microRNA expression in acute myeloid leukaemia: New targets for therapy?

Abstract Recent studies have shown that short non‐coding RNAs, known as microRNAs (miRNAs) and their dysregulation, are implicated in the pathogenesis of acute myeloid leukaemia (AML). This is due to their role in the control of gene expression in a variety of molecular pathways. Therapies involving miRNA suppression and replacement have been developed. The normalisation of expression and the subsequent impact on AML cells have been investigated for some miRNAs, demonstrating their potential to act as therapeutic targets. Focussing on miRs with therapeutic potential, we have reviewed those that have a significant impact on the aberrant biological processes associated with AML, and crucially, impact leukaemic stem cell survival. We describe six miRNAs in preclinical trials (miR‐21, miR‐29b, miR‐126, miR‐181a, miR‐223 and miR‐196b) and two miRNAs that are in clinical trials (miR‐29 and miR‐155). However none have been used to treat AML patients and greater efforts are needed to develop miRNA therapies that could benefit AML patients in the future.

Many of the biological therapies used to treat AML are inhibitory and block the aberrant gene product, cell cycle/proliferation and epigenetic regulators implicated in AML, such as the Fms-related receptor tyrosine kinase 3 (FLT3) inhibitor, Rydapt or the Bcl-2 inhibitor, venetoclax [5,6]. Recent advances in medicine herald a new era for genetic and epigenetic therapies and the opportunity to treat the genetic cause of disease rather than treating the product of a gene mutation [7]. While CRISPR/Cas9 gene editing solutions have shown success in minimising the formation of leukaemic stem cell (LSC) colonies, these therapies remain in the earliest stages of clinical application [8].
Recent studies have shown that microRNAs (miRNAs) and their dysregulation are implicated in the pathogenesis of AML due to their role in the control of gene expression that impacts on a variety of molecular pathways [9]. Therefore miRNAs represent a new therapeutic opportunity that can treat the cause of disease at the transcriptional level and unlike CRISPR/Cas9, miR therapy is unlikely to cause off-targets effects due to their specificity for a seed sequence [10].

miRNAs
miRNAs are short non-coding RNAs of approximately 20-25 nucleotides in length. Around 50% of miRNAs are intragenic, predominately transcribed from intronic sequences, while the remaining miRNAs are intergenic and subject to their own promoters, independent of host gene transcription [11]. miRNAs are pivotal in gene regulation through their gene silencing activity which causes mRNA decay, either through the formation of a complex or via interactions with specific proteins. Conversely miRNAs are implicated in translation activation and the upregulation of mRNA production [12,13].
While the canonical and non-canonical miRNA biogenesis pathways differ within the nucleus (Figure 1), once miRNAs have been transported to the cytoplasm by Exportin5/RanGTP, the mature miRNA complexes with Argonaut (AGO) family proteins to form the miRNAinduced silencing complex (miRISC) [11]. A 'seed' sequence at the 5' end of the miRNA binds to a complimentary sequence in the 3'UTR of the target mRNA and AGO interacts with GW182 family proteins. GW182 binds to the poly-A-binding protein and competitively inhibits its interaction with eukaryotic translation-initiation factor 4G (eIF4G), destabilising the closed loop structure of the mRNA and initiating deadenylation [14]. A miRISC/GW182 complex can remain in situ and form a silenced mRNA or the GW182 protein can recruit the decapping protein 1/2 to remove the mRNA 5' cap and proceed to complete mRNA degradation by the 5-3 exoribonuclease 1 (XRN1) [12,14].
miRNAs have numerous subcellular locations including the nucleus, the initiation site of miRNA biogenesis (Figure 1), but are also found in the cell cytoplasm of endosomes, exosomes, lysosomes, P-bodies, mitochondria, the Golgi network and the rough endoplasmic reticulum (RER) in addition to the circulatory system [11]. The RER membrane is rich in miRNA target messages and is a key site of de novo mRNA translation repression [15]. P-bodies form as a consequence of mRNA deadenylation and as such are rich in miRNA, GW182 proteins, decapping proteins and the CAF1-CCR4-NOT complex.
While miRNAs and their targets have numerous regulatory roles and functions under normal physiological conditions (Table 1), for a target to have therapeutic potential it needs to also have a functional effect in the disease. To this end some miRNAs have been shown to play a role in key features in leukaemia including stemness and therapy resistance.

miRNAs and stemness
Stemness of leukaemic cells is a major contributor to AML leukaemogenesis and LSCs play a significant role in disease relapse [16].

miRNAs and therapy resistance
Therapy resistance is a major hurdle in the pathogenesis of AML, with numerous mechanisms, including LSCs, attributed to chemoresistance [21,22]. miR-21 has been observed to work with other miRNAs in AML such as miR-15a to cause chemoresistance. Both miR-21 and miR-15a were found to downregulate programmed cell death 4 (PDCD4), BTG antiproliferation factor 2 (BTG2) and ADP ribosylation factor-like protein 2 (ARL2) and reduce chemotherapy-induced apoptosis [23].
miR-181b was found to inhibit high mobility group box 1 (HMGB1) and myeloid leukaemia (MCL-1) while HMGB1 was expressed at high levels in relapsed/refractory AML patients. Suppression of HMGB1 via RNA interference sensitized multidrug-resistant leukaemia cells to chemotherapy and induced apoptosis [24]. The downregulation of miR-451 was responsible for the increase in multidrug resistance protein 1 (MDR1) in FLT3-internal tandem duplication+ (ITD) AML, and as such contributed to the poor therapeutic response [25] seen in these patients.
F I G U R E 1 microRNA biogenesis. miRNA synthesis within the nucleus from intragenic genes is subject to host-gene promoter regulation.
Intergenic miRNA genes, those contained within non-coding regions, are subject to their own promoters and transcription is independent of host-genes. DGCR8 and Drosha recognition of primary-RNA, via DGCR8/N6-mth-GGAC motif recognition, processes the pri-miRNA and cleaves

miRNA targeting for cancer therapy
The clinical significance of miRNAs as targets for cancer therapy was highlighted by [26] who discovered that a small deletion in chromosome 13q14 in chronic lymphocytic leukaemia (CLL) resulted in a loss or downregulation of two miRNAs, miR-15a and miR-16-1, both of which target multiple known oncogenes [27]. Further investigation showed that miRNA genes commonly reside in genomic regions implicated in cancer and as such, are often deleted or over/underexpressed as a consequence of their close proximity to oncogenes.
miRNA therapies involving their suppression or replacement have been developed [28]. miRNA suppression therapies involve the binding of synthetic miRNAs to host miRNA or target mRNA in diseases where miRNA upregulation is implicated. Conversely, miRNA replacement therapy involves the integration of synthetic miRNA into target cells/tissues in disease where miRNA deletion or downregulation has occurred [29].

miR targets for therapy in AMLoverexpressed miRs
The potential of miRs to act as targets for therapy are based on their subversion. For example miR-21 is a prominent miRNA in a range of malignancies including oral squamous carcinoma and glioma, and in AML it provides an example of the complexity of targeting mRNA regulation [30][31][32] (Figure 2A). miR-21 is overexpressed in all French-American-British (FAB) subtypes of AML, with greater prevalence in M3, M4, M5 and NPM1-mutated AML [33,34] (Table 2) [37] discovered miR-21 negatively impacts T-cell fragility and potentially disrupts innate tumour defence mechanisms. miR-21 was found to be underexpressed in cytogenetically normal AML with high Tet methylecytosine deoxygenase 1 (TET1) and while some studies have found miR-21 to target TET1 in colorectal cancer, it was TET1 overexpression, and not miR-21 underexpression, that was suggested as the contributor to AML progression in this instance [38].
The oncogenic miR-155-5p plays a role in haematopoiesis and cell differentiation, while overexpression has been shown to be associated with lower survival rates in CLL patients (n = 88; p = 0.001) [39]. Upregulation of miR-155 inhibits SH2 domain-containing inositol 5'-phosphate 1 protein (SHIP1)-phosphoinositide 3-kinase (PI3K)alpha serine/threonine kinase (AKT) pathway and significantly reduces apoptosis in CLL and AML cells [40,41]. miR-210 also targets SHIP-1, which along with miR-155 has been associated with the loss of SHIP-1 in high-risk myelodysplastic syndrome patients [42]. Elevated expression of miR-210 was observed in AML patients where it correlated with poor outcomes (overall and event free survival) and was significantly lower in patients achieving CR. Hypoxia inducible factor-1α (HIF-1α) enhances miR-210 which in turn can stabilise HIF-1α. This positive feedback loop could be a factor in miR-210 overexpression given the hypoxic bone marrow microenvironment in AML [43].
The miR-125a cluster has a single promoter and is transcribed as one, whereas miR-125b contains an additional promoter which enables the transcription of miR-125b on its own [44]. An investigation of AML with t(2;11) in a murine model showed up to a 90-fold increase in expression of miR-125b, with miR-125b-transplanted mice developing myeloproliferative disorders that progressed to AML. It was noted that differing levels of miR-125b correlated with different leukaemic phenotypes, suggesting that the level of overexpression not only contributes to the disease state but also to the disease phenotype [45,46]. miR-125b2 is located on chromosome 21 and overexpressed in Down's syndrome (DS) patients with acute megakaryocytic leukaemia.

GATA binding protein 1 (GATA1) mutations that result in truncated
GATA1s are key drivers of leukaemogenesis in DS associated-AML and Klusmann et al. [47] determined that GATA1s-induced proliferation, self-renewal and megakaryocyte colony forming unit (CFU) numbers were more aggressive in the presence of miR-125b2. Interestingly, the study discovered DICER1, integral to miRNA biogenesis and the hairpin structure, and leaves a 3' overhang. The pre-miRNA is transported to the cytoplasm by Exportin 5/RanGTP. The endonuclease Dicer cleaves the pre-miRNA terminal loop to produce the mature miRNA duplex. The two strands of the mature duplex are 3p and 5p and only one strand is designated the guide strand and loaded into the AGO family protein to form the miRISC. The ratio of 5p and 3p loaded into AGO varies.  miRISC assembly, was a direct target of miR-125b2, suggesting that miR-125b2 may harbour the potential to disrupt/dysregulate overall miRNA synthesis.
During our own analysis of differential gene expression between risk subgroups in AML, miR-486 was found to be differentially expressed between intermediate versus good, and poor versus good risk subgroups of adult AML patients [48] and elevated in expression in high risk AML patients [49]. As detailed by Shaham et al. [50], miR-486 was also overexpressed in DS-AML and had the same synergy with GATA1s as miR-125b2, although with less acute effects on

Dysregulated miRNAs in AML -the 'double edged sword' of subtype expression
The varied expression patterns of miRNAs including the miR-9 and miR-181 families ( Figure 2C) demonstrate the delicate balance in the regulatory roles of miRNAs, targeting both tumour suppressors and oncogenes, and often in a disease-specific manner. This results in a regulatory 'double-edged sword' whereby aberrant expression (increased or decreased) contributes to different leukaemic subtypes. The miR-9 and miR-181 molecules highlight the dichotomous role of miRNAs in haematopoiesis and their varied expression across AML subtypes. In MLL-AML, miR-9 is overexpressed as a result of direct targeting by MLL-fusion genes. Induction of a more aggressive phenotype is attributed to miR-9 targeting of tumour suppressors, with the inhibition of miR-9 significantly increasing apoptosis in MLL-cells [63]. In contrast, miR-9 was downregulated in ectopic viral integration site 1 (EVI1) high AML compared to normal and EVI1 low cells, and forced expression of miR-9 by 5azacytidine correlated with reduced colony formation and increased apoptosis [64]. In paediatric AML with t(8;21), miR-9 was suggested to act as a tumour suppressor, its underexpression facilitating leukaemogenesis and differentiation arrest via its targets, Lin-28 homolog B (LIN28B) and high mobility group AT-hook 2 (HMGA2).
Reintroduction of miR-9 was able to reverse the leukaemic effects via growth reduction and differentiation initiation without affecting apoptosis [65]. miR-21 was found to be underexpressed in cytogenetically normal AML with high Tet methylecytosine deoxygenase 1 (TET1), while some studies have found miR-21 to target TET1 in colorectal cancer. It was TET1 overexpression and not miR-21 underexpression that was suggested as the contributor to AML progression in this study [38].
The miR-181 family are overexpressed in AML M1, M2 and M3 where downregulation of protein kinase C delta (PRKCD)-p38-CCAAT enhancer binding protein alpha (C/EBPα) contributed to proliferation and myeloid differentiation arrest. MiR-181a also directly targeted calcium/calmodulin-dependant protein kinase kinase 1 (CAMKK1) and tumour suppressor CTD small phosphatase-like protein (CTD-SPL), both of which are implicated in granulocyte and macrophagelike differentiation [66]. Conversely, downregulation of the miR-181 family members were identified in a specific subset of cytogenetically normal AML patients and moreover, the upregulation of miR-181 family was associated with better outcomes in C/EBPα -mutated AML, where the mutated C/EBPα directly targets miR-181 [67]. Reintroduction of miR-181b into MDR AML cells treated with doxorubicin or cytarabine significantly reduced cell growth and increased apoptosis.

1.8
Therapeutic potential of miRs in AML As highlighted above, miRNAs have complex and differing expression patterns, and a variable impact on leukaemogenesis and the prog-nosis of AML ( Table 2). The normalisation of expression to housekeeping genes such as 18S ribosomal subunit or RNU44 snoRNA, and the subsequent impact of miRNA expression in AML has been investigated using qPCR, elucidating their potential as therapeutic targets (Table 3). Of the five pre-clinical studies (detailed in Table 3

Clinical trials
Of the miRNAs reviewed here, only two are currently in clinical trials. Cobomarsen (MRG-106) is a miR-155 inhibitor developed by Viridian therapeutics and has demonstrated efficacy in the treatment of cutaneous T-cell lymphoma [68]. Viridian Therapeutics also developed Remlarsen and MRG-229, both of which are miR-29 mimics that show potential in the treatment of tissue injury/fibrotic disease and idiopathic pulmonary fibrosis respectively [69,70].
With limited data on miRNA therapy in AML, the potential role of miR-therapies in the treatment landscape of AML remains unclear.

DISCUSSION
The delivery system employed in miRNA therapies is equally as important as the miRNA itself. Particularly in ensuring efficient delivery of the miRNA therapy to its target system and minimising toxicity. Meyer [74] commented on the potential issue of toxicity in reference to  [77]. Of the six miR-therapies detailed here, five had a significant impact on LSCs, CFUs and self-renewal without altering normal HSCs (Table 2).
Another potential placement of miR-therapy is as a sensitizing agent for relapsed/refractory AML. A number of miRNAs are implicated in chemoresistance, either directly or via their targets.
A recent review by Fajardo-Orduña et al. [78] commented on the issue of chemoresistance with >60% of AML patients developing relapsed/refractory disease. Their recommendations for combating this involve multi-drug treatment, a response with potential increased toxicity and unknown efficacy in breaking resistance. While in contrast, miR-181a has shown efficacy in sensitising AML cells to chemotherapy treatment with no reported additional toxicity [24]. Of the miRNA-therapies discussed (Table 2)

SUMMARY
miRNAs are significantly dysregulated in AML and contribute to both disease progression and maintenance as well as treatment efficacy.
Several of the miRNAs reviewed here have been shown to inhibit the key characteristics of AML in vitro or in vivo in model systems.
However, to date, none of the miRNA therapies described have been developed for the treatment of AML. Greater efforts to develop miRNA therapies specifically for AML patients in the future may provide new treatment options and improve survival rates.