Wildtype (WT), transgenic and mutant zebrafish were maintained in accordance with standard husbandry practices (18). To generate socs3b mutants, WT embryos were injected with ~1 pL 12.5 ng/μL guide RNA (gRNA)1 (5’-GUCCGAGCUGUCCCGCACC) and gRNA2 (5’-GGCGCACGCCGCCUUCAAAG) targeting socs3b and 100 ng/μL Cas9 mRNA (Sigma-Aldrich) at the 1 cell stage. Zebrafish were screened using PCR of genomic DNA extracted from fin clips using primers flanking the gRNA1 (5’-ACAGGGGGTGGGCAGAGC and 5’-GCTCACGGTGCTCCAGTAGAA) and gRNA2 (5’-GAGCGGGAAGGAGGCCAG and 5’-GCCGTTTTCACACTGAGCGT) target sites, followed by Sanger sequencing (Australian Genome Research Facility). Zebrafish carrying a suitable mutant allele were subsequently out-crossed with WT zebrafish for two generations. In order to minimize potential bias a mixed pool of embryos obtained from heterozygote mutant in-crosses were used, with genotyping performed post-analysis, as described (19). All experiments compared homozygous mutant zebrafish with WT siblings generated from the same cross and were performed in a blinded manner and repeated at least 3 times, with representative data shown. The mutant alleles were also crossed onto two transgenic lines Tg(mpx::GFP) (20) and Tg(mpeg1.1::GFP) (21).

Zebrafish embryos at relevant timepoints were fixed with 4% (w/v) paraformaldehyde (PFA) at 4°C prior to whole-mount in situ hybridization (WISH) using DIG-labelled anti-sense probes as previously described (22), with embryos younger than 22 hours post fertilization (hpf) dechorionated prior to fixing, or subjected to Sudan black staining (23). For probes that identified discrete cells (spi1, lyz, lcp1, mpeg1.1, mpx, csf3r, Sudan black), manual counting was performed across the embryos. For probes that showed staining in specific loci (hbbe1.1, rag1), the area was measured using ImageJ as the average from imaging both sides of the embryo (24). The region covering the majority of cells are shown in the respective figures.

Total RNA was extracted from zebrafish embryos and various adult tissues with either RNeasy Mini Kit (Qiagen) or Trisure (Bioline) according to the respective manufacturer’s protocol, depending on the nature of the sample. This was subjected to quantitative real-time reverse-transcriptase PCR (qRT²-PCR) with the primers listed in Table 1 . Data were normalized relative to actb and fold-change calculated using the ΔCT method (31).

Adult zebrafish kidney cells were prepared in ice-cold phosphate buffered saline supplemented with 2 mM EDTA and 2% (v/v) fetal calf serum and passed through a 40 μm filter. For flow cytometry analysis, samples were analyzed on a FACSCanto II. Dead cells and doublets were removed by gating with lymphoid, erythroid, myeloid and precursor cells identified in a side scatter/forward scatter (SSC/FSC) plot and neutrophils and macrophages via GFP fluorescence using fish on the transgenic Tg(mpx::GFP) and Tg(mpeg1.1::GFP) backgrounds, respectively. Smears of kidney cells and adult blood were prepared using a cytospin and stained with Giemsa prior to differential counting and imaging using a Leica DME stereomicroscope and Olympus SC50 camera.

A wounding assay was utilized to identify relative neutrophil and macrophage migration using the transgenic lines and fluorescence microscopy (32). Embryos at 5 days post fertilization (dpf) were anaesthetized using 5 μg/mL benzocaine in system water. After the embryos stopped moving, an excision was made at the tip of the caudal fin with a number 21 scalpel. Embryos were then placed into 12 well plates containing 0.003% (w/v) PTU in system water and imaged at 2 hour intervals in the presence of 5 μg/mL benzocaine before being placed in fresh PTU solution to remove the benzocaine. Embryos were maintained on a heat block set at 28.5°C for the 24 hours of the assay and imaged on an Olympus MVX10 microscope with a DP74 camera using CellSens Dimension 1.6 software (Olympus) with UV excitation and a GFP filter for analysis. Cell numbers in the whole embryo prior to wounding and at the injury site across the time-course were manually determined. Cell areas were quantified with ImageJ using images obtained at the injury site during the peak of cell migration. The net rate of cell migration was calculated from the peak cell number divided by the corresponding time post-wounding, and the net rate of cell egress from the reduction from the peak until 24 hours post-wounding (hpw). In other experiments embryos were injected at the 1 cell stage with csf3a mRNA or at 3 dpf with lipopolysaccharide (LPS) as described (33).

Statistical analyses were performed using GraphPad Prism version 8. Normality of data was tested using the D’Agostino-Pearson omnibus normality test and unequal variances using the F-test. For normally distributed data, the difference between two groups was analyzed for significance with a student t-test, with Welch’s correction if unequal variances were present. Where multiple t-tests were performed, the Bonferroni-Dunn method was used to correct for multiple comparisons. The difference between more than two groups was tested for significance using ordinary one-way ANOVA coupled with Tukey’s multiple comparison test. If data did not follow a normal distribution, the Mann-Whitney U test was performed. Survival and pathology incidence were visualized with a Kaplan-Meier graph, with the Gehan-Breslow-Wilcoxon test used to compare between different groups.

Protein sequences of human SOCS3 (NP_003946) as well as zebrafish Socs3a (NP_956244) and Socs3b (NP_998469) were obtained from GenBank and subjected to constraint-based multiple sequence analysis with COBALT (34).

Both zebrafish Socs3a and Socs3b proteins share the same domain structure as human SOCS3 with very good overall conservation ( Figure 1A ), including many stretches of identical residues ( Figure 1B ). However, Socs3b shows higher conservation across the protein ( Figures 1A, B ) and so was chosen for further study. The socs3b gene also has the same exon structure as SOCS3 (17), with exon 2 targeted using two guide RNAs (gRNAs) directed at sequences encoding the N-terminus and SH2 domain ( Figure 1C ). These were injected along with Cas9 mRNA into embryos, which were raised to adulthood and out-crossed with wildtype (WT) fish. Screening of the resultant F1 fish identified a single mutant allele that impacted the socs3b reading frame. The carrier of this allele was again out-crossed before in-crossing to produce homozygous F3 mutant fish. Sequence analysis revealed that this allele, designated mdu24, possessed a complex insertion/deletion at both gRNA sites ( Figure 1D ). The first caused a frameshift resulting in 41 de novo residues after proline at position 15 followed by a stop codon, with the second occurring after this point and so not impacting the protein encoded, which represented a severely truncated knockout (KO) version of the Socs3b protein, lacking the SH2 and SOCS box domains ( Figure 1E ).

The potential effects of socs3b ablation on primitive hematopoiesis were assessed using WISH with lineage-specific markers at 22 hpf. No difference was observed between WT and KO embryos with spi1b, a marker of early myeloid progenitors (25) ( Figures 2A–C ), or the pan-leucocyte markers lyz (35) ( Figures 2D–F ) and lcp1 (36) ( Figures 2G–I ). Similarly, there was no difference between WT and KO embryos with the macrophage marker mpeg1.1 (37) ( Figures 2J–L ). However, the neutrophil population, identified using mpx (38), was significantly increased in KO embryos ( Figures 2M–O ). The number of cells expressing the alternate neutrophil marker csf3r (33) were elevated but not to a statistically significant level ( Figures 2P–R ).

Early definitive hematopoiesis was assessed using WISH at 5 dpf. A significant increase in mpx⁺ cells was again observed in KO compared to WT embryos ( Figures 3A, B, E ). There was also a significant elevation in cells expressing lcp1 ( Figures 3C, D, F ) or csf3r ( Figures 3G–I ), but not mpeg1.1 ( Figures 3J–L ). Analysis of mature erythrocytes using hbbe1.1 (39) ( Figures 3M–O ) and T lymphocytes using rag1 (40) ( Figures 3P–R ) revealed no differences between WT and KO embryos. To further investigate the potential impact of socs3b on innate immune cells fish carrying the KO allele were separately out-crossed with two transgenic lines Tg(mpx::GFP) (20) and Tg(mpeg1.1::GFP) (21) to facilitate visualization of neutrophils and macrophages, respectively. This confirmed a significant elevation in the number of mpx⁺ neutrophils ( Figures 3S–U ) but not mpeg1.1⁺ macrophages ( Figures 3V–X ) in KO compared to WT embryos.

Further analysis of the Tg(mpx::GFP) embryos by light microscopy failed to detect any differences between WT and KO genotypes ( Figures 4A, B ), while fluorescence microscopy revealed similar distribution of neutrophils across the embryo ( Figures 4C, D ). To assess neutrophil maturation, 5 dpf embryos were stained with Sudan black as a marker of differentiation (23) ( Figures 4E, F ). The number of stained cells was again significantly higher in KO compared to WT embryos with no difference in staining intensity observed ( Figure 4K ), suggesting the increased neutrophils were mature. To further investigate neutrophil responses, lipopolysaccharide (LPS) was also injected into WT and KO embryos on the Tg(mpx::GFP) transgenic background to simulate bacterial infection (33). In WT embryos LPS induction was able to significantly increase the mpx⁺ neutrophil population, but this increase was not observed in KO embryos ( Figures 4G–J, L ).

Embryos derived from both transgenic lines were also subjected to a tail wounding assay at 5 dpf ( Figure 5A ). No difference was observed in the neutrophil response to injury between WT and KO embryos, with similar numbers and kinetics of mpx ⁺ cells observed ( Figures 5B, C ). However, when this data was normalized with respect to total mpx⁺ cells, a significantly lower neutrophil response to wounding in KO embryos compared to WT was apparent ( Figure 5D ). In contrast, a significantly larger and more sustained response was seen for mpeg1.1⁺ macrophages in KO compared to WT embryos ( Figures 5E, F ). There was no difference between WT and KO embryos in relative size ( Supplementary Figures 1A, B ) or egress rate ( Supplementary Figures 1C, D ) for mpx⁺ or mpeg1.1⁺ cells during wounding, but there was a significant increase in the rate of migration for mpeg1.1⁺ ( Supplementary Figure 1F ) but not mpx⁺ ( Supplementary Figure 1E ) cells in KO embryos. Gene expression analysis revealed a significant increase in the inflammatory markers cxc8la, il1b, il6 and tgfb1b (41, 42) in both WT and KO embryos after wounding but two markers of inflammatory macrophages ccr2 and cxcr3.2 (42) were significantly upregulated only in KO embryos ( Figure 5G ). Comparing gene expression between genotypes at the various timepoints revealed elevation of mpeg1.1, ccr2 and il6 basally and all genes by 8 hpw in KO compared to WT fish ( Supplementary Figure 2 ).

Adult KO zebrafish had a major sex bias, with females drastically underrepresented, meaning it was only feasible to characterize adult male KO zebrafish. These were significantly smaller than male WT zebrafish at 5 months post fertilization (mpf), but otherwise anatomically normal ( Figures 6A–C ). Gene expression analysis of the kidney, which is the equivalent of mammalian bone marrow with respect to adult hematopoiesis (43), revealed a significant and specific increase in the neutrophil marker mpx in socs3b KO compared to WT fish, but not markers of macrophages (mpeg1.1) (21), T cells (tcra, cd4, cd8) (44), B cells (cd79a) (45), NK cells (nkl1.3) (46), thrombocytes (tpor) (47), RBC (hbaa1.1) (47), or precursors (cmyb, scl, ikzf1, runx1, gata1a) (48–50) ( Figure 6D ). Forward and side scatter analysis of kidney cells also identified a significantly larger myeloid cell compartment in KO zebrafish ( Figures 6E–G ). Moreover, analysis of fish on the Tg(mpx::GFP) and Tg(mpeg1.1::GFP) backgrounds revealed a significant increase in kidney neutrophils ( Figures 6H–J ) – but not macrophages ( Figures 6K–M ) – in KO compared to WT fish. Analysis of kidney smears confirmed a specific increase in mature neutrophils in KO fish that were morphologically indistinguishable from WT ( Figures 6N–P ), but this increase was not reflected in blood smears in which neutrophil numbers were not significantly different between genotypes ( Figures 6Q–S ).

A significant proportion of KO zebrafish developed an eye pathology, with these fish eventually becoming emaciated ( Figure 7A ). Affected individuals displayed significant tissue growth on the surface of the eye, in most cases covering the pupil and resulting in the development of so-called ‘bug eye’ (51) ( Figures 7A–C ). This tissue contained both neutrophils and macrophages as revealed in Tg(mpx::GFP) ( Figure 7B ) and Tg(mpeg1.1::GFP) ( Figure 7C ) lines, respectively. This eye pathology was first evident at around 1 mpf and reached almost 30% by 4 mpf ( Figure 7D ), and strongly correlated with early mortality in these fish ( Figure 7E ). Gene expression analysis of whole eyes confirmed an upregulation of neutrophil (mpx) (10) and macrophage (mpeg1.1) (37) markers in symptomatic KO fish compared to WT fish, as well as one T cell marker (cd4) (52), although the other T cell marker (cd8) and those for B cells (ighm) (53) and NK cells (nkl.3) (46) were not significantly altered ( Figure 7F ). Several inflammatory macrophage markers (cxcr3.2, tnfa, il1b) (42) were upregulated, along with inflammatory cytokines il4, il6, il21 (54) and ifng (55), but not cxcr4b, ccr2, tgfb1b or csf3b ( Figure 7F ). The gene expression profiles of various tissues extracted from normal WT and both symptomatic and asymptomatic KO adults were also assessed ( Supplementary Figure 3 ). The neutrophil marker mpx was found to be significantly upregulated in the spleen and liver of symptomatic KO compared to WT, and compared to asymptomatic KO for liver, but was down-regulated in muscle ( Supplementary Figures 3A-C ). No significant change was observed in macrophage marker mpeg1.1 in any of these tissues ( Supplementary Figures 3A–D ). The lymphocyte marker cd8 was also significantly upregulated in both the liver and spleen of symptomatic and asymptomatic KO compared to WT, with cd4 upregulated in asymptomatic KO only for these tissues ( Supplementary Figures 3A, B ).