Myelopoiesis involves differentiation of hematopoietic stem cells to cellular populations that are restricted in their self-renewal capacity, beginning with the common myeloid progenitor (CMP) and leading to mature cells including monocytes and granulocytes. This complex process is regulated by various extracellular and intracellular signals including microRNAs (miRNAs). We characterised the miRNA profile of human CD34+CD38+ myeloid progenitor cells, and mature monocytes and granulocytes isolated from cord blood using TaqMan Low Density Arrays. We identified 19 miRNAs that increased in both cell types relative to the CMP and 27 that decreased. miR-125b and miR-10a were decreased by 10-fold and 100-fold respectively in the mature cells. Using in vitro granulopoietic differentiation of human CD34+ cells we show that decreases in both miR-125b and miR-10a correlate with a loss of CD34 expression and gain of CD11b and CD15 expression. Candidate target mRNAs were identified by co-incident predictions between the miRanda algorithm and genes with increased expression during differentiation. Using luciferase assays we confirmed MCL1 and FUT4 as targets of miR-125b and the transcription factor KLF4 as a target of miR-10a. Together, our data identify miRNAs with differential expression during myeloid development and reveal some relevant miRNA-target pairs that may contribute to physiological differentiation.
Myelopoiesis results in the generation of mature myeloid cells: the monocytes, granulocytes, dendritic cells, megakaryocytes and erythrocytes. Differentiation follows a highly regulated series of phases from a hematopoietic stem cell (HSC) into increasingly self-renewal- and lineage-restricted progenitor cells. Myelopoiesis begins from the common myeloid progenitor (CMP) that gives rise to the mature myeloid cells, but not to the T, B and NK cells of the lymphoid lineage.
The classical, mouse-based model of HSC differentiation, devised by the work of Akashi, Kondo and Weissman1,2 is relevant to human physiology as confirmed using cells from the bone marrow and umbilical cord blood and a specific set of CD markers3,4,5. In this model, long-term repopulating HSCs give rise to short-term HSCs, which in turn lead to multipotent progenitors and then the common lymphoid and myeloid progenitors. Human cells with the phenotype CD34+CD38+ have no self-renewing or lymphoid potential and are believed to constitute the CMP as well as the further differentiated granulocyte/monocyte progenitor (GMP), the megakaryocyte/erythroid progenitor (MEP) and a potential monocyte/dendritic cell progenitor (MDP). These subsets can be further differentiated based on expression of CD45RA and CD1354. Together these represent ~1% of the mononuclear cell population of the bone marrow4,6. While this hierarchy is generally widely accepted for human myelopoiesis7,8, further studies have refined the model by demonstrating a potential for macrophage and dendritic cell development from a lymphoid progenitor9 and by defining in exquisite detail several sub-types of dendritic cells and monocytes10.
Several extrinsic and intrinsic factors work in concert to determine the fate of stem and progenitor cells as they differentiate into mature cells. The primary extrinsic determinants of myeloid differentiation are the cytokines granulocyte/macrophage colony stimulating factor (GM-CSF), granulocyte colony stimulating factor (GCSF), macrophage colony stimulating factor (MCSF), erythropoietin (EPO), and thrombopoietin (TPO)11. The most influential intrinsic factors are believed to be a group of transcription factors, beginning with SCL/TAL-1 and RUNX1 in the embryo, followed by PU.1, C/EBPα, IRF8 and GFI112. In addition to these, cell fate decisions may also be determined by other means including by post-transcriptional regulators exemplified by microRNAs (miRNAs).
MiRNAs are small, non-coding RNA molecules (19–22 nucleotides) that are genomically encoded and are involved in most aspects of cellular development and maintenance13. Initially transcribed as a long primary RNA, the transcript undergoes sequential cleavage by RNAse III family enzymes Drosha and Dicer to produce the mature miRNA which is incorporated into a protein complex named the RNA-induced silencing complex (RISC). The miRNA then guides the RISC to the 3′ untranslated regions (3′ UTRs) of messenger RNA (mRNA) molecules to which it binds while the RISC inhibits translation of the mRNA (reviewed in ref.14). Due to imperfect sequence complementarity each miRNA is predicted to bind hundreds to thousands of target mRNAs however, the degree of translational repression observed for each binding site is usually small15. Thus miRNAs are believed to impact specific phenotypes by broadly suppressing multiple targets within specific pathways rather than by large effects on a small number of targets16.
Determining the role of miRNAs in myelopoiesis has been approached from different angles previously, with the focus generally being on unilineage development or activation17,18,19 or on the transition from HSCs to committed progenitors20,21,22. Several of the reported studies have used cell lines such as NB4 or THP-1 differentiated in vitro using agents such as retinoic acid, which, it could be argued, may be a better model of myeloid malignancies rather than primary myelopoiesis23,24,25. Nonetheless, these studies have revealed interactions between key molecules in myelopoiesis and miRNAs. Myeloid transcription factors regulate expression of miRNAs and often form feedback loops e.g. C/EBP-α and miRs-34a, -223 and -30c26,27,28; Gfi1 and miRs-21 and -196b29; and several miRNAs regulated by PU.130. Conversely, differential expression of several miRNAs such as miR-223 and miR-17–5p, and consequent regulation of their targets, is required for normal myelopoiesis18,31.
While the transition of HSCs to progenitors has been previously examined, few studies have directly evaluated the influence of miRNAs on cell fate decisions subsequent to the progenitor, particularly in the human context. More generally, studies have explored miRNAs in bulk human CD34+ cells, which would encompass both stem and progenitor cells, versus mature cells from the same source to identify lineage specific miRs32,33. One previous study by Lu et al. (2008) examined human bone marrow derived MEPs to identify miRNAs that influence cell fate choice between megakaryocytes or erythrocytes with the identification of miR-150 as crucial in this process34.
To date a comparable investigation during monocytopoiesis or granulopoiesis has not been reported. In this study we examined the miRNA profiles of primary myeloid progenitors, monocytes and granulocytes derived from human cord blood. We also examined the mRNA and miRNA profiles of the same cells in combination with target prediction software. Here we identified and validated targets of two miRNAs with relevance to human myelopoiesis.
Isolation of myeloid populations
Myeloid progenitors and monocytes were isolated from human umbilical cord blood by fluorescence activated cell sorting (FACS), following the enrichment of mononuclear cells by density gradient centrifugation. We adopted the definition of myeloid progenitors as lin−CD34+CD38+ cells to incorporate the populations identified as the CMP, GMP and MEP. This conforms with classifications defined by Manz et al. (2002), based on colony forming potential and mRNA expression of relevant genes4. The myeloid progenitors in this study were purified using the same strategy, excluding mature cells with a lineage cocktail incorporating antibodies against the markers CD3, CD7, CD10, CD11b, CD14, CD19 and CD56, then sorting the CD34 and CD38 double positive cells (Fig. 1a). This lin−CD34+CD38+ population represents approximately 1% of the mononuclear cell fraction of cord blood. We further assessed their myeloid progenitor status using colony forming unit (CFU) assays in semi-solid Methocult medium supplemented with cytokines. Our results confirmed that the progenitor population isolated by us could give rise to CFU-granulocyte/erythrocyte/monocyte/macrophage (CFU-GEMM), CFU-granulocyte/macrophage (CFU-GM), CFU-granulocyte (CFU-G), CFU-macrophage (CFU-M), CFU-erythrocyte (CFU-E) and blast forming unit-erythrocyte (BFU-E) (Fig. 1d–f).
Expression of the integrin alpha M molecule (ITGAM), also known as CD11b, was used to identify monocytes and granulocytes in the enriched samples. Monocytes were identified as mononuclear cells with dual expression of CD11b and the bacterial lipopolysaccharide co-receptor CD14 (Fig. 1b). Granulocyte isolation was performed on the cells that separated from the mononuclear cells and settled with the red blood cells using a dextran gradient. During sorting granulocytes were identified following exclusion of contaminating red blood cells using CD235a and then isolating the cells positive for both CD11b and the granulocyte-specific carbohydrate adhesion molecule CD15 (Fig. 1c). The sorted cells were immediately stored for RNA extraction.
MicroRNA profiling of myeloid cells
Next, TaqMan Low Density Arrays (TLDAs) were used to profile the expression of 365 miRNAs, of which 113 were amplified in at least one of the three cell types. In a pairwise comparison of gene expression in progenitors vs. monocytes 33 miRNAs were increased in monocytes, while 29 miRNAs were reduced (≥2-fold) (Fig. 2a and Table 1). When comparing progenitors with granulocytes, 30 miRNAs were increased in granulocytes and 36 miRNAs were reduced (≥2-fold) (Fig. 2b and Table 1).
In order to identify miRNAs that are increased with differentiation in general, we combined expression data from monocytes and granulocytes. A subset of 19 miRNAs were increased in both monocytes and granulocytes compared to progenitors (Fig. 2c and Table 1). miR-223 showed the greatest increase, with levels in granulocytes and monocytes 24- and 9-fold higher than in progenitors (Fig. 2d, top panel). Among the commonly increased candidates were also several miRNAs that are clustered together in the genome and are often expressed in a similar pattern as each other (Fig. 2d, middle and lower panels). Interestingly, a similar comparison of commonly down-regulated genes revealed 27 miRNAs that were decreased in both monocytes and granulocytes relative to progenitors (Fig. 2e). These include five of the eight members of the miR-10 gene family (miRs-10a, -99a, -100, -125a and -125b, as classified by miRBase) (Fig. 2f, top panel) and also several clustered miRNAs (miRs-20a, -20b and -92, Fig. 2f middle and lower panels).
Five differentially expressed miRNAs were chosen for qRT-PCR validation, with expression normalised to a snoRNA RNU24 (Fig. 3a). miR-223, miR-125b, miR-10a, miR-196b and miR-135a showed similar patterns of expression to that observed in the TLDA data. When measured by qRT-PCR the mean level of miR-223 was ten-fold higher in monocytes relative to progenitors, and was sixty-fold higher in granulocytes. miR-125b was expressed at ten-fold lower levels in monocytes and granulocytes relative to progenitors, consistent with the TLDA results. The Hox-cluster encoded miR-10a could be amplified from all the progenitor samples but only from seven monocyte samples and eight granulocyte samples. Mean expression of miR-10a was a hundred-fold lower in both monocytes and granulocytes, making the levels too low to be detected from the remaining five monocyte and granulocyte samples.
The second Hox-cluster encoded miRNA, miR-196b, was expressed at thirty-fold lower levels in monocytes and fifteen-fold lower levels in granulocytes, relative to progenitors. miR-135a could be detected in all the progenitor samples, though the levels of miRNA were quite low. It could only be amplified from four monocyte samples and two granulocyte samples and showed approximately thirty-fold lower levels in the mature cells relative to the progenitor cells.
Two other studies have reported the miRNA expression profiles of differentiating hematopoietic stem and progenitor cells. Bissels et al. (2011) compared bone marrow CD133+ (more primitive) with CD133−CD34+ cells (less primitive)35 while Liao et al. (2008) studied more differentiated cells and compared CD34+CD38− cells with CD34+ cells36. Similarly, Cattaneo et al. (2015) compared miRNA profiles of CD34+CD38− cells (HSCs) and CD34+CD38+ cells (hematopoietic progenitor cells; HPCs)20. The miRNA profiles of the progenitors in our study are comparable as they could be placed at the next step with their more differentiated phenotype of CD34+CD38+. The miRNA profiles in these three studies and in this study were obtained using four different profiling platforms and therefore vary in the miRNAs included. Nonetheless, in order to identify miRNAs that are increased or decreased with differentiation we compared those that were reported as increasing or decreasing with differentiation in these three studies with those in ours.
We found ten miRNAs that decrease with differentiation in at least three of these four studies, suggesting that they may be involved in maintenance of a more undifferentiated phenotype and seven miRNAs that increase (Table 2). Interestingly, four of the reduced miRNAs, miR-99a, miR-100, miR-125b, miR-125a-5p, are members of homologous tricistronic clusters involved in stem and progenitor cell homeostasis37. Therefore there are 12 miRNAs that increase and 18 that decrease with differentiation from this study that were not reported by the previous studies of stem cells/progenitors. The miRNAs that increase with differentiation in our study may indicate some that are required for a terminally differentiated phenotype (e.g. miR-223 for granulocytes and miR-15b for megakaryocytes) and the ones that decrease tend to be those that help to maintain the stem/progenitor cell phenotype (e.g. let-7c).
These data together suggest a signature of miRNA expression associated with differentiation status and maturation within the myeloid lineage. They also provide some evidence of monocyte and granulocyte lineage specific miRNAs with varying expression in each of the mature populations.
Decreased expression of miR-10 family members is associated with increased lineage commitment
The miRBase database lists miR-s-10a, -10b, -125a, -125b, -99a, -99b and -100 as members of the miR-10 family. As there was a clear association of decreased expression of these miRNAs with more mature phenotypes, we further explored two family members, miR-125b and miR-10a in this context.
In initial screening experiments we sought to explore the change in levels of these miRNAs in response to short-term liquid culture. FACS purified cord blood CD34+ cells were cultured in medium with serum and SCF or IL-3 (Supplementary Figure 1). Cells were sampled at days 1, 2 and 3 and levels of miR-125b and miR-10a assessed by qRT-PCR. In response to SCF, levels of miR-125b decreased by an average of 40% by day 3. In response to IL-3, miR-125b levels decreased by an average of 60%. Levels of miR-10a mirrored this and decreased 50% by day 3 with SCF and 80% with IL-3.
While SCF and IL-3 by themselves can stimulate granulocyte/macrophage colony formation, the combination of SCF with GCSF is a more potent and specific inducer of myeloid commitment in CD34+ cells in liquid and semi-solid cultures38. In order to focus on granulocytic commitment using longer culture duration, MACS-purified cord blood CD34+ cells were differentiated in vitro using SCF and GCSF and their cell-surface expression of CD34, CD11b, CD14 and CD15 assessed. By day 6, expression of CD34 was decreased from approximately 70% to 15% of cells, with a concomitant increase in surface expression of the mature myeloid markers CD11b, CD14 and particularly of CD15 (Fig. 3b). Over a similar time course qRT-PCR showed a seven-fold decrease in miR-125b and approximately ten-fold decrease in miR-10a by day 9 (Fig. 3c). miR-223 expression was used as a positive control and showed an approximately three-fold increase by day 9 (Fig. 3c). These data confirm the decrease in expression of miR-10a and miR-125b during early myeloid differentiation.
Identification of predicted targets of miR-125b and miR-10a
Affymetrix microarrays were performed on progenitors, monocytes and granulocytes that had been FACS isolated as described above from three cord blood samples. Unsupervised hierarchical clustering showed that the gene expression of the two mature cell types was closer to each other than to the progenitors (Supplementary Figure 2). To assess the differences in gene expression between the progenitors and each of the mature populations, we examined the top 2000 genes that were increased and decreased in monocytes vs. progenitors and in granulocytes vs. progenitors (Fig. 4a and b). Gene ontology analysis of these sets showed that the differentially expressed genes were enriched in pathways pertaining to the acquisition of features of mature myeloid cells (Table 3). In the comparison of progenitors with mature monocytes the most significantly over-represented pathways include ‘regulation of myeloid leukocyte mediated immunity’, ‘toll-like receptor signalling pathway’ and ‘pattern recognition receptor signalling pathway’. In progenitors versus granulocytes the most significantly over-represented pathways included ‘granulocyte activation’, ‘rRNA processing’ and ‘nonsense-mediated decay’. These results further confirm the identity and maturation status of the cells used in our study.
In order to identify mRNA targets of our miRNAs of interest from among these differentially expressed genes, we first considered bioinformatically-predicted targets of the miRNAs using the miRanda algorithm39,40, which uses sequence composition, conservation and thermodynamic stability as criteria for predicting miRNA target sites. This algorithm applies a sensitive method that has a high relative overlap with predictions from other algorithms. The top 1000 gene candidates that were increased in monocytes and granulocytes versus myeloid progenitors were combined with the target predictions for miR-125b and miR-10a (Fig. 4c and d). The resulting 162 candidates for miR-125b and 161 candidates for miR-10a that were both predicted targets and had inverse expression patterns to the miRNAs were investigated further through literature searches. This process resulted in the selection of three putative targets for confirmatory testing using the luciferase reporter assay (MCL1, FUT4 and KLF4). The increase in MCL1, FUT4 and KLF4 expression during myeloid differentiation as determined in the microarray analysis is shown in Supplementary Figure 3.
Experimental confirmation of targets of miR-125b and miR-10a by luciferase assays
FUT4 mRNA has two predicted miR-125b binding sites (~100 nt apart) in its 3′ UTR, whilst MCL1 mRNA also contains a predicted miR-125b target site in its 3′ UTR (Fig. 5a). PhastCons scores for FUT4 and MCL1 miR-125b binding sites indicate that their seed regions are highly conserved in vertebrates (Fig. 5a). KLF4 mRNA contains a predicted miR-10a binding site, which is also highly conserved in vertebrate species (Fig. 5a). To experimentally validate these miRNA-target interactions we used HeLa cells in which we had confirmed the presence of endogenously expressed miR-125b and miR-10a by qRT-PCR (Supplementary Figure 4). HeLa cells were transfected with luciferase reporter vectors carrying either WT or mutated (mut) 3′ UTRs of interest located downstream of luciferase. If repression of the WT vector relative to the mut vector was observed, they were then co-transfected with miR-125b or miR-10a inhibitor (125I/10I) or negative control (NC) inhibitor to knock down the miRNA expression and ‘rescue’ the repression.
As the 3′ UTR of FUT4 has two predicted binding sites for miR-125b (Fig. 5a), repression of the WT 3′ UTR was tested alongside single mutants of each of the binding sites (Site 1 mut and Site 2 mut), and a construct with both seed regions disrupted (double mut). At the lower inhibitor concentration of 10 nM the luciferase signal from the WT vector with NC was only 66% of the double mut vector indicating repression by cell intrinsic miR-125b (Fig. 5b). This recovered to 90% when the WT vector was co-transfected with 10 nM 125I (p < 0.05). Upon mutation of miR-125b Site 1 in FUT4, the luciferase signal decreased to 79% of the double mut vector (i.e. 21% lower signal due to miR-125b binding the intact Site 2). This Site 1 mut vector was completely de-repressed when co-transfected with 125I (p < 0.05). Conversely, the vector containing the mutant Site 2 had a luciferase signal that was 90% of the double mut vector, i.e. only a modest 10% signal decrease due to miR-125b-mediated repression through the intact Site 1. Upon addition of 125I to the Site 2 mutant construct no de-repression was observed, confirming that minimal repression occurred through binding Site 1.
A similar pattern was observed when the same experiment was repeated at the higher concentration of the miRNA inhibitors. At 50 nM, the FUT4 WT vector was de-repressed by 39% with 125I vs. double mut (Fig. 5c, p < 0.01) and the Site 1 mut vector was de-repressed by 37% with 125I (p < 0.05), as a result of the effect of miR-125b/125I on the intact, conserved binding Site 2 in this vector. As at the lower inhibitor dose, there was no significant difference in luciferase expression in the Site 2 mut vector between NC or 125I transfected samples, due to minimal repression by miR-125b binding to the intact site 1 in this vector. These data together suggest that FUT4 expression is modulated by miR-125b by repression of the transcript through two binding sites, with most of the repression due to the more conserved of the two sites.
Luciferase vectors carrying either the WT or mut MCL1 3′ UTR were co-transfected with NC or 125I inhibitors at 10 nM or 50 nM. At 10 nM NC inhibitor, the luciferase signal of the WT vector was only 83% of the mut vector (Fig. 5d) indicating repression by miR-125b. When co-transfected with 125I the WT signal recovered to 96% of the mut, though this was not significant. At the higher inhibitor concentration of 50 nM a small non-specific inhibition was observed (31% in WT + NC vs. 10 nM and 35% in mut + 125I vs. 10 nM), however de-repression as a result of specific inhibition of miR-125b was still observed. At 50 nM, the WT luc signal was 63% of mut with NC, but recovered to 93% upon addition of 125I (p < 0.01), suggesting that miR-125b does regulate the expression of MCL1.
The KLF4 gene has one predicted binding site for miR-10a in its 3′ UTR. Luciferase assays comparing the 3′ UTR with intact (WT) or mutated (mut) versions of this binding site showed modest but consistent and significant repression through this binding site (Fig. 5e). Repression could be recovered upon addition of 10I (14% de-repression at 10 nM, p < 0.05; 11% de-repression at 50 nM). The dose of the inhibitor did not affect the degree of de-repression.
In summary, luciferase reporter assays confirmed both MCL1 and FUT4 as direct targets of miR-125b and KLF4 as a target of miR-10a.
MicroRNAs are crucial regulators of differentiation and lineage enforcement in the hematopoietic system41. While studies have examined miRNA profiles of bulk CD34+ cells, no study has specifically examined myeloid progenitors and their progeny concurrently. This study presents a miRNA profile of human myeloid progenitors and corresponding mature monocytes and granulocytes performed on highly pure and clearly defined primary cells. We identified miRNAs that are commonly increased and decreased upon differentiation in both lineages and also those that are unique to each lineage. For example, we demonstrate the pattern of expression of miR-125b in primary human cells, whereas previous data were obtained using murine cells. We also identified and experimentally validated some novel targets of miR-125b and another myeloid-relevant miRNA, miR-10a in this particular cellular context.
The role of miRNAs in myeloid leukaemia has been reported extensively42, however close examination of their role in primary human non-leukaemic stem and progenitor cells has been relatively limited. We observed a consistent pattern in the miRNA profile of our progenitor cells, and those previously reported in CD34+ cells21,32. This adds validation to the miRNAs generally identified as stem/progenitor-expressed (Table 2). In addition to this our data have highlighted human hematopoietic differentiation associated miRNAs (Table 1).
While studies by Bissels et al., Liao et al. and Cattaneo et al. focused on identifying miRNAs in more primitive stem cells20,35,36 our data identify miRNA changes after the cells pass these early stages and move towards maturity. The importance of the miRNAs we have identified in this manner are highlighted by the roles of some of them in other models. For instance, of the miRNAs that decrease with hematopoietic differentiation in our analysis, mouse studies have shown that a decrease in miR-146a is crucial for megakaryocytopoiesis43 and a decrease in the miR-99a/miR-125a cluster is important for loss of stemness as the cells mature44. Similarly, an increase in miR-142-3p is necessary for differentiation of myeloid cell lines45 and an increase in miR-16 is necessary for erythropoiesis46. Several of the miRNAs identified by us as relevant for monocyte or granulocyte differentiation have also been reported in in vitro models of human myeloid differentiation22,33.
The upregulation of miR-223 we observed during monocytopoiesis and granulopoiesis, has also been recently shown31, but with their additional observation of miR-223 being maintained at low levels during erythropoiesis. The authors further show that the lineage-specific expression levels of miR-223 are fine-tuned by the coordinated recruitment and function of different lineage-specific transcription factors to the miR-223 regulatory regions31. The functional role of miR-223 in granulopoiesis is realized through a regulatory circuit involving the transcription factors C/EBPβ, PU.1, NFI-A and C/EBPα47. miR-223 itself represses the expression of NFI-A and other transcription factors that are important in hematopoietic differentiation processes such as MEF2C, a transcription factor that promotes myeloid progenitor proliferation48, LMO2, a key regulator of erythropoiesis49, and E2F1, the master regulator of cell-cycle progression27. An overview summarising the miR-223 regulatory circuitry in myeloid differentiation is illustrated in Fig. 6a.
In addition to these miRNAs that have been previously reported, we have also identified some miRNAs that have not been studied in the context of myeloid differentiation, such as miR-127 and miR-484. Both of these miRNAs have roles defined predominantly in cancer50,51,52 and may provide future opportunities for further investigation.
The cytokines SCF and IL-3 are often used in combination and with other cytokines for ex vivo expansion of hematopoietic stem cells53. Prior to our experiments in primary CD34+ cells, we tested the short-term effects of these cytokines on levels of miR-125b and miR-10a and found them to cause a relatively modest decrease in their levels. These findings correlate well with previous reports of these cytokines having a similar modest effect on granulocyte/macrophage colony formation in semi-solid and liquid cultures38. To delve deeper into lineage commitment and the role of our miRNAs of interest, we focused on granulopoiesis and extended the culture for longer. Cytokine mediated in vitro differentiation of progenitors towards the granulocytic phenotype with GCSF (and SCF) resulted in the decrease of both miR-125b and miR-10a to a much greater degree (approx. 10-fold), and correlated well with acquisition of differentiation markers over the same period.
Several studies in the mouse have identified miR-125b and family member miR-125a as important for various aspects of primitive and leukaemic hemopoiesis54. In terms of the myeloid lineage, the earliest report by Bousquet et al. showed that a common translocation in AML results in miR-125b overexpression55. Overexpression of miR-125b resulted in various leukaemias in mice56. In fact, several groups have over-expressed miR-125b in the mouse hematopoietic system using different methods56,57,58 and have all observed various myeloproliferative disorders depending on the method and degree of overexpression. Together these studies also report that the highest level of miR-125b is found in long-term HSCs, with levels progressively decreasing in more differentiated progenitors. The important roles of miR-125b and miR-125a in human myelopoiesis is highlighted by their frequent dysregulation in various myeloid malignancies (reviewed in ref.54). However, little has been reported regarding miR-125b in lineage definition of primary cells such as granulocytes.
In mouse, levels of miR-125b (and miR-125a) decrease with increasing granulocyte maturity59. In order to understand dynamic changes, two studies have examined the role of miR-125b in the murine IL-3 dependent promyelocytic cell line 32D60,61. Both studies report that 32D cells have very low levels of miR-125b, but Surdziel et al. found that miR-125b levels increased upon GCSF-induced differentiation61. Our experiments, performed with primary human cells found the contrary; that miR-125b levels decreased during granulocytic differentiation with GCSF, agreeing with findings in primary murine granulocytes59. Our findings are further supported by two studies involving miRNA expression at different stages of human neutrophil maturation19,62. Surdziel et al. and Bousquet et al. observed that over-expression of miR-125b in murine 32D cells resulted in increased proliferation and maintenance of an immature blast cell phenotype60,61. These findings, coupled with our observation that miR-125b levels decrease with increasing maturity, suggest that reduced miR-125b is necessary for normal (i.e. non-dysplastic) granulocyte development.
We anticipate that miR-125b achieves its regulatory effects through repression of target mRNA molecules. MiRNA target prediction algorithms use various parameters including degree of complementarity with target mRNA site, miRNA and binding site conservation and number of sites per mRNA to make their predictions across all transcripts63. In order to identify transcripts of relevance, we validated transcripts that were both predicted targets and expressed.
The first miR-125b target we validated was FUT4. FUT4 encodes a member of the fucosyltransferase family, that produce carbohydrate moieties on the surface of cells that act as ligands for selectins and facilitate granulocyte adhesion to, and rolling on, endothelial cells. FUT4 is responsible for fucosylating the CD15 cell surface antigen in monocytes and promyelocytes64, as well as the production of ligands for P- and L- selectin in conjunction with another fucosyltransferase, FUT765,66. The levels of FUT4 (and FUT7) transcripts increase in response to GCSF, along with E-selectin ligands on immature and late mature granulocytes, as well as levels of E-selectin on endothelium67. These reports combined with our findings that FUT4 is a target of miR-125b also highlight an important relationship between GCSF, miR-125b and FUT4 in the development and function of myeloid cells. Interestingly, FUT7 which encodes a similar enzyme expressed in myeloid cells, also has a predicted binding site for miR-125b in its 3′ UTR, though this was not experimentally validated. These observations add further evidence that decreasing expression of miR-125b during myeloid differentiation contributes to the development of various lineage-specific and lineage-defining features.
The second validated miR-125b target was MCL-1, which encodes an anti-apoptotic member of the Bcl-2 family. A regulatory relationship between this miRNA/mRNA pair has been reported in various cancers68,69,70 but not in a myeloid context, where MCL-1 plays an important role. Conditional knock-out mice lacking Mcl-1 in neutrophils and macrophages develop with no neutrophils and severely depleted monocytes71,72. These mice have increased numbers of immature myeloid precursors in the bone marrow. The administration of GCSF did not lead to increased peripheral blood granulocytes, suggesting that this process requires MCL172. Our findings that miR-125b expression is decreased in response to GCSF and that miR-125b represses the mRNA for MCL1 suggests a regulatory relationship between this miRNA/mRNA pair and the process of GCSF-mediated granulopoiesis.
In summary, miR-125b targets are central to regulating important organismal processes including apoptosis, innate immunity, inflammation and hematopoietic differentiation54. During hematopoietic differentiation, miR-125b regulates a number of genes including cytokines and transcription factors critical for lymphoid and myeloid development (summarised in Fig. 6b). These include transcription factors BLIMP-1 and IRF-4 during human B-lineage differentiation54 and IFNG, IL2RB, IL10RA and PRDM1 during T-cell differentiation73. miR-125b also represses the expression of the RNA-binding protein Lin-28 which has been shown to be an important regulator of hematopoietic development74. In myelopoiesis, miR-125b is both induced by GCSF and negatively regulates GCSF expression via a feedback loop54 and also targets transcription factors STAT3, c-JUN and JUND and the pro-apoptotic effector BAK161. We have now shown that the miR-125b regulatory circuitry is de-repressed during myeloid differentiation leading to an upregulation of FUT4 and MCL1 (Fig. 6b).
The second miRNA examined in this study, miR-10a, also exhibits decreasing expression during differentiation. Previously, a decrease in miR-10a has been reported during megakaryocytopoiesis75 and knock down of miR-10a in CD133+ progenitors can induce megakaryocytic differentiation via the transcription factor HOX1A76. An outstanding potential target of this miRNA was the transcription factor KLF4. While KLF4 is best known as one of the four factors necessary for reprogramming somatic cells into induced pluripotent stem cells, KLF4 is also a crucial transcription factor for monocyte development from CMPs, acting downstream of PU.177. While the targeting of KLF4 by miR-10a has been previously suggested to be relevant in AML78, we have shown that the decrease in miR-10a expression upon primary myeloid differentiation and the consequent release of repression may be necessary for monocytopoiesis (Fig. 6c).
Other hematopoietic regulators that are targets of miR-10a repression include BCL-6 and NCOR2, transcriptional repressors involved in T cell differentiation79. In lymphopoiesis miR-10a expression is induced by retinoic acid and TGFß79. Apart from the validated target HOX1A, qRT-PCR-based target validation suggests the cell growth inducer USF2 as another target of miR-10a important in megakaryocytopoiesis80.
In summary, by examining primary human cells at defined points in myeloid differentiation we have identified miRNAs that direct specific targets, with potential for great impact in this cellular context. Our data identify a miRNA profile of myeloid progenitor cells, extending previous information in the field on more primitive stem cells. We further identified changes in the levels of these miRNAs as the cells mature into two separate myeloid lineages. We focused on two miRNAs miR-125b and miR-10a and uncovered their potential role in this process by identifying their likely mRNA targets. We provide biological validation of some of these targets revealing further their specific roles in the development of mature monocytes and granulocytes from the myeloid progenitor.
Materials and Methods
Cord blood collection and cell isolation
All research and procedures were performed in accordance with all relevant guidelines and regulations. Human cord blood from healthy donors was obtained from the Sydney Cord Blood Bank following informed consent according to approved protocols (X06-0153 and X09-0111) from the Sydney Local Health District Human Research Ethics Committee. Typically, samples contained 30–60 mL of cord blood with added citrate-dextrose anti-coagulant (ACD-A). Following separation using Ficoll-Paque the mononuclear cells were collected for isolation of monocytes and progenitors. The granulocytes were further isolated by centrifugation and red cell lysis following layering over a solution of 3% (w/v) dextran (Sigma-Aldrich; Sydney, Australia). In some experiments (indicated in text) CD34+ cells were enriched from the mononuclear cell fraction using MACS beads and a positive selection kit (Cat # 130-046-702; Miltenyi Biotec), following the manufacturer’s instructions.
All antibodies used for FACS staining were obtained from Biolegend (San Diego, California, USA). The lineage (lin) cocktail comprised antibodies against CD3 (clone HIT3a), CD7 (MEM-186), CD10 (HI10a), CD11b (ICRF44), CD14 (M5E2), CD19 (HIB19) and CD56 (MEM-188). The antibodies against CD34, CD38 and CD235a were from the following clones respectively: 4H11, HIT2 and HIR2. Cells were sorted to >90% purity on either a BD FACS Vantage or FACS Aria. Analytical flow cytometry was performed on a BD FACS Canto.
FACS purified myeloid progenitor cells were resuspended in Iscove’s Modified Dulbecco’s Medium (IMDM) then added to methylcellulose containing the cytokines GM-CSF, EPO, interleukin-3 (IL-3) and stem cell factor (SCF) (Methocult, H4434; Stem Cell Technologies) and plated in triplicate into 35 mm Petri dishes (1.1 mL/dish). The plates were incubated at 37 °C, 5% CO2 in a humidified environment for 10–16 days. Colony formation was observed using a MZFLIII microscope (Leica Microsystems, Wetzlar, Germany) and photographed with a DFC500 camera attached (Leica).
TaqMan® Low Density Arrays (TLDA) and miRNA assays
Ten cord blood samples were used for miRNA profiling. MiRNA profiling was performed using TaqMan Low Density Arrays (TLDAs – early access version; Applied Biosystems; Foster City, California, USA). Two TLDAs were performed for each cell type with a pool of 5 samples on each array. RNA was extracted from cells using Trizol reagent (Invitrogen; Sydney, Australia) following the manufacturer’s instructions. A total of 100 ng of RNA was reverse transcribed using Megaplex primers (Human Pool set v1, pre-release evaluation version; Applied Biosystems). The cDNA and master-mix was loaded on to the TLDA card and amplified using an Applied Biosystems 7900HT Fast Real-Time PCR System. The raw data were first analysed using RQ Manager software (version 1.2; Applied Biosystems) and then using Qiagen’s GeneGlobe Data Analysis Centre: (http://pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php). Validation of array results was performed using TaqMan miRNA Assays on the 10 samples used for the array individually. All qRT-PCRs performed in this study to amplify specific miRNAs were undertaken using TaqMan miRNA assays specific for the miRNA of interest on a Corbett Rotor Gene (RG 3000; Qiagen, Doncaster, Victoria, Australia).
Total RNA from myeloid progenitors, monocytes and granulocytes isolated from human cord blood was extracted using TRIzol reagent (Life Technologies) and profiled using Affymetrix GeneChip Gene 1.0 ST human arrays. The raw data were processed using the RMA method from the R Bioconductor ‘oligo’ package for background subtraction, normalization, and summarization of probe level data. They were then plotted as heat maps using the web-based tool Morpheus (software.broadinstitute.org/morpheus).
The sequences of 3′ UTRs used for luciferase reporter assays were PCR-amplified from normal human genomic DNA and sequence verified, then cloned into the dual luciferase vector pSiCHECK-2™ and sequence verified again. To create mutant miRNA binding site vectors, mutagenesis was performed using the enzyme Phusion® (Thermo Fisher). For the assay the plasmids were transfected into HeLa cells using Lipofectamine™2000 (Invitrogen; Carlsbad, CA, USA) according to the manufacturer’s instructions. miRIDIAN hairpin inhibitor control siRNA (IN-001005-01-05) or siRNA targeting miR-125b and miR-10a (IH-300595-05, IH-300549-05 respectively, all from Thermo Scientific Dharmacon) were used to suppress miRNA activity. Transfected cells were incubated for 24 h at 37 °C, 5% CO2 followed by lysis. A dual luciferase assay using Dual-Glo Stop & Glo substrate (Promega) was performed on a Victor2 Wallac Plate Reader.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors acknowledge grant funding from the Cancer Council NSW RG11-11 and NHMRC #1080530 both to JEJR. JEJR was funded by Cure the Future and an anonymous foundation. US holds a fellowship from the Cancer Institute of NSW. The authors would like to thank the Sydney Cord Blood Bank for the cord blood samples, the Centenary Institute Flow Cytometry Facility and Applied Biosystems for supplying the TLDAs used in this study.