|Home | About | Journals | Submit | Contact Us | Français|
Precise regulatory mechanisms are required to appropriately modulate the cellular levels of transcription factors controlling cell fate decisions during blood cell development. Here, we show that miR-126 is a novel physiological regulator of the proto-oncogene c-myb during definitive hematopoiesis. We show that knockdown of miR-126 results in increased c-Myb levels and promotes erythropoiesis at the expense of thrombopoiesis in vivo. We further provide evidence that specification of thrombocyte versus erythrocyte cell lineages is altered by the concerted activities of the miRNAs miR-126 and miR-150. Both microRNAs are required but not sufficient individually to precisely regulate the cell fate decision between erythroid and megakaryocytic lineages during definitive hematopoiesis in vivo. These results support the notion that microRNAs not only act to provide precision to developmental programs but also are essential determinants in the control of variable potential functions of a single gene during hematopoiesis.
The determination of hematopoietic lineages heavily relies upon regulation at the transcriptional level, and the functional roles of many transcription factors that are essential for hematopoietic specification have been well established. 1–4 Throughout hematopoietic development, the cellular levels of these transcription factors undergo dynamic changes that affect lineage specification as well as the proliferation and maturation of a given cell type 5. The c-Myb protein is a major hematopoietic transcription factor whose levels must be precisely controlled during normal blood cell development and whose aberrant expression contributes to leukemogenesis. 6–10 C-Myb functions at multiple stages of blood cell development from hematopoietic stem cells (HSCs) and progenitor cells to fully differentiated cell types, and can affect the definitive specification of lymphoid, myeloid, erythroid and megakaryocyte lineages. Given the spatio-temporal diversity of its developmental roles, the mechanisms controlling the timing and levels of c-Myb activity must be precise and tightly maintained. While the c-Myb protein has a very short half-life, the regulatory control of its degradation, synthesis and mRNA stability remains poorly understood.
MicroRNAs (miRNAs) provide an elegant mechanism that affords finely tuned control of protein levels by controlling translational efficacy and mRNA stability 11. These noncoding small RNA molecules are quickly becoming recognized as key regulators of diverse cellular functions, including the specification and differentiation of hematopoietic cells 12. The c-myb 3′UTR contains predicted target sites for more than a dozen miRNAs 13–15. Up to date, regulation of c-Myb by two of these miRNAs has been experimentally validated. The overexpression of miR-15 inhibits c-myb 3′UTR-dependent luciferase activity in vitro and contributes to the inhibition of erythroid and myeloid colony formation 16. Overexpression studies of miR-150 in vitro, using a luciferase reporter gene containing the c-myb 3′UTR, also result in a reduction in luciferase activity. miR-150 deficient mice show an increase in B1 cells which are known to require c-Myb 17; an increase in c-Myb protein level was also seen in miR-150 deficient activated B cells. These findings support the role of miR-150/c-myb interactions in B-cell maintenance and differentiation 18. Gain- and loss-of-function analyses have shown that the miR-150/c-myb interaction is also essential for the lineage decision of megakaryocyte-erythrocyte progenitor (MEP) cells to form either megakaryocytes or erythrocytes in vitro 19. Overall, these observations indicate that variations in the levels of c-Myb may be differentially controlled by multiple miRNAs to execute lineage–specific developmental programs at critical junctions during hematopoiesis. Nevertheless, possible combinatorial effects of multiple miRNAs regulating c-Myb within the same developmental context have yet to be addressed experimentally.
Zebrafish and mammalian hematopoiesis are highly conserved on molecular and cellular levels and the phases corresponding to mammalian embryonic and fetal hematopoiesis occur rapidly in the zebrafish with definitive hematopoiesis predominating by 5 days post-fertilization (dpf). Therefore zebrafish provide an ideal system to study the function of miRNAs on different cell lineages at specific hematopoietic stages 20. Furthermore, zebrafish are amenable to transient gene knockdown assays using antisense morpholinos and forced mRNA expression analyses, which provide major advantages for studying mechanisms of miRNA activities regulating embryonic development 21, 22. Hematopoietic development in zebrafish occurs in 4 distinct waves, the first two are termed ‘primitive’, generating macrophages and erythrocytes prior to 30 hours post fertilization (hpf). Two ‘definitive’ waves follow; first, erythromyeloid progenitors form at 30 hpf within the caudal hematopoietic tissue (CHT) or posterior blood island (PBI). At 32–36 hpf the second wave generates HSCs from endothelial cells lining the ventral wall of the dorsal aorta, a region thought to be comparable to the mammalian aorta-gonad-mesonephros (AGM).23, 24 These HSCs subsequently colonize the CHT and the adult hematopoietic organs, the kidney and thymus 25–27.
C-Myb is largely dispensable for primitive hematopoiesis but is essential for definitive blood formation and the normal development of erythroid and megakaryocyte lineages 28. Importantly, c-Myb is also required in non-hematopoietic tissues and c-Myb loss- or gain-of-function assays in vivo result in severe developmental malformation and embryonic lethality which can preclude hematopoietic analysis 29. One approach to circumvent this is by manipulating tissue-specific regulators of c-Myb, which allow analysis of c-Myb’s functions within a specific developmental context.
In this study we demonstrate the direct regulation of c-Myb by miR-126 and the importance of the miR-126/c-myb interaction during definitive hematopoiesis in the zebrafish in vivo. We show that precisely regulated levels of c-Myb are responsible for determining the lineage decision between megakaryocyte/thrombocyte and erythrocyte cell fates in the zebrafish. Furthermore, we show that regulation of this cell fate decision depends upon the concerted actions of both miR-126 and miR-150 assigning a permissive role to miR-126 to promote megakaryocyte/thrombocyte lineage commitment in vivo.
Morpholinos (Gene Tools, LLC, Philomath, OR, USA) were designed to target the mature miR-126 (MO126) and miR-150 (MO150), the Drosha and Dicer sites of pri-miR-126 (MOMB), the putative miR-126 seed sequence in the zf c-myb 3′UTR (c-myb-TPmiR126) and the translational start site of c-myb (MOmyb). The standard control MO from Gene Tools was used as a control (MOctrl). Synthetic miRNA duplexes for miR-126, 150 and 451 were purchased from Dharmacon (Dharmacon Inc, Lafayette, CO, USA) miR-451 was used as a negative control miRNA for target validation studies (miR-ctrl). MOp53 was co-injected with other morpholinos to abrogate non-specific morpholino toxicity 35. Morpholino and miRNA sequences are provided in Table 1. Morpholinos, miRNA duplexes and mRNAs were injected into the yolk of 1-cell stage embryos. MOs were injected at 8ng (MO126, MOctrl), 0.6ng (MO150), 0.8ng(for fluorescence quantification) or 2ng (for ISH analyses) (c-myb-TPmiR126), 3ng(MOMB), 2ng (MOmyb) and 1.6ng (MOp53); MiRNA duplexes were injected at 5-10μM; mRNAs were transcribed from linearized pCS2+ and injected at 350ng/μl (EGFP-c-myb-3′UTRwt and mut), 80ng/μl (mCherry), 100ng/μl (c-myb).
WISH for mRNAs was performed as described previously 36.
Full-length cDNA of zebrafish c-myb and c-myb 3′UTR were isolated from total RNA by one-step RTPCR (Qiagen, Valencia, CA, USA) and cloned into pCRII, verified by sequencing and subsequently subcloned. Primers for full-length zf-c-myb amplification included a 5′ BamHI site and a 3′ EcoRI site to be subcloned into the pCS2+ expression vector. Primers for the c-myb 3′UTR included a 5′ XhoI site and a 3′ XbaI site to be subcloned into a pCS2-EGFP expression vector C-terminal to the EGFP open reading frame. pCS2+. Mutated c-myb 3′UTR was generated using a site-directed mutagenesis kit (Stratagene, Santa Clara, CA, USA). Oligo sequences are provided in Table 2.
Embryos (~150) were collected from Tg(c-myb:EGFP), Tg(cd41:EGFP) Tg(gata1:dsRed) and Tg(gata1:dsRed/cd41:EGFP) at 36 hpf and 4 dpf 31, 32 respectively and disaggregated into a single cell suspension as previously described 26. Sorting of c-myb:EGFP+, cd41:EGFP+ and gata1:dsRed+ cells was performed on a BD FACSAria (Becton Dickenson, Franklin lakes, NJ, USA) and FACS analysis on gata1:dsRed/cd41:EGFP embryo suspensions was collected on a FACScaliber (Becton Dickenson, Franklin lakes, NJ, USA) and analyzed using Flowjo software (Treestar Inc, Ashland, OR, USA).
Total RNA from whole embryos or sorted cells was extracted with TRIzol (Invitrogen, Carlsbad, CA, USA). qRTPCR was performed using miRScipt Reverse Transcription and SYBR Green PCR and miRScript Primer Assays (Qiagen) according to the manufacturers protocol. All qRTPCR reactions were performed in triplicates in 2 independent experiments. Gene expression changes were quantified based on the delta-delta CT method using the relative expression software tool (REST) 37. Oligo sequences are provided in Table 2.
Embryos (50 or more) were lysed in 2μl/embryo lysis buffer (1% Nonidet P40, 0.1% SDS, 100mM NaCl, 50mM Tris pH 7.4, 10mM EDTA, 0.2mM PMSF, complete mini protein inhibitor cocktail (Roche, Indianapolis, IN, USA)), incubated on ice for 5 min. and spun down at 12500g and 4°C for 10 min. Protein concentration of the supernatant was determined using the DC Protein Assay Reagents Package (Bio-Rad, Hercules, CA, USA). For SDS-PAGE, 100μg protein was combined with 4x denaturing Laemmli buffer, heated at 70°C for 10 minutes and loaded on NuPAGE Novex Bis-Tris Mini Gels (Invitrogen) and separated according to the manufacturer’s protocol. Membranes were blocked with 5% milk powder in TBST (0.1%) and probed with primary antibody overnight at 4°C (mouse anti-c-Myb- MCA1793; 1:500 (AbD Serotec, Raleigh, NC), rabbit anti-beta-Tubulin; 1:10000- ab6046 (Abcam, Cambridge, MA, USA)). Membranes were then washed in TBST and probed with anti-rabbit or anti-mouse HRP-coupled secondary antibodies; 1:5000 (Perkin-Elmer Life Science, Waltham, MA, USA) for one hour. After washing in TBST proteins were detected by ECL (Pierce, Rockford, IL, USA).
To determine miRNA regulation of the c-myb 3′UTR in live zebrafish embryos we quantified the relative fluorescence upon co-injections of mRNA of the c-myb 3′UTR fused to EGFP, mCherry and various miRs and MOs. Co-injection of mCherry-encoding RNA with the c-myb 3′UTR:GFP mRNA was used in all conditions as an intra-embryo injection control. Relative fluorescence was calculated to account for normal variations observed in expression levels of injected mRNAs between embryos (related to minor variations in injection and translation efficiency). All experiments were performed in triplicate. Five representative embryos from each injection set were selected and imaged on a Zeiss Axio Imager Z1 microscope (Carl Zeiss, USA) using a Hamamatsu Orca-ER (Hamamatsu, Bridgewater, NJ, USA) digital camera and Openlab software (Improvision, Perkin-Elmer). To calculate relative fluorescence for each embryo, identical selections from three different body regions were chosen from individual embryos and measured for the sum of their pixel intensity using Volocity software (Improvision). Relative fluorescence was determined for each embryo (ΣGFP / ΣmCherry). These values were then averaged for all embryos analyzed for a given injection condition, normalized to the average relative fluorescence of the control miRNA-injected embryos, and compared to each other to determine the effect of a given injection condition on fluorescence intensity resulting in a readout for miRNA-mRNA interaction in vivo.
In order to identify novel miRNAs regulating c-Myb, we analyzed the c-myb 3′UTR for potential target sites of miRNAs expressed in HSCs. We were particularly interested in miR-126 because it is highly expressed in normal human and murine HSC and acute myeloid leukemia (AML) leukemia–initiating cells 38–41 and is part of a distinct micro-RNA overexpression signature in core binding factor leukemias 40. Our analysis revealed a putative miR-126 target site within 120 bp downstream of the c-myb stop codon with a perfect seed match that was conserved in three mammalian and two teleost fish species (Fig. 1A,B). Consistent with these findings, the sequence of mature miR-126 was also conserved among these organisms (Fig. 1B), as well as the sequence of miR-150 (Fig. 1C). Notably, mammalian 3′UTRs contained at least two putative miR-150 target sites, while zebrafish and medaka 3′UTRs had a single perfect seed match for miR-150 (Fig. 1A).
We evaluated the function of the putative miR-126 target site using an in vivo reporter assay in zebrafish embryos 22. Figure 2A illustrates our reporter assay where mRNA encoding the wild-type zebrafish c-myb 3′UTR fused to EGFP was co-injected with mCherry mRNA into embryos at the one cell stage. The co-injected mCherry mRNA controlled for injection variability. Changes in levels of EGFP expression relative to mCherry expression were assessed in the presence of various miRNA duplexes (Fig. 2B). Co-injection of miR-126 resulted in a statistically significant decrease (35%) in c-myb 3′UTR-EGFP fluorescence compared to co-injected control miRNAs and indicated potent negative regulatory function (Fig. 2B,C). Expression of miR-150 yielded a similar 30% reduction in EGFP fluorescence. To confirm the specificity of these observations we also performed assays using EGFP fused to a mutated 3′UTR that contained four nucleotide substitutions within the miR-126 target seed sequence, or a target protector morpholino (c-myb-TPmiR126)), which competes with miR-126 for the specific target sequence in the c-myb 3′UTR (Figs. 1D,E and and22)42. In both cases, GFP fluorescence was restored to control levels (Fig. 2B, C). Interestingly, the co-expression of both miR-126 and miR-150 showed no additive effect on the reduction of GFP fluorescence when compared with either of the miRs expressed independently. Target protector morpholinos provide an exquisite means to inhibit specific mRNA/miRNA interactions during zebrafish embryonic development without affecting other potential target mRNAs or miRNAs. Given that the miR-126 and miR-150 target sites in both fish species are in close physical proximity within the 3′UTR, we verified that c-myb-TPmiR126 blocked the miR-126 effects but did not interfere with miR-150-mediated regulation of c-Myb (Fig. 2B, C). Our findings show that miR-126 can interact in vivo with its predicted target site within the c-myb 3′UTR to attenuate translation of an upstream coding region.
To demonstrate the physiological relevance of the miR-126/c-myb interaction, we analyzed c-Myb expression in zebrafish embryos and sorted zebrafish hematopoietic stem/progenitor cells. As in mammals, c-Myb is broadly and dynamically expressed in many tissues, including HSCs of the developing hematopoietic system 20. Definitive c-myb+/runx1+ HSCs appear at approximately 36 hpf within the AGM 43, and 4 hours later begin to migrate into the kidney and thymus, the adult hematopoietic organs, and the posterior blood island (PBI) 25–27. In order to analyze c-myb mRNA levels in hematopoietic stem cells (HSCs) upon knockdown or forced expression of miR-126 in vivo, we used fluorescent-activated cell sorting (FACS) to isolate EGFP+ HSCs from embryos of the transgenic zebrafish line Tg(cmyb:EGFP), which drives EGFP under the control of the zebrafish c-myb promoter 30. MiR-126 knock-down was validated by qRT-PCR using a combination of morpholinos targeting either the mature form of miR-126 (MO126) or a multiblocker morpholino targeting the Drosha and Dicer cleavage sites within the primary miR-126 transcript (MOMB) or both. (Fig. 3B) The combination of both resulted in a robust reduction of mature miR-126 (Fig. 3A,B). A recent report indicated that miR-126 knockdown in zebrafish embryos resulted in a loss of vascular integrity and hemorrhage 44. However, at the morpholino concentrations used in this study, knockdown of miR-126 did not yield any obvious defects in circulation and embryos were viable through 4 dpf with intact circulation (Fig. 4; Suppl. Movies). In contrast, enforced expression of miR-126 by injection of a synthetic miR-126 duplex resulted in gross developmental malformations and circulation failure by 25 hpf, most likely due to the inhibition of c-Myb and other miR-126 target genes during development. Despite their morphological defects, c-myb:EGFP+ cells were formed in these embryos and were FACS sorted to provide data on c-Myb mRNA and protein levels 31 (Fig. 3A). The c-myb:EGFP+ cells from these embryos at 36 hpf revealed an expected inverse correlation of miR-126 to c-myb mRNA levels. However, the observed differences in mRNA levels were not statistically significant (Fig. 3A), suggesting that miR-126 primarily regulates c-Myb protein synthesis, rather than mRNA stability. To test this comprehensively, c-Myb protein levels were analyzed in lysates from whole zebrafish embryos at 24 hpf (Fig. 3C). Enforced expression of the miR-126 duplex reduced c-Myb protein to nearly undetectable levels, while the knockdown of miR-126 by morpholinos effectively increased c-Myb protein levels. A similar increase in c-Myb protein levels was observed when the c-myb-TPmiR126 was used to block the miR-126 target site in the c-myb 3′UTR to disrupt endogenous miR-126/c-myb interactions. We also assayed miR-150 function, which has been shown to regulate c-Myb in human and murine cells 19. The morpholino knockdown of miR-150 resulted in an increase in c-Myb protein while overexpression of miR-150 caused a reduction in c-Myb protein levels, validating a physiological role of the miR-150/c-myb interaction in the zebrafish. Our results indicate that both miR-126 and miR-150 directly regulate c-Myb protein levels in vivo during embryogenesis.
Zebrafish thrombocytes represent the functional equivalent of mammalian platelets 45. Despite this functional similarity, little is known about the ontogeny of thrombocytes in zebrafish. Megakaryocytes, the precursors of mammalian platelets, have not been identified in the zebrafish kidney marrow 20, 46. Nonetheless, many genes essential for megakaryocyte development are present in zebrafish, suggesting that their developmental program is conserved. As a novel regulator of c-Myb, we examined the role of the miR-126/c-myb interaction during thrombocyte-erythrocyte lineage commitment. In situ hybridization assays for mature markers of thrombocytes (tpoR), and erythrocytes (slc4a1) were performed after experimentally altering miR-126 and miR-150 levels (Fig. 4A–L). Functional knockdown of miR-126 by morpholinos targeting the microRNA or inhibition by the target protector morpholino, c-myb-TPmiR126, resulted in an increase in erythrocytes and a decrease in thrombocytes compared to control (Fig. 4A–F). To further demonstrate the dependence of this phenotype on the c-Myb pathway, we concurrently knocked down c-Myb expression by morpholino. While complete loss of c-Myb leads to severe morphological defects 47 titrating its knockdown demonstrated the ability to rescue the erythrocyte-thrombocyte phenotype (Fig. 4G,H). Consistent with studies in mammalian cells 48, forced expression of miR-150 resulted in an opposite effect, reducing c-Myb and increasing thrombocytes while decreasing erythrocyte numbers (Fig. 4I,J). Interestingly, the miR-126 loss-of-function phenotype was dominant over the miR-150 gain-of-function, in that over-expression of miR-150 could not rescue the miR-126 knockdown phenotype (Fig. 4K,L). These findings were confirmed independently by FACS analyses using a double transgenic reporter line Tg(cd41:EGFP):Tg(gata1:dsRed) that expressed EGFP in thrombocytes and dsRed in erythroid cells (Fig. 4M) The ratio of gata1:dsRed+ erythroid cells compared to CD41:EGFP+ thrombocytes was 7:1 in the control injected embryos. This increased to 10:1 when miR-126 was knocked down. Interestingly we see an even further increase in the ratio of gata1:dsRed/CD41:EGFP in the c-myb-TPmiR126 condition. This may reflect more complete inhibition of the interaction between miR126 and its binding site by blocking this directly compared to morpholino knockdown of miR-126 where residual miR may be present. Knockdown of c-Myb combined with knockdown of miR-126 completely abrogated the effects of the miR-126 knockdown, and supporting the in situ experiments, overexpression of miR-150 alone reduced the ratio of gata1:dsRed/CD41:EGFP by half (7:1 to 3.5:1), while it failed to do so in combination with miR-126 knockdown.
Unfortunately, the converse experiments could not be performed because either the overexpression of miR-126 or the knockdown of miR-150 resulted in severe developmental defects that precluded analysis at later developmental stages (Fig. S1 and data not shown). For each of these two miRNAs a number of additional mRNAs contain putative seed sequences. For example, the TargetScan algorithm13 for target prediction yields 18 putative targets for human miR-126 under the most stringent search parameters. For human miR-150 a total of 142 genes are predicted targets containing at least one seed sequence. Thus, the developmental effects of miR-126 overexpression and miR150 knockdown during embryogenesis are likely to arise from altered expression of not only c-Myb but one or more of their other target genes.
Since both miR-126 and miR150 target c-myb and result in altered megakaryocyte-erythroid cell fate, we addressed the relative expression levels of miR-126 and miR-150 in developing zebrafish thrombocytes and erythroid cells using quantitative RT-PCR on RNA extracted from sorted Tg(cd41:EGFP) and Tg(gata:dsRed) from 2dpf to 4dpf. We determined that the level of miR-126 is 6 fold higher at 3dpf and 10 fold higher at 4dpf in cd41-EGFP+ thrombocytes compared to miR-150 while in contrast the levels of both miRs were similar in gata1:dsRed erythroid cells (Fig 4N).
In this study, we have identified miR-126 as a physiological regulator of c-Myb. We show that it is the specific interaction of miR-126 and c-myb at the identified target site that permits post-transcriptional regulation of c-Myb levels. Our observations indicate that the level of c-Myb in hematopoietic tissue is dynamically regulated by the activities of miR-126 and miR-150, and that these effects on c-Myb result in control of the thrombocyte-erythrocyte lineage decision by multipotential progenitor cells. Thus, our data support the conservation of miRNA regulatory mechanisms of c-Myb function during mammalian and zebrafish hematopoiesis. Our data suggest that these microRNAs may act through a sequential rather than an additive mechanism to control the levels of c-Myb and that regulation of c-Myb during the thrombocyte-erythrocyte lineage decision depends preferentially on miR-126 or miR-150 at different developmental stages. Indeed, the level of miR-126 exceeds miR-150 levels at 3dpf six fold and is 10 fold higher at 4dpf in cd41-EGFP+ thrombocytes while in contrast the levels of both miRs were similar in gata1:dsRed erythroid cells. . While our study did not address more specific aspects of vessel formation it is notable that miR-126 has been demonstrated to play a role in vascular development 44, 49. Hemorrhages are a prominent feature of the phenotype observed in miR-126 knockdown in the studies by Fish et al and Nicoli et al. In view of our findings this may be a result of both loss of vascular integrity and thrombocytopenia. In addition, both microRNAs have many predicted targets, including others in the c-Myb pathway. For example, miR-126 has been suggested to target Tom1 (Target of myb 1), which in turn is suggested to activate c-Myb transcriptional activity via modulation of NF-kappaB signaling 50–52. On the other hand, miR-150 is predicted to target Pim1, another positive regulator of c-Myb 53. Hence, in addition to directly regulating c-Myb the complementary but different effects of these two miRNAs may be mediated, at least in part, by other target genes that in turn affect c-Myb function. In summary, this paper adds another level of complexity in the c-myb regulatory pathway, and underscores the importance of animal models where specific subsets of cells can be tracked during development.
We thank R. Hoffmans, U. Pyati and N. Bushati for critical comments on the manuscript and L. Zon for the Tg(c-myb:EGFP) zebrafish line. This work was supported by the NIH grant CA93152 (A.T.L). E.M.P. is the recipient of a Clinical Research Training Fellowship from Leukemia & Lymphoma Research UK. N.B is a Special Fellow of the Leukemia & Lymphoma Society.
Conflicts of Interest
The authors declare no conflicts of interest