Introduction

Genetic control over the development of hematopoietic stem cells (HSCs) is highly conserved in vertebrates, thus allowing the use of model organisms, such as zebrafish and mouse, to elucidate the regulatory mechanisms underlying HSC programming1,2. In zebrafish, HSCs emerge from the ventral wall of the dorsal aorta via an endothelial-to-hematopoietic transition (EHT)3,4, migrate to the caudal hematopoietic tissue (CHT) and finally home to the thymus and kidney to produce T cells and other blood cell lineages2. The self-renewal and differentiation of HSCs ensure a lifelong supply of hematopoietic cells to meet the demand and to renew blood cells in organs, which must be tightly controlled at the cellular and molecular levels. However, the underlying molecular mechanisms of HSC specification and differentiation during vertebrate embryogenesis are still poorly understood.

microRNAs (miRNAs) are a group of non-coding RNAs (∼22 nt) that can silence gene expression by binding to the 3′ untranslated region (UTR) or to the coding region of target mRNAs to promote mRNA destabilization or inhibit protein translation5,6. Importantly, many miRNAs have been shown to be expressed in hematopoietic lineages and to act as pivotal regulators of transcriptional programs for normal hematopoiesis, including HSC self-renewal, differentiation and functioning7,8. For example, miR-126 is highly expressed in human and mouse HSCs, and the attenuation of miR-126 activity by lentivirus-mediated knockdown leads to the inexhaustible expansion of HSCs in vivo in mouse and human9. Zebrafish miR-126 was reported to be necessary for the precise regulation of the erythroid/megakaryocytic lineage choice during definitive hematopoiesis in vivo10. In addition, miR-144 and miR-451, as direct targets of the critical hematopoietic master regulator GATA1, are required for erythropoiesis in zebrafish and mouse11,12,13. miR-142-3p is an evolutionally conserved miRNA of vertebrates, which is expressed in many different hematopoietic cells14,15,16,17,18,19,20,21. Chen et al.15 first reported that miR-142 is expressed in embryonic and adult hematopoietic tissues, including the fetal liver, bone marrow, spleen and thymus, in mice, indicating its role in both embryonic and adult hematopoiesis. Later studies using mouse and human cell lines have indicated that miR-142-3p is required for the development and function of terminally differentiated hematopoietic lineages, such as myeloid and dendritic cells. miR-142-3p can directly regulate the expression of tab2 in the myeloid lineage22 and that of IL-6 in dendritic cells(DCs)23. Very recently, zebrafish miR-142-3p (ZFIN ID: ZDB-GENE-090929-151, miR-142a-3p herein) has been shown to be required for the maturation of primitive erythrocytes, cardiogenesis and vessel integrity24,25. However, the mechanisms underlying miRNA regulation of HSC development in vivo during vertebrate embryogenesis remain largely unclear.

Here, we show that in zebrafish, miR-142a-3p controls HSC specification and differentiation by suppressing the expression of interferon regulatory factor 7 (irf7), revealing the pivotal roles of miR-142a-3p in HSC development. Moreover, miR-142-3p is also involved in the development of HSCs in mouse aorta-gonad-mesonephros (AGM) both in vitro and in vivo, suggesting that it has a well-conserved role in vertebrates.

Results

miR-142-3p is evolutionarily conserved and expressed in HSCs during zebrafish embryogenesis

miR-142-3p is highly conserved in vertebrates, from zebrafish to mammals (Supplementary information, Figure S1A). During zebrafish embryogenesis, miR-142a-3p expression was detected by RT-PCR from the tail-bud stage onward (Supplementary information, Figure S1B). Whole mount in situ hybridization (WISH) showed that miR-142a-3p is expressed at low levels in the AGM region at 24 hour post fertilization (hpf), where the first HSCs will emerge 6-10 h later3 (Figure 1A). Using the transgenic runx1:EGFP (Zhang P and Liu F, unpublished data) and cmyb:EGFP26 lines, miR-142a-3p was found to be highly expressed in GFP-positive cells, which represent the population of hematopoietic stem/progenitor cells, at 26-28 hpf (Figure 1B). Starting from 48 hpf and onwards, miR-142a-3p was specifically expressed in the CHT and at 4 day post fertilization (dpf) in the thymus (Figure 1A and 1C). Immunostaining of cryosections of 4 dpf embryos further confirmed that miR-142a-3p was expressed in the thymus (Figure 1C). In contrast, the expression of miR-142a-3p was not detected in either the thymus or the CHT region of cloche mutants that lack hematopoietic cells (Figure 1D), indicating that miR-142a-3p is indeed expressed in hematopoietic cells. The specific hematopoietic expression of miR-142a-3p suggests that it might play a critical role in definitive hematopoiesis.

Figure 1
figure 1

miR-142a-3p is specifically expressed in HSCs and T cells in zebrafish. (A) WISH images showing miR-142a-3p expression in the AGM (red arrowheads) at 24 hpf and in the CHT (red arrowheads) at 4 dpf. (B) qPCR results showing the expression of miR-142a-3p in sorted runx1:EGFP- or cmyb:EGFP-positive HSCs vs EGFP-negative cells. runx1 was used as a positive control. (C) Frozen sections showing the expression of rag1 and miR-142a-3p in the thymus at 4 dpf. The red square marks the thymus region. (D) WISH images showing the absence of miR-142a-3p expression in the thymus and CHT of the cloche mutants compared with WT siblings at 4 dpf.

Full size image

Loss of miR-142a-3p leads to defects in HSC development and T-cell differentiation

To interrogate the function of miR-142a-3p, we analyzed the effects of blocking the formation of mature miR-142a-3p by using two different antisense morpholinos (MOs) targeting different sites of the miR-142a-3p stem loop, 3p MO1 and 3p MO2, and a mismatch morpholino of miR-142a-3p, mis 3p MO, was used as a control (Figure 2A and Supplementary information, Table S1). To exclude the possibility that developmental delay was caused by the morpholino injections, we examined the expression of the somite marker myod and found that its expression levels were not obviously different in the controls and morphants at 24 hpf (Supplementary information, Figure S2A). Additionally, the expression of the neuron and pronephric duct marker pax2.1 was also not altered (Supplementary information, Figure S2A). These data suggest that at 24 hpf, the morphants grew relatively normally after the miR-142a-3p knockdown. qPCR showed that the endogenous expression of miR-142a-3p was reduced to various degrees (2-7 folds) by 3p MO1 or 3p MO2 injection (at 3.2 ng per embryo) (Figure 2A). WISH using a locked nucleic acid (LNA) antisense probe against mature miR-142a-3p revealed that the expression of miR-142a-3p in the thymus and the CHT was almost undetectable in the miR-142a-3p morphants and that the suppression effect continued through 4 dpf (Supplementary information, Figure S2B and S2C). Thus, the hematopoietic expression of miR-142a-3p was efficiently suppressed by both MOs (Figure 2A, Supplementary information, Figure S2B and S2C). Interestingly, the expression of miR-142a-5p, which originates from the opposite strand of the miR-142a gene, was detectable only in the thymus but not in the AGM region, and its expression in the thymus was also reduced by both of the miR-142a-3p MOs (data not shown).

Figure 2
figure 2

Loss of miR-142a-3p leads to defects in HSC development and T-cell differentiation in zebrafish. (A) Expression of miR-142a-3p in embryos injected with the miR-142a-3p mismatch MO, duplex, or two different miR-142a-3p MOs, at 2 and 4 dpf (mean ± SD, t-test, *P < 0.05, n = 3). Total RNA extracted from embryos at 2 and 4 dpf was subjected to qPCR, with U6 serving as the internal control. (B) WISH images showing the decreased expression of the HSC master regulators runx1 and cmyb in miR-142a-3p morphants, but no changes in the embryos injected with mis 3p MOs. (C) qPCR results showing the decreased runx1 expression in miR-142a-3p morphants, but no changes in embryos injected with mis 3p MOs at 24 and 36 hpf (mean ± SD, t-test, *P < 0.05, n = 3). (D) miR-142a-3p knockdown in cmyb:EGFP transgenic zebrafish embryos led to a decreased cmyb+ population. Left panel, representative fluorescence images showing the cmyb+ population in the CHT at 3 dpf. Right panels, quantitative analyses of the cmyb+ populations in the AGM (36 hpf) and CHT (3 dpf), showing a significant difference between the miR-142a-3p morphants and the controls (mean ± SD, t-test, P < 0.01, n = 3). (E) Decreased expression of the T-cell marker rag1 in miR-142a-3p morphants at 4 dpf. The red arrows indicate the thymus. (F) Unchanged expression of the thymic epithelial cell marker foxn1 in miR-142a-3p morphants at 4 dpf. The red arrows indicate the thymus. (G) Restoration of runx1 expression by miR-142a-3p duplex in miR-142a-3p morphants at 24 hpf. The red arrows indicate the HSCs in the AGM region.

Full size image

To test whether miR-142a-3p influences HSC development and differentiation, we examined the expression of multiple markers of HSCs and hematopoietic lineages, such as the HSC genes runx1 and cmyb and the T-cell markers rag1 and ikaros. The WISH and qPCR results showed that the miR-142a-3p morphants (generated by injection of 3p MO1 or 3p MO2) displayed decreased expression of runx1 and cmyb in the AGM and CHT at various stages (Figure 2B, 2C and Supplementary information, Figure S3A), whereas the miR-142a-5p knockdown did not have any effect on runx1 expression in the AGM (data not shown), suggesting that miR-142a-3p, but not miR-142a-5p, is required for HSC development in zebrafish. Moreover, at 4 dpf, the expression of the T -cell markers rag1 and ikaros was also decreased, whereas the expression of the thymic epithelial cell markers foxn1 and ccl25a remained unchanged (Figure 2E, 2F and Supplementary information, Figure S3A-S3C), suggesting that the decrease in the T-cell population resulting from the loss of miR-142a-3p expression is not attributable to a defective thymic epithelial microenvironment. Furthermore, knockdown experiments performed using the cmyb:EGFP transgenic line showed that the miR-142a-3p morphants had significantly reduced cmyb+ HSC populations in both the AGM and CHT regions (Figure 2D). In contrast, embryos injected with mismatch MOs displayed normal expression of the HSC markers runx1 and cmyb, and the T-cell marker rag1 (Figure 2B, 2C and Supplementary information, Figure S3A). To further confirm that the defects in HSC development in the morphants truly resulted from miR-142a-3p knockdown, we tested whether a miR-142a-3p duplex can rescue HSC gene expression in the miR-142a-3p morphants. Indeed, WISH showed that the decreased runx1 expression in the morphants was rescued by the duplex (Figure 2G). Time-course analysis revealed the continuous decrease in cmyb expression in the AGM and CHT of the morphants from 36 hpf to 5 dpf (Supplementary information, Figure S3D), further supporting that this defect in HSC development in the miR-142a-3p morphants was not due to developmental delay. Together, these data demonstrate that miR-142a-3p is required for HSC development and T-cell differentiation.

Hemogenic endothelium but not arterial programming is affected in miR-142a-3p morphants

The decreased expression of the HSC master regulators runx1 and cmyb in the miR-142a-3p morphants indicates that HSC emergence and development were severely disrupted. The first HSCs emerge in the ventral wall of the dorsal aorta (i.e., the hemogenic endothelium) via an EHT process, and arterial specification is a prerequisite for the normal emergence of HSCs1. To test whether the arterial programming was affected, we examined the expression of the arterial marker dltC and the venous marker flt4 using WISH. The results showed that the expression of dltC and flt4 was unchanged in the morphants (Figure 3A), suggesting that arterial/venous programming is not disrupted. Furthermore, a 3p MO1 injection experiment using fli1a:EGFP transgenic zebrafish showed that the dorsal aorta is normal in the morphants, compared with that of the controls (Figure 3B).

Figure 3
figure 3

Loss of miR-142-3p did not affect artery/vein identity, but affected hemogenic endothelial specification. (A) WISH images showing the unchanged expression of arterial marker dltC and venous marker flt4 in miR-142a-3p morphants at 24 hpf. (B) Confocal microscopy showing the unaffected artery and vein development in miR-142a-3p morphants using fli1a:EGFP transgenic zebrafish. DA, dorsal aorta. (C) Decreased number of hemogenic endothelial cells in the AGM as well as decreased HSCs in the CHT of miR-142a-3p morphants produced using the kdrl:mCherry/cmyb:EGFP transgenic line. Right panel, quantitative analyses of the kdrl+/cmyb+ populations showed a significant difference between the miR-142a-3p morphants and the controls (mean ± SD, t-test, *P < 0.05, n= 3).

Full size image

The decrease of runx1 expression at 24 hpf in the AGM region of miR-142a-3p morphants (Figure 2B and Supplementary information, Figure S3A) indicated that the specification of the hemogenic endothelium might be affected. To verify this hypothesis, we used the double-transgenic line kdrl:mCherry/cmyb:EGFP in which the double-positive cells (yellow cells) represent hemogenic endothelial cells4. Indeed, at 36 hpf, the number of kdrl+cmyb+ cells was reduced in the miR-142a-3p morphants, which persisted until 60 hpf, the latest stage that we monitored (Figure 3C). The reduction of the hemogenic endothelial cell population was unlikely due to dysregulated cell proliferation or cell-cycle arrest (data not shown). Therefore, considering the high expression of miR-142a-3p in the sorted cmyb+ or runx1+ cells at 26-28 hpf in wild-type (WT) zebrafish (Figure 1B), and the decrease of runx1 expression at 24 hpf (Figure 2B and Supplementary information, Figure S3A) as well as the reduced number of kdrl+cmyb+ cells at 36 hpf in miR-142a-3p morphants (Figure 3C), we concluded that the miR-142-3p knockdown affected hemogenic endothelial specification, which in turn attenuated the emergence of the earliest HSCs in the embryos.

Expression profiling identified downstream targets of miR-142a-3p

Direct comparison of the global gene expression patterns between control and miR-142a-3p-knockdown embryos might reveal further clues toward elucidating the biological effects of miR-142a-3p on HSC formation and differentiation. To identify the direct downstream targets of miR-142a-3p, we performed a combination of microarray, RT-PCR and miRNA target prediction (Figure 4A–4C and Supplementary information, Figure S4). The total RNAs of the AGM region in controls and miR-142a-3p morphants at both 2 dpf and 4 dpf were extracted and reverse-transcribed. The Cy3- or Cy5-labeled cDNAs were hybridized to a zebrafish oligo microarray. Gene set enrichment analysis was applied to the gene expression data to identify miR-142a-3p-modulated signaling and gene expression (Figure 4A). In total, 70 genes were upregulated at both 2 dpf and 4 dpf (Figure 4B). Based on combined target-prediction analyses using Pictar and TargetScan, four potential miR-142a-3p targets, irf7, krt18, foxp1a and rxrab were further analyzed in detail using RT-PCR, qPCR and WISH (Supplementary information, Figure S4A-S4C). The results showed that among these four genes, the expression of irf7 was dramatically upregulated in miR-142a-3p morphants compared to controls (Supplementary information, Figure S4A and S4B), suggesting that irf7 may be a primary target of miR-142a-3p in HSCs and their differentiated lineages. Although irf7 is one of the components of interferon signaling, its role in HSC development has not been reported. Therefore, we chose irf7 for functional analysis.

Figure 4
figure 4

irf7 is a direct target of miR-142a-3p. (A) Heat map analysis of gene expression in controls and miR-142a-3p morphants at 2 dpf and 4 dpf. (B) Fold changes of upregulated or downregulated genes at 2 dpf and 4 dpf in miR-142a-3p morphants compared to controls. (C) An imperfect match of irf7 3′ UTR with miR-142a-3p as determined by a calculation using RNAHybrid. (D) Scheme of the PGL3 constructs containing irf7 WT and mutated 3′ UTR. (E) Luciferase reporter activity of the reporter containing WT or mutated irf7 3′ UTR when co-transfected with the miR-142a-3p duplex into HEK293T cells (mean ± SD, t-test, *P < 0.05, n = 3). (F) Western blotting images showing that irf7 protein level was decreased in duplex-injected embryos at 4 dpf, compared to controls (left panel). Quantitative analysis of the western blotting results is shown in right panel.

Full size image

irf7 is a direct target of miR-142a-3p

irf7 has an imperfect match with miR-142a-3p, which potentially targets residues 312 to 339 of the irf7 3′ UTR (with a minimum free energy (MFE) of −24.3 kcal/mol; RNAHybrid Website: http://bibiserv.techfak.uni-bielefeld.de/rnahybrid/), but has no match with miR-142a-5p (Figure 4C and data not shown). qPCR results showed that irf7 expression level was markedly increased in the miR-142a-3p morphants at 2 dpf and 4 dpf (Supplementary information, Figure S4B). To determine whether miR-142a-3p regulates irf7 expression directly or indirectly, we fused the zebrafish irf7 3′ UTR containing the putative miR-142a-3p recognition site or mutated sites (mu1 or mu2) with a luciferase reporter (Figure 4D), and transfected the reporter together with the miR-142a-3p duplex into HEK293T cells. Luciferase activity was strongly repressed when miR-142a-3p was co-expressed with the WT pGL3-irf7-3′ UTR but not when it was co-expressed with the mutant ones (Figure 4E). Western blotting showed that the expression level of irf7 was decreased in duplex-injected embryos (Figure 4F). Taken together, these results suggest that miR-142a-3p directly binds to the irf7 3′ UTR region in a sequence-specific manner to repress irf7 expression.

To determine whether irf7 is responsible for the miR-142a-3p morphant phenotypes, we employed a double-knockdown approach. Importantly, double knockdown by co-injecting irf7 MO and 3p MO1 partially rescued the expression of runx1 at 24 hpf and that of cmyb at 36 hpf (Figure 5A). Furthermore, the double knockdown of miR-142a-3p and irf7 clearly rescued the expression of the T-cell marker rag1 (Figure 5A), suggesting that irf7 is required for miR-142a-3p-regulated lineage differentiation. Western blotting confirmed that co-injecting irf7 MO with 3p MO1 restored the Runx1 protein level at 24 hpf (Figure 5B), which provided further evidence for the rescue of defects in HSC development by knocking down irf7 in the miR-142a-3p morphants.

Figure 5
figure 5

irf7 knockdown can restore the expression of the HSC gene runx1 and T-cell gene rag1 in miR-142a-3p morphants. (A) irf7 knockdown restores the expression of HSC genes runx1 (24 hpf) and cmyb (36 hpf), and the T-cell marker rag1 (4 dpf) in miR-142a-3p morphants, as determined by WISH. (B) irf7 knockdown restores runx1 expression at the protein level, as shown by western blotting (left panel) and quantitative analysis of the western blotting results (right panel).

Full size image

Decreased Gcsfr-Nitric Oxide (NO) inflammation signaling might lead to defects in HSC development in the miR-142a-3p morphants

To elucidate the mechanism underlying the miR-142a-3p-mediated regulation of HSC formation and differentiation, we focused on the downstream events of miR-142a-3p-irf7 axis. irf7 is a member of the irf family, and irfmembers often function in host defense and inflammation27,28,29,30. An inflammation pathway might be involved in HSC and myeloid lineage differentiation in zebrafish31,32. Very recently, Hall et al.31 reported that infection-induced Gcsfr-NO signaling can enhance the expansion of the hematopoietic stem and progenitor cell compartment in zebrafish. To test whether this signaling was involved in the miR-142a-3p-mediated regulation of HSC development, we examined gcsfr, a very important upstream component in inflammation signaling. The expression of gcsfr in the CHT region at 36 hpf was downregulated in the miR-142a-3p morphants or the irf7-overexpressing embryos, and the decrease of gcsfr expression in the miR-142a-3p morphants was partially rescued by co-injection of irf7 MO (Figure 6A), suggesting that gcsfr acts downstream of irf7. The DAF-FM assay33 showed that NO production was decreased in the miR-142a-3p morphants but was restored by co-injection of irf7 MO (Figure 6B). NO production is also regulated by blood flow-induced Klf2a-eNOS signaling, as we reported previously34. To test whether blood flow-induced Klf2a signaling is involved in the defects in HSC development in the miR-142a-3p morphants, we evaluated the expression of klf2a. klf2a expression was not altered in the miR-142a-3p morphants compared to control embryos (Figure 6C), suggesting that klf2a-mediated blood flow did not contribute to the defects in HSC development in miR-142a-3p morphants. In addition, we also employed the rescue approach by treating miR-142a-3p morphants with the Gcsfr downstream NO agonist, S-nitroso N-acetylpenicillamine (SNAP). WISH showed that the expression of runx1 in the AGM was partially restored by treating miR-142a-3p morphants with SNAP (Figure 6D). As expected, an injection of gcsf mRNA can also rescue the expression of runx1 in miR-142a-3p morphants (Figure 6D). Taken together, these data suggest that miR-142a-3p regulates HSC formation and differentiation, at least in part, through Gcsfr-NO signaling.

Figure 6
figure 6

Decreased Gcsfr-NO inflammation signaling might lead to defects in HSC development in miR-142a-3p morphants. (A) Decreased expression of gcsfr in miR-142a-3p morphants and embryos injected with irf7 mRNA at 36 hpf. Co-injection of the irf7 MO partially rescued the gcsfr expression in miR-142a-3p morphants. Left panel, WISH; the red arrows indicate HSCs in the CHT region. Right panel, qPCR showing gcsfr expression in embryos injected with 3p MO1 alone or in combination with irf7 MO (mean ± SD, t-test, *P < 0.05, n = 3). (B) DAF-FM assay performed at 30 hpf showed the decreased production of NO in miR-142a-3p morphants and its restoration by co-injection of irf7 MO. (C) klf2a expression was unaffected in miR-142a-3p morphants at 36 hpf. (D) Partial rescue of runx1 expression was achieved by treating miR-142a-3p morphants with the NO agonist SNAP or by injecting gcsfmRNA. The red arrows indicate runx1 expression in the AGM region.

Full size image

miR-142-3p is involved in definitive hematopoiesis in mouse

Because the sequence of miR-142-3p is highly conserved in all vertebrates, including mammals (Supplementary information, Figure S1A), we hypothesized that mmu-miR-142-3p, the mouse homolog of miR-142-3p, might also function in the development of the HSCs in the AGM region of mouse embryos. To test this possibility, we first examined its expression in the mouse AGM region. The qPCR results clearly showed that miR-142-3p was highly expressed in E11.5 AGM tissues, but it was not detectable in the neighboring non-AGM tissues (Figure 7A).

Figure 7
figure 7

The involvement of miR-142-3p in HSC development in mouse. (A) Expression of mmu-miR-142-3p in dissected E11.5 AGM tissues and neighboring non-AGM tissues was determined by qPCR (mean ± SD, t-test, *P < 0.05, n = 3). runx1 was used as a positive control. (B) CFU-C assay. Isolated E11.5 AGM cells were transfected with the miR-142-3p duplex or inhibitor. Five hours after the transfection, the c-Kit+CD34+ cells were sorted, and cultured in CFU-C media for 7 days, and several types of colonies formed, including BFU-E, CFU-GM, and CFU-Mix colonies. The mmu-miR-142-3p duplex promoted colony formation, whereas the mmu-miR-142-3p inhibitor repressed colony formation. The colonies starting from a seeding density of 800 cells were counted to determine the CFU-C number. Data are presented as mean ± SD; *P < 0.05. (C) CFU-S assay. The in vivo splenic colony assay showed that the CFU-S colony number was increased by ∼1.5-fold after the transplantation of duplex-treated cells. However, the number decreased by ∼3-fold after the transplantation of inhibitor-treated cells. Data are presented as mean ± SD; * P < 0.05. (D) T-cell differentiation was assessed using FACS. The dissociated cells from the AGM region were treated with the duplex or the inhibitor for 5 h. Then, the sorted c-Kit+CD34+ cells were cultured under T-cell differentiation conditions. The T-lymphoid potential was determined by measuring CD44 and CD25 expression using FACS after 13 days of co-culture with OP9-DL1 cells. The number of lymphoid progenitors increased after duplex treatment while it decreased after inhibitor treatment (duplex control 39 ± 7%, duplex 53 ± 6%; inhibitor control 45.5 ± 7%, inhibitor 38.6 ± 6%) (n = 3, P < 0.05). (E) The expression of irf7 was decreased in c-Kit+CD34+ cells by treatment with the miR-142-3p duplex, while it was increased by treatment with the inhibitor. A representative quantitative RT-PCR result showing the expression of irf7 mRNA in miR-142-3p duplex- and inhibitor-treated c-Kit+CD34+ cells from the E11.5 AGM. Data are presented as mean ± SD; *P < 0.05.

Full size image

To determine the functional role of miR-142-3p in the HSCs of mouse embryos, we first performed a colony-forming unit-cell (CFU-C) assay in vitro. E11.5 AGMs (41-45 sp) were dissociated using collagenase and the cells were treated with the miR-142-3p duplex (20 pmol) or the miR-142-3p inhibitor (40 pmol). Then, the c-Kit+CD34+ cells were sorted and the indicated numbers of single cells were transferred to ultra-low attachment 24-well plates and cultured in CFU-C media. The cells were cultured for 7 days, and based on morphology, the colonies were categorized into BFU-E (burst forming unit-erythroid), CFU-GM (colony forming unit-granulocyte, macrophage), and mixed CFU-GEMM (colony forming unit-granulocyte, erythroid, macrophage, megakaryocyte) or CFU-Mix. We counted the colonies in each category, and found that the addition of the miR-142-3p inhibitor led to a 25% reduction of the total number of colonies, whereas the miR-142-3p duplex caused an approximately 2-fold increase of total colony number (Figure 7B). The proportions of each colony category showed little changes after either treatment (Figure 7B). To confirm the in vitro result, we performed a CFU-S (colony-forming unit-spleen) assay in vivo. As shown in Figure 7C, the number of CFU-S was increased in mice infused with duplex-treated AGM cells, whereas it was decreased in mice infused with inhibitor-treated AGM cells. These data suggest that miR-142-3p is important for the expansion of HSCs and their differentiation into lineage-restricted progenitors, such as erythroid and myeloid lineages, both in vitro and in vivo.

To determine whether miR-142-3p is also required for the differentiation of the T-lymphoid lineage in mouse, we performed a T-cell differentiation assay in vitro35. Dissociated cells from the AGM region were treated with the duplex or inhibitor for 5 h. Then, the sorted c-Kit+CD34+ cells were co-cultured with OP9-DL1 stromal cells and cytokines for 13 days, and the CD25+CD44+ T-cell progenitors were quantified by fluorescence-activated cell sorting (FACS). The results showed that the duplex increased, whereas the inhibitor decreased, the number of CD25+CD44+ cells (Figure 7D), indicating that miR-142-3p is required for T-cell differentiation. To investigate whether the expression of irf7 is also negatively regulated by miR-142-3p in mouse, we performed qPCR to quantify the level of irf7 expression after up- or down-regulation of miR-142-3p expression. As shown in Figure 7E, the mRNA level of irf7 was decreased after miR-142-3p overexpression, whereas it was increased after the inhibition of miR-142-3p expression in c-Kit+CD34+ cells, which were dissociated from the E11.5 AGM and co-cultured with OP9 stromal cells for 48 h. Taken together, these results confirm the hypothesis that miR-142-3p is involved in the formation and differentiation of HSCs in the AGM of mouse embryos both in vitro and in vivo, suggesting a highly conserved role of miR-142-3p in definitive hematopoiesis in vertebrates, from zebrafish to mouse.

Discussion

Our study reveals that miR-142-3p plays pivotal roles in regulating the formation and lineage differentiation of the earliest HSCs in vivo during vertebrate embryogenesis. We showed that irf7 acts as a negative regulator of HSC development and differentiation into the T-cell lineages, which can be inhibited by miR-142-3p.

To date, very few miRNAs have been shown to play important roles in HSC development in vertebrates in vivo. miR-142-3p is specifically expressed in HSCs (AGM and CHT) and T cells (thymus) in zebrafish, mouse and other species15,16,17,18,19,20,21. By knockdown of miR-142a-3p using 2 specific MOs, the numbers of HSCs and T cells were decreased. Interestingly, a recent report using miR-142−/− mice also showed that more miR-142−/− CD4+ DCs underwent apoptosis than WT ones36. However, defects in HSC development were not examined in that study. Based on our present findings, we speculate that a similar phenotype would be observed at the onset of definitive hematopoiesis in miR-142−/− KO mice. Interestingly, a very recent report showed that in Xenopus, miR-142-3p is required for HSC lineage specification37, which is consistent with the results for zebrafish and mouse embryos reported in this study. In contrast, an inverse correlation between miR-142-3p expression and HSC development has been suggested recently20. Bissels et al.20 reported that overexpression of miR-142-3p at a dose of 50 nM in human CD133+ cells had a negative effect on their overall colony-forming ability. The apparent difference between those results and ours might be due to the source and age of the cells used (human bone marrow vs mouse AGM; 79 years vs E11.5) and the concentration of miR-142-3p utilized for overexpression (50 nM vs 20 pmol), which awaits further investigation.

Identifying the targets of miR-142-3p is crucial for analyzing the biological function of this miRNA during hematopoiesis. Among the four genes that we investigated, the expression of irf7 was the most upregulated in miR-142a-3p morphants. Reporter assays and qPCR further indicated that irf7 is a direct target of miR-142a-3p. More importantly, the irf7 MO injection rescued the defects in HSC development and T-cell differentiation in the miR-142a-3p morphants. Members of the interferon regulatory factor (IRF) family of transcription factors mainly act as regulators of host defense and inflammation27,28,29,30. Inflammatory activation of early HSCs has been shown to be evolutionarily conserved38, and a recent report by Hall et al.31 demonstrated that, in zebrafish, Gcsfr-NO signaling is required for the expansion of HSCs following infection. Among the irf family members, irf8 was shown to be required for myeloid differentiation into macrophages as opposed to neutrophils39 and for the regulation of B-cell lineage specification and differentiation40. Here, we reveal that knocking down miR-142a-3p expression in zebrafish upregulated the expression of an irf family member, irf7, which might be responsible for the decreased HSC population in the miR-142a-3p morphants. We further showed that irf7 is indeed a direct target of miR-142a-3p, which plays a pivotal role in regulating the formation and differentiation of HSCs in zebrafish in vivo and possibly also in mouse, at least in vitro. Further work is required to elucidate how irf7 negatively regulates Gcsfr-NO signaling during HSC development under normal and stressful conditions. However, we cannot exclude the possibility that other target genes of miR-142-3p may also be involved in this process.

In summary, our results demonstrate that miR-142-3p regulates HSC development by directly targeting irf7. As miR-142-3p and irf7 are highly conserved across species, the newly identified miR-142-3p-irf7 axis provides new insights into miRNA modulation of HSC formation and differentiation in vertebrates and may provide new clues for the in vitro expansion of therapeutic HSCs.

Materials and Methods

Fish strains and embryos

Zebrafish embryos were obtained by the natural spawning of adult AB strain zebrafish. The embryos were raised and maintained at 28.5 °C in system water and staged as described previously41. The fli1a:EGFP42 transgenic line was kindly provided by Steve Wilson (King's College London). The cloche carrier43, cmyb:EGFP26, and kdrl:mCherry/cmyb:EGFP4 lines were kindly provided by Anming Meng (Tsinghua University). This study was approved by the Ethical Review Committee of the Institute of Zoology, Chinese Academy of Sciences, China.

MOs, duplex, mRNA synthesis and microinjection

Standard MOs and gene-specific antisense MOs were purchased from GeneTools (Philomath, OR) and prepared as 1 mM stock solutions using ddH2O. The miRNA duplex was purchased from Invitrogen. mRNAs were synthesized using an Ambion mMessage mMachine kit (Ambion, Austin, TX). Standard MOs were used as negative controls in all of the MO injection experiments along with gene-specific MOs. For the fish embryo injections, MOs (2-10 ng), the duplex (5-20 pmol) and capped mRNA were injected separately or in combination into 1- to 2-cell-stage zebrafish embryos at the yolk/blastomere boundary. The MO sequences are listed in Supplementary information, Table S1.

WISH and cryosectioning

WISH of zebrafish embryos was performed as described previously34 using probes for myod, pax2.1, ccl25a, klf2a, ikaros, runx1, cmyb, dltc, flt4, foxn1, rag1, irf7, krt18, foxp1a, rxrab and gcsfr. The LNA probe for miR-142a-3p was purchased from EXIQON. The WISH-labeled embryos were treated with 30% sucrose and then washed three times using PBST and embedded in O.C.T. medium (SAKURA). The embryos were sectioned using LEICA CM1900 cryostats.

Stem loop RT-PCR, quantitative RT-PCR and western blotting

The total RNA of embryos from the one-cell stage to 6 dpf was reverse-transcribed using the miR-142a-3p RT-primer 5′-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACTCCATA-3′, and the U6 RT primer 5′-AAAAATATGGAGCGCTTCACG-3′. Quantitative RT-PCR and western blotting were performed as described previously34. The PCR primers used are listed in Supplementary information, Table S2. All of the experiments were performed in at least triple duplicates in three different biological experiments. The data are given as mean ± SD. Student's t-test was used for statistical comparisons. P < 0.05 was considered significant and is indicated with one asterisk.

Zebrafish embryo dissociation and FACS

Embryo dissociation and FACS were performed as described previously by Covassin et al.44. The GFP+ and GFP− cells were separately collected from cmyb:EGFP and runx1:EGFP transgenic embryos (Zhang P and Liu F, unpublished data) using MoFlo XDP (Beckman). A plasmid containing a runx1 intronic enhancer together with a heat-shock minimal promoter fused with EGFP was injected into embryos at the one-cell stage to generate a stable transgenic line.

Chemical treatment

Embryos at bud stage were treated with SNAP (Sigma-Aldrich) at the concentration of 10 μM. The treated embryos were examined periodically to ensure that the pharmacologic effect remained constant over time.

Confocal microscopy

Confocal images were acquired using a Zeiss LSM 510 META confocal laser microscope, and 3D projections were generated using Zeiss LSM software (Carl Zeiss Inc.).

Microarray assay

The total RNA was extracted from the AGM region of 2- and 4-dpf embryos using TRIzol (Tiangen, Beijing), and Cy3- or Cy5-labeled cDNAs from the controls and morphants were hybridized to Agilent Zebrafish Oligo Microarrays (Shanghai Biotechnology Co. Ltd., Shanghai). The expression data were subjected to hierarchical clustering and subsequently depicted in a heat map format.

Reporter assay

The WT and mutated irf7 3′ UTR were cloned and inserted into the EcoRI and SpeI sites of the pGL3-luciferase vector, which was kindly provided by Dr Peifeng Li (Institute of Zoology, CAS, Beijing, China). The mutation sites of irf7 3′ UTR were designed as follows: from WT 5′-AGTGG-3′ to mut1 5′-GACAA-3′ and from WT 5′-AACACTATA-3′ to mut2 5′-GGTGTCGCC-3′. WT or mutated pGL3-irf7-3′ UTR with or without the miR-142a-3p duplex (Invitrogen) were transfected into HEK293T cells using Lipofectamine 2000 (Invitrogen). The luciferase activity was measured using the luciferase activity assay kit (Promega).

DAF-FM assay

Embryos at 30 hpf were incubated with 5 μM DAF-FM diacetate (4-amino-5-methylamino-2′,7′-difluorofluorescein diacetate) for 2 h at 28.5 °C and then washed with fish water and observed under a fluorescence microscope.

CFU-C assay and quantitative RT-PCR

E11.5 mouse AGMs (41-45 sp) were dissociated using collagenase, and the cells were treated with the miR-142-3p duplex (Invitrogen) at 20 pmol and the inhibitor (Invitrogen) at 40 pmol. The procedures for siRNA transfection were performed as described in the Lipofectamine 2000 Transfection Reagent protocol (Invitrogen). After 5 h of transfection, antibody staining was performed for 30 min at 4 °C using antibodies specific for CD34 (BD) and c-Kit (eBioscience). The c-Kit+CD34+ cells were sorted using a MoFlo XDP (Beckman Coulter). The indicated numbers of single cells were transferred to ultra-low attachment 24-well plates (COSTAR) and cultured in CFU-C media. The cells were incubated at 37 oC in 5% CO2 with 100% humidity for 7 days, and the number of each type of colony was counted according to their morphology. The experiment was repeated in triplicate. The sequences for the miR-142-3p inhibitor and duplex are listed in Supplementary information, Table S2.

After transfection with the miR-142-3p duplex or inhibitor, the FACS-sorted c-Kit+CD34+ cells were cultured on mouse OP9 stromal cells in medium supplemented with hematopoietic cytokines (50 ng/ml SCF, 10 ng/ml IL3, 10 ng/ml FL, 10 ng/ml IL-6, and 3 U/ml Epo). After being cultured for 48 h, the cells were harvested for quantitative PCR to evaluate the expression of miR-142-3p and irf7.

CFU-S assay

The AGM regions of E11.5 mouse embryos were dissected, and the cells were suspended as single cells in solution, and then were treated with the miR-142-3p duplex (20 pmol) or inhibitor (40 pmol). After 5 h transfection, one embryo equivalent was transplanted via the tail vein. There were four individual mice in each experimental group. The spleens were obtained 8 days posttransplantation, and the number of colonies formed was counted under a stereomicroscope.

T lymphoid progenitor differentiation assay

An equal number of c-Kit+CD34+ cells and OP9-DL1 stromal cells were co-cultured in α-MEM supplemented with 15% FBS, 50 ng/ml SCF, 10 ng/ml IL7, and 20 ng/ml FL-3. After 13 days of co-culture, the cells were labeled with CD25-PE and CD44-APC antibodies and analyzed using flow cytometry.

Statistical analysis

For statistical analysis, Student's unpaired two-tailed t-test was used for all of the comparisons unless otherwise indicated.