Environmentally relevant levels of four psychoactive compounds vary in their effects on freshwater fish condition: a brain concentration evidence approach

Experimental animals

The fish used in the experiment were hatchery-reared juvenile chubs obtained from a local fish supplier (Czech Fishery Ltd.; Czech Republic). A total of 320 similarly sized 0+ juvenile fish (five or ten months old for autumn and spring tests, respectively) were transported from the hatchery to the laboratory before testing each particular compound (i.e., A total of 1280 individuals were used for all compounds; the overall average standard length of the fish was 101 ± 13 mm (mean ± S.D.), and the overall average weight of the fish was 7.56 ± 3.46 g (mean ± S.D.). Two weeks prior to the start of the experiment, the fish were divided into four separate holding tanks (200 l with 80 randomly selected individuals each). Half of the identical tanks were intended for the exposed groups, and the second half were intended for the control groups. The fish were fed ad libitum on food pellets once a day and kept under a photoperiod of 12 h of light/12 h of darkness, maintaining the same regime to which they were accustomed in the hatchery. Two-thirds of the water volume was renewed with aged dechlorinated municipal tap water every day to habituate the fish to the experimental regime. The water temperature was controlled automatically using air conditioning throughout the whole experiment and held at an average of 20.8 ± 0.4 ^∘C (mean ± S.D.). Fish health, defined as normal appearance and behavior, including normal body position, movements and food intake (FAO, 1983), was monitored daily. Mortality during the experiment was also noted on a daily basis, but it was not primarily anticipated in the study design because environmentally relevant concentrations of the compounds were used in the present study, aiming to simulate the conditions in real-life polluted waters.

Pharmaceutical exposure

After acclimatization, fish in the exposed groups were separately exposed to the pollutants at an environmentally relevant concentration of 1 µg/L for 42 days, followed by a depuration period with tested compound-free water until the 56th day (two weeks of depuration). Individual psychoactive compounds (citalopram and sertraline from AK Scientific, Inc., USA, METH and tramadol from Sigma-Aldrich) were dissolved in ultra-pure water (AquaMax Basic 360 Series and Ultra 370 Series instrument, Younglin, purchased from Labicom, CR) to prepare a stock solution at a concentration of 10 mg/L. For the exposure bath, one mL of stock solution was added to every 10 L of aged dechlorinated municipal tap water in the holding tank. All environmental variables (i.e., the temperature, photoperiod and water renewal) were kept the same as those during the acclimatization period (see the Experimental Animals section). The compound was added during the water renewal to keep its concentration in the tanks at the required level. The control fish in the laboratory were kept under the same regime but without the psychoactive compound treatment. The tested compound concentration in the holding tanks was checked every other week during the exposure as well as during the depuration (32 water samples altogether) to verify the real concentrations in the exposed fish and exclude the possibility of cross-contamination in the control groups. The chemical analysis was performed according to Hossain et al. (2019), where isotopically labelled internal standards (citalopram-D₆, Toronto Research Chemicals; METH-D₅, Chiron Chemicals; sertraline-D₃ and tramadol-D₃, both from Lipomed) were added to filtered water (regenerated cellulose filter, 0.20 µm), and the samples were analyzed by liquid chromatography with tandem mass spectrometry (TSQ Quantiva, Thermo Fisher Scientific) for 10 min on a Hypersil Gold aQ column (50 ×  2.1 mm, 5 µm particles, Thermo Fisher Scientific) with heated electrospray ionization in positive mode (HESI+). The limits of quantification (LOQs) ranged from 0.011 to 0.021 µg/L for citalopram, from 0.010 to 0.012 µg/L for sertraline, from 0.0077 to 0.017 µg/L for methamphetamine, and 0.030 to 0.060 µg/L for tramadol. During the exposure, the average concentration of individual psychoactive compounds in the treated tanks was as follows: citalopram 1.3 ± 0.2 µg/L (mean ± S.D.); sertraline 0.23 ± 0.07 µg/L (mean ± S.D.); methamphetamine 1.1 ± 0.1 µg/L (mean ± S.D.); tramadol 0.99 ± 0.18 µg/L (mean ± S.D.). The concentrations in the tanks of the control fish as well as in the tanks of the exposed fish during depuration were below the limits of quantification.

Experimental design

The length and weight of eight randomly selected exposed and eight randomly selected control fish (four specimens per tank) were tested after 1, 7, 21 and 42 days of exposure and on the 56th day of the experiment (i.e., the day of exposure and two weeks after depuration). Thus, five trials with 80 specimens were conducted altogether. These specimens were killed according to valid law 246/1992, § 17 and the above-cited permit to analyze the psychoactive compound concentrations in their brains. The brains of freshly killed fish were dissected, weighed and stored frozen at −20 °C for later pharmaceutical content analyses. Before the analyses, the brain samples were defrosted and extracted according to the procedure described in the work of Grabicova et al. (2018). Briefly, isotopically labelled psychoactive compound and extraction solvent were added to approximately 0.1 g of brain tissue. The samples were homogenized, centrifuged, filtered and frozen for 24 h. Then, the extracts were centrifuged again, and the aliquots were analyzed using liquid chromatography with a high-resolution mass spectrometer (Q-Exactive; Thermo Fisher Scientific) on a Hypersil Gold aQ column for seven minutes in HESI+. The LOQs ranged from 0.12 to 0.34 ng/g wet weight (ww) for citalopram, from 0.10 to 0.44 ng/g ww for sertraline, from 0.16 to 0.43 ng/g ww for methamphetamine and from 0.19 to 0.69 ng/g ww for tramadol.

The analyzed concentration of psychoactive compounds was standardized by brain weight (ng/g). Please see Table 1 for details regarding the conditions of the chemical analyses.

Additional experiment (sertraline)

Data on food intake under sertraline exposure were collected regularly from the fourth to eighth week of the experiment from Friday to Sunday (three consecutive days every week, resulting in 15 trials). Food intake was not measured in the first three weeks because of the delayed therapeutic effects of SSRIs, which usually take two to three weeks (Rossi et al., 2008). Feeding experiments were conducted directly in the holding tanks to avoid causing distress to the fish and to prevent the results from being affected by the animal manipulation. Five grams of food pellets (i.e., 1% of the average weight of fish in the particular holding tank; food supply was adjusted according to mortality) were proffered to fish. The presence of non-ingested food in particular tanks was monitored for 15 min in one-minute intervals and noted every minute as a binomial value (0 –not all food pellets were ingested; 1 –all food pellets were ingested). A limit of 15 min was established based on the preliminary testing of food intake during regular feeding, which lasted from seven to eight minutes (i.e., the limit was double the regular feeding time).

Statistical analyses

‘Treatment’ was defined as the class variable distinguishing whether fish were exposed to sertraline. ‘Experimental phase’ distinguished between the exposure and depuration phases. ‘Compound in brain tissue’ expressed the relative concentration of particular psychoactive compound in the individual fish brain tissue. Fulton’s condition factor ‘K’ was evaluated as K = M/LS3, where M is mass and LS is the standard length. The compound concentration in brain tissue and Fulton’s condition factor were assigned for every sampled individual fish.

‘Food ingestion’ was assigned only for sertraline treatment as a count variable expressing number of one-minute intervals per feeding trial until all the food was eaten. Thus, it could acquire values from 1 to 15 (tanks where the all food was not eaten got a maximum score of 15 out of the 15 intervals). ‘Mortality’ was assigned as a count variable expressing a number of dead individuals in a particular holding tank per day. Food ingestion and mortality were assigned for every tank.

The statistical analyses were performed using the SAS software package (SAS Institute Inc., version 9.4, http://www.sas.com). The effects of all four psychoactive compounds (i.e., sertraline, citalopram, tramadol and methamphetamine) were tested separately.

Differences between Fulton’s condition factor ‘K’ of control and exposed fish and progress of ‘brain compound concentration’ of exposed fish during particular time samples (i.e., 1, 7, 21, 42, 56th day of the experiment) were determined with a separate t-tests and presented as boxplots of raw data. Generalized linear mixed models (GLMM; SAS function PROC GLIMMIX) with a Poisson distribution were used to analyze the variables ‘food ingestion’ and ‘mortality’. Particular holding tank and date of sampling were used as class random factors for both dependent variables. As feeding could be hypothetically influenced by fish weight, we included mean weight and its standard deviation obtained from individual condition measures as continuous random factors for analyses of ‘food ingestion’ as well. A separate model for each dependent variable was fitted, using interaction of ‘treatment’, and ‘experimental phase’ as the class explanatory variable. Generalized linear mixed model (GLMM; SAS function PROC MIXED) with a normal distribution was used to analyze Fulton’s condition factor ‘K’. Individual fish and date of sampling were used as a class random factors. Two models for dependent variable ‘K’ were fitted due to the ‘treatment’ and ‘compound in brain tissue’ variables overlap. Thus, first model contained ‘treatment’ fixed factor variable, while second model included ‘compound in brain tissue’ fixed factor variable. The significance of the exploratory variables was assessed using an F-test. Least-squares means (henceforth referred to and in bar charts presented as ‘adjusted means’ of model predictions) were subsequently computed for the particular classes of class variables. Differences between the classes were determined with a t-test, and a Tukey–Kramer adjustment was used for multiple comparisons. Association between the dependent variables ‘K’ and other continuous variable ‘compound in brain tissue’ was estimated by fitting a random factor model using PROC MIXED as described by Tao et al. (2002). With this random coefficient model, we calculated the predicted values for the dependent variable ‘K’ and plotted them against the continuous variable ‘compound in brain tissue’ by using the predicted regression lines in particular scatter charts. The degree of freedom was calculated using the Kenward–Roger method (Kenward & Roger, 1997).

Ethical note

All laboratory experimental procedures complied with valid legislative regulations (law no. 246/1992, § 19, art. 1, letter c) and were carried out with the relevant permission from the Ministry of Education, Youth and Sports of the Czech Republic (permit no. MSMT-1972/2016-5, registered by the Ministry of Education, Youth and Sports of the Czech Republic). The Departmental Expert Committee supervised the animal welfare in the present study. The research staff was trained according to valid legislative regulations on laboratory animal care, handling and welfare. The number of fish used corresponded to the reduction in laboratory animals used, and the use of animals could not be replaced by any other known method to obtain the relevant data.

Condition factor progression

There was no difference between the chub condition under the control and exposure treatments at the beginning of all experiments (1st day: citalopram, t = 0.71, P > 0.48, n = 16; Fig. 1A; methamphetamine t = 1.33, P > 0.2, n = 16; Fig. 1B; sertraline, t = 1.09, P > 0.29, n = 16; Fig. 1C; tramadol, t = 1.59, P > 0.14, n = 16; Fig. 1D). Citalopram (F_1,78 = 0.01, P > 0.93) and tramadol (F_1,78 = 0.96, P > 0.33) treatment did not cause any overall changes in chub condition when compared to the control. Additionally, comparison of condition during particular days did not show any significant differences (Fig. 1A, citalopram: 7th day, t = 0.57, P > 0.58, n = 16; 21st day, t = 1.79, P > 0.09, n = 16; 42nd day, t = 1.15, P > 0.27, n = 16; 56th day, t = 0.66, P > 0.52, n = 16; Fig. 1C, tramadol: 7th day, t = 0.21, P > 0.84, n = 16; 21st day, t = 0.96, P > 0.35, n = 16; 42nd day, t = 2.09, P > 0.06, n = 16; 56th day, t = 1.24, P > 0.23, n = 16). In contrast, methamphetamine (F_1,78 = 24.58, P < 0.0001; Fig. 2A) and sertraline (F_1,80 = 11.15, P < 0.0013; Fig. 2B) treatment induced significant overall condition alterations. Methamphetamine caused a condition increase (Adj. P < 0.0001; Fig. 2A), which was detectable from the 21st day of exposure (7th day, t = 0.64, P > 0.54, n = 16; 21st day, t = 2.52, P < 0.03, n = 16; 42nd day, t = 3.26, P < 0.01, n = 16) and persisted after the depuration phase (56th day, t = 9.51, P < 0.01, n = 16). In contrast, sertraline caused a decrease in condition (Adj. P < 0.0013; Fig. 2B), which was detectable from the 21st day of exposure (7th day, t = 0.54, P > 0.6, n = 16; 21st day, t = 2.31, P < 0.04, n = 16; 42nd day, t = 2.2, P < 0.04, n = 16) and disappeared during the depuration phase (56th day, t = 1.38, P > 0.19, n = 16).

Psychoactive substances in the fish brain

In all the control fish, particular psychoactive substances in the brain tissue were below the LOQ throughout the whole experiment, confirming that the control fish were unaffected in this respect. In contrast, fish exposed to environmentally relevant concentrations of psychoactive substances showed positive brain concentrations from the 1st day of exposure (Fig. 3A, citalopram: 1.7 ± 0.79 ng/g (mean ± S.D.); Fig. 3B, methamphetamine: 1.4 ± 0.62 ng/g (mean ± S.D.), Fig. 3C, sertraline: 590 ± 47 ng/g (mean ± S.D.); Fig. 3D, tramadol: 1.8 ± 0.78 ng/g (mean ± S.D.)). Citalopram, methamphetamine and tramadol brain tissue concentrations did not show any clear exposure-related progression, and they did not differ between the 1st and 42nd days of exposure (citalopram:, t = 0.53, P > 0.6, n = 16; methamphetamine: t = 0.45, P > 0.46, n = 16; tramadol: t = 0.34, P > 0.74, n = 16). After the depuration period (56th day), the brain concentration of tramadol was completely below the limit of quantification, and a positive response of citalopram was observed in only one of the eight individuals tested (mean 0.025 ± 0.070 ng/g (mean ± S.D.)), but methamphetamine was still detectable in all sampled fish (mean 0.82 ± 0.25 ng/g (mean ± S.D.)). In contrast to other compounds tested, the sertraline mean concentration in the exposed fish brain tissue subsequently increased during exposure (1st day—mean 590 ± 47 ng/g (mean ± S.D.); 7th day—mean 850 ± 120 ng/g (mean ± S.D.); 21st day—mean 1400 ± 660 ng/g (mean ± S.D.); 42nd day—mean 1800 ± 1000 ng/g (mean ± S.D.)) with significant differences between the 1st and 42nd days of exposure (t = 3.25, P < 0.01, n = 16). Despite this fact, the drug uptake was strongly individually driven as there were specimens reaching concentrations around 700 ng/g throughout the experiment. Low but still detectable levels of sertraline were also observed in all fish after depuration (56th day—mean 32 ± 17 ng/g (mean ± S.D.).

The direct link between methamphetamine (F_1,77 = 7.71, P < 0.0069; Fig. 4A) and sertraline (F_1,80 = 5.86, P < 0.0178; Fig. 4B) brain tissue concentrations and individual chub condition indicated that there is a strong influence of the individual amount of the specific drug received from the environmentally relevant water concentrations. The effect of methamphetamine and sertraline was opposite: increased methamphetamine concentrations were related to higher chub conditions (Fig. 4A) and vice versa for sertraline (Fig. 4B). In contrast, there was no significant relationship between tramadol (F_1,78 = 3.47, P > 0.07) and citalopram (F_1,78 = 0.00, P > 0.96) brain tissue concentration and fish condition, suggesting that the concentration-dependent effect was strongly compound-specific.

Mortality

Mortality varied across all control and most treatment groups (i.e., tramadol, citalopram and methamphetamine) from 0 to 8 individuals (i.e., 0–5%) throughout the whole experiment lasting 56 days. There was never found more than 1 dead individual during the daily controls. Dead fish showed no obvious signs of disease- or condition-related problems. In contrast, sertraline treatment caused mortality in 42 individuals (i.e., 26%). Up to 4 dead individuals were found during daily controls, and dead fish showed signs of lowered condition. The increased mortality of sertraline-treated fish was significantly higher than that of the control fish during exposure (F_3,230 = 6.22, P < 0.0004; Adj. P < 0.0008 Fig. 5A). This adverse effect disappeared during the depuration phase (Adj. P > 0.44).

Feeding under sertraline exposure (additional experiment)

Sertraline affected the feeding behavior of chub during the exposure phase. The food ingestion was lower in the exposed fish than in the control fish (F_3,8.571 = 5.49, P < 0.0218; Adj. P < 0.0206; Fig. 5B). In other words, sertraline caused a decrease in food consumption. The adverse effects disappeared during the depuration phase (Adj. P > 0.99).