Effects of repeated anaesthesia on gill and general health of Atlantic salmon, Salmo salar.:

Scottish Fish Immunology Research Centre, University of Aberdeen, Aberdeen, UK Marine Laboratory, Marine Scotland Science, Aberdeen, UK Rowett Institute of Nutrition and Health, University of Aberdeen, Aberdeen, UK Correspondence Rachel Chance, Scottish Fish Immunology Research Centre, University of Aberdeen, Zoology Building, Tillydrone Avenue, Aberdeen, AB24 2TZ, UK. Email: r01rc14@abdn.ac.uk; bertrand.collet@inra.fr Present address Bertrand Collet, Virologie et Immunologie Moléculaires, Institut National de la Recherche Agronomique (INRA), Université Paris-Saclay, Jouy-en-Josas, France Fish are the second most widely utilized vertebrate group used for scientific procedures in the United Kingdom, but the development and application of 3Rs (the principles of replacement, reduction, and refinement) in aquaculture disease research lags behind methodologies in place for mammalian studies. With a need for individual monitoring and non-lethal sampling, the effect of repeat anaesthesia on experimental fish needs to be better understood. This study analyses the effect of repeat anaesthesia with MS-222, metomidate and AQUI-S upon the gill and general health of post-smolt Atlantic salmon Salmo salar. A single, lethal dose of anaesthetic was compared with seven anaesthetizing time points over 28 days, terminating in a lethal dose. No anaesthetic showed significant differences in accumulation in the muscle tissue, or changes in plasma glucose after repeated or single dosing. Fish repeatedly anaesthetized with MS-222 or AQUI-S exhibited upregulation of osmoregulatory genes in the gill and AQUI-S-treated individuals showed, histologically, epithelial lifting from the lamellae capillary irrespective of whether they had a single or repeated dose history. No significant changes were seen in inflammatory or stress genes in the head kidney of fish repeatedly anaesthetized with AQUI-S or metomidate, however MS-222 treatment resulted in upregulation of tnfα3. Repeated anaesthesia with MS222 and metomidate gave a significant decrease and increase in peripheral blood neutrophils, respectively. This study concludes that no increase in cumulative stress or inflammation is induced by the repeated anaesthetization of S. salar with any of the tested anaesthetics, however gill osmotic regulation and blood parameters may be affected.


| INTRODUCTION
Fish are the second most widely used vertebrate group, after mice, for scientific procedures in the UK; a total of 287,000 procedures were carried out on fish in 2016, representing 14% of scientific procedures for that year (Home Office, 2017). While a significant number of these procedures (66%) correspond to studies involving the model species zebrafish Danio rerio (Hamilton 1822), a large proportion utilize foodfish species, such as the salmonids (Home Office, 2017), due to their importance in the growing aquaculture industry (FAO, 2016).
In terms of implementing the 3Rs principles of replacement, reduction, and refinement (Russell et al., 1959), aquaculture research methodologies are currently less developed than systems used for mammalian studies (Sneddon et al., 2017). However, recent work has shown a new focus on reduction and refinement for salmonid pathogen challenges in the framework of non-destructive individual monitoring (Braceland et al., 2015;Chance et al., 2018;Collet et al., 2015;Monte et al., 2016;Urquhart et al., 2016), with the scope that the approach allows to obtain more informative and robust data from the same number of individuals or that fewer fish can be used to obtain similar quality data, (Hall et al., 2018). Individual monitoring relies on the ability to obtain non-lethal samples, such as small amounts of blood or mucus, which in practice would be difficult and cause excessive stress to the fish, without the use of anaesthesia. Therefore, repeat anaesthesia is a key component of the individual monitoring approach. Anaesthetics are also commonly used in other areas of fish research and aquaculture industry, such as during transportation, fieldwork and disease control. Therefore, any contribution to the understanding of how repeat anaesthesia may affect fish health is relevant to multiple fields (Balamurugan et al., 2016;Nilsen et al., 2017;Scott, 2013).
While there are well-established guidelines in place to manage repeat anaesthesia of animals in the mammalian research sphere (Flecknell, 2016), currently, only a limited number of references reporting on teleost fish repeated anaesthetization are available and only few conducting analysis on any general effects. Available articles on the subject focus on ornamental or freshwater species, juvenile life stages, have lengthy intervals between anaesthesia or include no immunological observations in their analysis (Basrur et al., 2010;Braley & Anderson, 1992;de Lima Silva et al., 2012;Hoskonen & Pirhonen, 2004Kildea et al., 2004;Mitjana et al., 2014;Mylonas et al., 2005;Palmer & Mensinger, 2004;Posner et al., 2013;Smith et al., 1999;Soto, 1995).
In this study, the anaesthetics of interest were MS-222, metomidate (AquaCalm; www. syndel.ca) and isoeugenol (AQUI-S; www.aqui-s.com). MS-222, was introduced as an effective, powderbased anaesthetic in 1967 and has since become the most extensively used chemical for sedation, anaesthesia and euthanasia in fish research, aquaculture and ornamental fish rearing (Bourne, 1984;Popovic et al., 2012;Readman et al., 2013). A sulphonated analogue of benzocaine, MS-222 blocks the generation of action potentials through voltage-dependent Na + channels (Ramlochansingh et al., 2014). The use of MS-222 in food-fish is licensed in the United States and Norway under the provision of 21 day withdrawal time, 5 day withdrawal period in Canada and 70 days ( o D; a development value calculated by multiplying the average temperature by number of days incubated) in the United Kingdom (FDA 2007;Kiessling et al., 2009;Popovic et al., 2012).
Metomidate [DL-1-(1-phenylethyl)-5-(methoxycarbonyl) imidazole hydrochloride] is a powder-based, nonbarbiturate, hypnotic anaesthetic which is used in research but not currently approved for use with food fish (Iversen et al., 2003). This anaesthetic was chosen due to its ability to act directly on inter-renal cells, blocking the stress response of cortisol synthesis (Olsen et al., 1995), production of which is shown to inhibit the immune response (Barton & Iwama, 1991;Iversen et al., 2005;Pickering & Pottinger, 1988). Owing to the nature of repeated sampling, a cortisol-blocking anaesthetic may be more favourable in terms of avoiding or reducing extraneous effects when analysing the natural host immune responses to pathogen challenges.
It should be noted that while metomidate acts to block cortisol production, fish may still be experiencing stress, therefore the upmost care in animal handling and sampling methodology should be carried out while using this anaesthetic. Although metomidate is not currently licensed for use in food fish it can be used for research.
Isoeugenol is the active ingredient (50%) found in the gel-based anaesthetic AQUI-S. Isoeugenol is similar to clove oil (eugenol), which anaesthetizes by inhibiting Na + , K + and Ca 2+ channels, as well as Nmethyl-D-aspartate (NMDA) receptors (Lee et al., 2005;Li et al., 2007;Park et al., 2006;Wie et al., 1997). First developed in New Zealand in 1996, AQUI-S is increasing in popularity and licensed internationally; currently AQUI-S can be used with a zero-day withdrawal time for food fish in Australia, Chile, Costa Rica, Honduras, South Korea and New Zealand (AQUI-S, 2018) and as a stressreduction aid for transport and husbandry practices in Norway (Kolarevic & Terjesen, 2014).
This study compared the possible effects of seven repeat anaesthetizations utilizing MS-222, metomidate or AQUI-S over a 28 day period, simulating the sampling intervals suggested by Collet et al. (2015) for individual monitoring of Atlantic salmon Salmo salar L. 1758, with fish which experienced a single overdose of their respective anaesthetic. The approach taken was to emulate procedures around sampling at each time point, encompassing catching of fish, removal from and replacement in water, with the anaesthetic as a variable. The variables investigated were: the accumulation of anaesthetic residues in skeletal muscle tissue as determined by liquid chromatography tandem mass spectrometry (LC-MS/MS); peripheral blood leukocyte populations derived from blood smears; plasma glucose concentration measured by spectrophotometric assay; gill condition was evaluated through histopathology; selected inflammatory, osmotic regulatory and stress-related genes in the gills and head kidney were analysed with quantitative (q) PCR. To the authors' best knowledge, this is the first study to look at the possible effects of repeated anaesthesia in post-smolt S. salar, with a focus on gill health and immunological parameters.

| Fish
This study was carried out in strict accordance with the UK Animals (Scientific Procedures) Act 1986 (ASPA) under project licence 70/7897. The Marine Scotland Science Ethical Review Committee approved the protocols and all housing and procedures were carried out in a contained biosecure aquarium facility at Marine Scotland, Aberdeen (MSS). Freshwater S. salar parr were sourced from a commercial aquaculture company and carried through smoltification at the MSS aquarium. A total of 120 individuals (mean AE SD mass = 219 AE 42 g) were randomly distributed into twelve 1 m diameter (350 l) tanks (n = 10). Throughout the experiment, fish were held at 13 C, in full strength seawater (ca. 35 ppt) in a flow-through system with approximately 7.3 mg l −1 oxygen and ambient, natural photoperiod light conditions (57.15 N). Fish were acclimatised in experimental tanks for 1 week and hand-fed daily throughout with Nutra 50 (Skretting; www.skretting.com), except for a period of 24 h when food was withheld prior to terminal sampling.

| Anaesthetic exposures
Three tanks were allocated to each repeated exposure treatment for the three anaesthetics. Concentrations used for MS-222, metomidate and AQUI-S were 80, 12 and 17 mg l −1 respectively, selected due to previous effective protocols and standardized use at MSS or manufacture's recommendations (AQUI-S, 2018). AQUI-S is a viscous gel-like substance, therefore the required volume needed was calculated using the AQUI-S calculator, available on the manufacturers website (AQUI-S, 2018). In order to disperse AQUI-S evenly, it was first emulsified in 1 l of seawater and inverted until the solution turned an opaque white colour.
Anaesthetising points started at day 0 and continued every 4 days (Collet et al., 2015) until day 28 when all fish were terminally sampled. Throughout the experiment, to mitigate any stress experienced by the fish during the anaesthetizing process, an in-tank anaesthesia methodology was applied as described previously by Collet et al. (2015). Briefly, tanks were slowly drained to 120 l, liquid stock solutions of each anaesthetic were poured into their respective tanks in a figure of eight to ensure even dispersal throughout the tank and fish remained submerged until reaching Stage III, Plane 1 deep narcosis anaesthesia, as described by Burka et al. (1997). Induction times were recorded for each tank at each anaesthetizing time point. Once the required level of anaesthesia had been reached, fish were individually netted out and placed on a post-mortem to housing tanks once these had been refilled to 120 l; tanks continued to refill to full capacity of 350 l.

| Terminal sampling
The experiment was terminated at day 28 with a lethal dose of the respective anaesthetic for each treatment group, including the thus far undisturbed control tanks for each anaesthetic (n = 10). In-tank anaesthesia methodology used throughout the experiment was applied, with the same concentrations of anaesthetic as previously  Instrument control and peak integration and quantification were performed using Thermo Xcalibur 2.0.7 SP1 (Thermo Fisher Scientific).

| Histological examination
Metomidate and MS-222 calibration curves best fit a quadratic function with equal weighting whereas weighted least-squares linear regression with a weighting factor of x −1 was used to quantify isoeugenol concentration in unknown samples by comparing peak-area ratios (analyte:IS) with those obtained from a multi-level calibration curve.

| RNA extraction & cDNA synthesis
Once gill and head-kidney samples were defrosted, excess RNAlater (Sigma Aldrich) residues were removed from the tissues by compressing with sterilized tissue paper.

| Histological examination
No frequent or statistically relevant levels of inflammation, hyperplasia, lamellar fusion, filament fusion, or necrotic tissue were seen in any of the samples inspected for any of the anaesthetic treatments (data not shown). Epithelial capillary lifting from the lamellae was seen in all AQUI-S treated gills (Table 2 and Figure 2). Fish that had been repeatedly anaesthetized with AQUI-S had a mean (AESD) score of 0.6 AE 0.2, whereas fish which had a single overdose of

| Accumulation of residues in muscle tissue
There were notable differences between the levels of detected residues between the different anaesthetics used in this study after a single dose, particularly when evaluating the differences in concentrations required for an appropriate level of sedation; MS-222 (80 mg l −1 ) 45 mg g −1 AE 13, metomidate (12.5 mg l −1 ) 4 μg g −1 AE 1, AQUI-S (17 mg l −1 ) 112 μg g −1 AE 36 (Figure 4). No significant differences (P > 0.05) were found between detected skeletal muscle residues from fish that had received a single dose compared with those that saw multiple anaesthetizing timepoints for any of the anaesthetics tested (Figure 4).

| Gene expression in gill tissue
The abundantly expressed isoform of pro-inflammatory cytokines Metomidate treated fish had no significant differences in expression of any osmoregulatory marker genes, namely Na-K-ATPase α1a, α1b, α1c, α3, β1, Na-K-Cl cotransporter (nkcc) and cystic fibrosis transmembrane conductance regulator 1 and   Figure 6). Gene expression analysis of the inflammatory cytokines Il-1β, tnfα1 and tnfα3 also revealed no significant differences between single dose and repeated dose exposure for metomidate and AQUI-S, however upregulation of tnfα3 was seen in fish repeatedly anaesthetized with MS-222 when compared with single dose treated fish.

| Plasma glucose concentration
Plasma glucose concentrations showed no significant differences between fish that had been repeatedly anaesthetized and lethally dosed with their respective anaesthetic compared with undisturbed fish that had only a single lethal dose of respective anaesthetic ( Figure 7).

| Peripheral blood leukocytes
Analysis of counts of 200 leukocytes from each fish utilized in the study showed that there were no significant differences between fish that had been repeatedly anaesthetized with AQUI-S and then euthanized with a single lethal dose v. those that had been left undisturbed then euthanized with a single lethal dose (Figure 8).
Metomidate treated fish had no significant differences between the populations of monocytes, thrombocytes or lymphocytes when comparing single and repeated dosing, but repeat exposure of fish had a significantly increased proportion of neutrophils ( Figure 8). The opposite was the case for MS-222 treated fish, in which repeatedly exposed fish had a significantly lower proportion of neutrophils (P < 0.05), while the other leukocyte types showed no significant differences (Figure 8).

| DISCUSSION
Monitoring gill health is a priority when evaluating the effect on fish of non-lethal monitoring that involves repeated removal from water and exposure to anaesthesia. This is due to the importance of the gills from multiple standpoints: they are the site of uptake for immersion anaesthesia (Hunn & Allen, 1974); their biological importance in regulatory functions (Evans et al., 2005); the welfare of the fish (Pettersen et al., 2014). ), 25th and 75th percentiles ( ), range ( ) and outlier ( ) glucose concentrations of plasma from Salmo salar after single (n = 10) or repeat exposure (n = 30) to the anaesthetics MS-222 (80 mg l −1 ), metomidate (12.5 mg l −1 ) and AQUI-S (17 mg l −1 ). No significant differences were found using a Van der Waerden test Furthermore, the European aquaculture industry is facing increasing threat from gill-related health issues, such as amoebic (AGD) (Oldham et al., 2016), proliferative (PGD) (Durborow et al., 2015) gill disease and damage due to phyto and zooplanktonic species (Hallegraeff et al., 2017;Treasurer et al., 2003). Therefore, any challenge studies investigating individual fish responses to gill insults requires the assurance that hostinsult interactions are being analysed, as opposed to reactions to sampling processes . This study focused on assessing the effects of repeated anaesthesia with three different compounds, incorporating the simulation of sample collection wherein fish were removed from water and exposed to air for appropriate periods of time needed for the collection of blood or mucus (repeat sampling was not performed and any effect not assessed here).  (Gomulka et al., 2008). It should be noted that in-tank anaesthesia was not utilized in these studies, consequently, the stress induced by capture and dosing techniques cannot be ruled out as the underlying cause of the observed changes. Since sturgeons are chondrostean fish they are likely to have a different haematological profile in comparison with teleosts (Gomulka et al., 2008), however fish repeatedly exposed to MS-222 during this study also had a significant decrease in their peripheral neutrophil populations (Figure 8).
Throughout this experiment MS-222 was administered in an unbuffered form; the formation of methanesulphonic acid can reduce the pH of aquarium environments (Smith et al., 1999), which leads to adverse reactions in fish such as metabolic imbalances, increased blood pressure, haemoconcentration, epidermal and corneal irritation (Davis et al., 2008;Milligan & Wood, 1982;Packer 1979). The salinity of seawater offers some buffering protection from extreme dips in pH (Popovic et al., 2012), but the repetitive administration of this unbuffered form may be the cause of the significant upregulation of the pro-inflammatory cytokine tnfα3 in the head kidney and of Na-K-ATPase β1 and cftr2 in the gills (Figures 6 and 7). Any change in seawater pH was not monitored throughout this experiment. Moving forward, it is suggested that a protocol using buffered MS-222 for future individual monitoring challenges should be developed.
A lack of significant differences in the expression of secondary stress response indicators, such as upregulation of isoforms of hsp90 in the head kidney or significant changes in plasma glucose concentration, after repeat exposure to any of the anaesthetics tested in the present study may suggest that fish do not find the practice of repeated anaesthesia (compared with single overdose of such anaesthetic) cumulatively stressful. Iversen et al. (2003) reported that freshwater S. salar smolts also had no rise in plasma glucose after a single dose of metomidate (up to 10 mg l −1 ) or AQUI-S (up to 100 mg l −1 active ingredient). However, behavioural (Pirhonen & Schreck, 2003;Readman et al., 2013;Wong et al., 2014) and blood glucose stress responses (Bourne, 1984;Davis & Griffin, 2004;Larter & Rees, 2017;Molinero & Gonzalez, 1995;Small & Chatakondi, 2005;Soivio et al., 1977;Thomas & Robertson, 1991;Wagner et al., 2002) have been reported in a variety of fish species exposed to a single dose of the same anaesthetics tested in this study. While the role of in-tank anaesthesia in the reduction of stress has yet to be quantified, it seems likely this methodology plays an important role in minimizing any stress response. A rise in plasma glucose has been strongly linked to increased production of primary stress indicators such as the glucocorticoid steroid cortisol, which was not assessed during this study due to time constraints, lack of required resources and prioritisation of gill-health analyses (Farbridge & Leatherland, 1992;Iversen et al., 1998;Pottinger, 2008). Nevertheless, it may be useful to measure the cortisol response of S. salar undergoing repeat anaesthesia directly in future studies, especially as it has been shown to be a better indicator of chronic stress in other species, such as the red porgy Pagrus pagrus (L. 1758) (Rotllant & Tort, 1997). It should be noted that the level of response of plasma glucose and cortisol to both acute and chronic stressors differ widely between species (Barton, 2002), confounding comparisons of this variable when applying repeat anaesthesia methodology to different species.
Arguably the most concerning result from this study is the significant upregulation of osmotic control marker genes in the gills of fish repeatedly anaesthetized with AQUI-S ( Figure 5) and the epithelial lifting in fish following both repeat and single exposure (Table 2 and Figures 2 and 3). Na-K-ATPase activity is required for ion secretion, isoforms of which are upregulated in the fish gill during smoltification (Boeuf & Prunet, 1985;Pelis & McCormick 2001). Cortisol treatment has been shown to induce Na-K-ATPase α1a and α1b upregulation in vivo and in vitro in S. salar gill (Kiilerich et al., 2007;McCormick et al., 2008). A decrease of Na-K-ATPase enzyme activity in postsmolt (in seawater) salmonid gills is normally a sign of oxidative stress, often in response to environmental stressors and leads to osmoregulatory failure (Murray et al., 2017). Epithelial lifting from the gill capillary, which can be seen as a defence mechanism to reduce superficial surface area contact with the external environment (Mallatt, 1985), has been widely described in fish exposed to environmental stressors (Bruno & Ellis, 1988;Figueiredo-Fernandes et al., 2007;Frances et al., 2000;Good et al., 2017;Jiraungkoorskul et al., 2003;Martinez et al., 2004;Pretti et al., 2006;Skidmore & Tovell, 1972;Smart, 1976;Triebskorn et al., 2004) and after bath emersion to levamisoleadjuvanted Vibrio anguillarum vaccine (Morrison et al., 2001). In this study, fish treated with a single lethal dose of AQUI-S showed double the mean score of epithelial lifting when compared with fish which had been repeatedly anaesthetized with AQUI-S then lethally dosed (mean AE SD = 1.2 AE 0.2 and 0.6 AE 0.2, respectively), but both treatments' scores remain low and reflect light pathological signs (Table 2).
A common role of anaesthetic use in fish research is the application of a controlled overdose in humane killing, which leads to wide concentration ranges of AQUI-S reported in the literature, ranging from 12 mg l −1 -540 g l −1 specifically for post-smolt S. salar (Hevrøy et al., 2005;Sanden et al., 2004). A concentration of 17 mg l −1 of AQUI-S was used in this study as recommended for the sedation of mature S. salar by the manufacturer (AQUI-S New Zealand LTD), but the duration required to reach desired levels of sedation (mean AE SD = 21 AE 3 min) was, in our hands, double that expected based on guidelines detailed on the manufactures website (AQUI-S, 2018) and > seven and 10 fold that needed for MS-222 and metomidate, respectively ( Figure 1). Tanks were fully aerated while fish were exposed to anaesthetic, with approximately 7.3 mg l −1 oxygen maintained throughout the study. It is possible that the length of immersion in AQUI-S may be the cause of the epithelial disruption seen, which in turn may have affected hydromineral balance resulting in the significant upregulation of the Na-K-ATPase genes in the repeatedly anaesthetized fish. Interestingly, epithelial lifting has also been reported in gills of control fish euthanized by exposure to a high dose of AQUIS-S (5 ml l −1 , equivalent to 5.45 g l −1 ) for an unspecified period of time during an AGD microarray study, which the authors deemed smoltification or sampling responsible (Morrison & Nowak, 2008). However, in the context of the present results, AQUI-S exposure may be the true cause. It is important to note, however, that epithelial lifting is commonly falsely diagnosed when assessing gill histopathology (Wolf et al., 2015), with the separation seen being an artefact arising from conditions of sampling or fixation (Ferguson et al., 2006;Wolf et al., 2015), but it is unlikely in this case as little to no lifting was seen in MS-222 or metomidate treatments. Speare and Ferguson (1989) found gills fixed with 10% NBF, the fixative used in this study, were more likely to exhibit separation artefacts and therefore future studies seeking to further investigate this effect should make use of alternatives such as Bouin's, Dielrich's or Davidson's solutions (Wolf et al., 2015). Further study is needed to elucidate if the length of exposure to AQUI-S can indeed cause osmotic regulatory issues and furthermore, whether a higher concentration would reduce induction times to Stage III Plane 1 anaesthesia and mitigate these observed effects or cause further negative reactions. If individual monitoring is to be widely adopted to improve the investigation of the immune responses in a variety of different species, to a variety of different pathogens, more experiments are required to identify the effects of repeat sampling for each intended host species and challenge condition. Over time, a better picture should emerge to mitigate any extraneous effects due to repeated sampling from the same individual in a challenge model.
In conclusion, this study investigated the effects of repeat anaesthesia for post-smolt S. salar as required for individual monitoring, with a focus on gill health. Unbuffered MS-222 gave a significant decrease in peripheral neutrophils, upregulation of tnfα3 in the head kidney and of Na-K-ATPase β1 and cftr2 in the gills. Administration in buffered form for non-lethal sampling challenges may alleviate some of these effects. AQUI-S repeat exposure resulted in upregulation of most osmoregulatory genes examined in the gills and low-level epithelial lifting in both single and repeated dose treatment groups was seen.
AQUI-S also required substantially longer times for induction of anaesthesia compared with MS-222 and metomidate. Therefore, further studies on repeat exposure to anaesthetics should be performed with post-smolt salmonids using higher concentrations of AQUI-S to determine if this might mitigate against potentially harmful effects of the longer induction durations required to achieve an appropriate stage of anaesthesia.