Despite advances in defining the critical molecular determinants for leukemia stem cell (LSC) generation and maintenance, little is known about the roles of microRNAs in LSC biology. Here, we identify microRNAs that are differentially expressed in LSC-enriched cell fractions (c-kit+) in a mouse model of MLL leukemia. Members of the miR-17 family were notably more abundant in LSCs compared with their normal counterpart granulocyte-macrophage progenitors and myeloblast precursors. Expression of miR-17 family microRNAs was substantially reduced concomitant with leukemia cell differentiation and loss of self-renewal, whereas forced expression of a polycistron construct encoding miR-17-19b miRNAs significantly shortened the latency for MLL leukemia development. Leukemias expressing increased levels of the miR-17-19b construct displayed a higher frequency of LSCs, more stringent block of differentiation, and enhanced proliferation associated with reduced expression of p21, a cyclin-dependent kinase inhibitor previously implicated as a direct target of miR-17 microRNAs. Knockdown of p21 in MLL-transformed cells phenocopied the overexpression of the miR-17 polycistron, including a significant decrease in leukemia latency, validating p21 as a biologically relevant and direct in vivo target of the miR-17 polycistron in MLL leukemia. Expression of c-myc, a crucial upstream regulator of the miR-17 polycistron, correlated with miR-17-92 levels, enhanced self-renewal, and LSC potential. Thus, microRNAs quantitatively regulate LSC self-renewal in MLL-associated leukemia in part by modulating the expression of p21, a known regulator of normal stem cell function. Cancer Res; 70(9); 3833–42. ©2010 AACR.
MicroRNAs are small ∼22-nt regulatory RNAs (1) that serve important roles in a variety of normal and pathologic processes in a wide range of organisms. They function in association with the RNA-inducing silencing complex to control gene expression at the posttranscriptional level by mediating mRNA translational repression and/or triggering RNA degradation (2, 3). MicroRNAs may regulate the expression of approximately 25% to 50% of the mRNA transcriptome (4).
MicroRNAs have diverse and important roles in animals. They are differentially expressed during the development of multiple cellular lineages and respond to a variety of extracellular signals. Furthermore, misregulation of miRNAs has been implicated in the pathogenesis of various diseases, including hematopoietic malignancies (5). For example, miR-15 and miR-16, which are located in the commonly deleted region of chromosome 11 in B-cell chronic lymphocytic leukemia, are potential tumor suppressors whose decreased expression levels are associated with increased expression of their predicted target BCL2 (6), a suppressor of apoptosis. MicroRNAs also function as potent oncogenes, or so-called oncomirs. For example, enforced expression of miR-155 in hematopoietic progenitors leads to myeloproliferative disease and, in transgenic mice, induces B-cell lymphoma (7, 8).
The first oncomir to be characterized was miR-17-92 (called miR-17 polycistron), which encodes seven mature microRNAs: miR-17-5p, miR-17-3p, miR-18a, miR-19a, miR-20a, miR-19b, and miR-92-1. Increased expression of the miR-17 polycistron is a feature of lymphomas with amplification of the C13orf25 genes at chromosome 13q31.3. Experimental overexpression of the miR-17 polycistron in a c-Myc transgenic mouse model accelerates lymphoma onset, which was subsequently shown to result from the miR-19a repression of Pten to promote survival (9–11). Moreover, enforced expression of the miR-17 polycistron promotes proliferation in chronic myelogenous leukemia and B-cell lymphoma cell lines by targeting the expression of the cyclin-dependent kinase inhibitor p21 (12, 13). Increased dosage of individual members of the polycistron, including miR-17-5p, miR-20a, or miR-106a, which share the same seed sequences, effectively blocks human monocytic differentiation through suppression of the AML1 proto-oncogene (14). Taken together, these studies show that the contribution of individual miR-17 microRNAs and their respective targets vary in different hematologic lineage neoplasms. More recently, miR-17 polycistron miRNAs are highly expressed in human MLL leukemias, although the functional consequence of this remains to be determined (15, 16).
Here, we show that the miR-17 polycistron regulates leukemia stem cell (LSC) potential in a mouse model of MLL-associated acute myeloid leukemia (AML) by modulating the expression of the cyclin-dependent kinase inhibitor p21. Polycistron expression is downregulated upon the exit of LSCs from the self-renewing compartment, whereas forced expression blocks myeloid leukemia cell differentiation, enhances proliferation, and significantly decreases leukemia latency in vivo. Knockdown of p21 phenocopies miR-17 polycistron overexpression in MLL leukemia cells, validating p21 as a biologically relevant in vivo target and a downstream rate-limiting regulator of LSC potential.
Materials and Methods
C57BL/6 mice congenic for CD45 (Ly5.1/Ly5.2) were used for transplant studies. All experiments on mice in this study were performed with the approval of and in accordance with Stanford University's Administrative Panel on Laboratory Animal Care.
Retroviral constructs and hematopoietic progenitor transformation assays
Retroviral constructs encoding MLL-AF10 and the miR-17-19b polycistron have been previously described (9, 17, 18). Knockdown constructs for p21 (TRCN0000042583 and TRCN0000042586) were purchased from Open Biosystems. pBabe-c-myc-zeo was reported previously (19). Retroviral transductions and in vitro replating assays were performed essentially as previously described using primary murine myeloid progenitors harvested from the bone marrow (18). Cells transduced with retroviral vectors were selected for stable transduction in methylcellulose medium containing the appropriate antibiotic (1 μg/mL puromycin and/or 1 mg/mL neomycin and/or 0.3 mg zeocin/mL). The estrogen-inducible MLL-ENL system was used as described (20).
Transplantation experiments were performed as previously described (21) with the following minor modifications. For cotransduction experiments, transduced progenitors were incubated in 0.9% methycellulose medium containing cytokines [20 ng/mL SCF, 10 ng/mL interleukin (IL)-6, 10 ng/mL IL-3] in the presence of puromycin (1 μg/mL) and neomycin (1 mg/mL) for 5 d and were then transplanted (1 × 105 cells) together with a radioprotective dose of total bone marrow cells (1 × 105) into the retro-orbital venous sinus of 6- to 12-wk-old syngeneic C57BL/6 mice that had been lethally irradiated with 9.0 Gy of total body γ irradiation (135Cs). When transplanted mice exhibited signs of ill health (shortness of breath, lethargy, and hunched posture), they were euthanized. Donor and recipient cells were distinguished by fluorescence-activated cell sorting (FACS) analysis of CD45 congenic marker expression. Necropsy tissues were fixed in buffered formalin, sectioned, and stained with H&E for histologic analysis.
Flow cytometry analysis
Bone marrow and spleen cells were stained with fluorochrome-conjugated monoclonal antibodies against c-kit (2B8 clone), Mac-1 (M1/70 clone), and Gr-1 (RB6-8C5 clone) for the analysis of leukemic cell differentiation using antibodies purchased from Pharmingen or eBioscience and procedures described previously (18). FACS sorting conditions used to obtain highly purified populations for LSK (Lin−/c-kithigh/Sca-1high), CMP (Lin−/c-kithigh/Sca-1−/CD34+/FcγRlow), and GMP (Lin−/c-kithigh/Sca-1−/CD34high/FcγRhigh) were previously described (22). LSC-enriched and LSC-depleted leukemia cell populations were isolated by FACS based on c-kit expression using conditions previously described (23). DNA content analysis was performed by propidium iodide staining and analyzed by FACS. Apoptosis was quantified using an Annexin V apopotosis detection kit according to the manufacturer's instructions (BD Pharmingen).
Luciferase reporter assay
p21 wild-type and 3′ untranslated region mutant luciferase constructs were made as previously described (13). Two million MV4;11 cells were electroporated with 1 μg miR-17-19b DNA, 0.8 μg of the firefly luciferase reporter DNA, and 0.16 μg pRL-TK control DNA (Promega) using the Amaxa nucleofection technology (Amaxa) program A-30. Forty-eight hours later, firefly and Renilla luciferase activities were measured using dual-luciferase assays (Promega). Three independent experiments were carried out.
Leukemia cells were harvested and lysed in 250 μL 2× sample buffer. Proteins from ∼20 μL lysate were fractionated by electrophoresis through 12% SDS-polyacrylamide gels and were transferred to polyvinylidene fluoride membranes (Bio-Rad) using Tris-glycine SDS transfer buffer. After blocking with 5% milk, proteins were detected using an anti-p21 polyclonal antibody (c-19, Santa Cruz Biotechnology) or an anti-tubulin monoclonal antibody (DM-1a, Sigma).
Quantitative detection of coding RNAs and miRNAs
Reverse transcription-PCR (RT-PCR) profiling of individual miRNAs was performed according to the manufacturer's instructions (Applied Biosystems) using FACS purified c-kit+ and c-kit− cells from murine leukemias. cDNA was synthesized and subjected to real-time PCR using commercially prepared reagents according to the manufacturer's instructions. Taqman probes for the following miRNAs and genes were purchased from Applied Biosystems: has-miR142-5p (465), has-miR-14-23p (1189), has-miR17-5p (393), has-miR18a (2422), has-miR19a (395), has-miR19b (396), has-miR-20a (580), has-miR-16 (391), mmu-miR-106a (1128), has-miR10a (2288), has-miR196b (496), has-miR155 (479), has-miR-204 (508), snoRNA202 (1232), p19Arf (Mm01257348), Cdkn1a (p21; Mm01303209_m1); Cdkn1b (p27; Mm00438167_g1), Cdkn2a (Mm00494449_m1), Cdkn2c (Mm00483243_m1), Cdkn4b p15 (Mm00483241_m1), c-myc (Mm00501741_m1), E2F1 (Mm432939_m1), E2F3 (Mm01138833_m1), and β-Actin (Mm00607939_s1). Primers for mouse p16Ink4a were designed previously (24) and were purchased from Applied Biosystems. Expression levels of target mRNA transcripts and miRNAs were normalized using β-Actin and snoRNA 202, respectively, based on the crossing point values.
Expression profiling identifies differentially expressed microRNAs in MLL LSCs
Expression profiling was performed to identify miRNAs that are preferentially expressed in MLL LSCs. LSC-enriched subpopulations in AMLs initiated by the MLL-AF10 oncogene were prospectively isolated by flow cytometry based on their differential expression of c-kit (18, 23). High throughput RT-PCR profiling of >200 individual microRNAs revealed that ∼20% were expressed at >2-fold (P < 0.001) higher levels in LSC-enriched (c-kit+) versus LSC-depleted (c-kit−) AML cell fractions, whereas ∼10% displayed the opposite profile.4 The upregulated miRNAs included several (miR-10a, miR-20a, miR-17-5p, miR-106a, miR-155, miR-181a and miR-196b; Fig. 1A) that have previously been implicated in regulating normal and aberrant hematopoiesis (15, 25). MiR-17-5p and miR-20a are both encoded within the miR-17-92 polycistron, which has been implicated in B-cell lymphoma and the modulation of myeloid differentiation (9, 14). MiR-17 cluster miRNAs are also highly expressed in human MLL leukemias (16, 26), but the functional consequences are unknown. Therefore, we focused on the contributions of miR-17 cluster miRNAs in MLL-mediated leukemias.
Analysis of additional molecular subtypes of MLL leukemia (MLL-LAF4 and MLL-GAS7) showed a 2- to 3-fold higher level of miR-17-5p in c-kit+ versus c-kit- MLL leukemia cells (Fig. 1B) showing that miR-17 family microRNAs are consistently expressed at higher levels in the LSC-enriched fraction irrespective of the MLL fusion partner.
Using an estrogen-inducible system for controlling the activity of MLL-ENL, the levels of miR-17-5p, miR-106a, and miR-20a decreased dramatically (4- to 16-fold) 5 days after the inactivation of MLL-ENL following the withdrawal of tamoxifen, which caused loss of self-renewal and terminal differentiation (Fig. 1C). Thus, miR-17 miRNAs are preferentially expressed in self-renewing MLL leukemia cells.
Quantitative RT-PCR analysis was performed on prospectively isolated normal progenitor and differentiated myeloid cell populations to assess miRNA expression profiles within the hematopoietic compartment. Expression of miR-17-5p and miR-20a progressively increased with differentiation from multipotent progenitors (Lin- Sca-1+ c-kit+) through common myeloid progenitors (CMPs) and peaked at the granulocyte macrophage progenitor (GMP) stage. Expression levels dropped dramatically with further myeloid differentiation and were lowest in cells with the most mature myeloid immunophenotype (c-kit− Gr1+ Mac1+; Fig. 1D). The midmyeloid peak in expression was similar to that of genes critical for the maintenance of LSC self-renewal, suggesting that microRNAs of the miR-17 polycistron may be involved in regulating the transient self-renewal of normal myeloid progenitors as well as the enhanced self-renewal of MLL LSCs (27). Furthermore, the expression levels of miR-17-5p and miR-20a in the c-kit–positive leukemia cell population were 3- to 4-fold higher than in normal GMPs (Fig. 1D), indicating that abnormal expression of the miR-17 cluster is a feature of MLL LSCs.
Increased expression of a miR-17-19b construct accelerates the onset of MLL leukemia
To explore the possibility that the miR-17 polycistron may contribute to MLL-associated leukemia, a construct encoding all of its constituent miRNAs except miR-92b (miR-17-19b) was coexpressed along with MLL-AF10 to determine whether its enhanced expression affects MLL leukemogenesis. Myeloid progenitors (c-kit+) cotransduced with MLL-AF10 plus miR-17-19b (denoted MA10/miR17) exhibited miR-17-5p and miR-20a transcript levels that were 2- to 3-fold higher than progenitors cotransduced with MLL-AF10 plus empty vector (MA10/v; Fig. 2A). The LSC nonspecific miRNAs within the cluster (miR-18a, miR-19a, and miR-19b) showed minimal fold induction (<1.5); therefore, this construct faithfully expressed MLL LSC–specific miRNAs at a high level (Fig. 2A). Following the transplantation of 100,000 cotransduced cells with an equal number of total bone marrow radioprotective cells into lethally irradiated recipient mice, the MA10/miR17 cohort developed leukemia with substantially shortened latencies (median, 36 days) compared with mice transplanted with MA10/vcotransduced cells (median latency, 60 days; P < 0.001; Fig. 2B). Mice in both cohorts displayed similar pathologies with bone marrow effacement by blasts, enlarged spleens, and infiltration of the liver by leukemic cells (data not shown). Therefore, the miR-17 polycistron serves a rate-limiting role in the progression of MLL-associated leukemia.
The miR-17 polycistron blocks myeloid differentiation and increases LSC frequency
FACS analysis showed that all leukemias derived from transplanted cells exclusively displayed myeloid phenotypes (Mac1+ Gr1+), indicating that the miR-17 polycistron did not play an instructive role in influencing the lineage derivation of leukemia. However, MA10/miR17 leukemia cells displayed a more undifferentiated phenotype (high levels of c-kit and lower levels of Mac1), suggesting that the miR-17 polycistron blocked myeloid differentiation in MLL-induced leukemia (Fig. 3A). Morphologic assessment of splenocytes obtained at necropsy of MA10/miR17 leukemic mice revealed an ∼2-fold higher proportion of blast cells and substantially fewer differentiated macrophages and neutrophils (Fig. 3B) compared with splenocytes from MA10/v mice. Furthermore, in semisolid culture assays, the frequencies of colony-forming cells (CFCs), which correspond to LSCs (23, 27), in the spleens of leukemic mice were ∼2-fold more abundant for MA10/miR17 leukemias (Fig. 3C). Leukemia cells transduced by MA10/miR17 continued to self-renew at a higher rate generating ∼3-fold more colonies in a subsequent passage than cells transduced with MA10/v (Fig. 3C). Thus, MLL leukemias overexpressing the miR-17 polycistron exhibit less differentiation as well as a higher frequency of clonogenic leukemia cells with features of LSCs.
High-level expression of the miR-17 polycistron promotes cell cycle progression through the inhibition of p21
Cell cycle analysis of explanted leukemia cells growing in methylcellulose cultures revealed a ∼15% increase in the proportion of cells in S/G2/M for MA10/miR17 versus MA10/v leukemias (Fig. 4A). The higher fraction of cells in cycle correlated with substantially lower transcript and protein levels for the CDK inhibitor p21, but not other cell cycle regulators, across three pairwise, independently generated leukemias (Fig. 4C). The apoptosis rate as measured by Annexin V and 7-AAD staining was indistinguishable between MA10/miR17 versus MA10/v leukemias (data not shown). Because the seed regions (2–7 nucleotides of 5′ microRNA) of miR-17-5p and miR-20a share perfect complementarities with the 3′ UTR of p21 mRNA (Fig. 4B), this suggested that these microRNAs may directly regulate p21 to modulate cell cycle progression in MLL leukemia cells.
p21 is a critical downstream target of the miR-17 polycistron in MLL leukemia
p21 expression was knocked down in MLL-transduced cells to assess whether its silencing recapitulates the effect of mir-17-19b construct overexpression in MLL-mediated leukemia. Myeloid progenitors (c-kit+) cotransduced with an MLL oncogene plus p21 knockdown construct (MA10/p21kd) exhibited 40% to 60% lower p21 transcript and protein levels than progenitors cotransduced with the MLL-AF10 oncogene plus empty vector (MA10/v; Fig. 5A). Following transplantation of 50,000 cotransduced cells with 100,000 total bone marrow radioprotective cells into lethally irradiated recipient mice, the MA10/p21kd cohort developed leukemia with substantially shortened latencies (median, 58 days) compared with mice in the MA10/v cohort (median, 90 days; P < 0.001; Fig. 5B). Leukemias in both cohorts displayed similar pathologic features (data not shown), but explanted leukemia cells cotransduced with p21 kd constructs showed 10% to 15% increase in the percentage of cells in S/G2/M when compared with MA10/v leukemias (Fig. 5C). Moreover, reporter assays performed in the MV4;11 MLL leukemia cell line showed that expression of a luciferase reporter containing the p21 3′UTR was reduced by 40% when coexpressed with the miR-17-19b polycistron compared with the vector alone. The observed reduction was highly specific because mutations of the miR-17 seed sequences in the p21 3′UTR prevented comparable reporter activity reduction (Fig. 5D). Thus, p21 is a critical in vivo and direct target of the miR-17 polycistron in mediating MLL leukemias.
c-myc is a critical upstream regulator of miR-17 polycistron expression in MLL transformation
c-myc has been shown to directly bind and activate the expression of the miR-17 polycistron in a mouse lymphoma model (10). Therefore, we investigated if enforced expression of c-myc together with an MLL fusion oncogene would induce enhanced oncogenic properties through the miR-17 pathway. In a serial replating assay, coexpression of c-myc with MLL-AF10 resulted in at least 2-fold higher colony numbers in the second and third rounds of plating when compared with the MLL-AF10 + vector control–transduced cells (Fig. 6A). Moreover, the levels of MLL LSC–specific microRNAs miR-17-5p and miR-20a were upregulated by >3-fold (Fig. 6B). This is consistent with the finding that c-myc levels were 2-fold higher in LSC-enriched versus LSC-depleted leukemia cell populations (Fig. 6C). These results suggest that MLL fusion oncogenes interface with the c-myc network to deregulate the expression of the miR-17 cluster and maintain LSC self-renewal.
A single microRNA may regulate >100 transcripts and therefore could serve as a potential master regulator of the oncogenetic program of cancers, including leukemia (28). To identify candidate oncomirs that may contribute to MLL leukemogenesis, we screened for microRNAs differentially expressed in AML subpopulations enriched versus depleted for LSCs in a mouse model that recapitulates many of the pathologic and molecular features of human AML. Of several differentially expressed microRNAs, the miR-17 polycistron was found to be a rate-limiting regulator of MLL LSC potential through the modulation of p21 levels. Although the miR-17 polycistron has previously been associated with hematologic cancers and is implicated in targeting of p21, our current studies link this oncomir and CDKI in the molecular regulation of LSC self-renewal in AML for the first time.
MiR-17 family microRNAs contribute to MLL LSC maintenance
In AMLs induced by MLL oncoproteins, the c-kit+ fraction of leukemia cells is highly enriched for LSCs (23). In addition to marked differences in leukemogenic, clonogenic, and morphologic properties, over 5,000 genes are differentially expressed in the c-kit+ versus c-kit− cell fractions from AMLs induced by MLL-AF10 and several have been validated as critical for LSC maintenance (27). Therefore, we hypothesized that these functionally and molecularly distinct subpopulations might also differentially express miRNAs that contribute to LSC biology. Consistent with the substantial differences in gene expression, approximately one third of analyzed microRNAs displayed 2-fold or greater differences in expression (data not shown). One of the differentially expressed miRNAs is miR-155 (6-fold higher in the c-kit+ fraction), which has been shown to induce a myeloproliferative disease when sustainably expressed in hematopoietic stem cells (8). Its differential expression in LSCs in our model and its high level of expression in a subset of human AMLs including those with MLL aberrations (8) strongly support a pathogenic role in MLL leukemias. The largest group of differentially expressed miRNAs, however, comprised members of the miR-17 family, which are encoded in paralogous polycistrons such as miR-17-92. Although originally characterized for its role as an oncomir in human and mouse lymphomas, our studies implicate miR-17-92 and related microRNAs in AML pathogenesis consistent with their high expression in human leukemias containing MLL chromosomal aberrations (16, 26).
Previous studies have shown that the expression of miR-17 family members, including miR-17-5p, 20a, and 106a, decreases with progressive differentiation along the myeloid, megakaryocytic, and monocytic lineages (14, 29). Our studies further refine the normal expression profiles by showing that levels of miR-17-5p and miR-20a peak in GMPs and committed myeloid precursors, then rapidly decrease with terminal differentiation. This suggests that miR-17 family microRNAs may normally regulate the transient self-renewal and proliferative expansion of midmyeloid progenitors/precursors. However, expression of the miR-17-19b cluster alone in myeloid progenitors failed to induce immortalization in vitro or leukemia when transplanted in vivo (data not shown). Interestingly, the downstream progenitors that normally express the miR-17 polycistron constitute the normal cellular counterparts of MLL LSCs, which maintain their aberrant self-renewal following the conversion of a midmyeloid gene expression program into an LSC maintenance program (27). Thus, miR-17 family microRNAs may represent additional components of the LSC maintenance program rather than direct transcriptional targets of MLL oncoproteins.
c-myc regulates the miR-17 polycistron to modulate cell cycle control through p21
Ectopic expression of the miR-17-19b construct leads to the reduction of endogenous p21 levels in MLL leukemia cells as well as to the reduced expression of a transfected luciferase reporter gene harboring the p21 3′UTR region. Furthermore, knockdown of p21 partially recapitulates the phenotype of miR-17-19b polycistron overexpression, thus validating p21 as a physiologic and direct target in regulating LSC frequency and accelerating MLL leukemia. Because MLL oncoproteins can efficiently transform fetal liver progenitors harvested from miR-17 polycistron knockout mice, the polycistron is dispensable for MLL leukemogenesis (data not shown). MiR-17-5p family members miR-20a and miR-106b, which share an identical seed sequence with miR-17-5p, have also been shown to target p21 (30) and are both differentially expressed in MLL LSCs. As a CDKI, p21 inhibits the activity of cyclin-CDK2 or CDK4 complexes, thus regulating cell cycle progression at G1. Expression of p21 is normally induced during p53-dependent cell cycle arrest in response to a variety of stress stimuli such as DNA damage; however, MLL oncoproteins were reported to suppress p53-mediated p21 induction in response to ionizing radiation (31). MiRNAs of the miR-17 polycistron may therefore provide another level (posttranscriptional) of p21 suppression in addition to MLL oncoprotein blockade of p53 whose levels were not detectably altered comparing MA10/v- and MA10/miR17-induced leukemias (data not shown). p21 has recently been implicated in promoting LSC self-renewal through modulation of quiescence and DNA repair (32). However, its role in MLL LSCs is to antagonize self-renewal by inhibiting cell cycle progression. Recent studies show that miR-19 within the miR-17 polycistron is responsible for accelerating disease onset in the c-myc–mediated lymphoma mouse model (10, 11). Other proposed targets of miR-17 family members, including AML1 (14), were not downregulated in MLL-AF10 leukemias overexpressing the miR-17-19b polycistron (data not shown). Taken together, the findings suggest that different microRNAs within the cluster contribute to its oncogenicity and regulate different targets depending on cell type and the specific oncogenic stimulus.
Transcription of the miR-17 cluster is directly regulated by c-myc (33). Interestingly, c-myc is critical for MLL-mediated transformation (34). Indeed, c-myc expression is downregulated >2-fold in LSC-enriched versus LSC-depleted MLL leukemia cell populations consistent with comparable downregulation of the miR-17-19b cluster in these respective populations. Although E2F1 and E2F3 have been reported to be direct targets of the miR-17 polycistron and are capable of directly activating transcription of these miRNAs creating a negative feedback loop (33, 35, 36), levels of E2F1 and E2F3 were not downregulated in MLL-AF10 leukemias overexpressing the miR-17-19b polycistron (data not shown). These observations suggest a model in which MLL oncoproteins sustain the expression of c-myc, which in turn activates the miR-17 polycistron to help maintain LSC self-renewal.
MicroRNA associations with the Hox pathway
Hox proteins and their cofactors are integral components of the MLL leukemogenic pathway; therefore, microRNAs that regulate or are coexpressed with Hox genes are candidates for pathogenic roles in MLL leukemogenesis. Previous studies have shown that miR-204 targets Meis1 and Hoxa10, and its low expression levels inversely correlate with high levels of these targets in AMLs carrying NPM1 mutations (37). Our microRNA expression analyses are consistent with these observations because the levels of miR-204 were extremely low (Ct value, >35) in MLL leukemia cells expressing high levels of Meis1 and Hoxa10, but 2-fold higher in the cell fraction (c-kit−) depleted of LSCs (Fig. 1) that displays modest downregulation of Hox and Meis expression. We have shown previously that Meis1 is a critical and rate-limiting regulator of MLL LSC potential (18). MiR-10a, on the other hand, is highly coexpressed with Hox genes in AML with normal karyotypes (38). MiR-10a was the most differentially upregulated miRNA (∼20-fold) in MLL LSCs (Fig. 1), despite the fact that its neighboring HoxB genes were minimally expressed. MiR-10a has been implicated in affecting transformation susceptibility by positively controlling protein synthesis through stimulating ribosomal mRNA translation (39). MiR-196b, located between Hoxa9 and Hoxa10, is a common retroviral integration site and its expression is upregulated in MLL LSCs, consistent with its high level expression in leukemias with MLL translocations and its ability to induce proliferation and a partial block of myeloid cell differentiation (40). Based on their prior oncogenic associations and expression profiles, further studies are warranted to establish whether miR-10a and/or miR-204 may directly contribute to a broader genetic program that promotes or maintains MLL/Hox-mediated leukemogenesis.
In summary, miRNA expression profiling of MLL LSCs identified differentially expressed miRNAs, most notably members of the miR-17 family, which modulate differentiation state, self-renewal, cell cycle status, and LSC frequency at least in part through direct suppression of p21. This defines the mechanistic basis by which a hematopoietic oncomir influences LSC potential by modulating expression of a known regulator of stem cell self-renewal.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
We thank M. Ambrus, C. Nicolas, and K. Ochis for technical support; Lin He for providing the miR-17-19b construct; and Irene Blat, Andrea Ventura, and Tyler Jacks for providing and genotyping the miR-17 knockout fetal liver cells.
Grant Support: Children's Health Initiative of the Packard Foundation, grants from the NIH (CA55029 and CA116606) and the Leukemia and Lymphoma Society, and in part by a Croucher Foundation Research (P. Wong).
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.
P. Wong designed, performed and analyzed research and wrote the manuscript; M. Iwasaki, T.C.P. Somervaille, F. Ficara, C. Carico, and C. Arnold performed research; C-Z. Chen analyzed the research; and M.L. Cleary analyzed the research and edited the manuscript.
↵4C.A. & C-Z. C., in preparation.
- Received September 8, 2009.
- Revision received January 25, 2010.
- Accepted February 10, 2010.
- ©2010 American Association for Cancer Research.