
. 
Genomic and Epigenomic Landscapes of Adult de novo Acute Myeloid Leukemia
.
N Engl J Med
2013
;
368
:
2059
74
.
33.
Bennett
JM
,
Catovsky
D
,
Daniel
MT
,
Flandrin
G
,
Galton
DA
,
Gralnick
HR
,
Sultan
C
. 
Proposals for the classification of the acute leukemias. French-American-British (FAB) co-operative group
.
Br J Hem
1976
;
3
:
451
8
.
34.
Ranganathan
P
,
Yu
X
,
Na
C
,
Santhanam
R
,
Shacham
S
,
Kauffman
M
, et al
Preclinical activity of a novel CRM1 inhibitor in acute myeloid leukemia
.
Blood
2012
;
120
:
1765
73
.
35.
Huang
X
,
Schwind
S
,
Yu
B
,
Santhanam
R
,
Wang
H
,
Hoellerbauer
P
, et al
Targeted delivery of microRNA-29b by transferrin-conjugated anionic lipopolyplex nanoparticles: A novel therapeutic strategy in acute myeloid leukemia
.
Clin Cancer Res
2013
;
19
:
2355
67
.
36.
Bottero
V
,
Withoff
S
,
Verma
IM
. 
NF-KB and the regulation of hematopoiesis
.
Cell Death Differ
2006
;
13
:
785
97
.
37.
Gasparini
C
,
Celeghini
C
,
Monasta
L
,
Zauli
G
. 
NF-KB pathways in hematological malignancies
.
Cell Mol Life Sci
2014
;
71
:
2083
102
.
38.
Baumgartner
B
,
Weber
M
,
Quirling
M
,
Fischer
C
,
Page
S
,
Adam
M
, et al
Increased IkappaB kinase activity is associated with activated NF-kappaB in acute myeloid blasts
.
Leukemia
2002
;
16
:
2062
71
.
40.
Sankaran
VG
,
Orkin
SH
. 
Genome-wide association studies of hematologic phenotypes: a window into human hematopoiesis
.
Curr Opin Genet Dev
2013
;
23
:
339
43
.
41.
Ikegawa
S
. 
A short history of the genome-wide association study: where we were and where we are going
.
Genomics Inform
2012
;
10
:
220
5
.
42.
Petrie
K
,
Zelent
A
,
Waxman
S
. 
Differentiation therapy of acute myeloid leukemia: past, present and future
.
Curr Opin Hematol
2009
;
16
:
84
91
.
43.
Hoesel
B
,
Schmid
JA
. 
The complexity of NF-KB signaling in inflammation and cancer
.
Mol Cancer
2013
;
12
:
86
.
44.
Zhang
MY
,
Sun
SC
,
Bell
L
,
Miller
BA
. 
NF-kappaB transcription factors are involved in normal erythropoiesis
.
Blood
1998
;
91
:
4136
44
.

Supplementary data