Background
Acute myeloid leukaemia (AML) represents the convergent outcome a number of genetic abnormalities that have consequence in crucial cellular pathways of haematopoiesis. With an increasing knowledge of cellular processes and an accompanying improvement in our ability to interrogate these pathways, a spectrum of recurrent genetic abnormalities relevant to AML has become increasingly apparent. While most AML cases have at least one detectable genetic mutation potentially responsible for their pathogenesis, there remains a significant minority in which no abnormality is detectable [
1].
MicroRNA-mediated post-transcriptional control of gene expression is a relatively newly discovered mechanism of cellular regulation that could account for some of the gaps in our knowledge of AML pathogenesis [
2]. Through their repressive action on complementary sites in 3' untranslated regions (3'UTR) of target genes [
2], these short 19-25 nucleotide RNA species are important in numerous processes including haematopoietic stem cell maintenance [
3] and progenitor self-renewal [
4], myeloid differentiation [
5‐
7], cell cycle and proliferation [
8,
9], apoptosis [
10,
11] and gene methylation [
12]. All of these pathways are of potential relevance to AML pathogenesis if dysregulated.
NPM1-mutated
AML (NPM1
mut
-AML) accounts for approximately 30% of cases of adult AML (and up to 60% of AML with normal karyotype) [
13]. Recently
NPM1
mut
activation has been found to initiate a myeloproliferative disorder after knock-in into mouse haematopoietic stem cells, however co-expression with a secondary mutation was proposed to be needed for overt AML development [
14]. Global microRNA expression was assayed in a cohort of normal karyotype AML (NK-AML) using a stringent LNA-based microarray platform. We showed that among other differentially expressed microRNAs miR-10a, but not miR-10b, was significantly markedly over-expressed in
NPM1
mut
-AML versus AML not bearing
NPM1 gene insertions (
NPM1
WT
-AML). It is also demonstrated that the only available
NPM1 mutated cell line, OCI-AML3, exhibits high miR-10a expression. We have demonstrated that knockdown of over-expressed miR-10a in these cells resulted in reduced cellular survival and clonogenic growth. Using luciferase reporter analysis, we confirmed several miR-10a suppressible target genes located in key cellular pathways of significance to AML. Together, these findings suggest miR-10a may provide a pro-survival signal contributing to the pathogenesis of
NPM1mut-AML.
Discussion
miR-10 family members are now implicated in the malignant transformation across a range of tissues through altered expression including breast cancer [
18‐
21], colon and oesophageal cancer [
22,
23], glioblastoma [
24‐
26], hepatocellular carcinoma [
27], melanoma [
18,
28], neurofibromatosis [
29], pancreatic cancer [
30] and urothelial carcinoma [
31]. In murine models of breast carcinoma miR-10b has been demonstrated to have a crucial role in metastasis [
19] with the pro-metastatic effect repressible by systemic antisense inhibition of miR-10b [
21]. A similar role has also been demonstrated in oesophageal cell lines [
22]. miR-10a may have a role in initial oncogenic transformation events evidenced by the finding that miR-10a enhances the transformation of NIH-3T3 cells by RAS-V12 [
32]. Contrary to these findings miR-10a is actually down-regulated in chronic myeloid leukaemia (CML) CD34+ cells and over-expression of miR-10a retards the growth of KU812 cells (CML cell line) [
33].
miR-10a first appeared in reference to AML when it was individually selected along with miR-10b for expression analysis in a cohort of AML cases [
34], by virtue of its residence within HOX clusters known to be overexpressed in AML [
35]. In a subsequent study, miR-10a and miR-10b were found to be overexpressed (4.7 and 3.1 fold respectively) in NK-AML versus non NK-AML [
36,
37]. Thus, miR-10 family expression appeared to be reflective of the HOX over-expression characteristic of NK-AML. The HOX expression profile itself appears more specific to
NPM1
mut
-AML which has the gene expression reminiscent of the haematopoietic stem cell [
15].
Garzon et al., (2008) also compared microRNA expression between
NPM1
mut
-AML versus
NPM1
WT
-AML. They found that a microRNA signature, dominated by miR-10a, miR-10b and miR-100 over-expression in
NPM1
mut
-AML, was able to accurately discriminate these AML subtypes. miR-10 family over-expression was also noted in two subsequent studies undertaking a similar comparison [
38,
39], but each utilising the alternative platform of a multiplexed TaqMan
® MicroRNA Assay [
40].
In each of the earlier publications miR-10a and miR-10b were reported be co-overexpressed with a similar FC. However, it may be difficult to confidently discriminate miR-10a from miR-10b expression, since they differ at only a single nucleotide in the centre of the mature microRNA outside the seed site. While miR-10a and miR-10b share the same active 7 nucleotide seed site and therefore a vastly overlapping target profile, each is transcribed from distinct genomic locations and are likely to be controlled by their own cis-regulatory networks. The current study utilised a stringent LNA-based array which has a capacity to resolve miR-10a from miR-10b expression with cross-hybridisation of less than 10% for each probe. We found that while miR-10a was overexpressed in
NPM1
mut
-AML to a degree concordant with the earlier studies (13.4 fold), miR-10b was much more modestly DE (1.4 fold). Further to this, the expression ratio of miR-10a to miR-10b in the
NPM1-mutated samples was 54:1. This suggests that miR-10a is the dominantly expressed family member in
NPM1- AML cases, which verifies Garzon et al., finding [
41] and is confirmed by a recent study by Ovchrenko et al. [
42].
The normal physiological role of miR-10a is uncertain. miR-10a is deeply conserved both with respect to its sequence as well as its location within the HOXB cluster [
43,
44]. The finding that several HOX genes including HOXA1 [
45], HOXA3 [
46], HOXD4 [
47] and HOXD10 [
19] are repressible targets of miR-10 suggests that miR-10 and HOX genes form an interconnected regulatory network with an important role in embryonic development. To date, minimal evidence exists to indicate that miR-10 has a physiological role in normal haematopoiesis, substantiated only by the observation that miR-10a is down-regulated as haematopoietic precursors mature towards certain differentiated progeny, such as megakaryocytes [
48] or lymphocytes [
49].
To further evaluate the potential role of miR-10a over-expression in
NPM1
mut
-AML, functional studies are required to address the possibility the over-expression may represent a passenger phenomenon consequent on a permissive chromatin configuration encompassing the HOXB locus. This is particularly important, since
NPM1
mut
-AML is known to possess a HOX up-regulated expression signature, although it is noted that miR-10a is not overexpressed in
MLL- deregulated AMLs [
38] which similarly upregulate HOX genes [
50,
51].
We conducted functional studies in the
NPM1
mut
OCI-AML3 cell line. We have demonstrated in this study that knockdown of miR-10a overexpression in these cells lead to a decreased cell count after 48 hours in culture. Increased Annexin-V positive events but no change in cell cycling or proliferation was noted at 24 and 48 h. However, Caspase-3 was not activated, which suggests that the reduced cell survival occurs by a mechanism other than classic apoptosis, such as necrosis, or via activation of Apoptosis-Inducible Factor leading to Caspase 3- independent apoptosis [
52]. Interestingly, assessment of the clonogenic potential of OCI-AML3 cells by growth on semi-solid media found a decrease in the clonogenic potential of cells with miR-10a knockdown, suggesting a possible role for miR-10a in self renewal pathways.
We acknowledge that this study is constrained by the current lack of NPM1
mut
-AML cell lines, and future studies should determine the functional role of miR-10a by manipulating its expression levels in primary cells from AML patients, both with NPM1
wt
and NPM1
mut
, which has its own technical challenges. The interaction between miR-10a and NPM1
mut
needs to be defined. Evaluation of the leukaemogenic qualities of miR-10a, and its connection with the NPM
mut
, in murine models might address the unanswered issues raised by this study.
Prior to this study, only a short list of miR-10a repressible targets had been experimentally confirmed including
Pbp1, RAN [
32],
USF2 [
33] and the previously mentioned HOX genes. We wished to identify miR-10a gene targets which may explain the in vitro functional findings induced by suppression of miR-10a. Particular attention was paid to those genes that had previously been shown to contribute to AML (or other malignancies) by their down regulation.
RB1CC1 is a tumour suppressor and transcriptional promoter of key cell cycle regulator retinoblastoma-1 [
53] that has been found to be inactivated in breast cancer by the intriguing means of truncational mutation [
54]. Similarly,
TFAP2C transcriptionally activates p21 expression, retards breast cancer cell growth, and decreases clonogenic survival [
55].
ID4 is a putative tumour suppressor gene that is down-regulated by hypermethylation in numerous cancers including ALL [
56] and MDS [
57].
KLF4 is not only a critical regulator of monocytic differentiation [
58], but has been shown to be pathologically inactivated in medulloblastoma [
59], and recently to interact with miR-10b in transition to oesophageal cancer [
22].
NR4A3 has previously been demonstrated to be a critical tumor suppressor of myeloid leukemogenesis [
60] and possesses two 3'UTR seed matches to miR-10a, which may increases capacity for translational repression [
61].
ARNT was examined, since down-regulation of the
AHR/
ARNT pathway by epigenetic means in acute lymphoblastic leukaemia (ALL) [
62] may be extrapolatable to myeloid disease.
GTF2H1 is a general transcription factor involved in nucleotide excision repair, contributing to the maintenance of genomic stability [
63].
MAPRE1 is involved in microtubular physiology and maintenance of chromosomal stability [
64], processes often disturbed in AML. We have demonstrated that these genes were shown to be the suppressible targets of miR-10a by report assays and warrant further investigation. The mechanism of miR-10a over-expression is unknown. It could be consequent upon copy number amplification at the 17q21.32 locus where miR-10a resides, although this was not seen by high density SNP array (at 1 megabase resolution) in a cohort of AML patients which included 14 cases of
NPM1-AML [
65]. miR-10a expression is influenced by the methylation status of an upstream promoter [
46], therefore analysis of the methylation status and the broader epigenetic state of the miR-10a locus in
NPM1
mut
-AML would be helpful.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AB carried out qRT-PCR, apoptosis, cell cycle and cell growth assays, microRNA target prediction and validation by luciferace assay, and wrote the manuscript. CAP carried out the clonogenic, BrdU, differentiation assays and contributed to the apoptosis assays and wrote the manuscript. VY performed the bioinformatic microarray analysis. YWY contributed to statistical analysis. ML carried out the microarrays, participated in the design of the study and wrote the manuscript. DM conceived of the study, participated in its design, supervised the experiments and helped to write the manuscript. All authors read and approved the final manuscript.