Habitat selection disruption and lateralization impairment of cryptic flatfish in a warm, acid, and contaminated ocean
Anthropogenic release of greenhouse gases is leading to significant changes in ocean physicochemical properties. Although marine organisms will have to deal with combined effects of ocean warming and acidification, little is known about the impact of interactions between these climate change variables and contaminants. Nowadays, mercury emissions are mostly of anthropogenic origin, and part of these emissions is deposited in the ocean sediment. Within this context, our goal was to determine the acclimation potential of a benthic flatfish, Solea senegalensis, to future climate change scenarios and methylmercury (MeHg) neurotoxicity. After 28 days of exposure under three-factor crossed treatments of MeHg contamination (non-contaminated and contaminated feed, 0.08 ± 0.02 and 8.51 ± 0.15 mg kg−1 dry weight, respectively), high CO2 (ΔCO2 ≈ 500 ppm), and temperature (ΔT = 4 °C), we investigated brain mercury accumulation, habitat preference, and relative/absolute lateralization, as well as acetylcholinesterase (AChE) activity in five brain regions. Our results indicate a differential effect of hypercapnia (decrease) on brain mercury accumulation. MeHg-contaminated flatfish displayed decreased AChE activity, impaired lateralization, and bottom choosing judgment. Contaminated fish spent significantly higher amounts of time in the complex habitat, where they could neither bury nor match the background. While warming led to higher enzymatic activity, acidification decreased Hg accumulation, but also affected AChE activity and disrupted habitat selection. Present-day MeHg environmental concentrations may lead to severe disruption of behavioral and neurological functions, which, combined with ocean warming and acidification, might further jeopardize the ecological fitness of flatfish.
Since the industrial revolution, global climate is changing due to the rise in anthropogenic-related atmospheric CO2 concentrations, which are predicted to reach nearly 1000 ppm by 2100. As atmospheric CO2 dissolves in the ocean, seawater chemistry is altered, leading to a drop of mean ocean pH levels between 0.14 and 0.35 U (IPCC 2013). In parallel, average sea surface temperature has also been rising and a further increase of 2–4 °C in mean sea surface temperature is predicted by 2100 (IPCC 2013). Both these stressors, i.e., increased temperature and CO2, are expected to condition fish toxicological response in the future (Schiedek et al. 2007; Marques et al. 2010). In field and laboratory experiments (Dijkstra et al. 2013), increased temperature lead to higher MeHg accumulation in fish. In contrast, the toxicological potential of increased CO2 is less known, but lower fitness is expected due to suppression in animal physiology by hypercapnia (Langenbuch and Pörtner 2004; Michaelidis et al. 2007). Therefore, taking ocean warming and acidification into account is vital to pinpoint future toxicological effects in fish.
Methylmercury (MeHg) is the most toxic form of organic mercury and an ubiquitous environmental pollutant that has been subjected to intensive research (Chang 1977). This hydrophobic pollutant is easily capable of crossing blood–brain barrier, accumulating (and demethylating into mercury, i.e., Hg) in neural cells within specific parts of the brain, such as optic tectum, cerebellum, olfactory system, and rhombencephalon (Friberg and Mottet 1989; Rouleau et al. 1999). Such preferential accumulation triggers severe morphological abnormalities in key brain areas, posteriorly linked to loss of motor coordination, impaired predatory avoidance, and higher aggressive behavior (Beliles and Yule 1968; Baatrup 1991; Sakamoto et al. 2002).
Since organic forms of mercury accumulate in the sediment (Lee et al. 1998), MeHg biomagnification first dictates serious ecophysiological challenges to benthos-associated food webs (Mason 2001). By linking the pelagic and benthic environments, flatfish are key organisms in the food web of ecosystems worldwide (Gibson et al. 2014). These fish use the seabed as habitat to bury and camouflage (mimicking colors and shades), avoiding detection by both predators and prey (Sumner 1911; Saidel 1988; Healey 1999). Thus, as flatfish primarily choose benthic substrate by their cryptic potential (Ryer et al., 2008), i.e., temperate water flatfish, such as the Solidae family, avoid coarser sediments with strong distinct colors (Kelman et al. 2006), habitat selection is a vital process in this species’ ecology. In parallel, another relevant behavioral process reported in vertebrate species is lateralization (Jozet-Alves et al. 2012). Derived from brain and flatfish body asymmetry (Bisazza et al. 1998, 2000), lateralization yields higher cognitive performance (Domenici et al. 2014), resulting in faster and more efficient responses to external stimuli (e.g., escape responses). The loss of lateralization diminishes success in a number of cognitive tasks and predator response behaviors (Domenici et al. 2012). Combined with loss of habitat selection judgment and inability to crypsis, these impairments may severely decrease flatfish environmental fitness.
Processes like predator–prey response, spatial orientation, and color adaptation are mediated by sensory systems dependent on neurotransmitters, e.g., vision, lateral line, chemotaxis (Hara et al. 1976; Beauvais et al. 2001; Leclercq et al. 2010). Recent studies point out that high CO2 causes ionic balance deregulation in proton-based neurotransmitter receptors, e.g., GABA-A neurotransmitter, leading to increased fish anxiety and boldness (Nilsson et al. 2012; Hamilton et al. 2014). Across brain areas, there is ample evidence of connection between the GABAergic and the cholinergic neural systems (Decker and McGaugh 1991), the latter being a prominent neurotransmitter associated not only with learning and memory in the central nervous system (Donald 1998), but also with neuromuscular junction transmission in the peripheral nervous system (Castoldi et al. 2001). Hg inhibits acetylcholinesterase (AChE) activity by binding to its acetylcholine (ACh) receptor, which prevents electric signal propagation through postsynaptic neurons linked to target cells (Beauvais et al. 2001). Thus, severe cases of Hg-associated neurotoxicity in the cholinergic pathways may lead to mortality (Coccini et al. 2000; Vieira et al. 2009).
The physiological and behavioral consequences of MeHg exposure on biota are still unknown under a climate change context. Moreover, most of the ocean climate change-related research has been done using a single stressor approach, not taking into account potential synergistic effects. Within this context, here we undertook a novel multistressor experiment aimed to assess how behavioral (namely habitat selection, lateralization) and neurophysiological (AChE brain activity) responses of an important flatfish species, Solea senegalensis (Linnaeus, 1758), will be affected by MeHg contamination, ocean warming, and acidification.
Material and methods
Juvenile Solea senegalensis (mean ± SD; total length: 8.73 ± 0.89 cm; weight: 7.63 ± 2.08 g) were acquired from CRIPSul–IPMA (Centro Regional de Investigação Pesqueira do Sul–Instituto Português do Mar e da Atmosfera, Olhão, Portugal) and transported within thermal isolated containers, with constant aeration, to Laboratório Marítimo da Guia (Cascais). Upon arrival, fish were randomly placed within a total of 24 independent experimental tanks (30 fish per tank, 720 fish in total, corresponding to 28.6 fish m−2). Each experimental tank (0.35 m2 bottom area, 70 L capacity) was fit within an independent recirculating aquaculture system (RAS). Each RAS was equipped with biological (ouriço®, Fernando Ribeiro Lda, Portugal) and physical filtration (ReefSkimPro 850, TMC Iberia, Portugal), with additional UV disinfection (Vecton 120 nano, TMC Iberia, Portugal). Ammonium, nitrate, and nitrite were kept below detectable levels (Aquamerk). Fish were laboratory-acclimated for a time period of 1 week, at the following conditions: temperature (18 °C), salinity (35), and pH total scale (pHt 8.0). Photoperiod was kept at 14 h–10 h (light–dark), according to prevailing natural light conditions.
Natural seawater (NSW) was pumped straight from the sea into a storage seawater tank (8 m3 total capacity). Subsequently, seawater was 0.35-µm filtered (Fernando Ribeiro, Portugal) and UV-irradiated (Vecton 600, TMC Iberia, Portugal), before being supplied to mixing (n = 24) and experimental (n = 24) tanks/RAS, respectively. All the RAS were operated as semi-closed systems. Overhead tank illumination was provided, and salinity (35 ± 1) and dissolved oxygen (6.50 ± 0.52 mg L−1) were daily monitored and adjusted as needed. Temperature and pHt were daily checked in the holding tanks, using a handheld multiparametric device (Multi 3420 SET G, WTW, Germany). Temperature was kept by means of an automatic seawater chiller system (± 0.1 °C, Frimar, Fernando Ribeiro Lda, Portugal), connected to submerged 200-W digital heaters (V2Therm, TMCIberia, Portugal). A computerized controlling system (Profilux 3.1N, GHL, Germany), linked to individual pH probes (GHL, Germany), adjusted pH values automatically. Monitoring was performed every 2 s, and pH values were lowered by injection (submerged air stones) of a certified CO2 gas mixture (Air Liquide, Portugal) or upregulated by aeration with CO2-filtered air (using soda lime, Sigma-Aldrich). Experimental design of tank array followed the schematic setup described in Cornwall and Hurd (2015; Fig. 3d).
Following laboratory acclimation, fish were exposed for 28 days to the following experimental conditions: (1) control scenario (19 °C and pHt 8.0; pCO2 ≈ 350 ppm), (2) hypercapnic scenario (19 °C and pHt 7.6; pCO2 ≈ 850 ppm), (3) warming scenario (23 °C and pHt 8.0; pCO2 ≈ 350 ppm), and (4) combined warming and hypercapnic scenario (23 °C and pHt 7.6; pCO2 ≈ 850 ppm; Table S1). Seawater carbonate system speciation was calculated weekly from total alkalinity (spectrophometrically at 595 nm), temperature, salinity, and pHt measurements (Sarazin et al. 1999). Quantification of pHt was accomplished via a Metrohm pH meter (826 pH mobile, Metrohm, Filderstadt, Germany) connected to a glass electrode (Schott IoLine, SI analytics, ±0.001) and calibrated against TRIS–HCl (TRIS) and 2-aminopyridine-HCl (AMP; Mare, Liège, Belgium) seawater buffers (Dickson et al. 2007). Measurement of pHt was taken under temperature-controlled conditions using a water bath (Lauda, Germany, ±0.1 °C). Bicarbonate and pCO2 values (Supporting Table S1) were calculated using the CO2SYS software (Lewis and Wallace 1998), with dissociation constants (Mehrbach et al. 1973), as refitted by Dickson and Millero (1987).
Fish in contaminated treatments were fed a MeHg-contaminated feed (inserted in the form of MeHg(II) chloride, CH3ClHg, 99.8 %, Sigma-Aldrich, previously solubilized in ethanol; MeHg concentration: 8.51 ± 0.15 mg kg−1 dry weight), while fish in non-contaminated treatments were fed with a custom-made aquaculture feed for sole (MeHg concentration: 0.08 ± 0.02 mg kg−1 dry weight). Methylmercury concentrations used mimicked highly mercury-contaminated estuaries, e.g., Ria de Aveiro (Nunes et al. 2008). Feeds were prepared by a specialized fish feed producer (SPAROS Lda), and fish were daily fed 1.5 % of their body weight, divided into three portions per day. Each treatment was replicated in three tanks, and at the end of the 28-day acclimation period, behavioral trials were performed (n = 10 fish per tank, unrepeated between trials). Standard fish length was measured, while the brain was collected and divided into five macroareas. To assure the biochemical integrity of the samples, no anesthetic was used, and fish were euthanized by a swift severing of the spinal cord. Telencephalon, diencephalon, cerebellum, brainstem, and optic tectum were individually weighted and frozen at −80 °C, until they were analyzed for mercury (Hg) quantification (whole brain, n = 3 per tank) and acetylcholinesterase activity (five brain regions, n = 5 per tank).
Total mercury (Hg) brain concentration was calculated, as MeHg naturally demethylates in neural cells (Friberg and Mottet 1989). Methylmercury was extracted from the feed as described by Scerbo and Barghigiani (1998), i.e., freeze-dried samples (approximately 200 mg) were hydrolyzed in 10 mL of hydrobromic acid (47 % w/w, Merck), followed by MeHg extraction with 35 mL toluene (99.8 % w/w, Merck) and toluene removal with 6 mL cysteine aqueous solution (1 % l-cysteinium chloride in 12.5 % anhydrous sodium sulfate and 0.775 % sodium acetate; Merck). Total Hg was determined in all samples (10–15 mg for solids or 100–200 μL for liquids) by atomic absorption spectrometry (AAS), following the method 7473 of the EPA (2007), using an automatic Hg analyzer (AMA 254, LECO, USA). Hg concentrations were calculated from linear calibration (using, at least, five different standard concentrations), with an Hg(II) nitrate standard solution (1000 mg L−1, Merck) dissolved in nitric acid (0.5 mol L−1, Merck), and the detection limit was 0.005 mg kg−1, wet weight (ww). Accuracy was checked through the analysis of the certified reference material DORM-4, and results obtained in the present study were within the certified range of values (Table S2). A minimum of three measurements (replicates) were taken per sample. Blanks were always tested in the same conditions as the samples. Prior to utilization, all laboratory ware was cleaned with nitric acid (20 % v/v) for 24 h and rinsed with ultrapure water to avoid contamination. All standards and reagents were of analytical (pro analysis) or superior grade.
Brain regions were separated in 1.5 mL eppendorfs and manually homogenized in 150 µL of 50 mM saline phosphate buffer (PBS, pHt 7.0–7.4). Eppendorfs were then centrifuged for 10 min at 10,000 rpm. The samples were kept on ice throughout the homogenization process, and the centrifuge temperature was set to 4 °C. Supernatant was retrieved and kept at −80 °C. AChE activity was determined following the method described by Ellman et al. (1961), adapted to 96-well plates and microplate reader (Dizer et al. 2001). AChE measurements were duplicated by placing 50 µL of sample in two wells. Blank calibration consisted of 50 µL of PBS. 500 µL of 50 mM (3.96 mg mL−1) 5,5′-dithio-bis-(-2-nitrobenzoic acid) and 200 µL of 75 mM (21.67 mg mL−1) acetylthiocholine iodide were mixed in a 14.3 mL of PBS solution. 250 µL of this mixed solution was added to each well. The plate was incubated at 20 °C for 10 min and read in intervals of 1 min by a microplate reader (UVM-340, Asys) at 415 nm. Protein measurement was based on Bradford (1976) adapted to microplate reader, which indicates protein concentration through the absorption at 595 nm of the protein-binding Coomassie Brilliant Blue G-250. Sample measurements were duplicated by adding 200 µL of 5 % Bradford reagent solution to 20 µL of sample in two wells. Standards were obtained by adding increasing concentrations of bovine serum albumin (BSA) to distilled water. Blank calibration consisted of 200 µL of Bradford reagent. AChE measurements were standardized over protein concentration.
Lateralization index was determined using a detour test according to Bisazza et al. (1998). The detour test consisted in a two-way T-maze with a central runway and a movable barrier at the end. The apparatus is represented in the Supporting Information (Fig. S1). Before each trial, the animals were allowed 2 min to acclimate to the new tank (Domenici et al. 2014). At the start of every run, the animals were placed in the starting point and stimulated to swim forward by gently touching the tail region. When fish reached the barrier, a choice had to be made which way to turn. Ten consecutive runs per fish were executed in ten randomly chosen specimens per tank. Turning direction was scored visually. To minimize any irregularities of the detour tank, the tests were performed alternately between both ends of the T-maze. As flatfish rarely move vertically in the water column, water depth in the tank was 10 cm. Temperature and pH conditions of each treatment were maintained by using water from said treatments.
In order to determine turning bias, the relative lateralization index formula was used, according to Bisazza et al. (1998): LR = [(turn to the right − turn to the left)/(turn to the right + turn to the left)] × 100. This formula yields the animal’s turning preference, with values tending to 100 representing animals that turned right on all 10 trials, and −100 representing animals that turned left on all 10 times. Values near zero represent animals that equally turned left or right. Mean LR allowed to assess the turning bias on a population level. In addition, the absolute lateralization index (LA) of each animal was calculated. This corresponds to the absolute value of LR of each fish, thus ranging from 0 (animal turns equally to left and right) to 100 (animal turns either left or right on all 10 trials).
Determination of habitat preference was performed in individual rectangular tanks (area = 1092 cm2) and based on Stoner and Titgen (2003) habitat selection and burial experiments. Half of the tank bottom was composed of smooth sand (enabling crypsis), while the other half was characterized by a complex substrate composed by a mixture of smooth sand and multicolored coarse sediment, where flatfish could neither bury nor match the background (Fig. S2). In the previous 28 days, fish were maintained in glass tanks without the designated bottoms, in order to prevent habituation and biasing habitat preference. Water depth was 10 cm above the bottom, and experimental conditions were kept stable through water renewal. Before each trial, the animals were gently removed from their original tanks and placed randomly in dual-bottom tanks. On every run, 20 preference tanks were used, and a video camera was placed above the tanks for 3 h, avoiding any human contact. All tank walls were covered in white tape, and all sediment and water were replaced between runs, to eliminate cues left from previous fish. Videos were analyzed using a common video player, and cumulative time spent on each habitat was scored visually. Ten randomly chosen specimens per tank were tested.
GLM analysis of total mercury (total Hg) accumulation in S. senegalensis brain in the treatments exposed to MeHg, after 28 days
Total Hg quantification
GLM: total Hg in function of T and CO2
Family = Gaussian
AIC = 49.88
GLM analysis of AChE activity in five S. senegalensis brain regions (BR, 5 levels: DIE diencephalon; TEL telencephalon; BS brainstem; OT optic tectum; CE cerebellum) after 28 days under crossed treatments of MeHg exposure (2 levels, non-contaminated and contaminated), warming (T, 2 levels: 19 and 23 °C), and acidification (CO2, 2 levels: ambient CO2 = 400 ppm and CO2 = 900 ppm)
GLM: AChE in function of BR + MeHg × T × CO2
MeHg × T
MeHg × CO2
T × CO2
MeHg × T × CO2
Family = Gamma (log function)
AIC = 877.16
GLM analysis of S. senegalensis relative and absolute lateralization after 28 days under three-factor crossed treatments of MeHg exposure (2 levels, non-contaminated and contaminated), warming (T, 2 levels: 19 and 23 °C), and acidification (CO2, 2 levels: ambient CO2 = 400 ppm and CO2 = 900 ppm)
Relative lateralization (LR)
GLM: Lateralization in function of MeHg
Family = Gaussian
AIC = 786.99
Absolute lateralization (LA)
GLM: Lateralization in function of MeHg × T × CO2
MeHg × T
MeHg × CO2
T × CO2
MeHg × T × CO2
Family = Gamma (log function)
AIC = 688.18
GLM analysis of the cumulative time spent in complex habitat by S. senegalensis, after 28 days under three-factor crossed treatments of MeHg exposure (2 levels, non-contaminated and contaminated) and acidification (CO2, 2 levels: ambient CO2 = 400 ppm and CO2 = 900 ppm best fit modeling using AIC deemed warming (T) unnecessary to explain the data; thus, only MeHg and CO2 were used as factors
Habitat preference (time spent on complex habitat)
GLM: Time spent in complex habitat in function of MeHg × CO2
MeHg × CO2
Family = Gamma (inverse function)
AIC = 746.2
Our results showed that climate change stressors (ocean acidification and warming) interact with methylmercury exposure, affecting S. senegalensis neurophysiology and behavior. However, mortality rates were minimal in our 28-day multistressor experiment, showing that coastal/estuarine temperate flatfish accustomed to ever mutable and often contaminated habitats (Fonseca et al. 2011) are resilient (survival wise) to MeHg exposure and changes in abiotic parameters such as increased temperature and CO2.
Brain accumulation of MeHg decreased under high CO2, which may be related to reduced metabolic rates. Low-mobility organisms (e.g., gastropods, bivalves, crustaceans) show metabolic arrest in response to increased extracellular acid–base stress (Kroeker et al. 2010), which may have been displayed by juvenile soles given their low energy-demanding almost sessile lifestyle (Leakey et al. 2009). Theoretically, the likely prioritization of acid–base regulation and ion regulatory enzyme machinery for CO2 excretion (e.g., pyruvate kinase) leads to lower metabolic activity as reported in other fish (Perry et al. 1988). As MeHg accumulation rates are positively correlated with metabolic rates (Dijkstra et al. 2013), these results support the claim that acidification affects toxic compound accumulation rates (Schiedek et al. 2007).
AChE activity was highly affected by the tested stressors in all brain regions. The diencephalon and telencephalon are closely related and part of the limbic system (monoaminergic pathways) (Andén et al. 1966; Summers and Winberg 2006), while the brainstem acts as a bidirectional key communication point in the brain–body axis (Basbaum and Fields 1984; McClellan 1984), and the optic tectum plays a crucial role in visual acquisition and response, as well as pre-motor functions (Stevens 1973; Springer et al. 1977; Roeser and Baier 2002). Altered neural activity in these regions has the potential to impact sensorial acquisition systems, cognitive capabilities, psychological well-being, deliberate, and unconscious motor actions (e.g., escape behavior), overall severely affecting potential responses to environmental stimuli. It is worth noting that AChE activity was particularly low in the cerebellum, a brain region sensitive to Hg accumulation (Rouleau et al. 1999). The cerebellum is involved in many cognitive processes (reviewed in Rapoport et al. 2000) and, more importantly here, in exploratory and fear-related behaviors, which are sometimes interpreted as boldness (Caston et al. 1998; Frisch et al. 2000). Bearing in mind that this region is already characterized by low AChE activity (Toscano-Márquez et al. 2013), we argue that preferential Hg accumulation here (Rouleau et al. 1999) has the potential to be even more deleterious than in other brain regions, such as the diencephalon.
Likely due to increased overall metabolic rates (Dijkstra et al. 2013), AChE activity was increased by warming, while MeHg exposure decreased activity, due to competitive binding to its receptors (Coccini et al. 2000). Regarding the negative effects prompted by acidification, CO2 is a known disruptor of basal inhibitory potential on the GABA-A receptor, generating shifts in ionic concentrations of Cl− and HCO3− (Nilsson et al. 2012). This leads to increased GABA release, which is linked to inhibition of brain cholinergic activity (Giorgetti et al. 2000). Thus, we hypothesize that the CO2 effect detected could be an indirect consequence from altered concentrations of GABA neurotransmitter, which led to stronger AChE inhibition.
The results registered in the behavior trials support the AChE neurotoxicity patterns described above. Compared to non-contaminated subjects, fish under MeHg-exposed treatments spent higher amounts of time in a habitat where they were incapable of mimicking the background, thus becoming vulnerable. It is also noteworthy that increased CO2 was relevant (although non-significant, p = 0.086) in explaining the time spent in that complex habitat, which falls in line with previous studies, where CO2 enhanced boldness/exploratory behavior (Hamilton et al. 2014; Munday et al. 2014). MeHg neurotoxicity is widespread to other important neurological systems, e.g., glutamatergic and dopaminergic systems (Castoldi et al. 2001), which may help explain this stronger behavioral detrimental effect. Given the aforementioned lower cerebellar AChE activity, the disruption of S. senegalensis fear- and exploration-related behaviors may be related to suppression of cerebellar neural activity (Frisch et al. 2000).
Lateralization at an individual level, i.e., absolute lateralization, was generally increased by temperature and decreased by MeHg, while CO2 effects depended on cofactors. The attenuating effect of temperature on loss of lateralization is credited to its rushing effects on the cerebral processes determining right–left choice (Domenici et al. 2014). More importantly, at the population level, MeHg exposure led to complete loss of control-registered left turning preference. This preference is likely related to the right-side upwards morphology of S. senegalensis (Bisazza et al. 1998), as turning left from pursuing predators protects the ventral area. Thus, besides inhibiting essential neurotransmitters and disrupting sensory acquisition, MeHg also disrupts judgment and inherent motor responses, causing loss of evolutionary-gained competitive traits vital to flatfish ecological fitness. Moreover, ocean warming and acidification played a key role in mediating MeHg accumulation and neurotoxicity. Studies focusing on how other vital ecological processes, such as mating and prey capture success, can be affected by MeHg toxicity under future ocean conditions are necessary. Climate change-related studies encompassing interactions among multiple stressors are scarce, but in high demand since they can help managers and policy makers to take proactive regulation targeting the reduction of CO2 and pollutant levels in the ocean.
The research leading to these results (as well as ES, VB, and RA) was funded from the European Union Seventh Framework Programme (FP7/2007-2013) under the ECsafeSEAFOOD project (Grant Agreement No. 311820). Sparos Lda, team for preparing the contaminated and non-contaminated feed. IPMA DivAV team from Olhão (aquaculture facilities) for providing juvenile sole specimens for the trials. The Portuguese Foundation for Science and Technology (FCT) supported the contract of RR and AM in the framework of the IF 2013 and IF2014 programs. FCT also funded PhD scholarships (ALM, SFRH/BD/103569/2014; VML, SFRH/BD/97633/2013; JRP, SFRH/BD/111153/2015), and by a FCT Postdoctoral scholarship (TR, SFRH/BPD/98590/2013).
Compliance with ethical standards
Conflict of interest
The authors declare no conflict of interest.
All procedures followed ARRIVE guidelines, overseen by the Portuguese National Competent Authority (Direção-Geral de Alimentação e Veterinária, DGAV). Moreover, personnel involved in this experiment had professional certification in handling and humane killing of laboratory animals by the Federation of European Laboratory Animal Science Associations (FELASA).