An arginase 2 promoter transgenic line illuminates immune cell polarisation in zebrafish

Initial immune responses to inflammatory or infection stimuli are mediated by innate immune cells, of which leukocytes are major players. Two important leukocyte cell types, neutrophils and macrophages, work together to neutralise invading threats while promoting tissue healing and restoration of homeostasis. Neutrophils and macrophages have evolved together and co-exist in invertebrate and vertebrate species (Hartenstein, 2006). Neutrophils are often observed to be the first immune cell type to respond to immune challenge, rapidly migrating towards the stimuli and becoming activated to a pro-inflammatory and antimicrobial state, clearing damaged cells and invading pathogens. Macrophages also rapidly respond to immune challenge, with pro-inflammatory phenotypes emerging soon after immune challenge aiding microbial clearance (often termed M1 or classically activated). Dogmatically, neutrophils are considered a blunt first line of defence that, once activated, remain so until cleared from tissues, either by apoptosis and subsequent efferocytosis by macrophages or by migration (Duffin et al., 2010; Elks et al., 2011b; Mathias et al., 2006). Persistence of neutrophils during disease can cause considerable bystander damage to healthy surrounding tissues driving chronic pathologies. Macrophage phenotypes, on the other hand, are well characterised as being plastic throughout the pathogenesis of inflammation and infections in a process termed macrophage polarisation. Pro-inflammatory macrophage phenotypes are followed by anti-inflammatory phenotypes that promote healing and restoration of homeostasis (often termed M2 or alternatively activated). Many of the above observations come from in vitro approaches. In vivo exploration indicates that macrophage polarisation is not binary but rather a spectrum of phenotypes and behaviours, and that the neutrophil response is more plastic than previously thought, with emerging roles for neutrophil subsets in tissue protection and healing (Ballesteros et al., 2020; Giese et al., 2019; Phillipson and Kubes, 2019).

Both neutrophils and macrophages are activated by danger- and pathogen-associated molecular patterns (DAMPs and PAMPs, respectively), which trigger the production of pro-inflammatory signals (e.g. IL-1β and TNF-α) after immune challenge (Kato and Kitagawa, 2006; Shapouri-Moghaddam et al., 2018). These activate the production of antimicrobial molecules, including reactive nitrogen species (RNS), via the enzyme inducible nitric oxide synthase (iNOS) (Stuehr and Marletta, 1985). Immunomodulatory signals are required to limit and eventually turn off the pro-inflammatory response for tissues to regain homeostasis. The best characterised of these anti-inflammatory signalling molecules are cytokines, including the interleukins (ILs) IL-4, IL-13 and IL-10, alongside proteins that dampen the pro-inflammatory response, including arginase (Wiegertjes et al., 2016). Despite their classification as anti-inflammatory, these signals have also been identified as being upregulated in some pro-inflammatory situations (Munder et al., 2005). The production of antimicrobial RNS is negatively regulated by the enzyme arginase. Arginase competes with iNOS for a shared substrate, L-arginine, a limited resource within the cell. The immunomodulatory properties of arginase extend beyond the regulation of RNS production. Arginase 2 is essential for IL-10-mediated downregulation of pro-inflammatory factors, making it a key anti-inflammatory enzyme (Dowling et al., 2021). Arginase and iNOS have been well characterised in murine models as being strongly expressed in macrophage subtypes, with arginase being a wound-healing/anti-inflammatory macrophage marker and iNOS a pro-inflammatory macrophage marker (Munder et al., 1998). In human macrophages, the distinction between iNOS and arginase-expressing macrophage subtypes is less well defined, due, in part, to lower macrophage RNS in humans compared to that in mice. Interestingly, human neutrophils constitutively express arginase with levels increasing after infection in vitro, but its roles are poorly understood in vivo (Munder, 2009). Considering that arginase is not expressed highly by murine neutrophils (El Kasmi et al., 2008; MacMicking et al., 1997), alternative in vivo models are required to investigate the roles of arginase in immunomodulation and human disease.

The zebrafish has become a powerful model organism to determine the molecular mediators of immunity against injury and pathogenic challenges (Henry et al., 2013; Yoshida et al., 2017). Advantages of the zebrafish model include transparent larvae combined with fluorescent transgenic lines, allowing detailed microscopy in an intact organism in vivo. A key turning point in zebrafish immunity research came with the development of transgenic lines marking neutrophil and macrophage cell populations. Initially, transgenic lines were developed that labelled whole populations of immune cells in zebrafish larvae, e.g. TgBAC(mpx:GFP)i114 labelling the total neutrophil population and Tg(mpeg1:GFP) labelling the total macrophage population (Ellett et al., 2011; Renshaw et al., 2006). More recently, transgenic lines of important pro-inflammatory cytokines have become available [e.g. TgBAC(tnfa:GFP)pd1028 and TgBAC(il-1beta)sh445], allowing in-depth analysis of the cells producing these important signals following a variety of immune challenges in vivo (Marjoram et al., 2015; Ogryzko et al., 2019). These transgenic lines utilised bacterial artificial chromosome (BAC) technology, which allows tens of kilobases of promoter region to be used to drive expression of the fluorescent protein, ensuring that the expression of the transgene recapitulates endogenous expression patterns as closely as possible. A key gap in this zebrafish ‘transgenic toolbox’ is an anti-inflammatory fluorescent transgenic line. We therefore set out to develop an arginase 2 (arg2) promoter-driven BAC transgenic line to understand the arginase response to immune challenge in vivo.

Here, we compared existing expression data of arginase genes in zebrafish neutrophils and macrophages and show that arg2 is the most highly expressed arginase in zebrafish immune cells. We developed a new BAC transgenic line that drives GFP under the control of the arg2 promoter and demonstrate that the arg2:GFP transgene is expressed in ionocytes, a population of skin-resident cells, in resting conditions. Following a range of immune challenges, including tailfin transection (sterile injury) and Mycobacterium marinum (bacterial) and Cryptococcus neoformans/Candida albicans (fungal) infections, arg2:GFP expression was predominantly upregulated in neutrophils. We identify a small population of macrophages that express arg2 after injury and infection, suggesting the presence of anti-inflammatory macrophages in zebrafish. The arg2:GFP transgenic line has the potential to uncover new mechanisms behind innate immune regulation during in vivo immune challenge.

There are two isozymes of arginase in most mammals and fish, arginase 1 (ARG1) and arginase 2 (ARG2). In mice, ARG1 and ARG2 are both expressed by macrophages; however, it is ARG1 that is the most widely studied in macrophage polarisation, with increased expression of cytosolic ARG1 protein depleting intracellular stores of arginine (Rath et al., 2014). In fish, arginases have been studied in the common carp (Cyprinus carpio), a species phylogenetically close to zebrafish, in which arg2 is the most highly expressed gene of the arginase family in immune cells (Wentzel et al., 2020a; Wiegertjes et al., 2016). Head-kidney-derived macrophages of carp that have been polarised towards anti-inflammatory phenotypes using cAMP have a 16-fold upregulation of arg2 expression (Wentzel et al., 2020b; Wiegertjes et al., 2016).

Zebrafish have orthologues of mammalian ARG1 and ARG2 that share strong sequence homology with human (and mouse) orthologues (Fig. S1A-C). Zebrafish arg1 (NCBI accession number NM_001045197) and arg2 (NCBI accession number NM_199611) were compared using existing transcriptomics datasets to identify innate immune cell expression. Fluorescence-activated cell sorting (FACS)-purified neutrophils [Tg(mpx:GFP)i114] and macrophages [Tg(mpeg1:Gal4-VP16)gl24/(UAS-E1b:Kaede)s1999t] from unchallenged 5 days post fertilisation (dpf) larvae both expressed arg2, whereas arginase 1 (arg1) was not expressed at detectable levels (Fig. S1D,E; using raw data from Rougeot et al., 2019). arg2 expression was found at high levels in the bulk, non-immune-cell population and it was also expressed in the immune cell populations (Fig. S1F,G; data from Rougeot et al., 2019). arg2 expression was approximately 1.5-fold higher in neutrophils than that in macrophages in unchallenged zebrafish larvae (Fig. S1F,G; using raw data from Rougeot et al., 2019). In order to assess relative levels of arg2 expression in immune cells of zebrafish larvae, cDNA from FACS-isolated macrophages and neutrophils was analysed by real-time quantitative PCR (RT-qPCR) (Fig. S1H-J) (Nguyen-Chi et al., 2014). Sorted neutrophils expressed higher levels of arg2 compared to arg2 expression in sorted macrophages and the background of the larvae (Fig. S1J). In unchallenged adult zebrafish head kidneys (the site of haematopoiesis), single-cell RNA sequencing (RNAseq) identified arg2 expression in a population of neutrophils, a smaller population of monocytes/macrophages, with no detectable expression in other blood lineages (thrombocytes/erythrocytes) (Fig. S1K,L; data from The Zebrafish Blood Atlas, https://www.sanger.ac.uk/science/tools/basicz/basicz/; Athanasiadis et al., 2017). From the same dataset, arg1 expression was not detected in any immune cell lineage (Fig. S1K,L).

Based on the predominant expression of arg2 in fish innate immune cells, we chose to develop an arg2 transgenic zebrafish line to investigate its expression during immune challenge in vivo. We adopted a BAC transgenesis approach, using a BAC (CH-211-12d10) in which 11.5 kb of the arg2 promoter drives GFP expression, to generate two transgenic line alleles with the same expression [TgBAC(arg2:eGFP)sh571 and TgBAC(arg2:eGFP)sh572]. Owing to the higher fecundity in the sh571 line, this line was used in the following studies (hereon termed the arg2:GFP line).

In unchallenged arg2:GFP larvae, the transgene was expressed in cuboidal cells in the skin, distributed over the yolk and caudal vein regions at 2 dpf (Fig. 1A) and 3 dpf (Fig. 1B,C). A subset of ionocytes, cells in the skin responsible for the transport of sodium ions, also known as H⁺-ATPase-rich cells (HRCs), have been shown to express high levels of arg2 by in situ hybridisation in zebrafish (Eisenhoffer et al., 2017; Hsiao et al., 2007; Jänicke et al., 2007). arg2:GFP expression recapitulated the arg2 in situ hybridisation pattern, labelling ionocytes (Fig. 1D,E). To determine whether any of the cells over the yolk area expressing arg2:GFP were leukocytes, we crossed the arg2:GFP line with a macrophage transgenic line [Tg(mpeg1:mCherry)sh378, hereon termed mpeg:mCherry] and a neutrophil transgenic line [Tg(lyz:nsfB.mCherry)sh260, hereon termed lyz:mCherry]. Under resting conditions, there was no overlap between mpeg:mCherry-positive macrophages and arg2:GFP-positive cells (Fig. 1F) or between lyz:mCherry-positive neutrophils and arg2:GFP-positive cells (Fig. 1G) at 2, 3, 4 or 5 dpf. These data indicate that arg2:GFP is not expressed at detectable levels in immune cells in resting conditions at these developmental timepoints, matching in situ hybridisation data.

To assess whether arg2:GFP expression is upregulated in innate immune cells during an immune response, we challenged 2 dpf larvae with a sterile tailfin wound. Using a tailfin nick model, we identified that highly mobile immune cells migrating towards the wound expressed arg2:GFP within the first hour of timelapse microscopy (Fig. 2A). We reasoned that these cells were neutrophils, owing to their size and amoeboid shape alongside their rapid migration towards the wound. To investigate whether neutrophils expressed arg2:GFP after injury, we crossed the arg2:GFP line with the neutrophil lyz:mCherry transgenic line. Timelapse microscopy demonstrated that a subpopulation of 39% of neutrophils arriving at the tailfin-transection wound were arg2:GFP positive by 3 h post wound (hpw) (Fig. 2B-D), with arg2:GFP-positive neutrophils at the site of injury present during the recruitment phase of inflammation (1-3 hpw), whereas expression was not observable in neutrophils away from the wound. There was no expression of arg2:GFP observed in mpeg:mCherry-positive macrophages in the recruitment phase of inflammation between 1 and 6 hpw (Fig. 2E).

After the first 24-36 h of inflammation, neutrophils have been largely cleared from the tailfin to allow for tailfin regeneration (Renshaw et al., 2006). We therefore assessed the arg2:GFP expression status of macrophages and neutrophils at regenerative stages between 2 and 3 days post wound (dpw). arg2:GFP-positive neutrophils were observed in the regenerating tailfin at 2 and 3 dpw, whereas in uninjured larvae, few neutrophils were present and they were arg2:GFP negative (Fig. 2F,G). In order to observe potential anti-inflammatory macrophages during the tailfin regeneration stage, mpeg:mCherry larvae crossed into the arg2:GFP line were imaged at the 2-3 dpw timepoints, by which time the fin has partially regenerated. In uninjured larvae, there were few mpeg:mCherry-positive macrophages in the tailfin fold at 5 dpf and these did not express arg2:GFP (Fig. 2H, top panel). In tailfin-transected larvae, there were increased numbers of macrophages in the 2 and 3 dpw healing/regenerating tailfin fold, but the majority of these were arg2:GFP negative, whereas mpeg:mCherry-negative cells with the morphology of neutrophils did express arg2:GFP (Fig. 2H, bottom two panels). Upon closer investigation using confocal microscopy, mpeg:mCherry-positive cells expressing arg2:GFP were identified at the wound at 2 and 3 dpw (Fig. 2I), although these were outnumbered by arg2:GFP-negative macrophages. Taken together, these data hint at the presence of arg2-expressing anti-inflammatory macrophages during tailfin regeneration.

To investigate the expression of arg2 after bacterial infection, arg2:GFP embryos were injected with M. marinum (Mm) at 1 dpf and imaged at 1 day post injection (dpi) (2 dpf). Neutrophil arg2:GFP expression was not observed in mock-infected (injected with 2% polyvinylpyrrolidone 40, referred to as PVP) controls (Fig. 3A), with ionocyte-specific expression used to confirm arg2:GFP-positive larvae. We assessed the early response to Mm infection at 1 dpi and found that a subpopulation of neutrophils express arg2:GFP early in infection (Fig. 3B). arg2:GFP-positive neutrophils were present in the vicinity of Mm infection (Fig. 3B), with both infected and uninfected neutrophils expressing arg2:GFP (Fig. 3C). A subset of neutrophils had arg2:GFP expression, with 31.7% (of n=123 neutrophils) of lyz:mCherry-positive neutrophils in the region of infection expressing GFP, suggestive of differential immune responses between individual neutrophils (Fig. 3D,I). lyz:mCherry-positive cells expressed arg2:GFP to a higher level than lyz:mCherry-negative cells with an immune cell morphology (potential macrophages) in the same larvae (Fig. 3E). At the 1 dpi (2 dpf) timepoint, macrophage arg2:GFP expression was also not observed in mock-infected (PVP) controls (Fig. 3F), with ionocyte-specific expression used to confirm arg2:GFP-positive larvae. Only 6.1% (of n=98 macrophages) of mpeg:mCherry-positive macrophages were positive for arg2:GFP in Mm-infected individuals (Fig. 3G-I).

Fungal infections have been shown to modulate host arginine metabolism via arginase (Wagener et al., 2017); therefore, we assessed arg2:GFP expression in two well-characterised fungal zebrafish infection models – C. albicans (Brothers and Wheeler, 2012) and C. neoformans (Bojarczuk et al., 2016). Mock infection with PVP caused no neutrophil arg2:GFP expression (Fig. 4A). In both fungal infections, arg2:GFP was observed in a subpopulation of lyz:mCherry-positive neutrophils at 1 dpi (Fig. 4B,C). As with Mm infection, subsets of neutrophils both with or without internalised pathogen were arg2:GFP positive in C. albicans infection (Fig. 4C). arg2:GFP-positive mpeg:mCherry macrophages were also observed in C. neoformans infection at 1 dpi; however, these were outnumbered by arg2:GFP-negative macrophages and only three examples were identified (Fig. S2). In cryptococcal-infected larvae with a high fungal burden, arg2:GFP expression was observed in the liver (Fig. 4E) at timepoints when arg2:GFP expression was not present in PVP-injected larvae (Fig. 4D). This was especially apparent in Cryptococcus infections at 2 dpi (Fig. 4F,G) and was confirmed by in situ hybridisation (Fig. 4H). Arginase is a well-characterised liver enzyme (Wright et al., 2004), yet in unchallenged embryos (Fig. 1A-E) or PVP-injected larvae (Fig. 4D), there was no visible liver expression of arg2:GFP. Tissue-restricted arg2:GFP expression in the liver area was also observed in examples of heavily infected Mm larvae (Fig. S3). Like cryptococcal-infected larvae, liver expression occurred in individuals highly infected with Mm, but was not observed until later after infection (at 4 dpi), potentially reflecting the slower doubling time/pathogenesis of Mm compared to Cryptococcus.

arg2:GFP expression was assessed at a later stage of Mm infection (4 dpi), a stage at which innate immune granulomas are forming and when it is likely that leukocyte phenotypes are more diverse and polarised owing to immune modulation by mycobacteria (Cronan et al., 2021). PVP-control-injected larvae had minimal arg2:GFP expression, apart from ionocyte-specific expression and background green signal from the autofluorescence of pigment cells (Fig. 5A). In Mm-infected larvae, arg2:GFP was expressed in cells associated with developing granulomas at 4 dpi (Fig. 5B). The lyz:mCherry-positive cells appeared to express higher levels of arg2:GFP than other granuloma-associated cells (Fig. 5C,D). When quantified, the lyz:mCherry-positive cells had greater arg2:GFP fluorescence than lyz:mCherry-negative cells with an immune cell morphology (potential macrophages) in the same larvae (Fig. 5E). arg2:GFP-positive granuloma-associated neutrophils represented 37.9% of neutrophils imaged around granulomas, with the remaining 62.1% having no detectable arg2:GFP expression (Fig. 5F,G).

It was noted that there were also granuloma-associated arg2:GFP-expressing cells that were lyz:mCherry negative (Fig. 5C), some containing phagocytosed bacteria (Fig. 5C), and we hypothesised these to be macrophages. PVP-control-injected larvae had minimal arg2:GFP expression in macrophages, but did have ionocyte-specific expression and background green signal from the autofluorescence of pigment cells (Fig. 6A). There were few mpeg:mCherry-positive macrophages that were arg2:GFP positive at 4 dpi (4% of those imaged; Fig. 6B,C). At 4 dpi, mpeg:mCherry-positive macrophages had lower fluorescence levels of arg2:GFP compared to mpeg:mCherry-negative/arg2:GFP-positive granuloma-associated cells that would include neutrophils (Fig. 6D). As was the case with neutrophils, arg2:GFP expression was observed in both infected and non-infected mpeg:mCherry-positive macrophages around granulomas (Fig. 6E,F). The low number of mpeg:mCherry/arg2:GFP double-positive macrophages observed at 4 dpi, alongside arg2:GFP-positive neutrophils, did not appear to account for the many granuloma-associated arg2:GFP-positive cells that were observed, with many appearing to be mpeg:mCherry negative (Fig. 6B). As it has been previously shown that the mpeg promoter is downregulated by Mm at 4 dpi (Benard et al., 2015), the fluorescence brightness of mpeg:mCherry expression was assessed at 2, 3 and 4 dpi (3, 4 and 5 dpf, respectively) after Mm infection at 1 dpf. There was no difference in red fluorescence between 2, 3 or 4 dpi, suggesting that putative Mm-infection-induced downregulation of the mpeg promoter did not impact the brightness of the macrophage transgenic fluorescence at 4 dpi (Fig. 6G). The number of arg2:GFP-positive macrophages at 4 dpi was assessed using a second macrophage transgenic line [Tg(fms:Gal4.VP16)i186;Tg(UAS:nfsB.mCherry)i149, hereafter referred to as fms:mCherry for clarity]. The proportion of arg2:GFP-positive fms:mCherry macrophages at 4 dpi was approximately equivalent to that observed in mpeg:mCherry macrophages (Fig. 6H,I).

Taken together, these data show that Mm granulomas have arginase-expressing neutrophils and that other granuloma-associated cells, including potential anti-inflammatory macrophages, also express arginase but to a lower level.

Zebrafish transgenic lines have allowed in vivo exploration of the pro-inflammatory immune response (including il1b and tnfa) to a variety of stimuli; however, until now, there has been a lack of similar tools for anti-inflammatory factors (Marjoram et al., 2015; Nguyen-Chi et al., 2015; Ogryzko et al., 2019). Here, we describe a new transgenic line for arg2 and show that this transgene is upregulated shortly after immune challenge in neutrophils and a small subset of macrophages. These data correlate with a growing body of evidence that suggest that anti-inflammatory factors are expressed early after immune challenge to suppress hyper-inflammation, before a switch to an anti-inflammatory environment is required at later stages for tissue healing and restoration of homeostasis (Cicchese et al., 2018).

The arg2:GFP line shows that neutrophils are the predominant immune cell type that express arg2 early after immune challenge in zebrafish. The observation of arg2 expression in zebrafish neutrophils fits with our previous observations that zebrafish neutrophils are the primary innate immune cell that produce antimicrobial RNS after Mm infection and suggests a neutrophil iNOS/arginase balance (Elks et al., 2013). Human neutrophils also produce high levels of RNS (Munder et al., 2005), differing from expression in mice, in which it is primarily macrophages that produce RNS/arginase (Abebe et al., 2013; El Kasmi et al., 2008; Jacobsen et al., 2007; Munder et al., 2005). Human polymorphonuclear neutrophils produce high levels of arginase at transcript and protein levels following immune challenge and, in resting states, this may act as a negative regulator of the RNS response, as is the case in murine macrophages (Abebe et al., 2013; El Kasmi et al., 2008; Jacobsen et al., 2007; Munder et al., 2005).

Our analysis of the zebrafish arg2:GFP transgene and in situ hybridisation did not detect leukocyte arg2 expression in resting conditions; however, arg2 expression was detected by RNAseq and RT-qPCR of FACS-purified leukocyte populations. This suggests that either the transgenic and in situ hybridisation techniques were not as sensitive as RNAseq or that techniques to purify immune cells in RNAseq studies have led to upregulation of arg2 not present in the intact zebrafish. FACS-purified neutrophils expressed higher levels of arg2 compared to its levels in sorted macrophages and the background of the larvae, matching our observations in the arg2:GFP line once activated by immune challenge. In the original publication of these FACS-purified populations (Nguyen-Chi et al., 2014), it was noted that immune cells may have been activated by the dissociation/FACS process, as elevated expression of proinflammatory cytokines such as il1b was observed, highlighting the technical challenges involved in ex vivo experimentation on leukocyte polarisation (Nguyen-Chi et al., 2014). FACS and subsequent RNAseq have been extensively and successfully used to identify and assess activation states of both macrophages and neutrophils; however, immune cells are removed from their local microenvironment during dissociation and careful technique development is required to avoid inadvertent activation (Nguyen-Chi et al., 2015; Rougeot et al., 2014, 2019). The addition of intact transgenic lines such as the arg2:GFP line, alongside transcriptomics requiring dissociation, will be powerful tools to further dissect immune cell polarisation.

The mechanisms balancing innate immune regulation during inflammatory and infection responses are not well understood in vivo. Neutrophil arg2:GFP expression was observed at timepoints that are considered to be pro-inflammatory stages of inflammation and infection. arg2:GFP expression was reminiscent of our previous observations using the pro-inflammatory TgBAC(il-1β:GFP)sh445 transgenic line, in which neutrophil il1b was found at 1 hpw in the tailfin and 1 dpi in Mm infections, the same timepoints at which arg2:GFP expression was observed (Ogryzko et al., 2019). This suggests that anti-inflammatory arginase expression coincides with pro-inflammatory signals in neutrophils, providing evidence for a balanced response. The observation of immune cells producing anti-inflammatory signals upon immune challenge has been reported previously, but to date this has been mainly described in macrophages (Cicchese et al., 2018). It is becoming clear that a balanced response to immune challenge, including both pro- and anti-inflammatory signals, is beneficial in disease control (Cicchese et al., 2018). In murine macrophages, it has been demonstrated in vitro that ARG1 is produced soon after infection and can have immunomodulatory effects (El Kasmi et al., 2008). Our findings open up the possibility that a similar balance may exist in neutrophils and add to recent evidence suggesting that neutrophil phenotypes are more diverse and nuanced than previously appreciated (Ballesteros et al., 2020; Giese et al., 2019).

Some pathogens have evolved to disrupt the pro- and anti-inflammatory response, keeping pro-inflammatory factors low and increasing anti-inflammatory signals to allow for immune cell evasion and survival. One such pathogen is Mycobacterium tuberculosis, which suppresses an initial pro-inflammatory response, in part by upregulation of macrophage arginase, to allow for intra-phagocyte survival and intracellular growth and proliferation to form hallmark granuloma structures (El Kasmi et al., 2008). This makes arginase a potential target for therapeutic intervention during tuberculosis. In murine tuberculosis models, macrophage arginase expression is associated with decreased bacterial killing and is an immunomodulatory target of M. tuberculosis (El Kasmi et al., 2008). Similar observations have been described in fungal infections, with mice infected with Cryptococcus (Cryptococcus gattii) having elevated levels of arginase expression in lung tissues (Oliveira-Brito et al., 2020). In human-monocyte-derived macrophages, Candida albicans infection induces arginase expression, which blocks host nitric oxide (NO) production as a fungal survival mechanism via chitin exposure (Wagener et al., 2017). Our findings using the arg2:GFP line are consistent with these mammalian observations, with early arginase expression observed in innate immune cells after Mm and fungal infection. Further investigation is required to understand the molecular mechanisms of this intriguing host-pathogen interaction.

Arginase has been described as an anti-inflammatory macrophage marker due, in part, to high expression levels in experiments using anti-inflammatory stimuli such as IL-4 or IL-13 to drive monocytes towards M2/anti-inflammatory phenotypes, as well as observations in other murine macrophage models (Mantovani et al., 2004). We observed macrophages expressing arg2:GFP during infection and at wound-healing stages of tailfin transection. Our findings complement studies on a zebrafish pro-inflammatory macrophage tnfa transgenic line (Nguyen-Chi et al., 2014), which suggest that an anti-inflammatory population of macrophages exists based on the switching off of tnfa:GFP in some macrophages during the wound-healing phase of tailfin transection, and are consistent with anti-inflammatory macrophages being present in the developing zebrafish larvae (Nguyen-Chi et al., 2015). However, in both inflammation and infection, arginase-expressing macrophages were much less frequent than arginase-expressing neutrophils, and neutrophil arginase expression predominated. It is important to note that the mpeg promoter used to mark macrophages in our study is downregulated by Mm infection (Benard et al., 2015); therefore, it is possible that our observations using the mpeg:mCherry line is an underestimation of the population of macrophages that express arg2:GFP during Mm infection. The possibility of this downregulation is evident from our data as there are many arg2:GFP-positive granuloma-associated cells, some with phagocytosed bacteria, that expressed arg2:GFP but were neutrophil-marker negative, indicative of macrophages that lack a visible mpeg:mCherry marker. However, in our experiments, mpeg:mCherry fluorescence levels in the transgenic line were not decreased at 4 dpi, indicating that the mpeg promoter remained active or that any inactivation by Mm had not yet affected the mCherry levels. Furthermore, investigation using a separate macrophage marker, fms:mCherry, also showed few arg2:GFP-positive macrophages at 4 dpi, with numbers approximately equivalent to those observed in mpeg:mCherry larvae. The fms:mCherry line used was not a direct promoter driver, but uses the gal4:uas system, which is silenced in zebrafish over generations, so there remains the possibility that not all macrophages are labelled in these larvae (Akitake et al., 2011). The granuloma-associated arg2:GFP expression observed may also be from epithelioid-like cells that make up a large proportion of the zebrafish Mm granuloma structure, some of which are macrophage derived but may have lost macrophage markers (Cronan et al., 2021). Here, we have identified mpeg:mCherry-positive granuloma-associated macrophages that express arg2; however, it remains unclear as to exactly how many macrophages are polarised towards this potential anti-inflammatory phenotype in zebrafish. Characterising the macrophage polarisation infection response fully will require further study, for which the arg2:GFP zebrafish line will be an important tool.

Outside of immune cells, the arg2:GFP line has also illuminated arginase expression in the liver in highly infected individuals and in ionocytes, during both resting and inflammatory states. Arginase is a well-characterised liver enzyme (Berüter et al., 1978; Haraguchi et al., 1987); therefore, expression in the liver was not unexpected, although the function of the upregulated liver-specific expression in highly infected individuals remains unclear. Interestingly, ionocytes have recently been described as a new airway epithelial cell type in humans and mice and it is these cells that most highly express cystic fibrosis transmembrane conductance regulator (CFTR), the anion channel that is mutated in cystic fibrosis patients (Shah et al., 2022). The arg:GFP line is a new tool that could be used to investigate the roles of these intriguing cells in vivo.

Our data indicate that arg2, an important anti-inflammatory mediator, is produced early after immune challenge, predominantly by neutrophils. The arg2:GFP line is an exciting addition to the zebrafish transgenic toolbox with which to investigate innate immunity during infections. It has the potential to be applied to multiple zebrafish disease models of infection and inflammation and may also be relevant to any models with an inflammatory component, from ageing to cancer.

Animal work was carried out according to guidelines and legislation set out in UK law in the Animals (Scientific Procedures) Act 1986, under Project License P1A4A7A5E or PP7684817. Ethical approval was granted by the University of Sheffield Local Ethical Review Panel.

All zebrafish were raised in the Biological Services Unit (BSU) aquarium (University of Sheffield, UK) and maintained according to standard protocols (https://zfin.org/) in Home Office-approved facilities. Adult fish were maintained at 28°C with a 14 h/10 h light/dark cycle.

To investigate expression in immune cells, the TgBAC(arg2:eGFP)sh571 (arg2:GFP) reporter line was crossed with the macrophage reporters Tg(mpeg1:mCherryCAAX)sh378 (Bojarczuk et al., 2016) (mpeg:mCherry) and Tg(fms:Gal4.VP16)i186;Tg(UAS:nfsB.mCherry)i149 (Gray et al., 2011) (fms:mCherry), and the neutrophil reporter Tg(lyz:nfsB.mCherry)sh260 (Buchan et al., 2019) (lyz:mCherry) to generate embryos for experiments.

An eGFP SV40 polyadenylation cassette was inserted at the arg2 ATG start site of the zebrafish BAC CH-211-12d10 using established protocols (Renshaw et al., 2006). Inverted Tol2 elements were inserted into the chloramphenicol coding sequence and the resulting modified BAC containing 115,130 bp of the arg2 promoter region was used. We identified two founder zebrafish (allele codes sh571 and sh572) and raised colonies. The embryos of both alleles had the same GFP expression pattern. The data generated in this manuscript is from the TgBAC(arg2:eGFP)sh571 transgenic line (arg2:GFP) as this line had a higher fecundity than TgBAC(arg2:eGFP)sh572. The TgBAC(arg2:eGFP)sh571 strain can be requested by contacting the corresponding author.

To induce an inflammatory stimulus, 2- or 3-dpf zebrafish were anaesthetised in 0.168 mg/ml tricaine (MS-222, Sigma-Aldrich) in E3 medium and visualised under a dissecting microscope. Using a scalpel blade (5 mm depth, World Precision Instruments) the tailfin was transected after the circulatory loop as previously described, while ensuring that the circulation remained intact (Elks et al., 2011a).

Bacterial infection experiments were performed using Mycobacterium marinum strain M (American Type Culture Collection #BAA-535) containing the pSMT3-Crimson vector, with liquid cultures prepared from bacterial plates (van der Sar et al., 2009). Liquid cultures were washed and prepared in 2% polyvinylpyrrolidone 40 (PVP40) solution (Sigma-Aldrich) as previously described for injection (Benard et al., 2012; Cui et al., 2011). Injection inoculum was prepared to 100 colony forming units (cfu)/nl for all Mm experiments, which was injected into the circulation at 30 h post fertilisation (hpf) via the caudal vein.

Fungal infection experiments were performed using the Candida albicans strain TT21-mCherry (Seman et al., 2018). Overnight liquid cultures were grown from fungal plates, then prepared for injection as previously described (Seman et al., 2018). Cultures were counted using a haemocytometer and prepared in 10% PVP40 for 200 cfu/nl injection dose, which was injected into the circulation at 30 hpf via the caudal vein.

Fungal infection experiments were also performed using the Cryptococcus neoformans strain Kn99-mCherry (Gibson et al., 2018 preprint). Cryptococcal culture was performed as previously described (Bojarczuk et al., 2016) and, after counting on a haemocytometer, Kn99 was prepared in 10% PVP40 for 200 cfu/nl injection dose, which was injected into the circulation at 1-2 dpf.

Prior to injection, zebrafish were anaesthetised in 0.168 mg/ml tricaine in E3 medium and transferred onto 1% agarose in E3+Methylene Blue plates, removing excess medium. All pathogens were injected into the circulation to create systemic infection, using a microinjection rig (World Precision Instruments) attached to a dissecting microscope. A 10 mm graticule was used to measure 1 nl droplets for consistency, and droplets were tested every 5-10 fish and recalibrated if necessary. A final injection volume of 1 nl was injected to produce doses calculated for each pathogen. After injection, zebrafish were transferred to fresh E3 medium for recovery and maintained at 28°C.

RNA probes for zebrafish arginase type II (arg2, ENSDARG00000039269; plasmid obtained from Source Bioscience) were designed and synthesised after cloning into the pCR Blunt II-TOPO vector, according to the manufacturer's instructions (Thermo Fisher Scientific). Plasmids were linearised and probes synthesised according to the DIG RNA Labelling Kit (SP6/T7) (Roche). Zebrafish larvae were fixed in 4% paraformaldehyde solution (Thermo Fisher Scientific) overnight at 4°C. Whole-mount in situ hybridisation was performed as previously described (Thisse and Thisse, 2008).

Control, tailfin-transected and infected larvae were imaged using a Leica DMi8 SPE-TCS microscope using a HCX PL APO 40×/1,10 water immersion lens. For confocal microscopy, larvae were anaesthetised in 0.168 mg/ml tricaine and mounted in 1% low-melting agarose (Sigma-Aldrich) containing 0.168 mg/ml tricaine in 15 μ-Slide 4 well glass-bottomed slides (ibidi). Numerical data were determined using 40× confocal images.

Zebrafish larvae were anaesthetised in 0.168 mg/ml tricaine and transferred to a 50 mm glass-bottomed FluoroDish (ibidi). Zebrafish were imaged using a Leica DMi8 SPE-TCS microscope fitted with a Hamamatsu ORCA Flash 4.0 camera attachment using a HC FL PLAN 2.5×/0.07 and HC PLAN APO 20×/0.70 dry lens. Both transgenic zebrafish and whole-mount in situ staining was imaged using a Leica MZ10F stereo microscope fitted with a GXCAM-U3 series 5MP camera (GT Vision).

Larvae (2 and 3 dpf) were imaged using a Zeiss Z1 light-sheet microscope with Plan-Apochromat 20×/1.0 Corr nd=1.38 objective, dual-side illumination with online fusion and activated Pivot Scan at 28°C chamber incubation. Zebrafish were anaesthetised in 0.168 mg/ml tricaine and mounted vertically in 1% low-melting agarose in a glass capillary. Images were obtained using 16 bit image depth, 1400×1400 pixel field of view and GFP visualised with a 488 nm laser at 16% power, 49.94 ms exposure and user-defined z-stack depth (400-600 slices, 0.641 μm slices).

Embryos/larvae were randomly assigned to experimental groups and experimenters were blinded to groups where possible. Sample size (n of larvae) was determined by the number of healthy embryos that were laid in the batch and number of groups within the experiment. As the arg2:GFP line was a new transgenic line, it was not possible to use power calculations pre-experimentation.

Microscopy data were analysed using Leica Application Suite X (LAS X; Leica Microsystems) and Image J software. All numerical data were analysed (Prism 9.0, GraphPad Software) using unpaired two-tailed t-tests for comparisons between two groups and one-way ANOVA (with Bonferroni post-test adjustment) for other data. P-values shown are: *P<0.05, **P<0.01 and ***P<0.001.

Light-sheet imaging was carried out in the Wolfson Light Microscopy Facility, supported by a Biotechnology and Biological Sciences Research Council (BBSRC) ALERT14 award for light-sheet microscopy (BB/M012522/1). The authors would like to thank the BSU Aquarium Team for fish care and the Department of Infection, Immunity and Cardiovascular Disease (IICD) Technical Team for practical assistance (University of Sheffield). We thank the Renshaw and Johnston groups for sharing their expertise in BAC transgenesis, especially Miss Catherine Loynes and Dr Stone Elworthy. We also thank Dr Simon Johnston and Dr Stella Christou for invaluable help and expertise in fungal infection models, and Prof. Stephen Renshaw for his helpful comments and discussions on the manuscript. Lastly, we thank the University of Sheffield, which funded Dr Ffion Hammond with a PhD scholarship.