Noncoding RNAs in Acute Myeloid Leukemia: From Key Regulators to Clinical Players

Recent analyses have shown that human cells transcribe almost their entire genomes, implying the existence of a huge mass of ncRNAs. At the present, microRNAs are the most investigated regulative non-coding RNAs. Several studies have demonstrated that microRNAs play a crucial role in hematopoietic differentiation and hematological malignancies, including acute myeloid leukemia (AML). Aberrant expression of microRNAs has been associated with specific genetic abnormalities and clinical outcome of patients with AML. In addition, since microRNAs can function as either oncogenes or tumor suppressor genes, the potential of using these molecules as therapeutic targets opens up new opportunities in the future of AML therapy. The recent demonstration that other regulatory ncRNAs, in addition to microRNAs, are involved in hematopoietic cell differentiation and diseases, suggests that they may also have a biological relevance in AML. This paper will describe the role of ncRNAs in AML and discuss the expectations for the use of ncRNAs in diagnosis, prognosis, and therapy of AML.


Introduction
Traditionally biologists have concentrated their efforts on understanding the functions of coding genes. It may therefore be a little surprising that only a tiny fraction of the human genome encodes proteins, yet in contrast recent studies showed that the majority of our genome is transcribed into non-coding RNAs (ncRNAs) [1,2]. NcRNAs include highly abundant and functionally important RNAs, such as ribosomal RNAs (rRNAs), transfer (tRNAs), small nuclear RNAs (snRNAs), and small nucleolar RNAs (snoRNAs). However, two classes of recently discovered ncRNAs, microRltNAs (miRNAs) and long ncRNAs (lncRNAs), appear to play a sig-ni�cant role in the regulation of gene expression programmes that occur in higher eukaryotes [3][4][5]. ese ncRNAs may be involved in all levels of gene expression regulation within the cell and, eventually, they have also been implicated in many diseases, including cancers [6][7][8][9]. e role of these ncRNAs in normal and malignant myelopoiesis and their use as diagnostic and prognostic markers in acute myeloid leukemia (AML) are the subject of this paper.

MicroRNAs in Normal and Malignant Myelopoiesis
transcription factors [16,17]. miRNAs provide an additional level of control beyond the transcription factors. In particular, they play a crucial role in blood cell development by �ne-tuning differentiation and ad�usting the cell response to external stimuli [5,[18][19][20][21][22]. Acute myeloid leukemia (AML) is a heterogeneous hematopoietic malignancy in which immature myeloid progenitor cells accumulate in the bone marrow and eventually in blood and organs interfering with the production of normal blood cells [16,17]. In AML, the accumulation of leukemic cells (also referred as blasts) arises from a failure of myeloid progenitors to mature, and therefore AML used to be classi�ed in subtypes based on the stage at which normal differentiation is blocked in the leukemic blasts [16]. is failure is characterized by genetic and epigenetic alterations in progenitor cells that alter the expression or function of key transcription factors [17]. Noteworthy, cells derived from different AML subtypes can be induced to differentiate by speci�c agents into cells that resemble normal counterparts. In particular, acute promyelocytic leukemia (APL) represents a powerful in vitro model system to study granulopoiesis [16,23,24]. APL is characterized by chromosomal translocations involving the retinoic acid aeceptor (RAR ) gene resulting in clonal expansion of hematopoietic precursors blocked at the promyelocitic stage of differentiation [25]. e treatment of APL cells with all transretinoic acid (ATRA) overcome, this block and induces granulocytic differentiation [23]. First hints of the involvement of miRNAs in myeloid differentiation came from in vitro differentiation studies using APL cells [26]. e �rst miRNA found to play a critical role in APL differentiation was miR-223 [26]. miR-223 is preferentially expressed in myeloid cells [27] and is induced by ATRA treatment of APL cells through the transcription factors CCAAT/enhancer binding protein (C/EBP ) and PU.1 [26,[28][29][30]. ese proteins are key players in myelopoiesis as they regulate many myeloidcell-speci�c genes [17]. Conversely, miR-223 expression is negatively regulated by NFI-A and E2F1 [26,30]. NFI-A represses granulopoiesis and favours the development of erythrocytes [31], while E2F1, a critical regulator of the cell cycle, interferes with myeloid differentiation and promotes proliferation of myeloid progenitors [31,32]. Notably, these two genes are posttranscriptionally regulated by miR-223, thus generating a negative feedback loop [26,30]. Low levels of miR-223 have been reported in AML patients with the t(8; 21) chromosome translocation, which is responsible for the production of the fusion protein AML1-ETO (also known as RUNX1-ETO) and it has been shown that the AML1-ETO protein inhibits miR-223 expression through its binding to the promoter region and the recruitment of chromatin remodelling enzymes [33]. Importantly, treatment of cells carrying the AML1-ETO translocation with hypomethylating agents, AML1/ETO inhibitors, or ectopic expression of miR-223 was able to arrest proliferation and stimulate myeloid differentiation [26,31,33]. Altogether these data indicated that the deregulation of miR-223 might contribute to the differentiation block underlying myeloid leukemia pathogenesis.
Consistently with the data obtained in malignant myelopoiesis, in normal myeloid differentiation of human cord blood CD34 + hematopoietic progenitor cells (HPSs), miR-223 levels increased during granulocytic differentiation and decreased during erythroid maturation [34]. e decline of miR-223 is a critical event for the expansion of erythroblast cells. is is mediated, at least in part, by the translational repression of NFIA and LMO2 by miR-223 [31,34].
A miR-223 knock-out mouse has been produced [35]. e presence of miR-223 is dispensable for granulocyte cell fate speci�cation and its absence produced an increase of granulocyte progenitors and altered granulocyte immunological function. Interestingly, expansion of granulocyte progenitors has been also reported in conditional knock-out mice for C/EBP [36]. As C/EBP inhibits cell-cycle progression by interfering with E2F1 activity [37], these data suggest that downregulation of E2F1 by miR-223 (see above) could be important for the granulocytic maturation process.
e transcription factor Mef2c has been identi�ed as a crucial target of miR-223 in mouse myeloid precursors [35] and, indeed, conditional knock-out of Mef2c within the miR-223 knock-out mouse rescued the proliferation abnormality but not the differentiation defect and the functionality of granulocytes [35], suggesting that additional miR-223 targets might be responsible for these phenotypes.
AML cell lines have been also used to study the role of miRNAs in monocytic differentiation. Whereas ATRA treatment induces differentiation to morphologically and functionally mature granulocytes, 1,25-dihydroxyvitamin D3 (VitD3), phorbol myristate acetate (PMA), or 12-Otetradecanoylphorbol-13-acetate (TPA) induces a monocyte/macrophage phenotype [46][47][48]. Microarray analysis identi�ed different miRNAs modulated during monocytic differentiation of AML cell lines, in particular miR-424, miR-32, and miR-181 [49][50][51]. miR-424 is activated by the master myeloid transcription factor PU.1 and stimulated monocytic differentiation through translational repression of its target NFI-A [52]. In turn, the decrease in NFI-A levels is important for the activation of differentiation-speci�c genes such as M-CSF receptor (M-CSFr). In line with these data, both RNAi against NFI-A and ectopic expression of miR-424 enhanced monocytic differentiation [52,53]. e interplay among these three components was found in AML cell lines as well as in human CD34 + HPCs induced to differentiate into monocyte by cytokines [52,53]. Mir-32 is induced by VitD3 and negatively regulates the pro-apoptotic factor Bim [51]. Ectopic expression of miR-32 increased the differentiation response of AML cells to VitD3. In addition, miR-32 inhibition is sufficient to elevate Bim expression and sensitize AML cells to chemotherapy-induced apoptosis [51]. Conversely, miR-181 levels are decreased by VitD3 treatment [54]. is miRNA regulates the expression of the cell cycle regulator p27 kip1 , which is normally induced during monocytic differentiation, thus ectopic expression of miR-181 counteracts monocytopoiesis [54]. A relevant miRNAregulated molecular circuitry in monocytic differentiation has been characterised in human CD34 + HPCs [55]. is study showed that the miR-17-92 cluster was downregulated during cytokine-induced monocytopoiesis of human CD34 + HPCs. is produced upregulation of their target gene AML1 and the transcriptional activation of its transcriptional target M-CSF receptor. In addition, AML1 was shown to bind the miR-17-92 cluster promoter and to inhibit its transcription, thus generating a regulatory loop [55]. Notably, miR-17-92 downregulation is also observed upon PMAand TPA-induced differentiation of AML cells [56], and this cluster is also epigenetically silenced by PU.1 during monocytopoiesis in mouse [57], indicating the importance of miR-17-92 downregulation for myeloid differentiation. PU.1 also controls the expression of miR-146a, miR-342, miR-338, and miR-155 during monocytopoiesis of mouse myeloid progenitor cell lines [58]. erefore, in addition to controlling the expression of myeloid coding genes, the master transcription factor PU.1 exerts its crucial function in monocytic differentiation by modulating the expression of several miRNAs. Ectopic expression of miR-146a in mouse HSCs was able to direct the selective differentiation of these cells into functional macrophages [58]. In addition, miR-146a knock-out in mice led to a myeloproliferative disorder and eventually myeloid malignancies [59,60]. is phenotype has been attributed, at least in part, to NF-kB activation, a key event in in�ammation-induced carcinogenesis. us, miR-146 has been proposed to function as a tumor suppressor in the myeloid compartment [59,60].
e importance of miRNA in myeloid differentiation has been further proved by the deletion of the essential miRNA processing endoribonuclease Dicer in mouse myeloid progenitors [61]. is led to a block in monocytic differentiation and myeloid dysplasia, a cellular condition that may be considered as a pre-leukemic state.

Other Noncoding RNAs in Normal and Malignant Myelopoiesis
ere is a continually growing list of long non-coding RNAs (lncRNAs) that are associated with gene expression regulation and diseases [62]. However, very little is known about their precise function. LncRNAs generally indicate a RNA molecule longer than 200 nt without de�ned open reading frames, which can regulate gene expression through different molecular mechanisms [62]. To date, the number of human lncRNA genes is close to 9,000. However, only a few of them have been assigned a role in myelopoiesis. One of the �rst lncRNA to be identi�ed in the hematopoietic system was EGO [63]. is lncRNA was identi�ed in eosinophil differentiation of CD34 + HSCs where it stimulated differentiation and mature cell function by transcriptionally regulating eosinophil granule protein expression [62]. However, its mode of action has not yet been described. Subsequently, "antisense to PU.1 was discovered", a lncRNA produced by an antisense transcript to the master hematopoietic transcriptional regulator PU.1 that negatively regulates PU.1 mRNA translation [64]. PU.1 levels are critical for normal hematopoietic development and suppression of leukemia, and it was suggested that PU.1 antisense lncRNA contributes to keep them from being too high in expressing cells [64]. Mechanism and function of "antisense to PU.1" lncRNA resemble the ones of microRNAs in hematopoietic differentiation. However, even if many antisense transcripts have been identi�ed, this mechanism has not been described for the regulation of other genes. HOTAIRM1 was identi�ed as one of the lncRNAs induced during ATRA-mediated granulocytic differentiation of APL cell lines [65]. is lncRNA is one of the numerous lncRNAs produced from the HOXA cluster and its knockdown attenuated the expression of different HOXA genes, which are important transcriptional regulation in normal and malignant hematopoiesis. HOTAIRM1 also modulated the levels of CD11b and CD18 transcripts, two granulocytic differentiation genes. However, these effects did not produce evident defects in differentiation [65]. Conversely to other two well-characterized lncRNAs produced from the HOXA cluster, HOTAIR and HOTTIP, which regulates chromatin-modifying complexes [61], the molecular mechanism of HATAIRM1 action is still not known. Last to be identi�ed was "lincRNA erythroid prosurvival" (lincRNA-EPS). is lncRNA is one of the about 400 lncRNAs whose expression is modulated during erythropoiesis [66]. In particular, lincRNA-EPS promoted terminal differentiation and inhibited apoptosis of mature erythrocytes by inhibiting Picard expression, a pro-apoptotic gene, by a still not de�ned mechanism [66]. Like other studies on the identi�cation of lncRNAs in hematopoiesis, also this work raises many additional questions for future investigation. First of all, it will be important to determine how these lncRNAs regulate gene expression and if, similarly to microRNAs, they might be utilized as diagnostic and prognostic markers in AML. More recently, altered levels of another class of ncRNAs, the nucleolar snoRNAs, were found misregulated in AML cells [67]. In particular, increased levels of SNORD112-114 snoRNAs were found in primary AML blasts. e same study showed that ectopic expression of these snoRNAs promoted AML cell growth through Rb/p16 cell cycle regulation [67], indicating that, similarly to other ncRNAs, also snoRNAs may contribute to leukemogenesis.
Aberrant epigenetic regulation of miRNA genes as a consequence of molecular abnormalities has been described as one of the cause of miRNA misregulation. Fazi et al. observed a down-regulation of miR-223 in t(8; 21) AML samples [33] and showed that this was due to epigenetic silencing by the AML1-ETO oncoprotein (see above). e tumor suppressors let-7b and let-7c were also found downregulated in AML with t(8; 21) and inv16 [74]. Conversely, miR-126/126 * was speci�cally overexpressed in both t(8; 21) and inv (16) AMLs, the rearrangements resulting in the disruption of core binding factor (CBF) that, in turn, produces partial promoter demethylation of CpG island in which miR-126/126 * is embedded [75]. miR-196b, located between homeobox A9 (HOXA9) and HOXA10 genes, has been found to be speci�cally overexpressed in AML patients with MLL rearrangement [74][75][76][77][78]. Increased miR-196b expression depends on MLL fusion proteins [78], and its overexpression in bone marrow hematopoietic progenitor cells led to an increase in proliferation and survival capacity [78]. us, these data indicate that miR-196b deregulation might play an important role in the myeloid differentiation block that occurs in AML. Notably, inhibition of miR-196b activity by antisense LNA (locked nucleic acids) oligonucleotides in BM cells transformed with the MLL-AF9 fusion gene decreased their proliferative capacity [78]. In AML patients with the rare t(2; 11) translocation a speci�c upregulation of miR-125b was identi�ed [79]. Enforced expression of miR-125b in AML cell lines and CD34 + hematopoietic progenitor cells was able to inhibit myeloid differentiation and apoptosis, and conferred proliferation advantage to these cells [79][80][81]. In addition, increased expression of miR-125b in the bone marrow of mice was sufficient to induce a very aggressive and transplantable leukemia [81][82][83]. erein, these data indicated that high levels of miR-125b might contribute to leukemogenesis. Another relevant player in AML with MLL rearrangement is the miR-29 family [44,84]. In particular, members of this family were found downregulated in (11q23)/MLL and deleted in AML with loss of chromosome 7q, which encoded for the miR-29b-1 and miR-29a genes [44,84]. Importantly, ectopic expression of miR-29b in AML cell lines and primary AML blasts induced apoptosis and inhibited proliferation [84]. Furthermore, inoculation of miR-29b mimics into xenogra tumors decreased tumor growth [84]. Several targets were identi�ed for miR-29 function in AML, including MCL1, CDK6, IGFR, JAK2, and the DNA methyltransferases DNMT3a and DNMT3b [84]. Conversely, the mir-17-92 cluster was found to be particularly overexpressed in AMLs with MLL rearrangements [75]. Notably, this cluster was shown to inhibit monocytic differentiation of CD34 + hematopoietic progenitor cells through down-regulation of AML1 (see above).
Several miRNAs were found deregulated in APL, which is generally characterized by the t(15; 17) translocation [16]. e PML/RAR oncogenic fusion protein is directly responsible for the silencing of the tumor suppressors let-7c and miR-342. Notably, these two miRNAs were induced upon ATRA treatment of APL cell lines and primary t(15; 17) APL [38,40,44]; thus they might represent new markers in the therapeutic response in APL patients.
MiRNA expression has been also analysed in CN-AML patients. Speci�c miRNA signatures have been associated with mutations of the nucleophosmin (NPM1) and CEBPA genes [70][71][72][73]. AML with NPM1 mutations is characterized by high expression of the homeobox (HOX) genes and, notably, one of the down-regulated miRNAs identi�ed in this study (miR-204) controls the protein levels of the two HOX genes HOXA10 and MEIS-1 [70]. e same study identi�ed several upregulated miRNAs predicted to target CD34, a gene whose expression is frequently downmodulated in these leukemias. AMLs with CEBPA mutations are characterized by a more mature phenotype of malignant blasts and low expression of several homeobox genes [71]. Increased miR-181 levels characterized these leukemias [71], a miRNA �cienti�ca 5 T 1: MicroRNAs with a de�ne role in AML.

MicroRNAs in the Prognosis of Acute Myeloid Leukemia
e expression signature of some miRNAs has been found to be associated with clinical outcome and survival of patients with AML. Mutations in NPM1 and CEBPA genes are generally associated with favourable outcome while the FLT3-ITD has been linked to unfavourable outcome [103]. Increased expression levels of miR-181 were associated with favourable outcome in AML with both normal and abnormal karyotypes [75,104,105], and connected with CEBPA mutations [105]. In addition, it has been suggested that it may contribute to the partial erythroid differentiation reported in AML with CEBPA mutations [104,105]. Another miRNA signi�cantly associated with a prolonged overall survival is miR-212 [106]. However, the prognostic signi�cance of miR-212 did not correlate with speci�c AML subtype [106]. Conversely, miR-155 was found signi�cantly highly expressed in AML patients with FLT3-ITD and associated with poor prognosis [107,108]. Notably, sustained expression of miR-155 in mouse hematopoietic stem cells produced a myeloproliferative disorder, indicating that this miRNA may play a relevant role in leukemogenesis [109]. e C/EBP transcription factor, which plays an important function in myelopoiesis [17], has been identi�ed as a direct target of miR-155 [110]. However, the mechanism of miR-155 action in the myeloid lineages remains largely unknown. Different studies reported low expression of let-7b and miR-9 in patients with favourable cytogenetic translocations [76] while high expression of miR-191 and miR-199a adversely affected overall survival of newly diagnosed AML patients with predominantly intermediateand poor-risk cytogenetic [77].
Recently, high expression of miR-3151 was discovered an independent prognostic marker for poor outcome in CN-AML [111]. MiR-3151 is encoded in an intron of the coding gene BAALC, which also associate with poor outcome when highly expressed in CN-AML [112]. e two genes impact on different outcome endpoints: high miR-3151 expression associated with shorter disease-free and overall survival, while high BAALC expression predicted failure of complete remission and shorter overall survival [111]. us, the combination of both markers may be useful to identify patients with the poorest outcome.
In conclusion, miRNA expression will serve as a diagnostic and prognostic marker that adds valuable information beyond the cytogenetic.

Future Directions
Recent advances in the �eld demonstrated the feasibility of manipulating miRNA expression levels as a potential therapeutic strategy for cancer. erein, it is very likely that we will see the development of new therapeutic options based on miRNAs into the clinic in the next future. Beyond the understanding of the function of speci�c miRNAs that constitute therapeutic targets, a great effort will be put in the development of standard procedures for rapid and sensitive detection of miRNAs. Indeed, as miRNAs were found to be stably present in human serum [113] they might become novel noninvasive biomarkers for pathological conditions, including cancer. e study of other ncRNAs in normal and malignant hematopoiesis is still in its infancy, but new important progresses are expected in this �eld. us, it is clearly predictable that other ncRNAs, in addition to miRNAs, will become crucial new players in the diagnosis, prognosis, therapeutic responses, and even therapy of human leukemias.