Introduction

In mammals, tumour necrosis factor (TNF, previously TNF-α (alpha); TNFSF2 genenames.org) is a key orchestrator of inflammation, among other processes, and is expressed by and has effect on several cell types, including immune cells. The main producers of and responders to TNF are monocytes and macrophages (Aggarwal, 2003; Beutler and Ceramit, 1989; Dubravec et al., 1990). When a cell responds to TNF signalling, an intracellular signalling cascade is activated leading to a broad range of cellular activities, including proliferation, survival, differentiation, immune mediators production or cell death (Baud and Karin, 2001; Horiuchi et al., 2010; Preedy et al., 2024). These seemingly opposite activities are essential for maintaining a delicate balance between protective inflammatory responses and excessive tissue damage. In fact, numerous studies describe how a tight regulation of TNF during the early phase of inflammation is required to control inflammation or infection, as opposed to pathology related to uncontrolled TNF expression and subsequent exacerbated innate immune responses (Aggarwal, 2003; Aggarwal et al., 2012; Beutler and Ceramit, 1989; Dubravec et al., 1990; Feldmann and Maini, 2003).

Despite the substantial progress been made in the characterization of mammalian TNF biology, the pleotropic nature of TNF during inflammation or infection in other vertebrate species remains elusive. Interestingly, while in the genome of mammals, birds, reptiles and amphibians, only one copy of TNF has been identified (Secombes et al., 2016), in the majority of bony fish (teleost) genomes analysed so far, two or more tnf homologues can be identified (Cui et al., 2020; Grayfer et al., 2008; Hong et al., 2013; Kadowaki et al., 2009; Kajungiro et al., 2015; Y. Li et al., 2021; Milne et al., 2017; Nascimento et al., 2007; Pleić et al., 2015; Saeij et al., 2003). They have been subdivided into two groups based on distinct structural and functional properties (Hong et al., 2013). Zebrafish (Danio rerio) possess two TNF homologs (Kinoshita et al., 2014), tnfa belonging to group II located on chromosome 19, and tnfb belonging to group I located on chromosome 15 (Hong et al., 2013), and will be referred to as tnfa and tnfb, or tnf when referring to both. While tnfa has been extensively studied in the context of wounding or regeneration (Keatinge et al., 2021; Miskolci et al., 2019; Nelson et al., 2013; Nguyen-Chi et al., 2017, 2015; Tsarouchas et al., 2018), and during various bacterial infections (Bernut et al., 2016; Kam et al., 2022; Lewis and Elks, 2019; Miskolci et al., 2019), tnfb may have important roles in inflammation, yet has remained largely unexplored. Although direct functional evidence is sparse, zebrafish larvae pre-injected with anti-Tnfb antibodies showed increased survival after Vibrio vulnificus infection (Li et al., 2021), suggesting a role for Tnfb during infection-induced inflammation, and that regulation of Tnfb expression may determine susceptibility or resistance to the infection. Whether zebrafish Tnfa (group II) and Tnfb (group I) have overlapping as well as distinct roles during inflammation or infection, and how these relate to the better-known pleiotropic functions of mammalian TNF, is a gap in knowledge that we set out to address in this study.

Zebrafish are well-studied model organisms and offer several advantages, including larval transparency, a well annotated genome amenable to genetic manipulation, availability of several transgenic lines marking innate immune cells and cytokine-expressing cells (Benard et al., 2015; Bertrand et al., 2010; Dóró et al., 2019; Ellett et al., 2011; Jacobs et al., 2021; Lawson and Weinstein, 2002; Page et al., 2013; Petrie-Hanson et al., 2009; Renshaw et al., 2006; White et al., 2008), altogether making them suited to study the role of tnfb during inflammation. Furthermore, larval zebrafish lack a mature adaptive immune system for the first 2-3 weeks of development (Lam et al., 2004; Langenau et al., 2004; Torraca and Mostowy, 2018), offering the opportunity to focus on inflammatory responses driven by innate immune cells, especially macrophages and neutrophils.

For Tnfa and Il1b, a proinflammatory role and association with inflammatory macrophages has been established in the context of wound healing and bacterial challenges (Hasegawa et al., 2017; Lewis and Elks, 2019; Nguyen-Chi et al., 2017, 2014; Ogryzko et al., 2019). tnfa positive macrophages were shown to be essential during wound healing (Nguyen-Chi et al., 2017), to be highly phagocytic and to exert antimicrobial activities (Bernut et al., 2016; Clay et al., 2008; Nguyen-Chi et al., 2015). In such inflammatory contexts, also neutrophils express il1b alongside macrophages (Ogryzko et al., 2019). The role of Tnfb however, remains unclear. Previous studies that investigated tnfb gene expression, showed that it was enriched in macrophages during the proinflammatory phase of the response to caudal fin fold amputation (Nguyen-Chi et al., 2015). However, which cell-types express tnfb, its constitutive expression, and expression kinetics during inflammation in vivo, are largely unexplored.

In this study, we generated and characterized novel promotor-driven transgenic lines marking tnfb-expressing cells. We focused on the early phase of inflammation triggered by two well-established inflammation models: wounding, after caudal fin fold amputation (Begon-Pescia et al., 2022; Nguyen-Chi et al., 2017, 2015; Sipka et al., 2022, 2021), or infection, after Escherichia coli (E. coli) injection (Nguyen-Chi et al., 2015), to define, in a cell type-specific manner, the expression kinetics of tnfb compared to those of tnfa and il1b. We report that mantle cells of neuromast constitutively express tnfb from 32 hours post-fertilization onwards. After injury, we show that a subset of macrophages expressed both, tnfa and tnfb, whereas a subpopulation of neutrophils expressed tnfb and not tnfa. Interestingly, after wounding, tnfb expression kinetics differ in macrophages and neutrophils, with tnfb expression detected in macrophages preceding its detection in neutrophils. When analysing the kinetics of expression of tnfb, tnfa and il1b after wounding, our data indicate that, at least in macrophages, a sequential expression occurs, with tnfb expressed first followed by tnfa and il1b. Altogether, we demonstrate that tnfb is expressed by various cell types and has distinct expression patterns from tnfa, and propose that by studying both TNF homologues, the zebrafish model is particularly suited to dissect the pleiotropic nature of TNF during inflammation and infections.

Material and methods

Zebrafish line and maintenance

Zebrafish were kept and handled according to standard protocols (Westerfield, 2000; zfin.org) and local animal welfare regulations of The Netherlands. Zebrafish larvae were raised in E3 medium (4.96 mM NaCl, 0.18 mM KCl, 0.44 mM CaCl₂ and 0.40 mM MgCl₂·6H₂O in demi water at pH 7.2) at 28°C.

The following zebrafish lines were used in this study: transgenic Tg(mpx:eGFP)ⁱ¹¹⁴ (Renshaw et al., 2006), Tg(mpeg1:eGFP)^gl22 (Ellett et al., 2011), Tg(mpeg1:mCherry-F)^ump2Tg (Nguyen-Chi et al., 2014), TgBAC(il-1b:eGFP)^sh445 (Ogryzko et al., 2019) referred to as Tg(il1b:eGFP), Tg(tnfa:eGFP-F)^ump5Tg (Nguyen-Chi et al., 2015), or crosses thereof. Two transgenic zebrafish lines express a farnesylated (membrane-targeting sequence) mCherry (mCherry-F) or eGFP (eGFP-F) under the control of the mpeg1.1 or tnfa promoter, respectively.

tnfb transgenic line construction

The tnfb promotor (gene: ENSDARG00000013598) was amplified from zebrafish genomic DNA using primers Ztnfb-P3 FW: 5’-TAGCCAAGCCTACTTGTTGC-3’ or Ztnfb-P7 FW: 5’-AGGGTGACGATAATGTCAAC-3’ and ztnfb_E1N RV: 5’-TAATAGCGGCCGCTTTCGTATCTCACCATGCTG-3’. The resulting fragments were phosphorylated using T4PNK, digested with NotI and cloned in a derivative of pBSKI2 vector, upstream of either farnesylated eGFP (eGFP-F) (Thermes et al., 2002) or mCherry (mCherry-F). The resulting plasmids harbour either a 3.2-kb or 6.2-kb fragment of the zebrafish tnfb promoter, and each include part of the first coding exon using the endogenous ATG codon of tnfb to drive the translation of eGFP-F or mCherry-F. The expression cassette is flanked by two I-SceI sites. Either of these constructs was co-injected in fertilized eggs with the enzyme I-SceI (New England Biolabs, R0694S). Developing embryos underwent caudal fin fold amputation at 3 dpf (days post-fertilization), and those with clearly visible green or red fluorescence were raised as putative founders. Their offspring was tested in a similar manner to establish the stable transgenic lines Tg(-3.2tnfb:eGFP-F)^ump21Tg and Tg(-6.2tnfb:mCherry-F)^ump22Tg, referred to as Tg(tnfb:eGFP-F) and Tgtnfb:mCherry-F).

Caudal fin fold amputation and E. coli injection of zebrafish larvae

Caudal fin fold amputation was performed on 3 dpf larvae as described previously (Begon-Pescia et al., 2022). In brief, the caudal fin of anaesthetised larvae was transected with a sterile scalpel, posterior to the muscle and notochord.

For infection studies, 3 dpf larvae were injected with 2x10³ CFU Escherichia coli K12 harbouring an E2-Crimson expression plasmid, from here on referred to as E. coli crimson. Bacteria were grown overnight at 37 °C on Luria Broth (LB) agar plates (1.5% BactoTM Agar (BD, 214010), 1% BactoTM Trypton (GibcoTM, 211705), 0.5% BactoTM Yeast extract (GibcoTM, 212750) and 0.5% NaCl (Merck KGaA, 1.37017) in demi water) supplemented with 50 μg/ml ampicillin (A9518, Sigma-Aldrich), and 0.25 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG, Promega, V395A). Bacteria were harvested from the plate, washed twice and resuspended in sterile PBS (Gibco, 10010-015) to measure the OD600 (Camspec M107 Spectrophotometer), next bacteria were resuspended in sterile 2% (w/v) polyvinylpyrrolidone in PBS (PVP, Sigma-Aldrich, PVP40-50G) solution. Anesthetized larvae were injected intramuscularly with 2x10³ CFU (in 1 nl) of E. coli crimson solution or 2% PVP alone.

In vivo imaging of zebrafish

Prior to imaging, zebrafish larvae were anaesthetised in 0.017% MS-222 (ethyl 3-aminobenzoate methanesulfonate, Sigma-Aldrich, A5040-25G) and embedded in 1% UltraPure^TM Low Melting Point Agarose (Thermo Fisher, 16520-050) at the bottom of a 35 mm FluoroDish^TM (World precision instruments, 13012026). Andor-Revolution spinning disk confocal (Yokogawa) on a Nikon Ti Eclipse PFS3 microscope, 40x (1.15 NA, 0.61-0.59 mm WD) WI objective, 20x (0.75 NA, 1.0 mm WD) objective were used with the following settings: dual pass 523/561: GFP excitation: 488 nm, emission: 510-540 nm, EM gain: 150-300 ms, digitizer: 10 MHz (14-bit); RFP excitation: 561 nm; emission: 589-628 nm, EM gain: 200-450 ms, digitizer: 10 MHz (14-bit); CY5 excitation: 640 nm; emission: 665-705 nm, EM gain: 200-450 ms, digitizer: 10 MHz (14-bit); BF DIC EM gain: 180-250 ms, digitizer: 10 MHz (14-bit). Typically for time-lapse videos, Z-stacks spanning 20-60 µm at 1 µm steps were acquired every 3-18 min. For individual z-stack acquisitions the 20x or 40x objective was used, as indicated by the figure legends.

For detailed imaging of neuromasts, a Leica SP8-DIVE multiphoton confocal microscope (Leica Microsystems) equipped with a near infrared femtosecond sapphire pulse laser and a 20x (0.75 NA, 1.0 mm WD) objective was used with the following settings: GFP excitation 900 nm, mCherry excitation 1050 nm using preset GFP or mCherry detection modules available in the Leica Application Suite software for DIVE spectral multiphoton detectors. The image scan speed was 4 µs/pixel with an image size of 512 × 512 pixels.

Images were manually analysed with ImageJ-FIJI (version 2.1.0, Schindelin et al., 2012). After amputation, the number of recruited macrophages, neutrophils and cytokine-expressing cells was quantified after brightness and contrast adjustment in ImageJ-FIJI for better visualizations. Cells were counted at the site of injury, defined as a 105 μm width region centred around the posterior tip of the notochord.

For total cell fluorescence acquisition after E. coli injection, transgenic zebrafish were positioned on preheated flat agarose plates (1% agarose in E3 medium containing 0.017% MS-222) and imaged with a Leica M205FA fluorescence stereo microscope (Leica Microsystems). The image acquisition settings were as follows: Zoom: 58-61x, Gain: 2.5, Exposure time (ms): 300 (BF)/1000 (GFP)/6000 (mCherry)/ 6500 (Crimson), Intensity: 200 (BF)/100 (GFP)/230 (mCherry)/ 250 (Crimson), Contrast: 255/255 (BF)/ 70/255 (GFP)/ 15/255 (mCherry)/ 20/255 (Crimson). Quantification of total cell fluorescence was performed using the free-form selection tool and by accurately selecting the area around the fluorescent reporter, selected areas were saved using the ROI manager tool. At later timepoints or for PVP-injected larvae the equivalent area positioned around the site of injection. Area integrated intensity and mean grey values of each selected area were acquired. Analysis was performed using the following formula (Fitzpatrick, 2014): Corrected Total Cell Fluorescence (CTCF) = integrated density – (area x mean grey background value). The mean grey background value was the average of three random areas outside of the fish selected in each image.

Fluorescence-activated cell sorting, isolation of mRNA and gene expression

Fluorescence-activated cell sorting (FACS) and mRNA isolation were performed as described previously (Begon-Pescia et al., 2022). Briefly, the protocol is divided in five steps: (1) caudal fin fold amputation; (2) Enzyme-free cell dissociation using sorting buffer (0.9x Dulbecco’s Phosphate-Buffered Saline without calcium and magnesium (Life Technologies, 14190094), with 2% heat inactivated foetal bovine serum (Gibco, 10500064) and 2 mM EDTA (Invitrogen, 15575-020) kept at 4°C; (3) sorting of eGFP⁺ and eGFP^- cells form Tg(tnfb:eGFP-F) or sorting of mCherry⁺ and mCherry^- cells from Tg(tnfb:mCherry-F) larvae (3 dpf, 2-3 hpa) with FACS Aria™ IIu (BD biosciences) (Supplementary Fig 1); (4) total RNA extraction from sorted cells using the RNeasy micro Kit (Qiagen, 74004); (5) cDNA synthesis from 10 ng of total RNA using M-MLV Reverse Transcriptase Transcription kit (Invitrogen, 28025-013). Real-time quantitative PCR (RT-qPCR) was performed on a LightCycler 480 system (Roche) using SYBR Green (SensiFAST™ SYBR^® No-ROX Kit, 98050) as reaction mix according to the manufacturers instruction and primers listed in table 1. Program: 5 min at 95°C, 45 cycles of denaturation 15 s at 95°C, annealing 10 s at 64°C, and elongation 20 s at 72°C. Expression levels were determined with the LightCycler analysis software (version-1.5.1). The relative expression level of a given mRNA was calculated using the formula 2^−ΔCt with housekeeping gene eef1a1l1 (previously ef1a) as internal reference.

Table 1.

Whole-mount in situ hybridization

A tnfb probe was amplified from total cDNA derived from 3 dpf zebrafish larvae. The following primers were used for PCR: Forward 1_tnfb 5’-GCAGCATGGTGAGATACGAA-3’ and Reverse 1_tnfb 5’-ACTCTCACTGCATCGGCTTT-3’. The 890 bp amplicon was ligated into a pCRII-TOPO plasmid (TOPO™ TA Cloning™ Kit, Invitrogen, 45-0640). Agarose-gel electrophoresis analysis was performed to confirm the size of the insert by digesting the tnfb-pCRII-TOPO plasmid with EcoRI-HF (New England Biolabs, R3101S), which recognition sites flank the PCR insert. Sequencing was performed by Eurofins genetics using their standard SP6 primer: 5’-CATTTAGGTGACACTATAG-3’. Similarly, a 786 nt mpx probe was generated using primers: Forward 1_mpx 5’-AAACAGCCAGCTCGAAAGAC-3’ and Reverse 1_mpx: 5’-CCGCAGTTCCTTTGCTAG-3’.

To generate RNA probes, the plasmids were linearized by restriction digestion with HindIII (Thermo Scentific, FD0504) or NotI-FD (Thermo Scentific, FD0593) to facilitate in vitro transcription of antisense or sense digoxigenin (DIG)-labelled (Roche, 11277073910) probes, using T7 or Sp6 RNA polymerase (Promega, T7: M0251S or Sp6: P1085 RNA polymerase kit) according to the manufacturer’s instruction. T7 promotor was used to generate antisense probe for tnfb, whereas SP6 promotor was used to generate sense probe for tnfb and antisense probe for mpx. Probe integrity was confirmed by agarose-gel electrophoresis.

Larvae were euthanized with an overdose of anaesthetic (0.4% MS-222), and fixed overnight in 4% paraformaldehyde (PFA) at 4°C, followed by three washes with phosphate-buffered saline (PBS) with 0.1% (v/v) tween 20 (PBT), followed by gradual dehydration by 5 min incubations in increasing concentrations of methanol (25%, 50%, 75%, 100%). Larvae were stored in 100% methanol at -20°C for at least one week. Larvae were gradually re-hydrated using decreasing concentrations of methanol (75%, 50%, 25%), followed by three washes in PBT. In situ hybridization on whole-mount larvae was performed as previously described (Thisse and Thisse, 2008), with adjusted enzymatic digestion of larvae. To preserve tissue structure of neuromasts a mild enzymatic digestion, with 2 μg/mL Proteinase K (Promega, MC5005) for 17 min at room temperature (21°C), was performed (Steiner et al., 2014). Digestion was stopped by 20 min incubation in 4% paraformaldehyde (PFA) at room temperature, followed by three washes in PBT, and 6h incubation in hybridization buffer (nuclease free water containing 50% formamide (Sigma, S417117), 5x saline-sodium citrate (SSC, Sigma, 118K8403), 500 µg/ml tRNA (Sigma, R8759_500UN), 50 µg/ml heparin (Sigma, H3393-50KU), 0.1% tween, and 1M citric acid (Sigma, 71402) to adjusted to pH 6). Prior to hybridization of larvae, the probe was linearized by 3 min incubation at 80°C, directly followed by 5 min incubation on ice. Overnight hybridization was at 65°C, per ten larvae 150 ng linearized probe in 200 µl was used.

Several stringent washing steps were performed at 65°C using freshly prepared solutions, containing various concentration of formamide, SSC and tween (as described step-by-step by Thisse and Thisse, 2008). Hereafter, larvae were incubated in blocking buffer (PBT with 2% normal sheep serum (Roche, 013-000-121) and 2 mg/ml Bovine Serum Albumin (Sigma, A3059) for 5h at room temperature. The transcript’s expression pattern was visualized using anti-DIG antibody conjugated to alkaline phosphatase (1:500, Roche, 11093274910) overnight at 4°C in the dark, followed by six washed in PBT for 15 min each. Next a series of four stringent washed in freshly prepared NTMT (Final concentration: 100 mM NaCl, 100 mM Tris-HCL pH 9.5, 50 mM MgCl2, 0.1% tween in nuclease free water), followed by incubation in nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate (NBT/BCIP, Roche, 11681451001) until sufficient colour developed. Reactions were terminated by washing larvae in PBT and incubation in 4% PFA for 30 minutes. Incubation with tnfb sense probe was carried out in parallel as a negative control, and mpx antisense probe hybridization was carried out as a positive control. Prior to imaging, larvae were cleared sequentially, first larvae were incubated for five minutes in increasing concentrations of ethanol (30%, 50%, 70%) in PBS, and twice incubated in 100% ethanol. Finaly, larvae were incubated in increasing concentrations of glycerol (30%, 50%, 80%) in PBS.

Images were acquired using a Zeiss Stemi 508 stereo microscope (Zeiss Microscopy), equipped with Moticam 10+ 10.0 MP colour camera (Motic) and Motic Images Plus software (version 3.0 ML). Alternatively, images were acquired using Leica DM6b automated upright digital microscope (Leica Microsystems), controlled by Leica LASX software (version 3.6.0.), equipped with a 20x (NA 0.8, DIC) short distance objective and DFC450C CCD colour camera (Leica Microsystems).

Exploration of public single-cell RNA sequencing datasets

This paper used publicly available single-cell RNA sequencing (scRNAseq) datasets, explored through their respective web-based visualization platforms: I) data of homeostatic neuromasts retrieved from Zebrafish Neuromast scRNAseq (Lush et al., 2019), II) data from neuromasts at multiple timepoints after hair cell ablation retrieved from Neuromast regeneration scRNAseq (Baek et al. 2022), and III) data of multiple zebrafish developmental stages (3 – 120 hpf) retrieved from Daniocell database (Farrell et al., 2018; Sur et al., 2023). the latter, was explored using the companion application Daniocell Desktop (v1.0.3, Evans and Farrell, 2025). Each dataset provided pre-processed and annotated scRNAseq data, allowing exploration of gene expression across defined cell clusters.

Statistical analysis

Statistical analyses of gene expression, cell number and total fluorescence data were performed in GraphPad PRISM 10 (GraphPad Software). The specific statistical tests used are indicated per figure legend. In all cases, *p<0.05 was considered significant.

Results

tnfb is constitutively expressed in neuromast cells during early development

In the present study, two transgenic reporter lines marking tnfb were generated: one in which a 3.2-kb portion of the predicted tnfb promotor drives farnesylated eGFP (eGFP-F) expression and a second in which a 6.2-kb portion of the predicted tnfb promotor drives farnesylated mCherry (mCherry-F) expression. When the constructs were independently injected into wildtype embryos, reporter expression was observed along the lateral line (in neuromast cells) of F0, suggesting that both promotors could drive fluorescent protein expression in unchallenged conditions. Five stable transgenic zebrafish lines (three Tg(tnfb:eGFP-F) and two Tg(tnfb:mCherry-F)) were recovered that retained the fluorescent pattern observed in neuromasts. Due to higher fecundity in Tg(-3.2tnfb:eGFP-F)^ump21Tg and Tg(-6.2tnfb:mCherry-F)^ump22Tg, these two lines were used hereafter. In crosses of Tg(tnfb:eGFP-F);Tg(tnfb:mCherry-F) transgenic lines, earliest expression of both transgenes was detected concomitantly around 32 hpf (Fig 1A). From 3 dpf onward, eGFP-F or mCherry-F signal was visible in the neuromasts along the entire length of the lateral line (Fig 1B-C). Further characterization of both transgenic lines revealed that fluorescent protein expression recapitulated the pattern of basal tnfb expression in 3 dpf larvae as assessed by in situ hybridization (Fig 1D-E (upper panels), Supplementary Fig 2), confirming constitutive expression of tnfb mRNA in neuromasts during development (Fig 1E). In situ hybridization using an antisense probe targeting myeloperoxidase (mpx) mRNA was used as a positive control for immune cell staining as it marks a population of neutrophils present in the caudal haematopoietic tissue (CHT) (Fig 1D-E (lower panel), Supplementary Fig 2).

Figure 1.

Neuromasts are organs of the lateral line, each consisting of approximately 60 cells and of a combination of different cell types (Lush et al., 2019). We investigated which sub-population(s) of neuromast cells expressed tnfb:eGFP-F or tnfb:mCherry-F by multiphoton microscopy. Expression was observed on the outer perimeter of the neuromast in both transgenics, in a dome-like or volcano-like shape typical of mantle cells (Fig 2, Supplementary Video 1). The cells in the core and top of the dome-like structure, composed of support and sensory hair cells, which extend through to the top of the dome, did not show tnfb:eGFP-F nor tnfb:mCh-F expression (Fig 2, Supplementary Video 1). The specific expression of tnfb:eGFP-F and tnfb:mCh-F in mantle cells agrees with data retrieved from two publicly available datasets reporting on RNA expression in single cells of homeostatic neuromast. There, tnfb is a distinguishing marker of mantle cells ((Lush et al., 2019;Baek et al., 2022), Supplementary Fig 3). In addition to developmental expression, a transient increase in tnfb mRNA expression is observed following hair cell ablation ((Baek et al. 2022), Supplementary Fig 3D-F). This suggest that expression is not only developmentally regulated, but also responsive to wounding. Altogether these data suggest that both newly generated transgenic lines show a similar pattern of tnfb:eGFP-F and tnfb:mCh-F expression during development, that tnfb is constitutively expressed in neuromast mantle cells, and that it can be further induced in response to wounding.

Figure 2.

tnfb is transiently induced after caudal fin fold amputation

After having assessed that both transgenic lines mark constitutive tnfb expression in mantle cells of neuromasts, next, we assessed whether tnfb expression could be triggered in other cells during inflammation induced after caudal fin fold amputation. Using this wounding method, in both transgenic lines (Tg(tnfb:eGFP-F) and Tg(tnfb:mCherry-F)) or crosses thereof, cells expressing the transgene migrated towards or started to express the transgene near the site of injury (Fig 3A, Supplementary Video 2), matching the in situ hybridization pattern observed, showing tnfb-expressing cells localized at the wound site (Fig 3B, Supplementary Fig 4). Upon closer inspection, we found that cells at the wound site (within the white box) were positive for both reporter proteins (eGFP⁺mCherry⁺), while some cells were eGFP⁺mCherry^-, eGFP^-mCherry⁺ were never observed. Based on time-laps acquisitions we recorded cell motility over a period of 10 hours post-amputation (hpa); eGFP⁺mCherry⁺ cells were highly motile and migrated towards the wound, while in comparison, eGFP⁺mCherry^- cells showed minimal motility, remaining stationary while undergoing occasional changes in cell shape, which remained epithelial-like (Supplementary Video 2). Next, the number of cells expressing the transgene(s) at the wound site was quantified at different timepoints after amputation of Tg(tnfb:eGFP-F;tnfb:mCherry-F) larvae (Fig 3C-D). Timelapse microscopy over a period of 10h (Fig 3C), showed that in intact fins, the number of eGFP⁺mCherry⁺ or eGFP⁺mCherry^- cells remained constant at all investigated time points, while eGFP^-mCherry⁺ cells were never observed. In amputated fins, the number of eGFP⁺mCherry⁺ cells increased significantly from 2h onwards compared to the same cell type in the intact group, while the number of eGFP⁺mCherry^- cells did not change. Again, no eGFP^-mCherry⁺ cells were observed. Looking at cell number within individual larvae in the same experiment as in Fig 3C (Fig 3D), we observe that at the given timepoints after amputation, eGFP⁺mCherry⁺ cells are the most abundant cell type at the wound site. The combined observations on cell motility, morphology and kinetics of recruitment at the wound site, suggest that non-motile eGFP⁺mCherry^- cells with epithelial cell morphology are likely keratinocytes, whereas eGFP⁺mCherry⁺ are likely innate immune cells being recruited to the wound site. Thus, when considering motile cells, both transgenic lines mark similar innate immune cell populations, which increase their expression of tnfb after wounding. The Tg(tnfb:eGFP-F) line also marks a small subpopulation of epithelial cells, likely keratinocytes which, based on motility and morphology, can be distinguished from motile innate immune cells.

Figure 3.

To further confirm that eGFP⁺ or mCherry⁺ cells express tnfb, FACS analysis (Supplementary Fig 1) followed by RT-qPCR was performed on pools of Tg(tnfb:eGFP-F) or Tg(tnfb:mCherry-F) zebrafish larvae (3 dpf) 2-3 hours post-amputation (hpa) (Fig 3E-F). eGFP⁺ or mCherry⁺ cells were enriched in tnfb transcripts and to a lesser extend in tnfa transcripts (Fig 3E-F left panel), confirming that in both transgenic lines tnfb expression is induced in reporter protein-positive cells early during inflammation. Next, to gain insights into the types of cells expressing tnfb, expression of typical macrophage (mpeg1.1, mfap4.2) and neutrophil (mpx) markers was assessed in the same sorted cells (Fig. 3E-F). eGFP⁺ or mCherry⁺ cells were not only enriched in tnfb transcripts, but also in mpeg1.1, mfap4.2 (Fig 3E-F middle panel), and mpx transcripts (Fig 3E-F right panel). Altogether, these data suggest that, after wounding, macrophages and neutrophils are among the immune cells expressing tnfb. The specific expression of tnfb in innate immune cells agrees with data retrieved from the publicly available Daniocell dataset reporting on RNA expression in steady-state developing zebrafish ((Farrell et al., 2018; Sur et al., 2023), Supplementary Fig 5) and, within the dataset we additionally explored tnfa and il1b transcript expression. tnfb and il1b transcripts were expressed in subpopulations of neutrophils, macrophages, microglia and hematopoietic stem cells (myeloid), while tnfa transcripts were barely detectable, except in a few macrophages (Supplementary Fig 5A). In these tnfa⁺ macrophages coexpression of tnfb or il1b transcripts occurred (Supplementary Fig 5B left-middle panel). The neutrophil population in general, was predominantly composed of il1b⁺tnfb^- cells, with a small subpopulation il1b⁺tnfb⁺ cells. In nearly all tnfb⁺ neutrophils, macrophages, microglia and hematopoietic stem cell (myeloid), co-expression of il1b was observed (Supplementary Fig 5B right panel). Constitutive expression of tnfb could be detected in the lateral line cell cluster, however, the dataset did not facilitate distinction of mantle cells from other cell types within neuromast as separate populations (Supplementary Fig 5C). Prior to the emergence of any of the aforementioned immune cell types, tnfb was abundantly expressed in the periderm (Supplementary Fig 5C), suggesting that tnfb may play a role in early development. However, once innate immune cells are formed, they are the primary source of tnfb. Together, the data demonstrate that innate immune cells are the main cell sources of tnfb during development (Daniocell dataset) and in response to wounding (this study), thereby confirming that the motile tnfb-expressing cells recruited to the wound are indeed innate immune cells, and likely co-express tnfa and/or il1b. To investigate whether tnfb expression in innate immune cells can be triggered not only after wounding but also after infection, intramuscular injection of E. coli was performed in 3 dpf larvae and PVP injection served as no bacteria control (Fig 3G). To quantify tnfb:eGFP-F and tnfb:mCh-F expression in innate immune cells recruited to the site of injection, the injection area was carefully selected (Fig 3H, area within dashed line) and neuromast were excluded from total fluorescent analysis. In both, PVP and E. coli injected larvae, tnfb:eGFP-F and tnfb:mCh-F were induced at 2 hours post injection (hpi), with the total eGFP and mCherry fluorescence being higher in the E. coli compared to the PVP injected group (Fig 3H). This data suggests that in the context of a bacterial infection, tnfb is transcriptionally induced during the early, proinflammatory phase, and that both reporter lines can be used to evaluate tnfb expression during infection. Altogether, during early zebrafish development innate immune cells constitutively express tnfb, and its expression can be induced in these cells after wounding or E. coli injection. When considering tnfb expression in (motile) immune cells, both transgenic lines recapitulate transcriptional activation of tnfb. The Tg(tnfb:eGFP-F) line also marks a population of non-immune cells, likely keratinocytes, which is not marked by the Tg(tnfb:mCherry-F) line, but can be distinguished from immune cells base on their morphology and motility. Combined, these transgenic lines are new tools, useful to study tnfb kinetics of expression and cell sources in vivo during inflammation.

A subset of macrophages and of neutrophils express tnfb after wounding

The kinetics of tnfb expression were tracked in macrophages and neutrophils over a period of 6h after amputation, and the total number of macrophages, neutrophils and tnfb-expressing cells, was quantified at the 6h time point.

In intact caudal fins of Tg(mpeg1:mCherry-F;mpx:eGFP) larvae, macrophages (mCherry⁺) and neutrophils (eGFP⁺) were present in small number; after amputation, both significantly increased, with macrophages being generally more numerous than neutrophils (Fig 4A-B). When assessing tnfb expression in macrophages, in intact fins of Tg(tnfb:mCherry-F;mpeg1:eGFP) larvae, basal expression was observed in the majority of macrophages (Fig 4C-D); after amputation, the number of tnfb:mCh-F-expressing macrophages increased and a low number of mpeg1:eGFP^-tnfb:mCh-F⁺ cells (6.73% of tnfb⁺ cells) were also observed (Fig 4C-D). When looking at the kinetics of recruitment of tnfb:mCh-F-expressing macrophages (Supplementary Video 3), mpeg1:eGFP⁺tnfb:mCh-F ⁺ cells were observed as early as 1 hpa, the earliest timepoint assessed, whereas mpeg1:eGFP ^-tnfb:mCh-F ⁺ cells were not observed until 4 hpa.

Figure 4.

When assessing tnfb expression in neutrophils, in intact fins of Tg(tnfb:mCherry-F;mpx:eGFP) larvae, none of the neutrophils were tnfb:mCh-F ⁺ at the selected 6h timepoint, whereas after amputation, out of the total neutrophils recruited to the wound, a small subpopulation (8.75%) expressed tnfb:mCh-F (Fig 4E-F), suggesting that the small population of mpeg1:eGFP ^- tnfb:mCh-F ⁺ cells observed in Fig 4C-D, were likely neutrophils. When looking at the kinetics of tnfb expression in neutrophils, mpx:eGFP⁺tnfb:mCh-F⁺ cells were not observed at the wound site before 3 hpa (Supplementary Video 3), showing that neutrophils express tnfb:mCh-F to a detectable level later than macrophages, as mpeg1:eGFP⁺tnfb:mCh-F⁺ cells were observed from 1 hpa onwards. Together with our gene expression analysis (Fig 3E-F), these data show that after amputation both, macrophages and neutrophils, express tnfb, with macrophages being the main producers as tnfb-expressing macrophages were more numerous than tnfb-expressing neutrophils, and that tnfb expression was triggered earlier in macrophages than in neutrophils.

To compare cellular expression of tnfb to that of tnfa or il1b, we selected the 6 hpa timepoint, the time at which expression of all three transgenes is significantly increased after amputation. At this timepoint the total number of cytokine-expressing cells and the proportion of those that were macrophages, was quantified. To this end, we used double transgenic lines with macrophages in red and il1b or tnfa in green: Tg(tnfa:eGFP-F;mpeg1:mCherry-F) or Tg(il1b:eGFP;mpeg1:mCherry-F). When considering tnfa (Fig 4G-H), in the intact fin of Tg(tnfa:eGFP-F;mpeg1:mCherry-F) few macrophages were present, with the majority showing basal tnfa:eGFP-F expression. After amputation the number of recruited macrophages increased, with the majority being tnfa:eGFP-F⁺; only occasionally macrophages that were tnfa:eGFP-F^- were observed (1.5%). Mpeg1:mCh-F ^-tnfa:eGFP-F⁺ cells were not observed. When considering il1b (Fig 4I-J), in intact fins basal expression of il1b:eGFP was observed in Tg(il1b:eGFP;mpeg1:mCherry-F). After amputation the majority of recruited macrophages were il1b:eGFP⁺ although a subpopulation of mpeg1:mCh-F ^-il1b:eGFP⁺ cells (12.7% of total il1b⁺ cells) could be identified. Altogether, these data confirm that in response to injury tnfb, tnfa, and il1b are mainly expressed by macrophages. While tnfa is expressed exclusively by macrophages, tnfb and il1b are expressed by macrophages and also by a subpopulation of neutrophils.

Kinetics of tnfb, tnfa and il1b expression after amputation

After having characterised which innate immune cells express tnfb, tnfa and il1b, and having shown cell-specific expression kinetics of tnfb (Fig 4, Supplementary Video 3), we next characterised the kinetics of expression of all these transgenes and assessed whether they are expressed concomitantly or sequentially. To this end, crosses of lines with tnfb-expressing cells in fluorescent red and tnfa- or il1b-expressing cells in fluorescent green, Tg(tnfa:eGFP-F);Tg(tnfb:mCherry-F) or Tg(il1b:eGFP);Tg(tnfb:mCherry-F) were used. We imaged 3 dpf larvae over a period of 9h after amputation; to aid cell tracking and cell counting, images were acquired every 6 minutes. Owing to the availability of transgenic lines marking tnfa or il1b only in fluorescent green, it was not possible to assess co-expression of tnfa and il1b, or co-expression of all three cytokines within a cell.

When considering the kinetics of tnfb and tnfa expression (Fig 5A and 5C) in intact fins, only a few motile cells were observed and, if present, they were tnfa:eGFP-F^-tnfb:mCh-F⁺. In amputated fins, the majority of transgene-expressing motile cells were tnfa:eGFP-F⁺tnfb:mCh-F ⁺ and from 3 hpa onward their number significantly increased compared to the same cells type in the intact fins. A population of tnfa:eGFP-F^-tnfb:mCh-F⁺ cells was also identified, most prevalent at 2 hpa. When these cells were tracked (Supplementary Video 4), we observed that many transitioned into tnfa:eGFP-F⁺tnfb:mCh-F⁺ cells. This pattern of cytokine expression, combined with our observation that tnfa:eGFP-F is exclusively expressed by macrophages (Fig 4G-H), suggests that these are macrophages in which tnfb is expressed prior to tnfa. Conversely, tnfa:eGFP-F^-tnfb:mCh-F⁺ cells present after 4 hpa remained tnfa:eGFP-F^-tnfb:mCh-F⁺. tnfa:eGFP-F⁺tnfb:mCh-F^- cells were not observed.

Figure 5.

When considering the kinetics of tnfb and il1b expression (Fig 5B and 5D) in intact fins, only a few transgene-expressing motile cells were observed, the majority of which were il1b:eGFP^- tnfb:mCh-F ⁺, and a few were il1b:eGFP⁺tnfb:mCh-F ⁺. il1b:eGFP⁺tnfb:mCh-F ^- cells were not observed. After amputation, between 1-3 hpa, the majority of cells present at the wound were il1b:eGFP^-tnfb:mCh-F⁺ cells. Many of these cells transitioned into il1b:eGFP⁺tnfb:mCh-F ⁺ cells (Supplementary Video 4), the main population present by 4 hpa onwards. From 6 hpa onwards, a small population of il1b:eGFP⁺tnfb:mCh-F^- cells is present at the wound. Together with macrophages as the main cell type expressing il1b:eGFP (Fig 4J), this suggests that the largest population, il1b:eGFP⁺tnfb:mCh-F⁺ cells, is mostly composed of macrophages in which tnfb expression is detected earlier than il1b expression. Altogether, this suggests that the temporal dynamic of cytokine expression in macrophages is sequential, with of tnfb first, followed by tnfa and il1b.

tnfb expression by innate immune cells following E. coli injection

After having established that tnfb:eGFP-F and tnfb:mCherry-F are induced during the early phase of E. coli-induced inflammation (Fig 3G-H), we investigated whether tnfb is expressed with tnfa and il1b. To this end, transgenic larvae (3 dpf) were injected intramuscularly with E. coli or with PVP (control) and imaged at 6 hours post injection (hpi). Since at this stage it is challenging to manually count cells due to recruited cells crowding the injection area total fluorescence quantification instead of single cell count was performed. We made use of Tg(tnfa:eGFP-F);Tg(tnfb:mCherry-F) or Tg(il1b:eGFP);Tg(tnfb:mCherry-F) larvae. Based on total fluorescence of tnfa:eGFP-F, il1b:eGFP as well as tnfb:mCh-F, their expression increased significantly in the E. coli-injected compared to their corresponding PVP-injected group (Fig 6A-D). These data indicate that, like tnfa and il1b, tnfb is also induced following E. coli injection.

Figure 6.

Next, we assessed which innate immune cells expressed tnfb:mCh-F, tnfa:eGFP-F, or il1b:eGFP at 2.5 hours post injection with PVP or E. coli. In Tg(mpeg1:mCherry-F;mpx:eGFP) larvae, both macrophages (mCherry⁺) and neutrophils (eGFP⁺) were observed at the injection site (Fig 7A). When assessing tnfb expression, after PVP injection, tnfb:mCh-F expression was observed in macrophages (mpeg1:mCh-F⁺) (Fig 7B, Supplementary Video 5). When using the Tg(tnfb:mCherry-F;mpx:eGFP) line, no neutrophils were observed which express tnfb:mCh-F after PVP injection, whereas tnfb:mCh-F⁺ neutrophils could be observed after E. coli injection (Fig 7C). When assessing tnfa expression by macrophages, in both injected groups, mpeg1:mCh-F⁺tnfa:eGFP-F⁺ cells were observed (Fig 7D, Supplementary Video 5). When assessing il1b expression by macrophages, mpeg1:mCh-F⁺il1b:eGFP⁺ were observed after PVP injection, whilst after E. coli injection both mpeg1:mCh-F⁺il1b:eGFP⁺ and mpeg1:mCh-F ^-il1b:eGFP⁺ cells were observed (Fig 7E). This suggest that in response to PVP or E. coli, macrophages are the cells expressing tnfb, tnfa and il1b. In addition to macrophages, a subset of neutrophils expresses tnfb after E. coli injection.

Figure 7.

Combined, these data are consistent with and further extend our findings that during the early phase of inflammation tnfb is induced and modulated in vivo, in subpopulations of macrophages and neutrophils.

Discussion

Despite zebrafish possessing two homologues of human TNF, tnfa and tnfb (Kinoshita et al., 2014), research has largely focussed on tnfa, leaving tnfb underexplored. In this study, we generated two tnfb reporter lines and used them to identify tnfb-expressing cells and to describe tnfb expression kinetics in vivo during inflammation. We report on the constitutive tnfb expression in neuromast mantle cells, and inducible tnfb expression in innate immune cells during inflammation induced by caudal fin fold amputation and E. coli injection. By combining our newly generated tnfb reporters with available tnfa or il1b reporter lines and high-resolution microscopy, we report that macrophages are the main cells expressing tnfa, tnfb, and il1b, while neutrophils can express tnfb and il1b, but did not tnfa in response to wounding or to bacterial injection. We also show that the timing of tnfb expression differs between macrophages and neutrophils. Together, these results reveal cell type-specific expression of tnfb and distinct expression kinetics of tnfa and tnfb, setting the stage for future studies in which both homologs should be investigated. This will be essential to fully capture their overlapping as well as distinct roles during inflammation or infection, and how these relate to the better-known pleiotropic nature of mammalian TNF.

Here, we developed two transgenic reporter zebrafish lines (Tg(tnfb:eGFP-F) and Tg(tnfb:mCherry-F)); one, in which a 3.2-kb portion of the predicted tnfb promotor drives farnesylated eGFP (eGFP-F) expression and a second in which a 6.2-kb portion of the predicted tnfb promotor drives farnesylated mCherry (mCherry-F) expression. Direct comparison of fluorescent cells in crosses of these novel reporter lines, showed that both lines report constitutive expression of tnfb in neuromast mantle cells and inducible transcription of tnfb in innate immune cells during inflammation. Likely due to missing regulatory elements in the shorter promotor or an insertional effect, the Tg(tnfb:eGFP-F) line also marks a small subpopulation of keratinocyte-like cells that, based on motility and morphology, are easily distinguishable from motile innate immune cells. Nevertheless, eGFP-F and mCh-F expression kinetics were similar in immune cells. Together, Tg(tnfb:eGFP-F) and Tg(tnfb:mCherry-F) are novel complementary tools for characterising similarities and differences between tnfa and tnfb expression, in a range of cell types and inflammatory contexts.

By using two well-established inflammation models, we were able to contextualize tnfb expression and the immune cells involved. When focusing on macrophages in larval zebrafish, we report co-expression of tnfb, tnfa and il1b during the early phase of inflammation. These data agree with several previous studies reporting expression of the same cytokines in tnfa:eGFP-F⁺ sorted macrophages 6h after amputation (Nguyen-Chi et al., 2015), in RNA sequencing (RNAseq) data of pools of sorted macrophages after M. marinum infection (Rougeot et al., 2019), in single-cell RNA-seq (scRNAseq) data of macrophages 3h after amputation (García-López et al., 2023), and in gene expression and RNAseq data of FACS-sorted macrophages 2h after amputation (Begon-Pescia et al., 2025). These studies consistently point towards tnfb also being expressed by macrophages, but did not study cytokine expression kinetics during inflammation, as they analysed a limited (2-3) number of timepoints, and at each timepoint, individuals were pooled. In this study, leveraging the availability of tnfa:eGFP-F and il1b:eGFP transgenic lines, we show that tnfb:mCh-F⁺ cells can transition to tnfa:eGFP-F- or il1b:eGFP-expressing cells. Owing to the availability of transgenic lines marking tnfa or il1b only in fluorescent green, it was not possible to assess co-expression of tnfa and il1b, or co-expression of all three cytokines within a cell. However, our observation of distinct expression kinetics of the aforementioned cytokines combined with our observation that macrophages are the main tnfb-expressing cells after amputation, show that the temporal dynamics of the cytokines of interest, at least in macrophages, are sequential with expression of tnfb, followed by that of tnfa and il1b.

While we observed that tnfa:eGFP-F was expressed exclusively by macrophages, tnfb:mCh-F and il1b:eGFP were expressed by macrophages and by other cells. Others demonstrated that neutrophils indeed express il1b after amputation and during Mycobacterium marinum infection (Ogryzko et al., 2019). To our knowledge, this is the first report showing in vivo expression of tnfb by neutrophils and opens the possibility to use tnfb as a marker of neutrophil heterogeneity during inflammation. In mammals, it is increasingly appreciated that neutrophils occur with different activation states (Grieshaber-Bouyer et al., 2021; Rosales, 2018), among which a proinflammatory TNF-expressing subset has been observed (Khoyratty et al., 2021). tnfb expression as potential marker of neutrophils activation in response to amputation is reinforced by a study in larval zebrafish showing that, based on a photo-conversion lineage-tracing approach combined with single-cell RNAseq analysis, both rostral blood island- and caudal hematopoietic tissue-derived neutrophils are recruited to the wound and are enriched in tnfb (García-López et al., 2023). This suggests that tnfb expression is wound context-specific and independent of ontogeny. Furthermore, enrichment of tnfb in wound-responsive neutrophils, is conserved across developmental stages as demonstrated by heart-wound responsive neutrophils of adult zebrafish being enriched in tnfb (Wei et al., 2023). Based on co-expression with il1b, phagosome-related genes ncf1 and ncf2, and chemotaxis genes cxcr1, atf3 and illr4, it was suggested that these subsets of neutrophils promote inflammation by increasing recruitment and retention of neutrophils at the wound (Wei et al., 2023). Together, this supports that tnfb expression reflects a proinflammatory neutrophil activation state in larval and adult zebrafish and, when used as a marker, may provide novel insights into neutrophils heterogeneity during inflammation.

We observed that tnfb:mCh-F expression was detected earlier in macrophages than in neutrophils involved in the same inflammatory response. Cell-specific kinetics of tnfb expression may indicate that tnfb expressed by macrophages has a role in inflammation initiation, while tnfb expressed by neutrophils has a role during inflammation amplification. In mammals, TNF signalling occurs via TNFRSF1A (previously TNFR1) or TNFRSF1B (previously TNFR2) (Horiuchi et al., 2010). TNFRSF1A is characterized by an intracellular death domain, which eventually leads to proinflammatory or apoptotic signals, while TNFRSF1B does not harbour an intracellular death domain and instead leads to the induction of proinflammatory and survival signals (Holbrook et al., 2019; Preedy et al., 2024; Zelová and Hošek, 2013). Zebrafish also possess Tnfrsf1a and Tnfrsf1b. Similar to mammalian TNFRSF1 expression patterns, also in zebrafish, tnfrsf1a is broadly expressed across several cell types and tissues, whereas tnfrsf1b has limited expression, mostly restricted to immune cells (Smith et al., 1994). These expression patterns are consistent with those retrieved from publicly available RNAseq datasets of steady-state developing zebrafish larvae (Farrell et al., 2018; Lange et al., 2024; Sur et al., 2023). Although it is likely that distinct Tnf-Tnfr interactions contribute to the pleiotropic downstream effects attributed to Tnf, ligand-receptor preferences or how receptor expression is modulated on different cell types during inflammation, remain to be elucidated.

Besides innate immune cells, we (this study) and others (Baek et al., 2022; Lush et al., 2019) found that neuromast mantle cells constitutively express tnfb. Based on hair cell ablation experiments, it was shown that in mantle cells tnfb can be further induced in response to wounding (Baek et al., 2022). Combined, this suggests a role for tnfb in neuromast maintenance and/or a protective role during the response against environmental stimuli. From a neuromast maintenance perspective, Tnfb could be involved in directing hair cell polarity. Zebrafish modified to constitutively express Tnfrsf1b show asymmetry between rostrad- and caudad-oriented hair cells, whereas these are symmetrically distributed in steady-state neuromasts (Jacobo et al., 2019). Another possibility is that mantle cells are thought to serve as stem cells to various cell types within neuromast. In fact, a recent study has shown that maintenance of their stem cell identity is dependent on another TNF superfamily member Tnfsf10 (previously Trail) (Fan et al., 2025), but whether Tnfb (Tnfs2) plays a role in stem cell maintenance remains to be established.

Examination of tnfb expression in publicly available scRNAseq datasets of steady-state developing zebrafish (Farrell et al., 2018; Sur et al., 2023) revealed its expression in periderm prior to immune cell development. Initially, tnfb expression was suggested to be a hallmark during the gastrulation phase of zebrafish development (Ito et al., 2008). This was plausible in light of mammalian TNF playing a role during organogenesis (You et al., 2021). However, others also found that tnfb expression during this early developmental timeframe is derived from periderm rather than the inner cell mass (Hoijman et al., 2021) suggesting a non-developmental role for tnfb. Periderm, otherwise known as the outermost protective layer of developing larvae (Liu et al., 2020), constitutively expresses tnfb at least until 5 dpf (Farrell et al., 2018; Hoijman et al., 2021; Sur et al., 2023). The widespread constitutive expression of tnfb in periderm, in mantle cells and in cells lining the exterior of the developing larvae, suggests a putative role for Tnfb in the response to stimuli from the aquatic environment.

In addition to constitutive tnfb expression, transient upregulation of tnfb, but not tnfa, was shown in mantle cells after hair cell ablation (Baek et al., 2022), again suggesting that Tnfb in particular may be involved in the response to environmental stimuli or wounding. When considering bacterial challenges, others have shown that tnfb was triggers upon both, immersion and injection of 3 dpf larvae with Vibrio parahaemolyticus, while tnfa was only enriched in injected larvae (Guo et al., 2021). It is possible that after immersion and injection different cell types are responsible for tnfb expression, with cells lining the exterior of the fish playing a main role after immersion and innate immune cells playing a role after injection.

Interestingly, besides zebrafish (cyprinid fish), differential expression of group I and group II Tnf may also occur in meagre, belonging to perciform fish. Like zebrafish, meagre also possess two TNF homologs called Tnf1 (group I) and Tnf2 (group II) (Milne et al., 2017). In vitro stimulation of primary gill cultures and intestinal cell suspension with several pathogen-associated molecular patterns (PAMPs) elicited increased expression biased towards Tnf1 (group I) rather than Tnf2 (group II). In contrast, stimulation of primary head kidney cells or splenocytes elicited a relatively large increase in expression for Tnf2 (Milne et al., 2017). Similar results were observed in other teleost species (Cui et al., 2020; Hong et al., 2013; Kadowaki et al., 2009; Kajungiro et al., 2015; Pleić et al., 2015; Zou et al., 2002). Altogether, this not only confirms that group I and II Tnf in fish are differentially regulated, but is also indicative of a possible functional diversification between the two Tnf groups.

In conclusion, we report on the constitutive tnfb expression in neuromast mantle cells, and on the inducible tnfb expression in subpopulations of macrophages and of neutrophils after wound- and bacterial-induced inflammation. tnfb kinetics of expression are cell-dependent, with expression in macrophages occurring prior to that in neutrophils. We also show that, at least in macrophages, tnfb:mCh-F, tnfa:eGFP-F, and il1b:eGFP are expressed sequentially, indicative of a differential role of these cytokines. Altogether, we propose that with the addition of Tg(tnfb:eGFP-F) and Tg(tnfb:mCherry-F) reporter lines, the zebrafish model is especially suited to dissect the pleiotropic nature of Tnf during infection or inflammation.

Supplementary figures

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.

Data availability

Source data files are provided for figures 3 (C,D,E, F, H), 4 (B,D,F,H,J), 5 (A,B), 6 (B,D). Supplementary figure 2, provides raw data for main figure 1D; Supplementary figure 4, provides raw data for main figure 3B; Supplementary Video 1, provides raw data for main figure 2; Supplementary Video 2, provides raw data for main Figure 3A; Supplementary Video 3, provides raw data for main figure 4C,E; Supplementary Video 4, provides raw data for main figure 5C,D; Supplementary Video 5, provides raw data for main figure 7B,D; Previously published datasets Zebrafish Neuromast scRNA-seq, Neuromast Regeneration scRNA-seq, and Daniocell provide raw data for Supplementary figure 3 (A, B, C), Supplementary figure 3 (D, E, F), and Supplementary figure 5.

Acknowledgements

This work was supported by the WIAS PhD grant from Wageningen University and by the European Union’s Horizon 2020 research and innovation programme through the Marie-Curie Innovative Training Network INFLANET (955576). The authors like to thank the CARUS Aquatic Research Facility of Wageningen University for fish rearing and husbandry, and Wageningen Light Microscopy Centre and Microspectroscopy Research Facility, and the Aquatic model facility ZEFIX from LPHI, University of Montpellier and the imaging facility BioCampus Montpellier Ressources Imagerie (MRI), member of the national infrastructure France-BioImaging supported by the French National Research Agency (ANR-10-INSB-04, “Investments for the future”).

Additional files

Supplementary Video 1. 3D rendering of tnfb:mCh-F-expressing cells within a neuromast. Representative neuromast from Tg(tnfb:mCherry-F) generated as 3D project in ImageJ-FIJI. The rotating projection shows that mCherry⁺ cells form a hollow dome-like shape typical of neuromast mantle cells.

Supplementary Video 2. Tg(tnfb:eGFP-F) and Tg(tnfb:mCherry-F) report inducible tnfb-expression in the same motile (immune) cells. Caudal fins of 3 dpf Tg(tnfb:eGFP-F;tnfb:mCherry-F) larvae were amputated or left intact (control) and images were acquired over a period of 10 hpa as a 21 µm z-stack at a 18-minutes interval with an Andor spinning disk confocal microscope at 20x magnification. Representative time-lapse of the maximum projections rendered at 5 frames per second are shown. Please note that, in control (left time-lapse) and in amputated (right time-lapse) fins, tissue displacement is observed due to growth. Neuromast are therefore used as reference point within the tissue to distinguish motile from stationary cells within the tissue, independent of tissue growth. In both time-lapse, left green arrow indicates the relative position of a eGFP⁺mCherry^- cells to the nearest neuromast; the right green arrow indicates the relative position between two eGFP⁺mCherry^- cells. In all cases the relative positions do not change. In the intact fin, few eGFP⁺mCherry⁺ cells are observed and move non-directionally throughout the tissue; eGFP⁺mCherry^- cells however are stationary relative to the neuromast (green arrows). In the amputated fin, the number of eGFP⁺mCherry⁺ cells increase, and these cells are very motile, and most move directionally towards the wound; eGFP⁺mCherry^- cells however, although they change their morphology by protruding dendrites, do not move directionally towards the wound. As the tissue regenerates and grows, eGFP⁺mCherry^- cells appear to move towards the right, but they do not change their position relative to the neuromast (green arrows). Larvae as shown are the same as those in main Figure 3A.

Supplementary Video 3. Differential onset of tnfb expression in macrophages and neutrophils. Caudal fins of 3 dpf transgenic larvae were amputated or left intact as control. Images were acquired over a period of 6 hpa as a 21 µm z-stack at a 6-minutes interval with an Andor spinning disk confocal microscope at 20x magnification. Representative time-lapse of the maximum projections, rendered at 5 frames per second, are shown. The video first shows Tg(tnfb:mCherry-F;mpeg1:eGFP), followed by Tg(tnfb:mCherry-F;mpx:eGFP), together showing that tnfb:mCh-F expression is detectable in macrophages earlier than in neutrophils.

Supplementary Video 4. Kinetics of expression of tnfb, tnfa and il1b after amputation. Caudal fins of 3 dpf transgenic larvae were amputated or left intact and imaged at a 6-minute interval with an Andor spinning disk confocal microscope using 20x magnification. Representative time-lapse covering the 9 hpa period are shown, rendered at 5 frames per second. Each frame is a maximum projection of images acquired as 21 z-stacks. Tracks were added to indicate representative tnfb:mCh-F⁺ cells also expressing either tnfa:eGFP-F or il1b:eGFP-F. The video first shows Tg(tnfa:eGFP-F);Tg(tnfb:mCherry-F), followed by Tg(il1b:eGFP);Tg(tnfb:mCherry-F).

Supplementary Video 5. Kinetics of tnfa:eGFP-F and tnfb:mCh-F expression in macrophages following E. coli injection. Transgenic zebrafish (3 dpf) were injected intramuscularly with 1 nl PVP or with 2x10³ CFU E. coli crimson resuspended in PVP, and images were acquired over a period of 5 hpi as a 58 µm z-stack at a 10-minutes interval with an Andor spinning disk confocal microscope at 20x magnification. Representative time-lapse of the maximum projections rendered at 5 frames per second are shown. In the trunk of a PVP-injected Tg(tnfb:mCherry-F;mpeg1:eGFP) larvae, few macrophages (eGFP⁺) migrate to the site of injection and express tnfb (eGFP⁺mCherry⁺), the number of recruited macrophages expressing tnfb (eGFP⁺mCherry⁺) increases following to E. coli injection. In the trunk of a PVP-injected Tg(tnfa:eGFP-F;mpeg1:mCherry-F) larvae, few macrophages (mCherry⁺) are recruited to the site of injection express tnfa (eGFP⁺mCherry⁺), the number of recruited macrophages expressing tnfa (eGFP⁺mCherry⁺) increases following E. coli injection.

Figrue 3—source data 1.

Figrue 4—source data 1.

Figrue 5—source data 1.

Figrue 6—source data 1.

Key resources table.

Additional information

Funding

Marie-Curie Innovative Training Network INFLANET (955576)

• Mai E Nguyen-Chi

• Resul Özbilgiç

• Christina Begon-Pescia

WIAS PhD Grant from Wageningen University

• Kaylee van Dijk