Introduction
MicroRNAs (miRNAs) are 22-24 nucleotide non-coding RNAs that participate in the regulation of mRNA expression in eukaryotes[
1‐
3], and play critical roles in a wide range of biological processes including cell-cycle control[
4,
5], immune response[
6‐
8], and differentiation[
9‐
11]. One of the best-characterized differentiation processes is granulopoiesis, in which hemopoietic myeloid progenitor cells develop sequentially from myeloblasts into morphologically distinct promyelocytes, myelocytes and mature granulocytes. This process is tightly controlled by changes in the expression of hundreds of transcription factors[
12,
13], which are in turn regulated by a few highly expressed miRNAs, including miR-223 and miR-146a, both of which have been shown to promote granulopoiesis[
14‐
16]. Total loss of miR-223 does not completely block granulopoiesis[
14], suggesting that other miRNAs may also act in concert to maintain this process. Few studies, however, have sought to completely profile differentially expressed miRNAs during granulopoiesis in primary cells. Prior work has either been restricted to differentiated cell lines[
17,
18], in which accurate modelling of the specific stages of granulopoiesis is not entirely possible, or human neutrophil maturation[
19,
20].
In the most well-accepted models, miRNAs bind loosely to complementary sequences in the 3′UTR of their target mRNAs in the cytoplasm, and function by inhibiting translation, inducing mRNA cleavage or mRNA degradation following decapping and deadenylation[
21]. It is important to correlate the expression of miRNAs and their target mRNAs during granulopoiesis in order to obtain insights into the role of these molecules in the development of granulocytes. While previous studies have investigated this relationship in human cells[
22], the miRNA-mRNA interaction network in mouse granulopoiesis has been largely uncharacterized.
Several studies have reported localization of miRNAs in the nucleus[
23‐
31], suggesting these molecules may have other biological functions or mechanisms of action apart from their canonical role. For example, miRNAs have been shown to target gene promoters, potentially inducing overexpression (miR-373) or downregulation (miR-320) of target genes[
32,
33]. More recently, mouse-specific miR-709 was found to be enriched in the nucleus to target pri-miR-15a and pri-miR-16, thus regulating the expression of mature miR-15a and miR-16[
31]. With one exception[
31], studies on nuclear-enriched miRNAs have been performed in cell lines. It remains unclear whether nuclear enrichment of some miRNAs is a feature of transformed cells, and whether differential expression of nuclear miRNAs is important in regulating gene expression during cellular differentiation or transformation.
In this study, we analyzed miRNA expression in primary murine myeloid cells at four successive stages of hemopoietic differentiation; Lin- Sca1+ cKit+ stem/progenitor cells (LSK), promyelocytes, myelocytes and granulocytes. We not only performed analyses of miRNA expression levels in whole cells, allowing direct interrogation of miRNA-mRNA expression relationships, but also analyzed purified nuclear and cytoplasmic cell fractions to profile miRNA subcellular localization. We found four nuclear-enriched miRNAs in primary cells and further assessed their subcellular distribution in a range of mouse hemopoietic cell lines.
Discussion
Previous studies have shown the involvement of several miRNAs in granulopoiesis including miR-223[
14], miR-34a[
45], and miR-146a[
16]. However, a systematic expression profiling of miRNAs and their mRNA targets across successive populations of myeloid cells at defined stages of granulopoiesis is lacking. Comprehensive miRNA expression profiling of human granulocytic differentiation was recently published, and showed that the potential roles of many miRNAs in granulopoiesis, either individually or as a group, are currently ignored[
19,
20].
Our study sought to provide the first comprehensive characterization of miRNA expression across carefully purified cells at progressive stages of murine granulopoiesis. In addition, we correlated miRNA expression with their predicted and confirmed mRNA targets to determine the possible role of differentially expressed miRNAs during granulopoiesis.
Amongst miRNAs that showed differential expression between mouse promyelocytes and granulocytes, we noted many that have been shown to be associated with hematological malignancies (Figure
2A). Several miRNAs that were upregulated during granulopoiesis (miR-15a, miR-16 and miR-29) have previously been shown to be downregulated in acute myeloid leukemia[
46,
47]. miRNAs that were downregulated during granulopoiesis, including miR-17, miR-19a and miR-20a, were amongst those previously shown to be upregulated in leukemia[
48]. These data suggest the control of miRNA expression is crucial in ensuring proper cell differentiation, and failure of this control may lead to, or contribute to, cancer.
Many miRNAs that we identified as being differentially expressed in granulopoiesis were not previously implicated in this process including miR-19a, miR-19b miR-24, miR-26a, miR-26b, miR-93, miR-106b, miR-191, miR-139-5p, miR-140 and miR-195 (Figures
2A and
3, Additional file
1). Notably, several miRNAs shared common predicted targets with miRNAs that are known to be important for granulopoiesis (Additional file
3). For example, miR-106b and miR-194 were predicted to target the transcription factor gene,
Mef2c, which is a confirmed target of miR-223[
14,
49]. Repression of
Mef2c is important for early granulopoiesis and may also be involved in the regulation of granulocyte activation[
14]. Expression of miR-106b and miR-194 may act in concert with that of miR-223, which may explain a previous observation whereby granulopoiesis was not completely impaired in miR-223 null mice[
14].
A recent study reported a plethora of differentially expressed miRNAs during human granulopoiesis, many of which were predicted
in silico to target key transcription factors such as
Runx1 and
Pu.1[
20]. Our stringent analysis that considered anti-correlation between miRNA and mRNA target expression did not reveal any miRNAs predicted to regulate key transcription factors such as
Cebpα,
Pu. 1 or
Runx1 during granulopoiesis, suggesting that essential transcription factors are not likely to be directly regulated by miRNAs during this process (data not shown). However, the transcription factor
Myb was downregulated with concurrent overexpression of miR-15b, miR-16, miR-150 and miR-195 (Figure
3 and Additional file
3).
Myb is involved in hemopoietic progenitor proliferation[
50], and may need to be downregulated to allow terminal differentiation of granulocytes. We also observed downregulation of transcription factors such as
Mga and
Tcf4, and myeloid leukemia related genes including
Pml,
Abl1, and
Bcl11a in terminally differentiated granulocytes in conjunction with the expression of a group of granulocyte-enriched miRNAs (Figure
3). Future studies are required to determine whether downregulation of these genes is important for normal granulopoiesis and whether they are regulated by miRNAs.
Several members of the polycistronic miR-17-92 cluster and the homologous miR-106a-92 cluster (miR-17, miR-19a, miR-20a, miR-92a and miR-106a) were expressed at the highest levels in promyelocytes (Figure
3B). These miRNAs are known to have oncogenic potential and are overexpressed in hematological malignancies[
48,
51,
52]. Overexpression of these miRNAs is also known to confer a stem cell-like phenotype[
51]. While our data was consistent with others in that these miRNAs were highly expressed in early developmental lineages[
51], it was surprising that their expression levels were lower in LSK compared to promyelocytes. These data are the first to indicate that the miR-17-92 polycistron is predominantly over-expressed in promyelocytes during granulopoiesis. In relation to leukaemia, overexpression of miR-17-92 may reflect accumulation of blast cells and promyelocytes in myeloid malignancies such as acute and chronic myeloid leukemia[
53]. Furthermore, we and others have previously shown that miR-17-92 expression is downregulated following imatinib treatment in chronic myeloid leukaemia patients[
54,
55]; this observation may reflect the normalization of blood cell proportions following hematological response after treatment[
56]. In line with previous reports[
57,
58], we detected downregulation of confirmed miR-17-92 targets including
Pten and
E2f2 in promyelocytes compared to granulocytes (Additional file
3), although the significance of this observation in the context of granulopoiesis remains unclear.
Consistent with others[
59‐
61], we found the highest expression of miR-125a, mir-125b, let7d and let7e in LSK and promyelocytes compared to differentiated cells (Additional file
1). In addition, our present data indicate the downregulation of miR-17-92 and let-7 families in LSK compared to promyelocytes, in conjunction with the overexpression of their reciprocal targets including hepatic leukemia factor (
Hlf), V-myc myelocytomatosis viral related oncogene (
Mycn) and krüppel-like factor 12 (
Klf12) (Additional file
4).
Hlf is a transcription factor that can confer anti-apoptotic effects and prevent premature death of hemopoietic stem cells[
62].
Mycn is an oncogene which is important for the proliferation and homeostasis of stem cells[
63]. The role of the transcription factor,
Klf12 in stem cell development is as yet unknown. However, other members of krüppel-like factor family such as
Klf 4 and
Klf 5 are key players in embryonic stem cell renewal and somatic cell reprogramming[
64,
65]. Whether
Klf12 has a role similar to
Klf 4 and
Klf 5 remains to be elucidated.
We have also carefully purified the nuclear and cytoplasmic RNA from four populations of murine primary hemopoietic cells to study the localization of miRNAs in these cells. This step is important given the recent reports of nuclear-enriched miRNAs that could possibly be involved in non-canonical functions[
32,
33]. These miRNAs are unlikely to regulate gene expression by targeting the 3′ UTR of target mRNAs but may target the promoter of genes[
32,
33], or other pri-miRNAs[
31]. Alternatively, these miRNAs may be retained in the nucleus to prevent them from downregulating a potential target mRNA in the cytoplasm[
23].
It is important to consider the fact that perfect nuclear/cytoplasmic fractionation is technically difficult and thus rigid analysis is required to obtain meaningful data[
30]. We only considered the highly expressed miRNAs in each sample (expression was detected below 30 cycles of RT-qPCR amplification in at least one cell type) to minimize possible false positive results. Furthermore, by using cell equivalent volumes, and comparing to the mean expression of cytoplasmic enriched miRNAs, we further reduced false positive results. Our discovery of four nuclear-enriched miRNAs is similar to a recent study in primary mouse liver, where three nuclear-enriched miRNAs were found[
31]. These data suggest that the majority of miRNAs are acting in a canonical manner by targeting the 3′UTR of granulopoiesis-regulating genes. It is also noteworthy that the majority (73%) of the differentially expressed miRNAs are conserved between mouse and human (Figure
2B and Additional file
2), and this could mean they are likely to have common canonical mRNA targets[
66].
All four nuclear-enriched miRNAs, miR-709, miR-706, miR-690 and miR-467a* in hemopoietic cells are mouse specific miRNAs. It is interesting to note the enrichment of miR-706 and miR-467a* in the nucleus of primary mouse hemopoietic cells and cell lines because these miRNAs have not been reported as being nuclear-enriched in other cell types[
23‐
31]. Notably, the primary transcript of miR-142-3p, a miRNA recently found to be important in promoting granulopoieisis, was amongst the putative targets of miR-706 (Figure
6B). In contrast to the previous finding in mouse liver, we did not observe an anti-correlation between the expression of nuclear miR-709 and cytoplasmic miR-15a/16 in myeloid cells. We discovered two other predicted pri-miRNA targets of miR-709 based on our
in silico analysis using RNAhybrid and anti-correlation of miRNA expression. This result suggests that interaction between this nuclear-enriched miRNA and its pri-miRNA targets may be cell type specific. Inhibition of miR-706 in MEL and MPRO cells, however, did not result in significant accumulation of predicted miRNA targets, indicating either these miRNAs are not the actual targets or nuclear miR-706 may not function by targeting pri-miRNA transcripts. Alternatively, since the miRNA inhibitor was expressed in the cytoplasm, it is possible that there was little to no reduction in nuclear miR-706 expression levels in these cells. Whether miR-706 and other nuclear-enriched miRNAs target pri-miRNAs during hemopoiesis remains to be determined.
miRNAs may also be retained in the nucleus to fine-tune the expression of their mRNA target(s). Apart from the validated role of miR-709 in regulating the expression of
Myc, knockdown of miR-690 has recently been shown to increase the expression of a key myeloid transcription factor
Cebpα[
67]. Consistent with others[
44], we have shown that miR-706 regulates the expression of
Stat1; a transcription factor involved in myeloid differentiation. The retention of miR-709, miR-690 and miR-706 in the nucleus may be important to control the expression of transcription factors or other proteins during granulopoiesis.
Competing interests
The authors declared no competing interests.
Authors' contributions
JJ-LW, KAL, MG, AC, and JH performed experiments. WR and DG performed bioinformatic analyses. JJ-LW and JH designed the study with input from JEJR and RJT. JJ-LW, RJT, JEJR and JH wrote or edited the manuscript. All authors read and approved the final manuscript.