Retinoic acid–inducible gene I (RIG-I) is an important cytosolic pattern recognition receptor crucial for sensing RNA virus infection and initiating innate immune responses. However, the participation of RIG-I in cellular development under physiological conditions remains limited. In this study, the regulatory role of RIG-I in embryonic hematopoiesis was explored in a zebrafish model. Results showed that rig-I was ubiquitously expressed during embryogenesis at 24 h postfertilization (hpf). A defect in RIG-I remarkably disrupted the emergence of primitive hematopoietic precursors and subsequent myeloid and erythroid lineages. In contrast, RIG-I deficiency did not have an influence on the generation of endothelial precursors and angiogenesis and the development of mesoderm and adjacent tissues. The alteration in these phenotypes was confirmed by whole-mount in situ hybridization with lineage-specific markers. In addition, immunostaining and TUNEL assays excluded the abnormal proliferation and apoptosis of hematopoietic precursors in RIG-I–deficient embryos. Mechanistically, RIG-I regulates primitive hematopoiesis through downstream IFN signaling pathways, as shown by the decline in ifnφ2 and ifnφ3 expression, along with rig-I knockdown, and rescue of the defects of hematopoietic precursors in RIG-I–defective embryos after administration with ifnφ2 and ifnφ3 mRNAs. Additionally, the defects of hematopoietic precursors in RIG-I morphants could be efficiently rescued by the wild-type RIG-I but could not be restored by the RNA-binding–defective RIG-I with site mutations at the RNA-binding pocket, which are essential for association with RNAs. This finding suggested that endogenous RNAs may serve as agonists to activate RIG-I–modulated primitive hematopoiesis. This study revealed the functional diversity of RIG-I under physiological conditions far beyond that previously known.
All vertebrate organisms experience sequential hematopoietic waves, which can be divided into primitive and definitive (1). The primitive wave, which forms erythroid and myeloid cells, can facilitate tissue oxygenation to satisfy the rapid growth of embryos (2). The definitive wave generates erythromyeloid progenitors and long-term hematopoietic stem cells (HSCs), which have the potential to generate all blood lineages, including erythrocytes, megakaryocytes, myelocytes (such as monocytes and granulocytes), and lymphocytes (3, 4). In mammals and avians, the transitory primitive wave occurs in the extraembryonic yolk sac blood islands and mainly generates erythroid progenitor cells, which are not pluripotent and do not have renewal capability. In most organisms, a transient wave of definitive hematopoiesis occurs in the blood islands and produces erythroid-myeloid progenitors (5, 6). Later definitive hematopoiesis with HSC specification begins in a region termed the aorta-gonad-mesonephros region. Within this region, HSCs specifically arise from the specialized hemogenic endothelium found in the ventral wall of the dorsal aorta (DA) in a process termed the endothelial-to-hematopoietic transition (EHT), which is conserved across vertebrates (7, 8). Once formed, the HSCs first migrate to the fetal liver for self-expansion. Eventually, induced by several signaling pathways and chemokines, these HSCs lodge in the bone marrow, which serves as a lifelong hematopoietic organ (9).
The process of blood development in zebrafish is similar to that in mammals, involving waves of hematopoiesis (10). The first wave contributes to primitive erythropoiesis and myelopoiesis, in which primitive erythrocytes and macrophages arise, and the second wave produces HSCs. In zebrafish, three germ layers, that is, the ectoderm, mesoderm, and endoderm, are generated during gastrulation and are then specified into different tissues (11). The mesoderm is specified into a dorsal fate, in which somites and the notochord arise, and a ventral fate, in which blood, the vasculature, and the pronephros arise (12–14). The primitive wave begins at two intraembryonic sites, that is, the anterior lateral mesoderm (ALM) and the posterior lateral mesoderm (PLM) (15, 16). Primitive myelocytes are mainly generated in the ALM, whereas primitive erythrocytes are generated in the PLM, forming the intermediate cell mass (ICM) (17, 18). The existence of hemangioblasts, bipotential progenitors of both blood and endothelial cells, has been suggested in the ALM and the PLM (19). The circulation of these primitive cells begins at ∼24 h postfertilization (hpf). Erythroid-myeloid progenitors arise from the posterior ICM region. A sophisticated network of many transcription factors has been described to regulate primitive and definitive hematopoiesis. Among them, the stem cell leukemia scl/tal1 (the basic helix-loop-helix transcription factor stem cell leukemia) is a central regulator of primitive hematopoiesis (20). Gata1 (GATA-binding factor 1), a zinc finger protein, is specifically required for the maturation of proerythroblasts, and pu.1/spi1, a transcription factor that contains an ETS domain, plays an indispensable role in primitive myelopoiesis (21–23). In addition, scl/tal1 can form complexes with lmo2, gata2, and other transcription factors in hematopoietic cells (24). By 24 hpf, the specification of HSCs in the zebrafish aorta-gonad-mesonephros region indicates the initiation of definitive hematopoiesis. Then, at ∼30 hpf, the earliest HSCs are generated from the ventral wall of the DA through the EHT (7, 8). During the EHT, accompanied by changes in cell morphology, the HSCs egress into the mesenchyme between the DA and the posterior cardinal vein and sequentially enter the vein and then the caudal hematopoietic tissue through blood circulation. After a short stay in the caudal hematopoietic tissue, the expanded HSCs continue their migration journey either to the thymus, where some HSCs differentiate into T lymphocytes by 3 dpf (25, 26), or to the kidney marrow by 4 dpf to sustain lifelong hematopoiesis, which is analogous to the bone marrow in mammals (27, 28). Despite these unique characteristics, the development of hematopoiesis in zebrafish shares highly conservative genetic programs that regulate hematopoiesis with other vertebrate animals (10).
Retinoic acid–inducible gene I (RIG-I) is largely accepted as a cytosolic pattern recognition receptor (PRR) with a tandem caspase recruitment domain and DExD/H helicase domain (29, 30). Traditionally, RIG-I is responsible for sensing viral RNAs and initiating the downstream signaling pathway through the interaction with mitochondrial antiviral signaling protein, ultimately inducing IFNs and proinflammatory cytokines to activate immune responses (31, 32). Intriguingly, emerging evidence showed that RIG-I participates in various other cellular activities, such as tumor suppression and therapy resistance, by recognizing self-RNA molecules except viral RNAs, such as endogenous retrovirus–derived RNAs, microRNAs, small nuclear RNA, mitochondrial RNA, RNA in exosomes, transcripts from short interspersed elements, and other repetitive elements, including transposable elements and satellite repeats (33–40). Recently, RIG-I–like receptors were also characterized as important regulators of the definitive wave of hematopoiesis and enhance HSC formation in zebrafish (40). These findings suggested that RIG-I displays more diversified functions under physiological conditions than previously recognized, which was far beyond the function of a PRR (41). In the current study, RIG-I defect disrupted the generation of hematopoietic precursors and myeloid and erythroid lineages. However, it had no effect on the generation of endothelial precursors and angiogenesis. Meanwhile, the development of mesoderm and adjacent tissues was not affected in RIG-I morphants. Mechanistically, RIG-I controls the emergence of hematopoietic precursors in zebrafish embryos through downstream IFN signaling. In addition, endogenous RNAs may serve as agonists to activate RIG-I, which subsequently performs its roles in primitive hematopoiesis. Thus, the findings indicated that RIG-I is critical for primitive hematopoiesis.
Materials and methods
Wild-type AB zebrafish (Danio rerio) of both sexes, weighing 0.5–1 g with 3–4 cm body length, were maintained in circulating water at 28°C under standard conditions as previously described (42). All fish samples were acclimatized and evaluated for their overall health for at least 2 wk prior to use in experiments. Only healthy fish, determined by their general appearance and level of activity, were used in the study.
Morpholino and mRNA injections
ATG-morpholino (MO) RIG-I MO (5′-GATTCTCCTTCTCCAGCTCGTACAT-3′) and standard control MO (5′-CCTCTTACCTCAGTTACAATTTATA-3′) were chemically synthesized (Gene Tools). Eight nucleotides within the MO target site were changed without altering the encoding amino acids to generate RIG-I MO–resistant mRNA. MO-resistant RIG-I mRNAs were synthesized from the HindIII-linearized pcDNA3.1 constructs by using the mMESSAGE mMACHINE kit (Ambion) in accordance with the manufacturer’s instructions. RIG-I MO and control MO (1.6 ng) and RIG-I mRNA (150 pg), IFNφ2 mRNA (150 pg), and IFNφ3 mRNA (150 pg) were injected into each embryo.
Generation of RIG-I knockout zebrafish line
The single-guide RNA (sgRNA) for CRISPR-Cas9–based gene knockout was designed with a Web site program (CHOPCHOP, http://chopchop.cbu.uib.no/) (43). The target sequence of the third rig-I exon is 5′-GGAGTCTGAGGAGATTCAAGCGG-3′. The sgRNA was obtained by using the MEGAscript T7 high-yield transcription kit (Invitrogen) and MEGAclear kit (Invitrogen). A sgRNA mix for rig-I (500 ng/μl TrueCut Cas9 protein v2 and 250 ng/μl sgRNA) was injected into one-cell-stage embryos. The rig-I−/− zebrafish homozygote was acquired after two generations (44, 45). The following primers were used to detect the deletion of rig-I: forward, 5′-GTGTCACAATGCCTGTAATG-3′; reverse, 5′-GCCAGGTCACCCTTGTAAAT-3′.
Whole-mount in situ hybridization staining
Plasmids for probe synthesis were amplified and subcloned into the pBlueScript II SK vector. Digoxigenin-labeled probes were transcribed from linearized templates in the pBlueScript vector with the SP6/T7 promoter in accordance with the manufacturer’s protocol (Roche). Whole-mount in situ hybridization (WISH) was performed as described previously (46). Anti-digoxigenin Ab (1:5000; Roche) was added, and NBT–5-bromo-4-chloro-3-indolyl phosphate staining was performed to detect the signal in accordance with the manufacturer’s instructions (Promega). Finally, the embryos were visualized and imaged on an Olympus stereoscope (MVX10 MacroView).
Qualitative scoring of WISH staining was assessed by the number of embryos with altered primitive hematopoietic precursors per total number of embryos consisting of the sample for each case, which was conducted manually by visual observation and blindly as previously described (40). The individual observer categorized as “decreased” the embryos depicting a staining lower than the staining of most of the control embryos, and as “normal” the embryos showing a staining approximately equal to the staining of most of the control embryos. Then, the individual who categorized the phenotypes into normal and decreased turned the numbers of embryos into percentages (number of embryos depicting the particular staining intensity/total number of embryos examined in the experimental sample × 100%), and the comparisons made between samples were indicated by horizontal bars on each graph in the figures. On stacked columns including normal or decreased intensity groups, different groups were compared between samples each time, depending on the scientific question addressed.
Sudan Black B staining
Sudan Black B (SBB) staining was used to detect the neutrophil granules in neutrophils. Embryos (36 hpf) were fixed with 4% methanol-free formaldehyde (Polysciences, Warrington, PA) in PBS for 2 h at room temperature. Afterwards, they were rinsed in PBS, stained with SBB (Sigma-Aldrich, Saint-Quentin-Fallavier, France; 2.1 mg/ml, prepared with 73% ethanol and 3.5% phenol in 1.9 mM phosphate buffer) for 20 min, washed extensively in 70% ethanol in water, and then progressively rehydrated to PBS with 0.1% Tween 20 (PBST). Finally, the embryos were transferred to 100% glycerinum and examined by differential interference contrast microscopy (Carl Zeiss Axiovert 40 CFL).
O-Dianisidine staining was used to detect the expression of hemoglobin in RBCs. Embryos (36 hpf) were stained for 15 min in the dark in o-dianisidine (0.6 mg/ml), 10 mM sodium acetate (pH 4.5), 0.65% H2O2, and 40% (v/v) ethanol. The stained embryos were cleared with benzyl benzoate/benzyl alcohol (2:1, v/v) and examined by differential interference contrast microscopy (Carl Zeiss Axiovert 40 CFL).
TUNEL assay and Ab immunostaining
A TUNEL assay was carried out using a one-step TUNEL apoptosis assay kit (Beyotime) in accordance with the manufacturer’s instructions. Whole-mount Ab staining was performed as previously described (47). For the permeability of the whole-mount embryos, 24-hpf embryos were treated with 0.05% collagenase at 37°C for 5 min and then fixed in 4% PFA for 20 min at room temperature. The embryos were incubated with Ab against phospho–histone 3 (pH3, 1:500; Beyotime) and then with Alexa fluorescent-conjugated secondary Ab (1:500; Life Technology) diluted in a blocking solution at 4°C overnight. After the embryos were washed with PBST, they were prepared for mounting and imaging. Images were captured on an Olympus stereoscope (MVX10 MacroView).
The open reading frames of zebrafish RIG-I, IFNφ2 and IFNφ3, were inserted into pcDNA3.1/FLAG-His (Invitrogen) between NotI/SfaAI to construct eukaryotic expression vectors, designated as pcDNA3.1–RIG-I, pcDNA3.1-IFNφ2, and pcDNA3.1-IFNφ3. The RNA-binding mutant RIG-I with mutations of F859A, K864A, and K867A was also inserted into pcDNA3.1 between NotI/SfaAI and named pcDNA3.1–RIG-I mutant. The coding sequence for the full-length RIG-I protein was inserted into pEGFP-C1 (Clontech) to construct the eukaryotic expression vector for EGFP-fused RIG-I between EcoRI/KpnI and named pEGFP–RIG-I. All primers used in plasmid construction are shown in Supplemental Table I. All constructed sequences were confirmed by sequencing analysis. The plasmids for transfection and microinjection were prepared endotoxin-free using an EZNA plasmid mini kit (Omega Bio-Tek).
Luciferase reporter assay
Epithelioma papulosum cyprini cells were seeded in 24-well plates overnight and 20 ng of plasmids (pCMV-empty, pCMV–RIG-I wild-type (WT), pCMV–RIG-I mutant), as well as IFNφ3-pro (250 ng/ml) and pRL-TK Renilla luciferase internal control (25 ng/ml) were cotransfected. Transfection of poly(I:C) (0.1 μg/ml) was performed at 24 h before cell harvest. At 48 h posttransfection, the cells were washed three times with PBS and lysed for measuring luciferase activity by the Dual-Luciferase reporter assay system (Promega) according to the manufacturer’s instructions. The relative luciferase activity unit was determined as the ratio of firefly luciferase activity divided by the Renilla luciferase activity as previously described (48). Each trial was repeated at least three times.
Expression analysis of hematopoietic lineage markers
The expression patterns of hematopoietic lineage markers were examined in embryos at various developmental stages (10 or 24 hpf) by reverse transcription–quantitative PCR (RT-qPCR). Total RNA was isolated from embryos and tissues and reverse transcribed into cDNA, as described earlier (47). RT-qPCR was performed on a Mastercycler ep realplex PCR system using a SYBR Premix Ex Taq kit (Takara Bio) in accordance with the manufacturer’s instructions. The protocol consisted of the following: 1) 40 cycles of amplification at 95°C for 30 s and 60°C for 20 s; 2) melting curve analysis at 95°C for 5 s, 65°C for 15 s, and 95°C for 15 s; and 3) cooling at 40°C for 30 s. Relative gene expression was calculated using the 2−ΔΔCt method, with RIG-I initially normalized against β-actin. Each PCR trial was performed in triplicate and repeated at least three times.
Data from the three independent experiments were expressed as mean ± SD. The groups were compared statistically using Student t test for paired samples. The p values <0.05 were considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.001).
Deficiency of RIG-I impairs primitive hematopoiesis
Similar to that in mammals, zebrafish hematopoiesis consists of two successive waves, primitive (Fig. 1A) and definitive hematopoiesis. First, the spatiotemporal expression pattern of rig-I during the embryonic development was detected via WISH to explore whether RIG-I is involved in embryogenesis. The results showed that rig-I was expressed ubiquitously in all blastomeres before the midblastula transition, indicating that rig-I was a maternal mRNA (Supplemental Fig. 1). Additionally, RIG-I displayed a relatively high expression, with the tendency to accumulate over time during the early stage (24 hpf) of embryogenesis (Supplemental Fig. 1A–J). Thus, RIG-I may play some unexpected roles in regulating the development of zebrafish embryo. For evidence on this notion, knockdown of rig-I expression was performed by MO targeting its ATG region (RIG-I MO), whose effectiveness was ascertained by injecting a RIG-I–GFP chimeric plasmid containing a RIG-I MO–targeted sequence into one-cell-stage embryos with or without RIG-I MO (Supplemental Fig. 1Q, 1R). The WISH results showed that the myeloid lineage marker pu.1/spi1 in the ALM of RIG-I MO–injected embryos was significantly reduced compared with that in the control MO–injected embryos at 12 and 24 hpf (Fig. 1B–E, 1N). Furthermore, the macrophage marker lcp1/l-plastin at 24 hpf (Fig. 1F, 1G, 1O) and heterophil granulocyte marker mpo/mpx at 48 hpf (Fig. 1H–K, 1P) displayed a remarkable reduction. Meanwhile, the loss of neutrophil granulocytes was supported by SBB staining at 36 hpf (Fig. 1L, 1M, 1Q). The erythroid lineage marker gata-1 in the PLM at 18 hpf was diminished in the RIG-I MO embryos (Fig. 2A–F, 2I), accompanied with an extensive decrease in RBCs at 36 hpf, as shown in o-dianisidine staining (Fig. 2G, 2H, 2J). These data suggested that the disruption of RIG-I led to defects in primitive hematopoiesis in zebrafish embryos. Given the common origin of hematopoietic and endothelial precursors, the transcriptional expression of the endothelial-specific transcription factor fli-1 and the endothelial marker flk-1 was further examined to investigate the effects of RIG-I on the generation of endothelial precursors and angiogenesis. Conversely, the expression levels of fli-1 at 12 hpf and flk-1 at 24 hpf were comparable between RIG-I morphants and control siblings (Supplemental Fig. 2A–D). Altogether, these results demonstrated that RIG-I was required for the development of primitive hematopoietic cells, but it had no effect on the generation of endothelial precursors and angiogenesis.
RIG-I is required for specification of earliest hematopoietic precursors
On the basis of the effects of RIG-I MO on erythropoiesis and myelopoiesis, the regulatory role of RIG-I in the expression of genes that modulate the emergence of hemangioblasts/hematopoietic precursors and initiate primitive hematopoiesis was further examined. First, the expression levels of ntl, bmp4, fgf8, ndr1, and wnt3a, which are all essential for the signaling involved in mesoderm development, were detected by RT-qPCR. Minimal changes in the expression of these genes were found after RIG-I MO injection (Supplemental Fig. 2E). Additionally, the expression of bmp4, which is required for the development of ventral mesoderm that gives rise to primitive hemangioblasts/hematopoietic precursors, and that of myod, pax2.1, and nkx2.1, which are paraxial mesoderm/myogenic, pronephric duct, and cardiac markers, were evaluated using WISH. Similarly, these genes were not affected in RIG-I morphants (Supplemental Fig. 2F–M). In contrast to the effect on the mesoderm, the expression of the earliest hematopoietic marker scl/tal1 (the basic helix-loop-helix transcription factor stem cell leukemia), which is capable of converting mesoderm into hemangioblasts/hematopoietic precursors in zebrafish embryos, extensively decreased in RIG-I morphants at the six-somite stage (Fig. 2K–P, 2W). Furthermore, the expression of lmo2 (the LIM domain only 2 transcription factor binding to scl/tal1), a lineage-specific transcription factor expressed in hematopoietic and endothelial precursors, significantly diminished in RIG-I morphants at the six-somite stage (Fig. 2Q–V, 2X). Consistent with the WISH results, RT-qPCR analysis showed that the expression levels of scl/tal1, lmo2, gata-1, and pu.1/spi1 were remarkably reduced after rig-I knockdown (Fig. 2Y, 2Z). The embryos with decreased expression levels of pu.1/spi1, gata-1, scl/tal1, and lmo2 accounted for 90, 80, 82, and 79% of all counted embryos in RIG-I morphants, respectively (Figs. 1N, 2I, 2W, 2X). Considering that early scl-positive cells could contribute to hematopoietic lineages and endothelial precursors but the generation of endothelial precursors was not affected by rig-I knockdown according to fli-1 WISH results, these observations indicated that RIG-I is essential for the specification of the earliest hematopoietic precursors. For further confirmation, rig-I mRNA (150 pg) and RIG-I MO (1.6 ng) were coinjected into the embryos to compare the expression of pu.1 and gata-1 with that of embryos only injected with RIG-I MO. In this case, we used an optimized dose of rig-I mRNA (150 pg) that did not have an impact on the phenotype of WT embryos (Supplemental Fig. 3A, 3B). Consequently, coinjection of rig-I mRNA was able to restore the expression of pu.1/spi1 and gata-1 at 12 hpf, lcp1/l-plastin at 24 hpf, and mpo/mpx at 36 hpf to normal levels, indicating that RIG-I MO specifically targeted rig-I, which plays a pivotal role in primitive hematopoiesis, especially the generation of hematopoietic cells (Fig. 3). Next, we verified the requirement of RIG-I for specification of primitive hematopoietic precursors. For this procedure, rig-I mRNA (150 pg) and RIG-I MO were injected at the one-cell stage, and the expression of scl/tal1 and lmo2 during early somitogenesis was examined. The results revealed that the expression of both in the PLM in rig-I mRNA-injected embryos was expanded in comparison with that in rig-I MO-injected embryos to an extent comparable with their expression levels in control morphants (Fig. 4). Furthermore, a gRNA targeting exon3 of rig-I was constructed to generate a RIG-I mutant, resulting in transcripts with a nonsense codon (Fig. 5A, 5B). The rig-I−/− embryos showed a significant impairment of primitive hematopoiesis, as evidenced by the decrease in the expression of gata-1, pu.1/spi1, and scl/tal1 in mutant embryos during early somitogenesis compared with that in WT embryos (Fig. 5C–K). These data demonstrated that the phenotype observed in rig-I mutants was indeed RIG-I–dependent.
RIG-I regulates primitive hematopoiesis without intervening cell proliferation and apoptosis
The loss of hematopoietic cells in RIG-I morphants could be attributed to three mechanisms, including extensive apoptosis of hematopoietic cells, aberrant proliferation of hematopoietic cells, and lack of hematopoietic precursors at the very beginning of primitive hematopoiesis. To investigate whether the defects of primitive hematopoiesis in RIG-I–deficient embryos were due to abnormal cell proliferation or apoptosis of hematopoietic precursors, we exploited pH3 immunostaining, which marks condensed chromosomes in metaphase and early anaphase cells using a pH3 Ab and TUNEL assay to examine DNA fragmentations in apoptotic cells, respectively. The immunostaining results revealed that the number of pH3-positive cells in the ICM of RIG-I MO–injected embryos was comparable to that in control MO–injected embryos (Fig. 6A, 6B, 6E). Similarly, no significant differences in apoptotic cell numbers in the ICM of embryos were observed between the control siblings and RIG-I–deficient embryos (Fig. 6C, 6D, 6F). Taken together, these data suggested that RIG-I functions in the specification of primitive hematopoietic precursors rather than in their survival or proliferation.
RIG-I–mediated IFN signaling pathway is implicated in primitive hematopoiesis
RIG-I has been reported to participate in several different signaling pathways. In hepatocellular carcinoma and acute myeloid leukemia, RIG-I acts as a tumor suppressor through either augmenting STAT1 activation by competitively binding STAT1 against its negative regulator SHP1 or inhibiting the AKT–mTOR signaling pathway by directly interacting with Src (49, 50). In BrCa cells, RIG-I triggers the STAT1-dependent signaling pathway through the stimulation of endogenous RNAs in the exosomes released by stromal cells. Subsequently, the activated STAT1 facilitates the transcription of NOTCH3 downstream target genes to help with the therapy resistance and expansion of breast cancer cells (38). Most importantly, RIG-I is widely known to initiate IFN signaling pathways as a cytosolic PRR by detecting exogenous viral RNAs (51). Thus, to elucidate the mechanism underlying RIG-I regulation of hematopoietic precursor development, the expression changes in key molecules involved in the above signaling pathways after rig-I knockdown were examined by RT-qPCR. A significant reduction in the expression of ifnφ2 and ifnφ3 (group II of type I IFNs in zebrafish) was observed in RIG-I morphants, whereas the expression of ifnφ1 and ifnφ4 (group I of type I IFNs in zebrafish) was unchanged in RIG-I MO–injected embryos (Fig. 7A, 7C). However, the expression levels of STAT1a, STAT1b, and HEY1 (NOTCH3 downstream target genes) in RIG-I MO embryos did not significantly differ compared with those in control embryos (Fig. 7B, 7D). The PxxP motif used to interact with Src in mammals was not conserved in zebrafish according to bioinformatics analysis. Taken together, these results indicated that RIG-I possibly exerts its regulation of primitive hematopoiesis through the downstream IFN signaling pathway. Next, whether overexpression of ifnφ2 and ifnφ3 mRNAs could rescue the defects of the hematopoietic precursors in RIG-I morphants was investigated. We first identified an optimal dose of ifnφ2 and ifnφ3 mRNAs (150 pg) that did not have an impact on the phenotype of WT embryos (Supplemental Fig. 3C–F). WISH results showed that the overexpression of ifnφ2 and ifnφ3 mRNAs with RIG-I MO efficiently rescued the number of scl/tal1-positive hematopoietic precursors in the ALM and PLM regions in RIG-I–defective embryos (Fig. 7E–M). The embryos with normal expression of scl/tal1 in the control, RIG-I morphants, ifnφ2 mRNA rescue, and ifnφ3 mRNA rescue accounted for 90, 17, 78, and 67% of all the counted embryos, respectively (Fig. 7N). Similar results were verified in the lmo2-positive hematopoietic precursors in rig-I−/− embryos (Fig. 5L–X). Rescue experiments indicated that the overexpression of ifnφ2/3 considerably rescued the defect of hematopoietic precursors in RIG-I morphants. Collectively, RIG-I controls the emergence of primitive hematopoietic precursors in zebrafish embryos through the downstream IFN signaling pathway.
RNA binding activity of RIG-I is essential for hematopoietic cell development
RIG-I is known to be held in a self-closed conformation in resting state without RNA-ligand stimulation, thereby making it unavailable for the initiation of downstream IFN signaling. However, emerging new findings demonstrated that RIG-I is closely associated with various endogenous RNAs in many important cellular activities (34, 41, 52). The results showed that RIG-I regulated the generation of hematopoietic precursors during embryogenesis through the activation of downstream IFN signaling. Thus, some endogenous RNAs may serve as ligands of RIG-I to initiate the IFN signaling pathways and modulate primitive hematopoiesis. To support this hypothesis, we assessed the significance of the RNA binding ability of zebrafish RIG-I in regulating primitive hematopoiesis. Zebrafish RIG-I shares high amino acid sequence identity and conserved functional domains and motifs with those of human RIG-I homolog (53). In particular, three functional amino acid residues (F853, K858, and K861 in humans, which correspond to F859, K864, and K867 in zebrafish) are completely conserved from fish to humans and other mammalian species (38, 54, 55). These amino acid residues play an essential role in the interaction of RIG-I with RNAs. Thus, an RNA-binding–defective RIG-I with mutations of F859A, K864A, and K867A in the C-terminal repressor domain was constructed to investigate whether the overexpression of this mutant mRNA could rescue the defects of the hematopoietic precursors in RIG-I morphants (Fig. 8A). As expected, this RNA-binding–defective RIG-I failed to activate the IFN-inducing pathway, as shown by its significantly decline in triggering IFNφ3-pro luciferase reporter activity upon poly(I:C) stimulation (Fig. 8B). Correspondingly, overexpression of WT rig-I, not the RNA-binding–deficient mutant, restored the expression of hematopoietic lineage markers to the normal level (Fig. 8C–L). This finding implied that the RNA-binding pocket with F859, K864, and K867 residues is the key for RIG-I to exert its regulatory functions in hematopoiesis, indicating that endogenous RNAs may serve as agonists to activate the RIG-I–mediated IFN-signaling pathway that contributes to the primitive hematopoiesis of zebrafish embryos.
During the past two decades, RIG-I has been well documented to serve as an important cytosolic PRR for detecting viral RNAs and inducing IFN signaling and other inflammatory responses in innate antiviral immunity (56–60). However, this long-held paradigm has been challenged because emerging new investigations have shown that the biological function of RIG-I is broader and more complicated than previously recognized (50, 61–63). For example, the recognition ligands of RIG-I have been considerably extended from foreign viral RNAs to endogenous cellular RNAs, including various microRNAs, small nuclear RNAs, and transcriptional RNAs from endogenous retroviruses and repetitive elements (34–36, 38, 40). These findings suggested the potential involvement of RIG-I in many other physiological and/or pathological processes through association with cellular self-RNAs, in addition to its functional role in innate immunity. Consistent with this notion, RIG-I has exhibited its extensive participation in cellular development and oncogenesis of multiple cancers, such as acute myeloid leukemia, acute promyelocytic leukemia, and hepatocellular carcinoma (34, 35, 49, 50, 64). In these cases, RIG-I augmented STAT1 activation to upregulate the expression of numerous IFN stimulatory genes, ultimately promoting retinoic acid– and/or IFN-mediated growth inhibition and differentiation of acute myeloid leukemia cells (49, 50). Similarly, RIG-I competitively bound STAT1 against its negative regulator SHP1 to amplify the antitumor effect of IFN-α on hepatocellular carcinoma (64). The regulatory functions of RIG-I in cancers implied that this protein may play key roles in monitoring cellular growth and differentiation. Actually, RIG-I was initially identified as an all-trans retinoic acid–modulated RNA helicase protein in an acute promyelocytic leukemia cell line NB4 (65). Subsequently, RIG-I was found to be an essential regulator for myeloid development, in which its expression was developmentally upregulated along with myelopoiesis (61). RIG-I defect in mice disrupted physiological myelopoiesis, especially granulopoiesis, resulting in a progressive myeloproliferative disorder (61). Additionally, a rig-I–targeting procedure led most rig-I−/− mice to die before birth. In this case, massive cell death was observed in the embryonic hepatic tissue (66). These observations partially supported the notion that RIG-I may harbor an intrinsic role in regulating cellular development and survival in the absence of viral infection.
In this study, insights into the role of RIG-I in embryonic hematopoiesis were provided in a zebrafish model. The spatiotemporal expression pattern of RIG-I during embryonic development was initially examined using WISH. The results showed that rig-I mRNAs were detected ubiquitously in all blastomeres before the midblastula transition, indicating that rig-I mRNAs may be maternally transferred into embryos at the early stage. Additionally, rig-I displayed a relatively high expression with the tendency to accumulate over time during the period of 24-hpf embryogenesis. These data implied that RIG-I may play some unexpected roles in regulating zebrafish embryonic development. In addition, the knockdown of rig-I significantly reduced embryonic hematopoiesis at stem and lineage-committed cell levels. Disruption of RIG-I hinders the generation of hematopoietic precursors and the development of myeloid and erythroid lineages. However, the generation of endothelial precursors and angiogenesis were unaffected after the knockdown of rig-I. Similarly, the development of mesoderm and adjacent tissues in RIG-I morphants was comparable to that of control morphants. These findings implied that RIG-I may preferentially regulate the development of primitive hematopoiesis. Thus, the involvement of RIG-I in embryonic hematopoiesis was further explored. WISH analysis demonstrated that the lineage-specific markers, including the myeloid-lineage marker pu.1/spi1, the macrophage marker lcp1/l-plastin, and the heterophil-granulocyte marker mpo/mpx, were significantly reduced in RIG-I morphant embryos compared with those in the control embryos. Meanwhile, a remarkable decrease in granulocytes was reconfirmed by SBB staining. In addition, the erythroid lineage marker gata-1 displayed a remarkable reduction, which was accompanied with loss of RBCs in the RIG-I morphants, as examined by o-dianisidine staining. These data suggested that RIG-I is essential for primitive hematopoiesis, including the development of myeloid and erythroid lineages in zebrafish embryos. However, the expression of fli-1 and flk-1 in RIG-I–deficient embryos did not significantly change, suggesting that RIG-I had no effect on the generation of endothelial precursors and angiogenesis. According to the RT-qPCR and WISH results of key genes involved in the development of mesoderm and tissues adjacent to hematopoietic regions, the mesoderm development was preserved intact in RIG-I morphants. Intriguingly, the expression of the earliest hematopoietic markers scl/tal1 and lmo2 decreased extensively in the RIG-I morphant embryos. Early scl/tal1-positive cells could give rise to hematopoietic lineages and endothelial precursors (20, 67). The results showed that the generation of endothelial precursors was not affected by knockdown of rig-I, whereas the development of myeloid and erythroid lineages was disrupted. Altogether, these findings indicated that RIG-I is essential for the specification of the earliest hematopoietic precursors, which differentiate into myeloid and erythroid lineages.
Finally, pH3 immunostaining and TUNEL assays were performed to evaluate whether the observed deficiency of RIG-I–mediated phenotypes was due to the abnormal cell proliferation or apoptosis of hematopoietic precursors. The results revealed that RIG-I deficiency had no effect on the apoptosis and proliferation of hematopoietic cells. Moreover, the expression of ifnφ2 and ifnφ3 in zebrafish reduced significantly with the knockdown of rig-I, implying that IFNφ2 and IFNφ3 may be the downstream effectors of RIG-I in hematopoiesis. Further investigations proved that the overexpression of ifnφ2 and ifnφ3 mRNAs efficiently rescued the defects of hematopoietic precursors in the ALM and PLM regions in RIG-I–defective embryos. Therefore, all of these results indicated that RIG-I controls the emergence of hematopoietic precursors in zebrafish embryos through the downstream IFN-signaling pathways. In teleost fish, the type I IFNs possess two or four cysteine-containing subgroups, namely group I type I IFNs (IFNφ1 and IFNφ4) and group II type I IFNs (IFNφ2 and IFNφ3), respectively. These two groups of IFNs activate downstream signaling through two different receptor complexes (CRFB1/CRFB5 for group I IFNs and CRFB2/CRFB5 for group II IFNs), which is different from that of mammalian type I IFNs that stimulate downstream signaling through a common receptor complex (IFNAR1/IFNAR2) (68). The use of two receptors suggests that the functional behavior and regulation of the cellular activities by fish type I IFNs may be more complicated than in mammals; in addition, the two groups of fish IFNs may function differently. It has been reported that RIG-I specifically induces zebrafish group II, but not group I, type I IFNs under viral RNA stimulation, which implies that the two groups of IFNs are induced via different RNA ligands differentially associated with diverse PRR signaling pathways, such as RIG-I, MDA5, and TLR7 pathways. In this case, RIG-I was supposed to be more related to the regulation of zebrafish group II IFNs (69). Our results showing the preferential regulation of RIG-I for IFNφ2 and IFNφ3 expression was consistent with that observed in previous investigations; however, the detailed mechanism underlying this process remains to be further explored. Additionally, in the resting state without RNA ligands, RIG-I is held in a “closed” conformation by autoregulatory intramolecular interactions with the C-terminal repressor domain within the C-terminal domain that holds the caspase recruitment domains unavailable for signaling. Consequently, RIG-I is unable to activate downstream signaling and induce the expression of IFNs (70). Given the emerging findings about endogenous ligands of RIG-I, some endogenous RNAs may function as ligands to activate RIG-I and thus trigger RIG-I–IFN-mediated hematopoietic regulation. An RNA-binding–defective RIG-I with mutation at sites of F859, K864, and K867, which are essential for association with RNAs, was constructed in the RNA-binding pocket. In contrast to WT rig-I, the overexpression of this RNA-binding–defective rig-I was unable to restore the expression of hematopoietic lineage markers to the normal level. This finding implied that the RNA-binding pocket is crucial for RIG-I to exert its regulatory functions in hematopoiesis. Certain endogenous RNAs served as agonists to activate RIG-I and its downstream IFN signaling, which is essential for the embryonic primitive hematopoiesis of zebrafish. Recently, Lefkopoulos et al. (40) reported that repetitive elements transcribed during zebrafish embryonic development drive RLR-mediated inflammatory signals that regulate hematopoietic stem and progenitor cell formation during the definitive wave of hematopoiesis, and the function of RLRs in developmental hematopoiesis is conserved in mammals. However, they demonstrated the lack of a phenotype in primitive hematopoiesis by flow cytometry for mpx:GFP-, gata1:GFP-, and mpeg1:GFP-positive cells in control and RIG-I morphant embryos, which might not be sensitive enough to assess regional differences in expression (40). Hence, by using WISH, our present study can identify more subtle effects of RIG-I knockdown on primitive hematopoiesis. All of these findings largely supported the hypothesis that RIG-I plays intrinsic roles in regulating cellular development activities, including developmental hematopoiesis, with the activation of endogenous RNAs and the IFN signaling pathways. Further study is needed to clarify which types of endogenous RNAs are responsible for RIG-I–regulated primitive hematopoiesis and how these RNAs are transcribed upon hematopoietic cell fate change under epigenetic regulation. Overall, our present study highlights a new functional role of RIG-I in primitive hematopoiesis by sensing endogenous cellular RNAs. Given that zebrafish embryonic hematopoiesis is highly homologous to that of mammalian species, this finding makes zebrafish an integral part for understanding the mechanisms of vertebrate primitive hematopoietic regulation and disorders. It would also be beneficial in mapping the molecular evolutionary history of primitive hematopoiesis from fish to mammals as a whole.
This work was supported by National Key Research and Development Program of China Grants 2018YFD0900503 and 2018YFD0900505, National Natural Science Foundation of China Grant 31630083, and by the Open Funding Project of the State Key Laboratory of Bioreactor Engineering.
The online version of this article contains supplemental material.
Abbreviations used in this article
anterior lateral mesoderm
hematopoietic stem cell
intermediate cell mass
posterior lateral mesoderm
pattern recognition receptor
retinoic acid–inducible gene I
reverse transcription–quantitative PCR
Sudan Black B
whole-mount in situ hybridization
The authors have no financial conflicts of interest.