Inflammatory signaling regulates hematopoietic stem and progenitor cell emergence in vertebrates

Hematopoietic stem cells (HSCs) have the abilities of self-renewal and multilineage differentiation to maintain the supply of all mature blood cells for the lifetime. In vertebrate embryos, the earliest definitive hematopoietic stem and progenitor cells (HSPCs) are originated from a group of specialized endothelial cells, hemogenic endothelium (HE),^1-4  through the endothelial-to-hematopoietic transition (EHT) process.^5-7  In vitro generation and/or expansion of functional and transplantable bona fide HSCs with self-renewal and multilineage potential hold great promise in regenerative medicine. However, this has not been achieved yet, at least partially because a complex interplay between cell-extrinsic cues and cell-intrinsic regulatory signaling governing HE formation and HSPC specification remains incompletely understood.

Inflammatory signaling is generally thought to be active in the presence of infection and inflammation. In response to infection stimuli, innate immune cell–secreted inflammatory cytokines directly or indirectly act on myeloid precursors to differentiate and expand to meet the acute demand of effector myeloid cells. However, recent reports also showed that inflammatory cytokine can also directly act on HSPCs to instruct their fate, either increased proliferation or directed differentiation toward myeloid⁸  or lymphoid⁹  lineages, at the expense of HSC self-renewal and/or maintenance in the bone marrow.^10,11  Upon Toll-like receptor (TLR) stimulation, nuclear factor κ-light-chain enhancer of activated B (NF-κB), a central player in innate immune response, integrates multiple inflammatory signaling pathways including tumor necrosis factor α (TNFα) and granulocyte colony-stimulating factor (G-CSF) pathways. NF-κB has been shown to regulate both innate and adaptive immune systems including myeloid cells and B and T cells in homeostasis and differentiation. Interestingly, the role of inflammatory signals during HSPC emergence in both zebrafish and mouse embryos has been described very recently.^12-14 

Inflammatory signaling pathways are well conserved in vertebrates. Previous studies in mice showed that cytokines including interleukin-1 (IL-1) and IL-3 are involved in the embryonic development of HSPCs in the aorta-gonad-mesonephros (AGM) region.^15,16  We previously reported that a hematopoietic microRNA, miR-142a-3p, regulates the formation and differentiation of HSPCs by regulating the inflammatory signaling cascade irf7-gcsfr,¹⁷  indicating a possible role of inflammatory signaling in HSPC emergence. In this study, we demonstrated that in both zebrafish and mouse embryos, inflammatory signaling is required for HSPC specification. Ablation of the TLR4–myeloid differentiation primary response 88 (MyD88) pathway and inflammatory cytokine signaling pathways including TNFα and G-CSF, decreased the population of the earliest HSPCs through downregulation of their core effector NF-κB, which in turn reduced the Notch activity in endothelial cells. Attenuation of Notch activity in endothelial cells then compromised the HE and the following HSPC emergence via the EHT process. Therefore, we discovered, for the first time, an unexpected essential role of proinflammatory signaling in HSPC emergence through a conserved NF-κB–Notch pathway.

Zebrafish strains including Tubingen, cmyb:green fluorescent protein (GFP),¹⁸ kdrl:mCherry,⁵ fli1a:enhanced GFP (EGFP),¹⁹  heat shock protein (HSP)-il1b-EGFP,²⁰ CD41:GFP,²¹ mpo:GFP,²² lyz:DsRed,²³ myd88 mutants,²⁴ tp1:DsRed,²⁵  and runx1:en-GFP (P.Z. and F.L., manuscript in preparation) were raised and maintained in system water at 28.5°C. The HSP-il1b-EGFP embryos were heat-shocked from 20 hours post fertilization (hpf) for 30 minutes at 42°C. The embryos were obtained by natural spawning. This study was approved by the Ethical Review Committee in the Institute of Zoology, Chinese Academy of Sciences, China.

Antisense Morpholinos (MOs) used in this study (including runx1 MO,⁷ tlr4bb MO,²⁶ myd88 MO,²⁷ tnfr2 MO,²⁸ gcsfr MO,²⁹ ikbaa MO, pu.1 MO,³⁰ irf8 MO,³¹ cebp1 MO,³²  mismatch MOs (mis MOs), and standard control MO) were purchased from GeneTools and dissolved with distilled H₂O into 1 mM as stock solutions. MOs (4 ng for runx1 MO; 1 ng for tlr4bb MO and tlr4bb mis MO; 2 ng for ikbaa MO and ikbaa mis MO; 4 ng for myd88 MO and myd88 mis MO; 3 ng for tnfr2 MO and tnfr2 mis MO; 8 ng for gcsfr MO; 6 ng for pu.1 MO; 6.4 ng for irf8 MO; 4 ng for cebp1 MO and 8 ng for control MO) were injected into 1-cell-stage zebrafish embryos at the yolk/blastomere boundary. The MO sequences are listed in supplemental Table 1 (available on the Blood Web site). For overexpression of dominant-negative ikbaa (dnikbaa) or NICD specifically in the blood vessel region, the zebrafish ikbaa lacking the first 117 N-terminal nucleotides or full-length coding sequences of zebrafish NICD were cloned into pDONR221 vector by attB and attP recombination reaction (Gateway System) to generate the entry clones, and then subcloned into a pDestTol2pA2 vector with P5E-fli1ep (fli1 enhancer/promoter)³³  and p3E-V2AEGFP-pA by attL and attR recombination reaction (MultiSite Gateway Technology; Invitrogen). The pEGFP-N1-ikbaa fusion construct containing the ikbaa MO recognition site was coinjected with both ikbaa MO and ikbaa mis MO.

Whole-mount in situ hybridization (WISH) for embryos was performed using the ZF-A4 in situ hybridization machine (Zfand) with probes including runx1, cmyb, rag1, foxn1, ephrinB2, gata1, scl, and kdrl.^34-36 

The trunk region of Tg(runx1:en-GFP) embryos at 26 to 28 hpf was dissected to sort the GFP⁺ cells for RNA-Seq. The runx1 GFP⁺ cells were sorted by AriaI (BD Biosciences). Total RNA from the sorted cells was purified and subjected to RNA-Sequencing (RNA-Seq). Beijing Genomics Institute provided the deep-sequencing service and subsequent gene ontology (GO) analysis.

Embryos were treated with dimethylsulfoxide (AMRESCO) or JSH-23 (300 μM; Selleck) from 10-somite stage to 24 hpf or 36 hpf.

Total RNA from trunk region of zebrafish embryos and the AGM regions of tlr4^+/− and tlr4^−/− mouse embryos at embryonic day 10.5 (E10.5) were extracted by TRNzol Reagent (Tiangen) and were reversely transcribed using reverse transcriptase (RT). The complementary DNA was diluted 5 times as the templates. Quantitative polymerase chain reaction (qPCR) was performed with GoTaq qPCR Master Mix (Promega) using the Bio-Rad CFX96 Real-Time PCR system. The PCR primers used are listed in supplemental Table 2.

The trunk regions of zebrafish embryos were dissected for the protein extraction. The nuclear protein was extracted using the Nucl-Cyto-Mem Preparation kit (Applygen). The whole-cell protein extraction and western blot were performed as previously reported³⁴  using anti-p65 (1:800; Cell Signaling Technology), anti-pIκBα (1:800; Cell Signaling Technology), anti-LaminB1 (1:4000; Abcam), and anti-Tubulin β (1:1000; Proteintech) antibodies.

TUNEL (terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling) and BrdU (5-bromo-2′-deoxyuridine) assays were performed as described previously.³⁷ 

The tail fin of Tg(mpo:GFP) embryos was cut at 24 hpf as described previously.²⁰ 

Confocal images were acquired as previously reported³⁸  with a Nikon confocal A1 laser microscope, and 3-dimensional projections were generated using Nikon confocal software. The supplemental Videos from 33 to 54 hpf were generated with PerkinElmer spinning disk confocal microscopy³⁸  and edited by ImageJ.

Embryo dissociation and fluorescence-activated cell sorter (FACS) were performed as described previously.³⁹  The cmyb⁻kdrl⁺ and cmyb⁺kdrl⁺ cells were separately collected from cmyb:GFP/kdrl:mCherry transgenic embryos and lyz⁺ cells from lyz:DsRed transgenic embryos using AriaI (BD Biosciences).

The plasmid with the full length of a human codon-optimized, catalytically inactive version of Cas9 (dCas9), fused with Krüppel-associated box (KRAB) repressor domain, was kindly provided by Liu Dong (Peking University). Capped dCas9-KRAB messenger RNA was synthesized using the mMessage mMachine T7 kit (Ambion) and purified with an RNA purification kit (Tiangen). The guide RNAs (gRNAs) for zebrafish genes were designed according to the Web site (http://zifit.partners.org/ZiFiT/), synthesized by T7 RNA polymerase, and purified by the mirVanaTM miRNA Isolation kit (Ambion) in vitro. Five different gRNAs targeting the downstream transcription start site at the antisense strand were designed for each gene to inhibit the transcription elongation. Cas9 messenger RNA (200 ng/μL) and 5 gRNAs (100 ng/μL per gRNA) were coinjected into 1-cell stage wild-type embryos. Injected embryos were incubated at 28.5°C and collected for quantitative RT-PCR (qRT-PCR), confocal imaging, and WISH at different stages.

An immunofluorescence assay for mouse embryos was performed as previously reported.³⁴ tlr4^+/− and tlr4^−/− embryos at E10.5 were fixed with 4% paraformaldehyde in phosphate-buffered saline for 6 to 8 hours at 4°C. The slides were blocked with 5% bovine serum albumin for 1 hour at room temperature and incubated with Runx1 (Abcam) antibody diluted in 1% bovine serum albumin overnight at 4°C. After being washed 3 times in phosphate-buffered saline with Tween 20, the slides were incubated with anti-rabbit-immunoglobulin-fluorescein for 1 hour at room temperature. The sections were counterstained with 4,6 diamidino-2-phenylindole and images were acquired by Nikon confocal A1.

E10.5 AGMs (36-40 somite pairs) were dissociated by collagenase. FACS was performed as previously reported³⁴  using MoFlo XDP. The CD31⁺CD41⁻CD45⁻TER119⁻ cells of the indicated numbers from tlr4^+/− or tlr4^−/− were cultured on mouse OP9 stromal cells and supplemented with hematopoietic cytokines (50 ng/mL stem cell factor, 50 ng/mL IL3, 20 ng/mL FLT3 ligand). After being cultured for 4 days, semiadherent cells were carefully harvested for flow cytometry.

A colony-forming cell (CFC) assay was performed as previously reported.³⁴  The experiment was repeated in triplicate.

Semiadherent cells were carefully harvested. Then, antibody staining was performed as described previously,³⁴  using antibodies specific to C-Kit–phycoerythrin [PE]–CY7, CD45-PE, CD45.2-PE-CY7, Gr1-PE, CD3–fluorescein isothiocyanate, B220, CD4–fluorescein isothiocyanate, and CD8a-PE (eBioscience).

Eight-week-old male CD45.1 mice were exposed to a split dose of 9 Gy x-ray irradiation. The 2 embryo equivalent cells from tlr4^+/− (CD45.2) or tlr4^−/− (CD45.2) AGM regions were injected into irradiated adult recipients via tail vein, together with 20 000 CD45.1 bone marrow cells. Bone marrow, spleen, and thymus of recipients were collected at the indicated time points. Greater than 10% CD45.2-positive cells as determined by flow cytometry was considered successful reconstitution.

All experiments were performed at least 3 times. Data were given as the mean ± standard error of the mean (SEM). The Student t test was used for statistical comparisons.

Previous studies in zebrafish and mouse embryos^15-17  implicate the potential involvement of inflammatory signaling in HSPC emergence. To further test this possibility, we examined the expression profiling of the earliest HSPC population in zebrafish embryos. A mouse runx1 intronic enhancer was used to drive GFP in zebrafish embryos which can mimic zebrafish runx1 endogenous expression.⁴⁰  The newly generated runx1:en-GFP line can faithfully mark the earliest HSPCs in zebrafish embryos (supplemental Figure 1). We first performed RNA-Seq with hemogenic endothelial cells by isolating runx1⁺ GFP cells (P.Z. and F.L., manuscript in preparation) using FACS at 26 to 28 hpf, at which stage the hemogenic endothelial cells were just specified and underwent the EHT in the ventral wall of the dorsal aorta.^5,7,41  Interestingly, GO analysis showed that inflammatory signaling was highly enriched among the upregulated genes in runx1⁺ cells (Figure 1A-B). The qRT-PCR result confirmed the RNA-Seq data (Figure 1C). Notably, the sorted hemogenic endothelial cells expressed a high level of inflammatory signaling components including cytokine receptors and NF-κB3 (p65); the myeloid cells were used as the positive control (Figure 1D). Together, these data strongly suggest that inflammatory signaling transduced by HE and HSPCs might be involved in definitive hematopoiesis in zebrafish.

To investigate whether proinflammatory signaling is required for HSPC specification, we first performed loss-of-function experiments for the core of inflammatory signaling players, the TLR4–MyD88–NF-κB axis.⁴²  The AGM expression of runx1, which is a well-known HSPC marker in zebrafish and mammals, was drastically reduced in tlr4bb or myd88 morphants (Figure 2A, supplemental Figure 2A-B). Additionally, the number of cmyb:GFP/kdrl:mCherry double-positive HSPCs⁵  in the AGM region was significantly reduced in tlr4bb- or myd88-deficient embryos, compared with control embryos (Figure 2B, supplemental Figure 2C). The number of CD41⁺ cells, that is, the hematopoietic progenitor cells, was also significantly reduced in deficient embryos at 48 hpf (Figure 2C). A well-known readout for HSPC function in zebrafish is its capability to differentiate into T lymphoid. We thus examined expression of T-cell marker rag1. WISH showed that rag1 expression was remarkably reduced in tlr4bb or myd88 morphants, whereas the expression of thymic epithelial cell marker foxn1 was normal (Figure 2A), further supporting that T-cell defects in these deficient embryos resulted from early HSPC defects. We further used the newly developed clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 technology to manipulate expression of these signaling components. Guide RNAs targeting gene-specific promoters, together with a dead form of Cas9 fused with transcriptional repressor KRAB⁴³  (supplemental Figure 3A), were coinjected into zebrafish embryos at the 1-cell stage to achieve conditional repression of genes of interest. Transcriptional repression of TLR4-MyD88 (supplemental Figure 3B) remarkably reduced runx1 expression and the number of cmyb⁺kdrl⁺ cells in the AGM region and CD41⁺ cells in the caudal hematopoietic tissue (CHT) region (supplemental Figure 3C-E).

To further support that inflammatory signaling is required for HSPC development, we treated embryos with JSH-23, a commonly used inhibitor for the nuclear translocation of NF-κB.⁴⁴  First, the efficiency of JSH-23 treatment was validated in zebrafish embryos. The nuclear p65 level was decreased in JSH-23–treated embryos (supplemental Figure 4A). Furthermore, the tail amputation assay also showed that the mpo⁺ myeloid cell number was decreased at the wound region when treated with JSH-23 (supplemental Figure 4B), suggesting that the injury-induced inflammation response was attenuated by JSH-23 treatment. Then, similar HSPC defects were observed in JSH-23–treated embryos (Figure 2D-E). Given that we demonstrated that inflammatory signaling was necessary for HSPC emergence, we wondered whether it was also sufficient to promote this process. Knockdown of IκBaa, the NF-κB inhibitory protein that normally binds to NF-κB and prevents the translocation of NF-κB to the nucleus,⁴⁵  increased the number of HSPCs (Figure 2A-B). Consistently, the decreased expression of HSPC markers was also observed in the myd88 mutants (Figure 2F).

To pinpoint the stage of HSPC development at which inflammatory signaling was required, we performed time-lapse live confocal imaging with cmyb:GFP/kdrl:mCherry double transgenic lines. The number of cmyb⁺kdrl⁺ cells was reduced at the onset of HSPC emergence at 30 to 36 hpf in inflammatory signaling-deficient embryos (supplemental Videos 1-3). In addition, runx1 expression was reduced at 24 hpf, agreeing well with a defect in HE specification (Figure 2A,F). Taken together, these data demonstrated that proinflammatory signaling is required for HE-derived HSPC emergence in zebrafish embryos.

We went further to determine whether TLR4–MyD88–NF-κB regulation of HSPC development is mediated by NF-κB derived from endothelial cells rather than other sources. First, the western blotting result showed that the nuclear localization of NF-κB was indeed decreased in the tlr4bb or myd88 morphants, whereas increased in ikbaa morphants (supplemental Figure 4C). Then, we generated a gateway construct driven by the fli1a promoter to express dominant-negative IκBaa (dnikbaa), which lacks the first 117 N-terminal nucleotides,⁴⁶  fused with GFP. The truncated IκBaa cannot be phosphorylated by the IκB kinase, and therefore prevents the nucleus translocation of NF-κB.¹⁴  The endothelial GFP signals indicated the specific expression of dnikbaa (Figure 3A), and the phosphorylated level of IκBa was decreased in these GFP⁺ embryos (Figure 3A). This endothelial-specific attenuation of NF-κB signaling mimicked the HSPC defects observed in tlr4bb or myd88 morphants (Figure 3B-C). Importantly, coinjection of ikbaa MOs efficiently restored the decreased population of HSPCs and il1b expression in tlr4bb or myd88 morphants (Figure 3D-F, supplemental Figure 4D). Furthermore, overexpression of il1b by heat shock treatment of a transgenic line with heat-shock–inducible il1b embryos [Tg(HSP-il1b-EGFP),²⁰  from 20 hpf, when the artery-vein specification has been accomplished] can also rescue the HSPC defects in the absence of tlr4bb or myd88 (Figure 3G). Together, these data demonstrated that endothelial TLR4–MyD88–NF-κB signaling is both necessary and sufficient for HSPCs in zebrafish.

Because the earliest HSPCs are derived from the hemogenic endothelial cells in the dorsal aorta,^47,48  the observed HSPC defects in inflammatory signaling-deficient embryos were likely attributed to the defective vessel or artery specification. To explore this possibility, we examined expression of endothelial marker kdrl and arterial marker ephrinB2. Expression of these markers was not altered (supplemental Figure 5A-B), indicating a proper vessel development and artery-vein establishment. Similarly, we examined primitive hematopoiesis in these inflammatory signaling-deficient embryos. We found that expression of hemangioblast marker scl, erythroid marker gata1, and myeloid marker l-plastin was relatively normal, suggesting that primitive hematopoiesis was not affected (supplemental Figure 5C). In addition, TUNEL and BrdU staining showed no significant changes in the AGM region in the morphants, compared with control embryos (supplemental Figure 5D-E), implying that the observed HSPC defects were not caused by abnormal cell proliferation or excessive apoptosis.

TLR signaling has been reported to cooperate with Notch signaling to promote production of inflammatory cytokines, and Notch signaling acts upstream of Runx1 to specify HSPCs.^49-51  To explore the possible link between Notch signaling and inflammatory signaling in HSPC emergence, we examined expression of a panel of Notch target genes by qRT-PCR in the AGM region of inflammatory signaling-deficient zebrafish embryos. A significant reduction in expression of Notch target genes was observed specifically in tlr4bb or myd88 morphants (Figure 4A, supplemental Figure 6A). Furthermore, the expression of hey2 was also confirmed in the myd88 mutants (Figure 4B). Consistently, using a Notch reporter transgenic line tp1:DsRed/fli1a:EGFP,²⁵ tp1⁺fli1a⁺ positive cells in the AGM region were greatly reduced (Figure 4C), suggesting that Notch activity was attenuated in these morphants. Repression of these inflammatory genes using dead Cas9 technology further confirmed this result (supplemental Figure 6B). Next, we investigated whether overexpression of Notch can rescue the HSPC defects in the morphants. Endothelial-derived Notch overexpression with fli1a promoter-driven Notch intracellular domain (NICD) efficiently rescued the number of runx1⁺ cells in the AGM region in inflammatory signaling-defective embryos (Figure 4D). Together, endothelial Notch signaling is downstream of TLR4–MyD88–NF-κB–mediated inflammatory signaling to control the emergence of HSPCs in zebrafish embryos.

NF-κB can respond to multiple inflammatory cytokines in innate immune responses. We thus examined whether attenuation of this cytokine-mediated signaling, including TNFα and G-CSF,^52,53  would cause similar HSPC defects as observed in TLR–NF-κB–deficient embryos. runx1 expression was remarkably decreased in tnfr2- and gcsfr-deficient embryos by MO knockdown, compared with control embryos (Figure 5A). The number of cmyb⁺kdrl⁺ HSPCs in the AGM region was significantly reduced (Figure 5B). The HSPC phenotype in gcsfr morphants was consistent with a recent report by another group.⁵⁴  To demonstrate whether the cytokine signaling is mediated by NF-κB, we examined expression of il1b and found that il1b expression was decreased to the similar level as that in NF-κB–deficient embryos (Figure 5C), suggesting that the HSPC defects observed in tnfr2- and gcsfr-deficient embryos were likely attributed to dysregulated NF-κB signaling. Importantly, coinjection of ikbaa MO efficiently restored the decreased population of HSPCs in these morphants, indicating that NF-κB acts downstream of inflammatory cytokine signaling (Figure 5D). Taken together, these data support that inflammatory cytokines activate NF-κB signaling to control HE and HSPC specification.

Consistently, both the expression of Notch targets and the number of tp1⁺fli1a⁺ cells in the AGM region were greatly reduced in these inflammatory receptor-defective embryos (supplemental Figure 7A-C). Furthermore, the endothelial-derived NICD overexpression efficiently rescued the population of runx1⁺ cells in the AGM region in inflammatory signaling-defective embryos (supplemental Figure 7D). Together, endothelial Notch signaling also functions downstream of inflammatory signaling to control the emergence of HSPCs in zebrafish embryos.

To determine which subset of the myeloid lineages is responsible for the production of inflammatory cytokines, we used pu.1 MO which can eliminate all myeloid lineages in zebrafish, irf8 MO which attenuates macrophages, and cebp1 MO which decreases neutrophils. The runx1 expression was not much affected in irf8 knockdown embryos, whereas decreased in pu.1 and cebp1 knockdown embryos (supplemental Figure 8A-B), suggesting that neutrophils are the main source for inflammatory cytokines responsible for HSPC emergence, consistent with the findings published recently.^13,14 

To determine whether inflammatory signaling is evolutionarily conserved in HSPC emergence in mammals, we examined the AGM region of tlr4^−/− knockout mice.⁵⁵  Immunofluorescence analysis showed that Runx1 expression was severely attenuated in the AGM (Figure 6A). qRT-PCR confirmed decreased levels of runx1 and cmyb, as well as downregulated expression of il1b and il6, in NF-κB signaling (Figure 6B-C). As observed in zebrafish, expression of Notch target genes was also decreased (Figure 6D). The colony-forming unit-cell (CFU-C) assay showed compromised colony formation abilities of AGM tissues in tlr4^−/− mice (Figure 6E). To test whether inflammatory signaling acts on endothelial cells cell-autonomously in mice, endothelial cells (CD31⁺CD41⁻CD45⁻TER119⁻)⁵⁶  from the AGM tissues in tlr4^+/− or tlr4^−/− were cultured for 4 days, then flow cytometry was performed (Figure 6F). The C-Kit⁺CD45⁺ hematopoietic cells were significantly decreased in tlr4^−/− embryos (Figure 6G). The HSC transplantation is widely accepted as the most reliable assay for HSC activity. Five of 6 recipients were reconstituted after 2 months when transplanted with cells from tlr4^+/− embryos, whereas only 1 of 6 reconstituted when transplanted with cells from tlr4^−/− embryos (Figure 6H-I). Together, these data suggest that the role of inflammatory signaling in HSPC emergence is evolutionarily conserved in mammals.

Many if not all developmental signaling pathways during developmental hematopoiesis and homeostasis have been identified. Inflammatory signaling has been well studied in infection and inflammation, but its physiological and developmental roles are poorly understood. In the present study, we have demonstrated that inflammatory signaling plays a highly conserved role in HSPC emergence in zebrafish and mice by investigating the expression pattern, requirement for HSPC specification, and regulatory mechanisms. Inflammatory signaling functions upstream of Notch signaling pathway via NF-κB. Our findings strongly support that proinflammatory signaling is a common master regulator of HSPC emergence and is shared by normal hematopoiesis and stress hematopoiesis in vertebrates.

The relationship between NF-κB and Notch is controversial and has been studied in different systems. The Notch signaling pathway is essential to maintain HSPC homeostasis by regulating the inflammatory signaling level in the bone marrow niche because inhibition of Notch signaling activates NF-κB via mir-155 and leads to myeloproliferation.⁵⁷  On the one hand, Notch may prevent the NF-κB activation by directly interacting with NF-κB in the nucleus.⁵⁸  On the other hand, Notch signaling has been reported to activate NF-κB^59,60  as well. In hematopoietic progenitor cells, Notch1 has been shown to regulate NF-κB activity through transcriptional regulation of NF-κB subunits.⁶¹  In the T-cell leukemia, the Notch/Hes1 pathway sustains the NF-κB activation by negatively regulating IκB kinase.⁶²  Furthermore, the cooperation of Notch3 and canonical NF-κB signaling pathways can regulate Foxp3 transcription to maintain the regulatory T-cell homeostasis and function.⁶³  In this work, we identified the NF-κB regulation of the Notch signaling pathway by upregulating Notch signaling at the onset of definitive hematopoiesis, which is consistent with the NF-κB–dependent activation of Notch in endothelial cells⁶⁴  and peripheral lymphoid tissues.⁶⁵  Further studies are warranted to fully characterize the interaction between these 2 pathways in HE and HSPC specification.

Baseline inflammatory signaling might be crucial for the emergence of HSPCs under the physiological condition in that a low level of inflammatory signaling promotes HSPC specification or “priming.” Once in demand, that is, in response to infection, inflammatory signaling will rapidly induce the programming for cytokine production within the earliest HSPCs, and efficiently instruct HSPC differentiation into effector immune cells, such as myeloid and lymphoid cells.^8,66  However, the sources of endogenous TLR ligands and inflammatory cytokines during normal hematopoiesis remain unclear. It is possible that the reported heat shock proteins and signals from damaged or stressed cells⁶⁷  firstly induce the TLR4–NF-κB pathway during normal embryonic hematopoiesis, and then activate the Notch pathway to regulate the HSPC emergence.

In conclusion, our results support that proinflammatory signaling is a novel and potent regulator during HSPC emergence and is shared by normal and stress hematopoiesis in vertebrates. This may prove to be a useful theoretical basis in treating patients with bone marrow failure or following transplantation with clinically applicable cytokines and point to a new strategy for developing a new protocol for in vitro generation and expansion of inducible HSPCs by modulation of inflammatory signaling.

Note added in proof. During the preparation of the revision for this work, 3 recent publications reported that proinflammatory signaling mediated by TNFα or interferons plays an essential role in HSPC development in zebrafish and mouse embryos,^12-14  consistent with our findings here.

The authors thank Dangsheng Li and Zilong Wen for helpful discussions and/or critical reading of the paper. The authors also thank Zilong Wen and Jing-Wei Xiong for providing the reagents used in this work.

This work was supported by grants from the National Basic Research Program of China (2010CB945300, 2011CB943900), the National Natural Science Foundation of China (31271570, 31425016), and the Strategic Priority Research Program of the Chinese Academy of Sciences (CAS) (XDA01010110).

Contribution: Q.H., C.Z., L.W., P.Z., D.M., and J.L. performed the experiments; F.L. conceived the project, analyzed the data, and wrote the paper; and all authors read and approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Feng Liu, State Key Laboratory of Biomembrane and Membrane Biotechnology, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China; e-mail: liuf@ioz.ac.cn.