Yolk sac-derived Pdcd11-positive cells modulate zebrafish microglia differentiation through the NF-κB-Tgfβ1 pathway

Microglia are the primary immune cells in the central nervous system, which plays a vital role in neuron development and neurodegenerative diseases. Microglial precursors in peripheral hematopoietic tissues colonize the central nervous system during early embryogenesis. However, how intrinsic and extrinsic signals integrate to regulate microglia’s differentiation remains undefined. In this study, we identified the cerebral white matter hyperintensities susceptibility gene, programmed cell death protein 11 (PDCD11), as an essential factor regulating microglia differentiation. In zebrafish, pdcd11 deficiency prevents the differentiation of the precursors to mature brain microglia. Although, the inflammatory featured macrophage brain colonization is augmented. At 22 h post fertilization, the Pdcd11-positive cells on the yolk sac are distinct from macrophages and neutrophils. Mechanistically, PDCD11 exerts its physiological role by differentially regulating the functions of nuclear factor-kappa B family members, P65 and c-Rel, suppressing P65-mediated expression of inflammatory cytokines, such as tnfα, and enhancing the c-Rel-dependent appearance of tgfβ1. The present study provides novel insights in understanding microglia differentiation during zebrafish development.


Introduction
Microglia, the innate immune cells of the central nervous system (CNS), enter the brain during early embryogenesis and play important roles in white matter pathology [8]. As influenced by their environment, microglia assume different molecular patterns ranging from M1 phenotype in secretion pro-inflammatory cytokines to M2 phenotype in inflammation resolution and tissue repair [9]. In the zebrafish embryo, macrophages develop early on the yolk sac (YS) expression L-plastin and spread in the cephalic mesenchyme at about 24 h post fertilization (hpf). After which they invaded brain and differentiated into microglia featured by lower expression of the L-plastin, and higher expression of the neurotrophic lipid carrier, apolipoprotein E at about 60 hpf [10,11]. Factors like Colony-Stimulating Factor 1 Receptor (CSF1R) and neuronal apoptosis modulate the seeding of microglia precursors in the brain [12,13]. But how intrinsic and extrinsic signals integrate in microglia differentiation remains largely unknown. Transforming growth factor-beta (TGF-β) family members are cytokine important for both innate and adaptive immune functions [14]. It is required for the maintenance of the microglia-specific homeostatic gene signature. Deletion of the TGF-β receptor in microglia resulted in rapid conversion of microglia toward an inflammatory macrophage phenotype [15,16]. Although the importance of brain-derived TGF-β1 in shaping immune response in pathogenic conditions has been demonstrated, the origin of embryonic TGF-β1, especially for inducing microglia differentiation, remains unknown.   and C-Terminal Coil domain (1408-1871). b Gross morphology of 3 dpf WT and pdcd11 mutants. c qPCR confirmation of hyperactivated inflammatory pathway genes, including tnfa, il1b, and il6, in 22 hpf pdcd11 mutants, which could not be restored by combined p53 mutation. d-e Il6 and il1b expression in the brain of WT or pdcd11 mutants with p53 mutated or not examined by WISH at 52 hpf. f WISH analysis of lplastin, mfap4, and csf1ra expression in 24 hpf WT and pdcd11 mutant embryos. Red arrows indicate csf1ra-positive cells in the brain. g Microglia development in 3 dpf WT and pdcd11 mutants by WISH assessment of apoeb and Neutral Red staining. h L-plastin and mfap4 expression pattern in 3 dpf WT and pdcd11 mutants. i Morphology of macrophages in the brain and caudal hematopoietic tissue (CHT) were examined using the Tg(mpeg1:GFP) transgenic line. White arrows indicate the vacuolated macrophages found in the pdcd11 mutant. j Increased tnfa mRNA expression in sorted macrophages Tg(mpeg1: GFP) from 60 hpf pdcd11 mutants as compared with WT controls. In this study, helped by zebrafish model organism, we demonstrated that PDCD11, a WHM susceptibility gene, helps promote embryonic microglia differentiation through Tgfβ1. Macrophages in pdcd11 mutants showed increased expression of inflammatory genes and impaired upregulation of microglia signature genes. This imbalanced macrophage generation, likely caused by uncontrolled P65 activation, causes inflammatory gene upregulation and suppresses c-Rel-dependent Tgfβ1 expression under pdcd11 deficiency.

P53 independent inflammatory pathway activation in pdcd11 mutants
Pdcd11 is highly conserved between species and mainly harbors two functional domains, including an N-terminal ribosome and Coil (Fig. 1a). To define the role of Pdcd11 in embryonic development, we generated a pdcd11 mutant zebrafish line with a 7-bp deletion in the fourth exon by CRISPR-Cas9 nuclease (Fig. S1a). Antibody targeting the ribosome domain of Pdcd11 confirmed the complete loss of functional domains in pdcd11 mutants (Fig. 1a). Perturbations of ribosome biogenesis often stabilize and activate genes related to the P53 apoptosis pathway [17][18][19]. Common with other reported ribosome defects, obvious developmental retardation and craniofacial malformation were observed in 3 dpf pdcd11 mutants (Fig. 1b). Not surprisingly, P53 pathway genes were hyperactivated (Fig. S1b) and acridine orange (AO)-positive dead cells were also significantly increased in pdcd11 deficient embryos from 22 hpf (Fig. S1c). To determine the contribution of the induction of apoptosis in the developmental defects of pdcd11 mutants, pdcd11 mutants in a p53 knockout background were also generated [20]. P53 depletion favorably recovered AO cell numbers and P53 pathway genes in pdcd11 mutants (Fig. S1d, e). However, cranial deformities, shorter body lengths, and survival rates were only partially restored (Fig. S1f, g), possibly suggesting that other P53 independent factors also contribute to the pathology of pdcd11 mutants.
Interestingly, consistent with the perivascular inflammation a prominent pathological feature in WMH, we found that genes associated with inflammatory responses, like il6, il1b, and tnfa, were significantly increased, which otherwise failed to be restored by combined p53 mutation (Fig. 1c). Using whole-mount in situ hybridization (WISH), the apparent increased expression of inflammatory genes in the brain of 52 hpf pdcd11 mutants were detected, which also could not be recovered by the p53 mutation (Fig. 1d, e).

Macrophage generation imbalance in pdcd11 mutants
Inflammatory gene expression was elevated in pdcd11 mutants as early as 22 hpf. At this time, macrophages and neutrophils were the two major types of immune-related inflammatory cells identified in zebrafish. Macrophages initiate migration in the cephalic mesenchyme at about 24 hpf. Compared with that in WT embryos, pdcd11 deficiency induced increased and earlier accumulation of macrophages in brain tissue as revealed by mfap4 or pan-leukocyte marker l-plastin staining (Fig. 1f). The distribution pattern of mpo-positive leukocytes remained unaltered at this time (Fig. S2a). A previous study showed that apoptotic neuronal death drives the colonization of microglia precursors in the optic tectum [21]. Thus, we speculated that the advanced macrophage migration might be stimulated by increased neural apoptotic signaling. However, the combined p53 mutation failed to restore the increased macrophage accumulation in the brains of pdcd11 mutants, with accumulation actually being exacerbated (Fig. S2b). Interestingly, distinct from nonspecific macrophage markers, the expression of csf1ra, which was previously shown to be required for the migration and proliferation of microglia precursors [13,22], was contrarily reduced in pdcd11 mutants (Fig. 1f). Consistently, staining by either apoeb or Neutral Red revealed that mature microglia, which are tissueresident macrophages in the brain, were completely absent in 3 dpf pdcd11 mutants (Fig. 1g). At this time, macrophages stained for mfap4 or l-plastin were abundant in the brain of pdcd11 mutants, with an evidently enlarged and vacuolated morphology, which was also verified using a macrophage transgenic line Tg(mpeg1:GFP) (Fig. 1h, i). Macrophages acquire specific functional phenotypes that are manifest by secretion of inflammatory gene products and their morphological appearance. We found that the expression of tnfa, a well-established marker of M1 macrophages in zebrafish [23], was obviously increased in macrophages sorted from 60 hpf pdcd11-deficient embryos (Fig. 1j). In teleost fish, cxcr3.1 and cxcr3.2 macrophages show functional polarization towards M1 and M2, respectively [24]. We found that in 22 hpf pdcd11 mutants, cxcr3.1 expression was profoundly increased while that of cxcr3.2 was reduced as compared with that in WT controls (Fig. 1k, l).
Similar to the failed restoration of inflammatory genes with p53 mutation, simply inhibiting apoptosis could not restore the macrophage generation abnormalities observed in pdcd11 mutant embryos (Fig. S2c, d). Moreover, no obvious aberrations in proliferation or apoptosis of myeloid progenitors were detected in pdcd11 mutants at 22 hpf ( Fig. S2e-h).

Pdcd11-positive cells yield Tgfβ1 required for microglia differentiation
Microglia differentiation requires step-wise-activated intrinsic transcription factors and circumstance-derived inducing factors, like Tgfβ1. Through comparing the RNA-seq data carried between 22 hpf WT and pdcd11 mutants, we noticed that tgfb1a mRNA, a homolog of human TGFB1, was profoundly reduced in 22 hpf pdcd11 mutant embryos (Table S1). The finding was verified by qPCR and western blot (Fig. 2a, b). Using WISH, we found that tgfb1a was highly expressed in cells on the YS at 22 hpf, at a location similar to Pdcd11, and that the expression of tgfb1a was obviously reduced with pdcd11 deficiency (Fig. 2c). Otherwise the expression of tgfb1b, another orthologue of human TGFB1 were lowly expressed at 22 hpf and no difference was detected with pdcd11 depletion ( Fig. S2i, j). Moreover, specially enriched Tgfβ1 in Pdcd11 expression cells was revealed using an immunofluorescence assay (Fig. 2d). Tgfβ1-induced signature genes in mouse microglia have been studied [25,26]. Supporting the role of pdcd11 in mediating Tgfβ1 activation in microglia, brain macrophages sorted from pdcd11-deficient Tg(mpeg1:GFP) line showed reduced expression of Tgfβ1-stimulated genes, including sall1a, tgfbr1a and tgfbr1b, and increased expression of Tgfβ1-suppressed genes like tlr5a and ctsl.1 as compared with that in WT embryos (Fig. 2e). The elevation of csf1b and reduction of ctsba in microglia were also verified by WISH (Fig. 2f). Confirming the importance of tgfb1a reduction in contributing to microglia depletion in pdcd11 mutants, tgfb1a overexpression restored both precursor and mature microglia numbers in pdcd11 mutants ( Fig. 2g-i). However, this rescue was compromised by the simultaneous repression of Tgfβ activation in macrophages by either LY364947, a TGFβR-I inhibitor addition, or overexpressing inhibitory cofactor smad7 under control of the mpeg1 promoter, implying the importance of Pdcd11derived Tgfβ in regulating macrophage development ( Fig. 2h-j).

Elevated P65 transcriptional activity in pdcd11 mutants
Aside from the large N-terminal ribosome domain (Rib), Pdcd11 contains C-Terminal Coil domain (Coil), the function of which has not been defined (Fig. 3a). To investigate the molecular mechanism, we first overexpressed Rib and Coil. While Rib overexpression faithfully restored P53 pathway genes, the disturbances of macrophage generation and inflammation pathway activation could only be reversed with Coil overexpression in pdcd11 mutants (Fig. 3b, c and S3a-c). A short C-Terminal fragment (Ter) reportedly could interact with P65 ( Fig. 3a), one member of the NF-κB family [27,28]. The NF-κB signaling pathway participates in modulating tissue immune homeostasis. Under normal circumstances, P65 is retained in the cytoplasm by the inhibitory protein IkappB and is available to the proteasome degradation pathway, which avoids uncontrolled inflammation activation. Upon stimulation, P65 is post-translationally modified by phosphorylation at the 536-serine residue, which obviates the interaction with IκB. P65 can then translocate into the nucleus to activate the downstream signaling pathway. We found that 536-serine phosphorylated P65 was obviously increased in pdcd11 mutants (Fig. 3d). In vitro overexpression of Coil or simply the terminal fragment Ter efficiently reduced nuclear 536serine phosphorylated P65 (Fig. 3e). Interestingly, we found that in vitro transfected P65 and Coil mostly colocalized in the cytoplasm or in the nucleolus, which correlated with reduced P65 transcriptional activity [29,30] (Fig. 3f). The cytoplasm and nuclear P65 content with Coil overexpression were then examined respectively by western blot. Increasing Coil transfection obviously increased the P65 cytoplasm content, while correspondingly reducing the nuclear P65 level (Fig. 3g). Moreover, increased 536serine phosphorylated P65 in the pdcd11 mutants were restored by either Coil or Ter domain overexpression, but not by Rib domain overexpression (Fig. 3h). To determine whether the rescuing effects of Coil relied on its capacity in suppressing P65 activation, two NF-κB inhibitors, BMS-345541 and BAY 11-7821, were used [31]. Treatment with either BMS-345541 or BAY 11-7821 reduced 536-serine phosphorylated P65, and inflammatory genes and macrophages also returned to normal in pdcd11 mutants ( Fig. 3i and S3d, e). A correctly regulated inflammasome cascade is required for the maintenance of microglia homeostasis [32]. However, microglia number or the TGFβ1 level could not be rescued by BMS-345541 or BAY 11-7821 treatment (Fig. 3j, k), suggesting that other P65 independent mechanisms might be utilized by Coil in regulating microglia generation.
Interestingly, while Coil overexpression restored both macrophage appearance and inflammatory genes in the pdcd11 mutants, similar efficacy was not detected with either Coil overexpression in pan neuron cells under Huc promoter (also named embryonic lethal, abnormal visionlike, neuron-specific RNA-binding protein 3, elavl3) [33] or macrophages under control of the mpeg1 promoter [34] (Fig. S3c, f).

Pdcd11 enhances c-Rel/p105 nuclear transport
Transgenic zebrafish with NF-κB recognition sequences driven fluorescent protein was utilized to study NF-κB activity in zebrafish [35]. Helped by this transgenic line, we were surprised to find that although PDCD11 suppressed P65 activation, NF-κB was otherwise activated in Pdcd11expressing cells on the YS at 22 hpf (Fig. 4a). In comparison with cytoplasm Coil expressed cells, cells with nuclear Coil expression were prone to activate EGFP under control of the NF-κB recognition sequences ( Fig. 4b and S4a).
Among the five members of NF-κB family, P65, c-Rel, and Rel-B harbor the transcription activation domain (TAD), which binds NF-κB recognition sequences and activates the expression of downstream genes [36,37]. The findings prompted us to speculate that this activation of the NF-κBbinding element relied on the transcriptional activity of other members of the NF-κB family. Indeed, different from P65, Coil colocalized with c-Rel and P105 in the nucleus and Coil overexpression promoted the nuclear retention of c-Rel (Fig. 4c-e and S4b, c). A luciferase assay showed that Coil co-expression enhanced the capacity of c-Rel in activating the NF-κB-binding element (Fig. 4f). Confirming the importance of reduced nuclear c-Rel/P105 in leading the microglia deficiency in pdcd11 mutants, we found that c-Rel and p105 co-overexpression recovered microglia numbers as manifested by either apoeb or Neutral Red staining (Fig. 4g, h). These rescue effects were largely compromised when repressive smad7 was co-expressed in macrophages, suggesting that c-Rel/P105 functions upstream of TGFβ signaling (Fig. 4g, h). Consistently, the expression level of Tgfβ1 was also increased with c-Rel and p105 mRNA overexpression (Fig. 4i, j). Injection of the xanthine derivative pentoxifylline (PTXF), which selectively degrades c-Rel [38], also reduced the number of microglia in WT embryos. The reduction could also be recovered by the . Under normal circumstances, on one hand, the Ter fragment of PDCD11 (yellow) binds and masks the phosphorylation site of P65 (pink), which is required for its nuclear translocation and transcriptional activation; on the other hand, the only existing interaction between the coil-coil of PDCD11 and the RHD domain of c-Rel contrarily promotes c-Rel (green) nuclear retention for activating TGFβ1 expression, which drives macrophage differentiation toward microglia with "M2" properties. However, cells deficient in PDCD11 show increased P65 nuclear translocation, which tends to lead to activation of inflammation-related genes, such as IL-1, IL-6, and TNFα. These cytokines then facilitate "M1" macrophage differentiation.
injection of tgfb1a mRNA (Fig. 4k). In a previous study, chromatin immunoprecipitation combined with microarray technology (ChIP/Chip) examination of mitogen-stimulated T cells Rel/NF-kappa-B revealed potential Rel/NF-κB recognition motifs in the promoter region of TGFB1 [39]. To further determine the regulatory role on TGFB1 by Coil and c-Rel, HEK293 cells were examined using a luciferase reporter assay. Coil co-expression enhanced the capacity of c-Rel in stimulating the TGFB1 promoter. This enhancement was abolished by the combined M1 and M2 c-Rel-binding site mutation (Fig. 4l). These results suggest that Coil and c-Rel promotes tgfb1 expression in early zebrafish embryos.
Since P65 and c-Rel are structurally similar and both harbor an N-terminal Rel homology domain (RHD) and C-Terminal TAD domain, it was of interest to investigate how PDCD11 distinguished the two (Fig. 5a). Co-IP examination revealed that both P65 and c-Rel could interact with the Coil domain of PDCD11 (Fig. 5b, c). The Coil domain was further divided into the Coil-coil and Ter fragment, which displayed a different interaction pattern with P65 and c-Rel. Since the RHD domain of P65 and c-Rel interact with the Coil-coil region of PDCD11, the Ter fragment failed to bind to the TAD domain of c-Rel as observed with P65 ( Fig. 5d-g). This result was reasonable since, because the RHD domains of NF-κB family members are similar, their TAD domains are much more distinct. The interaction of Ter with the TAD domain of P65 might cover the phosphorylation site required for P65 transcriptional activation. The mere overexpression the Ter fragment could suppress the P65 activation in pdcd11 mutants, otherwise the Coil-coil fragment overexpression failed to do so (Fig. 5h). Confirming the importance of Ter fragment in inhibiting P65 activation, the overexpression of this fragment also inhibited P65 nuclear translocation (Fig. 5i). Thus, the differential interactions between the TAD domains and PDCD11 might contribute to the different regulation mode of P65 and c-Rel by PDCD11 (Fig. 5j).

YS Pdcd11-expressing cells are derived from Pu.1progenitors
By WISH, we found that aside from the obvious expression signal in the neuron epithelium, scattered cells on the YS were also detected from 7-somite stages (ss) (Fig. 6a). At 32 hpf, pdcd11-positive cells had accumulated in the pericardial cavity, the site of inflow of all blood cells (Fig. 6a). In zebrafish, myeloid progenitor cells expressing Pu.1 emerged within the anterior lateral plate mesoderm (ALPM) at the 5-ss and then migrated into the rostral blood island (RBI) and onto the YS. This was followed by differentiation of the cells into macrophages and neutrophils, which then seeded the brain from approximately 24 hpf. Through use of the zebrafish Tg(Pu.1:GFP) transgenic line, we found that at 10-ss in the ALPM region, Pdcd11 was activated in a subset of Pu.1-GFP-positive cells (Fig. 6b). However, this co-expression pattern reduced at 22 hpf and thereafter remained absent, suggesting later silencing Pu.1 expression (Fig. 6b). The expression of Pdcd11 in macrophages and neutrophils was also examined using Tg(mpeg1:GFP) and Tg(mpo:GFP) transgenic lines, respectively. Mpeg1positive macrophages appeared to be closely surrounded by Pdcd11-positive cells in the pericardial cavity underneath the brain at 32 hpf, but no colocalizations were detected (Fig. 6c). Similarly, no colocalizations between Pdcd11 and mpo-positive neutrophils were detected (Fig. S4d). Intriguingly, in pu.1-deficient embryos, while pdcd11 expression in the neuron epithelium remained unchanged, its expression in the pericardial cavity was disappeared, implying that the scattered pdcd11-expressing cells observed on the YS were Pu.1-dependent (Fig. 6d). To verify the pdcd11 expression pattern, the 1.86 kb genomic region upstream of the translation initiation site of zebrafish pdcd11 was cloned into a transposon vector containing EGFP. The pdcd11 promoter-driven GFP-expressing cells on the YS strongly colocalized with endogenous Pdcd11 at 24 hpf in an immunofluorescence assay (Fig. 6e). Overexpression of Coil under the control of the 1.86 kb pdcd11 promoter reversed the macrophage imbalance in pdcd11 mutants, with the number of inflammatory macrophages reduced and that of microglia increased (Fig. 6f, g). This rescue effect was not observed in embryos overexpressing Coil overexpression under the control of neuron (Huc), macrophage (Mpeg1), or neutrophil (Mpo) specific promoters (Fig. 6g). Collectively, these results suggest that scattered Pdcd11-expressing cells may be derivatives of Pu.1-positive progenitors, but that they are distinct from previously identified neutrophils or macrophages.

Discussion
WMH examined by magnetic resonance imaging have been used to describe WM changes related with cognitive dysfunction, stroke injury, cerebral small vessel disease, and neurodegenerative disorders [40]. Key cellular components of the gliovascular unit including astrocytes, oligodendrocytes, pericytes and microglia. It was suggested that microglia within the WM be associated functions like surveillance at the blood-brain barrier and clearance debris [41,42]. In this study, helped by zebrafish model organism, we found that deficiency of pdcd11, a WMH susceptibility gene, led to completely lost of mature microglia in the brain. In mouse, helped by in vivo lineage tracing, it was found that adult microglia derive from primitive myeloid progenitors that arise before embryonic day 8 [43]. However, as with zebrafish, multiple waves of microglia exist and originate from distinct hematopoietic precursors were identified [44]. Despite the different ontogeny of primitive and adult microglia between mouse and zebrafish identified, the regulators underlying the differentiation of EMP progenitors toward microglia remain undefined. TGF-β family members are cytokine required for the maintenance of the microglia-specific homeostatic gene signature. Deletion of the TGF-β receptor in microglia resulted in rapid conversion of microglia toward an inflammatory macrophage phenotype [15,16]. Mice harboring mutated TGFB1 exhibit an acute wasting syndrome followed by death [45]. The most prominent lesions are tissue necrosis in specific organs and multifocal, mixed inflammatory cell infiltration into numerous organs [46]. Despite accumulating evidence for local expression of TGF-β1 in the nervous system during inflammation, the cellular source of TGF-β1, which drives microglia differentiation is unknown. Pdcd11-deficient embryos exhibit brain necrosis and organ inflammatory cell infiltration, which is very similar to what is observed with TGFB1-mutated mice. Moreover, Pdcd11-expressing cells on the YS colocalized with TGF-β1-positive cells at 22 hpf on the YS. Both the mRNA and protein levels of TGF-β1 were significantly reduced with pdcd11 depletion. Importantly, overexpression of TGF-β1 under the control of the pdcd11 promoter faithfully recovered the microglia numbers in pdcd11 mutants. Thus, our observations support the notion that during the early development in zebrafish, Pdcd11-expressing cells might be one of the major source of TGF-β1 to induce microglia maturation.
In studying the property of YS Pdcd11-expressing cells, we found that these cells are distinct from initially identified macrophages and neutrophils, but they are also derived from Pu.1-expressing progenitors and require Pu.1. The NF-κB family member possesses distinct biological functions and controls the expression of different target genes, manifesting as unique developmental phenotypes between NF-κB/Rel family member knock-out mice. The distinct and opposing functions of P65 and c-Rel in regulating cellular functions, such as proliferation and apoptosis, have been revealed in several different models [47,48]. In P105 and c-Rel double knock-out mice, while peripheral T-cell populations developed normally, follicular, marginal zone, and CD5 + peritoneal B-cell populations were all reduced, suggesting the particular requirement of P105 and c-Rel in these cell types [49]. It was reported that B1a cells are abundant in the neonatal mouse brain and promote oligodendrogenesis by regulating the proliferation of oligodendrocyte-precursor cells and the number of microglia [50]. As the definitive markers for B1a lymphocytes in zebrafish are elusive, other model organism like mice might be useful for our understanding Pdcd11-positive cells in the future. A recent report in studying the mpeg1.1:GFP reporter line, it found that beside macrophages, a subpopulation of Blymphocytes is marked by mpeg1.1 reporters in most adult zebrafish organs, with highly expressed cd79a, ighm, and ighd [51]. This would also be helpful for our future examination of this transgenic line in study the property of Pdcd11-positive cells.

Experimental model
Zebrafish maintenance and staging were performed as previously described [52]. The zebrafish facility and study were approved by the Institutional Review Board of the Institute of Health Sciences, Shanghai Institutes of Biological Sciences, Chinese Academy of Sciences (Shanghai, China), and the methods were carried out in accordance with the approved guidelines. The Tg(pu.1: GFP) [53], Tg (mpeg1: GFP) [54], Tg(mpo: GFP) [55], P53 zdf1/zdf1 mutants [20,56], pu.1 Δ839 zebrafish lines were used.
mRNA synthesis, and microinjection mMACHINE Kit (Ambion) was used for mRNA transcription. For promoter-driven Coil or smad7 expression in zebrafish, 30 ng/μl transposase mRNA was coinjected.

qPCR analysis
After treatment with DNAaseI (Ambion), RNA was prepared using Trizol reagent (Invitrogen, H10522) and then subjected to cDNA synthesis using a cDNA synthesis kit (ABI). Real-time PCR was performed using a Fast Start Universal SYBR ® Green Master Rox probe (Roche Applied Science, 13800300) and Mastercycler thermal cycler (Eppendorf, 22331). Primers are summarized in Table S2.

Cell culture and dual-luciferase reporter assay
Cells were cultured in a humidified incubator at 37°C in the presence of 5% CO 2 . 293T and Hela cells obtained from ATCC were cultured in DMEM medium supplemented with 10% fetal bovine serum, 100 IU/mL penicillin, and 100 μg/ mL streptomycin (Gibco). Identity of cell lines were validated by STR analysis. Mycoplasma testing was routinely performed on all cells used in the study, and confirmed to test negative. For cell transfection, Effectene transfection reagent (QlAGEN, 1012829) was used according to the manufacturer's instructions. NF-κΒ binding motif driven luciferase reporter and human TGFB1 (1.3-kb upstream the transcription start site) was used. C-Coil (150 ng) and/or c-Rel (150 ng) as well as Renilla constructs (5 ng) were transfected into 293 T cells (60%-70% confluence per well in 24-well plates) and luciferase activities were assessed according to the manufacturer's instructions (Promega). Site mutagenesis was carried on the TGFB1 promoter (Invitrogen, A13282)

Cell sorting and cytology
Tg (mpeg1:EGFP) transgenic embryos at 60 hpf were connected and digested with 0.5% trypsin (Gibco) for 15 min at 37°C to analyze microglia gene signature. Sixty hours post fertilization zebrafish brain was cut before digestion. After centrifugation at 400 × g for 5 min, two washes with PBS, and passage through a 40 μm nylon mesh filter, the single cell suspension was subjected to fluorescence-activated cell sorting using a MoFlo device (DakoCytomation). EGFP positive cells were used for downstream qPCR assay.

Cell IF assay
Cell IF were carried as previously described. Cells grown to 60% confluence on coverslips were first fixed and blocked and then incubated with mouse anti-FLAG (Sigma-Aldrich, F7425). Secondary antibody goat anti-mouse, Alexa Fluor 594 (Invitrogen).

Imaging
Zebrafish IF stain images were taken using an Olympus FV1000 scanning confocal microscope. The confocal images were captured with an UPLSAPO 40× or 60× objective. Immuno-stained cell images were collected using a Leika with an UPLSAPO 60× objective.

RNA-sequencing (RNA-seq)
Total RNA from 22 hpf WT and pdcd11 mutants was collected respectively and subjected to RNA-seq. Shanghai Novel Bioinformatics Company provided the deepsequencing service.

Statistical analysis
For all experiments carried, zebrafish embryos were randomized distributed to each group. The outcome of experiments were assessed according to phenotypic changes. Statistical analyses were conducted using the Student's t test (two-sided) or log-rank tests. Statistical significance was taken to be P < 0.05.