Rough and smooth variants of Mycobacterium abscessus are differentially controlled by host immunity during chronic infection of adult zebrafish

Prevalence of Mycobacterium abscessus infections is increasing in patients with respiratory comorbidities. After initial colonisation, M. abscessus smooth colony (S) variants can undergo an irreversible genetic switch into highly inflammatory, rough colony (R) variants, often associated with a decline in pulmonary function. Here, we use an adult zebrafish model of chronic infection with R and S variants to study M. abscessus pathogenesis in the context of fully functioning host immunity. We show that infection with an R variant causes an inflammatory immune response that drives necrotic granuloma formation through host TNF signalling, mediated by the tnfa, tnfr1 and tnfr2 gene products. T cell-dependent immunity is stronger against the R variant early in infection, and regulatory T cells associate with R variant granulomas and limit bacterial growth. In comparison, an S variant proliferates to high burdens but appears to be controlled by TNF-dependent innate immunity early during infection, resulting in delayed granuloma formation. Thus, our work demonstrates the applicability of adult zebrafish to model persistent M. abscessus infection, and illustrates differences in the immunopathogenesis induced by R and S variants during granulomatous infection. The pathogen Mycobacterium abscessus can switch from a smooth colony form (S) into a highly inflammatory, rough colony form (R) during infection. Here, Kam et al. use an adult zebrafish model of M. abscessus chronic infection to illustrate differences in the immunopathogenesis induced by R and S variants.

M ycobacterium abscessus is an increasingly recognized human pathogen responsible for a wide array of clinical manifestations including muco-cutaneous infections and disseminated or chronic pulmonary diseases 1 . The latter is mostly encountered in patients with underlying lung disorders, such as bronchiectasis or cystic fibrosis (CF). Irrespective of being a rapid-growing mycobacteria (RGM), M. abscessus displays many pathophysiological traits with slowgrowing mycobacteria (SGM), such as Mycobacterium tuberculosis. These include the capacity to persist silently within granulomatous structures and to produce pulmonary caseous lesions 2,3 . In addition, M. abscessus is notorious for being one of the mostdrug resistant mycobacterial species, characterized by a wide panel of acquired and innate drug resistance mechanisms against nearly all anti-tubercular drugs, as well as many different classes of antibiotics 1,4 . Consequently, this explains the complexity and duration of the treatments and the high level of therapeutic failure 5 .
M. abscessus exists either as smooth (S) or a rough (R) colony morphotype variants associated with distinct clinical outcomes 6 . Previous epidemiological studies have highlighted the association of the R variant that can persist for many years in the infected host, with a rapid decline in pulmonary function [7][8][9] . It is well established that these morphological differences between S and R variants are dependent on the presence or absence of surfaceexposed glycopeptidolipids (GPL), respectively, which are reduced or lost in R variants by transcriptional downregulation or loss of function mutations 6,[10][11][12][13][14] . However, our knowledge of the pathophysiological characteristics and interactions between R or S variants with the host immune cells remains largely incomplete and is hampered by the lack of animal models that are permissive to persistent M. abscessus infection 15 .
Intravenous injection or aerosol administration of M. abscessus in immunocompetent BALB/c mice fails to establish a persistent infection, typified by the rapid clearance of the bacilli from the liver, spleen and lungs within 4 weeks 16 . Immunosuppression is required to produce a progressive high level of infection with M. abscessus in mice, either genetically, as evident in nude, severe combined immunodeficiency (SCID), interferon-gamma (GKO) and granulocyte-macrophage colony-stimulating factor (GM-CSF) knock-out mice, or chemically by dexamethasone treatment 17,18 .
Recently, an improved model of intratracheal agar bead inoculation has been adapted for M. abscessus causing a persistent infection for at least two months post implantation 17,19 .
The contribution of B and T cells in the control of M. abscessus infection has been studied in C57BL/6 mice with Rag2 −/− , Cd3e −/− and µMT −/− knockouts 20 . These studies indicated that infection control was primarily T cell-dependent in the spleen, and both B and T cell-dependent in the liver. In addition, IFNγ-receptor KO mice (ifngr1 −/− ) were found to be significantly impaired in their control of M. abscessus both in the spleen and in the liver, as were TNF −/− mice that had more pronounced and markedly dysregulated granulomas 20 . Collectively, these findings highlight the central role of T cell immunity, IFNγ and TNF for the control of M. abscessus in C57BL/6 mice, similarly to the control of M. tuberculosis infection.
In recent years, alternative non-mammalian models, such as Drosophila 21 , Galleria larvae 22 , and zebrafish embryos 15 have been developed to study the chronology and pathology of M. abscessus infection and for in vivo therapeutic assessment of drugs active against M. abscessus. In particular, zebrafish embryos have delivered important insights into the pathogenesis of M. abscessus and the participation of innate immunity in controlling infection 10,23 . The optical transparency of zebrafish embryos has been used to visualise the formation of large extracellular cords by the R form in vivo, representing a mechanism of immune subversion by preventing phagocytic destruction and highlighting the importance of mycobacterial virulence factors such as the dehydratase MAB_4780 and the MmpL8 MAB lipid transporter 10,24,25 . Other studies in zebrafish embryos have demonstrated the contribution of host TNF signalling and IL8-mediated neutrophil recruitment for protective granulomatous immunity against M. abscessus 23 , and the link between dysfunctional cystic fibrosis transmembrane conductance regulator (CFTR) and vulnerability to M. abscessus infection via phagocytic oxidative responses 26 .
Adult zebrafish models have been well-described for the study of mycobacterial pathogenesis by Mycobacterium marinum, used as a surrogate for the closely related M. tuberculosis, and the human pathogen Mycobacterium leprae [27][28][29][30] . Encompassing a fully functional immune system, previous studies in adult zebrafish with pathogenic mycobacteria, such as M. marinum, have unravelled the interplay between innate and adaptive immunity in mycobacterial granuloma formation and function.
Here, we show adult zebrafish are a useful host to analyse and compare the chronology of infection with M. abscessus S and R variants and to study the contribution of the T cell-mediated immunity in the granulomatous response in M. abscessus infection.

Results
Adult zebrafish can be chronically infected with M. abscessus. We infected adult zebrafish with approximately 10 5 CFU per animal with the rough (R) and smooth (S) variants of the reference strain CIP104536 T , scaled for the smaller size of zebrafish from 10 6 −10 7 used in mouse intravenous infections 16,17,20 . To determine if M. abscessus produces a persistent infection in adult zebrafish, we performed CFU recovery on animals across 28 days of infection (Fig. 1A). Variation in the initial inoculum ranging from 10 4 to 10 6 CFU did not appear to impact the course of infection burden with stable burden of the R variant within a 1-log window either side of the inoculation dose up to 28 days post infection (dpi) and progressive growth of the S variant to approximately 1-log above the inoculation dose at 28 dpi in each of three experiments.
Normalising burdens across three independent experiments per M. abscessus variant to perform statistical testing, we observed statistically significant increases in the proliferation of M. abscessus S compared to R variant at 7, 14, and 28 dpi (Fig. 1B). Furthermore, comparison of the Day 0 and Day 28 burdens demonstrated M. abscessus R burdens were statistically unchanged (P > 0.99, ANOVA) while M. abscessus S burdens increased by approximately 20-fold across the 4 weeks (P = 0.024, ANOVA).
Granuloma histopathology is accelerated during M. abscessus R infection compared to S. We next performed histology on adult zebrafish infected with fluorescent M. abscessus R or S carrying tdTomato or Wasabi encoding plasmids 10 . From 10 dpi, we noted a heterogeneous mix of free bacteria around the peritoneal cavity or cellular granulomas diffuse foci of bacteria spread throughout host abdominal organ tissue, and necrotic granulomas with stereotypical host nuclei ringing around a central necrotic core containing M. abscessus in all animals ( Fig. 2A).
We also observed the appearance of very large necrotic granulomas filled with fluorescent bacteria expressing tdTomato and necrotic debris measuring over 500 µm in M. abscessus R-infected animals from 14 dpi onwards ( Fig. 2A). These large granulomas were observed only occasionally and at a rate of no more than 1 per infected animal at 14 and 28 dpi (n = 2 with single abscess, 9 without abscess). Three M. abscessus R-infected animals were maintained until 70 dpi and appeared to be outwardly healthy. All three were found to have multiple necrotic granulomas containing fluorescent M. abscessus R, demonstrating that persistent infection is possible in adult zebrafish, and two were found to have necrotic granulomas measuring over 500 µm suggesting an increase in the rate of large granuloma formation as infections progress (Supplemental Fig. 1). Granulomas in M. abscessus S-infected animals did not reach this size at the 14 and 28 dpi timepoints sampled (n = 17 at 14 dpi, 16 at 28 dpi without abscess).
We next quantified the number of necrotic granulomas and "unorganised lesions", consisting of either diffuse foci of bacteria spread throughout host tissue or free bacteria around the peritoneal cavity. The proportion of necrotic granulomas in M. abscessus R-infected adult zebrafish increased between 10 and 14 dpi then remained similar between 14 and 28 dpi (Fig. 2B). M. abscessus S was observed to grow freely in mesenteric spaces and the rate of necrotic granulomas increased across all times with a statistically significant difference between 10 and 28 dpi (Fig. 2B). The proportion of necrotic granulomas was higher in M. abscessus R-infected than in M. abscessus S-infected animals at 14 dpi but similar at 10 and 28 dpi, suggesting granuloma formation is accelerated in R infections compared to S infections.
These patterns were recapitulated in our quantification of fluorescent bacterial burden in each type of lesion per animal. Significantly more M. abscessus R was observed within necrotic granulomas at 14 and 28 dpi than at 10 dpi, and an increase in the proportion of M. abscessus S within necrotic granulomas was only observed at 28 dpi compared to 10 dpi (Fig. 2C). The proportion of M. abscessus R within necrotic granulomas was higher than the proportion of M. abscessus S within necrotic granulomas at 14 dpi but similar at 10 and 28 dpi, again suggesting accelerated granuloma formation in R variant infections compared to S.
Tumour necrosis factor-mediated immunity differentially controls infection by M. abscessus variants. The cytokine TNF is essential for the granulomatous containment of M. abscessus in zebrafish embryos 23 . To visualize tnfa transcription, we next took advantage of the TgBAC(tnfa:GFP) pd1028 zebrafish line 31 , where GFP expression is driven by the tnfa promoter, to investigate if TNF expression is linked to granuloma formation. Expression of GFP was observed in adult zebrafish infected with either variant of M. abscessus. GFP was expressed by host cells in close contact with either M. abscessus R (Fig. 3A) or S (Fig. 3B).
To examine the role of host inflammatory signalling in the adult zebrafish-M. abscessus infection model, we first treated infected adults with the potent anti-inflammatory drug dexamethasone 32 . Dexamethasone treatment reduced tnfa promoter activity around M. abscessus lesions in TgBAC(tnfa:GFP) pd1028 zebrafish (Fig. 3C). Additionally, and as expected, dexamethasone treatment increased both R and S M. abscessus burdens (Fig. 3D), indicating that ablation of the inflammatory response led to uncontrolled bacterial expansion.
As dexamethasone acts broadly, we next specifically examined the role of the TNF-TNFR axis by knocking down expression of tnfa using a pooled gRNA CRISPR-Cas9 strategy. Comparison of WT and scrambled gRNA/Cas9-injected clutchmates indicated no effect of scrambled gRNA/Cas9-injection on M. abscessus R or S burdens (Supplemental Fig. 2). The efficacy of knockdown was validated by reduced RT-qPCR detection of tnfa transcripts in CRISPR-Cas9 injected adults (Fig. 4A). Knockdown of tnfa reduced the burden of R M. abscessus and increased the burden of S M. abscessus at 14 dpi (Fig. 4B).
While tnfa-depleted animals formed necrotic granulomas at the same rate as control animals in response to R M. abscessus infection (Fig. 4C), there was a sharply reduced proportion of M. abscessus R variant found within necrotic granulomas in tnfa-depleted animals (Fig. 4D). These parameters were unchanged in S M. abscessus infections suggesting the increased burden in S-infected tnfa knockdown animals was due to non-granulomatous mechanisms.
To further investigate the role of TNF signalling in our M. abscessus infection model, we next individually targeted the genes encoding the two zebrafish TNF receptors tnfr1 and tnfr2 for knockdown using CRISPR-Cas9 injections. Knockdown was confirmed by high resolution melt analysis of amplicons containing the first CRISPR target site demonstrating efficient editing of the loci (Fig. 4E). Knockdown of tnfr1 or tnfr2 again reduced the burden of R M. abscessus and increased the burden of S M. abscessus at 14 dpi (Fig. 4F).
Necrotic granuloma formation was reduced in both tnfr1 and tnfr2 knockdown animals infected with R M. abscessus, but unaffected in S M. abscessus-infected animals (Fig. 4G). The proportion of R M. abscessus, but not S M. abscessus, found within necrotic granulomas was also reduced in both tnfr1 and tnfr2 knockdown animals (Fig. 4H).
These observations demonstrate conservation of protective TNFmediated immunity against the S variant in adult infections and uncover a detrimental role for the TNF-TNFR axis in promoting R variant immunopathology through granuloma necrosis.
T cell-mediated immunity differentially controls infection by M. abscessus variants. Given the requirement for T cells to maintain granuloma structure in adult zebrafish M. marinum infection 29 , we next asked if there was T cell involvement around M. abscessus granulomas using TgBAC(lck:EGFP) vcc4 zebrafish 33 . We observed T cell association and penetration throughout cellular and necrotic M. abscessus R granulomas (Fig. 5A). We did not observe T cell interaction with M. abscessus S growing "freely" around peritoneal organs early in infection (Fig. 5B) and the T cell response to tissue-invasive M. abscessus S was noticeably less than that for equivalent sized M. abscessus R granulomas (Fig. 5C).
To directly test the requirement of T cells for containing M. abscessus, we next utilised the T cell-deficient lck −/− sa410 mutant line. We infected wild type (WT) control and lck −/− sa410 mutant adult zebrafish with both the S and R variants. T cell-deficient adult zebrafish were significantly more susceptible to M. abscessus R infection with reduced survival over 28 days of infection (P = 0.0005, Log-rank test) (Fig. 5D). T cell deficiency had a less pronounced effect on the survival of animals infected with M. abscessus S compared to M. abscessus R infection (WT S versus lck −/− S; P = 0.03, Log-rank test). Within the T cell-deficient animals, there was a 5.5 day increased median survival for M. abscessus S-infected animals (34 dpi) compared to M. abscessus R (28.5 dpi), although both groups eventually succumbed to infection at the same rate after 35 dpi (P = 0.78, Log-rank test). Bacterial burden was significantly increased at 14 dpi in lck −/− sa410 animals infected with the R, but not the S variant compared to burdens in WT adult zebrafish (Fig. 5E).
Surprisingly, we found necrotic granulomas in a survivor 56 dpi lck −/− sa410 fish infected with M. abscessus R (Supplemental Fig. 3). These granulomas were all relatively small, with the largest having necrotic cores of approximately 100 µm, compared to the large granulomas seen in our 70 dpi WT animals at more than 500 µm (Supplemental Fig. 1).
These observations demonstrate that the control of M. abscessus R infection is more reliant on T cell-mediated immunity abscessus R granulomas from 14 dpi (Fig. 6A), and to necrotic M. abscessus S granulomas from 28 dpi (Fig. 6B). The Treg response to M. abscessus R granulomas was significantly higher than for equivalent sized M. abscessus S granulomas at 14 dpi (Fig. 6C). Although we observed a higher Treg response to M. abscessus S than R at 28 dpi (Fig. 6C), this was driven by the significantly higher burden of M. abscessus R than S per granuloma rather than an absolute increase in Treg fluorescence per granuloma ( (Fig. 6D). However, there were similar percentages of necrotic granulomas and similar burdens of M. abscessus within necrotic granulomas for each group at 14 dpi (Fig. 6E, F).

Discussion
In this study, we report the use of adult zebrafish to probe both host and mycobacterial determinants of pathogenesis during persistent infection with M. abscessus. Infection with the R and S variants was maintained at high levels up to one month post infection in genetically intact animals, a major improvement on traditional mouse models of M. abscessus infection.
It is well known that the intracellular lifestyle of the R and S morphotypes differ significantly, resulting in entirely distinct infection scenarios that, we hypothesise, underlie the accelerated granuloma formation by the R variant in adult zebrafish 10,34 . The absence of GPL on the outer mycobacterial membrane causes corded growth of R variants, resulting in multiple bacilli being simultaneously phagocytosed by macrophages and overloaded phagosomes that rapidly activate autophagy pathways 12,34 . Comparatively, the S variant is able to survive for an extended period of time within the phagosome, producing a chronic and persistent infection 10,15 . As such, these polar infection responses may explain why the R variant displays widespread necrotic granuloma formation by 14 dpi, compared to the S variant that shows delayed onset of granuloma formation after 14 dpi. Moreover, this observation matches the superior in vivo growth performance of S bacilli compared to R, suggesting that the R variant is at an overall disadvantage because of its intrinsic hyperinflammatory status and the activation of T cell-mediated immunity that occurs concomitantly with granuloma formation. Interestingly, earlier reports using the zebrafish embryo demonstrated that both bacterial burden and granuloma formation dynamics were similar between both the S and R variants 10,26 , highlighting the critical role of adaptive immunity in divergent responses to the M. abscessus variants. Taken together, our data provide additional evidence for the distinct intracellular fates of both S and R variants in vivo, and further implicate the role of adaptive immunity in granuloma formation and the control of M. abscessus infection in adult zebrafish. While dexamethasone-sensitive immunity was necessary for the control of both variants and Treg immunomodulation was necessary for the control of R infection, we discovered a striking pathogenic role for tnfa, tnfr1, and tnfr2 during R infections and a protective role during S infections. Although dexamethasone treatment marginally reduced host tnfa promoter activity around sites of infection, dexamethasone treatment has wider immunosuppressive effects. Our data points towards a specific role for host TNF signalling being co-opted by M. abscessus R infection to drive to the formation of necrotic granulomas as a permissive niche for bacterial growth within the host. The detrimental role of tnfa, tnfr1, and tnfr2 in adult zebrafish-M. abscessus R infection is the opposite of previously published work in the zebrafish embryo using knockdown of tnfr1 23 , which points to the importance of using multiple models systems to investigate different stages of infection. Our data suggest human studies of increased susceptibility to M. abscessus in patients receiving TNF inhibition should be stratified by strain morphology to investigate the clinical relevance of this effect. We hypothesise that the increased susceptibility of patients receiving TNF inhibition therapy to M. abscessus disease will be driven by the loss of control of S variant infections 35 .
Our finding that Tregs are closely associated with M. abscessus R granulomas and act to restrain M. abscessus R growth provides complementary evidence that M. abscessus R takes advantage of host immunopathogenesis to establish chronic infection in adult zebrafish. While the assays proved adept at detecting reduced necrosis in the tnfa, tnfr1, and tnfr2 crispants, Treg-deficient mutants had comparable proportions of necrotic granulomas and proportions of M. abscessus R within necrotic granulomas to WT clutchmates indicating no increase in necrotic granuloma phenotypes. We hypothesise that this is because our granuloma necrosis assays for M. abscessus R are running at the upper limits of signal. Zebrafish Treg cells produce context-specific regulatory factors in a Foxp3a-dependent manner 36 , and future experiments are warranted to identify the protective Foxp3a-dependent factors produced in the context of M. abscessus R granulomas.
T cells are critical host determinants in the control of mycobacterial infection 37 . Recruitment of T cells into granulomas is thought to be essential in containing persistent infection, while T Herein, we examined the recruitment of T cells within granulomas and identified that S variant granuloma were characterised by a relative paucity of T cell infiltration, suggesting that T cells play a less significant role in S variant infections. Consistent with this, we observed increased in vivo growth of the R, but not S, variant in the absence of T cells when compared to WT animals and in our co-infection experiments where R but not S burdens were increased in lck −/− zebrafish. Following the hypothesis that patients acquire an environmental S variant of M. abscessus, our findings provide insight into the clinical observation that AIDS patients are not at increased risk of M. abscessus infection to the same degree that AIDS is a risk factor for M. tuberculosis and other non-tuberculous mycobacterium infections, such as Mycobacterium avium, as T cell deficiency has a limited effect on the control of S variant infection 37 .
The extended maintenance of R variant burden for at least 4 weeks in zebrafish is comparable to our recent data from the C3HeB/FeJ mouse 40 , but the proliferation of the S variant up to a log above inoculation dose is unprecedented in a genetically intact vertebrate host. The granulomatous immunopathology in mycobacterium-infected C3HeB/FeJ mice is due to an exaggerated type I interferon response suppressing protective IL-1 41 . Further analysis of IFN and IL-1 responses to M. abscessus infection of mice and zebrafish will help translate our understanding of these dichotomous responses into host directed therapies.
We did not observe switching of S M. abscessus into a rough colony morphotype at any timepoint during this study. In vivo switching is a rare event that has only been documented in immunocompromised mice or after months-to-years of infection in patients 13,42 . The high S morphotype burdens achieved in adult zebrafish suggest this platform may be useful for future studies of switching during extended infections, with the potential to model responses to chemotherapy and transcriptional adaptations to host residence that may include rough-like surface morphologies 13,43 .
To date, our understanding of the diverse immune responses between S and R variants have essentially been thoroughly described with respect to innate immunity, and currently our knowledge pertaining to adaptive immunity in M. abscessus infection has been poorly characterised 20   M. abscessus strains and handling. Rough (R) and smooth (S) variants of M. abscessus strain CIP104536 T were grown at 37°C in Middlebrook 7H9 broth supplemented with 10% Oleic acid/Albumin/Dextrose/Catalase (OADC) enrichment and 0.05% Tween 80 or on Middlebrook 7H10 agar containing 10% OADC (7H10 OADC). Recombinant M. abscessus strains expressing tdTomato or Wasabi were grown in the presence of 500 µg/ml hygromycin 10,23 . Homogenous bacterial suspensions for intraperitoneal injection in adult fish were prepared as previously reported 45 . Briefly, M. abscessus was plated from freezer stocks onto 7H10 OADC supplemented with 500 µg/ml hygromycin. Colonies were picked and outgrown in 7H9 OADC 500 µg/ml hygromycin for three days. M. abscessus was harvested by pelleting, resuspended in 7H9 and sheered through at 27 G needle prior to aliquoting and freezing.
The fluorescent M. abscessus strains are available from the corresponding author or their creator Dr Laurent Kremer at IRIM (Montpellier, France).
Bacterial recovery. Animals were killed by tricaine anaesthetic overdose (>300 μg/ ml) and rinsed in sterile water. Individual carcasses were mechanically homogenised and serially diluted into sterile water. Homogenates were plated onto 7H10 supplemented with OADC and 300 μg/ml hygromycin. Plates were grown for at least 4 days at 37°C.
Histology. Animals were subjected to cryosectioning as previously described 44 . Briefly, euthanasia was performed by tricaine anaesthetic overdose and specimens were fixed for 2-4 days in 10% neutral buffered formalin at 4°C. Specimens were then rinsed in PBS, incubated overnight in 30% sucrose, incubated overnight in 50/ 50 30% sucrose and Tissue-Tek O.C.T. compound (OCT, Sakura Finetek), and finally incubated overnight in OCT prior to freezing at −80°C. Cryosectioning was performed to produce 20 µm thick sections. Sections were post-fixed in 10% neutral buffered formalin and rinsed in PBS prior to further processing. Slides for fluorescent imaging were mounted with coverslips using Fluoromount G containing DAPI (Proscitech).
T cells were detected in sections from TgBAC(lck:EGFP) vcc4 zebrafish by anti-GFP staining to enhance visible fluorescent green signal (primary antibody: ab13970, Abcam; secondary antibody: ab150173, Abcam), both antibodies were diluted 1:500 in 5% goat serum (Thermofisher). Stained slides were then mounted with coverslips using Fluoromount G containing DAPI. All imaging was carried out on a Leica DM6000B microscope with LAS software version X (Leica) and image analysis was performed in ImageJ version 1.53n (NIH).
Drug treatment. Animals for dexamethasone treatment were infected with 1-5 × 10 4 CFU to ensure survival following immunosuppression. Water soluble dexamethasone (Sigma, D2915) was added to a final concentration of 40 μg/ml immediately following infection and changed every second day.
Genotyping of tnfr1 and tnfr2 crispants was performed by high resolution melt analysis of amplicons produced by amplification of gDNA from diluted homogenate using oligos: tnfr1 gFw GCAGTGCAGAAAACATGAGG, GRv CGTTTTGTGCATTGCTGGC; tnfr2 gFw ACTCGCTTGTCTGTGCAATG, gRv TGGACACTTGAAACAATTGGGA. Amplification and analysis was carried out with MeltDoctor™ HRM Master Mix (Thermofisher) reagent on a CFX thermocycler with CFX Maestro software version 2.2 (Biorad) according to MeltDoctor product specifications.
Phenotyping of tnfa crispants was performed by RT-qPCR using oligos: tnfa qFw GCTTATGAGCCATGCAGTGA, qRv AAGTGCTGTGGTCGTGTCTG; 18 s qFw TCGCTAGTTGGCATCGTTTATG, qRv CGGAGGTTCGAAGACGATCA. Homogenate was processed for RNA extraction following product specification for Trizol LS (Thermofisher). Reverse transcription was carried out from 2 μg total RNA with the Applied Biosystems™ High Capacity cDNA Reverse Transcription Kit (Thermofisher). Quantitative PCR was carried out with PowerUp™ SYBR™ Green Master Mix (Thermofisher) on a CFX thermocycler with CFX Maestro software version 2.2 (Biorad).
Statistics. All statistical testing was carried out using Graphpad Prism version 9.3.1. Statistical tests are indicated in the figure legends and include both T tests for pairwise comparisons and ANOVA analyses with post hoc analysis for comparison of three or more groups. Each data point indicates a single animal unless otherwise stated.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The processed analysis data are included in the Source Data file that is provided with this paper. Raw image files will be archived for 10 years (2022-2032) by The Centenary Institute (Sydney, Australia). The raw image and analysis data are available under restricted access due to their large size and prohibitive cost of public hosting: they can be obtained upon request by email from the corresponding author, who will mediate access from the Centenary Institute servers. Expected response times will be within one week. Source data are provided with this paper.